Computer Science A Structured Approach Using C by Behrouz A. Forouzan Richard F. Gilberg

COMPUTE R SCIENCE
A Structured Programming Approach Using C

Behrouz A . Forouzan & Richard F. Gilberg
— THIRD EDITION
Operator Description Example * Assoc Pr
Identifiers amount
3.14159 N N /A 16
' Constants
Parenthetical Expressions (a + b)
[ ]. Array Index ary[i] N
ninction Call doIt(x, y) Y
Direct Member Selection str.mem N Left - Right 16
-> Indirect Member Selection ptr >mem - N
++ Postfix Increment • Decrement a++ Y
++ Prefix Increment • Decrement ++a Y
sizeof Size in Bytes sizeof(int) N
!
Ones Complement
Not
-a
la
N
N Right - Left 15
+ Plus • Minus +a N
& Address &a N
* Dereference / Indirection *ptr N
( )Type Cast (int)ptr N Right -Left 14
* /% Multiply • Divide • Modulus a * b N Left - Right 13
+ - Addition • Subtraction a + b N Left - Right 12
« » Bit Shift Left • Bit Shift Right a << 3 N Left - Right 1 1
< <= > >= Comparison a < 5 N Left - Right 10
ss ! = Equal • Not Equal N Left - Right 9
a == b
& Bitwise And a & b N Left - Right 8
A
Bitwise Exclusive Or a
A
b N Left - Right 7
Bitwise Or a b N Left - Right 6
&& Logical And a b ScSc N Left - Right 5
Logical Or a b N Left - Right 4
? : Conditional a ? x • Y * N Right - Left 3
= + = -= a = 5
*= / = % =
Assignment a %= b
>> =
A
<< =
a &= c Y Right - Left 2
& = = a |= d
> Comma a, b, c N Left - Right 1 i
* Side Effects ( Yes / No )
Precedence Table
>H
' . '
i -t X .U 6
\/o
*+ - - M i | J, -
' X AH .v A f\ £ w \/ dUe
Computer Science:
A Structured Programming
Approach Using C,
Third Edition
DATE DUE
Nov o 5 rm
LI 2o / H
1ft
March
/tprt I a ao\ ( p
2-6 ^6
AA 3 Z7
î y
Wt ~ 7 IS7 %
JW 3 0 2( 17 <15
i m 1 o : Qi8 Clb
DEC 17 BtS
fa
IAUG 0 ! 2012
»7
Singapore • Spam • United Kingdom •United States

Australia •Brazil • Japan • Korea •Mexico •
SAN JACIN1U COLLEGE LIBHAHY- SOUTH

Preface to Third Edition
basic principles ol program -

This text has two primary objectives: to teach the
_ curriculum for a CS 1 class and to teach the
ming as outlined in the ACM
of the C Language . Our text puts these objectives in the con -
basic constructs
have developed through
text of good software engineering concepts that
we
more than 30 years of experience in industry and academia .
Major Changes in Third Edition

This is a major change from the second edition . In addition to significant
changes in the presentation and material covered , it also updates the lan -
guage to the ISO / IEC 9899 ( 1999 ) , or as it is more commonly known , C 99 .
While we have made major changes, we have held true to our primary
objectives as outlined above. I he text remains a comprehensive introduction
to computer programming in a software engineering context.
The major changes are outlined in the following sections.
Standard C ( 1999 )
I he following topics, in alphabetical order, are explicitly discussed in the text .
• the Boolean type
• character set extensions ( including Unicode and wide-characters)
• complex arithmetic
• extended math library
• for statement declarations
• integer type extensions
• line comments
Extended Chapter Material

In addition to the changes required by the inclusion of the new C99 exten
sions , several chapters have been revised or reorganized .
-
-
A new section in Chapter 4, Inter Function Communication , ” provides a
better understanding of how functions communicate.
IV
Preface to Third Edition v
• The concept of “ incremental program development ’ is presented through

an example that starts in Chapter 4 and is extended in C hapter 5 . A sec -
ond example in Chapter 8 completes the discussion .
• The discussion of text and binary files has been reworked to present the
materials in a more logical, and less redundant, fashion .
• The presentation of structured types is changed to conform to C 99 and
common usage.
• Chapter 14 . “ Bitwise Operators , " has been extended and includes more
examples. In particular, we added several applications related to the
Internet and network programming.
• Chapter 15 , " Lists , " has also been extended . It now includes an overview
of the basic elements of data structures, including introductory discus-
sions of stack, queues, lists , trees , and graphs.
• Additional flowcharts and structure charts are included throughout the
text to provide better design considerations for some of the more complex
examples.
• Appendix D , " Numbering Systems," provides an expanded presentation of
the material.
• Appendix 1, " Pointers to void and to Functions ," has been expanded to
include a discussion of using pointer to void types.
• Appendix k . " Program Development , ” is a new discussion of program
deve lop men t co n cepts.
• Appendix L , “ Understanding Complex Declarations ," contains the discus-
sion on reading complex declarations, which has been removed from
Chapter 10.
A C Language Perspective
While C is a complex and professional language, our classroom experience
using the previous editions of this book has shown that beginning students can
easily understand it . We believe that if the language is put into a perspective
that allows the student to understand its design , C is not a difficult language.
There are two aspects of C that separate it from most other languages:
expressions and pointers. The concept of expressions, as used in C , is unique.
The first half of this text builds a firm understanding of expressions. We intro-
duce pointers only to the extent necessary to cover passing them to functions
7
in Chapter 4 and arrays in Chapter 8. Our experiences have shown that with
a firm grasp of the expression concept , much of the mystery of C disappears.
In Chapters 9 , we begin to develop the concept of pointers, which are
further developed in Chapter 10. Chapter 1 1 provides a discussion of C’s
approach to strings. Finally, Chapter 14 provides an extensive discussion of
bitwise operators, a subject that is needed to make the book complete.
The last chapter is a simple introduction to data structures. While not
all courses will have time to cover this chapter, those that do will give the
ce to Third Edition
students a head start into data structures

. Motivated students will find that
over the term break.
they can "get a leg up” on data structures
a rich set of references to
The appendices at the end of the text comprise
subjects required in the complete C language
. These will be of use to stu -
dents who go on to take other courses in the C language .
Features of the Book

Several features of this book make it unique and easy for beginning students
to understand .
Structure and Style

One of our basic tenets is that good habits are formed early. The corollary is
that bad habits are hard to break. Therefore, we consistently emphasize the
principles of structured programming and software engineering. Throughout
each chapter, short code snippets and programs are used to demonstrate
techniques.
Complete programs, generally found in the last section of the chapter,
are written in a well - document consistent style. As programs are analyzed,
style and standards are further explained . While we acknowledge that there
are many good styles, our experience has shown that if students are exposed
to one good style and adopt it , they will be better able to adapt to other
good styles.
Principle Before Practice

Whenever possible, we develop the principle ot a subject before we introduce
the language implementation . For example , in Chapter 5 we first introduce
the concept of logical data and selection and then we introduce the if . . . else
and suitch statements. This approach gives the student an understanding of
selection before introducing the nuances of the language.
Visual Approach
A brief scan of the hook will demonstrate that our approach is visual . 1 here
are more than 400 figures, plus more than 70 tables and 180 program exam -
pies. While this amount of material tends to create a large hook, the visual
approach makes it easy for students to follow the material .
Examples
While the programming examples vary in
style. Our experience
complexity, each uses a consistent
working with productional programs that live for 10 to
20 years convinced us that
readable and understandable programs are easier
to work with than programs
written in a terse, cryptic manner. For that reason ,
Preface to Third Edition vii
and to emphasize the structure of the language, we label the sections in a

function with comments. We consistently follow a style that places only one
declaration, definition, or statement on a line.
I he source code for these programs is available on the Course Technology
Web site ( see ’Web Site Support Materials” later in this Preface). Students are
able to run these programs and to explore related topics through modification
and experimentation.
Coding Techniques
Throughout the text we include coding techniques that make programs more
readable and often more efficient. For example, in the analyTsis of Program 5 - 4
you will find the following discussion:
Where do we check for greater than } The answer is that we
default the greater than condition to the outer else . . . . When cod -
ing a two - way selection statement , try to code the most probable
condition first; with nested selection statements, code the most
probable first and the least probable last.
These techniques are drawn from our extensive industry and classroom
experience.
Software Engineering
A discussion of software engineering principles concludes each chapter. Our
intent is not to replace a separate course in software engineering. Rather, we
firmly believe that by incorporating basic software engineering principles
early in their studies, students will be better prepared for a formal treatment
of the subject . Even more important, by writing well-engineered programs
from the beginning, students will not be forced to unlearn and relearn. I hey
will better understand software discussions in their subsequent classes.
While the software engineering sections are found at the end of each
chapter, they are most successfully taught by introducing them as the chapter
unfolds. Then, a short review at the end of the chapter summarizes the prin-
ciples that have been demonstrated during the lectures.
These sections are visually distinguishable from the rest of the chapter.
They have been set apart for several reasons. First, they are in reality a small
book within a book. While these sections contain important material, the book
stands on its own without them. You may, therefore, decide to cover the soft -
ware engineering sections as formal lecture topics or informally while the
chapter material is being covered. You may decide to assign them to the stu -
dent as additional reading, or, you may decide to exclude them entirely from
the course.
In general, software engineering sections directly or indirectly pertain to
the chapter material. Where they don’t, they discuss general software engi-
neering subjects, such as cohesion, coupling, and quality.
vjjj Preface to Third Edition
Pedagogical Material
first , it helps the stu -
End chapter material meets two pedagogical objectives:
dents to review or summarize what they have learned , and second , it tests the
students’ mastery of the chapter material .
Review Material
• Tips and Common Programming Errors points out helpful hints and
possible problem areas.
• Key Terms provides a list of the important terms introduced in the chapter.
• Summary contains a concise overview of the key points for students to
understand in the chapter.
Practice Sets
• Review Questions contain several multiple choice questions that are
similar to the questions found in the examination database. The answers
to the odd - numbered questions are included in the solutions on the
Course Technology web site .
• Exercises are short questions covering the material in the chapter. The
answers to the odd - numbered exercises are also included in the solutions
on the Course Technology web site.
• Problems are short coding problems, generally intended to be run on a
computer. They can usually be developed in two to three hours. Once
again , odd - numbered solutions are found on the web site .
• Projects are longer, major assignments that may take the average student
six to nine hours to develop.
Appendices and Cover Material

Ilu appendices are intended to provide quick reference material , such as the
Unicode Character Set , or to provide a review of material, such as numbering
systems , usually covered in a general computer class .
I he inside covers contain two important references
that are used con -
I / O Codes . —
tinually throughout the course the Precedence Table and the
Formatted
Web Site Support Materials

'1 <~ Hl" St lechn logy Web site ( http://www.course.com ) are
twn sets off materials
|
, one for the°professor and one for
the student .
1
Preface to Third Edition ix
Professor Materials
The professors materials include the solutions to all review questions, exer-
cises , and problems. Because the projects are more complex assignments lor
which there is no standard answer, we do not provide solutions for them .
In addition to the end - material solutions , there are three other sets ol
support materials: ( 1 ) Copies of all programs in the text. The programs have
been extensively tested on at least two different computers with different
operating systems. All of them conform to the C99 Standard.
( 2 ) PowerPoint materials covering objectives, figures, programs, impor -
tant points , and chapter summaries. The PowerPoint materials can be used as
is or edited to suit individual class needs. This flexibility allows professors to
argument the text materials with their own . It also allows them to rearrange
the materials to suit their individual needs and style .
( 3) New with the third edition is the availability of ExamView®. This
objective- based test generator lets the professor create paper, LAN , or Web-
based tests from testhanks designed specifically for this text . Using the
QuickTest Wizard , they can easily and quickly create true/false and multiple -
choice tests. It is also possible to add questions that cover supplemental
material provided by the professor.
Student Materials
I he student materials include the solutions to the odd - numbered review
questions , exercises , and problems. They also have access to the copies of the
programs used in the text
Acknowledgments
No book of this scope can he developed without the support ol many people.
Reviewers
To anyone who has not been through the process, the value ol peer reviews
cannot he appreciated enough . Writing a text rapidly becomes a myopic pro -
cess. The important guidance of reviewers who can stand hack and review the
text as a whole cannot be measured. We would especially like to acknowledge
the contributions of the reviewers of all three editions.
Stephen Allen , Utah State University
Mary Astone, / roy State University
Ali Behforooz, Towson State University
George Berrv, Wentworth Institute oj technology
Ernest Carey, Utah Valley State College
Ping Chu Chu , Fayetteville State University
Preface tojhird Edition
r
of San Francisco
Constance Conner, City College
Connecticut State University
John S. DaPonte, Southern University
Maurice L. Eggen , Trinity
/
Peter Gabrovsky, CSU Northridge

Robert Gann , Hartwick College
Henry Gordon , Kutztown University
Rick Graziani , Cabrillo College
Barbara Guillott , Louisiana State University
Jerzy Jaromczyk, University of Kentucky
John Kinio, Humber College
Roberta Klibaner, College of Staten Island
Joseph A. Konstan , University of Minnesota
Krishna Kulkarni, Rust College
John Lowther, Michigan Technological University
Mike Michaelson , Palomar College
—
Kara Nance, University of Alaska Fairbanks
Ali Nikzad , Huston -Tillotson College
Jo .Ann Parikh , Southern Connecticut State University
Mark Parker, Shoreline Community College
Savitha Pinnepalli, Louisiana State University
Oskar Reiksts, Kutztown University
Jim Roberts, Carnegie Mellon University
Ali Salenia, South Dakota State University
Larry Sells, Oklahoma City University
Shashi Shekhar, University of Minnesota
Robert Signorile, Boston College
Brenda Sonderegger, Montana State University
Deborah Sturm , College of Staten Island
Venkat Subramanian . University of Houston
John B. Iappen , University of Southern Colorado
Marc Thomasm , California State University, Bakersfield
John Irono, St . Michael 's College
K C. Wong, Fayetteville State University
Course Technology Staff

Our thanks to our editors , Alyssa Pratt , Senior Product Manager, and Mary
hranz , Senior Acquisitions Editor, for helping us produce the book. We are
also indebted to the Quality Assurance staff who diligently double checked
-
each c aP er anc program. Our thanks to Burt LaFountain , Serge Palladino,
^ ^ ’ .^
and Chris Sc river *
• ^^
W< U aS
? Mitchell
cially Sandra
° to diank the staff at GEX Publishing Services , espe-
, who ably guided the book through production .
BehrouzA. Forouzan
Richard F. Gilberg
1
Chapter 1 Introduction to Computers 1

1.1 Computer Systems 2
Computer Hardware 2
Computer Software 3
1.2 Computing Environments 5
Personal Computing Environment 5
Time-Sharing Environment 5
Client/Server Environment 6
Distributed Computing 7
1.3 Computer Languages 7
Machine Languages 8
Symbolic Languages 9
High-Level Languages 10
1.4 Creating and Running Programs 11
Writing and Editing Programs 12
Compiling Programs 12
Linking Programs 12
Executing Programs 13
1.5 System Development 13
System Development Life Cycle 13
Program Development 14
1.6 Software Engineering 22
1.7 Tips and Common Errors 24
1.8 Key Terms 24
1.9 Summary 25
1.10 Practice Sets 26
Review Questions 26
Exercises 28
Problems 28
29
Chapter 2 Introduction to the C Language
2.1 Background 30
2.2 C Programs 31
Structure of o C Program 31
Your First C Program 32
Comments 34
The Greeting Program 35
2.3 Identifiers 36
2.4 Types 38
Void Type 38
Integral Type 38
Floating-Point Types 41
Type Summary 42
2.5 Variables 42
Variable Declaration 43
Variable Initialization 44
2.6 Constants 4/
Constant Representation 47
Coding Constants 51
2.7 Input/Output 53
Streams 53
Formatting Input/Output 54
2.8 Programming Examples 68
Program Documentation 77
Data Names 78
Data Hiding 79
2.10 Tips and Common Programming Errors 81
2.11 Key Terms 82
2.12 Summary 82
Review Questions 84
Exercises 86
Problems 89
Projects 90
Chapter 3 Structure of a C Program 93

3.1 Expressions 94
Primary Expressions 95
Postfix Expressions 95
Prefix Expressions 97
1
Contents xiii
Unary Expressions 99
Binary Expressions 101
3.2 Precedence and Associativity 106
Precedence 107
Associativity 108
3.3 Side Effects 110
3.4 Evaluating Expressions 111
Expressions without Side Effects 111
Expressions with Side Effects 112
Warning 113
3.5 Type Conversion 114
Implicit Type Conversion 114
Explicit Type Conversion (Cast) 118
3.6 Statements 120
Statement Type 120
The Role of the Semicolon 124
Statements and Defined Constants 124
3.7 Sample Programs 125
KISS 135
Parentheses 135
User Communication 136
3.9 Tips and Common Errors 138
3.10 Key Terms 138
3.11 Summary 139
Review Questions 140
Exercises 142
Problems 144
Projects 146
Chapter 4 Functions 149

4.1 Designing Structured Programs 150
4.2 Functions in C 151
4.3 User -Defined functions 155
Basic Function Designs 156
Function Definition 162
Function Declaration 164
The Function Call 165
Function Examples 166
Contents xv
'
5.3 Multiway Selection 254
The switch Statement 255
The else-if 261
5.4 More Standard Functions 264
Standard Characters Functions 264
A Classification Program 266
Handling Major Errors 268
5.5 Incremental Development Part II 268
Calculator Design 269
Calculator Incremental Design 269
Dependent Statements 280
Negative Logic 281
Rules for Selection Statements 282
Selection in Structure Charts 283
5.8 Key Terms 286
5.9 Summary 286
Exercises 289
Programs 294
Projects 295
Chapter 6 Repetition 303

6.1 Concept of a loop 304
6.2 Pretest and Post-test Loops 304
6.3 Initialization and Updating 306
Loop Initialization 306
Loop Update 306
6.4 Event- and Counter-Controlled Loops 308
Event-Controlled Loops 308
Counter-Controlled Loops 308
Loop Comparison 309
6.5 Loops in C 309
The while Loop 310
The for Loop 315
The do . ..while Loop 319
The Comma Expression 323
6.6 Loop Examples 325
for Loops 325
Contents xvii
7.3 Standard Library Input/Output Functions 397

File Open and Close 398
7.4 Formatting Input/Output Functions 403
Stream Pointer 404
Format Control Strings 404
Input Formatting ( scantand fsconf ) 406
Output Formatting ( printf and fprintf) 418
File Sample Programs 425
7.5 Character Input/Output Functions 432
Terminal Character I/O 433
Terminal and File Character I/O 434
Character Input/Output Examples 436
Testing Files 442
Data Terminology 445
7.8 Key Terms 448
7.9 Summary 448
Exercises 451
Problems 453
Projects 455
Chapter 8 Arrays 459

8.1 Concepts 460
8.2 Using Arrays in C 463
Declaration and Definition 463
Accessing Elements in Arrays 464
Storing Values in Arrays 465
Precedence of Array References 469
Index Range Checking 470
8.3 Inter -Function Communication 473
Passing Individual Elements 473
Passing the Whole Array 475
8.4 Array Applications 481
Frequency Arrays 481
Histograms 482
Random Number Permutations 487
8.5 Sorting 490
Selection Sort 491
Bubble Sort 493
Insertion Sort 496
Testing Sorts 499
Sorts Compared 500
Sort Conclusions 500
8.6 Searching 501
Sequential Search 501
Binary Search 505
8.7 Two-Dimensional Arrays 509
Declaration 510
Passing a Two-Dimensional Array 516
8.8 Multidimensional Arrays 519
Declaring Multidimensional Arrays 520
8.9 Programming Example-Calculate Averages 522
First Increment: main 523
Second Increment: Get Data 524
Third Increment: Calculate Row Averages 526
Fourth Increment: Calculate Column Averages 528
Fifth Increment: Print Tables 528
Testing Sorts 532
Testing Searches 532
Analyzing Sort Algorithms 533
Analyzing Search Algorithms 535
8.11 Tips ond Common Programming Errors 537
8.12 Key Terms 538
8.13 Summory 538
Exercises 542
Problems 545
Projects 550
Chapter 9 Pointers 557

9.1 Introduction 558
Pointer Constants 558
Pointer Values 559
Pointer Variables 561
Accessing Variables Through Pointers 562
Pointer Declaration and Definition 563
Declaration versus Redirection S65
Initialization of Pointer Variables 566
9.2 Pointers for Inter-Function Communication 573
Passing Addresses 573
Functions Returning Pointers 575
Contents xix
9.3 Pointers to Pointers 576

9.4 Compatibility 578
Pointer Size Compatibility 579
Dereference Type Compatibility 580
Dereference Level Compatibility 583
9.5 Lvalue and Rvalue 583
Pointer Examples 585
Quality Defined 593
Quality Factors 594
The Quality Circle 597
Conclusion 598
9.8 Key Terms 600
9.9 Summary 600
Exercises 604
Problems 607
Projects 608
Chapter 10 Pointer Applications 611

10.1 Arrays and Pointers 612
10.2 Pointer Arithmetic and Arrays 614
Pointers and One-Dimensional Arrays 614
Arithmetic Operations on Pointers 617
Using Pointer Arithmetic 618
Pointers and Two-Dimensional Arrays 621
10.3 Passing an Array to a Function 623
10.4 Memory Allocation Functions 627
Memory Usage 627
Static Memory Allocation 628
Dynamic Memory Allocation 628
Memory Allocation Functions 628
Reallocation Of Memory ( realloc) 631
Releasing Memory ( fee) 631
10.5 Array of Pointers 633
10.6 Programming Applications 634
Selection Sort Revisited 634
Dynamic Array 639
Contents xxi
1
String to Number 704
String Examples 707
11.6 String/Data Conversion 712
String to Data Conversion 713
Data to String Conversion 714
11.7 A Programming Example — Morse Code 718
Program Design Concepts 728
Information Hiding 728
Cohesion 728
Summary 732
11.10 Key Terms 734
11.11 Summary 734
Exercises 737
Problems 740
Projects 742
Chapter 12 Enumerated, Structure, and Union Types 745

12.1 The Type Definition ( lypedef ) 746
12.2 Enumerated Types 746
Declaring an Enumerated Type 74 7
Operations on Enumerated Types 748
Enumeration Type Conversion 749
Initializing Enumerated Constants 749
Anonymous Enumeration: Constants 750
Input/Output Operations 752
12.3 Structure 752
Structure Type Declaration 753
Initialization 755
Accessing Structures 756
Operations on Structures 759
Complex Structures 764
Structures and Functions 774
Sending the Whole Structure 7 7 5
Passing Structures Through Pointers 7 7 8
12.4 Unions 782
Referencing Unions 782
Initializers 782
Contents xxiii
13.7 Key Terms 867

13.8 Summary 867
Exercises 870
Problems 876
Projects 877
Chapter 14 Bitwise Operators 881

14.1 Exact Size Integer Types 882
14.2 Logical Bitwise Operators 882
Bitwise and Operator 882
Bitwise Inclusive or Operator 884
Bitwise Exclusive or Operator 885
One's Complement Operator 886
14.3 Shift Operators 889
Rotation 894
14.4 Masks 896
Creating Masks 896
Using Masks 898
Payroll Case Study 907
Program Design Steps 908
Structure Chart Design 908
14.7 Key Terms 915
14.8 Summary 915
Exercises 917
Problems 920
Projects 921
Chapter 15 Lists 927

15.1 List Implementations 928
Array Implementation 928
Linked List Implementation 928
15.2 General Linear Lists 930
Insert a Node 930
Delete a Node 935
Locating Data in Linear Lists 938
Traversing linear Lists 941
Building a linear List 943
Build List 944
Remove a Node 945
Linear List Test Driver 946
15.3 Stacks 949
Stack Structures 950
Stack Algorithms 951
Stack Demonstration 954
15.4 Queues 958
Queue Operations 958
Queue Linked List Design 960
Queue Functions 961
Queue Demonstration 964
15.5 Trees 96 /
Basic Tree Concepts 96 /
Terminology 968
Binary Trees 969
Binary Search Trees 9 / 5
Binary Tree Example 9 / 8
15.6 Graphs 981
Graph Traversal 983
15. / Software Engineering 986
Atomic and Composite Data 986
Data Structure 98 /
Abstract Data Type 987
A Model for an Abstract Data Type 989
ADT Data Structure 990
15.8 Key Terms 992
15.9 Summary 992
15.10 Practice Set 994
Exercises 996
Problems 1000
Projects 1002
Appendix A Character Sets 1005

A .l Unicode 1006
Planes 1006
A.2 ASCII 1009
Some Properties of ASCII 1014
A. 3 Universal Encoding in C 1015
Contents xxv
Appendix B Keywords 1017
Appendix C Flowcharting 1019

C.l Auxiliary Symbols 1019
Flow Lines 1020
Connectors 1021
C.2 Primary Symbols 1021
Sequence 1022
Selection Statements 1024
Looping Statements 1026
Appendix D Numbering Systems 1033

D.l Computer Numbering System 1033
Decimal Numbers ( Base 10) 1034
Binary Numbers (Base 2) 1034
Hexadecimal Numbers (Base 16) 1036
Base 256 1037
A Comparison 1038
Other Conversions 1039
D.2 Storing Integers 1040
Unsigned Integers 1040
Signed Integers 1041
Overflow 1047
D.3 Storing Real Numbers 1049
Normalization 1049
Sign, Exponent, and Mantissa 1051
IEEE Standards 1051
Storing and Retrieving Algorithm 1051
Appendix E Integer and Float Libraries 1055

E. l limits . h 1055
E .2 float .h 1056
Appendix F Function Libraries 1059

F.l Function Index 1059
F.2 Type Library 1062
F.3 Math Library 1063
F.4 Standard 1/0 Library 1065
General 1/0 1066
Formatted 1/0 1066
xxvi Contents
Character I/O 1066

File I/O 1067
String I/O 1067
System File Control 1067
F.5 Standard Library 1067
Math Functions 1067
Memory Functions 1068
Program Control 1068
System Communication 1068
Conversion Functions 1068
F.6 String Library 1068
Memory Functions 1068
String Functions 1069
F.7 Time Library 1069
Appendix G Preprocessor Commands 1071

G.l File Inclusion 1072
G.2 Macro Definition 1072
Coding Defined Constants 1073
Macros that Simulate Functions 1074
Nested Macros 1077
Undefining Macros 1078
Predefined Macros 1078
Operators Related to Macros 1079
G.3 Conditional Compilation 1080
Two-Way Commands 1081
Multi-Way Commands 1085
Summary of Conditional Commands 1086
G.4 Other Commands 1087
Line Command 1087
Error Command 1088
Pragma Command 1089
Null Command 1089
Appendix H Command-Line Arguments 1091

H.l Defining Command-Line Arguments 1091
H.2 Using Command-Line Arguments 1093
Appendix I Pointers to Void ond to Functions 1095

1.1 Pointer to void 1095
1.2 Pointer to Function 1097
Contents xxvii
Appendix J Storage Classes and Type Qualifiers 1103

J.l Storage Classes 1103
Object Storage Attributes 1103
Storage Class Specifiers 1105
J.2 Type Qualifiers 1111
Constants 1112
Volatile 1113
Restricted 1114
Appendix K Program Development 1115

K .l Process 1115
First Case: A Simple File 1115
Second Case: Two Functions 1117
Third Case: Two Source Files 1118
Case 4 : Separate Compilation of Function 1120
K.2 Coding 1121
UNIX 1121
Other Platforms 1124
Appendix L Understanding Complex Declarations 1125
Glossary 1129
Index 1145
t
I
Introduction to Computers
and
Welcome to computer science! You are about to explore a wonderful
exciting world — a world that offers many challengin g and
In this chapter, we introduce you to the concepts of computer
exciting careers .
science ,
especially as they pertain to computer programming . You will study the con -
cept of a computer system and how it relates to computer hardware
and soft -
ware. We will also present a short history of computer programmi ng
languages so that you understand how they have evolved and how the C lan -
guage fits into the picture.
We will then describe how to write a program , lirst with a review ot the
tools and steps involved , and then with a review of a system development
methodology.
Objectives
To review basic computer systems concepts
To be able to understand the different computing environments and
their
components
To review the history of computer languages
To be able to list and describe the classifications of computer language
s
To understand the steps in the development of a computer program
To review the system development life cycle
1
2 Section 1.1 Computer Systems
1.1 Computer Systems

everywhere. Computers have become
Today computer systems are found
. But what is a computer? A computer is a
almost as common as Ltelevisions
_
The com - .
components: hardware and software
system made of two major the collection of
puter hardware is the physical
equipment. The software is
the hardware to do its job . Figure I - 1 rep -
programs (instructions) that allow
resents a computer system .
Computer
System
Hardware Software
FIGURE 1 - 1 A Computer System
Computer Hardware
The hardware component of the computer system consists of five parts:
input devices, central processing unit (CPU ), primary storage, output
devices, and auxiliary storage devices (Figure 1 - 2 ).
Primary Monitor
Keyboard CPU
Storage
Input Devices
t
Printer
Auxiliary Storage Devices Output Devices
FIGURE 1 - 2 Basic Hardware Components
I he input device is usually a keyboard

where programs and data are
entered into the computer. Examples of other input
a pen or stylus, a touch screen, or an
devices include a mouse,
audio input unit.
The central processing unit (CPU) is
responsible for executing
instructions such as arithmetic calculations, comparisons among
data, and
one, tyvo ,
movement of data inside the system . Today’s computers may have
main memory is a place
or more CPUs. Primary storage , also known as
,
where the programs and data are stored temporarily during processing. The
data in primary storage are erased when we turn off a personal computer or
when we log off from a time -sharing computer.
The output device is usually a monitor or a printer to show output . If the
output is shown on the monitor, we say we have a soft copy. If it is printed
on
the printer, we say we have a hard copy .
Auxiliary storage, also known as secondary storage, is used for both
input and output. It is the place where the programs and data are stored per-
manently. When we turn off the computer, our programs and data remain in
the secondary storage, ready for the next time we need them .
Computer Software
Computer software is divided into two broad categories: system software and
application software. This is true regardless of the hardware system architec -
ture. System software manages the computer resources. It provides the inter-
face between the hardware and the users hut does nothing to directly serve
the users’ needs. Application software, on the other hand , is directly responsi -
ble for helping users solve their problems. Figure 1 - 3 shows this breakdown
of computer software .
Software
System Application
Software Software
Operating System System General Application

Systems Support Development Purpose Specific
FIGURE 1 - 3 Types of Software
System Software
System software consists of programs that manage the hardware resources
of a computer and perform required information processing tasks . These pro -
grams are divided into three classes: the operating system , system support
,
and system developme nt .
The operating system provides services such as a user interface, file and
database access, and interfaces to communication systems such as Internet
protocols. The primary purpose of this software is to keep the system operat -
.
ing in an efficient manner while allowing the users access to the system
System support software provides system utilities and other operating
services . Examples of system utilities are sort programs and disk format
4 Section 1.1 Computer Systems
perlormance
consist of programs that provide
programs. Operating services and security monitors to protect the system
statistics for the operational
staff
and data. software ,
, system development
The last system software category lan -
convert programs into machine
includes the language translators that ensure that the programs are error-
guage for execution , debugging tools
to
engineering ( CASE ) systems that
are
free , and computer-assisted software
beyond the scope of this hook .
Application Software
classes: general-purpose software
Application software is broken into two
purpose software is purchased
and application -specific software. General-
he used lor more than one application .
from a software developer and can
word processors , database
Examples of general - purpose software include
management systems, and computer-aided
design systems. I hey are labeled
computing problems.
general purpose because they can solve a variety of user
purpose.
Application -specific software can he used only for its intended
requirements
A general ledger system used by accountants and a material
are examples of
planning system used by a manufacturing organization
only for the task for which
application -specific software. They can he used
they were designed ; they cannot he used for other generaliz ed tasks .
The relationship between system and application software is seen in

Figure 1 - 4 . In this figure, each circle represents an interface point. The inner
core is the hardware . The user is represented by the outer layer. To work with
the system , the typical user uses some form of application software. The
application software in turn interacts with the operating system , which is a
part of the system software layer. The system software provides the direct
interaction with the hardware. Note the opening at the bottom of the fig ure .
This is the path followed hv the user who interacts directly with the operating
system when necessary.
User
Software
W
FIGURE 1 - 4 Relationship Between System and Application Software

Chapter 1 Introduction toComputers 5
If users cannot buy software that supports their needs , then a custom -
developed application must be built . In today’s computing environment , one
of the tools used to develop software is the C language that we will he study-
ing in this text .
1.2 Computing Environments

In the early days of computers , there was only one environment : the main -
frame computer hidden in a central computing department . With the advent
of minicomputers and personal computers, the environment changed , result -
ing in computers on virtually every desktop. In this section we describe sev -
eral different environments.
Personal Computing Environment

In 1971 , Marcian E. Hoff , working for Intel , combined the basic elements of
the central processing unit into the microprocessor. This first computer on a
chip was the Intel 4004 and was the grandparent many times removed of
Intel ’s current system .
If we are using a personal computer, all of the computer hardware com -
ponents are tied together in our personal computer ( or PC for short ) . In
1
this situation , we have the whole computer for ourself ; we can do whatever
we want . A typical personal computer is shown in Figure 1 - 5 .
Monitor
Hard Drive
(Internal)
DVD ^
CD
Keyboard Mouse
FIGURE 1 - 5 Personal Computing Environment
Time- Sharing Environment

Employees in large companies often work in what is know n as a time-sharing
environment . In the time-sharing environment , many users are connected to
one or more computers. These computers may be minicomputers or central
mainframes. The terminals they use are often nonprogrammable, although
today we see more and more microcomputers being used to simulate termi -
nals. Also, in the time sharing environment , the output devices ( such as
- ol the
printers ) and auxiliary storage devices (such as disks) are shared by all
operating
1 . PC is now generally accepted to mean any hardware using one of Microsoft s Windows
’
systems, as opposed to Apple ’s Macintosh. We use it in the original , generic

sense meaning any per -
sonal computer.
6 Section 1.2 Computing Environments
is shared by many stu -
in which a minicomputer
users. A typical college lab
dents is shown in Figure 1 -
6.
Central Computer
Central Storage
Shared Printers
FIGURE 1 - 6 Time- sharing Environment
In the time-sharing environment , all computing must he done

hy the cen -
the central computer has many duties : It must
tral computer. In other words ,
control the shared resources ; it must manage the shared data and printing;
and it must do the computin g. All of this work tends to keep the computer
busy. In fact , it is sometimes so busy that the user becomes frustrated hy the
computers slow responses.
Client/Server Environment
A client/server computing environment splits the computing function
between a central computer and users’ computers. The users are given per-
sonal computers or workstations so that some of the computation responsi -
bility can he moved Irom the central computer and assigned to the
workstations. In the client/server environment , the users’ microcomputers or
workstations are called the client . The central computer, which may be a
powerful microcomputer, minicomputer, or central mainframe system , is
known as the server. Because the work is now shared between the users
computers and the central computer, response time and monitor display are
aster and the users are more productive. Figure 1 - 7 shows a tvpical client /
server environment.
Server
Shared Printers
Central Storage
Clients
FIGURE 1 -7 The Client / Server Environment
Distributed Computing
A distributed computing environment provides a seamless integration of
computing I unctions between different servers and clients. The Internet pro-
vides connectivity to different servers throughout the world . For example ,
eBay uses several computers to provide its auction service. This environment
provides a reliable, scalable, and highly available network. Figure 1 -8 shows a
distributed computing system.
Server
Internet
Server Clients
(browsers)
FIGURE 1 - 8 Distributed Computing
1.3 Computer Languages

To write a program for a computer, we must use a computer language. Over
the years computer languages have evolved from machine languages to natural
languages. A summary of computer languages is seen in Figure 1 - 9 .
8 Section 1.3 Computer longuoges
1940s 1950s
FIGURE 1 - 9 Computer Language Evolution
Machine Languages
only programming languages available
In the earliest days of computers , the
computer has its own machine language ,
were machine languages. Each
1 - 1 shows an example of a
which is made of streams of Os and fs. Program
two numbers and prints the
machine language. This program multiplies
results.
PROGRAM 1 - 1 The Multiplication Program in Machine Language

1 00000000 00000100 0000000000000 000
2 01011110 00001100 11000010 0000000000000010
3 11101111 00010110 0000000000000101
4 11101111 10011110 0000000000001011
5 11111000 10101101 11011111 0000000000010010
6 01100010 11011111 0000000000010101
7 l i i o i n i ooooooio u n i o n oooooooooooiom
8 11110100 10101101 11011111 0000000000011110
9 00000011 10100010 11011111 0000000000100001
10 l i i o i n i ooooooio u n i o n ooooooooooiooio o
11 0 1 1 1 1 1 1 0 1 1 1 1 0 1 0 0 1 0 1 0 1 1 0 1
12 1 1 1 1 1 0 0 0 1 0 1 0 1 1 1 0 1 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 1
13 ooooono IOIOOOIO u n i o n oooooooooonooo i
14 l i i o i n i ooooooio u n i o n oooooooooonoio o
15 01010000 11010100 0000000000111011
16 00000100 0000000000111101
The instructions in machine language must he in streams of O s and I s

because the internal circuits of a computer are made of switches, transistors,
and other electronic devices that can be in one ol two states: off or on . 1 he
off state is represented by 0; the on state is represented by 1 .
The only language understood by computer hardware is

machine language.
Symbolic Languages
It became obvious that few programs would be written if programmers contin-
.
ued to work in machine language In the early 1950s, Admiral Grace I topper, a
mathematician and naval officer, developed the concept of a special computer
program that would convert programs into machine language. I hese early
programming languages simply mirrored the machine languages using
symbols, or mnemonics, to represent the various machine language instructions.
Because they used symbols, these languages were known as symbolic
.
languages Program 1 - 2 shows the multiplication program in a symbolic
language. 2
PROGRAM 1 - 2 The Multiplication Program in Symbolic Language

1 entry main,~ m<r2>
2 subl 2 # 12 ,sp
3 jsb C$MAIN ARGS_
4 movab $CHAR STRING CON
5
6 pushal -8( fp)
7 pushal (r2)
8 calls # 2 ,SCANF
9 pushal -12( fp)
10 pushal 3( r2)
11 calls # 2 ,SCANF
12 mull3 - -
8( fp), 12( fp )
13 pusha 6( r 2 )
14 calls # 2 ,PRINTF
15 clrl rO
16 ret
Because a computer does not understand symbolic language, it must be

translated to the machine language. A special program called an assembler
translates symbolic code into machine language. Because symbolic languages
had to he assembled into machine language, they soon became known as
assembly languages. This name is still used today for symbolic languages
that closely represent the machine language of their computer.
Symbolic language uses symbols, or mnemonics, to represent the various

machine language instructions.
2. The symbolic language format is label , operators, and operands. There are no labels in this example,
so the left column is empty.
10 Section 1.3 Computer Languages
High- Level Languages

Although symbolic languages greatly improved programming efficiency, they
still required programmers to concentrate on the hardware that they were
using. Working with symbolic languages was also very tedious because each
machine instruction had to be individually coded . The desire to improve pro-
grammer efficiency and to change the focus from the computer to the prob-
lem being solved led to the development ol high - level languages .
High - level languages are portable to many different computers, allowing
the programmer to concentrate on the application problem at hand rather
than the intricacies of the computer. High -level languages are designed to
relieve the programmer from the details of the assembly language. High - level
languages share one thing with symbolic languages , however: I hey must be
converted to machine language. The process of converting them is known as
compilation .
I he first widely used high -level language, FORTRAN / was created by
John Backus and an IBM team in 1957; it is still widely used today in scien -
tific and engineering applications. Following soon after FORTRAN was
COBOL.4 Admiral Hopper was again a key figure in the development of the
COBOL business language.
C , with its standard libraries, is considered a high - level language used for
-
system software and new application code. Program 1 3 shows the multipli -
cation program as it would appear in the C language.
PROGRAM 1 - 3 The Multiplication Program in C

1 /* This program reads two integers from the
keyboard
2 and prints their product.
3 Written by:
4 Date:
5 */
6 # include <stdio.h >
7
8 int main ( void )
9 {
10 // Local Definitions
11 int numberl ;
12 int number2;
13 int result;
14
15 // Statements
continued
3. FORTRAN is an acronym for
4. COBOL is an acronym FORmula TRANslation.
for COmmon Business-Ori
rientcd Language.
PROGRAM 1 - 3 The Multiplication Program in C (continued )

16 scanf ( " %d " , & number 1 );
17 scanf ( "%d " , &number2);
18 result = numberl * number2;
19 printf ( " %d " , result );
20 return 0;
21 > // main
1.4 Creating and Running Programs

As we learned in the previous section, computer hardware understands a pro-
gram only it it is coded in its machine language. In this section, we explain
the procedure for turning a program written in C into machine language. The
process is presented in a straightforward, linear fashion, but you should rec -
ognize that these steps are repeated many times during development to cor -
rect errors and make improvements to the code.
It is the job of the programmer to write and test the program. There are
lour steps in this process: ( I ) writing and editing the program, ( 2 ) compiling
the program, ( 3 ) linking the program with the required library modules, and
( 4 ) executing the program . These steps are seen in Figure 1 - 10.
/tv ainclude < stdio.h>

Programmer int main ( void )
Text Editor {
} // main
Source
00110 100
10101 010
Compiler
01001 011
10110 100
is Object
/00110 1001 01010 io\

0011001101001001
Linker
Library h" ' \piooi 0111011010 oj Executable
Loader
Results
FIGURE 1 - 10 Building a C Program

12 Section 1.4 Creating and Running Programs
Writing and Editing Programs
reports, or write programs.

our system , we could use it to write letters create
,
text and program writing is that pro-

The main difference between processing
text processing is clone with
grams are written using lines of code, while most
characters and lines.
, but it is more
Our text editor could be a generalized word processor
often a special editor included with the compiler . Some of the features you
should look for in your editor are search commands to locate and replace
statements, copy and paste commands to copy or move statements
I mm one
part of a program to another, and formatting commands that allow you to set
tabs to align statements.
After we complete a program , we save our file to disk. I his file will be
input to the compiler ; it is known as a source file .
Compiling Programs
The code in a source file stored on the disk must he translated into machine
language . This is the job of the compiler. The C compiler is actually tw o sep -
arate programs: the preprocessor and the translator.
The preprocessor reads the source code and prepares it for the transla -
tor. While preparing the code , it scans for special instructions known as
preprocessor commands. These commands tell the preprocessor to look for
special code libraries, make substitutions in the code, and in other ways pre -
pare the code for translation into machine language . The result of prepro-
cessing is called the translation unit .
After the preprocessor has prepared the code for compilation , the trans-
lator does the actual work of converting the program into machine language.
I he translator reads the translation unit and writes the resulting object
module to a file that can then be combined with other precompiled units to
form the final program . An object module is the code in machine language .
Even though the output of the compiler is machine language
code, it is not
yet ready to run ; that is, it is not yet executable
because it does not have the
required C and other functions included .
Linking Programs
fomTofof theseSeVatT
omc th ’
functions
3
uam maJe UPof °f source> Unctions. However
C P 8
and?they
iS man ’ We write
processesând m.them..icai
, are a part our program
W L , " ,T that " ."elsewhere
libra
°
functions
" T1' f •
^
exist and
The linker assembles be attached
must our program
to .
all of these functio ns, ours and
final executable program . the system s , into our
Executing Programs
Once our program has been linked , it is ready for execution . To execute a pro-
gram , we use an operating system command , such as run , to load the pro-
gram into primary memory and execute it . Getting the program into memory
is the function of an operating system program known as the loader. It
locates the executable program and reads it into memory. When everything is
loaded , the program takes control and it begins execution . In todays inte -
grated development environments, these steps are combined under one
mouse click or pull-down window.
In a typical program execution , the program reads data lor processing,
either from the user or from a file . After the program processes the data , it
prepares the output. Data output can he to the user ’s monitor or to a file .
When the program has finished its job, it tells the operating system , which
then removes the program from memory.
1.5 System Development

We ’ve now seen the steps that are necessary to build a program . In this sec -
tion , we discuss how we go about developing a program. I his critical process
determines the overall quality and success of our program . II we carefully
design each program using good structured development techniques , our pro-
grams will be efficient , error-free, "* and easy to maintain .
System Development Life Cycle

Today’s large -scale, modern programming projects are built using a series of
interrelated phases commonly referred to as the system development life
cycle . Although the exact number and names of the phases differ depend -
ing on the environment , there is general agreement as to the steps that
must he followed . Whatever the methodology, however, today ’s software
engineering concepts require a rigorous and systematic approach to soft -
ware development .6
One very popular development life cycle is the waterfall model . Depend -
ing on the company and the type of software being developed , this model
consists of between five and seven phases. Figure 1 - 1 I is one possible varia -
tion on the model .
—
5. Many computer scientists believe that all programs contain at least one hug an undetected error —
that is just waiting to cause problems, given the right set of circumstances. Programs have run for
years without problems only to fail when an unusual situation occurs. Perhaps the most famous bug
was the one known as Y 2 K because it caused programs to fail on January 1.20( H ).
6. For a discussion of various models, see Software Engineering : A Practitioner ' s Approach , 5 th. ed . by
Roger S . Pressman , ( New York . NY: McGraw- Hill , 2001 ).
/
14 Section 1.5 System Development
Systems
Requirements
Analysis
Design
Maintenance
FIGURE 1 - 11 Waterfall Model
The waterfall model starts with systems requirements . In this phase, the
systems analyst defines requirements that specif ) what the proposed system
is to accomplish. The requirements are usually stated in terms that the user
understands. The analysis phase looks at different alternatives from a system ’s
point of view, while the design phase determines how the system will be built .
In the design phase, the functions of the individual programs that will make
up the system are determined and the design of the liles and/or the databases
is completed . Finally, in the fourth phase, code , we write the programs. This
is the phase that is explained in this book. After the programs have been writ -
ten and tested to the programmer’s satisfaction , the project proceeds to sys -
tem test . All of the programs arc tested together to make sure the system
works as a whole . The final phase, maintenance , keeps the system working
once it has been put into production .
Although the implication of the waterfall approach is that the phases
( low in a continuous stream from the first
to the last, this is not really the
case. Note the iteration as indicated by the backward - flowing arrows in
Figure 1 - 1 1 . As each phase is developed , errors and omissions will often be
found in the previous work. When this happens, it is necessary to go back to
the previous phase to rework it lor consistency and to analyze the
impact
caused hv the changes. Hopefully, this is a short rework. We are aware of at
least three major projects, however, that were in the code
and test phases
when it was determined that they could not be implemented
and had to be
canceled. When this happens, millions of dollars and years
of development
are lost .
Program Development
Program Development is a multistep process
stand the problem , develop a solution , that requires that we under-
write the program , and then test it.
When we are given the assignment to develop a program, we will be given a

program requirements statement and the design of any program interfaces .
We should also receive an overview of the complete project so that we will
.
understand how our part fits in to the whole Our job is to determine how to
take the inputs we are given and convert them into the outputs that have
been specified. This is known as program design . To give us an idea of how
this process works, let s look at a simple problem: Calculate the square foot -
age of a house. I low do we go about doing this ?
Understand the Problem

The first step in solving any problem is to understand it. We begin by reading
the requirements statement carefully. When we think that we fully under -
stand it , we review our understanding with the user and the systems analyst.
Often this involves asking questions to confirm our understanding.
For example, after reading our simple requirements statement, we should
ask several clarifying questions.
What is the definition of square footage?

How is the square footage going to be used?
for insurance purposes?
to paint the inside or outside of the house?
to carpet the whole house?
Is the garage included?

Are closets and hallways included?
Each of these potential uses requires a different measure. II we don’t

clarify the exact purpose— that is, if we make assumptions about how the out -
put is going to be used—we could supply the wrong answer.
As this example shows, even the simplest problem statements need clarifi -
cation. Imagine how many questions must be asked for a programmer to write
a program that will contain hundreds or thousands of detailed statements.
Develop the Solution

Once we fully understand the problem and have clarified any questions we
.
may have, we need to develop our solution Three tools will help us in this
task: ( 1 ) structure charts, ( 2 ) pseudocode, and ( 3) flowcharts. Generally, we
will use only two of them—a structure chart and either pseudocode or a
flowchart.
The structure chart is used to design the whole program. Pseudocode
and flowcharts, on the other hand, are used to design the individual parts of
the program. These parts are known as modules in pseudocode or functions
in the C language.
Structure Chart the functional

as a hierarchy chart , shows
A structure chart, also known structures consisting
flow through our program . Large
programs arc complex . I his task is
they must he carefully laid out
of many interrelated parts ; thus,
design engineer who is responsible lor the operational
similar to that of a
difference between the design built by
design of any complex item . The major programmers
a programmer and the design
built by an engineer is that the
the computer , whereas the engi -
product is software that exists only inside and touched .
neer’s product is something that can
be seen
going to break our program into
The structure chart shows how we are
logical steps; each step will he a separate module
. The structure chart shows
) of our program .
the interaction between all the parts ( modules
It is important to realize that the design
, as represented by the structure
. In this respect , it is like the archi-
chart, is done before we write our program
to build a house without a detailed set of
tect ’s blueprint . We would not start
and new pro-
plans. Yet one of the most common errors of both experienced
design is complete and
grammers alike is to start coding a program before the
fully documented . fully
This rush to start is due in part to programmers thinking they
’
understand the problem and also due to their excitement about solving new
a
problem . In the first case , what they find is that they did not fully understand
the problem. By taking the time to design the program , they will raise more
questions that must be answered and therefore will gain a better understand -
ing ol the problem .
An old programming proverb: Restet the temptation to code .

The second reason programmers code before completing the design is
just human nature. Programming is a tremendously exciting task , lo see our
design begin to take shape, to see our program creation working lor the lirst
time , brings a form of personal satisfaction that is a natural high .
In the business world , when we complete a structure chart design , we
-
will convene a review panel for a structured walk through of our program .
Such a panel usually consists of a representative from the user community,
one or two peer programmers, the systems analyst , and possibly a representa
live from the testing organization. In the review, we will walk our review team
through our structure chart to show how we plan to solve the objectives of
our program. The team will then offer constructive suggestions as to how to
improve our design.
The primary intent of the review is to increase quality and save time. The
earlier a mistake is detected , the easier it is to fix it . If we can
one or two problems with the structured walk- through ,
eliminate only
the time will he well
spent. Naturally, in a programming class, you will not
be able to convene a
lull panel and conduct a formal walk- through .
What you can do, how'ever, is
review your design with some of your classmates
and with your professor.
1
Looking again at our problem to calculate the square footage of a house,

let ’s assume the following answers to the questions raised in the previous
section.
.
1 The purpose of calculating the square footage is to install new floor cov-
ering.
. .
2 Only the living space will be carpeted The garage and closets will not he
considered.
.
3 The kitchen and bathrooms will he covered with linoleum; the rest ol the
house is to he carpeted.
With this understanding, we decide to write separate modules lor the
kitchen, bathroom( s), bedrooms, family room, and living room. We use sepa -
rate modules because the various rooms may require a dilferent quality ol lino-
leum and carpeting. The structure chart for our design is shown in Figure 1 - 12.
flooring
Cost
get print
calculate Report
Userlnfo
calc calc
Linoleum Carpeting
calc calc calc calc

Kitchen BathRooms BedRooms LivingAreas
calc calc
FamilyRoom DineLiving
FIGURE 1 - 1 2 Structure Chart for Calculating Square Footage
W hether you use a flowchart or pseudocode to complete the design ol

your program will depend on your experience, the difficulty ol the process you
are designing, and the culture and standards of the organization where you
.
are working We believe that new programmers should first learn program
design by flowcharting because a flowchart is a visual tool that is easier to
.
create than pseudocode On the other hand, pseudocode is more common
among professional programmers.
Pseudocode . Its purpose is to describe, in

, part program logic
Pseudo code is part English designed is to do. I his
what the program being
precise algorithmic detail , accomplish the task in sufficient detail so that
requires defining the steps to excels at this
computer program Pseudocode
they can he converted into a for determini ng the linoleum for the
type of precise logic. The pseudo code
is shown in Algorithm 1 - 1
.
bathroom
Pseudocode
are used
a loosely defined syntax and
to
English-like statements that follow
convey the design of an algorithm.
are easy to understand. A

Most of the statements in the pseudocode
telling the user what data are to he
prompt is simply a displayed message
three statements that follow it
entered. The uhile is a loop that repeats the
t 2 to tell when to stop.
and uses the number of bathrooms read in statemen
allows us to repeat a block of code.
Looping is a programming concept that
6. In this case , it allows us to process the informa -
We will study it in Chapter
tion for one or more bathrooms.
ALGORITHM 1 - 1 Pseudocode for Calculate Bathrooms

Algorithm Calculate BathRooms
1 prompt user and read linoleum price
2 prompt user and read number of bathrooms
3 set total bath area and baths processed to zero
4 while ( baths processed < number of bathrooms )
1 prompt user and read bath length and width
2 total bath area =
3 total bath area + bath length * bath width
4 add 1 to baths processed
5 bath cost - total bath area * linoleum price
6 return bath cost
end Algorithm Calculate BathRooms
Flowchart
A flowchart is a program design tool in which standard graphical symbols arc
used to represent the logical flow of data through a function. Appendix C
contains complete instructions lor creating flowcharts. If you are not f amiliar
with flowcharts, we suggest you read Appendix C now.
I In flowchart in Figure 1 - 13 shows the design for
*
calculating the area
and cost for the bathrooms . A few points merit
comment here. This flow-
chart is basically the same as the pseudocode. We begin with
prompts for the
Pnce °f the bnoleum and the number of bathrooms and read these two
pieces of data. ( As a general rule ,
flowcharts do not explicitly show standard
1
concepts such as prompts ) The loop reads the dimensions for each bath -
.
.
room Finally, when we know the total area, we calculate the price and return
to calcLinoleum.
O
/ bathProc < \ no
f calcbathRooms ^\ numbath w
7
yes
READ
/
( linoPrice) READ
( bathLngth,
bathWdth )
bathArea
READ
( numbath )
/
-*— zero
bathArea
bathArea +
bathLngth * bathWdth
bathProc
bathProc zero bathProc + 1
o
»
linoCost ^
bathArea ‘ linoPrice
RETURN
FIGURE 1 - 13 Flowchart for Calculate Bathrooms
Write the Program

Now it ’s time to write the program! But first, let ’s review the steps that
we ’ve used.
.
1 Understand the problem .
2. Develop a solution.
a. Design the program—create the structure chart.
h. Design the algorithms lor the program using either flowcharting or
pseudocode or both.
.
3 Write the program.
When we write a program, we start with the top hox on the structure
chart and work our way to the bottom. This is known as top-down implemen -
tation. You will find that it is a very easy and natural way to write programs,
especially if you have done a solid job on your design.
For our first few programs, there will he only one module, representing
the top hox of the structure chart . The first programs are quite simple and do
n . When .
we g f 4 however
rlumter
.
, we will begin to
„
subdivisio
.
tjme we wi
not require
write functions an our
struc urt
writing structured
^^
programs . In the
.
point out some more techn | d
good pseudocodede or flowchart s for the
meantime , concentrate on writing
main part of our programs.
Test the Program can be a

must test it. Program testing
After we’ve written our program, we developm ent . As the pro-
of program
very tedious and time-consuming part testing our program . In large
completel y
grammer, we are responsible for known as test engineers who
development projects, there are often specialist s
are responsible for testing the system
work together.
—
as a whole that is , for testing to make
sure all the programs
There are two types of testing: blackbox and
whitebox. Blackbox testing is
the user . Whitebox testing is the
done by the system test engineer and
responsibility of the programmer .
Blackbox Jesting
with -
Blackbox testing gets its name from the concept of testing the program
out knowing what is inside it
like a black
—
box
without
that we
knowing
can ’t see
how
into.
it works . In other words ,
the program is
Blackbox test plans are developed by looking only at the requirements
statement ( this is only one reason why it is so important to have a good set ol
requirements ) . The test engineer uses these requirements and his or her
knowledge of systems development and the user working environment to cre-
ate a test plan that will then he used when the system is tested as a whole. We
should ask to see this test plan before we write our program . The test engi -
neer 's plan will help us make sure we fully understand the requirements and
also help us create our own test plan .
Whitebox Jesting
Whereas blackbox testing assumes that the tester knows nothing about the
program , whitebox testing assumes that the tester knows everything about
the program . In this case , the program is like a glass house in which every -
thing is visible.
Whitebox testing is our responsibility. As the programmer, we know
exactly what is going on inside the program . We must make sure that every
instruction and even possible situation has been tested . That iis not a
simple task!
fmnf„ iTTi r, Wil' hClP” UuV g00d tCSt data but one thing we can do
f Witi S « ’P ans. We should star the,
st T ‘
n
ask oursetf
S gC
° " '
dt'sign stage. As we build your structure chart ,
““
and make a note of th m
hour later.
^^ WC eed l teSt f r
" °them an
we won ’t remember
°
When we are writing our flowcharts or pseudocode, we need to review

them with an eye toward test cases and make additional notes of the cases we
may need. Finally, while we are coding, we need to keep paper handy ( or a
test document open in our word processor ) to make notes about test cases
we need .
Except for the most simple program, one set of test data will not completely
validate a program.
When it is time to construct our test cases, we review our notes and orga -
nize them into logical sets. Except for very simple student programs, one set
of test data will never completely validate a program . For large -scale develop -
ment projects, 20, 30, or even more test cases may need to he run to validate
a program .
Finally, while we arc testing, we will think of more test cases. Again , write
them down and incorporate them into our test plan . After our program is fin -
ished and in production , we will need the test plan again when we make mod -
ifications to the program .
I low do we know when our program is completely tested ? In reality, there
is no way to know lor sure. But there are a few things we can do to help the
odds. While some of these concepts will not be clear until you have read
other chapters , we include them here for completeness.
1 . Verify that every line of code has been executed at least once . Fortu -
nately, there are programming tools on the market today that will do this
for us.
2. Verify that every conditional statement in our program has executed both
the true and false branches, even if one of them is null . ( See Chapter 5 . )
3. For every condition that has a range, make sure the tests include the first
and last items in the range, as well as items below the first and above the
—
last the most common mistakes in array range tests occur at the
extremes of the range. (See Chapter 8. )
4 . If error conditions are being checked , make sure all error logic is tested .
I bis may require us to make temporary modifications to our program to
force the errors. ( For instance, an input /output error usually cannot be
—
created it must he simulated . )
22 Section 1.6 Software Engineering
1.6 Software Engineering and use of sound engineering

ing is the establishment
.
Software engineer and that works on
methods and
*
principles
to obtain software that is reliable
, from the first interna
tional conference on
7 !
real machines This definition
was proposed 30 years
after the first computer
software engineering in 969 , art than a science. In
, software was more of
an
was built . During that period ing desenbes it as
treatments of programm
fact, one of the most authoritative . This three -volume senes, origi -
an art : The Art of Computer
Programming
1970 s , is con -
nally written by Donald E. Knuth in
the late 1960s and early
of many computer science concepts .
sidered the most complete discussion base for building reliable softwar e
Because the science and engineering 1960 s were a maze of
the 1950 s and
did not yet exist , programs written in
known as “ spaghet ti code .” It was not until Edsger Dijkstra wrote
complexity
a letter to the editor of the Communication
s of the ACM ( Association of Com -
Machinery ) 9 in 1968 that the concept
of structured programming
puting
began to emerge.
Dijkstra was working to develop algorithms
that would mathematically
d that any program could he \ \ i ittcn w ith
prove program accuracy. He propose
es, ( 2 ) the if ...else
only three constructs or types of instructions: 1 sequenc
( )
selection statement , and ( 3 ) the while loop . As we will
see , language develop -
and the switch in C . These
ers have added constructs, such as the for loop
ments to Dijkstra s basic constructs
additional statements are simply enhance
that make programming easier . Today , virtually all program ming languages
offer structure d program ming capabili ties.
Throughout this text we will he emphasizing the concepts of good soft -
ware engineering. Chief among them is the concept of structured program -
ming and a sound programming style. A section in each chapter will include a
discussion of these concepts with specific emphasis on the application of the
material in the chapter.
The tools of programming design have also changed over the years . In the
first generation of programming, one primary tool was a block diagram. This
tool provided boxes, diamonds, and other flowchart symbols to represent dif -
ferent instructions in a program . Each instruction was contained in a sepa -
rate symbol . Ibis concept allowed programmers to write a program on paper
and check its logic flow before they entered it in the computer.
With the advance of symbolic programming, the block diagram gave way
to the flowchart. Although the block diagram and flowchart look similar, the
flowchart does not contain the detail of the block diagram . Many instructions
are implied by the descriptive names put into the boxes ; for example, the read
7 . F. L Bauer, Technical University. Munich . Germany ( 1969).
1 r~
9 . Edsger W . Dijkstra , "Go To Slatemenl
no. 3 ( March 1968 ).
Considered Harmful r
Harmful ," Commun
'
,
2 < Third Edi ion '
ications of the ACM . vol . 11 ,

1
.
statements in Figure 1 - 1 3 imply the prompt Flowcharts have largely given
way to other techniques in program design, but they are still used today by
many programmers for working on a difficult logic problem .
Today’s programmers are most likely to use a high-level design tool such
as tight English or pseudocode. We will use pseudocode throughout the text
to describe many of the algorithms we will he developing.
Finally, the last several years have seen the automation of programming
through the use of computer-assisted software engineering ( CASE) tools.
These tools make it possible to determine requirements, design software, and
develop and test software in an automated environment using programming
workstations. The discussion of the CASE environment is beyond the scope
of this text and is left for courses in systems engineering.
I
24 Section 1.8 Key Terms
1.7 Tips and Common Errors

so you will be able to
1 . Become familiar with the text editor in your system
The time spent learning differ-
create and edit your programs efficiently.
ent techniques and shortcuts in a text editor will save
time in the future.
2. Also, become familiar with the compiler commands and keyboard short -
cuts. On most computers, a variety ol options are available to be used
with the compiler. Make yourself familiar with all ol these options.
3. Read the compiler’s error messages. Becoming familiar with the types ol
error messages and their meanings will he a big help as you learn C .
4. Remember to save and compile your program each time you make
changes or corrections in your source file. When your program has been
saved , you won ’t lose your changes if a program error causes the system
to fail during testing.
5. Run your program many times with different sets of data to be sure it
does what you want.
6. The most common programming error is not following the old proverb to
" resist the urge to code. ” Make sure you understand the
requirements
and take the time to design a solution before you start writing code .
1.8 Key Terms

application software operating system
application -specific software output device
assembly language personal computer ( PC )
auxiliary storage preprocessor
blackbox testing preprocessor commands
central processing unit ( CPU ) primary storage
client program development
client /server pseudocode
compilation secondary storage
compiler server
computer language soft copy
computer system
software
distributed environment source file
executable program
flowchart structure chart
symbolic language
general - purpose software
system development life cycle
hard copy
system development
hardware software
high - level language system software
system support software
input device
linker text editor
time-sharing environment
loader translation unit

machine language translator
main memory waterfall model
object module whitebox testing
1.9 Summary
A computer system consists of hardware and software .
Computer hardware consists of a central processing unit ( CPU ) , primary
memory, input devices, output devices , and auxiliary storage .
Software consists of two broad categories: system software and application
software.
The components of system software are the operating system , system sup-
port , and system development.
Application software is divided into general- purpose applications and
application -specific software.
Over the years , programming languages have evolved from machine lan -
guage , to symbolic language , and to high - level languages.
The C language is a high - level language .
The software used to write programs is known as a text editor.
The file created from a text editor is known as a source fde.
J The code in a source file must be translated into machine language using
the C compiler, which is made of two separate programs: the preprocessor
and the translator.
The file created from the compiler is known as an object module.
J An object module is linked to the standard functions necessary for running
the program by the linker.
A linked program is run using a loader.
The system development life cycle is a series of interrelated steps that pro -
vide a rigorous and systematic approach to software development.
To develop a program , a programmer must complete the iollowing steps:
a. Understand the problem .
b . Develop a solution using structure charts and either flowcharts or
pseudocode.
c. Write the program .
d . Test the program .
The development of a test plan starts with the design of the program and
continues through all steps in program development .
Blackbox testing consists primarily of testing based on user requirements.
26 Section 1.10 Practice Sets
, tests the program with full

Whitebox testing, executed by the programmer
knowledge of its operational weaknesses.
task, 'lou
Testing is one of the most important parts of your programming
testing the systems analyst and user are
are responsible for whitebox ;
responsible tor blackbox testing.
methods and
Software engineering is the application of sound engineering
principles to the design and development of application programs .
1.10 Practice Sets

Review Questions
.
1 Computer software is divided into two broad categories: system software
and operational software.
a. True
b. False
.
2 The operating system provides services such as a user interface, file and
database access, and interfaces to communications systems.
a. True
b. False
.
3 The first step in system development is to create a source program .
a .
True
b. False
.
4 The programmer design tool used to design the whole program is the
flowchart.
a. True
b. False
5. Blackbox testing gets its name from the
concept that the program is
being tested without knowing how it works.
a. True
b. False
6. YY Inch of the following iis a component (s) ol a
computer system?
a. Hardware
b. Software
c. Both hardware and software
d. Pseudocode
.
e System test
.
7 Which of the following is an example of application
a . Database rmanagement system software?
b. Language translator
.
c Operating system
d. Sort
e. Security monitor
8. Which of the following is not a computer language ?

a. Assembly/symbolic language
b. Binary language
c. High - level languages
d. Machine language
e. Natural language
9. The computer language that most closely resembles machine language is

a. Assembly/symbolic
.
b COBOL
c. FORTRAN
d. High-level
.
10 The tool used by a programmer to convert a source program to a machine
language object module is a
. Compiler
a
I). Language translator
c. Linker
d. Preprocessor
.
e Text editor
.
1 1 The contains the programmer ’s original program code .
a .
Application file
b. Executable file
c. Object file
.
d Source file
e. Text file
.
12 The series of interrelated phases that is used to develop computer soft -
ware is known as
a. Program development
b. Software engineering
c. System development life cycle
d. System analysis
e. System design
.
13 The is a program design tool that is a visual repre -
sentation of the logic in a function within a program.
a . Flowchart
b. Program map
c . Pseudocode
d. Structure chart
e . Waterfall model
,4 ensuring that all of its statements have

' 4'
Wen d" ,
The test that validates a program by ,,
is, b luiowing « l bow
ibo
^
a . Blackbox testing
b. Destructive testing
c. Nondestructive testing
d . System testing
e. Whitebox testing
Exercises
system .
15. Describe the two major components ol a computer
16. Computer hardware is made up of five parts.
List and describe them .
17. Describe the major differences between a time -

sharing and a client/
server environment .
18. Describe the two major categories of software .
19 . What is the purpose ol an operating system ?
20. Identify' at least two types of system software that you will use when you
write programs.
21 . Give at least one example of general - purpose and one example of
appIication -specific software .
.
22 List the levels of computer languages discussed in the text .
23. What are the primary differences between symbolic and high-level languages?
24. What is the difference between a source program and an object module?
25. Describe the basic steps in the system development life cycle.
26. What documentation should a programmer receive to be able to write a
program ?
27. List andexplain the steps that a programmer follows in writing a program .
28. Describe the three tools that a programmer may use to develop a program
solution.
29. What is meant by the old programming proverb, “ Resist the temptation
to code ”?
30. What is the difference between blackbox and whitebox testing?
31 . What is soltware engineering?
Problems
32. Write pseudocode for calcLivingAreas , Figure
1 - 12 , “Structure Chart for
C alculating Square Footage.
33. Create a flowchart for a routine task ,
such as calling a friend , that you do
on a regular basis.
34. Write pseudocode for the
flowchart you created in Problem 33.
Introduction to the C Language
In Chapter 1 , we traced the evolution of computer languages Irom the
machine languages to high - level languages. As mentioned , C ( the language
used exclusively in this hook ) is a high - level language. Since you are going to
spend considerable time working with the language, you should have some
idea of its origins and evolution .
In this chapter we introduce the basics ol the C language. You will write
your first program , which is traditionally known in C as the ' Hello World , ” or
“ Greeting" program . Along the way we will introduce you to the concepts ol
data types , constants , and variables. Finally, you will see two C library func-
tions that read and write data . Since this chapter is just an introduction to C ,
most of these topics are covered only in sufficient detail to enable you to
write your first program. They will he fully developed in luture chapters.
Objectives
To understand the structure of a C -language program
To write your first C program
To introduce the include preprocessor command
To be able to create good identifiers for objects in a program
To be able to list, describe, and use the C basic data types
To be able to create and use variables and constants in a program
To understand input and output concepts as they apply to C programs
To be able to use simple input and output statements
To understand the software engineering role in documentation, data nam -
ing, and data hiding
29
p
30 Section 2.1 Bockgtound
2.1 Background
It is considered a high -level lan-
C is a structured programming language
to concentrate on the problem at
guage because it allows the programmer
program wdl be using. VVhde
hand and not worry about the machine that the
many languages claim to be machine independent , C is one of the closest to
why it is used by software develop-
achieving that goal. That is another reason
hardware platforms.
ers whose applications have to run on many different
ALGOL, the first lan-
C, like most modern languages, is derived from
gained wide acceptance in the
guage to use a block structure. ALGOL never
United States, but it was widely used in Europe .
ALGOLs introduction in the early 1960s paved the way for the develop -
ment of structured programming concepts. Some ol the first
work was done
by two computer scientists, Corrado Bohm and Guiseppe Jacopini, who pub -
lished a paper in 1966 that defined the concept of structured programming .
Another computer scientist, Edsger Dijkstra, popularized the concept. Mis
letter to the editors of the Communications of the ACM ( Association of Com -
puting Machinery) brought the structured programming concept to the atten -
tion of the computer science community.
Several obscure languages preceded the development ol C. In 1967,
Martin Richards developed a language he called Basic Combined Program-
ming Language, or BCPL. Ken Thompson followed in 1970 with a similar
language he simply called B. B was used to develop the first version of UNIX,
one of the popular network operating systems in use today. Finally, in 1972,
Dennis Ritchie developed C, which took many concepts from ALGOL,
BCPL. This path, along with several others, is shown in Figure 2 - 1 .
ALGOL
BCPL ALGOL -68 Pascal
B ALGOL - W Modula - 2
Traditional
C Modula-3
ANSI/ISO
C99
C
FIGURE 2 - 1 Taxonomy of the C Language

Chapter 2 Introduction to the C Language 31
What is known as traditional C is this 1972 version of the language, as

documented and popularized in a 1978 hook by Brian W. Kernighan and
Dennis Ritchie. 1 In 1983, the American National Standards Institute (ANSI )
began the definition of a standard for C . It was approved in December 1989 .
In 1990 , the International Standards Organization ( ISO ) adopted the ANSI
standard . This version of C is known as C89 .
In 1995 , minor changes were made to the standard . I bis version is
known as C 95 . A much more significant update was made in 1999 . I he
changes incorporated into the standard , now known as C 99 , are summarized
in the following list .
1 . Extensions to the character type to support non - English characters

2. A Boolean type
3. Extensions to the integer type
4 . Inclusion of type definitions in the for statement .
5. Addition of imaginary and complex types
.
6 Incorporation of the C + + style line comment ( //)
We use the Standard C in this hook .
2.2 ( Programs
It s time to write your first C program ! This section will take you through all
the basic parts of a C program so that you will be able to write it .
Structure of a C Program
Every C program is made of one or more preprocessor commands, a global dec -
laration section , and one or more functions. The global declaration section
comes at the beginning of the program . We will talk more about it later, hut
the basic idea of global declarations is that they are visible to all parts of the
program.
The work of the program is carried out by its functions, blocks of code
that accomplish a task within a program . One, and only one, of the functions
must be named main . The main function is the starting point tor the pro -
gram . .All functions in a program , including main , are divided into two sec -
tions: the declaration section and the statement section. The declaration
section is at the beginning of the function . It describes the data that you will
1 . Brian Kernighan and Dennis Ritchie, The C Programming Language , 2 nd cd . ( Lnglewood

Cliffs, N .J .: Prentice Hall , 1989 ).
32 Section 2.2 C Programs
known as local decla-
the function . Declara tions in .a function are
be using in are visible only to the
rations (as opposed to global
declarations) because they
function that contains them .
the declaration section.
It contains the
The statement section follows do someth ing, such as add two
cause it to
instructions to the computer that in the form of statements,
.
numbers In C, these instructions
the
are
section .
written
which gives us the name for

Figure 2- 2 shows the parts of a
simple C program. We have explained
the preprocessor commands They . are special
everything in this program but the program for
tell how to prepare
instructions to the preprocessor that
it
of the preprocessor comma nds, and
compilation. One of the most important
one that is used in virtually all program s .
, is include I lie include command
from selected libraries known
tells the preprocessor that we need information
ming environments, it is almost
as header files. In today’s complex program
smalles t of program s w ithout at least one library
impossible to write even the
include command to tell C
function. In your first program, you will use one
to the monitor.
that you need the input /output library to write data
Preprocessor Directives
Global Declarations
int main ( void )
Local Declarations
Statements
} // main
Other functions as required.
FIGURE 2- 2 Structure of a C Program
Your First C Program

Your first C program will he very simple (see Figure 2- 3 ). It will have only one
preprocessor command, no global declarations, and no local definitions. Its
purpose w ill he simply to print a greeting to the user. Therefore, its statement
section will have only two statements: one that prints a
greeting and one that
stops the program.
Preprocessor directive to
include <stdio.h> include standard input/output
^ functions in the program.
int main ( void )

{
printf( " Hello World!\n" ); Hello World
return 0 ;
} // main
FIGURE 2 - 3 The Greeting Program
Preprocessor Commands
The preprocessor commands come at the beginning of the program. All pre-
processor commands start with a pound sign ( # ); this is just one of the rules
.
of C known as its syntax Preprocessor commands can start in any column,
but they traditionally start in column 1 .
The preprocessor command tells the compiler to include the standard
input /output library file in the program. You need this library file to print a
message to the terminal. Printing is one ol the input/output processes identi-
fied in this library. The complete syntax for this command is shown below.
# include <stdio.h>
The syntax of this command must he exact . Since it is a preprocessor

command, it starts with the pound sign. There can he no space between the
pound sign and the keyword, include . Include means just what you would
think it does. It tells the preprocessor that you want the library file in the
pointed brackets (< >) included in your program. The name ol the header file
is stdio.h . I bis is an abbreviation for " standard input /output header file."
mam
The executable part of your program begins with the function main, which is
identified hv the function header shown below. We explore the meaning ol
the function syntax in Chapter 4. For now, all you need to understand is that
int says that the function will return an integer value to the operating system,
that the function's name is main, and that it has no parameters (the parame-
ter list is void ). Note that there is no punctuation alter the function
header.
int main ( void )
J
34 Section 2.2 C Programs
.. , function one
—
w,,hi„ here ar
S.“EL
, ,
writingttfthTmonitor To invoke or vr ru v h» print
^
,
^
""' £',„l, ,. tor-

ihe atlua
you va » it . All function call » n p e
this case printf , followed by a parameter what you want dis
,
you simple program , the
marks t ••
““ „
° u llle enll f the „
played , enclosed in two double quote .
to the next line in the output
message tells the computer to advance 0 terminates the program
, return
The last statement in your program
. One last thing: The unction
and returns control to the operating system
main starts with an open brace ( { and
) terminates with a close brace ( } ) .
Comments
Although it is reasonable to expect that a good programme
r should he able to
read code, sometimes the meaning of a section of
code is not entirely clear.
who writes the
This is especially true in C . Thus , it is helpful if the person
the reader. Such comments
code places some comments in the code to help
on . The compiler ignores these
are merely internal program documentati
comments when it translates the program into executable code . Io identify a
comment , C uses two different formats : block comments and line comments .
Block Comment
A block comment is used when the comment will span several lines. We call
this comment format block comment . It uses opening and closing comment
tokens. A token is one or more symbols understood by the compiler that help
it interpret code . Each comment token is made of two characters that , taken
together, form the token ; there can be no space between them . The opening
token is /* and the closing token is * /. Everything between the opening and
closing comment tokens is ignored by the compiler. The tokens can start in
any column , and they do not have to be on the same line. The only require -
ment is that the opening token must precede the closing token . Figure 2 - 4
shows two examples of block comments.
/* This is a block comment that

covers two lines. V
/*
* * It is a very common style to put the opening token
** on
** programmers also like to put

asterisks at the beginning
of each line to clearly mark the
*/
. comment
FIGURE 2 - 4 Examples of Block Comments

Line Comment
The second format, the line comment , uses two slashes ( //) to identify a
comment. This format docs not require an end-of-comment token; the end of
the line automatically ends the comment . Programmers generally use this
format for short comments. The line - comment token can start anywhere on
the line. Figure 2 - 5 contains two examples ol line comments.
/ / This is a whole line comment
a = 5; // This is a partial line comment
FIGURE 2 - 5 Examples of Line Comments
Although they can appear anywhere, comments cannot be nested. In

.
other words, we cannot have comments inside comments Once the compiler
sees an opening block-comment token, it ignores everything it sees until it
finds the closing token. Therefore, the opening token ot the nested comment
is not recognized, and the ending token that matches the first opening token
is left standing on its own. This error is shown in Figure 2 - 6.
Inner ^
^
Comment not Closing
v Allowed / Token
/* /* */ */
Left On
Ignored Its Own
FIGURE 2 - 6 Nested Block Comments Are Invalid
The Greeting Program

Program 2 - 1 shows the greeting program just as we would write it. We have
included some comments at the beginning that explain what the program is
going to do. Each program we write begins with documentation explaining
the purpose of the program. We have also shown comments to identily the
declaration and statement sections ol our program. I he numbers on the left
in Program 2 - 1 and the other programs in the text are for discussion refer -
ence. T hey are not part of the program.
36 Section 2.3 Identifiers
PROGRAM 2- 1 The Greeting Program

1 /* The greeting
program. This program demonstrates
a simple C program.
2 some of the components of
3 Written by: your name here
4 Date: date program written
5 */
6 # include <stdio.h>
7
8 int main (void )
9 {
10 // Local Declarations
11
12 // statements
13
14 printf( "Hello World!\n" );
15
16 return 0;
17 } // main
2.3 Identifiers
One feature present in all computer languages is the identifier. Identifiers
allow us to name data and other objects in the program . Each identified
object in the computer is stored at a unique address. If we didn ’t have identi -
fiers that we could use to symbolically represent data locations, we would
have to know and use object’s addresses. Instead , we simply give data identifi -
ers and let the compiler keep track of where they are physically located .
Different programming languages use different syntactical rules to form
identifiers. In C , the rules for identifiers are very simple . The only valid name
symbols are the capital letters A through Z, the lowercase letters a through z ,
the digits 0 through 9 , and the underscore. The first character of the identi -
fier cannot he a digit .
Typically, application programs do not use the underscore for the first
character either, because many of the identifiers in the C system libraries
start with an underscore. In this way, we make sure that
our names do not
duplicate system names , which could become very confusing. The last rule is
that the name we create cannot he keywords. Keywords, also
known as
reserved words , include syntactical words, such as if and
xvhile . For a list of
the reserved words, see Appendix R .
Good identifier names are descriptive but short . To
make them short we
often use abbreviations.2 C allows names to he
up to 63 characters long. II
2. One way to abbreviate an identifier is to

remove any vowels in the middle of the word . For
exam l >le * student could be abbreviated stdnt .
the names are longer than 63 characters, then only the first 63 are used.
Table 2- 1 summarizes the rules for identifiers.
.
1 First character must be alphabetic character or underscore .
2. Must consist only of alphabetic characters, digits, or underscores.
3. First 63 characters of an identifier are significant.
4. Cannot duplicate a keyword.
TABLE 2 - 1 Rules for Identifiers

You might he curious as to why the underscore is included among the
possible characters that can be used for an identifier. It is there so that we
can separate different parts of an identifier. To make identifiers descriptive,
we often combine two or more words. When the names contain multiple
words, the underscore makes it easier to read the name.
An identifier must start with a letter or underscore: it may not have a space
or a hyphen .
Another way to separate the words in a name is to capitalize the first let -
ter in each word. The traditional method of separation in C uses the under -
score. A growing group of programmers, however, prefer to capitalize the first
letter of each word. Table 2 - 2 contains examples of valid and invalid names.
C is a case-sensitive language.
Two more comments about identifiers. Note that some of the identifiers
in Table 2 - 2 are capitalized. Typically, capitalized names are reserved lor pre -
processor - defined names. T he second comment is that C is case sensitive.
This means that even though two identifiers are spelled the same, if the case
of each corresponding letter doesn’t match, C thinks of them as different
names. Under this rule, num, Num, and NUM are three different identifiers.
Valid Names Invalid Name

a // Valid but poor style / / $ is illegal
$ sum
student_name 2names / / First char digit
_aSystemName sum-salary / / Contains hyphen
Bool / / Boolean System id stdnt Nmbr // Contains spaces
INT MIN / / System Defined Value int / / Keyword
TABLE 2 - 2 Examples of Valid and Invalid Names
38 Section 2.4 Types
2.4 Types can be applied on

A type defines a set of values
and a set ol operations that
compar ed to a computer
switch can he
those values. For example, a light ns can he
values, on and off. Only two operatio
type. It has a set of two off .
applied to a light switch : turn on and
turn
into four
C languag e has defined a set of types that can he divided
The derived as shown in
floating- point, and ,
general categories: void , integral ,
Figure 2 -7.
C Types
Floating-point Derived
Void Integral
Integer Real Imaginary Complex

Boolean Character
FIGURE 2 -7 Data Types
In this chapter, we concentrate only on the first three types. The derived
type will be discussed in future chapters.
Void Type
The void type , designated by the keyword void , has no values and no opera -
tions, Although having no values and operations might seem unusual , the void
type is a very uselul data type. For example, it is used to designate that a func-
tion has no parameters as we saw in the main function. It can also he used to
define that a function has no return value as we see in Chapter 4. It can also
be used to define a pointer to generic data as we will see in Chapter 9.
Integral Type
The C language has three integral types: Boolean , character, and integer,
Integral types cannot contain a fraction part; they are whole numbers.
Boolean
With the release of C99, the C language incorporated a
Boolean type.
Named after the French mathematician /philosopher George Boole, a Bool-
ean type can represent only two values: true or false .
Prior to C99 , C used
integers to represent the Boolean values: a o
nonzer number (positive or
negative ) was used to represent true, and
zero was
For backward compatibility, integers can still he used to represent false
,
used to represent Boolean
values; however, we recommend that new programs use the Boolean type.
The Boolean type, which is referred to by the keyword bool , is stored in mem -
ory as 0 ( false ) or 1 ( true ) .
Character
The third type is character. Although we think of characters as the letters of
the alphabet , a computer has another definition . To a computer, a character
is any value that can be represented in the computer’s alphabet , or as it is bet -
ter known , its character set . The C standard provides two character types:
char and wchar t .
char
wcharj
FIGURE 2 -8 Character Types
Most computers use the American Standard Code for Information Inter-
—
change ( ASCII pronounced “ ask-key ) alphabet . You do not need to memo-
rize this alphabet as you did when you learned your natural languages;
however, you will learn many of the special values by using them . The ASCII
code is included in Appendix A.
Most of the personal, mini - , and mainframe computers use 1 byte to
store the char data types. A byte is 8 bits. With 8 bits , there are 256 different
values in the char set . ( Note in Appendix A that ASCII uses only hall of these
possible values.) Although the size of char is machine dependent and varies
from computer to computer, normally it is 1 byte, or 8 bits.
If you examine the ASCII code carefully, you will notice that there is a
pattern to its alphabet that corresponds to the English alphabet . The first
32 ASCII characters and the last ASCII character are control characters.
They are used to control physical devices, such as monitors and printers, and
in telecommunication systems. The rest are characters that we use to com -
pose words and sentences.
All the lowercase letters are grouped together, as are all the uppercase
letters and the digits. Many of the special characters, such as the shift char-
acters on the top row of the keyboard , are grouped together, but some are
found spread throughout the alphabet.
What makes the letter u different from the letter x? In English , it is the
visual formation of the graphic associated with the letter. In the computer, it
is the underlying value of the hit configuration for the letter. The letter a is
binary 0110 0001. The letter x is 0111 1000 . The decimal values of these
two binary numbers are 97 and 120 , respectively.
To support non - English languages and languages that don’t use the
Roman alphabet , the C99 standard created the wide character type
40 Section 2.4 Types
( u’charj )
to »« < >
. Without going in> our for two - byte - chTrac
national standards, one or , YP traditional characters found in
ters. B„« h of those standard character. The otfgi -
ASCII ; that is, all extensions occur as the basic Latin character set . Gen-
nal ASCII characters are now known beyond the scope of an introductory
set is
erally speaking, the wide-character text.
programming text and is not covered in this
Integer
int , int , long int , and long long m . A short
sizes of the integer data type: short
, long int can be referred to as long and
,
int can also be referred to as short
long . C defines these data types so
long long int can be referred to as long , as shown in
be organized from the smallest to the largest
that they can
Figure 2- 9 . The type also defines the size of
the field in which data can be
stored. In C, this is true even though the size is machine
dependent and var -
ies from computer to computer.
short int
int
long int
long long int
FIGURE 2 -9 Integer Types
II we need to know the size ol any data type , C provides an operator,

sizeof , that will tell us the exact size in bytes . We will discuss this operator in
detail in C hapter 3. Although the size is machine dependent, C requires that
the following relationship always be true:
sizeof (short) < sizeof (int) £ sizeof (long) < sizeof (long long)
S ,gnedEach
’
integer size can be a signed or an unsigned integer. If the integer is
tben one bit must be u ^d for a signed (0 is plus, 1 is minus ) . The
unsigned integer can store a positive number that is twice as large as the signed
integer of the same size.* Table 2 -3 contains typical
values for the integer
08"''6’
hardware aCtUal
^
are dePende on the physical
"‘
3. For a complete discussion , see Appendix D ,
“ Numbering Systems.”
Type Minimum Value Maximum Value
short int 2 -32,768 32,767
int 4 -2,147,483,648 2,147,483,647
long int 4 -2,147,483,648 2,147,483,647
long long int 8 -9,223,372,036,854,775,807 9,223,372,036,854,775,806
TABLE 2 - 3 Typical Integer Sizes and Values for Signed Integers

To provide flexibility across different hardware platforms, C has a library,
. _
limits h, that contains size information about integers. For example, the mini-
mum integer value for the computer is defined as INT MIN , and the maxi -
_
mum value is defined as INT MAX . See Appendix E, Integer and Float
"
Libraries for a complete list of these named values.

"
Floating- Point Types

The C standard recognizes three floating-point types : real, imaginary, and
complex. Like the limits library for integer values, there is a standard library,
flout . h , for the floating-point values ( see Appendix E. " Integer and Float
Libraries”). Unlike the integral type, real type values are always signed.
Real
The real type holds values that consist of an integral and a fractional part,
.
such as 43.32 The C language supports three different sizes of real types:
flout , double, and long double . As was the case for the integer type, real num-
bers are defined so that they can be organized from smallest to largest. I he
relationship among the real types is seen in Figure 2 - 10.
float
double
long double
FIGURE 2- 10 Floating -point Types
Regardless of machine size, C requires that the following relationship

must he true:
sizeof ( float) < sizeof (double) < sizeof ( long double)

42 Section 2.5 Variables
Imaginary Type extensively in

imaginary number is used
C defines an :imaginary type. An number multi -
^ . ^*
_‘ number is a real
mathematics and engineering.
a
!!j maginary tvpe, like the real type,
P
'
can ,
double imaginary.
r
beyonhS «e” m (h lm m *
the imaginary type yet and the
•"*
Most C implementations do not: support . We mention them here
functions to handle them are not pa rt of the standard
because the imaginary type is one of
-the components of the complex type.
Complex
by most compilers. A com -
C defines a complex type, which is implemented
plex number is a combination of a real and an imaginary ' number. 1 he com -
of three different sizes : float complex
plex type, like the real type, can be
complex . The size needs to be the same in both
double complex , and long long
the real and the imaginary' part . We provide two program examples that use
com plex numbers at the end ot this chapter.
Type Summary
A summary of the four standard data types is shown in I able 2 - 4.
Category
Void Void
a void
C Implementation
Integral Boolean bool

Character char , wcharj
Integer short int, int, long int, long long int
Floating-Point Real float, double, long double
Imaginary float imaginary, double imaginary, long
double imaginary
Complex float complex, double complex, long double
complex
TABLE 2 - 4 Type Summary
2.5 Variables
Variables are named memory locations that
have a type , such as integer or
character, which is inherited from their type.
that a variable may contain the operations
The type determines the values
that may he used with its values.
Variable Declaration
Each variable in your program must be declared and defined. In C, a declara -
tion is used to name an object, such as a variable. Definitions are used to
.
create the object With one exception, a variable is declared and defined at
the same time. The exception, which we will see later, declares them first and
then defines them at a later time. For variables, definition assumes that the
declaration has been done or is being done at the same time. W hile this dis -
tinction is somewhat of an oversimplification, it works in most situations. We
won’t worry' about the exception at this time.
When we create variables, the declaration gives them a symbolic name
and the definition reserves memory for them. Once defined, variables are
used to hold the data that are required by the program lor its operation. Gen -
erally speaking, where the variable is located in memory is not a program -
mer ’s concern; it is a concern only of the compiler. From our perspective, all
we are concerned with is being able to access the data through their symbolic
names, their identifiers. The concept of variables in memory is illustrated in
Figure 2 - 11 .
Variable' s Variable's
. type „ identifier
char code;
int i ;
long long national debt ;
float payRate ;
double pi;
Program
FIGURE 2 - 11 Variables
A variable’s type can he any of the data types, such as character, integer ,
or real . The one exception to this rule is the type void : a variable cannot be
type void .
To create a variable, we first specify the type, which automatically speci -
fies it size ( precision ), and then its identifier, as shown below in the definition
of a real variable named price ol type float.
float price ;
Table 2 - 5 shows some examples of variable declarations and definitions .

As you study the variable identifiers, note the dillerent styles used to make
them readable. You should select a style and use it consistently. We prefer the
use of an uppercase letter to identify the beginning of each word after the
first one, although we do include examples using underscores.
^
bool
short
long
fact ;
maxltems ;
long national debt ;
_ / / Word separator : Capital
/ / word separator : underscore
/ / Word separator : Capital
float payRate ;
double tax;
;
float complex voltage
—
/ / Poor style see text
char
int
code , kind ;
a, b; —
/ / Poor style see text
TABLE 2 - 5 Examples of Variable Declarations

and Definitions
type to be defined in one state-
C allows multiple variables of the same
ment The last two entries in Table 2 - 5 use this format. Even though many
style.
professional programmers use it, we consider it to be poor programming
if they are defined on sepa -
It is much easier to find and work with variables
work a little harder , but the resulting code
rate lines. This makes the compiler
is no different. This is one situation in which ease of
reading the program and
programmer efficiency are more important than the convenience of coding
multiple declarations on the same line.
Variable Initialization
We can initialize a variable at the same time that we declare it hv including an
initializer. When present , the initializer establishes the first value that the
variable will contain. To initialize a variable when it is defined , the identifier
is followed by the assignment operator 4 and then the initializer, which is the
value the variable is to have when the function starts. This simple initializa -
tion format is shown below.
int count = 0;
Every time the function containing count is entered, count is set to zero.
Now, what will be the result of the following initialization ? Are both count
and sum initialized or is only sum initialized ?
int count , sum = 0;
The answer is that the initializer applies only

to the variable defined
immediately before it. Therefore, only sum is initialized! If
you wanted both
variables initialized , you would have to provide
two initializers.
int count = 0 , sum = 0;
4 . The assignment operator is

the equal sign ( =).
Again, to avoid confusion and error, we prefer using only one variable
definitionto a line. The preferred code in this case would he
int count = 0;
int sum = 0;
Figure 2 - 12 repeats Figure 2 - 11, initializing the values in each of the

variables.
char code = 'B ‘ ; [ B|code

int i = 14; 14 i
long long natl _debt = 1000000000000 ; 1000000000000 natl debt
float payRate = 14.25; 14.25|payRate
double Pi = 3.1415926536 ; 3.1415926536 pi
Program Memory
FIGURE 2- 12 Variable Initialization
It is important to remember that with a few exceptions that we will see

later, variables are not initialized automatically. When variables are defined,
they usually contain garbage ( meaningless values left over from a previous
use), so we need to initialize them or store data in them ( using run- time
statements ) before accessing their values . Many compilers display a warning
message when a variable is accessed before it is initialized.
.
When a variable is defined, it is not initialized We must initialize any variable
requiring prescribed data when the function starts.
One final point about initializing variables when they are defined:
Although the practice is convenient and saves you a line of code, it also can
lead to errors. It is better, therefore, to initialize the variable with an assign-
ment statement at the proper place in the body ol the code. This may take
another statement, but the efficiency of the resulting program is exactly the
same, and you will make fewer errors in vour code.
EXAMPLE 2 - 1 Print Sum

At this point you might like to see what a more complex program looks like.
As you read Program 2 - 2, note the blank lines to separate different groups ol
code I bis is a good technique for making programs more readable. You
.
should use blank lines in your programs the same way you use them to sepa -
rate the paragraphs in a report.
Ill
PROGRAM 2- 2 Print Sum of Three Numbers

and prints the sum of
1 /* This programi calculates user at the keyboard.
2 three -numbers input by the
3 Written by:
4 Date:
5 */
7
8 int main (void )
9 {
11 int a ;
12 int b;
13 int c ;
14 int sum ;
15
16 // Statements
printf("\nWelcome. This program adds\n ) ;
"
17
numbers \n " );
18 printf("three numbers. Enter three
19 printf(" in the form: nnn nnn nnn < return >\ n" );
20 scanf( " %d %d % ,
d " &a , & b , &c );
21
22 // Numbers are now in a, b , and c. Add them ,
23 sum = a + b + c ;
24
25 printf("The total is: %d\n\n" , sum);
26
27 printf("Thank you. Have a good day.\n" );
28 return 0;
29 } // main
Results:
Welcome. This program adds
three numbers. Enter three numbers
in the form: nnn nnn nnn <return>
11 22 33
The total is: 66
Thank you. Have a good day.
Program 2- 2 Analysis Study the style of this program carefully. First, note
how we start with a welcome
message.
This program contains three different processes. First it reads three numbers. The
code to read the numbers includes the printed instructions and a read ( scanf ] statement.
The second process adds the three numbers. While this process consists of only a com-
ment and one statement, we separate it from the read process. This makes it easier for
the reader to follow the program. Finally, we print the result. Again, the print process is
separated from the calculate process by a blank line.
2.6 Constants
Constants are data values that cannot be changed during the execution of a
program. Like variables, constants have a type. In this section, we discuss
Boolean, character, integer, real, complex, and string constants.
Constant Representation
In this section, we show how we use symbols to represent constants In the .
next section, we show how we can code these constants in our program .
Boolean Constants
A Boolean data type can take only two values. Therefore, we expect that we
have only two symbols to represent a Boolean type. The values are true and
false. As we mentioned before, a Boolean value can have only one of the two
values: 0 ( false ) and 1 ( true ). We use the constant true or false in our program.
To do so, however, requires that we include the Boolean library; stdbool .h .
Character Constants
Character constants are enclosed between two single quotes ( apostrophes).
In addition to the character, we can also use a backslash ( \ ) before the char -
acter. The backslash is known as the escape character. It is used when the
character we need to represent does not have any graphic associated with it —
that is, when it cannot he printed or when it cannot he entered from the key-
hoard. The escape character says that what follows is not the normal charac -
ter hut something else. For example, ' \ n ' represents the newline character
( line feed ). So, even though there may he multiple symbols in the character
constant , they always represent only one character.
A character constant is enclosed in single quotes.
Wide-character constants are coded by prefixing the constant with an L ,

as shown in the following example.
L'x
48 Section 2.6 Constants
the character set
constant comes from
The character in .the character . Most computers use the ASCII char-
nufacturer
supplied by the hardware ma calied , the ASCII alphabet. The ASC II charac-
’
acter set , or as it is sometimes

A.
ter set is shown in Appendix we can refer to them
character values so that
C has named the critical characters use the escape character fob
symbolically. Note that these control
lowed by a symbolic character
. They are shown in Table 2-6.
Symbolic Name
ASCII Character
\0 *
null character
' \a
alert (bell) \b
backspace \t
horizontal tab \n
newline \v
vertical tab \f
form feed \r
carriage return V•
single quote \ •• •
double quote w
backslash
TABLE 2 -6 Symbolic Names for Control Characters
Integer Constants
Although integers arc always stored in their binary form , they are simply
coded as we would use them in everyday life. Thus, the value 1 5 is simply
coded as 15.
II we code the number as a series ol digits, its type is signed integer, or
signc ong integer il the number is large. We can override this default by
specifying unsigned ( u or u ) , and long ( l or L ) or long long ( n or LL ) , after
the number I he codes may be combined and may be coded in any order.
. ote that there is no way to specify a short int constant. When we omit the
.
sulhx on a hterai , lt defaults to int While both upper- and lowercase codes
r C0 1men l lal ou always use uppercase to avoid confusion
*
( esnechllv
^" >
able 2- / shows several examples of integer
typical lor a personal computer.
constants The default types are
Representation Value Type
+ 123 123 int

-378 -378 int
-32271 L -32,271 long int
76542LU 76,542 unsigned long int
12789845 LL 12,789, 845 long long int
TABLE 2 - 7 Examples of Integer Constants

Real Constants
The default form for real constants is double . If we want the resulting data
type to be flout or long double, we must use a code to specify the desired
data
type. As you might anticipate , f and F are used for float and 1 and L are used
for long double . Again, do not use the lowercase / for long double; it is too eas -
ily confused with the number 1.
Table 2 - 8 shows several examples of real constants.
Representation Value Type
0. 0.0 double
.0 0.0 double
2.0 2.0 double
3.1416 3.1416 double
-2 .Of -2.0 float
3.1415926536 L 3.1415926536 long double
TABLE 2 - 8 Examples of Real Constants
Complex Constants
Although we do not discuss the imaginary constants, we need to talk about
complex constants that are widely used in engineering.
Complex constants are coded as two parts, the real part and the imagi -
nary part, separated by a plus sign. The real part is coded using the real
lor -
( ) the
mat rules. The imaginary part is coded as a real number times *
.h ) is
imaginary constant ( _Complex _l). If the complex library ( complex
included, the imaginary constant can he abbreviated as I. Examples are
shown in Table 2 - 9 .
Value Type
Representation
12.3 + 14.4 * I
,
12.3 + 14.4 * H ) /
2 double complex
14F + 16F * I 14 + 16 * H ) 1 / 2 float complex
1.4736L+ 4.56756L * I
1 2
1.4736 + 4.56756 * H ) / long double
complex
TABLE 2 - 9 Examples of Complex Constants

The default form for complex constants is double . II we want the result -
ing data type to be float or long double , we must use a code to specif \ the
.
desired data type As you might anticipate, f and F arc used for float and 1
and L are used for long double. Again, do not use the lowercase / for long dou -
ble; it is too easily confused with the number 1 .
The two components of a complex constant must be of the same precision ,

that is, if the real part is type double, then the imaginary time must also be
type double.
Table 2 -9 shows several examples of complex constants. Note that we use

the abbreviated form for the imaginary part.
String Constants
A string constant is a sequence of zero or more characters enclosed in double
.
quotes You used a string in your first program without even knowing that it
-
was a string! Look at Program 2 1 to see if you can identify
the string.
Listed in Figure 2- 13 arc several strings, including the one from
Program 2 - 1 .
The first cxample, an empty string, is simply two double
quotes iin succes-
sion. The second example, a string containing only
the letter h, differs from a
character constant in that it is enclosed in double .
quotes When we study
strings, we will see that there is also a big
difference in how h is stored in
memory as a character and as string. I he last
wide characters.
example is a string that uses
it II
"h" // A n u l l string
"Hello World\n"
"HOW ARE YOU"
"Good Morning!"
L"This string
contains wide characters."
FIGURE 2 - 1 3 Some Strings
It is important to understand the difference between the null character

( see Table 2 - 6 ) and an empty string. The null character represents no value.
As a character, it is 8 zero hits. An empty string, on the other hand, is a string
containing nothing. Figure 2 - 1 4 shows the difference between these two con -
stant types .
' \0• Null character

II ii
Empty string
FIGURE 2 - 14 Null Characters and Null Strings
At this point, this is all you need to know about strings. We talk more
about them and how they are stored in the computer when we study strings
in Chapter I I .
Use single quotes for character constants. Use double quotes for string
constants .
Coding Constants
In this section we discuss three different ways we code constants in our pro -
grams: literal constants, defined constants, and memory constants.
Literal Constants
A literal is an unnamed constant used to specify data. Ii we know that the
data cannot he changed, then we can simply code the data value itsell in a
statement.
Literals are coded as part of a statement using the constant formats
described in the previous section. For example, the literal 5 is used in the fol -
lowing statement.
a = b + 5;
Defined Constants
Another way to designate a constant is to use the preprocessor command
define. Like all preprocessor commands, it is prefaced with the pound sign ( # .
)
The define commands are usually placed at the beginning ol the program ,
although they are legal anywhere. Placing them at the beginning ol the pro -
gram makes them easy to find and change . A typical define command
might he
# define SALES TAX RATE .0825

sales tax ratt changes more

In the preceding example, for instance, the
and other similar constants at the
often than we would like. By placing it
find and change them easily.
beginning of the program , we can
program for the language translator, it
As the preprocessor
replaces each defined name
reformats
, SALES_the
TAX_ RATE in the previous example with
. I his action is
its value ( . 0825 ) wherever it is found in the source program
just like the search -and - replace command found in a text
editor. I he prepro-
cessor does not evaluate the code in any way— it just blindly makes the sub -
stitution. For a complete discussion of defined constants, see Appendix G ,
" Preprocessor Commands.”
Memory Constants
The third wav to use a constant is with memory constants. Memory constants
use a C type qualifier, const , to indicate that the data cannot be changed . Its
format is:
const type identifier = value ;
We have seen how to define a variable, which does nothing more than
give a type and size to a named object in memory. Now let us assume that we
want to fix the contents of this memory location so that they cannot he
changed . This is the same concept as a literal , only now we give it a name.
The following code creates a memory constant , cPi . To help us remember
that it is a constant, we preface the identifier name with c .
const float cPi = 3.14159 ;
Three points merit discussion: ( 1 ) The type qualifier comes first . ( )

2 Then
there must be an initializer. If we didn 't have an
initializer, then our named
constant would he whatever happened to he
in memory at cPi 's location
when our program started . ( 3 ) Finally, since we
have said that cPi is a con-
stant , we cannot change it .
Program 2 - 3 demonstrates the three different ways
constant . to code pi as a
PROGRAM 2- 3 Memory Constants

/ * This program demonstrates
three ways to use
constants.
Written by:
Date :
/
^define
include <stdio.h>
PI 3.1415926536
continued
PROGRAM 2- 3 Memory Constants ( continued )

9 int main ( void)
10 {
12 const double cPi = PI;
13
14 // Statements
15 printf( " Defined constant PI: %f \n " , PI ) ?
16 printf( " Memory constant cPi: %f\n " , cPI);
17 printf( "Literal constant: % f \n " , 3 . 1 4 1 5 9 2 6 5 3 6 ) ;
18 return 0 ;
19 > // main
Results:
Defined constant PI: 3.141593
Memory constant cPi: 3.141593
Literal constant: 3.141593
2.7 Input/Output
Although our programs have implicitly shown how to print messages , we have
not formally discussed how we use C facilities to input and output data
. We
devote two chapters, Chapters 7 and 13, to fully explain the C input /output
facilities and how to use them . In this section , we describe simple input and
output formatting.
Streams
In C , data is input to and output from a stream . A stream is a source ol or
destination for data . It is associated with a physical device, such as a termi -
nal , or with a file stored in auxiliary memory.
C uses two forms of streams: text and binary. A text stream consists of a
sequence of characters divided into lines with each line terminated hv a new -
-
line ( \ n ). A binary stream consists of a sequence ol data values such as inte
ger, real , or complex using their memory representati on . In this chapter , we
briefly discuss only text streams. A more detailed discussion ol text streams is
found in Chapter 7 , “Text Input /Output ,“ and a detailed discussion of binary'
streams is found in Chapter 13, Binary Input /Output .
"
A terminal can be associated only with a text stream because a keyboard

can only
can only send a stream of characters into a program and a monitor
display a sequence ol characters. A file, on the other hand can he associated
with a text or binary stream . We can store data in a file and later retrieve
them as a sequence of characters ( text stream ) or as a sequence of data -
val
ues ( binary streams . )
54 Section 2.7 Input/Output
we
^
. In C ,
use produce or consume text streams
monitor is known as standard output .
dard input and the
can be associated only with a text stream .

A terminal keyboard and monitor for a text
; a monitor is a destination
A keyboard is a source for a text stream
stream .
Fioure 2 - 15 shows the concept of streams

and the two physical devices asso-
dated with input and output text streams .
Data Source Program
( Input Text Stream » Data
keyboard
Output Text Stream - Data

^
monitor
Data Destination
FIGURE 2 - 15 Stream Physical Devices
Formatting Input/Output
I he previous section discussed the terminal as a text stream source and desti -
nation . We can only receive text streams from a terminal ( keyboard ) and send
.
text streams to a terminal ( monitor ) However, these text streams often repre -
sent different data types, such as integer, real , and Boolean . The C language
provides two formatting functions: printj for output formatting and sccinf for
input formatting. The printf function converts data stored in the program
into a text stream for output to the monitor; the scan
text stream coming from the keyboard to data
/ function converts the
values and stores them in pro-
gram variables. In other words, the printf and
scanf functions are data to text
stream and text stream to data converters.
Output Formatting: printf

The output formatting function is print .
--
/ The print/ function takes a set of
data values, converts them to a text stream
,-
ncinn f
^T.
* • •
tained in a format control string, and 8 mStmCtl°nS con
sends he
standard output ( monitor ) . For example
m 1 a"
8
" f T
l
°
““* “ • / 352SS ix 5!35
and then is sent to the monitor. What we see on the monitor is these three
characters, not the integer 234. However, we interpret the three characters
together as an integer value. Figure 2 - 16 shows the concept.
r 234 IBtDEHH " prinlf (- ) | 234

Stream
Text Integer
monitor
Data Destination Program
FIGURE 2- 16 Output Formatting Concept
Basic Concept
The printf function uses an interesting design to convert data into text
streams. We describe how the text stream should he formatted using a format
control string containing zero or more conversion specifications. In addi-
tion to the conversion specifications, the control string may contain textual data
and control characters to be displayed.
Each data value to he formatted into the text stream is described as a sep -
arate conversion specification in the control string. The specifications describe
the data values’ type, size, and specific format information, such as how wide
the display width should he. The location of the conversion specification
within the format control string determines its position within the text stream.
The control string and data values are passed to the print function
( print/ ) as parameters, the control string as the first parameter and one
parameter for each value to be printed. In other words, we supply the follow-
ing information to the print function:
.
1 The format control string including any textual data to be inserted into
the text stream .
.
2 A set of zero or more data values to he formatted.
Figure 2 - 17 is a conceptional representation of the format control string
and two conversion specifications .
Figure 2 - 17 ( a ) shows the format string and the data values as parameters
for the print function. Within the control string we have specified quantity
( Qty : ) and total ( Tot : ) as textual data and two conversion specifications ( % d
and % f ). The first specification requires an integer type value; the second
requires a real type value. We discuss the conversion specifications in detail
in the following section.
Figure 2 - 17 ( h) shows the formatting operation and the resulting text
stream. The first data value is a literal integer ; the second data value is
the
contents of a variable named tot . I bis part of Figure 2 - 1 7 shows how
the
print function expands the control stream and inserts the data values and
/
text characters.
56 Section 2.7 Input /Output
(a) Basic Concept

%d Tot:
Format Control String
,
Text StrearnQ^ n - printf (...) I
J
23 48.53
Data Values
~ ~
7r
723 sum ) ; sum 48.53
Iprintf( "Qty r ^rTot7 $ %F
~
/
A i —,
(b) Implementation
FIGURE 2 - 17 Output Stream Formatting Example
Format Control String Jext

, such as instructions to
The control string may also contain text to be printed
the user, captions or other identifiers, and other text intended to make the
output more readable. In fact as we
, have already seen , the format string may
contain nothing hut text in , which case the text will be printed exactly as
shown . We used this concept when we wrote the greeting program . In addi-
( n ),
tion . we can also print control characters, such as tabs \ t ) , newlines \
(
and alerts ( \ a ) , by including them in the format string . Tabs are used to for-
.
mat the output into columns Newlines terminate the current line and con -
tinue formatting on the next line . Alerts sound an audio signal to alert ,
usually to alert the user to a condition that needs attention . These control
characters are seen in Table 2-6.
Conversion Specification
To insert data into the stream , we use a conversion specification that contains
a start token ( % ), a conversion code , and up to four optional modifiers as
-
shown in Figure 2 18. Only the field -specification token ( % ) and the conver-
sion code arc required.
% Flag Minimum Precision Size Code

Width
FIGURE 2 - 18 Conversion Specification
Approximately 30 different conversion codes are used to describe data

types . For now,however, we are concerned with only three: character (c),
integer ( d ) , and floating point ( f ) . These codes with some examples are
shown in Table 2 - 10.
Type Size0 Code Example
char None c %c
short int h d % hd
int None d %d
long int None d % ld
long long int 11 d % lld
float None f %f
double None f %f
long double L f % Lf
a . Size is discussed in the next section.
TABLE 2- 10 Format Codes for Output

The size modifier is used to modify the type specified by the conversion
code. There are lour different sizes: h, 1 ( el ) , 11 ( el el ) , and L. The h is used
with the integer codes to indicate a short integer value ; * the 1 is used to indi -
'
cate a long integer value; the 11 is used to indicate a long long integer value;
and the L is used with floating-point numbers to indicate a long double value.
A width modifier may be used to specify the minimum number of posi -
tions in the output. ( If the data require using more space than we allow, then
print ] overrides the width .) It is very useful to align output in columns, such
as when we need to print a column of numbers. II we don ’t use a width mod -
ifier, each output value will take just enough room lor the data.
If a floating- point number is being printed , then we may specif ) the num -
ber of decimal places to he printed with the precision modifier. The preci -
sion modifier has the format
.m
where m is the number of decimal digits. If no precision is specified , printf

prints six decimal positions. These six decimal positions are often more than
is necessary.
When both width and precision arc used , the width must he large enough
to contain the integral value of the number, the decimal point , and the num -
ber of digits in the decimal position . Thus, a conversion specification
5. The h code is a carry over from assembler language where it meant hall word .
-
58 Section 2.7 Input/ Output
a maximum value
of 9999.99. Some examples of
of % 7.2 f is designed to print are shown below .
width specifications and precision
2 print positions
% 2 hd // short integer-
positions
% 4d // integer- 4 print
81 ) positions
% 81d // long int-print positions
8 ( not
: nnnn.dd
% 7.2 f // float 7 -
// long double-10 positions
: nnnnnn.ddd
% 10.3Lf
print modifications: justification , pad-

The flag modifier is used for four
variants. The first three are discussed
ding, sign , and numeric conversion
in Chapter .
,
here; the conversion variants are discussed
of a value when it is shorter than
Justificati on controls the placement be left or right . If there is no flag and
the specified width , justification can
, the value is right justihed . The
the defined width is larger than required
default is right justification . To left justify a value
, the flag is set to minus ( ). -
unused space when the value is
Padding defines the character that fills the
the default , or zero. If there is no
smaller than the print width. It can be a space ,
width is filled with spaces ; if the flag is 0 ,
flag defined for padding, the unused
with zeroes . Note that the zero flag is ignored if it is
the unused width is filled
changes its value.
used with left justification because adding zeros after a number
The sign flag defines the use or absence of a sign in a numeric value . We
can specify one of three formats: default formatting , print signed values , or
prefix positive values with a leading space . Default formatting inserts a sign
only when the value is negative. Positive values are formatted without a sign.
When the flag is set to a plus ( + ), signs are printed lor both positive and neg-
ative values. If the llag is a space , then positive numbers are printed with a
leading space and negative numbers with a minus sign .
Table 2 - 1 I documents three of the more common llag options.
Flag Type Flag Code Formatting

Justification None right justified
left justified
Padding None space padding
0 zero padding
Sign None positive value: no sign
negative value: -
positive value: +
negative value : -
None positive value: space
negative value: -
TABLE 2-11 Flag Formatting Options
Output Examples
This section contains several output examples. We show the printf statement,
followed by what would be printed. Cover up the solution and try to predict
the results.
1. printf( "%d %c%f " , 23 , 'z', 4.1);
23z4.100000
Note that because there are no spaces between the conversion speci-
fications, the data are formatted without spaces between the values.
2. printf( "%d %c %f " , 23, 'z * , 4.1 );
23 z 4.100000
This is a repeat of Output Example 1 with spaces between the con -

versionspecifications.
3. int numl = 23 ;
char zee = 'z ' ;
float num2 = 4.1;
printf( " %d %c %f " , numl , zee , num 2 );
23 z 4.100000
Again, the same example, this time using variables.

.
4 printf( " %d \t%c \t%5.If \n " , 23 , ' Z ' , 14.2 );
printf( " %d \t%c \t%5.If \n " , 107 , ' A ' , 53.6 );
printf( "%d \t%c\t%5.1f \n" , 1754 , 'F ' , 122.0);
printf( " %d \t%c\t%5.If \n" , 3 , 'P', 0.1 );
23 Z 14.2
107 A 53.6
1754 F 122.0
3 P 0.1
In addition to the conversion specifications, note the tab character

( \ t ) between the first and second, and second and third conversion spec -
ifications. Since the data are to he printed in separate lines, each format
string ends with a newline ( \ n) .
.
5 printf( "The number%dis my favorite number." , 23);
The number23is my favorite number.
Since there are no spaces before and after the format code ( % d), the
number 23 is run together with the text before and alter .
60 Section 2.7 Input /Output
—
is %6d " , 23);
6. printf("The number
23
.
The number is
A A A A
A A A A A
A A A A A
A A A A A A
H you count
co spaces fully . 1
It
J comes from the space
,
after is and before
t tbf "
the format » rinE
conversion specification .
. 11« «1» fou contc from the width in the
this year." , 233.12);

7. printf("The tax is %6 2f -
The tax is 233.12 this year
.
the precision two. Since the

In this example , the width is six and
totals five ( three for the integral portion and two
number of digits printed
takes one print position,
for the decimal portion ) , and the decimal point
are the spaces before
the full width is filled with data . The only spaces
and after the conversion code in the format string .
8. printf("The tax is %8.2f this year." , 233.12 )
;
The tax is 233.12 this year.

A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A
9. printf( "The tax is %08.2f this year." , 233.12 );
The tax is 00233.12 this year.

A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A
This example uses the zero ( lag to print leading zeros. Note that the
width is eight positions. I hree of these positions are taken up by the pre-
cision oi two digits and the decimal point. This leaves live positions tor
the integral portion of the number. Since there are only three digits
( 233 ) , printf inserts two leading zeros.
10. printf( "\"%8c %d \"" , 'h' , 23 );
h 23"
A A A A A A A A A A A A A A A
In this example we want to print the data

within quotes. Since
quotes are used to identify the format string, we
can ’t use them as print
characters . To print them , therefore- we must use the escape character
., . , >
wi h the quote ( \ " ), which
.
tells printf that what follows is not the end of

the string but a character to be printed ,
in this case , a quote mark.
printf ( A null '

characterrXOkills
\ 0ki 1 1 c the rest of the line\n")?
printf ( " \nThis is \'it\ ' in single quotes\n " );

printf ( "This is \ " it\ " in double quotes\n " );
printf ("This is \\ the escape character itself\n " );
...A new line

This is the bell character
A null character
This is 'it ' in single quotes
This is " it" in double quotes
This is \ the escape character itself
These examples use some of the control character names found in

Table 2 - 6. Two of them give unexpected results. In Output Example I 1 ,
the return character ( \ r ) repositions the output at the beginning of the
current line without advancing the line. Therefore, all data that were
placed in the output stream are erased.
The null character effectively kills the rest of the line. Had we not
put a newline character ( \ n) at the beginning of the next line, it would
have started immediately after character.
.
12 New example with multiple flags.
printf( "|% +8.2f||%0+8.2f||% 0+8.2f|" , 1.2, 2.3 , 3.4);
- -
1 +1.20 | 1 +0002.30||+3.40
A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A
This example uses multiple Hags. So that we can see the justification,
each value is enclosed in vertical bars. The first value is printed left justi-
fied with the positive flag set. The second example uses zero fill with a
space for the sign. Note that there is a leading space in the output. I bis
represents the plus value. It is then followed hv the leading zeros. The last
example demonstrates that the zero fill is ignored when a numeric value
is printed with left justification.
Common Output Errors

Each of the following examples has at least one error. Try to find each one
before you look at the solutions. Your results may vary depending on your
compiler and hardware.
1. printf ( " %d %d %d \n" , 44 , 55 );
44 55 0
This example has three conversion specifications hut only two values.
2. printf ( "%d % d \n" , 44 , 55 , 66);
44 55
specifications with three values. In
This example has two conversion .
value
this case , printf ignores the third
.
3 float x = 123.45 ;
d\n" , x );
printf( "The data are: %
The data are: 1079958732

in which the format specification ( integer )
This is a very common error in
( real ).
does not match the data type
Input Formatting: scanf

scanf ( scan formatting ). I his
The standard input formatting function in C is
, extracts and formats data
function takes a text stream from the keyboard
, and then stores the data
from the stream according to a format control string
in specified program variables. For example , the stream ot 5 characters 2 , 3 ,
'4 , 7, and ‘2 are extracted as
' ’ the real 234.2 . Figure 2 - 19 shows the concept .
Data Source
' 2' '3' ' 4 ' scanf (...) 234.2

Text Stream Real
Keyboard
Standard Input Program
FIGURE 2 - 19 Formatting Text from an Input Stream
The scanf function is the reverse of the printf function .

1 . A format control string describes the data to be extracted from the stream
and reformatted .
2 . Rather than data values as in the printf lunction , scanf requires the
variable addresses were each piece of data are to be stored . Unlike the
printf function , the destination of the data items cannot be literal values,
they must store in the variables.
3. With the exception of the character specification
, leading whites!jaces
are discarded .
n >n
conversionspecification characters in the format string must be
exac y matched by the next characters
in the input stream .
e car ul about extra
!
Extra characters
rha ,
^ characters
n the control string can be
non -whitespace and whitespace.
in the control stream .
divided i
into two categories:
a. Non -whitespace characters in the control string must exactly match

characters entered by the user and are discarded by the scanf alter
they are read . If they don ’t match , then scanj goes into an error state
and the program must he manually terminated .
We recommend that you don ’t use non - whitespace characters in
the format string, at least until you learn how to recover from errors in
Chapter 13. However, there are some uses for it . For example, il the
users want to enter dates with slashes, such as 5 / 10/06. the slashes
must either he read and discarded using the character format specifica -
tion (see the discussion of the assignment suppression flag in the later
section , “Conversion Specification ” ) or coded as non - whitespace in the
format specification . We prefer the option to read and discard them ,
.
h Whitespace characters in the format string are matched by zero or
more whitespace characters in the input stream and discarded . There
are two exceptions to this rule: the character conversion code and the
scan set (see Chapter 1 1 ) do not discard whitespace. It is easy, how-
ever, to manually discard whitespace characters when we want to read
a character. We simply code a space before the conversion specifica -
tion or as a part of it as shown below. Either one works.
" %c" o r " % c "
Remember that whenever we read data from the keyboard , there is a

return character from a previous read . If we don ’t flush the whitespace char-
acters when we read a character, therefore , we will get the whitespace from
the previous read. To read a character, we should always code at least one
whitespace character in the conversion specification; otherwise the
whitespace remaining in the input stream is read as the input character. For
example, to read three characters , we should code the following format
string. Note the spaces before each conversion specification .
s c a n f ( " %c % c %d " , &c l , &c 2 , &c 3 ) ;
Figure 2 - 20 demonstrates the input format string concept with a control

string having two fields ( %d and %f ) . The first one defines that a character
will he inserted here; the second defines that a real will be inserted there. We
w ill discuss these place holders, or format specifiers , later in the chapter.

Like the control string for print ] , the control string for scanj is enclosed in a
set of quotation marks and contains one or more conversion specifications
that describe the data types and indicate any special formatting rules and /or
characters.
(a) Basic Concept

" %c % f ^ n
"
r B 18.23 1
(" Text Stream
^ » scanf (...)
'
Data Values
printf(" %c % f " , &code , & price); |
Ll
18.23 B
price code
L — Discarded —-
(b) Implementation
FIGURE 2 - 20 Input Stream Formatting Examp e
Conversion Specification
To format data from the input stream , we use a conversion specification
that contains a start token ( % ) , a conversion code , and up to three optional
modifiers as shown in Figure 2 - 21 . Only the field -specification token ( % ) and
the conversion code are required .
Maximum
Flag Size
Width
FIGURE 2 - 21 Conversion Specification
I here are only three differences between the conversion codes for input
loi matting and output lormatting. First , there is no precision in an input con -
\ ersion specification . It is an error to
include a precision; if scanif finds a pre-
cision it stops processing and the input stream is in the
error state .
There is only one f° r input lormatting, the assignment suppression
Hag ( * ). More commonly associated with text files (
see Chapter 7 ) , the assign -
ment suppression flag tells scan that the next
/
stored . It is discarded . The following scan
input field is to be read but not
acter, and a floating- point number from
/ statement reads an integer, a char-
the input stream . The character is
read and discarded. The other fields are
read ,
here is no matching address parameter for formatted and stored . Note
,
the data to be discarded .
s c a n f ( " %d % * c % f " , sx , Sy);
The third difference is the width specification ; with input formatting it is

a maximum , not a minimum , width . The width modifier specifies the maxi -
mum number of characters that are to be read for one format code. \ \ hen a
width specification is included , therefore , scanf reads until the maximum
number of characters have been processed or until scan) finds a whitespace
character. If scanf finds a whitespace character before the maximum is
reached , it stops.
Input Parameters
For every' conversion specification there must he a matching variable in the
address list . The address list contains the address of the matching variable.
How do we specify' an address ? It ’s quite simple: Addresses are indicated by
prefixing the variable name with an ampersand ( & ). In C , the ampersand is
known as the address operator. Using the address operator, if the variable
name is price , then the address is & price . Forgetting the ampersand is one
of the most common errors for beginning C programmers, so you will have to
concentrate on it when you use the scanf function .
scanf requires variable addresses in the address list.
Remember that the first conversion specification matches the first vari-
able address , the second conversion specification matches the second variable
address, and so on . This correspondence is very' important . It is also very
important that the variable’s type match the conversion specification type.
The C compiler does not verify that they match. If they don ’t , the input data
will not he properly formatted when they are stored in the variable.
Pnd of File and Errors

In addition to whitespace and width specifications , two other events stop the
scanf function . If the user signals that there is no more input by keying end
of file ( EOF ) , then scanf terminates the input process. While there is no EOF
on the keyboard , it can be simulated in most systems. For example , Windows
uses the < ctrl + z > key combination to signal EOF. Unix and Apple Macin -
tosh use < ctrl + d > for EOF. The C user ’s manual for your system should
specify the key sequence for EOF.
Second , if scanf encounters an invalid character when it is try ing to con -
vert the input to the stored data type, it stops. The most common error is
finding a nonnumeric character when it is trying to read a number. The valid
characters are leading plus or minus , digits , and one decimal point . Any other
combination , including any alphabetic characters , will cause an error.
Although it is possible to detect this error and ask the user to re- input the
data , we will not he able to cover the conventions for this logic until
Chapter 7. Until then , he very careful when you enter data into your program .
66 Section 2.7 Input/ Output
Input Formatting Summary .

using scanf
Table 2- 12 summarizes the rules for
1 . The conversion operation processe

s until :
.
a End of file is reached .
has been processed .
b. The maximum number of characters
after a digit in a numeric specification.
c. A whitespace character is found
d. An error is detected.
2 . There must be a conversion specification
for each variable to be read .
type for each conversion
3. There must be a variable address of the proper
specification .
4. Any character in the format string other than whitespa
ce or a conver-
sion specification must be exactly matched by the user during input. If
the input stream does not match the character specified , an error is sig-
naled and scanf stops.
5 . It is a fatal error to end the format string with a whitespace character.
Your program will not run correctly if you do.
TABLE 2- 12 sconf Rules
Input Examples
This section contains several examples. We list the data that will be input
first . T his allows you to cover up the function and try to formulate your own
scan! statement.
1 . 214 156 14 Z
scanf ( " %d %d % d%c" , &a , & b, &c , &d ) ;
Note that if there were a space between the 14 and the Z . it would
create an error because % c does not skip whitespace! To prevent this prob
lem , put a space before the % c code as shown below. This will cause it to
-
skip leading whitespace.
scanf ( " % d % d % d % c " , & a , & b , &c , &d ) ;
2 . 2314 15 2.14
scanf ( " % d % d % f " , & b , &c ) ;
Note the whitespace between the

spaces are not conversion specifications. T hese
necessary with numeric iinput, but it is a good idea to
include them.
3. 14/26 25/66
scanf( " %2d /%2d %2d /% 2d " ,

& numl , &denl , &num2 / &den 2 );
Note the slashes ( / ) in the format string. Since they are not a part ol
the conversion specification, the user must enter them exactly as shown
or sccitif will stop reading.
4. 11-25-56
- -
scanf ( "%d %d %d " , &a , & b , &c );
Again, we see some required user input, this time dashes between
the month, day, and year. While this is a common date format, it can
.
cause problems A better solution would be to prompt the user separately
for the month, the day, and the year.
Common Input Errors

Each of the following examples has at least one error, fry to find it before you
look at the solution. Your results may vary depending on your compiler and
hardware.
1. int a = 0;
scanf ( "%d" , a);
printf( "%d\n " , a ) ;
234 ( Input )
0 ( Output )
This example has no address token on the variable ( & a ). II the pro -
gram runs at all, the data are read into an unidentified area in memory.
What is printed is the original contents of the variable, in this case 0.
2. float a = 2.1 ;
scanf ( "%5.2f " , & a );
printf ("%5.2f" , a ) ;
74.35 ( Input )
2.10 ( Output )
This example has no precision in the input conversion specification.

When scanf finds a precision, it stops processing and returns to the func -
tion that called it . The input variable is unchanged.
3. int a ;
int b ;
scanf ( " %d %d %d " , &a , & b );
printf ( " %d %d \n ” , a , b );
( input )
5 10 (output)
5 10
This example has three

addresses. Therefore , scanf
.
.
conversion specifications but only two
reads the first two values and qu because
ts
no third address if found
4. int a = 1
int b 2=
int c 3= &c ) i
scanf ( "%d%d" , &a t & b /
printf ( "%d %d\n", a , b
, c);
5 10 15 ( input)
5 10 3 (output)
, but it has three

This example has only two conversion specifications
and ignores the
addresses. Therefore, scanf reads the first two values
stream waiting to be read.
third address. The value 15 is still in the input
2.8 Programming Examples

In this section , we show some programming example to emphasize the ideas
and concepts we have discussed about input /output .
EXAMPLE 2 - 2 Print "Nothing !"

Program 2- 4 is a very simple program that prints “Nothing!”
PROGRAM 2 - 4 A Program That Prints "Nothing!"

1 /* Prints the message "Nothing 1 ".
2 Written by:
3 Date:
4 */
5 ftinclude <stdio.h>
6
int main (void )
{
// Statements
printf( "This program prints\n\n\t\"Nothing!\ •• M
);
return 0 ;
} // main
continued
PROGRAM 2 - 4 A Program That Prints "Nothing!" ( continued )

Results:
This program prints
" Nothing!"
EXAMPLE 2 - 3 Print Boolean Constants

Program 2- 5 demonstrates printing Boolean values. As the program shows,
however, while a Boolean literal contains either true or false , when it is
printed , it is printed as 0 or 1 . This is because there is no conversion code tor
Boolean . To print it , we must use the integer type , which prints its stored
value, 0 or 1 .
PROGRAM 2 - 5 Demonstrate Printing Boolean Constants

1 /* Demonstrate printing Boolean constants.
2 Written by:
3 Date:
4 */
6 # include <stdbool.h>
7
8 int main ( void )
9 {
11 bool x = true ;
12 bool y = false;
13
14 // Statements
15 printf ( "The Boolean values are: %d %d \n" , x , y );
16 return 0 ;
17 > // main
Results:
The Boolean values are: 1 0
EXAMPLE 2 - 4 Print Character Values

Program 2-6 demonstrates that all characters are stored in the computer as
integers. We define some character variables and initialize them with values ,
and then we print them as integers. As you study the output , note that the
ASCII values of the characters are printed . The program also shows the value
of some nonprintable characters. All values can be verified by referring to
Appendix A.
Section 2.8 Programming Examples
PROGRAM 2-6 Print Value of Selected Characters

of selected characters ,
1 /* Display the decimal value
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 <
10 char A = ’ A'
11
12
char a
char B
= a'
= 'B
-
13 char b = * b*
14 char Zed = 'z’
15 char zed z'
16 char zero = '0'
17 char eight = '8';
18 char NL = '\n'? // newline
19 char HT = '\t'; // horizontal tab
20 char VT = ' \v ' ; // vertical tab
21 char SP // blank or space
22 char BEL = '\a'; // alert ( bell)
23 char dblQuote • it
// double quote
24 char backslash = • ; // backslash itself
25 char oneQuote = '\ ; // single quote itself
26
27 // Statements
28 printf( " ASCII for char 'A ' is: %d\n", A)
29 printf( " ASCII for char 'a' is: %d\n " , a)
30 printf( " ASCII for char B is: %d \n", B)
31 printf("ASCII for char
* > • is: %d \n", b)
32 printf( " ASCII for char 'Z' is: %d\n" , Zed );
33 printf(" ASCII for char 'z 1
is: %d\n", zed );
34 printf("ASCII for char '0' is: %d \ "
35 printf( "ASCII for char
n , zero);
'8 is: %d \n" ,
36 Printf("ASCII for char eight );
• Wn * is: %d\n " ,
37 Printf( "ASCII for char NL);
' Wt ' is: %d\n " ,
38 printf("ASCII for HT);
char '\\ v' is
i : %d\n " , VT);
39 Printf( " ASCII for char
40 ' is: %d\n", SP );
Printf("ASCII for char • Wa' is: %d\n " ,
41 printf( " ASCII for BEL );
42
char '\ n i
is: %d\n " ,
printf( " ASCII for dblQuote );
43
char • \\ is: %d\
n" ,
printf(" ASCII
for char backslash );
'\ i
is: %d\n " ,
44 oneQuote );
continued
PROGRAM 2 -6 Print Value of Selected Characters ( continued )

45 return 0 ;
46 > // main
Results:
ASCII for character 'A' is: 65
ASCII for character 'a ' is: 97
ASCII for character 'B ' is: 66
ASCII for character ' b' is: 98
ASCII for character 'Z' is: 90
ASCII for character 'z' is: 122
ASCII for character ' 0' is: 48
ASCII for character '8' is: 56
ASCII for character ' \n ' is: 10
ASCII for character '\t' is: 9
ASCII for character '\v' is: 11
ASCII for character ' ' is: 32
ASCII for character ' \a ' is: 7
ASCII for character H i
is: 34
ASCII for character * \' is: 92
ASCII for character i i i
is: 39
EXAMPLE 2- 5 Define Constants

Let’s write a program that calculates the area and circumference ol a circle
using a preprocessor-defined constant for n . Although we haven t shown you
’
how to make calculations in C if you know

, algebra you will have no problem
reading the code in Program 2 - 7.
PROGRAM 2 - 7 Calculate a Circle's Area and Circumference

1 /* This program calculates the area and circumference
2 of a circle using PI as a defined constant.
3 Written by:
4 Date:
5 */
7 # define PI 3.1416
8
9 int main ( void )
10 {
12 float circ ;
13 float area ;
14 float radius ;
continues
72 Section 2.8 Programming Examples
( continued )
's Area and Circumference
PROGRAM 2-7 Calculate a Circle
15
16 / / statements the value of t h e radius "
: );
17 printf("\nPlease enter
18 scanf("%f" i &radius);
19
* PI * radius;
20 circ = 2
;
21 area = PI * radius * radius
22 % 10.2f " , radius );
23 printf("\nRadius is :
: %10.2f " f circ );
24 printf("\nCircumference is
%10.2f " area );
printf("\nArea is :
/
25
26
27 return 0 ;
28 } // main
Results:
: 23
Please enter the value of the radius
Radius is : 23.00
Circumference is : 144.51
Area is : 1661.91
EXAMPLE 2- 6 Print a Report

You are assigned to a new project that is currently being designed . Io give the
customer an idea of what a proposed report might look like , the project
leader has asked you to write a small program to print a sample. 1 he specifi-
cations for the report are shown in Figure 2 - 22 , and the code is shown in
Program 2 -8.
Part Numbers musTN

have leading zeros Z
^
Part Number Qty On Hand Qty On Order Price
031235 22 86 $ 45.62
000321 55 21 $1 2 2 . 0 0
028764 0 24 $ 0.75
003232 12 0 $ 10.91
End of Report
Decimal points
Leading zeros must be aligned.
suppressed.
FIGURE 2- 22 Output Specifications for Inventory Report

The report contains four fields: a part number, which must he printed
with leading zeros; the current quantity on hand; the current quantity on
order; and the price of the item, printed to two decimal points. All data
should he aligned in columns with captions indicating the type of data in
each column. The report should he closed with an “ End of Report message.
'
PROGRAM 2 - 8 A Sample Inventory Report

1 /* This program will print four lines of inventory data
2 on an inventory report to give the user an idea of
3 what a new report will look like. Since this is not
4 a real report , no input is required. The data are
5 all specified as constants
6 Written by:
7 Date:
8 */
10
11 int main ( void )
12 {
13 // Statements
14 // Print captions
15 printf( "\tPart Number\tQty On Hand ” );
16 printf( "\tQty On Order \tPrice\n" );
17
18 // Print data
19 printf( "\t %06 d \t\t% 7d \t\t %7d \t\t $%7.2f \n " ,
20 31235 , 22, 86 , 45.62);
21 printf( " \t %06d \t\t%7d \t\t%7d \t\t $ %7.2f \n " ,
22 321 , 55, 21 , 122.);
23 printf( " \t %06d \t\t%7d \t\t%7d \t\t $%7.2f \n" ,
24 28764 , 0, 24 , .75);
25 printf( \t %06d \t\t%7d \t\t%7d \t\t
" $ %7.2f \n " ,
26 3232, 12 , 0, 10.91 );
27
28 // Print end message
29 printf( " \n\tEnd of ReportAn");
30 return 0 ;
31 > // main
Program 2 - 8 Analysis There are a few things about Program 2 - 8 that you should note. First, it is fully docu
-
mented. Professional programmers often ignore documentatio n on "one - time - only"
programs, thinking they will throw them away, only to find that they end up using
them over and over. It only takes a few minutes to document a program, and it is
always time well spent. If nothing else, it helps clarify the program in your mind.
Next, look carefully at the formatting for the print statements . Spacing is controlled
by a combination of tabs and format code widths. The double spacing for the end of
/4 Section 2.8 Programming Examples
a newline command \ n
( ) at the beginning of
report message is controlled by placing
.
the message in Statement 29. nrlnrU with a return statement that informs the
-
I ^
»’EEd
good programmer.
^ "
m !mo pr -
°
grams, is the sign of a
EXAMPLE 2 - 7 Printing The Attributes of o Complex Number

: a real part and an imaginary
A complex number is made of two components
vector with two components.
part. In mathematics , it can be represented as a
The real part is the projection of the vector on the horizontal axis ( x ) and the
imaginary part is the projection of the vector on the vertical axis ( \ ) . I n C , we
use complex number and a predefined library function to print
the real and
imaginary values. We can also find the length of the vector , which is the abso-
lute value of the complex number and the angle of the vector , which is the
argument of the vector . These four attributes arc shown in Figure 2 - 23 .
Imaginary
^ Argument
Real
y
FIGURE 2 - 23 Complex Number Attributes
As the figure shows, the absolute value of the complex a + b * I can be found
as ( a + b ) 1 /2 . The argument can be found as arctan ( b/ ) .
a The conjugate of a
complex number is another complex number defined as a - b * .
I
Program 2 -9 shows how we print the different attributes
of a complex
number using the predefined functions creal , cimag ,
cabs and carg .
,
PROGRAM 2 - 9 Print Complex Number Attributes

1 /* Print attributes of a complex number.
2 Written by:
3 Date
4 */
6 # include <math.h>
7 #include <complex.h>
8
9 int main (void )
continued
PROGRAM 2 - 9 Print Complex Number Attributes ( continued )

10 {
12 double complex x = 4 + 4 * I;
13 double complex xc ;
14
15 // Statements
16 xc = conj (x);
17 printf( "% f % f %f %f \n" , creal( x ) cimag( x ),
18 cabs( x ), carg( x ));
19
20 printf( " %f %f % f %f \n" / creal( xc ), cimag( xc ),
21 cabs( xc ), carg( xc ));
22 return 0;
23 > // main
Results:
4.000000 4.000000 5.656854 0.785398
-
4.000000 4.000000 5.656854 -0.785398
EXAMPLE 2-8 Mathematics with Complex Numbers.

C allows us to add, subtract, multiply, and divide two complex numbers using
the same operators ( +, -, *, f ) that we use lor real numbers. Program 2 - 1 0
demonstrates the arithmetic use of operators with complex numbers .
PROGRAM 2 - 10 Complex Number Arithmetic
1 /* Demonstrate complex number arithmetic.
2 Written by:
3 Date:
4 */
6 # include < math.h >
7 # include <complex.h >
8
9 int main ( void )
10 {
12 double complex x = 3 + 4 * I;
13 double complex y = 3 4 * 1;
14 double complex sum;
15 double complex dif ;
16 double complex mul ;
17 double complex div ;
continitec
76 Section 2.8 Programming Examples
PROGRAM 2 - 10 Complex Number Arithmetic ( continued

)
18
19 // statements
20 sum = x + y ;
21
22
dif = x -
y;
mul = x * y;
23 div = x / y ;
24
25 printf("%f %f %f %f\n" , creal(sum), cimag(sum ),
26 cabs(sum), carg( sum ));
27 printf( ” %f %f %f %f\n", creal(dif ), cimag(dif ),
28 cabs(dif), carg( dif ));
29 printf("%f %f %f %f \n" , creal(mul), cimag( mul ),
30 cabs( mul), carg( mul ));
31 printf("%f %f %f %f\n ” creal(div ),
/
cimag( div ),
32 cabs(div), carg(div ));
33 return 0;
34 > // main
Results:
6.000000 0.000000 6.000000 0.000000
0.000000 8.000000 8.000000 1.570796
25.000000 0.000000 25.000000 0.000000
-0.280000 0.960000 1.000000 1.854590
J
2.9 Software Engineering

Although this chapter introduces only a few programming concepts
, there is
still much to he said from a software engineering point ol view. We will dis -
, data naming , and data hiding.
cuss the concepts of program documentation
Program Documentation
There are two levels of program documentation. The first is the general docu -
mentation at the start of the program. The second level is found within each
function.
General Documentation
. Each
Program 2 - 1 1 shows what we recommend lor program documentation
program should start with a general description of the program . Followin g the
of the author and the date the program was
general description is the name
's change history, which documents
written. Following the date is the program
the reason and authority for all changes. For a production program whose
use
spans several years, the change history can become extensiv e.
PROGRAM 2 - 11 Sample of General Program Documentation

1 /* A sample of program documentation. Each program
2 starts with a general description of the program.
3 Often , this description can be taken from the
4 requirements specification given to the programmer.
5 Written by: original author
6 Date: Date first released to production
7 Change History:
8 <date> Included in this documentation is a short
9 description of each change.
10 */
Module Documentation
Whenever necessary, we include a brief comment lor blocks of
code. A block
of code is much like a paragraph in report
a . It contains one thought — that is,
one sel of statements that accomp lish a specific task. Blocks ol code in our
ed by blank program lines , just as we skip blank lines
program are separat
between paragraphs in reports.
nt,
If the block of code is difficult, or if the logic is especially significa
line — descripti on ol the blocks
then we give the reader a short —one- or two -
purpose and/or operation. We will provide many example
s of this type ol doc -
umentation through out the text .
be docu -
Sometimes a textbook suggests that each variable in a program
mented. We disagree with this approach. First , the proper location tor vari-
able documentation is in a data dictionar y. A data dictiona ry is a system
78 Section 2.9 Software Engineerii
document
's.iT,“r “ * « " •
need for the comment .
.
tool W
» »
— -- — ,
» p
"ante
,
""8
«1
,hc nsme eiimin “' ,ht

ei
Data Names programming is the use of intelligent

Another principle of good structured ,
data are used .
,
..
good idea about what data it e ntain
and
name itself should »« 11* reader
data names. This mean that the variable maybe even an idea about how the
, , . . -
keeping names short , the ad van
. .
Althouoh there are obvious advantages to .

so cryptic that they are unintelligible
tage is quickly lost if the names become
to find a bug, only to discover
We have seen programmers struggle for hours
that the problem was the wrong variable was used . The time saved keying
lost - or a hundredfold in debugging time.
short , cryptic names is often ten
several guidelines to help you construct good , intelli-
We have formulated
gent data names .
1 . The name should match the terminology of the user as closely as possible
.
Let ’s suppose that you are writing a program to calculate the area of
a rectangle. Mathematicians often label the sides of a rectangle a and b }
but their real names are length and width. Therefore , your program
should call the sides of the rectangle length and width . These names
are commonly used by anyone describing a rectangle.
2 . When necessary for readability, and to separate similar variables from
each other, combine terms to form a variable name.
Suppose that you are working on a project to compute a payroll.
I here are many different types of taxes. Each of the different taxes
should he clearly distinguished from the others by good data names.
Table 2- 13 shows both good and bad names for this programming situa -
tion . Most of the poor names are either too abbreviated to be meaningful
( such as ftr ) or are generic names ( such as
rate ) that could apply to
many different pieces of data .
Good Names Poor Names

ficaTaxRate ficajax rate rate ftr frate fica
ficaWitholding fica_witholding fwh ficaw wh
ficaWthldng fica _ wthldng fcwthldng wthldng
ficaMax ficaDIrMax ax fmax
TABLE 2 - 13 Examples of Good and Poor Data
Names
Note the two different concepts for separating the words in a vari -
ables name demonstrated in Table 2- 13. In the first example, we capital-
ized the first letter of each word . In the second example, we separated the
words with an underscore . Both are good techniques for making a com -
pound name readable. If you use capitalization , keep in mind that C is
case sensitive , so you must he careful to use the same cases for the name
each time you use it .
.
3 Do not create variable names that are different by only one or two
. Names
letters,
that are
especially if the differences are at the end of the word
too similar create confusion . On the other hand , a naming pattern makes
it easier to recall the names. This is especially true when user terminol -
ogy is being used. Thus , we see that the good names in Table 2 - 13
all
start with fica .
4. Abbreviations, when used , should clearly indicate the word
being
abbreviated .
Table 2- 13 also contains several examples of good abbreviations.
Whenever possible, use abbreviations created by the users. They will
often have a glossary of abbreviations and acronyms that they use.
Short words are usually not abbreviated . II they are short in the first
place, they don ’t need to be made shorter.
5. Avoid the use of generic names.
Generic names are programming or user jargon . For example, count’
and sum are both generic names. They tell you their purpose but don t
give you any clue as to the type ol data they are associated with. Better
names would be emplyCnt and ficaSum. Programmers are especially
fond of using generic names , but they tend to make the program confus -
ing. Several of the poor names in Table 2 - 13 arc generic.
6 . Use memory constants or defined constants rather than literals lor values
that are hard to read or that might change from system to system .
Some constants are nearly impossible to read . We pointed out the
lor it .
space earlier. If you need a space often , create a defined constant
Table 2 - 14 contains several example s of constants that are better when
coded as defined constants.
# define SPACE # define BANG ' !
# define DBL QTE

m i
# define QUOTE ' V
# define COMMA # define COLON
TABLE 2- 14 Examples of Defined Constants
Data Hiding
of
In “ Structure of a C Program ” in Section 2.2 , we discussed the concept
before main
global and local variables. We pointed out that anything placed
and data encapsulate.

connectio n withprincipl
this concept are data hiding
es have as their objective protect
,ng data from acct-
Both of these
dental destruction by parts of your program doesn t
that don access to the
. . .
requtre
, if a part of your program require data do , ts
to
data . In other words
the data Unt l you learn to use
job it shouldn 't be able to see or modify
, you will not be able to prov de th s data-
functions in Chapter 4, however
hiding capability.
Nevertheless, you should start your programming
with good practices .
structur ed program ming, we now
And since our ultimate objective is good
formulate our first programming standard :
Programming Standard
.
No variables are to be placed in the global area of a program
Any variables placed in the global area of your program — that is, before
main an be used and changed by every part of your program . This is unde-
sirable and is in direct conflict with the structured programming principles of
data hiding and data encapsulation .
2.10 Tips and Common Programming Errors

1 . Well -structured programs use global (defined ) constants but do not use
global variables.
2. The function header for main should be complete . We recommend the
following format :
int main (void)
a. If you forget the parentheses after main , you will get a compile error.
.
b If you put a semicolon after the parentheses, you will get a compile error.
c. If you misspell main you will not get a compile error, but you will get
an error when you try to link the program . All programs must have a
function named main .
3. If you forget to close the format string in the scanj or printj statement,
you will get a compile error.
4. Using an incorrect conversion code for the data type being read or writ -
ten is a run - time error. You can ’t read an integer with afloat conversion
code. Your program will compile with this error, but it won 't run correctly.
3. Not separating read and write parameters with commas is a compile error.
6. Forgetting the comma after the format string in a read or write statement
is a compile error.
7. Not terminating a block comment with a close token ( * / ) is a compile error.
8. Not including required libraries, such as stilio . h , at the beginning of your
program is an error. Your program may compile, but the linker cannot
find the required functions in the system library.
9. If you misspell the name of a function , you will get an error when you
link the program . For example, if you misspell scanj or printj , your pro-
gram will compile without errors , but you will get a linker error. Using
the wrong case is a form of spelling error. For example, each of the fol -
lowing function names are different:
scanf , Scanf , SCANF printf , Printf , PRINTF
10. Forgetting the address operator ( & ) on a scanf parameter is a logic ( run -
time ) error.
I 1 . Do not use commas or other characters in the format string lor a scanf
statement . This will most likely lead to a run - time error when the user
does not enter matching commas or characters. For example, the comma
in the following statement will create a run - time problem if the user
doesn ’t enter it exactly as coded .
scanf ( "%d , %d " , &a, & b);

82 Section 2.12 Summary
, put a space
read a whitespace character
statement .
in a scan/
13. Using an address

operator
usually a run - time error.
whitespace at the end of a format string in scunf .
14. Do not put a trailing
- .
This is a fatal run time error
2.11 Key Terms

floating- point type precision modifier
address list program
address operator format control string
function documentation
ASCII real type
binary stream global declaration
section reserved word
block comment sign flag
Boolean header file
identifier size modifier
call
character constant imaginary type standard input
character set include standard output
comment initializer statement
complex type integral type standard types
constant intelligent data name statement section
conversion code justification stream
conversion keyword string
specification Latin character set string constant
data encapsulation line comment syntax
data hiding literal text stream
declaration logical data token
definition memory constant type
derived types padding type qualifier
end of file ( EOF) parameter variable
escape character parameter list width modifier
flag modifier
2.12 Summary
J In 1972. Dennis Ritchie designed C at Bell Laboratories.
° l '?!?
" the Ameritan National Standards Institute ( ANSI ) approved
ANSI C ;’.in 1990, the ISO standard was approved ,
fhe basic component of a C program is the function .
Every C function iis made of declarations,
statements.
,
definitions and one or more
J One and only one of the

functions in a C program must be called main .
To make a program more readable, use comments. A comment

is a
. C of com -
sequence of characters ignored by the compiler uses two types
ments: block and line. A block comment starts with the token / * and ends
the
with the token * / . A line comment starts with the / / token; the rest ol
line is ignored.
Identifiers are used in a language to name objects.
C types include void , integral, floating point, and derived .
A void type is used when C needs to define a lack of data.
An integral type in C is f urther divided into Boolean, character
, and integer .
is desig-
A Boolean data type takes only two values: true and false . It
nated by the keyword bool .
A character data type uses values from the standard alphabet ol
the lan-
. There characte r type sizes,
guage, such as ASCII or Unicode are two
char and wjchar .

An integer data type is a number without a fraction. C uses lour
dilfer -
ent integer sizes: short int , int , long int , and long long int .
The floating-point type is further divided into real, imaginary, and complex
.
,
A real number is a number with a fraction. It has three sizes: float
double , and long double.
The imaginary type represents the imaginary part of a complex number
.
It has three sizes, float imaginary , double imaginary , and long double
imaginary .
The complex type contains a real and an imaginary part. C uses three
complex sizes: float complex , double complex , and long double complex
.
A constant is data whose value cannot be changed.
com-
Constants can be coded in three different ways: as literals, as define
mands, and as memory constants .
Variables are named areas of memory used to hold data.
in C.
IJ Variables must be declared and defined before being used
To input data through the keyboard and to output data through
the moni -
tor, use the standard formatted input /output functions.
data through the
scanf is a standard input function for inputting formatted
keyboard.
d data the
IJ printf is a standard output function for outputting formatte
to
monitor .
the reader
As necessary, programs should contain comments that provide
with in-line documentation lor blocks ol code.
and understand.
Programs that use “intelligent names are easier to read
"
2.13 Practice Sets
Review Questions is to store a programs
file, such as stdio.h ,
1. The purpose of a header
source code.
a. True
b. False
can he used in an identifier.
2. Any valid printable ASCII character
a. True
b. False
data from the keyboard is printf .
3. The C standard function that receives
a. True
b. False
the structure of a C program
4. Which of the following statements about
is false?
section.
a. A C program starts with a global declaration
b. Declaration sections contain instructio ns to the .
computer
c. Every program must have at least one function .
d . One and only one function may be named main .
e. Within each function there is a local declaration section .
5. Which of the following statements about block comments is false ?
a . Comments are internal documentation for programmers.
b. Comments are used by the preprocessor to help format the program .
c. Comments begin with a / * token .
.
d Comments cannot he nested .
e. Comments end with a * / token .
.
6 Which of the following identifiers is not valid ?
a . _option
.
b amount
c. sales_amount
d . salesAmount
e. $salesAmount
/ . W Inch ol the following is not a data type?
a. char
b. float
.
c int
d . logical
e. void
(
8. The code that establishes the original value for a variable is known as a n ) :
a. assignment
b. constant
c. initializer
d . originator
e. value
9. Which of the following statements about a constant is true?
a. Character constants are coded using double quotes ( " ) •
b. It is impossible to tell the computer that a constant should he a float
or a long double .
c. Like variables, constants have a type and may he named .
d . Only integer values can be used in a constant .
e. The value of a constant may be changed during a program s execution .
10. The conversion specification is used to read or write
a short integer.
a. % c
b. %d
.
c %f
d . % hd
e. % lf
1 1 . To print data left justified , you would use a in the
conversion specification .
.
a flag
.
b precision
c. size
d. width
e. width and precision
12. The function reads data from the kevboard .
a. displayf
b. printf
.
c read
d . scanj
e. write
13. One of the most common errors for new programmers is forgetting to use
the address operator for variables in a scanf statement . What is the
address operator ?
a . The address modifier ( @ ) in the conversion specification
b. The ampersand ( & )
c. The caret ( * )
d . The percent ( % )
e. The pound sign ( # )
Exercises in C ?
a character constant
14. Which of the following is not
a. 'C *
b. 'bb '
c. "C"
d. ' ?
e.
an integer constant in C?
15. Which of the following is not
a. -320
b. +45
c. -31.80
d. 1456
e. 2, 456
-
16. Which of the following is not a floating point
constant in C?
a. 45.6
b. -14.05
c. ' a '
d. pi
e. 40
17. What is the type of each of the following constants?
.
a 15
b. -14.24
.
c ' b'
d. "l"
e. " 16 ”
18. Which of the following is not a valid identifier in C ?
a. A3
I). 4 A
c. if
d. IF
e. tax-rate
19. What is the type of each of the following constants?
a. "7 "
b. 3
.
C " 3.14159 "
d. '2'
e. 5.1
20. What is the type of each of the
following constants?
a. " Hello"
I). 15L
C. 8.5 L
d. 8 . 5 f
e. '\a '
.
21 Which of the following identifiers are valid and which are invalid ?
Explain your answer.
a. num
b. num2
c. 2dNum
_
d . 2d num
e. num# 2
.
22 Which of the following identifiers are valid and which are invalid ?
Explain your answer.
a. num-2
b. num 2
c. num 2
d. num2
e. num 2
23. What is output from the following program fragment ? To show your out -
.
put draw a grid of at least 8 lines with at least I 3 characters per line.
// Local Declarations
int x = 10;
char w = 'Y';
float z = 5.1234 ;
// Statements
printf( "\nFirst\nExample\ n:" );
printf( " %5d \n , w is %c\n" , x , w );
printf( "\nz is %8.2f \n" , z );
24. Find any errors in the following program .
// This program does nothing

int main
{
return 0;
>
# include ( stdio.h )
int main ( void )
{
print ( "Hello World " );
return 0 ;
{
Section 2.13 Practice Sets
the following program

.
.
26 Find any errors in
include <stdio>
int main (void )
{ correct ' );
printf('We are to learn');
here
printf('C language
return 0;
} // main
.
.
27 Find any errors in the following
program
/* This is a program with some errors

in it to be corrected.
*/
int main ( void )
{
integer a;
-
floating point b;
c;
character
// Statements
printf("The end of the program." );
return 0;
} // main
/* This is another program with some

errors in it to be corrected.
*/
int main (void )
{
a int;
b float, double ;
c, d char;
// Statements
printf( "The end of the program." );
return 0;
> // main
29. Find any errors in the following

program.
/* This is the last 7program to be
corrected in these-1 exercises.
*/
continued
int main ( void )

<
a int;
b : c : d char;
d , e , f double float;
// Statements
printf( "The end of the program." );
return 0 ;
> // main
Problems
.
30 Code the variable declarations for each of the following:
a. a character variable named option
b. an integer variable, sum, initialized to 0
c. a floating-point variable, product , initialized to I
.
3 1 Code the variable declarations for each ol the following:
.
a a short integer variable named code
b. a constant named salesTax initialized to .0825
c. a floating-point named sum of size double initialized to 0
.
32 Write a statement to print the following line. Assume the total value is
contained in a variable named cost.
The sales total is: $ 172.53

A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A
.
33 Write a program that uses four print statements to print the pattern of
asterisks shown below.
******
******
******
******
.
34 Write a program that uses four print statements to print the pattern of
asterisks shown below.
*
**
***
****
in the alpha-
that uses defined constants for the vowels
( 0 , 2 , 4 , 6, 8 ). It
35. Write a program decimal digits
memory constants for the even inp constants for the odd
bet and lines using literal
then printsthe following three
digits.
a e i o u
2 4 6 8
0
integer va riables and

initializes them to
36. Write a program that
defines five on a single line sepa-
1 , 10 , 100 , 1000 ,
10000. It then prints them
and code ( % d ) , and on
using the decimal conversion
rated by space characters conversion ion code ( % f ) . Note the differences
the next line with the float them ?
you explain
between the results. How do a cost. The
the user to enter a quantity and
37. Write a program that prompts named quantity and a float named
values are to be read into an integer use only one statement to read the
, and
unitPrice. Define the variables
the values , skip one line and print each value, with
values. .After reading
line.
an appropriate name , on a separate
user to enter an integer and then
38. Write a program that prompts the
prints the integer first as a character
, then as a decimal, and finally as a
run is shown below.
float. Use separate print statements. A sample
The number as a character: K

The number as a decimal : 75
The number as a float : 0.000000
Projects
39. Write a C program using printf statements to print the three f irst letters
of your first name in big blocks. This program does not read anything
from the keyboard. Each letter is formed using seven rows and five col-
umns using the letter itself . For example , the letter B is formed using 17
Bs, as shown below as part of the initials BEF.
BBB EEEEE FFFFF

B B E F
B B E F
BBB EEE FFF
B B E F
B B E F
BBB EEEEE F
l his is just an example. Your program must the first three let -
print
ters ol your first name. Design your print
/ statements carefully to create
enough blank lines at the beginning and end to make your initials read-
able. Use comments in your program to enhance readability as shown in
this chapter.
40. Write a program that reads a character, an integer, and a floating-point
.
number It then prints the character, first using a character format speci -
fication ( % c ) and then using an integer specification ( % d ) . After printing
the character, it prints the integer and floating-point numbers on separate
lines. Be sure to provide complete instructions (prompts) lor the user.
41. Write a program that prompts the user to enter three numbers and
then
prints them vertically ( each on one line ) , first forward and then reversed
( the last one first), as shown in the following design.
Please enter three numbers: 15 35 72

Your numbers forward:
15
35
72
Your numbers reversed:

72
35
15
42 Write a program that reads 10 integers and prints the first and the last on
.
one line, the second and the ninth on the next line, the third and the sev-
enth on the next line, and so forth. Sample input and the results are
shown below.
Please enter 10 numbers:

10 31 2 73 24 65 6 87 18 9
Your numbers are:

10 9
31 18
2 87
73 6
24 65
43. Write aprogram that reads nine integers and prints them three in a line
separated by commas as shown below.
Input:
10 31 2 73 24 65 6 87 18
Output
10 , 31 , 2
73 , 24, 65
6 , 87, 18
Structure of a C Program
-
Two features set the C language apart from many other languages: expres
heart oi the lan -
sions and pointers. Both of these concepts lie at the very
guage , giving C its unique look and feel .
This chapter explores the first of these concepts: expressions. Expres-
sions are not new to you ; you have used them in mathematics.
However, C’s
use of expressions is unique to the C language.
Closely tied to the concept of expressions are operators, precedence and
associativity, and statements, all of which are discussed in this chapter
. This
chapter also introduces a concept known as side effects and explains in detail
how it affects statements in C .
Objectives
To be able to list and describe the six expression categories
To understand the rules of precedence and associativity in evaluat
ing
expressions
To understand the result of side effects in expression evaluation
To be able to predict the results when an expression is evaluated
To understand implicit and explicit type conversion
To understand and use the first four statement types: null, expression
,
return, and compound
93
Section 3.1 Expressions
3.1 Expressions that reduces to a sin -

ofoperands and p
An expression is a sequenee operator is a syntactical
gle value . Expressions can
e
^^ ^
^ operand is an object on which an
° Lee;:I:whose
PeA:eTpLs
^ ssion contains 'nly one o
.
operator. For example 2 + S is a
, a s mplc express,on . A
simple value is 7; similarly, -a
plex expression is 2 + 5 * 7. To
expressions .
that one operator. An example of a com -
complex expression contains more a complex express,on , we reduce to
evaluate
In the previous example we first evaluate
. the
a series of simple 2 + 3 % gmng a result
then the expression
simple expression 5 * 7 (35) and
of 37. are clearly specified in the
Every language has operators whose actions
language syntax. The order in which
the operators in a complex expression
of known as precedence; the
are evaluated is determined by a set priorities
the expression containing the operator is
higher the precedence, the earlier
the front cover contains the precedence of each
evaluated . The table inside
operator. Looking at that table, we see that in
the expression 2 + 5 * 7, multi-
is eval -
plication has a higher priority than addition the multiply expression
so
detail in the next section .
uated first. We discuss precedence in more
If two operators with the same precedence occur in a complex expression,
another attribute of an operator, its associativity, takes control. Associativity
is the parsing direction used to evaluate an expression . It can be either lelt - to-
right or right- to- left . When two operators with the same precedence occur in
-
an expression and their associativity is left to- right, the left operator is evalu -
ated first . For example, in the expression 3 * 4 / 6, there are two operators,
multiplication and division , with the same precedence and left -to- right asso-
ciativity. Therefore, the multiplication is evaluated before the division . We
also discuss associativity in more detail in the next section .
An expression always reduces to a single value.
We can divide simple expressions into six categories based on the number
ol operands, relative positions of the operand and operator, and the prece-
dence of operator. Figure 3- 1 shows the categories.
Expression
Categories
Primary Postfix Prefix Unary Binary Ternary
FIGURE 3- 1 Expression Categories

Primary Expressions
The most elementary type of expression is a primary expression . A primary
expression consists of only one operand with no operator. In C , the operand
in the primary expression can he a name, a constant, or a parenthesized
expression . Although a primary expression has no operator, the null operator
in this expression has the precedence of 16 according to the table ol
prece -
dence. In other words , a primary expression is evaluated first in a complex
expression .
Names
the
A name is any identifier for a variable , a function , or any other object in
language. The following are examples of some names used as primary
expressions:
b!2 price calc INT MAX SIZE
Literal Constants
The second type of primary expression is the literal constant . As discussed in
Chapter 2 , a constant is a piece of data whose value can t change during the
’
execution of the program . The following are examples of literal constants

used as primary expressions :
5 123.98 A' " Welcome"
Parenthetical Expressions
The third type of primary expression is the parenthetical expression. Any
value enclosed in parentheses must be reducible to a single value and is
therefore a primary expression . I bis includes any of the complex expressions
when they are enclosed in parentheses. Thus, a complex expression can be
enclosed in parentheses to make it a primary expression . The following are
primary expressions:
* 3 + 4) (a = 23 b * 6)
(2
Postfix Expressions
The postfix expression consists of one operand followed by one operator
. Its
category is shown in Figure 3-2 . There are several operators
that create a
table . We discuss only
postfix expression as you can see in the precedence .
three of them here: function call , postfix increment , and postfix decrement
96 Section 3.1 Expressions
Operand Operator
FIGURE 3-2 Postfix Expressions
Function Coll
We have already
. In the hello world program , we
the printf function. Function calls are
wrote a message on
postfix expressions. The function name is the
parentheses that follow the name. The
or be empty. When present , the
operand and the operator is the
parentheses
arguments are part of
may conta n arguments
the operator.
.
Postfix Increment/Decrement
The postfix increment and postfix decrement are also postfix operators Virtu -
,
ally all programs require somewhere in their code that the \ ilue 1 be added
«
to a variable. In early languages, this additive operation could only be repre-

sented as a binary expression . C provides the same functionality in two
expressions : postfix and prefix.
In the postfix increment , the variable is increased by 1 . Thus, a + + results
in the variable a being increased by 1. The effect is the same as a = a + 1.
( a ++ ) has the same effect as ( a = a + 1)
Although both the postfix increment ( a ++ ) and binary expression ( a = a +

1) add 1 to the variable , there is a major difference. The value of the postfix
increment expression is determined before the variable is increased . For
instance , if the variable a contains 4 before the expression is evaluated , the
value of the expression a + + is 4. As a result of evaluating the expression and
its side effect , a contains 5. The value and side effect of the postfix increment
are graphically shown in Figure 3 3. -
x = a
Q value of expression is a
x — a+ +
0 value of a is incremented by 1
a = a +l
FIGURE 3 - 3 Result of Postfi x a + +
Chapter 3 Structure of aiCProgram 97
—
the postfix decrement ( a ) also has a value and a side effect . As with
the increment , the value of the expression is the value of a before the decre-
ment ; the side effect is the variable is decremented by I .
The operand in a postfix expression must be a variable .

-
Program 3 1 demonstrates the effect of the postfix increment expression .
PROGRAM 3 - 1 Demonstrate Postfix Increment

1 /* Example of postfix increment.
2 Written by:
3 Date:
4 */
6 int main ( void )
7 {
9 int a;
10
11 // Statements
12 a = 4;
13 printf( " value of a : % 2d \n" , a );
14 printf( " value of a++ : %2d \n" , a++ );
15 printf( "new value of a: %2d \n\n" , a );
16 return 0;
17 > / / main
Results:
value of a 4
value of a++ 4
new value of a: 5
Prefix Expressions
In prefix expressions , the operator comes before the operand as seen in
Figure 3 - 4 .
^ ^
( OperatorJ ) Operand
Variable
FIGURE 3 - 4 Prefix Expression

decrement operators, subtracting 1 from a variable.

shorthand notations for adding or
must be a variable.
The operand of a prefix expression
the postfix and prefix operators,

There is one major difference between
, the effect takes place k/ore the expres-
however: with the prefix operators that this is he reverse of
sion that contains the operator is evaluated . Note
graphically.
the postfix operation . Figure 3- 5 shows the operation
a = a + 1
Q value of a is increment by 1
X = ++ a
Q value of expression is a after increment
FIGURE 3 - 5 Result of Prefix ++ a
The effect of both the postfix and prefix increment is the same: The vari -
able is incremented by 1 . If we don 't need the value of the expression — that
is, if all we need is the effect of incrementing the value of a variable by I
then it makes no difference which one we use. You will find that programmers
—
use the postfix increment and decrement more often , if for no other reason
than that the variable is shown first and is therefore easier to read .
( ++ a ) has the same effect as ( a

= a + 1)
On the other hand , if we require both the value and the

effect , then our
application determines which one we need to use. When
we need the value of
expression to he the current value of the
variable, we use the postfix oper-
tor; when we
nef
"been incremented orthe val e to be the new value of the variable ( after it has
“
decremented ), we use the prefix operator. Program 3- 2
d monstra es the prefix increment
expression . Studv it carefully, and com -
pare it to the results from Program 3
-1 .
PROGRAM 3 - 2 Demonstrate Prefix Increment
1 / * Example of prefix
2
increment.
Written by:
continuei
PROGRAM 3 - 2 Demonstrate Prefix Increment (continued )

3 Date:
4 */
6 int main ( void )
7 {
9 int a ;
10
11 // Statements
12 a = 4;
13 printf( "value of a : % 2d\n " , a );
14 printf( "value of ++a : % 2d\n" , ++a );
15 printf( " new value of a: % 2d \n" , a );
16 return 0;
17 > // main
Results:
value of a 4
value of ++a 5
new value of a: 5
Program 3 - 2 Analysis The only difference in the printouts between Program 3- 1 and Program 3 - 2 is the use
of the increment operators. The first program uses the postfix increment; the second
uses the unary prefix increment. In both cases, we start with the same value for a and
it has the same value at the end. But the value of the expression itself is different. To
help remember the difference, use this rule: If the ++ is before the operand, the incre-
ment takes place before the expression is evaluated; if it is after the operand, the
increment takes place after the expression is evaluated.
If ++ is after the operand, as in a++, the increment takes place after the
expression is evaluated. If ++ is before the operand , as in ++a, the increment
takes place before the expression is evaluated .
Unary Expressions
A unary expression , like a prefix expression , consist ol one operator and one
operand . Also like the prefix expression , the operator comes before the oper-
and . Although prefix expressions and unary expressions look the same, they
belong to different expression categories because the prefix expression needs
a variable as the operand while the unary expression can have an expression
or a variable as the operand . Many of the unary expressions are also familiar
to you from mathematics and will require little explanation . In this chapter
we discuss the sizeof operator, the plus/minus operators, and the cast opera -
tor. The others will be discussed in later chapters. The format of the unary
expressions is demonstrated in Figure 3-6.
Operator
Operand
FIGURE 3 -6 Unary Expressions
. , ,, of
.. .
typ« o a prtmary expres-
.... - , “ “ *“'“.‘ .I
tells us the sfte, 1« byte
s”I
.
'
m
pdnt On
.
The sixeof operator
“Specifying
po« W « he
some
t
the «» «f
*hardwa
d rin "le' wrll rllus rare<The
" °ki re. A simple examp
of the m eg r type 2 by
personal comparers, rhe sire the very large supercomputers,
, i is 4 bytes. On
On ,,
'
1
some mainframe compoters
it can be as large as 16 bytes
know the exact size (in
. If it is important to with the integer type as
use the sizeof operator
bytes) of an integer, we can
shown below.
sizeof (int)
It is also possible to find the size of a primary expression. I lere are two
examples.
sizeof -345.23 sizeof x
Unary Plus/Minus
The unary plus and unary minus operators are what we think ol as
simply the
plus and minus signs . In C , however , they are actually operators . Because
, they can he used to compute the arithme tic value of an
they are operators
operand.
The plus operator does not change the value of the expression. II the
expression’s value is negative, it remains negative; if the expression’s value is
positive, it remains positive.
The minus operator changes the sign of a value algebraically— that is, to
change it from plus to minus or minus to plus. Note, however, that the value
ol the stored variable is unchanged. The operation of these operators is seen
in Table 3 1. -
Expression Contents of a Before Expression
ond After Expression Value
+a 3 +3
-a 3 -3
continued
TABLE 3 - 1 Examples of Unary Plus And Minus Expressions
Chapter 3 Structure of a C Program J 01
Expression Contents of a Before Expression

and After Expression Value
-5 -5
+a
-5 +5
-a
TABLE 3 - 1 Examples of Unary Plus And Minus Expressions (continued
)
Cast Operator
The third unary operator we discuss in this chapter is cast . The cast operator
converts one expression type to another. For example to
, convert an integer to
a real numbe r, we would use the followin g unary express ion.
( float ) x
It is important to note that, as with all of the unary operators, only

the
, unchan ged . We dis -
expression value is changed. T he integer variable x , is
” in
Explicit Type Conver sion ( Cast )
cuss the cast operator in detail in “
Section 3.5 .
Binary Expressions
a-
Binary expressions are formed by an operand-operator -operand combin
tion. T hey are perhaps the most common expression category
. Any two num -
bers added, subtracted, multiplied, or divided are usually formed in algebra ic
notation, which is a binary expression. There are many binary express
ions.
We cover the first two in this chapte r. Figure 3 - 7 shows the format of a binary
expression.
Operand I
^ ^
( OperatoT ) Operand
I
FIGURE 3 - 7 Binary Expressions
Multiplicative Expressions
The first binary expression we study, multiplicative expressions
, which take
its name from the first operator, include the multiply
, divide , and modulus
( 1 3 ) among the binary
operators. These operators have the highest priority
operators and are therefore evaluat ed first among them .
ds.
The result of a multiply operator ( * ) is the product of the two operan
The operands can he any arithmetic type ( integral or floating point
- ) . The type
in the chapter.
of the residt depends on the conversion rule that we discuss later
// evaluates t o 30
10 * 3 / / evaluates t o 4
true * 4 // evaluates to 130
A' * 2 // evaluates to 44.6
22.3 * 2
,,
o II P "r
' * 7'; I*
one or j
quotient • H
•'of the"division
w <«examples
«• ,
. ‘"
'
the quotient. The following shows some operator
// evaluates to 3
10 / 3 // evaluates to 0
true / 4 to 32
// evaluates
•A' / 2 to 11.15
// evaluates
22.3 / 2
with the
Multiply and divide are well known , hut you may not be familiar
modulus operator ( % ) , more commonly known as modulo . I his operator
divides the first operand by the second and returns the remainder
rather than
the quotient . Both operands must be integral types and the operator returns
the remainder as an integer type . The following examples demonstrate the
modulo operator .
10 % 3 // evaluates to 1
true % 4 // evaluates to 1
• A ' % 10 // evaluates to 5
22.3 % 2 // -
Error : Modulo cannot be f l o a t i n g point
Because the division and modulus operators are related , they are olten
confused. Remember: The value of an expression with the division operator is
the quotient ; the value of a modulus operator is the remainder. Study the
effect of these two operators in the following expressions:
3 / 5 / / evaluates to 0
3 % 5 / / evaluates to 3
Another important point to remember is, if the first integral operand is

smaller than the second integral operand, and the result of division is 0, the
result of the modulo operator is the first operand
as shown below:
3 / 7
3 % 7 / / evaluates to 0
/ / evaluates to 3
Both operandsofthem oMo

.
operator (7 ) must be integral types .
Additive Expressions
In additive expressions, the second operand is added to or subtracted
from
the first operand, depending on the operator used. I lie operands in an addi -
tive expression can be any arithmetic types ( integral or floating-point ). Addi -

( 13 );
tive operators have lower precedence ( 12 ) than multiplicative operators
therefore, they are evaluated after multiplicative expressions. Two simple
examples are shown below:
3 + 7 // evaluates to 10
3 7- // evaluates to 4 -
EXAMPLE 3 - 1 Binary Expressions

Now lets look at short program that uses some of these expressions. Pro-
a
gram 3 - 3 contains several binary expressions.
PROGRAM 3 - 3 Binary Expressions

1 /* This program demonstrates binary expressions.
2 Written by:
3 Date:
4 */
6 int main ( void )
7 {
9 int a = 17 ;
10 int b = 5;
11 float x = 17.67 ;
12 float y = 5.1;
13
14 // Statements
15 printf("Integral calculations\n " );
16 printf( " %d + %d %d \n " , a , b , a + b );
17 printf("%d - %d = %d\n" , a , b , a -
b);
18 printf( " %d * %d = % d \n " , a , b , a * b );
19 printf( " %d / %d %d \n " / a , b , a / b );
20 printf( " %d % % %d = %d \n", a, b , a % b );
21 printf( " \ n " );
22
" );
23
24
printf( "Floating- point calculations
printf( " %f + % f = %f \n " , x , y , x +
^yj;
%f = %f \n" , x , Y f x - y);
25 printf("%f -
26 printf( " % f * % f = % f \n " , x , Y , x * y);
27 printf( "%f / %f = %f\n" , x , y , x / y);

continue d
(continuedJ
PROGRAM 3- 3 Binary Expressions
28 return 0;
29 > // main
Results:
integral calculations
17 + 5 = 22
17 5 = 12
17 * 5 = 85
17 / 5 = 3
17 % 5 = 2
Floating point calculations

-
17.670000 + 5.100000 = 22.770000
17.670000 - 5.100000 = 12.570000
17.670000 * 5.100000 = 90.116997
17.670000 / 5.100000 = 3.464706
comments. ( 1 ) Note that even

Proqram 3 - 3 Analysis This simple program requires only three explanatory documentation comments. [2]
for asimple program we include all of the standard
statements as we have
We do not recommend that you include calculations in print
done in this program -it is not a good structured programming technique.
We
because we haven ' t yet shown you how to save the
include them in this program
results of a calculation. (3) Study the format string in statement 20. To print a percent
sign as text in the format string, we need to code two percent signs.
Assignment Expressions
The assignment expression evaluates the operand on the right side of the
operator ( = ) and places its value in the variable on the left . The assignment
expression has a value and a side effect .
• The value of the total expression is the value of the expression on the
right of the assignment operator ( = ).
• I he side effect places the expression value in the variable on the left of
the assignment operator.
The left operand in an assignment expression must be a single variable .

-
There are two forms of assignment , simple and compound .
Simple Assignment
Simple assignment is found iin algebraic expressions
. Three examples of sim -
pie assignments are shown below !
a = 5 b = x + l
i = i + 1
Of course, for the effect to take place, the left variable must be able to
receive it; that is, it must he a variable, not a constant. II the left operand
cannot receive a value and we assign one to it, we get a compile error.
Compound Assignment
A compound assignment is a shorthand notation for a simple assignment It .
requires that the left operand he repeated as a part of the right expression.
Five compound assignment operators are discussed in this chapter: * =, / = , % =,
+=, and
To evaluate a compound assignment expression, first change it to a sim-
ple assignment, as shown in Table 3 - 2. Then perform the operation to deter -
mine the value of the expression.
Compound Expression Equivalent Simple Expression

x *= expression x = x * expression
x / - expression x = x / expression
x %= expression x = x % expression
x += expression x = x + expression
x -= expression x = x - expression
TABLE 3 - 2 Expansion of Compound Expressions

When a compound assignment is used with an expression, the expression
is evaluated first . Thus, the expression
x *= y + 3
is evaluated as
x x * (y + 3)
which, given the values x is 10 and y is 5 . evaluates to 80.

Program 3 - 4 demonstrates the first three compound expressions.
PROGRAM 3 - 4 Demonstration of Compound Assignments

1 /* Demonstrate examples of compound assignments.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
continued
106 Section 3.2 Precedence and Associativity
Assignments (continued)
PROGRAM 3 - 4 Demonstration of Compound
10 int x;
11 int y;
12
13 // statements
14 x = 10?
15 y = 5;
16
x, y ) ;
17 printf("x: %2d | y: %2d " ,
printf("|x *= y + 2: %2d " , x *= y + 2 );
18
19 printf( |
" x is now: %2d\n " , x ) ;
20
21 x = 10?
22 printf("x: %2d | y: %2d " , x , y )?
23 printf( " | x /= y + 1: %2d " , x /= y + 1)?
24 printf(" | x is now: %2d\n", x)?
25
26 x = 10?
27 printf( " x: %2d | y: %2d ", x, y ) ;
28 -
printf(" | x %% = y 3: %2d ", x %= y 3);
29 printf( " | x is now: %2d \n", x )?
30
31 return 0;
32 } // main
Results:
x: 10 | y 5 x *= y + 2: 70 x is now: 70
x: 10 I
y: 5 | x /= y + 1: 1 x is now: 1
x: 10 | y 5 | x %= y 3: 0 - I x is now: 0
Program 3 - 4 Analysis Note that we have used an assignment statement in the printf statements to demon-
strate that an assignment expression has a value . As we
said before, this is not good
programming style, but we use it here to match the format we used in
Do not hide calculations in print statements . Also, Program 3-3.
since we are changing the value of
x with each assignment, even though it
is in a printf statement, we need to reset it to
10 for each of the print series.
3.2 Precedence and Associativity

cPo3exneCl e ssi o end
-“ -
t0
°r d e r i n w h i c h d i f fd eUn» t doepteerramt oi nr se i tnh ae
o Z t Z r"
plex expression.
Another
'
P,rCCedence are evaluated in a corn -
wav of a ng l
how operators with the S e
" “ ’"
, '
Precedence ^ 'ls s associativity determines
grouped together to form complex
expressions. Precedence is applied before associativity to determine the order

in which expressions are evaluated . Associativity is then applied , if necessary.
Precedence
The concept of precedence is well founded in mathematics. For example, in
algebra , multiplication and division are performed before addition and sub-
traction . C extends the concept to 16 levels , as shown in the Precedence
Iahle inside the front cover.
I he following is a simple example of precedence:
2 + 3 * 4
This expression is actually two binary expressions, with one addition and
one multiplication operator. Addition has a precedence ol 12 . Multiplication
has a precedence of 13. This results in the multiplication being done first ,
followed by the addition , as shown below in the same expression with the
default parentheses added. The value of the complete expression is 14 .
(2 + (3 * 4) ) —> 14
As another example consider the following expression :
-b+ +
Two different operators are in this expression . Ihe first is the unary
minus the second is the postfix increment. The postfix increment has the
higher precedence ( 16 ) , so it is evaluated first . Then the unary minus , with a
precedence of 15 , is evaluated . To reflect the precedence, we have recoded
the expression using parentheses.
( - ( b+ + ) )
Assuming that the value of b is 5 initially, the expression is evaluated to

-5 . What is the value of b after the expression is completer It is 6 because the
operator has an effect that is separate from the value of the expression.
Program 3- 5 demonstrates precedence by printing the same expres-
sion , once without parentheses and once with parentheses to change the
precedence. Because the parentheses create a primary expression that must
he evaluated before the binary multiply, the answer is diflerent .
PROGRAM 3-5 Precedence

1 /* Examine the effect of precedence on an expression.
2 Written by:
continued
108 Section 3.2 Precedence and Associativity
PROGRAM 3- 5 Precedence ( continued )

3 Date:
4 */
5 # include <stdio.h
>
6
7 int main ( void)
8 {
10 int a = 10;
11 int b = 20;
12 int c 30; =
13
14 // Statements
+ c );
15 printf ("a * b + c is: %d\n" , a * b
: %d\n", a * (b + c) );
16 printf ("a * (b + c) is
17 return 0;
18 > // main
Results:
a * b + c is: 230
a * ( b + c) is: 500
Associativity
Associativity can be left - to- right or right - to- left. Left-to- right associativity
evaluates the expression by starting on the left and moving to the right . Con -
versely, right - to -left associativity evaluates the expression by proceeding
Irom the right to the left . Remember, however, that associativity is used only
when the operators all have the same precedence.
Associativity is applied when we have

more than one operator of the
s. same precedence level >
in an expression.
Left-to-right Associativity
The following shows an example of left- to
four operators of the same precedence ( -
right associativity. Here we have
* / % *) .
3 * 8 / 4 % 4
* 5
LL?V
Associativity determines how the subexpressions are grouped together.

All of these operators have the same precedence ( 13 ). Their associativity is
from left to right. So they are grouped as follows:
(((( 3 * 8) / 4) % 4 ) * 5 )
The value of this expression is 10. A graphical representation of this

expression is shown in Figure 3 - 8.
3 8 / 4 % 4 * 5
FIGURE 3 - 8 Left - to- right Associativity
Right-to-left Associativity
Several operators have right- to-left associativity as shown in the precedence
table. For example, when more than one assignment operator occurs in an
assignment expression, the assignment operators must he interpreted from
right to left. This means that the rightmost expression will he evaluated first ;
then its value will he assigned to the operand on the left of the assignment
operator and the next expression will he evaluated. Under these rules, the
expression
a + = b *= c -= 5
is evaluated as
(a += ( b *= (c -= 5 )))
which is expanded to
(a = a + (b = b * (c = c - 5)))
If a has an initial value of 3, b has an initial value of 3, and c has an ini-

tial value of 8, these expressions become
(a = 3 + (b = (5 * (c = 8 - 5 )))
which results in c being assigned a value of 3, b being assigned a value of 1 3 ,

and a being assigned a value of 18. The value of the complete expression is
also 18. A diagram of this expression is shown in Figure 3 -9.
L
110 Section 3.3 Side Effects
FIGURE 3 - 9 Right- to-left Associativity

-
a + b *= c
— 5
assignment is shown below. Suppose we

A simple but common form of
to be initialized to zero . Rather than ini -
have several variables that all need complex statement to do it .
tializing each separately, we can use a
a = b = c = d = 0;
3.3 Side Effects

A side effect is an action that results from the evaluation
of an expression.
, C first evaluates the expression on the right (,f
For example, in an assignment
the assignment operator and then places the value in the left variable. Chang-
ing the value of the left variable is a side effect. Consider the
following
expression :
x = 4;
This simple expression has three parts. First, on the right ol the assign -
ment operator is a primary expression that has the value 4 . Second , the whole
expression ( x = 4 ) also has a value of 4 . And third , as a side effect , x receives
the value 4 .
Let 's modify the expression slightly and see the same three parts.
x = x + 4;
Assuming that x has an initial value of 3, the value of the expression on

the right of the assignment operator has a value of 7 . The whole expression
also has a value of 7. And as a side effect, x receives the value ol 7 . Fo prove
these three steps to yourself, write and run the following code fragment:
i n t x = 3;
——
p r i n t f ( " Step 1 Value of x: %d
p r i n t f ( " Step 2 Value of x
\n" , x ) ;
—
p r i n t f ( " Step 3 Value of x
Now, let’s consider the side

=
x + 4 : %d \ n " , x
now: %d \ n " , x ) ;
= x + 4);
effect in the postfix increment expression,

This expression is typically coded
as shown below.
a ++
As we saw earlier, the value of this expression is the value of a before the
expression is evaluated . As a side effect , however, the value of a is incre -
mented by 1 .
In C , six operators generate side effects: prefix increment and decrement ,
postfix increment and decrement , assignment , and function call.
3.4 Evaluating Expressions

Now that we have introduced the concepts of precedence, associativity, and
side effects, let s work through some examples.
Expressions without Side Effects

EXAMPLE 3 - 2 Expression with no Side Effects
The first expression is shown below. It has no side effects, so the values of all
ol its variables are unchanged .
a 4 + b / 2 c b
For this example, assume that the values of the variables are
3 4 5
a b c
Io evaluate an expression without side effects, follow the simple rules

shown below.
1 . Replace the variables by their values. I bis gives us the following expression :
3 * 4 + 4 / 2 5 * 4
2. Evaluate the highest precedence operators, and replace them with the
resulting value. In the above expression , the operators with the highest
precedence are the multiply and divide ( 13). We therefore evaluate them
first from the left and replace them with the resulting values. The expres-
sion is now
(3 * 4) + (4 / 2) (5 * 4) —> 12 + 2 - 20
3. Repeat step 2 until the result is a single value.

In this example , there is only one more precedence , binary addition and
subtraction . After they are evaluated , the final value is -6 . Since this
expression had no side effects, all of the variables have the same values after
the expression has been evaluated that they had at the beginning.
112 Section 3.4 Evaluating Expressions
Expressions with Side Effects

EXAMPLE 3- 3 Example with Side Effects
that has side effects and paren-
Now let 's look at the rules for an expression
example , consider the expression
thesized expressions. For this
__ a * ( 3 + b) / 2 - C ++ * b
Assume that the variables have the values used above, a is 3, b is 4 , c is 5.

To evaluate this expression , use the following rules:
.
1 Calculate the value of the parenthesized expression ( 3 + b ) first ( prece-
dence 16 ) . The expression now reads
— a * 7 / 2 - C ++ * b
2 . Evaluate the postfix expression ( c ++ ) next ( precedence 16 ) . Remember

that as a postfix expression , the value of C ++ is the same as the value of c;
the increment takes place after the evaluation . The expression is now
— a * 7 / 2 - 5 * b
——
3. Evaluate the prefix expression ( a ) next ( priority 1 3 ) . Remember that as
a prefix expression , the value of a is the value after the side effect ,
which means that we first decrement a and then use its decremented
value. The expression is now
2 * 7 / 2 - 5 * b
4. 1 he multiply and division are now evaluated using their associativity rule
lelt to right , as shown below.
14 / 2 - 5 * b -» 7 - 5 * 4 -> 7 - 20
5 ' The lasl step is to
evaluate the subtraction. The final expression value is
13 as shown in the final example.
7 “ 20 -> 13
-
Alter the side effects, the
variables h ave the values shown below.
2
a b
6
Program 3- 6 evaluates
the tw0 expressions
in this section .
PROGRAM 3 - 6 Evaluating Expressions

1 /* Evaluate two complex expressions.
2 Written by:
3 Date:
4 */
5 # include <stdio. h>
6 int main ( void )
7 {
9 int a = 3 ;
10 int b = 4;
11 int c 5
12 int x ;
13 int y ;
14
15 // Statements
16 printf( "Initial values of the variables: \n " );
17 printf( " a = %d \tb = %d \tc = %d\n\n " a, b , c );
/
18
19 x = a * 4 + b / 2 - c * b;
20 printf
21 ( " Value of a * 4 + b / 2 c * b: %d \n" , x );
22
23 y = — a * ( 3 + b ) / 2 - C ++ * b ;
24 printf
25 ( " Value of --
a * ( 3 + b) / 2 - C ++ * b: %d \n" , y );
26 printf( "\nValues of the variables are now: \n" );
27 printf( " a = %d \tb = %d\tc = %d \n\n " , a, b , c );
28
29 return 0;
30 } // main
Results:
Initial values of the variables:
a = 3 b = 4 c = 5
Value of a * 4 + b / 2 c * b: -6
Value of
— a * ( 3 + b) / 2 - C ++ * b: -13
Values of the variables are now:
a = 2 b = 4 c = 6
Warning
A warning is in order: In C , if an expression variable is modified more than
once during its evaluation , the result is undefined . C has no specific rule to
114 Section 3.5 Type Conversion
give different expression results

.
In
/ / C, if a
/ single variable \\
is modified more than \\
once in an expression,
the r e s u l t i s undefined .
WARNING
3.5 Type Conversion

Up to this point, we have assumed that all of our expressions involved data of
the same type. But, what happens when we write an expression that involves
two different data types, such as multiplying an integer and a floating- point
number? To perform these evaluations, one of the types must be converted.
Implicit Type Conversion

When the types of the two operands in a binary expression are different, C
automatically converts one type to another. This is known as implicit type
conversion. I or implicit type conversion, C has several complicated rules
that we gradually introduce throughout the book. We mention some of the
simple conversions in this section.
Conversion Rank
Before we discuss how conversions are handled,
we need to discuss the con-
cept ol conversion rank. In C, we can
assign a rank to the integral and
floating-point arithmetic types. Figure 3- 10
for conversion in this chapter. While
shows the ranks as we use them
the 1 to 9 scale we use is conceptually
correct , the actual implementation
is much more complex
intevt " Fi§Ure 3' , 3 l0 d uble
u integer
integer andI a short 10 US
° »
real »s a higher rank than a long
has a higher rank than a character.
——
Conversions in Assignment Expressions
,Depending
he
* ^- ^ *•-.-K
on^thed fference n the
»
rank" t"'
'
^ tW
°
occurs if the right expression has lower rank; demotion occurs il the right
expression has a higher rank.
Real
9. long double
8. double
Integer 7. float
6. long long
5. long
4. int
Character 3. short
Boolean 2. char
1. bool
FIGURE 3 - 10 Conversion Rank
Promotion
.
There is normally no problem with promotion The rank of the right expres -
.
sion is elevated to the rank of the left variable The value of the expression is
the value of the right expression after the promotion. The following examples
demonstrate some simple promotions.
bool b = true ;
char c = 'A';
int i = 1234 ;
long double d = 3458.0004 ;
c = b; // value of c is SOH ( ASCII 1 )
I c; // value of i is 65
d = b; // value of d is 1.0
d = i; // value of d is 1234.0
Demotion
Demotion may or may not create problems. If the size of the variable at the
left side can accommodate the value of the expression, there is no problem;
however, some of the results may surprise you . »
Any integral or real value can be assigned to a Boolean type. II the value
of the expression on the right is zeroy false ( 0 ) is stored; if the result is not
zero, either positive or negative, true ( 1 ) is stored.
When an integer or a real is assigned to a variable ol type character, the
least significant byte of the number is converted to a character and stored.
When a real is stored in an integer, the fraction part is dropped. However, if
the integral part is larger than the maximum value that can he stored, the
results are invalid and unpredictable. Similarly, when we try to store a long
double in a variable of type flout , the results are valid if the value fits or
invalid if it is too large.
.
demonstrate demotion
The following examples
bool b = false;
char c = A' ;
short s = 78;
int j = INT MAX ;
int k = 65;
1 (true)
// value of b is
b = c; // value of s is unpred ictable
s = j; : value of c is B
// demotion
c = k + 1;
Conversion in Other Binary Expressions ons. The

of rules for the other binary expressi
Conversion has a different set , but we can summarize
them in three
rules are sometimes very complicated
:
steps, which cover most cases
1 . The operand with the higher
rank is determined using the ranking in
Figure 3- 10.
to the rank defined in step 1.
2. The lower-ranked operand is promoted
After the promotion, both expressions have the same rank.
3. The operation is performed with the expression
value having the type of
the promoted rank.
I he lollowing examples demonstrate some common conversions
.
bool b = true;
char c = 'A ' ;
int i = 3650;
short s = 78;
long double d = 3458.0004 ;
b + c // b promoted ; result is 'B' ('A ' + 1)
i * s; // result is an int
d * c; // result is long double
Let’s look at a small program to see the effect of implicit conversions. In

Program 3- 7 we add a character, an integer, and a float . We can add charac -
ters to integers and lloating- point values because all characters have an
ASCII value that can be promoted.
PROGRAM 3-7 Implicit Type Conversion

1|/* Demonstrate .automatic promotion of numeric types .
2 Written by:
3 Date:
4 */
continue d
PROGRAM 3 -7 Implicit Type Conversion (continued )

6 I # include <stdbool.h >
7
8 int main ( void )
9 {
11 bool b = true;
12 char c = ' A';
13 float d = 245.3 ;
14 int i = 3650;
15 short s = 7 8;
16
17 // Statements
18 printf( " bool + char is char: %c\n" , b + c );
19 printf( "int * short is int: %d \n" , i * s);
20 printf( "float * char is float: %f \n" , d * c);
21
22 c = c + b; // bool promoted to char
23 d = d + c; // char promoted to float
24 b = false;
25 b = - d; // float demoted to bool
26
27 printf( "\nAfter execution...\n" );
28 printf( "char + true: %c\n" , c )
29 printf( " float + char: %f \n" , d )
30 -
printf( " bool = float: %f \n " , b)
31
32 return 0 ;
33 > // main
Results:
bool + char is char: B
int * short is int: 284700
float * char is float: 15944.500000 !
After execution...
char + true: B
float + char: 311.299988
-
bool = float: 1
Program 3 -7 Analysis Several points in this program require explanation. First, as we stated before, it is not
a good programming practice to code an expression in a print statement. We do it in
this program, however, to demonstrate the promotion or demotion of the expression.
The first print series displays the value of mixed type expressions. As you exam -
ine each result, note that the value printed is in the form of the higher ranked vari -
able. For example, in statement 18 the Boolean in the expression (b) is promoted to a
of the character expression ( c ) . The result is

character and then added to the value the printf function where it is printed using
the character B, which is then passed to
the format specification %c. . t
assignments. In the hrst one, we add
The second print series displays the results of
true ( 1 ) to the letter A. The result is B as you
would expect. In the second assignment, we
real number. The new value of the real
add the letter B from the previous assignment to a . The difference occurs because real
number is almost 66 greater than the original value
value stored was 245.299988.
numbers are not exact. Rather than storing 245.3, the , real number to a boolean .
Finally, note what happens when we assign a negative
The result is true .
Explicit Type Conversion (Cast )

Rather than let the compiler implicitly convert data , we can convert data
from one type to another ourself using explicit type conversion . Explicit
type conversion uses the unary cast operator, which has a precedence of 14.
To cast data from one type to another, we specify the new type in parentheses
before the value we want converted . For example, to convert an integer, a , to
afloat , we code the expression shown below.
( float ) a
Note that in this operation, like any other unary operation , the value stored
in a is still of type int , hut the value of the expression is promoted to float .
One use of the cast is to ensure that the result of a divide is a real num -
ber. For example, il we calculated the average of a series of integer test scores
without a cast , the result would he an integer. To force a real result, we cast
the calculation as shown below .
average ( float ) totalScores / numScores ;
In this statement, there is an explicit conversion of

totalScores to float ,
and then an implicit conversion ol numScores
so that it will match . I he
lesult ol the divide is then a lloating- point
number to he assigned to average.
But beware! \\ hat would he the result of
is 3?
the following expression when a
(float ) (a / 10 )
Are you surprised to find that
the result is ().()? Since no conversions are
required to divide integer 3 by
integer 10, C simply divides with an integer
result, 0. 1 he integer 0 is then
To get afloat result, c must explicitly converted to the lloating- point 0.0.
" cast one of the
numbers as shown below.
( float ) a / io
One final thought about casts: Even when the compiler can correctly cast
for you, it is sometimes better to code the cast explicitly as a reminder that
the cast is taking place.
Program 3 - 8 demonstrates the use of explicit casts. In this program, we
divide several mixed types. While the results are nonsense, they demonstrate
the effect of casting.
PROGRAM 3 - 8 Explicit Casts

1 /* Demonstrate casting of numeric types.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
10 char aChar = ’ NO ' ;
11 int intNuml = 100 ;
12 int intNum2 = 45 ;
13 double fltNuml = 100.0;
14 double fltNum2 = 45.0 ;
15 double fltNum3 ;
16
17 // Statements
18 printf( "aChar numeric : %3d \n" , aChar );
19 printf( " intNuml contains: %3d \n " , intNuml );
20 printf( " intNum2 contains: % 3d\n" , intNum2);
21 printf( " fltNuml contains: %6.2f \n " , fltNuml );
22 printf( " fltNum2 contains: %6.2f \n" , fltNum2 );
23
24 fltNum3 = ( double)( intNuml / intNum2 );
25 printf
26 ( " \n( double )( intNuml / intNum2 ): %6.2f \ n " ,
27 fltNum 3 );
28
29 fltNum 3 = ( double)intNuml / intNum2 ;
30 printf( "( double) intNuml / intNum2 : %6.2f \n " ,
31 fltNum3 );
32
33 aChar = ( char )( fltNuml / fltNum2 );
34 printf( " ( char )( fltNuml / fltNum2 ): %3d \n " , aChar );
35
continued
120 Section 3.6 Statements
PROGRAM 3 -8 Explicit Casts ( continued

)
36 return 0;
37 y // main
Results:
aChar numeric : 0
intNuml contains: 100
intNum2 contains: 45
fltNuml contains: 100.00
fltNum2 contains: 45.00
2.00
(double)(intNuml / intNum2)
2.22
(double) intNuml / intNum2
2
(char)(fltNuml / fltNum2)
statements 24 and 29 is the use

Program 3- 8 Analysis Study the casts carefully. The only difference between
of parentheses around the calculation . In statement 24, both operands are integers so
the result of the division is integer, which is then cast to a double . In statement 29,
intNuml is cast to a double. The compiler automatically casts intNum 2 to a double
before the division. The result is therefore a double. Finally, in statement 33, we cast
the result of the integer division into a character.
3.6 Statements
A statement causes an action to be performed by the program . It translates
directly into one or more executable computer instructions.
You may have noticed that we have used a semicolon at the end of the
statements in our programs. Most statements need a semicolon at the end ;
some do not . When we discuss statements, we identify those that do.
Statement Type
C defines eleven types of statements , which are shown in Figure 3 - 1 1 . In this
chapter we will discuss the first four; the other types will be covered in the
future chapters.
Null Statement
The null statement 15
is Just
i a semicolon ( the terminator ) as shown below:
// null statement
Although they do not arise often,
we must have a statement there are syntactical situations where
but no action is required. In
use the null statement. these situations, we
Expression Statement
An expression is turned into a statement by placing a semicolon ( ; ) after it .
expression ; // expression statement
Null
Expression
Return
Compound
Conditional
Statement Labeled
Switch
Iterative
Break
Continue
Goto
FIGURE 3 - 1 1 Types of Statements
When C sees the semicolon , it completes any pending side effects and
discards the expression value before continuing with the next statement . An
expression without side effects does not cause an action . Its value exists and
it can be used , but unless a statement has a side effect , it does nothing.
Let us look at some expression statements. First , consider the expres-
sion statement
a = 2;
The effect of the expression statement is to store the value , 2 , in the vari -
able a . The value of the expression is 2. After the value has been stored , the
expression is terminated ( because there is a semicolon ) , and the value is dis-
carded . C then continues with the next statement .
The next expression statement is a little more complex.
a = b = 3;
This statement actually has two expressions. II we put parentheses

around them , you will be able to see them clearly.
122 Section 3.6 Statements
a * (b = 3);
its value, 3, is dis-

a . Since the expression is
is terminated by the semicolon ,
statement therefore, is that 3 has been
carded. The effect of the expression
stored in both a and b.
In Chapter 2 we examined the scanf and print/ functions
. These state-
of expressions and side
ments present interesting insights into the concepts
effects. Consider the following scan/ function call :
ioResult = scanf( " % d ", & x );
This statement has two side effects. The first is found in the scanf func-
tion. Reading an integer value from the keyboard and placing it into the vari -
able x ( note the address operator before the variable ) is a side eflect . I he
second side effect is storing the value returned by scanf which represents the
number of values that were converted correctly. In this case , the return value
could be EOF, 0, or 1 . Assuming that the user correctly keys the integer, the
value will he 1. The assignment operator stores the return value in ioResult .
I he expression then terminates, and the scant value is discarded .
In a similar fashion , printj has the effect of displaying data on the moni -
tor and returning a value , the number of characters displayed . This is seen in
the statement below.
numDisplayed = printf
( " x contains % d , y contains % d \ n " , x , y ) ;
As a general rule , however, the number of

characters displayed is dis-
carded without being stored . Therefore we normally
code the above state-
ment as shown below.
print
^ "x contains %d , y contains % d \ n ", x , y ) ;
value
\ a l u c of ^ ôwin8 exPression statement. Assume that a has a
^ helore the expression is evaluated .
a + +;
In this postfix expression , the
. value
i changed. .by the
variable , a , before it 15 ,
of the expression is 5 , the value of the
side effect . Upon completion of the
expression statement a is
,
which is still 5 is discarded sntrtmumed
l
°55 The Vyluc of ^-expression ,
"
^ ^^ ^
fon W '3"re*a so expression
COmPlete
Although they are useless 0
I hey are useless because
they hav ° S C c
T statements,
assigned to a variable. We I K H I 1 . * ect ancI their values are not
> °n t use them , hut it is important to know
they are syntactically correct expression statements. C will evaluate them ,

determine their value, and then discard the value. 1
b; 3;
Return Statement
A .
return statement terminates a function All functions, including main ,
.
must have a return statement When there is no return statement at the end of
the function, the system inserts one with a void return value.
return expression ; // return statement
In addition, the return statement can return a value to the calling Iunc -
tion. For example, while we have not talked about it or used it, the scanj func -
.
tion returns the number of variables successfully read In the case of main , it
returns a value to the operating system rather than to another function. In all
of our programs, main returns 0 to the operating system. A return value of
zero tells the operating system that the program executed successfully.
Compound Statements
A compound statement is a unit of code consisting of zero or more state-
.
ments It is also known as a block. The compound statement allows a group
of statements to become one single entity. You used a compound statement in
your first program when you formed the body of the function main . All C
functions contain a compound statement known as the function body .
A compound statement consists ol an opening brace, an optional declara-
tion and definition section, and an optional statement section, followed by a
closing brace. Although both the declaration section and the statement sec -
.
tions are optional, one should be present If neither is present, then we have
a null statement , which doesn’t make much sense. Figure 3 - 1 2 shows the
makeup of a compound statement.
{
Opening int x;
Brace int y;
int z;
// Statements
Closing
x = 1;
y 2;
Brace
> // End Block
FIGURE 3 - 12 Compound Statement
1 . In an optimized compiler, the translator determines that there is no effect and generates no code.
— —
One important point ° r
not need a semicolon . The open
iters lor the compoun s a
ZI.
“X
.
.„
compile error, although »

^
The compound statement does not need
-
C requires that the declaration section be before any
a com pound statement block. The code in the declaration
ment section cannot be intermixed
The Role of the Semicolon

.
that a compound statement does
is
Pg
«
-
amUclosinggFparentheses are simply delim-
^
"*'
a
statements 2 within
section and state-
The semicolon plays an important role in the syntax of the C language. It is

used in two different auscultations.
• Every declaration in C is terminated by a semicolon.
^^^
extra null statement after the
semicolon.
yc
,
» reaming me sage .
„ *,
• Most statements in C are terminated by a semicolon .

On the other hand , we must he careful not to use a semicolon when it is
not needed . A semicolon should not be used with a preprocessor directive
such as the include and define. In particular, a semicolon at the end of a
define statement can create a problem as discussed in the next section .
Statements and Defined Constants

When we use preprocessor- defined commands, we must be very careful to
make sure that we do not create an error. Remember that the defined con -
stant is an automatic substitution . This can
cause subtle problems. One com -
mon mistake is to place a semicolon at the end of
the definition . Since the
preprocessor uses a simple text replacement of
the name with whatever
expression follows , the compiler will generate
a compile error if it finds a
example
^
-” ^ ^ ^^ definition' This Prob|em is seen in the following
#define SALES TAX RATE 0.825 ;

salesTax = SALES TAX RATE * _ _salesAmount;
After the substitution, the
coded a semicolon after the following erroneous code occurs because we
constant value:
salesTax = 0.0825; * salesAmount;

2. The one exception to
this rule is in the for
statement. which we study in Chapter 6.
This can be an extremely difficult compile error to figure out because we

see the original statement and not the erroneous substitution error. One of
the reasons programmers use uppercase for defined constant identifiers is to
provide an automatic warning to readers that they are not looking at the
real code.
3.7 Sample Programs

This section contains several programs that you should study for program -
ming technique and style.
EXAMPLE 3 - 4 Calculate Quotient and Remainder

Lets write a program that calculates and prints the quotient and remainder
oftwo integer numbers. The code is shown in Program 3-9.
PROGRAM 3 - 9 Calculate Quotient and Remainder

1 /* Calculate and print quotient and remainder of two
2 numbers.
3 Written by:
4 Date:
5 */
7
8 int main ( void )
9 {
11 int intNuml ;
12 int intNum2;
13 int intCalc ;
14
15 // Statements
16 printf( " Enter two integral numbers: " );
17 scanf ( " %d %d " , sintNuml , &intNum2 );
18
19 intCalc = intNuml / intNum2;
20 printf( " %d / %d is %d " , intNuml , intNum2 , intCalc );
21
22 intCalc = intNuml % intNum2;
23 printf( " with a remainder of: %d \n" , intCalc );
24
25 return 0;
26 > // main
continued
\j \
126 Section 3.7 Sample Programs
Calculate Quotient and Remainder (continued

)
PROGRAM 3 - 9
Results:
: 13 2
Enter two integral numbers
with a remainder of: 1
13 / 2 is 6
begins with documentation about what

Program 3 - 9 Analysis Using good programming style, the program
it does, who created it , and when it was created .
Program 3-9 has no global variable declarations, so after including the standard
immediately with main. Following main is the opening
input/output library, we start
brace. The matching closing brace is found on line 26 .
The most difficult part of this program is figuring out how to get the remainder. For -
tunately, C has a modulo operator ( % ) that does the job for us. The rest of the problem is
straightforward.
EXAMPLE 3 - 5 Print Right Digit

Another problem that requires the use of the modulo operator is to print a
digit contained in an integer. Program 3- 10 prints the least significant ( right -
most ) digit of an integer.
PROGRAM 3 - 10 Print Right Digit of Integer

1 /* Print rightmost digit of an integer.
2 Written by:
3 Date:
4 */
5 linclude <stdio.h>
6
7 int main (void )
8 {
10 int intNum;
11 int oneDigit;
12
13 // Statements
14 printf("Enter an integral
15 number:
scanf ("%d" , &intNum);
16
17 oneDigit = intNum % 10;
18 printf("\nThe right
digit is: oneDigit );
19 /
20 return 0;
21 } // main
Results:
continued
PROGRAM 3 - 10 Print Right Digit of Integer ( continued )

Enter an integral number: 185
The right digit is: 5
EXAMPLE 3 -6 Calculate Average

Program 3- 1 1 reads Four integers from the keyboard, calculates their average,
and then prints the numbers with their average and the deviation (not the
standard deviation, just the difference plus or minus) from the average.
PROGRAM 3 - 1 1 Calculate Average of Four Numbers

1 /* Calculate the average of four integers and print
2 the numbers and their deviation from the average.
3 Written by:
4 Date:
5 */
6 tinclude <stdio.h>
7 int main ( void )
8 {
10 int numl ;
11 int num2;
12 int num3 ;
13 int num4;
14 int sum ;
15 float average ;
16
17 // Statements
18 printf( " \ nEnter the first number : " );
19 scanf("%d", & numl );
20 printf( " Enter the second number : " );
21 scanf( " %d " , & num2);
22 printf( " Enter the third number : " );
23 scanf("%d" , & num3 );
24 printf( " Enter the fourth number : " );
25 scanf( " %d " , & num4 );
26
27 sum = numl + num2 + num3 + num4;
28 average = sum / 4.0 ;
29
30
31
32
printf( "\n * * **** average is %6.2f **** **** " ,
average );
printf( "\n " );
I
continued
of Four Numbers (continued)

PROGRAM 3 - 11 Calculate Average
33
34
: %6d
printf("\nfirst numberaverage);
numl, numl -
—
deviation: %8.2f " ,
35
36 printf nsecond number
( " \ : %6d --
deviation: % 8.2f " ,
-
num2, num2 average);
37
38
39
printf("\nthird number:
num3, num3
%6d
-
average);
—
deviation: %8.2f" ,
40 printf( "\nfourth number:

%6d --
deviation: %8.2f " ,
41 -
num4, num4 average);
42
43 return 0;
44 } // main
Results:
Enter the first number: 23
Enter the second number: 12
Enter the third number: 45
Enter the fourth number: 23
*** * * * * * average is 25.75 * * * * * * * *
first number:
second number:
23
12
—— deviation:
deviation:
-2.75
-13.75
third number:
fourth number:
45
23
—— deviation:
deviation:
19.25
-2.75
Proaram 3 * 11 Analysis Program 3-11 is a little more complex than the previous ones. At the beginning o
^ main are several variable declarations, five integers, and a floating- point number
The first four are for the variables read from the keyboard, the fifth is for the sum, and
the floating-point number is for the average.
The statements section starts by reading the data. Each read is preceded by a dis-
play so the user will know what to do. The specific instructions about what to input are
known as user prompts You should always tell the user what input is expected from the
keyboard. After the user has keyed the data, the program continues by adding the
numbers, placing the total in sum, and computing average. It then
displays the results.
Notice that the program displays the results in a format
verify that the program ran correctly. The program not
that allows the user to easily
repeats each input with its deviation from the
only prints the average but also
average. After completing its work, the
program concludes by returning to the
operating system.
Look at the results carefully. Note how each
they can be easily read. Taking the time to series of numbers is aligned so that
guishes a good programmer from an align output is one of the things that distin-
how your program presents its results toaverage programmer. Always pay attention to
pays off in the long run.
the user. Paying attention to these little details
EXAMPLE 3 - 7 Degrees to Radians

One way to measure an angle in a circle is in degrees. For example , the angle
formed by a clock at 3 o’clock is 90 . Another way to measure the angle is in
radians. One radian is equal to 57.2957793 degrees. In Program 3- 12 we ask
the user to input an angle in radians , and we convert it to degrees.
PROGRAM 3 - 1 2 Convert Radians to Degrees

1 /* This program prompts the user to enter an angle
2 measured in radians and converts it into degrees.
3 Written by:
4 Date:
5 */
7
8 # define DEGREE FACTOR 57.295779
9
11 {
13 double radians;
14 double degrees;
15
16 // Statements
17 printf( " Enter the angle in radians: " );
18 scanf( " %lf " , &radians);
19
20 degrees = _
radians * DEGREE FACTOR ;
21
22 printf( " %6.3f radians is %6.3f degrees\n " ,
23 radians , degrees);
24 return 0;
25 } // main
Results:
Enter the angle in radians: 1.57080
1.571 radians is 90.000 degrees
Program 3 - 1 2 Analysis In this short program we introduce the defined constant. Defined constants are an excel -
lent way to document factors in a program. We could have just as easily used a mem-
ory constant. Factors are usually placed at the beginning of the program where they
can easily be found and changed as necessary. Remember that the area before main is
global and that it should be used only for declarations and constants — not variables.
3. More precisely, a radian is ( 180 / n ) .

EXAMPLE 3 - 8 Calculate Sales Total unit price , quantity, discount rate,
a sale given the
Program 3-13 calculates
and sales tax rate.
PROGRAM 3 - 13 Calculate Sales Total the unit price,

1 /* Calculates
the total sale given
tax rate.
2 quantity, discount, and
3 Written by:
4 Date:
5 */
7
_
8 tdefine TAX RATE 8.50
9
10 int main (void )
11 {
13 int quantity;
14
15 float discountRate;
16 float discountAm;
17 float unitPrice;
18 float subTotal;
19 float subTaxable;
20 float taxAm;
21 float total ;
22
23 // Statements
24 printf( "\nEnter number of items sold:
25 scanf("%d", quantity);
26 ^
27 printf("Enter the unit price: ")?
28 scanf("%f " , &unitPrice);
29
30 printf("Enter the discount rate ( per cent ): " );
31 scanf("%f ", discountRate);
32 ^
33 subTotal = quantity * unitPrice;
34 discountAm = subTotal * discountRate / 100.0;
35 subTaxable = subTotal discountAm ;
36 taxAm = subTaxable * TAX RATE /
-_
100.00 ;
37 total = subTaxable + taxAm;
38
39 printf(" \nQuantity
sold: %6d\n" , quantity );
continued
PROGRAM 3 - 1 3 Calculate Sales Total (continued )

40 printf( " Unit Price of items: %9.2f \n " , unitPrice );
41 printf( " \n" ) ;
42
43 printf( "Subtotal : %9.2f \n" , subTotal);
44 printf( " Discount: -%9.2f\n" discountAm );
/
45 printf( "Discounted total: %9.2f \n " subTaxable );

/
46 printf( "Sales tax: +%9.2f \n " taxAm );

/
47 printf( "Total sale: %9.2f \n " , total);

48
49 return 0;
50 > // main
Results:
Enter number of items sold: 34
Enter the unit price: 12.89
Enter the discount rate ( per cent ): 7
Quantity sold: 34
Unit Price of items: 12.89
Subtotal : 438.26
Discount: 30.68
Discounted total: 407.58
Sales tax: + 34.64
Total sale: 442.23
Program 3 - 1 3 Analysis Look at the results of this program carefully. Do you see any problems? Just because a
program runs doesn't mean that it is running correctly. In this case, the total is incor -
rect ( 407.58 + 34.64 is not equal to 442.23 ). The problem is created by the floating-
point arithmetic and rounding errors. If we wanted absolute accuracy, we would have
to do the arithmetic in integer ( cents ) and then divide by 100 to print the report .
EXAMPLE 3 - 9 Calculate Student Score

Program 3- 1 4 calculates the average score for a student. The class has ( our
quizzes ( 30% ) , two midterms ( 40% ), and a final ( 30%) . The maximum score
for all quizzes and exams is 100 points.
PROGRAM 3 - 14 Calculate Student Score

1 /* Calculate a student's average score for a course
2 with 4 quizzes , 2 midterms, and a final. The quizzes
3 are weighted 30% , the midterms 40% , & the final 30%.
4 Written by:
continued
Calculate Student Score (continued

)
PROGRAM 3 - 14
5 Date:
6 */
7 #include <stdio
.h>
8 __
9 tdefine QUIZ WEIGHT
30
40
10 #define MIDTERM WEIGHT
__ _
11 #define FINAL WEIGHT
12 #define QUIZ MAX
30
400.00
200.00
13 #define MIDTERM MAX
14 # define FINAL_ MAX 100.00
15
16 int main (void)
17 {
19 int quizl ;
20 int quiz2;
21 int quiz3;
22 int quiz4;
23 int totalQuiz;
24 int midterml ;
25 int midterm2;
26 int totalMidterm;
27 int final;
28
29 float quizPercent;
30 float midtermPercent ;
31 float finalPercent;
32 float totalPercent ;
33
34 / / Statements
35 =
printf(" == =
=== QUIZZES = = =\n" );
36 printf("Enter the score for the first quiz: ")?
37 scanf("%d", &quizl);
38 printf("Enter the score for the second quiz: ");
39 scanf("%d" , &quiz2);
40 printf("Enter the score for the third quiz:
41 scanf("%d " , &quiz3);
42 printf("Enter the rscore for the
fourth quiz: " );
43 scanf("%d" , &quiz4);
44
45 printf(" ====== == **=
=
46
MIDTERM ==== ====== \ n" ) ;
printf("Enter the
score for the first midterm: " )/
47 scanf("%d", ,5< midterml);
'
48 printf("Enter the
score for the second midterm: " );
continued
PROGRAM 3- 1 4 Calculate Student Score (continued )

49 scanf( " %d ", &midterm2);
50
51 printf( "===== === FINAL === ======\n" );
52 printf( "Enter the score for the final: " );
53 scanf( "%d " , &final);
54 printf( "\n" );
55
56 totalQuiz = quizl + quiz2 + quiz3 + quiz4 ;
57 totalMidterm = midterml + midterm2;
58
59 quizPercent =
60 _
(float )totalQuiz * QUIZ WEIGHT / QUIZ MAX ; _
61 midtermPercent =
62 _
( float )totalMidterm * MIDTERM WEIGHT / MIDTERM _MAX ;
63 finalPercent =
64 (float )final * FINAL WEIGHT / FINAL MAX ; _
65
66 totalPercent
67 quizPercent + midtermPercent + finalPercent ;
68
69 printf( "First Quiz %4d \n" , quizl );
70 printf( "Second Quiz %4d \n " , quiz2);
71 printf( "Third Quiz %4d \n" , quiz3 );
72 printf( "Fourth Quiz %4d \n " , quiz4 );
73 printf( "Quiz Total %4d \n\n " , totalQuiz);
74
75 printf( "First Midterm %4d \n" , midterml );
76 printf( "Second Midterm %4d \n " , midterm2);
77 printf( "Total Midterms %4d \n\n " , totalMidterm );
78
79 printf( "Final %4d \n\n " , final );
80
81 printf( "Quiz %6.1f %% \n" , quizPercent );
82 printf( " Midterm %6.1f % % \n " , midtermPercent );
83 printf( "Final %6.1f %%\n" , finalPercent );
84 printf( " \n");
85 printf( "Total %6.1f %%\n " , totalPercent );
86
87 return 0;
88 } // main
Results:
Enter the score for the first quiz: 98
continued
Student Score ( continued )

PROGRAM 3- 14 Calculate quiz: 89
for the second
Enter the score quiz: 78
for the third
Enter the score quiz. 79
for the fourth
Enter the score
= MIDTERM midterm: 90
score for the first
Enter the : 100
the second midterm
Enter the score for
========= FINAL = final: 92
:
the
Enter the score for
First Quiz 98
Second Quiz 89
Third Quiz 78
Fourth Quiz 79
Quiz Total 344
First Midterm 90
Second Midterm 100
Total Midterms 190
Final 92
Quiz 25.8%
Midterm 38.0%
Final 27.6%
Total 91.4%
First, note how the pro-

Program 3 - 1 4 Analysis This rather long program contains several points to consider. val-
gram starts with a series of defined constants. Putting the definitions of constant
ues at the beginning of the program does two things: ( 1 ) It gives them names that we
can use in the program, and ( 2) it makes them easy to change .
Now study the statements. Notice how they are grouped? By putting a blank line
between a group of related statements, we separate them, much as we would separate
paragraphs in a report. This makes it easy for the user to follow the program.
Finally, study the input and output. Notice that the user was prompted for all input
with clear instructions. We even divided the input with headings. The output is also
divided, making it easy to read. It would be even easier to read if we aligned all the
amounts, but the techniques for doing so will not be introduced until Chapter 7.

In this section we discuss three concepts that , although technically not
engineering principles , are important to writing clear and understandable
programs.
KISS
Keep It Simple and Short ( KISS 4 ) is an old programming principle. Unfortu -
nately, many programmers tend to forget it , especially the simple part . They
seem to feel that just because they are working on a complex problem , the
solution must be complex, too. That is simply not true. Good programmers
solve the problem in the simplest possible way; they do not contribute to a
complex situation by writing obscure and complex code .
A trivial example will make the point . If you were writing a program that
reads floating- point numbers from the keyboard , you would not program it so
that the user had to enter the integral portion of the number first and then
the fractional part . Although this would work , it is unnecessarily complex,
even though it might be a fun way to solve the problem .
Unfortunately, C provides many operators and expression rules that make
it easy for a programmer to write obscure and difficult to follow code . Your
job as a programmer is to make sure that your code is always easy to read .
Your code should be unambiguous: It should not he written so that it is easy
to misread it .
Another old structured programming principle is that a function should
not be larger than one page of code. Now, for online programming in a work -
station environment , a function should be no longer than one screen about
20 lines of code . By breaking a problem down into small , easily understood
—
parts, we simplify it . Then we reassemble the simple components into a sim -
ple solution to a complex problem .
3 \ ocke of code should be no longer than one screen.
One element of the C language that tends to complicate programs , espe -

cially for new programmers, is side effects. We explained in Section 3.4 ,
“ Evaluating Expressions , ” that side effects can lead to confusing and different
results depending on the code. You need to fully understand the effects when
you write C code . If you are unsure of the effects, then simplify your logic
until you are sure.
Parentheses
One programming technique to simplify code is to use parentheses, even
when unnecessary'. While this may lead to a few extra keystrokes, it can save
hours of debugging time created by a misunderstanding ol the precedence
4. KISS originally had a different , rather insulting , meaning . We prefer this interpretation .
and associativity rules. Whenever a
statement contains multiple expressions
the compiler will interpret it as you intended .
use parentheses to ensure that
do, not what you Intended to tell them

Computers do what you tell them to
to do. Make sure your code Is as clear
and simple as possible.
User Communication
You should always make sure you communicate with your user from the very
first statement in your program to the very last . As mentioned previously, we
recommend that you start your program with a message that identifies the
program and end with a display that says the program is done.
When you give your user instructions , make sure that they are clear and
understandable. In Program 3- 13, we used three statements to give the user
complete and detailed instructions on what we wanted entered . We could
have simply said
"Enter data"
but that would have been vague and subject to interpretation . How would the
user know what specific data were required? For each input in Program 3- 13,
we told the users exactly what data we needed in terms that they understand .
If you don’t tell users exactly what data to input , they may do anything they
feel like, which is usually not what you wanted or expected .
One common mistake made by new programmers is to forget to tell the
user anything. What do you think would be the user ’s
response to
-
Program 3 15 when confronted with a blank screen and a
computer that is
doing nothing?
PROGRAM 3 - 15 Program That Will Confuse the User
1 linclude <stdio.h>
2 int main (void)
3 {
4 int i;
5 int j;
6 int sum ;
7
8 scant("%d%d", Si , S j ) ;
9 sum =
i + j;
10 printf("The SUm
of %d & is %d\n" , i
11 return 0; i , j, sum);
12 > // main
We now rewTite
i the program
addition of one print with clear uuser
communication. With the
statement, the user knows exactly
what is to be done.
PROGRAM 3 - 16 Program That Will Not Confuse the User

2 int main (void )
3 {
4 int i;
5 int j;
6 int sum;
7
8 printf( " Enter two integers and key <return>\n " );
9 scant( " %d %d " , &i, &j);
10 sum = i + j;
11 printf( "The sum of %d & %d is %d\n" / i, j, sum );
12 return 0;
13 > // main
Results:
Enter two integers and key <return>
4 5
The sum of 4 & 5 is 9
We will return to these three concepts from time to time when we intro-
duce new structures that tend to be confusing or misunderstood.
3.9 Tips and Common Errors
. They are one of the main sources of
1. Be aware of expression side effects
confusion and logical errors in a program .
2 Use decrement/increment operators wisely
. . Understand the difference
between postfix and prefix decrement /increment operators before
using them .
.
3 Add parentheses whenever you feel they will help to make an expression
clearer.
4 It is a compile error to use a variable that has not been defined .
.
5. It is a compile error to forget the semicolon at the end of an expression
statement .
6. It is a compile error to code a variable declaration or definition once you
have started the statement section. To help yourself remember this rule,
use comments to separate these two sections within a function .
.
7 It is most likely a compile error to terminate a defined constant (#define)
with a semicolon . This is an especially difficult error to decipher because
—
you will not see it in your code you see the code you wrote, not the code
that the preprocessor substituted .
8. It is a compile error when the operand on the left of the assignment oper
ator is not a variable. For example, a + 3 is not a variable and cannot
-
receive the value of b * c.
(a + 3) = b * c;
.
9 It is a compile error to use the
increment or decrement operators with
any expression other than a variable identifier.
For example, the following
code is an error:
(a + 3 ) ++
10. It is a compile error to use
the modulus operator ( % ) with anything other
than integers.
11. It is a logic error to use i
a variable before it has
12. It is a logic error to
been assigned a value.
modifv> a variable li n an expression when
'
appears more than once. the variable
3.10 Key Terms

additive expression
assignment expression left - to- right associativity
associativity multiplicative expression
name
binary expression operand

block operator
cast postfix expression
complex expression precedence
compound statement primary expression
conversion rank promotion
demotion right - to-left associativity
explicit type conversion side effect
expression simple expression
expression statement unary expression
implicit type conversion user prompts
KISS
3.11 Summary
An expression is a sequence of operators and operands that reduces to a
single value .
An operator is a language-specific token that requires an action to
be taken .
An operand is the recipient of the action .
C has six kinds ol expressions: primary, postfix, prefix, unary, binary, and
ternary.
The most elementary type of expression is a primary expression . A primary
expression is an expression made up of only one operand . It can he a
name, a constant , or a parenthesized expression .
A postfix expression is an expression made up of an operand followed by an
operator. You studied function call and postfix increment /decrement
expressions in this chapter.
A unary expression is an expression made up of an operator followed by an
operand . You studied six in this chapter: prefix increment /decrement ,
sizeofy plus/minus , and cast expressions.
A binary expression is an expression made up of two operands with an
operator between them . You studied multiplicative , additive, and assign -
ment expressions in this chapter.
Precedence is a concept that determines the order in which different oper-
complex expression act on their operands.
ators in a
Associativity defines the order of evaluation when operators have the same
precedence.
The side effect of an expression is one ol the unique phenomena in C. An
expression can have a side effect in addition to a value.
the rules of precedence and

To evaluate an expression , we must follow
associativity.
A statement causes an action to be performed by
the
program .
Although C has eleven different types of statements, you studied only four-
types in this chapter:
.
1 A null statement is just a semicolon .
2. An expression statement is an expression converted to a statement bv
keeping the side effect and discarding the value.
3. A return statement terminates a function.
4. A compound statement is a combination of statements enclosed in two
braces.
KISS means " Keep It Simple and Short . "
One of the important recommendations in software engineering is the use
of parentheses when they can help clarify your code .
J Another recommendation in software engineering is to communicate
clearly with the user.
3.12 Practice Sets

Review Questions
I . A unary expression consists of only one
operand w ith no operator.
a. True
.
b False
2. The left operand in an assignment
expression must be a single variable.
.
a True
b. False
3. Associativity is used to determine which of
is evaluated first. several different expressions
a. True
b. False
4. Side effect i
action that results from
the evaluation of an expression .
.
a True
b. False
5. An expression
statement is terminated with a
.
a True period .
.
b False
.
6 A(n) is a sequence of operands and operators that
reduces to a single value.
.
a expression
b. category
c. formula
.
d function
.
e value
.
7 Which of the following is a unary expression?
.
a i+ j
.
b +a
c. ++ C
d. scanf ( ... )
e. x * 5 =
8 . The expression evaluates the operand on the right

side of the operator and places its value in the variable on the left side of
the operator.
a. additive
b. assignment
c . multiplicative
d. postfix
e. primary
9 . is used to determine the order in which different
operators in a complex expression are evaluated.
a. associativity
b. evaluation
.
c category
d. precedence
e. side effect
10. is an action that results from the evaluation of an
expression .
.
a associativity
.
b evaluation
.
c category
d. precedence
e. side effect
.
1 1 Which of the following statements about mixed expressions is false ?
a. A cast cannot be used to change an assigned value.
b. An explicit cast can be used to change the expression type.
c. An explicit cast on a variable changes its type in memory.
.
d An implicit cast is generated by the compiler automatically when
necessary.
.
e Constant casting is done by the compiler automatically.
stateme nts about compound false?

statements is
12. Which of the following
_ known as a block .
is also
a. A compound statement is
b. A compound statement is enc
losed in a set of braces.
be terminated by a semicolon .
c. A com pound statement must
section in a compound statement is
d . The declaration and definition
optional.
statement is optional.
e. The statement section in a compound
Exercises
13. Which of the following expressions are not postfix expressions ?
a . x ++
b. x
—
c. scanf ( ... )
.
d x * y
e. + + x
.
14 Which of the following are not unary expressions?
.
a ++x
b. x
. —
c sizeof
d. +5
(x)
e. x = 4
15. Which of the following is not a binary expression ?
a. 3 * 5
b. x + = 6
c. y = 5 + 2
d. z - 2
e. y % z
16. Which ol the following is not a valid
assignment expression ?
a. x = 23
b. 4 = x
c. y % = 5
d. x = 8 = 3
e. x = r = 5
17. It originally x - 4, what
is the value of x after the
lowing expression? evaluation of the fol-
a. x = 2
b. x += 4
.
C X += x + 3
d. x * = 2
e. x / = x + 2
18. If originally x = 3 and y = 5 , what is the value of x and y after each of
the following expressions ?
a. x + + + y
.
b ++ x
c. x ++ + y ++
.
d ++ x + 2
.
e x
— y —
19. What is the value of each of the following expressions?
a. 2 4 6 * 2
b. -15 * 2 + 3
c. 72 / 5
d. 72 % 5
e. 5 * 2 / 6 + 1 5 % 4
20. What is the value ol each of the following expressions?
.
a 6.2 + 5.1 * 3.2
b. 2 . 0 + 3 . 0 / 1 . 2
c. 4 . 0 * ( 3 . 0 + 2 . 0 / 6 . 0 )
d. 6 . 0 / ( 2 . 0 + 4 . 0 * 1 . 2 )
e. 2 . 7 + 3 . 2 5.3 * 1.1
21 . Given the following definitions, which of the following statements are
valid assignments?
# d e f i n e NUM 10 10
int x;
i n t y = 15 ;
.
a x = 5;
.
b y = 5;
.
C X Y = 50 ;
.
d x = 50 = y ;
.
e x = x + 1;
.
f y = 1 + NUM 10 ;
-
g 5 = y;
22. II originally x = 2 , y = 3 , and z = 2 , what is the value of each of the
following expressions?
a . x ++ + y++
b. + + x - -- z
c.
d. x
—— —
x + y++
+ x - y —
e. x +
— =y
23. If originally x
- X + X++ y
2 , y = 3 , and z = 1, what is the value of each of the
.
a x + 2 / 6 + y
b. y - 3 * z + 2
c. z (x + z) % 2 + 4
d. x - 2 * ( 3 + z ) + y
—
e. y++ + z + x++
2 4. I f x = 2 9 4 5 , what is the
value of each of the following expressions?
a. x % 10
b. x / 10
c.(x / 10) % 10
d . x / 100
e.(x / 100) % 10
25. What is the output from the following
code fragment ?
int a;
int b;
a = b = 50;
printf ("%4d %4d", a, b);
a = a * 2;
b = b / 2;
printf ("%4d %4d", a, b);
Problems
26. Given the following pseudocode , write a program that executes it . Use
floating- point types for all values.
Algorithm Problem26
1 read x
2 read y
3 compute p = x * y
4 compute s = x + y
5 total = s2 + p * (s
6 print total
- x) * (p + y )
End Problem26
27. Write a program that reads two integers

from the keyboard , multiplies
them , and then prints the two numbers and their product .
28. Write a program that extracts and
prints the rightmost digit of the inte-
.
gral portion of a float
29'
.
rSa7pZoanmofhaSrCtS^^^
30 Write a program that
SeC
°nd ri 8htmost di &il of the
.
from a user-ciirsrti .length and width.* Perimeter of a rectangle
calculates the area ant
suppliedi (/ scanfn)
.
31 We are all familiar
with the fart tW anges
i
minutes, and seconds Annthr " 0i afe measured in dc8rees’
3n8 C S 3 rad an ' A radian
is the angle formed by
two radii
‘ ' ^ '°
^^
radiUS
of their circle. One
‘ ^ * Vs
radian cquals / * 5779Q degrees l
. Write a program
M
that converts degrees into radians. Provide good user prompts. Include
the following test data in your run :
90° is 1.57080 radians
32 . I he formula lor converting centigrade temperatures to Fahrenheit is:
F = 32 + C x 180.0
100.0
Write a program that asks the user to enter a temperature reading in

centigrade and then prints the equivalent Fahrenheit value. Be sure to
include at least one negative centigrade number in your test cases.
33. Write a program that changes a temperature reading from Fahrenheit to
Celsius using the following formula:
Celsius = ( 100.0 / 180.0) * (Fahrenheit - 32 )
Your program should prompt the user to enter a Fahrenheit tempera -

ture . It then calculates the equivalent Celsius temperature and displays
the results as shown below .
Enter the temperature in Fahrenheit: 98.6
Fahrenheit temperature is: 98.6
Celsius temperature is: 37.0
34 . \ \ rite the C code lor each of the following formulas. Assume that all vari -
ables are defined as double .
a.
2
HIV
K •i n E n
w r r<
= -y-
b.
b+c
res = 2be
35. Write the C code to calculate and print the next two numbers in each of
the following series. You may use only one variable in each problem .
a . 0, 5 , 10, 15 , 20 , 25 , ? , ?
b. 0, 2, 4, 6, 8, 10, ?, ?
.
c 1 , 2 , 4 , 8, 1 6 , 3 2 , ? , ?
—
Projects
36. Write a program that converts
inches into
and prin s a user-supplied
' re ..,, ,
a. foot ( 12 inches )
b. yard ( 36 inches )
c. centimeter ( 2.54/inch
)
d . meter (39.37 inches )

a set in which each number is the
37’ A Fibonacci number is a member of
sum of the previous two numbers . (The Fibonacci senes describes a form
of a spiral . ) The series begins
0 , 1 , 1 , 2 , 3 , 5 , 8 , 1 3 , 2 1 , ...
Write a program that calculates and prints the next three numbers in
the Fibonacci series. You are to use only three variables, fibl , fib2 .
and fib 3.
38. Write a program that prompts a user for an integer value in the range 0 to
32,767 and then prints the individual digits of the numbers on a line
with three spaces between the digits. The first line is to start with the
leftmost digit and print all five digits; the second line is to start with the
second digit from the left and print four digits, and so forth . For example,
if the user enters 1234 , your program should print
0 1 2 3 4
1 2 3 4
2 3 4
3 4
4
39. Write a program to create a customers

bill lor a company. The company
_
sells only five different products: TV, VCR ,
Remote Controller, CD
TaPc Recorder The unit prices are $400.00, $ 220 , $ 35.20,
-
.00, and $150.00, respectively. The program
eac P,ece oi equipment purchased
must read the quantity
from the keyboard . It then calcu -
lax ^^ ^
sales ^ °
subtotal» and the total cost after an 8.25%
*^
iris:i su?., tr
SCt
‘ntcgers
U
representing the quantities
^
of each if
w ;
“ p,om* ,*
|> ,h
in
answers. iheZ2rin“ boldface show

' the user s '
Chapter 3 Structure of a C Program 14 7
How Many TVs Were Sold? 3

How Many VCRs Were Sold ? 5
How Many Remote Controllers Were Sold? 1
How Many CDs Were Sold? 2
How Many Tape Recorders Were Sold? 4
The format for the output is shown in Figure 3- 13.
QTY DESCRIPTION UNIT PRICE TOTAL PRICE
XX TV 400.00 XXXX XX .
XX VCR 220.00 XXXX XX .
XX REMOTE CTRLR 35.20 X X X X .X X
XX CD PLAYER 300.00 X X X X. X X
XX TAPE RECORDER 150.00 XXXX.XX
SUBTOTAL XXXXX XX .
TAX XXXX XX .
TOTAL XXXXX XX .
FIGURE 3 - 1 3 Output Format for Project 39
Use either defined constants or memory constants for the unit prices
and the tax rate . Use integer variables to store the quantities for each
item . Use floating- point variables to store the total price for each item ,
the bill subtotal, the tax amount , and the total amount of the bill . Run
your program twice with the following data:
SET 1 -> 2 1 4 1 2
SET 2 3 0 2 021
« •
Functions
In Chapter 2, we introduced data types. In the preceding chapters, we used
only the three non - derived types, void , integral, and floating-point . Although
these types are very usef ul, they can solve only a limited number of problems.
There are six derived types in C, as shown in Figure 4 - 1 .
In this chapter we introduce the first derived type, the function type. The
function type is derived from its return type. The return type can be any type
except an array, which we introduce in Chapter 8, or a function type .
Derived
Types
Function
Type i Array
Type
Chapter 8
i Pointer
Type
Chapter 9
i Structure
Type
Chapter 12
Union
Type
Chapter 12
Enumerated
Type
Chapter 12
FIGURE 4- 1 Derived Types
Objectives
To design and implement programs with more than one function
To be able to design multi - function programs using structure charts
To understand the purpose of the function declaration, call, and definition
To understand the four basic function designs
To understand how two functions communicate through parameters
To understand the differences between global and local scope
To understand the software engineering principles of functional cohesion
and top-down development
149
150 Section 4.1 Designing Structured Programs
4.1 Designing Structured Programs

far have been very simple. They solved
The programs
problems that cTuldTeunderswodwithout too
, you will
much effort. As we consider
discover that it is not possible to
larger and larger programs, however without somehow reducing them to
understand all aspects of such programs
more elementary parts.
. .. ..
parts is a common practice. For
Breaking a complex problem into smaller
this year you dec de to dnve n a cir-
example, suppose that for your vacation onal parks as poss ble in
visit as many nat
cular route that will allow you to are very s mple: Visit as many
two weeks Your requirements for this problem
will you do it ? You might first gather
parks as possible in two weeks. But how
some data about national parks and then
calculate the distance between each
. Next , you would estimate how much
of them to figure out the travel time
time it would take to visit each park . Finally , you would put all your data
together to plan your itinerary . Then you would make your motel and camp
reservations and any other arrangements that had to be in place in advance.
The planning for large programs is similar f irst , you must understand
,
the problem as a whole; then you must break it into simpler, understandable
parts. We call each of these parts of a program a module and the process of
subdividing a problem into manageable parts top-down design .
The principles of top-down design and structured programming dictate
that a program should be divided into a main module and its related modules.
Each module should also be divided into submodules according to software
engineering principles that we discuss in Section 4.8 , “ Software Engineer-
ing. The division of modules proceeds until the module consists only of ele-
mentary' processes that are intrinsically understood and cannot be further
subdivided . This process is known as factoring .
In top-down design, a program is divided into a main module and its related
.
modules Each module is in turn divided into submodules until the resulting
modules are intrinsic; that is, until they are
implicitly understood without
further division .
^rf "° ”deSign USUallyc* donc usinstructure

iS
Wn
m 8 a visual representation of the
between T \ StrUfUre ^ - Th ruleschart shows the relation
lart ’ e
,
structured "^ ° °“ ;ed Software Engineering but andthis point
Submodules S
for «
reading creating
we only nee
left-right.
/aV f S
ReferrinoTp 6
Ve ln " ," at
FlrSt the structure charl is read top-down,
,
’
^ ^^ ^ Mai Motlule' 1,1 this caSC'
Main Module represents our "
entire set of code to solve the problem .
Main
Module
Module 1 Module 2 Module 3
Module 1a Module 1b Module 1c Module 2a Module 3a Module 3b
FIGURE 4 - 2 Structure Chart
Moving down and left, we then read Module 1 . On the same level with
Module 1 are Module 2 and Module 3. The main module consists of three
submodules. At this point, however, we are dealing only with Module 1. We
now note that Module 1 is further subdivided into three modules, Module la,
Module lb, and Module lc. To write the code for Module 1 , therefore, we
will need to write code lor its three submodules. What does this concept say
about writing the code for the Main Module?
Now lor some more terminology. The Main Module is known as a calling
module because it has submodules. Each of the submodules is known as a
called module. But, because Modules 1, 2, and 3 also have submodules, they
are also calling modules; they are both called and calling modules.
Communication between modules in a structure chart is allowed only
through a calling module. If Module 1 needs to send data to Module 2, the
data must be passed through the calling module, Main Module. No commu -
nication can take place directly between modules that do not have a calling-
called relationship.
With this understanding, how can Module la send data to Module 3b? It
first sends the data to Module 1, which in turn sends it to the Main Module,
which passes it to Module 3, and then on to Module 3b. Although this
sounds complex, it is easily done.
The technique used to pass data to a function is known as parameter
passing. The parameters are contained in a list that is a definition of the data
passed to the function by the caller. The list serves as the formal declaration
of the data types and names.
4.2 Functions in C
.
In C, the idea of top-down design is done using functions A C program is
made of one or more functions, one and only one of which must he named
main . The execution of the program always starts and ends with main , but it
152 Section 4.2 Functions in C
to do special tasks. Figure 4-3 shows a C program

can call other functions
structure chart .
of one or more functions, one

and only one of which
In C, a program is made always starts with main,
must be ca Zd main. The execution of the program job.
to do some part of the
but it can 1call other functions
) is an independent module that will be

A function in C (including main
function receives control from a calling
called to do a specific task. A called
function . When the called function
completes its task , it returns control to
may not return a value to the caller. The function
the calling function . It may or
operating system ; main in turn calls other functions.
main is called bv the
system .
When main is complete , control returns to the operating
^ ^j
nair
Functions called by main
B
Function| Function Function
2 3
FunctionI FunctionI Function Function Function! I Function I

a b c d e ll f
Functions called by Function 1 Function called Functions called
by Function 2 by Function 3
FIGURE 4- 3 Structure Chart for a C Program
In general , the purpose of a function is to receive zero or more pieces of

data , operate on them , and return at most one piece of data. 1 At the same
time a function can have a side effect . A
function side effect is an action that
results in a change in the state of the
program . If a side effect occurs, it
occurs while the function is executing and before the function returns. The
0 tdn n
I
f
* ' '° 'e
fcceP ‘ ^
* n8 ata from outside the program , sending data
inn the Call , n8 f
“- ^
monitor or a e > or changing the value of a variable
" The function concept is shown in Figure 4- 4 .
-
1 In Chapter 12 we
discuss how to turn multiple data values
using a structure.
Zero or more data

values can be passed.
H 1
<b <b> ••• <b b
t t t t
Side Effects
J -
CD
t
At most one data value
or structure can be returned.
FIGURE 4 - 4 Function Concept
Several advantages are associated with the use of functions in C or any

.
other language The major advantages are:
.
1 As already described, problems can be factored into understandable and
manageable steps.
.
2 Functions provide a way to reuse code that is required in more than one
place in a program. Assume, for instance, that a program requires that we
compute the average ol a scries of numbers in five different parts of the
program. Each time the data arc different . We could write the code to
compute the average five times, but this would take a lot of effort. Also, if
we needed to change the calculation, we would have to find all five places
.
that use it to change each of them It is much easier to write the code
once as a I unction and then call it whenever we need to compute the
average .
A function in C can have a return value, a side effect, or both The side .
.
effect occurs before the value is returned The function’s value is the
.
value in the expression of the return statement A function can be called
for its value, its side effect, or both .
.
3 This advantage to using functions is closely tied to reusing code. Like
many languages, C comes with a rich and valuable library. For example,
there is a math library, math.hy that contains almost any mathematical or
statistical function that we will ever need. T hese C libraries provide stan-
dard functions that make our work as a programmer much easier. We can
also create personal and project libraries that make developing systems
easier. Appendix F, “Function Libraries,” documents many of the func -
tions included with the C language.
154 Section 4.2 Functions in C
„.„ rather complex idea that cen -

4. We use functions to pro t data. This is a
Local data consist of data
ters around the conceP , ^^
jata
described in a function.
on1» »
'"
the data are
«
" “ /SS .
are available only to the function and
when the function is not running,
not accessible. Uata in o function, the. cannot be seen or
changed by a function outside of its scope.
. „4 cpveral functions, such as scan/ and printf , in our
-
J
We have already *veral fun
i
program . »'
jil^ ^ ^ by
>B rating our own functions. ISeW ^ c
rSTJaSl -
let * M at Program 4 1, which demo,,
strates how we write and call a function .
PROGRAM 4- 1 Sample Program with Subfunction

1 / * This program demonstrates function calls by
calling
2 a small function to multiply two numbers .
3 Written by :
4 Date :
5 */
6 iinclude < stdio. h >
7
8 / / Function Declarations
9 int multiply ( int numl , int num 2 ) ;
10
12 {
13 / / Local Declarations
14 int multiplier ;
15 int multiplicand ;
16 int product ;
17
18 // Statements
19 printf ( " Enter
two integers: " ) ;
20 scanf ( "%d%d " ,
21 ^ multiplier , Multiplicand ) ;
22 product = multiply ( multiplier ,
23 multiplicand ) ;
24
25
printf ( "
Product of %
d & % d is_
i %d \ n ,
26
multiplier, multiplicand "
return 0 ; , product ) ;
27 F / / main
28
29 /* ===.
30 === == === multiply
Multipies two ====
31 numbers and
Pre numl & returns product .
« 32 num2
are values to be
Post Product multiplied
returned
continued
PROGRAM 4 - 1 Sample Program with Subfunction ( continued )

33 */
34 int multiply (int numl , int num2)
35 {
36 // Statements
37 return ( numl * num2);
38 > // multiply
Results:
Enter two integers: 17 21
Product of 17 & 21 is 357
4.3 User- Defined Functions

Like every other object in C, functions must be both declared and defined.
I he Function declaration, which needs to be done before the function call, gives
the whole picture ol the Function that needs to be defined later. I he declaration
mentions the name ol the function, the return type, and the type and order of
formal parameters. In other words, the declaration uses only the header of the
function definition ended in a semicolon. The function definition, which is
traditionally coded alter the function that makes the call, contains the code
needed to complete the task .
A function name is used three times: for declaration, in a call, and for definition .
Figure 4 - 5 shows the interrelationships among these function compo-
nents. As you study Figure 4- 5, note that the function name is used three
times: when the function is declared, when it is called, and when it is
defined.
// Function Declaratiol void greeting ( void )

void greeting ( void );
int main ( void ) printf( " Hello World!" );
{ return ;
// Statements
greeting( );
\
/ / call
,
} // greeting
return 0 ; i
} // main
* pHello World
Side Effect
FIGURE 4- 5 Declaring, Calling, and Defining Functions

156 Section 4.3 User-Defined Functions
Basic Function Designs

. .
designs by their return values and their para
*,
,
We classify the basic function
value or they don t. Funct ons that don'
return a
eter lists Functions either
void functions. Some funct ons have parameters
return a value are known as
and parameter lists results in four
and some don’t. Combining return types
basic designs: void functions with no parameters
eters functions that return a
value but
that return a value and have parameters .
have
We
, void functions with param-
no parameters, and funct ons
discuss these four designs in the
.
sections that follow.
void Functions without Parameters

A void function can be written with no parameters. The greeting I unction
in Figure 4- 5 receives nothing and returns nothing. It has only a side effect,
displaying the message, and is called only lor that side effect.
The call still requires parentheses, however, even when no parameters
are present. When we make a call to a function with no parameters, it is
tempting to leave the parentheses off the call. Although this is valid syn-
tax, it is not what we intended. Without the parentheses, it is not a func -
tion call.
Because a void function does not have a value, it can be used only as a
statement; it cannot he used in an expression. Examine the call to the
greeting function in Figure 4 - 5. This call stands alone as a statement.
Including this call in an expression, as shown below, would he an error.
result = greeting!); // Error. Void function
void Functions with Parameters
.
Now let s call a function that has parameters but
still returns void . The func -
r pr ; -nt0ne> as seen in Figure 4-6,
*
this funet on returns nothing to the
receives an integer parameter Since .
calling function, main , its return type is
s uit 11 c greeting
function discussed previously, this function must
e coded as a stand-alone call
because it does not return a value; it cannot be
re
the
lmmoni
s noSP
tor
art
|
eS
’'^
"T LreXpreSSi effect
“ * 3 : The ’ -
n Note however that while printone
0 '
parameter value is printed to
and theynaUIUodfnFigUrC ^ ^ the "a °f the variable in main (a)

to he the same. On
easier to understand ^ ^^ be the same if that makeS il
called multiple times ^ print0ne to demonstrate that a function can be

Chapter 4 Functions 15 /
// Function Declarations
void printOne ( int x ); ^ a
int main ( void ) 5
{
int a = 5;
// Statements
printOne ( a ); // call
return 0;
> // main
void printOne ( int x )

{
printf( " %d\n " , x );
5 n
return; x
> // printOne
Side Effect
FIGURE 4 - 6 void Function with Parameters
PROGRAM 4 - 2 void Function with a Parameter

1 /* This program demonstrates that one function can be
2 called multiple times.
3 Written by:
4 Date:
5 */
7
8 // Function Declarations
9 void printOne ( int x );
10
11 int main ( void)
12 {
14 int a ;
15
16 // Statements
17 a = 5; // First call
18 printOne (a);
19
20 a = 33 ;
21 printOne ( a ); // Second call
continued
J
\m
( continued )
PROGRAM 4 - 2 void Function with a Parameter
22 operating system .
23 // Done. Return to
24 return 0;
25 > // main
26
27 / * =: = printOne
28 Print one integer value.
to be printed
29 Pre x contains number
30 Post value in x printed
31 * /
32 void printOne (int x)
33 {
34 // Statements
35 printf("%d\n" , x);
36 return ;
37 > // printOne
Results:
5
33
Non-void Functions without Porameters

Some functions return a value but don’t have any parameters. The most com-
mon use for this design reads data from the keyboard or a file and returns the
data to the calling program. We show this design in Figure 4 - 7. Note that the
called lunction contains all of the code to read the data. In this simple exam-
ple, only a prompt and a read are required. Later we introduce concepts that
require much more code.
// Function Declaration
int getQuantity (void);
P\
int main ( void )
int getQuantity ( void )
// Local Declarations // Local Declarations
int amt ;
int qty ;
// Statements // Statements
amt = getQuantity ( ); printf( " Enter Quantity " )?
return 0; scanf ( " %d " , &qty );
} // main return qty ;
} // getQuantity
FIGURE 4- 7 Non- void Function

without Parameters
t.
Non-void Functions with Parameters

Figure 4 -8 contains a function that passes parameters and returns a value— in
this case, the square of the parameter. Note how the returned value is placed
in the variable, b . This is not done by the call ; it is a result of expression eval -
—
uation . Since the call is a postfix expression , it has a value whatever is
returned from the Function . After the function has been executed and the
value returned , the value on the right side of the assignment expression is the
returned value, which is then assigned to b. Thus, again we see the power of
expressions in the C language . Note that the function , sqr , has no side effect .
In main , the function call is evaluated first because it is a postfix expression ,
which has a higher precedence than the assignment expression . To evaluate it ,
the function call is executed . Its value is the value returned by the called func -
tion . The expression value is then used in the binary expression (assignment).
The side effect of the binary expression stores the returned value in the variable .
int sqr ( int x ); Returned
int main ( void ) stored here
{
int a ;
int b ; a b
// Statements
scant( " %d " , &a );
b = sqr (a);
printf( " %d squared: %d \ n " , a , b) ;
return 0;
} // main
int sqr ( int x )

{
// Statements
return ( x * x );
> // sqr
FIGURE 4- 8 Calling a Function That Returns a Value
In Chapter 3 we saw that an expression becomes an expression statement

when it is terminated by a semicolon . The function call , which is a postfix
expression and is terminated by a semicolon , is an expression statement .
Functions that do not return a value ( void functions ) can only be used in an
expression statement as shown below.
printOne( ); // See Figure 4 6 -

Functions that return a value can be used either in an expression state-
.
ment or as an expression When it is used as an expression statement , the
, as in Figure 4-8,
return value is discarded .
When it is used as an expression
from the function . The following example dem-
its value is the value returned statement and as an expression .
Note
onstrates calling sqr as an expression the function call nor
that the expression statement is useless because neither
the statement has a side effect.
/ / Expression Statement
. No side effect
sqr ( a ) ; b
b = sqr ( a ) ; / / Return value is stored in
only function calls, lo demon-
In large programs, main is written with
be done with our simple example, we combine our
strate how this would
square and print functions into Program 4 - 3.
PROGRAM 4- 3 Read a Number and Square It

1 /* This program reads a number and prints its square.
2 Written by:
3 Date:
4 */
5 #include <stdio.h>
6
8 int getNum (void );
9 int sqr ( int x);
10 void printOne ( int x );
11
13 {
15 int a ;
16 int b;
17
18 / / Statements
19 // Get number and square it
20 a = getNum ();
21
22 // Square the number just
read
23 b = sqr (a);
24
25 // Now print it
26 printone( b);
27
28 return 0;
29 } // main
30
31 /* === :
i
=== getNum ===
continued
PROGRAM 4- 3 Read a Number and Square It (continued )

32 Read number from keyboard and return it.
33 Pre nothing
34 Post number read and returned
35 */
36 int getNum ( void )
37 {
39 int numln;
40
41 // Statements
42 printf( " Enter a number to be squared:
43 scanf ( "%d " , &numln);
44 return numln ;
45 > // getNum
46
47
48 Return the square of the parameter.
49 Pre x contains number to be squared
50 Post squared value returned
51 */
52 int sqr ( int x )
53 {
54 // Statements
55 return ( x * x );
56 } // sqr
57
58 ===== printOne ====
59 Print one integer value.
60 Pre x contains number to be printed
61 Post value in x printed '
62 */
63 void printOne ( int x )
64 {
65 // Statements
66 printf( "The value is: %d \n " , x );
67 return ;
68 > // printOne
Results:
Enter a number to be squared: 81
The value is: 6561
Program 4- 3 Analysis Our simple program has grown to four functions, including main . This is an example
of decomposition, the process of breaking a complex problem into simple parts.
While our example is not really complex, it still demonstrates the concept.
J
S,'!*,p .^ . '
o
* mete tolJ 9"«
“ ”*<"*•“’P"'i
Function Definition
Now that we have seen examples of the
functions in more detail. We
The function definition
begin
contains
with
the
the
code
.
basic function formats , let ’s look at
function dehmt on .
for a function . It is made up of
two parts: the function header and the function
body, which is a compound
statement must have opening and
statement . Remember that a compound
closing braces, and it has declaration and statement .
sections The function
definition format is shown in Figure 4 - 9 .
Function Header
return type functionname ( formal parameter list )
{
/ / Local Declarations
/ / Statements
} / / function name_
Function Body
FIGURE 4- 9 Function Definition
Function Header
A function header consists of three parts :
the return type, the function
name , and the formal parameter list . A
semicolon is not used at the end of
the function definition header.
With the implementation of the C 99
standard , the return type became a
Ivpe istJ ‘men WKen nothinS is to be returned , the return
^
Function Body
ments. Thebodvsl^ l 0Cal dctlaralions and the function state-
by the function. A f t e r S p C C f y the variables neede!<
‘ -
stateml f T J
n statements. terminal
ing with a return
be written without a
return
ments should be explicitly
every function , even void fimrc C
I i
’
>
,^? ^
statement are cod d'' ( wction
*
' WCU°retu
C arity we
> ons, have a return
>
type is void , it can
® u s e we believe that default state-
strongIy recommend that

statement.
Figure 4 - 10 shows two functions, first and second. The function

first lias been declared to return an integer value. Its return statement
therefore contains the expression x + 2 . When the return statement is exe -
cuted, the expression is evaluated and the resulting value is returned. The
function second returns nothing; its return type is void . It therefore needs no
return statement— the end of the function acts as a void return . Again, we
strongly recommend that you include a return statement even for void func -
tions. In this case, the return statement has no expression; it is just com-
pleted with a semicolon.
Function return type

should be explicitly
\ defined ^
int first (...) void second (...)
{
return statement
{ should be used even if
^ nothing is returned
^
return ( x + 2 ); return;
} // first } // second
FIGURE 4 - 10 Function Return Statements
Formal Parameter List

In the definition ol a function, the parameters are contained in the formal
parameter list . This list defines and declares the variables that will contain
.
the data received by the function The parameter list is always required. If the
Function has no parameters—that is, if it does not receive any data from the
calling function— then the fact that the parameter list is empty is declared
with the keyword void .
In C each variable must he defined and declared fully with multiple
parameters separated by commas. To make the parameter list easier to read
when there are multiple lines, we align the parameter types and their names
with tabs. Attention to these little details makes the code much easier to read
and understand.
In Figure 4 - 1 1 , the variables x and y are formal parameters that receive
data from the calling Functions parameters. Since they are value parame -
ters, copies of the values being passed are stored in the called Function ’s
memory area. If the Function changes either of these values, only the copies
will be changed. The original values in the calling Function remain
unchanged.
Two values received
from calling function
,int y ) parameter variables

double average (int x
{ y
double sum;
-
sum x + y ;
return (sum / 2);
local variable
sum 1
} // average
One value returned
to calling function
FIGURE 4- 11 Function Local Variables
Local Variables
a function and used with -
A local variable is a variable that is defined inside
between functions. Figure 4-11
out having any role in the communication
shows an example of a function with both formal parameters and a local vari-
able, sum.
Formal and Actual Parameters

Formal parameters are variables that are declared in the header of the func-
tion definition .
Actual parameters are the expressions in the calling statement .
Formal and actual parameters must match exactly in type, order, and num-
.
ber Their names, however, do not need to match .
Function Declaration
Function declarations consist only ol a function header ; they contain no
code. Like I unction definition headers , function declaration headers consist
of three parts: the return type, the function name , and the formal parameter
list . Unlike the header lor the function definition , function declarations are
terminated with a semicolon .
I he return type and the parameter list are required entries. If the pro *
gram has no parameters, we code void in the parentheses.

If the program has
multiple parameters, we separate each type- identifier set with commas.
I he C standard does not require
identifier names for the function decla
ration's formal parameters . This does not
-
prevent our using names in a lunc *
tion declaration , however. In fact ,

readability and understandahility are
improved if names are used. One point
to note , however, is that the names
do not need to he the same in the
function declaration and the function
definition; on the other hand, if the types are not the same, we get a compile
error. The compiler checks the types in the function statements
with the
in the call to ensure that they are the same or at least compatihle. I he
types
major reason to include the identifiers in the function
declaration is docu -
mentation; therefore, their names should be meaning ful. Don’t use generic
identifiers such as a x or .
Declarations are placed in the global declaration section before m a i n .
Grouping all function declarations at the beginning of the program
makes
them available whenever they are needed. Function declaratio ns also provide
an excellent quick reference for functions used in the program, making
them
excellent documentation .
Figure 4 - 12 demonstrates several of these concepts. The function decla
-
tells the that a function named multiply , which accepts two

ration program
needs;
integers and returns one integer, will he called. That is all the compiler
it does not need to know anything else to evaluate the call .
int multiply (int multiplier, int multiplicand );
int main (void )
{
int product ;
product = multiply ( 6 , 7 );
return 0 ;
} // main
7
42
int multiply
{
lint
^
x, i n t y ) |
X
5 6
return x * y ;
y 7
} // multiply
Function Definition
FIGURE 4 - 12 Parts of a Function Call
in the
Figure 4 - 12 also demonstrates that the formal parameter names
declaration do not need to be the same as the actual parameter names. In this
the function declaratio ns are much more meaningf ul and
case the names in
therefore should have been used in the function definition.
The Function Call

A function call is a postfix expression with precedence .
16 The operand in a
function call is the function name; the operator is the parenthe ses set, ( .. ),.
which contains the actual parameters . The actual parameter s identify the
function. They match the function's

values that are to be sent to the called
d order in the parameter list . If the program has
commas.
mu
,
and he fifth shows the function multiply ( a b as Us
,
own first parameter.
The last example sums it all up: Any expression that reduces to a smgle value
can be passed as a parameter .
multiply ( 6, 7 ) multiply ( a, 7 )
multiply ( 6, b ) multiply ( a + 6, 7 )
multiply ( multiply ( a, b ), 7 ) multiply ( )
expression ; expression
FIGURE 4 - 13 Examples of Function Calls
Functions can be classified by the presence or absence of a return value.

Functions that cannot return a value have a return type of void . Since these
types of functions do not have a value, they can be used only as a stand -alone
statement; that is, they cannot be included as part of an expression . All other
functions return a value and can be used either as part of an expression or as
a stand-alone statement , in which case the value is simply discarded .
Function Examples
This section contains four examples of programs
in which functions call
functions. Look for the points they demonstrate.
EXAMPLE 4- 1 Print Least Significant Digit

Program 4-4 prints the least significant (
from the keyboard. rightmost ) digit of any integer read
PROGRAM 4 - 4 Print Least Significant Digit

1 /* This program prints the first
2 read from the keyboard digits of an integer
3 Written by:
4 Date:
5 */
6 tinclude < stdio . h >
7
continued
PROGRAM 4 - 4 Print Least Significant Digit (continued )

9 int firstDigit ( int num );
10
12 {
14 int number ;
15 int digit;
16
17 // Statements
18 printf(" Enter an integer: " );
19 scanf ( "%d " , &number );
20
21 digit = firstDigit ( number );
22 printf("\nLeast significant digit is: %d \n" , digit );
23
24 return 0;
25 > // main
26
27 ==== firstDigit ====
28 This function extracts the least significant digit
29 of an integer.
30 Pre num contains an integer
31 Post Returns least significant digit.
32 * /
33 int firstDigit ( int num )
34 {
35 // Statements
36 return ( num % 10);
37 } // firstDigit
Results:
Enter an integer: 27
Least significant digit is: 7
from main . Note

Program 4 - 4 Analysis This extremely simple program demonstrates how to call a function is at the
that even though firstDigit is used only in main , its function declaration
when executed , it returned 7,
global level. In the sample run, firstDigi t was
which was then put into digit and printed.
EXAMPLE 4 - 2 Add Two Digits

digits of
Lets write a function that extracts and adds the two least significant
- . It consists of four
any integer number. The design is shown in figure 4 14
secondDigit . The func-
firstDigit, and
, mam. addTwoDigits
,
functions
charts.)
tro uble
understanding the structure
main
addTwo
Digits
secondDigit
Two Digits
FIGURE 4- 14 Design for Add
The code is shown in Program -

4 5.
PROGRAM 4 - 5 Add Two Digits

least
1 /* This program extracts and adds the two
significant digits of an integer.
2
3 Written by:
4 Date:
5 */
7
9 int addTwoDigits (int num);
10 int firstDigit (int num);
11 int secondDigit (int num);
12
13 int main (void )
14 {
16 int number ;
17 int sum;
18
19 // Statements
20 printf(" Enter an integer: ");
21 scanf ("%d", number);
22 ^
23 sum = addTwoDigits (number);
24 printf ("\nSum of last two digits is: %d " , sum);
continued
PROGRAM 4 - 5 Add Two Digits (continued )

25
26 return 0;
27 } // main
28
29 /* == addTwoDigits =========
30 Adds the first two digits of an integer ,
32 Post returns sum of least significant digits
33 */
34 int addTwoDigits (int number )
35 {
37 int result;
38
39 // Statements
40 result = firstDigit( number ) + secondDigit( number );
41 return result ;
42 > // addTwoDigits
43
44 ===== firstDigit ====
45 Extract the least significant digit of an integer ,
47 Post Returns least significant digit.
48 */
49 int firstDigit ( int num )
50 {
51 // Statements
52 return ( num % 10 );
53 } // firstDigit
54
55 /* =====
56 Extract second least significant ( 10s ) digit
57 Pre num is an integer
58 Post Returns digit in 10s position
59 */
60 int secondDigit ( int num )
61 {
63 int result ;
64
65 // Statements
66 result = ( num / 10) % 10;
67 return result ;
68 } // secondDigit
continuec
PROGRAM 4 - 5 Add Two Digits (continued
)
Results:
Run 1
Enter an integer:
23
two digits is : 5
Sum of last
Run 2
Enter an integer : 8
is : 8
Sum of last two digits
they first read this program is, "Why not

Program 4 - 5 Analysis A no turd question asked by students when - code in addTwoDigits?" This
put firstDigit and secondDigit as in line
. And aHer all, each or the
may seem to be an obvious way to code the problem
called functions is only one statement .
The answer is that, although each function is only one statement, it does a job that
can be used in other places. One of the principles of structured programming is that
processes should appear in a program in only one place. For example, we have used
the same code for firstDigit that we used in the first program . If a function is to be
reusable in this way, it must do only one thing. The short answer, then, is that it is better
structured programming. It is the nature of the task to be performed, not the amount of
code, that determines if a function should be used.
An interesting point to note is the way these two different digits were calculated. To
get the least significant digit, we took the 1 Os modulus of the number. But to get the sec -
ond digit, we had to divide by 10. Can you figure out how to sum the digits in a three-
digit number? We will give you a chance to do this in the problems at the end of the
chapter.
Note that we tested the program with two different numbers, one containing only
one digit. It is often necessary to run the program with more than one test case. Another
test case that should be run is a negative number. What do you think would happen?
As a programmer, not only should you run several tests but you should predict the
results before you run the program.
EXAMPLE 4 - 3 Format Long Integer

Program 4 -6 reads a long integer and prints it with a
comma after the first
three digits, such as 123,456. The number is
printed with leading zeros (sec
C Cati n % 03 d in
° Statement 40) in ease the value is less than
100 000
PROGRAM 4 -6 Print Six Digits with Comma

1 I / * This program reads long integers from the keyboard
2 and prints them
with leading zeros in the form
3 123 , 456 with ra
comma between 3rd & 4 th digit .
4 Written by:
5 Date:
continued
PROGRAM 4 - 6 Print Six Digits with Comma (continued )

6 */
9 void printWithComma ( long num);
10
12 {
14 long number ;
15
16 // Statements
17 printf( "\nEnter a number with up to 6 digits: " );
18 scant ( "%ld " , &number);
19 printWithComma ( number);
20
21 return 0;
22 > // main
23
24 /* =============== printWithComma ================
25 This function divides num into two three-digit
26 numbers and prints them with a comma inserted ,
27 Pre num is a six digit number
28 Post num has been printed with a comma inserted
29 */
30 void printWithComma (long num )
31 {
33 int thousands ;
34 int hundreds;
35
36 // Statements
37 thousands = num / 1000 ;
38 hundreds = num % 1000 ;
39
40 printf( "\nThe number you entered is \t %03d ,%03dM ,
41 thousands , hundreds );
42 return ;
43 } // printWithComma
Results:
Run 1
Enter a number with up to 6 digits: 123456
The number you entered is 123 ,456

continuec
(continued!)
PROGRAM 4 - 6 Print Six Digits with Comma
Run 2
up to 6 digits: 12
Enter a number with
entered is 000, 012

The number you
Prnnrnm 4 - 6
. ~ aaain• , we Lnave „a cimnlp
Anolvsis Once
_ nroaram
simple p 9 that has the. , of a very useful func-
makings
9 as commas
,
S Land t S? [;
s i c e his °g c wiN be
^
'
, " ,
over again, it must be in its own fund on
Note, however that more work s
over
needed to print numbers less than 100 ,000 correctly. However, we have not yet unre-
duced the tools to do the complete job. , ,
Again, we have used two test cases to show that more work must be done. An even
bigger problem occurs if we try to format a small negative number. Can you see what
the problem is? If not, code the problem and run it to see. We will explain how to han-
die these problems in Chapter 5.
EXAMPLE 4- 4 Print Tuition for Strange College

Our next example calculates and prints the annual tuition for a student
enrolled in Strange College. In this college , students can take an unlimited *
number of units each term . Each term , the students are charged S 10 per unit
plus a $ 10 registration fee. To discourage them from overloading, the college
charges $ 50 extra for each 12 units, or fraction thereof , a student takes after
the first 12 units. For example, if a student takes 13 units, the tuition is $ 190
( $ 10 for registration , plus 13 times $ 10 for units, plus a $ 50 penalty for the
one extra unit ). II a student takes 25 units, the tuition is $ 360 ( $ 10 for regis-
tration , plus 25 times $ 10 for units , plus $ 100 for two penalty fees ) . The
design lor this problem is shown in Figure 4 - 15.
Strange
College Fees
calculateFee
termFee termFee termFee
FIGURE 4-15 Design for Strange

College fees
The code for Strange College

is shown in Program 4 7 .
-
\\l
PROGRAM 4 -7 Strange College Fees

1 /* This program prints the tuition at Strange College.
2 Strange charges $ 10 for registration , plus $ 10 per
3 unit and a penalty of $50 for each 12 units, or
4 fraction of 12, over 12.
5 Written by:
6 Date:
7 */
9
10 _
# define REG FEE 10
11 #define UNIT FEE _ 10
12 # define EXCESS FEE 50
13
15 int calculateFee ( int firstTerm , int secondTerm ,
16 int thirdTerm );
17 int termFee ( int units );
18
20 {
22 int firstTerm ;
23 int secondTerm;
24 int thirdTerm ;
25 int totalFee;
26
27 // Statements
28 printf( " Enter units for first term: " );
29 scanf ( "%d " , &firstTerm );
30
31 printf( " Enter units for second term: " );
32 scanf ( " %d " , &secondTerm );
33
34 printf( " Enter units for third term:
35 scanf ( " %d " , &thirdTerm );
36
37 totalFee = calculateFee
38 ( firstTerm , secondTerm , thirdTerm );
39 printf( " \nThe total tuition is : %8d\n " , totalFee);
40
41 return 0 ;
42 > // main
43
continual
(continued )
PROGRAM 4- 7 Strange College Fees
44 =========== calculateFee the year.
45 Calculate the total fees for
to be taken each term.
46 pre The number of units
the annual fees.
47 Post Returns
48 */ , int secondTerm ,
49 int calculateFee (int firstTerm
50 int thirdTerm )
51 {
53 int fee;
54
55 // Statements
56 fee (firstTerm)
= termFee
57 + termFee (secondTerm )
58 + termFee (thirdTerm );
59 return fee;
60 } // calculateFee
61
62 /* === === termFee ============== =
63 Calculate the tuition for one term
64 Pre units contains units for the term
65 Post The fee is calculated and returned
66 */
67 int termFee ( int units )
68 {
70 int totalFees;
71
72 // Statements
73 totalFees =
REG FEE
74 + (( units - _
75
76
+ (units * UNIT FEE ); _
1 )/ 12 * EXCESS FEE )
return ( totalFees);
77 } // termFee
Results:
Enter units for first term: 10
Enter units for second term: 20
Enter units for third term:
30
The total tuition ii s :
780
Program 4 - 7 Analysis The most interesting aspect

of Prnnm AT S L
c
times in one function• Let's look ,
at how t works
| ^° W We
. The key ^
ca termFee three different
statement foliows.
fee = termFee ( firstTerm )

+ termFee ( secondTerm)
+ termFee (thirdTerm);
A function call is a postfix expression. Therefore, it evaluates from the left. To eval-
uate the expression ( three function calls) on the right of the assignment operator, we first
evaluate the first expression, the call to termFee with the number of units for the first
term. When termFee completes the first time, the return value ( 110 ) replaces the call .
At this point, we have the expression shown below.
110 + termFee (secondTerm ) + termFee (thirdTerm )
When termFee is executed a second time, its return value ( 260) becomes the
value of the second expression and we have
110 + 260 + termFee ( thirdTerm )
After the third call to termFee, the expression on the right of the assignment oper -
ator is ready for evaluation. Its value is 780, which is assigned to fee .
At least two more tests are needed to completely evaluate this program. We would
run it with all three terms having 0 units and then do another test of 11, 12, and
1 3 units.
4.4 Inter- Function Communication

Although the calling and called function are two separate entities, they need
to communicate to exchange data. The data flow between the calling and
called functions can he divided into three strategies: a downward flow Irom
the calling to the called function, an upward flow Irom the called to the call-
ing function, and a bi -directional flow in both directions. Figure 4 - 16 shows
the flow of data in these cases .
Calling Function Calling Function Calling Function
Called Function Called Function

:
Called Function
a . Downward b. Upward c . Bi-direction
FIGURE 4 - 16 Data Flow Strategies

176 Section 4.4 Inter-Function Communication
Basic Concept
concept before we discuss how they are done
Let us first discuss the general
in C .
Downward Flow
the calling function sends data to the called
“ rivsinThe
^ ^ opposite direction . In this strategy, copies of
^
o
the data items are passed from the calling function to the called I unction.
The called function may change the values
passed , but the original values in
example of this type of communi-
the calling function remain untouched . An
cation is passing data to a print function .
Upward Flow
Upward communication occurs when the called function sends data hack to
the called function without receiving any data from it . A good example of this
type of communication is when the called function reads data from the key-
board that needs to be passed hack to the called function .
Bi-directional Flow
Bi - directional communication occurs when the calling function sends data
down to the called function . During or at the end of its processing, the called
function then sends data up to the calling function . For example, the calling
function may send data to the called function , which it manipulates and
sends up to the calling function .
C Implementation
Now let us examine how these three types of communication are imple-
mented in C . In the discussion that follows, it is important to remember that
the flow refers to the data . The data flow s down to the called function. The
'
data flows from the called function hack to the calling
function .
Most programming languages have three strategies for inter function
communication: pass by value , pass by
-
guage , unfortunately, uses only the first
reference, and return . The C lan-
and last strategies; there iis no pass by
reference in C.
,ue
communirat
communications between ° ^3SS
^ Va
a calling and a called
and return achieve three types of
function.
Downward Communication
e rct (dved from the calling function. The

JLV v
calling function sends a copy of each value to the called function; no data
( lows upward.
Figure 4 - 1 7 demonstrates downward communication. Two data items are
passed from main to the downFun function. One data value is a literal, the
other is the value of a variable.
int main ( void ) void downFun ( int x , int y )

{ {
int a;
return ;
downFun ( a , 15);
> // downFun
} // main
FIGURE 4 - 1 7 Downward Communication in C
Downward communication is one - way communication. I he calling

function can send data to the called function, hut the called function cannot
send any data to the calling function. Figure 4 - 1 8 demonstrates one - way
communication.
void downFun ( int x , int y );
int main ( void )
{
// Local Definitions
int a = 5; a 5 15
// Statements C prints 5 I T
downFun ( a , 15 );,
printf("%d\n" , a ) ; I I
return 0;
One - way
} // main communication
I
void downFun (int x , int y )
{
// Statements
tw X X 20 y 15
x = x + y;
return; Only a copy
} // downFun
FIGURE 4 - 1 8 Downward Communication
Upward Communication
C provides only one upward direction flow, the return statement. \ \ hile it
works well, only one data item can he returned. The only way that a called
function can pass multiple data items up to the calling function is to access
-
F'
not allow us o
I herefore , we
»
.
words, we cannot access a
needf
The solution is 0 h(
to the called function . Given the
then put the data in the calling
1st
variab|e in the calling function by its identifier.
solve the problem .
frallingfanction
variable
function
's
t0 Pass thc address of thc variable
.
address, the called function can
1 he calling function needs to
variable to receive the data . The called function needs to
declare a data
that it receives from the calling
declare a variable to store the address
function .
In C , a variable that can store data is different from
the variable that can
store an address. So the called function needs a special type of variable, an
variable , or as it is called in C , a pointer variable , that points to the
address
variable in the called function .
The calling function needs to pass the address of a variable to receive thc
data from the called function . This action is a pass - by-value mechanism
although what is passed is not data , it is an address.
To get the address of a variable, we use the address operator ( & ). If the
name of the variable in the calling function is x , we extract its address using
the address operator ( & x ) and pass it as a parameter in the function call. Thc
following example passes the addresses of a and b to a function named upFun.
upFun ( & a , & b ) ;
I he called function needs to declare that the parameter is to receive an

address; in other words, it needs to create an address variable. To declare an
address variable , we use an asterisk ( * ) after the type. In other words, il x in
the calling I unction is ol type int , we need to declare and define a variable in
the called function of type int *. Ihis is done in the header of the called
function. For example, to define our upFun function , we use the following
function header.
void upFun ( int * ax , int * ay )
I5 u ^ JJ
MTRESS
‘ aster sk signifies that the variables
variables
ax and ay are not data
variables holding the address of int variables. Note that the
asterisk as used , n the declaration
and ay ).
belongs to the type not the variables ( ax
address parameter ^ function > we need use the variable's

l(>
indirect access - rh •
^* t lrou
^ an address variable is known as
asterisk is known as the inrli change the data For this * reason, the
*
use an asterisk. This

time le astt nstT
pe
? ?is put
ltor To change -
the data , we again
address variable’s name F
*
îmmediately in front ol the

function to 23, we wouldu use the
^ l C ange tFe
°
following statement. ^
variable in the calling
*ax = 23 ;
It is important to remember that these two uses of the asterisk, declaring

an address variable and indirectly accessing the data, play completely differ-
ent roles. The first asterisk, in the declaration, belongs to the type; the second
asterisk is an expression that indirectly accesses the variable in the called pro-
gram as shown in Figure 4 - 19.
int main ( void )

{
tv void upFun ( int * ax , int* ay )
{
int a ; *ax = 23;
int b ; *ay = 8;
return ;
upFun (&a , & b ); } // upFun
> // main
FIGURE 4 - 19 Upward Communication in C
We need to emphasize that when only one data item needs to he

returned, we use the standard return statement. Only use upward communi -
cation when multiple items need to be returned. Figure 4 - 20 demonstrates
the techniques for upward communication.
void upFun ( int* ax, int* ay
int main ( void )
Type includes
{
int a ;
Address 23 8
int b; operators. a b
// Statements
upFun (&a , & b );
printf( " %d %d \n " , a, b);
return 0 ; Dereference
} // main Prints 23 andjjT
void upFun ( int* ax, int* ay )
{
// Statements ax ay
*ax = 23;
__ jay Address
— L
= 8;
>eturn;
^
(pointer )
FIGURE 4 - 20 Upward Communication

s symbol in front of the data variable when we call the
1. Weneed to use the
function. when we declare the
sy mbol after the data type
2. We need to use the *
address variable , , .
we store data
. . . ..
front of the variable when indirectly.
3. We need to use the * in
Bi-direction Communication
The strategy described for the upward
direction can easily be augmented !
in both directions . The only difference is that tin
allow the communication
indirect reference must be used in both sides of the assignment statement
The variable in the called function first is accessed for retrieving using tk
side . The same parameter is accessed agar
address variable in the right - hand
to store a value in the left hand- side . We demonstrate bi -
directional accessii
Figure 4- 21 .
i n t main ( void )
{
void biFun ( i n t * ax , i n t * ay )
{
4
int a; * ax = * ax + 2 ;
i n t b; *ay = * ay / * ax ;
return;
biFun ( & a , & b ) ; } / / biFun
} / / main
FIGURE 4 - 21 Bi - directional Communication in C
Figure 4 - 22 demonstrates bi-directional communication flow.
Communication Summary
Let ’s summarize the rules for inter-function communication . Like man
aspects ol learning a language, they must be
memorized.
.
1 Rules lor downward
communication
a. Use values in the functiion call to pass data ,
.
h Use ,appropriate data types in
the functiion parameter list to receive tk
data values.
on to access the loe
2. Rules for upward and bi-directional communication

a. Use &variableName in the function call to pass a reference to the
variable.
b . Use type * in the function parameter list to receive the variables
address.
c. Use * parameterName in the function to reference the original variable.
void biFun (int* ax, int* ay);
int main ( void )
{
int a = 2; & 4 £ 2
int b = 6;
a b
// Statements
biFun (&a, & b);
return 0 ;
} // main
Dereferences
^
)
void biFun ( int* ax, int* ay)

{
*ax = *ax + 2;
k ax ay
*ay = *ay / *ax;

return;
} // biFun
FIGURE 4 - 22 Bi -directional Communication
Communication Examples
In this section we write some short programs to demonstrate function com -
munication .
EXAMPLE 4 - 5 Exchange Function

Let s look at a program that uses the indirection operator to dereference data.
One common process that occurs often in programming is exchanging two
pieces of data . Let ’s write a function that, given two integer variables ,
exchanges them . Since two variables are being changed , we cannot use the
return statement . Instead we pass addresses to make the changes.
j
not
.
F rst, make sure you
First
simply assign va
understand how to exchange
^ ^ as shown
two
below.
variables: We can-
/ / This won ' t work

y
-
x = y;
x;
/ / Result i s y i n both
statements, you will see that the original

If you carefully trace these two Therefore, to exchange variables, we
value of y ends up in both variables
.
variable to hold the first value while the exchange
need to create a temporary
is shown below.
is being made. The correct logic
/ / value of y s a v e d
hold = y;
/ / x now i n y
y = x;
x = hold ; / / o r i g i n a l y now i n x
The exchange function and its data flow are shown in Figure 23. First,
4-
examine the function declarations in Figure 4 - 23 carefully. Note the asterisk
in the declaration of numl and num 2 . The asterisk is used with the type decla-
—
ration to specify that the type is an address in this case , the address of an
integer. Now, look at the call statement in main. Since we will change the val-
ues of a and b in main, we need to pass their addresses. The address operators
( & ) tell the compiler that we want their addresses passed, not their values.
void exchange (int* numl, int* num2); Note that the type
includes an asterisk.
int main ( void )
{
int a ;
int b;
a b
// Statements
exchange (&a, &b);

Dereferences
return 0; Address
} // main operators
K
void exchange (int* numl, int
* num2)
numl num2
int hold; numl and num2
// Statements are addresses
hold = *numl; Note the indirection )
* numl = *num2; operator is used for
*num2 = hold; dereferencing.
return;
} // exchange
t* FIGURE 4- 23 Exchange Function

V
Now, look at the statements in exchange. First , we copy numl ’s value

to hold. Since hold is a local variable, it is treated the same as any local
— .
variable no special operators are required However, numl contains the
address of the data we want , not the data itself. To get the value it is referring
to, we must dereference it . This is done with the indirection operator, the
asterisk ( * ). The statement is shown below.
hold = *numl ;
In the same manner, we can now dereference num2 and copy it to numl
and then complete the exchange hy copying hold to num2. These statements
are seen below'.
*numl = *num2 ;
*num2 = hold ;
Note that with the exception of hold , all of the data movement is done in
the calling program ’s area . This is the power of the indirection operator.
- -
Here is another simple example, one that uses both pass by value and
pass- by address parameters. We need to write a function that , given two
-
numbers, calculates both the quotient and the remainder. Since we can t
’
return two values, we will pass the addresses where we want the quotient and
remainder stored . This problem is shown in Figure 4 - 24.
void divide ( int divnd , int divsr ,
tv a b quot rem
int* quot , int* rem ); K] [Z

int main ( void )
{
int a ;
int b ;
int quot;
int rem ;
// Statements
divide (a, b, &quot, &rem);
return 0 ;
} // main
void divide ( int divnd , int divsr

int* quot , int* rem )
// Statements 14
{ divnd quot rem
*quot = divnd / divsr ;
*rem = divnd % divsr ;
return ;
> // divide
FIGURE 4 - 24 Calculate Quotient and Remainder
j
Note that the first two parameters are pass by

Let ’s exa minedlV1
^ their types are just inf there

indicating that they are addresses
, any references to them
Their types are int *; therefore
use the indirection operator
.
are no asterisks
value. You can tell this cause . The last two parameters are addresses.
in the function must
Now look at the call in main. Note that a and b are simply passed by
using their identifiers. They are
primary expressions whose value is the con-
tents of the variable. On the
other hand , quot and rem are passed as
. In this case, the value of the primary
addresses by using the address operator
and the identifier is the address of
expression created by the address operator
value rather than a data value.
the variable. Therefore, we pass an address
what needs to be passed , a value
As the programmer, it is our job to know
. Similarly, when we use the parameters in
or an address , when we write a call
the called program , we must remember to use the indirection operator when
we have an address .
Let ’s use divide in a program that reads two integers , divides them , and
then prints the quotient and the remainder. The design for this program is
shown in Figure 4 -25.
Quotient
Remainder
getData divide print
FIGURE 4- 25 Quotient and Remainder Design
The code is shown in Program 4 -8.
PROGRAM 4 - 8 Quotient and Remainder

1 /* This program reads two integers
and then prints the
2 quotient and remainder of
the first number divided
3 by the second .
4 Written by:
5 Date:
6 */
7 include <stdio.h >
8
^
9 / / Function
10
Declarations
void divide ( int
11 dividend , int divisor ,
int * quotient , int *
remainder ) ;
-
c
- ^
continue
PROGRAM 4 - 8 Quotient and Remainder ( continued )

12 void getData ( int* dividend , int* divisor );
13 void print ( int quotient , int remainder );
14
16 {
18 int dividend ;
19 int divisor ;
20 int quot ;
21 int rem;
22
23 // Statements
24 getData ( &dividend , &divisor );
25 divide ( dividend , divisor , &quot , &rem );
26 print ( quot , rem );
27
28 return 0 ;
29 } // main
30
31 /* ================== getData ==================
32 This function reads two numbers into variables
33 specified in the parameter list.
34 Pre Nothing.
35 Post Data read and placed in calling function.
36 */
37 void getData ( int* dividend , int* divisor )
38 {
39 // Statements
40 printf( " Enter two integers and return: " );
41 scanf ( " %d %d " , dividend , divisor );
42 return;
43 > // getData
44
45 /* ===
46 This function divides two integers and places the
47 quotient/remainder in calling program variables
48 Pre dividend & divisor contain integer values
49 Post quotient & remainder calc'd
50 */
51 void divide ( int dividend , int divisor ,
52 int* quotient, int* remainder )
53 {
54 // Statements
55 quotient = dividend / divisor ;
continued
j
186 Section 4.5 Standard Functions
(continued )
PROGRAM 4 - 8 Quotient and Remainder
dividend % divisor;
56 remainder =
57 return ;
58 } // divide
=== print ========= the

-
59
60 / * remainder
the quotient and
61 This function prints quotient
the
62 Pre quot contains
rem contains the remainder
63 printed
64 Post Quotient and remainder
65 */
66 void print ( int quot
, int rem)
67 {
68 // Statements
;
69 printf ("Quotient : %3d \n" , quot )
" Remainder: % 3d \n " , rem );
70 printf (
71 return ;
72 > // print
main contains only calls to sub-

Program 4- 8 Analysis First, look at the design of this program. Note how
functions. It does no work itself; like a good manager, it delegates all work to lower
levels in the program . This is how you should design your programs.
Study the getData function carefully. First note that the parameters identify the
variables as addresses. Verify that they contain an asterisk as a part of their type. Now
look at the scanf statement carefully. What is missing? There are no address operators
for the variables! We don't need the address operators because the parameters are
already addresses. We have used the address operator ( & ) up to this example to tell the
compiler that we want the address of the variable, not its contents. Since the parame-
ters were passed as addresses ( see Statement 24), we don't need the address operator.
Now study the way we use quotient and remainder in divide. We pass
them as addresses by using the address operator. Since we are passing addresses, we
can change their values in main . In the function definition, the formal parameters show
these two parameters to be addresses to integers by showing the type as int * ( see
Statement 52). To change the value back in main, we use the indirection operator when
we assign the results of the divide and modulo operations.
4.5 Standard Functions

C pro \ idts a rich collection of standard
functions whose definitions have
ecn written and arc reads to be
used in our programs. To use these func-
tions . we must include their
function declarations. The function declarations
or these
Junctions
are grouped together and collected in several header files.
°r "
he function declaration, f
• . ^
'
Chapter 4 Functions 18 /
Figure 4 - 26 shows how two of the C standard functions that we have

used several times are brought into our program . The include statement
causes the library header file for standard input and output ( stdio .h ) to be
copied into our program . It contains the declarations for print/ and scunf .
Then , when the program is linked , the object code for these functions is com -
bined with our code to build the complete program .
înclude
int main ( void)
^-
<stdio.h> t r
= Generated from
stdio header file Mint scanf(...);
int printf(...);
int main ( void )

{
{ ...
scant(...); scant (...);
Our program
printf (...); printf(...);
return 0; return 0;
} // main
} // main
int scant(...)
{
return ... f
Definitions } // scant
added by linker
int printf(...)
{
return ... ,
} // printf
FIGURE 4-26 Library Functions and the Linker
Math Functions
Many important library functions are available for mathematical calculations.
Most of the function declarations for these functions are in either the math
header file ( math .h ) or standard library ( stdlib.h ). In general , the integer
functions are found in stdlib.h .
Absolute Volue Functions

The functions described in this section return the absolute value of a num -
ber. An absolute value is the positive rendering of the value regardless ol its
sign . There are three integer functions and three real functions.
The integer functions are absf labs , and Hubs . For abs the parameter must
he an int and it returns an int . For labs the parameter must be a long int , and
it returns a long int . For Hubs the parameter must be a long long int , and it
returns a long long int .
// stdlib.h
number);
int abs (int // stdlib.h
labs (long
number);
long // stdlib.h
long long number);
long long 11abs (
jw .=:«ss£Kites;
double version is named fabsl .
number); // math.h
double fabs (double // math.h
fabsf (float number );
float / / math.h
long double number );
long double fabsl (
EXAMPLE 4-6 Absolute Numbers
abs ( 3) returns 3
-
fabs ( 3.4 ) returns 3.4
Complex Number Functions

The functions for manipulating complex numbers are collected in the complexJi
header file. Some of the complex number functions are shown below.
double cabs (double complex number ); // absolute

float cabsf (float complex number);
long double cabsl (long double complex number );
double carg (double complex number );
float cargf (float complex number ); // argument
long double cabsl (long double complex number );
double creal (double complex number ); // real
float crealf (float complex number );
long double creall (long double complex number );
double cimag (double complex number ); // imaginary
float cimagf (float complex number );
long double cimagl (long double complex
number );
Ceiling Functions
A ceiling is the smallest integral value
greater than or equal to a numher. For
e \am|) e , t e cei ing o(
3.0000001 is 4. If we consider all numbers as a con-
muous range from minus infinity to plus
moves the number right to an integral infinity ( Figure 4-27 ) , this function
value.
-
t
u
ceil (-1.9) ceil (+1.1 )

. is -1.(K . is +2.C)
. -2.0 -1.0 0 1.0 2.0 +
FIGURE 4 - 27 Ceiling Function
Although the ceiling functions determine an integral value, the return

type is defined as a real value that corresponds to the argument. The ceiling
function declarations are
double ceil (double number);

float ceilf ( float number );
long double ceill ( long double number );
EXAMPLE 4-7 Ceiling Function
-
ceil ( 1.9 ) returns -2.0
1.0
ceil ( 1.1 ) returns
Floor Functions
A floor is the largest integral value that is equal to or less than a number ( see
Figure 4 - 28). For example, the floor of 3.99999 is 3.0. Again, looking at num -
bers as a continuum, this function moves the number left to an integral value.
floor (-1.1)
^ is -2.0 ^ .flooris +(+1.01.9^)
. -2.0 -1.0 0 1.0 2.0 +
FIGURE 4 - 28 Floor Function
The floor function declarations are
double floor ( double number );

float floorf ( float number );
long double floorl ( long double number );
EXAMPLE 4- 8 Floor Function
-
floor ( 1.1) returns
floor ( 1.9) returns
2.0
1.0
-
Truncate Functions
The truncate functions return the
same as floor function for positive
..
integral in the d rect on of 0. They are the
numbers and the same as ceiling function
are
for negative numbers. Their function declarations
double trunc (double number );

float truncf (float number );
long double truncl ( long double number );
EXAMPLE 4-9 Truncate Function
-
trunc ( 1.1) returns 1.0
trunc ( 1.9 ) returns 1.0
-
Round Functions
The round functions return the nearest integral value . Their function declara -
tions are
double round (double number );

float roundf (float number);
long double roundl (long double number);
In addition to the real round functions, C provides two sets that return a long
int or a long long int . Note that there is not a round function that returns an
int .1 heir function declarations are
long int lround (double number );

long int lroundf (float number);
long int lroundl (long double number );
long long int llround (double
long number );
long int llroundf (float
long number);
long int llroundl (long
double number );
EXAMPLE 4 - 10 Round Function

round ( 1.1)
round ( 1.9 )
- returns -1.0
returns 2.0
round ( 1.5)
- returns 2.0-
Power Function
The power ( pow ) function returns the value of the x raised to the power y
that is, xy . An error occurs if the base ( x) is negative and the exponent ( y ) is
not an integer, or if the base is zero and the exponent is not positive. The
power function declarations are
double pow ( double nl , double n2 );

float powf (float nl, float n2);
long double powlf ( long double nl , long double n 2 );
EXAMPLE 4- 11 Power Function
pow ( 3.0 , 4.0 ) returns 81.0

pow ( 3.4, 2.3 ) returns 16.687893
Square Root Function

The square root functions return the non - negative square root of a number.
An error occurs if the number is negative. The square root function declara-
tions are
double sqrt (double nl ) ;

float sqrtf ( float nl );
long double sqrtlf ( long double nl );
EXAMPLE 4- 12 Square Root Function
sqrt ( 25 ) returns 5.0
Random Numbers
A random number is a number selected from a set in which all members have
the same probability of being selected. Random numbers are useful in many
areas of computer science. Two examples are application testing and gaming.
Random Number Generation

Because of their usefulness, most languages provide functions to generate
random numbers. Each time the function is called, it generates another num-
ber. The function that generates random numbers uses a formula that has
been mathematically designed to ensure that each number in the set in fact
has the same probability of being selected; in other words, to ensure that
each number is truly random. As a part of the formula, the function uses the
previous random number. 1 he design is shown in Figure 4 - 29.
192 Section 4.5 Stondord Functions
Next
Previous Random Number Number
Number
Function
FIGURE 4- 29 Random Number Generation
design of a random number generator,

Now that we un derstand the basic
the first random number generated?" The
a logical question is “How is
can use the default seed provrded by the sys-
answer is from a seed . In C, we
tem ( 1 ) or we can specify our own seed .
Because C uses only one algorithm to generate
random numbers, the
pseudorandom number series. Only by using differ-
series it produces is a
ent seeds can we generate different random series . Calling the random
function several times generates a series of random numbers as shown in
Figure 4- 30.
Random Number First

Seed Function ^ Number
Random Number Second

^ Function ^ Number
FIGURE 4 - 30 Generating a Random Number Series
_
We can create the same series or a different series
— seri of random numbers
each time we run our program . Some programs require that the series be the
same each time the program runs; others require that the series be different
each time. We control the type of series by our choice of
seeds.
Random Numbers in C
c provides two functions
^ and rand m
° ^
Seed Random Number Function
to
-
build a random number series, seed random
Th e functions are found in stdlib.h .
Hie seed random functionn, ( srand ),

Its I unction declaration is creates the starting seed for a number series
•
void srand ( unsigned
int seed );
EXAMPLE 4- 13 Generate the same series in each run

To generate the same number series in each run , we can either omit srand or
we can provide a constant seed random , preferably a prime number, such
as 997 .
srand (997 );
EXAMPLE 4- 14 Generate a different series in each run

To generate different series in each run , we use the time of day as the seed .
.
The C call for the time, which requires the time h library, is used in the fol -
lowing example.
srand(time ( NULL));
Whichever series we need , srand shoidd be called only once lor each ran -
dom number series, usually only once in a program.
srand must be called only once for each random number series.
Random Number Function

The random number function , ( rand ) returns a pseudorandom integer
between 0 and RAND_MAX, which is defined in the standard library as the larg-
est number that rand can generate. The C standard requires that it be at least
32 , 767. Each call generates the next number in a random number series. The
random number function declaration is
int rand (void);
EXAMPLE 4 - 15 Program to create a different series in each run

Program 4 -9 is a simple program that prints three random numbers. Because
we seed the random number generator with the time of day, each program
execution generates a different series. We have run it twice to demonstrate
that the random numbers are different .
PROGRAM 4 - 9 Creating Temporal Random Numbers

1 /* Demonstrate the use of the time function to generate
2 a temporal random number series.
3 Written by:
4 Date:
5 */
continuec
Numbers (continued )
PROGRAM 4 -9 Creating Tempora Random
|
7 tinclude < stdlib

. h>
8 # include < time . h >
9
10 i n t main ( void )
11 {
12 // statements / / Seed temporally
13 srand ( time ( NULL ) ) ;
14
15 printf ( " % d \ n " , rand ( ) ) ;
16 printf ( " % d \ n " , rand ( ) ) ;
17 p r i n t f r % d \ n " , rand ( ) ) ;
18
19 return 0 ;
20 } / / main
Results:
F i r s t Run
9641
16041
6350
Second Run
31390
31457
21438
EXAMPLE 4- 16 Program to create the some series

Program 4- 10 is a simple program that prints three random numbers.
Because we seed the random number generator with a constant, each pro-
gram execution generates the same series. We would have
also generated the
same series if we didn 't use srand . We have run it
twice to demonstrate that
the random numbers are the same.
PROGRAM 4 -10 Creating Pseudorandom Numbers

l / * Demonstrate the use of
the srand function t o
2 g e n e r a t e a pseudorandom
number s e r i e s .
3 Written by :
4 Date :
5 */
6 # include <stdio h > .
7 # include < stdlib
h> .
8 tinclude < time h > .
9
c-
continue*
\\
PROGRAM 4- 10 Creating Pseudorandom Numbers ( continued )

11 {
12 // Statements
13 srand(997);
14
15 printf( " %d \n " , rand( ));
16 printf( " %d \n " , rand( ));
17 printf("%d\n" , rand());
18
19 return 0 ;
20 } // main
Results:
First Run
10575
22303
4276
Second Run
10575
22303
4276
Using Random Numbers

The numbers returned by the random number generator are in a large range.
The C standard specifies that it must he at least 0 to 32,767. W hen we need
a random number in a different range, we must map the standard range into
the required range. Two common ranges are a real - number series, usually 0.0
to 1.0, and an integral range, such as 0 to 25 or 1 1 to 20.
2
! !! ]
'
Generating Random Integrals
To generate a random integral in a range x to y, we must first scale the num -
ber and then, if x is greater than 0, shift the number within the range. We
scale the number using the modulus operator. For example, to produce a ran-
dom number in the range 0... 50, w'e simply scale the random number as
shown below. Note that the scaling factor must be one greater than the high-
est random number needed.
rand ( ) % 51
Modulus works well when our range starts at 0. But what il we need a
different range? In that case, we must shift the result . For example, suppose
is 0.99999999.
2. Note that 0.0 is included in the range but 1.0 is not. The maximum number
.r between 3 and 7 ( Figure 4-31 ). If we call rand

L
we want a random nu
bc 0 through 7. To convert to the
and then use modu
' ^
-
“ factor by subtracting the start-
.-
correct range we determine our modulus
divjsor ( g) and then adding the starting point
« number. Th» , f«
makes the modulus di\isor 5.
I " *. 3 8'
oh RAND
Scaling
Shifting
4 RAND MAX *
7 RAND _ MAX
FIGURE 4 -31 Random Number Scaling for 3-7
Generalizing the algorithm , we get
rand ( ) % range + minimum
where range is ( maximum - minimum ) + 1 , minimum is the

minimum num -
ber and maximum is the maximum number in the desired
range. For example,
to create a random number in the range 10-20 ,
we would use the following
expression :
range = ( 20 - 10 ) + 1;
/ / 11
randNo = rand( ) % range + 10;
Program 4- 11 generates a random

number series in the range 10 to 20.
PROGRAM 4 - 11 Generating Random
I
Numbers in the Range
i 10 to 20
'
l /* Generate a random series in

2 the range 10 to 20.
Written by:
3 Date:
4 */
5 iinclude <stdio.h >
6 include <stdlib . h >
7
^# include <time.h >
8
9 int main ( void )
continued
iii
PROGRAM 4-11 Generating Random Numbers in the Range 10 to 20 ( continued )

10 {
12 int range ;
13
14 // Statements
15 srand(time( NULL));
16 range = ( 20 -
10 ) + 1 ;
17
18 printf( "%d " , rand() % range + 10 );
19 printf( " % d ” , rand( ) % range + 10);
20 printf( " %d \n" , rand() % range + 10 );
21
22 return 0;
23 > // main
Results:
10 11 16
Generating a Real Random Number Series

To generate a real random number series between 0.0 and 1.0 we divide the
_
number returned by the random number generator by the maximum random
number(RAND MAX), which is found in stdlib . h , and then store it in a real num-
ber type . Program Program 4 - 12 demonstrates the generation of a real ran -
dom number series in the range ol 0 to 1 .
PROGRAM 4 - 12 Generating Random Real Numbers

1 /* Generate a real random series in the range 0 to 1.
2 Written by:
3 Date:
4 */
6 # include <stdlib.h >
7 # include <time.h >
8
9 int main (void )
10 {
12 float x;
13
14 // Statements
15 srand(time( NULL));
16
continues
198 Section 4.6 Scope
Real Numbers (continued )

PROGRAM 4-12 Generating Random _
( float ) rand ( )
/ RAND MAX ;
17 x =
18
19
printf ( " % f " r x ) ;
x ( float ) rand ( )
=
_
/ RAND MAX ;
p r i n t f ( " % f" i x ) '
20
21 x = ( float ) rand ( ) / RAND
_
MAX ;
22 printf ( " % f \ n " » x ) ?

23
24 return 0;
25 } / / main
Results :
0.782006 0.264260 0.348460
To scale a random real number into a range other than 0.0 to 1.0 ( exclu-
sive ) , we first generate the random number as we did in Program 4 - 12. Then
we re- scale it to the new range. For example , to generate random real num -
bers in the range 100.0 to 300.0 ( exclusive ), we multiply the random number
by the range ( 200.0 ) and then add the minimum ( 100.0 ). Using the results
from Program 4- 12, our random real numbers would be 256.4012, 152.852,
and 169.692.
4.6 Scope
Scope determines the region of the program in which a defined object is
—
visible that is, the part of the program in which we can use the object’s
name. Scope pertains to any object that can be declared , such as a variable or
a function declaration. It does not pertain directly to precompiler directives,
such as define statements; they have separate rules. Scope is a source pro-
gram concept: It has no direct bearing on the run - time program .
To discuss the concept of scope, we need to review two concepts. First , a
block is zero or more statements enclosed iin a set of braces . Recall that a
functions body is enclosed in a set of braces; thus, a body is also a block. A
block has a declarations section and a statement
section . This concept gives
us the ability to nest blocks within the
body of a function and allows each
block to be an independent group of
statements with its own isolated defini-
tions. Second , the global area of
our program consists of all statements that
are outside functions. Figure 4
-32 isi a graphical representation of the con-
cept of global area and blocks.
An object 's scope extends from its
variable is in scope if it is visible - ion until the end of its block . A
declarati
to the statement being examined . Variables
are in scope from their point
of declaration until the end of their block.
.
k i V V|i
/* This is a sample to demonstrate scope. The techniques

used in this program should never be used in practice.
*/
#include <stdio.h>
int fun (int a, int b); Global area
i
int main ( void )

{
int a; main
i
s area
int b;
float y ;
{ // Beginning of nested block

float a = y / 2 ;
float y ;
float z; Nested block
area
z = a * b;
} // End of nested block
} // End of main
int fun (int i, int j)

{
int a ; fun's area
int y ;
} // fun
FIGURE 4 - 32 Scope for Global and Block Areas

i
Global Scope
The global scope is easily defined . Any object defined in the global area of a
program is visible from its definition until the end of the program . Referring
to Figure 4- 32 , the function declaration for fun is a global definition . It is vis-
ible everywhere in the program .
Local Scope
Variables defined within a block have local scope . They exist only from the
point of their declaration until the end ol the block ( usually a function ) in
which they are declared. Outside the block they are invisible.
In Figure 4 - 32 , we see two blocks in main . The first block is all of main .
Since the second block is nested within main , all definitions in main are visible
J
200 Section 4.7 Programming Example — IncrementalDeveJopmerU
, L local variables with an

identical name are defined. In the
M
inner
\ aridutc ara -
these circumstances, the integer floating-point variable a m the nested block.
of the
ion until the declaration the nested block. Any
At that point , main’s a can
no longer be referenced m
a will get the float version . Once we
statement in the block that references float a is no longer in scope and the
, the
reach the end of the nested block
integer a becomes visible again .
until the end of their block.

Variables are in scope from declaration
y. Note, however, that before we

We have also defined a new variable
y to set the initial value for a. Although
defined the local y, we used main s
this is flagrant disregard for structured programmin g principles and should
, it demonstrate s that a variable is in scope until it is
never be used in practice
. Immediately after using y , we defined the local version , so main' s
redefined
b is not redeclared in
version of y is no longer available. Since the variable
the block , it is in scope throughout the entire block.
It is poor programming style to reuse identifiers within the same scope.
Although mains variables are visible inside the nested block, the reverse
is not true. The variables defined in the block, a , y , and z , exist only for the
duration of the block and are no longer visible after the end of the block.
,
Within the function fun , which is coded after main only its variables
and any global objects are visible. Thus, we are free to use any names we
want . In Figure 4- 32, we chose to use the names a and y even though they
had been used in main . This is an acceptable practice; there is nothing wrong
with it .
4.7 Programming Example — Incremental Development

In Section 4 8, “Software Engineering. we discuss a
concept known as “Top-
"
n eve opment . Top-down development , a concept inherent to modu -
WS US t0 deve
uTdmm, ln ,
T‘ » , ' *
-
8'
m L ° °P programs incrementally. By writing
”malt,, T r
88 08 Ca i0 n Separatd we are able to solve the program in
, '
* “ “ •*
P e « To
b ta the development*of a calculator • demonstra
vve begin te the conce
program in this section . In later *
chapters we add functionality.
then calls aTttrtu.dttt

sum of the two numbers. numbers !T tW numbers - The Pr gram
“‘"t ” ’ . It *concludes
° °
by displaying the
The design is shown in Figure
( mam ) , and three
,
4 -33 1 consists ol a calculator function
subfunctions’ 9etData > add , and printRes .
Calculator
i
getData
i add
I printRes
i
FIGURE 4 - 33 Calculator Program Design
First Increment: main and getData

When we develop programs incrementally, we begin by writing an abbreviated
main function and one subfunction. Our first program also contains the nec -
essary include statements and any global declarations that are required. It is
shown in Program 4- 13. Note that the program includes documentarx com -
ments as we understand them at this point in the development .
PROGRAM 4 - 1 3 Calculator Program — First Increment

1 /* This program adds two integers read from the
2 keyboard and prints the results.
3 Written by:
4 Date:
5 */
7
9 void getData (int* a, int * b);
10
12 {
14 int a;
15 int b;
16
17 // Statements
18 getData ( &a , & b);
19
20 printf( " **main: a = %d ; b = %d\n" , a , b);
21
22 return 0;
23 > // main
24
25 ========== getData ====
26 This function reads two integers from the keyboard.
continued
(continued )
PROGRAM 4- 13 Calculator Program -First Increment
parameters a and b
are addresses
Pre
27 parameter addresses
28 Post Data read into
29 */ , mt * *> )
30 void getData ( int * a
31 { ");
i n t e g e r numbers :
32 printf( "Please enter two
33 scant( " %d % d" , a, b );
34 %d \ n " , *a , *b ) ;
35 printf( "** getData: a = % d
; b =
36 return ;
37 } // getData
Results:
: 8 13
Please enter two integer numbers
* * getData: a = 8 ; b = 13
* * main: a = 8 ; b = 13
Program 4- 13 Analysis It is very rare that we don't make some syntactical errors
when writing even the sim-
the
plest programs. By compiling program in its simplest state, we are able to easily
correct minor errors. We have also verified that the communication between functions
is valid; that is, we verify that the variable types match the parameter types and that
the data are being properly passed down and back up .
In this first incremental program, we pass the addresses of the two input variables
to the get data function. It then reads and prints the input to verify that the data are cor-
rectly read. When we are back in main, we again print the input data to verify that it
was properly passed up to main. Because we were careful to align the print formatting,
we can easily check that the values are the same.
One point to note in the program is that we prefix temporary print statements with
two asterisks. This allows us to tell the difference between permanent output and debug-
ging output. As we move to the next increment, we remove the temporary statements
from the debugged functions.
Second Increment: add

Once the first compile has been run and verified , we are ready to write the
next function . While it is sometimes better to
ma \ \ \ e se ect the next function in the
work in a different order, nor -
program flow as shown in the struc-
art , in t is case the add function .
This incremental version of the
program is seen in Program 4- 14 .
PROGRAM 4- 14 Calculator Program- Second Increment

1 / * This
2
and p r i n t s t h e r e s u l t s .
con tinned
PROGRAM 4 - 14 Calculator Program — Second Increment ( continued )

3 Written by:
4 Date:
5 */
7
9 void getData ( int* a , int* b );
10 int add ( int a , int b);
11
13 {
15 int a;
16 int b;
17 int sum ;
18
19 // Statements
20 getData ( & a , & b );
21
22 sum = add ( a , b );
23 printf( "** main: %d + %d = %d \n " , a , b , sum );
24 return 0 ;
25 > // main
26
27 ===== getData ====
29 Pre Parameters a and b are addresses
30 Post Data read into parameter addresses
3 1 */
3 2 void getData ( int* a , int* b )
33 {
34 printf( "Please enter two integer numbers: " );
35 scanf("%d %d " , a, b);
36 return ;
3 7 } // getData
38
3 9 /* ==== === add ===
40 This function adds two numbers and returns the sum .
41 Pre a and b contain values to be added
42 Post Returns a + b
43 */
44 int add ( int a , int b )
45 {
continuec
204 Section 47 Programming Example lncrei
^ ^^
( continued )
PROGRAM 4- 14 Calculator Program-Second Increment
47 int sum;
48
49 // Statements
50 sum = a + b;
% d \ n " , a , b , sum ) ;
51 printf("**add: % d + % d =
52
53 return sum;
54 > // add
Results:
Please enter two integer numbers: 8 13
* * add : 8 + 13 = 21
* *main : 8 + 13 = 21
Program 4- 14 Analysis Once again, we include print statements to verify that the add function received the
correct data and properly returned it to main.
Final Increment: Print Results

The third and final increment includes the print results function . We do not
need any debugging statements , however, because the print results function
prints everything as a part of its requirements. The final program is shown in
Program 4-15.
PROGRAM 4- 15 Calculator Program - Final Increment

1 / * This program adds two integers read
2 from the keyboard and prints the results.
3 Written by:
4 Date:
5 */
7
9 void getData (int* a , int* b );
10 int add (int a, int b);
11 void printRes (int a' int b, int sum );
12
14 {
15 // Local
Declarations
continue
PROGRAM 4 - 1 5 Calculator Program — Final Increment ( continued )

18 int sum = 0;
19
20 / / Statements
21 getData ( & a, & b );
22
23 sum = add ( a, b );
24
25 printRes ( a , b, sum );
26 return 0 ;
27 > / / main
28
29 /*
31 Pre Parameters a and b
33 */
34 void getData ( int * a , int * b )
35 {
36 printf ( "Please enter two integer numbers: " ) ;
37 scanf( ''% d % d " , a, b );
38 return;
39 > / / getData
40
41 1
42 This function adds two integers and returns the sum.

45 */
46 int add ( int a , int b )
47 {
48 / / Local Definitions
49 int sum ;
50
51 / / Statements
52 sum = a + b ;
53 return sum ;
54 > // a d d
55
56 /*
57 Prints the calculated results.
58 Pre a and b contain input ; sum the results
59 Post Data printed
60 */
contin net
PROGRAM 4- 15 Calculator Program — Final Increment (continued )
61 void printRes (int a, int b, int sum )
62 {
63 printf( " %4d + %4d = %4d \n " / a , b, sum );
64 return ;
65 > // printRes
Results:
8 + 13 = 21

In this section we discuss three different but related aspects of software engi -
neering design: the structure chart , functional cohesion , and top-down
development .
Structure Charts
The structure chart is the primary design tool for a program . Therefore, you
should create it before you start writing your program . An analogy will help
you understand the importance of designing before you start coding.
Assume that you have decided to build a house. You will spend a lot ol
time thinking about exactly what you want . How many rooms will it need ? Do
you want a family room or a great room ? Should the laundry he inside the
house or in the garage? To make sure everyone understands what you want ,
you will prepare formal blueprints that describe everything in detail. Even if
you are building something small , like a dollhouse for a child or a toolshed
for your back yard , you will make some sketches or plans.
Figuring out what you want in your house is comparable to determining
the requirements for a large system. Drawing up a set of blueprints parallels
the structure chart in the design of a program . All require advance planning;
only the level of detail changes.
Professional programmers use the structure chart for another purpose.
In a project team environment , before programmers start writing a program ,
they must have its design reviewed . This review process is called a structured
walk- through . The review' team consists of the systems analyst responsible for
the area of the project , a representative of the user community, a system test
engineer, and one or two programmers from the project.
The design walk- through serves three purposes: First , it ensures that you
understand how your program fits into the system by communicating your
design to the team . If the design has any omissions or communication errors ,
]
the team should detect them now . If you invite programmers w ho must inter-
! ?
face with your program , you will also ensure that the interprogram communi -
cation linkages are correct.
Second , the walk- through validates your design . In creating your design ,
you will have considered several alternative approaches to writing your pro-
gram . The review team will expect to see and understand the different
designs you considered and hear why you chose the design you are proposing.
They will challenge aspects of the design and suggest approaches you may
not have considered . The result of the review will be the best possible design .
Finally, the walk- through gives the test engineer the opportunity to assess
your programs testability. Ibis in turn ensures that the final program will be
robust and as error-free as possible.
Structure Chart Rules and Symbols that we use to write a structure chart.
Figure 4-34 shows the various
Although we include all symbols
symbols
here for completeness, we w ll d scuss only ..
fh two coloredto thesymbols
Indesigning
addition
. We discuss the other symbols m Chapters 5 and 6„.
symbols, we will discuss several
rules that we follow i
our structure chart .
Function Symbol
( Figure .4.-34 ) represents ,a function
Each rectangle in a structure chart see .
C libraries are not shown . The
that we write . Functions found in the standard
name in the rectangle is the name we
will give to the function when we write
the program . It should be meaningful . The software engineering principle
known as intelligent names states that the names used in a program should be
self - documenting ; that is , they should convey their intended usage to the
reader. In general , we use intelligent names for both functions and for data
names within our program .
(a) Function
(b) Common (c) conditional (d) loop (e) conditional

function loop
T +T
( f ) exclusive or
it t:
( g) data flow ( h) flag
FIGURE 4- 34 Structure Chart Symbols
ou that we
have explained that all names should be descriptive, we will
ture
" cZriTU tbeCaUSe
*
^ M j ‘
WC W3nt t 0 concen rate on the format of a struc-
3 ) drt cu ar Program .
*
The names in Figure 4-35 iden-
*
tifv the various
tity vaHn modules for discussion .
Reading Structure Charts
consisT
( main )
According
oflbeelb W ^
to the left-right rule
’ In FiSure 4 35> Programme
'
PrOC 6SS and

first call in the program is’to initialize-
.
k
Alter initialize is complete, the program calls process. When process is

complete, the program calls endOfJob. The functions on the same level of a
structure chart are called in order from the left to the right.
Program i
Name
initialize process endOfJob
A B C
A1 A2
FIGURE 4 - 35 Structure Chart Design
The concept of top-down is demonstrated by process. When process

is called, it calls A, B , and C in turn. Function B does not start running, how-
ever, until A is finished. While A is running, it calls A1 and A 2 in turn. In
other words, all functions in a line from process to A 2 must be called before
Function B can start.
No code is contained in a structure chart. A structure chart shows only
I
the function flow through the program. It is not a block diagram or a flow -
chart. As a map of our program, the structure chart shows only the logical
.
flow of the functions Exactly how each function does its job is shown by
algorithm design ( flowchart or pseudocode). A structure chart shows the big
picture; the details are left to algorithm design.
Structure charts show only function flow; they contain no code .
Often a program contains several calls to a common function ( see

Figure 4- 34b). These calls arc usually scattered throughout the program.
The structure chart shows the call wherever it logically occurs in the pro-
gram. To identify common structures, the lower right corner ol the rectangle
contains a cross-hatch or is shaded. If the common function is complex and
contains subfunctions, these subfunctions must he shown only once. An
indication that the incomplete references contain additional structure
should he shown. This is usually done with a line below the function rectan-
gle and a cut (~) symbol. This concept is shown in Figure 4 - 36, which uses a
common function, average, in two different places in the program
. Note,
show a function connected to two calling

however, that we never gra phically
functions.
average sun average print

fun
T
GetData calculate
FIGURE 4- 36 Common Functions in a Structure Chart
It is not necessary to show data flows ( Figure 4 - 34 g) and flags ( see Fig-
ure 4 - 34 h ) , both of which represent parameters, although it may be helpful
in certain circumstances. If they are shown , inputs are on the left of the ver-
tical line and outputs are on the right . When they are included , the name of
the data or flag should also be indicated .
The structure chart rules described in this section are summarized in
Table 4 - 1 .
1. Each rectangle in a structure chart represents a function written by the

programmer. Standard C functions are not included .
2. The name in the rectangle is an intelligent name that communicates
the purpose of the function. It is the name that will be used in the cod -
ing of the function.
.
3 The function chart contains only function flow. No
code is indicated.
4. Common functions are indicated by a cross
lower right corner of the function rectangle.
- hatch or shading in the
>.
•
Common calls are shown in a structure wherever they will be found in
program , t ey contain subfunction
calls, the complete structure
need be shown only once
- .
6. Data flows and flags are optional . When used ,
they should be named .
^
andT HagU Sh Wn n the left of the vertical
3re
nous and Hags are shown on°the °
flows right . ^ °utPut
TABLE 4- 1 Structure Chart Rules
Functional Cohesion
One of the most difficult structured programming concepts for new program -
mers is knowing when and how to create a function .
Functional cohesion is a measure of how closely the processes in a
function are related . A function that does one and only one process is func -
tionally cohesive. A function that contains totally unrelated processes is coin -
cidentally cohesive. We provide a few rules here to help you write cohesive
functions. For a complete discussion of the topic , see Page -Jones. *
Before we discuss the rules, however, you should understand why the
concept is important . Following are the three primary reasons for using struc -
turally cohesive functions:
1 . Correctness : By concentrating on only one thing as you write a function ,
you will be less apt to make an error. It is much easier to get a simple task
right than a complex task .
2. Maintainability: Production programs can live for years. The better struc -
tured a program , the easier it is to change. When programs are not well
structured , making a change in one part of the program often leads to
errors in other parts.
3. Reusability: Some processes are so common that they are found in many
programs. Good programmers build libraries of these functions so that
they don ’t have to reinvent the function each time it is needed . I his not
only leads to quicker program development hut also reduces debugging
time , since the library functions have already been debugged .
f
Only One Thing

Each function shoidd do only one thing. Furthermore, all of the statements
in the function should contribute only to that one thing. For example, assume
that we are writing a program that requires the statistical measures of average
and standard deviation. The two statistical measures are obviously related , il
for no other reason than they are both measures of the same series of num -
bers. But we would not calculate both measures in one function. That would
he calculating two things, and each function should do only one thing.
One way to determine if our function is doing more than one thing is to
count the number of objects that it handles. An object in this sense is anything
that exists separately from the other elements of the function . In the previous
exam pie, the average and the standard deviation are two different objects
.
As another example , to compute the taxes for a payroll program in the
state of California , we would deal with FICA taxes, state disability insurance ,
state unemployment taxes, state withholding taxes , and federal withholding
taxes. Each of these is a different object. Our design should group
all ol these
.
3 IVleilir Page Jones, The Practical Guide to Structured Systems
- Design , 2nd ed., Chap. 6 .
( Englewood Cliffs , N .J .: Prentice Hall , 1988 ) .
i
212 Section 4.8 Softwore Engineering
it should call subfunctions
to calculate taxes, and
taxes together in a function design seen in Figure 4 - 37.
tax. This is
to calculate each individual
1
calculate
Taxes
State Federal
FICA SDI SUI Witholding Witholding
FIGURE 4 - 37 Calculate Taxes Design
In One Place
The corollary rule is that the one thing a function does should be done in
only one place. If the code for a process is scattered in several different and
unrelated parts of the program , it is very difficult to change. T herefore, all the
processing for a task should be placed in one function and , il necessary, its
subfunctions. This is the reason we created the function calculateTaxes in
Figure 4- 37.
An example of scattered code common among programmers is found in
printing reports. Suppose that we needed to write a program that , among
other things, prints a report that includes a heading, some data with a total,
and then an end -of - report message. A well -structured solution is seen in Fig-
ure 4 - 38. It is quite common , however, to find the statements for each of
these subtasks scattered in main and other parts of the program .
print
report
print print print

heading data end- of -report
FIGURE 4- 38 Design For Print Report
Testability
aoSependentlyL WrS°? tecbnitlueshould be able We to test its funC
-
"
.
next section.
^ for this the
3
in
we -s,
Let us simnlv m |
a J
" 77from« TO
separately
| * 7/ a
h program to be
"
tested
the rest of
the program.
Top-Down Development
It we have designed our program using structured programming concepts and
a structure chart , we can then proceed to implement it in a top-down fashion .
Referring again to Figure 4 - 35 , “Structure Chart Design , ” a top-down
implementation starts with the code for main only, shown in the structure
chart as ProgramName. The code for the first compile and test is shown in
Program 4 - 16.
PROGRAM 4- 16 Top-down Development Example

1 -
/* Sample of top down development using stubs.
2 Written by:
3 Date:
4 */
6
8 int initialize ( void );
9 int process ( void );
10 int endOfJob ( void );
11
13 {
14 // Statements
15 printf( " Begin program \n\n " );
16
17 initialize ();
18 process ( );
19 endOfJob ( );
20 return 0 ;
21 > // main
22
23 /* initialize
24 Stub for initialize.
25 */
26 int initialize ( void )
27 {
28 // Statements
29 printf( "In initialize: \n");
30 return 0;
31 } // initialize
32 === process ===
33 Stub for process
34 */
35 int process ( void )
continues
Example (continued )
PROGRAM 4- 16 Top-down Development
36 {
37 / / statements
printf ( " In process : \ n
");
38
39 return 0 ;
40 } / / process
41 / * ========== =====
endOfJob
42 Stub for endOfJob
43 */
44 int endOfJob (void )
45 {
46 // Statements
47 printf( "In endOfJob: \n" );
48 return 0 ;
49 > // endOfJob
Results:
Begin program
In initialize:
In process:
In endOfJob:
Program 4 - 16 Analysis Note that this program has only the first four boxes from the structure chart:
ProgramName ( main), initialize, process, and endOfJob. For each of
main' s subfunctions, all that is included is a stub . A stub is the skeleton of a function
that is called and immediately returns. Although it is a complete function, it does
nothing other than to establish and verify the linkage between the caller and itself. But
this is a very important part of testing and verifying a program. At this point the pro-
gram should be compiled, linked, and run. Chances are that you will find some minor
problems, such as missing semicolons or errors between the function declarations
and the function definitions. Before you continue with the program, you should cor-
rect these problems.
The top-down development then continues with the coding of

initialize, process , or endOfJob. Normally, you
e t to rig t , but it is not necessary to do
develop the functions
so. To develop process , you again
,tu ltspment
su unct ons , A, B , and C, and
' then test the program . This top-down
COntinues u til the complete program has coded and
" been
t

I . Several possible errors are related to passing parameters.
a. It is a compile error if the types in the function declaration and Iunc -
tion definition are incompatible. For example, the types in the follow -
ing statements are incompatible:
double divide ( int dividend , int divisor );
double divide ( float dividend , float divisor )

{
> // divide
h. It is a compile error to have a different number of actual parameters in

the function call than there are in the function declaration,
c. It is a logic error if you code the parameters in the wrong order. I heir
meaning will be inconsistent in the called program. For example, in
the following statements, the types are the same but the meaning ol
the variables is reversed:
double divide ( float dividend , float divisor );
double divide ( float divisor, float dividend )

{
> // divide
.
2 It is a compile errordefine local variables with the same identifiers as
to
formal parameters, as shown below.
double divide ( float dividend , float divisor )

{
float dividend ;
} // divide
.
3 Using a void return with a function that expects a return value or using a
with a function that expects a void return is a compile error.
return value
.
4 Each parameter’s type must he individually specified; you cannot use
multiple definitions like you can in variables. For example, the following
is a compile error because y does not have a type:
double fun ( float x , y );

5 Forgetting the *n»colon

?'
err Similarly, n »g . .' ,
the end of funeliondech ration is a eotnpj
semicolon « < he end of the header . „„ ,
f cta
definition is a compile error.
6' It is most likely a logic error
its called functions. (This is .
to call a function from within itself or one of
known as recursion , and ts correct use is
covered in Chapter 6.)
to define a function within the body of
7. It is a compile error to attempt
another function .
call without the parentheses,
8. It is a run - time error to code a function
even when the function has no parameters
.
printHello ; / / Not a c a l l
printHello ( ) ; / / A valid call
9. It is a compile error if the type of data in the return statement does not
match the function return type.
10. It is a logic error to call srand every time you call rand .
4.10 Key Terms

absolute value function definition
actual parameters function header
bi - directional communication indirection operator
called function local variables
calling function parameter passing
ceiling pass by reference
data flow pass by value
decomposition pseudorandom number series
dereferencing random number series
downward communication return
flag scaling (with random numbers)
floor scope
formal parameter list random number seed
function body stub
function call
testability
functional cohesion top-down design
function declaration
upward communication
4.11 Summary
In structured programming,
a program is divjded
Each module is designed to d ^ modu cs
,
0 a specific
task ,
Modules in C are written as
functions.
Each C program must have one and only one function called main .
A function can return only one value .
A function can he called for its returned value or for its side effect .
The function call includes the function name and the values of the actual
parameters to provide the called function with the data it needs to perform
its job .
Each actual parameter of the function is an expression . The expression
must have a value that can he evaluated at the time the function is called .
A local variable is known only in a function definition . Ihe local variables
do not take part in communication between the calling and the called
functions.
The general format for a function definition is
_
return type function name (parameter list )
{
// Statements
} _
// function name
If a function returns no value, the return type must he declared as void .

If a function has no parameters, the parameter list must he declared void .
J The actual parameters passed to a function must match in number, type ,
and order with the formal parameters in the function definition .
When a function is called , control is passed to the called function . The
calling function " rests" until the called Function finishes its job.
We highly recommend that every function have a return statement . A
return statement is required if the return type is anything other than void . i
Control returns to the caller when the return statement is encountered .

A function declaration requires only the return type ol the function , the
function name, and the number, types , and order of the formal parame -
ters. Parameter identifiers may be added for documentation hut arc not
required .
The scope of a parameter is the block following the header.
A local variable is a variable declared inside a block . I he scope ol a local
variable is the block in which it is declared .
Structure charts are used to design a program .
Structure charts contain no code.
Functional cohesion is a measure of how closely the processes in a func -
tion are related.
IJ Programs should he written using a top-down implementation with stubs
for functions not yet coded .
4.12 Practice Sets

Review Questions
structured programming dictate
principles of top-down design and mam module and its
1 ' The a related
that a program should be divided
into
modules.
.
a True
b. False
2. The function definition contains the
code for a function .
a. True
b. False
as a part of an expression.
3. Function calls that return void may not be used
.
a True
b . False
.
4 The address operator ( s ) is used to tell the compiler to store data at an
address.
.
a True
b. False
5. Variables defined within a block have global scope .
a. True
b. False
—
6. The process of dividing a program into functions which in turn are
divided into functions until they consist of only elementary processing
that is intrinsically understood and cannot be further subdivided is
known as
—
a. charting
b. flow charting
c. factoring
.
d programming
e. structuring
/ . Which of the following statements about function declaration and defini-
tion is true?
a . The function call is found in
the called function .
b. The function declaration
c. 1 he function definition is
requires that the parameters be named.
done with a function declaration .
d. 1 he function
definition contains executable statements that perform
the function ’s task.
e. The function definition
header concludes with a semicolon ( ; ).
.
8 Which of the following is not a part of a function header ?
.
a name
b. parameter list
.
c return type
d. title
9. Which of the following statements about function parameters is true?
a. Empty parameter lists are declared with the keyword void .
b. If there is only one parameter, the function list parentheses are not
required.
.
c In the definition of a function, the parameters are known as actual
parameters.
.
d Parameters are separated by semicolons.
.
e The parameters in a function definition are defined in the function’s
body ( local declaration section ).
10 . Which of the following statements about local variables is false?
a A .local variable’s value may be returned through a return statement .
.
b Local variables are defined inside a function.
c. Local variables cannot he referenced through their identifiers outside
the function.
.
d Local variables may be initialized with an initializer.
.
e Local variables’ names can he the same as the function s para-
’
meter names .
1 1 To tell the compiler to store data at an address, use the
.
.
address operator ( & )
a
.
b array operator ( [ ] )
.
c dereference operator ( # )
d. indirection operator ( * )
.
e pointer operator ( )
A
.
12 The function that returns the absolute value of a long integer is
.
a abs
b. dabs
.
c fabs
d. labs
.
e tabs
13. Which of the following statements will generate a random number in the
range 30-50 ?
.
a rand ( 33 )
.
b ( rand ( ) % 20 ) + 1
.
c ( rand ( ) % 21 ) + 20
.
d ( rand ( ) % 21 ) + 30
.
e ( rand ( ) % 51 ) + 1
about structure charts is false?
14. Which of the following statements
replacement for flowcharts.
a . Structure charts are a design tool for a program .
b. Structure charts are the primary
in a structured walk- through to validate the
c. Structure charts are used
design.
to assess the testability of a program.
d. Structure charts can be used before you start writing a program
e. Structure charts should be
created
Exercises
function definition :
1 5. Find any errors in the following
void fun ( int x, int y )

{
int z;
return z;
> // fun
16. Find any errors in the following function definition :
int fun ( int x, y )

{
int z;
return z;
> // fun
17. Find any errors in the following function definition:
int fun ( int x, int y )

{
int sun ( int t )
return ( t + 3 );
}
return z;
} / / fun
1 S. Find any errors in the
following function definition :
void fun ( int, x )

{
1
continuei
return ;
> // fun
.
19 Find any errors in the following function declarations:
a. i n t sun ( i n t x , y ) ;
b. i n t sun ( i n t x , i n t y )
c. void sun ( v o i d , v o i d ) ;
d . void sun ( x i n t , y f l o a t ) ;
.
20 Find any errors in the following function calls:
a . void f u n ( ) ;
b. fun ( void ) ;
c. void f u n ( i n t x , i n t y ) ;
d . fun ( );
2 I . Evaluate the value of the following expressions:

a. fabs ( 9 . 5 )
b. f a b s ( - 2 . 4 )
c. f a b s ( - 3 . 4 )
d. fabs ( -7 )
e. f a b s (7)
22. Evaluate the value of the following expressions:

a . floor ( 9.5 )
b. f l o o r ( -2 . 4 )
.
c floor ( -3 . 4 )
d. c e i l ( 9.5 )
e. c e i l ( -2 . 4 )
f. ceil ( -3 . 4 )
23. Evaluate the value of the following expressions when x is 3.5 , 3.45 , 3.76 ,
3.234 , and 3.4567 :
a. f l o o r ( x * 10 + 0 . 5 ) / 10
b. f l o o r ( x * 100 + 0 . 5 ) / 100
c. f l o o r ( x * 1000 + 0 . 5 ) / 1000
.
24 Define the range of the random numbers generated by the following
expressions:
a . rand ( ) % 10
b.rand ( ) % 4
.
c rand ( ) % 10 + 1
.
d rand ( ) % 52
.
e rand ( ) % 2 + 1
.
f rand ( ) % 52 5 -
data?
Jill
5 as
25. What would he printed from Program 4- 17 when run using 3
25
PROGRAM 4-17 Program for Exercise
1 # include <stdio
.h>
2
, int y );
4 int strange ( int x
5
6 int main (void )
7 {
9 int a;
10 int b;
11 r; int
12 s; int
13
14 // Statements
15 scant("%d %d ", & a, & b );
16 r = strange ( a, b );
17 s = strange ( b, a );
18 printf( "%d %d ", r, s );
19 return 0;
20 > // main
21 / / === strange =====
22 int strange (int x , int y)
23 {
24 // Statements
25 return (x
// strange
- y);
26 >
26. What would be printed Irom Program 4- 18 when run using 3 5 4 6 as data?
PROGRAM 4- 18 Program for Exercise 26

1 # include <stdio h> .
2
4 int strange (int x, int y );
5
6 int main (void )
7 {
9 int a ;
10 int b
11 int c;
12 int d;
13 int r;
continued
1
PROGRAM 4- 1 8 Program for Exercise 26 (continued )
14 int s;
15 int t;
16 int u;
17 int v;
18 i
19 // Statements
20 scanf( " % d %d %d %d " , & a , & b , &c , &d );
21
22 r = strange ( a , b );
23 s = strange ( r , c );
24 t = strange ( strange ( s, d ), strange ( 4 , 2 ));
25 u = strange (t + 3 , s + 2);
26 v = strange ( strange (strange ( u , a ), b ), c );
27
28 printf( " % d %d %d %d %d " , r, s, t , u, v);
29 return 0;
30 > // main
31 // strange
32 int strange ( int x , int y )
33 {
35 int t ;
36 int z;
37
38 // Statements
39 t = x + y?
40 z = x * y;
41 return ( t + z);
42 > // strange
.
27 Draw the structure chart for Program 4 - 19. What output does it produce ?
PROGRAM 4- 19 Program for Exercise 27

2
4 int funA ( int x );
5 void funB ( int x );
6
7 int main ( void )
8 {
10 int a ;
continuec
PROGRAM 4- 19 Program for Exercise 27 ( continued

)
11 int b;
12 int c;
13
14 // Statements
15 a = 10;
16 funB (a);
17 b = 5;
18 c = funA ( b);
19 funB(c );
20 printf("%3d %3d %3d'\ a, b, c );
21 return 0;
22 } // main
23
24 int funA (int x )
25 {
26 // Statements
27 return x * x;
28 > // funA
29
30 void funB (int x)
31 {
33 int y ;
34
35 // Statements
36 y = x % 2;
37 x /= 2;
38 printf ("\n%3d %3d\n" , x , y);
39 return ;
40 > // funB
Problems
28. Write a Junction to print your name, as shown below.
Write a call as it
would be coded in a calling function, such as w a i n .
*************** i<'
i'
1'iii i i ii i i +
'''''''' **
*
*
* Your Name Here *
*
*
**************** ***
***********
29. Write a program that generates a random number from the following set :
1, 2, 3, 4, 5, 6
30. Write a program that generates a random number from the following set :
1, 4 , 7 , 10, 13, 16
31. Explain what is meant by the statement “ a function should do only

one thing."
32. Code and run Program 4 - 16 , “Top-down Development Example, ” to

demonstrate how stubs work .
.
33 Write a function to convert inches into centimeters. ( One inch is 2.54
centimeters.) Then write a program that prompts the user to input a
measure in inches, calls the conversion function, and prints out the mea -
surement in centimeters .
34 . Write a program that reads three integers and then prints them in the
order read and reversed . Use four functions: main , one to read the data ,
one to print them in the order read , and one to print them reversed .
35. Expand the calculator program , Program 4 - 15 , to calculate the differ-

ence, product , quotient , and modulus of the numbers. Calculate the
quotient and modulus in one function.
36. Modify Program 4 - 5 , “ Add Two Digits," to add the least significant three
digits ( hundreds, tens, and ones) .
37. Write a function that receives a positive floating- point number and
rounds it to two decimal places. For example, 127.565031 rounds to
127.570000. Hint: To round, you must convert the floating- point num -
ber to an integer and then hack to a floating point number. Print the
-
rounded numbers to six decimal places. lest the function with the fol -
lowing data:
123.456789 123.499999 123.500001
38 Write a program that reads a floating-point number and prints the

. ceil -
ing, floor and
, rounded value . Use the function in Problem 37 for the
average and test it with the same values as Problem 37 .
39 Write a function to compute the perimeter and area of a

. right triangle
length of the two sides ( a and b ).
( Figure 4-39) when given the
c
a
FIGURE 4- 39 Problem 39
The following formulas may be helpful:
c2
area =
Projects
40. Prepare a payroll earnings statement For the sales Force at the Arctic Ice
Company. All of Arctics employees are on a straight commission basis oi
12.5% of sales. Each month , they also receive a bonus that varies
depending on the profit For the month and theii length of service. The
sales manager calculates the bonus separately and enters it with the
salesperson’s total sales for the month. Your program is also to calculate
the withholding taxes and retirement lor the month based on the lollow -
ing rates:
a . 25% Federal withholding
h . 10% State withholding
c . 8% Retirement plan
Use the test data in Table 4 - 2 to test the program .
Salesperson Sales Bonus

1 53,500 425
2 41,300 300
3 56,800 350
4 36,200 175
TABLE 4 - 2 Test Data For Project 40
.
41 \ \ rite a program that , given a
beginning balance in your savings account ,
calculates the balance at the end of 1 year. The
interest is 5.3% com -
pounded quarterly. Show the interest
earned and balance at the end of
each quarter. Present the data in tabular
columns with appropriate head-
ings. Use separate functions to
compute the interest and print the balance.
T
!
42. I he formul a for convert ing centigr ade temper atures to fahren heit is
180.0
F = 32 + C
100.0
Write a program that asks the user to enter a tempera ture readin g
in centigr ade and then prints the equiva lent Fahren heit value. It then
asks the user to enter a Fahren heit value and prints out the equiva lent
centigr ade value. Run the progra m severa l times . Be sure to include at
least one negativ e tempera ture readin g in your test cases. Provid e sepa -
rate functio ns as needed by your design. One possib le design is shown
in Figure 4 - 40 . ( Your main functi on should have only functio n calls
.)
temperatures
Celsius
i Fahrenheit
l
get
Celsius i Celsius to
Fahrenheit i print
Fahrenheit i get
Fahrenheit i Fahrenheit
to Celsius i print
Celsius i
FIGURE 4 - 40 A Possible Design for Project 44
may he written
43 Write a program that uses standar d functio ns. The program
.
entirel y in main and must follow the pseudo code shown in Algori thm 4- 1 .
riate caption s , and align the data .
Give the output approp
ALGORITHM 4 - 1 Pseudocode for Project 43

1 Prompt the user to enter a number
2 Read number
3 Display number
3...37
4 Get a random number and scale to range
5 Display random number
6 Set product to number * random number
7 Display product
8 Display ceiling of random number
9 Display floor of product
random number
10 Display number raised to power of
Displa y square root of random number
11
customers’ bills for a carpet company

44 Write a C program that creates
. is given:
when the following information
a. the length and the wiidth of
the carpet in feet
b. the carpet price per square foot
c. the percent of discount for each
customer
square foot . It is to be defined as
The labor cost is fixed at $0.35 per
a constant . The tax rate is 8.5%
applied after the discount Us also to be
defined as a constant . The input data consist to be carpeted the per
room
^
of a set of three integers
-
representing the length and width of the
owner to a customer , and a real number
centage of the discount the gives
representing the unit price of the carpet
The program is to prompt the
below . ( Colored numbers are typical
user for this input as shown
responses. )
Length of room (feet)? 30

Width of room ( feet )? 18
Customer discount (percent)? 9
Cost per square foot ( xxx.xx)? 8.23
The output is shown below. Be careful to align the decimal points.
MEASUREMENT
Length XXX ft
Width XXX ft
Area XXX square ft
CHARGES
DESCRIPTION COST/SQ.FT. CHARGE
Carpet XXX.XX $XXXX.XX

Labor 0.35 XXXX.XX
INSTALLED PRICE $ xxxx . xx

Discount xx % XXXX.XX
SUBTOTAL $ xxxx . xx
Tax XXXX.XX
TOTAL $ xxxx . xx
I he program's design should use main and at least the six functions
described below:
n
a. Read data from the keyboard. This function is to use addresses to read
all data and place them in the calling function’s variables.
b. Calculate values. This function calls three subfunctions. Each func -
tion is to use addresses to store their results.
Calculate the installed price. This function calculates area, carpet
cost , labor cost, and installed price. The installed price is the cost
of the carpet and the cost of the labor.
Calculate the subtotal. This function calculates the discount and
subtotal.
Calculate the total price with discount and tax. This function calcu -
lates the tax and the total price.
c . Print the result. Use two subfunctions to print the results: one to print
the measurements, and one to print the charges.
lest your program with the test data shown in Table 4 - 3.
Test Length Width Discount Cost
1 23 13 12 $ 14.20
2 35 8 0 $ 8.00
3 14 11 10 $22.25
TABLE 4 - 3 Test Data for Project 44

—
Selection — Making Decisions !
In Chapter Is “ Software Engineering” section, we mentioned that Dijkstra

proposed that any program could he written with three constructs: sequence,
selection, and loop. The C language implementation of the first construct,
.
sequence, is covered in the previous chapters In this chapter, we turn our
attention to the second construct, selection. Selection allows us to choose
between two or more alternatives: It allows us to make decisions.
.
What a dull world it would he if we didn’t have any choices Vanilla ice
cream for everybody! Uniforms all around! And no debates or arguments to
keep things interesting Fortunately, our world is filled with choices. And
.
since our programs must reflect the world in which they are designed to oper -
ate, they, too, are filled with choices and opportunities lor decision making
.
I low are decisions made by a computer? In this chapter, you will find out .
One of the main points to keep in mind is that the decisions made by a com-
puter must be very simple, since everything in the computer ultimately
.
reduces to either true or false If complex decisions are required, it is the pro-
grammer ’s job to reduce them to a series of simple decisions that
the com -
puter can handle .
Objectives
To understand how decisions are made in a computer
To understand the logical operators: and, or, and not
To understand how a C program evaluates a logical expression
To write programs using logical and comparative operators
To write programs that use two - way selection: if...else statements
To write programs that use multi - way selection: switch and else. .. if
To understand Cs classification of characters
To write programs that use Cs character functions
To be able to design a structure chart that specifies function selection
231
232 Section 5.1 Logical Data and Operators
5.1 Logical Data and Operators .

if it conveys the tdea of true false We need
or
A piece of data is called logical . In real life , logical data
logical data in real life as well
as in programming
to a question that needs a yes - no answer.
( true or false ) are created in answer
if an item is on sale or not. We ask if a business is open or
For example , we ask
is a piece of data that is usually yes or no.
not . The answer to these questions
"Is x greater than y? The answer is again
We can also ask questions such as
or no. In computer science , we
do not use yes or no , we use true or false.
yes
Logical Data in C
Traditionally, C had no logical data type C
; programmers used other data
data . If data y’alue is zero, it is consid
types , such as hit , to represent logical
3 *
ered false. If it is nonzero, it is considered true . This concept of true and false
on a numeric scale is seen in Figure 5 - 1.
false
true
J true
FIGURE 5- 1 True and False on the Arithmetic Scale
The C 99 standard introduced the Boolean data type, _bool , which is

declared as an unsigned integer in the stdbool .h header file . Also defined in
the header files are the identifiers bool , true , and false. While the standard
still supports the traditional use of integers for logical data , we recommend
that the Boolean type, bool , be used .
Logical Operators
t has three logical operators for combining logical values and creating ..new .
logical values: not , and , and or. A common way to show' logical relationships is in
trill ta es '
^rutJ ^
1 ta es l st he values that each operand can assume and the
* *
resulting value. The truth tables for logical operators are shown in Figure 5- 2.
|
not Operator
^ e 15 in the Prece-
^ hT 01^
denct T
^
A r - operator vvit
^ precedenc
tme value to false and
false value to tme ' ' ^
3 3
Chapter 5 Selection — Making Decisions 233
not and or
x y xlly
false true false false false false false false
true false false true false false true true
true false false true false true
true true true true true true i
FIGURE 5 - 2 Logical Operators Truth Table
and Operator
The and operator ( & & ) is a binary operator with a precedence of 5 . Since and
is abinary operator, four distinct combinations of values in its operands are
possible. The result is true only when both operands are true ; it is false in all
other cases.
or Operator
The or operator ( | | ) is a binary operator with a precedence of 4. Again, since
it is a binary operator, four distinct combinations of values in its operands are
possible. The result is false if both operands are false ; it is true in all
other cases.
5j!
Evaluating Logical Expressions

Computer languages can use two methods to evaluate the binary' logical rela -
tionships. In the first method, the expression must be completely evaluated
before the result is determined. This means that the and expression must be
completely evaluated, even when the first operand is false and it is therefore
known that the result must be false . Likewise, in the or expression, the whole
expression must be evaluated even when the first operand is true and the
obvious result of the expression must be true .
The second method sets the resulting value as soon as it is known. It does
not need to complete the evaluation. In other words, it operates
in a short -
circuit fashion and stops the evaluation when it knows for sure what the final
result will be. Under this method if , the first operand of a logical and expres -
sion is false , the second half of the expression is not evaluated because it is
that the result must be false . Again , with the or expression, if the
apparent
first operand is true , there is no need to evaluate the second half ol the
expression, so the resulting value is set
true immediately. C uses this short -
circuit method, which is graphically shown in Figure 5 - 3.
7
true II ( anything)
4 4
true
false
FIGURE 5 -3 Short-circuit Methods for and / or
Although the C method is more efficient , it can

cause problems when
the
poor programming practice.
second operand contains side effects, which is
expression in which a programmer wants
Consider for example, the following i
i_
and at the same time wants to
to find the value of the logical expression
increment the value of the second operand :
x && y ++
Everything works fine when the first operand is true . But if the first oper-
and is false , the second operand will never be evaluated and therefore will
never be incremented. It is the same with the following or example. If the
first operand is true , the second operand will never be incremented .
x || y + +
Program 5 - 1 demonstrates the use of logical data in expressions.
PROGRAM 5 - 1 Logical Expressions

1 I /* Demonstrate the results of logical operators.
2 Written by:
3 Date:
4 */
6 #include <stdbool.h>
7
8 int main ( void )
9 {
11 bool a = true;
12 bool b = true;
13 bool c = false;
14
15 // Statements
continued
l
A
PROGRAM 5 - 1 Logical Expressions ( continued )

16 printf( " %2d AND % 2d: % 2d \n " , a, b, a && b );
17 printf(" %2d AND %2d: %2d\n" , a, c, a && c);
18 printf( " % 2d AND %2d: % 2d \n " , c, a, c && a );
19 printf( " %2d OR %2d: %2d\n" , a, c, a *
20 printf( " % 2d OR %2d: %2d \n " , c, a, c
21 printf( " %2d OR %2d: %2d\n" , c, c, c
22 printf( " NOT %2d AND NOT % 2d: %2d \n " , a, c , !a && !c );
23 printf( " NOT % 2d AND %2d: %2d\n " , a, c , !a && c);
24 printf( " % 2d AND NOT % 2d: % 2d \n " , a, c, a && !c );
25 return 0 ;
26 } // main
Results:
1 AND Is 1
1 AND 0: 0
0 AND Is 0
1 OR 0: 1
0 OR Is 1
0 OR 0: 0
NOT 1 AND NOT 0: 0
NOT 1 AND 0: 0
1 AND NOT 0: 1
Program 5 - 1 Analysis Each print statement in Program 5- 1 contains a logical expression that evaluates
either to true or false. Make sure you understand why each of the expressions evalu-
ates as shown in the results . Note that even though the results are of type Boolean
,
they must be printed as integer d .
(% ) This is because while the expressions evaluate to
true or false, there is no conversion code for Boolean. Therefore, when we print
them,
we must use the integer field specification .
Comparative Operators
In addition to logical operators, C provides six comparative operators. Com
-
parative operators are divided into two categories, relational

and equality .
They are all binary operators that accept two operands and compare them .
always true false , fhe
The result is a Boolean type— that , the is result is or
operators are shown in Figure 5 - 4 .

The relational operators—less than, less than or equal, greater than
,
equal — have a higher priority ( 10 ) in the Precedenc e Table
greater than or
( 9 ) . This
( see inside front cover ) than the equality and not equal operators
means that relational operators are evaluated before the equality operators
when they appear together in the same expression.
It is important to recognize that each operator is a complement
ot another
. But , surprising ly, the complem ent is not the one that
operator in the group
you might expect . Figure 5 - 5 shows each operator
and its complement.
t
Meaning Precedence
Type Operator
< less than
<= less than or equal 10
Relational > greater than
>= greater than or equal
== equal 9
Equality not equal
FIGURE 5- 4 Relational and Equality Operators
complement >=
< 4 r
complement r
> 4
_ complement
t
\ -
FIGURE 5 - 5 Comparative Operator Complements
II we want to simplify an expression involving the not and the less than
operator, we use the greater than or equal operator. This concept is impor -
tant lor simplifying expressions and coding expressions in good, clear
style. Table 5 - 1 shows an example of each expression and its simplified
version.
Original Expression Simplified Expression

!(X < y) x >= y
!( x > y) x <= y
!( x 1= y) x == y
! ( x <= y ) x > y
! ( X >= y )
x < y
I ( X == y )
x 1= y
TABLE 5 - 1 Examples of Simplifying Operator

Complements
Program 5 - 2 demonstrates the use of
the comparative operators.
PROGRAM 5 - 2 Comparative Operators

1 /* Demonstrates the results of relational operators.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 <
10 int a = 5;
11
12
int b = 3; -
13 // Statements *
14 printf( " %2d < % 2d is % 2d \ n " , a, b, a < b)
15 printf( " %2d == %2d is % 2d \ n ” , a, b, a == b )
16 printf( " % 2d != % 2d is % 2d \n " , a, b, a != b )
17 printf( " %2d > %2d is % 2d \ n " , a, b, a > b)
18 printf( " % 2d <= % 2d is % 2d \n " , a, b, a <= b )
19 printf( " % 2d >= % 2d is % 2d \n " , a, b, a >= b )
20 return 0;
21 > // main
Results:
5 < -
3 is 0
-
5 == 3 is 0
-
5 != 3 is 1
5 > --
3 is 1
5 <= 3 is 0
-
5 >= 3 is 1
, make
Program 5 - 2 Analysis Program 5 - 2 follows the same patterns we saw in Program 5 1 . Once again
-
sure that you understand why each of the representati ons in the results evaluates to
true or false .
5.2 Two- Way Selection

. I he
The basic decision statement in the computer is the two- way selection
al that can he
decision is described to the computer as a condition statement
the true , one or more action state -
answered either true or false . II answer is
false , then different action or set ol
ments are executed . IT the answer is a
is executed, the program
actions is executed . Regardless of which set of actions
continues with the next statement after the selection
. The flowchart for two-
way decision logic is shown in Figure 5
- 6 .
238 Section 5.2 Two-Way Selection
false decisionX true

conditionJ
false action true action
FIGURE 5 - 6 Two- way Decision Logic
if . . . else
C implements two-way selection with the if . . . else statement. An if . . . else state-
ment is a composite statement used to make a decision between two alterna -
tives. Figure 5-7 shows the logic flow for an if . . else statement . The expression
can be any C expression. After it has been evaluated, if its value is true ,
statement 1 is executed: otherwise, statement 2 is executed. It is impossible
for both statements to be executed in the same evaluation.
false true
(zero) (Izero) if (expression )
expression
statement 1
else
statement2
i statementl
i statement2
«#
(a) Logical Flow (b) Code
FIGURE 5 - 7 if .. .else Logic Flow
I here are some syntactical

.
statements The points,, „sumZSdlS
1. t he expression must be enclosed in parentheses.

.
2. No semicolon ( ; ) is needed for an if . . else statement ; statement 1 and
statement 2 may have a semicolon as required by their types.
.
3 The expression can have a side effect .
4. Both the true and the false statements can he any statement ( even
.
another if . . else statement ) or they can he a null statement .
5. Both statement 1 and statement 2 must he one and only one statement.
Remember, however, that multiple statements can he combined into a
compound statement through the use of braces.
6. We can swap the position of statement 1 and statement 2 if we use the
complement of the original expression .
TABLE 5 - 2 Syntactical Rules for if . . . else Statements
The first rule in Table 5 - 2 —that the expression must he enclosed in

parentheses— is simple and needs no further discussion . The second rule is
also simple, hut it tends to cause more problems. We have therefore pro -
vided an example in Figure 5 - 8. In this example, each action is a single state -
ment that either adds or subtracts 1 from the variable a . Note that the
semicolons belong to the arithmetic statements, not the if ...else statement .
if ( i
a ++ ;
— 3) The semicolons |
belong to the
expression statements, I
else not to the
if ... else statement
a — ;
FIGURE 5 - 8 A Simple if . .. else Statement
The third rule in Table 5 - 2 requires more discussion . It is quite common

in C to code expressions that have side effects. For example
, we often use
expressions that read data as a side effect . As an example , consider what hap -
pens when we are writing a series of numbers, and when have
we written ten
a line
numbers we want to go to a new line. A simple solution increments
count and tests the limit in the same statement . The
code for this logic could
he written as shown in the following example .
240 Section 5.2 Two-Wayj>elecfion
if (++lineCnt > 10)

{
printf("\n");
lineCnt = 0 ;
} // end true
printf( ... );
. The fact that any state-
Rules 4 and 5 in Table 5-2 are closely related
straightforward , but often new C program -
ment can be used in an if ...else is
compound statement for complex logic. The use of
mers will forget to use a
compound statements is seen in Figure 5 - 9. The first example shows a com -
pound statement only for the true condition . The second example shows
compound statements for both conditions. Note that the compound state-
ments begin with an open brace and end with a close brace
.
if '( j != 3) if (j != 5 && d == 2)
r< b++;
{
j++;
printf("%d\ b);
} // if Compound
;
d
—
printf("%d%d " , j, d )
else statements } // if
printf( " %d \ j ); are treated as else
one statement r< D --;
X - d++;
printf( ” %d%d " , j, d )
i } // else
FIGURE 5- 9 Compound Statements in an if...else
I he sixth rule states that the true and false statements can he exchanged
hy complementing the expression . Recall from our discussion of relational
operators on Comparative Operators in Section 5.1 that any expression can
he complemented . When we find that we need to complement an if . ..else
statement, all we have to do is to switch the true and false statements. An
example ol this operation is shown in Figure 5 - 10.
Null else Statement

Although there are always two possible
we d „ not care about both of them .
actions after a decision , sometimes
In this ease, the else action is usually left
Hnwpv
‘lml’ c’ assunu vve are processing numbers as they are being read .
However for some reason , the logic requires that we process only numbers
greater than zero. As we read a
number, we test it lor greater than zero. II the
test is true , we include it in
the process. II it is false, we do nothing.
These two statements are the same

because the expressions are the
complements of each other!
if (expression)
i\
else else
(a) Original (b) Complemented
FIGURE 5 - 10 Complemented if...else Statements
— —
If the else condition is not required that is, if it is null it can be omit -
ted . This omission can he shown as a null else statement ( a null statement
consists of only a semicolon ) ; more commonly, the else statement is simply
omitted entirely, as shown in Figure 5 - 11 .
if ( expression)
{
\ if ( expression )
{
N
} // if } // if
else
/
FIGURE 5 - 11 A Null else Statement
De Morgans Rule
When we design the logical flow of a program, we often have a situation in which the
not operator is in front of a logical expression enclosed in the parentheses
. Human
engineering studies tell us, however , that positive logic is easier to read and under -
stand than negative logic. In these cases, therefore , if we want to make the total
expression positive by removing the parentheses, we apply the not operator in this
directly
to each operand. De Morgan 's rule governs the complemen ting of operators
situation. This rule is defined as follows:
-
When we remove the parentheses in a logical expression preceded by the not opera
tor, we must apply the not operator to each expression
while changing the logical
or ( | | ) to and ( & & ).
operators — that is, changing and ( & & ) to or ( | | ) and changing
continued
t
Consider the expression shown below.

i ( x && y ) ix || iy
, (x || y ) -> ix && iy
the result of the first

^
forx nd y U is important to recognize, however,
ETo^&fiol the C rules of precedence can affect De Morgan's rule.
,
, but the true branch cannot be

It is possible to omit the false branch
_ coded as a null statement ; normally, however, we do not
omitted . It can be
use null in the true branch of an if ...else statement
.
. To eliminate it we can
use Rule 6 in Table 5- 2 , “ Syntactical Rules for if . . .else Statements," which
allows us to complement the expression and swap the two statements The
.
result is known as a complemented if ...else and is shown in Figure 5 - 12 .
if (expression) if ( lexpression ) if (!expression)

{ {
else
{ } // if } // if
* else *
} // elseQ
Null
Statement
FIGURE 5- 12 A Null i f Statement
Program 5 - 3 contains an example of a simple two- way selection. It dis-

plays the relationship between two numbers read from the keyboard .
PROGRAM 5-3 Two-way Selection

1 /* Two-way selection.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
i ; continued
PROGRAM 5 - 3 Two - way Selection (continued )

10 int a ;
11 int b;
12
13 // Statements
14 printf( "Please enter two integers: " );
15 scant ( " %d %d " , &a, & b );
16
17 if ( a <= b )
18 printf( " % d < = %d \n " , a , b );
19 else
20 printf( " %d > %d \n " , a , b );
21
22 return 0;
23 > // main
Results:
Please enter two integers: 10 15
10 <= 15
Nested if Statements
As we stated previously, for the if . ..else , the statements may be any statement,
including another if . ..else . When an if .. .else is included within an if . . else , it
is known as a nested if statement . Figure 5 - 1 3 shows a nested ij statement.
There is no limit to how many levels can be nested, but if there are more than
three, they can become difficult to read.
expression
1 if ( expression 1 )
if ( expression 2 )
statement 31 expression statement 1

2
else
statement 2
statement 2 | statement 1
J else
statement 3
( a ) Logic flow (b ) Code
FIGURE 5 - 13 Nested if Statements

. It uses a nested if state-
modification of Program 5 - 3 than b.
Program 5 -4 is a , greater
ment to determine if a is less
than, equal to or
PROGRAM 5- 4 Nested if Statements

.
1 /*
Written by:
-
Nested if in two way selection
2
3 Date:
4 */
6
7 int main ( void )
8 {
10 int a;
11 int b;
12
13 // Statements
14 printf( "Please enter two integers: ” )?
15 scanf ("%d%d", &a, &b);
16
17 if (a <= b)
18 if (a < b)
19 printf("%d < %d\n" , a, b);
20 else
21 printf( "%d == %d \n" , a, b);
22 else
23 printf("%d > %d \n" , a , b);
24
25 return 0;
26 I} // main
Results:
Please enter two integers: 10 10
10 == 10
Program 5 - 4 Analysis You should be able to follow this simple program with relative ease . However, it does
contain a subtle software engineering principle.
^ '^
greater than condition to the outer else (see
, Study the if statements ( 17 and 18)
, er 6 j- ° We C[ eC or 9rea er than? The answer is that we default the
,
Statement 22). When coding a two-way
" ' t T11 *
Y
^
statements, code the most probable
Dangling else Problem

probable co di i
Tfirst and the least probable
" "° ast with nested selection
is no matching else for every if . C ’s solution to this problem is a simple

rule: Alxvuys pair an else to the most recent unpaired if in the current block .
I his rule may result in some if statement ’s being left unpaired. Since such
an arbitrary rule often does not match our intent, we must take care to
.
ensure that the resulting code is what we intended Take, for instance, the
example shown in Figure 5 - 14. From the code alignment , we conclude
that the programmer intended the else statement to he paired with the first
if . However, the compiler will pair it with the second i/, as shown in the
flowchart .
else is always paired with the most recent unpaired if.
expression
if (expression!) 1
if ( expression2 )
expression
2
statement 1
else
statement 2 | statement 1
statement 2
Compiler pairs
'
this if and else
( a) Code
T ( b ) Logic Flow
FIGURE 5 - 1 4 Dangling else
Figure 5 - 15 shows a solution to the dangling else problem, a compound

statement. In the compound statement, we simply enclose the true actions
in
braces to make the second if a compound statement . Since the closing brace
completes the body ol the compound statement, the if statement is also
closed from further consideration and the else is automatically paired with
the correct if .
Simplifying if Statements
By now you should recognize that if ..else statements can become quite
com-
plex. This discussion gives you some ideas on simplifying if statements.

Usually, the purpose of simplification is to provide more readable code.
Fhe block closes

expression .the if statement
1 if (expression 1 )
expression
if (expression 2 )
2 statement 1
statement 2 > // if
else
statement 1
statement 2
T
( a) Logic Flow (b) Code
FIGURE 5 - 15 Dangling else Solution
Sometimes the control expression itsell can be simplified . For example,

the two statements in Table 5- 3 are exactly the same. The simplified state-
ments, however, are much preferred by experienced C programmers. W hen
the simplified code becomes a natural way of thinking, you have begun to
internalize the C concepts; that is, you are beginning to think in C!
Original Statement Simplified Statement

if (a != 0) if (a)
statement statement
if (a = = 0) if (!a)
statement statement
TABLE 5 - 3 Simplifying the Condition
Since the simplified statements in Table 5 - 3 are new, let’s look at them a
little more carefully. The expression a ! = 0 evaluates to either true or false. If
a is anything other than zero, then the
expression is true and statements is
executed . However, any integer can he used to
represent true or false . In this
ease , it a contains any value other
than zero, it is true ; otherwise, it is false.
et igure 1 . ) Therefore , since we want to
a ,s not zero, and since anything
execute statement 2 whenever
( ) t at is , as a is true .
other than zero is true , we code the expres -
Similarly, if we want to test for a equal to
z-ero, we stmply complement the
expression , making it ! a .
Conditional Expressions
( provides a convenient alternative to the traditional if ..else for two- way
selection—the ternary' conditional expression has a precedence of 3 ( see
inside front cover ).
I he conditional expression has three operands and a two- token opera -
tor. Each operand is an expression. The first token, a question mark ( ? ) , sepa -
rates the first two expressions. The second token, a colon ( : ), separates the
last two expressions. This gives it the following format:
expression ? expression!: expression2
To evaluate this expression, C first evaluates the leftmost expression. If

the expression is true, then the value of the conditional expression is the
value of expression! . If the expression is false , then the value of the condi -
tional expression is the value of expression2.
Let s look at an example.
a == b ? C++ : d++;
.
In this expression, only one of the two side effects take place II a is equal
to b, C++ is evaluated and one is added to c; expression2 is ignored On the .
other hand, if a is not equal to b, then d++ is evaluated and 1 is added to d ;
expression!is ignored If this sounds much like a simplified if . else , its
. .
because it is! Figure 5 - 16 shows the flowchart for the expression, which
could easily he coded as an if ...else.
a == b ? C++ : d++ ;
(a) Logic Flow (b) Code
FIGURE 5- 16 Conditional Expression
The conditional expression, like any other expression, can he used in an

write either
assignment expression. Suppose that we have a program that can
to a printer or to the system monitor. When we write to the monitor, we can
i
line. When we write to

the printer, we can write
, i
number «
.
tn
write ten a variable that indicates either the
is
fiieFlag
,, , ,, „ ,
5
STp«u bTr re .. nol

h number per line a how bd
M' ? 10 : 15 );
, ,
necessary but make he Women more readable.)
numPerLine = (fiieFlag
nest conditional expressions, it is not
One final note: Although we can
complex, remember the MSS
recommended . When the logic begins to get
principle and use nested if statements .
Two-Way Selection Example

To demonstrate two-way selection , let s look at a program
' that calculates
tax brackets may be helpful.
income taxes. A brief explanation of progressive
In a progressive bracket system , the higher the income , the higher the tax
rate . However, the higher rates are applied only to the income in the bracket
level . Thus, if you examine two incomes, they will both pay the same amount
of taxes at the lower rates. This concept of marginal tax rates is shown in
Table 5 -4.
Case 1 : Total Income 23, 000 Case 2: Total Income 18,000
Income in Bracket Tax Rate Tax Income in Bracket Tax Rate Tax
(1 ) 10,000 2% 200 ( 1 ) 10,000 2% 200
( 2) 10,000 5% 500 (2) 8,000 5% 400
(3) 3,000 7% 210 (3 ) none 7% 0
Total Tax 910 Total Tax 600
TABLE 5 - 4 Examples of Marginal Tax Rates
The design ( or the program to calculate taxes is shown in the structure
chart in Figure 5 - 17. The design has only four functions besides main . The
notation lor bracketTax is somewhat unusual: In the final code, it is called
live times in one expression . Therefore, we show it as a nested set of calls.
I he design lor each ol the functions is shown in
Figure 5 - 18. The logic
tor calculating the tax information requires three
major steps: ( 1 ) Calculate
, , ,
the taxable income. Taxable income is total
income adjusted for the number
° dcl n' c lts - n °ur example, we subtract $ 1,000 for each dependent .
. , â cu ate du taxes The tax due
- calculation uses the progressive
at thc be8inning of the section . Our example uses
fivebracket
^
( 3) Calculate the taxes due
or to be refunded , Note
withheld than are due , the taxes due is
a negative
that if more taxes were
number, indicating a refund.
calculate
tax
getData calcTaxes printlnfo
r
bracketTax j bracketTax
\ bracketTax bracketTax bracketTax
i
FIGURE 5 - 1 7 Design for Calculate Taxes
( Start ) ( getData ) ( calcTaxes )

getData
/ Read Total / Calculate tax lor each bracket
/ Income / ( bracketTax )
Calc / Read / Add brackets

Taxes / taxPaid
J
/ Read Num / taxDue = totTax - taxPaid
Printlnfo / Dpndnts /
T f T
( Stop ) c Return ) ( Return )
( printlnfo ) c bracketTax )
~
/ print all jj income
/ data except t
/ dueTax / startLimit
taxDue
>= tax = 0.0
o.o
/ print Refund / /print Tax Due M

/ -dueTax / / dueTax /
tax = tax =
( income - startLimit ) -
( stopLimit startLimit )
• rate / 100.0 * rate / 100.0
?
c Return ) ?
?
expr: startLimit < income <= stopLimit ( Return )
FIGURE 5 - 18 Design for Program 5 - 5
The code is seen in Program v 5.

PROGRAM 5 - 5 Calculate Taxes

or the refund for a family based
Calculate the tax due
1
2
/*
on the following imaginary -
formula.
deduct $1,000 from income ,
3 1. For each dependent
from the following brackets:
4 2. Determine tax rate tax rate
bracket taxable income
5 2%
1 <= 10000
6
7 2 10001 - 20000 5%
8 3 20001 - 30000 7%
9 4 30001 - 50000 10%

15%
10 5 50001 and up
the refund.
11 Then print the amount of tax or
12
13 Written by:
14 Date:
15 */
17
18 #define LOWEST 0000000.00
19 #define HIGHEST 1000000.00
20
21 #define LIMIT1 10000.00
22 # define LIMIT2 20000.00
25
26 #define RATE 1 02
27 #define RATE2 05
28 #define RATE3 07
29 #define RATE4 10
30 #define RATE5 15
31
32 # define DEDN PER DPNDNT 1000
33
35 void getData (double* totallncome , double* taxPaid ,
36 int* numOfDpndnts);
37
38 void calcTaxes (double
totallncome ,
39 double taxPaid,
40 int numOfDpndnts,
41 double* taxablelncome,
42 double* totalTax,
43 double* taxDue );
44
i
con tinned
\ it
PROGRAM 5 - 5 Calculate Taxes (continued )

45 void printlnformation ( double totallncome ,
46 double taxPaid ,
47 int numOfDpndnts ,
48 double totalTax ,
49 double paidTax ,
50 double taxDue);
51
52 double bracketTax ( double income ,
53 double startLimit ,
54 double stopLimit,
55 int rate);
56
58 {
60 int numOfDpndnts;
61 double taxDue;
62 double taxPaid ;
63 double totallncome ;
64 double taxablelncome;
65 double totalTax ;
66
67 // Statements
68 getData ( & totalIncome , &taxPaid , &numOfDpndnts );
69 calcTaxes ( totallncome, taxPaid , numOfDpndnts,
70 & taxableInco me , & totalTax , &taxDue );
71 printlnformation (totallncome, taxablelncome,
72 numOfDpndnts , totalTax ,
73 taxPaid , taxDue );
74 return 0 ;
75 > // main
76
77 ===== getData =====
78 This function reads tax data from the keyboard.
79 Pre Nothing
80 Post Reads totallncome , taxPaid , & numOfDpndnts
81 */
82 void getData ( double* totallncome, double* taxPaid ,
83 int* numOfDpndnts )
84 {
85 // Statements
printf("Enter your total income for last year: );
"
86
87 scanf ( " % lf " , totallncome );
88
continued
252 Section 5.2 Two-Woy Selection

payroll deductions • " \) •r
89 printf("Enter total of
taxPaid );
90 scanf
91
of dependents .• ).
II V
printf(" Enter the number

/
92
numOfDpndnts );
93 scanf ( "%d ",
94 return ;
95 } // getData
96
97 /* === ===== calcTaxes === —
due.
98 This function calculates the taxes
& taxPaid
99 Pre Given income , numOfDpndnts ,
- tax , and tax due
100 Post Tax income, total
101 calculated
102 */
103 void calcTaxes (double totlnc ,
104 double taxPaid ,
105 int numOfDpndnts,
106 double* taxablelnc ,
107 double* totTax ,
108 double* taxDue )
109 {
110 // Statements
111 *taxablelnc = totlnc
112 ( numOfDpndnts* DEDN PER DPNDNT);
113 *totTax
114 bracketTax( *taxablelnc , LOWEST , LIMIT 1 , RATE1 )
115 + bracketTax(*taxablelnc , LIMIT1 , LIMIT2 , RATE2)
116 + bracketTax(*taxablelnc , LIMIT2 , LIMIT3 , RATE3 )
117 + bracketTax(*taxablelnc, LIMIT3, LIMIT4, RATE4)
118 + bracketTax( *taxablelnc , LIMIT4 , HIGHEST, RATE5 );
119
120
121
*taxDue = *totTax
return ;
taxPaid ;-
122 } // calcTaxes
123
124 /* - printlnformation =====
125 This function prints a table showing all information.
126 Pre The parameter list
127 Post Prints the table
128 */
129 void printlnformation
(double totallncome ,
130
double income ,
131
int numDpndnts ,
132
double totalTax ,
i
continue

133 double paidTax ,
134 double dueTax )
135 {
136 // Statements
137 printf( " \nTotal income :% 10.2f \n " .
138 totallncome);
139 printf( " Number of dependents :%7d \n " , numDpndnts);
140 printf( "Taxable income % 10.2f\n " , income);
141 printf( "Total tax % 10.2f\n " , totalTax );
142 printf( "Tax already paid % 10.2f \n " , paidTax );
143
144 if ( dueTax >= 0.0 )
145 printf("Tax due :%10.2f\n" , dueTax);
146 else
147 printf( " Refund :% 10.2f\n" , dueTax ); -
148 return;
149 > // printlnformation
150
151 bracketTax ===
152 Calculates the tax for a particular bracket.
153 Pre The taxablelncome
154 Post Returns the tax for a particular bracket
155 * /
156 double bracketTax ( double income , double startLimit ,
157 double stopLimit , int rate)
158 {
160 double tax ;
161
162 // Statements
163 if (income <= startLimit )
164 tax = 0.0 ;
165 else
166 if ( income > startLimit && income < = stopLimit)
167
else
tax = ( income- startLimit) * rate / 100.00;
168
169 tax = (stopLimit startLimit ) * rate / 100.00 ;
-
170
171 return tax ;
172 } // bracketTax
Results:
Enter your total income for last year: 15000
Enter total of payroll deductions: 250
continued
254 Section 5.3 Multiway Selection
PROGRAM 5 - 5 Calculate Taxes (continued

)
Enter the number of

dependents: 2
Total income: 15000.00
Number of dependents: 2
Taxable income: 13000.00
Total Tax: 350.00
Tax already paid: 250.00
Tax due: 100.00
Program 5 - 5 Analysis Note that Program 5-5 contains extensive interna

includes a series of comments at the beginning of
.
documentat on This documentation
the program, the define statements
used to set some of the key values in program and the prototype statements .
the ,
Next examine the structure of the program; main contains no detail code; it simply
calls three functions to get the job done. Since two of the functions must pass data back
to main, they use the address ( & ) and indirection ( * ) operators. The call
to getData in
Statement 68 uses the address operator to pass the address of the three variables that
need to be read from the keyboard. Then, Statements 82-83 in the getData header
statement use the asterisk to specify that the type is an address. Because the parameters
are already addresses, the scanf statements do not need an address operator. You must
always consider what type of parameter you are using, data or address, and use the
correct operators for it. Do not automatically use the address operator with scanf .
Examine the code for calcTaxes. The function header ( starting at Statement 103)
specifies that the first three formal parameters are passed as values and the last three
are passed by address. The last three are address parameters because they are calcu-
lated values that need to be passed back to main for later printing . Note that all refer-
ences to them in the function are prefaced with the indirection operator, which tells the
compiler that it must use the variables in main.
Note the use of the type double throughout the program . If you examine the con-
version specifications in the format strings, you will note that they are If for input and
f for output.
Finally, and the main point of this example, note how we used the function
bracketTax to calculate the tax . This function was designed so that it could calculate
the tax for any bracket . This is a much simpler design than writing complex code for
different brackets and demonstrates how keeping it simple and short ( KISS ) makes for
better programs.
5.3 Multiway Selection

,
In addition to two- way selection
.
another selection concept known as
. most programming languages provide
multiway selection . Multiway selection
chooses among several alternatives. The
decision logic for the multiway state-
ment is seen in Figure 5 - 19.
different ways to implement multiway
aS tuo
• .i selection. The first is by
else if thr , ^ ^
. fatement
S 3
C lc ot er ls a
*
programming technique known as the
Style t0 nest V aments. The
an be rrent switch state-
Yi ^
expression. Manvf "
selection
" CVCr suc 1 as
^ condition reduces to an integral
,
* Y, „ *
' ',°"
„„ „,
hen the selection is based on a range
teognJ,"
of values the con 1 '
\ . ' not
t
c
multiway
expression
valuel value2 value3 value4
valuel value 2 value3 value4

action action action action
FIGURE 5- 19 switch Decision Logic
The switch Statement

Snitch is a composite statement used to make a decision between many
alternatives. Figure 5-20 shows the sn itch statement syntax.
l
switch ( expression )
{
ik
-
case constant 1: statement
:
statement
case constant-2: statement
statement
case constant-n: statement
:
statement
default : statement
:
statement
} // end switch
FIGURE 5 - 20 switch Statement Syntax
Although the switch expression can use any expression that reduces
to
common a unary expression in the lorm of an
an integral value, the most is
I . Syntactically, the block in the switch statement is not needed

if only one case is required . I lowever,
is available.
a simple if statement should lx* used il only one choice
arate case label isdefined . Figure 5- 20 also shows the format for the case
case is zero or more statements. Every-
label Associated with each possible label is a sequence. The case label
thing from a case label to the next case
executing the code .
simply provides an entry point to start
the case label. It is executed when-
The default label is a special form of
matches the value in the switch expression.
ever none of the other case values
Note , however, that default is not required . If we do not provide a default , the
compiler will simply continue with the statement after the closing hrace in
the switch .
The switch statement is a puzzle that must he solved carefully to avoid
confusion. Think of the switch statement as a series of drawbridges, one for
each case and one for the default . As a result of the switch evaluation, one
and only one of the drawbridges will he closed , so that there will be a path for
the program to follow. ( II none of the drawbridges is closed , then the state-
ment is skipped and the program continues with the next statement after the
switch . ) The switch flow is shown in Figure 5 - 2 I .
fen feF)
^
sej)
switch
{
.case 2 .
FIGURE 5- 21 switch Flow
case
as
- -- *
7F
shortly, but let ’s look at an example first . Program 5-6 demonstrates the
switch statement . Can you figure out what it prints?
PROGRAM 5-6 Demonstrate the switch Statement

l // Program fragment to demonstrate switch
2 switch ( printFlag )
3 {
4 case Is printf("This is case l \n " );
5
6 case 2: printf( "This is case 2\n" );
7
8 default: printf( "This is default \n " );
9 > // switch
Program 5 -6 has three different case labels. The first case identifies the
entry point to he used when printFlag is a 1 . The second case identifies the
entry point when printFlag is a 2. And finally; the default identifies the entry
point when printFlag is neither a 1 nor a 2. While default is not a required
condition in a switch statement , it should be included when all possible situa -
tions have not been covered by the case labels.
Have you figured out what is printed by Program 5 - 6 ? The answers are in
Figure 5 - 22 . Three results are possible , depending on the value in printFlag.
If printFlag is a 1 , then all three print statements are executed . If printFlag
is a 2 , then the first print statement is skipped and the last two are executed
.
if is neither a 1 nor a 2 , then only the statements defined
Finally , printFl ag
by the default are executed . This results in the first two print statements
being skipped and only the last one being executed .
pTtisis case 1
This is case 2
l^ This is case 2
This is default
l^This is default
This is default
(a) printFlag is 1 (b) printFlag is 2 (c) printFlag is not 1 or 2
FIGURE 5 - 22 switch Results
? To
But what if we want to execute only one of the case - label sequences
. The break statement causes the pro-
do so, we must use break statements
out of the sn itch statement — that is , to go to the closing brace
gram to jump
and continue with the code that follows the switch. If we add
a break as the
5 - 23. Now only one print state-
last statement in each case , we have Figure
,
ment will be executed , regardles s of the value of printFl ag .

switch ( printFlag)
{
case 1:
printf
printFlag ( " This is case 1 " );
break ;
case 2:
printf
("This is case 2");
1 2 default break ;
This is This is This is default:
default printf
case 1 case 2 ( "This is default " );
break ;
} // switch
(a) Logic Flow

(b)Code
~
pThis is case 1 l^
This is case 2 PrhiTis default
FIGURE 5- 23 A switch with break Statements
Two or more case labels can be associated with the same set of actions.
In Program 5 - 7, for example , we print a message depending on whether
printFlag is even or odd .
PROGRAM 5 -7 Multivalued case Statements

1 /* Program fragment that demonstrates multiple
2 cases for one set of statements
3 */
4 switch (printFlag)
5 {
6 case 1:
7 case 3: printf( "Good Day \ n " );
8 printf( "Odds have it!\n");
9 break;
10 case 2:
11 case 4: printf( "Good Day \n " );
12 printf( " Evens have it!\n "
13
);
break;
14 default: printf( " Good Day , I'm
15 confused!\ n" );
printf( "Bye!\n" );
16 break ;
17 } // switch
t
! |[
As a matter of style, the last statement in the switch does not require a
break . We recommend , however, that you get in the habit of using it , especially
when the last statement is not the default . This good habit will eventually
save you hours of debugging time because you will not forget to add it when
you add a new case to the switch statement.
Table 5-5 summarizes some points you must remember about the switch
statement .
1 . The control expression that follows the keyword switch must be an inte-
gral type.
2 Each case label is the keyword case followed by a constant expression .
.
3. No two case labels can have the same constant expression value.
4. But two case labels can be associated with the same set ol actions.
5 The default label is not required. If the value of the expression does not
.
match with any labeled constant expression , the control transfers out -
side of the switch statement . However, we recommend that all switch
statements have a default label .
6 . The switch statement can include at most one default label. The default
label may be coded anywhere , but it is traditionally coded last .
TABLE 5 - 5 Summary of switch Statement Rules
switch Example
Program 5 -8 converts a numeric score to a letter grade. The grading scale is
the rather typical ‘‘absolute scale, ” in which 90 % or more is an A, 80 %—89 % is
a B , 70%-79 % is a C , and 60%— 69% is a I ). Anything below 60% is an F
.
PROGRAM 5 - 8 Student Grading

1 / * This program reads a test score, calculates the letter
2 grade for the score, and prints the grade.
3 Written by:
4 Date:
5 */
7
9 char scoreToGrade ( int score );
10
12 {
continued
PROGRAM 5 - 8 Student Grading ( continued

)
14 int score;
15 char grade;
16
17 // Statements
18 -
printf(" Enter the test score ( 0 100 )
: " );
19 scant ("%d", &score );

20
21 grade = scoreToGrade (score );
22 printf( "The grade is: %c\n" , grade );
23
24 return 0;
25 } // main
26 /* ========== ========== scoreToGrade ==== = ========= =====
27 This function calculates the letter grade for a score.
28 Pre the parameter score
29 Post returns the grade
30 */
31 char scoreToGrade (int score)
32 {
34 char grade;
35 int temp;
36
37 // Statements
38 temp = score / 10;
39 switch ( temp )
40 {
41 case 10:
42 case 9 : grade = A * ;
43 break ;
44 case 8 : grade = B' ;
45 break ;
46 case 7 : grade = * C';
47 break ;
48 case 6 : grade = D';
49 break ;
50 default: grade = F ';
51 } // switch
52 return grade;
53 } // scoreToGrade
Results:
Enter the test score (0
The grade is: 3
-
100): 89
i
\U
n IT
Program 5 - 8 Analysis This example shows how we can use the integer division operator ( / ) to change a
of
range numbers to individual points to be used by the switch statement . The problem
definition requires that if the score is between 80% and 89%, it must be changed to
letter grade ' B ' . This condition cannot be used in a switch statement. But if we divide
the score by 10 ( integer division), the entire range ( such as 80-89 ) can be changed
to one single number (8), which can be used as a constant in the case label.
Note how the break statement works. This is an important part of the logic for
switch statements. Without the break , we would have determined and assigned the
score, and then proceeded to assign all of the lower scores down to ' F ' , with the result
that everyone would have failed. The break allows us to leave the body of the switch as
soon as we have completed the grade assignment.
One word of caution. If the user enters an invalid score, such as 110, this program
gives invalid results. We describe how to prevent this problem in Chapter 6.
The else -if

The switch statement only works when the constant expression in the case
labels are integral . What if we need to make a multiway decision on the basis
of a value that is not integral ? The answer is the else if . There is no such -
C construct as the else - if . Rather, it is a style of coding that we use when we
need a multiway selection based on a value that is not integral.
Suppose we need a selection based on a range of values. We code the lirst ij
condition and its associated statements, and then we follow it with all other pos-
-
sible values using else if . The last test in the series concludes with an else. This is
the default condition ; that is , it is the condition that is to be executed il all other
-
statements are false. A sample of the else if logic design is shown in Figure 5 - 24.
score
=
> 90
score grade = 'A '
grade = 'B
1
score
C
=
> 60
grade = *F' | grade = 'D

1
j
v
Y ?
FIGURE 5 - 24 The else-if Logic Design for Program 5 -9
..
if code . As we said , it is really nothin g
-
What is different about the else indenting each / statement , we code
than
more than a style change. Rather
align it with the previous / I n tins way we sim-
-
the else if on a single line and
we see in the switch and its associated case
ulate the same formatting that
.
label. This style format is shown below
if ( score >= 90 )
grade = 'A';
else if (score >= 80)
grade = 'B' ;
One important point about the else if: Use - it only when the same basic
"> - 24 , the expressions < ue all based on
expression is being evaluated . In Figure
the variable score. If different variables were being evaluated
, we would use
the normal nesting associated with the if . .

else statement . Do not use the
-
else if format with nested if statements .
-
The else if is an artificial C construct that is only used when
.
1 The selection variable is not an integral, and
2. The same variable is being tested in the expressions .
else -if Example

Program 5 - 9 is the same as the switch example in Program 5 -8 hut this time
we use the else-if to solve the problem . It shows how we can use multiway
selection and the else-if construct to change a numeric score to a letter grade.
PROGRAM 5 - 9 Convert Score to Grade

1 /* This program reads a test score, calculates the letter
2 grade based on the absolute scale, and prints it.
3 Written by:
4 Date:
5 */
6 # include <stdio h> .
7
9 char scoreToGrade ( int score );
10
12 {
14 int score;
15 char grade;
16
17 // Statements
continue
. uli
PROGRAM 5 - 9 Convert Score to Grade (continued )

18
19
printf( " Enter the test score (0 100 ): " );
scant ( "%d " , &score);
-
20
21 grade = scoreToGrade (score);
22 printf( "The grade is: %c \n " / grade );
23
24 return 0 ;
25 > // main
26
27 / * ==== =============== scoreToGrade =================
28 This function calculates letter grade for a score ,
29 Pre the parameter score
30 Post returns the grade
31 */
32 char scoreToGrade ( int score )
33 {
35 char grade ;
36
37 // Statements
38 if (score >= 90)
39 grade = ' A ' ;
40 else if ( score >= 80 )
41 grade = 'B';
43 grade = C * ;
45 grade = ' D';
46 else
47 grade = 'F' ;
48 return grade;
49 } // scoreToGrade
Results:
Enter the test score (0 100): 90 -
The grade is: A
Program 5 - 9 Analysis We used the else- if construct because our condition was not an integral; rather, it
tested several ranges of the same variable, score. Study the code carefully. Note
that, once the correct range is located, none of the following conditions will be tested.
For instance, if a score of 85 is entered, the test against 90% is false, so
we execute
than 80% Since this condition is true, we set grade
the else- if test for a score greater .
to ' B ' and skip all the remaining tests .
Also, note how the tests are ordered. In this case we first eliminate those scores
equal to or greater than 90%; then we check 80%, 70%, and 60% in turn. Sincefirst we
were checking for greater than, we could not have coded it in the reverse, 60% .
264 Section 5.4 More Standard Functions
eS'L
^ wil the lowest Calue
®
JSh LTaigeClueSn chSg than start
5.4 More Standard Functions

One of the assets of the C language is its
rich set of standard functions that make
we introduced some of these standard
programming much easier. In Chapter 4
have studied selection , we can discuss two other libraries
functions. Now that we
functions that are closely related to selection statements.
of standard
toi manipulating charac-
C 99 has two parallel but separate header lies
(
set . The second ,
ters . The first , ctype.h, supports the ASCII character
wctype .h , supports wide characters . While the wide - character library
( wctype .h ) contains some functions not found in the ASCII character library
( ctype.h ) they follow a common naming format . For example , the ASCII
j
library contains a function, islower, to test for lowercase alphabetic charac-

ters. The equivalent function in the wide-character library is isulower . There-
fore, if we know the name for one library; we know it for the other.
Standard Characters Functions

The character libraries are divided into two major groups: classifying func-
tions and converting functions. The prototypes of these functions are in the
ctype .h or wctype .h header files. Before looking at these functions, make sure
you understand the classification of characters that C uses. I bis breakdown
of classes is shown in Figure 5- 25 , which uses a tree to show how characters
are classified. You read the tree much like a structure chart , starting at the
top and following the branches to the bottom .
characters
%
control printable
space graphical
alphanumeric punctuation
alphabetic digit |
| upper lower
FIGURE 5- 25 Classifications of the Character Type

ll
Characters are first broken down into control characters , such as car-
riage return and end of file , or into printable characters. This tells us that
control characters are not printable. The printable characters are either a
space or the rest of the printable characters, which are classified as graphical .
In turn , the graphical characters are broken down into alphanumeric and
punctuation characters. Alphanumeric means either an alphabetic character
or a digit. Finally, alphabetic characters are either upper- or lowercase.
Classifying Functions
Classifying functions examine a character and tell if it belongs to a given
classification as described previously. They all start with the prefix is and return
true if the actual parameter is in the specified class and false if it is not . The
prototypes of these functions are found in the ctype.h and cwtype .h files.
The general form of the prototype function is
int is... ( int testChar )
where the function name starts with is or isw ; 2 for example , iscntrl ,
w hich stands for " is a control character.
"
All of the classifying functions return true or false . If the character

matches the set being tested by the function , it returns true .; if it doesn t , it
’
returns false . For example , the isdigit function tests the character against the
decimal digits ( 0 through 9 ). If the character is a decimal digit , it returns
.
true ; if the character is not a decimal digit , it returns false fable 5 -6 summa -
rizes each function with a brief explanation .
Function Description
iscntrl Control characters

isprint Printable character, that is character with an assigned graphic
isspace Whitespace character: space character (32), horizontal tab (9),
line feed ( 10) , vertical tab ( 11 ), form feed ( 12), and carriage
return ( 13 )
isgraph Character with printable graphic; all printable characters except

space
isalnum Alphanumeric: any alphabetic or numeric character
ispunct Any graphic character that is not alphanumeric
isalpha Any alphabetic character, upper- or lowercase
continued
TABLE 5 -6 Classifying Functions
-
2 . For wide character functions.
266 Section 5.4 More Standard Functions
Description
Function
isupper Only uppercase alphabetic
islower Only lowercase alphabetic
isdigit Decimal digits (0...9)
Hexadecimal digits (0...9, A...F )
isxdigit
isodigit Octal digits (0...7)
TABLE 5-6 Classifying Functions (continued )
Character Conversion Functions

from one case to
Two converting functions are used to convert a character
another. These functions start with prefix to or tow ( for wide characters ) and
return an integer that is the value of the converted character . Their basic for-
mat is
int to... ( int oldChar )
Table 5 - 7 summarizes each function with a brief explanation .
Function Description
toupper Converts lower- to uppercase. If not lowercase, returns it
unchanged.
tolower Converts upper- to lowercase. If not uppercase, returns it
unchanged.
TABLE 5-7 Conversion Functions
A Classification Program
Let s write a program that uses classification functions to examine a character
input from the keyboard. If you study Figure 5- 25, carefully, you will note that
most characters fall into more than one
classification . For instance, a digit is
printable, graphical, alphanumeric, and a digit.
Our program tests from the bottom up, so that
only one classification will
be printed lor each character. For a digit ,
only the fact that it is a digit will he
prime . ecausi it is a relatively
simple demonstration of the classification
functions, we wrote it using only main . The
solution is shown in Program 5 10. -
PROGRAM 5 - 10 Demonstrate Classification Functions

1 /* This program demonstrates the use of the character
2 classification functions found in the c type library. -
3 Given a character , it displays the highest
4 classification for the character.
5 Written by:
6 Date:
7 */
8
10 # include <ctype.h>
11
12 int main (void)
13 {
1 4 // Local Declarations
15 char charln ;
16
1 7 / / Statements
18 printf( " Enter a character to be examined: " );
19 scanf ( " %c" , &charln );
20
21 if (islower(charln ))
22 printf( " You entered a lowercase letter.Xn " );
23 else if (isupper(charln ))
24 printf( " You entered an uppercase character.\n " );
25 else if (isdigit(charln ))
26 printf( " You entered a digit.Xn " );
27 else if (ispunct(charln ))
28 printf( " You entered a punctuation character.\n " );
29 else if ( isspace( charln ))
30 printf( " You entered a whitespace character.\n " );
31 else
32 printf( " You entered a control character \n " );
33 return 0 ;
34 > // main
Results:
Enter a character to be examined: a
You entered a lowercase letter.
so at first analy -
Program 5 - 10 Analysis Note the use of the else - if in this program . While it may not appear for multiway
sis, this is actually a range analysis problem, which makes it suitable
selection.
268 Section 5.5 Incremental Development PartJl
Handling Major Errors

acronyms is GIGO-garbage in , garbage
One of the better known computer
out. In writing programs, we
must decide how to handle errors to prevent gar-
, as we will see m Chapter 6, we
bage from corrupting the data . Sometimes re enter tt. Other times, there is
can recover from the error by having the user -
no way to recover.
functions that allow us to termi-
When we can’t recover, C provides two
and abort . Both functions arc found in the standard
nate the functions: exit
library (stdlib.h) .
exit
Whereas return terminates a function , exit terminates the program regardless
of where in the program it is executed . While we use it to terminate the pro-
gram because we detected an error, C considers it a normal termination. For
this reason , the termination is orderly; any pending Hie stream writes are first
completed and all streams are closed . The exit prototype statement is
shown below.
void exit ( int terminationStatus);
There is one parameter to the exit call , an integer value to he passed to

the operating system . While any integer is acceptable, it is usually a non -
zero, indicating that the program did not complete successfully. We demon -
—
strate the exit function in Program 5 - 12 , “ Menu -driven Calculator Third
Increment ."
abort
The abort Junction is used to terminate a program abnormally. It is consid -
* ered a non -orderly termination because the output streams are not Hushed
and they are not closed . It is like drawing the "Go to Jail" card in Monopoly
you go directly to jail , you do not pass GO, and you do not collect $ 200.
—
\\ hen abort is called , the program immediately goes to the operating system .
I he abort function has no parameters. Its prototype statement is
shown below.
void abort ( void );
5.5 Incremental Development Part II

Ihifplc emulator de
;f °Pmt nt ^ J
'
calculator subfunctions. ^ dlScusslon by adding a menu and

Calculator Design
Although it is an elementary example, our calculator program illustrates two
concepts , hirst , it shows how we can communicate with a user through a
menu . Menu ’s are a common design for programs that interact with users.
Second , it demonstrates incremental development through three levels in a i
-
program . Figure 5 26 shows the design of the program .
calculator
getOption
I getData calc
I printResult
add sub mul dvd
FIGURE 5 - 26 Design for Menu-driven Calculator
Calculator Incremental Design

Good structured programming requires that each I unction do only one thing. .i
In this program , we are doing four different things: ( 1 ) We must ask the user
what function is desired . Then we need to ( 2 ) get the data tor the operation ,
( 3 ) make the calculation , and finally ( 4 ) print the result . These four processes
are seen as called functions in main . In its turn , calc calls one ot lour I unc -
tions to perform an arithmetic operation . Finally, the print result function
displays the result .
First Increment: main and Get Option

Program 5 11 begins the development with the main function and the get
-
option function .
PROGRAM 5 - 11 Menu -driven Calculator — First Increment

/* This program uses a menu to allow the user to add
,
1
2 multiply , subtract , or divide two integers.
3 Written by:
4 Date:
5 */
8
continued
270 Section 5.5 Incremental Development Part II _
— ——
Increment (continued )
PROGRAM 5 - 11 Menu -driven Calculator -
First
10 int getOption (void );

11
13 {
15 int option ;
16
17 // Statements
18 option = getOption( );
19
%d\n , option );
20 printf( "**You selected option
21
22 return 0 ;
23 } // main
24
25 ====== getOption ====================
26 This function shows a menu and reads the user option
27 Pre Nothing
28 Post returns the option
29 */
30 int getOption (void )
31 {
33 int option ;
34
35 // Statements
36 printf ( "\t********************************** " )
37 printf("\n\t* MENU * ••
38 printf( "\n\t* *»
39 printf( "\n\t* 1 ADD . *'* )
40 printf( "\n\t* 2. SUBTRACT * •• )
41 printf( "\n\t* 3. MULTIPLY
* ••
42 printf( "\n\t* 4. DIVIDE •• )
43 printf( "\n\t*
* •)
•
44 printf( "\n\t************** ***** ************* * * "
45
46 piintf( \ n\ nPlease type your
M
choice " );
47 printf( " and key return: " );
48 scanf ( "%d " , option);
49 ^
piintf( " * * You selected option
%d \n" , option );
50 return option ;
51 > // getOption
Results:
************ *******
111
***************
A continuei
PROGRAM 5 - 1 1 Menu - driven Calculator — First Increment ( continued )

MENU *
* *
* 1. ADD *
* 2. SUBTRACT * i
* 3. MULTIPLY
* 4. DIVIDE *
* *
********************************* *
*
Please type your choice and key return: 3

**You selected option 3
**You selected option 3
Program 5 - 1 1 Analysis We use the same approach that we used previously. In main we call the get option
function. Just before returning, we print the option and then we re-print it in main so
that we can easily verify the results.
We incorporate one new style. Because it can be difficult to find the debugging
we leave
statements in a large function, we do not indent them in the function. Rather
them flush to the left margin so that they are easy to find and remove and we continue
the incremental development.
Second Increment: getData

Following the structure chart, we next write and debug get data. Because we
wrote and debugged the get data function in Chapter 4 we
, do not need to
include it here. We simply copy it into our new program.
Third Increment: Calculate

The third function is calculate(calc). This is the first example of a subfunc -
tion that calls other subfunctions: add, subtract, multiply, and divide
. In this
case, we write the complete function except for the subfunctio n calls.
The design for the calculate function requires a switch statement
to
select the correct subfunctio n In.place of the actual calls , we return dummy
divide. You
values: 1.0 for add, 2.0 for subtract, 3.0 for multiply, and 4.0 for
, use
should recognize these as the option numbers. When testing a program
be easily verified . I he code is
test data that is easy to remember and that can
shown in Program S - 12.
PROGRAM 5 - 12 Menu -driven Calculator — Third Increment

to add ,
1 /* This program uses a menu to allow the user
multiply , subtract , or divide two integers.
2
3 Written by:
4 Date:
5 */
continued
272 Section 5.5 Incremental Development ParjJL
( continued )
PROGRAM 5 - 12 Menu -driven Calculator -Third
Increment
7 # include <stdlib.h>
8
11 void getData ( int* numl ,
int* num2 );
12 float calc (int option, int numl , int num2);
13
15 {
17 int option ;
18 int numl;
19 int num2 ;
20 float result;
21
22 // Statements
23 option = getOption();
24 getData (&numl , &num2 );
25 result = calc (option, numl , num2 );
26 printf( "**In main result is: %6.2f " , result );
27
28 return 0;
29 > // main
30
31 /* === getOption ===
32 This function shows a menu and reads the user option.
33 Pre Nothing
35 */
36 int getOption ( void )
37 {
39 int option ;
40
41 // Statements
42 printf ( "\t************* *** **
******** ** *** * ** ).
i
43 printf("\n \t*
MENU *»
44 printf( "\n\t*
45 *» )
printf("\n\t* 1. ADD
"
46 printf( "\n\t* 2.
SUBTRACT * ")
47 printf( M \n\t* 3.
48
MULTIPLY * •• )
printf( \n\t* 4.
M
49
DIVIDE * <• )
printf("\n\t*
*» )
i ;
continue
\i
PROGRAM 5-12 Menu-driven Calculator — Third Increment (continued )

-
printf " \j}\ £***************************** ***** " j •
50
51 ^
52 printf( " \n\nPlease type your choice " );
53 printf("and key return: ");
54 scant ( " %d " , &option );
55 return option;
56 > // getOption
57
58
60 Pre Parameters a and b are addresses
61 Post Data read into parameter addresses
62 */
63 void getData ( int* a , int * b)
64 {
65 printf( " Please enter two integer numbers: " );
66 scanf("%d %d " , a , b);
67 return ;
68 > // getData
69
70 /*
71 This function determines the type of operation
72 and calls a function to perform it.
73 Pre option contains the operation
74 numl & num2 contains data
75 Post returns the results
76 */
77 float calc ( int option , int numl , int num2 )
78 {
80 float result ;
81
82 // Statements
83 printf( •• * *In calc input is: %d %d %d\n " ,
84 option , numl , num 2 );
85 switch(option)
86 {
case 1: result 1.0 ; // Add
87
88 break ;
case 2: result = 2.0 ; // Subtract
89
90 break ;
case 3: result = 3.0 ; // Multiply
91
92 break ;
== 0.0 ) // Divide
93 case 4: if ( num 2
contmuec
274 Section 5.5 Incremental Development Part II
PROGRAM 5 - 12 Menu-driven Calculator

Third Increment (continued )
{
94
95
printf("\n\a\aError: " )?
96
printf( "division by zero * **\n" );
97
exit (100);
> / / if
98
99 else
100 result = 4.0;
101 break;
102 /* Better structured programming would validate
103 option in getOption. However, we have not
104 yet learned the technique to code it there.
105 */
06 default: printf("\aOption not available\n");
07 exit (101);
08 > // switch
09 printf(" **In calc result is: %6.2f \n " , result );
10 return result;
111 return result;
112 > // calc
Results:
**************************** ******
* MENU *
* *
* 1. ADD *
* 2. SUBTRACT *
* 3. MULTIPLY *
* 4. DIVIDE *
* *
**********************************
Please type your choice and key return: 1

**In calc input is: 1 13 8
**In calc result is: 1.00
**In main result is: 1.00
Program 5- 12 Analysis Once again, note how we verify both the

downward and upward communication.
The calculate function displays the parameter values it
receives and the result before it
returns. In main the value that was returned
is displayed .
Before we can begin to write the subfunctions
tests, one for subtract, one for multiply
, however, we must run four more
, and two for divide . To test divide, we must first
test it with a zero divisor and then with
a valid divisor.
i
Fourth Increment: Add

We wrote the code For the add function in Chapter 4 so we don’t need to write
and debug it again We can simply copy it to our new program. (Because ot
.
the design change required to pass back a real number From divide, we must
change the add Function to return a flout rather than an int . This change is
minor, however, so we don’t show the incremental program For it here.
Fifth Increment: Subtract

At this point we have verified that the calculate parameters are passed cor -
rectly and the correct value is being returned. We are therefore ready to
develop and debug its subfunctions. Note that the debugging code is still in
place For the multiply and divide Functions. We do not remove it until the
subfunctions have been written and debugged.
On the other hand, we have removed the input parameter displays
because they have been debugged. The upward communication verification
display remains, however, until the calculate function is complete. The code
is shown in Program 5 - 13.
PROGRAM 5 - 1 3 Menu - driven Calculator — Fifth Increment

1 /* This program uses a menu to allow the user to
add ,
2 multiply , subtract, or divide two integers.
3 Written by:
4 Date:
5 */
8
10 int getOption (void);
11 void getData ( int * numl , int* num 2 );
12 float calc ( int option , int numl , int num2 );
13 float add ( int numl , int num2 );
14 float sub ( int numl , int num 2 );
15
17 {
19 int option;
20 int numl ;
21 int num2;
22 float result;
23
continuec
276 Section 5.5 Incremental Development Part II
(continued )
PROGRAM 5- 13 Menu -driven Calculator -Fifth
Increment
24 // Statements
26 getData (&numl, &num2);
, num 2 );
27 result = calc (option, numl
28
29
// Temporary code to be —
removed
printf("**In main result is:
%6.2f \n " , result );
30
31 return 0;
32 } // main
33
34 /* ===== ======== getOption ===== =======
35 This function shows a menu and reads the user option.
36 Pre Nothing
38 */
39 int getOption (void)
40 {
42 int option;
43
44 // Statements
45 P3 "1ritf ( "\£******* ** A:* * ***** * **** *
• •
*** * * * ,, J
' ' '
^
46 printf( n t*
" \ \ MENU * •• )
47 printf( "\n\t* *"
48 printf( "\n\t* 1. ADD *» )
49 printf( "\n\t* 2. SUBTRACT »
50 printf("\n \t* 3. MULTIPLY * •' )

51 printf( "\n\t* 4. DIVIDE *"
52 printf(" \n \t* *» )
53 printf( "\n\t******** ************************** *
54
55 pr int f( " \n\nPlease type your choice " );
56 printf( " and key return: " );
57 scanf ( "%d\ Soption);
58 return option;
59 > // getOption
60
61 /* ===
= getData ====
62 This function reads two integers
from the keyboard.
are addresses
64 ost Data read into parameter
65 */ addresses
66 void getData ( int* a
, int* b)
67 {
68 printf( "Please enter
i. two integer numbers: " );
continued
PROGRAM 5- 1 3 Menu-driven Calculator — Fifth Increment (continued )

68 scanf( " %d %d " , a , b );
69 return;
70 > // getData
71
72 /* ==================== calc ============ ========
73 This function determines the type of operation
74 and calls a function to perform it.
75 Pre option contains the operation
76 numl & num 2 contains data
77 Post returns the results
78 */
79 float calc ( int option , int numl , int num2)
80 {
82 float result;
83
84 // Statements
85 switch(option)
86 {
87 case 1: result = add ( numl , num 2 );
88 break ;
89 case 2: result = sub ( numl , num 2 );
90 break ;
91 case 3: result = 3.0 ; // Multiply
92 break ;
93 case 4: if (num2 == 0.0) // Divide
94 {
95 printf( " \n\a\aError: " );
96 printf( " division by zero ***\n " );
97 exit ( 100);
98 > // if
99 else
100 result = 4.0 ;
101 break ;
102 / * Better structured programming would validate
103 option in getOption. However , we have not
104 yet learned the technique to code it there.
105 * /
106 default: printf( " \aOption not available\n");
107 exit ( 101 );
108 } // switch
109 printf( "**In calc result is: %6.2f\n , result)
" ;
110 return result ;
111 > // calc
continued
J
278 Section 5.5 Incrementol Development Port II
Increment (continued )
PROGRAM 5- 13 Menu-driven Calculator-Fifth
112
113 /* === and returns the sum.
114 This function adds two numbers
b contain values to be added
115 Pre a and
Post Returns a + b
116
117 */
118 float add ( int a , int b)
119 {
121 float sum;
122
123 // Statements
124 sum = a + b;
125 return sum ;
126 > // add
127
128 /* === ==
129 This function subtracts two numbers
130 Pre a and b contain values to be subtracted
132 */
133 float sub ( int a , int b)
134 {
136 float dif;
137
138 // Statements
139 dif = a b; -
140 printf("**ln sub result is: %6.2f \n" , dif );
141 return dif ;
142 } // sub
Results:
*************************
*
^^^
MENU *
*
*
* 1. ADD *
* 2. SUBTRACT
*
* 3. MULTIPLY
*
* 4. DIVIDE
*
*
*
**************************
** ******
Please type your choice and
key return: 2
i
continue*
Chapter 5 Selection — Making Decisions 2 / 9
n
-
ft I
n
PROGRAM 5 - 13 Menu -driven Calculator — Fifth Increment ( continued )
**In sub result is: 5.00
**In calc result is: 5.00
**In main result is: 5.00 i
Program 5 - 1 3 Analysis Note how we display the results up the program to main. This verifies that the upward
communication is correct. If there should be a problem, we will know exactly where it
occurred.
Remaining Increments
At this point, we have demonstrated the incremental development concepts for
programs using multiple levels of subprograms. To complete the program, we
must write and debug the final three increments:
• Increment 6: multiply
• Increment 7: divide
• Increment 8 : print results
We leave the completion of the program for you as a problem at the end
of the chapter.

Dependent Statements
other statements that
Several statements in the C language control that we have seen . follow
them The if. else is the first of these statements When-
statements that follow it , good
ever one statement controls or influences
statements to show
structured programming style indents the dependent
code is dependent on the controlling statement . The com-
that the indented
of how a program is formatted-but

—
piler does not need the indentation it follows its syntactical
good style makes for
rules regardless
readable programs.
To illustrate the point , consider the two versions of the code for the func-
tion in Program - 5 14 . They both accomplish the same task . To make this
exercise even more meaningful , cover up the right half of the program and
predict the results that will be produced when the ill-formed code executes.
Then look at the well-structured code.
PROGRAM 5 - 14 Examples of Poor and Good Nesting Styles

Poor Style Good Style
1 int someFun ( int a, int b) int someFun (int a , int b)
2 { {
3 int x ; int x ;
4
5 if ( a < b) if (a < b )
6 x = a; x = a;
7 else else
8 x b; x = b;
9 x 5f ; X * = •5 f ;
10 return x; return x ;
11 > // someFun > // someFun
Assume that in this example, a has a value of 10 and b has a value of 20.
W hat value will he returned: First , look at Statement 6. The
assignment of x
in this example is dependent on the i in Statement
/
not indented , however, it is difficult to see the
5. Since Statement 6 is
dependency. Because the value
o a ( 10 ) is less than the value of b ( 20 ) , x
will he assigned the value 10.
Now examine Statement 9. It is
indented and therefore appears to be
epen ent on t e else statement . But is it ?
The answer is no. It just looks that
way and is therefore misleading.
-
Statement 9 will therefore execute regard -
expression in the if statement .
This relationship is much more
2l See ln
"
/
0 f
' tyW code on the right . The code on the right is
,W the
, rclationshiPs
foree the chance of misreading among the statements , and there-
the code is minimal .
L, ,
I he indentation rules are summarized in Table 5-8.
1 . Indent statements that are dependent on previous statements. The

indentations are at least three spaces from the left end of the control -
ling statement.
2. Align else statements with their corresponding if statements. ( See
Figure 5- 1 3. )
3. Place the opening brace identifying a body of code on a separate line.
Indent the statements in the body of the code one space to the right of
the opening brace.
4. Align the closing brace identifying a body of code with the opening
brace, and place the closing brace on a separate line. Use a comment to
identify the block being terminated .
5. Align all code on the same level , which is dependent on the same con -
trol statement.
6. Further indent nested statements according to the above rules.
7. Surround operators with whitespace .
8. Code only one definition or statement on a single line .
9 . Make your comments meaningful at the block level . Comments should
not simply parrot the code .
TABLE 5- 8 Indentation Rules
Negative Logic
Negative logic refers to any expression that begins with not or that contains
multiple not expressions within . Like the proverbial double- negative in the
English language , interpreting negative logic can be difficult . We therefore
try to avoid it .
In the discussion of Figure 5- 12 and in the section on simplifying the if
statement , one technique that we proposed was complementing a condi -
tional statement. This requires making a positive statement negative and a
5- 2,
negative statement positive. This can be done by following Rule 6 in Fable
” which states that the positions of
“ Syntactical Rules for if ...else Statements ,
the statements in an if ..else can be swapped if the original control expression

is com plemented . The concept of complementi
ng the if statement was shown
in Figure 5 - 10 , “Complemen ted if ...else Statements .”
Remember, however, that simple code is the clearest code. T his concept
has been formulated into an acronym : KISS , which stands for “ Keep It Sim -
ple and Short!” ( see Chapter 3 ) . Unfortunatel y, negative logic is not always
extremely confusing. We have seen professional

simple. In fact, it can" get;
to debug negative logic .
programmers work for hours trying
!
Avoid compound negative statements
, make sure that the resulting state-

When you complement an expression
ment is readable. Complementing
an expression can he more difficult than
simply making the condition negative
. Examine the third statement in
the complement of the not ( ! ) is uot - not ( !|),
Table 5-9 carefully. Note that
should avoid compound nega-
which in effect cancels the not . In general , you
, the complemented statement is
tive statements. In this case , therefore
greatly preferred .
Original Statement Complemented Statement

if (x > 0)
i f ( x <= 0 )
if (x != 5) if (x == 5)
i f (! (x <= 0 || !flag) ) if (x > 0 && f l a g )
TABLE 5 - 9 Complementing Expressions
Rules for Selection Statements

Tor selection statements, you need to consider three other rules, shown in
Table 5- 10. Since these rules are sometimes conflicting, they are listed in
their order of importance.
.
1 Code positive statements whenever possible.
2. Code the normal /expected condition first .

.5 . Code the most probable conditions first .
TABLE 5- 10 Selection Rules
Human engineering studies have shown that people make fewer errors
en rca * n 8 positive statements than
^ us is
i
when reading negative statements.
especially true when complex, compound Boolean statements are
—
involved. Therefore, the first rule in Table 5 10 is, whenever possible, code
your selection statements using positive
-
conditions.
low P, ! 7Td TCemS
FUle
‘he human expectations about what will fol -
confused if H
f ' to an cipate things. They will therefore be less
'
*
Finally, the third rule concerns the efficiency of the resulting program.
Coding the most probable conditions first is especially important in a multi-
way selection, such as the else - if . When you code the most probable test first,
then the program can skip the rest of the statements. Obviously, the more
statements skipped, the more efficient the resulting program.
As we mentioned previously, these rules often conflict with each other.
!
We have listed them in their order of importance from a human engineering

point of view. Unless there are overriding circumstances, you shoidd select
the higher option ( Rule 1 before Rule 2 before Rule 3) in case of conflicts.
But remember the overriding principle: KISS — Keep It Simple and Short .
Selection in Structure Charts

We introduced the basic concepts of structure charts in Chapter 4. In ibis sec -
tion we extend the discussion to selection in structure charts. Figure 3 - 27
shows two symbols for a function that is called by a selection statement,
the condition and the exclusive or.
int select (...)

int dolt (...) {
dolt { select :
: if (...)
5 if (...)
fun (...);
I
doA
(+)
doB
else
doA (~);
fun
I :
} // dolt
doB (...);
// select
}
Design Code
Design Code
(a ) conditional ( b) exclusive or
FIGURE 5 - 27 Structure Chart Symbols for Selection
a sub-
In Figure 5 - 27 ( a ) , the function dolt contains a conditional call to
function, fun. If the condition in the // statement is true , we call dolt . II it is
. represent ed in a structure chart as a
not true, we skip dolt This situation is
small diamond on the vertical line between the two function blocks .
Figure 3 - 27 ( b ) represents the selection between two different functions
.
between and doB One and
In this example, the function select chooses doA .
is exe-
only one of them will be called each time the conditional statement
the alternative s is exe-
cuted. This is known as an exclusive or ; one of two
cuted to the exclusion of the other. The exclusive or is represent ed by a plus
sign between the processes.

can be called
Now consider the design found when a series of functions
calls to several
exclusively. This occurs when a multiway selection contains
7
different functions. Figure 5 -28 contains an example of a switch statement

that calls different functions based on color.
switch (color)
selectColoJ
{
case R': colorRed (...);
break ;
(+j U+ ) case ' B ':
colorBlue (...);
break ;
colorRed colorBlue I otherColor default : otherColor (...);
} // switch
( a) Design (b) Code
FIGURE 5 - 28 Multiway Selection in a Structure Chart
The structure chart rules described in this section follow:

1. Conditional calls are indicated by a diamond above the called function
rectangle.
2. Exclusive or calls are indicated by a (+) between functions.

1 . The complement of < is > =, and the complement of > is < = .
2. Dangling else statements are easily created and difficult to dehug. One
technique for avoiding dangling else statements is to use braces, even
when they are not needed .
.
3 The expression in the control expression in the if . . . else statement may
have a side effect , as shown below.
if ( a++)
4. Encapsulate the statements inside braces if you have more than one
..
statement after ij or else in an if else statement .
3. Do not use the equal operator with a floating- point number, f loating-
point numbers are seldom exactly equal to a required value. When you
need to test for equality, such as a = = b , use the expression shown below.
if (fabs (a - b) < .0000001 )
.
6 Do not forget to use a break statement when the cases in a switch state -
ment are exclusive.
7. While not necessarily an error, it is poor programming practice to write a
switch statement without a clejault label . If the logic doesn ’t require one ,
code it with an error message to guard against unanticipated conditions.
This is shown below.
default: printf(" Nalmpossible default\n");

exit ( 100 );
8. The most common C error is using the assignment operator ( = ) in place

of the equal operator ( = = ) . One way to minimize this error is to get in the
habit of using the term “ assignment operator when
" reading the code.
"
" a is assigned b , " not " a equals b.
For example, say
9. It may be an error to place a semicolon after the if expression .
a. The semicolon terminates the if statement, and any statement that
fol -
lows it is not part of the if
b. It is a compile error to code an else without a matching if . This error is
most likely created by a misplaced semicolon .
// if terminated here
if (a == b);
printf(...);
else // No matching if
printf(...);
10. It is a compile error to forget

the parentheses in the if expression.
11 . It is a com pile error to put a
space between the lollowing relational oper-
ators: = = , !=, > = < =. It is also
a compile error to reverse them.
rather than an integral constant as
12. It is a compile error to use a variable
the value in a case label .
13. It is a compile error to use the same
constant in two case labels.
( & & ) or two bars i | ). It is
14. The logical operators require two ampersands
one. (Single operators are bitwise
a logic error to code them with only
operators and are therefore valid code . )
1 5. It isgenerally a logic error to use side effects in the second operand in a

logical binary expression , as shown below, because the second operand
may not be evaluated.
( a ++ & & — b)
5.8 Key Terms

break statement exclusive or
case logical operator
case- label multiway selection
classifying function negative logic
comparative operator nested if statement
complemented if ...else null else statement
conditional expression selection
converting function selection in structure chart
dangling else simplifying if statements
default label switch
De Morgan s rule
dependent statement
-
two way selection
5.9 Summary
Data are called logical if they convey the idea of true or false.
C 99 implements the logical type, bool . It also
supports integer logical
d t ata item is
nonzero, it is considered true ; if it is zero, it is
considered false .
three operators lor combining logical values to create new' values:
wot, and , or.
Six comparative operators are
used in C: <, <= > , > =, = = , and !=.
Selection in C is done usi
using two statements: if . .else and switch.
Chapter 5 Selection — Making Decisions 28 /
used for selection between two alternatives.

J I he if . . .else construct is
IP!
J You can swap the statements in the true and false branches if you use the
.
complement of an expression in an if .. else statement .
II the false statement is not required in an if . .. else , it is omitted and the
keyword else dropped . J,
If an else is dangling, it will he paired with the last unpaired if .

Multiway selection can be coded using either the suitch statement or an
-
else if construct.
The switch statement is used to make a decision between many alterna -
tives when the different conditions can be expressed as integral values.
The else- if format is used to make multiple decisions when the item being
tested is not an integral and therefore a switch statement cannot be used .
A labeled statement is used for selection in a switch statement.
A default label is used as the last label in a switch statement , to be exe -
cuted when none of the case alternatives match the tested value.
Indenting the controlled statements in C is good style that enhances the
readability of a program.
Selection is used in a structure chart only when it involves a call to
another function.
The structure chart for selection shows the paths taken by the logic flow.
You cannot always tell by looking at the structure chart which selection
will be used ( two- way or multiway ) .
a . A simple if is indicated by a diamond below the calling function ,
h . An if .. . else and switch arc indicated by the exclusive or ( + ) .
5.10 Practice Sets

Review Questions
1 . Logical data are data that can be interpreted as true or false .
a. True
b . False
2. The expression in a selection statement can have no side effects
.
a. True
b. False
3. Each labeled statement may identify one or more statements
.
a. True
b. False
in the standard library
4. The character classification functions are found
(stdlib.h).
.
a True
b. False
5. To ensure that a character is uppercase
, the toupper conversion function
is used .
a. True
b. False
6. The logical operator is true only when both operands
are true.
a. and ( & & )
b. greater than (> )
c. less than ( <)
d. or (||)
e. not (!)
.
7 Which of the following is not a comparative operator in C ?
a. <
b. <=
c. =
d. >
e. >=
8. Two-way selection is implemented with the statement.
a. case
b. else if
c. switch
d. the if . ..else and the switch
e. if . . .else
.
9 Which of the following is not a syntactical rule for the if . ..else statement?
a. Any expression can be used for the if expression .
b. Only one statement is allowed for the true and the false actions.
c. I he true and the false statements can he another if ... else statement.
d . 1 he expression must he enclosed in parentheses.
e. The selection expression cannot have a side
effect .
10. \ \ hich ol the following statements
creates the ?
“dangling else problem
a . a nested if statement without a false
statement
b. a nested if statement without a
true statement
c. a switch statement without a
default
d . an if statement without a true or
a false statement
e. any nested if statement
It
11 . there are two different ways to implement a multiway selection in C .

They are
.
a if . .. else and switch
b. else -if and switch
.
c if ...else and else if
d . else - ij and case
e. switch and case
12. Which of the following statements about switch statements is lalse?
.
a No two case labels can have the same value.
b. The switch control expression must be an integral type.
c. The case - labeled constant can be a constant or a variable .
d . Two case labels can he associated with the same statement series.
.
e A sii itch statement can have at most one default statement .
’
-
1 3. Which of the following statements about the else if is false?
-
a . Each expression in the else if must test the same variable .
b. The else if is a coding style rather than a C construct .
-
.
c The else // requires integral values in its expression .
-
.
d The else if is used for multiway selections.
-
.
e The last test in the else if series concludes with a single else , which is
-
the default condition . '
14 . Which of the following is not a character classification in the C language?

a . ascii
b. control
.
c digit
d . graphical
e. space
Exercises
1 5. Evaluate the following expressions to true or false . Show how you
arrived
at your answer by first adding the default parentheses and then showing
-
the value of each expression as it would be evaluated by C , one expres
sion to a line.
.
a ! ( 3 + 3 >= 6 )
.
b 1 + 6 = = 7 || 3 + 2 = = 1
.
c 1 > 5 | | 6 < 50 & & 2 < 5
d . 1 4 ! = 5 5 & & I ( 1 3 < 2 9 ) || 3 1 > 5 2
e. 6 < 7 > 5
16. If x = 0, y = 5, z 5, what is the value of x, y, and z for each of the follow

=
ing code fragments ? ( Assume that x, y, and /. are their o r i g i n a l values for
each fragment .)
a. i f ( z != 0 )
y = 2 9 5;
else
x = 10;
b. i f (y + z > 10 )
y = 99;
z = 8;
x = ++ z ;
c. i f ( x == 0 )
{
x = x - 3;
z = z + 3;
}
else
y = 99;
17. If x - 3, y - 0, and z
a. x & & y || z
— 4 , what is the value of the followimg expressions?
b. x | | y & & z
c. ( x & & y ) || z
d. ( x || y ) & & z
e. ( x & & z ) || y
18. Simplify the following expressions by removing the !
parentheses: operator and the
a. ! ( x < y )
b. ! ( x > = y )
c . ( x == y )
* d . ( x != y )
e. ! ( l ( x > y ) )
19. If x = -2, y = 5, z 0, and t -4,
ing expressions?
= =
- what is the value of each of the follow-
.
a x + y < z + t
. -
b x 2 * y + y < z * 2 / 3
.
c 3 * y / 4 % 5 && y
d. t | | z < ( y + 5 ) && y
e. ! ( 4 + 5 * y >= z 4 ) && ( Z _2 )
20. 11 originally x 4,
= y = 0, and z - what is the value of x, y, and z after
»
executing the following code ?
if (x 1 = 0)
y = 3;
continued
r
Chapter 5 Selection — Making Decisions 291 ]

else
z = 2;
21 . II originally x = 4, y 0, and z
= = 2, what is the value of x, y, and z alter
executing the following code? »
if ( z == 2)
y = 1;
else
x = 3;
.
22 If originally x = 4 , y = 0, and z = 2, what is the value of x, y, and z after
executing the following code?
if ( x && y )
x = 3;
else
y = 2;
23. If originally x = 4 , y 0 , and z

= = 2, what is the value of x, y, and z alter
|y
if ( x | z)
y = l;
else
z = 3;
.
if ( x )
if ( y )
z = 3;
else
z = 2;
= 2 , what is the value of x , y, and z after

25. If originally x = 4 , y = 0, and z
if (z == 0| |x && !y )
if ( iz)
y = 1?
else
x = 2;
26. If originally x = 0, y = 0, and z = 1 , what is the value of x, y, and z after

executing the following coder
if ( x )
if ( y )
if (z)
z = 3;
else
z = 2;
.
27 If originally x = 0, y = 0, and z = 1, what is the value of x, y, and z after
if ( z < x||y >= z && z == 1)

if (z && y)
y = 1?
else
x = 1;
28. If originally x = 0, y = 0, and z = 1 , what is the value of x, y, and z after

if (z = y )
{
y ++;
>
—
z ;
else
— x;
29. II originally x = 0, y = 0, and z = 1 , what is the value of x , y, and z after

if ( z = x < y)
—
{
x += 3;
y 1?
>
else
x = y++;
3 . If originally x
° executing
= 0, y = 0, and z = 1 , what is the value of x, y, and z after
the following code?
switch (x)
continued
case 0 : x = 2
y = 3
case 1 : x 4
default : y = 3
x = 1 n
>
. =
31 II originally x = 2 , y 1 , and z = 1 , what is the value of x, y, and z after
switch ( x )
{
case 0 : x = 2;
Y = 3;
case 1 : x = 4;
break ;
default:
y = 3;
x = 1;
>
.
switch ( x )
{
case 0 : x = 2
y = 3
break
case 1 : x 4
break
default: y = 3
x = 1
}
33. Evaluate the value of the following expressions:

a. tolower ( 'C' )
b. tolower ('?')
c. tolower ( 'c' )
d . tolower ( '5 ' )
34 . Evaluate the value of the following expressions:
a. toupper ( 'c • )
b. toupper ( 'C')
c. toupper ( '?' )
d . toupper ('7')
Programs
assign the value 1 to the variable best if
35. Write an if statement that will
i
the integer variable score is 90 or greater.

expression .
36 Repeat Problem 35 using a conditional
.
variable, num, if a float variable
37. Write the code to add 4 to an integer
amount , is greater than 5.4 .
variable flag is true .
38. Print the value of the integer num if the
39. Write the code to print either zero or not zero
based on the integer vari -
able num .
40. If the variable divisor is not zero, divide the variable dividend by
divisor , and store the result in quotient . II divisor is zero, assign it
to quotient . Then print all three variables . Assume that dividend and
divisor are integers and quotient is a double .
.
41 If the variable flag is true , read the integer variables a and b. Then cal -
culate and print the sum and average of both inputs.
42. Rewrite the following code using one ij statement :
if (aChar == • E * )
C ++ ;
if (aChar == 'E' )
printf ( "Value is E\n" );
43. Rewrite the following code fragment using one switch statement:
if (ch == 'E * |
|ch == e ' )
countE++ ;
else if (ch = = 'A'|
|ch == 'a 1 )
< 4? countA++
else if (ch == '1' |
|Ch == • i’ )
countI++;
else
print ("Error Not A , E ,
— o r I \a \ n" ) ;
.
44 Write a code fragment that tests the value of an
integer numl . If the value
is 10 , square numl . If it is 9 ,
read a new' value into numl . If it is 2 or 3,
multiply numl by 99 and print out the result. Implement your code using
nested if statements , not a switch .
45. Rewrite Problem 44 using a switch
statement .
46
' fhTtU °de fragment for the flowchart shown
C
that the variables x and y are integers and z is a
in Figure 5- 29. Assume
-
float point number.
i.
yread x & yj
x>0
y = x-1
z=2 * x
l z=y z=x
y=x+1
l
z=z+1
9l +
/ /
write
/ ** /
'
2
FIGURE 5 - 29 Flowchart for Problem 46
.
47 Write a function called smallest that, given three integers, returns the
smallest value.
_
48. Write a function called day_of week that , given an integer between 0
and 6, prints the corresponding day of the week. Assume that the first day
of the week (0 ) is Sunday.
49. Write a function called month _of _ year that , given an integer between 1
and 12, prints the corresponding month of the year.
50. Write a function called parkingCharge that , given the type of vehicle
( c for car, b for bus, t for truck ) and the hours a vehicle spent in the
parking lot , returns the parking charge based on the rates shown below.
car $2 per hour

bus $ 3 per hour
truck $4 per hour
5 1 Write a function to find the smallest of four integers .

.
Projects
52. Complete the incremental implementatio n of Program 5 - 13. f
irst write
, then the divide function ,
the code to implement the multiply function
and finally the print results function . The print results function should
option was requested and then print the

use a m’itch to determine which
results in the format shown below
.
number operator number = result

results in Chapter 4, but
You may pattern the function after the print
it will need extensive modification .
’s grade . It reads three test
53. Write a program that determines a student
100 ) and calls a function that calculates and
scores ( between 0 and
returns a student ’s grade based on the following rules :
a. If the average score is 90% or more , the grade A

is .
b. If the average score is 70 % or more and less than 90%, it checks the
third score. If the third score is more than 90 % , the grade is A; other-
wise, the grade is B.
.
c If the average score is 50% or more and less than 0 %, it checks the
"
average of the second and third scores. If the average of the two is
greater than 70%, the grade is C; otherwise, it is D .
d . If the average score is less than 50 percent , then the grade is F.
The program’s main is to contain only call statements. At least three
subfunctions are required: one to read scores, one to determine the
grade, and one to print the results.
54. In Program 4 - 7 , "Strange College Fees, ” we wrote a program to calculate
college fees. Modify this program for Typical College. At Typical College,
the students pay a fee of $ 10 per unit for up to 12 units; once they have
paid for 12 units, they have no additional per- unit fee . The registration
fee remains $ 10 but is assessed only if courses are taken in the term .
55 . Given a point , a line from the point forms an angle with the horizontal axis
to the right of the line. The line is said to terminate in one of four quad -
rants based on its angle ( a ) from the horizontal , as shown in Figure 5- 30.
II I
Quadrants
I 0 < a < 90
90 < a < 180
III: 180 < a < 270
III IV IV: 270 < a < 360
FIGURE 5 - 30 Quadrants for Project 55

Write a program that determines the quadrant , given a user input -

angle. Use a function to read and validate the angle . Note: If the angle is
exactly 0°, it is not in a quadrant hut lies on the positive X-axis ; if it is
-
exactly 90°, it lies on the positive Y axis; if it is exactly 180°, it lies on the
negative X-axis ; and if it is exactly 270°, it lies on the negative Y-axis. Test
your program with the following data:
0°, 48.3°, 90°, 179.8°, 180°, 186°, 270°, 300°, and 360°
36. I low many values of the variable num must be used to completely test all
branches of the following code fragment ?
if (num > 0)
if ( value < 25 )
{
value = 10 * num ;
if ( num < 12 )
value = value / 10;
} // if value
else
value = 20 * num ;
else
value = 30 * num;
57. Write a program that asks the user to enter the current date and a per -
son’s birth date in the form month , day, year. The program then calcu -
lates the person ’s age in integral years. Use separate lunctions to enter
the dates ( pass by address) , calculate the person s age, and print the
’
results. lest your program with the following dates: 1 1 / 14 / 1957 , 5 / 10/
1989, and 1/5/2000.
58. Write a C program to calculate the parking fare for customers who park
their cars in a parking lot when the following information is given :
a. A character showing the type of vehicle: C for car, B for bus ,
T for truck .
b. An integer between 0 and 24 showing the hour the vehicle entered
the lot .
c. An integer between 0 and 60 showing the minute the vehicle entered
the lot .
d. An integer between 0 and 24 showing the hour the vehicle leit the lot .
e. An integer between 0 and 60 showing the minute the vehicle
leit
the lot.
This is a public lot . To encourage people to park for a short period ot
time, the management uses two different rates for each type
of vehicle,
as shown in fable 5 11- .
First Rate Second Rate

Vehicle
$0.OO/hr first 3 hr $ 1.50/hr after 3 hr
CAR
$ 1.OO/hr first 2 hr $ 2.30 / hr after 2 hr
TRUCK
$2.OO/hr for first hr $3.70/hr after first hr
BUS
TABLE 5- 11 Rates for Project 58
No vehicle is allowed to stay in the parking lot later than midnight; it

will be towed away .
The input data consist of a character and a set of four integers rep-
resenting the type of vehicle and the entering and leaving hours and
minutes. But these pieces of data must he input into the computer in a
.
user - friendly way In other words, the computer must prompt the user to
enter each piece of data as shown below. ( Color indicates typical data.)
Type of vehicle? C
Hour vehicle entered lot (0
Minute vehicle entered lot (0
-- 2460))?? 2314
Hour vehicle left lot
Minute vehicle left lot
(0
(0
-- 6024 )?)? 818
The output format is shown below.
PARKING LOT CHARGE
Type of vehicle: Car or Bus or Truck

TIME IN- XX : XX
-
TIME OUT XX : XX
PARKING TIME XX:XX

ROUNDED TOTAL XX
TOTAL CHARGE $XX.XX
Pro8ram , ,st first calculate the actual time spent in the park-
in I /
°. ,
f re
TU
° C
Jc e* ^*caliscumeans using modulo arithmetic to handle
ti
tZ
^
shownV I
* "
\
'
Y ,* * ^
11 ate is in many ways, one of which is
“ h e P H i n, l o, u* h e - y. .
.
a Compare the minute portion of the leaving and
the entering time.
If the first one is smaller than the
second,
Add 60 to the miminute portion of the leaving time.
Subtract 1 from the hour portion of the
leaving time.
b. Subtract the hour portions.
c. Subtract the minute portions.

d . Since there are no fractional hour charges, the program must also
round the parking time up to the next hour before calculating the
charge . The program should use the switch statement to distinguish
between the different types of vehicles.
A well -structured program design is required . A typical solution will
use several functions besides main . Before you start programming, pre-
pare a structure chart . Run your program six times with the data shown
in Table 5 - 12.
Test Type Hour In Minute In Hour Out Minute

Out
1 C 12 40 14 22
2 B 8 20 8 40
3 T 2 0 3 59
4 C 12 40 16 22
5 B 8 20 14 20
6 T 2 0 12 0
TABLE 5- 12 Test Data for Project 58
59. This program is a simple guessing game . The computer is to generate a

random number between 1 and 20. The user is given up to five tries to
guess the exact number. After each guess, you are to tell the user if the
guessed number is greater than , less than , or equal to the random num’-
ber. If it is equal , no more guesses should he made . If the user hasn t
guessed the number after five tries, display the number with a message
that the user should know it by now and terminate the game.
A possible successful dialog:
I am thinking of a number between 1 and 20.

Can you guess what it is? 10
Your guess is low. Try again: 15
Your guess is low. Try again: 17
Your guess is high. Try again: 16
Congratulations! You did it.
A possible unsuccessfu l dialog :
I am thinking of a number between 1 and 20.

Can you guess what it is? 10
continued
: 20
Your guess is low. Try again
high. Try again: 10
Your guess is
Your guess is low. Try again
: 18
Your guess is high. Try again
: 12
Sorry. The number was 15.

You should have gotten it by now.
Better luck next time.
Your design For this program should include a separate function to

get the user’s guess, a function to print the unsuccessful message, one to
print the successful message , and one to print the sorry message.
60. Write a program that , given a person ’s birth date ( or any other date in the
Gregorian calendar ), will display the day of the week the person was horn.
To determine the day of the week , you will first need to calculate the
day of the week for December 31 of the previous year. To calculate the
day for December 31 .> use the following formula .
year - 1 year - 1 fyear - 1

( y e a r - l ) x 365 + + %7
4 100 400
The formula determines the day based on the values as shown below.
Day 0: Sunday
Day 1: Monday
Day 2: Tuesday
Day 3: Wednesday
Day 4: Thursday
Day 5: Friday
Day 6: Saturday
Once you know the day for December 31 , you simply calculate the
days in the year before the month in question. Use a switch statement to
make this calculation. ( Hint: Use case 12 first , and then fall into case 11,
10 . . . 2 . ) If the desired month is 12 , add the number of
days for November
( 30 ) . If it is 11, add the
number of days for October ( 31 ) . If it is 3, add
the number of days for February ( 28).'lf it is 2, add the number of days
or January ( 31 ) . If you do not
use a break between the months, the
switch will add the days in each month
before the current month .
o t is gure, add the day in the
current month and then add the
resu t to t e day code for
December 31 . I his number modulo seven is
the day of the week.
There is one more refinement. If the
current year is a leap year, and
Th
tSlrt
I he ffollowing
i| r
3te S a ter February
' *
, -
you need to add 1 to the day code.
formula can be used to dete rmine if the year is a leap year.
( l (year % 4) && (year
1 100 ) )
» II 1 ( year % 400 )
iitl
x
Your program should have a function to get data from the user,
1
another to calculate the day of the week , and a third to print the result.
To test your program , run it with the following dates:
a . February 28, 1900, and March 1 , 1900
.
h February 28, 1955 , and March 1 , 1955 iw
c. February 28, 1996 and March

, 1 , 1996
d . February 28, 2000, and March 1 , 2000
e. December 31, 1996
f . The first and last dates of the current week.
6 I . Write a program that calculates the change due a customer by denomina -
tion ; that is , how many pennies, nickels , dimes, etc. are needed in
change. The input is to be the purchase price and the size of the hill ten -
dered by the customer ( $100, $ 50 , $ 20, $ 10 , $ 5 , SI ).
-
62. Write a menu driven program that allows a user to enter five numbers
and then choose between finding the smallest , largest , sum , or average.
The menu and all the choices are to be functions. Use a switch statement
to determine what action to take. Provide an error message il an invalid
choice is entered .
Run the program five times, once with each option and once w i t h an
invalid option. Each run is to use the following set ol data:
16, 21, 7, 54 , 9
63. Write a program that tests a user entered character and displays its clas -
-
sification according to the ASCII classifications shown in Figure 5 - 25 ,
“ Classifications of the Character Type.” Write the program starting at the
top of the classification tree and display all classifications that the char-
acter belongs to. For instance, if the user enters a digit , you should
dis-
play that it is printable, graphical , alphanumeric , and a digit .
(
64. Write a program to compute the real roots of a quadratic equation ax
+
bx + c = 0 ) . The roots can he calculated using the following formulas:
ip
xl
-b+ Jb 2
- 4 ac and \2 =-
b - Jb 2
- 4 ac
2a 2a
Your program is to prompt the user to enter the constants ( a . b. c . It

)
is then to display the roots based on the following rules :
a . If both a and b are zero, there is no solution.

b . If a is zero, there is only one root ( -c / b ) .
are no real roots.
c. If the discriminate ( b - 4 ac ) is negative, there
2
d . For all other combinations, there are two roots.

Test your program with the data in fable 5 - 13.
a b
3 8 5
-6 7 8
0 9 -10
0 0 11
TABLE 5- 13 Test Data for Project 64
4
Repetition
The real power of computers is in their ability to repeat an operation or a
series of operations many times. This repetition , called looping, is one of the
basic structured programming concepts. In this chapter, we discuss looping
and introduce different looping constructs. First we define the basic concepts
of loops, including a most important concept: how to stop a loop. We then
present the three different loop constructs and take you through the C imple -
mentation of these three constructs. In this discussion , we include some
basic loop applications. After discussing the basic loop constructs, we intro -
duce an implicit loop construct , recursion . We conclude with a discussion of
some of the software engineering implications of loops.
Objectives
To understand basic loop concepts:
pretest loops and post- test loops
loop initialization and updating
event and counter -controlled loops
To understand and be able to select the best loop construct for a given
problem.
To write programs that use the while, for , or do. . . while statements.
To understand the basic concepts and usage of recursion algorithm
s.
To understand and be able to determine the efficiency of an algorithm
through an analysis of its looping constructs.
303
304 Section 6.2 Pretest and Post-test Loops
6.1 Concept of a loop

The concent of a loop is shown in the flowchart -
segment in Figure 6 1. b
is repeated over and over again . It never stops.
this flowchart , the action
Since the loop in Figure 6- 1 never stops , it is the computer version of the
perpetual motion machine. The action ( or actions ) will be repeated forever.
We definitely do not want this to happen . We want our loop to end when the
work is done . To make sure that it ends , we must have a condition that con-
trols the loop. In other words, we must design the loop so that before or after
each iteration , it checks to see if the task is done. II it is not done, the loop
repeats one more time ) if the task is done, the loop terminates. I his test is
known as a loop control expression.
An action or a
series of actions
FIGURE 6 - 1 Concept of a Loop
6.2 Pretest and Post- test Loops

We have established that we need to test lor the end of a loop, hut where
—
should we check it before or after each iteration ? Programming languages
allow us to check the loop control expression either before or after each itera -
tion ol the loop. In other words, we can have either a
-
pre or a post - test ter-
minating condition. In a pretest loop , the condition is checked before we
start and at the beginning of each iteration .
If the test condition is true, we
execute the code; if the test condition is false , we terminate the loop.
Pretest Loop
In each iteration, the control
.
expression is tested first If it is true, the loop
continues; otherwise, the loop is
terminated .
Post-test Loop
In each iteration, the loop action
(s) are executed. Then the control expression is
tested. If t is true, a new iteration
is started; otherwise, the loop terminates .
In the post -test loop , we always
execute the action at least once. The loop
» 1 control expression is then tested If tU .
the expression
• . . f
is true , the loop repeats; if
the expression is false, the loop terminates. The flowcharts in Figure 6- 2 show
these two loop types.
V An action or
series of actions
true
An action or
series of actions
—(
true \
Condition \M
\ \ false
( a) Pretest Loop ( b) Post - test Loop
FIGURE 6 - 2 Pretest and Post- test Loops
Consider an example of pretest and post-test looping. Imagine that you

are ready to start your daily exercises. Your exercise program requires you to
do as many push - ups as possible. You can check your limit using either a pre-
test or a post-test condition . In the pretest strategy, you first check to see if
you have enough energy to start . In the post- test strategy, you do one push - up
and then you test to see if you have enough energy to do another one. Note
that in both cases the question is phrased so that if the answer is true you
continue the loop. The two strategies are shown in Figure 6-3.
I
Energy
? Do one
push up -
true
I
Do one / \
push-up
=\ Energy
|false
/
(a) Pretest Loop (b) Post- test Loop
FIGURE 6 - 3 Two Different Strategies for Doing Exercises
As you can see, in the first strategy, you may not do any push - ups
. If you
for at least one push - up ,
are tired when you start and don ' t have the energy
you are done. In the second strategy , you must do at least one push - up. In
loop , the action may he done zero , one , or more
other words , in a pretest
306 Section 6.3 Initialization and Updating
is done one or more times. This major

-
times in a post - test loop, the action
test loop , which you must clearly
difference between a pretest and a post -
understand , is shown in Figure 6-4 .
[V
Body 1
exit
In a pretest loop, the body

may not be executed.
/
\
true _I_J\Test
false
In a post -test loop, the body
must be executed at least once.
exit
(a) Pretest ( b ) Post - test
FIGURE 6 - 4 Minimum Number of Iterations in Two Loops
6.3 Initialization and Updating

In addition to the loop control expression, two other processes , initialization
and updating, are associated with almost all loops.
Loop Initialization
Belore a loop can start , some preparation is usually required . We call this
preparation loop initialization . Initialization must be done before the first
execution of the loop body. It sets the stage lor the loop actions. Figure 6- 5
shows the initialization as a process box before the loop.
Initialization may be explicit or implicit . Explicit initialization is much
more common. When the initialization is explicit , we include code to set the
beginning values ol key loop variables. Implicit initialization provides no
direct initialization code; rather, it relies on a preexisting situation to control
the loop.
Loop Update
How can the condition that controls the
loop be true for a while and then
change to falser I he answer is that
something must happen inside the body
u oo|) to t tange the condition .
oop. For example , in the loops shown
Otherwise , we would have an infinite
'
izg2l “ 'Tr Jrfr
chanaes th
;“ j “T “ “ “
^
' ar l
' " " P
°
in Figure 6- 3, you gradually lose vour
y U Cannot continue the push - ups. This
update . Updating
The
J „.
each
h» cu*
deaden ,
, .
£;£iar
u | is nt in
t ^^ . «* *
ail
7H!
Initialization
I
tv Initialization
1
I
*
\ false
/
( Test Y- Action(s)
i
I true
Action(s)
Updating
I
Updating
s—(\ _7\ Test
false
exit exit
(a) Pretest Loop (b) Post -test Loop
FIGURE 6 - 5 Loop Initialization and Updating
The concepts of initialization and updating can be applied to our previ-

ous push - up example. In this case, initialization is created by nutrition
, an
implicit initialization. During each push - up, some of the initial energy is con -
sumed in the process and your energy is reduced, which updates your energy
level. This is shown in Figure 6 - 6.
I tv
1
Create k
Create
Energy Energy I
*
( y r
Do One
Push - up
Energy
I
r
[true Reduced
Do One
Push - up I
- gy
1
Energy Ener
Reduced true OK ?
false
exit
exit
( a) Pretest Loop (b) Post -test Loop
FIGURE 6 - 6 Initialization and Updating for Exercise

308 Section 6.4 Event- and Counter-Controlled Loops
6.4 Event- and Counter- Controlled Loops

in a loop limit test can be sum -
AII the possible expressions that can be used
marized into two general categories : event - controlled loops and counter-
controlled loops.
Event- Controlled Loops

In an event -controlled loop , an event changes the control expression from
true to false. For example, when reading data , reaching the end of the data
changes the expression from true to false. In event -controlled loops , the
updating process can be explicit or implicit. If it is explicit , such as finding a
specific piece of information , it is controlled by the loop. If it is implicit, such
as the temperature of a batch of chemicals reaching a certain point , it is con -
trolled by some external condition . The event -controlled loop is shown in
Figure 6- 7.
Initialize Initialize
condition condition
false
Condition
true
Action( s )
Update
I
Action(s)
I condition I
Update
condition Condition
true
false
!! I
exit exit
(a) Pretest Loop (b) Post- test Loop
FIGURE 6 - 7 Event-controlled Loop Concept
Counter Controlled Loops

-
we use a
UP , or ,, WhiCh
controlled loop is shown in Figure 6-8.
““” “
counting down . The counter-
L
"
I
Set count
I
Set count K
to 0 toO
— </ count < n V -

I true
Action( s)
i
Action( s)
i
increment
count i
increment
count i true
count < n
false
exit exit
(a) Pretest Loop ( b) Post- test Loop
FIGURE 6 - 8 Counter - controlled Loop Concept
Loop Comparison
The number of iterations of a loop is given as w. In a pretest loop, when we
come out of the loop, the limit test has been executed n + 1 times. In a
post - test loop, when we come out of the loop, the limit test has been executed
only n times. A summary of the two different loop concepts is shown in Table 6- 1 .
Pretest Loop Post- test Loop
Initialization: 1 Initialization: 1
Number of tests: n+ 1 Number of tests : n
Action executed: n Action executed: n
Updating executed: n Updating executed: n
Minimum iterations: 0 Minimum iterations: 1
TABLE 6 - 1 Loop Comparisons
6.5 Loops in C
C has three loop statements: the while, the for , and the do. . . while
. I he first two
are pretest loops, and the do. . . while is a post - test loop . We can use all ol them
for event - controlled and counter -controlled loops . I he while and do . .. while are
most commonly used for event-controlled loops and
, the for is usually used for
counter-controlled loops. These loop constructs
are shown in Figure 6 -9 .
310 Section 6.5 Loops in C
for do . ..while
Pretest Loop Pretest Loop Post -test Loop
FIGURE 6 - 9 C Loop Constructs
Note that all three of these loop constructs continue when the limit con -
trol test is true anti terminate when it is false. This consistency of design
makes it easy to write the limit test in C. On the other hand , general algo-
rithms are usually written just the opposite because analysts tend to think
about what will terminate the loop rather than what will continue it . There-
fore, we must complement or otherwise modify the limit test when we write
our program . This is one place where " De Morgan ’s Rule , ” discussed in
Chapter 5 , is very handy.
The while Loop

The while loop is a pretest loop. It uses an expression to control the loop.
Because it is a pretest loop, it tests the expression before every iteration of the
loop . The basic syntax of the while statement is shown in Figure 6- 10. No
semicolon is needed at the end of the while statement. When you see a semi -
colon , it actually belongs to the statement within the while statement .
i
I
.false
expression while ( expression)
true statement
statement
I
(a) Flowchart
(b) Sample Code
FIGURE 6 - 10 The while Statement
L
single
^
sinole statement; that is, the *
n Figure 6 - 1 shows that the loop body iis a
body of the loo p must be one , and only one
TH
.
statement II we want to include multiple statements in the body, we must put
them in a compound statement ( block) This concept is shown in Figure 6- 1 1.
.
1\
while ( expression )
{
Action
Action
Action
} // while
( a) Flowchart (b) C Language
FIGURE 6 - 11 Compound while Statement
Process-Control Loops
Perhaps the simplest loop is the loop that never ends. While it is virtually
never used in data processing, it is common in process -control loops such
as
network servers, environmental systems , and manufacturing systems. A sim -

ple process-control system that might he used to control the temperature in a
building is shown in Program 6- 1.
PROGRAM 6 - 1 Process- control System Example

1 while (true )
2 {
3 temp = getTemperature();
4 if (temp < 68 )
5 turnOnHeater( ); turnOffAirCond( )
6 else if ( temp > 78 )
7 turnOnAirCond( ); turnOffHeater()
8 else
9 {
10 turnOffHeater( );
11 turnOffAirCond();
12 } // else
13 } // while true
, |
^ êffectlîr
a |it cons
Program 6- 1 Analysis The limit test in this s mp ejpjro
m
' ^ statement strl, is always
L : 6
con
however, that this is not a good
^
^ ^
f
°r
ment were used?

'
would be ie Program 6- 1 if the following while state-
while (0)
Because the limit condition is a constant false, the loop

would never start. Obvi-
ously, this would be a logic error.
Print Loops
A common while loop is shown in Program 6- 2. In this case we want to print
a series of numbers in descending order. To keep from running off the end of
the line, we have added a test to write a newline when ten numbers have
been written.
PROGRAM 6- 2 A while Loop to Print Numbers

1 /* Simple while loop that prints numbers 10 per line.
2 Written by:
3 Date:
4 */
6
7 int main (void )
8 {
10 int num;
11 int lineCount;
12
13 // Statements
14 printf ("Enter an integer between 1 and 100: " );
15 scanf ("%d" , &num); // Initialization
16
17 // Test number
18 if (num > 100)
19 num = 100;
20
21 lineCount = 0;
22 while (num > 0)
23 {
24 if (lineCount < 10 )
25
lineCount++;
continued
i
1
PROGRAM 6- 2 A while Loop to Print Numbers (continued )
26 else
27 {
28 printf( " \n " );
29 lineCount = 1 ;
30 > // else
31
32
printf("%4d" , num );
> // while
— // num -- updates loop
33 return 0;
34 > // main
Results:
Enter an integer between 1 and 100: 15
15 14 13 12 11 10 9 8 7 6
5 4 3 2 1
Program 6 - 2 Analysis Find the basic elements of a loop in this program. First, look for loop initialization,
then the limit test, and finally the update.
The initialization is done by asking the user to enter num in Statement 14 Note
how we make sure that the user enters a number that is not too large. If the number is
over 100, we simply set it to 100, which is the maximum we told the user to enter. The
limit test is in Statement 22. As long as num is greater than 0, we continue printing out
the number series. The update is hidden in Statement 31; after printing num, we sub-
tract 1 from it. Although this is a common C programmer trick, the program would
have been easier to read if we had put the update (num — ) on the line after the print.
That way it would have been obvious.
File Loops
One of the most common loop applications in any language is reading data
until all the data have been processed—that is , until the end ol file is
reached. In C, the scan/ function returns the system constant EOF when it
detects an end- of- file.
Consider this application of the scanf feature: suppose we want to read
and process a list of numbers from the keyboard. We type all the numbers,
each one on a separate line . At the end, we can type end- of- file ( <ctrl+d> in
UNIX or <ctrl+ z> in DOS ) . This loop logic is shown in the next example.
ioResult = scanf ( "%d" , & a);

while ( ioResult 1 = EOF )
{
// Action: Process data
ioResult = scanf ( "%d" , &a );

> // while
7
this. Lets simplify it . In the revised

But no one codes a while loop like
version of the loop, we have moved the scan / to the conditional expression in
itself . We can do this because the initialization and the
the while statement
are now both self -contained
update are identical. The initialization and updating .
change is shown below
parts of the while statement . This
while ((ioResult = scanf ("%d", &a)) != EOF )

{
// Process data
> // while
But we can simplify this loop even further. Because the scanf function
returns a value , we can test for end-of -file in the while expression . We don ' t
need the variable, ioResult, so we can simply use the function value and
discard it after it has been checked. The result is the standard C loop for
reading and processing data from a file shown in Program 6- 3, which adds a
list of integers read from the keyboard and displays their sum .
PROGRAM 6- 3 Adding a List of Numbers

1 / * Add a list of integers from the standard input unit
2 Written by:
3 Date:
4 */
5 iinclude < stdio.h >
6 int main ( void )
7 {
9 int x;
10 int sum = 0;
11
12 // Statements
13 printf( "Enter your numbers: < EOF > to stop. xn " );
14 while ( scanf( "%d", & x ) 1 = EOF )
15 sum += x ;
16 printf ( "\nThe total is: % d \ n ", sum ) ;
17 return 0 ;
18 } / / main
Results:
Enter your numbers: < EOF > to stop.
15
22
3 Ad
The total is: 40
Chapter 6 Repetition ^
315
Program 6 - 3 Analysis Note that a compound statement (block) is not needed in the while loop, because
the
addition can be done in one statement. Another important point is that the statement
to print the sum is outside the loop. Since the user can see the input values on the
screen, all we need to show is the sum.
The for Loop

I he for statement is a pretest loop that uses three expressions. The first expres-
sion contains any initialization statements, the second contains the limit test -
expression , and the third contains the updating expression . Figure 6- 12 shows
a flowchart , and an expanded interpretation , for a sample for statement .
\
expri
expr2 M
\ exprl
"
| true expr2 \Vais-

statement | true
statement
t expr3
f
( a ) Flowchart (b) Expanded Flowchart
for (exprl; expr2; expr3)

statement
FIGURE 6 - 12 for Statement
Expression 1 is executed when the for starts. Expression 2 is the limit test
-
expression. As shown in the expanded flowchart , it is tested before every itera
tion . Remember that since the /or is a pretest loop, the body is not executed if
the limit condition is false at the start of the loop.
Finally, Expression 3 is the update expression . It is executed at the end of
each loop. Note that the code in the for statement must be expressions. This
means that you cannot use statements, such as return , in the for statement
.
itself. Like the while statement , the for statement does not need a semicolon
C allows the limit test in Expression 2 to he a variable . It also allows us to
change the value of the variable during the execution of the loop . We
do not
recommend changing it in the loop , however . It is not a good structured pro-
gramming style and can lead to errors and perpetual loops .
The body of the for loop must he one, and only one, statement
. II we
the body , we must code them in
want to include more than one statement in
nt with a compound statement is shown

a com pound statement . A for stateme
in Figure 6- 13.
exprl \fafse for (exprl ;

expr2 ;
Ik
ixpr3
expr2 expr3)
{
true
Action Action
Action Action
Action
} // for
exit
(a) Flowchart ( b ) C Language
FIGURE 6 - 1 3 Compound for Statement
Unlike some languages, C allows the loop control expression to he con -

trolled inside the for statement itself . This means that the updating of the
limit condition can be done in the body of the for statement. In I act , Expres-
sion 3 can he null and the updating controlled entirely within the body of the
loop. This is not a recommended structured programming coding technique,
although it is required in some situations.
Although the for loop can be used anywhere a pretest loop is required , it
is most naturally used for counter-controlled loops ( see “ Counter-Controlled
Loops in Section 6.4 ). Its self -contained design makes it ideal for
count logic .
A for loop is used when a loop is to be executed a .

known number of times We
can do the same thing with a while
loop, but the for loop is easier to read and
more natural for counting loops .
1 et s compare the while and the for loops
.
representation
/ ation update code,
. Figure 6- 14 shows a graphical
ol each side by side. Note that the
/or loop contains the initial
and limit test in one statement . This makes for very
readable code . All the control steps,
*
updating are done in one place. This is a -

initialization , end of - loop test , and
variation of the structured program-
ming concepts of encapsulation,
in which all code for a process is placed in
one module. Another way of
looking at it is that a for loop communicates bet -
ter and is more compact . * 1
k
U
initialize
for ( initialize ;
while ( expression ) expression ;
update)
1
{ {
action action
action action
update
// while i
L
> // for
FIGURE 6 - 14 Comparing for and while Loops
Now let ' s solve the same problem using both a while and a for loop. I he
code shown below contains a loop to read 20 numbers f rom the keyboard and
find their sum. This can he done both in a while loop and a for loop. But as
you can see, the for loop is more self-documenting.
l = 1?
sum = 0; sum = 0;
while ( i <= 20 ) for ( i = 1 ; i <= 20; i++)
{ {
scanf( "%d " , &a); scanf ("%d" , &a);
sum += a ; sum += a ;
i ++; } // for
} // while
C99 modified the /or syntax to allow the definition of loop variables to be
done in the for statement itself. This syntax is shown in the next example.
Note, however, that definitions of the for statement are local to the loop body;
they are not in scope when the loop terminates. If we need them after the
loop, they must be defined in the local declarations section.
Example: Print Number Series

To demonstrate the for loop, let’s write a program that asks the user
for a
the series of numbers starting at one and continuing
number and then prints
is in Program 6 - 4.
up to and including the user-entered number. The code
PROGRAM 6 - 4 Example of a for Loop

limit.
-
1 /* Print number series from 1 to user specified
2 Written by:
3 Date:
continued
PROGRAM 6 - 4 Example of a for Loop (continuedI
4 */
6 int main ( void )
7 {
9 int limit;
10
11 // Statements
12 printf ( "\nPlease enter the limit:
13 scant ("%d", &limit);
14 for (int i = 1; i <= limit ; i++ )
15 printf("\t%d\n" , i);
16 return 0 ;
17 } // main
Results:
Please enter the limit: 3
1
2
3
Program 6 - 4 Analysis This simple program is the model for many looping functions. Let's look at three sim-
ple modifications to it. First, how would you print only odd numbers? This requires a
change only to the update in the for statement.
for ( i = 1 ; i <= limit ; i += 2)
Now let's change it to print the numbers backward. In this case, all the statements
in the for statement must be changed, but the rest of the program is still unchanged.
for (i
For the
= limit; i > 1; i )
—
final example, let's print the numbers in two columns, the odd numbers in
the first column and the even numbers in the second column. We must modify the
update statement in the for statement and also the print statement as shown below.
-
for (i 1; i <= limit ; i += 2 )
printf("\t%2d\t%2d\n" , i, i +
i );
We have used tabs and width specifications to
a e secon print va ue (i + l) is an
align the output in columns. Note
expression . It does not change the value of i.
There is no side effect, it is just a value.
Nested for Loops
.
Any statement, even another or
ement . Ls ng nested loops
/ loop, can be included in the body of a for
can create some interesting applications. Let’s
I\
look at a very simple one here . We will give other examples in Section 6.8
"Looping Applications,” later in the chapter.
Program 6- 5 uses a nested loop to print a series of numbers on
multiple lines.
11
PROGRAM 6 - 5 A Simple Nested for Loop
1 /* Print numbers on a line.
2 Written by:
3 Date:
4 */
6
7 int main (void )
8 {
9 // Statements
10 for ( int i = 1; i <= 3 ; i++ )
11 {
12 printf( "Row %d: ” , i );
13 for (int j = 1; j<= 5; j++)
14 printf( "% 3d ” , j);
15 printf("\n");
16 > // for i
17 return 0;
18 } // main
Results:
Row 1: 1 2 3 4 5
Row 2: 1 2 3 4 5
Row 3: 1 2 3 4 5
The do . . . while Loop

The do . .. while statement is a post - test loop. Like the while and for loops it
,
also uses an expression to control the loop, hut it tests this expression after
the execution of the body. The format of the do. . . while statement is shown
in
Figure 6- 15.
The body of the do. .. while loop must he one, and only one, statement If
.
in the body , we must put them in a
we need to include multiple statements
com pound statement . The second example in Figure 6 - 15 shows the logic
statement .
flow and code for a do. ..while statement that uses a compound
Look carefully at the code block in the second example . Note that the do and
follows the
the while braces are aligned . Note also that the while expression
the
line . This is a good style because it makes it easy for
brace on the same
the statement . Finally, note that the do .. . while is concluded
reader to see
constructs that we
with a semicolon. This differs from the other looping
have seen.
!
Flowchart Sample Code
do
statement
statement
expression
true
while (expression ) ;
[false
do
tv
Action {
Action
Action
Action
Action
T
J
Action
expression
true'
} while ( expression);
false
FIGURE 6 - 15 do. . . while Statement
Lets look at a simple program that uses a loop to print five numbers We .
code two loops, first a while loop and then a do. . . while . Although one loop is
a pretest and the other is a post- test, both print the same series. The code is
seen in Program 6- 6.
PROGRAM 6 - 6 Two Simple Loops

1 / * Demonstrate while and do...while loops.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 <
10 int loopCount ;
11
12 // Statements
continued
PROGRAM 6 - 6 Two Simple Loops ( continued )

13 loopCount = 5 ;
14 printf(" while loop : " );
15 while ( loopCount > 0)
16
17
printf ( "%3d" , loopCount );
printf( " \n\n" );
—
18
19 loopCount = 5;
20 printf("do...while loop: " );
21 do
22
23
printf ("%3d " , loopCount );
while ( loopCount > 0);
—
24 printf("\n");
25 return 0 ;
26 > // main
Results
while loop 5 4 3 2 1
do...while loop: 5 4 3 2 1
Program 6-6 Analysis Because the do. ..while limit test isn't done until the end of the loop, we use it when we
know that the body of the loop must be done at least once. To demonstrate the impact
of the two loops, study the code in Figure 6- 16. in the while loop, the message is not
printed, because the limit condition is tested first. In the do...while loop, even though
the limit test is false, the message is printed because the message is printed before the
limit test.
^ Pretest
nothing
^
prints
while ( false) do
{ {
printf( " Hello World " ); printf( " Hello World " );
} // while } while (false);
^Post -tesP' N
Hello... prints,
FIGURE 6 - 16 Pre - and Post- test Loops
Programmers commonly use the do ...while loop in data validation

to
that requires
make a program robust . For example, consider an application
do . . .while loop as
that we read an integer between 10 and 20. We can use the
shown in the next example .
do
{
printf ("Enter a number between 10 & 20:
" );
scant ( "%d " , &a);

} while (a < 10 | | a > 20);
To demonstrate the do...while , let’s rewrite Program 6-3, “ Adding a List
of Numbers. ” The modified code is shown in Program 6- 7.
PROGRAM 6 -7 Adding a List with the do. . . while

1 / * Add a list of integers from the standard input unit
2 Written by:
3 Date:
4 */
6
7 int main (void )
8 {
10 int x;
11 int sum = 0;
12 int testEOF ;
13
14 // Statements
15 printf( " Enter your numbers: <EOF> to stop. Vn" );
16 do
17 {
18 testEOF = scanf( " %d " , & x );
19 if ( testEOF != EOF )
20 sum += x ;
21 } while (testEOF != EOF );
22 printf ( "\nTotal: %d\n " , sum );
23 return 0 ;
24 } // main
Results:
Run 1:
Enter your numbers: <EOF> to stop.
10 15 20 25 Ad
Total: 70
Run 2:
Enter your numbers: <EOF> to
stop.
Ad
Total: 0
rn
Program 6 - 7 Analysis Since the do. ..while always executes the body of the loop at -
least once, we had to
make some changes. Compare Program 6- 3 and Program 6 7 carefully. Note that
the scanf is no longer in the loop limit test. Since the limit test is after the loop body,
the scanf must be moved to the beginning of the loop. The result of the input is saved
in a new variable, testEOF. Finally, before we can add the value we read to the
accumulator, we must ensure we are not at the end of the file. The add statement is
therefore guarded by an if statement. Although Program 6-7 is a little less efficient
than Program 6 - 3, it does the same job.
The Comma Expression

A comma expression is a complex expression made up of two expressions
separated by a comma. Although it can legally he used in many places, it is
most often used in for statements. The expressions are evaluated left to right
.
The value and type of the expression s are the value and type of the right
expression; the left expression is included only for its side effect . The comma
operator has the lowest priority of all operators, priority 1 .
The following statement is a modification of the for statement code
shown on page 317. It uses a comma expression to initialize the accumulator,
sum, and the index, i, in the loop. In this example, the value ol the comma
expression is discarded. This is a common use of the comma operator.
for (sum = 0, i 1; i <= 20; i++)

{
scanf( " %d " , &a);
sum + = a ;
} // for
Comma expressions can he nested. When they are nested, all expression
values other than the last are discarded. Figure 6- 17 shows the format of a
nested comma expression .
expression , expression , expression
FIGURE 6 - 17 Nested Comma Expression
n is the
A final word of caution. Remember that the value of the expressio
nded, if use
value of the rightmost expression. Although recomme
it is not you
second expressio n in a for loop , make sure that
a comma expression for the
the loop control is the last expression.
Lets use the comma expression to demonstrate
the difference between
. , Loop Comparisons,"
the while and the do . . .while As shown in Table 6
- 1
the only difference is the number of limit tests

that are made. Program 6-6
, ,
shows that the same job can be done by either loop . Program 6-8 uses both
loops to count from 1 to 10. It uses the comma expression to count the num-
ber of limit tests in each loop.
PROGRAM 6 - 8 Comparison of while and do...while

1 /* Demonstrate while and do...while loops.
2 Written by:
3 Date:
4 */
6
7 int main (void )
8 {
10 int loopCount ;
11 int testCount;
12
13 // Statements
14 loopCount = 1 ;
15 testCount = 0;
16 printf( "while loop:
17 while ( testCount++, loopCount <= 10)
18 printf( " % 3d " , loopCount++ );
19 printf( "Loop Count: %3d \n " , loopCount );
20 printf( " Number of tests: % 3d \n " , testCount );
21
22 loopCount = 1;
23 testCount = 0;
24 printf( "\ndo...while loop: " );
25 do
26 printf( "%3d " , loopCount++ );
27 while ( testCount++, loopCount <= 10);
28
29 printf( "\nLoop Count: % 3d \n " , loopCount );
30 printf( " Number of tests: % 3d \ "
n , testCount );
31 return 0;
32 > // main
Results:
while loop: 12 3 4 5 6
Loop Count: 7 8 9 10
11
Number of tests: 11
do...while loop:
.
12 3 4 5 6 7 8 9 10
> continue
A
PROGRAM 6 - 8 Comparison of while and do . . .while ( continued )

Loop Count: 11
Number of tests: 10
Program 6 - 8 Analysis Look at Statements 17 and 27 carefully. They both contain comma expressions. This
technique of combining the counter and the limit test in one expression ensures that
the count will be accurate. Because the value of the whole comma expression is the
value of its last expression, however, the limit test must be coded last.
The results demonstrate that both loops count from one to ten. Since they are doing
exactly the same job, we expect that the loop bodies would also execute the same num-
ber of times. As predicted in Table 6 - 1 the only difference is in the number of tests: the
while loop control expression was evaluated 11 times; the do...while control expres-
sion was evaluated only 10 times.
6.6 Loop Examples

This section contains several short examples of loop applications. Each pro-
gram demonstrates one or more programming concepts that you will find
helpful in solving other problems.
for Loops
Example: Compound Interest
One classic loop problem is calculating the value of an investment . We want
to know the value of an investment over time, given its initial value and
annual interest rate. Program 6 - 9 displays a compound interest table.
PROGRAM 6 - 9 Compound Interest

1 /* Print report showing value of investment.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
10 double presVal ;
11 double futureVal ;
12 double rate;
13 int years ;
14 int looper ;
continuec
326 Section 6.6 Loop Examples
PROGRAM 6- 9 Compound Interest (continued )

15
16 // statements
17 printf("Enter value of investment:
18 scanf ("%lf", &presVal);
19 printf(" Enter rate of return ( nn.n ): '• ) ;
20 scanf ("%lf", &rate);
21 printf( "Enter number of years:
22 scanf ("%d", &years);
23
24 printf("\nYear ValueXn" );
25 printf("===* ======\n");
26 for (futureVal = presVal, looper = 1?
27 looper <= years;
28 looper++)
29 {
30 futureVal = futureVal * ( 1 + rate / 100.0);
31 printf( "%3d % 11.21f\n" , looper , futureVal );
32 > // for
33 return 0 ;
34 | > // main
Results:
Enter value of investment: 10000
Enter rate of return (nn.n ): 7.2
Enter number of years: 5
Year Value
1 10720.00
2 11491.84
3 12319.25
4 13206.24
5 14157.09
Program 6 - 9 Analysis This program uses a for loop to calculate

the value of the investment at the end of
each year. Each iteration adds the current year's
interest to the investment and then
prints its current value.
... how we prompted the user far input, especially the decimal return rate.
Note
Things like percentage rates can be confusin
Make sure you give the user a sample
g to enter. Is 7.2% entered as 7.2 or .072?
of how the data should be entered .
Now study the way we created a
underscore the captions and spaces withcaption for the reports, using equal signs to
umns. This rather simple technique
width specifications to align the values in col
makes the results quite readable
One final point. When defining multiple variables iin one statement, they must all
have the same type. Notice that we defined the for loop variables in the definition
sections. To define multiple variables in a for statement, they must all be the same
type because they are treated as a type list.
I' d
Example: Right Triangle
Lets write a program that will print a scries of numbers in the form of a right
triangle. We ask the user to enter a one-digit number. Each line, from the
first to the limit entered by the user, then prints a number series from one to
the current line number. For example, if a user enters 6, the program prints
1
12
123
1234
12345
123456
The flowchart and pseudocode for the loop are shown in Figure 6 - 18.
/
1
line = l false \
\ 1 set line to 1 k
2 loop (line not greater than limit)
^
» ine++ line <=
\ limit I
| true
> 1 set num to 1
2 loop (num not greater than line)
1 print num
ium = false 2 increment num
num <= M 3 end loop
line 4 advance line
I true 5 increment line
j print r\un\
J 3 end loop
I
j print "\n“ jf
( a) Flowchart ( b ) Pseudocode
FIGURE 6 - 18 Print Right Triangle Flowchart and Pseudocode
The completed program is shown in Program 6 - 10.

PROGRAM 6- 10 Print Right Triangle Using Nested for

1 /* Print a number series from 1
right triangle.
-
to a user specified limit
2 in the form of a
3 Written by:
4 Date:
5 * /
7
8 int main (void )
9 {
11 int limit;
12
13 // Statements
14 // Read limit
15 printf("\nPlease enter a number between 1 and 9: ");
16 scanf("%d" , &limit);
17
18 for (int lineCtrl = 1; lineCtrl <= limit ; lineCtrl++)
19 {
20 for (int numCtrl = 1;
21 numCtrl <= lineCtrl ;
22 numCtrl++)
23 printf( " % ld" , numCtrl );
24
25 printf("\n");
26 } // for lineCtrl
27 return 0;
28 > // main
Results:
Please enter a number between 1 and 9: 6
1
12
I 123
1234
12345
123456
Program 6 - 10 Analysis Program 6- 10 demonstrates the

concept of a loop within a loop. Note how we use
two for loops to print the triangle.
The first or outer for controls how many lines we
will print. The second or inner for writes
nested loops is a very important
the number series on one line. This use of
programming concept.
Also note the name we used in the
control for loops. We often do this
loops. Often programmers will use i and j to
code is when meaningful names
ourselves. But notice how much more readable the
are used. When we use lineCtrl
rather than i in
u
the outer loop, we know that the for is controlling the number of lines we are printing.
Likewise, the name numCtrl clearly tells the reader that the for loop is controlling the
numbers.
EXAMPLE 6- 1 Print Rectangle

Now lets write a program that prints the triangle pattern in the previous
example, but now filled out with asterisks to form a rectangle . For example, if
a user enters 6 , Program 6 - 1 1 prints the rectangle shown in the results.
PROGRAM 6 - 11 Print Number Series Using User - specified Limit

1 /* Print number series from 1 to a user-specified limit
2 in the form of a rectangle.
3 Written by:
4 Date:
5 */
7
8 int main (void)
9 {
11 int limit ;
12
13 //Statements
14 // Read limit
15 printf( "Please enter a number between 1 and 9: **)?
16 scanf ( " %d " / &limit );

17
18 for ( int row = 1; row <= limit ; row++)
19 {
20 for ( int col = 1 ; col <= limit ; col++)
21 if (row >= col )
22 printf( "%d " , col );
23 else
24 printf( » * ••)?
25 printf ( " \ n " );
26 } // for row ...
27 return 0 ;
28 > // main
Results:
Please enter a number between 1 and 9: 6
!*** *
22****
123* * *
continuer
( continued )
PROGRAM 6 - 11 Print Number Series Using User - specified
Limit
1234**
12345*
123456
Program 6 - 1 1 Analysis This program is an interesting variation of the previous program. Compare its inner
loop to the inner loop in Program 6- 10. This is the only part of the program that is
different. The first thing you should note is that the limit test expression is different; it
always goes to the maximum number of print positions. Within the inner loop, we test
the column number (col) to determine how many digits we print on the line. If the
expression is true, we print a digit. If it is false, we print an asterisk .
Example: Print Month

.
As a final example of a for loop, lets print a calendar month In Program 6- 12,
the function printMonth receives only the start day of the month—Sunday is
!
0, Monday is 1 ... Saturday is 6—and the number of days in the month. This
is all that the program needs to print any month of the year.
PROGRAM 6 - 12 Print Calendar Month

1 /* Test driver for function to print a calendar month.
2 Written by:
3 Date:
4 */
6
7 // Prototype Declarations
8 void printMonth (int startDay , int days);
9
<4 11 {
12 // Statements
13 printMonth ( 2, 29 ); // Day 2 is Tuesday
14 return 0;
15 > // main
16
17 /* === printMonth
18
============
Print one calendar month.
19 Pre startDay is day of week
20
relative
to Sunday (0)
21 days is number of days in month
22 Post Calendar printed
23 */
24 void printMonth (int
25
startDay , int days )
{
i con l i MM
i!i
w
1
PROGRAM 6 - 12 Print Calendar Month ( continued )

27 int weekDay ;
28
29 // Statements
30 / / print day header
31 printf("Sun Mon Tue Wed Thu Fri Sat\n" );
32 printf( " \n " );
33
34 // position first day
35 for (weekDay = 0; weekDay < startDay ; weekDay++)
36 printf( " ")?
37
38 for (int dayCount = 1 ; dayCount < = days ; dayCount++ )
39 {
40 if ( weekDay > 6 )
41 {
42 printf( " \n " );
43 weekDay = 1 ;
44 > // if
45 else
46 weekDay++;
47 printf( " % 3d " , dayCount );
48 } // for
49 printf( " \n \n " );
50 return;
51 } // printMonth
Results:
Sun Mon Tue Wed Thu Fri Sat
1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29
-
Program 6 - 12 Analysis This program is interesting for two reasons: first, it requires two for loops, one to posi
tion the printing for the first day of the month and one to print the dates. Second , the
logic to control the days of the week is simple yet efficient. In an effort
to eliminate
would use the modulo statement
one variable (weekDay), many programmers
shown below to determine the day of the week .
(dayCount + startDay ) % 7
-j
Although this logic works, it is very inefficient. Using a separate variable to control
is much more efficient. It also
the days of the week requires simple addition, which
avoids another poor programming style: using one variable to control two processes.
Although using one variable to control two processes may work initially, it often leads
to subtle bugs when the program is changed.
One final point. Examine the initialization in the two for loops. Do you see a differ-
ence? Because weekday is used in both loops, it cannot be declared local to the first
loop. We must declare it in the definition section.
Never use one variable to control two processes.
while LOOPS
This section contains two examples of while loops. The first prints the sum of
the digits in an integer entered from the keyboard. The second prints a num-
her backwards.
Example: Print Sum of the Digits

Program 6- 13 accepts an integer from the keyboard and then prints the num-
her of digits in the integer and the sum of the digits.
PROGRAM 6 - 13 Print Sum of Digits

1 /* Print the number and sum of digits in an integer.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8
I?
10 int number;
11 int count = 0 ;
12 int sum = 0;
13
14 // Statements
15 printf("Enter an integer: )
" ?
16 scanf ("%d" , number );
17 ^
printf("Your number is: % \
d n\n " , number );
18
19 while (number •* 0)
20 {
21 count++;
22 sum += number % 10 ;
23 number /= IQ ;
continued
PROGRAM 6- 1 3 Print Sum of Digits ( continued )

24 } // while
25 printf( "The number of digits is: %3d \n " , count );
26 printf( "The sum of the digits is: %3d\n" , sum);
27 return 0 ;
28 > // main
Results:
Enter an integer: 12345
Your number is: 12345
The number of digits is: 5

The sum of the digits is: 15
of
Program 6 - 1 3 Analysis This problem requires that we "peel" off one digit at a time and add it to the total
the previous digits. We can solve this problem in a couple of ways, but by far the
most straightforward is to use the modulus operator ( % ) to extract the rightmost digit
and then to divide the number by ten to remove the right digit. For example, given the
number 123, we first extract the 3 by
123 % 10
and then eliminate it by dividing by 10 to give 1 2. Note that since both the dividend
and the divisor are integers, the result is an integer. We loop until there is only one digit
left, at which time any digit divided by ten will result in zero and the loop terminates.
Example: Print Number Backward

Now let’s look at an example of a loop that prints a number backward . We can
solve this problem in several ways. Perhaps the easiest is to simply use mod -
ulo division , as shown in Program 6 - 14 .
PROGRAM 6- 14 Print Number Backward

1 /* Use a loop to print a number backward.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
10 long num ;
11 int digit ;
12
continued
PROGRAM 6- 14 Print Number Backward (continued)

13 I I Statements
14 printf("Enter a number and I'll print it backward: ");
15 scant ("Id" , &num);
16
17 while (num > 0)
18 {
19 = num % 10;
digit
20 printf("%d" , digit);
21 num = num / 10 ;
22 > // while
23 printf("\nHave a good day.\n");
24 return 0;
25 > // main
Results:
Enter a number and I'll print it backward: 12345678
87654321
Have a good day.
do . . . while LOOPS
This sections contains an example of a program that uses a do . . . while
-
loop . While it is a simple problem , it demonstrates a classic data validation
technique.
EXAMPLE 6 - 2 Data Validation

Now we ll write a program that reads an integer consisting ol only zeros and
ones ( a binary number ) and converts it to its decimal equivalent Provide a
function that ensures that the number entered is a binary number. The
.
design for this program is shown in Figure 6 19. -
convertTo
Decimal
getNum binaryTo
Decimal
Validate
Binary
firstDigit powerTwo
FIGURE 6 - 19 Design for Binary to Decimal
ii
HU *
1 his design follows a classic program design: input process output. The
— —
input function is getNum; the process function is binaryToDecimal; and the
output function is printResults. Study the design of binaryToDecimal.
Note that it uses a loop and calls firstDigit and powerTwo in turn within
the loop . How would you implement this loop? Think about it for a minute ,
VI
and then look at Statement 65 in Program 6- 15 to see how we did it . II you
are not sure of your binary arithmetic and binary c o n v e r s i o n s, study
Appendix I ), " Numbering Systems. "
PROGRAM 6 - 1 5 Convert Binary to Decimal

1 /* Convert a binary number to a decimal number .
2 Written by:
3 Date:
4 */
6 # include <stdbool.h >
7
9 long long getNum ( void );
10 long long binaryToDecimal ( long long binary );
11 long long powerTwo ( long long num );
12 long long firstDigit (long long num);
13 bool validateBinary ( long long binary );
14
16 {
18 long long binary ;
19 long long decimal;
20
21 // Statements
22 binary = getNum ( );
23 decimal = binaryToDecimal ( binary );
24 printf("The binary number was: %lld " , binary);
25 printf( " \nThe decimal number is: % lld " , decimal );
26 return 0;
27 > // main
28
29
30 This function reads and validates a binary number
31 from the keyboard.
32 Pre nothing
33 Post a valid binary number is returned
34 */
continued
PROGRAM 6 - 15 Convert Binary to Decimal (continued )

35 long long getNum (void)
36 {
38 bool isValid ;
39 long long binary ;
40
41 // Statements
42 do
43 {
44 printf( "Enter a binary number ( zeros and ones ): " );
45 scant ( "%lld" , sbinary );
46 isValid = validateBinary ( binary );
47 if ( lisValid )
48 printf( "\a\aNot binary. Zeros/ones only.\n\n");
49 > while ( lisValid );
50 return binary ;
51 > // getNum
52
53 /*
54 Change a binary number to a decimal number.
55 Pre binary is a number containing only 0 or 1
56 Post returns decimal number
57 */
58 long long binaryToDecimal ( long long binary )
59 {
61 long long decimal;
62
63 // Statements
64 decimal = 0;
65 for (int i = 0; binary != 0; i++ )
66 {
67 decimal += firstDigit ( binary ) * powerTwo ( i);
68 binary /= 10 ;
69 } // for i
70 return decimal ;
71 > // binaryToDecimal
72
73
74
validateBinary ====
Check the digits in a binary
75 number for only 0 and 1
Pre binary is a number to be validated
76 Post returns true if valid
77 ; false if not
*/
continued
PROGRAM 6- 15 Convert Binary to Decimal (continued )

78 bool validateBinary (long long binary )
79 {
80 // Statements
81 while ( binary != 0 )
82 {
83 if (!( binary % 10 == 0 || binary % 10 == 1 ))
84 return false;
85 binary /= 10 ;
86 > // while
87 return true;
88 > // validateBinary
89
90 /* === ===== powerTwo ==========
91 This function raises 2 to the power num
92 Pre num is exponent
93 Post Returns 2 to the power of num
94 */
95 long long powerTwo ( long long num )
96 {
98 long long power = 1;
99
100 // Statements
101 = 1 ; i <= num; i++ )
for ( int i
102 power *= 2;
103 return power ;
104 > // powerTwo
105
106
107 This function returns the right most digit of num
108 Pre the integer num
109 Post the right digit of num returned
110 */
111 long long firstDigit ( long long num )
112 {
113 / / Statements
114 return (num % 10);
115 > // firstDigit
Results:
Enter a binary number (zeros and ones): 10001
The binary number was: 10001
The decimal number is: 17
338 Section 6.7 Other Statements Related to Looping
, note the data validation that we use

Program 6- 1 5 Analysis This program has several interesting aspects. First of nothing but zeros and ones.
to ensure that the binary number that we read consists
We used this series of modulus and divide statements previously in Program 6 - 13 and
^Thei
allows us
^note how we enclosed the call to the validation
to keep reading input until the user gives us
function in a do. . .while that
a binary number. Again, note
how we display an error message when the number is not valid. This is a standard data
validation technique.
Next, study the binaryToDecimal function that converts the binary number to
its decimal value. Note that when we extract a digit, it is either a zero or a one. We
then multiply the extracted digit by 2 raised to the digit position we are currently evalu-
ating, which gives us the binary value of that digit's position in the binary number. The
value is then added to the decimal number. Of course, if the digit is a zero, then the
product is zero and the value is unchanged. It is only when the digit is a one that we
add to the decimal number.
Finally, note that throughout this program we used long long for the binary and
decimal number. We did this because the decimal representation of a binary number
can get very big very fast. On a personal computer, long might not be able to hold a
large binary number.
6.7 Other Statements Related to Looping

Three other C statements are related to loops: break , continue , and goto. The
last statements, the goto , is not valid for structured programs and therefore is
not discussed in this text .
break
I he first statement is break . We previously saw the break statement when we
discussed snitch in Chapter S . In a loop, the break statement causes a loop
to terminate. It is the same as setting the loop ’s limit test
to false . If we are in
a series of nested loops , break terminates only the
inner loop —the one we are
currently in. Figure 6 - 20 shows how break transfers control out of an inner
for loop and continues with the next statement in the while . Note that break
statement needs a semicolon.
I he bieak statement can he used in any of the
loop statements—while ,
’ —
for and do. ..while and in the selection switch
structured programming limits its use to the switch
statement . However, good
statement. It should not
)e use in an \ of the looping statements. If
you feel you must use break , reex-
amine your resign. You will
usually find that you have not properly structured
your logic . 1 1 1 J
u
while (condition )
{
for ( •••i
The break statement takes\
us out of the inner loop
(the for loop ). The while I
{ loop is still active. 1 S' '
if (otherCondition)
—
break ; !
} // for
// more while processing
> 7/ while
FIGURE 6 - 20 break and Inner Loops
Program 6- 16 shows two examples of poor loop situations and how to

restructure them so that the break is not needed . The for statement shows a
never-ending loop . As coded , there is no way to terminate the loop without
the break . Although the break works, better documentation and style puts the
limiting condition as the second expression in the for statement . After all,
that is the use of the limit expression in the first place .
PROGRAM 6 - 16 The for and while as Perpetual Loops

1 // A bad loop style // A better loop style
2 for ( ; ? ) for ( ; !condition ; )
3 { {
4
5 if (condition) } // for
6 break ;
7 } // for
1 while ( x ) while ( x && Icondition )

2 { {
3
4 if (condition ) if ( Icondition )
5 break ; ... ,
6 else } // while
7
8 } // while
Even if the break condition is in the middle of a block , it can he removed

to the limit condition , as shown in the while example
in Program 6 16. Note -
that in this case, we complement the condition , which then executes the else
logic when the limit condition has not been reached .
340 Section 6.7 Other Stotements Related to Looping
Sometimes the limit condition is so

complex that it cannot easily be put
. In these cases , a flag is used . A flag is a logical variable
in the limit condition ,
that tracks the presence or absence of an event
. In this case, the flag s set to
1 to indicate that the end of the loop has been reached. This is shown in the
while example in Program - 6 17 .
PROGRAM 6 -17 Using a break Flag

1 breakFlag = 0 ;
2 while ( 1 breakFlag )
3 {
4
5 if ( x & & ! y || z ) / / Complex limit test
6 breakFlag = 1;
7 else
8 ... i
9 > // while
On the other hand , too many flags can make a function overly complex.
The use of break and flags needs to be tempered with simplicity and clarity of
logic . Finally, note that the flag is called breakFlag , which is a generic name.
You should choose a more descriptive name, such as accountFlag or
timeLimitFlag .
continue
I he continue statement does not terminate the loop but simply transfers to
the testing expression in while and do. .. while statements and transfers to the
updating expression in a for statement . These jumps are shown in Figure 6-21 .
Although the transfer is to different positions in pretest and post - test loops,
both can be logically thought of as a jump to the end of the loop s body.
while ( expr) do for (exprl ; expr2; expr3)

{ { {
continue ; continue; continue ;
} // while } while (expr );

7 } // for
FIGURE 6 - 21 The continue Statement
I ht use of the continue statement

is also considered unstructured pro-
grammmg. If you think that you need a
well structured. A little study will
continue , vour algorithm may not be
show how to eliminate it . Program 6- 18
contains a common continue
example found in i many textbooks . In this
-
T
I unction , the assignment is to read data and return the average of nonzero
numbers read . In other words, it skips zeros. Note how simply reversing the
conditional test eliminates the need for the continue .
PROGRAM 6 - 18 continue Example

1 float readAverage ( void ) float readAverage ( void )
2 { {
3 // Local Declarations // Local Declarations
4 int count = 0 ; int count = 0;
5
6 int n; int n;
7 float sum = 0; float sum = 0;
8
9 // Statements // Statements
10 while( scanf( "%d " ,&n ) while( scanf( " %d " ,& n )
11 != EOF ) != EOF )
12 { {
13 if ( n == 0) if ( n != 0)
14 continue; {
15 sum += n ; sum += n ;
16 count++; count++ ;
17 > // while > // if
18 > // while
19 return ( sum / count); return ( sum / count );
20 } // readAverage > // readAverage
6.8 Looping Applications

In this section , we examine four common applications for loops: summation ,
product , smallest or largest , and inquiries. Although the uses for loops are vir-
tually endless , these problems illustrate many common applications. Note
that a common design runs through all looping applications. With few excep -
tions , each loop contains initialization code, looping code,
and disposition
code . Disposition code handles the result of the loop , often by printing it ,
other times by simply returning it to the calling function .
Summation
very easily. But how can
As you have seen , we can add two or three numbers
we add many numbers or a variable series of numbers ? The solution is sim -
add operator in a loop . The concept of summation is graphically
ple: use the
shown in Figure 6- 22 .
342 Section 6.8 Looping Applications
sum
^ conditiony^
exit
Summation Product
FIGURE 6 - 22 Summation and Product Loops
A sum function has three logical parts: ( I ) initialization of any necessary

working variables, such as the sum accumulator; ( 2 ) the loop, which includes
the summation code and any data validation code ( for example, “only nonzero
numbers are to be considered ’); and ( 3 ) the disposition code to print or
return the result. Program 6 - 19 is a loop function that reads a series of num-
bers from the keyboard and returns their sum. In each loop, we read the next
number and add it to the accumulator, sum. A similar application, counting,
is a special case of summation in which we add 1 to a counter instead of add-
ing the number we read to an accumulator.
PROGRAM 6 - 19 Sum to EOF Function

1 /* Read a series of numbers , terminated by EOF , and
i* 2 return their sum to the calling program.
3 Pre nothing
4 Post data read and sum returned
5 */
6 int sumEOF (void )
7 {
9 int nmbr;
10 int sum ;
11
12 // Statements
13 sum = 0;
14 printf ("Please enter
an integer: ")?
15
16 while (scanf( "%d " , &nmbr ) !
= EOF)
tiki continue*
PROGRAM 6- 19 Sum to EOF Function ( continued )

17 {
18 sum += nmbr;
19 printf( " Next integer < EOF> to stop: " ) ?
20 > // while
21 return sum ;
22 } // sumEOF
Powers
Just as we can add a series of numbers in a loop, we can perform any mathe-
matical operation . A product loop is useful for two common applications:
raising a number to a power and calculating the factorial of a number. I or
example, Program 6-20 shows a function to return xn . Notice that this func -
tion also includes initialization logic to validate the parameter list . If either of
the parameters is invalid , we return zero as an error indicator.
PROGRAM 6- 20 Powers Function

1 /* Raise base to an integral power , exp. If the
2 exponent is zero, return 1.
3 Pre base & exp are both positive integer values
4 Post return either (a) the result of raising the
5 base to the exp power
6 or ( b) zero if the parameters are invalid
1 */
8 int powers ( int base, int exp )
9 (
11 int result = 1;
12
13 // Statements
14 | exp < 0 )
if ( base < 1 |
15 // Error Condition
16 result = 0 ;
17 else
18 ( int i = 1 ; i <= exp ; i++)
for
19 result *= base;
20 return result;
21 } // powers
: initial -
The summation and powers examples demonstrate a subtle point
is not the same for every problem .
ization must he based on the application ; it
For summation , the initialization sets the accumulator to 0 . I or product -
to 1 . II we use 0.
based applications, such as powers, we must initialize it
then the result is 0 because 0 multiplied by anything is 0 .
To find the sum of a series, the result is

initialized to 0 ; to find the product
.
of a series, the result is initialized to 1
Smallest and Largest

We often encounter situations in which we must determine the smallest or
largest among a series of data . This is also a natural looping structure.
We can write a statement to find the smaller of two numbers. For example,
result = a < b ? a : b;
But how can we find the smallest of several numbers? We simply put the
same statement inside a loop . Each iteration then tests the current smallest
to the next number. If this new number is smaller than the current smallest ,
we replace the smallest . In other words, we determine the smallest number
by looping through a series while remembering the smallest number the loop
has found . This concept is shown in Figure 6- 23.
smallest
INT MAX —
exit
smallest largest
FIGURE 6 - 23 Smallest and Largest Loops
Io find the smallest ol a series, the

o smallest to
initialization sets the initial value
the largest possible value for its type . In a C program , the
argest value lor an integer is given
lmits i rarv ( units . h ) . The
_ .
as INT MAX which is found in the
loop then proceeds to read a series of num -
ers and tests each one against
the previously stored smallest number,
bmce smallest starts with the
maximum integer value , the first read
*
entirety
entirely
^
on the data being read .
smallest value found .
-
CCOmcs smallest Thereafter, the result depends
The disposition simply returns the
Program 6- 21 is the C implementation
of smallest .
u
PROGRAM 6 - 21 Smallest to EOF Function

1 /* Read a series of numbers, terminated by EOF , and
2 pass the smallest to the calling program ,
3 Pre nothing
4 Post data read and smallest returned
5 */
6 int smallestEOF ( void )
7 {
9 int numln;
10 int smallest;
11
12 // Statements
13 smallest = INT MAX ; // requires < limits.h >
14
15 printf( "Please enter an integer: " );
16
17 while ( scanf( " % d " , &numln ) != EOF )
18 {
19 if ( numln < smallest )
20 smallest = numln ;
21 printf( "Enter next integer <E0F> to stop: ");
22 } // while
23 return smallest;
24 } // smallestEOF
We can find the largest number by simply reversing the less than operator
in the expression, making it greater than. We would also need to set the vari -
able, renamed largest, to INT MIN. _
To find the largest, we need to initialize the result ( a variable named

largest)variable to a very small number, such as INT_ MIN.
To find the smallest, we need to initialize the result (a variable named
smallest)to a very large number, such as INT_ MAX.
Inquiries
An inquiry is simply a question asked of the computer program. In program
-
ming, we often encounter one of two basic inquiry types any : and all. We use
the inquiry type any when we have a list of data and we want to know at
il
least one of them meet a given criteria . The answer to the inquiry is yes if one
or more data meet the criteria. The answer is no if none of the data meet the
criteria.
(I
... when we have a

We use .all list of data and we want to make sure that
each and every one of them meet some specified criteria. I he concept of an}
and all in an inquiry is shown in Figure 6- 24.
-
result *- fe/se|
false
t
exit
any all
FIGURE 6 - 24 any and all Inquiries
The any Inquiry

lo determine il one item in a series satisfies a requirement , we can test them
using or as shown in the next example.
result = ( a == condition ) | | ( b == condition ) ;
II either or both ol the equal expressions are true , result is set to true.
Ibis works fine lor testing the first two items. To test the third item in the
series , however, we need to include the result ol the first two. This is easily
done with the next example.
result - result || (c == condition ) ;
Study this test carefully. If the previous tests were all false , then result
contains Juke hen the statement is executed . If
" the equal expression is false ,
tien resu t is assigned false , however, -
if the equal expression is true , then
result is assigned true. By placing this expression
in a loop , we can test all ol
the tallies in the series. For example ,
to test a series to determine if any of the
values are positive , we can use the following
code.
result = false ;
while ( s c a n f ( " % d M ,
{ ^ number )
result = result number > 0 ;
> / / while
111
,
‘ TV
Chapter 6 Repetition 34 /
I his code produces an accurate answer to the inquiry. However, it can be

very inefficient especially ii the series is long. For example, suppose that the
first item tested is greater than 0. In this case, it is not necessary to test the
rest of the list ; we already know the answer.
Program 6- 22 is an example of an any inquiry. It reads a series of num -
bers and checks to see if any of the numbers in the list are greater than zero.
I he function terminates and returns true as soon as a positive number
is read .
PROGRAM 6- 22 anyPositive to EOF Function

1 /* Read number series to determine if any positive.
2 Pre nothing
3 Post return true if any number > zero
4 return false if all numbers <= zero
5 */
6 bool anyPositiveEOF ( void )
7 {
9 bool anyPositive = false;
10 int numln ;
11
12 // Statements
13 printf( "Determine if any number are positive\n " );
14 printf( " Enter first number: " );
15 while ( scanf( " %d " , &numln ) != EOF )
16 {
17 anyPositive = numln > 0 ;
18 if ( anyPositive)
19 return true;
20 printf( "Enter next number:
21 > // while
22 return false;
23 } // anyPositiveEOF
Program 6 - 22 Analysis Note that anyPositive is a logical var iable [ bool] and is initialized to falseto
.
do not need to read the entire number series , we can use a simple test
Because we
we return true . If not, we
set anyPositive. If a positive number was entered,
prompt for the next number and loop back to the while expression .
To use the bool type , we must include the stdbool . h library in the program. This
bool and it to the C Boolean
library uses a define macro to declare the type equate
type, £>oo/. It also defines
_
true and false.
The all Inquiry

, we can test them
To determine if two items in a series satisfy a requirement
using and as shown in the next example.
7
result = (a -= condition) && (b

— condition );
set to true ; if either of

If both of the equal expressions are true, result is
them is false , the result is false . This works fine for testing the first two items.
To test the third item in the series , however , we need to include the result of
the first two. This is easily done with the next example .
result = result && (c

— condition );
Study this test carefully. If the previous tests were all true , then
result contains true when the statement is executed . If the equal expres-
sion is also true , then result is assigned true ; however, if the equal
expression is false , then result is assigned false . By placing this expres-
sion in a loop , we can test all of the values in the series . For example , to
test a series to determine if all of the values are positive , we can use the
following code.
result = true;
while (scant( " %d " , Snumber )
{
result = result && number > 0;
} / / while
Once again , for efficient processing we may not need to examine all of
the numbers in the series. As soon as a nonpositive number is found , we
know the result and can return false .
This function is seen in Program 6- 23.
PROGRAM 6 - 23 All Positive Function

1 / * Read number series , and determine if all are positive.
2 Pre nothing
3 Post return true if all numbers > zero
4 return false if any numbers <= zero
5 */
6 bool allPositiveEOF (void )
7 {
9 bool allPositive = true ;
10 int numln ;
11
12 // Statements
13 printf( "Determine if all
numbers are positive\n ");
14 printf( Enter first number: "
);
15 while ( allPositive && (scanf( "%d " , &
numln) != EOF ))
16 {
'
i cont mucL
UL
T
"
PROGRAM 6 - 23 All Positive Function ( continued )

17 allPositive = numln > 0;
18 if (!allPositive)
19 return false;
20 printf( " Enter next number:
21 > // while
22 return true;
23 > // allPositiveEOF
Once again you should study how we initialize the Boolean ( lags in these
two programs . To test for any item , we start with the flag false If and when it .
becomes true , the result has been determined . Similarly, to test for all items,
we initialize the Boolean flag to true . If and when it becomes false , the result
has been determined .
To answer an any inquiry, the result is initialized to false; to answer an all

inquiry, the result is initialized to true .
6.9 Recursion
In general , programmers use two approaches to writing repetitive algorithms.
One approach uses loops ; the other uses recursion . Recursion is a repetitive
process in which a function calls itself . Some older languages do not support
recursion . One major language that does not is COBOL.
Iterative Definition
To study a simple example , consider the calculation of a factorial. The facto -
rial of a number is the product of the integral values from 1 to the number.
This definition is shown in Formula 6- 1 .
factorial ( /2 ) = [I * ( /2 - !) * ( /? - 2) ... 3 * 2 * 1
if /2 = 0
if / 2 > 0 ]
FORMULA 6- 1 Iterative Factorial Definition
This definition is iterative. A repetitive function is defined iteratively

whenever the definition involves only the parameter ( s ) and not the func -
6 - 1 as
tion itself . We can calculate the value of factorial ( 4 using Formula
)
follows:
factorial (4) = 4 * 3 * 2 * 1 24
350 Section 6.9 Recursion
Recursive Definition
whenever the function appears
A repetitive function is defined recursively
itself . For example , the factorial function can be defined
within the definition
recursively, as shown in Formula 6 - 2 .
factorial ( n ) = [I * factorial (n - 1)
if n = 0
if n > 0 ]
FORMULA 6 - 2 Recursive Factorial Definition
The decomposition of factorial ( 3 ) , using Formula 2 , is shown in

Figure 6 - 25 . Study this figure carefully, and note that the recursive solu -
tion for a problem involves a two - way journey : first we decompose the
problem from the top to the bottom , and then we solve it f rom the bot -
tom to the top .
Factorial (3) = 3 * Factorial (2) Factorial ( 3) = 3 * 2=6

T
Factorial (2) = 2 * Factorial (1) Factorial (2) = 2 * 1=2
V
Factorial ( 1) = 1 * Factorial (0) Factorial ( 1 ) = 1 * 1=1
T
Factorial (0) = 1
FIGURE 6 - 25 Factorial ( 3 ) Recursively
Judging by the above example, the recursive calculation looks much

longer and more difficult. So why would we want to use the recursive
method ? We use it because, although the recursive calculation looks more
difficult when using paper and pencil , it is often a much easier and more ele-
gant solution when we use computers.
Additionally, it offers a conceptual
simplicity to the creator and the reader.
Iterative Solution
Let’s write a function to solve the
factorial problem iteratively. This solution
usually involves using a loop, as shown in
I
Program 6- 24 .
PROGRAM 6 - 24 Iterative Factorial Function

1 /* Calculate the factorial of a number using a loop.
2 There is no test that the result fits in a long.
3 Pre n is the number to be raised factorially
4 Post result is returned
5 */
6 long factorial ( int n )
7 {
9 long factN = 1;
10
11 // Statements
12 for ( int i = 1; i <= n; i++)
13 factN = factN * i;
14 return factN ;
15 > // factorial
Recursive Solution
I he recursive solution to factorial is shown in Program 6- 25 . I his program
does not need a loop ; the concept itself involves repetition .
PROGRAM 6 - 25 Recursive Factorial

1 /* Calculate factorial of a number using recursion.
2 There is no test that the result fits in a long.
3 Pre n is the number being raised factorially
4 Post result is returned
5 */
6 long factorial ( int n )
7 {
8 // Statements
9 if ( n == 0 )
10 return 1 ;
11 else
12 return ( n * factorial ( n - 1 ));
13 > // factorial
In the recursive version , we let the function factorial call itself , each
time with a different set of parameters. Figure 6- 26 shows the recursive exe -
cution with the parameters for each individual call.
Designing Recursive Functions

ol
All recursive functions have two elements: each call either solves one part
the problem or it reduces the size of the problem . In Program 6 - 25 ,
10 solves a small piece of the problem factorial ( 0 ) is I .
—
statement
the size of the problem by recur-

Statement 12, on the other hand , reduces
siveiy calling the
ively caning factorial with* n* - 1 . Once the solution to factorial '( n l )
me laciunai win r * -
is known , Statement 12 provides part of
the solution to the general problem
by returning a value to the calling function .
As we see in Statement 12 , the general part of the solution is the recur-
call Statement 12 calls its own function to solve the problem . We also
sive :
see this in Figure 6- 26. At each recursive call , the size of the problem is
reduced from the factorial of 3 , to 2 , 1 , and finally to factorial 0.
The statement that “ solves ” the problem is known as the base case. Every
recursive function must have a base case . The rest ol the function is known as
the general case. In our factorial example, the base case is factorial ( 0 ) ;
the general case is n * factorial ( n - 1 ) . The general case contains the
logic needed to reduce the size of the problem .
Every recursive call must either solve part of the problem or reduce the size
of the problem .
In the factorial problem , once the base case has been reached , the solu -
tion begins. The program has found one part of the answer and can return
that part to the next more general statement. Thus, in Program 6- 25, after
the program has calculated that factorial ( 0 ) is 1 , then it returns value 1.
That leads to solving the next general case .
factorial ( 1 ) -> 1 * factorial ( 0 ) -> 1 * 1 -» 1
The program now returns the value of factorial ( 1 ) to the more gen -
era I case , factorial ( 2 ) , which we know to be
factorial ( 2 ) —> 2 * factorial ( 1 ) -> 2 * 1 -> 2
As the program solves each general case in turn , the program can solve
the next higher general case, until it finally solves the most general case, the
original problem .
The following are rules for designing a recursive function:
1 . First ,
determine the base case.
2. I hen ,determine the general case.
3. Finally, combine the base case and general
case into a function .
In combining the base and general cases into a function , we must pay
careful attention to the logic . Each call must reduce the size of the problem
and move , t toward the base case. The base case,
when reached , must termi-
te without a call to the recursive function ; that is,
it must execute a return.
int main (void) tv

<
int n = 3;
long f ;
3
f = factorial( 3);
printfl4" %d\n" , f );
return 0;
> // ma .n int factorial( int n )
{
// Statements
6
if (n == 0)
return 1 ;
else
return 2
(n*factorial(n 1); -
} // fac
int factorial( int n )
{
// Statements
2
=
if ( n = 0)
return 1;
else
return 1
-
( n * factorial( n 1 );
} // facti
int factorial(int
{
*3k
// Statements
1
if (n == 0)
return 1 ;
else
return 0
-
( n*factorial( n 1 );
} // fact'
int factorial(int
{
tlk
// Statements
1
if (n == 0)
return 1;
else
return
-
(n*factorial(n 1);
} // factorial
FIGURE 6 - 26 Calling a Recursive Function

Fibonacci Numbers
Another example of recursion is a function that generates Fibonacci num-
bers. Named after an Italian mathematician , Leonardo Fibonacci, who lived
in the early thirteenth century, Fibonacci numbers are a series in which each
number is the sum of the previous two numbers ( Figure 6 - 27 .
)
Fib(4)
Fibn
Fibn 1- 0 Fibn-2
Fib n-2 0 Fib n-3 Fibn-3 0 Fibn-4
-
Fib n 3 [T| F b n-4 Fib(1) [T] Fib(O )
' 1 0
(a) Fib(n) (b) Fib( 4)
FIGURE 6 - 27 Fibonacci Numbers
I he first few numbers in the Fibonacci series are
0 , 1, 1 , 2 , 3, 5 , 8 , 13 , 21 , 34
lo start the series, however, we need to know the first two numbers. As
you can see from the above series, they are 0 and 1 . Since we are discussing
recursion , you should recognize these two numbers as the base cases.
We can generalize the Fibonacci series as follows:
Given:
Fibonacci0 = 0
Fibonaccij = 1
Then
Fibonacci = Fibonacci - i + Fibonaccin _ 2
The generalization of Fibonacci is shown in

Figure 6- 27. Figure 6-27 (a )
shows the components of Fibonacci, using a
general notation . Figure 6 - 27 ( b)
shows the components as they would be
called to generate the numbers in
the series.
Fo determine Fibonacci 4 we can
start at 0 and move up until we have
,.
’
.
the number
. .
U C can slart ut
. ihonacci4 and move down to zero. 1 he first
technique s used in the iterative solution ; *
the second is used in the recursive
solution , which is shown in Program
6-26.
PROGRAM 6- 26 Recursive Fibonacci

1 /* This program prints out a Fibonacci series.
2 Written by:
3 Date:
4 */
6
7 // Function Declaration
8 long fib ( long num );
9
11 {
13 int seriesSize ;
14
15 // Statements
16 printf( "This program prints a Fibonacci series.Xn " );
17 printf( "How many numbers do you want? " );
18 scanf ( "%d " , &seriesSize);
19 if ( seriesSize < 2 )
20 seriesSize = 2 ;
21
22 printf( "First %d Fibonacci numbers: \n " , seriesSize );
23 for (int looper = 0; looper < seriesSize ; looper++ )
24 {
25 if ( looper % 5 )
26 printf( " , %81d " , fib( looper ));
27 else
28 printf( "\n%81d " , fib( looper ));
29 >
30 printf( "\n" );
31 return 0;
32 } // main
33
34 / ===== ==== fib ===
35 Calculates the nth Fibonacci number.
36 Pre num identifies Fibonacci number
37 Post returns nth Fibonacci number
38 */
39 long fib ( long num )
40 {
41 // Statements
42 if ( num == 0|| num == 1 )
43 // Base Case
44 return num ;
continued
PROGRAM 6 - 26 Recursive Fibonacci (continued )
45
46
47
return ( fib (num
} // fib
- 1) + fib ( num - 2));
Results:
This program prints a Fibonacci series.
How many numbers do you want? 30
First 30 Fibonacci numbers:
0,
5,
1
8,
/1
13,
2,
21 ,
. 3
34
55, 89, 144 , 233, 377
610, 987, 1597, 2584, 4181
6765, 10946, 17711 , 28657, 46368
75025 , 121393, 196418 , 317811 , 514229
Program 6- 26 Analysis Compare fib in Program 6-26 with the solution in Figure 6- 27. To determine the
fourth number in the series, we call fib with num set to 4. To determine the answer
requires that fib be called recursively eight times, as shown in Figure 6-27, which
with the original call gives us a total of nine calls.
This sounds reasonable. Now, how many calls does it take to determine
Fibonacci?
The answer is 15 ( Table 6- 2). As you can see from Table 6 - 2, the number of calls goes
up quickly as we increase the size of the Fibonacci number we are calculating.
Calls Calls
1 1 11 287
2 3 12 465
3 5 13 753
4 9 14 1,219
5 15 15 1,973
6 25 20 21,891
7 41 25 242,785
8 67 30 2,692,573
9 109 35 29,860,703
10 177 40 331,160,281
TABLE 6- 2 Fibonacci Run Time
Table 6-2 leads us to the conclusion

numbers is not realistic. that a recursive solution to calculate Fibonacci
Chapter 6 Repetition 35 /
Limitations of Recursion
We have introduced only the briefest explanation of recursion in this section .
VVe have not attempted to demonstrate how recursion works. To understand
how it works, you will need to study data structures and concepts that are
beyond the scope of this text . 1
On the other hand , you do need to understand the two major limitations
ol recursion . F irst , recursive solutions may involve extensive overhead
because they use function calls. Second , each time you make a call , you use
—
up some ol your memory allocation . If the recursion is deep that is , if the
—
program has a large number of recursive calls then you may run out ol
memory. Both the factorial and Fibonacci numbers solutions are better devel -
oped iteratively.
Does this mean that iterative solutions are always better than recursive
functions? The answer is definitely no. Many algorithms are easier to imple -
ment recursively and are efficient. When you study data structures, you will
study many ol them . Unfortunately, most of those algorithms require data
structures beyond the scope of this text .
The Towers Of Hanoi

One classic recursion problem , the Towers of Hanoi , is relatively easy to fol -
low, is efficient , and uses no complex data structures. Let ’s look at it .
According to legend , the monks in a remote mountain monastery knew
how to predict when the world would end . They had a set ol three diamond
needles. Stacked on the first diamond needle were 64 gold disks of decreasing
size. The monks moved one disk to another needle each hour, subject to the
following rules:
.
1 Only one disk could he moved at a time.
2 . A larger disk must never be stacked above a smaller one .
3. One and only one auxiliary needle could be used for the intermediate
storage of disks.
The legend said that when all 64 disks had been transferred to the desti -
nation needle, the stars would be extinguished and the world would end .
Today we know' that we need to have 264 - 1 moves to do this task. Figure 6- 28
shows the Towers of Hanoi with only three disks.
This problem is interesting for two reasons. First , the recursive solution
is much easier to code than the iterative solution would he . This is often the
case with good recursive solutions. Second , its solution pattern differs from
the simple examples we have been discussing. As you study the Towers solu -
tion , note that after each base case, we return to a decomposition of the gen -
eral case for several steps. In other words, the problem is divided into several
subproblems, each of which has a base case, moving one disk .
Source Auxiliary Destination
FIGURE 6 - 28 Towers of Hanoi — Start Position
Recursive Solution For Towers Of Hanoi

To solve this problem we must study the moves to sec if we can find a pat-
tern . We will use only three disks because we do not want the world to end!
First , imagine that we have only one disk to move. This very simple case
involves only one step as shown in Case 1 .
Case 1 : Move one disk from source to destination needle .
Now imagine that we have to move two disks. First , the top disk is moved
to the auxiliary needle . Then the second disk is moved to the destination.
Finally, the first disk is moved to the top of the second disk on the destina-
tion . These three steps are shown in Case 2.
Case 2: Move one disk to auxiliary needle.

Move one disk to destination needle.
Move one disk from auxiliary to destination needle .
-
Figure 6 29 traces the steps for two disks.
A B C
Step 1
A B C
Step 3
FIGURE 6 - 29 Towers Solution for Two Disks
We are now ready to study the case for three

disks. Its solution is seen in
Figure 6- 30.
Towers (3,A ,C,B )
Towers ( 2, A ,B,C ) I Step 41 Towers ( 2,B,C, A )
Towers Towers Towers Towers

...
(1 A C B) ... ...
( 1 C B A ) ( 1 B A C) Step 6 . .
(1 A .C B )
Step 1 Step 3 Step 5 Step 7
n n n
A B C A
A
B C
Step 2 Step 3
Move one disk from source to destination.
Step 4
A B
Step 5
C A B
Step 6
C A 3
Step 7
4
FIGURE 6 - 30 Towers of Hanoi Solution for Three Disks
I he first three steps move the top two disks from the source to the auxil -
iary needle. (To see how to do this, refer to Case 2 . ) In Step 4 . we move the
bottom disk to the destination . We now have one disk in place. This is an
example of Case 1 . It then takes three more steps to move the two disks on
the auxiliary needle to the destination . These steps are shown in Case 3.
Case 3: Move two disks from source to auxiliary needle.

Move one disk from source to destination needle.
Move two disks from auxiliary to destination needle.
We are now ready to generalize the problem .

auxiliary needle. General Case

1 . Move n - 1 disks from source to . Base Case
destination needle.
2. Move one disk from source to
needle. General Case
3. Move n - 1 disks from auxiliary to destination
Our solution requires a function with

four parameters: the number of
disks to be moved , the source needle, the destination
needle, and the auxil -
moves in the generalization shown
iary needle. Using pseudocode, the three
above are then
1 Call Towers ( n 1, source , auxiliary , destination)

-
2 Move one disk from source to destination
3 Call Towers ( n - 1, auxiliary , destination source)
,
Study the third step carefully. After we complete the move of the first
disk , the remaining disks are on the auxiliary needle. We need to move them
-
from the auxiliary needle to the destination . In this case , the original source
needle becomes the auxiliary needle. Remember that the positions of the
-
parameters in the called function are source , destination , auxiliary . The call -
ing function must remember which of the three needles is the source and
which is the destination for each call .
We can now put these three calls together with the appropriate
print statements to show the moves . The complete function is shown in
Program 6 - 27 .
PROGRAM 6 - 27 Towers of Hanoi

1 /* Move one disk from source to destination through
2 the use of recursion.
3
4
Pre The tower consists of n disks source ,
destination , & auxiliary towers given
—
5 Post Steps for moves printed
6 */
7 void towers ( int n, char source ,
8 char dest, char auxiliary )
9 {
11 static int step = 0;
L
12
13 // Statements
14 printf( "Towers ( %d , %c , %c , %c )\n " ,
15 n , source, dest, auxiliary);
16 if ( n == 1 )
17 printf( "\t\t\t\tStep % 3d: Move from %c to %c\n" ,
18 ++step, source, dest);
19 else
continue
PROGRAM 6- 27 Towers of Hanoi ( continued )

20 {
21
22
towers ( n - 1 , source , auxiliary , dest);
printf( " \t\t\t\tStep %3d: Move from %c to %c \n ” ,
23 ++step, source, dest);
24 towers ( n - 1 , auxiliary , dest , source );
25 > // if ... else
26 return ;
27 > // towers
The output from Program 6- 27 is shown in Tabic 6-3.
Calls Output
Towers (3 , A , C, B )
Towers ( 2, A , B , C )
Towers ( 1 , A, C , B)
Step Is Move from A to C

Step 2: Move from A to B
Towers ( 1 , C, B , A)
Step 3: Move from C to B
Step 4: Move from A to C
Towers ( 2, B, C, A )
Towers ( 1 , B, A , C )
Step 5: Move from B to A
Step 6: Move from B to C
Towers ( 1 , A, C , B)
Step 7: Move from A to C
TABLE 6 - 3 Tracing of Program 6- 27, Towers of Hanoi
6.10 Programming Example — The Calculator Program

Lets look at our calculator program one more time . In Chapter 5 , we gave
users the capability ol selecting one of four options: add , subtract , multiply,
or divide. However, if users needed to make two calculations , they had to run
the program twice. We now add a loop that allows users to make as many
—
362 Section 6.10 Programming Example The Calculator Program
) . We include only two functions, main

calculations as needed ( Program 6 - 28
and getOption , since, at this point , all of the others are the same .
PROGRAM 6- 28 The Complete Calculator

1 /* This program adds , subtrac
ts , multiplies, and divides
2 two integers.
3 Written by:
4 Date:
5 */
8
11 void getData ( int* a , int* b );
12 float calc (int option , int numl , int num2);
13
14 float add (float numl , float num 2 );
15 float sub (float numl , float num2 );
16 float mul (float numl, float num2);
17 float divn ( float numl , float num2 );
18
19 void printResult (float numl , float num2,
20 float result , int option );
21
23 {
25 int done = 0;
26 int option;
27 float numl ;
28 float num 2;
29 float result ;
30
31 // Statements
32 while ( Idone)
33 {
35 if (option = = 5)
36 done = 1;
37 else
38 {
39 do
40 {
41 printf( "\n \nEnter two
numbers:
i continued
PROGRAM 6- 28 The Complete Calculator (continued )

42 scanf( "%f %f " , &numl , &num2);
43 if ( option = = 4 && num2 == 0)
44 {
45 printf( " \a\n *** Error *** " )?
46 printf( "Second number cannot be 0\n" );
47 > // if
48 > while (option = = 4 && num2 = = 0);
49
50 switch ( option )
51 {
52 case Is result = add ( numl , num2);
53 break ;
54 case 2: result = sub ( numl , num2 );
55 break ;
56 case 3: result = mul ( numl , num2 );
57 break ;
58 case 4: result = divn ( numl , num2 );
59 > // switch
60
61 printResult ( numl , num2 , result, option );
62 > // else option != 5
63 } // while
64 printf( " \nThank you for using Calculator.\n " );
65 return 0;
66 > // main
67
68 /*
69 This function shows a menu and reads the user option ,
70 Pre nothing
71 Post returns a valid option
72 * /
73 int getOption ( void )
74 {
76 int option ;
77
78 // Statements
79 do
80 {
" \n**************** *** " •
81
82 printf( " \n* MENU ^
•• ) ?
83 printf("\n* * •• );
84 printf( " \ n * 1. ADD * »
);
85 printf( " \ n* 2. SUBTRACT * »
) ;
continued
364 Section 6.10 Programming Example — The Calculate Program
PROGRAM 6 - 28 The Complete Calculator (continued

)
»
printf(”\n* 3. MULTIPLY * );
86
87 printf( "\n* 4. DIVIDE * •• );
88 printf( " \n* 5. QUIT * "» ) ;
);
89 printf("\n*
*** * **** " ) ?
90 printf( "\n***********
91
92 printf( "\n\n\nPlease type your choice ")?
93 printf("and press the return key :
94 scanf(" %d " , &option );
95
96 if (option < 1 | |option > 5)
97 -
printf( " Invalid option. Please re enter.\n " );
98
99 > while (option < 1|
|option > 5);
100 return option ;
101 } // getOption
Program 6 - 28 Analysis As you look at the changes in this version of our program, first note the two loops in
main. The first loop ( statement 32) continues the calculator until the user says it's time
to quit. The second loop ( statement 39) gets and validates the numbers, making sure
that the user isn' t trying to divide by zero. ( Your computer will get very upset if you
divide by zero and will stop!)
We also modified our getOption function to add the quit option and to vali-
date the options. If the user makes a mistake, we correct it in getOption. Extending
the concept to a general principle, whenever we write a function to get data from a
user, the function should handle all data validation. This makes for much simpler code
in the rest of the program.
This simplification is also seen in the switch statement. Since we have validated the
numbers before the switch, we no longer need to test for a valid divisor in the fourth
case option. We also no longer need a default, since we know the options are valid.
The result is a simpler statement, much more in line with the KISS principle.

Loops in Structure Charts
Now that you understand how to write loops, let ’s look at how they are shown
in a structure chart . The symbols are very simple. Loops go in circles, so
the symbol we use is a circle. Programmers use two basic looping symbols.
I he first is a simple loop. It is represented by Figure 6 - 3 1 ( a). The other is the
conditional loop, shown in Figure 6- 31 ( b).
int dolt ( ... ) int select ( ... )

{
dolt select
while ( expression ) while ( expression )
fun ( ... ); (
if (condition )
} // dolt doA ( ... );
fun doA } // while
} // select
( a) loop (b) conditional loop
FIGURE 6 - 31 Structure Chart Symbols for Loops
When the function is called unconditionally, as in a while loop, the circle

flows around the line above the called function. On the other hand, il the call
is conditional, as in a function called in an if ...else statement inside a loop,
then the circle includes a decision diamond on the line.
Figure 6- 32 shows the basic structure for a function called process. The
circle is below the function that controls the loop. In this example, the loop -
ing statement is contained in process, and it calls three functions, A, B, and
C . The exact nature of the loop cannot be determined from the structure
chart . It could he any of the three basic looping constructs. To help you bet -
ter visualize the process, however, let ' s further assume that the loop is a while
loop that contains a scanf that reads until the end of file. Within the while
loop are three calls: the first to A, the second to B, and the third, condition-
ally, to C.
Determining Algorithm Efficiency

There is seldom a single algorithm for any problem. When comparing two dif-
ferent algorithms that solve the same problem, we often find that one algo-
rithm will be an order of magnitude more efficient than the other. In this
case, it only makes sense that programmers have some way to recognize and
choose the more efficient algorithm.
ion 6.11 Software Engineering
1 while ( ... )
{
Process ... ) ;
A (
B ( ... ) ;
<> if (
C (
)
... );
A B C
} / / while
( a ) Design ( b ) Code
FIGURE 6- 32 Structure Chart for Process
Although computer scientists have studied algorithms and algorithm effi-

ciency extensively, the field has not been given an official name . Brassard and
Bratley coined the term algorithmic^, which they define as “ the systematic
study of the fundamental techniques used to design and analyse efficient
algorithms. ” 2 We will use this term .
—
II a function is linear—that is, if it contains no loops then its efficiency
is a function of the number of instructions it contains. In this case, the func-
tion ’s efficiency depends on the speed of the computer and is generally not a
factor in its overall efficiency in a program . On the other hand , functions that
loop will vary widely in their efficiency. The study of algorithm efficiency is
therefore largely devoted to the study of loops.
As we study specific examples, we will develop a formula that describes
the algorithm ’s efficiency. We will then generalize the algorithm so that the
efficiency can be stated as a function of the number of elements to he pro-
cessed . The general format is
/( n ) = efficiency
Linear Loops
Let ’s start with a simple loop. We want to know how many times the body of
the loop is repeated in the following code:
for ( int i = 1 ; i < = 1000 ; i ++ )

{
BODY ;
} // for
.
2. Gilles Brassard and Paul Bratley, Algorithmic* theory amt Practice ( Upper Saddle River NJ Prenti
Hall . 1988 ), p. xiii.
The answer is 1000 times. But the answer is not always straightforward
as it isin the previous example. Consider the following loop. How many times
is the body repeated in this loop? I lore the answer is 500 times. Why ?
for (int i = 1 ; i <= 1000; i += 2 ;)

{
BODY ;
> // for
In both cases, the number of iterations is directly proportionate to the

limit test in the for loop. If we were to plot either of these loops, we would get
a straight line. Thus, they are known as linear loops.
Since the efficiency is proportionate to the number of iterations , it is
fiu ) = n
Logarithmic Loops
Now consider a loop in which the controlling variable is multiplied or divided
in each loop. How many times will the body of the loops he repeated in the
following program segments?
Multiply Loops Divide Loops

i = 1; i = 1000;
while ( i < 1000) while ( i >= 1 )
{ {
BODY ; BODY ;
i *= 2 ; i /= 2;
} // while } // while
To help us understand this problem, Table 6- 4 analyzes the values ol i lor

each iteration.
Multiply Divide
Iteration i Iteration
1 1 1 1000
2 2 2 500
3 4 3 250
4 8 4 125
continued
TABLE 6 - 4 Analysis of Multiply / Divide Loops
)
Multiply Divide
Iteration Iteration i
5 16 5 62
6 32 6 31
7 64 7 15
8 128 8 7
9 256 9 3
10 512 10
(exit) 1024 ( exit) 0
TABLE 6 - 4 Analysis of Multiply / Divide Loops ( continued )
As can be seen, the number of iterations is 10 in both cases. The reason

is that in each iteration the value of i doubles for the multiplication and is cut
in half for the division. Thus, the number of iterations is a function of the
multiplier or divisor; in this case, two. That is, the loop continues while the
condition shown below is true .
2Iterations < JQQQ

multiply
divide
,
1000 / 2l crafions >= 1
Generalizing the analysis, we can say that the iterations in loops that
multiply or div ide are determined by the following formula:
/( / / ) = ceil (log2/i)
Nested Loops
When we analyze loops that contain loops, we must determine how many
iterations each loop completes. The total is then the product of the number
of iterations for the inner loop and the number of iterations in the outer loop.
Iterations = outer loop iterations * inner loop iterations
We now look at three nested loops: linear logarithmic, dependent qua-

dratic, and quadratic.
Linear Logarithmic
The inner loop in the following code is a loop that multiplies. (To see the
multiplication , look at the update expression in the inner for statement . )
for (int i = 1 ; i < 10; i ++)

for ( int j = 1; j <= 10 ; j *= 2)
{
BODY ;
> // for j
I he number of iterations in the inner loop is therefore
ceil (log210)
I lowever, since the inner loop is controlled by an outer loop, the above
formula must be multiplied by the number of times the outer loop executes,
which is 10. This gives us
10 (ceiV( log2 l ()))
which is generalized as
/('» ) = /i ( ceil (log2/i ))
Dependent Quadratic
Now consider the nested loop shown in the following example:
for (int i = 1; i < = 10; i ++)

for ( int j = 1 ; j <= i; j ++)
{
BODY ;
> // for j
The outer loop is the same as the previous loop. However, the inner loop
is executed only once in the first iteration , twice in the second iteration , three
times in the third iteration , and so forth . The number of iterations in the
body of the inner loop is mathematically stated as
1 + 2 + 3 + . . . + 9 + 10 = 55
which is generalized to
( n + 1)
f( n ) = n
2
Quadratic
In the final nested loop, each loop executes the same number of times, as
seen in the following example:
for (int i = 1 ; i < 10 i ++) -

for ( int j = 1 ; j <= 10; j ++)
{
BODY ;
> // for j
The outer — —
loop that is, the loop at the first for statement is executed
.
ten times For each iteration , the inner loop is also executed ten times. The
answer, therefore, is 100 , which is 10 * 10, the square of the loops. This for-
mula generalizes to
M = /i 2
Big -0 Notation
With the speed of computers today, we are not concerned with an exact mea -
surement of an algorithm’s efficiency as much as we are with its general mag-
nitude . If the analysis of two algorithms shows that one executes 1 5 iterations
while the other executes 25 iterations, they are both so fast that we can ’t see
the difference. On the other hand , il one iterates 15 times and the other
1500 times, we should he concerned .
We have shown that the number of statements executed in the function
for n elements ol data is a function of the number of elements, expressed as
/( » ) . While the equation derived for a function may be complex, usually, a
dominant factor in the equation determines the order of magnitude of the
result . I herefore, we don t need to determine the complete measure of effi-
ciency, only the factor that determines the magnitude . This factor is the big-
0, as in On - the-Order-Of , and expressed as O ( w ) , that is , on -the-order-of n.
-
I his simplification ol efficiency is known as big O analysis , For example,
if an algorithm is quadratic , we would say its efficiency is
0( n2 )
-order of w-squared .
or on - the
I he big- O notation can be derived from (w ) using the following steps:
/
1 . In each term , set the coefficient of the term to 1 .
2. Keep the largest term in the function , and discard the others. Terms are
ranked from lowest to highest, as shown below.
log n n n log n n1 3
/i nk 2"
- '
11
tor example, to calculate the big-0 notation For
„ = nielli = i„ 2 + ^1 n
f( )
2 2 2
M
we first remove all coefficients. This gives us
/i 2 +n
which , alter removing the smaller Factors, gives us
/i
2
which , in Big-O notation , is stated as
0( f(/i )) = 0( /i 2 )
Io consider another example, let’s look at the polynomial expression
f ( /l ) = 371
k
+ 3y _, /1
k-1
+
— + 32 /I + 3
2
, /1
I
+ 3n
We first eliminate all of the coefficients, as shown below.
= nk + n - + •• +
k 1 2
f ( /# ) • /i +n+1
The largest term in this expression is the first one, so we can say that the
order of a polynomial expression is
0( f ( /» )) = 0(„*)
Standard Measures of Efficiency

Computer scientists have defined seven categories oF algorithm efficiency. We
list them in Table 6-5 in order of decreasing efficiency. Any measure of effi -
ciency presumes that a sufficiently large sample is being considered. If we
are
with ten elements and the time required is a fraction of a sec -
only dealing
ond , there will he no meaningful difference between two algorithms. On the
other hand , as the number of elements processed grows, the difference
between algorithms can he staggering. In I able 6- 5 , n is 10,000.
about effi -
Returning to the question ol why we should he concerned
the situation in which we can solve a problem in three ways:
ciency, consider
, the third is quadratic . I he
one is linear, another is linear- logarithmic and
372 Section 6.1II Software Engineering
order of their efficiency for a problem containing 10 ,000 elements is shown

in Table 6-5 , along with the other algorithmics. Obviously, we wouldn ’t want
to use the quadratic solution .
Efficiency -
Big 0 Iterations
logarithmic 0(log n) 14
linear O(n) 10,000
linear logarithmic 0( n(log n) 140,000
quadratic Ofn2) 10 ,0002
polynomial Q [ nk ) 10, 000fe
exponential 0( cn ) 2 io, ooo
factorial O(nl ) 10,000!
TABLE 6 - 5 Measures of Efficiency
Looking at the problem from the other end , if we use a computer that
executes a million instructions per second and the loop contains ten instruc-
tions , then we spend .00001 second for each iteration of the loop. Table 6- 5
also contains an estimate of the time to solve the problem given different
efficiencies.

1 . Be aware that the while and /or loops are pretest loops. Their body may
never he executed . If you want your loop to he executed at least once , use
a do . ..while .
2. Do not use equality and inequality for the control expression in loops ;
use limits that include less than or greater than . You may accidentally
if
create an infinite loop, as shown helow.
i = 0;
while (i != 13 )
{
i++; // sets i to 1, 3, 5 , ..., 13
i++; // sets i to 2 , 4, 6, ... , 14
> // while
3. It is a compile error to omit the semicolon after the expression in the

do. .. while statement .
4 . It is most likely a logic error to place a semicolon after the expression in a
while or for statement. ( Some compilers warn you when you do. )
5. It is a compile error to code a for statement with commas rather than
semicolons, as shown helow.
for ( int i = 0, i < 10, i++)
.
6. It is a logic error to omit the update in the body of a while or do . . while loop.
Without an update, either explicit or implicit , the loop never terminates.
7 It is a common logic error to miscode the limit test in for statements. I he
.
result is usually a loop that executes one extra time or terminates one
iteration short . For example, the following statement executes nine times,
not ten:
for ( int i = 1; i < 10 ; i++)
8. It is generally a logic error to update the terminating variable in both the

for statement and in the body ol the loop, as follows.
for ( int i = 0; i < 10; i++)

{
i += 1;
} // for
9. A recursive function must have a base case. Therefore

, it is most likely an
error if a recursive function does not have an if statement that prevents
the recursive call and allows the function to return . For example, the fol -
— —
lowing code based on Program 6-26 would never terminate.
long fib ( long num )

{
// Statements
return (fib ( num - 1) + fib ( num - 2));
} // fib
6.13 Key Terms

all inquiry initialization
any inquiry inquiry
base case limit - test expression
big-O analysis loop control expression
body of loop loop update
comma expression post - test loop
comma operator pretest loop
continue process-control loops
counter-controlled loop recursion
-
event controlled loop summation
general case while
infinite loop
6.14 Summary
The real power of computers is in their ability to repeat an operation ora
series of operations many times.
J lo control the loop, we need a condition to determine if more processing
is needed .
J In a pretest loop, in each iteration , we check the condition first . II it is
true, we iterate once more; otherwise, we exit the loop.
J In a post - test loop, in each iteration , we do the processing. Then we check
the condition . It it is true, we start a new iteration ; otherwise, we exit
the loop.
J In a pretest loop, the processing is done zero or more times.
J In a post- test loop, the processing is done
one or more times.
J In a pretest loop , if the body is executed n times, the limit test is executed
n + 1 times .
1
J In a post -test loop, it the body is executed n times, the limit test is exe -
cuted n times.
J 1 he control expression in a loop must be explicitly or implicitly initialized .
IJ It you know exactly the number of times the body must he repeated , use a
counter-controlled loop ; it an event must occur to terminate a loop, use an
event-controlled loop.
C has three loop statements: while , for , and do ...while .
1 he while loop is a pretest loop. It can he used for a counter- controlled or
-
event controlled loop, hut it is usually used only for event control.
I he for loop is a pretest loop. It can be used for both counter-controlled
and event -controlled loops , hut it is used mostly in the first case.
The loop variable in a for loop may be locally defined in the for statement .
The do...while loop is a post -test loop. It is usually used when the body
must be executed at least once.
VVe discussed two C statements that are related to looping, break and
continue.
The break statement is used to terminate a loop prematurely. We strongly
recommend that you use the break statement only within switch statements.
The continue statement is used to skip the rest of the statements in a loop
and start a new iteration without terminating the loop. VVe strongly recom -
mend that you never use the continue statement .
The best loop for data validation is the do. .. .while loop.
Recursion is a repetitive process in which a I unction calls itself .
The statement that solves a recursive problem is known as the base case;
the rest of the function is known as the general case.
A loop in a structure chart is indicated by a circle on the line connecting it
to the called functions. Only loops that call other functions are shown .
6.15 Practice Sets

Review Questions
1 . In a pretest loop, the limit test condition is tested first .
.
a True
b. False
2. The action that causes the loop limit test to change from true to false is
the loop update.
a. True
b . False
3. The value of a comma expression is the value ol

the first expression.
a . True
b. False
4. Recursion is a repetitive process in which a function calls itself .
a. True
.
b False
.
5 Which of the following statements about pretest loops is true?
a. If a pretest loop limit test is false, the loop executes one more time.
b . Pretest loop initialization is clone first in the loop body.
c. Pretest loops execute a minimum ol one time .
.
d Pretest loops test the limit condition after each execution of the
loop body.
e. The update for a pretest loop must be a part of the loop body.
6. Which of the following statements about loop initialization is falser
a. Explicit initialization includes code to set the initial values of loop
variables.
b. Implicit initialization relies on preexisting values for loop variables.
c. Initialization code is explicitly required in all loops.
d . Initialization is preparation required for proper execution of a loop.
.
e Initialization must he done before the first execution ol the loop.
7. Which of the following statements about loop updates is false?
a . A loop update changes key variable(s) in a loop, thus allowing the loop
to terminate .
b. Loop updates may be made before or after a loop iteration .
c . Loops may use explicit or implicit updates.
d . In a for loop, updates are generally found in the for statement itself .
e. I he number ol updates always equals the number of loop iterations.
8. Which of the following statements about counter-controlled loops is false?
a . Counter-controlled loops are generally pretest loops.
b. Counter-controlled loops generally increment or decrement a counter.
c. Counter-controlled loops require a limit test .
d . The number of times a loop iterates must be a constant .
e. I he update in a counter-controlled loop is generally explicit .
9. Which ol the C loops is a pretest loop?
a . do... while
.
b for
c. while
d. both the do. . . while and the for
e .
both the for and the while
10. \ \ hich ol the lollowing statements about the while
statement is true?
a . Multiple statements are allowed in a while
loop.
b. I he limit test in a while loop is made before
each iteration .
r T
c .I he update in a while statement is contained in the while statement

expression itself .
d . The while statement is a post - test loop.
e. I he while statement must be terminated with a semicolon.
11 . Which of the following statements about for and while statements 41
is false?
a . Both statements allow only one statement in the loop.
b. Both statements are pretest loops.
c . Both statements can he used for counter-controlled loops.
d. Both statements include initialization within the statement .
e. Both statements require an update statement.
..
12. Which of the following statements about the do . while loop is false ?
a. A do. . .while loop executes one or more iterations.
b . Anv statement may he used as the action in a do ... while .
c. The do ... while is best suited for use as an event-controlled loop.
d . The do...while is the only loop that requires a semicolon.
e. The limit test in a do...while loop is executed at the beginning ot each
iteration .
13. Nested loops have a standard measure of efficiency.
a. exponential
h . linear
.
c linear logarithmic
d . quadratic
e. linear logarithmic or quadratic
14. The standard measure of efficiency is considered the
most efficient .
a . exponential
b. linear
c . logarithmic
d . quadratic
e. polynomial
Exercises
13. What would he printed from each of the following program segments?
Compare and contrast your answers to parts a , b, and c.
a.
x = 12 ;
while ( x > 7)
printf("%d\n " , x );
3/ 8 Section 6.15 Practice Sets
b.
for ( int x = 12 ; x > 7 ; )

printf( " %d\n" , x );
c.
x = 12;
do
printf( " %d\n " , x );
while (x > 7);
16. What would he printed from each of the following program segments?
a.
x = 12;
while ( x > 7)
{
printf( "%d\n " , x);
x --;
>
b.
for (int x = 12; x > 7 ; x

printf( " %d\n " , x ) ; — )
c.
x =12;
do
{
—
printf( "%d\n" , x ) ;
x ;
} while (x > 7);
17. What would be printed from each ol the following program segments?
Compare and contrast your answers to parts a and b.
a.
x = 12 ;
while ( x > 7)
{
printf( " %d\n" , x ) ;
-
x = 2;
>
Chapter 6 Repetition 3/9
b.
for ( int x = 12; x > 7; x

printf ( "%d\n " , x ) ;
-= 2)
18. What would be printed from each of the following program segments ?
a.
x = 12 ;
while ( x < 7 )
{
printf( " %d \n " , x );
x
;
—
} // while
b.
for ( int x = 12 ; x < 7;

printf ( "%d\n" , x ); —
x )
c.
x 12;
do
{
printf( "%d\n " , x );
x
— ;
} while ( x < 7);
19. Change the following while loops to for loops.

a.
x = 0;
while ( x < 10)
{
printf( "%d\n " , x );
x++ ;
}
b.
scanf( " %d " , &x );

while ( x != 9999 )
{
printf( " %d\n " , x );
scanf("%d" , &x );
>
20. Change the while loops in Exercise 19 to do...while loops.

2 1 . Change the following for loops to while loops:
a.
for (int x = 1; x < 100; x++ )

printf ( " %d\n" , x);
b.
-
for ( ; scanf( " %d " , &x) ! EOF ; )
printf ( "%d\n" , x);
22. Change the /or loops in Exercise 21 to do ...while loops.

...
23. Change the following do while loops to while loops.
a. i
x = 0;
do
{
x++;
> while ( x < 100);
b.
do
{
res = scanf("%d " , &x );
> while (res != EOF );
24. C hange the do...while loops in Exercise 23 to for loops.

25 . A programmer writes the following /or loop to print the numbers 1 to 10.
W hat is the output? If the output is incorrect , how would you correct it:
for ( num = 0; num < 10; num++)

printf( " %d " , num);
26. Another programmer writes the following lor loop to print the numbers 1
to 10 . What is the output ? It the output is
incorrect, how would you cor-
rect it ?
for (int num = 0; num < 10; num++)

{
numOut = num + 1;
printf( " %d " , numOut );
> // for
27 . What will be printed from the following program segments?

a.
for ( int x = 1 ; x <= 20 ; x++)

b.
for ( int x = 1 ; x <= 20 ; x++)

{
printf( "%d\n", x );
x++ ;
} // for
28. What will he printed from the following program segments?

a.
for ( int x = 20; x >= 10; x

printf( " %d\n " , x );
— )
b.
for ( int x = 20; x >= 1; x

{
— )
x ;
—
printf( " %d\n " , x );
} // for
29. What will he printed from the following program segments?

a.
for ( int x = 1; x <= 20; x++)

{
for ( int y = 1; y <= 5; y++ )
printf("%d " , x );
printf(" \n " );
} // for
b.
for ( int x = 20; x >= 1 ; x

— )
{
for ( int y = x ; y >= 1 ; y
printf( " %3d " , x);
— )
printf( " \n" );

} // for
.
30 What will be printed from the following program segments ?
a.
for ( int x = 1 ; x <= 20; x++)

{
for (int y = 1; y < Y++)
printf( " ");
> // for
b.
for (int x
{
= 20; x >= 1; x
— )
for (int y = x; y >=1; y

printf ( " M );
—)
printf ( " %d\n " , x);
> // for
31. You find the statement shown below in a program you are maintaining.
for ( ; ; )
{
> // for
a. Describe the implications behind the null expressions in the for

statement.
b . Since there is no limit condition , how can this statement he exited ?
c. Is this good structured programming style? Explain your answer .
Problems
32 . Write a program that uses a for loop to print a line of 60 asterisks.
.
33 Write a Jor loop that will produce each of following sequences:
.
a 6 , 8, 10 , 12 , ... , 66
b. 7 , 9 , 1 1 , 13, ..., 67
c. 1 he sum ol the numbers between 1 and 1 5 inclusive
.
d I he sum ot the odd numbers between 1 5 and 45 inclusive
.
e The first 50 numbers in the series 1 , 4 , 7 , 10 , . . . (
calculate the total
.
1 + 4 + 7 + 10 + . . )
.
34 \ \ rite a program that prompts the user to enter
an integer, n , and then n
floating- point numbers. As the numbers are read , the program will calcu -
late the average of the positive numbers.
35. Rewrite Problem 34 to average the
negative numbers .
— Chapter 6 Repetition 383
.
36 Write a program that asks the user to enter a list of integers. The pro-
gram is to determine the largest value entered and the number of times it
was entered , l or example , if the following series is entered
5 2 15 3 7 15 8 9 5 2 15 3 7
it would output the largest value is 15 and it was entered 3 times.

37 . Write a program that creates the following pattern:
1 2 3 4 5 6 7 8 9
1 2 3 4 5 6 7 8
1 2 3 4 5 6 7
1 2 3 4 5 6
1 2 3 4 5
1 2 3 4
1 2 3
1 2
1
38. Write a function that creates the following pattern , given the height
( number of rows ) and the width (asterisks per row ) :
* ******* *****
*** **** ** * * * *
**** * * ** * * ** *
* * * * * * ** * * * * * ,
( number of rows ) and the width ( print characters per line ):
* *
* *
*
* *
( number of rows ):
*
* **
* * ***
* * * ** * *
* ***** * **
** * ** * * * * * *
. Jd
( number of rows ) :
*** * * * * * * * *
* ** * ** * * *
* * * * * **
** * * *
** *
.
42 Write a function that creates the following pattern , given the height
( number of rows ):
*
** *
* ****
* ** * * * *
** * * * * ***
* * * ** * * * *
* *** * **
** ***
* **
*
43. Modify Program 6 - 3 to display the total as each number is entered. I he

format should he
Enter your numbers: <EOF> to stop.

5
Total: 5
17
Total: 22
8
Total: 30
44 . Write a program that reads integer data from the standard input unit and
prints the minimum integer read , maximum integer read , and the average
of the list . Test vour program with the data shown below.
{24 7 31 -5 64 0 57 -23 7 63 31 15 7 -3 2 4 6}
45 . In “ I he do. ..while Loop" in Section 6.5, we
demonstrated the use of a
do ...while to validate input. I he code fragment
contains no message to
tell the user that an invalid number has been
entered . Write a function
that reads only positive even numbers from the keyboard . If a negative or
odd number is entered , it should print an
error message and ask the user
to enter another number. Each call to
the function is to read only one
'l l
number. The valid number read is to be returned to the calling program .

II the user enters EOF, the function should return it . Then write a short
program to test the function using the data shown below. The valid num -
bers should he printed , either in a separate function or in main .
{ 2 18 -18 5 7 100 1 -1 }
46. Write a function that reads integers from the keyboard . If any of the
numbers are negative, it returns a negative number. If all the numbers
are positive , it returns their average. ( Hint: See Program 6- 22. )
47 . Program 6 - 19 uses a while loop to read a series of numbers from the key-
board . Since you will always have at least one number in this program ,
rewrite it to use the do. ..while .
_ .
48. Program 6- 21 uses INT MAX from the limits h library to initialize the
smallest variable. Another solution is to read the first number and put
its value in smallest , then go into the loop to read the rest of the num -
bers. Modify Program 6- 21 to make this change .
49. Euler 's number, e , is used as the base of natural logarithms. It can be
approximated using the following formula :
e - l + I + I + I + I + I + I + ••• +
1! 2! 3! 4! 5! 6!
1
(w - 1)
+I
n\
Write a program that approximates e using a loop that terminates when

the difference between two successive values of e differ by less than
0.0000001 .
.
50 The value of pi ( 7t ) can be calculated by the following formula :
1
0 = 6
lh h h
+ + + . .. +
limit 2
Write a function that uses this formula to calculate pi . Then write a test
driver, and run it once to test your function with a limit of 5 terms and a
limit of 10 terms. Display the result of each test.
5 1 . Write a program that reads an integer from the keyboard and then calls a
recursive function to print it out in reverse. For example, if the user
enters 4762 , it prints 2674.
52. Rewrite Program 6- 26 using an iterative solution .
Projects
to the arithmetic
53. Statisticians use many different algorithms in addition
average. Two other averages are thegeometric and the harmonic mean .
The geometric mean of a set of n numbers, x\ > x 2 , xn - 1 » *„ is

defined by the following formula:
7xixx 2 x - • * xn
n •
The harmonic mean is defined by the following formula:
I !
n
+ + ... +
1 _
X, X 2 x *
Write a program that reads a series of numbers and calculates the aver -
age , geometric mean , and harmonic mean .
54. Write a C program that can create four different patterns of different
sizes. The size of each pattern is determined by the number of columns
or rows. For example , a pattern of size 5 has 5 columns and 5 rows. Each
pattern is made of character $ and a digit , which shows the size . The size
must he between 2 and 9. Table 6 - 6 shows the four patterns in size 5 .
Pattern 1 Pattern 2 Pattern 3 Pattern 4

5$$$$ $$$$5 $ $$$ $ $ $$$$
$5$$$ $$$5$ $$$$5 5$$$$

$$5$$ $$5$$ $$$55 55$$$
$$$5$ $5$$$ $$555 555$$
$$$$5 5$$$$ $5555 5555$
TABLE 6 - 6 Patterns for Problem 54

Tour program displays a menu and asks the user to choose a pattern and
si / e . But note that it must he robust ; it must prompt the user to choose
an option only between 1 and 5 , and a pattern size only between 2 and 9.
You arc to print the menu and the user 's response. The following example
shows all user menu responses, including potential errors ( user
responses are in color ):
M E N U
1. Pattern One
2. Pattern Two
3. Pattern Three
i
continued
I
4. Pattern Four
5. Quit
Choose an Option ( between 1 and 5 ): 11
Your Option is incorrect. Please try again.
Choose an Option ( between 1 and 5 ): 3
Choose a Pattern Size ( between 2 and 9 ): 12
Your Pattern size is incorrect. Try again.
Choose a Pattern Size ( between 2 and 9 ): 4
I he program must consist of a main function and six other functions

called getOption, getSize, patternOne, patternTwo, patternThree.
and patternFour.
Run your program once with the options and sizes shown in Table 6 - 7.
Note that some options and sizes are missing because either the previous
option or the size is invalid.
Option
SET 1 1
SET 2 2 3
SET 3 3 4
SET 4 4 5
SET 5 6
SET 5 3 6
SET 6 2 10
SET 6 7
SET 7 5
TABLE 6 -7 Test Data for Problem 54

55. Write a C program to create a calendar for a year. The program reads the
vear from the keyboard. It then calculates which day ol the week SUN
( ,
MON, TUE, WED, T 11U , FR 1 . SAT) is the first day of the year and prints
the calendar for that year. After printing the year, it should ask il the user
wants to continue. II the answer is yes, it will print
the calendai lor
another year until the user is done.
The program prompts the user for the input, as shown in the next
example.
2000
Enter the year for your calendar :
i
W
The output is a calendar for the whole year ( 12 months ) . One month is
shown in the next example.
JANUARY 2000
SUN MON TUE WED THU FRI SAT
1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31
To print the correct calendar for the requested year, you must first find which
day of the week is the first day of the that year. This can he done with the fol-
lowing formula . ( For a complete explanation , see Chapter 5, Project 61.)
(«»• ' FF1] Far1] * FSr] * ‘)

- %7
You also must calculate leap years. The formula for this calculation is
also found in Chapter 5 , Project 61 .
Run your program once with the following sets of data 3:
SET Is 2005
SET 2: 0
SET 3: 2000
SET 4: 123
56. \ \ rite a C program to help a prospective borrower calculate the monthly

payment for a loan . I he program also prints the amortization ( payoff )
table to show the balance of the loan after each monthly payment.
1 he program prompts the user for input , as shown in the following
example:
Amount of the loan (Principal)? 10000.00

Interest rate per year ( per cent)? 12
Number of years? 10
I he program then creates an information summary and amortiza *
tion table.
3. The Julian calendar was changed to our current (

Gregorian ) calendar in 1752. Although calendars
e ore this date are valid Gregorian calendars,
they do not represent the Julian calendar in use at
that time.
*
use a simple formula
NM = ( NY * 12)
Banks and financial institutions use different formulas to calculate

tlu monthly payment of a loan . For the purpose of this assignment , we
H
'
IM = (IY / 12 ) / 100
P = (1 + IM)NM
Q = ( P / (P - 1) )
MP = (PR * IM * Q)
where
NY: Scheduled number ol years to amortize the loan
NM : Scheduled number of months lor the loan
IY: Interest rate per year ( as a percentage)
IM : Interest rate /month ( decimal )
PR : Principal ( the amount of the loan )
P: Fhe value of ( 1 + IM )NM
Q: The value of P / ( P - 1 )
MP: Monthly payment
main must call three other functions: c a l c u l a t e M o n t h l y P a y m e n t ,

p r i n t l n f o r m a t i o n, and p r i n t A m o r t i z a t i o n T a b l e. Of course, the
first function may also call other functions if necessary.
Because ol the approximation used in calculation , the value of the
new balance at the end of the last month may become nonzero. To pre-
vent this, the last payment must be adjusted. It may he less or greater
than the other months. It must be calculated by adding the principal paid
to the interest paid for that month . The new balance at the end of the
last month must be zero.
I he following example shows the concept. The program has been
run for a loan of $ 5000.00 at 11 % interest rate for a period of 1 year. The
input was
Amount of the loan ( Principal)? 5000.00

Interest rate / year ( per cent )? 11
Number of years? 1
Run your program once with the test data shown in Table 6-8.
Amount Interest Years
10 , 000.00 12 l
Set 1
Set 2 5, 0 0 0 . 0 0 10 2
Set 3 1, 000.00
TABLE 6 - 8 Test data for Problem 56

The output is shown in Table 6 - 9. Note: Your answer may look a little dif -
ferent ( a few pennies ) because ol different precision .
The amount of the loan ( principal ): 5000.00
Interest rate/year ( percent ): 11.0
Interest rate/month (decimal): 0.009167
Number of years: 1
Number of months: 12
Monthly payment: 441.91
Old Monthly Interest Principal New

Month Payment Paid Paid Balance
Balance
1 5000.00 441.91 45.83 396.08 4603.92
2 4603.92 441.91 42.20 399.71 4204.21
3 4204.21 441.91 38.54 403.37 3800.84
4 3800.84 441.91 34.84 407.07 3393.77
5 3393.77 441.91 31.11 410.80 2982.97
6 2982.97 441.91 27.34 414.57 2568.40
7 2568.40 441.91 23.54 418.37 2150.03

8 2150.03 441.91 19.71 422.20 1727.83
9 1727.83 441.91 15.84 426.07 1301.76
10 1301.76 441.91 11.93 429.98 871.78
11 871.78 441.91 7.99 433.92 437.86
12 437.86 441.87 4.01 437.86 0.00
T o t a l amount p a i d: 5 3 0 2 . 8 8
TABLE 6 - 9 Sample Output from Project 57
.
57 Write a program that reads a list ol integers from
the keyboard and cre -
ates the following information :
a. 1 inds and prints the sum and the average of
the integers
b. Finds and prints the largest and the smallest integer
.
c Prints a Boolean ( true or false ) if some of
them are less than 20
d . Prints a Boolean ( true or false ) if all of
them are between 10 and 90
raw
I V'
.
I he input data consist of a list of integers with a sentinel The program
must prompt the user to enter the integers, one by one, and enter the
sentinel when the end of the list has been reached. The prompt should
look like the following:
Enter numbers with <return> (99999 to stop):
I he output should he formatted as shown below.
The number of integers is: XXX

The sum of the integers is: xxxx
The average of the integers is: X X X.X X
The smallest integer is: XXX
The largest integer is: XXX
At least one number was < 20: <true or false>
All numbers were ( 10 <= n >= 90): <true or false>\
.
58 The formula for converting centigrade temperatures to Fahrenheit is
180.0
F = 32 + C x
100.0
Write a program that prints out conversion tables for Celsius to Fahren-
heit (0° to 100°) and Fahrenheit to Celsius ( 32° to 212°). Use separate
functions to convert Fahrenheit to Celsius and Celsius to Fahrenheit.
The output format for each table is to ( it on a standard monitor display,
80 columns by 20 rows.
59. Rewrite “Towers of Hanoi” Program 6 - 27 using an iterative solution.
1
I
Text Input/Oulput
A program is a data processor: It accepts input data, processes data, and cre -
ates output data. Handling input /output is a complex task, largely because of
the variety of devices and data formats. Data may come from many different
sources and may go to different destinations. For example, data may come
from such diverse sources as a keyboard, a file on the disk, a heating or air-
conditioning system, or a communication channel ( network or Internet ).
Data may also go to many destinations, such as a monitor or a file on the disk,
or to a communication channel. In this chapter, we concentrate on the key-
board and files as input entities and the monitor and files as output entities.
Objectives
To understand the basic properties and characteristics of external files
To understand the C implementation of file I /O using streams
To write programs that read and write text files using the formatting functions
To write programs that read and write text files using the C character I / O
functions
To write programs that handle simple I /O errors
To understand and implement basic data validation concepts
394 Section 7.1 Files
7.1 Files
.
A file is an external collection of related data treated as a unit 1he primary
purpose of a file is to keep a record of data. Since the contents of primary
memory are lost when the computer is shut down, we need files to store our
data in a more permanent form. Additionally, the collection ol data is often
too large to reside entirely in main memory at one time. I herefore, we must
have the ability to read and write portions of the data while the rest remain in
the file.
A file is an external collection of related data treated as a unit .

Files are stored in auxiliary or secondary storage devices. The two most
common forms of secondary storage are disk ( hard disk, CD, and DVD)
and tape.
When the computer reads, the data move Irom the external device to
memory; when it writes, the data move Irom memory to the external device.
This data movement often uses a special work area known as a buffer. A
buffer is a temporary' storage area that holds data while they are being trans-
ferred to or from memory. The primary purpose of a buffer is to synchronize
the physical devices with a program ’s needs, especially on large storage
devices such as disk and tape. Because of the physical requirements of these
devices, more data can be input at one time than a program can use. The
buffer holds the extra data until the program or the user is ready for it. Con-
versely, the buffer collects data until there are enough to write efficiently.
I hese buffering activities are taken care of by software known as device driv-
ers or access methods provided by the supplier of the computer s operating '
system .
As a file is being read, there eventually comes a point when all the data
have been input. At this point, the file is said to be at end of file . The end ol
file in an auxiliary device is detected automatically by the device and passed
to the program. It is the programmer s job to
test for end of file. This test is
often done in a loop control statement.
File Name
Lu i \ operating system uses a set ol rules for
to read or write auxiliary storage
.
naming its files When we want
files, therefore, we must use the operating
systems rules when we name the files We . refer to the operating systems
name as the file name in this text .
File Information Table

A program that reads or write files needs to
know several pieces of information,
such as the operating system’s name for
the file, the position of the current
r
Chapter 7 Text Input/Output 395
character in the file, and so on. C has predefined a file structure 1 to hold this
information. I he stdio.h header file defines the file structure; its identifier is
F I L E . When we need a file in our program, we declare it using the F I L E type .
7.2 Streams
Although the source and destination of data is a file or a physical device in C,
as we briefly discussed in Chapter 2, data are input to and output from a
stream. A stream can be associated with a physical device, such as a terminal,
or with a file stored in auxiliary memory. Figure 7 - 1 shows this idea.
Data Source Program
File f Input Text Stream Data
|Output Text Stream +# Data

File
Monitor
Data Destination
FIGURE 7- 1 Streams
Text And Binary Streams

C uses two types of streams: text and binary. A text stream consists of a
sequence of characters divided into lines with each line terminated by a new-
line ( \ n). A binary stream consists of a sequence of data values such as inte -
ger, real, or complex using their memory representation. In this chapter, we
discuss only text streams; binary streams are discussed in Chapter 13.
Stream- File Processing

A file exists as an independent entity with a name known to the operating sys-
tem. A stream is an entity created by the program. To use a file
in our pro -
gram, we must associate the program ’s stream name with the operating
system s file name.
In general, there are four steps to processing a file. We first create the
stream . We then open the file, which associates the stream name with the file
in Chapter 12 .
I . Technically, a structure ( struct type ). We discuss structures
396 Section 7.2 Streams
name Once the file is opened, we read or write data; that is, we process the
file. When the processing is complete, we close the file. We describe the steps
to make this association in this section.
Creating a Stream
We create a stream when we declare it . I he declaration uses the FILE type as
shown in the following example. The FILE type is a structure that contains
the information needed lor reading and writing a file.
FILE * spData ;
In this example, spData is a pointer to the stream. Because of the impor-

tance of streams in a program, we provide a special notation to help us recog-
nize them. The sp in the name stands for stream-pointer. We then follow it
with a descriptive name for the data that flows through the stream. In this
case we used the generic name, data . In our examples, we use more descrip-
tive names.
Note that there is an asterisk after the file type(FILE). We saw this nota -
tion in Chapter 4. It indicates that spData is a pointer variable that contains
the address of the stream we are creating. It is an error to omit the asterisk.
Opening a File
Once the stream has been created, we are ready to associate the stream to a
file. I his is done, as we discuss in detail in the next section, through the stan-
dard open function. When the file is opened, the stream and the file are asso -
ciated with each other, and the FILE type is filled with the pertinent file
information. I he open function returns the address of the file type, which is
stored in the stream pointer variable, spData. The file open function creates
the stream, which we then refer to by its name.
Using the Stream Name

Alter we create the stream, we can use the stream pointer in all functions
that need to access the corresponding file lor input or output. For example, a
function can use the stream pointer to read from the file through the corre-
sponding stream.
Closing the Stream

When the file processing is complete, we close the file. Closing the file breaks
the association between the stream name and
the file name. After the close,
the stream is no longer available and any
attempt to use it results in an error,
lo close the association, we need to use
a close function to release the file.
m
System- Created Streams

YVc discussed that a terminal—that is, the keyboard or monitor—can be the
source or destination of a text stream. C provides standard streams to com-
.
municate with a terminal I hese streams must be created and associated with
.
their terminal devices just like files The difference is that C does it automat -
ically for us.
C declares and defines three stream pointers in the stdio.h header file .
I he first, stdin , points to the standard input stream; the second stdout, points
to the standard output stream; and the third, stderr , points to the standard
error stream. Note that these streams are created when the program starts
and we cannot declare them.
Standard stream names have already been declared in the stdio.h header file
and cannot be declared again in our program .
The association between the three standard streams and the keyboard
and the monitor is also done automatically when the program starts. I here-
fore, we cannot open any of the standard streams in our code. Like file
streams, the standard streams must be closed at the end ol the program.
I lowever, they are closed automatically when the program terminates.
There is no need to open and close the standard streams. It is done auto -
matically by the operating system .
Using Stream Names

C includes many standard functions to input data from the keyboard and out -
put data to the monitor automatically without the need lor explicitly using
these standard streams. For these functions, the stream that connects our
programs to the terminal is automatically created by the system and we do
not have to do anything more. Examples of such functions are
scan) and
print / , which we defined and used in C hapter 2.
7.3 Standard Library Input/Output Functions

The stdio . h header file contains several different input/output function declara-
as shown in I igure 7 - 2.
tions. They are grouped into eight different categories,
The first three will be discussed in the following sections. Those shown in
Chapters 1 1 and 13.
shaded boxes will he discussed in
398 Section 7.3 Standard Library Input/ Output Functions
File
Open/Close
Formatted
Input/Output
Character
Input/Output
Categories of
I/O Functions
Line
Input/Output
l
Block
Input/Output
File
Positioning
System
File Operations
File
Status
FIGURE 7- 2 Categories of Standard Input/ Output Functions
File Open and Close

In this section we discuss the C functions to open and close streams .
File Open ( fopen)
.
I he Junction that prepares a file lor processing is fopen It does two things:
I irst, it makes the connection between the physical file and the file stream in
.
the program Second, it creates a program file structure to store the informa-
tion needed to process the file .
lo open a file, we need to specify the physical filename and its mode, as
shown in the following statement.
fopen( " filename ” , " mode" );
Let s examine this statement by starting on the

string that tells C how we intend to use the file: Are
.
right The file mode is a
we going to read an exist -
ing file, write a new file, or append to a file? We discuss
file modes later in
this section.
A filename is a string that supplies the name
of the physical file as it is known to
t he external world. I or example, when we work in
a Microsoft Windows system, the
file name consists ol a name, a dot (period), and
a three-character file extension.
I he address of the file :structure
that contains the file information is
returned by fopen. The actual contents
of FILE are hidden from our view
uyl
Vr
•
Chapter 7 Text Input/ Output 399
because we do not need to see them. All we need to know is that we can store
the address of the file structure and use it to read or write the file . A complete
open statement is shown below. Continuing with our temperatures example,
we could open it for output ( writing ) with the following open statements. The
first is the basic format as it might he used for the current directory in UNIX
or Windows; the second is the Windows version to open a file for drive A. 2
spData = fopen( " MYFILE.DAT", V );

spData = fopen( " A:WMYFILE.DAT" , "w");
We see this open statement in the program in Figure 7- 3. Some explana -

tion of the program segment in the figure is in order. You should first note
that the standard input/output library is called out at the beginning of the
program .
tfinclude <stdio .h >
Internal
int main (void) ile Variable ,
FILE* spData ;
spData = fopen(" MYDATA .DAT" , " W ");
} // main External
File Name
FILE
~ L,
I | J spData
-|
MYDATA . DAT
Physical File ^
C Stream Data
FIGURE 7 - 3 File Open Results
Next , note the name we have used for the file address. I he sp stands for
“stream pointer”; we use it in all of our stream identifiers. To the stream
pointer abbreviation we add the name of the file. Because this is a generic
example, we use “ Data . ” This combination gives a readable name that is easy
to remember.
Our stream pointer variable, spData, is assigned the address of the file
structure when we open the file; the return value from the fopen
function
contains the file structure address. Later in the program , when we need to
read or write the file, we use this pointer.
2 . Note the two backslashes in the filename. Because

the backslash is the escape character, to
« i t one in the output , we must
* code two.
400 Section 7.3 Standard Library Input/Output Functions
File Modes
When we open a file, we explicitly define its mode . The mode shows how we
will use the file: for reading, for writing, or for appending, which means add-
mg new data at the end of the current file. C has six different file modes. The
first three, which we discuss here , are used to read , write, or append text files.
In Chapter 13 we discuss the other modes, which allow both reading and
writing in the same file. The mode codes discussed in this chapter are shown
in Iable 7- 1.3
Mode Meaning
r Open text file in read mode
•If file exists, the marker is positioned at beginning .
•If file doesn't exist, error returned.
w Open text file in write mode
•If file exists, it is erased.
•If file doesn't exist, it is created.
a Open text file in append mode
•If file exists, the marker is positioned at end.
•If file doesn't exist, it is created.
TABLE 7- 1 Text File Modes
Figure 7- 4 describes the simple open modes.
Mode Mode Mode
Open
Open existing file Open new file
for reading existing file for writing
for writing vor create new file^
EOF EOF EOF
File marker File marker ^

positioned at File marker
positioned at positioned at
beginning of file beginning of file end of file
( a) Read Mode (b) Write Mode (c) Append Mode
FIGURE 7- 4 File -Opening Modes
Read Mode
I he read mode ( r ) opens an existing file lor
reading. When a file is opened in
t M S mode , the file marker is
positioned at the ( the first
beginning of the file
3. There are more modes. We discuss them
in Chapter 13.
character ). I he file marker is a logical element in the file structure that keeps
track of our current position in the file.
Hie file must already exist : If it does not , NULL is returned as an error.
Files opened for reading are shown in Figure 7- 4a . If we try to write to a file
1
'^
opened in the read mode, we get an error message.
Write Mode
I he write mode ( w) opens a file for writing. If the file doesn ’t exist , it is cre-
ated . II it already exists, it is opened and all its data are deleted ; that is , it
assumes the status of an empty file. It is an error to try to read from a file
opened in write mode . A file opened for writing is shown in Figure 7 - 4 b.
Append Mode
I he append mode ( a ) also opens an existing file for writing. Instead ol creat -
ing a new file, however, the writing starts after the last character; that is, new
data are added , or appended , at the end of the file.
If the file doesn ’t exist , it is created and opened . In this case, the writing
will start at the beginning of the file ; the result will he logically the same as
opening a new file for writing. Files opened in append mode are shown in
Figure 7- 4 c . If we try to read a file opened for write append , we get an error
message.
File Close [ fclose)

When we no longer need a file , we should close it to free system resources,
such as buffer space.4 A file is closed using the close function , Jclose , as
shown in the following section .
# include <stdio. h>
int main ( void )

{
FILE* spTemp ;
// Statements
spTemp = fopen( " MYDATA .DAT" , " w " ) ;
fclose( spTemps );
4 . Some operating systems allow a file to be opened

by only one application program at a time.
application programs to have access to the hie.
Closing the file also allows other
i
402 Section 7.3 Standard Library Input/Output Functions
Open and Close Errors

What if the open or close fails? Open and close errors occur for a number of
reasons. One of the most common errors occurs when the external filename
in the open function call does not match a name on the disk . W hen we create
a new file, the open fails if there isn ’t enough room on the disk .
Always check to make sure that a stream has opened successfully. If it
has, then we have a valid address in the file variable. But if it failed for any
reason , the stream pointer variable contains NULL, which is a G -defined con -
stant for no address in stdio.h.
Similarly, we can test the return value from the close to make sure it suc-
ceeded . The /close function returns an integer that is zero if the close suc-
ceeds and EOF if there is an error. EOF is defined in the standard input/output
—
header file . Traditionally, it is 1 , but the standard defines it as any nonchar-
acter value . To ensure that the file opened or closed successfully, we use the
function as an expression in an if statement , as shown in Program 7- 1 .
PROGRAM 7- 1 Testing for Open and Close Errors

3
4
5 int main ( void )
6 {
8 FILE* spTemps ;
9
10 // Statements
11
12
13 if ((spTemps = fopen( "TEMPS.DAT" , "r " )) == NULL)
14 {
15 printf( "\aERROR opening TEMPS.DAT\n" );
16 exit ( 100);
17 > // if open
18
19
20 if (fclose(spTemps) == EOF)
21 {
22 printf( "\aERROR closing TEMPS.DAT\n " ;
)
23 exit ( 102 );
24 > // if close
25
26
27 > // main
Program 7- 1 Analysis The most common mistake in testing for a successful open is getting the parentheses
wrong, as shown below.
vlj
if (spTemps = fopen("TEMPS.DAT" , "w") == NULL)
The above statement is syntactically correct ( we do not get a compile error) but
invalid. You must test the address returned by fopen after it has been assigned to
spTemps. In the above statement, spTemps is assigned the logical value of the
expression
fopen( "TEMPS.DAT" , " w" ) == NULL
because the equal operator has a higher precedence than the assignment operator.
Study the open in Program 7- 1 carefully, and make sure you understand the difference
between it and the incorrect version shown in the previous example.
The error testing for the close is much simpler; we can use a simple test for the error
code, EOF. Note that we have assigned a different return code to distinguish the open
failure from the close failure.
7.4 Formatting Input/Output Functions

In Chapter 2 we introduced two formatting input/output functions, scan/ and
printif The scan) function receives a text stream from the keyboard and con-
.
verts it to data values to be stored in variables. The printJ function receives
data values from the program and converts them into text stream to be dis-
played on the monitor.
These two functions can he used only with the keyboard and monitor.
The C library defines two more general functions, fscanf and fprintf , that can
be used with any text stream. Table 7 - 2 compares these four input/output
functions.
Terminal Input / Output

scanf ("control string ", ...);
printf ("control string ", ... );
General Input /Output

fscanf ( streamjpointer , "control string " ,
",
fprintf(stream _pointer , "control string
TABLE 7- 2 Formatting Functions

404 Section 7.4 Formatting Input/Output Functions
Stream Pointer
The first argument in the text input/output function , stream pointer, is the
pointer to the streams that has been declared and associated with a text file.
The following example demonstrates the use of an input stream .
FILE* spin ;
spin = fopen( "file name" , " r " ) ;
fscanf( spin , "format string " , address list );
While we normally read data from the terminal keyboard using scanf , we
can also use fscanf . \\ hen we use fscanf we must specify that the stream
pointer is std m , as shown in the following example.
'
fscanf(stdin , "format string " , address list );
The following example demonstrates the use of an output stream .
FILE* spOut;
spOut = fopen( "file name" , "w" );
fprintf( spOut, " format string " , address list );
Similarly, we can use fprintf to print to the terminal monitor by specifying

that the stream is stdout , as shown in the following example.
fprintf(stdout , "format string " , value list );
Format Control Strings

Input and output functions for text files use a format string to describe bow
data are to be lormatted when read or written . The format control string con -
sists ol three types ol data , which may be repeated : whitespace , text charac-
ters , and the most important ol the three, the conversion specification that
describes how the data are to be formatted as they are read or written .
Whitespace
Format control string whitespace is handled differently for input and output.
In an input function , one or more whitespaces in
the format string cause
zero, one , or moie whitespaces in the input stream
to be read and discarded ,
Thus, ,am sequence ol consecutive whitespace
characters in the format string
will match any sequence ol consecutive
whitesjnice characters, possibly ol
different length , in the input stream .
FI
A whitespace character in an input format string causes leading whitespace

characters in the input to be discarded. A whitespace character in an output
format string is copied to the output stream .
W hitespace in an output function is simply copied to the output stream.

I hus, a space character is placed in the output stream for every space charac -
ter in the format string. Likewise, tabs in the format string are copied to the
output stream. I his is not a good idea, however, because we can t see tabs in
the format string. It is better to use the tab escape character ( \ t ) so that we
can see the tabs.
Text
Any text character other than a whitespace in an input format string must
match exactly the next character of the input stream. If it does not match, a
.
conflict occurs that causes the operation to be terminated The conflicting
input character remains in the input stream to be read by the next input oper -
ation on that stream. We therefore recommend that you do not use text char-
acters in an input format string.
Text characters in an output format string are copied to the output

stream. They are usually used to display messages to the user or to label data
being output.
Conversion Specificotion
The conversion specification consists of a percent character ( % ), optional for-
matting instructions, and a conversion code. With one exception, each con-
version specification must have a matching parameter in the parameter list
that follows the format string. The type in the conversion specification and
the type of the parameter must match.
The number, order, and type of the conversion specifications must match the
number, order, and type of the parameters in the list. Otherwise,
the result
will be unpredictable and may terminate the input /output function .
Conversion specifications can have up to six elements, as shown

in
Moure 7 - 5 . ( Note that for input there are only five; precision is not allowed .
)
specification token (. % ) . I he last element is
The first element is a conversion
the other ele -
the conversion code. Both of these elements are required;
of each element is the
ments are optional. Generally, the meaning
and usage
sa me for both input and output. The exceptions are noted in the following
discussion.
406 Section 7.4 Formatting Input/ Output Functions
scanf /fscanf
maximum size conversion
width code
. Suppress h *
short
.
4
d i, u, o, x,
J I
I
L
long int
double
long double
c, s, p, n,
..
a t e, g, [
- left justify
+ sign ( + or - ) hh char
.. . .
d i, u, o x, X,
. . .
space if positive h short
I long int f, e E, g G, a A,
0 zero padding c s, p n, %
# alternate L long double
T T T
minimum
precision size conversion
% flag width code
printf /fprintf
FIGURE 7- 5 Conversion Specifications
Conversion Codes
The conversion code specifies the type of data that are being formatted. For
input, it specifies the type of variable into which the formatted data arc
stored. For output, it specifies the type of data in the parameter associated
with the specification. It is our responsibility to ensure that the data are of the
right type. II they are not, strange formatting may result. Conversion codes
are discussed in detail in the input Formatting { scanf andfscanf )" and " Out -
put Formatting ( printf and fprintf )" sections that follow.
Input Formatting ( scanf and fscanf )

I he scanf and fscanf functions read text data and convert the data to the types
specified by a format string. I he only dilference between them is that scanf
reads data from the standard input unit ( the keyboard by default ) and fscanf
reads the input from a file specified by the first parameter. This file can be
standard input ( stdin ).
scanf reads from stdin; fscanf reads from a user - specified stream.
I he name scanf stands for “scan formatted ; the stands lor

” name fscanf
“ file scan formatted.” These functions have the follow ing formats:
scanf ( " format string" , address list

);
fscanf(sp, "format string " , address list
);
\\ here sp is the address of a stream defined as ”
type FILE*, “format string is
a string containing formatting instructions, and the address list specifies the
n >
addresses where the data are to be stored alter they have been formatted. A
comma must separate the format string from the variable list . If more than one
vaiiable is present, then separate the variables from each other by commas.
We must include a variable address in the address list for every Conver -
sion specification in the format string that requires data. If we do not, the ts
result is unpredictable and undefined.” This is a standard disclaimer that
means anything can happen. I he usual result is that the program doesn ’t do
what we expect and often crashes.
Input Data Formatting

I he conversion operation processes input characters until any of the follow -
ing occur:
. .
1 End ol file is reached
2. An inappropriate character is encountered.
3. I he number of characters read is equal to an explicitly specified maxi -
mum field width.
Input Conversion Specification

In C hapter 2 we introduced some ol the conversion specifications We dis - .
cuss the rest here. Table 7 - 3 shows the type of the pointer argument, size,
.
and conversion codes for scanf family Note that we have not shown a Hag
,
specification in the table, because there is only one flag suppress ( * ) , which
can be used with all input format codes.
Argument Type Size Specifier Code
integral hh (char), h ( short), none (int), I (long), II

(long long)
integer h ( short), none (int) , I (long) . II (long long) d
unsigned int hh (char), h ( short), none ( int) , I (long), II

(long long)
character octal hh ( unsigned char)
integer hexadecimal h ( short) , none (int), I (long) , II (long long )
real none (float), I (double) , L (long double) f
real (scientific ) none ( float), I ( double) , L ( long double)
real ( scientific ) none ( float), I ( double) , L ( long double) 9
real (hexadecimal) none (float) , I (double) , L (long double) a

continued
TABLE 7- 3 Sizes and Conversion Code for scanf Family

character none (char), I (wcharj) c
string none (char string), I ( wcharj string) s
pointer P
integer (for count) none int), hh (char ), h ( short), I (long) , II n
(long ong)
set none (char), I ( wcharj) [
TABLE 7- 3 Sizes and Conversion Code For sconf Family ( continued )
Flag
There is only one (lag for input formatting, the assignment suppression Hag (*).
More commonly associated with text files, the assignment suppression flag
tells scanf that the next input field is to he read but not stored. It is discarded.
The following scanif statement reads an integer, a character, and a floating-
point number from the input stream. The character is read and discarded.
The other fields are read, formatted, and stored. Note that there is no match-
ing address parameter for the data to be discarded.
scanf ( " % d % *c % f ", &x , & y ) ;
Width
I he width specifies the maximum width of the input ( in characters) This .
allows us to break out a code that may he stored in the input without spaces .
Consider what happens when we read a Social Security number from a file
anti the number has no formatting; it is just nine digits in a row, followed by a
space. We could read it into three variables, thus allowing us to format it with
dashes in our output , with the following format specifications:
scanf( " % 3 d % 2 d % 4 d..." , & ssnl , & ssn 2, & ssn 3 , ...);
Note that the width is a maximum. If the amount of data is less than
required, the scan terminates. What determines the end of the data depends
on the type of data, but generally whitespace will
terminate most scans.
Size
The si /.e specification is a modifier for the conversion code. Used in combi-
nation with the conversion code, it specifies the
type of the associated vari-
able, lor example, a long double (Lf ) . The
size codes, along with their
associated conversion codes, are explained in Table 7 3, Sizes and Conversion
Code tor scan ) Family.”
- "
II
Conversion Codes
\\ ith the exception of string and pointer, this section discusses the input con -
.
version codes I he pointer conversion code is discussed in Chapter 9 ; the
string conversion code is discussed in Chapter 1 1 .
' 4«
Integer (d ) I he decimal ( d) format code accepts a value from the input stream
and formats it into the specified variables. It reads only decimal digits and an
optional plus or minus sign as the first character of the value.
Integer ( i ) The integer format code (i) allows the user to key a decimal, octal,
or hexadecimal number. Numbers starting with any digit other than zero are
read and stored as decimal values. Numbers starting with / ero are interpreted
as octal values and are converted to decimal and stored. Hexadecimal num-
bers must he prefixed with Ox or OX; the hexadecimal value is converted to
decimal and stored. We strongly recommend that the integer format code be
used only when non-decimal data is being entered—such as when program -
mers are entering technical data; in all other cases, the decimal format code
( d) should be used.
Octal and Hexadecimal (o, x ) The octal ( o) and hexadecimal ( x ) conversion codes
perform unsigned conversion. For octal, the only valid input digits are 0...7.
For hexadecimal input, the valid digits are 0... 9, a... f, and A...F. If you are
not familiar with octal or hexadecimal numbers, sec Appendix I). Numbering
"
Systems."
Scientific Notation (e , g , a ) The C language uses three real format codes lor scien -
tific notation. In scientific notation, the significand and exponent are speci -
fied separately. The significand part is a floating-point number that contains
as many significant digits as possible. For example, if it contains six digits,
then the number is significant only to six digits; if it has twelve digits, then it
is significant to twelve digits. The larger the significance, the greater the pre -
cision. Therefore, long double may he more precise than double, which may
he more precise than float .
The exponent specifies the magnitude of the number. It may be either
positive or negative. If it is positive, then the number is the significand times
ten to the power of the exponent, which may be a very large
number. If it is
negative, then the number is the significant times the reciprocal of the base
ten exponent , which may be a very small number . These forms are
shown below.
123e03 123* 103 -

123e 03 -> -
123* 10 3
All of the following numbers are in scientific notation:
2e2
3e -1.0e-3 0.1+el
d
For input , all conversion codes ( f , e , g , a ) accept all three scientific nota -
tions as well as algebraic notation.
Count ( n ) To verify the number of input characters, we use the n conversion

code. This code requires a matching variable address into which scanf places
the count of the characters input. In the following example, count is a short
integer that is to receive the number ot characters read :
s c a n f ( " %d % 8 . 2 f % d % h n " , &i , &x , & j , &c o u n t ) ;
Input Side Effect and Value

When the scanf function is called , it can create a side effect and return a
value. The side effect is to read characters from a file and format them into
variables supplied in the parameter list . It continues until the end of the line
is reached or until there is a conflict between the format string and a charac-
ter read from the input stream . In either case , the scan returns the number of
successfully formatted values before termination. If the scan detects the end -
of -file before any conflict or assignment is performed , then the function
returns EOF. The difference between the side effect and the value is seen in
Figure 7- 6.
Reads and converts a stream of characters from the

input file, and stores the converted values in the list
of variables found in the address list .
scant side effect

or
fscant
Returns the number of successful data conversions.

If end of file is reached before any data are
converted, it returns EOF .
value
FIGURE 7- 6 Side Effect and Value of scanf and fscanf
I bis leads us to a common example of a value

error. Suppose we want to
read two numbers with one scan statement. If
we succeed, the scan function
returns a 2 , indicating that both were read
correctly. If the user makes a mis *
take with the first number, the scan function

returns a 0. If there is a mistake
with the second entry, scanf returns a 1 .
This allows us to validate that at least
the correct amount and type of data were
read . An error rea ding the second
piece of data is shown in Figure 7 - 7 .
TTI
Input stream Should have

before > been a digit .
printf( " \nEnter price and amount: " ); 2 3 r ^n
Ui *
result = scanf( "%d %d " , &a , & b);
:
^ scant aborted.
23
a
?
b
D
result
Value of a is 23!
FIGURE 7-7 Another Common Error
When we read data, we should verify that the data were read correctly.
This is easily done by checking the return value from the scan function.
Program 7 - 2 is a program fragment that shows how the two numbers in
Figure 7 - 7 could he checked.
PROGRAM 7- 2 Checking sconf Results

1 # define FLUSH while ( getchar() != \n * )
2 -
# define ERR 1 " \aPrice incorrect. Re enter both fields\n"
3 # define ERR 2 " \aAmount incorrect. Re-enter both fields\n "
4
5 // Read price and amount
6 do
7 {
8 printf( "\nEnter amount and price: " )?
9 ioResult = scanf( " %d %f " , & amount , & price);
10
11 if ( ioResult != 2 )
12 {
13 FLUSH ;
14 if (ioResult == 1)
15 printf( ERR 1);
16 else
17 printf( ERR2);
18 } // if
19 } while ( ioResult 1 = 2 );
Results:
Enter amount and price: ? 15.25
-
Amount incorrect. Re enter both fields
Enter amount and price: 100 ?

-
Price incorrect. Re enter both fields
Enter amount and price: 100 15.25

Program 7- 2 Analysis We have introduced a new define statement, FLUSH , in this program segment.
# define FLUSH while ( getchar ( ) !- ’ \ n ' )
Why is this statement necessary? Recall that when scant encounters an error, it
leaves the invalid data in the input stream. If we simply print the user messages and
then return to the scant statement, the invalid data are still there. We need to get rid of
them somehow! This is the purpose of the FLUSH statement. It reads all data from the
invalid input to the end of the line. For a complete discussion of the flush statement, see
"Testing for Valid User Input" in the software engineering section.
Also note how the error messages tell the user exactly what went wrong, rather
than giving a general "The numbers entered are incorrect" type of message. Whenever
possible, you should tell the user exactly what went wrong.
Two Common Input Format Mistakes

There are as many mistakes to be made as there are programmers program -
ming. Two are so common that they need to be emphasized .
Invalid Address
Fhe first common mistake is to use a data value rather than an address for
the input parameter. The scan function places the formatted input at the
address specified in the parameter list . When we pass data rather than an
address , C interprets the data as an address . This causes a part of memory to
be destroyed . II we are lucky, our program fails immediately. If we are
unlucky, it runs until the destroyed part of the program is needed and then
fails . This problem is shown in Figure 7 - 8 .
printf( " \ nPlease enter number of dependents:

scanf ( " %4d " , a ) ;
A missing address
operator will cause
vour program to fail.
FIGURE 7- 8 Missing Address Operator in scanf
Data type Conflict

I he second common mistake is a conflict between
the format string and the
input stream . This occurs, for example , when
the format string calls for a
numciic \ alue and the input stream
contains an alphabetic character. II any
conflict occurs between the format string and
the input stream , the scanf
operation aborts . The character that
caused the conflict stays in the input
stream , and the rest of the input
stream ( including the troublemaker )
remains waiting to be processed . The problem
is that the input stream cannot
n
he filied again until it has been completely processed or flushed by the pro-
gram. I he next input operation will read from the unused stream instead of
1
reading from a new input stream. The result : Everything is a mess!
In figure 7 -9, the user meant to enter the number 235 but instead
entered 23r. ( I his is a common error. Note that the r is just below and to the
lei t of the 5 on the keyboard.) When scanf reads the input stream, it interprets
the number 23 but stops with the r and returns 1 , indicating that one decimal
number was successfully formatted. The r remains in the input stream, wait -
ing lor the next scanf . The typical result of this error is a perpetual loop look -
ing for an integer value. Even it we detect and flush the error when the r is
read, we still have processed the wrong data ( 23, not 233 ) in the first read!
Input stream Should have

before scantl . been a 5 ,
2 3 r \n
printf(" \nPlease enter the price : ");
scanf( " %d " , &a); Input stream
after scanf ,
s scanf completed x) ,
V V a l u e o f a i s 2 3! v 2 3 r \n
FIGURE 7 - 9 Data Type Conflict
Figure 7 - 10 contains three more examples that show the operation of a

fscanf function. In the first example, the format string requests that two inte -
gers be read, and two are successfully read. The value returned by the scanf
function is 2. In the second example, two more numbers are requested, but
only one is available in the input stream. In this case, the value of the scanf
expression is 1 . The third example shows the result ii all the data have been
read and the input file is at end of file.
Input
n=2 Stream
n = fscanf ( fp, " %d %d " , &a , & b); 1234 752 -
i
n=1
%d ", &a , & b); 1234 F -•
n = fscanf ( fp , " %d
n = EOF
<EOF>
n = fscanf ( fp, " %d %d " , &a, & b);
FIGURE 7- 10 fscanf Examples

Input Examples
In this section , we present several examples of scanf loimatting. Some ol
them are rarely used , hut they illustrate some of the finer points of format-
ting. We suggest that you cover up the answer until you have tried to write
the statement yourself .
—
EXAMPLE 7- 1 Write the scanf statement that will read three values a real number, an inte-
—
ger, and a character from the standard input file . I he input is
3.1416 31416 x
scanf( " % f %d %c ", & fNum, & iNum, & aChr );
—
EXAMPLE 7- 2 Write the scanf statement that will read three values a real number, an inte-
—
ger, and a character from the standard input file. All data are on different
lines. The input is
3.1416
31416
X
scanf ( " % f %d %c", & fNum, & iNum , & aChr );
Note that this statement is the same as the first example. Since the new-
line is whitespace, it makes no difference if the data are separated by new-
lines, spaces, or tabs. However, the following code fails when it tries to read
the character!
scanf ( " % f % d % c ", & fNum, & iNum, & aChr );
Do you see the mistake - Recall that the character conversion code ( c )
J
does not skip whitespace. I herefore, it reads the return character from the
integei input. Io prevent this problem , code a space before the character for-
mat specification. The space discards all whitespace
characters before it
reads the character.
EXAMPLE 7- 3 Write the scanf statement that will read two fractions, such as 14 /26 and 25/
66 , 11 om tlu standard input unit and place the numerators
and denominators
in integer variables.
scanf ( " % d / % d % d / % d ", & nl, & dl,

& n 2, & d 2 );
EXAMPLE 7- 4
-
\ \ rite the fscanf statement that will read a date , such as 5 10- 1936, Format -
ted with dashes ( - ) into three different integer fields. The input is to he read
1
1
from a file opened as spData.
fscanf ( spData, " %d -% d -% dM , & month, & day, & year );
Note that with this statement , an error results if the user enters 5 / 10/ 1997 .
EXAMPLE 7-5 \ \ rite the fscanf statement that will read three integers from a file opened as
spData and assign the number of successfully formatted variables to the
identifier result .
result = fscanf( spData, " %d %d %d ", & i, & j, & k );
EXAMPLE 7-6 Given a file with four integers in each line , write a fscanf statement to read
only the first , second , and fourth integer from a line; that is, discard the third
element . Read the data from the file spData.
result = fscanf ( spData , " %d % d % * d % d " , & a, & b , & d );
EXAMPLE 7- 7 Write the fscanf statement that will read three integers from a file opened as
spData and assign the number of successfully formatted variables to the
identifier result. Use a defined constant for the format string.
# define FORMAT 3 D " %d % d %d "
result = fscanf( spData, F0RMAT_ 3 D, & i , & j, & k );
Input Stream Concerns

There are two more points we need to discuss about scanf and how it handles
the input stream .
1 . The input stream is buffered. The operating system does not pass the
data in the input stream until we press the return key. I bis means that
there is always a return character at the end of the stream.
2 . The scanf function leaves the return character in the buffer by default.
If
we want the buffer to he empty, we must read the return character and
discard it . We read it using a character conversion specification with a
suppress flag. When the return character remains in the buffer, the next
scanf discards it for all conversion codes except character and scan set
.
Alternatively, we can force it to he discarded by placing a space at the
beginning of the format string or before a conversion specification
.
3. The scanf function does not terminate until the 1 ormut sii ing is exhausted ,
that is , until all specified operations are complete. W hitespace characters
in the format string cause whitespace characters in the buffer to be read
and discarded. When a whitespace character occurs at the end of the for-
mat string, scanf reads and discards the return character in the buffer
and then waits for more input. It cannot terminate until it finds a non -
whitespace character.
Lets look at some examples.
EXAMPLE 7-8 In this example, we have just one scanf call in our program that reads two
integers. In this case , the return character remains in the buffer, but it is not
a problem. When the program terminates, the operating system flushes all
buffers.
scanf ( " %d %d ", &a, &b);
EXAMPLE 7-9 In this example, we explicitly consume the return character by reading and
suppressing it.
scanf (" %d %d %*c", &a, & b);
EXAMPLE 7- 10 In this example, we let the return key stay in the buffer and let the second
call discard it . Note that in this case, it does not matter if the next scanf fol -
lows immediately or after several lines of code. The first conversion specifica-
tion ( % d ) at the beginning ol the second scanf discards the return character
left hv the first scanf I he return character for the second input stream , how-
ever, remains in the buffer.
scanf ( " %d %d " , &a , &b);

scanf ( " %d %d " , &c, &d );
EXAMPLE 7- 1 1 When the first format specification in a scanf format string uses a character
oi edit set conversion code , any
return character left in the buffer by a previ-
H I S scanf operation must he
(
manually consumed . This can be problematic
because the leftover return character is read first as shown below.
// Does not work because %c

-
reads the left over return
scanf ( " %d %d ", r&numl , &num2 );
scanf ( "%c %d " &aChr , &numl );
One solution would be to put a whitespace

character at the beginning ol
t second scanf to discard the previous return
ic
character. The whitespace
before tc in the second scanf discards the return
character.
1
1 \ 1
// Preferred Solution
scanf ( " %d %d " , &numl , &num2);
scanf (" %c %d" , &aChr, &numl);
Another solution would be to discard the return character at the end ot

the previous format string. Note that as the number of scanf statements in a
program grow, the safer and preferred method is to discard the return charac -
ter at the beginning of the format string. In the following code, the character
conversion specification ( % * c ) consumes the newline character at the end of
the stream .
Alternate Solution:
scanf ( " %d %d%*c" , &numl, &num2);
scanf ( " %c %d " , &aChr , &numl );
EXAMPLE 7- 1 2 The scanf function can hang our program . For example, in the following frag-
ment of code , we expect that the printf function be called after we entered
two integers . The program , however, hangs . The reason is that the whitespace
at the end of the control string tells scanf to discard any whitespace it sees in
the buffer. Therefore, scanf deletes our return character. But , to complete its
operation , scanf must find a return character. It does not matter how many
return keys we enter on the keyboard , scanj discards all ol them looking lor a
non - whitespace character.
// Note space at end of first format string ,

scanf ( " %d %d " , &a , & b);
printf( "Data entered: %d %d \n " / a , b );
scanf (" %d %d" , &c, &d );
Sample run:
10 20 <return>
<return>
We can force scanf to complete by entering any non - whitespace charac -

ter. Because scanj wants to format only two numbers, any character
causes it
to com plete successfully. The following example demonstrates how we can
force scanj to complete.
// Note space at end of first format string

,
scanf ( "%d %d " , &a , & b);

printf( " Data entered: %d %d\n" , a , b);
scanf ( "%d %d " , &c , &d );
Sample run:
10 20<return> // scanf hangs here
x<return> // extra character to force end
Data entered: 10 20
Note, however, that the extra character we enter ( x ) remains in the buffer
and will be read next time scanf is executed . This extra character is undoubt -
edly not valid for the next read , which means that our program will produce
invalid results.
Discarding the return character is different from consuming it. Discarding

can be done by whitespaces in the control string; consuming can only be done
by reading the return character with a %c.
Output Formatting ( printfand fprint! )

Two print formatted functions display output in human readable form under
the control of a format string that is very similar to the format string in scanf .
When necessary, internal formats are converted to character formats. In
other words, it is the opposite of the scan formatted functions.
These functions have the following format:
printf( "format string" , value list )
fprintf( sp , " format string " , value list )
One of the first differences to note is that the value list is optional.
Whereas you always need a variable when you are reading, in many situa-
tions , such as user prompts , you display strings without a value list .
Three examples of printf output that might be found in a program arc
shown below. The first is a greeting message at the start of the program, the
second displays the result of a calculation, and the last iis a closing message.
printf (" \nWelcome to Calculator.\n ");

printf ( " \nThe answer is %6.2f \n " , x ) ;
printf ( "Thank you for using Calculator " );
The fprintf function works just like printf except that it specifies the file
.
in which the data will be displayed The file can be the
standard output ( stdout )
.
or standard error ( stderr ) files For example, to write the three
previous lines
to a report file , we use the following code:
fprintf (spReport, "\nWelcome to

Calculator.\n" ) ;
fprintf (spReport, "\nThe answer is
%6.2f\n" , x );
fprintf (spReport, "Thank you for
using Calculator " );
Print Conversion Specifications

In Chapter 2 we introduced some of the
conversion codes. Table 7 - 4 shows
the flags, sizes, and conversion code for
printf family.
' 1 1'
Argument Type Size Specifier code

integer - +, 0, space
/ hh (char ) , h ( short) , none (int) d, i
I (long ) , II (long long)
1
v' l
unsigned integer +, 0, space hh (char), h ( short), none (int), u
I (long), II (long long)
integer ( octal) +, 0, #, hh (char), h ( short), none (int), o
space I (long), II (long long)
integer (hex ) -, + , 0, # , hh (char ) , h ( short) , none ( int) , x, X
space I (long ) , II (long long)
real +, 0, #, none (float), I ( double) , f
space L ( long double)
real (scientific) +, 0, #, none ( float), I (double), e, E
space L (long double)
real ( scientific ) +, 0, #, none ( float) , I ( double) ,

L (long double)
g G .
space
real (hexadecimal) +, 0, # , none ( float), I ( double), a, A
space L (long double)
character none (char ) , I ( w-char) c
string none ( char string), I ( w-char s

string)
pointer P
integer (for count) none (int), h (short), I (long) n
to print % %
TABLE 7 - 4 Flags, Sizes, and Conversion Codes for printf Family
Flog
There are four output flags: justification , padding, sign , and alternate form
.
Table 7 - 5 shows the output flag options.
Justification none right justified

left justified
continued
TABLE 7- 5 Flag Formatting Options

Padding none space padding
0 zero padding
Sign none positive value: no sign
negative value: -
positive value: +
negative value: -
space positive value: space
negative value: -
Alternate # print alternative format
for real, hexadecimal,
and octal.
TABLE 7- 5 Flag Formatting Options (continued )

Justification The justification flag controls the placement of a value when it is
shorter than the specified width. Justification can be left or right. If there is no
flag and the defined width is larger than required, the value is right justified. The
default is right justification. To left justify a value, the flag is set to minus ( -).
Podding The padding flag defines the character that fills the unused space
when the value is smaller than the print width. It can he a space, the default,
or zero. If there is no flag defined lor padding, the unused width is filled with
spaces; if the flag is 0, the unused width is filled with zeros. Note that the
zero flag is ignored if it is used with left justification because adding zeros
after a number changes its value.
Sign Flog I he sign flag defines the use or absence of a sign in a numeric value .
We can specify one of three formats: default formatting, print signed values,
or prefix positive values with a leading space. Default formatting inserts a
.
sign only when the value is negative Positive values are formatted without a
sign . \ \ hen the flag is set to a plus ( + ), signs are printed for both positive and
negative values. If the flag is a space, then positive numbers are printed with
a leading space and negative numbers with a
minus sign.
Alternate Flag The alternate flag ( # ) is used with the real, engineering, hexadeci-
mal, and octal conversion codes. The alternate flag is discussed with the con-
version codes to which it applies.
Width
.
I he meaning of the width field is
In t in the opposite direction.
conceptually the same for input and output,
For input, the width specifies the maximum
> p
width. I or output specifications, the width provides the minimum output width.
However, if the data are wider than the specified width, C prints all the data .
Precision
Precision is specified as a period followed by an integer. Precision has mean -
ing only lor output fields, and it is an error to use it for input. Precision can
control the following: 5
1 . I he minimum number ol digits for integers. If the number has fewer sig-
nificant digits than the precision specified, leading zeros will be printed.
.
2 The number of digits after the decimal point in float.
3. The number of significant digits in g and G output .
Size
The size specification is a modifier for the conversion code. Used in combi -
nation with the conversion code, it specifies the type of the associated vari -
able, lor example, a long double (Lf ) . The size codes with their associated
conversion codes are explained in Table 7 - 4, “Flags, Sizes, and Conversion
Codes for printf Family.”
Conversion Codes
With the exception of string and pointer, this section discusses the input con -
version codes. I he pointer conversion code is discussed in Chapter 9 ; the
string conversion code is discussed in Chapter I 1 .
Integer (d and i ) The value is formatted as a signed decimal number. I he integer

format code (i) is identical to the decimal code ( d) for output .
Unsigned (u ) The value is formatted as a unsigned decimal number.
Octal ond Hexadecimal (o, x, xj The value is formatted in either octal or hexadecimal,
as indicated by the conversion code. The alpha hexadecimal codes are printed
in lowercase if the x code is used and in uppercase for the X code. If you are
not familiar with octal or hexadecimal numbers, see Appendix I , Numbering
) “
Systems.”
When the alternative format flag is used, the hexadecimal values are pre-
fixed with ox or OX and octal numbers are printed with a zero prefix.
Real ( f ) The real format code ( f ) is discussed in Chapter 2. If the precision is
0, the decimal point is printed only if the alternate llag ( // ) is also present.
Scientific notation (e , E, g, G, a, AJ These codes are the same as described for the scanf .
The e and E codes format the number using ( f ), precision may be used
to control the number of digits after the decimal point
. If no precision is
used. II the precision is 0, no digits aie used altei the
specified, six digits are
5 . In Chapter I I , " Strings." we introduce another use of precision

.
decimal point and the decimal point itself is shown only when the alternate
form flag ( # ) is used .
The g and G codes are also scientific notation. However, the scientific
notation is used only when the exponent is less than -4 or greater than the
specified precision . Otherwise, the real formatting rules are used . When the
value is Formatted in scientific notation , the uppercase G iormats the number
with an uppercase E .
The a and A codes generate signed hexadecimal fractional representa-
-
tions with a decimal power-of two exponent. The formatted result with A is
formatted as ±0Xh.hhhP +dd, where h represents hexadecimal digits and d
represents decimal digits.
Count (n ) To verify the number of output characters, we use the n conversion

code . This code requires a matching variable into which printf places the
count of the characters output . Since the operation places the results in the
variable, its address must he used in the parameter list . In the following
example, count is a short integer that is to receive the number of characters
written ( note the address operator ):
printf( " %d %8.2f %d % hn" , i, x , j , & count ) ;
Pefcent (% ) The percent sign as a conversion code is used to output a percent

sign , l or example , assume that grade contains a student ’s score expressed as
a percentage. The following statement could then he used to print the grade
in the format 93.5 %. Note that there is no output value for the percent con-
version code.
printf( " %4.If %% " , grade );
EXAMPLE 7- 13 Demonstrate Scientific Formatting

I he code fragment shown below demonstrates the use of scientific iormat-
ting. We print the same real number first in real number format ( %f ) , then in
three scientific notations(%e, %G, and %A).
double x = 1234.5678;
printf( "|% #.4 |
f |% #.4e||% #.4G
||% #.4A|" , X , x, x, x ) ;
Results:
|1234.5678| ||1235.||0X 1.34A4P+10|
1.2346e+03
I lie fourth output needs a little discussion . To verify that it is correct , we

must convert the hexadecimal number to decimal. Using the following calcu -
lation , we see that 1.34 A4 is approximately equal to 1 234.56.
1 2
1 x 16 + 3 x 16 3
(
+ 4 x 16 + A x 16 4
+ 4 x 16 ) x 2* °
F
Output Side Effect and Value

I he side effect of the print functions is to write text data to the output file.
I he value it returns is the number of characters written . If an error occurs,
EOF , which is defined in the stdio.h header file, is returned . These concepts,
which are summarized in Figure 7- 11 , are similar to the side effects and value
discussed with scan functions.
Converted internal data, as required, to strings of

characters and writes the converted values to a file,
which may be the standard output or error file.
side effect
printf
and
fprintf
Returns the number of characters written to the

output file. In case of an error, it returns EOF.
value
FIGURE 7- 11 Side Effect and Value of printf
Output Examples
In this section we present several examples of print formatting. Although
some of them are rarely used , they do illustrate some of the finer points of
formatting. The sample output is shown in hold characters. You may want to
cover up the answer until you have answered the question yourself .
EXAMPLE 7- 14 Write the print statement that prints three columns of numbers. The first col -
umn contains a two-digit integer, the second column contains up to seven
digits, and the third column contains a float with four integral numbers and
three decimal places.
15 10 15.010
78 1234567 1234.123
printf( "% 2d %71d %8.3f" , i, j, x );
EXAMPLE 7- 15 Write the print statement that prints the tax due as stored in a float named x.
The output is
The tax is: 233.12
printf( "The tax is: %8.2f\n " , x ) ;

EXAMPLE 7 - 16 Modify the previous example to add the words ‘ dollars this year after the
”
numeric value. The output will then be

The tax is 233.12 dollars this year.
printf( "The tax is %8.2f dollars this year \n , x );
EXAMPLE 7- 17 Write the print statement that prints three integer variables to a file opened
as spOut. Each write is to append a newline character ( -» ) to the output.
100 200 300 * “
fprintf(spOut, " %d %d %d\n " , i, j, k);
EXAMPLE 7- 1 8 Write the print statement that prints three floating-point values. The first is a
float with two places of precision; the second is a double , printed in scientific
notation (G); and the third is long double , printed using scientific notation
(E). Each value prints on a separate line.
3.14
200
4.408946E+00
fprintf( spOut, " %.2f \n%G\n%E\n " , i, j, k );
EXAMPLE 7 - 1 9 Write the print statement that prints three integral values. The first is a short ,
printed in octal: the second is an int, printed using lowercase hexadecimal
digits; and the third is a long , printed using uppercase hexadecimal digits.
End the line with a newline character.
25547 7e44 7864CB
fprintf( spOut, " %ho %x %lX\n" , > j / k );
EXAMPLE 7 - 20 W rite the print statement that prints three real numbers in scientific notation
( e oi g ) as shown
below. Use the alternate format flag so that the value prints
with a decimal point even if it is zero.
3.1 7.893E 05 - 0.
printf("%#5.2g % #9.4G % # 5.0g\n " , i , j

k ); /
EXAMPLE 7 - 21 Write the print statement that prints a decimal

number in decimal, hexadeci-
mal, and octal format.
256 0X 100 0400
printf( " %5d % # 5X %#5o\n" , x , x , x ) ;
File Sample Programs

In this section we develop four small programs that demonstrate various
aspects of reading and writing files.
Read and Print File

Program 7- 3 is a simple program that reads a file of integers and prints
them out .
PROGRAM 7- 3 Read and Print Text File of Integers

1 /* Read a text file of integers , and print the contents.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
11 FILE* spin ;
12 int numln;
13
14 // Statements
15 spin = -
fopen( "P07 03.DAT" , " r" );
16 if (spin != NULL)
17 {
18 printf("Could not open file\a\n" );
19 exit ( 101 );
20 > // if open fail
21
22 while (( fscanf( spin , " %d " , &numln )) == 1)
23 printf( "%d " , numln );
24
25 return 0 ;
26 } // main
Results:
1 2 3 4 5 6 7 8 9 10
Program 7- 3 Analysis Note how we open the file in this small program. To make it easier to read the code,
into two statements. This does
we siplit the file open and the test for successful open
not affect the program 's efficiency and makes it much easier to code and read the
statements.
I
Now look at the read in the while loop. We loop as long as the fscanf succeeds, as
indicated by a return value of 1 . When the end of file is reached, the while statement
will be false and the program terminates. There is one slight risk to this program: If a
read error is caused by a bad disk, we get the same results. In Chapter 1 3 we show
you how to detect this type of error.
Copy File
In Program 7-4, we copy a text file of integers to a new file .
PROGRAM 7- 4 Copy Text File of Integers

1 /* Copy a text file of integers.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
11 FILE* spin ;
12 FILE* spOut;
13 int numln ;
14 int closeResult ;
15
16 // Statements
17 printf( " Running file copy \n" );
18
19 if (!spin )
-
spin = fopen( "P07 03.DAT" , "r " );
20 {
21 printf( " Could not open input file\a\n " );
22 exit ( 101 );
24
25 spOut = fopen( "P07 04.DAT" , "w" );
26 if (!spOut )
-
27 {
28 printf( "Could not open output file\a\n" );
29 exit ( 102 );
31
32 while (( fscanf(spin , " %d " , &numln
33
)) == 1)
fprintf( spOut, " %d\n " , numln
);
34
35 closeResult = fclose( spOut);
continued
n1
PROGRAM 7- 4 Copy Text File of Integers ( continued )

36 if ( closeResult == EOF )
37 {
38 printf( "Could not close output file\a \n " );
39 exit ( 201 ) ;
40 } // if close fail
41
42 printf( "File copy complete\n " );
43 return 0;
44 } // main
Results:
Running file copy
File copy complete
Program 7 4 Analysis Because this program has no display output, we start with a message that says we
*
will copy a file, and we end with a message that says the copy succeeded. Users
need to know that the program ran. If we don't print these messages, users will not
know if the program ran. In fact, it is a good idea to print start and end messages
with every program. This is a common standard in large production systems.
Now study the format string in statement 33 carefully. Note that it requires a new-
line after the data have been written. It is important to use whitespace between the data
in the file or it will not be readable. To verify that the data are correctly written, the file
should be read and printed by Program 7- 3 or a similar program .
Finally, note that we close the output file at the end of the program. While input
files don't need to be closed, you should always close output files and check to make
sure that they closed successfully. If the close fails, we print an error message. If it
doesn't, we print the successful copy message.
Append File
Now lets append some data to the end of a hie. After we wrote Program 7 - 5 ,
we ran it .
without a file Because append creates a file i ( it doesn t exist , the
’
first run created our file. Then we ran it a second time and verified that the
new data were appended to the original file .
PROGRAM 7- 5 Append Data to File
1 /* Copy a text file of integers.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
continuec
PROGRAM 7- 5 Append Data to File (continued )

11 FILE* spAppnd ;
12 int numln;
13 int closeResult ;
14
15 // Statements
16 printf( "This program appends data to a file\n " );
17 spAppnd = fopen( " P07-05.DAT " , "a" );
18 if (!spAppnd )
19 {
20 printf( "Could not open input file\a\n " );
21 exit ( 101);
22 } // if open fail
23
24 printf( "Please enter first number: " )?
25 while ((scanf( " %d", &numln )) != EOF )
26 <
27 fprintf( spAppnd , " %d \n " , numln );
28 printf( " Enter next number or <EOF>: " );
29 } // while
30
31 closeResult = fclose( spAppnd );
32 if (closeResult == EOF )
33 {
34 printf("Could not close output file\a\n ");
35 exit ( 201 );
36 } // if close fail
37
38 printf( "\nFile append complete\n" );
39 return 0;
40 > // main
Results:
This program appends data to a file
Please enter first number: 1
Enter next number or <EOF>: 2
Enter next number or <E0F>: 3
Enter next number or <EOF>:Ad
File append complete
Program 7 5 Analysis This program differs from Program 7-4 in two

ways. First, this program has only one
disk file . While we could have read the data to be
appended from a file, it is more
interesting and more flexible to get the data from the keyboard
. The second change is
the read loop to get the data. Because we are
reading from the keyboard, we need to
provide prompts for the user.
Chapter 7 Text Input/Output 429 1
H
Student Grades
Programs 7 - 3 to 7 - 5 were all written in main . As our last example, let ’s write a
program that reads and writes a student grades file to be turned in at the end
of the term. The code is shown in Program 7-6.
PROGRAM 7-6 Student Grades

lr
1 /* Create a grades file for transmission to Registrar.
2 Written by:
3 Date:
4 */
6
8 int getStu (FILE* spStu ,
9 int* stuID , int * examl ,
10 int* exam 2 , int* final );
11 int writeStu (FILE* spGrades,
12 int stuID , int avrg , char grade );
13 void calcGrade ( int examl , int exam2 , int final,
14 int* avrg , char* grade );
15
17 {
19 FILE* spStu ;
20 FILE* spGrades ;
21
22 int stuID ;
23 int examl ;
24 int exam 2 ;
25 int final ;
26 int avrg ;
27
28 char grade;
29
30 // Statements
31 printf(" Begin student grading\n" );
32 -
if (!( spStu = fopen ( "P07 06.DAT " , "r " )))
33 {
34 printf( " \aError opening student file\n " );
35 return 100;
36 } // if open input
37
38 if (!(spGrades = fopen ( " P07 06Gr.DAT" , "w " )))
-
39 {
continued
PROGRAM 7- 6 Student Grades ( continued )

40 printf( " \aError opening grades file\n" );
41 return 102 ;
42 > // if open output
43
44 while ( getStu
45 ( spStu , &stuID, &examl , &exam 2 , &final))
46 {
47 calcGrade ( examl , 6X 31112 , final, &avrg , &grade);
48 writeStu (spGrades , stuID , avrg , grade );
49 > // while
50
51 fclose (spStu );
52 fclose ( spGrades );
53
54 printf( "End student grading \n " );
55 return 0;
56 > // main
57
58 /* =======================getStu======================-
59 Reads data from student file.
60 Pre spStu is an open file.
61 stuID , examl , exam2 , final pass by address
62 Post reads student ID and exam scores
63
64
if data read
—— returns 1
if EOF or error returns 0
65 */
66 int getStu (FILE* spStu , int* stuID , int* examl ,
67 int* exam2 , int* final)
68 {
70 int ioResult;
71
72 // Statements
73 ioResult = fscanf( spStu, "%d %d %d %d" , stuID ,
74 examl , exam2 , final);
75 if ( ioResult == EOF )
76 return 0 ;
77 else if ( ioResult != 4 )
78 {
79 printf( " \aError reading data \n " );
80 return 0 ;
81 } // if
82 else
83 return 1 ;
con tinned
PROGRAM 7-6 Student Grades (continued )
84 } // getStu
85
86 /* ==================== calcGrade ===================

87
88
Determine student grade based on absolute scale.
Pre examl , exam 2, and final contain scores
n
89 avrg and grade are addresses of variables
90 Post Average and grade copied to addresses
91 */
92 void calcGrade ( int examl , int exam2 , int final ,
93 int * avrg , char * grade )
94 {
95 // Statements
96 *avrg = (examl + exam2 + final ) / 3;
97 if ( *avrg >= 90)
98 grade = 'A';
99 else if ( *avrg >= 80 )
100 grade = ' B';
101 else if (*avrg >= 70)
102 grade = 'C' ;
103 else if ( *avrg >= 60)
104 grade = 'D' ;
105 else
106 grade = 'F';
107 return;
108 > // calcGrade
109
110 /* ==== == writeStu ====
111 Writes student grade data to output file.
112 Pre spGrades is an open file
113 stuID , avrg , and grade have values to write
114 Post Data written to file
115 */
116 int writeStu (FILE* spGrades , int stuID ,
117 int avrg , char grade)
118 {
119 // Statements
120 fprintf(spGrades , "%04d %d %c\n" ,
121 stuID , avrg , grade );
122 return 0 ;
123 } // writeStu
Results:
Begin student grading
End student grading
con t i n net
432 Section 7.5 Character Input/Output Functions
PROGRAM 7- 6 Student Grades (continued )
Input
0090 90 90 90
0089 88 90 89
0081 80 82 81
0079 79 79 79
0070 70 70 70
0069 69 69 69
0060 60 60 60
0059 59 59 59
Output
0090 90 A \ n
0089 89 B \ n
0 0 8 1 81 B \ n
0079 79 C \ n
0070 70 C \ n
0069 69 D \ n
0060 60 D \ n
0059 59 F \ n
Program 7- 6 Analysis This program has several points to study. Let's start at the top. We open and close the
files in main . Often, programmers write subfunctions for program initialization and
conclusion, but in this case we decided to write them in main . The processing loop
uses the return value from getStu to control the loop ( see statement 44). if data are
read, then the return value is true ( 1 ); if there is an error or all data have been read,
the return value is false (0). This design results in a very simple while loop.
Within the loop are three calls: one to read the student data (in the while limit test),
one to calculate the grade, and one to write the grades file. Study the parameters care-
fully. We have used a combination of data values and addresses. Note especially the file
parameters. In the function headers, the file parameters are coded as FILE*, which you
should recognize as a file address. We do not use an address operator when we pass
the files, however, because they are already addresses. (See statements 44 and 48 .)
Now study the results. The student ID is written as four digits with leading zeros
because it is really a code, not a number. Note how we created the test data to verify
that all of the boundaries in our multiway selection work correctly. To make it easy to
verify, we made the student ID the same as the average . Witb these clues to the
expected output, it only takes a quick review of the results to confirm that the program
ran successfully.
7.5 Character Input/Output Functions

Character input functions read one character at
a time from a text stream .
Character output functions write one character at
the time to a text
stream . These functions can he divided into
two general groups: input/output
functions used exclusively with a terminal device and input/output func-

tions that can be used with both terminal devices and text files. They are
summarized in Figure 7- 12 .
1
Character
I/O
Terminal Any
Only Stream
Input Output Input Output Push Back
getchar putchar getc / fgetc putc / fputc ungetc
FIGURE 7 - 1 2 Character Input / Output Functions
C provides two parallel sets of functions , one for characters and one for
wide-characters. They are virtually identical except for the type. Because
wide-characters are not commonly used and because their functions operate
identically, we limit our discussion to the character type. The wide-character
functions are listed in Appendix F.
Terminal Character I/O

C declares a set of character input /output functions that can only be used
with the standard streams: standard input ( stdin) , standard output ( stdout ) .
Read a Character: getchor

The get character function (getchar) reads the next character from the stan -
dard input stream and returns its value. Only two input conditions stop it :
end of file or a read error. If the end of file condition results, or if an error is
detected , the functions return EOF. It is our responsibility to determine il
some condition other than end of file stopped the reading. The function
dec -
la rat ion is shown below.
int getchar ( void ) ;
might
Note that the return type is an integer and not a character as you
expect. The return type is an integer because EOF is defined as an integer
files . I here is
( i n t ) in the standard definition stddef .h and other headei
( lag is not a character.
another reason for this: C guarantees that the EOF
, or
set it is using : ASCII , EBCDIC
This is true regardless of what character
control characters in ASCII you will find none
whatever. If you examine the
,
for end of file. Traditionally, EOF is defined as -1, but this is not prescribed
by ANSI/ISO. An individual implementation could therefore choose a differ -
ent value.
Write a Character: putchor

The put character function ( putchar ) writes one character to the monitor. If
anv error occurs during the write operation, it returns EOF . This may sound
somewhat unusual, since EOF is normally thought of as an input file consider-
ation, but in this case it simply means that the character couldn’t be written.
The function declaration is shown below.
int putchar (int out char ); _

Again the type is integer. An interesting result occurs with this function:
When it succeeds, it returns the character it wrote as an integer !
Terminal and File Character I/O

The terminal character input/output functions are designed for convenience;
we don’t need to specify the stream. In this section, we describe a more
general set of functions that can be used with both the standard streams
and a file.
These functions require an argument that specifies the stream associated
with a terminal device or a file.
1 . \ \ hen used with a terminal device, the streams are declared and opened
by the system— the standard input stream ( stdin ) for the keyboard and
the standard output stream ( stdout ) for the monitor.
.
2 \ \ hen used with a file, we need to explicitly declare the stream. It is our
responsibility to open the stream and associate it with the file.
Read a Character: getc and fgetc

I he get functions read the next character from the file stream, which can be a user-
defined stream or stdin, and convert it to an integer. If the read detects an end of file,
the functions return EOF. EOF is also returned if any error occurs. We discuss howto
tell the difference in C hapter 13. I he prototype functions are shown below.
int getc ( FILE* spin);

int fgetc ( FILE* spin );
\ \ bile the getc and Jgetc functions

operate basically the same, for techni-
cal reasons we recommend that only fgetc be used.
An example of fgetc is
shown in the following section.
nextChar = fgetc (spMyFile);

m
Chapter 7 Text Input/Output 435 1
Write a Character: putc and fputc
I lie put functions write a character to the file stream specified , which can he
a user-defined stream , stdout , or stclerr. For fputc , the first parameter is the
character to he written and the second parameter is the file. If the character
is successfully written , the functions return it. If any error occurs, they return
EOF. The prototype functions are shown below.
int putc ( int oneChar , FILE* spOut);

int fputc ( int oneChar , FILE* spOut);
Once again , we recommend that only fputc be used . An example is

shown below.
fputc ( oneChar , spMyFile );
Note that we have discarded the return value. Although we may occa -
sionally find a use for it , it is almost universally discarded .
Push D Character Back: ungetc

rhe push hack functions insert one or more characters into a stream that has
been opened for input; that is, it writes to an input stream . When multiple
characters are pushed back , they are then read in reverse order ; that is the
last character pushed is the first character read . While multiple characters
can be pushed hack, we cannot push back more characters than we read .
After a successful push back, when the stream is next read , the character
pushed back will he read .
If the operation succeeds, the functions return the character; il they do
not succeed , as with too many characters being pushed into the stream , they
return EOF.
The unget character functions require that the first parameter be a char-
acter and that the second parameter be an input stream . I he function decla -
rations are shown below.
int ungetc ( int oneChar , FILE* spData );
The intent of these functions is to return one or more characters to the

input stream for a su bsequent read . This situation might occur when
we are
characters that prompt action , such as in a menu system , and the
reading
to rein -
wrong character is read ( see below). In these situations, it is cleaner
than to set a flag or otherwise pass
sert the character into the input stream
the character through several functions .
is a charac-
In the following example, we want to process any option that
ter. Numeric options are processed in another function
. Therefore , il we
it back on the input stream and exit
detect that an option is numeric, we put
the function .
'
option = fgetc (stdin);

if ( isdigit(option))
ungetc ( option , stdin );
else
{ >-
Character Input/Output Examples
This section contains examples of common text file applications.
Create Text File

Program 7 - 7 reads text from the keyboard and creates a text file. All data are i
stored in text format with newlines only where input by the user.
PROGRAM 7-7 Create Text File

1 /* This program creates a text file.
2 Written by:
3 Date: i
4 */
6 int main (void )
7 { i
9 FILE* spText ;
10 int c;
11 int closeStatus;
12
13 // Statements
14 printf( "This program copies input to a file.Xn ");
15 printf( "When you are through , enter < EOF>.\ n\n " );
16
17 if (!(spText = fopen( "P07-07.DAT ” ,"w " )))
18 {
19 printf( " Error opening P07 07.DAT for writing " );
-
20 return ( 1 );
21 } // if open
22
23 while ((c = getchar( )) != EOF )
24 fputc(c , spText );
25
26 closeStatus = fclose(spText );
27 if ( closeStatus == EOF )
28 {
continued
iff
PROGRAM 7-7 Create Text File ( continued )
29 printf( "Error closing file\a \n " );
30 return 100;
31 > // if
32
33 printf( " \n \nYour file is complete\n " );
34 return 0 ;
35 > // main
Results:
This program copies input to a file.
When you are through, enter <EOF>.
Now is the time for all good students

To come to the aid of their college.*d
Your file is complete
Program 7- 7 Analysis This simple program is the beginning of a text editor. The biggest element missing is
word wrap. Word wrap prevents a word from being split between two lines on a page.
Study the program carefully and note that we used integer for the character type
(c ) in statement 10. This is an important point whenever you are checking for end of
file, as we did in statement 23. Because EOF will not be correctly represented in a char -
acter, we must use integer. But, what about when we have a character ? C automatically
casts the integer to a character wherever it is necessary.
Copy Text File

Program 7-8 will copy one text file to another text file . It is more generalized
than Program 7- 4 , "Copy Text File of Integers ” because it will copy any file
data , not just integers.
PROGRAM 7- 8 Copy Text File

1 /* This program copies one text file into another.
2 Written by:
3 Date:
4 */
6 int main (void )
7 {
9 int c;
10 int closeStatus;
11 FILE* spl ;
12 FILE* sp2 ;
continuec
PROGRAM 7- 8 Copy Text File ( continued )

13
14 // Statements
15 printf("Begin file copy \n " );
16
17 if (!(sp1 = -
fopen ( "P07 07.DAT " , " r " )))
18 {
19
20 return ( 1);
-
printf( " Error opening P07 07.DAT for reading" );
21 > // if open input

22 -
if (!( sp2 = fopen ( "P07 08.DAT" , "w " )))
23 {
24
25 return ( 2 );
-
printf( " Error opening P07 08.DAT for writing" );
26 } // if open output
27
28 while ((c = fgetc( spl )) 1 = EOF )
29 fputc(c, sp2);
30
31 fclose(spl );
32 closeStatus = fclose( sp2 );
33 if ( closeStatus == EOF )
34 {
35 printf( "File close error.\a\n " );
36 return 201 ;
37 > // if close error
38 printf( "File successfully created \n" );
39 return 0;
40 > // main
Program 7-8 Analysis This program contains two style points that need comments. First, we have used
generic names for the files, spl and sp2. Since this program simply copies and cre-
ates text files, we cannot give the files names that reflect their data Better names
.
might be spin and spOut, but they are also generic.
Second, the program has two potential file open errors. We use different error
codes for those operating systems whose job control can distinguish between different
completion codes. We have also included the file names in the error messages so that
the user knows exactly which file had the problem .
Finally, a subtle point: Note that we have arranged the local declarations in order
of increasing complexity. First, we define the character and integer
types: then we
define the files. It is good practice to group the definitions by
type and in order of
increasing complexity.
Count Characters and Lines

I rogram / - 9 counts the number of
characters and lines in a program. Lines
arc designated by a newline. Also note that the
program guards against a file
that ends without a newline for the last line.
1
PROGRAM 7- 9 Count Characters and Lines

1 / * This program counts characters and lines in a program.
2 Written by:
3 Date:
4 */ '1
6 int main (void )
7 {
9 int curCh ;
10 int preCh ;
11 int = 0;
countLn
12 int countCh = 0;
13 FILE*spl ;
14
15 // Statements
16 -
if (!(spl = fopen("P07 07.DAT" , "r" )))
17 {
18 -
printf( " Error opening P07 07.DAT for reading " );
19 return ( 1 );
20 > // if open error
21
22 while ((curCh = fgetc(spl )) != EOF )
23 {
24 if (curCh != ' \n ' )
25 countCh++;
26 else
27 countLn++;
28 preCh = curCh ;
29 } // while
30
31 if ( preCh != * \n ' )
32 countLn++;
33
34 printf( "Number of characters: %d \n " countCh ); /
35 printf( " Number of lines : %d\n" countLn ); /
36 fclose ( spl );
37 return 0;
38 > // main
Results:
Number of characters: 74
Number of lines: 2
Program 7- 9 Analysis Program 7- 9 is rather straightforward. The only real problem is in making sure that
the end of the
the last line is counted even if there is no newline. We can check this at
file by making sure that the last character we read, stored in preCh, was a newline.
If it isn't, we add 1 to the line count.
Once again, we have used the integer type for our input character. Everything
works correctly, even statement 31, which compares the integer type (preCh) to a
character literal.
Count Words In File

Program 7- 10 counts the number of words in a file . A word is defined as one
or more characters separated by one or more whitespace characters that is,
by a space, a tab, or a newline.
—
PROGRAM 7- 10 Count Words
1 /* Count number of words in file. Words are separated by
2 whitespace characters: space , tab , and newline.
3 Written by:
4 Date:
5 */
7
8 #define WHT SPC\_
9 (cur == cur == '\n ' cur == '\f )
11 {
12 //Local Declarations
13 int cur ;
14 int countWd = 0;
15 char word = 'O' ; // 0 out of word: I in word
16 FILE* spl ;
17
18 // Statements
19 if (!( spl = fopen( "P07-07.DAT" , "r " )))
20 {
21
22 return ( 1 );
-
printf( " Error opening P07 07.DAT for reading");
23 } // if file open error

24 while (( cur = fgetc( spl )) != EOF )
25 {
26 _
if ( WHT SPC )
27 word = 'O ’ ;
28 else
29 if ( word = = • O ' )
30 {
31 countWd ++ ;
32 word = i ’ ;
continual
'
V
PROGRAM 7- 10 Count Words (continued )

33
34
35
>
> // else
// while
printf("The number of words is: %d \n" , countWd );

I
36
37 fclose( s p l ) ;
38 return 0;
39 } // main
Results:
The number of words is: 15
Program 7- 10 Analysis The selection logic for this problem is similar to the previous program. We must deter -
mine when we are in a word and when we are between words. We are between
words when we start the program and whenever we are at whitespace. To keep track
of where we are, we use a flag, word. When word contains the letter I, we are in a
word; when it contains the letter 0, we are out of a word. We increment the counter
only at the beginning of a word, when we set the word flag toI.
Note that the problem handles multiple whitespace characters by simply setting the
word flag to 0. Note also how we use a preprocessor define statement to define
whitespace. This has no effect on the efficiency of the program, but it makes it easier
to read.

Testing Files
Testing files can be a very difficult task for two reasons. First , many errors
cannot be created through normal testing. Among these types of errors arc
those created by bad physical media , such as disk and tapes. Often , disks and
tapes become unreadable. Good programs test for these types of errors , but
they are difficult to create.
Second , there are so many ways things can go wrong. The test plan there-
fore needs to ensure that all possibilities have been tested . Chief among them
is the case where a user enters a non - numeric character while trying to enter
a number. If we don ’t provide for this situation , our program will not work
properly. In fact , in some situations, it will not work at all .
Testing for Volid User Input

Consider the following code to read an integer value. If the user miskeys the
data , the program will go into an infinite loop.
printf( " \nPlease enter Number of Units Sold: " );

while (scanf( " %d " , &unitsSold ) != 1 ) \
// scanf returns 1 if number read correctly
printf( " \nlnvalid number. Please re-enter.\n " );
In this case , it looks as if the programmer has done everything correctly.

I he program checks for valid data being returned. A user prompt shows what
data should be entered ; the program checks for valid data being read ; and if
there were errors, provides the user with a good error message and repeats
the scanf . What is wrong?
I he problem lies in what scanj does when it finds an invalid first digit . If
the number is entirely wrong, it leaves the data pending in the input stream.
I his means that while the user sees the error message, the invalid data are
still in the input stream waiting lor a scan] that will properly read and process
them. Remember that scanf thinks that these “ invalid data are the beginning
ol the next field . It is our job as a programmer to get rid of the “ bad data .’ We
can do this with a small piece of code that is commonly named FLUSH. Its
purpose is to read through the input stream looking for the end of a line.
\ \ hen it finds the end ol the line, it terminates. Let 's first look at the code,
then we will look at an easy way to implement it.
while (getchar( ) != '\n')
Examine this statement carefully. All it does is

get a character and then
throw it away. I hat is, the character it reads is not
saved . Now examine the
expression that is controlling the while. If getchar
reads any character other
lit
j
'
than a newline, the statement is true. It it reads a newline, then the state-
ment is false. The while statement will therefore read the input stream and
throw away all characters until it sees a newline. When it finds the newline, it
will throw it away too, but at that point, the loop stops. In other words, it
flushes the input stream to a newline character. This is exactly what we want
‘II '
to do when the user accidentally keys the wrong character when the program
expects a digit.
One more point . Note that we coded the null statement on its own line.
I his is to make it clear that we intended to code a null statement . It also
eliminates a warning messagefrom the compiler —good compilers look for
common errors like unintended semicolons.
Now, what ’s the easiest way to implement this handy statement ? Well, we
could simply code it everywhere you need it, but chances are that we would
make some mistakes that would have to be found and debugged. A better
solution is to use the preprocessor define declarative and code the statement
only once, as shown below.
# define FLUSH while ( getchar( ) != '\n')
Note that there is no semicolon at the end of the define declarative. We

could have put one there, but it ’s better to put the semicolon after the FLUSH
in the program. Our error code can now be changed to handle this type of
error, as shown in Program 7 - 11.
PROGRAM 7- 11 Handling Errors — the Right Way

1 # define FLUSH while ( getchar() != ’ \n ' )
2
3 printf( " \nPlease enter Number of Units Sold: " );
4 while ( scanf( " %d " , &unitsSold ) != 1 )
5 {
6 // scanf returns 1 if number read corrrectly
7 FLUSH ;
8 -
printf( " \alnvalid number. Please re enter: " );
9 } // while
Value Errors
In Chapter 6, we discussed some of the techniques for data validation. But
the subtleties of data validation can be quite complex. Consider another type
of human error, the partially correct input. In our program above, we
assumed that the user entered invalid data for the first digit of the number.
What if the error occurs on the second or third digit r Then the scanf function
is happy, for the time being anyway, and returns a 1,
indicating that it read
one number successfully. How do we guard against this ?
The best way is to echo the input to the user and ask for verification that
it is correct . Although this greatly slows down the input process , for critical
data it is necessary. The code for this situation is shown in Program 7- 12.
Here we present it as a complete function.
PROGRAM 7- 1 2 Handling Errors with Explanations

1 /* This function reads the units sold from the keyboard
2 and verifies the data with the user ,
3 Pre nothing
4 Post units Sold read , verified , and returned
5 */
6 int getUnitsSold ( void )
7 {
9 int unitsSold ;
10 bool valid ;
11
12 // Statements
13 do
14 {
15 printf( "\nPlease enter Number of Units Sold: " );
16 while (scanf( " %d " , & unitsSold ) != 1 )
17 {
18 FLUSH;
19
20 > // while
-
printf( " \alnvalid number. Please re enter: " );
21 printf( " \nVerify Units Sold: %d: " , unitsSold );

22 printf( "<Y> correct: <N> not correct: \n " );
23 FLUSH ;
24 if (toupper(getchar ( )) = = Y*)
25 valid = true ;
26 else
27 {
28 FLUSH ;
29 printf( " \nYou responded ' no.' " );
30
31 valid = false;
-
printf( "Please re enter Units Sold \n " );
32 > // if
33 > while (!valid );
34 return unitsSold ;
35 } // getUnitsSold
Program 7- 1 2 Analysis This function merits some discussion. First, note the good
user communication
throughout the function. It begins with a complete prompt and provides clear error
messages whenever problems are detected.
We have implemented this logic with a do. . .while
statement, which always loops
at least once, since we know that there will
always be input. This is the standard loop
for validating user input. Within the loop are two
different validation steps. The first
n
l; '
tests for totally invalid input, the second asks the user to verify the input. If either test
indicates a problem, the input is flushed. Note that the user messages are different
j
depending on the circumstances.
The function cannot end unless the valid flag is true. The if statement in the loop
will set it true if the user replies positively. Otherwise, it is set false and the input is
flushed. This code again demonstrates two principles of good human engineering. The N
if statement is coded in a positive manner, and the expected event, good input, is
checked first.
Many other things can go wrong when we are reading a file, but these two exam-
ples cover most of them.
Data Terminology
Computer specialists who deal with data use a set of specific terms to
describe their data. These terms deal with data that are stored in files. What
we call a variable in our program , they call a field or a data item . A field is
the smallest named unit of data in a file . If we were working with data about
the ten western states in the continental United States, we would have data
like Name, Capital , Number, Square Miles , Population , and Number of
Counties. The first two fields arc strings ( delimited arrays of characters that
we study in Chapter 1 1 ) and the last four are integers.
These six fields grouped together make up a state record . A record is a
collection of related data , in this case state data , treated as a unit . Each
record has a key, one or more fields that uniquely identify the record . In our
state s record , the key could be the name. Names normally do not make good
keys , because they are not guaranteed to be unique. A better choice is the
state number, which represents the order in which the states entered the
union . This field is guaranteed to be unique.
With text files we cannot create a record . We must wait for binary files
and structures to do that . But we can simulate a record by grouping these
data on the same line, with each field separated from the next by whitespace.
The data for the ten western states are shown in Table / -6.
State Capital No . .
Sq Miles Population No .
Counties
Arizona Phoenix 48 113,508 3,447,100 15
California Sacramento 31 156,299 33, 871,648 58
Colorado Denver 38 103,595 4,301,251 63
Idaho Boise 43 82,412 1 ,293,953 44
Montana Helena 41 145,388 902, 195 56
Carson City 36 109,894 1 ,998,257 16

Nevada
continual
TABLE 7-6 Ten Western States ( 2000 Census )
Section 7.6 Software Engineering
State Capital No. .

Sq Miles Population No .
Counties
47 121,335 1,819,046 33
New Mexico Santa Fe
33 96,184 3,421,399 36
Oregon Salem
Olympia 42 66,511 5,894,121 39
Washington
Cheyenne 44 96,989 493,782 23
Wyoming
TABLE 7- 6 Ten Western States (2000 Census continued

)( )
more meaning. Although corn -
Data can be logically organized to provide
this type of data using binary files, there is no
puter scientists normally store
file as well . We will return to the discussion
reason it can’t be stored in a text
binary files in Chapter 13.
again w hen we study strings in Chapter 1 1 and
'I '

'
!l

1 . Io print a percent sign , you need to use two tokens ( % % ) .
2. After you have finished working with a file, you should close it .
3. IT you open a file lor writing, a new file will be created . This means that if tj I
you already have an existing file with the same name , it could he deleted .
4 . 11 you want to write at the end of an existing text file, open the file for
appending, not for writing.
5. C output is right justified when using printf and fprintf functions. If you
—
want your data to he left justified , use the left - justify flag ( ) explicitly.
6. It is a compile error to misplace the file pointer in the parameter list . The
file pointer is coded last in all file functions except fscanf and fprintf , in
which it is coded first.
7. An error that may cause your program run to terminate is to use a format
code that does not match the variable type being read .
.
8 An error that will give invalid output is to use a format code that does not
match the variable being printed .
.
9 Several common errors are created when a file operation is included in a
selection statement . Three are shown below.
a . The following code does not properly test to make sure that the file
was opened correctly:
if (sp = fopen() != NULL) // logic error

if ((sp = fopen( )) != NULL) // good code
h. The following code does not store the character read in ch . It stores
the logical result of the compare in ch .
if (ch = getchar( ) != ’ \n' ) // logic error

if ((ch = getchar( )) != '\n' ) // good code
.
e In a similar way, the following code is invalid :
// logic error
while ( ioResult = scanf(...) != EOF )
// good code
while (( ioResult = scanf(...)) != EOF )
as a char-
d . It is a logic error to define ioResult in the above statement
—
acter because EOF ( 1 ) cannot he stored in a character.
10. It is a logic error to code a whitespace character at the end of a scanf

statement .
11 . It is a run - time error to attempt to open a file that is already opened.
12 . It is a run - time error to attempt to read a file opened in the write mode or
to write a file opened in the read mode.
7.8 Key Terms

alternate flag key
append mode leading zero padding
auxiliary storage devices open function
binary file precision
buffer read mode
character input record
output functions right justification
close function secondary storage device
data item signed
field size
field width space ( lag
file standard error
file mode standard file stream
file type text file
filename value error
hexadecimal conversion write mode
7.9 Summary
A file is a collection of related data treated as a unit .
Data in a text file are stored as human- readable characters. The data in a
text file arc usually divided into lines separated by a newline character.
A stream is a sequence of elements in time.
A stream is associated with a file in C.
J I here are three standard file streams in C: standard input , standard out-
put , and standard error.
I he standard input is associated with the keyboard . The standard output
and the standard error are associated with the monitor.
-J I he standard file streams can he accessed respectively using stdin , stclout ,
and stderr. I hese are pointers (addresses) to tables (structures ) containing
the information about the standard streams.
To manipulate and access files, there are different
types of input/output
functions.
nil
J I he open /close functions, /open and /close , are used to open and close the
association between external files and internal streams.
J A (Tie in C can be any of the three basic modes: reading, writing, and
appending.
When a file is opened in reading mode, the file marker is positioned at the
beginning of the existing file . The file is ready to be read .
When a file is opened for writing , the file marker is positioned at the
beginning of a newly created empty file, before the end -of -file character.
The file is ready to be written .
When a file is opened for appending , the marker is positioned at the end of
the existing file ( or at the beginning of a newly created file ) , before the
end - of -file marker. The file is then ready to be written .
Formatted input/output functions allow us to read data from and write
data to files character by character while formatting them to the desired
data types such as char , int , and float . The functions scan / and /scan/ are
used for reading. The functions print/ and / print/ are used for w riting.
Character input/output functions allow' us to read or write files character
by character. The functions fgetcy getc , and getchar can be used for read -
ing. The functions fputcy putc , and pntchar can be used lor writing.
C99 added a set of wide character functions. They are fgetwc , getuc ,
getwchar , fputwc , putuc , and putivchar .
When we are reading data from files , we should validate the data.
7.10 Practice Sets

Review Questions
1 . User files are created using the FILE type.
a. True
b. False
2 . If a file does not exist , the append mode returns an error.
a. True
b. False
3. The conversion specification starts w ith an asterisk ( * and ends
) with the
conversion code.
a . True
b. False
-
4 . The space flag is used to left justify output in
the printf function .
.
a True
b. False
5. One of the problems with testing files is that some errors, such as physi-
cally bad media , cannot be created so that they can be tested .
a . True
.
b False
6. Which of the following is considered auxiliary storage?
a. disk
b. random access memory ( RAM )
.
c read only memory ( ROM )
d . tape
.
e both disk and tape
.
7 Which of the following is not a standard file stream ?
a . stdin
b. stderr
c. stdfile
d . stdont
8. The C library that contains the prototype statements for file operations is
a. ftle.h
h . proto , h
c. stdfile .h
d . stdio .h
e. stdlib . h
9. II a file is opened in the r mode , C
a. opens the file for reading and sets the file marker at the beginning,
h . opens the file for reading and sets the file marker at the end .
c . opens the file lor writing and sets the file marker at the beginning.
d . opens the file lor writing and sets the file marker at the end .
e. returns a null file address indicating an error if the file exists .
10. II a file is not opened successfully, C
a. aborts the program .
b. continues processing with the next statement after the open .
c . changes the file mode to write and opens a new file.
d . displays a message and waits for an answer.
e. returns a file address.
1 1 . The specifies the type of data that are being format -
ted in formatted input/output .
a . conversion code
b. data code
c. field code
d . format code
e. pcode
rn
12. I he two input functions that read text data and convert the data to the
types specified by a format string are
a. frecid and readd . read and readf
h . fscanf and fprintfe. scanf and printf
c. fscanf and scanf
13. Which of the following statements about the formatted input function
( scanf ) is false?
a . scanj implicitly casts data to match the conversion code and its corre -
sponding address parameter.
b. scanf reads data from the keyboard and formats it according to conver-
sion specifications.
c . scanf returns an integer containing the number of variables success-
fully formatted for EOF.
d . scanf stops formatting data when an inappropriate character lor the
type being formatted is encountered .
e. To function properly, scanj requires at least one address.
14. The function reads the next character from the stan -
dard input stream and returns its value.
a. fgetc
b. getc
.
c getchar
.
d readchar
e. ungetc
1 5 . When scanf finds invalid data in the input stream , it
.
a flushes the input to a newline.
.
b prints an error message and terminates the program.
c. prints an error message and terminates the scanf function .
d . skips the data and formats the next conversion specification , if any.
e. terminates the scanf function and leaves the invalid character in the
input stream .
Exercises
16. Given the following declaration :
int il ?
int i2 ;
float fl ;
char cl ;
char c2 ;
char c3;
and the following line oi data
14 23 76 CD
what would be the value ol i l , i 2 , f 1, c l , c 2 , and c 3 after the following

statement ?
scanf( " %d %d % f %c %c %c " ,

&il , &i2, &fl , &cl , &c2 , &c3);
int il ;
int i2 ;
float fl ;
char cl ;
char c2;
and the following line of data
14.2 67 67.9
what would be the value ol i l , i 2 , fl, cl, and c 2 after the following
statement?
scanf( " %d %c %c %i %f ",

&il , &cl, &c2 , &i2 , &f 1 );
18. Given the following declaration:
int il ;
int i2 ;
int i3;
char cl ;
char c2 ;
char c3 ;
and the following line of data :
c l 45 d l 23 34.7
what would be the value of i l , i 2 , i 3, c l , c 2 , and c 3 after the following

statement?
scanf( ” %c%c%d %c%d %d " ,

&cl , &c 2 , &il , &c3 , &i2 , & i
3 );
1
^
' If
19. \ \ hat would be printed Irom the following program segment?
int 11 = 123;
int 12 = -234;
int 13 = -7;
float fl = 23.5; m
float f2 = 12.09 ;
float f 3 = 98.34;
char cl = 65 ;
char c2 = '\n' ;
char c3 = 'E';
printf( " %06d , %06d , %06d %c" ,
il , i2 , i3, c2);
- -
printf("% 6d , % % , \" , \\t, % 06d " , i2, il );
printf( " %c %d ", cl, c2 );
printf( "%c %c % # x", cl + 32, c3 + 3 , c 2 + 5 );
20. W hat output is produced Irom the following program ? Include in your
answer the contents of any files created .
# include <stdio.h >

int main (void )
{
FILE* sp;
int i;
charch ;
sp = fopen( "TEST.DAT", " w" );
for ( i = 2; i < 20; i += 2)
fprintf( sp , " %3d " , i );
fclose(sp );
sp = fopen( "TEST.DAT" , " r" );
while ((ch = fgetc( sp )) != EOF )
if (ch == )
fputc( * • , stdout );
else
fputc( ch , stdout );
return 0;
} // main
Problems
For the problems and projects in this chapter, you need to create a test file
using your text editor. End each line with a newline
character; do not write in
paragraph format. We suggest the Gettysburg Address as a suitable subject
for these files:
Four score and seven years ago our fathers brought forth on this continent
a new nation , conceived in liberty, and dedicated to the proposition that all
continued
men are created equal. Now we are engaged in a great civil war, testing
whether that nation , or any nation so conceived and so dedicated , can long
endure . We are met on a great battlefield of that war. We have come to
dedicate a portion of that field as a final resting-place for those who here
gave their lives that that nation might live. It is altogether fitting and
proper that we should do this. But in a larger sense, we cannot dedicate,
we cannot consecrate, we cannot hallow, this ground . The brave men , liv-
ing and dead , who struggled here have consecrated it far above our poor
power to add or detract . The world will little note, nor long remember what
we say here, but it can never forget what they did here . It is for us the liv-
ing , rather to be dedicated here to the unfinished work which they who
fought here have thus far so nobly advanced . It is rather for us, to he here
dedicated to the great task remaining before us, that from these honored
dead we take increased devotion to that cause for which they gave the last
full measure of devotion ; that we here highly resolve that these dead shall
not have died in vain ; that this nation , under God , shall have a new birth of
freedom , and that government of the people, by the people, for the people,
shall not perish from the earth .
21. Write a function that appends one file at the end ol the other.
.
22 Write a function that appends a file to itself .
23. Write a function that accepts a file of varying- length lines and changes it
to a formatted file with 60 characters in each line.
24. Write a function that calculates the average number of characters per
line in a file.
25 . Write a function that deletes the last line of any file.
26. Write a I unction that deletes the blank lines in a file. A blank line is a line
with only one single character in it : newline .
.
27 Write a program that prints itself .
28. Write a program that copies one text file to another and inserts blank
lines between paragraphs in the new' file. Paragraphs are identified by a
newline character.
29. W rite a program to copy only lines beginning with a user- specified
character.
.
30 Write a program to parse words onto separate lines ; that is , locate and
write each word to its own line . Words are defined as one or more charac-
ters separated by whitespace .
31 . W'hen scanf encounters an error, the invalid data are left in the input
stream , sometimes making it impossible to continue . Write a function
that reads three pieces of numeric data . If an error is detected ( return
value not EOF but less than 3) , flush the erroneous data . ( See "Softw'are
Engineering" for details on how to write a flush statement . ) Then test the
function by entering numeric data and alphabetic data in different

sequences.
32. W rite a program to insert a blank line after the seventh line in a file.
.
33 \ \ rite a program to delete the sixth line in a file. Do not change the sixth
line to a blank line; delete it completely. f
.
34 Write a program to insert a blank line after each line in a file. In other
words, double - space the text.
.
35 Write a program to duplicate the fourth line in a file.
.
36 Write a program to copy a file, deleting the first two characters of each
line. ( Do not replace the characters with blanks ) .
.
37 Write a program to copy a file, inserting two space characters at the
beginning of each line. In other words, each line w ill he shifted two char -
acters to the right.
.
38 W rite a program that copies the 21 st character of each line in a file to a
new file. All extracted characters are to be on the same line. If a line in
the input file has fewer than 21 characters, write the last character. If a
line is blank—that is, if it consists of only whitespace— then copy noth-
ing. At the end ol file, write a newline to the new file and close it .
.
39 W rite a program that writes the odd - numbers between 300 and 500 to a
text file.
40. Write a function that writes the multiples of num between lowLimit and
highLimit file. Assume that lowLimit < highLimit and that
to a text
num is not equal to zero. Then write a test driver to validate your function.
41. W rite a program to read a set of scores from a text file, count the scores
over 90, copy the scores over 90 to a new file, and print the number ol
scores over 90 to the monitor.
42. Modify Problem 41 to count and create a file for the ranges 0 to 30, 31 to
60, 61 to 90, and over 90. Print the statistics for each range at the end of
the program.
Projects
43. Write a program that will read a text file and count the number of alpha-
betic characters (isalpha), digits (isdigit ), punctuation characters
( ispunct ), and whitespace characters ( isspace ) in the file. At the end of
the file, the program is to display an appropriate report. (The classifying
functions are covered in Chapter 5.)
44. Write a text analyzer program that reads any text file. The program prints
a menu that gives the user the options of counting
lines, words, charac -
more words ending in a period ) , or all of the
ters , or sentences ( one or
above. Provide separate
a function for each option . At the end of the
analysis, w rite an appropriate report.
45. Write a menu -driven text utility program. The program will have the fol-
lowing capabilities:
a. Copy a user- named hie to a new file .
b Append a user- named file to another user- named file.
.
c. Change the hie format to be double spaced.
-
d Remove all blank lines (change a double-spaced file to single-spaced ).
.
e. Display the contents of the file as a series of 60 character lines with no
words split between lines.
.
46 Using an editor, create an inventory file using the data shown in Table 7-7
( do not include the column captions, just the data ) .
Part No. Price Quantity On Reorder Minimum

Hand Point Order
0123 1.23 23 20 20
0234 2.34 34 50 25
3456 34.56 56 50 10
4567 45.67 7 10 5
5678 6.78 75 75 25
TABLE 7- 7 Inventory file data for Project 46

Write a program to read the inventory file and create an inventory report.
The report will contain the part number, price, quantity on hand , reorder
point, minimum order, and order amount . The order amount is calcu -
lated when the quantity on hand falls below the reorder point. It is calcu -
lated as the sum of the reorder point and the minimum order less the
quantity on hand . Provide a report heading, such as “ Inventory Report ,”
captions for each column , and an " End of Report " message at the end of
the report . Print the part number with leading zeros.
47 . Using an editor, create the employee file shown in Table 7 - 8.
Employee No. Department Pay Rate Exempt Hours Worked

0101 41 8.11 Y 49
0722 32 7.22 N 40
1273 23 5.43 Y 39
2584 14 6.74 N 45
TABLE 7 - 8 Employee File Data for Project 47

PS
W rite a program to read the employee file and create a payroll register.
The register will contain the following data :
a . Employee number ( print left - justified )
b. Department
c. Pay rate
.
d Exempt IT:
e. I lours worked
.
f Base pay ( pay rate * hours worked )
1
Arrays
In Chapter 4 we discussed the first derived data type , the function. In this
chapter we discuss the second derived type, the array. Figure 8- 1 recaps the
six derived types and the chapters in which they are covered .
Derived
Types
Function
Type
Chapter 4
Array
Type i Pointer
Type
Chapter 9
Structure
Type
Chapter 12
Union
Type
Chapter 12
Enumerated
Type
Chapter 12
i
FIGURE 8 - 1 Derived Types
Objectives
To understand the basic concepts and uses of arrays
To be able to define C arrays
To be able to pass arrays and array elements to functions
To understand the classical approaches to sorting arrays: selection, bubble,
and insertion sorting
To write programs that sort data using the three classical algorithms
To be able to analyze the efficiency of a sort algorithm
To understand the two classical search algorithms: sequential and binary
To write programs that search arrays
To be able to design test cases for sorting and searching algorithms
To be able to analyze the efficiency of searching algorithms
it
459
460 Section 8.1 Concepts
8.1 Concepts
Imagine we have a problem that requires us to read, process, and print
10 integers. We must also keep the integers in memory for the duration of the
program . To begin , we can declare and define 10 variables, each with a differ-
ent name, as shown in Figure 8- 2 .
scoreO score5
score1 score6
score2 score7
score3 score8
score 4 score9
FIGURE 8 - 2 Ten Variables
Having 10 different names, however, creates a problem : I low can we read

10 integers from the keyboard and store them ? To read 10 integers from the
keyboard , we need 10 read statements, each to a different variable. Further-
more , once we have them in memory, how can we print them ? To print them ,
we need 10 write statements. The flowchart in Figure 8- 3 shows a design for
reading, processing, and printing these 10 integers.
Although this approach may he acceptable for 10 variables, it is definitely
not acceptable lor 100 or 1 ,000 or 10,000. To process large amounts of data
we need a powerful data structure, the array. An array is a collection ol ele-
ments ol the same data type.
Since an array is a sequenced collection , we can refer to the elements in
the array as the first element , the second element , and so forth until we get to
the last element . Thus, when we put the 10 integers of our .problem into an
array, the address of the first element is 0 as shown below.
scoreso
In a similar fashion , we refer to the second score as scores and the third
\
-
score as scores2 Continuing the series , the last score would be scores We
can geneialize this concept in the following fashion where the subscripts^
indicate the ordinal number of the element counting from the beginning of
the array:
scores0 scoresj,
/ scores n -1
nr
/
(
/ ReadX
/
START
scoresO
^
X
X
/ scoresO
/
Print X
X
X
I
//
Read Print
/ scores 1 X scores 1 X
/ Read X / Print X
/ s c o r e s2 X / scores2 X
/ Read X / Print X
/ scores3 X / scores3 X
/ Read X Process / Print X
/ scores 4 X 10 scores
/ scores 4 X
/ Read X / Print //
/ scores5 g / scores5 X
~
/ i /
g/
Read Print
/ scores6 X ! scores6
~
/ t
/
/
/
Read
scores7
Read
^
X
/
/
/
Print
scores7
Print
X
~
t
/ scoresS X / scores8 X
/
/
/ Read
scores9
^
X /
(
Print
scores9
STOP
/
X
^
FIGURE 8 - 3 Process 10 variables
What we have seen is that the elements of the array are individually
addressed through their subscripts, a concept shown graphically in Figure 8 - 4.
The array as a whole has a name, scores, hut each member can he accessed
individually using its subscript.
The advantages of the array would he limited if we didn’t also have pro-
gramming constructs that would allow us to process the data more conve-
niently. Fortunately, there is a powerful set of programming constructs—
loops— that makes array processing easy.
We can use loops to read and write the elements in an array; to add, sub -
tract , multiply, and divide the elements; and even for more complex process
-
ing such as calculating averages, searching, or sorting. Now it does not
to he processed,
matter if there are I , 10, 100, 1,000, or 10,000 elements
because loops make it easy to handle them all.
One question remains: How can we write an instruction so that one
time
to another
it refers to the first element of an array and the next time it refers
element? The answer is really quite simple : We simply borrow from the sub -
, howcvcr , we
script concept we have been using. Rather than using
subscripts
will place the subscript value in brackets. This format is known as indexing .
Using the index notation, we would refer to scoresg as
scores [ 0 ]
462 Section 8.1 Concepts
scores0 scores[0]
scores 1 scores[1 ]
scores2 scores[2]
scores3 scores[ 3 ]
scores4 scores[ 4 ]
scores 5 scores[ 5]
scores6 scores[ 6]
scores7 scores[ 7 ]
scores8 scores[8]
scores 9 scores[9]
scores scores
(a) Subscript Format (b) Index Format
FIGURE 8 - 4 An Array of Scores
Following the convention, scores0 becomes scoresfO] anti scores,

becomes scores[ 9 ]. Using a typical reference, we now refer to our array using
the variable i .
scores [ i ]
I he flowchart showing the loop used to process our 10 scores using an

array is seen in Figure 8- 5.
i++
Start
i = 0
o
Process /
©
i = 0
i < 10 10 scores
r i < 10
/
/ Read
~
scores [ i ] m
k
o / scores
/
Write
/
[ i ]M
o Stop
FIGURE 8 - 5 Loop for 10 Scores
ui
n
8.2 Using Arrays in C

In this section, we first show how to declare and define arrays. Then we
present several typical applications using arrays including reading values into
arrays, accessing and exchanging elements in arrays, and printing arrays. Fig-
ure 8 - 6 shows a typical array, named scores , and its values.
Indexes
scores [ 0]
scores [1]
scores [2]
scores [3]
( Elements scores [4]

scores [5] 45
scores [6] 56
scores [ 7] 34
scores [8 ] 83
rTJame of
scores [9] 16 \ t h e array
scores
FIGURE 8 - 6 The Scores Array
C provides for two different array types, fixed-length array and variable -
length array. In a fixed -length array, the size of the array is known when the
program is written. In a variable -length array, inroduced in C 99, the size ol
the array is not known until the program is run.
Declaration and Definition

An array must be declared and defined before it can he used. Array
declara-
tell the compiler the name of the array, the type of each
lion and definition
element, and the size or number of elements in the array . In a fixed-length
have value compilation
array, the size of the array is a constant and must a at
time. The declaration format is shown in the lollowing

example.
type arrayName [ arraySize ]
array declarations: one for

Figure 8- 7 shows three different fixed-length
, and one for floating-point numbers.
integers, one for characters
r
464 Section 8.2 Using Arrays in C
int scores [ 10 ] ;
[0] [1 ] [ 2] [3] [4] [5] [6] [7] [8] [ 9]
type of each scores
. element .
char name [ 10 ] ;
[0] [1 ] [2 ] [3] [4] [5] [6] [ 7] [8] [9 ]
name of name
the array
float gpa [ 4 0 ] ; •••

0] [ 1 ] [ 2 ] [37][38][39]
number of gpa
elements
FIGURE 8 -7 Declaring and Defining Arrays
The declaration format lor a variable- length array is the same as for a
fixed-length array except that the array size is a variable. When the program is
executed, the array size is determined and the array is defined. Once defined,
its size cannot be changed. Following standard C syntax rules, the array size
must he declared and initialized before it is used in the variable - length array
definition.
float salessAry [arySize];
Accessing Elements in Arrays

C uses an index to access individual elements in an array. The index must be
an integral value or an expression that evaluates to an integral value The .
simplest form for accessing an element is a numeric constant . For example,
given the scores array in Figure 8 - 6, we could access the first element as
shown in the next example:
scores[ 0 ]
I he index is an expression, typically the value of a variable. To process all

the elements in scores, a loop similar to the following code is used:
for ( i = 0 ; i < 1 0 ; i ++ )
process ( scores[i]);
V>u might he wondering how C knows

where an individual element is
located in memory. In scores, for example, there are nine elements. How
does it find just one? The answer is simple. The
array’s name is a symbolic
reference lor the address to the first byte of the array Whenever we use the
7
T
array’s name, therefore, we are actually referring to the first byte of the array.
1 he index represents an offset from the beginning of the array to the element
.
being referenced With these two pieces of data, C can calculate the address
ol any element in the array using the following simple formula:
element address = array address

+ ( sizeof ( element) * index )
1 or example, assume that scores is stored in memory at location

10,000. II scores is an integer of type int, the size of one element is the size
ol int . Assuming the size of an int is 4, the address of the element at index 3 is
element address = 10,000 + 4 * 3 = 10 ,012
Storing Values in Arrays

Declaration and definition only reserve space for the elements in the array.
No values are stored. II we want to store values in the array, we must either
initialize the elements, read values from the keyboard, or assign values to
each individual element .
Initialization
Just as with variables, initialization of the elements in a fixed-length array can
he done when it is defined. Variable-length arrays cannot be initialized when
they are defined.
For each element in the array, we provide a value. The only difference is
that the values must be enclosed in braces and, if there is more than one,
separated by commas. It is a compile error to specify more values than there
are elements in the array.
Figure 8 - 8 contains four examples of array initialization. The first exam -
ple (Figure 8- 8( a)) is a simple array declaration of five integers and is typical
initial -
of the way array initialization is coded. When the array is completely
ized, the programmer does not need to specify the size of the array . This case
is seen in Figure 8 - 8 ( b ) . It is a good idea, however, to define

the size explicitly ,
checking for errors and is also good
because it allows the compiler to do some
documentation.
If the number of values provided is fewer than the number
of elements in
the array, the unassigned elements are filled with zeros . I bis case is seen in
initialize an array to all
Figure 8 - 8 c . We
( ) can use this rule to easily
zeros by supplying just the first zero value
, as shown in the last example in
Figure 8 - 8 ( d ).
when they are defined. Variable -

-
Only fixed length arrays can be initialized
or assigning the values.
length arrays must be initialized by inputting
(a) Basic Initialization (b) Initialization without Size

int numbers[5] = (3,7,12,24,45); int numbers[ ] = (3,7,12,24,45);
/ n 111
12124[ 45
(c) Partial Initialization
3 | 7 12 24 45 j 3 7
(d) Initialization to All Zeros

^
int numbers[5] = (3,7);
3 1 7|0|0|
0
int lotsOfNumbers [1000]
r-r
0 0
—.
= {0};
The rest are

filled with Os All filled with Os
FIGURE 8 - 8 Initializing Arrays
Inputting Values
Another way to fill the array is to read the values from the keyboard or a file.
I his method of inputting values can be done using a loop . When the array is
to be completely filled , the most appropriate loop is the for loop because the
number of elements is fixed and known .
for ( i = 0; i < 10; i++)

scanf ("%d " , &scores [i ]);
Several concepts need to be studied in this simple statement. First , we

start the index, i, at 0. Since the array has 9 elements , we must load the val-
ues Irom index locations 0 through 8 . 1 he limit test , therefore, is set at i < 9
which conveniently is the number of elements in the array. Then, even
though we are dealing with array elements , the address operator ( & ) is still
necessary in the scanf call.
Finally, il all the elements might not be filled, then we use one of the
event- controlled loops ( while or do... while ) . Which loop we use depends on
the application.
Assigning Values
We can assign values to individual elements using the assignment operator.
\ n> alue that reduces to the proper type can he assigned to an individual
'
array element. A simple assignment statement for scores is seen below.
scores [ 4 ] = 23;
1
On the other hand, we cannot assign one array to another array, even if
they match fully in type and size. We have to copy arrays at the individual ele-
ment level . For example, to copy an array of 25 integers to a second array of
25 integers, we could use a loop, as shown below.
for ( i = 0; i < 25; i++ )

second[ i ] = first[ i ];
II the values ol an array follow a pattern, we can use a loop to assign val -
ues. For example, the following loop assigns a value that is twice the index
number to array scores:
for ( i = 0; i < 10; i++ )

scores [i] = i * 2 ;
For another example, the following code assigns the odd numbers 1
through I 7 to the elements of an array named value:
for ( i = 0; i < 9 ; i++)

value [i ] = ( i * 2 ) + 1 ;
One array cannot be copied to another using assignment .
Exchanging Values
A common application is to exchange the contents of two elements. We see
this operation later in the chapter when we talk about sorting arrays. When
we exchange variables, we swap the values of elements without knowing
what ’s in them.
For example, imagine we want to swap numbers[ 3] and numbers [ 1 ] in
Figure 8 -8. A common beginner’s mistake would be simply to assign
each ele-
ment to the other, as shown below.
numbers [ 1 ]; // Error
numbers [ 3 ] =
numbers [ 1 ] = numbers [ 3 ];
code care-
Although this code looks as if it will do the job, if we trace the
. is moved to numbers [ 3 ],
fully, we find that it does only half the job numbers !1 ]
both elements have the same
but the second half isn’t done. The result is that
value. Figure 8- 9 traces the steps.
[0] [1] [2] [3] [4]

numbersf ] 3 12 24 451
Before
[01 [11 [21

1
[31 [4]
i
numbers [3] = numbers[1]; 3 7 12 7 45

I
'
1
10] [1] [2] [3] [4]

numbers [1 ] = numbers [3] ; 3 7 12 7 45 |
After
FIGURE 8 - 9 Exchanging Scores — the Wrong Way
Figure 8-9 shows that the original value of numbers!3 ] is destroyed

before we can move it . The solution is to use a temporary variable to store the
value in numbers[ 3 ] before moving the data from numbers[ 1 ].
temp = numbers[3 ];
numbers [3 ] = numbers [ 1 ];
numbers [ 1 ] = temp ;
This technique is shown in Figure 8- 10.
IQ) IQ 12) [3] 14 ]

numbers! ] ; [3 7 12 24 45 |
Before
[0] [1] [2l [3] [4]

temp = numbers [ 3] ; 3 7 12 24 I 45 I 24
temp
[01 [11 [ 2] [3] [4]

numbers !3 ] = numbers {1] ; 3 7 12 7 | 45 | 24
temp
[0]
f
11 ] !2] !31 [ 4] l
numbers !1 ] = temp ; I 12 I I 45
3 24
After
7
1 temp
24
FIGURE 8 - 10 Exchanging Scores with Temporary Variable

\ : 'T i
Printing Values
Another common application is printing the contents of an array. This is eas -
ily done with a for loop, as shown below.
for (i = 0; i < 9; i++)

printf ( " %d " , scores [i ]);
printf ("\n");
In this example, all the data are printed on one line. After the for loop
completes, a final printf statement advances to the next line. But what if we
had 100 values to be printed? In that case, we couldn’t put them all on one
line. Given a relatively small number width, however, we could put 10 on a
line. We would then need 10 lines to print all the data. This rather common
situation is easily handled by adding a counter to track the number of ele -
ments we have printed on one line. This logic is shown in Program 8 - 1 .
PROGRAM 8 - 1 Print Ten Numbers per Line

1 // a program fragment
2 # define MAX SIZE 25
3
5 int list [ MAX SIZE];
6
7 // Statements
8
9 numPrinted = 0;
10 for ( int i = 0; _
i < MAX SIZE; i++ )
11 {
12 printf( " % 3d " , list[i]);
13 if ( numPrinted < 9 )
14 numPrinted ++ ;
15 else
16 {
17 printf( " W );
18 numPrinted = 0;
19 > // else
20 > // for
Precedence of Array References

References to elements in arrays are postfix expressions. By looking
at the
of 16 , which is
Precedence Table, we see that array references have a priority
very high. What is not apparent from the table, however , is that the opening
and closing brackets are actually operators. They create postfix
a expression
from a primary expression . With a little thought , you should recognize that
4 / 0 Section 8.2 Using Arrays in C
this is exactly as it must be: When we index an array element in an expres-

sion , the value must he determined immediately.
Referring to the original array in Figure 8 - 10, what will be the result of
the following code?
numbers[3] = numbers[4] + 15;
In this case, numbers [ 4 ] has a higher precedence ( 16) than the addition
operator ( 12 ) , so it is evaluated first . The result is then
numbers[3] = 45 + 15;
After this statement has been executed , numbers[3] has been changed from
24 to 60.
Index Range Checking

The C language does not check the boundary of an array. It is our job as pro-
grammers to ensure that all references to indexed elements are valid and
within the range of the array. If we use an invalid index in an expression, we
get unpredictable results. (The results are unpredictable because we have no
way of knowing the value of the indexed reference. )
When we use an invalid index in an assignment , we destroy some unde-
termined portion of the program . Usually, but not always, the program con -
tinues to run and either produces unpredictable results or eventually aborts.
An example of a common out - of - range index is seen in the code to fill an
array from the keyboard . For this example, we have reproduced the code we
used to fill the scores array earlier, only this time we have made a common
mistake! Can you spot it ?
for (i = 1; i <= 9; i++) // Error

scanf ( " %d " , &scores [i ]);
\ \ hen dealing with array processing, be very careful at the beginning and
end of the array. A careful examination of the previous code discloses that we
erroneously started at 1 instead of 0. So we fix it as shown below, only to find
that it still doesn ’t work!
for (i = 0; i <= 9 ; i++ )

scanf ( " %d " , &scores [ i]);
The moral of this example is to examine our logic. If we made one mis -
take , we may well have made two. Although we corrected the error lor initial -
ization ( the beginning of the array ) , there is still an error at the other end . If
you can ’t see it , check the original code in “ Inputting Values. ”
r -
.
1
' V'l

'
I he result ol both versions of this error is that the data stored in memory
.
alter the scores array are erroneously destroyed In the first version of the
error, the first element of the array was not initialized.
The problems created by unmanaged indexes are among the most diffi-
cult to solve, even with todays powerful programming workbenches. So we
want to plan our array logic carefully and fully test it .
EXAMPLE 8 - 1 Print Array

L e t s write a program that uses arrays. Program 8 - 2 uses a for loop to initial -
ize each element in an array to the square of the index value and then prints
the arrav. j
PROGRAM 8 - 2 Squares Array

1 /* Initialize array with square of index and print it.
2 Written by:
3 Date:
4 */
5 tinclude <stdio.h >
6 # define ARY SIZE 5
7
8 int main ( void )
9 {
11 _
int sqrAry[ ARY SIZE ];
12
13
14
// Statements
_
for ( int i = 0; i < ARY SIZE; i++)
15 sqrAry[ i] = i * i ;
16
17 printf( "Element\tSquare\n " );
18 printf( "=======\t======\n" );
19 for ( int i = 0; i < ARY SIZE ; i++)
20 printf("%5d\t%4d\n", i, sqrAry[i]);
21 return 0;
22 > // main
Results:
Element Square
0 0
1 1
2 4
3 9
4 16
1
Example: Read and Print Reversed

As another example of an array program , let s read a series of numbers from
’
the keyboard and print them in reverse order ; that is , il we read 12 3 4 , we
want to print them 4 3 2 1 . With a little thought , it should be obvious that we

must read all of the numbers before we can begin printing them. This deli-
nitely sounds like a problem lor an array.
In Program 8 - 3, we read up to 50 integers and then print them reversed,
10 to a line.
PROGRAM 8 - 3 Print Input Reversed

1 /* Read a number series and print it reversed .
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
10 int readNum ;
11 int numbers[50 ];
12
13 // Statements
14 printf( " You may enter up to 50 integers:\n" );
15 printf(" How many would you like to enter? " );
16 scanf ( " %d", &readNum );
17
18 if (readNum > 50 )
19 readNum = 50;
20
21 / / Fill the array
22 printf("\nEnter your numbers: \n " );
23 for ( int i = 0; i < readNum; i++)
24 scanf("%d " , & numbers[i]);
25
26 / / Print the array
27 printf("\nYour numbers reversed are: \n");
28 for ( int i = readNum - 1, numPrinted = 0;
29 i > = 0;
30 i~ )
31 {
32 printf(" %3d ", numbers[i]);
33 if (numPrinted < 9)
continued
TTTI
PROGRAM 8 - 3 Print Input Reversed ( continued )

34 numPrinted ++;
35 else
36 {
37 printf( "\n" ); V
38 numPrinted = 0;
39 > // else
40 > // for
41 return 0 ;
42 > // main
Results:
You may enter up to 50 integers:
How many would you like to enter? 12
Enter your numbers:

1 2 3 4 5 6 7 8 9 10 11 12
Your numbers reversed are:

12 11 10 9 8 7 6 5 4 3
2 1
Program 8 - 3 Analysis First, note how we validate the number of integers to be read. If the user requests
more than 50, we simply set readNum to 50 . This ensures that we will not run off the
end of the array.
This is an interesting loop because we start at the end and work to the front. While
many arrays are processed from the beginning, it is often necessary to process from the
end. You will see more in the sorting functions later in the chapter.
Finally, note that we incorporate the logic from Program 8 - 1 to print ten numbers
per line. Had we made Program 8 - 1 a function, we could have reused the code rather
than repeating it.
8.3 Inter- function Communication

To process arrays in a large program, we have to be able to pass them to func
-
ways: pass individual elements or pass the

tions . We can pass arrays in two
whole array. In this section we discuss first how to pass individual elements
and then how to pass the whole array.
Passing Individual Elements

by either passing
As we saw in Chapter 4 , we can pass individual elements
their data values or by passing their addresses.
474 Section 8.3 Inter-function Communication
Passing Data Values

We pass data values , that is individual array elements, just like we pass any
data value .
As long as the array element type matches the function parameter type, it
can he passed . The called function cannot tell whether the value it receives
comes from an array, a variable, or an expression . Figure 8- 1 1 demonstrates
how an array value and a variable value can he passed to the same function .
int a; int ary( 10 ];
fun (a); fun ( ary[ 3 ]);
void fun ( int x )

{
process x
} // fun
FIGURE 8 - 11 Passing Array Elements
Note how only one element is passed by using the indexed expression ,
ary [ 3 ] .
Since the value of this expression is a single integer, it matches the
formal parameter type in fun . As far as fun is concerned , it doesn 't know or
care that the value it is working with came from an array.
Passing Addresses
In C hapter 4 , we saw that we can use two - way communication by passing an
address. We can pass the address ol an individual element in an array just like
we can pass the address of any variable.
To pass an array element’s address , we prefix the address operator to the
element’s indexed reference . Thus, to pass the address of ary [ 3 ] we use the
code shown in the following example.
& ary [ 3 ]
fussing an address of an array element requires two changes in the called

function . First , it must declare that it is receiving an address. Second , it must
use the indirection operator ( * ) to refer to the elements value . These con -
cepts are shown in Figure 8- 12. Note that when we pass an address, it is two-
way communication .
~
'1
Chopter 8 Arroys 475
int a ; int ary[ 10];

fun ( &a ); fun (&ary[3]);
* t
void fun (int* x)

{
process x
} // fun
FIGURE 8 - 1 2 Passing the Address of an Array Element
Passing the Whole Array

Here we see the first situation in which C does not pass values to a function .
Hie reason for this change is that a lot of memory' and time would he used in
passing large arrays every time we wanted to use one in a function. For exam -
ple , if an array containing 20,000 elements were passed by value to a func -
tion , another 20,000 elements would have to be allocated in the called
function and each element would have to be copied from one array to the
other. So, instead of passing the whole array, C passes the address of the array.
In C , the name of an array is a primary expression whose value is the
address of the first element in the array. Since indexed references are simply
calculated addresses, all we need to refer to any of the elements in the array
is the address of the array. Because the name of the array is in fact its
address, passing the array name, as opposed to a single element , allows the
called function to refer to the array back in the calling function. The design
for passing the whole array is shown in Figure 8 13.-
int ary[ 10 ]; int ary[size];
fun ( ary );
void fun ( int fAry[ ])

{
process x
{
fun ( ary );
process x
.
void fun ( int fAry[ * ])
} // fun
> // fun
Fixed- size Array Variable -size Array
FIGURE 8 - 1 3 Passing the Whole Array

Fixed-length Arrays
To pass the whole array, we simply use the array name as the actual parame-
ter. In the called function , we declare that the corresponding formal parame-
ter is an cirra ) We do not need to specify the number of elements in a fixed-
length array. Since the array is actually defined elsewhere, all that is impor-
tant is that the compiler knows it s an array.
The function declaration for an array can declare the array in two ways.
First , it can use an array declaration with an empty set of brackets in the
parameter list. Second , it can specify that the array parameter is a pointer. We
prefer the array declaration over the pointer declaration because it is clearer.
Roth are shown in the following example, which declares arrays of integers .
void fun (int ary[], ...)/*
void fun (int* ary , ...);
i
Variable length Arrays

-
When the called function receives a variable- length array, we must declare
and define it as variable length . In a function declaration , we declare an array
as variable using an asterisk for the array size or by using a variable for the
array size . In the function definition , we must use the variable name, and fol -
lowing standard syntax rules, it must be defined before the array. The follow-
ing example demonstrates these points.
float calcArrayAvg (int size, float avgArray [ * ]);
// Function Definition
float calcArrayAvg ( int size, float avgArray [size])
{
I
} // calcArrayAvg
Note , however, that the array passed to a function declaring a variable-

-
length array can he either a fixed length or variable - length array. As long as
the array types are correct and the correct size is specified , the passed array
can be processed .
In summary, we must observe the following rules to pass the whole array
to a function:
1 . I he function must be called passing only the name of the array.

2 . In the function definition , the formal parameter must be an array type,
either fixed length or variable length.
T I he size ol a fixed -length array does not need to be specified.
4 . I he size ol a variable-length array in the function
prototype must be an
asterisk or a variable.
S. I he size of a variable-length array in the function definition must be a

variable that is in the function ’s scope, as a previously specified variable
parameter .
Passing an Array as a Constant
When a function receives an array and doesn’t change it , the array should be
received as a constant array. This prevents the function from accidentally
changing the array. To declare an array as a constant , we prefix its type with
the type qualifier const , as shown in the following function definition . Once
an array is declared constant , the complier will flag any statement that tries
to change it as an error.
double average ( const int ary[ ], int size );
Function Communication Examples

EXAMPLE 8 - 2 Average Array Elements
We can use a function to calculate the average of the integers in an array. In
this case, we pass the name of the array to the function and it returns the
average as a real number. This concept is shown in Program 8 - 4.
PROGRAM 8 - 4 Calculate Array Average

1 /* Calculate the average of the number in an array.
2 Written by:
3 Date:
4 */
6
8 double average ( int ary[ ]);
9
11 {
13 double ave ;
14 int base[5] = {3 , 1 , 2 , 4 , 5 > ;
15
16 // Statement
17 ave = average( base );
18 printf( " Average is: % lf \n " / ave);
19 return 0;
20 > // main
21
22 === average ===
continued
PROGRAM 8 - 4 Calculate Array Average ( continued )

23 Calculate and return average of values in array.
24 Pre Array contains values
25 Post Average calculated and returned
26 * /
27 double average ( int ary[ ])
28 {
30 int sum = 0;
31
32 // Statement
33 for (int i = 0; i < 5 ; i++)
34 sum += ary[i ];
35
36 return ( sum / 5.0);
37 } // average
EXAMPLE 8- 3 Average Elements in Variable-length Array

In this example, we average the elements in a variable- length array. However,
because a variable - length array cannot be initialized , we read the data from
the keyboard . Also, note that to keep the program as simple as possible , we
create the array in a block. This is necessary because the size of the array
must he known before it is declared .
PROGRAM 8 - 5 Average Elements in Variable - length Array

1 /* Calculate average of numbers in variable-length
2 array.
3 Written by:
4 Date:
5 */
6 # include <stdio. h >
7
9 double average (int size, int ary[*]);
10
1 1 int main ( void )
12 {
14 int size;
15 double ave;
16
17 // Statements
18 printf ( " How many numbers do you want to average? " )?
continued
H1
PROGRAM 8 - 5 Average Elements in Variable-length Array (continued )

19 scanf ( "%d " , &size );
20
21 -
// Create and fill variable length array
22 {
23 // Local Declaration
24 int ary[size];
25
26 // Fill array
27 for ( int i = 0; i < size; i++)
28 {
29 printf( " Enter number %2d: " , i + 1 );
30 scanf ( "%d " , &ary[i]);
31 > // for
32 ave = average(size, ary );
33 > // Fill array block
34
35 printf("Average is: %lf" , ave);
36 return 0 ;
37 } // main
38
39 /* ====
40 Calculate and return average of values in array.
41 Pre Array contains values
42 Post Average calculated and returned
43 */
44 double average ( int size , int ary[size ])
45 {
47 int sum = 0 ;
48 double ave;
49
50 // Statement
51 for ( int i = 0; i < size; i++ )
52 sum += ary[ i];
53
54 ave = (double)sum / size;
55 return ave;
56 > // average
Results
How many numbers do you want to average? 5
Enter number 1: 3
Enter number 2: 6
Enter number 3: 9
continued
PROGRAM 8 - 5 Average Elements in Variable-length Array ( continued )

Enter number 4: 2
Enter number 5: 8
Average is: 5.600000
Program 8 - 5 Analysis There are two points to study in this program . First, in main , we use a variable- length
array. This requires that the array size be known before the array is declared. We
accomplish this by using a block within main. The array size is declared in main' s
local declarations; the array itself is declared in the block local declarations.
Second, note how the variable array is declared in the function definition. We can
do it by just declaring it as a variable array with an asterisk in the array brackets, or
we can put the variable size in the brackets. Using size is better documentation
because it declares our intent. When we use size, however, it must be declared
before the array, in this case as the first parameter.
EXAMPLE 8 - 4 Changing Values in an Array

The previous examples were one-way communication. In Program 8 -6 we
demonstrate two-way communication by changing the values in the array. As
you study it, note that no special code is required when we want to change
the values in an array. Because the array s address is passed, we can simply
use index notation to change it .
PROGRAM 8 - 6 Change Values in an Array

1 /* Multiply each element in an array by 2.
2 Written by:
3 Date:
4 */
f 7 void multiply 2 (int x[ ]);
8
9 int main ( void )
10 {
12 int base[ 5 ] = {3, 7, 2 , 4 , 5 >;
13
14 // Statements
15 multiply 2 ( base );
16
17 printf( " Array now contains: ");
18 for (int i = 0; i < 5; i++)
19 printf( " % 3d " , base[ i ]);
20 printf("\n");
21 return 0;
continue
n1
PROGRAM 8 -6 Change Values in an Array
2 2 } // main
23
24 /* ==================== multiply 2 =====
25 Multiply each element in array by 2.
26 Pre array contains data
27 Post each element doubled
28 * /
29 void multiply2 (int ary[ ])
30 {
31 // Statements
32 for ( int i = 0 ; i < 5; i++ )
33 ary[i] * = 2 ;
34 return ;
35 > // multiply2
8.4 Array Applications

In this section we study two array applications: frequency arrays with their
graphical representations and random number permutations.
Frequency Arrays
Two common statistical applications that use arrays are frequency distribu -
tions and histograms . A frequency array shows the number of elements with
an identical value found in a series of numbers. For example , suppose we
have taken a sample of 100 values between 0 and 19 . We want to know how
many of the values are 0 , how many are 1 , how many are 2 , and so forth up
through 19.
We can read these numbers into an array called numbers. Then we create
an array of 20 elements that will show the frequency of each
number in the
series. This design is shown in figure 8 - 14 .
do we write
With the data structure shown in Figure 8 - 14 in mind , how
there exactly 100 elements , we can
the application ? Since we know that arc
examine each value in the array . But how can we relate the
use a for loop to
value in numbers to a location in the frequency " .
to an index and
One way to do it is to assign the value from the data array
then use the index to access the frequency array. This technique is
shown below'.
f = numbers[i];
frequency [ f ]++ ;
, we can simply use the value

Since an index is an expression, however
into the frequency array as shown in the next
from our data array to index us
482 Section 8.4 Army Applications
exam pie. The value of numbers [ i ] is determined first , and then that value is
used to index into frequency.
frequency [numbers [i]]++;
numbers [0] frequency [0]
numbers [1 ] frequency [ 1 ]
This value shows
how many Os are
numbers [2]
numbers [3]
frequency [ 2] in numbers
-
numbers [ 4]
frequency[ 18]
frequency[19 ]
requency
This value shows
numbers[98] how many 19s are
v. in numbers
numbers[99] ^
numbers
FIGURE 8 - 14 Frequency Array
The complete function is shown in Program 8-7 as makeFrequency. The

function first initializes the frequency array and then scans the data array to
count the number of occurrences of each value.
Histograms
A histogram is a pictorial representation of a frequency array. Instead of
printing the values of the elements to show the frequency of each number, we
print a histogram in the form of a bar chart . For example, Figure 8- 15 is a his *
togram lor a set of numbers in the range 0... 19. In this example, asterisks ( *
are used to build the bar. Each asterisk represents one occurrence of the '
data value .
four 1s
0 0
1 4 •* » •
2 7 seven 3s
3 7
18 2 •• zero 19 s
19 0
FIGURE 8 - 1 5 Frequency Histogram
L
'• n
Let’s write a program that builds a frequency array for data values in the
range 0 .. . 19 and then prints their histogram . The data are read from a file. To
provide flexibility, the getData function may only partially fill the array. The
function that loads it also guards against too much data. The design for the
program is shown in Figure 8- 16.
Frequency
Histogram
make make
getData printData Frequency Histogram
FIGURE 8 - 16 Histogram Program Design
The code is shown in Program 8-7.
PROGRAM 8 - 7 Frequency and Histogram

1 /* Read data from a file into an array.
2 Build frequency array & print data in histogram.
3 Written by:
4 Data:
5 */
8
9 _
# define MAX ELMNTS 100
10 #define ANLYS RNG 20
11
13 int getData (int numbers[], int size , int range );
14
15 void printData ( int numbers[ ],
int size, int lineSize);
16
void makeFrequency ( int numbers[ ], int size,
17
int frequency [ ], int range );
18
range);
19 void makeHistogram ( int frequencyf ], int
20
22 {
24 int size;
continued
484 Section 8.4 Array Applications
PROGRAM 8 - 7 Frequency and Histogram ( continued )

25 int nums _
[ MAX ELMNTS ];
26 _
int frequency [ ANLYS RNG ];
27
28 // Statements
29 size = getData ( nums, MAX_ELMNTS, ANLYS_RNG );
30 printData ( nums , size, 10);
31
32 _
makeFrequency( nums , size, frequency , ANLYS RNG );
33 _
makeHistogram( frequency , ANLYS RNG);
34 return 0 ;
35 > // main
36
37
38 Read data from file into array. The array
39 does not have to be completely filled ,
40 Pre data is an empty array
41 size is maximum elements in array
42 range is highest value allowed
43 Post array is filled. Return number of elements
44 */
45 int getData ( int data [ ], int size , int range )
46 {
48 int dataln ;
49 int loader = 0;
50 FILE* fpData ;
51
52
53
54 // Statements
55
56
-
if (!(fpData = fopen ( "P08 07.dat " , " r" )))
printf( " Error opening file\a\a\n" ) , exit ( 100);
57
58 while ( loader < size
59 && fscanf( fpData , " %d " , &dataln ) != EOF)
60 if (dataln >= 0 && dataln < range )
61 data[ loader++] = dataln ;
62 else
63 printf( " \nData point %d invalid. Ignored. \n" f
64 dataln );
65
66 // Test to see what stopped while
67 if ( loader = = size)
68 printf( " \nToo much data. Process what read. Xn " );
continue
It
I 1
PROGRAM 8- 7 Frequency and Histogram (continued )

69 return loader ;
70 } // getData
71
72 /* === printData =====
73 Prints the data as a two dimensional array.
-
74 Pre data: a filled array
75 size: number of elements in array
76 lineSize: number of elements printed / line
77 Post the data have been printed
78 */
79 void printData ( int data[], int size , int lineSize)
80 {
82 int numPrinted = 0;
83
84 // Statements
85 printf( " \n\n " );
86 for ( int i = 0; i < size; i++)
87 {
88 numPrinted ++;
89 printf( "%2d " , data[ i ]);
90 if ( numPrinted >= lineSize)
91 {
92 printf( "\n " );
93 numPrinted = 0 ;
94 } // if
95 } // for
96 printf( "\n\n " );
97 return ;
98 } // printData
99
100 /* ================= makeFrequency ====
101 analyze the data in nums and build their frequency
102 distribution
103 Pre nums: array of validated data to be analyzed
104 last: number of elements in array
105 frequency: array for accumulation ,
106 range: maximum index/value for frequency
107 Post Frequency array has been built.
108 */
void makeFrequency ( int nums[ ], int last ,
109
110 int frequency[ ], int range )
111 {
112 // Statements
continued
486 Section 8.4 Array Applications
PROGRAM 8 - 7 Frequency and Histogram ( continued )

113 // First initialize the frequency array
114 for (int f = 0; f < range ; f++)
115 frequency [ f ] = 0 ;
116
117 // Scan numbers and build frequency array
118 for (int i = 0; i < last; i++ )
119 frequency [ nums [ i ]]++ ;
120 return;
121 > // makeFrequency
122
123 === makeHistogram ==================
124 Print a histogram representing analyzed data.
125 Pre freq contains times each value occurred
126 size represents elements in frequency array
127 Post histogram has been printed
128 */
129 void makeHistogram ( int freq[ ], int range )
130 {
131 // Statements
132 for ( int i = 0; i < range; i++)
133 {
134 printf ( " %2d %2d " , i, freq[ i ]);
135 for (int j = 1 ; j <= freq[i]; j++)
136 printf ( M * H );
137 printf ( "\n" );
138 > // for i...
139 return;
140 } // makeHistogram
141 // === === End of Program =====
Results:
Data point 20 invalid. Ignored.
Data point 25 invalid. Ignored.
1 2 3 4 5 6 7 8 7 10
2 12 13 13 15 16 17 18 17 7
3 4 6 8 10 2 4 6 8 10
4 3 5 7 1 3 7 7 11 13
5 10 11 12 13 16 18 11 12 7
6 1 2 2 3 3 3 4 4 4
7 7 8 7 6 5 4 1 2 2
8 11 11 13 13 13 17 17 7 7
continued
in
PROGRAM 8 - 7 Frequency and Histogram (continued )

13 17 17 15 15
0 0
1 4 * ***
2 7 *******
3 7 *******
4 8 * * * * ** * *
5 4 ****
6 5 *****
7 12 * * * ** * * * * * * *
8 5 *****
9 0
10 4 ****
11 5 *****
12 3 ***
13 8 ********
14 0
15 3 ***
16 2 **
17 8 ******
18 2 **
19 0
Program 8 - 7 Analysis Remember our discussion of what happens when an index gets out of range? What if
one of the numbers in our data is greater than 19? We would destroy some other
part of our program ! To protect against this possibility, we test each data value to
make sure that it is within the indexing range of frequency . If it is not, we display
an error message and read the next number.
Similarly, we guard against too much data . If the file contains more than 100 valid
numbers, then we stop reading. After the read loop, we test to make sure we are pro-
cessing all of the data. If there were too many numbers, we print an error message
.
Finally, note that we print the data to make it easy to verify the results.
Random Number Permutations

no num-
A random number permutation is a set of random numbers in which
bers are repeated. For example, given a random number permutation of 10
with no duplicates.
numbers, the values from 0 to 9 would all be included
4. To gen-
We saw how to generate a set of random numbers in Chapter
erate a permutation, we need to eliminate the duplicates . Borrowing from the
the previous section , we can solve the problem most
histogram concept in
efficiently by using two arrays. The first array contains
the random numbers .
that indicates whether or not the
The second array contains a logical value
in the random number
number represented by its index has been placed
array. I bis design is shown in Figure 8 - 1 ./
488 Section 8.4 Army Applications
randNos[0] 8 haveRand[0] 0
O means random
randNos[1] 3 haveRand[ 1] 1 not generated
randNos[2] 5 haveRand[2 ] 0
randNos[3] 1 haveRand[3] 1
1 means random
7 haveRand[ 4 ] 0 (v
^^
randNos[4] generated
randNos[5] haveRand[ 5] 1
randNos[6] haveRand[6] 0
randNos haveRand
After first five random numbers generated |
FIGURE 8 - 17 Design for random number permutations
As you study Figure 8- 17, note that only the first five random numbers
have been placed in the permutation. For each random number in the ran-
dom number array, its corresponding location in the have- random array is set
to 1 . Those locations representing numbers that have not yet been generated
are still set to 0. The implementation of the design is shown in Program 8- 81.
PROGRAM 8 - 8 Generate a Permutation

1 /* Generate a random number permutation.
2 Written by:
3 Date:
4 */
6 tinclude <stdlib.h >
7
8 # define ARY SIZE 20
9
11 void bldPerm ( int randNos[ ]);
12 void printData ( int data[ ], int size, int lineSize);
13
continued
1 . I lie numbers generated by this program are a

repeating series. To make them different with each
run , we would use the srand function and set
it to the time of day as explained in Chapter 4.
i v1
PROGRAM 8-8 Generate a Permutation ( continued )

15 {
17 int randNos [ARY SIZE];
18
19 // Statements
20 printf( " Begin Random Permutation Generation\n " );
21
22 bldPerm ( randNos );
23 _
printData (randNos, ARY SIZE, 10);
24
25 return 0;
26 > // main
27
28 /* ===
29 Generate a random number permutation in array.
30 Pre randNos is array to receive permutations
31 Post randNos filled
32 */
33 void bldPerm (int randNos[])
34 {
36 int oneRandNo;
37 _
int haveRand[ ARY SIZEJ = {0 > ;
38
39 // Statements
40 _
for (int i = 0; i < ARY SIZE; i++)
41 {
42 do
43 {
44 oneRandNo = rand( ) % ARY SIZE; _
45 } while ( haveRand[ oneRandNo ] == 1 );
46 haveRand[oneRandNo ] = 1;
47 randNos[i] = oneRandNo;
48 } // for
49 return;
50 } // bldPerm
51
52 ===== printData ===
53 Prints the data -
as a two dimensional array ,
54 Pre data: a filled array
55 last: index to last element to be printed
56 lineSize: number of elements on a line
57 Post data printed
58 */
continued
490 Section 8.5 Sorting
PROGRAM 8 - 8 Generate a Permutation ( continued )

59 void printData (int data[ ], int size , int lineSize)
60 {
62 int numPrinted = 0;
63
64 // Statements
65 printf( " \n " );
66 for ( int i = 0 ; i < size; i++)
67 {
68 numPrinted ++;
69 printf("%2d " , data[i]);
70 if ( numPrinted >= lineSize)
71 {
72 printf( " \n " );
73 numPrinted = 0 ;
74 > // if
75 > // for
76 printf( " \n " );
77 return ;
78 > // printData
Results:
Begin Random Permutation Generation
18 13 15 11 7 10 19 12 6 9
4 0 5 3 17 14 2 16 1 8
8.5 Sorting
One of the most common applications in computer science is sorting the
process through which data arc arranged according to their values. We are
—
surrounded by data. II the data are not ordered , we would spend hours trying
to find a single piece of information . Imagine the difficulty of finding some -
one’s telephone number in a telephone book that is not ordered in name
sequence!
In this chapter we introduce three sorting algorithms: the selection sort ,
bubble sort , and insertion sort. In each section we will first introduce the
basic concept, then use the idea in an example, and finally develop the code
for the algorithm .
One programming concept common to the sorting algorithms we discuss
in this section is the swapping of data between two elements in a list . You
might want to review our discussion in "Exchanging Values ” in Section 8.2.
m
Chapter 8 Arrays 491 1
Selection Sort
In the selection sort, the list is divided into two sublists, sorted and
.
unsorted, which are divided by an imaginary wall We find the smallest ele -
ment from the unsorted sublist and swap it with the element at the beginning
.
of the unsorted sublist After each selection and swapping, the wall between
the two sublists moves one element ahead, increasing the number of sorted
elements and decreasing the number of unsorted ones Each time we move .
one element from the unsorted sublist to the sorted sublist, we say that we
have completed a .
sort pass If we have a list of n elements, we need n - 1
passes to completely rearrange the data. The selection sort is graphically pre -
sented in Figure 8- 18.
Swap y
m smallest (a[k]...a[n - 1])
0 n-1
4
Sorted Unsorted
FIGURE 8 - 1 8 Selection Sort Concept
Figure 8 - 19 traces a set of six integers as we sort them. It shows how the
wall between the sorted and unsorted sublists moves in each pass. As you
study the figure, you will see that the array is sorted after five passes, which is
one fewer than the number of elements in the array. This means
that our sort
loop has one less iteration than the number of elements in the array.
BJ .
3 n
23 78 45 8 32 56 Original list 8 23 32 78 45 56 After pass 3
c4
4 4
4 Sorted Unsorted
Unsorted
PLi 8
an
23 32 45 78 56 After pass 4
8 78 45 23 32 56 After pass 1
4 4 4 4 »
4
Unsorted Sorted
8 23 32 45 56 78 After pass 5
8 23 45 78 32 56 After pass 2
4 4*
Sorted
4
^ ^ Unsorted
FIGURE 8 - 1 9 Selection Sort Example

The program design is a rather straightforward algorithm. Starting with

the first item in the list , it uses an inner loop to examine the unsorted items in
the list for the smallest element . In each pass, the smallest element is
“selected ” and exchanged with the first unsorted element. The outer loop is
repeated until the list is completely sorted . The design is shown in Figure 8-20.
/
(Selectior
^ ^\
cur = 0
ort
/
0
walk = \
/ cur X
*
/waik EHLL! \
\ ++ cur < last M \ ++ walk <= last m
smlst = cur
/ |ist[ walk ] \
o < list [smlst ]
0
Exchange
smlst = walk
cur & smlst Y

Return
o
FIGURE 8 - 20 Design for Selection Sort
I he code is shown in Program 8 - 9 . We have highlighted the inner loop to

make it easy to follow the logic .
PROGRAM 8 - 9 Selection Sort

l ===== selectionSort ======== ===========
2 Sorts by selecting smallest element in unsorted
3 portion of array and exchanging it with element at
4 the beginning of the unsorted list.
5 Pre list must contain at least one item
6 last contains index to last element in list
7 Post list rearranged smallest to largest
8 */
9 void selectionSort ( int list[], int last )
10 {
12 int smallest ;
13 int tempData ;
14
continuei
1n
l Ti
If |
Selection Sort (continued )

15 // Statements
16 // Outer Loop
17 for (int current = 0; current < last; current++)
18 {
19 smallest = current ;
20 // Inner Loop: One sort pass each loop
21 for (int walk = current + 1 ;
22 walk <= last ;
23 walk++ )
24 if ( list[walk] < list[smallest])
25 smallest = walk ;
26 // Smallest selected: exchange with current
27 tempData = list[current];
28 list[ current ] = list[smallest];
29 list[ smallest ] = tempData ;
30 > // for current
31 return ;
32 > // selectionSort
discussed in
Program 8 -9 Analysis In this algorithm we see two elements that are common to all three sorts e the
this section. First, each algorithm makes use of an inner loop either to determin
proper location of the next elemen t or to identify and exchang e two element s . Each
iteration of the inner loop is one sort pass.
Second, each time we need to move data, we must use a temporary storage area.
This technique is found in every sort algorithm except those that use two sorting areas.
In the selection sort, the temporary area is used to exchange the two element
s.
Bubble Sort
unsorted.
In the bubble sort, the list is divided into two sublists, sorted
and
the unsorted sublist and moved to the
I he smallest element is bubbled from
the sorted list , the wall
sorted sublist. After moving the smalles t element to
ahead, increasin g the number of sorted element s and

moves one element
from
decreasing the number of unsorted ones. Each time element
an moves
,
the unsorted sublist to the sorted sublist one sort pass is complet ed . Given a
up to n - 1 passes to sort the data.
list of n elements, the bubble sort requires
The bubble concept is seen in Figure 8 - 21 .
in each pass. Look-
Figure 8 - 22 shows how the wall moves one element
compare it to 32. Since 56 is not
ing at the first pass, we start with 56 and
one element . No exchanges
less than 32, it is not moved and we step down
than 45 , the two elements
take place until we compare 45 to 8. Since
8 is less
elemen t. Because 8 was moved down, it is
are exchan ged and we step down I
elements are exchanged. Finally, 8 is com
now compared to 78 and these two
exchanges places 8 in the first loca -
pared to 23 and exchanged. This series of
tion and the wall is moved up one position
.
Bubble up
Sorted Unsorted
FIGURE 8 - 21 Bubble Sort Concept
c 23 78 45 8 32 56 Original list Q 23 45 78 56 After pass 3

bu Unsorted *—Sorted Unsorted*
[ jj 23 78 45 32 56 After pass 1 8
—
23 32 45 56 78 After pass 4
— — Sorted!
Sorted ^ *
4 *
Unsorted
8 32 78 45 56 After pass 2
* Unsorted
FIGURE 8 - 22 Bubble Sort Example
I he bubble sort was originally written to “bubble up " the highest element
in the list . From an efficiency point ol view, it makes no difference whether
the high element is bubbled or the low element is bubbled. From a consis-
tency point of view, however, it makes comparisons between the sorts easier il
all three of them work in the same manner. For that reason, we have chosen
to bubble the lowest key in each pass.
Like the selection sort, the bubble sort is quite simple. In each pass
through the data, controlled by the outer for loop, the lowest element is
bubbled to the beginning ol the unsorted segment of the array. The bubbling
process is done in the inner loop.
1 ach time it is executed, the inner for loop makes one pass through the
data. Whenever it finds two elements out ol sequence, it exchanges them. It
then continues with the next element.1bis process allows the smallest element
to be bubbled to the beginning of the
array, while at the same time adjacent ele -
ments along the way are rearranged. The
design is shown in Figure 8 - 23.
The bubble sort is shown in Program 8 - 10.
pi
1
( Bubble Sort
^\ /
0 \X
//cur
cur = 0 walk = last
X wwalk
Vf + cur < last M V~ “ walk > cur M
0 list[walk ]
,< list [ walk-1
o
Exchange
( Return
^ cur & smlst
?
0
FIGURE 8 - 23 Bubble Sort Design
PROGRAM 8 - 10 Bubble Sort

1 === bubbleSort ===================:
2 Sort list using bubble sort. Adjacent elements are
3 compared and exchanged until list is ordered.
4 Pre the list must contain at least one item
6 Post list rearranged in sequence low to high
7 */
8 void bubbleSort ( int list [ ], int last )
9 {
11 int temp;
12
13 / / Statements
14 // Outer loop
++
15 for( int current = 0; current < last; current )
16 {
// Inner loop: Bubble up one element each
pass
17
18 for ( int walker = last ;
19 walker > current;
20
21
walker )
—
if ( list[walker] < list[ walker 1 ]) -
22 {
continues
PROGRAM 8 - 10 Bubble Sort ( continued )

23 temp = list[walker];
24 list[walker ] = list[walker 1]; -
25 list[walker - 1 ] = temp ;
26 > // if
27 > // for current
28 return ;
29 > // bubbleSort
Program 8 - 1 0 Analysis If the data being sorted are already in sequence, the inner loop still goes through the
array element by element. One common modification is to stop the sort if there are
no exchanges in the inner loop . This change requires the addition of a " sorted" flag.
We leave this modification to the practice sets at the end of the chapter.
Insertion Sort
The insertion sort algorithm is one of the most common sorting techniques
used by card players. As they pick up each card , they insert it into the proper
sequence in their hand.2
In the insertion sort , the list is divided into two parts: sorted and
unsorted. In each pass, the first element of the unsorted sublist is picked up
and transferred into the sorted sublist by inserting it at the appropriate place.
—
If we have a list of n elements, it will take at most n 1 passes to sort the data
( Figure 8- 24 ).
5
:
0 n-1
5r 4k
* Sorted
L-
Unsorted
FIGURE 8 - 24 Insertion Sort Concept
Figure 8- 2 S traces the insertion sort through our list of six numbers,
l ach pass moves the wall as an element is removed from the unsorted sublist
and inserted into the sorted sublists.
2. As an aside, card sorting is an example of a sort that

uses two pieces of data to sort: suit
and rank .
1
•
n
:
~
[ 23 E
78 45 8 32 56
Original
List
32 56
After
pass 3
Unsorted Sorted
After
45 8 32 56
After
pass 1 8 | 23 32 45 78
I pass 4
Unsorted Sorted
3 After After
23 45 8 32 56 8 23 32 45 56 78
3 pass 2 pass 5
Sorted Unsorted Sorted
FIGURE 8 - 25 Insertion Sort Example
The design of the insertion sort follows the same pattern we saw in both
the selection sort and the bubble sort ; each iteration of the outer loop is a
sort pass. The inner loop inserts the first element from the unsorted list into
its proper position relative to the rest of the data in the sorted list . I he design
is shown in Figure 8 - 26.
Insertion Sort
\ /
o
walk = \
cur = 1
V-
cur <= last' t M
M
I K\ _ cur ' 1
walk >= 0 & &#
JlocatedJ
\
located = false
temp <
list [walk]
list[ walk + 1 ]
= temp temp = list(cur) list [walk + 1 ]
= list[ walk ]
o
located = true
walk - -
o ?
I
Return
J
o
FIGURE 8 - 26 Insertion Sort Design
Program 8 - 1 I shows the insertion soil .

PROGRAM 8 - 11 Insertion Sort

1 ===== insertionSort === : =======
2 Sort list using Insertion Sort. The list is divided
3 into sorted and unsorted lists. With each pass, first
4 element in unsorted list is inserted into sorted list ,
5 Pre list must contain at least one element
7 Post list has been rearranged
8 */
9 void insertionSort ( int list[ ], int last)
10 {
11 // Statements
13 int walk ;
14 int temp;
15 bool located ;
16
17 // Statements
18 // Outer loop
19 for ( int current = 1; current <= last; current+t)
20 {
21 // Inner loop: Select and move one element
22 located = false;
23 temp = list[current ];
24 for ( walk = current 1; walk >= 0 && !located ;)
25 if (temp < list[walk ])
26 {
27 list[walk + 1 ] = list[walk];
28 walk ; —
29 > // if
30 else
31 located = true;
32 list ( walk + 1] = temp ;
33 > // for
34 return ;
35 > // insertionSort
Program 811 Analysis Note how the exchange is worked in this sort. Before the inner loop starts, we put the
data from the current element into the hold area (temp). This is the first step in the
exchange . The loop then determines the correct position to place the element by start -
ing with the largest element in the sorted list and working toward the beginning. As it
searches, it spreads the sorted portion of the list by shifting each element one position
higher in the list. When it locates the correct position, therefore, the data have
already been moved right one position and the current location is ''empty"; the sort
therefore simply places the saved element in its proper location, completing the
exchange.
Testing Sorts
Now that we have written three sorts, howr do we test them? The answer is
that we write a simple test driver. The same test driver can he used for all three
sorts. The only changes necessary are in the include statement, the prototype
statement, and the function call ( see highlighted statements). Program 8- 1 2
is set up to test the insertion sort .
PROGRAM 8 - 1 2 Testing Sorts

1 /* Test driver for insertion sort.
2 Written by:
3 Date:
4 */
7 # include "P08-11.C "
8
9 Idefine MAX ARY SIZE 15
10
12 void insertionSort (int list[ ], int last );
13
15 {
16
17 _ _
int ary[ MAX ARY SIZE] = { 89, 72, 3 , 15, 21,
18 57 , 61 , 44, 19 , 98 ,
19 5 , 77 , 39 , 59 , 61 };
20 // Statements
printf( " Unsorted: " );
21
22
_ _
for ( int i = 0; i < MAX ARY SIZE; i++)
23 printf( " % 3d " , ary[ i ]);
24
25
_
insertionSort ( ary , MAX ARY SIZE _ - 1 );
26
printf( " \nSorted : " );
27
28
_ _
for ( int i = 0; i < MAX ARY SIZE; i++ )
29 printf( " %3d " , ary[i]);
30 printf( " \n " );
31 return 0;
32 } // main
Results:
19 98 5 77 39 59 61
Unsorted: 89 72 3 15 21 57 61 44
57 59 61 61 72 77 89 98
Sorted : 3 5 15 19 21 39 44
Sorts Compared
It is interesting to compare the three sorting algorithms. All three use two
nes ted loops. The outer loop is the same in all three; it moves the wall one
element toward the beginning of the array. The processing in the inner loop,
however, changes with each sort .
In the selection sort , the inner loop starts at the unsorted portion of the
array and finds the smallest element in the unsorted area . Once located , it
swaps the smallest element with the first unsorted element . It then moves the
wall and continues the sort . Note that all exchanges take place in the
unsorted portion of the array; the sorted elements are never touched .
While the selection sort starts at the beginning of the unsorted portion of
the array, the bubble sort starts at the end . Another difference between the
two is that the bubble sort may make multiple exchanges in each pass. While
bubbling its way to the beginning of the array, it exchanges all elements that
are out of sequence. Like the selection sort , once the wall is moved , the
sorted data is never touched .
The insertion sort takes a completely different approach. In each pass,
the first unsorted clement is inserted into the sorted portion of the array.
While it might appear that there are no exchanges in the insertion sort , there
are . As explained previously, the first unsorted element is moved to a hold
area . This is the first step in an exchange. Then , as it loops through the
unsorted portion of the array, elements are shifted higher. Each shift is the
equivalent of the second step in an exchange. Finally, when the correct loca-
tion is found , the third step ol the exchange moves the temporarily stored ele -
ment to its correct position .
I able 8- 1 summarizes the exchange and shift characteristics of these
three sorts.
Sort Exchanges Shifts

Selection Once for each pass
Bubble Several in each pass
Insertion Only partially in loop Zero or more in each pass
TABLE 8 - 1 Sort Exchanges and Passes
Sort Conclusions
In this section , we have covered three classic sorts. With the exception of the
insertion sort , you generally will not find them implemented in production
systems. I he insertion sort is used as a subfunction
in both Quicksort and
m
Singleton ’s variation , Quickersort , which are considered the best general -

purpose sorts.3
Historically, however, these three sorts are the foundation of improved
and faster sorting methods that you will study in a data structures course .
I he selection sort is the foundation of a sorting method called the heap sort ;
the bubble sort is the foundation for Quicksort and Quickersort ; and the
insertion sort is the foundation of a sorting method called Shell Sort .
8.6 Searching
Another common operation in computer science is searching, which is the
process used to find the location of a target among a list of objects. In the
case of an array, searching means that given a value , we want to find the loca -
tion ( index) of the first element in the array that contains that value. The
search concept is shown in Figure 8- 27 .
Location wanted
(4)
a[ 0] a[ 1] a[ 2) a[3] a[4 ] a[ 5] a[6] a[7] a[8) a[9] a[ 10] a[11]

4 21 36 14 62 91 8 22 7 81 77 10
Target given
( 62 )
FIGURE 8 - 27 Search Concept
The algorithm used to search a list depends to a large extent on the struc-
ture of the list . Since our structure is currently limited to arrays , we will
study
searches that work with arrays . As we study other structures , we will study
different search techniques .
There are two basic searches for arrays: the sequential search and the
binary search . The sequential search can be used to locate an item in any
to be sorted .
arrav. The binary search , on the other hand, requires the list
Sequential Search
The sequential search is used whenever the list is not ordered
. Generally, we
lists lists that are not searched often. In
use the technique only for small or
Gilberg and Behrou / A. Forouzan , Data Structures:

3. For a discussion of Quicksort , see Richard F.
. Pacific Grove , CA: Brooks/Cole , 1998.
A Pseudocode Approach With C
I
502 Section 8.6 Searching
other cases we would first sort the list and then search it using the binary
search discussed later.
In the sequential search, we start searching for the target from the begin-
ning of the list, and we continue until we find the target or until we are sure
that it is not in the list . This gives us two possibilities; either we find it or we
reach the end of the list. Figure 8 - 28 traces the steps to find the value 62. We
.
first check the data at index 0 then 1, 2, and 3 before finding the 62 in the
fifth element ( index 4 ) .
location wanted'
4
index \
a(0] a[1] a[ 2] a[ 31 a[ 4 ] a [ 5 ] a[ 6] a [ 7 ] a(8 ] a[ 9 ] a [ 10 ) a [ 11 ]
4 l
21 36 14 62 91 8 22 7 81 77 10
1 Target given
62 != 4
index ^ < 621
afO] a[ 1) a[ 2] a[ 3] a[ 4 ] a[ 5] a[ 6] a( 7 ] a[ 8) a[ 9] a[10] a|11 ]
4 | 21 ; 36 14 62 91 8 22 7 81 77 10
2
62 != 21
index
afO) a[l ) a[ 2 ) a[ 3 ] a[ 4 ] a[ 5 ] a (6 ) a[ 7 ] a [ 8 ] a [ 9] a [ 10 ) a [ 11 )
4 21 [ 36» 14 62 91 8 22 7 81 77 10
3
62 != 36
index
a [ 0] a[ 1 ] a[ 2 j a[ 3] a [ 4 ] a[ 5) a [ 6 ] a [ 7 ] 3( 8 ] a[ 9 ] a[ 10] a [ 11 ]
4 21 36 14 62 91 8 22 7 81 77 10
4
62 != 14
index
a[0] a[ 1 ] a[2] a[3T a[ 4 ] a[ 5 ] a[ 6] a[7 ] a|8) a[ 9 j a[ 10) a [ 11 ]
4 21 36 14 62 | 91 8 22 7 81 77 10
62 == 62
FIGURE 8 - 28 Locating Data in Unordered List
But what il the target is not in the list ? Then we have to examine each cle-
ment until we reach the end ol the list . Figure 8 - 29 traces the search fora target
ol 2. W hen we detect the end ol the list, we know that the target does not exist.
Let s write the sequential search function. A search function needs to tell
the calling function two things: Did it find the data it was looking lor? If it
did. what is the index at which the data were found?
But a function can return only one value. For search functions, we use
the ietuin value to designate whether we found the
target or not. To ‘‘ return
the index location where the data were found, we will use call by-address
- .
I he search Junction requires four parameters: the
list we are searching,
the index to the last element in the list, the
target, and the address where the
found element 's index location will he stored. Althougl1 we could write it
without passing the index to the last element, that
would mean the search
would have to know how many elements are in the list. To make the function
as flexible as possible, therefore, we pass the index of the last data value in
the array. I his is also a good structured design technique. With this informa-
tion, we are now ready to create the design. It is shown in Figure 8- 30.
\
I Target
index
.a[0]
A
a( 1) a( 2] a[3] a[ 4) a[ 5 ] a[6 ] a[ 71 a[ 01 a[9 ] a( 10 ) a [ 11 ]
4 21 36 14 62 91 8 22 7 81 77 10
1 72 1= 4
index
a[0Pa[ 1 ) a|2] a[ 3] a[ 4 ] a|5 ] a [ 6 ] a[7 ] a(8) a(9) a[10 ] a[ 11 ]
4 21 ] 36 14 62 91 0 22 7 81 77 10
5 72 != 21
index
a(0) a|i) a|2) a [ 3 ] a]
^
a[ 5) a [ 6 ] a { 7) a[8 ] a[ 9) a|10] a[ll]
4 I 21 I 36 I 14 I 62 I 91 I 0 | 22 I 7 I 81 1 77 I 10
12 72 != 91
index
a (0 ] a ( 1 ] a [ 2 ] a [ 3! a( 4 ] a[ 5 j a(h ] n • r
4 21 36 14 62 91 8 22 7 81 77 10
Index off
Note : Not all test points are shown end of list
FIGURE 8 - 29 Unsuccessful Search in Unordered List
( SeqSearch
looker = 0
^
/
/ looker < last
&& target !=
\\
\ list[ looker ] M
looker++
* locn = looker
set found
( Return
J
FIGURE 8 - 30 Sequential Search Design
The code is shown in Program 8 - 13.

PROGRAM 8 - 13 Sequential Search

1 ==== seqSearch ===
2 Locate target in an unordered list of size elements ,
3 Pre list must contain at least one item
4 last is index to last element in list
5 target contains the data to be located
6 locn is address for located target index
7 Post Found: matching index stored in locn
8 return true (found )
9 Not Found: last stored in locn
10 return false ( not found )
11 */
12 bool seqSearch ( int list[ ], int last ,
13 int target , int * locn)
14 {
16 int looker;
17 bool found ;
18
19 // Statements
20 looker = 0 ;
21 while (looker < last && target != list[ looker])
22 looker++ ;
23
24 *locn = looker ;
25 found = ( target == list[ looker]);
26 return found ;
27 > // seqSearch
Program 8 - 1 3 Analysis Program 8 - 13 is simple, but it does merit some discussion. First, why did we use a
while statement? Even though we know the limits of the array, it is still an event -
controlled loop. We search until we find what we are looking for or reach the end of
the list. Finding something is an event, so we use an event loop.
Next, note that there are two tests in the limit expression of the loop. We have
coded the test for the end of the array first. In this case, it doesn' t make any difference
which test is first from an execution point of view, but in other search loops it might. You
should get in the habit of coding the limit test first because it doesn' t use an indexed
value and is therefore safer.
The call-by - address use for locn also merits discussion. Since we need to pass the
found location back to the variable in the calling program, we need to pass its address
to the function. A typical call to the search would look like the
statement shown below.
found = seqSearch (stuAry , lastStu , stuID, &locn);
Notice how succinct this function is. In fact, there are more lines of documentation
than there are lines of code. The entire search is contained in one while statement. With
this short code, you might be tempted to ask, "Why write the function at all? Why not
j
\ \l
'
just put the one line of code wherever it is needed?'' The answer lies in the structured
programming concepts that each function should do only one thing and in the concept
of reusability. By isolating the search process in its own function, we separate it from the
process that needs the search . This is better structured programming. This also makes
the code reusable in other parts of the program and portable to other programs that
need searching. S ’!
One final point: This function assumes that the list is not ordered. If the list were
ordered, we could improve the search slightly when the data we were looking for were
not in the list. We will leave this improvement for a problem at the end of the chapter.
Binary Search
The sequential search algorithm is very slow. If we have an array of 1 million
elements, we must do 1 million comparisons in the worst case. If the array is
not sorted, this is the only solution. But if the array is sorted, we can use a
more efficient algorithm called the binary search. Generally speaking, we
should use a binary search whenever the list starts to become large. The defi -
nition of large is vague. We suggest that you consider binary searches when -
ever the list contains more than 50 elements.
The binary search starts by testing the data in the element at the middle
of the array. This determines if the target is in the first half or the second half
of the list. If it is in the first half, we do not need to check the second half. If
it is in the second half, w'e don’t need to test the first half. In other words,
either way we eliminate half the list from further consideration. We repeat
this process until we find the target or satisfy ourselves that it is not in the list .
To find the middle of the list, we need three variables, one to identify the
beginning of the list, one to identify the middle of the list, and one to identify
the end of the list . We will analyze two cases: the target is in the list , and the
target is not in the list .
Target Found
Figure 8 - 51 shows how .
we find 22 in a sorted array We descriptively call our
I I we
three indexes first , mid, and last. Given first as 0 and last as
4
can calculate mid as follows :
mid = ( first + last ) / 2 ;
integral value of
Since the index mid is an integer, the result will be the
the quotient; that is, it truncates rather than rounds the calculation. Given
the data in Figure -
8 3 I , mid becomes 5 as a result of the first calculation.
mid = (0 + 11) / 2 - 11 / 2 5
of elements in the array is greater than half

4 This formula does not work if the number
.
In that case ,
the correct formula is: mid first + ( last first) 2.
=
MAX INT.
first mid last

m m RT (JargetT22 )
~
afOl a [1 ] a[2 l a[3] a [4 ] a[5] a[6] a[7] a [8] a [9] a [TO ]a[11 ]
4 I 7 I 8 10 I 14 21 I 22 I 36 I 62 I 77 I 81 91
first mid last
22 > 21
rinrinrTr
a[0] a [1 ] a [2] a[3] a [4 ] a[5] a [6] a [ 7] a[8] a[9] a[1 0 ) a [ 11 ]
4 7 [ 8 10 | 14721 | 22 | 36 | 62 | 77 | 81 | 91
first mid last 22 < 62
; ] a [8] a [9] a [ 10 ]a [11 ]

a[0 ] a [1 ] a [2] a [3] a [4 ] a [5] a[6] a [7
4 7 8 10 14 21 22 36 62 77 81 91
first mid last 22 = = 22

CDCD CD
Function terminates
FIGURE 8 - 31 Binary Search Example
At index location 5 , we discover that the target is greater than the list
value ( 22 > 21 ) . We can therefore eliminate the array locations 0 through 5.
( Note that mid is automatically eliminated . ) To narrow our search , we assign
mid + 1 to first and repeat the search .
I he next loop calculates mid with the new value for first and deter -
mines that the midpoint is now 8.
mid = (6 + 1 1) / 2 = 17 / 2 = 8
Again we test the target to the value at mid , and this time we discover
that the target is less than the list value ( 22 < 62 ) . This time we adjust the
—
ends ol the list by setting last to mid 1 and recalculate mid . This elimi -
nates elements 8 through 1 1 from consideration . We have now arrived at
index location 6 , whose value matches our target . This stops the search ^ . Fig-
ure 8 - 31 traces the logic we have just described .
Target Not Found

\ more interesting case occurs when the target is not in
the list . We must con -
struct oui search algorithm so that it
stops when we have checked all possible
locations. I his is done in the binary search by testing for first and last
crossing; that is , we are done when first becomes greater than last. Thus,
l 5. To terminate the search , we set first to last

+ I . See Program 8- 14.
T
’ ?
!
only two condition s terminate the binary search algorithm: Either the target is
( ound or first becomes larger than last . For example , imagine we want to
lind I 1 in our binary search array. This situation is shown in Figure 8 - 32 .
In this example, the loop continues to narrow the range as we saw in the
successfu l search , until we are examining the data at index locations 3 and 4.
These settings of first and last set the mid index to 3.
mid = (3 + 4) / 2 = 7 / 2 3
I he test at index location 3 indicates that the target is greater than the
list value, so we set first to mid + 1 or 4. We now test the data at location 4
and discover that 1 1 < 1 4.
mid = (4 + 4 ) / 2 = 8/ 2 = 4
At this point , we have discovere d that the target should be between two
adjacent values; in other words, it is not in the list . We see this algorithm i-
cally because last is set to m i d - 1, which makes first greater than last ,
the signal that the value we are looking for is not in the list .
first~
~
mid~ last
0 11 5 l 111
a[0 ] a[ 1 ] a[ 2 ] a [3 ] a[ 4 ] a[5 ] a[ 6 ] a [ 7 ] a[ 8 ] a[9 ] a[ 10)a [ 11 ]
4 I 7 I 8 I 10 I 14 I 21 I 22 I 36 62 I 77 81 91
first mid last 11 < 21
^ nnrinr
a[ 0 ] a[ 1 ] a[ 2 ] a[ 3] a[ 4) a [5 ] a [ 6) ' a[ 7 ] a [8 ] a [9 ] a[ 10)a[ 11 ]
4 I 7 I 8 I 10 I 14 I 21 22 36 I 62 ' 77 81 T 91
first mid last
[ ] [ 10 ]a [ 11 ]
a[ 0] a[ 11 a[ 2 ] a[ 3 ] a[ 4 ] a[ 5 ] a[ 6 ] a [ 7 ] a[ 8 ] a 9 a
4 7 8 I 10 14 I 21 I 22 I 36 I 62 I 77 | 81 | 91
first mid last
11 > 10
npnpr jj
T
a[ 3) a[ 4 ] a[ 5 ] a[ 6 ] a [ 7 ] a[ 8 ] a [9 ] a[ 10 ]a[ 11 ]
| 4 | 7 |8 to ] 14 [ 2?] 22 I 36 I 62 I 77 l 81 ] 91 |
first mid last
11 < 14
(2JCDC D
Function terminates
FIGURE 8 32 Unsuccessful Binary Search Example

*
'1
Once we fully understand the logic, we can design the program Figure 8- 33 .
contains the design in the form of a flowchart .
Binary
o
mid =
first = 0
last = end ( first + last) / 2
first <= Iasi —
o target <
list [mid] first = mid + 1
o first = last = mid - 1
* locn = mid
set found ?
( Return
^ 6
FIGURE 8 - 33 Design for Binary Search
Program 8 - 14 contains the implementation of the binary search algo-

rithm we have been describing. It is constructed along the same design we
saw for the sequential search. The first three parameters describe the list and
the target we are looking lor, and the last parameter contains the address into
which we place the located index. One point worth noting: When we termi-
nate the loop with a not- found condition, the index returned is unpredict -
able —it may indicate the node greater than or less than the value in target.
PROGRAM 8 - l 4 Binary Search

1 / * == === binarySearch ==== :
====
2 Search an ordered list using Binary Search
4 end is index to the largest element in list
5 target is the value of element being sought
6 locn is address for located target index
7 Post Found: locn = index to target element
8 return 1 ( found )
9 Not Found: locn = element below or above target
10 return 0 ( not found )
continuei
Fy n
PROGRAM 8- 14 Binary Search ( continued )
11 */
12 bool binarySearch ( int list[ ], int end ,
13 int target, int* locn)
14 {
16 int first;
17 int mid ;
18 int last;
19
20 // Statements
21 first = 0;
22 last = end ;
23 while ( first <= last)
24
25 mid = ( first + last ) / 2 ;
26 if (target > list[mid ])
27 // look in upper half
28 first = mid + 1;
29 else if (target < list[ mid ])
30 // look in lower half
31 last = mid 1; -
32 else
33 // found equal: force exit
34 first = last + 1 ;
35 } // end while
36 * locn = mid ;
37 return target == list [ mid ];
38 } // binarySearch
87 Two- Dimensional Arrays

The arrays we have discussed so far are known as one -dimensional
arrays
because the data are organized linearly in only one direction . Many appli -
cations require that data he stored in more than one dimension . One corn -
is a table , which is an array that consists of rows and
mon example
called a two-
columns. Figure 8 - 34 shows a table , which is commonly
dimensional array.
by Figure 8-34 ,
Although a two-dimensional array is exactly what is shown
C looks at it in a different way . It looks at the two - dimension al array as an
words , a two - dimensio nal array in C is an array of
array of arrays. In other
one-dimensional arrays . This concept is shown
in Figure 8 - 35 .
510 Section 8.7 Two-Dimensional Arrays
c
.2
w>
c) ^
a </>
it
il
Second Dimension
v (columns) >
FIGURE 8 - 34 Two-dimensional Array
i
WBBKm
table [0][0] table [0][1] table [0 ][2] table [0][3]
table [0]
J I I
table [1][0] table [1][1] table [1][2] table [1][3]
table [1]
table [2][0] table [2][1] table [2][2] table [2][3]

table [2 ]
I I [
table [3][0] table [3](1] table [3][2] table [3][3]
table [3]
table [4][0] table [4][1] table [4][2] table [ 4][3]

table [ 4]
table
FIGURE 8 - 35 Array Of Arrays
Declaration
Iwo - dimensional arrays, like one-dimensional arrays, must be declared
before be ing used. Declaration tells the compiler the name of the array, the
t \ pe < > l each element, and the
size of each dimension. Two-dimensional
arrays can he either fixed
length or variable length.
n
m
-
As we saw with the one dimensional array, the size of a fixed - length array
is a constant and must have a value at compilation time. For example , the
array shown in Figure 8 - 35 can he declared and defined as follows:
int table[5 ][4]; 1

*1
By convention , the first dimension specifies the number of rows in the

array. The second dimension specifies the number of columns in each row.
-
The rules for variable length arrays follow the same concepts. To define a
variable length array, we would use the following statement.
-
int table5x4 [ rows][cols ];
Remember, however, that the dimensions must be set before the declara -
tion or a compiler error occurs.
Initialization
As we noted before, the definition of a fixed - length array only reserves mem -
ory for the elements in the array. No values will he stored. If we don t initial -
'
ize the array, the contents are unpredictable. Generally speaking , fixed -length
arrays should be initialized .
Initialization of array elements can he done when the array is defined. As
previously noted , variable-length arrays may not he initialized when they are
defined. As with one dimensional arrays, the values must he enclosed in braces.
-
This time, however, there is a set of data for each dimension in the array. So for
table , we will need 20 values . One way to initialize it is shown below.
int table[5][ 4 ] =
{ 0 , 1 , 2 , 3, 10, 11 , 12 , 13 , 20, 21 ,
22 , 23 , 30, 31 , 32 , 33 , 40, 41 , 42 , 43};
We highly recommend , however, that nested braces he used to show the

exact nature of the array. For example, array table is better initialized
as
shown below.
int table [5][4 ] =

{
{ 0 , 1 , 2, 3},
{10 , 11, 12 , 13},
{20 , 21 , 22 , 23},
{30, 31 , 32 , 33},
{40 , 41 , 42, 43}
> ; // table
In this example, we define each row as a one-
dimensional array of four
. The array of five rows also has its set of braces.
elements enclosed in braces
Note that we use commas between the elements in the rows and also com-
mas between the rows.
When we discussed one- dimensional arrays , we said that if the array is
com pletely initialized with supplied values , we do not need to specify the size
of the array. This concept carries forward to multidimensional arrays, except
that only the first dimension can be omitted. All others must he specified.
The format is shown below.
int table [][4] =

{
{ 0, 1, 2 , 3} ,
{10, 11, 12, 13},
{20, 21 , 22 , 23},
{30 , 31, 32, 33},
{40, 41 , 42, 43}
}} // table
To initialize the whole array to zeros , we need only specify the first value
as shown below.
int table [5][4] = {0};
Inputting Values
Another way to ( ill up the values is to read them from the keyboard . For a
two-dimensional array, this usually requires nested for loops. If the array is an
—
n by m array, the first loop varies the row from zero to n 1 . The second loop
varies the column from zero to m - 1 . The code to fill the array in Figure 8- 35
is shown below .
for ( row = 0; row < 5; row++)

for ( column = 0; column < 4 ; column++ )
scanf ( " %d " , stable [ row][column]);
W lien the program runs, we enter the 20 values for the elements and
they are stored in the appropriate locations.
Outputting Values
We can also print the value of the elements one by one < , using two nested
loops. Again , the first loop controls the printing of the rows and the second
loop controls the printing ol the columns. To print the table in its table format ,
a newline is printed at the end of each row. The code to
print Figure 8- 35 is
shown below .
for (row = 0; row < 5 ; row++ )
continual
\ 'TTl / l
n
for ( column = 0; column < 4; column++ )

printf( " %8d " , table [row][ column ]);
printf("\n");
> // for
Accessing Values
Individual elements can be initialized using the assignment operator.
table[ 2][0] = 23 ;
table[ 0 ][ 1 ] = table[ 3 ][ 2] + 15;
Let s assume that we want to initialize our 5 x 4 array as shown below.
00 01 02 03
10 11 12 13
20 21 22 23
30 31 32 33
40 41 42 43
One way to do this would be to code the values hv hand . However, it is

much more interesting to examine the pattern and then assign values to the
elements in the array using an algorithm . What pattern do you see ? One is that
the value in each element increases by one from its predecessor in the row.
Another is that the first element in each row is the row index times 10. With
these two patterns, we should he able to w rite nested loops to fill the array. The
code to initialize the patterns for the array is shown in Program 8- 15.
PROGRAM 8 - 15 Fill Two - dimensional Array

1
2 This function fills array such that each array element
3 contains a number that , when viewed as a two digit -
4 integer , the first digit is the row number and the
5 second digit is the column number ,
6 Pre table is array in memory
7 numRows is number of rows in array
8 Post array has been initialized
9 */
10
_
void fillArray ( int table[ ][ MAX COLS], int numRows)
11 {
12 // Statements
13 for ( int row = 0; row < numRows; row++ )
14 {
15 table [row ](0] = row * 10;
continued
PROGRAM 8 - 15 Fill Two-dimensional Array [ continued )

16 for ( int col —
1 ; col < MAX COLS ; col++ ) _
17 table [ row][col ] = table [ row][col 1] + i; -
18 > // for
19 return;
20 } // fillArray
Memory Layout
As discussed earlier, the indexes in the definition of a two-dimensional array
.
represent rows and columns This format maps to the way the data are laid
out in memory. If we were to consider memory as a row of bytes with the low-
est address on the left and the highest address on the right , then an array
would be placed in memory with the first element to the left and the last ele -
ment to the right . Similarly, if the array is a two- dimensional array, then the
first dimension is a row of elements that are stored to the leit. This is known
as “row- major " storage and is seen in Figure 8 - 36.
00 01 02 03 04
10 11 12 13 14
User’s View
row 0 row 1
00 01 02 03 04 10 11 12 13 14
[ 0][ 0] 0][ 1] [0][2 0][ 3] 0][ 4] 1][ 0] 1][1] 1 ][ 2] 1][ 3] [ 1 ][ 4]
Memory View
FIGURE 8 - 36 Memory Layout
Memory Example: Map One Array to Another

lo Iurther examine how data arc laid out in memory, look at Program 8 - 16,
which converts a two-dimensional array to a one-dimensional array.
PROGRAM 8 - 16 Convert Table to One - dimensional Array

1 / * This program changes a two-dimensional array to the
2
3
-
corresponding one dimensional array.
Written by:
continuei
.
6 At least one \language, FOR IRAN, reverses
the data values placement in memory. It stores
data by columns.
m1
PROGRAM 8 - 16 Convert Table to One-dimensional Array (continued )
4 Date:
5 */
7 #define ROWS 2
8 tdefine COLS 5
9
11 {
13 int table [ ROWS ] [ COLS ] =
14 {
15 {00 , 01 , 02 , 03 , 04},
16 {10, 11 , 12, 13 , 14}
17 }; // table
18 int line [ROWS * COLS];
19
20 // Statements
21 for ( int row = 0; row < ROWS ; row++ )
22 for ( int column = 0; column < COLS ; column++ )
23 line[ row * COLS + column ] = table[ row ][ column ];
24
25 for ( int row = 0; row < ROWS * COLS; row++ )
26 printf( " %02d " , line[row]);
27
28 return 0 ;
29 } // main
Results:
00 01 02 03 04 10 11 12 13 14
Program 8- 16 Analysis In Program 8 - 16 we use nested for loops to make the conversion . The first loop con-
trols the table rows, and the second loop controls the columns within each row. Since
we know how many elements are in each row of the two dimension array
- al , we can
by simply multiplying the row index by the num -
map it to the one - dimensional array
each When we are in row zero , we multiply the number of
ber of elements in row .
defined , ) by the row number ( 0 )
elements in one row ( designed by the constant COLS
result the column This maps the elements in the two - dimensional array
and add the to .
row 1 , we add the number of
to the beginning of the receiving array. When we are in
times the row ( 1 ) to the current column , which maps the elements
elements in one row
we add the product of the row and
to the next set ( in this case, 5 .. . 9) . To generalize,
location.
the number of elements in a row to the column to determine the receiving the and col-
Study the technique we used to define the number of elements in rows
only define the arrays but also to control the loop
umns. Note how they are used not to
with different row - column sizes to help
execution. Copy this program, and then run it also . You might
will need to change the initializatio n
understand the differences. You (
want to consider initializing the values by assignment
s within for loops. )
Passing A Two- Dimensional Array

With two-dimensional arrays, we have three choices for passing parts of the
array to a function . ( 1 ) We can pass individual elements, as shown in “ Passing
Individual Elements” in Section 8.3. ( 2 ) We can pass a row of the array. This
is similar to passing an array, as we saw in Passing the Whole Array” in
"
Section 8.3. ( 3 Finally

) , we can pass the whole array.
Passing A Row
Passing a row of the array is rather interesting. We pass a whole row by
indexing the array name with only the row number. Referring to Figure 8- 37,
we see that a row is four integers. When we pass the row, therefore, the
receiving function receives a one-dimensional array of four integers. The for
_
loop in print square prints the square ol each ol the lour elements. After
printing all the values, the function advances to the next line on the console
_
and returns. The for loop in main calls print square five times so that the
final result is a table of the values squared shown on the monitor.
#d e f i n e MAXROWS 5
#d e f i n e M A X C O L S 4
/ / Function Declarations
k
void print square ( int [ ] ) ; table
i n t main ( void )
{ 0 1 2 3
i n t t a b l e ( MAXROWS ) ( M A X C O L S ) = 10 11 12 13
{
{ 0 , 1, 2 , 3 >, 22 23
{ 1 0 , 1 1, 1 2 , 13 } , fy 20 21
{
{
2 0 , 2 1 , 22 ,
3 0 , 3 1, 3 2 ,
23 } ,
33 },
\ 30 31 32 33
{
};
40 , 41, 42 ,
/* table */
43 >
f o r ( i n t r o w = 0 ; row < MAX ROWS ; r o w ++ )

r 40 41 42 43
_
p r i n t s q u a r e ( t a b l e [ row ] ) ;
return 0; address
) / / main
of a row
_
void print square ( int x ( ] )
{
f o r ( i n t c o l = 0 ; c o l < MAXCOLS ; c o l + + )
k X
p r i n t f ( " % 6d ” , x [ c o l ] * x [ c o l ] ) ; 0 14 9
100 121 144 169
printf ( "\ n" ); 400 441 484 529
900 961 1024 1089
return ; 1600 1681 1764 1849
} / / printsquare
FIGURE 8 - 37 Passing a Row

'I '
Note that we have declared the array dimensions as defined constants.

I his allows us to symbolically refer to the limits of the array in both the array
definition and the Jor loops. Now, if we need to change the si/e of the array,
all that is necessary is to change the define declarations, and any array initial -
izers, and recompile the program.
We could code the print function array in Figure 8 - 37 as a variable -
length array of one dimension. Io do so, we would need to change the func -
tion definition and the calling statement to pass the array size as shown in the
following example.
_
// Function call
_
print square (MAX COLS, table);
// Function definition
_
void print square ( int size, x[ size])
> _
// print square
Passing the Whole Array

When we pass a two-dimensional array to a function, we use the array name
as the actual parameter just as we did with one-dimensional arrays. The for -
mal parameter in the called function header, however, must indicate that the
array has two dimensions. I bis is done hv including two sets of brackets, one
for each dimension, as shown below.
_
double average ( int table[][ MAX COLS ])
Note that again we do not need to specify the number of rows in a fixed -
length array. It is necessary, however, to specify the size of the second dimen -
sion. Thus, we specified the number of columns in the second dimension
_
( MAX COLS). In summary, to pass two-dimensional arrays to functions:
1 The function must he called by passing only the array name.

.
2. In the function definition, the formal parameter is a two-dimensional array,
.
with the size of the second dimension required for a fixed- length array
size of all
3 In the function definition for a variable-length array, the
.
dimensions must be specified.
For example, we can use a function to calculate the average of the
integers in
.
an array In this case, we pass the name of the array to the function as seen in
Figure 8 - 38.
__
#define MAX ROWS 5 table
#define MAX COLS 4 1 2 3
_
double average (int [ ][ MAX COLS ]); 11 12 13
int main (void ) 21 22 23
{
double ave;
int table[MAXROWS][ MAX
__COLS] = 31 32 33
{ 41 42 43
{ 0, 1, 2, 3 },
{ 10, 11, 12, 13 },
{ 20, 21 , 22 , 23 },
{ 30, 31, 32, 33 },
{ 40, 41 , 42 , 43 >
}; // table
ave = average (table);
return 0;
} // main
Address
double average (int x( ]( MAX _COLS ])

of table
{
double sum = 0;
for (int i = 0 ; i < MAXROWS; i++) X
for (int j = 0; j < MAX_COLS; j++)
sum += x [i] (j];
return (sum / ( MAX ROWS * MAXCOLS)); sum
} // average
FIGURE 8 - 38 Calculate Average of Integers in Array
EXAMPLE 8 - 5 Two-dimensional Array

Write a program that Fills the left - to- right diagonal of a square matrix (a two-
dimensional array with an equal number of rows and columns ) with zeros,
the lower left triangle with - Is , and the upper right triangle with + ls. The
-
output ol the program , assuming a six - by six matrix is shown in Figure 8- 39.
FIGURE 8 - 39 Example of Filled Matrix

t he program code is shown in Program 8- 17.
PROGRAM 8 - 1 7 Fill Matrix

1 / * This program fills the diagonal of a matrix (square
2 array ) with 0 , the lower left triangle with - 1 , and si
3 the upper right triangle with 1.
4 Written by:
5 Date:
6 */
8
9 int main ( void )
10 {
12 int table [6](6];
13
14 // Statements
15 for ( int row = 0; row < 6 ; row++)
16 for (int column = 0; column < 6; column++)
17 if ( row == column )
18 table [row][column ] = 0;
19 else if ( row > column )
20
21 else
table [row][column ] = 1; -
22 table [row][column] = 1 ;
23
24 for (int row = 0; row < 6 ; row++ )
25 {
26 for ( int column = 0 ; column < 6 ; column++ )
27 printf( "% 3d " , table[ row ][column ]);
28 printf( "\n" );
29 } // for row
30 return 0;
31 } // main
Program 8 - 1 7 Analysis This iis a rather simple pattern problem, similar to many we saw in Chapter 6 . The
there
only difference is that now we are creating the pattern in array elements. Since the
, need loops to control the pattern . If the row equals
are two dimensions we two
, then we are
column, we assign 0 to the element. If the row is greater than the column
, assign - 1 . And if the row is less than the col-
in the lower half of the matrix so we
umn, then we are in the upper half of the matrix
and we assign + 1 .
8.8 Multidimensional Arrays

Multidimensional arrays can have three, four, or more
dimensions. Figure 8 40 -
describe
shows an array of three dimensions. Note the
terminology used to
520 Section 8.8 Multidimensional Arrays
the array. The hrst dimension is called a plane, which consists of rows and
columns. Arrays of four or more dimensions can be created and used, but
they are difficult to draw.
Second Dimension) 2
.
s ( rows)
3
2
0
0 1 2 3
&&
/" Third Dimension"/
x )
^ fcolumns ^ ^
FIGURE 8 - 40 A Three- dimensional Array ( 3 x 5 x 4)
.
Although a three-dimensional array is exactly what is shown in Figure 8- 40.
the C language looks at it in a different way. It takes the three-dimensional array
to be an array ol two-dimensional arrays. It considers the two - dimensional
array to he an array ol one-dimensional arrays . In other words, a three -
dimensional array in C is an array of arrays of arrays. This concept also holds
true lor arrays ol more than three dimensions. The C view of a three-
dimensional array is seen in Figure 8 - 41.
Declaring Multidimensional Arrays

Multidimensional arrays, like one -dimensional arrays, must he declared
before being used. Declaration tells the compiler the name of the array, the
type of each element, and the size of each dimension. The size of the fixed-
length array is a constant and must have a value at compilation time. The
three -dimensional array seen in Figure 8 - 4 1 can he declared as follows f o r a
fixed-length array.
int table[ 3][ 5][4 ];
I he definition for the same table as a variable- length array requires three
variables, one for the rows, one for the columns, and one for the planes.
( » i \ c n these variables, it
would be defined as shown in the following example.
int table[ planes][rows][cols];

HJI
'
200 201 202 203

table [2][0][0] table [ 2][0][1] table [ 2 ][0][ 2 ] table [ 2][0][3]
100 l 101 102 103

table [1][0][0] table [1][0][1] table [1][0][2] table [1][0][3] 213
[ 2][ 1 ][3]
0 1 2 I 3
113
table [0][0][0] table [0][0][1] table [0][0][2] table [0][0][3]
1][1][3)
223
table [0][0] [2][2][3]
10 11 12 13
123
table [0][1][0] table [0][1][1] table [0][1 ][2] table [0][1][3] 233
[ 1 ][ 2 ][ 3 ]
table [01(1] [2](3][3]
20 21 22 23
133
table [0][2][0] table [0][ 2][1] table [0][ 2][2] table [0][2][3] 243
[ 1][3][3]
table (0][2) [2][4][3]
30 31 32 33 table[2]
143
table [0][3][0] table [0][3)[1] table [0][3][2] table [0][3][3]
MM3]
table [0][3]
table[1]
40 41 42 | 43
table [0][ 4][0] table [0][4][1] table [0) [4][ 2] table [0][ 4][3]
table [0][ 4] table[0]
FIGURE 8 - 41 C View of Three - dimensional Array
Initialization
As we said before, declaration and definition only reserve space for the ele-
ments in the array. No values will he stored in the array. If we want to
store
,
values we must either initialize the elements , read values from the keyboard ,
or assign values to each individual element . Following the rules we discussed
for one- and two-dimensional arrays, only fixed - length multidimensional
arrays can be initialized when they are defined .
saw
Once more, initialization is simply an extension of the concept we )
Section 8 ./ .
for initializing a two-dimensional array ( see Initialization in
For the three-dimensional array, we nest each plane in a set of brackets . For
as we did for the two -dimensional array .
each plane, we bracket the rows
plane and row as we have done in the following
When we group the data by
code , the reader will be able to visualize the array with ease
. We have added
comments to make it even easier to read the values
.
int tablet 3 ][ 5][ 4 ] -

{
{ // Plane 0
continued
If
522 Section 8.9 Programming Example — Calculate Averages
{0, 1, 2, 3}, // Row O

{10, 11 , 12, 13}, // Row 1
{20, 21, 22, 23}, // Row 2
{30, 31, 32 , 33}, // Row 3
{40, 41, 42, 43} // Row 4
>,
{ // Plane 1
{100, 101, 102, 103}, // Row 0
{110, 111, 112, 113}, // Row 1
{120, 121 , 122, 123}, // Row 2
{130, 131 , 132 , 133}, // Row 3
{140, 141, 142 , 143} // Row 4
>,
{ // Plane 2
{200, 201, 202, 203}, // Row 0
{210, 211, 212, 213}, // Row 1
{220, 221 , 222 , 223}, // Row 2
{230, 231, 232, 233}, // Row 3
{240, 241, 242, 243} // Row 4
>
}; // table
As we saw previously, the planes size, and only the plane’s size, does not
need to he specified when we use explicit initialization . The size of all dimen-
sions after the first must he explicitly stated .
Ot course, if we want to initialize all the elements to zero, we can simply
initialize only the first element to zero and let the compiler generate the code
to initialize the rest of the array.
int table [ 3 ][ 5 ][ 4] = {0};
8.9 Programming Example — Calculate Averages

In Chapters 4 and 5 we introduced the software engineering concept ot top-
down development and a programming concept known as incremental devel-
opment . In this chapter, we develop a n example that contains many of the
programming techniques found in array problems and implement it imcre-
mentally. It contains three arrays: a two-dimensional array of integers and two
one-dimensional arrays of averages , one for rows and one for columns.
When we rk with a lurge program with many different data structures,
in this case three
see how the *
strut ir TT
,
to raw a picture of the arrays . We can then
.^ ^ t 0 SO ve
the arra> structures and the r relationships is
problem. A picture of
shown in Figure 8- 42
T
i \
rowAve
columnAve
FIGURE 8 - 42 Data Structures For Calculate Row-Column Averages
The program begins by requesting the user provide data for a two-
dimensional array. Once the array has been filled , the program calculates the
average of each row and places it in a parallel array of row averages. It then
calculates the average lor each column and places it in an array of column
averages. Although we have represented the column -average array horizon -
tally and the row average array vertically, they are both one-dimensional
-
arrays .
When all the calculations are complete, the program calls a I unction to
print the array with the row averages at the end of each row and the column
averages at the bottom of each column. The structure chart for the program
is seen in Figure 8 - 43.
calculate
Averages
row column print

getData Average Average Tables
FIGURE 8 - 43 Calculate Row-Column Average Design
First Increment: main

main and its First sub-
In our previous examples, the hrst increment included
, limit it to just main in this
function. Because our data are more complex we
statements we
contains the global include
example. The First increment main we define our
the known data delinitions . In
anticipate needing and the program
. We then compile
local structures , including the three arrays
and correct any coding errors found by the compiler. Note that we don’t run
the program because it produces no output at this stage. Program 8- 18 con-
tains the code.
PROGRAM 8 - 18 Calculate Row and Column Averages: main

1 /* Read values from keyboard into a two dimensional -
2 - averages.
array. Create two one dimensional arrays that
3 contain row and column
4 Written by:
5 Date:
6 */
8
9 # define MAX ROWS 5_
10 #define MAX COLS 6
11
12 int main (void )
13 {
15 int table [ MAX ROWS][ MAX COLS ];
16
17 float rowAve _
[ MAX ROWS] = {0};
18 _
float columnAve [ MAX COLS ] = {0};
19
20 // Statements
21 return 0;
22 > // main
Second Increment: Get Data

I laving verified that mum is error free > we are ready to write get data . At this
point we add the necessary function declaration and a print loop in main to
display the table. I he code is shown in Program 8 - 19.
PROGRAM 8 - 1 9 Calculate Row and Column Averages: Get Data

1 / * Read values from
2 array. Create two one dimensional arrays that
-
keyboard into a two dimensional
3
-
contain row and column averages.
4 Written by:
5 Date:
6 */
8
9 #define MAX ROWS 5
continue
ft]if
PROGRAM 8 - 19 Calculate Row and Column Averages: Get Data (continued )

11
12
13
void getData ( int _
table[][ MAX COLS ]);
14
15 int main ( void)
16 {
17
18
int table _ _
[ MAX ROWS ][ MAX COLS ];
19
20 float rowAve [MAX_ROWS] = {0};
21 float columnAve ( MAX COLS ] = {0};
22
23 // Statements
24 getData ( table);
25
26 printf( " \n**Tables built\n " );
27 _
for ( int i = 0; i < MAX ROWS; i++ )
28 {
29 for ( int y _
= 0; y < MAX COLS; y++)
30 printf(" %4d" , table[i](y]);
31 printf( "\n")?
32 } // for i
33
34 return 0 ;
35 } // main
36
37 /* == ===== getData ==
38 Reads data and fills two dimensional array.
-
39 Pre table is empty array to be filled
40 Post array filled
41 */
42 void getData ( int table[ ][ MAX COLS ])
_
43 {
44 // Statements
45 for (int row = 0; row
_
< MAX ROWS ; row++ )
46 for ( int col =
_
0; col < MAX COLS ; col++ )
47 {
printf( "\nEnter integer and <return : ")?
>
48
scanf( "%d " , &table[ row ][ col ]);
49
50 } // for col
51 return ;
52 } // getData
53 ==== End of Program ====
continuec
PROGRAM 8 - 19 Calculate Row and Column Averages: Get Data (continued )

Results:
**Tables built
10 12 14 16 18 20
22 24 26 28 30 23
25 27 29 31 33 35
39 41 43 45 47 49
51 53 55 57 59 61
Program 8 - 1 9 Analysis After running the program we carefully examine the results to make sure that the new
function worked correctly. Because we created our test data before we ran the pro-
gram, we knew what the results should be. A simple comparison of the test data and
the results verifies that the table was built correctly.
Third Increment: Calculate Row Averages

We are now ready to write the code lor the averages. We begin with the calcu-
lation for the rows. This requires nested for loops, the outer loop steps
through the rows and the inner loop steps through the columns within a row.
PROGRAM 8 - 20 Calculate Row and Column Averages: Row Averages

1 /* Read values from keyboard into a two-dimensional
2 array. Create two one dimensional arrays that
-
3 contain row and column averages.
4 Written by:
5 Date:
6 */
8
9 _
# define MAX ROWS 5
10 # define MAX COLS 6
11
13 void getData ( int _
table[ ][ MAX COLS ]);
14 void rowAverage ( int _
table[ ][ MAX COLS ],
15 float rowAvrg []);
16
18 {
20 int table _ _
[ MAX ROWS ][ MAX COLS];
21
22 float rowAve [ MAX_ROWS ] = {0 > ;
continued
PROGRAM 8- 20 Calculate Row and Column Averages: Row Averages (continued)

23 _
float columnAve [MAX COLS] = {0};
24
25 // Statements
26 getData (table);
27 rowAverage ( table, rowAve );
28
printf( " \n**Tables built\n " );
29
30 _
for ( int i = 0; i < MAX ROWS; i++ )
31
32
{
_
for ( int y = 0; y < MAX COLS; y++ )
33 printf( ,,%4d" , table( i][y]);
34 printf( " \n " );
35 > // for i
36 printf(" W );
37
printf( " **Row averages\n ");
38
39
_
for ( int i = 0; i < MAX ROWS; i++)
40 printf("%6.If " , rowAve[i]);
41 printf ( " \nM );
42
43 return 0 ;
44 > // main
45
46 /*
47 Reads data and fills two dimensional array.
-
48 Pre table is empty array to be filled
50 */
51 void getData (int table[ ][ MAX COLS])
_
52 {
53 // Statements
54
_
for (int row = 0 ; row < MAX ROWS ; row++ )
55
_
for ( int col = 0; col < MAX COLS ; col++)
56 {
printf( "\nEnter integer and <return>: );
"
57
58 scanf( " %d " , &table[ row][col ]);
59 } // for col
60 return;
61 } // getData
62
63 /* === == ===== rowAverage =====
This function calculates the row averages for a table
64
65 Pre table has been filled with values
66 Post averages calculated and in average array
continuec
PROGRAM 8 - 20 Calculate Row and Column Averages: Row Averages ( continued )

67
68
*/
void rowAverage ( int
_
table[ ][ MAX COLS],
69 float rowAvrg [ ])
70 {
71 // Statements
72 for ( int row = 0; _
row < MAX ROWS ; row++ )
73 {
74 _
for (int col = 0; col < MAX COLS; col++)
75 rowAvrg[row ] += table [ row ][col];
76 rowAvrg [ row ] /= MAX COLS ;
77 } // for row
78 return ;
79 > // rowAverage
80 / / = = = == ===== End of Program =====
Results:
**Tables built
10 12 14 16 18 20
22 24 26 28 30 23
25 27 29 31 33 35
39 41 43 45 47 49
51 53 55 57 59 61
**Row averages
15.0 25.5 30.0 44.0 56.0
Fourth Increment: Calculate Column Averages

I he calculation ol the column averages is very similar to the calculation of
the row averages. We leave its incremental development for a problem at the
end of the chapter.
Fifth Increment: Print Tables

1 he final increment completes the program by printing the three tables. Note
that we present the results with the row averages at the end ol the rows anti
the column averages at the bottom of the columns. This is a natural and logi-
cal presentation from a user’s perspective. The code for the fifth increment is
shown in Program 8- 21 . Note that the column totals will he displayed as zero
until the fourth increment is completed .
PROGRAM 8 - 21 Calculate Row and Column Averages: Print

Tables
1 / * Read values from keyboard into a two-dimensional
2 array. Create two one-dimensional arrays that
continue
'
1 7
1
PROGRAM 8 -21 Calculate Row and Column Averages: Print Tables (continued )
3 contain row and column averages.
4 Written by:
5 Date:
6 */
8
9 _
# define MAX ROWS 5
11
13 void getData ( int _
table[ ][ MAX COLS ]);
15 float rowAvrg [ ]);
16 void colAverage ( int _
table[][ MAX COLS ],
17 float colAvrg [ ]);
18 void printTables ( int _
19 float rowAvrg[],
20 float colAvrg[ ]);
21
23 {
25 int _ _
table [ MAX ROWS ][ MAX COLS];
26
27 float rowAve [ MAX_ROWS] = {0};
28 float colAve [ MAX_COLS ] = {0 > ;
29
30 // Statements
31 getData ( table);
32 rowAverage (table, rowAve);
33 // colAverage ( table , colAve );
34
35 printf( " \n " );
36 printTables ( table, rowAve , colAve);
37 return 0;
38 > // main
39
40 ===== getData ====
- array.
41 Reads data and fills two dimensional
table is empty array to be filled
42 Pre
*/
44
45 void getData (int tablet ][ MAX
_
COLS])
46 {
continued
PROGRAM 8 - 21 Calculate Row and Column Averages: Print Tables ( continued )

47 / / Statements
48 _
for (int row = 0; row < MAX ROWS ; row++)
49 for (int col = 0 ; col < MAX_COLS ; col++)
50 {
51 printf("\nEnter integer and <return>: " );
52 scanf("%d " , stable[row][col ]);
53 > // for col
54 return ;
55 > // getData
56
57 =====
: rowAverage ===
58 This function calculates the row averages for a table
59 Pre table has been filled with values
60 Post averages calculated and in average array
61 */
63 float rowAvrg [ ])
64 {
65 // Statements
66 for ( int row = 0; row < MAX ROWS; row++)
67 {
68 _
for ( int col = 0 ; col < MAX COLS ; col++)
69 rowAvrg[ row] += table ( row][ col ];
70 rowAvrg [ row ] /= MAX COLS ;
71 } // for row
72 return ;
73 } // rowAverage
74
75 /*
76 Print data table, with row average at end of each
77 row and average of columns below each column ,
78 Pre each table filled with its data
79 Post tables printed
80 */
81 void printTables ( int _
82 float rowAvrg[],
83 float colAvrg[ ])
84 {
85 // Statements
86 for ( int row = 0; _
row < MAX ROWS; row++ )
87 {
88 _
for ( int col = 0 ; col < MAX COLS ; col++)
89 printf( " %6d " , table[ row ][col]);
90 printf( " I %6.2f\n " , rowAvrg [row]);
continue
PROGRAM 8 - 21 Calculate Row and Column Averages: Print Tables (continued )

91 } // for row
92
93 printf( " \n" );
94
95
printf(" ")?
_
for ( int col = 0 ; col < MAX COLS; col++)
96 printf( " %6.2f " , colAvrg[col ]);
97 return;
98 > // printTables
99 // ======== ===== End of Program =====
Results:
10 12 14 16 18 20 15.00
22 24 26 28 30 23 25.50
25 27 29 31 33 35 30.00
39 41 43 45 47 49 44.00
51 53 55 57 59 61 56.00
0.00 0.00 0.00 0.00 0.00 0.00


In this section , we discuss two basic concepts: testing and algorithm effi -
ciency. To be effective , testing must he clearly thought out . We provide some
concepts for testing array algorithms by studying sorting and searching. We
then continue the algorithm efficiency discussion started in Chapter 6 by :
studying sort and search algorithms as case studies.
Testing Sorts
As our programs become more complex, we need to spend more time creating
test data that will completely validate them . In this section , we examine some
techniques for testing sorts.
In general , we should conduct four tests: ( 1 ) sort a list of random values,
( 2 ) sort a list that is already in sequence, (3) sort a list that is in reverse order,
( 4 ) sort a nearly ordered list , such as one in which every tenth item is one
position out of sequence . Table 8- 2 contains a summary of the tests that we
should conduct and some sample test data to show the points.
Test Case Sample Data

Random data 5 23 7 78 22 6 19 33 51 11 93 31
Nearly ordered 5 6 7 21 19 22 23 31 29 33 51 93
Ordered — ascending 5 6 7 11 19 22 23 31 33 51 78 93
Ordered — descending 93 78 51 33 31 23 22 19 11 7 6 5
TABLE 8 - 2 Recommended Sort Test Cases
Testing Searches
When we test the sequential search , we need only four tests. Three deal with
—
finding an element in the table find the first , last , and any element in the
middle. The last case deals with trying to find an element that is not in the
list ; look for any value that is not in the middle of the list. Whenever testing
an array, always he sure that the test data access the first and last elements of
the array. Many array processing errors occur at these locations.
When testing an array search, always access the first and last elements .
I he binary search requires the lour tests discussed
above, plus three
more. Since it uses an ordered list , we should try to find a target lower than
the first element in the list and another target greater than the last element.
Finally, we should find two elements that are in adjacent array locations,
such as l i s t [ 0 ] a n d l i s t [ l ] . The reason for this
test is that the binary
search includes logic that divides by two . An error
could result in being able
\ / \ ml
.
to find only even - or only odd - numbered elements Testing for adjacent loca -
tions ensures that such an error won ’t happen . These test cases are seen in
Table 8- 3.
Expected Results
found
Index
0
B
target = = list[0] all
Search
found 1 target = = list[ l ] binary only

found n- 1 target = = list[ n— 1 ] all
found 0<i<n target = = listfi] all
not found 0 target < list [0] binary only
not found n- 1 target > list[n] binary only
not found 0<i<n target ! = Iist[i] all
TABLE 8 - 3 Test cases for searches
Analyzing Sort Algorithms

VVe have developed three sort algorithms. We examine each of them in this
section .
Bubble Sort Analysis

The bubble sort essentially contains the block of code shown below. Note
that “exchange” in the code could be a function call or the three statements
shown in Program 8- 10.
for(current = 0 ; current < last; current++)

for ( walker = last ; walker > current ; walker--)
if ( list[ walker] < list[walker 1 ]) -
exchange (walker , walker 1); -
I
To determine the relative efficiency, we need to analyze the two for state-
in the sort array. It
mcnts. The first /or , the outer loop, examines each entry
—
will therefore loop n 1 times .
The inner loop starts at the end of the array and works its way
toward the
called , it
current node as established by the outer loop. The
first time it is
time , n - 2 elements , and so forth until
examines n - 1 elements ; the second
elements examined ,
it examines only one element. The average number of
therefore, is determined as shown below .
( n - l ) + ( n - 2 ) + . .. + 2 + 1 = n
This is the nested dependent loop. Simplifying the previous formula

we get
-
(41) ' !<"• - >
Discarding the coefficient and selecting the larger factor, we see that the
-
dominant factor in the bubble sort is n 2 t which in big O notation would be
stated as 0( w 2 ).
The efficiency of the bubble sort is 0( n2 ) .
Selection Sort Analysis

Now' lets examine the efficiency of the selection sort shown in Program 8 9. -
Again , its pivotal logic is essentially shown in the next example.
for (current = 0; current < last; current *- -*- ) -

{
for (walker = current + 1 ; walker <= last; walker++)
if ( list[walker] < list[smallest ])
smallest = walker;
> // for current
This algorithm bears a strong resemblance to the bubble sort algorithm
we discussed previously. Its first loop looks at every element in the array from
the first element (current = 0) to the one just before the last . The inner loop
moves from the current element, as determined by walker , to the end of the
list . I his is similar to the bubble sort , except that it works from the lower por-
tion of the array toward the end . Using the same analysis, we see that it will
test n( n - 1 ) / 2 elements, which means that the selection sort is also 0( »r ).
The efficiency of the selection sort is 0( n2 ) .

Insertion Sort Analysis
I he last sort we covered was the insertion sort , Program 8 1 1 . The nucleus of
its logic is shown as follows.
-
for ( current - 1 ; current <= last ; current++

)
for (walker = current 1?
walker >= 0 && !located ;)
if ( temp < list[walker ])
continued
{
list[walker + 1 ]
walker ;
> // if
—
= list[ walker];
1
Ilf
Does the pattern look familiar? It should . Again we have the same basic
nested for loop logic that we saw in the bubble sort and the selection sort .
I he outer loop is executed n times, and the inner loop is executed ( n 1 ) / 2 —
times , giving us 0( w 2 ).
The efficiency of the insertion sort is 0( nz ).

As we have demonstrated , all three of these sorts are 0(w 2 ) , which means
that they should be used only for small lists or lists that are nearly ordered .
You will eventually study sorts that are O( wlogn ) , which is much more effi -
cient for large lists.
Analyzing Search Algorithms

All of the sort algorithms involved nested loops. We now turn our attention to
two algorithms that have only one loop, the sequential search and the binary
search . Recall that a search is used when we need to find something in an
array or other list structure. The target is a value obtained from some external
source.
Sequential Search Analysis

The basic loop for the sequential search is shown below.
while (looker < last && target != list[looker])

looker++;
This is a classic example of a linear algorithm . In fact, in some ol the lit -

erature , this search is known as a linear search . Since the algorithm is linear,
its efficiency is O ( n ) .
The efficiency of the sequential search is 0( n).
Binary Search Analysis

The binary search locates an item by repeatedly dividing the list in half
. Its
loop is
while (first <= last )

{
mid = (first + last) / 2 ;
continued
if (target > list[mid ])

first = mid + 1 ;
else if (target < list[mid ])
last = mid 1;
else
first = last + 1;
> // while
This is obviously a loop that divides, and it is therefore a logarithmic
loop. This makes the efficiency O(logn), which you should recognize as one
of the more efficient of all the measures.
The efficiency of the binary search is O( logn) .
Comparing the sequential search and binary search, we see that, disre-
garding the time required to order the list , the binary' search is obviously bet -
ter for a list of any significant size ( see Table 8 - 4 ). We therefore recommend
the binary search for all but the smallest of lists—that is, lists with less than
50 elements.
Size Binary Sequential Sequential

( Average) ( Worst Case)
16 4 8 16
50 6 25 50
256 8 128 256
1 ,000 10 500 1 ,000
10,000 14 5,000 10,000
100,000 17 50,000 100,000
1 ,000,000 20 500,000 1 ,000,000
TABLE 8 4 Comparison of
*
binary and sequential searches

I he big- O concept isgenerally interested only in the largest factor. This
tends tosignificantly distort the efficiency of the sequential sort, in that it is
always the worst case. II the search is always successful, it turns out that
the efficiency of the sequential search is 1/2«. We include the average in
1 rogram 8 - 3 for comparison. (The average for the binary search is only one
less than the maximum, so it is less interesting.)
l
<
m
Chapter 8 Arrays 53 /

1 . In an array declared as array [ n ] , the index goes from 0 ( not 1 ) to n — 1.
2. I hree things are needed to declare and define an array: its name, type,
and size.
‘a
3. I he elements of arrays are not initialized automatically. You must initial -
ize them if you want them to start with known values.
4. Id initialize all elements in an array to zero, all you need to do is initialize
the first element to zero.
5 . To exchange the value of two elements in an array, you need a temporary
variable.
6. You cannot copy all elements of one array into another with an assign -
ment statement. You need to use a loop.
7. To pass the whole array to a function, you only use the name of the array
as an actual parameter.
8. The most common logic error associated with arrays is an invalid index.
An invalid index used with an assignment operator either causes the pro-
gram to fail immediately or destroys data or code in another part of the
program and causes it to fail later.
9. Invalid indexes are often created by invalid coding in a for statement. For
example, given an array of ten elements, the following /or statement logic
error results in an index value of ten being used . Although it loops 10
times, the indexes are 1 through 10, not zero through 9 .
for (i = 1; i <= 10; i++ )
.
10. Another cause of invalid indexes is an uninitialized index. Make sure
your indexes are always properly initialized.
1 1 When initializing an array when it is defined , it is a compile error to pro-
.
vide more initializers than there are elements.
12 . It is a compile error to leave out the index operators in an assignment
statement, as shown below.
float costAry[20];
•••
costAry = quantity * price;
i a scanj
13. It is most likely a logic error to leave out the index
operators in
statement , as shown in the following example . In this case , costAry is
array and the input will be placed in the first element
the address of the
code it with
of the array. Even when this is the desired result you should
,
the index operators as shown.
float costAry[20];
•••
scanf("%f", costAry ); // Poor Style
scanf( " %f " , ScostAry[ 0 ]); // Clear code
1 4. It is a compile error to omit the array size in the parameter declaration

for any array dimension other than the first .
J J
-
1 5. To pass a variable size array to a function , you need either to include the
variable defining the array size or use an asterisk .
8.12 Key Terms

array passing a row
binary search pass the whole array
bubble sort pass individual elements
colu mn plane
exchanging reusable code
-
fixed length array row
frequency array searching
histogram selection sort
index sequential search
invalid index sort pass
insertion sort subscript
linear search table
multidimensional array two-dimensional array
one-dimensional array variable- length array
8.13 Summary
-
A one dimensional array is a fixed sequence of elements of the same type.
We use indexes in C to show the position of the elements in an array.
J An array must be declared and defined before being used. Declaration and
definition tell the compiler the name of the array, the type of each ele-
ment , and the size of the array.
Initialization of all elements of a fixed length array can be done at the time
of the declaration and definition.
LI II a one - dimensional array is completely initialized when it is declared , it is
not necessary to specify the size , hut it is recommended
.
When an a i i a y is partially initialized , the rest of the elements are set
to zero.
1
iff
We can (ill the elements of an array by using a loop to read the values from
the keyboard or a file.
We can access the individual elements of an array using the array name
and the index.
We can read or write the values of an array using a loop.
ti
An array reference is a postfix expression with the opening and closing

brackets as operators.
C does not do boundary checking on the elements of an array.
We can pass an individual element of an array to a function either by value
or by address.
We can also pass the whole array to a function. In this case, only the
address ol the array will he passed. When this happens, the function has
access to all elements in the array.
A irequency array is an array whose elements show the number of occur-
rences ol data values in another array.
A histogram is a pictorial representation of a frequency array.

A two -dimensional array is a representation of a table with rows and
columns.
We can pass either a single element, a row, or the whole two-dimensional
array to a function.
A multidimensional array is an extension of a two-dimensional array to
three, four, or more dimensions.
An array can be sorted using a sorting algorithm.
The selection sort divides the array into sorted and unsorted sublists. In
each pass, the algorithm chooses the smallest element from the unsorted
sublist and swaps it with the element at the beginning of the unsorted
sublist.
The bubble sort divides the array into sorted and unsorted sublists. In each
pass, the algorithm bubbles the smallest element from the
unsorted list
into the sorted sublist .
The insertion sort divides the array into sorted and unsorted suhlists. In
list
each pass, the algorithm inserts the first element from the unsorted
into the appropriate place in the sorted sublist.
a list of
Searching is the process of finding the location of a target among
objects.
searches
The set juential search starts at the beginning of the array and
the data hits the end of the list . The data may he ordered or
until it finds or
unordered.
A binary search is a much faster searching algorithm. In the binary search ,

each test removes half of the list from further analysis. The data must be
ordered.
-
To pass a whole variable length array to a function , you need to also pass
its size .
8.14 Practice Sets

Review Questions
1 . The type of all elements in an array must he the same .
a . True
b. False
2. Any expression that evaluates to an integral value may he used as an index.
a . True
h. False
3. When an array is defined , C automatically sets the value of its elements
to zero.
a. True
.
b False
4 . When an array is passed to a function , C passes the value for each clement.
a. True
.
h False
5 . Because ol its efficiency, the binary search is the best search for any
array, regardless of its size and order.
. True
a
h . False
6 . I he selection , insertion , and bubble sort are all O ( n ) sorts .
^
.
a True
b. False
.
7 A( n ) is an integral value used to access an element
in an array.
.
a constant
h. element
c. index
d . number
e . variable
rr
8.
Which of the following array initialization statements is valid ?

^
a. int ary{ > = {1, 2 , 3, 4> ;
b. int ary[ ] = [1 , 2, 3, 4];
c. int ary[ ] = {1, 2 , 3, 4>;
d . int ary{4 > = [ 1, 2, 3, 4]?
e. int ary[ 4 ] = [ 1 , 2, 3, 4];
9. Which ol the lollowing statements assigns the value stored in x to the
first element on an array, ary?
a. ary = x;
b. ary = *[ 0 ] ;
.
c ary = x[ 1 ] ;
d . ary [ 0 ] = x;
e. ary[ 1 ] = x;
10 . Which of the following statements concerning passing array elements
is true?
a . Arrays cannot be passed to functions, because their structure is too
complex.
b . It is not possible to pass just a row of a two-dimensional array to a
function .
c . Only the size of the first dimension is needed when a two-dimensional
array is declared in a parameter list .
d . When an array is passed to a function , it is always passed by reference
( only its address is passed ) .
e . When a two-dimensional array is passed to a function , the size ol the
second dimension must be passed as a value parameter.
I I . The process through which data are arranged according to their values is
known as
a. arranging
b . listing
c . parsing
d . searching
e . sorting
12 . The finds the smallest element from the
sort
unsorted sublist and swaps it with the element at the beginning of the
unsorted data .
a . bubble
b . exchange
c . insertion
d . quick
e . selection
.
13 The search locates the target item by starting at the
beginning and moving toward the end of the list .
a. ascending
b. binary
c. bubble
d. selection
e. sequential
14. Which of the following statements about a sequential search is false?
a. Any array can be searched using the sequential search .
b. If the target is not found , every element in the list is tested .
c. The efficiency of the sequential search is O ( n ) .
d. The list must be ordered.
e. The sequential search is generally recommended only for small lists.
15. Which of the following statements about two-dimensional arrays is true?
a. A two - dimensional array can he thought of as an array of one-
dimensional arrays.
b. Only the size ol the second dimension needs to be declared when the
array is used as a parameter.
c. I wo different types can be stored in a two-dimensional array.
d. I he first dimension is known as the column dimension .
e. \\ hen passed to a function , the size of the second dimension must be
passed as a value parameter.
Exercises
16. What would be printed by the following program ?
#include <stdio.h>
int main (void )
{
int list [10] = {0};
// Statements
for (int i = 0; i < 5; i++ )
list [2 * i + l ] = i + 2 ;
for (int i = 0; i < 10 ; i++ )
printf( " %d\n " , list [i]);
return 0 ;
} // main
17. What would be printed by the following program ?

int main (void )
{
continued
V' 11 'iT,| ft1
iTy 1
int list [ 10 ] = {2, 1, 2, 1 , 1, 2, 3, 2 , 1, 2 > ;
/ / Statements
printf( " %d \ n ", list [ 2 ] );
printf( "% d \ n ", list [ list [ 2 ] ] ); M fl ,
printf( " % d \n ”, list [ list [ 2 ] + list [ 3 ] ] );
printf( " %d \ n", list [ list [ list [ 2 ] ]] );
return 0;
> / / main
18. \ \ hat would be printed by the following program ?

int main ( void )
{
int list [ 10 ] = { 2 , 1, 2 , 4, 1, 2 , 0, 2 , 1, 2};
int line [ 10 ];
// Statements
for ( int i = 0 ; i < 10; i + + )
line [ i ] = list [ 9 - i ];
for ( int i = 0; i < 10; i++ )
printf( " % d % d \ n " , list [ i ], line [ i ] );
return 0;
} / / main
19. An array contains the elements shown below. The first two elements
have been sorted using a selection sort . What would be the value of
the elements in the array after three more passes of the selection sort
algorithm ?
7 8 26 44 13 23 98 57
20 An array contains the elements shown below. The first two elements have
.
been sorted using a bubble sort . What would be the value of the elements
in the array after three more passes of the bubble sort algorithm ? Use the
version of bubble sort that starts from the end and bubbles the smallest
element.
7 8 26 44 13 23 57 98
21 . An array contains the elements shown below. The

first two elements
sort . What would be the value of
have been sorted using an insertion
the elements in the array after three more passes of the insertion sort
algorithm ?
22. We have the following array:
47 3 21 32 56 92
After two passes of a sorting algorithm , the array has been rearranged as
shown below.
3 21 47 32 56 92
Which sorting algorithm is being used (selection , bubble , insertion )?

Defend vour answer.
j
80 72 66 44 21 33
After two passes of a sorting algorithm , the array has been rearranged
as shown below .
21 33 80 72 66 44
Which sorting algorithm is being used ( selection , bubble , insertion )?

Defend your answer.
47 3 66 32 56 92
After two passes of a sorting algorithm , the array has been rearranged
as shown below.
3 47 66 32 56 92
\ \ hich sorting algorithm is being used ( selection , bubble , insertion ) ?

Defend your answer.
25. An array contains the elements shown below. Using the binary search
algorithm , trace the steps followed to find 88. At each loop iteration ,
including the last , show the contents of first , last , and mid .
8 13 17 26 44 56 88 97
26. An array contains the elements shown below.

Using the binary search
algorithm , trace the steps followed to find 20. At each loop iteration ,
inc hiding the last , show the contents of first , last ,
and mid .
8 13 17 26 44 56 88 97
m1
2 / . Both the selection and bubble sorts exchange elements. The insertion
sort does not . Explain how the insertion sort rearranges the data without
exchanges.
Problems
28. We have two arrays A and B, each of 10 integers. Write a function that
tests if every element of array A is equal to its corresponding element in
array B. In other words, the function must check if A[0 ] is equal to B[ 0],
A[1 ] is equal to B[ 1 ] , and so forth. The function is to return true if all
elements are equal and false if at least one element is not equal.
29. Write a function that reverses the elements of an array so that the last
element becomes the first , the second from the last becomes the second ,
—
and so forth . The function is to reverse the elements in place that is ,
without using another array. ( It is permissible to use a variable to hold an
element temporarily.) Then write a test driver to test your function . Test
it twice , once with an even number of elements in the array and once
with an odd number of elements in the array.
.
30 The Pascal triangle can be used to compute the coefficients of the terms in
the expansion of ( a + b ) n. Write a function that creates a two-dimensional
matrix representing the Pascal triangle. In a Pascal triangle , each element is
the sum of the element directly above it and the element to the left ol the
element directly above it ( if any). A Pascal triangle ol size 7 is shown below.
1
1 1
1 2 1
1 3 3 1
1 4 6 4 1
1 5 10 10 5 1
6 15 20 15 6 1
1
In the above example , the first element ol row[0] and the first two -
ele
[ ] are set to 1 . Then each of the following rows up to the
ments of row l
maximum size , are set with a loop as
shown in the following pseudocode:
Algorithm PascalTriangle
1 pascal[0][0] = 1
2 pascal[l ][0] = 1
3 pascal[ l ][ l ] = 1
4 prevRow = 1
5 currRow = 2
6 loop ( currRow <= size)
1 pascal[ row ][0 ] = 1
2 col = 1
continued
3 loop (col <= currRow)

1 pascal[ row][ col] =
pascal[row -
l][col 1] -
+ pascal[ row -
l ][col]
2 col = col + 1
4 end loop
7 end loop
end PascalTriangle
Your program must IK* able to create the triangle of any size .
.
31 An international standard book number (ISBN) is used to uniquely iden-
tify a book. It is made of 10 digits, as shown in Figure 8 - 44. Write a func -
.
tion that tests an ISBN to see if it is valid For an ISBN number to be
valid, the weighted sum of the 10 digits must be evenly divisible by 11.
The tenth digit may hex, which indicates 10.
Publisher Check digit
0-07-881809-5
Country Book Number
FIGURE 8 - 44 ISBN for Problem 31
Fo determine the weighted sum, the value of each position is multi-

plied by its relative position, starting from the right, and the sum of the
products is determined. I he calculation of the weighted sum lor the
ISBN shown above is demonstrated in Table 8 - 5.
Code Weight Weighted Value

0 10 0
0 9 0
7 8 56
8 7 56
8 6 48
1 5 5
continued
TABLE 8 - 5 Demonstration of ISBN Calculatiion
I
', ;
n
Code Weight Weighted Value
8 4 32
0 3 0
9 2 18
5 1 5
Weighted Sum 220
TABLE 8 - 5 Demonstration of ISBN Calculation ( continued )

Since the weighted sum modulus 1 1 is 0, the ISBN number is valid.
Test your function with the above example, the ISBN number for this
text ( see the copyright page), and 0-08- 781809- 5 (an invalid ISBN—the
third and fourth digits are reversed).
.
32 Another technique to validate the ISBN in Problem 3 1 is to calculate the
.
sum of the sums modulus 1 1 The sum of the sums is calculated by add -
ing each digit to the sum of the previous digits, as shown in Table 8 -6.
Code Sum of Digits Sum of Sums
0 0 0
0 0 0
7 7 7
8 15 22
8 23 45
1 24 69
8 32 101
0 32 133
9 41 174
46 220
5
TABLE 8 - 6 ISBN Calculation: Sum of Sums Method

sums method.
Rewrite the function to use the sum-of- the-
33 Write a function that copies a one-dimensiona
. l array ol n elements into a
two- dimensional array ol k rows
.
and j columns I he rows and columns
factor of the number of elements in the one-dimensional
must he a valid
array; that is, k * j = n .
.
34 Write a program that creates an array of 100 random integers in the
range 1 to 200 ( see Chapter 4 for a discussion of random numbers ) and
then , using the sequential search , searches the array 100 times using
randomly generated targets in the same range. At the end of the program,
display the following statistics:
.
a The number of searches completed .
b. The number of successful searches.
.
c The percentage of successful searches.
d . The average number of tests per search .
To determine the average tests per search , you will need to count the
number of tests for each search . 1 lint: Use the comma operator to count
the compares.
35. Repeat Problem 34 using the binary search.
36. The sequential search assumes that a list is unordered . 11 it is used when
the list is in fact ordered , the search can he terminated with the target
not found whenever the target is less than the current element . Modify
Program 8- 13 to incorporate this logic .
37. Repeat Problem 34 with the modified search you created in Problem 36.
38. I he binary search is a good algorithm to implement using recursion.
Rewrite the binary search function using recursion . Then repeat
Problem 35 using the recursive function.
39. Modify the bubble sort to stop as soon as the list is sorted . ( See discus-
sion in the analysis of Program 8- 10 . )
40. Modify the selection sort function to count the number of exchanges
needed to order an array ot 50 random numbers (see Chapter 4 for a dis-
cussion of random numbers). Display the array before and after the sort .
At the end of the program , display the total exchanges needed to sort
the array.
41 . Repeat Problem 40 using the bubble sort .
42. Program 8- builds a frequency array and its histogram. If there is an
/
invalid data point in the input, it is displayed and ignored. Modify' the
program to make the frequency array one element larger than the data
range, and use the last element of the array as a count of numbers not in
the specified range.
.
43 Write a program that fills the right - to left diagonal of a square matrix
-
-
with zeros, the lower right triangle with Is, and the upper- left triangle
—
with + Is. The output ol the program , assuming a six by-six matrix, is
shown in Figure 8- 45.
-
FIGURE 8 - 45 Matrix for Problem 43
44. Given an array of 100 random numbers in the range 1 ...999, write a
function for each of the following processes. In building the array, il the
random number is evenly divided by 3 or 7 , store it as a negative number.
a . Print the array ten values to a line . Make sure that the values are
aligned in rows.
b. Print the odd values, ten to a line .
c. Print the values at the odd numbered index locations, ten to a line .
d . Return a count of the number of even values.
e. Return the sum of all values in the array.
f . Return the location of the smallest value in the array.
g. Return the location of the largest value in the array.
h . Copy all positive values to a new array. Then use process "a " above to
print the new array.
i. Copy all negative values to a new array. Then use process “ a ” above to
print the new array.
45. Modify the sort of your choice to sort two parallel arrays. Parallel arrays
are two or more arrays that contain related data . For example, one array
might contain identification numbers, and a related array might contain
corresponding scores in a tournament . The sort will sort only the array
containing the identifiers ; however, as each element in the sorted array is
moved, its corresponding element in the data array must also he moved .
After the arrays have been sorted , print out the data as shown below.
ID Score
1 72
2 89
3 98
4 105
—
Write a test driver to test your sort . Use the following data the first value
is an identifier and the second value is a score: { 18, 90}, { 237 , 47}, {35, 105},
}.
{ 5 , 25}, {76, 739}, { 103, 26}, { 189 , 38}, {22 , 110}, {156, 31}, {49 , 245
.
46 Write the incremental implementation to calculate the column averages
—
for Section 8.9, “ Programming Example Calculate Averages.” Before
testing the column average increment , determine the averages so that
you will be able to verify that the correct values were calculated.
Projects
47. Write a C program that simulates a guessing game. Each turn , you
choose among nine possible guesses. As many as five guesses may be
made in a turn . For each turn , the program will generate a random num -
ber between 1 and 36. Each correct guess will be rewarded with points
based on how many of your current points you risked .
A game board divides the numbers into rows and columns, as shown
in Figure 8- 46. This board provides the basis for your guesses.
01 02 03
04 05 06
LOW
07 08 09
10 11 12
13 14 15
16 17 18
MEDIUM
19 20 21
22 23 24
25 26 27
28 29 30
HIGH
31 32 33
34 35 36
LEFT CENTER RIGHT

FIGURE 8 - 46 The Guessing Game Board
You can guess whether the random number is even or odd. In this
case, you get I point for each point risked when you guess correctly. You
can guess whether the number is low ( 1 -12 ) , medium ( 1 3-24 ) , or high
( 23-36 ) . In this
case, you will get 2 points for each point risked . You can
TT
-
also guess left , center, or right , as shown in Figure 8 46. In this case, you
get 2 points lor each point risked when your guess is correct. Finally, you
can guess a specific number between 0 and 36. In ibis case you get
36 points for each point risked when your guess is correct .
Io make the game more interesting, each round allows up to five 4
guesses. None ol the five may he correct , or any number up to all five
may he correct. The program stops when the player quits or when the
player is out of points.
The program first asks the number of points the user wants to start
with , as shown below.
How many points would you like? 2000
The program then prints the guess menu and allows up to five
guesses , as shown in the following example.
Guesses Choices
0 Odd
E Even
L Low
M Med
H High
F Left
C Center
R Right
N Number
How many guesses would you like? 5

Guess 1
Enter your choice? L
Points at risk? 20
Guess 2 :
Enter your choice? H
Points at risk? 15
Guess 3 :
Enter your choice? N
Enter your number: 18
Points at risk? 20
Guess 4 :
Enter your choice? 0
Points at risk? 120
Guess 5 :
Enter your choice? L
Points at risk? 0
After all guesses have been made, the program generates the random
number and displays the following message:
My number is: 31
The program then prints the situation of the player:
Previous Points: 2000

Guess Type Number Amount Won/Loss
1 L 20 -20
2 H 15 +30
3 N 18 20 -2 0
4 O 120 +120
5 N 20 10 -10
You won 100 points in this turn.
Your new balance is : 2100 points
Do you want to play again (Y or N ) ? Y
Some special rules: The minimum amount risked on a guess is 0.

The maximum is the players current balance . You need to verify that at
no time are the points risked more than the players current balance. .Any
combinations of guesses are allowed on a round as long as the total does
not exceed the player’s balance.
Some hints: You must use at least four arrays, each of five elements.
1 he arrays hold the guess information for the kind of guess , chosen num-
ber ( in case the player chooses a number ), amount of the guess , and points
won or lost .
Run your program twice , first with 2000 points and then with 500 points.
I ry different situations. Each run is to exercise each guess at least twice.
48. Write a program to keep records and perform statistical analysis lor a
class ol students. The class may have up to 40 students. There are five
quizzes during the term . Each student is identified by a four- digit student
number.
The program will print the student scores and calculate and print the
statistics lor each quiz. The output is in the same order as the input; no
sorting is needed . I he input will be read from a text file. The output from
the program should be similar to the following:
Student Quiz 1 Quiz 2 Quiz 3 Quiz 4 Quiz 5

1234 78 83 87 91 86
2134 67 77 84 82 79
3124 77 89 93 87 71
High Score 78 89 93 91 86
Low Score 67 77 84 82 71
Average 73.4 83.0 88.2 86.6 78.6
171
Use one - and two- dimensional arrays only. Test your program with
the quiz data in Table 8- 7 .
1234 52 7 100 78 34
2134 90 36 90 77 30
3124 100 45 20 90 70
4532 11 17 81 32 77
5678 20 12 45 78 34
6134 34 80 55 78 45
7874 60 100 56 78 78
8026 70 10 66 78 56
9893 34 9 77 78 20
1947 45 40 88 78 55
2877 55 50 99 78 80
3189 22 70 100 78 77
4602 89 50 91 78 60
5405 11 11 0 78 10
6999 0 98 89 78 20
TABLE 8 - 7 Test Data for Project 48

49. Rework Project 48, creating statistics for each student . Print the stu -
dents’ high , low, and average scores to the right of Quiz 5. Provide appro-
priate column headings.
.
50 Program 8- 7 builds a frequency array. In the discussion of the algorithm ,
.
we noted that a potential problem occurs if any of the data are invalid
Write a program that uses the random number generator, rand , to
generate 100 numbers between 1 and 22. The array is then to be passed
to a modified version of Program 8- 7 that will count the number
of val-
all num -
ues bctween 0 and 19 . Add a 21 st element in the array to count
bers not in the valid range (0 to 19 ) .
—
Print the input data in a 20 x 5 array that is, 20 numbers
in five
—
rows and then print the frequency diagram
hers as shown below.
with a heading of the nu ni -
— 0 1 - 2
—
- 3- 4- ... -18- -19- Invalid
51. Modify the program you wrote in Project 50 to include a histogram print-
out of the data. Its format should be similar to Figure 8- 15.
52. Using the data from Project 48, build a two-dimensional array of stu -
dents. Then write a search function that uses the sequential search to
find a student in the array and prints the students’ scores, average score,
and grade based on an absolute scale ( 90% is A, 80% is B, 70% is C, 60%
is D , less than 60% is F ). After each printout , give the user the opportu -
nity to continue or stop.
53. Modify the program you wrote in Project 52 to sort the two-dimensional
array of students. Then rewrite the search function to use a binary
search .
54. Write a program that sorts a 50-element array using the selection sort .
the bubble sort , and the insertion sort . Each sort is to be executed twice.
a . For the first sort , ( ill the array with random numbers between
1 and 1000.
b. For the second sort , ( ill the array with a nearly ordered list . Create the
nearly ordered list by filling the array with sequential numbers and
then subtracting 5 from every 10 th number in the array.
c. Each sort ( selection , bubble , insertion ) is to sort the same data . For
each sort , count the number of comparisons and moves necessary to
order this list .
d . After each sort execution , print the unsorted data followed by the sort
-
data in 5 - by 10 matrixes ( 5 rows of 10 numbers each ). After the sorted
data , print the number of comparisons and the number of moves
required to order the data. Provide appropriate headings for each
printout .
e. Io make sure your statistics are as accurate as possible , you must ana -
lyze each loop limit condition test and each selection statement in your
sort ( unctions. The best way to count them is with a comma expres -
sion , as shown below. Use similar code for the selection statements.
while (( count++, a ) && (count++ , b ))
( . Analyze the heuristics you generated , and write a few lines concerning
what you discovered about these sorts. Put your comments in a box
( asterisks ) alter your program
documentation at the beginning o! your
program.
55 . \ \ iite a program to compute the arithmetic
mean ( average ) , median , and
-
mock loi up to 50 test scores. I he data are contained in a text file. The
program w ill also print a histogram of the scores.
I he piogram should start with a function to
read the data file and ( ill
the array. Note that there may he fewer than
50 scores. This will require
that the read ( unction return the index for
the last element in the array.
Iff
Io determine the average, write a function similar to Program 8 - 4 .
To determine the median , you must first sort the array. The median is the
score in the middle ol the range . This can be determined by selecting the
score at last / 2 il last is an even index and by averaging the scores at
the floor and ceiling ol last / 2 if last is odd . The mode is the score that
occurs the most often . It can be determined as a byproduct of building
the histogram. After building the frequency array, use it to determine
which score occurred most often . ( Note that two scores can occur the
same number of times.)
56. Write a menu - driven program that allows the user to fill an array of
50 integers with random numbers in the range 1 . . .999, sort it , and then
search it to determine if a given random number was generated . The
menu is shown below.
MENU
Select one of the following options:
F. Fill array with a random number series
P. Print the array
S. Sort the array
Q. Query the array
Z. Terminate the program
Each time the fill option is executed , it will fill the array with a new
random number series.
You may use any of the sorts discussed in the text .
If the query locates the number, it will print a message that the num -
ber was located and the index location where it was found.
If the query does not locate the number, it will print a message that
the number was not in the list and then print the value and index loca -
tion of the largest number less than the target and the smallest value
greater than the requested number. Note that the first and
last ele-
ments will not have both a smaller and larger value.
If the array has been sorted , the query will use a binary search . If it is
not sorted , it will use the sequential search .
To test the program
a . Fill ( F ) the array.
b. Print ( P ) the unsorted array.
c. Search ( Q ) the first , last , and a middle element.
d. Refill ( F ) the array.
e. Sort ( S ) the array.
f. Print ( P ) the sorted array.
g - Search ( Q ) the first , last , and a middle element
.
for a number less than the first , greater than the last , and
h . Search Q ( )
not found in the middle ol the list .
11
Pointers
Every computer has addressable memory locations. In previous chapters, all
of our data manipulations, whether for inspection or alteration, used variable
names to locate the data. In other words, we assigned identifiers to data and
then manipulated their contents through the identifiers.
Pointers have many uses in C. Besides being a very efficient method of
accessing data, they provide efficient techniques for manipulating data in
arrays, they are used in functions as pass- bv- address parameters, and they are
the basis for dynamic allocations of memory.
As shown in Figure 9- 1 , pointers are the third of the derived types.
r Derived
Types i
Function
Type
Array
Type
Chapter 8
Pointer
Type
Structure
Type
Chapter 12
1 Union
Type
Chapter 12
Enumerated
Type
Chapter 12
i
Chapter 4
FIGURE 9- 1 Derived Types
Objectives
To understand the concept and use of pointers
To be able to declare, define, and initialize pointers
To write programs that access data through pointers
To use pointers as parameters and return types
To understand pointer compatibility, especially regardin
g pointers to pointers
To understand the role of quality in software engineer
ing
558 Section 9.1 Introduction
9.1 Introduction
A pointer is a constant or variable that contains an address that can be used
to access data. Pointers are built on the basic concept of pointer constants .
To understand and use pointers, we must first understand this concept .
Pointer Constants
First, let ’s compare character constants and pointer constants. We know that
we can have a character constant , such as any letter of the alphabet, that is
drawn from a universe of all characters. In most computers, this universe is
ASCII. A character constant is a value and can be stored in a variable.
Although the character constant is unnamed, the variable has a name that is
declared in the program. This concept is shown in Figure 9 - 2.
’\n’
address
value
character
constants ’ A’ 145600
’G’ aChar
’X’
variable
variable . name.
’c’
’k’
’x’
FIGURE 9 - 2 Character Constants and Variables
In Figure 9 - 2 we have a character variable, a C h a r . At this point, aChar

contains the value 'G ' that was drawn from the universe of character con-
stants. The variable a C h a r has an address as well as a name. The name is
created by the programmer; the address is the relative location of the variable
.
with respect to the program’s memory space Assume, for example, that a
computer has only I megabyte ol memory ( 220 bvtes ) . Assume also that the
computer has chosen the memory location 145600 as the byte to store this
.
variable I bis gives us the picture we see in Figure 9- 2 .
l ike character constants, pointer constants cannot be changed. In Fig-
ure 9- 4 we see that the address for our character
variable, a C h a r , was drawn
from the set ol pointer constants for our computer.
Although the addresses within a computer cannot change, be aware that
the address ol the variable, a C h a r , can and will
change from one run ol our
program to another. Ibis is because today’s modern
operating systems can
Put a program in memory wherever it is convenient when the program is
started. I hus, w hile a C h a r is stored at
memory location 145600 now, the
next time the program is run, it
could be located at 876050. It should be
m
obvious, therefore, that even though addresses are constant , we cannot know
w hat they will be . It is still necessary to refer to them symbolically.
000000
V
variable
name
address
character 145595
constants ' A'
•G
X*
aChar
a
- 145600 145600
145603
pointer
'c' constants
k' 1048575
'x '
Memory
FIGURE 9 - 3 Pointer Constants
Pointer constants, drawn from the set of addresses for a computer, exist by
themselves We cannot change them; we can only use them.
.
Pointer Values
I laving defined a pointer constant as an address in memory, we now turn our
attention to saving this address. If we have a pointer constant , we should be
able to save its value if we can somehow identify it . Not only is this possible ,
but we have been doing it since we wrote our first scanf statement with an
address operator.
The address operator ( & ) extracts the address for a variable . The result is
a unary expression , which is also known as an address expression
. The Prece-
dence Table ( see inside front cover ) lists the address operator in the unary
category, precedence IS.
in the unary expression

An address expression, one of the expression types
category, consists of an ampersand ( & ) and a variable name .
The address operator format is seen below.
&variable name
character variables and prints

Now let’s write a program that defines two
their addresses as pointers ( conversion code
% p ) . Depending on the operating
numbers each time we run it , as
system , this program may print different
previously explained . The addresses would also be dillerent in different com -

puters. However, most of the time , the computer allocates two adjacent
memory locations because we defined the two variables one after the other. If
you are at your computer, take a moment to code and run the program in
Figure 9- 4 to demonstrate the concept of address constants.
// Print character addesses

#include <stdio.h>
IV * 142300 b I I 142301
int main ( void )
{
char a ;
char b; KU 2300
// Statements 142301
printf ( " % p\ n % p\ n " , &a , & b );
return 0 ;
> // main
FIGURE 9- 4 Print Character Addresses
rhe situation changes slightly when we talk about integers. In most com-
puters , the int type occupy 4 bytes . This means that each int variable occu
pies four memory locations. Which of these memory locations is used to find
-
the address of the variable? In most computers, the location of the first byte
is used as the memory address . For characters , there is only l byte , so its
location is the address. For integers, the address is the first byte. This is
shown in Figure 9- 5.
I he same design applies to real and other data types. The address of a
variable is the address ol the first byte occupied by that variable.
A variable’s address is the first byte occupied by the variable .
variable
name integer
value
integer \ a
constants ) j -123 -123 234560
b 145 234564
145 integer addresses

variables
FIGURE 9 - 5 Integer Constants and Variables

m
Chapter 9 Pointers 561 1
Pointer Variables
If we have a pointer constant and a pointer value, then we can have a pointer
variable. I hus, we can store the address of a variable into another variable,
which is called a pointer variable. This concept is shown in Figure 9 - 6 .
address of
.variable a ^
a -123 1234560
& a stored pointer

in variable. variable
p 234560 |
Physical representation Logical representation
FIGURE 9 - 6 Pointer Variable
We must distinguish between a variable and its value. Figure 9 - 6 details

.
the differences In this figure we see a variable, a , with its value, — 123. The
variable a is found at location 234560 in memory. Although the variable s
’
name and location are constant , the value may change as the program exe -
cutes . This figure also has a pointer variable , p. The pointer has a name and a
location, both of which are constant. Its value at this point is the memory
location 234560. This means that p is pointing to a. In Figure 9 - 6, the physi-
cal representation shows how the data and pointer variables exist in memory.
The logical representation shows the relationship between them without the
physical details.
We can go even further and store a variable’s address in two or more dif
-
ferent pointer variables, as is shown in Figure 9 - 7. This figure has a variable ,

a, and two pointers, p and q. The pointers each have a name and a location ,
both of which are constant . Their value at this point is the memory location
234560. This means that both p and q are pointing to a.
There is no limit to
the number of pointer variables that can point to a variable.
value of p
( address of a)
&a
p | 234560 | ^ -123
234560
q 234560 |
FIGURE 9 -7 Multiple Pointers to a Variable

A final thought : If we have a pointer variable, but we don ’t want it t 0

point anywhere , what is its value? C provides a special null pointer constant,
.
NULL , in the standard input/output stdio h header file for this purpose.1
NULL is usually defined as a macro with a value of integer 0 or 0 cast to a
void pointer ( void * ) .
A pointer that points to no variable contains the special null-pointer con-

stant, NULL .
Accessing Variables Through Pointers

Now that we have a variable and a pointer to the variable, how can we relate
the two; that is , how can we use the pointer? Once again C has provided an
operator for us. Right below the address operator in the unary expressions
portion of the Precedence Table ( see inside front cover ) , we find the indirec-
tion operator ( * ). When we dereference a pointer, we are using its value to
reference ( address ) another variable. The indirection operator is a unary
operator whose operand must he a pointer value. The result is an expression
that can he used to access the pointed variable lor the purpose of inspection
or alteration . To access the variable a through the pointer p, we simply code
* p. The indirection operator is shown below.
*p
Let 's assume that we need to add 1 to the variable a . We can do this with
any ol the following statements, assuming that the pointer, p , is properly ini -
tialized ( p = & a ).
a ++ ; a = a + 1; *p = * p + 1; ( * p ) ++ ;
In the last example ,( * p ) ++ , we need the parentheses. The postfix incre-

ment has a precedence ol 16 in the Precedence Table while indirection ,
which is a unary operator, has a precedence of 1 5. The parentheses therefore
force the dereference to occur before the addition so that we add to the data
variable and not to the pointer. Without the parentheses, we would add to the
pointer first , which would change the address.
An indirect expression, one of the expression expression

types in the unary
category, is coded with an asterisk ( ) and an identifier
* .
f igure 9-8 expands the discussion. Let’s assume that the variable x is
pointed to by two pointers, p and q . Fhe expressions x , * p, * q all are expres-
sions that allow the variable to he either inspected or changed. When they are
1 . Technically, NULL is found in the stddefh library

. Traditionally, iit ma y also be defined in stdio.h -
- V1
I 1
used in the righthand side of the assignment operator, they can only inspect
( copy or compare ). When they are used in the left hand side of the assign -
-
ment operator, they change the value of x .
Before Statement After

P I ? I x = 4; P 4 I
q X q X
p
± ) x = x + 3; P 7 i
q X q X
P 7 i *p = 8; P 8 I
q x q x
P 8 I + * p; P i6 i
q x fmultiply N,
q X
Voperatory
P 16 I x = *p *q; P 256 \
q X q X
FIGURE 9 - 8 Accessing Variables Through Pointers
The indirection and address operators are the inverse of each other, and
when combined in an expression, such as * & x, they cancel each other. To see
this, let ’s break down the expression. These two unary operators are evaluated
from the right . T he first expression is therefore & x, the address of x , which as
we have seen, is a pointer value. The second expression, * ( & x ) , dereferences
the pointer constant, giving the variable ( x) itself. Therefore, the operators
effectively cancel each other ( Figure 9 -9). Of course, we would never code
the expression * & a in a program; we use it in Figure 9- 8 for illustration only.
&
J
inverse
>
•
J
FIGURE 9 - 9 Address and Indirection Operators
Pointer Declaration and Definition

As shown in Figure 9 - 10, we use the asterisk to declare pointer
variables .
, it is really an operator but rather a
When we use the asterisk in this way not
compilcr syntactical notation .

Figure 9 - 1 1 shows how we declare different pointer
.
variables Their cor -
responding data variables are shown for comparison
Note .that in each case
the pointer is declared to be of a given type Thus ,.p is a pointer to characters,
, and r is a pointer to floating-point variables.
while q is a pointer to integers
data declaration
type identifier
1
pointer declaration
type
D identifier ]
FIGURE 9 - 10 Pointer Variable Declaration
char a ; Z char* p
int n ;
GD int* q;
float x ; 3.3 float* r;
FIGURE 9- 1 1 Declaring Pointer Variables
EXAMPLE 9 - 1 Print with a Pointer

Program 9 - 1 stores the address of a variable in a pointer and then prints the
data using the variable value and a pointer.
PROGRAM 9- 1 Demonstrate Use of Pointers

1 /* Demonstrate pointer use
2 Written by:
3 Date:
4 */
5 ffinclude <stdio.h >
6
7 int main ( void )
8 {
10 int a ; a 14
11 int* p; 135760
12
13 // Statements p 135760
14 a =
14 ;
15 P = &a ;
16
17 printf( " %d % p\n " , a / &a);
continue
^
PROGRAM 9 - 1 Demonstrate Use of Pointers (continued )

18 printf( " % p %d % d \ n ",
19 P# *P f a);
20
21 return 0 ;
22 > / / main
Results:
14 00135760
00135760 14 14
Program 9- 1 Analysis Program 9- 1 requires a little explanation. First, we have defined an integer variable,
a, and a pointer to integer, p, to which we assign a's address. We then print twice .
The first print displays the contents of the variable a and its address (note the pointer
conversion code). The second print uses the pointer, p. It prints the pointer value con-
taining the address of a, followed by the contents of a, first referenced as a pointer
and then as a variable. This demonstrates two ways to access the data . We suggest
that you run this program for yourself. Of course, when you do, you will get a differ -
ent address . Also, some compilers display addresses as integers, and some use hexa -
decimal numbers.
Declaration versus Redirection

We have used the* asterisk operator in two different contexts: for declaration
.
and for redirection When as asterisk is used for declaration, it is associated
with a type. For example, we define a pointer to an integer as
int * pa ;
int * pb;
In this usage, the asterisk declares that the type of pa and pb is a pointer to
integer.
!
On the other hand, we also use the asterisk for redirection. When used
for redirection, the asterisk is an operator that redirects the operation from
the pointer variable to a data variable. For example, given two pointers to
integers, pa and pb, sum is computed as
sum = * pa + * pb;
The first expression, * pa, uses the pointer value stored in the pointer
variable, pa, to locate its value. The address in pa is not added; it is redirected
through the asterisk operator to the desired value. Likewise, the address
in
the second expression, pb, is not added but used to locate the second expres -
sion’s value. Once both of the expressions have been evaluated, they can be
added and the resulting expression value stored in sum.
Initialization of Pointer Variables

As we discussed in the previous section , the C language does not initialize
variables. Thus, when we start our program , all of our uninitialized variables
have unknown garbage in them .
The same is true for pointers. When the program starts, uninitialized
pointers will have some unknown memory address in them. More precisely,
they will have an unknown value that will be interpreted as a memory loea -
tion. Most likely the value will not he valid for the computer you are using, or
if it is, will not be valid for the memory you have been allocated. If the
address does not exist , you will get an immediate run - time error. If it is a valid
address, you often , but unfortunately not always , get a run - time error. ( It is
better to get the error when you use the invalid pointer than to have the pro-
gram produce garbage.)
One of the most common causes of errors in programming, by novices
and professionals alike, is uninitialized pointers. These errors can he very dif -
ficult to debug because the effect of the error is often delayed until later in
the program execution . Figure 9- 12 shows both an uninitialized variable and
an uninitialized pointer.
unknown
value
int a; a ??? fc ?
int* p ; P ???
^ pointer to \
unknown location
FIGURE 9- 12 Uninitialized Pointers
lo correct the problem shown in Figure 9- 12 , we must always assign a

valid memory address to the pointer, as shown in the next example.
int a ;
int* p = & a ; -
// int variable value unknown
// p has valid address
* p = 89 ; // a assigned value 89
As \ \ c saw with variables, it is also possible to initialize pointers when

they are declared and defined . All that is needed is that the data variable be
defined belore the pointer variable. For example , if we have an integer vari -
able , a , and a pointer to integer, p, then to set p to
point to a at declaration
time, we can code it as shown in Figure 9 - 13.
I 1'
Chapter 9 Pointers 56 /
int a ; int* p ;
declaration
int* p = &a;
P = &a ;
initialization
FIGURE 9 - 1 3 Initializing Pointer Variables
Note that in Figure 9- 13, the initialization involves two different steps.
First the variable is declared. Then the assignment statement to initialize the
variable is generated. Some style experts suggest that you should not use an
.
initializer in this way 2 Their argument is that it saves no code; that is, that
the initializer statement is required either as a part of the declaration and ini-
tialization or as a separately coded statement in the statement section of the
function. Putting the initialization in the declaration section tends to hide it
and make program maintenance more difficult . We tend to agree with them.
We can also set a pointer to NULL, either during definition or during exe-
cution. The following statement demonstrates how we could define a pointer
with an initial value of NULL.
int* p = NULL;
In most implementations, a null pointer contains address 0, which may

be avalid or invalid address depending on the operating system. II we deref -
erence the pointer p when it is NULL, we will most likely get a run- time error
.
because NULL is not a valid address The type of error we get depends on the
system we are using .
EXAMPLE 9 - 2 Change Variables- Fun with Pointers

Now let’s write a program and have with pointers. Our code is
some fun
shown in Program 9 - 2. Do figure out
not try to why this program is doing
what it is doing; there no
is reason. Rather, just try to trace the different vari -
ables and pointers as we change them.
PROGRAM 9 - 2 Fun with Pointers

1 / * Fun with pointers
2 Written by:
3 Date:
4 */
con l in i tec
of Style ( Mountain View, Calif .: M &T Books , 1992 ).

2 . For example, see Steve Oualline, C Elements
PROGRAM 9 - 2 Fun with Pointers (continued )

6
7 int main (void )
8 {
10 int a;
11
12
int b;
int c;
aQbGcQ
13
14
int* p;
int * q ;
p O q Or Q
15 int * r;
16
17 // Statements
bJ
18
19
20
21
a = 6;
b = 2;
p = &b ;
a
P qQrO^
22 q = p;
a [TJ b
23 r = &c; P
3
24
25 P = &a;
26 *q = 8;
27
28 *r = *p;
29
30 *r = a + *q + *&c;
31
32 printf( " %d %d %d \n" ,
33 ar b , c );
34 printf( " %d %d %d " ,
35 * P * q , * r );
'
36 return 0;
37 } // main
Results:
6 8 20
6 8 20
L
I
are uninitialized . The

Program 9-2 Analysis When the program starts, all the variables and their pointers
var iables have garbage values, and the pointers have invalid memory locations .
The first thing the program does, therefore, is to assign values to a and b, and to ini-
tialize all three pointers. After statement 23, both p and q point to b, and r points to c.
Statement 25 assigns the address of a to p. All three pointers now point to differ -
ent variables . Using the indirection operator, we dereference q ( * q), and assign b the
value 8.
Statement 28 demonstrates that both operands can be dereferenced when it
assigns the contents of a ( * p ) to c ( * r ).
Finally, in statement 30, we use three different formats to sum the values in the vari-
ables, a variable name, a dereferenced pointer, and a dereferenced address operator.
Using the figures in the program, trace these statements carefully to assure yourself that
you understand how they work .
EXAMPLE 9 - 3 Add Two Numbers

This example shows how we can use pointers to add two numbers. It explores
the concept of using pointers to manipulate data. A graphic representation ol
the variables is shown in Figure 9 - 14. The code is shown in Program 9 - 3.
a b
'
pa pb Pr
FIGURE 9 - 1 4 Add Two Numbers Using Pointers

l
PROGRAM 9 - 3 Add Two Numbers Using Pointers
to
1 / This program adds two numbers using pointers
demonstrate the concept of pointers.
2
3 Written by:
4 Date:
5 */
7
8 int main ( void )
9 {
11 int a;
12 int b;
13 int r;
14 int* pa = &a ;
continual
PROGRAM 9 - 3 Add Two Numbers Using Pointers ( continued )

15 int* pb = & b ;
16 int* pr = &r;
17
18 // Statements
19 printf( "Enter the first number : " );
20 scanf ( " %d " , pa);
21 printf( "Enter the second number: " );
22 scanf ( "%d " , pb);
23 * pr = * pa + * pb ;
24 printf("\n%d + %d is %d " , *pa , * pb , * pr);
25 return 0;
26 } // main
Results:
Enter the first number : 15
Enter the second number: 51
15 + 51 is 66
Program 9 - 3 Analysis This program is rather straightforward except for one thing. Look at statements 20
and 22 carefully. What is missing? When we discussed pass-by address in
Chapter 4, we mentioned that when input areas are passed by address, we don't use
the address operator in the sconf statement . This is because the pointer already con-
tains an address. It would be an error to use the address operator on the pointers in
this program.
EXAMPLE 9 - 4 Pointer Flexibility

I b i s example shows how we can use the same pointer to print the valued
different variables. T he variables and their pointer are shown in f igure 9 - 15 .
FIGURE 9 - 15 Demonstrate Pointer Flexibility

'\
The code is in Program 9-4.
PROGRAM 9 - 4 Using One Pointer for Many Variables

1 /* This program shows how the same pointer can point to
2 different data variables in different statements.
3 Written by:
4 Date:
5 */
7
8 int main ( void)
9 {
11 int a;
12 int b;
13 int c;
14 int* p;
15
16 // Statements
17 printf( " Enter three numbers and key return:
18 scanf ( " %d %d %d " , &a , & b , &c );
19 P = &a ;
20 printf( "% 3d\n " , * p );
21 P = & b;
22 printf( " % 3d\n" , * p );
23 p = &c ;
24 printf( "%3d\n" , *p);
25 return 0 ;
26 > // main
Results:
Enter three numbers and key return: 10 20 30
10
20
30
EXAMPLE 9- 5 Multiple Pointers For One Variable

to print the value of
This example shows how we can use different pointers
shown in Figure 9- 16 .
the same variable. The variable and its pointers are
FIGURE 9 - 16 One Variable with Many Pointers
The code is shown in Program 9- 5.
PROGRAM 9 - 5 Using A Variable with Many Pointers

1 /* This program shows how we can use different pointers
2 to point to the same data variable.
3 Written by:
4 Date:
5 */
7
8 int main ( void )
9 {
11 int a ;
12 int* p = &a ;
13 int* q = & a ;
14 int* r = &a ;
15
16 // Statements
17 printf( "Enter a number: ")?
18 scanf ( " %d " , & a ) ;
19 printf( " %d\n " , *P );
20 printf("%d \n " , * q ) ;
21 printf( " %d \n " , *r);
22
23 return 0;
24 > // main
Results:
Enter a number: 15
15
15
15
n1
9.2 Pointers for Inter- function Communication

One of the most useful applications of pointers is in functions. When we dis-
cussed functions in Chapter 4 , we saw that C uses the pass- by-value for
downward communication . For upward communication , the only direct way
to send something back from a function is through the return value. We also
saw that we can use upward and bi-directional communication hv passing an
address and using it to refer back to data in the calling program . When we
pass an address, we are actually passing a pointer to a variable . In this sec -
tion , we fully develop the bi - directional communication . We use two exam -
ples to demonstrate how pointers can be used in functions.
Passing Addresses
Figures 9 - 17 and 9 - 18 demonstrate the use of pointers. In Figure 9 - 17 , we
call the exchange function, passing it two variables whose values arc to be
exchanged . When we use downward communication , the data are exchanged
in the called function, hut nothing changes in the calling program .
Rather than passing data to the called function , we need to pass
addresses ( pointers). Once we have a pointer to a variable, it doesn t make
’
any difference if it is local to the active function , defined in main , or even if it

—
is a global variable we can change it ! In other words, given a pointer to a
variable anywhere in our memory space, we can change the contents of the
variable.
Figure 9 - 18 shows the exchange using pointers. To create the pointers,
we use the address operator ( & ) in the call , as show n below. We are now pass-
ing the address of a and b rather than their values.
exchange ( & a , & b ) ;
To pass addresses, the formal parameters in the called function are

defined as a pointer to variables. This definition , w hich is shown below , com -
pletes the connection and allows us to indirectly reference the variables in
the calling program through the pointers.
void exchange ( int * px , int * p y ) ;
To assign a value to a , all that we need to use is the indirection operator

us to access data in the
in exchange . Using the indirection operator allows
calling program .
We then call exchange using the address operator for
the variables that
. Note that exchange uses the two formal parameters,
we want to exchange
the parameters , we
px and py , and one local variable, temp. By
dereferencing
main and the local variable , temp ,
make the exchange using the variables in
in exchange . The correct logic is shown in Figure 9 - 18 .
574 Section 9.2 Pointers for Inter -function Communication
void exchange ( int x , int y );
K a b
5 I 7
int main ( void )
No Change
{
int a = 5;
int b = 7;
exchange (a , b);
printf( " %d %d\n\ a , b );
return 0;
} // main
{
void exchange (int x , int y )
int temp;
lk i
i
5
x
l 7
y i
i
As Passed
temp = x;
x = y? 7 i
y = temp ; I X y i
return ;
} // exchange When Exchanged
FIGURE 9 - 17 An Unworkable Exchange
void exchange ( int* , int* ); tk a b
X5
int main ( void )
{
int a = 5;
int b = 7 ;
exchange (&a , & b);

printf( " %d %d\n " , a , b);
return 0;
} // main
void exchange (int * px , int* py ) \

{
int temp;
temp = * px ;
^ x y
* px = * py ; temp
* py = temp;
return ; 5
} // exchange
FIGURE 9 - 18 Exchange Using Pointers

I
111 '
In summary, when we need to send more than one value back from a
-
function , we use pointers. Using either upward or bi directional communica -
tion . we can change any number of variables .
Every time we want a called function to have access to a variable in the calling
function, we pass the address of that variable to the called function and use
the indirection operator to access it .
Functions Returning Pointers

We have shown you many examples of functions using pointers , hut so far we
have shown none that return a pointer. Nothing prevents a function from
returning a pointer to the calling function . In fact , as you will see. it is quite
common for functions to return pointers.
As an example, let s write a rather trivial function to determine the smaller
of two numbers. In this case, we need a pointer to the smaller of two variables,
a and b . Since we are looking for a pointer, we pass two pointers to the func -
tion , which uses a conditional expression to determine which value is smaller.
Once we know the smaller value, we can return the address ol its location as a
'
pointer. The return value is then placed in the calling function s pointer, p, so
that after the call it points to either a or b based
, on their values . Both the
code and a diagram of the variables and pointers are seen in f igure 9- 19.
// Prototype Declarations
int* smaller ( int* pi , int* p2 ) a b
int main ( void )

t
\
int a ; X
X /
/
int b; x /
^
int * p ;
& a or &b |
scant ( "%d %d " , &a , &b );
p = smaller (&a , &b); P
int* smaller (int* px, int* py )

{
px py
return ( * px < * py ? px : py );
} // smaller
FIGURE 9 - 19 Functions Returning Pointers
the calling I unction or

When we return a pointer, it must point to data in
return a pointer to a local variable in
a higher level function . It is an error to
function terminates its memory mav
the called function , because when the
. Although a simple program might not
be used by other parts of the program
I!
576 Section 9.3 Pointers to Pointers
notice the error because the space was not reused , a large program would
either get the wrong answer or fail when the memory being referenced by the
pointer was changed .
It is a serious error to return a pointer to a local variable.
9.3 Pointers to Pointers

—
So far, all our pointers have been pointing directly to data . It is possible and
with advanced data structures often necessary—to use pointers that point to
other pointers. For example, we can have a pointer pointing to a pointer to an
integer. This two - level indirection is seen in Figure 9 - 20. There is no limit as
to how many levels of indirection we can use but , practically, we seldom use
more than two.
int a;
int * p;
int * * q;
pointer t o "" pointer to

inter to integer integer
q P a
J
234560 - »1 287650 I *
397870 234560 287650
/ / Statements
a = 58;
P = & a;
q = & P;
printf( " % 3 d ", a);
printf( " % 3 d ", * P );
printf( " % 3 d ", * * q ) ;
FIGURE 9 - 20 Pointers to Pointers
I ach level of pointer indirection requires a separate indirection opera -

totwhen it is dereferenced . In Figure 9- 20, to refer to a using the pointer p,
we have to dereference it once , as shown below.
*P
lo refer to the variable a using the pointer q , we have to dereference it

twice to gi t to the integer a because there are two levels of indirection ( point-
ers ) involved. If we dereference it only
once, we are referencing p , which is a
n
pointer to an integer. Another way to say this is that q is a pointer to a pointer
to an integer. The double dereference is shown below.
**q
MU
Figure 9- 20 also shows how we use these concepts in a C code fragment .
All three references in the printf statements refer to the variable a . The first
print statement prints the value of the variable a directly; the second uses the
pointer p; the third uses the pointer q . The result is the value 58 printed
three times, as shown below.
58 58 58
EXAMPLE 9- 6 Pointer-to-Pointer
The last example in this section shows how we can use different pointers with
pointers to pointers and pointers to pointers to pointers to read the value ot
——
the same variable. A graphic representation of the variables is shown in
Figure 9 - 2 1 . The code is shown in Program 9-6.
o a
r q p a
FIGURE 9 - 21 Using Pointers to Pointers
PROGRAM 9 -6 Using pointers to pointers

1 /* Show how pointers to pointers can be used by different
2 scanf functions to read data to the same variable.
3 Written by:
4 Date:
5 */
7
8 int main ( void )
9 {
11 int a;
12 int* p;
13 int* * q ;
14 int*** r ;
15
16 // Statements
17 p =&a ;
continuec
578 Section 9.4 Compatibility
PROGRAM 9- 6 Using pointers to pointers (continued )

18 q = & p;
19 r = &q ;
20
21 printf( "Enter a number: " );
22 scant ( " %d ", &a ); // Using a
23 printf("The number is : %d\n " , a );
24
25 printf( " \nEnter a number: " );
26 scant ("%d" , p); // Using p
27 printf( "The number is : %d\n" , a );
28
29 printf( "\nEnter a number: " );
30 scant ("%d" , *q ); // Using q
31 printf("The number is : %d\n", a );
32
33 printf("\nEnter a number: ");
34 scant ( " %d " , **r ); / / Using r
35 printf( "The number is : %d\n" / a);
36
37 return 0;
38 } // main
Results:
Enter a number: 1
The number is : 1
Enter a number: 2
The number is : 2
Enter a number: 3
The number is : 3
Enter a number: 4
The number is : 4
Program 9 - 6 Analysis In each successive read we use a higher level of indirection. In the print statements,
owever, we a ways use the integer variable, a, to
prove that the reads were successful.
9.4 Compatibility
It is important to recognize that pointers
have a type associated with them.
They are not just pointer types, hut rather are
pointers to a specific type, sue
as character. Each pointer
therefore takes on the attributes of the type to
which it refers in addition to its own
attributes.
Chapter 9 Pointers 5 / 9
Pointer Size Compatibility

I he size of all pointers is the same. Every pointer variable holds the address
of one memory location in the computer. On the other hand, the size of the
variable that the pointer references can be different; it takes the attributes of
the type being referenced. This is demonstrated by Program 9 - 7, which prints
the size of each pointer and the size of the referenced object.
PROGRAM 9 - 7 Demonstrate Size of Pointers

1 /* Demonstrate size of pointers.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
10 char c;
11 char* pc ;
12 int sizeofc = sizeof(c );
13 int sizeofpc = sizeof( pc );
14 int sizeofStarpc = sizeof( * pc );
15
16 int a;
17 int* pa ;
18 int sizeofa = sizeof(a);
19 int sizeofpa = sizeof( pa );
20 int sizeofStarpa = sizeof( *pa );
21
22 double x ;
23 double* px ;
24 int sizeofx = sizeof( x );
25 int sizeofpx = sizeof( px);
26 int sizeofStarpx = sizeof( *px );
27
28 // Statements
29 printf( "sizeof(c): % 3d " , sizeofc );
30 printf( "sizeof( pc ): % 3d | " , sizeofpc );
31 printf( "sizeof( * pc ): % 3d \n " , sizeofStarpc );
32
33 printf( " sizeof(a ): % 3d | " , sizeofa );
34 printf( "sizeof( pa ): %3d | " , sizeofpa );
35 printf( "sizeof( * pa ): % 3d\n " , sizeofStarpa );
36
contimiec
r 580 Section 9.4 Compatibility
PROGRAM 9 -7 Demonstrate Size of Pointers (continued )

37 printf( "sizeof( x ): % 3d | " , sizeofx );
38 printf("sizeof(px): % 3d | " , sizeofpx);
39 printf( "sizeof(* px ): %3d \n" , sizeofStarpx );
40
41 return 0;
42 } // main
Results:
sizeof(c): 1 sizeof(pc): 4|sizeof(*pc): 1
sizeof(a): 4 sizeof( pa ): 4 | sizeof( * pa ): 4
sizeof( x ): 8 | sizeof( px ): 4 | sizeof( * px ): 8
Program 9 -7 Analysis What is this code telling us? First, note that the variables a, c, and x are never
assigned values. This means that the sizes are independent of whatever value may be
in a variable. In other words, the sizes are dependent on the type and not its values.
Now look at the size of the pointers. It is 4 in all cases, which is the size of an
address in the computer on which this program was run. This makes sense: All com-
puters today have more than 65,535 bytes, which is the maximum address that could
be stored in 2 bytes.
But note what happens when we print the size of the type that the pointer is refer -
ring to: The size is the same as the data size! This means that in addition to the size of
the pointer, the system also knows the size of whatever the pointer is pointing to. To
confirm this, look at the size of the pointer, px, and what it is pointing to when deref -
erenced ( * px ) .
Dereference Type Compatibility

I he second issue in compatibility is the dereference type . The dereference
type is the type ol the variable that the pointer is referencing. With one
exception ( discussed later ) , it is invalid to assign a pointer of one type to a
pointer ol another type , even though the values in both cases are memory
addresses and would therefore seem to be fully compatible. Although the
addresses may he compatible because they are drawn from the same set , what
is not com patihle is the underlying data type ol
the referenced object .
In C, we can ’t use the assignment operator with pointers to different
types; il we try to, we get a compile error. A pointer to a char is only compati -
ble with a pointer to a char ; and a pointer to an int is
only compatible with a
pointer to an int . We cannot assign a pointer to a char
to a pointer to an int .
l ets construct an example in which we have
two variables: one int and one
char . We also define one pointer to char and
one pointer to int as shown in
Figure 9-22.
Type: pointer to char 123450 \ 123450

pc c
Type: pointer to int 234560 1 +* 58 | 234560
pa a
char c;
char * pc ;
int a;
int* pa ;
pc = &c; // Good and valid
pa = &a; // Good and valid
pc = &a; // Error: Different types

pa a; // Error: Different levels
FIGURE 9 - 22 Dereference Type Compatibility
The first pair of assignments are valid . In the first assignment, we store the
address of a character variable in a pointer to character variable. In the second
assignment , we store the address of an integer ( int ) variable in a pointer to an
integer ( int ) variable. We create an error in the third assignment because we
try to store the address of a character variable into a pointer variable whose
type is pointer to integer ( int ) . We also get an error in the fourth assignment.
Pointer to Void
The exception to the reference type compatibility rule is the pointer to void .
As we discussed in Chapter 2 , one of the types defined in C is void . Although
the void type defines the lack of any other type when used with functions, its
use with pointers is different . A pointer to void is a generic type that is not
associated with a reference type; that is, it is not the address of a character,
an integer, a real , or any other type. However, it is compatible for assignment
purposes only with all other pointer types. Thus, a pointer of any reference
type can be assigned to a pointer to void type and a pointer to void type can be
assigned to a pointer of any reference type . There is, however, one restriction ;
since a void pointer has no o bject tvpe , it cannot be dereferenced unless it is
cast ( see next section ). The following shows how we can declare a variable of
pointer to void type.
void * pVoid ;
A
582 Section 9.4 Compatibility
It is important to understand the difference between a null pointer and a

variable of type pointer to void . A null pointer is a pointer of any type that is
assigned the constant NULL. The reference type of the pointer will not change
with the null assignment . A variable of pointer to void is a pointer with no ref -
erence type that can store only the address of any variable. The following
examples show' the difference
void * pVoid ; // Pointer to void type

int* pint = NULL; // Null pointer of type int
char* pChar = NULL; // Null pointer of type char
A void pointer cannot be dereferenced.
Casting Pointers
The problem of type incompatibility can be solved if we use casting. We can
make an explicit assignment between incompatible pointer types by using a
cast , just as we can cast an int to a char . For example, if we decided that we
needed to use the char pointer, pc in the previous example, to point to an int
( a ) , we could cast it as shown below.
pc = (char* )&a ;
But in this case, user beware! Unless we cast all operations that use pc ,
we have a great chance ol creating serious errors. In fact , with the exception
ol the void pointer, we should never cast a pointer. The following assignments
are all valid , but they are extremely dangerous and must be used with a very
carclully thought -out design .
void * pVoid ;
char* pChar ;
int* pint;
// Statements
pVoid = pChar ;
pint = pVoid ;
pint = (int* ) pChar ;
Another use ol the cast is to provide a type for the void pointer. As noted
in the previous section , a void pointer cannot be dereferenced because it has
.
no object type \ \ hen we cast it , however, we provide
the type .
Dereference Level Compatibility

Compatibility also includes dereference level compatibility. For example , a
pointer to int is not compatible with a pointer- to - pointer to int . The pointer to
int has a reference type ol int , while a pointer- to- pointer to int has a reference
type ol pointer to int . Figure 9- 23 shows two pointers declared at different * '1
levels . I he pointer pa is a pointer to int ; the pointer ppa is a pointer- to -

pointer to int .
ppa
h-H pa a
int
int
a;
b;
//
//
type
type
int
int tv
int* pa ; // type pointer to int
int** ppa ; // type pointer to pointer to int
pa = &a ; // Valid: same level
ppa = & pa ; // Valid: same level
b = ** pa ; // Valid: same level
pa = a; // Invalid: different level
ppa = pa ; // Invalid: different level
b = * ppa ; // Invalid: different level
FIGURE 9 - 23 Dereference Level Compatibility
9.5 Lvalue and Rvalue

In C , an expression is either an lvalue or an rvalue. As you know, every expres-
sion has a value. But the value in an expression (after evaluation ) can he used
in two different ways.
1 . An lvalue expression must he used whenever the object is receiving a

value ; that is , it is being modified .
2. An rvalue expression can be used to supply a value for further use ; that
is, to examine or copy its value.
But how do you know when an expression is an lvalue and when it is an

rvalue? Fortunately, only seven types of expressions are lvalue expressions.
They are shown in Table 9- 1 .
I Hi
^
"
584 Section 9.5 Lvalue and Rvalue
Expression Type 0 Comments
1. identifier Variable identifier

2. expression [...] Array indexing
3. (expression) Expression must already be lvalue
4. * expression Dereferenced expression
5. expression name . Structure selection
6. expression-> name Structure indirect selection
7. function call If function uses return by address
a
Expression types 5, 6, and 7 have not yet been covered.
TABLE 9 - 1 lvalue Expressions
For example , the following are lvalue expressions:
a = ... a[5] = ... (a) = ... *p = ...
All expressions that are not lvalue expressions are rvalues. The following
show some rvalue expressions:
5 a + 2 a * 6 a[2 ] + 3 a++
Note that even if an expression is an lvalue, if it is used as part of a larger

expression in which the operators create only rvalue expressions, then the
whole expression is an rvalue. 1 or example, a [ 2 ] is an lvalue. But when it is
used in the expression a [ 2 ] + 3 , the whole expression is an rvalue, not an
lvalue. Similarly, in the expression a + + , the variable a is an lvalue while the
whole expression ( a + + ) is an rvalue .
You may ask , “ Why worry so much about lvalues and rvalues ? The rea-
son is that some operators need an lvalue as their operand . If we use one of
these operators and use an rvalue in place of the operand, we will get a com-
pile error.
The right operand of an assignment operator must be an rvalue expression .

( )nl \
six operators need an lvalue expression as an operand : address
operator, postfix increment, postfix decrement , prefix increment , prefix dec -
rement , and assignment. They are shown
of each .
in Table 9 2 with examples -
i ll
Type of Expression Examples

Address operator Sscore
Postfix increment/ decrement x++ y--
Prefix increment/ decrement ++ x -- y

Assignment (left operand) x 1 Y += 3
TABLE 9 - 2 Operators That Require lvalue Expressions

Table 9-3 contains several examples of invalid expressions that will create
syntax errors because an rvalue is used when an lvalue is needed.
Expression Problem
a + 2 = 6; a + 2 is an rvalue and cannot be the left operand in an
assignment; it is a temporary value that does not have an
address; no place to store 6.
& (a + 2) ; a + 2 is an rvalue, and the address operator needs an
lvalue; rvalues are temporary values and do not have
addresses.
& 4; Same as above (4 is an rvalue) .
( a + 2 ) ++ ; Postfix and prefix operators require lvalues; (a + 2 ) is an
++ ( a + 2 ) ; rvalue.
TABLE 9 - 3 Invalid rvalue Expressions

One Pinal thought: A variable name can assume the role of either an
Ivalue or an rvalue depending on how it is used in an expression . In the fol -
lowing expression , a is an lvalue because it is on the left of the assignment
and b is an rvalue because it is on the right of the assignment.
a = b
Pointer Examples
In this section we demonstrate two ways we can use pointers when calling
functions.
EXAMPLE 9 -7 Convert Seconds to Hours

We begin with a simple function that converts time in seconds to hours, min -
utes, and seconds. While the function is simple, it does require three address
parameters to return the values. The code is shown in Program 9-8.
PROGRAM 9- 8 Convert Seconds to Hours, Minutes, and Seconds

1 /* ===== === secToHours
2 Given time in seconds, convert it to hours , minutes,
3 and seconds.
4 Pre time in seconds
5 addresses of hours, minutes, seconds
6 Post hours, minutes, seconds calculated
7
8 */
—
Return error indicator 1 success , 0 bad time
9 int secToHours ( long time ,

10 int* hours, int* minutes, int* seconds )
11 {
13 long localTime;
14
15 // Statements
16 localTime = time;
17 seconds = localTime % 60 ;
18 localTime = localTime / 60;
19
20 minutes = localTime % 60;
21
22 hours = localTime / 60 ;
23
24 if (*hours > 24)
25 return 0;
26 else
27 return 1;
28 } // secToHours
Program 9 - 8 Analysis The first question you might ask when reading this simple function is "Why define a
local variable for the time?" In this short function, it really wasn't necessary. However,
a good programmer does not change a value parameter
within the function, because
its original value may be needed later. We have seen
times when "later" turned out to
be a maintenance change and hours were spent debugging the error when the
wrong value was computed.
Varia e 3 w en 3 va ue
ti
6
^ ^ parameter will be changed within a func-
*
n 50 that the original value will always be
.
available for processing
function do ^ y°a nee

^ throughSfatUS
u < to send back two values from a called
^ and return the other. Either use the
return for somp nth°
SS
^ pointer
—In ° maê return vo c^ your

S0
desiqn consistent ° r /
' 'V
Chapter 9 Pointers Jj87
When several values need to be sent back to the calling function, use address
.
parameters for all of them Do not return one value and use address parame -
ters for the others .
EXAMPLE 9 - 8 Quadratic Equations

Let ’s look at a typical program design that reads, processes, and prints data.
This is a processing cycle that is common to many, many programs. Figure 9 - 24
shows its structure chart.
program
getData compute printResults
FIGURE 9 - 24 A Common Program Design
To demonstrate the universality of this design, let ’s compute the real

roots for a quadratic equation. Recall that a quadratic equation has the form
2
a\ + b\ +c = 0
Four possible situations can occur when you solve lor the roots in a qua -
dratic equation. First, it is an error if both a and b are 0: 1 here is no solution.
Second, if a is 0 and b is not 0, there is only one root:
-c
x
b
Third, if b 2 - 4ac is 0 or positive, there are two , possibly equal, roots

derived from the following equation:
2
x =
J
- b± b — 4ac
2a
Finally, if b 2 - 4ac is negative, the roots are imaginary.

Figure 9 - 25 diagrams the interaction of the variables and pointers
lor
Program 9 - 9. In this short program , we use pointers to pass data irom a read
function , pass both values and pointers to a compute function, and finally
pass the values to a print function.
main
a b c rootl root2 numRoots
trSniJ
pa
getData
pb pc
(a) Calling getData

main
a
quadratic
b c
#n
pRootl pRoot2
(b) Calling quadratic

main
a b c rootl root 2 numRoots
LJ
printResults
(c) Calling printResults
FIGURE 9- 25 Using Pointers as Parameters
PROGRAM 9- 9 Quadratic Roots

1 /* Test driver for quadratic function.
2 Written by:
3 Date:
4 */
6 # include < math.h >
7
9 void getData ( int* int* pb , int* pc);
pa ,
10 int quadratic ( int a,
int b, int c,
11 double* pRootl , double* pRoot2);
12 void printResults ( int
numRoots ,
continued
PROGRAM 9 - 9 Quadratic Roots ( continued )

13 int a, int b, int c,
14 double rootl , double root 2 );
15
16 int main ( void)
17 {
19 int a;
20 int b;
21 int c;
22 int numRoots;
23 double rootl ;
24 double root2 ;
25 char again = 'Y' ;
26
27 // Statements
28 printf( "Solve quadratic equations.\n\n " );
29 while ( again == 'Y ' || again == Y * )
30 {
31 getData (&a , &b , &c);
32 numRoots = quadratic (a , b , c, &rootl , &root2 );
33 printResults ( numRoots, a , b , c , rootl , root2 );
34
35 printf( "\nDo you have another equation (Y/N): ");
36 scanf ( " %c" , &again );
37 > // while
38 printf( " \nThank you.Xn " );
39 return 0;
40 } // main
41
42 /* ===: ===== getData ====
43 Read coefficients for quadratic equation.
44 Pre a , b , and c contains addresses
45 Post data read into addresses in main
46 */
47 void getData ( int* pa , int* pb , int* pc )
48 {
49 // Statements
50 printf( "Please enter coefficients a, b , & c: " );
51 scanf ( " %d %d %d " , pa , pb , pc );
52
53 return ;
54 > // getData
55
56 j * ========== ==== quadratic =====
continued
jr
PROGRAM 9- 9 Quadratic Roots (continued )

57 Compute the roots for a quadratic equation.
58 Pre a , b, & c are the coefficients
59 pRootl & pRoot2 are variable pointers
60 Post roots computed , stored in calling function
61 Return 2 two roots,
62 1 one root,
63 0 imaginary roots
64
65 */
-
1 not quadratic coefficients.
66 int quadratic ( int a , int b , int c ,

67 double* pRootl , double* pRoot2 )
68 {
70 int result;
71
72 double discriminant;
73 double root;
74
75 // Statements
76 if (a == 0 && b == 0)
77
78
result = 1;
else
-
79 if (a = = 0)
80 {
81 * pRootl = -c / ( double) b;
82 result = 1;
83 > // a == 0
84 else
85 {
86 discriminant = b * b (4 * a * c );
87 if ( discriminant >= 0 )
88 {
89
90
root
*pRootl =
= _
sqrt( discriminant );
( b + root) / ( 2 * a);
91
92
* pRoot2 =
result =
- -
( b root) / ( 2 * a);
2;
93 > // if >= 0
94 else
95 result = 0;
96 > // else
97 return result;
98 } // quadratic
99
100
—=== printResults ====
continue

101 Prints the factors for the quadratic equation.
102 Pre numRoots contains 0, 1, 2
103 a , b , c contains original coefficients
104 rootl and root2 contains roots ; K
105 Post roots have been printed
106 */
107 void printResults (int numRoots,
108 int a , int b , int c,
109 double rootl , double root2 )
110 {
111 // Statements
112 printf( " Your equation: %dx**2 + %dx + %d \n " ,
113 a , b, c);
114 switch ( numRoots)
115 {
116 case 2: printf( "Roots are: %6.3f & %6.3f \n " ,
117 rootl , root2 );
118 break;
119 case 1: printf( "Only one root: %6.3 f \n " ,
120 rootl );
121 break ;
122 case 0: printf( " Roots are imaginary.\n " );
123 break ;
124
125
default: printf("Invalid coefficients ");
break ; ^
126 > // switch
127 return ;
128 } // printResults
129 ===== End of Program ==== =
Results:
Solve quadratic equations.
Please enter the coefficients a, b , & c: 2 4 2

Your equation: 2x**2 + 4x + 2
-
Roots are: 1.000 & 1.000 -
Do you have another equation ( Y/N ): y
Please enter the coefficients a , b, & c: 0 4 2
-
Only one root: 0.500
Do you have another equation (Y/N): y

Please enter the coefficients a , b, & c: 2 2 2
continuec

Roots are imaginary.

Please enter the coefficients a, b , & c: 002
Your equation: 0x**2 + Ox + 2
Invalid coefficients

Please enter coefficients a, b , & c: 1 -5 6
Your equation: lx**2 + 5 x + 6
Roots are: 3.000 & 2 . 0 0 0
-
Do you have another equation ( Y/N ): n
Thank you.
Program 9-9 Analysis This problem has many interesting points. The function main is a test driver — that is,
code that will not be put into production . Therefore, we code much of the test logic in
mam rather than providing separate functions for it.
The scanf function used in getData does not have the usual address operator ( & )
in front of variables a, b, and c. Why? Because they are already pointer values point-
ing to variables in main and therefore are already addresses.
The variables in these examples are either integers or pointers to integers. Parame-
ters that receive something from main are integers whose values will be
filled when the
call is made. Those that send data back to main are pointers to integers that will be
filled with the addresses of the corresponding variables in main. As a general rule, if a
value will be changed, it must be passed as a pointer. If it will not be changed, it should
be passed as a value. This protects data from accidental destruction.
In quadratic, note the extensive testing to make sure that the coefficients are
valid. To ensure valid code, they are all necessary. Look at how we calculated the square
root of the discriminant separately ( statement 89) .
Since square root is a complex func-
tion, it is more efficient to call it just once and save the
value for later use. Note also that
the function has only one return statement. This is proper
structured code, though many
professional programmers would simply return at statements 77,
82, 92, and 95.
Study our test data carefully. Note that this set of test data executes
every line of
code in the program. Designing test data that completely a function is not an
validates
easy task. Ensuring that all code has been executed is even
more difficult and tedious.
One way to make sure that all code has been
tested is to use your debugger to set a
break point at every statement and then clear them as the all
break points have been cleared, you know every instructionprogram executes. When
has been executed.
Executing every line of code does not ensure that the
ever. With large programs, it is virtually
function has no bugs, how-
impossible to test every possible combination
of data. One of the advantages of structured
gram down into separate functions we can
programming is that by breaking the pro-
test it better.
This program has a potential problem. Do
you see it? Hint: What if the user enters
invalid data? There is no error checking in getData
. If this were a production pro-
gram, it would contain code to check for
errors. It would then return a status flag to
indicate if getData was successful or not.

In the previous chapters, software engineering has always been related to the
material in the chapter. This chapter is an exception . In this chapter, we dis-
cuss a general software engineering topic , quality, which can he applied to
any topic , including pointers.
You will find no one who would even consider minimizing software qual -
ity, at least publicly. Everyone wants the best software available , and to listen
to the creators of systems on the market , their systems are all perfect . Yet , as
users of software , we often feel that quality software is a contradiction in
terms. We all have our favorite software products, but not one of them is
without a wart or two.
Since you are now moving into the world to be one of those software cre -
ators, you need to be aware of the basic concepts of software quality. In this
chapter, we discuss some of the attributes of a quality product and how you
go about achieving quality.
Quality Defined
Quality software is defined as
Software that satisfies the user’s explicit and implicit requirements, is well
documented, meets the operating standards of the organization, and runs
efficiently on the hardware for which it was developed.
Every one of these attributes of good software falls squarely on you , the sys -
tem designer and programmer. Note that we place on you the burden ol satis -
fying not only the users ' explicit requirements, but also their implicit needs.
Often , users don 't fully know what they need . When this happens, it is your
job to determine their implicit requirements, which arc hidden in the back -
ground . This is a formidable task indeed .
Of course, it is also your job to document the software. If you are lucky,
you will have a technical writer to help, and even if you do, the final product
is still your responsibility. And as an analyst and programmer, you are
expected to know the standards of your organization and to implement them
properly.
Finally, it is your program , so you are responsible for its efficiency. This
means that you are expected to use appropriate and efficient algorithms. I bis
was the focus of our discussions in Chapters 6 and 8 when we talked about
analyzing algorithms and the big-O theory.
But quality software is not just a vague concept. If we want to attain it
,
we have to be able to measure it . Whenever possible, these measurements
should be quantitative; they should he numerically measurable. For example
,
, should he able to tell you the
if an organization is serious about quality it
number of errors ( bugs ) per thousand lines of code and the mean
time
between failures lor every software system it maintains
. I hese are measur-
able statistics.
r ~
On the other hand , some of the measurements may he qualitative, mean -

ing that they cannot be numerically measured . Flexibility and testability are
exam pies of qualitative software measurements. This does not mean that they
can t be measured , but rather that they rely on someone ’s judgment in assess-
ing the quality of a system .
Quality Factors
Software quality can be divided into three broad measures: operability, main-
tainability, and transferability. These measures can be broken down as shown
in Figure 9-26.
Software
Quality
operability maintainability transferability
• accuracy • changeability • code reusability

• efficiency • correctability • interoperability
• reliability • flexibility • portability
• security • testability
• timeliness
• usability
FIGURE 9 - 26 Software Quality
Operability
Operability refers to the basic operation of a system . The first thing a user
notices about a system is its “ look and feel . ” This means, especially for an
online , interactive system , how easy and intuitive it is to use. Does it fit well
into the operating system it is running under? For example , if
it is running in
a Windows environment , its pull -down and
pop - up menus should work the
same way the operating system ’s menus do. In short , operability
answers the
question , " How does it drive ?”
But these factors are subjective; they are not measurable.
So let’s look at
the factors that comprise operability. I hey are listed
alphabetically.
Accuracy
A system that is not accurate is worse
than no system at all. Most workers
would rather rely on intuition and experience
than a system that they know
gives false and misleading information .
Am s \ stem that you develop, therefore ,
must be thoroughly tested , both
b \ you ( whitebox ) and by a systems
test engineer and the user ( blackbox ). If
you get the opportunity, take a course on
software testing.
Accuracy can be measured by such metrics as mean time between fail-

ures, number of bugs per thousand lines of code , and number of user
requests for change.
Efficiency
Efficiency is, by and large, a subjective term . In some cases the user will
specify a performance standard , such as that a real - time response must be
received within 1 second , 95 % of the time . This is certainly measurable.
Reliability
Reliability is really the sum of the other factors. If users count on the system
to gettheir job done and are confident in it , then it is most likely reliable. On
the other hand , some measures speak directly to a system’s reliability, most
notably, mean time between failures.
Security
The security of a system refers to how easy it is for unauthorized persons to
access the system ’s data . Although this is a subjective area , there are check-
lists that assist in assessing the system ’s security. For example, does the sys-
tem have and require passwords ?
Timeliness
Does the system deliver its output in a timely fashion ? For online systems ,
does the response time satisfy the users’ requirements? For batch systems,
are the reports delivered in a timely fashion ? It is also possible, il the system
has good auditability, to determine if the data in the system are timely ; that is,
are data recorded within a reasonable time after the activity that creates them
takes place ?
Usability
Usability is also highly subjective. The best measure of usability is to watch
the users and see if they are using the system . User interviews will often
revea 1 problems with the usability of a system .
Maintainability
Maintainability refers to keeping a system running correctly and up to date.
Many systems require regular changes, not because they were poorly imple-
mented but because of changes in external factors. For example, the payroll
system for a company must be changed yearly, il not more often
, to meet
changes in government laws and regulations .
Changeability
The ease of changing a system is a subjective factor. Experienced project
will take. If it
leaders , however, can estimate how long a requested change
takes too long, it may well be because the system is difficult to change. This is
especially true of older systems.
Current software measurement tools estimate a program ’s complexity
and structure . These tools should be used regularly, and if a program ’s com-
plexity is high , we should consider rewriting the program . Programs that have
been changed many times may have often lost their structured focus and
become difficult to change. They also should be rewritten .
Correctability
One measure of correctability is mean time to recovery— how long it takes to
get a program back in operation when it fails. Although this is a reactive defi-
nition , there are currently no predictors of how long it will take to correct a
program before it fails .
Flexibility
Users are constantly requesting changes in systems. This qualitative attribute
attempts to measure how easy it is to make these changes. If a program
needs to be completely rewritten to effect a change, it is not flexible. Fortu -
nately, this factor became less of a problem with the advent of structured
programming.
Testability
You might think that testability is a highly subjective area , but a test engineer
has a checklist ol factors that can be used to assess a program ’s testability.
Transferability
Transferability refers to the ability to move data and /or a system from one
platform to another and to reuse code. In many situations, it is not an impor-
tant factor. On the other hand , il you are writing generalized software, it can
be critical.
Code Reusability
II ( unctions are written so that they can be reused in
different programs and
on different projects, then they are highly reusable.
Good programmers build
libraries of reusable functions that they can use when they need to solve a
similar problem .
Interoperability
Interoperability addresses the capability of sending data to other systems. In
todays highly integrated systems, interoperability
( act , it has become so is a desirable attribute. In
important that operating systems now support the abil-
ity to move data between applications,
such as between a word processor and
a spreadsheet.
Portability
Portability addresses the ability to move software from one hardware platform
to another; for example , from a Macintosh to a Windows environment or
Irom an IBM mainframe to a VAX environment .
The Quality Circle

The first and most important point to recognize is that quality must be
designed into a system . It can ’t be added as an afterthought. It begins at
Step 1 , determining the user requirements, and continues throughout the life
of the system . Since quality is a continuous concept that, like a circle, never
ends , we refer to it as the quality circle .
I here are six steps to developing quality software: quality tools, technical
reviews, formal testing, change control, standards, and measurement and
reporting. These steps are shown in Figure 9- 27.
FIGURE 9- 27 The Quality Circle
While no one can deny that quality begins with the software engineers
: igned to the team , they need quality tools to develop a quality product.
assi
Fortunately, today’s development tools are excellent . A whole suite of quality
tools known as computer-assisted software engineering ( CASE ) guides soft -
ware development through requirements, design , programming and testing,
and into production . Programmers can use workstations that not only assist
in writing the program but also in testing and debugging. For example , some
CASE tools track tests through a program and then determine which state-
ments were executed and which were not . Tools such as these are invaluable
for whitebox testing.
Another major step in quality software is the technical review. T hese
reviews should be conducted at every step in the development process includ -
ing requirements , design , programming, and testing. A typical program review
begins after the programmer has designed the data structures and structure
chart for a program . A design review hoard consisting of the systems analyst ,
test engineer, user representative, and one or two peers is then convened .
Note that no one from management is allowed to attend a technical review.
During the review, the programmer explains the approach and discusses
interfaces to other programs while the reviewers ask questions and make
suggestions.
Quality also requires formal testing. Formal testing ensures that the pro-
grams work together as a system and meet the defined requirements. After
the programmer has completed unit testing, the program is turned over to
another software engineer for integration and system testing. On a small
project, this is most likely the systems analyst and /or the user. A large project
will have a separate testing team.
Large systems take months and sometimes years to develop. It is only
natural that , over extended periods of time, changes to the requirements and
design become necessary. To ensure quality, each change should he reviewed
and approved by a change control board . The impact of a change on each
program needs to be assessed and properly planned . Uncontrolled change
causes schedule and budget overruns and poor- quality products.
Finally, a good -quality environment measures all aspects of quality and
regularly reports the results. Without measurement , you cannot tell if quality
is good or bad , improving or deteriorating. At the same time , published stan -
dards provide the yardstick for many of the quality measurements.
Conclusion
In this short discussion , we have only introduced the concept of software qual-
ity. Consider these points as you design and program systems in the future. For
a more complete discussion of the subject , we recommend Chapter 12 in
Roger Pressmans Software Engineering , A Practitioner’s App roac Ji. 3
3. Roger Pressman , Software Ertgitneerimg ,

A Practitioner's Approach . Sixth Edition ( New York,
'
N .Y., McGraw- Hill Rook Company, 2005 ) .


1 . I he address of a memory location is a pointer constant and cannot he
changed.
.
2 Only an address (pointer constant) can he stored iin a pointer variable.
3. Remember compatibility. Do not store the address of a data variable of
one type into a pointer variable ol another type. In other words, a variable
ol pointer to int can only store the address of an int variable, and a vari-
able ol pointer to char can only store the address of a char variable.
4. I he value of a data variable cannot he assigned to a pointer variable. In
other words, the following code creates an error :
int* p;
int a ;
P = a; // ERROR
.
5 You must not use a pointer variable before it is assigned the address of a
variable. In other words, the following lines create an error unless p is
assigned an address:
int* p ;
x *p; // ERROR
*p = x; // ERROR
.
6 A pointer variable cannot be used to refer to a nonexistent pointed vari-
able. For example, the following lines create a run- time error because p .
exists, hut * p does not exist until it is assigned a value.
int* p ;
*p = ..
.
7 You cannot dereference a pointer variable of type void * .
8. Remember that the definition for a pointer variable allocates memory
only for the pointer variable, not for the variable to which it is pointing.
9. The address operator ( & ) must be used only with an lvalue.
.
10 A function that uses addresses as parameters needs a pointer to data as a
formal parameter; the actual parameter in the function call must be a
pointer value ( address ).
11. If you want a called function to change the value of a variable in the calling
function, you must pass the address of that variable to the called function.
12 When using multiple definitions in one statement —a practice we do not
.
recommend—the pointer token is recognized only with one variable .
Therefore, in the following definition, only the first variable is a pointer
to an integer; the rest are integers.
A
ir~n
int* ptrA, ptrB , ptrC ;
13. It is a compile error to initialize a pointer to a numeric constant.
int* ptr = 59;
14. Similarly, it is a compile error to assign an address to any variable other

than a pointer.
int x = &y ; / / ERROR
15. It is a compile error to assign a pointer ol one type to a pointer of

another type without a cast . ( Exception: II one pointer is a void pointer,
it is permitted . )
16. It is a common compile error to pass values when addresses are required
in actual parameters. Remember to use the address operator when pass-
ing identifiers to pointers in the called function .
17. It is a logic error to use a pointer before it has been initialized.
18. It is a logic error to dereference a pointer whose value is NULL.
9.8 Key Terms

indirection operator ( * ) pointer indirection
lvalue pointer variable
pointer rvalue
pointer constant
9.9 Summary
Every computer has addressable memory locations.
J A pointer is a derived data type consisting of addresses available in the
computer.
J A pointer constant is an address that exists by itself . We cannot change it.
We can only use it .
J I he address operator ( & ) makes a pointer value f rom a .
pointer constant To
get the address ol a variable, we simply use this operator in front ol the
variable name.
Ll I he address operator can only be used in
front of an lvalue. The result is
an rvalue .
A pointer variable is a variable that can store an
address.
^
I he indirection operator ( * ) accesses a variable through a pointer contain -

ing its address.
I he indirection operator can only be used in front of a pointer value.
Pointer variables can be initialized just like data variables. The initializer
must be the address ol a previously defined variable or another pointer
variable.
J A pointer can be defined to point to other pointer variables ( pointers to
pointers) .
The value ol
a pointer variable can be stored in another variable it they are
compatible— that is, if they are of the same type.
One ol the most useful applications of pointers is in functions .
II we want a called function to access a variable in the calling function, we
pass the address of that variable to the called function and use the indirec -
tion operator to access the variable.
\ \ hen we need to return more than one value from a function, we must
use pointers.
II a data item is not to be changed, it is passed to the called function by

value. II a data item is to be changed, its address is passed to let the called
function change its value.
A function can also return a pointer value.
In software engineering, quality factors refer to characteristics that a piece
of software must have to become quality software.
Quality factors are defined as operability, maintainability; and transferability.
One of the most important points about the quality of a piece of software
.
is that quality must he designed into the system It cannot be added as an
afterthought. To design quality into a system, we can use a tool called the
quality circle.
9.10 Practice Sets
Review Questions
.
1 Pointer constants are drawn from the set of addresses for a computer.
a. True
b. False
2. Pointers are declared using the address operator as a token.
.
a True
.
b False
3. The value of a pointer type is always an address.

a. True
b. False
4. An rvalue expression can be used only to supply a value.
a . True
b. False
5. Which of the following statements about pointers is false?
a . Pointers are built on the standard type, address.
b. Pointers are machine addresses.
.
c Pointers are derived data types.
d . Pointers can be defined variables.
e. Pointers can be constants.
6. The operator is used with a pointer to dereference the
address contained in the pointer.
a . address ( & )
b. assignment ( = )
c. indirection ( * )
d . pointer ( * )
e. selection ( > ) -
7. Which of the following statements will not add 1 to a variable?
a. a + + ;
b. a + = 1 ;
c. a = a + 1;
d. * p = *p + 1 ;
.
e * p++ ;
8. \ \ hich ol the following defines a pointer variable to an integer?
a . int & ptr ;
b. int & & ptr ;
c. int * Ptr ;
d . int * * ptr ;
e. intA Ptr ;
9. \ \ hich ol the following defines and initializes a pointer to the address of x?
a . int * ptr = *x ;
b. int * ptr = ;
c. int * ptr = &x;
d. int & ptr = *x;
e. int & ptr = Ax ;
10. Pointers to pointers is a term used to describe
a. any pointer that points to a variable.
b. any two pointers that point to the same
variable.
c. any two pointers that point to variables of
the same type.
d. pointers used as formal parameters in a function header.

e. pointers whose contents are the address of another pointer.
1 1 . Given the following definitions,
int x ;
int* P = &x ;
int** pp = &p;
which answer can access the value stored in x ?

a. p
b. pp
c. & p
.
d * pp
e. ** pp
12. Given the definitions shown below,
int i;
float f ;
int* pd ;
float* pf ;
which answer is not valid ?

a. i = 5;
b. f = 5;
c. pd = & i;
d. pf
e. pd
-= &f ?
pf ?
13. Which of the following statements about pointer compatibility is true?
a . Because all pointers are addresses, pointers to different types can be
assigned without a cast .
b. If pointers of different types are assigned , C uses an implicit cast .
c. When a pointer is cast , C automatically reformats the data to reflect
the correct type.
d . When a void pointer is dereferenced , it must he cast.
14 . Which of the following operators does not require an lvalue as its operand ?
a. address operator ( & total )
.
b assignment ( x = ...)
c . indirection ( * ptr )
d. postfix increment ( x + + )
.
e prefix increment ( + + x )
1 5. Which of the following is not an attribute of quality software ?
a . Meets operating standards of organization .
b. Runs efficiently on anv hardware .
c. Satisfies the user’s explicit requirements .

d . Satisfies the user’s implicit requirements.
e. Well documented .
Exercises
16. Declare and define the following:
a . A pointer variable pi pointing to an integer
b . A pointer variable ppi pointing to a pointer to an integer
.
c A pointer variable pf pointing to a float
d . A pointer variable ppc pointing to a pointer to a char
17. If a is declared as integer, which of the following statements is true and
which is false?
a . The expression * & a and a are the same.
b. The expression * & a and & * a are the same.
18. Given the following declarations:
int x;
double d;
int* p;
double* q;
\\ hich ol the following expressions are not allowed ?

.
a p = &x;
b. p = &d ;
c. q = &x ;
d . q = &d ;
e. p = x ;
19. Given the following declarations:
int a = 5;
int b 7;
int * p = &a ;
int * q = & b;
\ \ hat is the value of each of the following expressions?

a. ++a
b . ++( * p)
-
c. - (*q )
d. -- b
20. What is the error ( il any ) in each of the
a. int a = 5 ;
b. int* p = 5;
c. int a ;
int* p = &a ;
d. int a;
int** q = &a ;
21 . \\ hich
ot the following program segments is valid ? Describe each error in
the invalid statements. 1'
a. int* p ;
scant("%d " , &p);
b. int* p;
scanf( " %d " , &* p );
c. int* p ;
scant( "%d " , * p);
d . int a ;
int* p = &a;
scant( "%d " , p );
22. Which ol the following program segments has a logical error in it ?
a. int** p ;
int* q ?
q = &P;
b. int** p;
int* q ;
p = &q ;
c. int** p;
int** q ;
p = &q ;
d. char c = ' A ’ ;
char** p ;
char* q ;
q = & c;
.
printf("%c " , * p);
int*** p;
What is the type of each of the following expressions?

.
a p
b. * p
c. ** p
d. ***p
is an
24. If p is a name of a variable, which of the following expressions
lvalue and which one is an rvalue ? Explain .
a. p
b. * p
.
c &p
d, * p + 2
. i an array, which of the foi -
25 If p and q are pointer variable names and a ls
lowing expressions are not syntactically correct , because they violate the
rules concerning lvalues and rvalues?
a. *p = *p + 2 ;
b. &p = &a [ 0 ] ;
.
c q = &(p + 2);
.
d a [ 5 ] = 5;
26. Write a function prototype statement for a function named calc that
returns void and contains a reference parameter to iwt , x , and a reference
parameter to a lotto double , y .
27. Write a function prototype statement for a function named spin that
returns a pointer to an integer and contains a reference parameter to an
int , x , and a pointer parameter to the address of a long double , py .
28. Assuming all variables are integers, and all pointers are typed appropri-
ately, show the final values of the variables in Figure 9- 28 after the fol -
lowing assignments.
a p;
s * * p;
t p;
b = ** r ;
* *q = b;
HJ- R
a —a Q5
«
——
sQ
t
a
•
22
32
61
|
FIGURE 9- 28 Exercise 28
29. Assuming all variables are integer, and

all pointers are typed appropri -
ately, show the final values of the variables in Figure 9- 29 alter the fol -
lowing assignments.
Chapter 9 Pointers 60 /
t = **p;
b = ***q;
*t * c ;
v = r;
w = *s;
a **v
*u = *w
PQ—
—
a 40
b 5 r
E3 Q s
c 30 t
1
191 18
[T
y
20
> I
VU w
FIGURE 9 - 29 Exercise 29
.
30 In the following program , show the configuration of all the variables and
the output.
# include <stdio.h>
int main (void )
{
int a;
int* p;
int** q ;
// Statements
a = 14 ;
p = &a ;
q = & p;
printf( " %d\n" , a);
printf( " %d\n " , * p);
printf( " %d\n ", * * q ) ;
printf( " % p\t" , p);
printfC % pM , q ) ;
return 0;
} // main
Problems
31. Write a function that converts a Julian date to a month
and day. A Julian
relative January 1 . For
date consists of a year and the day of the year to
example, day 41 is February' 10. The month and day are to he stored in
integer variables whose addresses are passed as parameters. The function
is to handle leap years. ( For a discussion of the calculation of a leap year,
see Chapter 5 , Project 60.) If there is an error in a parameter, such as a
day greater than 366, the function will return zero. Otherwise, it returns
a positive number.
.
32 Modify' Program 9 - 3 to include subtraction, multiplication, division, and
remainder.
.
33 Write a function that receives a floating-point number representing the
.
change from a purchase The function will pass hack the breakdown of the
change in dollar hills, half-dollars, quarters, dimes, nickels, and pennies.
.
34 Write a function that receives a floating-point number and sends hack
the integer and fraction parts.
35. Write a function that receives two integers and passes back the greatest
common divisor and the least common multiplier. The calculation of the
greatest common divisor can he done using Euclid’s method of repetitively
dividing one number by the other and using the remainder (modulo).
When the remainder is zero, the greatest common divisor has been
found. For example, the greatest common divisor of 247 and 39 is 13, as
.
in Table 9 - 4
numl num2 Remainder

247 39 13
39 (ged ) 13 0
TABLE 9- 4 Greatest Common Divisors
Once you know the greatest common divisor ( ged), the least corn-
mon multiplier (1cm ) is determined as shown below.
|cm _ ( numl x num 2 )

ged
Projects
.
36 \ \ iite a program that creates the structure
shown in Figure 9- 30 and
then reads an integer into variable a and
prints it using each pointer in
turn. I hat is, the program must
read an integer into variable a and print
it using p, q, r, s, t, u, and v.
.
37 Write a program that creates the
structure shown in Figure 9 - 31. It then
.
re ids data into a, b, and c
using the pointers p, q, and r .
FIGURE 9 - 30 Structure for Project 36
P q
5)
a b
a c
FIGURE 9 - 31 Data Structure for Project 37
After the data have been read, the program reassigns the pointers so
thatp points to c, q points to a, and r points to b. After making the reas-
signments, the program prints the variables using the pointers. For each
variable, print both its contents and its address.
38. Write a program that creates the structure shown in Figure 9 - 32 and
reads data into a and b using the pointers x and y.
FIGURE 9- 32 Data Structure for Project 38

The program then multiplies the value of a by b and stores the result
in c using the pointers x , y , and z . finally, it prints all three variables
using the pointers x , y , and z .
39. Write an exploratory program to study pointers. In this program , prompt
the user to enter a character, an integer, a long integer, and a floating-
point number. The program then calls a print function that accepts a void
pointer as its only parameter and prints the address of the pointer fol-
lowed by the data , first as a character, then as an integer, a long integer,
and a floating- point number. Since void pointers have no type , you will
need to cast the pointer appropriately in each of your print statements.
A proposed output format is shown below’. It is the result of calling
the print function w ith a pointer to integer. You will most likely get differ-
ent results. Call the print function for each of the user inputs, each time
printing the results shown below. (1 he first line may be printed before
calling the print function .)
Printing int 123

Printing data at location: 01D70C44
Data as char
Data as int 123
Data as long 8069185
Data as float 0.00
Using the addresses that you printed , build a memory map of the
variables as they existed in your program. Then , by analyzing the results
of the printout , write a short explanation of what happened when your
program ran . To fully understand the results , you may have to refer to
your system documentation to determine the physical memory sizes of
the various types.
40. Assume an environment in which transactions and responses are being
sent across a network. When a transaction is sent , a time stamp is cre-
ated and attached to the transaction. When the response is received , the
time is noted . Using an integral time stamp, as shown in Program 9-8,
write a function to determine the elapsed time in hours, minutes, and
seconds between the transaction submittal and its response. Assume that
no transaction takes longer than 24 hours. Create a test driver to validate
your I unction . lest the function with at least five test cases, including
one that spans midnight and one that is longer
than 24 hours.
Chapter 10
Pointer Applications
In this chapter we explore two basic uses of pointers: with arrays and in
dynamic memory.
We begin with the use of pointers in implementing arrays. This discus -
sion includes the important topic of pointer arithmetic, which allows us to
process data in an array by adding or subtracting from pointer addresses.
Following a discussion of passing arrays to functions, we discuss one of
the most powerful aspects of C, dynamic memory. We examine two different
approaches to dynamic memory, static allocation and dynamic allocation.
We conclude with two applications, sorting with pointers and ragged
arrays.
Objectives
To understand the relationship between arrays and pointers
To understand the design and concepts behind pointer arithmetic
To write programs using arrays and pointer arithmetic
J To better understand the design behind passing arrays to functions
To understand the C implementation of dynamic memory
To write programs using static and dynamic memory allocation
To understand and implement ragged arrays ( arrays of pointers)
611
612 Section 10.1 Arrays and Pointers
10.1 Arrays and Pointers

The name of an array is a pointer constant to the first element. Because the
array’s name is a pointer constant , its value cannot be changed. Figure 10- 1
shows an array with the array name as a pointer constant.
a[ 0 ] [ a
a[ 1 ] [
a[ 2 ] [
The name of an array is a
a[ 3 ] [ pointer constant to its first
V element v
a[ 4 ] [
a
FIGURE 10 - 1 Pointers to Arrays
Since the array name is a pointer constant to the first element , the
address of the first element and the name of the array both represent the
same location in memory. We can , therefore, use the array name anywhere
we can use a pointer, as long as it is being used as an rvalue. Specifically, this
means that we can use it with the indirection operator. When we dereference
an array name, we are dereferencing the first element of the array; that is, we
are referring to array [ 01 . However, when the array name is dereferenced , it
refers only to the first element , not the whole array.
same
a - &3 [ 0 ] .
a is a pointer only to the first element—not the whole array.
Prove this to yourself by writing a program with the code shown below.
1 he code prints the address of the first element of the array ( & a [ 0 ] ) and the
array name , which is a pointer constant . Note that we have used the format
conversion code for a pointer ( % p ).
// Demonstrate that array name is pointer constant

int a [ 5];
printf("% p % p" , &a[0], a);
We cannot tell you the values that this code will print , hut they will be
addresses in your computer. Furthermore, the first printed address ( the first
element in the array ) and the second printed address ( the array pointer) will
be the same , proving our point.
The name of an array is a pointer

constant; it cannot be used as an lvalue .
A simple variation on this code is to print the value in the first element
ol the array using both a pointer and an index. This code is demonstrated in
1 igure 10- 2 . Note that the same value, 2 , is printed in both cases , again prov-
ing our point that the array name is a pointer constant to the beginning ol
the array.
his element is called <stdio.h>
^
a[0] or * a ^x întinclude
{
main ( void )
int a[ 5 ] = {2 ,4,6 ,8 , 22};
a[ 0 ] 2 a printf( " %d %d " , *a, a[ 0]);
a[ 1 ] 4
return 0;
a[ 2 ] 6 } // main
a[ 3 ] 8
a[ 4 ] 22
2 2
a
FIGURE 10- 2 Dereference of Array Name
Let ’s investigate another point . If the name oF an array is really a pointer,

let ’s see il we can store this pointer in a pointer variable and use it in the
same way we use the name of the array. The program that demonstrates this
is shown in Figure 10-3.
#include <stdio.h>
int main ( void )
a[ 0 ] {
int a[5] = {2, 4, 6 , 8, 22 ) 2 2
a[ 1 ] int* p a; ^
a[ 2 ]
printf( "%d %d \n " , a[0 ), * p);
a[ 3 ]
a[ 4 ] return 0;
} // main
FIGURE 10 - 3 Array Names as Pointers
Right after we define and initialize the array, we define a pointer and ini -
tialize it to point to the first element of the array by assigning the array name.
Note especially that the array name is unqualified ; that is, there is no address
operator or index specification . We then print the first element in the array,
first using an index notation and then pointer notation .
Let’s look at another example that explores the close relationship
between an array and a pointer. We store the address of the second element
of the array in a pointer variable . Now we can use two dilferent names to
614 Section 10.2 Pointer Arithmetic and Arroys
accesseach element. This does not mean that we have two arrays; rather it .
shows that a single array can be accessed through different pointers.
include <stdio.h>
^
int main ( void )
{
int a[5 ] = {2, 4 , 6, 8 , 22};
a[ 0 ] int* p;
a[ 1 1
a[ 2 ] P = &a[ 1 ];
a[ 3 ] This is p[0] printf(" %d % d " , a [ 0 ] , p[-1 ]);
a[ 4 ] printf( " \ n " );
a printf( "%d %d " , a [ l ) , p [ 0 ] ) ;
} // main
FIGURE 10 - 4 Multiple Array Pointers
Figure 10- 4 demonstrates the use of multiple names for an array to refer-
ence different locations at the same time. First, we have the array name. We
then create a pointer to integer and set it to the second element of the array
( a [ 1 ] ). Now, even though it is a pointer, we can use it as an array name and
index it to point to different elements in the array. We demonstrate this by
printing the first two elements using first the array name and then the
pointer. Note especially that , when a pointer is not referencing the first ele-
ment ol an array, the index can he negative [ - 1 ] . This is shown in the refer-
ence to p [ -1 ] .
To access an array, any pointer to the first element can be used instead of
the name of the array.
10.2 Pointer Arithmetic and Arrays

Besides indexing, programmers use another powerful method ol moving
through an array: pointer arithmetic. Pointer arithmetic offers a restricted
set ol arithmetic operators for manipulating the addresses in
pointers. It is
especially powerful when we need to move through an array from element to
element , such as when we are searching an array sequentially.
Pointers and One- Dimensional Arrays

II we have an array, a, then a is a constant the first element and
pointing to
a + 1 is a constant pointing to the
second element . Again, il we have a
pointer , p, pointing to the second element of an
array ( see Figure 10- 4), then
P - 1 is a pointer to the previous ( first ) element and p + 1 is a pointer to the
next ( third) element , furthermore,
given a, a + 2 is the address two elements
Irom a , and a + 3 is the address three elements from a . We can generalize the
notation as follows:
Given pointer, p, p ± n is a pointer to the value n elements away.
It does not matter how a and p are defined or initialized ; as long as they
are pointing to one of the elements of the array, we can add or subtract to get
the address ol the other elements of the array. This concept is portrayed in
Figure 10- 5 .
a 2 P “ 1
a + 1 4 P
a + 2 6 P + 1
a + 3 8 p + 2
a + 4 22 «
* p + 3
a
FIGURE 10 - 5 Poinfer Arithmetic
But the meaning of adding or subtracting here is different from normal

arithmetic. When we add an integer n to a pointer value, we get a value that
corresponds to another index location , n elements away: In other words, n is
an offset from the original pointer. To determine the new value, C must know
the size of one element. The size of the element is determined by the type of
the pointer. I bis is one of the prime reasons that pointers of different types
cannot he assigned to each other.
If the offset is 1 , then C can simply add or subtract one element size from
the current pointer value. This may make the access more efficient than the
corresponding index notation . If it is more than 1 , then C must compute the
offset by multiplying the offset by the size of one array element and adding it
to the pointer value. This calculation is shown below.
address = pointer + ( offset * size of element)
Depending on the hardware, the multiplication in this formula can make

simply adding 1 , and the efficiency advantage of pointer
it less efficient than
arithmetic over indexing may be lost .
a+n
<
a + n * (sizeof ( one element ))

!
616 Section 10.2 Pointer Arithmetic and Arrays
We see the result of pointer arithmetic on different-sized elements in

Figure 10 -6 . For char , which is usually implemented as 1 byte, adding 1
moves us to the next memory address ( 101 ) . Assuming that integers are
4 bytes, adding 1 to the array pointer b moves us 4 bytes in memory ( 104 ).
Finally, assuming the size of float is 6 bytes, adding 1 to array pointer c moves
us 6 bytes in memory ( 106 ). In other words, a + 1 means different things in
different situations.
a 100 b 100
a +1 101
a+2 102
b+1 104
memory
addresses
b+2 108
char a[3];
int b[3];
fV
float c[ 3];
FIGURE 10 - 6 Pointer Arithmetic and Different Types
We ’ve seen how to get the address of an array element using a pointer
and an offset , now let ’s sec how we can use that value . We have two choices:
First , we can assign it to another pointer. This is a rather elementary opera -
tion that uses the assignment operator as shown below.
p = aryName + 5 ;
Second , we can use it with the indirection operator to access or change the
value ol the element we are pointing to. This possibility is seen in Figure 10-7.
a[0J or * ( a + 0 ) a
a[1 ] or *(a + 1 ) a + 1
a[2l or * ( a + 2 ) a + 2
a [ 3] or *(a + 3) a + 3
a[4] or *(a + 4 ) a + 4
FIGURE 10 - 7 Dereferencing Array Pointers
lor practice, lets use pointers to find the smallest number among
integers stored in an array. Figure 10- 8 tracks the code as it works its way
through the array.
vm
The following expressions are identical .
* (a + n) and a [ n ]
We start with the smallest pointer (pSm) set to the first element of the
array. T he functions job is to see if any of the remaining elements are smaller.
Since we know that the first element is not smaller than itself , we set the
working pointer ( pWalk) to the second element. The working pointer
advances through the remaining elements, each time checking the element it
is currently addressing against the smallest to that point ( pSm). If the current
element is smaller, its location is assigned to pSm.
Arithmetic Operations on Pointers

Arithmetic operations involving pointers are very limited . Addition can be
used when one operand is a pointer and the other is an integer. Subtraction
can he used only when both operands are pointers or when the first operand
is a pointer and the second operand is an index integer. We can also manipu -
late a pointer with the postfix and unary increment and decrement operators.
All of the following pointer arithmetic operations are valid :
p+ 5 5 + p P - 5 pl - p2 p++ —p
When one pointer is subtracted from another, the result is an index rep-
resenting the number of elements between the two pointers. Note, however,
that the result is meaningful only if the two pointers are associated with the
same array structure.
The relational operators ( such as less than and equal ) are allowed only il
both operands are pointers of the same tvpe . Two pointer relational expres-
sions are shown below.
p i >= p2 p i != P2
The most common comparison is a pointer and the N U L L constant , as

shown in Table 10 - 1 .
Long Form Short Form
if (ptr == NULL) if (!ptr)

if (ptr != NULL) if (ptr)
TABLE 10 - 1 Pointers and Relational Operators

~~
ary 32 U Q PSm psm is smallest , pwaik is walked

It tracks the It moves to
J pWalk smallest value. find smallest.
22
^ pLast
After initialization
ary 3? ary-
PSm pSm
^QpWalk u ^ rjPWalk
22 22
20 20
r* pLast - Fj PLast
After first iteration After second iteration
ary 32 pSm ary-
TT-
22
pWalk
nz / m
22
pWalk
20 pLast
PLast
After third iteration After fourth iteration
pLast = ary + arySize - 1?
for ( pSm = ary , pWalk = ary + 1 ;
pWalk <= pLast ;
pWalk+t)
if ( * pWalk < *pSm)
pSm = pWalk ;
FIGURE 10- 8 Find Smallest
f
Using Pointer Arithmetic
In this section, we write two small programs that demonstrate moving
through arrays using pointers .
Printing an Array
Program 10- 1 prints an array, first by adding I to advance through the array
and then subtracting 1 to print it backward.
PROGRAM 10 - 1 Print Array with Pointers

1 /* Print an array forward by adding 1 to a pointer. Then
2
3
print it backward by subtracting one .
Written by ;
continue*
1 M
PROGRAM 10- 1 Print Array with Pointers (continued )

4 Date:
5 */
7
8 #define MAX SIZE 10
9
11 {
13 int ary[ 1 = {1 / 2, 3, 4, 5, 6, 7, 8 , 9, 10};
14 int * pWalk ;
15 int * pEnd ;
16
17 // Statements
18 // Print array forward
19 printf( " Array forward : " );
20 for ( pWalk = ary , pEnd = ary + MAX SIZE; _
21 pWalk < pEnd ;
22 pWalk++ )
23 printf ( " % 3d " , * pWalk );
24 printf ( "\n" );
25
26 // Print array backward
27 printf ( " Array backward: " );
28
29
for ( pWalk = pEnd -
1 ; pWalk >= ary ; pWalk
printf ( " %3d " , * pWalk );
— )
30 printf ( " \n" );

31
32 return 0;
33 } // main
Results:
Array forward : 1 2 3 4 5 6 7 8 9 1 0
Array backward: 1 0 9 8 7 6 5 4 3 2 1
Program 10 - 1 Analysis While this program is quite simple, it does demonstrate one efficiency point. Note
that before we printed the forward loop, we set pEnd . Many programmers would
use the following test in the limit condition:
pWalk < ary + MAX _SIZE
While this logic works, some compilers will calculate the end address in
each loop, an obvious inefficiency. We recommend, therefore, that whenever
possible you calculate the end pointer in the loop initialization.
Searching with Pointers

The logic in Program 10- 1 works well for the sequential search hut not for
the binary search . 1 Recall that the binary search requires the calculation of
the index or the address of the entry in the middle of a table. When we wrote
the program using indexes, the calculation of the midpoint was done with the
statement shown below.
mid = ( first + last ) / 2 ;
Since we cannot use addition with two pointers, this formula will not
work. We need to come up with the pointer arithmetic equivalent . Another
formula determines the midpoint in an array by calculating the number of
elements from the beginning of the array. I bis method , known as the offset
method , is shown below.
mid = first + ( last - first) / 2;

The offset calculation works with pointers also. The subtraction of the
first pointer from the last pointer will give us the number of elements in the
array. The offset from the beginning of the array is determined by dividing
the number of elements in the array by 2 . We can then add the offset to the
pointer lor the beginning of the list to arrive at the midpoint. The pointer
code is shown below.
midPtr = firstPtr + ( lastPtr - firstPtr ) / 2;
The pointer implementation of the binary search is shown in Program 10-2.
PROGRAM 10 - 2 Pointers and the Binary Search

1 / * == == == ====== binary Search===
2 Search an ordered list using Binary Search
4 endPtr is pointer to largest element in list
5 target is value of element being sought
6 Post FOUND: locnPtr pointer to target element
7 return 1 ( found )
8 !FOUND: locnPtr = element below or above target
9 return 0 (not found )
10 */
11 int binarySearch ( int list[], int* endPtr ,
12 int target , int** locnPtr )
13 {
continuei
1 . We discussed the binary search in Program 8- 14 in Chapter 8.

^1!
V 'l
PROGRAM 10- 2 Pointers and the Binary Search (continued )

15 int* firstPtr;
16 int* midPtr ;
17 int* lastPtr ;
18
19 // Statements
20 firstPtr = list;
21 lastPtr = endPtr ;
22 while ( firstPtr < = lastPtr )
23 {
24
25
midPtr = firstPtr + ( lastPtr
if (target > * midPtr)
firstPtr ) / 2;-
26 // look in upper half
27 firstPtr = midPtr + 1 ;
28 else if (target < *midPtr )
29 // look in lower half
30 lastPtr = midPtr 1?
31 else
32 // found equal: force exit
33 firstPtr = lastPtr + 1 ;
34 } // end while
35 * locnPtr = midPtr;
36 return ( target == * midPtr );
37 > // binarySearch
Program 10 - 2 Analysis Although the code in this function is relatively simple, the coding for locnPtr merits
some discussion. In the calling function, locnPtr is a pointer to the found location.
To store the pointer in locnPtr, therefore, we need to pass a pointer to a pointer to
an integer ( see statement 12). To create a pointer to pointer, the calling function must
pass the address of its location pointer.
In addition to demonstrating the subtraction of two pointers, we also see the use of
a relational operator with two pointers in statement 22.
Pointers and Two - Dimensional Arrays

The first thing to notice about two-dimensional arrays is that , just as in a one -
dimensional array, the name of the array is a pointer constant to the first ele-
ment of the array. In this case , however, the first element
is another array!
Assume that we have a two -dimensional array of integers . When we derefer-
ence the array name , we don ’t get one integer , we get an array of integers. In
other words, the dereference of the array name of a two - dimensional array is
a pointer to a one-dimensional array. Figure 10 - 9 contains a two - dimensional
array and a code fragment to print the array.
.
Each element in the figure is shown in both index and pointer notation
Note that tablet 0 ] refers to an array of four integer values . The equivalent
pointer notation is the dereference of the array name plus 0, * ( table + 0)

which also refers to an array of four integers.
table [ 0 ] is identical to * ( table + 0 )
To demonstrate pointer manipulation with a two-dimensional array, let’s

print the table in Figure 10-9. To print the array requires nested for loops.
When dealing with multidimensional arrays, however, there is no simple
pointer notation . To refer to a row, we dereference the array pointer, which
gives us a pointer to a row. Given a pointer to a row, to refer to an individual ele-
ment , we dereference the row pointer. This double dereference is shown below.
*(*( table) )
table ] [
table [0] or * ( table + 0 )
3
table+1 ( [
table [ 1 ] or * ( table + 1 )
table+2
](
table[2 ] or * ( table + 2 )
int table[ 3 ][ 4 ] ;
for ( i = 0; i < 3; i++ )

{
for ( j = 0; j < 4; j++)
printf( " %6d " , *( *( table + i ) + j));
printf( " \n " );
} // for i
Print Table
FIGURE 10 - 9 Pointers to Two - dimensional Arrays
lUit the previous expression refers only to

the first element of the first
row. Io step through all the elements, we
need to add two offsets, one for the
row and one lor the element within
the row. We use loop counters, i and j ,
as the offsets. I b i s is the same
logic we saw when we printed a two-dimen -
sional array using indexes. To print an element
, we use the array name,
table , and adjust it w i t h the loop
indexes. This gives us the relatively com -
plex expression shown below.
*(*( table + i ) + j)
u
I his pointer notation is equivalent to the index syntax, table [ i ] [ j ] .

With multidimensional arrays , the pointer arithmetic has no efficiency advan -
tage over indexing. Because the pointer notation for multidimensional arrays
is so complex and there is no efficiency advantage , most programmers find it
easier to use the index notation.
We recommend index notation for two- dimensional arrays.
10.3 Passing an Array to a Function

Now that we have discovered that the name of an array is actually a pointer to
the first element , we can send the array name to a function for processing.
When we pass the array, we do not use the address operator. Remember, the
array name is a pointer constant , so the name is already the address of the
first element in the array. A typical call would look like the following:
dolt (aryName );
The called program can declare the array in one of two ways. I irst , it can
use the traditional array notation . This format has the advantage of telling the
user very clearly that we are dealing with an array rather than a single pointer.
This is an advantage from a structured programming and human engineering
point of view.
int dolt ( int ary[ ])
We can also declare the array in the function header as a simple pointer.
The disadvantage to this format is that , while it is technically correct , it actu -
ally masks the data structure ( array ). For one-dimensional arrays , it is the
code of choice among professional programmers.
int dolt (int* arySalary)
If you choose to code this way, use a good descriptive name for the
parameter to minimize any reader confusion. The function documentation
should also indicate clearly that an array is being passed.
Note, however, that when we pass a multidimensional array, we must use
the array" syntax in the header declaration and definition . The compiler needs
the offset for
to know the size of the dimensions after the first to calculate
pointer arithmetic . Thus , to receive a three- dimensional array, we would use
the following declaration in the function’s header:
float dolt (int bigAry[][12][5])

7'
624 Section 10.3 Possing an Array to a Function
To see how passing array pointers works, let ’s write a program that calls
a function to multiply each element of a one-dimensional array by 2. The
program s variables are shown in Figure 10- 10 , and the code is shown in
Program 10 - 3.
main
multiply
5 |
pAry pWalk pLast size
FIGURE 10- 10 Variables for Multiply Array Elements By 2
PROGRAM 10 - 3 Multiply Array Elements by 2

1 /* Read from keyboard & print integers multiplied by 2.
2 Written by:
3 Date:
4 */
5 include <stdio.h >
^
6 # define SIZE 5
7
9 void multiply ( int* pAry , int size );
10
1 1 int main ( void )
12 {
14 int ary [SIZE ];
15 int* pLast ;
16 int* pWalk;
17
18 // Statements
19 pLast = ary + SIZE i;
20 for ( pWalk = ary ; pWalk < pLast
21
= ; pWalk++)
{
22 printf( "Please enter an integer "
: );
continue
IIP
PROGRAM 10- 3 Multiply Array Elements by 2 (continued )

23
24
scanf ( "%d", pWalk );
} // for
1
25
26 multiply ( ary , SIZE);
27
28 printf ( "Doubled value is: \n " );
29 for ( pWalk = ary ; pWalk < = pLast ; pWalk++ )
30 printf ( " % 3d " , *pWalk );
31
32 return 0;
33 > // main
34
35
36 Multiply elements in an array by 2
37 Pre array has been filled
38 size indicates number of elements in array
39 Post values in array doubled
40 */
41 void multiply ( int* pAry , int size )
42 {
44 int* pWalk ;
45 int * pLast ;
46
47 // Statements
48 pLast = pAry + size 1; -
49 for ( pWalk = pAry ; pWalk <= pLast ; pWalk++ )
50 *pWalk = *pWalk * 2;
51 return ;
52 } // multiply
53 ===== End of Program ====
Results:
Please enter an integer: 1
Doubled value is:
2 4 6 8 10
Program 10 - 3 Analysis This program contains several points of interest. First, we have declared the array in
have
the function declaration using the more common pointer notation, but we
given
( ).
it a name that indicates it is a pointer to an array pAry
626 Section 10.3 Passing an Array to a Function
In the multiply function, we use a separate pointer ( pWalk) to walk through the list.
We could have used pAry, since it was not being used other than to identify the begin-
ning of the array. All too often, however, this type of " shortcut" saves a line or two of
code only to create hours of debugging when the program is changed later. As we
have pointed out before, do not use formal parameters as variables unless their intent is
to change a value in the calling program. This rule is especially important when the
parameter is a pointer, as in this case.
Finally, note that we have passed the size of the array to the multiply function. We
still need to know how much data we need to process, and we use the size to calculate
the address of the last element in the list. As a variation on the limit test, however, we
could have passed a pointer to the last element of the array, &ary [ SIZE - 1 ] . This
would save the calculation of pLast. From a style and efficiency point of view, neither
method has an advantage over the other. The structure and needs of other parts of the
program usually dictate which method is used.
Understanding Complicated Declarations

To help you read and understand complicated declarations, we have developed the
right-left rule. Using this rule to interpret a declaration, you start with the identifier in
the center of a declaration and "read" the declaration by alternatively going right
and then left until all entities have been read. The basic concept is shown below.
identifier
t t t t t t3 t
6 4 2 start here 1 5
Consider the simple declaration

int x;
This is read as "x is an integer.' ^
int X ;
T T T
2 0 1
Since there is nothing on the right, we simply go left.
Now consider the example of a pointer declaratiion. This
"p is
example is read as
a pointer to integer."
int ;
T t T T
4 2 0 1 3
Note that we keep going right even when there
is nothing there until all the entities on
_
the left have been exhausted. For a more
complete discussion of complex declara-
tions, see Appendix L.
a The box ( ) is just a place holder to show that there

ignore it when you read each declaration. is no entry to be considered. Simply
L
^
10.4 Memory Allocation Functions

C gives us two choices when we want to reserve memory locations for an
object : static allocation and dynamic allocation. Figure 10- 1 I shows the char -
acteristics of memory allocation.
Memory
Allocation
Static Dynamic
FIGURE 1 0- 1 1 Memory Allocation
Memory Usage
To understand how dynamic memory allocation works, we must study how
memory is used. Conceptually, memory is divided into program memory and
.
data memory Program memory consists of the memory used for main and all
.
called functions Data memory consists of permanent definitions, such as glo -
bal data and constants, local declarations, and dynamic data memory Exactly .
how C handles these different needs is a function of the operating system and
the compiler writer ’s skills. We can, however, generalize the concepts.
.
Obviously, main must he in memory at all times Beyond main, each
called function must he in memory only while it or any ol its called functions
are active. As a practical matter, most systems keep all functions in memory
while the program is running.
Although the program code for a function may he in memory at all
times, the local variables for the function are available only when it is active .
Furthermore, more than one version of the function can he active at a time.
( See the discussion of recursion in Chapter 6.) In this case, multiple copies
of the local variables arc allocated, although only one copy of the function is
.
present The memory facility for these capabilities is known as
the stack
memory.
In addition to the stack, a memory allocation known as the heap is avail-

able. Heap memory is unused memory allocated to the program and avail -
able to he assigned during its execution. It is the memory pool from which
functions.
memory is allocated when requested by the memory allocation
conceptual view of memory is shown in Figure 10 - 12.
Phis
It is important to recognize that this is a conceptual view of memory
. As we
the software engineers who
said before, implementation of memory is up to
the heap from
design the system. For example, nothing prevents the stack and
.
sharing the same pool of memory In fact, it would he a good design concept
.
628 Section 10.4 Memory Allocation Functions
main function
Program Memory
global heap stack
Data Memory
Memory
FIGURE 10- 12 A Conceptual View of Memory
Static Memory Allocation

Static memory allocation requires that the declaration and definition of
memory be fully specified in the source program. The number of bytes
reserved cannot he changed during run time. This is the technique we have
used to this point to define variables, arrays, pointers, and streams.
Dynamic Memory Allocation

Dynamic memory allocation uses predefined functions to allocate and
release memory ( or data while the program is running. It effectively post-
pones the data definition, but not the data declaration, to run time.
pointer.
We can refer to memory allocated in the heap only through a
lo use dynamic memory allocation, we use either standard data types or

derived types that we have previously declared. Unlike static memory alloca-
tion, dynamic memory allocation has no identifier
associated with it; it has only
an address that must be used to access
it . Io access data in dynamic memory;
therefore, we must us a pointer. Ihis concept is shown in Figure 10- 13.
Memory Allocation Functions

lour memory management functions are used with dynamic memory. Three ol
them, malloc, calloc, and realloc, are used for
memory allocation. The fourth,
free , is used to return memory when it is
no longer needed. All the memory'
management functions are found in the
standard library file ( stdlib .h ). The
collection ol memory functions is shown in Figure
10 - 14.
ary
int x; x int ary[ 3 );
Stack Stack
(a) Static Memory Allocation
tv
int* x ; int* ary ;
x = malloc(...); ary = calloc(...);
X
I
Stack Stack
(b) Dynamic Memory Allocation
FIGURE 10 - 13 Accessing Dynamic Memory
Memory
Management
malloc calloc realloc free
FIGURE 10- 14 Memory Management Functions
Block Memory Allocation (malloc )

The malloc function allocates a block of memory that contains the number ol
bytes specified in its parameter. It returns a void pointer to the first byte ol
the allocated memory The allocated memory is not initialized. We should
.
therefore assume that it will contain unknown values and initialize it as
required by our program.
The malloc function declaration is shown below.
void *malloc (size t size);

_
The type, sizej , is defined in several header files including stdio.h . The
it is guaranteed to be
type is usually an unsigned integer, and by the standard
large enough to hold the maximum address ol the computei .
630 Section 10.4 Memory Allocation Functions
To provide portability, the size specification in malloc's actual parameter

is generally computed using the sizeof operator. For example, if we want to
allocate an integer in the heap, we code the call as shown below.
pint = malloc (sizeof ( int));
As mentioned above, malloc returns the address ol the first byte in the
memory space allocated. However, if it is not successful , malloc returns a
NULL pointer. An attempt to allocate memory from the heap when memory
is insufficient is known as overflow. It is up to the program to check for mem-
ory overflow. If it doesn’t , the program produces invalid results or aborts with
an invalid address the first time the pointer is used .
Exactly what action should he taken when memory overflow is encoun -
tered is application dependent . If memory might he released by another por-
tion of the program , the memory request can be held. Generally, however, the
programmer must terminate the program and allocate more memory' to
the heap.
Memory Allocation Casting

Prior to C99, it was necessary to cast the pointer returned from a memory allocation
function. While it is no longer necessary, it does no harm as long as the cast is cor-
rect. If you should be working with an earlier standard, the casting format is:
pointer = (type* ) malloc(size)
The malloc function has one more potential error. If we call malloc with
a zero size, the results are unpredictable. It may return a N U L L pointer, or it
may return some other implementation -dependent value . Never call malloc
with a zero size .
Figure 10- 1 5 shows a typical malloc call. In this example, we are allocat -
ing one integer object. II the memory is allocated successfully, ptr containsa
value. II it doesn t , there is no memory and we exit the program with error
'
code 100.
if ( I ( pint = malloc(sizeof( int))))\

// No memory available
|
exit ( 100) ;
// Memory available
pint
Stack
FIGURE 10- 15 malloc

m
Contiguous Memory Allocation ( calloc )

I he second memory allocation function, calloc, is primarily used to allocate
.
memory lor arrays It differs from mailoc only in that it sets memory to null
.
characters I he calloc function declaration is shown below.
__
void *calloc ( size t element-count ,
size t element size ); _
I he result is the same for both mailoc and calloc when overflow occurs
and when a zero size is given.
A sample calloc call is shown in Figure 10- 16. In this example, we allo-
cate memory for an array of 200 integers.
Ptr
P GE
if (!( ptr = ( int*)calloc ( 200, sizeof( int))))
// No memory available
exit ( 100) ;
200 integers // Memory available
FIGURE 10- 16 calloc
Reallocation Of Memory ( realloc)

The realloc function can he highly inefficient and therefore should be used
advisedly. When given a pointer to a previously allocated block of memory,
realloc changes the size of the block by deleting or extending the memory at
the end of the block. If the memory cannot be extended because of other allo -
cations, realloc allocates a completely new block, copies the existing memory
allocation to the new allocation, and deletes the old allocation. The program-
mer must ensure that any other pointers to the data are correctly changed.
The operation of realloc is shown in Figure 10 - 17 .
_
void* realloc ( void * ptr , size t newSize);
Releasing Memory [ free )

When memory locations allocated by mailoc, calloc, or i calloc arc no longer
needed, they should be freed using the predefined function free . It is an error
to free memory with a null pointer, a pointer to other than the
first element of
a n allocated block, a pointer that is a different type than
the pointer that
allocated the memory; it is also a potential error to refer to memory alter it
r 632 Section 10.4 Memory Allocation Functions
has been released . The function declaration statement for free is

shown below.
void free (void* ptr );
Before
GL
ptr
1181551331121|64 | 1 |90 | 3l [ 5 177
10 integers
ptr = realloc ( ptr , 15 * sizeof(int));
new elements
.
.not initialized
ptr
18155133|l 21 641 1 190| 31 | 5 771 ? ? ? ? | ?|
15 integers
After
FIGURE 10 - 17 realloc
Figure 10- 18 shows two examples. The first one releases a single element, allo-
cated with a malloc , hack to the heap. In the second example, the 200 elements
were allocated with calloc . When we free the pointer in this case, all 200 elements
arc returned to the heap. You should note two things in this figure. First , it is not
the pointers that are being released hut rather what they point to. Second, to
release an array of memory that was allocated by calloc , you need only release the
I
pointer once. It is an error to attempt to release each element individually.
BEFORE AFTER
ptr
a-ptr
free ( p t r ) ;
BEFORE AFTER
QH i i ITH T \wmm\
~
... mi
ptr 200 integers ptr 200 integers
free ( p t r ) ;
FIGURE 10- 18 Freeing Memory
Releasing memory does not change the value iin a pointer. It still contains
the address in the heap. It is a look error to use
the pointer after memory has
'1
been released . The program may continue to run , but the data may be
destroyed it the memory area is allocated for another use. This logic error is
very difficult to trace . We suggest that immediately after you free memory you
also clear the pointer by setting it to NULL .
Using a pointer after its memory has been released is a common program -
.
ming error Guard against it by clearing the pointer.
One final thought: You should free memory whenever it is no longer

needed . It is not necessary, however, to clear memory at the end of the program.
The operating system will release all memory when your program terminates.
The pointer used to free memory must be of the same type as the pointer
used to allocate the memory .
10.5 Array of Pointers

Another useful structure that uses arrays and pointers is an array of pointers.
This structure is especially helpful when the number of elements in the array
is variable .
-
To look at an example, Table 10- 2 is a two dimensional array in which
only one row ( 1 ) is full . The rest of the rows contain from one to four ele-
ments. This array is also known as a ragged array because the right elements
in each row may he empty, giving it an uneven ( ragged ) right border.
32 18 12 24
13 11 16 12 42 19 14
22
13 13 14
11 18
TABLE 10- 2 A Ragged Table
If we use a two dimensional array for storing these numbers, we are wast -
-
ing a lot of memory. The solution in this case -
is to create five one dimensional
arrays that are joined through an array of pointers. One implementation
of
this concept is seen in Figure 10- 19 along with the statements needed to allo -
heap. Note that table is a pointer to a pointer to an
cate the arrays in the
integer and must be declared as shown below, not as an
array.
int * * table
634 Section 10.6 Programming Applications
Program 10-6 in Section 10.6 , " Programming Applications," demon-

strates the concept in a complete program. We will show other variations on
this data structure in the next Few chapters.
table [ 0 ]
table [ 1 ]
table [ 2 ]
table [ 3 ]
table [ 4 ]
table [ 5 ]
table = ( int** )calloc (rowNum + 1 , sizeof( int* ));

table[0] = ( int * )calloc ( 4, sizeof( int));
table[ l ] = ( int* )calloc ( 7, sizeof( int ));
table( 2] = ( int * )calloc ( 1, sizeof( int ));
table[ 3 ] = ( int*)calloc ( 3, sizeof( int ));
tablet 4 ] = ( int*)calloc (2 , sizeof( int));
tablet 5] = NULL;
FIGURE 10- 19 A Ragged Array
10.6 Programming Applications

This section contains two applications. The first isi a rewrite of the selection
sort , this time using pointers . The second uses dynamic arrays .
Selection Sort Revisited

Let ’s revisit the selection sort we developed in Chapter 8. Now that we know
how to use pointers, we can improve it in several ways. First , and perhaps
most important, it is structured . I he
structure chart is shown in Figure 10- 20.
Note that main contains no detailed code. It simply calls the three Functions
tint will get the job done , first , getData reads data from the keyboard and
puts it into an array. I hen selectSort calls two functions
to sort the data .
Finally, printData displays the result , l he complete algorithm is shown in
Program 10- 4.
W
Selection Sort
(pointers)
getData selectSort printData
smallest exchange
FIGURE 10- 20 Selection Sort with Pointers — Structure Chart
PROGRAM 10- 4 Selection Sort Revisited

1 /* Demonstrate pointers with Selection Sort
2 Written by:
3 Date written:
4 */
6 # define SIZE 25
7
9 int* getData (int* pAry , int arySize );
10 void selectSort ( int* pAry , int* last);
11 void printData ( int* pAry , int* last );
12 int* smallest ( int* pAry , int* pLast );
13 void exchange ( int* current , int* smallest );
14
16
18 int ary[ SIZE ];
19 int* pLast ;
20
21 // Statements
22 pLast = getData (ary , SIZE);
23 selectSort ( ary , pLast );
24 printData ( ary , pLast );
continued
PROGRAM 10- 4 Selection Sort Revisited ( continued )

25 return 0;
26 > // main
27
28 /* ==================== getData ===== = == = =
29 Reads data from keyboard into array for sorting.
30 Pre pAry is pointer to array to be filled
31 arySize is integer with maximum array size
32 Post array filled. Returns address of last element
33 */
34 int* getData ( int* pAry , int arySize )
35 {
37 int ioResult;
38 int readCnt = 0 ;
39 int * pFill = pAry ;
40
41 // Statements
42 do
43 {
44 printf( "Please enter number or <EOF>: " );
45 ioResult = scanf( "%d " , pFill);
46 if ( ioResult == 1 )
47 {
48 pFill++ ;
49 readCnt++;
50 > // if
51 > while ( ioResult == 1 && readCnt < arySize);
52
53 printf( "\n\n%d numbers read." , readCnt);
54
55
--
return ( pFill );
} I I getData
56
57 /* ==== ==== selectSort === :
========
58 Sorts by selecting smallest element in unsorted
59 portion of the array and exchanging it with element
60 at the beginning of the unsorted list.
61 Pre array must contain at least one item
62 pLast is pointer to last element in array
63 Post array rearranged smallest to largest
64 */
65 void selectSort (int * pAry , int* pLast)
66 {
continue
\
Chapter 10 Pointer Applications 63 /
PROGRAM 10 - 4 Selection Sort Revisited ( continued )

68 int* pWalker ;
69 int* pSmallest;
70
71 // Statements
72 for ( pWalker = pAry ; pWalker < pLast ; pWalker++)
73 {
74 pSmallest = smallest ( pWalker , pLast );
75 exchange ( pWalker, pSmallest );
76 > // for
77 return;
78 } // selectSort
79
80 /* ==================== smallest ===
81 Find smallest element starting at current pointer.
82 Pre pAry points to first unsorted element
83 Post smallest element identified and returned
84 */
85 int* smallest ( int* pAry , int* pLast)
86 {
88 int* pLooker;
89 int* pSmallest;
90
91 // Statements
92 for (pSmallest = pAry, pLooker = pAry + 1;
93 pLooker <= pLast;
94 pLooker-M- )
95 if (* pLooker < *pSmallest )
96 pSmallest = pLooker;
97 return pSmallest ;
98 > // smallest
99
100 /* === ==== exchange ===
101 Given pointers to two array elements, exchange them
102 Pre pi & p2 are pointers to exchange values
103 Post exchange is completed
104 */
105 void exchange ( int* pi , int* p2 )
106 {
108 int temp;
109
110 // Statements
continued
PROGRAM 10- 4 Selection Sort Revisited (continued )

111 temp = *pl ;
112 *pl = *p2;
113 *p 2 = temp ;
114 return ;
115 > // exchange
116
117 ===== printData ====== ========
118 Given a pointer to an array , print the data.
119 Pre pAry points to array to be filled
120 pLast identifies last element in the array
121 Post data have been printed
122 */
123 void printData ( int* pAry , int * pLast )
124 {
126 int nmbrPrt;
127 int* pPrint;
128
129 // Statements
130 printf("\n\nYour data sorted are: \n");
131 for (pPrint = pAry, nmbrPrt = 0;
132 pPrint <= pLast;
133 nmbrPrt++ , pPrint++ )
134 printf ( " \n # %02d %4d " , nmbrPrt , * pPrint );
135 printf("\n\nEnd of List " );
136 return;
137 > // PrintData
138 // = End of Program
Program 10- 4 Analysis Here are a few points you should note as you study this program. Note that we have
used pointers and pointer arithmetic in all functions.
The getData function fills the array. Since the pointer, pFill , is always one
ahead of the read, when we reach the end of file, it is pointing to an empty element.
Therefore, when we return it we subtract 1.
Sf ^ , it selects
^*
< rt nc on advances through the array using a for statement. For
eachL iteration eCfS
? the smallest
' element in the unsorted portion of the array and
,
exchanges t with the first element in the unsorted portion of the . Each loop, there-
array
hpfnrp . ?m[ nes ° TecousnumêrGr unordered elements. We stop at the element just
- - *• -
® smaHest always
tests the first element and at least one
±TotlXS
element
1
*
Finally, the style used to code the for statement in both smallest and
note
,
separate tine ^ StaterTlen s are more readable if you put each expression on a
V
T1J
Dynamic Array
I his program creates a dynamic table that can store a ragged array. The col -
umn and the width of the dynamic array are tailored to the needs of the user.
1 he program starts by asking the user for the number of rows that must he
stored. After allocating the row pointers (using culloc ) the program asks for
y
the number ol entries in each row. The table is then filled with data supplied
.
by the user from the keyboard To demonstrate the applications that could be
used with this type of structure, we then determine the minimum, maximum,
and average ol each row oi data. The design is shown in Figure 10-2 I .
Dynamic
Arrays i
build
Table i fill
Table i process
Table
row
Mimimum i row
Maximum
row
Average i
smaller j larger
FIGURE 10- 21 Dynamic Array Structure Chart
The data structure is shown in Figure 10 - 22. The table pointer points to
the first pointer in an array of pointers. Each array pointer points to a second
array of integers, the first element of which is the number of elements in the
list . All arrays arc allocated out of the heap, giving us a structure that is lim-
ited only by the computer s memory.
3 12 3 45
table
5 10 6 21 78 61
K 4 2 31 40 70
FIGURE 10 - 22 Ragged Array Structure
The complete set of programs to build and fill the table and some sample
applications are shown in Programs 10- 5 through 10 - 13. To
compile these
need he put in one source file or combined using include
programs , they to
statements .
PROGRAM 10- 5 Dynamic Arrays: main

1 /* Demonstrate storing arrays in the heap. This program
2 builds and manipulates a variable number of ragged
3 arrays. It then calculates the minimum, maximum, and
4 average of the numbers in the arrays.
5 Written by:
6 Date:
7 */
10 # include <limits.h>
11
13 int** buildTable ( void );
14 void fillTable ( int ** table );
15 void processTable (int** table);
16 int smaller ( int first , int second );
17 int larger ( int first , int second );
18 int rowMinimum ( int* rowPtr );
19 int rowMaximum ( int * rowPtr );
20 float rowAverage ( int * rowPtr );
21
23 {
25 int ** table;
26
27 // Statements
28 table = buildTable( );
29 fillTable ( table);
30 processTable (table);
31 return 0 ;
32 > // main
Program 10 - 5 Analysis The main function in Program 10- 5 is a classic example

of a well - designed program.
It contains only one variable: the pointer to the
array. There are only three functions
calls in main : The first returns the address of the array; the other two use the array,
and only the array. All detail processing is done in subfunctions
.
PROGRAM 10- 6 Dynamic Arrays: buildTable
1 /* == buildTable ===============-=:=:==
2 Create backbone of the table by creating an array of
3 pointers, each pointing to an array
of integers ,
L
4 Pre nothing
continue1
PROGRAM 10- 6 Dynamic Arrays: buildTable (continued )

5 Post returns pointer to the table
6 */
7 int** buildTable ( void )
8 {
10 int rowNum ;
11 int colNum;
12 int** table;
13 int row;
14
15 // Statements
16 printf( " \nEnter the number of rows in the table: " );
17 scanf ( " %d " , &rowNum);
18 table = ( int** ) calloc( rowNum + 1 , sizeof( int* ));
19 for (row = 0; row < rowNum ; row++)
20 {
21 printf("Enter number of integers in row %d: " ,
22 row + 1 );
23 scanf ( " %d ", &colNum );
24 table[ row ] = ( int* Jcalloc(colNum + 1 ,
25 sizeof( int ));
26 table[ row ][ ]0 = colNum ;
27 > // for
28 table[row ] = NULL;
29 return table;
30 } // buildTable
We begin Program 10- 6 by asking the user how many rows of data must be entered
.
Program 10 - 6 Analysis plus
Using the calloc function, we then allocate the memory for an array of pointers
one extra pointer at the end. Each entry in the allocated table is
used to point to an
array of integers, also stored in the heap .
PROGRAM 10- 7 Dynamic Arrays: fillTable

1 /* ====== fillTable ==============
This function fills the array rows with data
,
2
3 Pre array of pointers
4 Post array filled
5 */
6 void fillTable ( int** table )
7 {
9 int row = 0;
10
continuec
r 642 Section 10.6 Programming Applications
PROGRAM 10- 7 Dynamic Arrays: f illTable ( continued )

11 // Statements
12 printf( " \n ==== ==== • );
13 printf( "\n Now we fill the table.\n" );
14 printf("\n For each row enter the data");
15 printf( " \n and press return: " );
16 printf( " \n ==== ======\n " );
17
18 while ( table[ row] != NULL )
19 {
20 printf( " \n row %d ( %d integers ) > \
21 row + 1 , table[ row ][0]);
22 for ( int column = 1;
23 column <= * table[ row];
24 column++ )
25 scanf( " %d " , table[ row] + column );
26 row++;
27 > // while
28 return ;
29 > // fillTable
Program 10 - 7 Analysis Filling the rows, Program 10 - 7 requires a while statement to loop through the
array
pointers and a for statement to enter the data. We use the while statement because
the pointer array is designed with a null pointer at the end, and we use it to tell we
are at the end of the array. We use the for statement for filling the row because the
user has already told us how many elements are in each row.
PROGRAM 10- 8 Dynamic Arrays: Process Table

1 /* = ===== processTable =====
2 Process the table to create the statistics.
3 Pre table
4 Post row statistics ( min , max , and average )
5 */
6 void processTable ( int** table)
7 <
9 int row = 0;
10 int rowMin ;
11 int rowMax;
12 float rowAve;
13
14 // Statements
15 while ( table[row] != NULL
)
16 {
L continue
PROGRAM 10- 8 Dynamic Arrays: Process Table ( continued )

17 rowMin = rowMinimum ( table[row ]);
18 rowMax = rowMaximum ( table[row ]);
19 rowAve = rowAverage ( tablet row ]);
20 printf( "\nThe statistics for row %d " , row + 1 );
21 printf( "\nThe minimum: % 5d " , rowMin );
22 printf( " \nThe maximum: % 5d " , rowMax );
23 printf( " \nThe average: %8.2f " , rowAve);
24 row++;
25 > // while
26 return;
27 > // processTable
Program 10 - 8 Analysis Processing the table in Program 10- 8 calls three functions to show how you could use
a dynamic structure such as this . Obviously, many more applications could be used.
Remember the structure of this example. You will be able to use this structure in future
applications. Programs 10- 9 through 10- 13 continue the code for the functions.
PROGRAM 10-9 Dynamic Arrays: Find Row Minimum

1 ==== rowMinimum =====
2 Determines the minimum of the data in a row.
3 Pre given pointer to the row
4 Post returns the minimum for that row
5 */
6 int rowMinimum ( int* rowPtr )
7 {
8
10 int rowMin = INT MAX ;
11
12 // Statements
13 for ( int column = 1 ; column < = *rowPtr ; column*-*- )
14 rowMin = smaller ( rowMin , *( rowPtr + column ));
15 return rowMin ;
16 > // rowMinimum
PROGRAM 10- 10 Dynamic Arrays: Find Row Maximum

1 ==== rowMaximum ===
2 Calculates the maximum of the data in a row.
3 Pre given pointer to the row
4 Post returns the maximum for that row
5 */
continued
r 644 Section 10.6 Programming Applications
PROGRAM 10- 10 Dynamic Arrays: Find Row Maximum ( continued )

int rowMaximum ( int* rowPtr )
6
7 {
9 int rowMax = INT MIN;
10
11 // Statements
12 = 1; column <= *rowPtr; column++)
for ( int column
13 rowMax = larger ( rowMax , *(rowPtr + column));
14 return rowMax ;
15 } // rowMaximum
PROGRAM 10 - 11 Dynamic Arrays: Find Row Average

1 === rowAverage ==== ========
2 This function calculates the average of data in a
row.
3 Pre pointer to the row
4 Post returns the average for that row
5 */
6 float rowAverage ( int * rowPtr )
7 {
9 float total = 0 ;
10 float rowAve;
11
12 // Statements
13 for (int column = 1; column <= *rowPtr; column++)
14 total += ( float)*( rowPtr + column );
15 rowAve = total / *rowPtr ;
16 return rowAve;
17 } II rowAverage
PROGRAM 10 - 12 Dynamic Arrays: Find Smaller

1 ===== smaller ===
2 This function returns the smaller of two numbers.
3 Pre two numbers
4 Post returns the smaller
5 */
6 int smaller ( int first, int second )
7 {
8 // Statements
9 return ( first < second ? first : second
);
10 > // smaller
^
PROGRAM 10- 13 Dynamic Arrays: Find Larger

1 /* === larger =====
2 This function returns the larger of two numbers.
3 Pre two numbers
4 Post returns the larger
5 */
6 int larger (int first, int second )
7 {
8 // Statements
9 return ( first > second ? first : second );
10 > // larger

Pointer applications need careful design to ensure that they work correctly
.
and efficiently The programmer not only must take great care in the program
design hut also must carefully consider the data structures that are inherent
with pointer applications. I he design of the data structures is beyond the
scope of this text, but we can discuss the design of the pointers.
Before we discuss specific aspects of pointer applications, a word of cau-
tion: Remember the KISS principle. The complexity of pointers grows rapidly
as you move from single references to double references to triple references.
Keep it short and simple!
Pointers and Function Calls

You should always pass by value when possible. It you have to use a pointer to
pass hack a value, however, whenever possible, pass a pointer to the ultimate
.
object to be referenced When the pointer refers to the data variable, it is a
single dereference. Despite your best efforts, at times you will have to pass a
.
pointer to a pointer When a function opens a file whose file pointer is in the
calling function, you must pass a pointer to the pointer to the file table. In
.
Figure 10- 16 we allocated a dynamic array of 200 integers If the allocation is
performed in a subfunction, then it must receive a pointer to the pointer to
the array in memory so that it can store the address of the array.
Whenever possible, use value parameters.
Pointers and Arrays

\ \ hen you combine pointers and arrays, the complexity again quickly
becomes difficult. This is especially true when the array is multidimensional.
\\ henever possible, therefore, rather than passing a multidimensional array,
pass just one row. 1 his reduces the complexity significantly, because the func-
tion is now dealing w ith a one- dimensional array. Not only are the references
easier to work with, but passing a row allows simple pointer arithmetic, which
is usually more efficient.
\ \ hen you must work with a two-dimensional array, use index rather than
pointer notation. Index notation is much simpler to work with, and there is
n difference in efficiency, li you are not sure of this recommendation, con-
° the following equivalent
sider expressions. Which one would you rather find
in a strange program?
*( *( ary + i) + j) or a [i1 [ j 1
Array Index Commutativity

C ommutativity is a principle in
mathematics that says the results of an
expression do not depend on the order in which the evaluated. For
factors are
example, a + b is identical to b + a. Pointer addition is commutative; subtrac -

tion is not. I luis, the following two expressions are identical.
a + i i + a
But we also know that a + i is identical to a [i]. Similarly, i + a would be

equivalent to i[ a ] . Therefore, using the principle of commutativity, we
see that
a[ i] is identical to i[ a ]
A word ol caution. Commutavity works in C because of the pointer con -

cept and pointer arithmetic. Do not try this in another language.
Dynamic Memory: Theory versus Practice

Do not get carried away with dynamic memory. The programming complexity
of dynamically managing memory is very high. What you will often find,
therefore, is that memory is not fully reused. To test memory reusability in
your system, run Program 10- 14. Note: I bis may not be a problem with
.
today’s compilers When we reran this program for this edition, the memory
was reused.
PROGRAM 10- 14 Testing Memory Reuse

1 /* This program tests the reusability of dynamic memory.
2 Written by:
3 Date:
4 */
6 tinclude <stdlib.h>
7
8 int main ( void )
9 {
11 int looper ;
12 int* ptr;
13
14 // Statements
15 for ( looper = 0; looper < 5 ; looper++ )
16 {
17 ptr =
malloc( 16 );
18 printf " Memory allocated at: % p\n " , ptr );
(
19
continued
PROGRAM 10- 14 Testing Memory Reuse ( continued )

20 free ( ptr);
21 > // for
22 return 0;
23 } // main
Results in Personal Computer:

Memory allocated at: 0x00e80238
Memory allocated at: 0x00e8024a
Memory allocated at: 0x00e8025c
Memory allocated at: 0x00e8026e
Memory allocated at: 0x00380280
Results in UNIX system:

Memory allocated at: 0x00300f70
Memory allocated at: 0x00300f 70
Memory allocated at: 0x00300f 70
Program 10- 14 Analysis First look at the logic of this simple program. It loops five times, each time allocating
16 bytes and immediately freeing them . The program was run on two different sys-
tems, a personal computer and a UNIX network.
If the memory management is doing its job, the same 16 bytes would be allocated
each time. What we see, however, is that for the personal computer, each address is
different. It is also interesting to note that the difference between all the allocations
except the first was 18 bytes, not 16.
The larger UNIX network system is more sophisticated, and it appears to reuse
dynamic memory efficiently.

1 . If ary is an array, then
ary is the same as &ary[0]
2. II ary is the name of an array, then
ary[i] is the same as *( ary + i)
.
3 Remember that we usually pass the name of an array as a pointer value to
a function that needs to access the elements of the array.
4 . Remember that
int* a[5]; is different from int (*a)[5];
.
5 Similar to array indexes, the most common pointer error is referencing a
nonexistent element in an array. This is especially easy to do with pointer
arithmetic.
6. It is a compile error to use pointer arithmetic with a pointer that does not
reference an array.
.
7 It is a logic error to subtract two pointers that are referencing dillerent
arrays.
8. It is a compile error to subtract a pointer from an index.
9. It is a compile error to attempt to modify the name of an array using
pointer arithmetic , such as
table++; // Error: table is constant

table = ... , // Error: table is constant
10. The header file stdlib.h is required when using memory allocation functions.
11 . It is a compile error to assign the return value from malloc or calloc
to
anything other than a pointer.
12. It is a logic error to set a pointer to the heap to NULL before the memory
has been released.
13. It is a compile error to use pointer arithmetic with multiply
, divide , or
modulo operators.
10.9 Key Terms

array of pointers pointer arithmetic
dynamic array ragged array
dynamic memory allocation right —left rule
offset stack memory
heap memory static memory allocation
overflow
10.10 Summary
Arrays and pointers have a close relationship. The name of an array is a
pointer constant to the first clement of the array.
The name of an array and the address of the first element in the array rep-
resent the same thing: an rvalue pointer.
The name of an array is a pointer only to the first element , not the
whole array.
A pointer variable to the first element of an array can be used anywhere
the name of the array is permitted , such as with an index.
In pointer arithmetic, if ptr is pointing to a specific element in an array,
ptr + n is the pointer value n elements away.
LI I he name of a two-dimensional array is a pointer to a one-dimensional
—
array the first row.
-J In a multidimensional array, the following two expressions arc equivalent .
*( *( a + i ) + j ) a[ i][ j]
J We can pass an array to a function in many ways. One way is to pass the
name of the array as a pointer.
J A
— —
ragged array that is, an array of pointers can be used to save space
when not all rows of the array are full .
I he memory in a computer can be divided into program memory and data
memory. Data memory can be partitioned into global area , heap, and stack.
Q Static allocation ol memory
requires that the declaration and definition of
memory be I ully specified at compilation time.
J Dynamic allocation of memory is done during
run time through the use of
predefined functions.
J C has lour predefined memory allocation functions: malloc , calloc , realloc ,
an d free .
J lo read and interpret a complex declaration , we can use the right-left rule.
10.11 Practice Sets
^
T |
Review Questions
1. Given a pointer to an array element , ptr, ptr - 5 is a pointer to the
value 5 elements toward the beginning of the array.
a . True
.
b False
2. Adding 1 to a pointer increases the address stored in it by 1 byte.
a . True
b. False
3. 1 he parameter declaration int * a can be used to declare an array of
integers passed to a function.
a . True
b. False
4 . Dynamically allocated memory can only be referred to through pointers.
a. True
.
b False
5. Which of the following statements about pointers and arrays is true?
.
a The following expressions are identical when ary is an array: * ary
and &ary[ 0 ] .
b. The following expressions are identical when ary is an array: *ary
and *ary[0].
c The name of an array can be used with the indirection operator to ref -
.
erence data .
.
d The name of the array is a pointer variable.
e. The only way to reference data in an array is with the index operator.
6. Which of the following statements about pointer arithmetic is true ?
a Any arithmetic operator can be used to change the value ol a pointer.
.
b Given a pointer ptr, ptr + n is a pointer to the value u elements away.
.
-
c. Pointer arithmetic is a short hand notation that changes the value that
a pointer is referencing.
d Pointer arithmetic is valid only with pointers to arithmetic variables,
.
such as pointer to integer.
e Pointer arithmetic is valid only with the name of the array.
.
?
7. Which of the following parameters is a two-dimensional array of integers
.
a int ary
.
b int ary[ ][SIZE2 ]
c. int* ary
d . int* ary[ ][SIZE2]
.
e int* ary [ SIZE2 ]
652 SectionlO . il Practice Sets
8. Which of the following is not a C memory allocation function ?

a. alloc ( )
b . calloc ( )
.
c free( )
d . malloc( )
e. realloc ( )
9. Which of the following statements about memory allocation is true?
a . Allocated memory can he referred to only through pointers; it does not
have its own identifier.
b . calloc (change allocation ) is used to change the allocation to memory
previously allocated through malloc .
c. malloc (memory allocation ) is used to allocate blocks of memory for
arrays.
d . realloc ( release allocation ) is used to release memory when it is no
longer needed .
e. Underflow can occur only with calloc and malloc .
10. The function that returns memory to the heap is:
.
a allocf )
b . callocf )
c . free( )
d. mallocf )
e. realloc { )
1 1 . Which ol the following statements about releasing memory allocation
is false?
a. It is an error to dereference a pointer to allocated memory after the
memory' has been released .
.
b It is an error to free memory with a pointer to other than the first ele-
ment ol an allocated array.
c . Memory' should be freed as soon as it is no longer needed .
d. Only one call to free is necessary' to release an entire array allocated
with calloc .
e. lo ensure that it is released , allocated memory should be freed before
the program ends.
1 2 . \ \ hich ol the following statements about ragged arrays is false?
a . Ragged arrays are two-dimensional arrays in which the right elements
of a row may be empty.
b. Ragged arrays can Ibe implemented using an array of pointers to one-
dimensional arrays.
c. Ragged arrays can only be used with
arrays of integers.
d . Ragged arrays implemented as an array of
pointers save memory.
e. Ragged arrays implemented as an
array of pointers can be created in
dynamic memory.
HI
Exercises
13. Rewrite each ot the following expressions by replacing the index operator
( [ . . . ] ) with the indirection operator ( * ) .
a. tax [ 6 ]
b. score [ 7 ]
.
c num [ 4 ]
d . prices [ 9 ]
14. Rewrite each ot the following expressions by replacing the indirection
operator ( * ) with the index operator ( [ . . . ] ). Each identifier refers to
an array.
.
a * ( tax + 4 )
b. * ( score + 2 )
.
c * ( number + 0 )
d. * prices
1 5 . Imagine you have the following declarations:
int ary[10];
int* p = &ary[3 ];
Show how you can access the sixth element of ary using the pointer p.
.
16 Given the following declarations:
float ary[200];
Write the function declaration for a function named fun that can manip-
ulate a one-dimensional array of floating- point numbers, ary . Provide an
additional parameter, a pointer to the last element in the array. Code a
call to the function .
.
1 7 Show what would be printed from the following block:
{
int num[ 5] = { 3, 4, 6, 2 , 1 > ;
int* p = num ;
int* q = num +2 ;
int* r = &num[ 1 ];
printf( " \n%d %d ", num[ 2], *( num + 2 ));
printf( "\n%d %d " , *p, *(P + 1 ));
printf( "\n%d %d " , *q , *( q + 1 )),*
printf( "\n %d %d " , *r, *(r + 1 ));
} // Block
18. Show what would be printed from the following block.
void printOne ( int* );
void printTwo ( int* );
void printThree (int* );
{
int num [ 5 ] = {3 , 4 , 6 , 2 , 1};
// Statements
printOne ( num );
printTwo ( num + 2 );
printThree (&num [2 ]);
return 0 ;
> // Block
void printOne ( int* x )
{
printf( " \n%d " , x[ 2]);
return ;
} // printOne
void printTwo ( int* x )
{
printf( " \n %d " , x[2 ]);
return ;
> // printTwo
void printThree ( int* x )
{
printf( " \n %d " , * x ) ;
return ;
} // printThree
19. Given the following definition :
int table [ 4 ][5 ];
write the function declaration for a function named fun that can accept the
whole array using a pointer. Write the statement that calls this function.
20. Given the following definition:
int table [ 4 ][ 5];
w i i t e the function declaration ( or a function named torture that

accepts one row of an array at a time.
21 . Draw pic tures to show the memory configuration for each of the follow -
ing declarations.
.
a int* x [5];
b. int ( *x ) (5);
\ \r
22. Show what would be printed from the following block:
{
int x [ 2 ][ 3 ] =
{
{ 4 , 5 , 2 > ,
{ 7 , 6 , 9 }
>;
int (*p) [3] = &x [1];
int (*q ) [3] = x;
// Statements
printf( "\n%d %d %d " , ( * P )[ 0 ],( * p )[ l ],( * p )[ 2 ] );
printf( " \n%d %d " , * q [ o ], * q [ i 1 );
} // Block
23. Show what would be printed from the following block:
{
int x [2][3] = {
{ 4 , 5 , 2 } ,
{ 7 , 6 , 9 }
>;
// Statements
fun (x);
fun (x + 1 );
return 0;
} // Block
void fun ( int ( * p )[ 3 J )
{
printf( " \n%d %d %d " , ( * P )[ 0 ],( * p )[ l ] ( * p )[ 2 ] );
/
return ;
>
24. Given the following definitions:
int num[ 26] = {23 , 3 , 5, 7 , 4, -1, 6};

int* n = num ;
int i = 2;
int j = 4;
show the value of the following expressions:

a. n
h. *n
c. *n + 1
d. * ( n + 1 )
e. *n + j
if
f. *&i
656 SectionlO. il Practice Sets
char a [ 2 0 ] {'z' = , ’x' , 'm* , 's' , 'e' , ' h ' };

c h a r * pa = a ;
int i = 2;
int j = 4;
int* pi = & i;

.
a * ( pa + j )
b. * ( pa + *p i )
.
26 Given the following definition:
i n t d a t a [ 15 ] = { 5, 2 , 3 , 4 , 1, 3, 7 , 2 , 4 , 3, 2 , 9 , 1 2 };

.
a data + 4
.
b * ( data + 4)
c. * d a t a + 4
d. * ( d a t a + ( *d a t a + 2))
.
27 Given the following definitions:
int i = 2;
int j = 4;
int* pi = & i;
i n t* pj = & j ;
show the value ol the following expressions:

a. * & j
b. *&*& j
c. * & pi
d. **& pj
e. & * * & p i
char a[ 20] = {' z', x ' , 'm ' , 's * , e * , ’ h ' };

int i = 2;
int j = 4;
write pointer expressions that evaluate to the same value as each of the
following:
a. a [ 0 ]
b. a [ 5 ]
c . The address ol the element just
before a [ 0 ] .
mm
'
d . The address of the last element in a .

.
e I he address of the element just after the last element in a .
f . The next element after a [ 3 ] .
.
g The next element after a [ 12 ] .
h . The next element after a [ j ] .
int num[ 10]

int i = 2;
= -
{23,3,5,7 ,4 , 1,6 ,12, 10 ,-23};
int j = 4 ;
write index expressions that evaluate to the same value as each oi the
following:
a. *( num + 2)
b. *(num + i + j)
c. *( num + *( num + 1 ))
d. * ( num + j)
e. *( num + i) + *( num + j)
.
30 Given the following definition :
int num[2000 ] = {23, 3 , 5, 7 , 4,
write two pointer expressions for the address ol num [ 0 ] .

3 1 . Given the follow ing definitions:
int num[ 26] = {23 , 3, 5, 7, 4, -1, 6 > ;

int i 2;
int j = 4;
int* n = num;
write the equivalent expressions in index notation .

a n .
b *n.
c. *n + 1
.
d * ( n + 1)
.
e *(n + j)
int num[ 26] = {23, 3, 5,

int* pn ;
write a test to check whether pn points beyond

the end of num.
658 SeclionlO .il Practice Sets
.
33 Given the following function for mushem and the definitions shown below
int mushem (int*, int*);

int i = 2;
int j = 4;
int* pi = &i ;
int* pj = &j;
indicate whether each of the following calls to mushem is valid:

.
a i = mushem ( 2 , 10 );
b. j= mushem (& i, & j);
c. i =mushem ( pi, & j);
d. j =mushem ( i , j);
.
e mushem (pi, pj);
.
34 What is the output from the following program?
#include <stdio.h>
int fun ( int* , int , int* );
int main ( void )
{
int a = 4 ;
int b = 17 ;
int c[5] = {9 , 14 , 3, 15, 6 > ;
// Statements
a = fun(&a , b, c );
printf( " 2. %d %d %d %d %d %d %d \n" ,
a , b, c[0], c[ l ], c[2], c(3], c[4]);
return 0;
> // main
int fun (int* px , int y , int* pz )
{
int a = 5;
int* p;
// Statements
printf( " 1. %d %d %d \n" , * px, y , * pz ;
)
for ( p = pz; P < pz + 5. ++p )
*p = a + *p;
return (* px + * pz + y );
> // fun
.
35 \ \ hat is the output Irom the
following program?
# include <stdio.h>
int sun ( int* , int , int* );
continued
Til
int main ( void )

{
int a = 4;
int b = 17;
int c[5) = {9 , 14 , 3, 15, 6 > ;
int* pc = c;
// Statements
a = sun( pc , a, &b);
printf( "2. %d %d %d %d %d %d %d \n " ,
a, b, c[0], c[ l ], c[2], c[ 3 ], c[4]);
return 0;
} // main
int sun (int* px, int y , int* pz)
{
int i = 5;
int* p;
// Statements
printf( " 1. %d %d %d \n " , * px , * pz );
for ( p = px ; p < px + 5; p++)
* p = y + *p ;
* px = 2 * i;
return (* pz + * px + y );
> // sun
Problems
36. We have two arrays, A and B, each containing 10 integers. Write a Function
that checks if every element of array A is equal to its corresponding element
in array B. In other words, the function must check if A[0 ] is equal to B[ 0 ] ,
A[ 1 ] is equal to B[ 1 ] , and so on. The function must accept only two pointer
values and return a Boolean, true for equal and false for unequal.
.
37 Generalize the function in Problem 36 to include the number of ele -
ments to be compared as a parameter.
38 Write a function that reverses the elements of an array in place. In other

.
words, the last element must become the first, the second Irom last must
become the second, and so on. The function must accept only one
pointer value and return void .
.
39 The Pascal triangle can be used to compute the coefficients of the terms
in the expansion of ( a + b ) . Write a function that creates a ragged array-
n
representing the Pascal triangle. In a Pascal triangle, each element is the

sum of the element directly above it and the element to the left of the ele-
ment directly above it ( if any). A Pascal triangle of size 7
is shown in the
.
JjJjl
following
,
wr 660 SectionlO. il Practice Sets
1
1 1
1 2 1
1 3 3 1
1 4 6 4 1
1 5 10 10 5 1
1 6 15 20 15 6 1
*
Your function must be able to create the triangle of any size . The func-
tion should accept an integer representing the size of the triangle and
return a pointer to the array it created .
40. Write a function that tests an International Standard Book Number

( ISBN ) to see if it is valid . The ISBN is used to define a hook uniquely. It
is made of 10 digits, as shown below. For an ISBN to he valid , the
weighted sum of the 10 digits must be evenly divisible by 11. The 10 th
digit may be X, which indicates 10. ( If you are not familiar with the algo-
rithm for the weighted sum , it is explained in Chapter 8 , Problem 31 ).
The ISBN format is shown in Figure 10- 2.3.
Publisher Check digit
0-07-881809- :
i
Country Book Number
;
FIGURE 10 - 23 Project 40 Structure
I he function must accept a pointer value ( the name of the array ) and
return a Boolean , true for valid and false for invalid .
41. W rite a Junction that copies a one-dimensional array of n elements into a

two-dimensional array of j rows and k columns. The resulting array will
he placed in the heap. The data will he inserted into the array in row
order ; that is , the first k items will he placed in row 0 , the second k items
in row 1 , and so lorth until all rows have been filled .
— —
If j and k are not factors of n that is , if n j * k the function returns a
*
null pointer. Otherwise , it returns the pointer to the two-dimensional
i array. I he input array and j, k , and n will he passed as parameters.
i i r
n
. .
42 C iven the following definition:
int num[20];
1
and using only pointer notation, write a for loop to read integer values
from the keyboard to fill the array.
.
43 Given the following definition
char a[ 40];
and using only pointer notation, write a for loop to read characters from
the keyboard to fill the array.
.
44 Given the following declaration and definition:
char a[6] = {'z', ' x ' , 'm', 's', e ' , ' h' };
and using only pointer notation, write a loop to rotate all values in a to
the right ( toward the end) by one element.
.
43 Repeat Problem 44, with the rotation one element to the left.
. -
46 Write a function named addem with two call-by address integer parame -
ters. The function will add 2 to the first parameter and 5 to the second
parameter. Test the function by calling it with the values of I 7 and 23 . i
Projects
.
47 Write a program that reads integers from the keyboard and place them in
.
an array The program then will sort the array into ascending and
descending order and print the sorted lists. The program must not
change the original array or create any other integer arrays.
The solution to this problem requires two pointer arrays as shown in
the example in Figure 10- 24. The first pointer array is rearranged so that
it points to the data in ascending sequence. The second pointer array is
rearranged so that it points to the data in descending sequence.
26
14 14
57
#
-
33
41
C
-/ >
^
—
41
Ascending Descending
Before Sorting After Sorting
FIGURE 10 - 24 Sample Structure for Project 47

- f
662 SectionlO.il Practice Sets
Your output should be formatted with the three arrays printed as a

10 3.
vertical list next to each other, as shown in Table -
Ascending Original Descending

14 26 57
26 14 41
33 57 33
41 33 26
57 41 14
TABLE 10- 3 Format for Project 47
.
48 Write a program that creates a two-dimensional array in the heap and
then analyzes it to determine the minimum, maximum, and average of
each column.
The data are to he read from a file. The first two elements are the
number of rows in the array and the number ol columns in each row. The
file data for a 1 2 x 8 array are shown below.
12 8
838 758 113 515 51 627 10 419
212 86 749 767 84 60 225 543
89 183 137 566 966 978 495 311
367 54 31 145 882 736 524 505
394 102 851 67 754 653 561 96
628 188 85 143 967 406 165 403
562 834 353 920 444 803 962 318
422 327 457 945 479 983 751 894
670 259 248 757 629 306 606 990
738 516 414 262 116 825 181 134
343 22 233 536 760 979 71 201

336 61 160 5 729 644 475 993
49 Rewrite the straight insertion sort from Chapter 8 using pointer arith -
.
.
metic The data to he sorted are to he read from a file. I he array is
dynamically allocated in the heap after reading the file to determine the
number of elements. While reading the data to determine the size of
array you will require, print them 10 integers to a line. Use the test data
shown below.
838 758 113 515 51 627 10 419 212 86

749 767 84 60 225 543 89 183 137 566
966 978 495 311 367 54 31 145 882 736
524 505 394 102 851 67 754 653 561 96
628 188 85 143 967 406 165 403 562 834
353 920 444 803 962 318 422 327 457 945
479 983 751 894 670 259 248 757 629 306
606 990 738 516 414 262 116 825 181 134
343 22 233 536 760 979 71 201 336 61
The data will he sorted as they are read into the array. Do not fill the array
and then sort the data . After the array has been sorted , print the data
again using the same format you used for the unsorted data .
50. Write a program to answer inquiries about student data . Using a menu -
driven user interface, provide the capability to print out the scores, aver-
age, or grade lor a student . A fourth menu option is to provide all infor-
mation about a given student . All array functions are to receive the array
as a pointer and use pointer arithmetic.
The data in Table 10- 4 will he stored in a two-dimensional array.
1234 52 7 100 78 34
1947 45 40 88 78 55
2134 90 36 90 77 30
2877 55 50 99 78 80
3124 100 45 20 90 70
3189 22 70 100 78 77
4532 11 17 81 32 77
4602 89 50 91 78 60
continued
TABLE 10 - 4 Student Data for Project 50
664 SectionlO.il Practice Sets

5405 11 11 0 78 10
5678 20 12 45 78 34
6134 34 80 55 78 45
6999 0 98 89 78 20
7874 60 100 56 78 78
8026 70 10 66 78 56
9893 34 9 77 78 20
TABLE 10 - 4 Student Data for Project 50 ( continued )
.
51 Contract bridge is a popular card game played by millions of people
throughout the world. It began in the 1920 s as a variation of an old
.
English card game, whist In bridge, the entire deck is dealt to four play-
ers named North, South, East, and West. In tournament bridge, teams of
players (North- South versus East - West ) compete with other players using
.
the same hands ( sets of 13 cards) Today it is common for large tourna-
ments to use computer-generated hands. Write a program to shuffle and
deal the hands for one game.
To simulate the bridge deck, use an array of 52 integers initialized
from 1 to 52. To shuffle the deck, loop through the array, exchanging the
current card element with a random element ( use the random number
generator discussed in Chapter 4 ) . After the deck has been shuffled,
print the hands in four columns using the player ’s position as a heading.
Use the following interpretation for the cards suits:
1 to 13 Clubs
14 to 26 Diamonds
27 to 39 Hearts
40 to 52 Spades
To determine the rank of the card, use its number (modulo 13 ) + l. The
interpretation of the rank follows: 1 is an ace, 2 through 10 have their
value, 1 1 is a jack, 12 is a queen, and 13 is a king.
.
the h in M
n turn
p
^*K
’ ^^
13 cards tIle or North, the next 13 cards
*
Then use a Unction to sort each hand
II
'
Chapter 11
Strings
II is impossible to write a well- structured and human-engineered program
without using strings. Although you probably weren’t aware of it, you have
been using strings ever since you wrote your first C program .
Some languages, such as Pascal, Ada, and C++ provide intrinsic string
types, hut C does not In C, the programmer implements strings. Because
.
strings are so important, however, functions to manipulate them have been
defined in an ad hoc standard library.
In this chapter we first consider how strings are defined and stored, and
then we explore the standard string f unctions that are available in C.
Objectives
To understand design concepts for fixed -length and variable- length strings
To understand the design implementation for C - language delimited strings
To write programs that read, write, and manipulate strings
To write programs that use the string functions in the string library
IJ To write programs that use arrays of strings
To write programs that parse a string into separate variables
To understand the software engineering concepts of information hiding and
cohesion
665
666 Section 11.1 String Concepts
11.1 String Concepts

In general, a string is a series of characters treated as a unit. Computer sci-
ence has long recognized the importance of strings, but it has not adapted a
.
standard for their implementation We find, therefore, that a string created in
Pascal differs from a string created in C.
Virtually all string implementations treat a string as a variable- length
.
piece of data Consider, for example, one of the most common of all strings, a
. .
name Names, by their very nature, vary in length It makes no difference if
we are looking at the name of a person, a textbook, or an automobile .
Given that we have data that can vary in size, how do we accommodate
them in our programs ? We can store them in fixed-length objects, or we can
.
store them in variable-length objects I bis breakdown ol strings is seen in
Figure 11- 1 .
string
fixed
length
i variable
length
length delimited
controlled
C Strings
FIGURE 11 - 1 String Taxonomy
Fixed- Length Strings

When implementing a fixed -length string format, the first decision is the
size ol the variable. If we make it too small, we can't store all the data. II we
I, make it too big, we waste memory.
Another problem associated with storing variable data in a fixed-length
data structure is how to tell the data from the nondata. A common solution is
to add nondata characters, such as spaces, at the end of the data. Of course,
this means ithat the character selected to represent the nondata value cannot
be used as data.
Variable- Length Strings

A much preferred solution is to create a structure that can expand and con-
tract toaccommodate the data. Thus, to store a person’s name that consists
ol only one letter, we would provide only enough
storage lor one character.
1
.
I The shortest name that we are aware of is (). To accommodate the computers of credit card
and other companies, however, Mr. () was Forced to .
legally change his name to Oh
Chapter 11 Strings 667
Io store a person ’s name that consists of 30 characters, the structure would

he expanded to provide storage lor 30 characters.
I his flexibility does not come without a cost, however. There must he
some way to tell when we get to the end of the data. Two common techniques
are to use length-controlled strings and delimited strings.
Length-Controlled Strings
Length-controlled strings add a count that specifies the number ol charac -
ters in the string. I his count is then used by the string manipulation functions
to determine the actual length of the data.
Delimited Strings
Another technique used to identify the end of the string is the delimiter at
the ends of delimited strings. You are already familiar with the concept of
delimiters, although you probably don’t recognize them as such. In English,
each sentence, which is a variable-length string, ends with a delimiter, the
period. Commas, semicolons, colons, and dashes are other common delimit -
ers found in English.
I he major disadvantage of the delimiter is that it eliminates one charac -
ter from being used for data. The most common delimiter is the ASCII null
character, which is the first character in the ASCII character sequence ( \ 0 ) .
This is the technique used by C.
Figure 1 1 - 2 shows length-controlled and delimited strings in memory.
length delimiter
9 C O N F U C I U S B 0 O L E \0
Length-controlled string Delimited string
FIGURE 11 - 2 String Formats
11.2 C Strings
is delimited by the null
A C string is a variable-length array of characters that
character. Generally, string character s are selected only from the printable
, any character , other than the
character set . Nothing in C however , prevents
fact quite common to use
null delimiter, from being used in a string. In is
, it
formatting characters, such as tabs , in strings.
C uses variable-length, delimited strings.

668 Section 11.2 C Strings
Storing Strings
In C , a string is stored in an array of characters. It is terminated by the null
character ( ' \ 0 • ) • Figure 1 1 -3 shows how a string is stored in memory. What
precedes the string and what follows it is not important. What is important is
that the string is stored in an array of characters that ends with a null delim-
iter. Because a string is stored in an array, the name of the string is a pointer
to the beginning of the string.
ieginning of end of string

string delimiter
H e I I o \0
FIGURE 11 -3 Storing Strings
Figure 1 1 - 4 shows the difference between a character stored in memory

and a one-character string stored in memory. The character requires only one
memory location . The one-character string requires two memory locations:
one for the data and one for the delimiter. The figure also shows how an
-
empty string is stored. Empty strings require only the end -of string marker.
char 'H' string "H" ( Empty String)
H H \0 \0
FIGURE 1 1 - 4 Storing Strings and Characters

\
The String Delimiter

At this point, you may he wondering, “ Why do we need a null character at the
end of a string: I he answer is that a string is not a data type but a data struc-
ture. This means that its implementation is logical , not physical . The physical
structure is the array in which the string is stored . Since the string, by its def -
inition , is a variable-length structure, we need to identify the logical end of
the data within the physical structure.
Looking at it another way, if the data are not variable in length , then we
don ’t need the string data structure to store them . They are easily stored in an
array, and the end of the data is always the last element in the array. But , if
the data length is variable, then we need some other way to determine the
end of the data.
I he null character is used as an end -of -string marker. It is the sentinel
used by the standard string functions. In other words, the null character at
the end lets us treat the string as a sequence of objects (characters ) with a
defined object at the end that can be used as a sentinel. Figure 1 1 - 5 shows
the difference between an array of characters and a string.
H e I I o \0 H e l l o
end-of -string array —no

character end of string
FIGURE 1 1 - 5 Differences Between Strings and Character Arrays
Because strings are variable- length structures, we must provide enough

room for the maximum - length string we will have to store, plus one for the
delimiter. It is possible that the structure will not he filled , so we can have an
array with the null character in the middle. In this case, w'e treat the part of the
array from the beginning to the null character as the string and ignore the rest .
In other words, any part of an array of characters can he treated as a string as
long as the string ends in a null character. This is shown in Figure I I -6.
Part of the array,

but not part of the
- string
^
char str[ ll ]; G o o d D a y ? )
FIGURE 11 -6 Strings in Arrays
String Literals
— —
A string literal or as it is also known , string constant is a sequence of char-
acters enclosed in double quotes. For example , each of the following
is a
string literal :
"C is a high level language."

-
" Hello"
"abed "
When string literals are used in a program , C automatically

creates an
array of characters, initializes it to a null-delimited
string , and stores it ,
because we use the double quotes
remembering its address. It does all this
that immediately identify the data as a string value .
A string literal is enclosed in double quotes.

r i
I
670 Section 11.2 C Strings
Strings and Characters

When all we need to store is a single character, we have two options: We can
store the data as a character literal or as a string literal. To store it as a char-
acter literal, we use single quote murks. To store it as a string literal, we use
double quote murks . Although the difference when we code the literal is only
a shift -key operation on most keyboards, the difference in memory is great.
The character occupies a single memory location. Ihe data portion of the
string also occupies a single memory location, but there is an extra memory
location required for the delimiter.
The differences in the ways we manipulate the data are even greater. For
example, moving a character from one location to another requires only an
assignment. Moving a string requires a function call. It is important, there-
fore, that you clearly understand the differences. Figure 1 1 - 7 shows examples
of both character literals and string literals.
nII
a "a "
an empty
a character a string
string
FIGURE 11 -7 Character Literals and String Literals
Another important difference between a string and a character is how we

represent the absence of data. Technically, there is no such thing as an empty
character. Logically, we often specify that a space ( ' ' ) or a null character
( ' \ 0 ' ) represents the absence of data. Since the character exists in all cases,
however, both of these concepts require that we program for the interpreta-
tion of no data.
tl A string, on the other hand, can he empty. That is, since it is a variable-
length structure, a string can exist with no data in it . A string that contains no
data consists ol only a delimiter. I his concept is specified in the definition of
a string and is programmed into all the string-handling functions. We can,
therefore, move or compare an empty string without knowing that we arc
dealing with no data. An example of a null string is also shown in Figure 11-7.
Referencing String Literals

\ stiing literal is stored in memory.
Just like any object stored in memory, it
has an address. I bus, we can refer to a string literal by using pointers.
Fet s first examine addressing a string literal. I he literal, since it is an
anay ol characters, is itsell a pointer constant to the first element of the
stiing. Generally, when we use it, we are referring to the entire string. It is
possible, however, to refer to only one of the characters in the string, as
shown in Figure 1 1 - 8.
Tv
1
#include <stdio.h>
"Hello" [ 0 ] H -"Hello"
int main (void) "Hello"! 1 l e "
{ "Hello"[ 2 ] i FTe
printf( " %c\n " , " Hello "( 1 ); "Hello "! 3 ]
return 0;
"Hello"! 4 ] O
} // main
"Hello"! 5 ] \0
FIGURE 1 1 - 8 String Literal References
Declaring Strings
As we said, C has no string type. To declare a string, therefore, we must use one
of the other available structures. Since strings are a sequence of characters, it is
only natural that the structure used to store string variables is the character array.
In defining the array to store a string, we must provide enough room for
the data and the delimiter. The storage structure, therefore, must he one byte
larger than the maximum data si/e. A string declaration for an eight -character
string, including its delimiter, is shown below.
char s t r [ 9 ] ;
As we have seen, string declaration defines memory lor a string when it is

declared as an array in local memory ( the stack). I lowever, we can also declare a
string as a pointer. When we declare the pointer, however, memory is allocated
only for the pointer; no memory is allocated for the string itself. In this case, we
must allocate memory for the string either dynamically or using a string
literal.
Figure I 1 -9 demonstrates two different ways to declare and define a
string. Let’s examine each carefully. In the first case (Figure 1 l -9a), memory
is allocated for future characters. The name of the string is a pointer
con -
stant . We don’t need to worry about allocating memory because the declara -
tion is for an array and memory allocation is automatic . We can read data into
the string and change data in the string as necessary.
// Local Declarations str

char str( 9 ];
(a) String Declaration
char* pStr ;
pStr Q >
(b) String Pointer Declaration
FIGURE 11 - 9 Defining Strings

6/2 Sectional .2 Strings
The second case ( Figure 1 1 -9 b ) allocates memory for a pointer variable.

In this case, however, no memory is allocated for the string itself . Before we
can use the string in any way we need to allocate memory for it . Any attempt
to use the string before memory' is allocated is a logic error that may destroy
memory contents and cause our program to fail.
Memory for strings must be allocated before the string can be used.
Initializing Strings
We can initialize a string the same way that we initialize any storage structure
by assigning a value to it when it is defined. In this case, the value is a string
literal . For example, to assign " Good Day to a string, we would code
char str[9 ] = "Good Day " ;
Since a string is stored in an array of characters , we do not need to indi-

cate the size of the array if we initialize it w hen it is defined . For instance we
could define a string to store the month January, as shown below .
char month[] = "January " ;
In this case, the compiler will create an array of 8 bytes and initialize it
with January and a null character. We must he careful, however, because
month is an array. II we now tried to store " December " in it , we would over-
run the array and destroy whatever came after the array. This example points
out one ol the dangers of strings: We must make them large enough to hold
the longest value we will place in the variable.
C provides two more ways to initialize strings. A common method is to
assign a string literal to a character pointer, as shown below. This creates a
string for the literal and then stores its address in the string pointer variable,
pStr. To clearly see the structure , refer to Figure 11 10.
-
char* pStr = "Good Day" ;
We can also initialize a string as an array of characters. This method is

not used too often because it is so tedious to code. Note that in this example ,
we must ensure that the null character is at the end of the string.
char str[9] =
i|
-
{’ G' ,'o' ,'o' , d ' , ' f ' D ' ' a ' ' y ' , ' \ 0 ' };
j j
I he structures created by these three examples are shown in Figure 11 10 * *

HI
month J aTn uTa r y \0
"Good Day" — |G|o|o 1 d D a y \0
str — ! BHHEMBBnEl
FIGURE 11 - 10 Initializing Strings
Strings and the Assignment Operator

Since the string is an array, the name of the string is a pointer constant. As a
pointer constant, it is an rvalue and therefore cannot be used as the left oper-
and of the assignment operator. This is one of the most common errors in
writing a C program ; fortunately, it is a compile error, so it cannot affect our
program.
char strl[6 ] = " Hello" ;

char str2[6];
str2 = strl ; // Compile error
Although we could write a loop to assign characters individually, there is

a better way. C provides a rich library of functions to manipulate strings ,
including moving one string to another. We discuss this library in the section
“String Manipulation Functions."
Reading and Writing Strings

A string can be read and written . C provides several string functions tor input
"String Input/Output
and output . We discuss them in the following section ,
Functions."
11.3 String Input/Output Functions

C provides two basic ways to read and write strings. First ,
we can read and
write strings with the formatted input /output functions , scanf / fscanf and
- only functions, get
printf / fprintf . Second , we can use a special set of string
string ( gets / fgets ) and put string ( puts / fputs .
) 2
for wide characters. They function the same except

2 . C provides a parallel set of read functions
for the type.
:
674 Section 11.3 String Input/Output Functions
Formatted String Input/Output

In this section, we cover the string-related portions of the formatted input
and output functions.
Formatted String Input: sconf/ fsconf

We have already discussed the basic operations of the format input functions
( see Chapter 7 ). However, two conversion codes pertain uniquely to strings,
and we discuss them here.
The String Conversion Specification

We read strings using the read- formatted function ( scan/ ). The conversion
code for a string is “ s . " The scanf functions then do all the work for us. First ,
they skip any leading whitespace. Once they find a character, they read until
they find whitespace, putting each character in the array in order. When they
find a trailing whitespace character, they end the string with a null character.
The whitespace character is left in the input stream. Io delete the whitespace
from the input stream, we use a space in the format string before the next
conversion code or FLUSH the input stream, whichever is more appropriate.
The conversion specification strings can use only three options fields,
flag, maximum field si/e, and size.
The string conversion code( s) skips whitespace.
flag
Like all inputs, the only flag that can he used is the asterisk ( * ), which dis-
cards the data read.
Maximum Field Size

\\ hen present, the maximum field size specifies the maximum number of
characters that can he read.
Size
.
II no si / e option is specified, we read normal characters To read wide charac-
ters, we use size 1 ( ell ).
Examples
For example, to read a string, such as month, from the keyboard, we could
simply write the statement shown below.
s c a n f ( " % s " , month ) ;
An address operator is not required for month, since it is already a

.
pointer constant In fact , it would be an error to use one. The only thing ve
'
\
1
need to worry about is to make sure that the array is large enough to store all
the data. If it isn ’t , then we destroy whatever follows the array in memory.
I herelore, we must make sure we don ’t exceed the length of the data . Assum -
ing that month has been defined as
char month[ 10 ];
we can protect against the user entering too much data by using a width in
the field specification . ( Recall that the width specifics the maximum number
.
ol characters to be read ) The modified scanf statement is shown below.
scanf( " %9s" , month );
Note that we set the maximum number of characters at nine while the
array size is ten . This is because scanf will read up to nine characters and
then insert the null character. Now, if the user accidentally enters more than
nine characters, the extra characters will he left in the input stream . But this
can cause a problem. Assuming that the data are being entered as a separate
—
line that is , that there is only one piece of data on the line we use the —
—
preprocessor-defined statement, FLUSH see Chapter 7 , to eliminate any
extra characters that were entered . I bis function also flushes the newline
that is left in the input stream by scanf when the user correctly enters data.
The complete block of code to read a month is shown in Program 11 - 1 .
PROGRAM 11 - 1 Reading Strings

1 { // Read Month
2 #define FLUSH while ( getchar( ) 1 = \n ' )
3 char month[ 10 ];
4
5 printf( "Please enter a month , ")?
6 scanf( " %9s" , month );
7 FLUSH ;
8 > // Read Month
The Scon Set Conversion Code (t J )

scan set conver-
In addition to the string conversion code, we can also use a
specification consists of
sion code to read a string. The scan set conversion
edit characters , and terminated hv the
the open bracket ( [ ) , followed by the ,
identify the valid characters
close bracket ( 1 ). The characters in the scan set
set . that arc to he allowed in the string . All characters
known as the scan
except the close bracket can he included in
the set .
Edited conversion reads the input stream as a string. Each character read
the scan set . If the character just read is
by scanfIfscanf is compared against
it is placed in the string and the scan continues. The first
in the scan set ,
the read. The nonmatching
character that does not match the scan set stops
6 / 6 Section 11.3 String Input/Output Functions
character remains in the input stream for the next read operation If the first .
character read is not in the scan set , the scanf / fscanf terminates and a null
string is returned .
A major difference between the scan set and the string conversion codes
is that the scan set does not skip leading whitespace. Leading whitespace is
either put into the string being read when the scan set contains the corre-
sponding whitespace character, or stops the conversion when it is not .
The edit set does not skip whitespace.
In addition to reading a character that is not in the scan set , there are
two other terminating conditions. First , the read will stop if an end -of -file is
.
detected Second , the read will stop if a field width specification is included
and the maximum number of characters has been read .
For example, lets assume we have an application that requires we read a
string containing only digits, commas, periods , the minus sign , and a dollar
sign ; in other words, we want to read a dollar value as a string. No other char-
acters are allowed . Let 's also assume that the maximum number of characters
in the resulting string is 10. The format string for this operation would be3
s c a n f ( " %10 [ 0123456789 . str) ;
Sometimes it is easier to specify what is not to be included in the scan set

rather than what is valid hor instance, suppose that we want to read a whole
,
line . We can do this by stating that all characters except the newline ( \ n ) are
valid . To specify invalid characters, we start the scan set with the caret ( ) A
symbol . I he caret is the negation symbol and in effect says that the following
characters are not allowed in the string. ( If you know UNIX, this should
sound lamiliar. ) Io read a line , we would code the scanf as shown below.
scanf ( "%81[ Wn ]" , l i n e ) ;
In this example, scanf reads until it finds the newline and then stops.
Note that we have again set the width ol the data to prevent our string, line,
Irom being overrun . We would never use this code, however. As we see in the
next section , an intrinsic string function does it for us.
I or the last example, let ’s read a 1 5-character string that can have any
character except the special characters on the top of the keyboard . In this
case , we again specify w hat is not valid . This conversion code is shown in the
following example.
s c a n f ( " %1 5 [ v ~ ! @ # $ % *&* ( ) + j
/
. f str) ;
3. UNIX users note that the dash ( - ) does not have

the same meaning in the scan set that it
does in UNIX.
i
r
^l
Note that the caret can he included in the scan set , as long as it is not the
first character. Similarly, the closing bracket is treated as text as long as it is
the first character after the opening bracket. However, if it follows the nega -
tion symbol ( ) , then it ends the scan set . This means that the close bracket
A
cannot he used when the scan set is being negated . The following example
reads a string containing brackets and digits only.
scanf( " % 15[][0123456789 ]" , str );
Always use a width in the field specification when reading strings.
Formatted String Output: printf/ fprintf

Formatted string output is provided in the print/ and / print/ f unctions. I hey
use the same string conversion codes that we used for string input .
C has four options of interest when we write strings using these print
functions: the left-justify flag, width , precision , and size. I he lelt - justify flag (- )
and the width are almost always used together.
Justification Flog
The justification flag ( - ) is used to left justify the output . It has meaning only
when a width is also specified , and then only if the length of the string is less
than the format width . Using the justification flag results in the output being
left justified , as shown below. If no flag is used , the justification is right .
printf( "|% 30s|\n", "This is the string " );

-
Output:
|This is the string
Minimum Width
The width sets the minimum size of the string in the output . If it is used with-
out a flag, the string is printed right justified as shown below
.
s \n", "This is the string");

" %30|
printf(|
Output:
This is the string|
Precision
is the minimum number of
Strings can get quite long. Because the width
characters that can be printed , a long string can easily exceed the width spec-
ified , which destroys column formatting . Therefore , C also uses the precision
that will be written. In the
option to set the maximum number of characters
following typical example we set the maximum characters to one less than
the width to ensure a space between the string and the next column .
printf( "|% 15.14s|", " 12345678901234567890" );

-
Output:
|12345678901234 |
The maximum number of characters to be printed is specified by the preci -

sion in the format string of the field specification .
Size
If no size option is specified , we write normal characters. To write wide char-
acters , we use size 1 ( ell ) .
String Input and Output Examples
EXAMPLE 11 - 1 Read Part of Line

When data are stored in secondary files , the data we need are often only a
part of a file . Program 11 - 2 demonstrates how to use the scan set to read part
of a line and discard the rest . In the following example, the input contains
numeric data , with each line containing two integers and a float. We are
interested only in the second integer ; the rest of each line will be discarded.
As we wrote the program , it could contain any type or amount of data after
the second integer. It is important to note, however, that if there are not two
integers at the beginning of the line , the program fails.
PROGRAM 11 - 2 Demonstrate String Scan Set

1 /* Read only second integer.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
11 int amount;
12 FILE* spData ;
13
14 // Statements
15 if (!( spData = -
fopen ( "P 11 03.TXT" , " r " ) ) )
continued
PROGRAM 1 1 - 2 Demonstrate String Scan Set (continued )

16 {
17 printf( " \aCould not open input file.Xn " );
18 exit ( 100);
19 } // if
20 // Read and print only second integer
21 while ( fscanf( spData ,
22 " %*d%d %* p\n]" , &amount ) != EOF )
23 printf( "Second integer: %4d \n " , amount );
24
25 printf( "End of program\n " );
26 fclose ( spData);
27 return 0;
28 > // main
Input:
123 456 7.89
987 654 3.21
Results:
Second integer: 456
Second integer: 654
End of program
Program 1 1 - 2 Analysis As long as two numbers are at the beginning of each line, this program works just
fine. It skips the first number, reads the number into amount, and then discards the
rest of the line. Study the flags contained in the format string in statement 22 carefully.
It consists of only three field specifications. The first one reads and discards an integer.
The second one reads amount. The third one considers the rest of the line as a long
string; fscanf reads until it finds a newline. Because we are not interested in these data
for the first and third parts of the line, we used the suppress flag to discard them.
EXAMPLE 11 - 2 Delete Leading Whitespace from Line

A similar problem occurs when a file contains leading whitespace at the begin-
ning of a line. In Program 11-3, we first determine if there is a w hitespace char-
acter at the beginning of the line. If there is, we delete all leading spaces using
the scan set and then read the rest of the line. Its only limitation is that the non-
space portion of the line cannot he longer than 80 characters.
PROGRAM 11 - 3 Delete Leading Whitespace

1 /* Delete leading spaces at beginning of line.
2 Written by:
3 Date:
4 */
continued
> 80 Section 11.3 String Input/Output Functions
PROGRAM 11 - 3 Delete Leading Whitespace ( continued )

7
8 int main ( void )
9 { PRC
11 char line[ 80 ];
12
13 // Statements
14 printf( "Enter data: ");
A
15 while (( fscanf(stdin , " %*[ \t\v \f ]%79[ \n ]" , line))
16 != EOF )
17 {
18 printf(" You entered: %s\n " line); /
19
20 // Discard newline and set line to null string
21 fgetc (stdin);
22 *( line) = '\0' ;
23 printf( " Enter data: ");
24 } // while
25
26 printf( " \nThank you\n " );
27 return 0;
28 I > // main
Results:
Enter data: No whitespace here.
You entered: No whitespace here.
Enter data: Only one whitespace character.
You entered: Only one whitespace character.
Enter data: Tabs and spaces here.
You entered: Tabs and spaces here.
Enter data: Next line is only one space.
You entered: Next line is only one space.
Enter data:
You entered:
Enter data: Ad
Thank you
EXAMPLE 11 - 3 Read Student Names and Scores

Very often reading names requires that the names be parsed into first and last
names. Parsing names presents unique problems, hirst , we must skip the
whitespace between the parts of the name and the newline at the end or
line. I he string format code ( s ) stops when it reads a whitespace character.
Second , some names have special characters such as a dash in them .
rr
.
In Program 1 1 - 4 we read a file of student names and scores As each
name is read, it is printed. We read the first and last names using the string
.
conversion code. We read the students score as an integer As an alternative,
we could have used the scanset conversion code, but it is simpler to use the
basic string code ( remember KISS).
PROGRAM 1 1 - 4 Read Student Names and Scores

1 /* Demonstrate reading names from a file.
2 Written by:
3 Date:
4 */
7 # include <string.h >
8
9 int main ( void )
10 {
1 1 // Local Declarations
12 char first[ 80 | ;
13 char last[ 80 ];
14 int score;
15 FILE* spStuScores;
16
17 // Statements
18 -
if (!( spStuScores = fopen ( "PI1 04.TXT" , "r" )))
19 {
20 printf( " \aCould not open student file. Nn " );
21 exit ( 100);
22 > // if
23
24 // Read and print first name , last name, and score
25 while (fscanf( spStuScores, " %s %s %d " ,
26 first , last, &score) == 3 )
27 printf("%s %s % 3d \n " , first , last, score);
28
29 printf( " End of Student List\n " );
30 fclose (spStuScores );
31 return 0;
32 > // main
Results:
George Washington 95
Benedict Arnold 53
-
Mary Todd Lincoln 91
End of Student List
Program 1 1 - 4 Analysis There is only one significant point to note in this program how we handle end of
data. As long as we read three pieces of data, we continue looping. When the last
data have been read, the scan returns 0 and the while statement becomes false.
String Input/Output
In addition to the formatted string functions , C has two sets of string func -
tions that read and write strings without reformatting any data. These func-
tions convert text -file lines to strings and strings to text -file lines. A line
consists of a string of characters terminated by a newline character.
C provides two parallel sets of functions, one for characters and one for
wide characters. They are virtually identical except for the type. Because wide
characters are not commonly used and because the functions operate identi -
cally, we limit our discussion to the character type. The wide - character func-
tions are listed in Appendix F.
Line to String
The gets and fgets functions take a line ( terminated by a newline ) from the
input stream and make a null - terminated string out of it . They are therefore
sometimes called line- to-string input functions.
The function declarations for get string are shown below.
char* gets (char* strPtr );

char* fgets (char* strPtr , int size , FILE* sp);
Figure 11 - 11 shows the concept . As you can see, gets and fgets do not
work the same. The gets function converts the return ( newline character ) to
-
the end of -string character ( \ 0 ), while fgets puts it in the string and appends
an end - of -string delimiter.
\n changed
to null
HelloNn
..
gets( .) H e I I o \0
Keyboard
Memory
\n kept
and null added .
Hello\n
fgets(...) I I o \n \0
* He
Memory
FIGURE 11 - 11 gets and fgets Functions

"
1
The source o( data for the gets is standard input ; the source of data lor
fgets can he a file or standard input . Both accept a string pointer and return
the same pointer if the input is successful. If any input problems occur, such
as detecting end -of - fde before reading any data , they return NULL . If no data
were read , the input area is unchanged. If an error occurs after some data * ,
have been read , the contents of the read - in area cannot be determined . The
current string may or may not have a valid null character.
Note that since no size is specified in gets, it reads data until it finds a
-
newline or until the end -of file. If a newline character is read , it is discarded
and replaced with a null character.
The fgets function requires two additional parameters: one specifying the
array size that is available to receive the data , and the other a stream . It can
be used with the keyboard by specifying the stdin . In addition to newline and
end of file, the reading will stop when size - 1 characters have been read.
Since no length checking is possible with gets , we recommend that you
never use it. Should a user enter too much data , you will destroy the data
after the string input area , and your program will not run correctly. Rather,
use fgetSy and specify the standard input file pointer ( stdin ) .
EXAMPLE 11 - 4 Demonstrate fgets Operation

-
Now let us write a simple program that uses fgets . In Program 1 1 - 5 , we use
fgets to read a string and then print it .
PROGRAM 11 - 5 Demonstrate fgets Operation

1 /* Demonstrate the use of fgets in a program
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
10 char str(81];
11
12 // Statements
13 printf( "Please enter a string: " );
14 fgets (str, sizeof (str), stdin );
15 printf( " Here is your string: \n\t%s" , str );
16 return 0;
17 > // main
Results:
the time for all students
Please enter a string: Now is
continuec
684 Section 11.3 String Input/ Output Functions
PROGRAM 1 1 - 5 Demonstrate fgets Operation (continued )

Here is your string:
Now is the time for all students
String to Line
The puts/ fputs functions take a null-terminated string from memory and write
it to a file or the keyboard. Thus, they are sometimes called string- to- line out -
put functions.
Figure 11 - 12 shows how these functions work. All change the string to a
line . The null character is replaced with a newline in puts; it is dropped in
fputs. Because puts is writing to the standard output unit , usually a display,
this is entirely logical. On the other hand , fputs is assumed to be writing toa
file where newlines are not necessarily required . It is the programmer’s
responsibility to make sure the newline is present at the appropriate place.
Note how the newline is handled in these functions. I hen compare their
use of the newline to the gets and /gets functions. While the output functions
treat the newline the opposite of the input functions, they are compatible. As
long as we are reading from a file and writing to a file, the newlines will be
handled automatically. Care must he taken , however, when we read from the
keyboard and write to a file .
The declarations for these functions are shown below.
int puts (const char * strPtr );

int fputs (const char* strPtr , FILE* sp);
The string pointed to by strPtr is written to the indicated file as

explained above. If the write is successful , it returns a non - negative integer; if
any transmission errors occur, it returns EOF. Note that the absence of a null
character to terminate the string is not an error; however, it will most likely
cause your program to fail.
r Hello
Hello\n
puts( ... )
NULL is changed
v to newline y
H e I I o \0
Memory
k.
Hello NULL is dropped
fputs( ... )
No newline
(unless in H e I I o \0
string data)
Memory
FIGURE 11 - 1 2 puts and fputs Operations

1
Examples
I his section contains examples to demonstrate the use of these string functions.
EXAMPLE 11 - 5 I he following program demonstrates the use of / puts with stdout . It calls / puts
three times. The first time, we pass a pointer to the beginning of an array. The
third time, we pass a pointer that is at the middle of the array. If you were to
run this code, you would see that it does not matter where a string pointer
starts , fhe function starts at the address in the pointer and writes until
it
finds a null characte r. Note that we had to include a newline put because
/ puts does not insert a newline. The output is shown in Program 1 1 -6.4
PROGRAM 1 1 - 6 Demonstration of Put String
1 /* Demonstrate fput string
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
on." ;
10 char str[ ] = " Necessity Is the Mother of Inventi
11 char* pStr = str ;
12
13 // Statements
14 fputs( pStr , stdout );
15 fputs( "\n", stdout );
16 fputs( pStr + 13 , stdout);
17 return 0;
18 > // main
Results:
Necessity Is the Mother of Invention
the Mother of Invention.
EXAMPLE 11 - 6 Typewriter Program

a time typewriter. In other words, it
Program 1 1 - 7 plays the role of a line-at- -
accepts text, line by line , from the keyboar
d and writes it to a text file. The
program stops when it detects an end - .
of - file
- is attributed to the American economist Thorstein Veblen .

4 . The quote in Program I 1 6
PROGRAM 11 - 7 Typewriter Program

1 /* This program creates a text file from the keyboard.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
11 char str[100];
12 FILE* spOut;
13
14 // Statements
15 -
if (!(spOut = fopen ("PI1 07.TXT", "w")))
16 {
17 printf( " \aCould not open output file.Xn " );
18 exit (100);
19 } // if
20 while ( fgets(str , sizeof ( str ), stdin ))
21 fputs(str, spOut);
22 fclose ( spOut);
23 return 0;
24 > // main
Program 1 1 - 7 Analysis Note that we are reading the keyboard [ stdin ] using /gets ( see statement 20). This
requires that we specify the file name, even though we do not have to declare it or
open it. By using fgets we are able to ensure that the user will not overrun our string
variable area (str). But even if the user enters too much data, nothing is lost. The
data are left in the input stream buffer and are read in the next loop iteration.
This little program has two problems, one stylistic and one technical. The style prob-
lem is that it contains no instructions to the user, who must therefore guess what is to be
done. This leads directly to the second problem. Some systems have a problem if the
user should end the program with an end - of - file but no return. Depending on the system,
the last line is lost or the system may just wait for an end- of - file at the beginning of the
line. This problem can be prevented by adding good user instructions and a prompt.
EXAMPLE 11 - 7 Print Selected Lines

Program 1 1 -8 reads text from the keyboard , line by line , and prints only the
lines that start with uppercase letters. In this case we will write to the stan -
dard output ( stclout ) file . This will allow us to direct the output to the printer
hv assigning standard output to a printer. If standard output is assigned to the
monitor, then the input and output will he interleaved, as shown in the
results. I he output lines are green .
I
m
1
PROGRAM 1 1 - 8 Print Selected Sentences
1 /* Echo keyboard input that begins with capital letter.
2 Written by:
3 Date written:
4 */
7
8 int main ( void )
9 {
11 char strng[81];
12
13 // Statements
14 while ( fgets ( strng , sizeof( strng ), stdin ))
15 if ( isupper(*strng ))
16 fputs(strng, stdout );
17 return 0;
18 > // main
Results:
Now is the time
Now is the time
for all good students
to come to the aid
of their school .
Amen
Amen
, to determine which lines we

Program 1 1 - 8 Analysis In this program, we use the character function, isupper Althoug h we are guarding against
want to write . The output lines are printed in color .
If the long lines, they will be
excessive input, we do not flush the line . user enters very
obvious when the program runs.
, you will need
if you use this program or any variation of it to write to your printer for your sys-
file Refer the docume ntation
to assign the printer to the standard output
. to
tem to determine how to do this.
EXAMPLE 11 8 Print File Double Spaced

-
text Irom a file and prints the text double

Program 1 1 -9 reads a single- spaced
line after each line. In this program
spaced. In other words, it inserts a blank file , stdout , which is usually the
we direct the output to the
standar d output
, you need to redirect the output or
monitor. To get the output to printer
a
assign stdout to the printer.

688 Section 11.4 Arrays of Strings
PROGRAM 11 - 9 Print File Double spaced

1 /* Write file double spaced.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
11 char strng[81];
12 FILE* textln ;
13
14 // Statements
15 if (!(textln = -
fopen( " PI 1 07.TXT" , "r")))
16 {
17 printf( "\aCan ' t open textdata\n " );
18 exit ( 100 );
19 } // if
20 while ( fgets( strng , sizeof( strng ), textln))
21 {
22 fputs( strng , stdout );
23 putchar ( '\n' );
24 > // while
25 return 0 ;
26 > // main
Program 1 1 - 9 Analysis Because we are reading data from a text file, we use the fgets function. This function
guarantees that we will not overrun our input string variable. Note how we used the
sizeof operator to set the maximum number of characters to be read. Since fgets
adds a newline character at the end of the data being read, we don' t need to worry
about adding one when we write. After writing the data string, we use a putchar to
write the blank line.
11.4 Arrays of Strings

When we discussed arrays of pointers in Chapter 10, we introduced the con-
cept of a ragged array. Ragged arrays are very common with strings. Consider,
'
for example, the need to store the days of the week in their textual format. We
could create a two- dimensional array of seven days by ten characters
(Wednesday requires nine characters) , but this
wastes space.
It is much easier and more efficient to create ragged array using
a
array of string pointers. Each pointer points to a
day of the week . In this way
each string is independent , but at the same time, they are grouped together
rr 1
through the array. In other words, although each string is independent, we

can pass them as a group to a function by passing only the name ol the array
of pointers. Program 1 1 - 10 demonstrates this structure .
PROGRAM 11 - 10 Print Days of the Week

1 /* Demonstrates an array of pointers to strings.
2 Written by:
3 Date written:
4 */
6
7 int main ( void )
8 {
10 char* pDays( 7];
11 char** pLast ;
12
13 // Statements
14 pDays[0] = "Sunday " ;
15 pDays[ l ] = " Monday " ;
16 pDays[ 2] = "Tuesday " ;
17 pDays[ 3] = "Wednesday " ;
18 pDays[ 4 ] = "Thursday " ;
19 pDays[ 5 ] = "Friday " ;
20 pDays[6 ] = "Saturday" ;
21
22 printf( "The days of the week \n " );
23 pLast = pDays + 6;
24 for (char* * pWalker = pDays ;
25 pWalker <= pLast ;
26 pWalker++)
27 printf( " %s\n" , * pWalker );
28 return 0;
29 } // main
is a pointer constant,
Program 11 - 1 0 Analysis To print the days of the week, we use a for looparithmetic . The pointer variable must
. Since pDays
we also need a pointer variable to use pointer

to the strings through pDays, which means
that it will be a pointer to a pointer,
point
as seen in statement 11.
A point of efficiency is the way we handled the limit test in the for loop. We could
have simply coded it as
pWalker <= pDays + 6
address be recalculated for each limit

This would require, however, that the ending
should recogniz e that pDays is a pointer constant
test. ( A good optimizing compiler
690 Section 11.5 String Manipulation Functions
and that the calculation needs to be done only once, but we can't be sure that such effi-
cient code would in fact be generated.) Therefore, we calculate the ending address just
once, before the while loop, and then we can be sure that the limit test will be efficient.
Study this code carefully. Note first that pWalker is a pointer to a pointer to a
character. Then notice how it is used in the for statement. It is initialized to the first ele-
ment in pDays, then it is incremented until it is no longer less than or equal to pLast.
Finally, note how it is used in the printf statement. The printf syntax requires that the
variable list contain the address of the string to be printed. But pWalker is a pointer to
an address that in turn points to the string ( a pointer to a pointer ). Therefore, when we
dereference pWalker, we get the pointer to the string, which is what printf requires.
This example is diagrammed in Figure 11 - 13.
pWalker
4 S u n d a y \0
M | o n d a y \0
\\ \ 'x
X'\ v
V
T | u | e | s | d | a | y \0
*v W | e | d | n | e | s | d | a | y \0
T | h | u | r | s | d a y \0
\
pLast F r i | d a y \0
pDays
S | a | t | u | r | d | a | y | \0
FIGURE 1 1 - 1 3 Pointers to Strings
11.5 String Manipulation Functions

Because a wstring is not a standard type, we cannot use it directly with most
C operators. For example, to move one string to another, we must move the
individual elements ol the sending string to the receiving string. We cannot
simply assign one string to another. II w 4 e were to write the move, we would
have to put it in a loop.
C has provided a rich set ol string functions. Besides making it easier for
us to write programs, putting the string
operations in functions provides the
opportunity to make them more efficient when the operation is supported by
hardware instructions. For example, computers often have a machine instruc -
tion that moves characters until a token, such as a null character, is reached.
When this instruction is available, it allows a string to he moved in one
instruction rather than in a loop.
In addition to the string character functions, C provides a parallel set ol
functions for wide characters. They are virtually identical except lor the type.
Because wide characters are not commonly used and because the functions
operate identically, we limit our discussion to the
character type. The wide-
character functions are listed in Appendix F. The traditional string functions
hu \ e a prefix ol str. I he basic format is shown
in the following.
nr
int str... ( parameters)
I he string character functions are found in the string library ( string .h ) .
String Length
I he string length function, strlen, returns the length of a string, specified as
the number of characters in the string excluding the null character. If the
string is empty, it returns zero The function declaration is shown below.
.
int strlen (const char* string );
EXAMPLE 11 -9 String Length Demonstration

To demonstrate the use of string length, lets write a program that reads text
from the keyboard, line by line, and adds two blanks ( spaces) at the beginning
of each line before writing it to a file. In other words, it shifts each line two
characters to the right . The code is shown in Program 1 1 - 1 1 .
PROGRAM 11 - 11 Add Left Margin

1 / * Typewriter program: adds two spaces to the left
2 margin and writes line to file
3 Written by:
4 Date:
5 * /
8 # include <string.h>
9
10 int main (void )
11 {
13 FILE* spOutFile;
14 char strng[ 81 J ;
15
16 // Statements
17 if (!( spOutFile = fopen("Pll 11.TXT -
" , "w" )))
18 {
19 printf( " \aCould not open output file Xn )
" ; .
20 exit ( 100 );
21 > // if
22
, stdin ))
23 while ( fgets(strng , sizeof( strng )
24 {
continued
PROGRAM 1 1 - 1 1 Add Left Margin ( continued )

25 fputc( , spOutFile );
26 fputc(1
spOutFile);
27 fputs( strng , spOutFile );
28 if (strng[ strlen(strng ) -
1 ] != \n ' )
29 fputs("\n" , spOutFile );
30 > // while
31 fclose (spOutFile );
32 return 0;
33 } // main
Program 1 1 - 1 1 Analysis To ensure that the user doesn't overrun the input area, we use the fgets function to
read the keyboard. Some data may be left in the buffer, but we don' t flush the buffer,
because we want to write them to the file on the next line.
Also, since fputs does not add a newline when it writes the file, we need to ensure
that there will be a new line at the end of each line. Therefore, in statement 28 we test
the last character of the input string and if it isn' t a newline, we write one.
Because we want to add characters to the beginning of the line, we must use a
character operation. The function fputc writes one character to a designated file. To
insert two characters, therefore, we use it twice, and then use fputs to write the line
read by fgets .
String Copy
C has two string copy functions. The first , strepy, copies the contents of one
string to another. I he second , strnepy , also copies the contents of one string
to another, hut it sets a maximum number of characters that can be moved.
I herd ore, strnepy , is a safer function .
Basic String Copy

I he string copy function , strepy, copies the contents of the from string,
including the null character, to the string . Its function declaration is
shown below.
char* strepy (char* toStr , const char* fromStr);
II fromStr is longer than toStr , the data in memory after toStr are
destroyed . It is our responsibility to ensure that the destination string array is
large enough to hold the sending string. This should not be a problem, since
we control the definition oi both string variables. The address of toStr is
returned , which allows string functions to he used as arguments inside other
string functions. We will demonstrate the use of these returned pointers later
in the chapter.
Figure 11 - 14 shows two exam pies ol string copy. In the first example, the
source string is shorter than the destination variable. The result is that , after
the stling has been copied , the contents of
the last three bytes of si are
unchanged ; si is a valid string, however.
7
G o o d D a | y \Q
s1- before s2 - before
(a ) strcpy( s1, s2 ) ;
G| o | o | dl |D|aly [ \0 |G|o| o|d| lD| a | y|\Q[

s 1 - after s2 - after
Copying Strings
S h o r t \0 0 t h e r \0 G o o d D | a l y |\Q |
s1- before s3 - before s 2 - before
(b) strcpy( s 1, s2 ) ;
Goo d | D|a y \0 e r \0 Good D| a y \0

s1 - after s3- after s 2 - after
Copying Long Strings
FIGURE 11 - 1 4 String Copy
In the second example in Figure 1 1 - 14 , the destination variable, si , is

only 6 bytes, which is too small to hold all the data being copied Irom s2
(8 bytes plus the delimiter ) . Furthermore, si is immediately followed by
another string, s3. Even though si is too small , the entire string from s2 is
copied , partially destroying the data in s3. Although si is a valid string, any
attempt to access the data in s 3 will result in a string containing only ay.
If the source and destination strings overlap—that is, if they share corn -
mon memory locations, then the results of the copy are unpredictable
. In
Figure 11 - 14 , this would occur if we tried to execute the following statement :
strcpy(( s2 + 2 ), s2 ); // Invalid copy overlap -

String Copy — Length Controlled
Many of the problems associated with unequal string-array sizes
can he con -
trolled with the string- number copy function , strncpy . I bis function can he
contains
that specifies the maximum number of characters that
a parameter
.
moved at a time, as shown in its function declaration below
char* fromStr ,
char* strncpy (char* toStr, const
int size );
number of characters that

In this function , size specifies the maximum
a little more complex. If the size of
can be moved. Actually, the operation is
or greater than size then size characters are
the from string is equal to
,
moved . If the from string is smaller than size , the entire string is copied and
then null characters are inserted into the destination string until exactly size
characters have been copied. Thus, it is more correct to think of size as the
destination characters that must be filled .
If the sending string is longer than size , the copy stops after size bytes
have been copied . In this case, the destination variable may not be a valid
string; that is , it may not have a delimiter. The string number copy functions
do not insert a delimiter if the from string is longer than size . On the other
hand , the data following the destination variable will be intact , assuming the
size was properly specified . Figure 11 - IS shows the operation of strncpj
under these two conditions.
G o o d D a y \0
s1 - before s2 - before
(a) strncpy ( s1, s2, sizeof (s1)) ;
G| o | oldl |D| a | y |\0 \0 \0 \0 |G| o | o | d | |D| a | y \0

s1 - after s2 - after
Copying Strings
S| h| o | r 11 |\0|O| 11hle| r |\o| | G| o | o | d Da y \0

s1- before s3- before s 2 - before
(b) strncpy ( s1, s2 , sizeof ( s1 )) ;
G| o | o | d | |D 0| 11 h | e | r \0 |G| o | o | d | |D | a | y |\0
s1 - after s3- after s 2 - after
Copying Long Strings
FIGURE 1 1 - 1 5 String -number Copy
We recommend that you always use strncpy ; do not use strcpy. To prevent
invalid strings , we also recommend that you move one fewer character than
the maximum and then automatically pi ace a null character in the last posi-
tion . The code for this technique is shown below.
strncpy ( si , s 2 , s i z e o f ( s l ) -1 ) ;
* ( s l + ( sizeof ( s i ) -
1 ) ) = ' \o ' ;
Since the strncpy places null characters in all unfilled characters, we are
guaranteed that the last character in the string array is a null character. If it is
not , then the copy was short . By executing the above statements, we are assured
that si will be a valid string, even if it doesn ’t have the desired contents.
A closing note: If size is zero or negative , nothing is copied. The destina-
tion string is unchanged .
1
Always use stmcpy to copy one string to another.
EXAMPLE 1 1 - 1 0 Build Name Array

Let ’s write a small program that uses the strep)' function. Program 11 - 12
builds an array of strings in dynamic memory using the calloc function ( see
Chapter 10). It then fills the array from strings entered at the keyboard.
When the array is full , it displays the strings to show that they were entered
and stored correctly.
PROGRAM 11 - 1 2 Build Name Array in Heap

1 /* Build a dynamic array of names.
2 Written by:
3 Date:
4 */
8
9 # define FLUSH while ( getchar( ) != ’ \n ' )
10
12 {
14 char input(81];
15 char** pNames; // array of pointers to char
16
17 int size;
18 int nameslndex ;
19
20 // Statements
21 printf("How many names do you plan to input?
22 scanf ("%d", &size);
23 FLUSH ;
24
25 // Allocate array in heap.
26 // One extra element added for loop control
27 pNames = calloc ( size + 1 , sizeof (char* ));
28 printf( " Enter names:\n " );
29
30 nameslndex = 0;
31 while ( nameslndex < size
32 && fgets( input, sizeof( input ), stdin ))
33 {
continuei
PROGRAM 11 - 12 Build Name Array in Heap ( continued )

34 *(pNames + nameslndex) = (char* )
35 calloc (strlen( input ) + 1 , sizeof(char ));
36 strcpy (*( pNames + nameslndex ), input);
37 nameslndex++;
38 } // while
39
40 *( pNames + nameslndex ) = NULL ;
41 printf("\nYour names are: \n" );
42 nameslndex = 0;
43 while (*( pNames + nameslndex ))
44 {
45 printf( " % 3d: %s" ,
46 nameslndex , *( pNames + nameslndex ));
47 nameslndex++;
48 > // while
49 return 0;
50 } // main
Results:
How many names do you plan to input? 3
Enter names:
Tom
Rico
Huang
Your names are:

0: Tom
1: Rico
2: Huang
Program 11 - 12 Analysis This little 50- line program contains some rather difficult code. To understand it better,
let's look at the array structure, shown in Figure 11 - 16.
pNames
fB *
>
-
*
Last pointei
.always null
(heap)
FIGURE 11 - 16 Structure for Names Array

^
T1
The identifier, pNames, is a pointer to an array of pointers to a character that is
dynamically allocated from the heap. Then, as each name is read, space is allocated
from the heap, and its pointer is placed in the next location in the pNames array. The
only way to refer to the names is by dereferencing pNames. To access an individual
element, we use pNames and index it to get to an individual string pointer in the array.
This code is shown in statement 36. Since the first parameter in the string copy is a
pointer to a string, only one dereference is required.
To build the pointer array, we use a while loop with two limit tests: end - of - file and a
full array. Either condition will stop the loading of the array. To print the array, however,
we only need to test for a null pointer, since the pointer array is allocated with one extra
element. Using an extra element is a common programming technique that makes pro-
cessing arrays of pointers easier and more efficient.
String Compare
As with the string copy functions, C has two string compare functions.^ The
first, strcmp, compares two strings until unequal characters are found or until
.
the end of the strings is reached The second, strncmp, compares until
unequal characters are found, a specified number of characters have been
tested, or until the end of a string is reached.
Roth functions return an integer to indicate the results of the compare.
Unfortunately, the results returned do not map well to the true-false logical
values that we see in the if .. .else statement (see Chapter 5 ), so you will need
to memorize a new set of rules:
. .
1 If the two strings are equal, the return value is zero Two strings are con -
sidered equal if they are the same length and all characters in the same
relative positions are equal.
2. If the first parameter is less than the second parameter, the return value
is less than zero . A string, si, is less than another string, s 2 , if starting
from the first character, we can find a character in si that is less than the
character in s 2 . Note that when the end of either string is reached, the
null character is compared with the corresponding character in the other
string.
.
3 If the first parameter is greater than the second parameter
, the return
. A string si , is greater than string, s 2 , if start -
value is greater than zero ,
first character, we can find a character in s 2 that is greater
ing from the
than the corresponding character in si. Note that when the
end of either
compared with the correspond ing
string is reached, the null character is
character in the other string .
Note that the not -equal values are specified as range

a . If the first param-
eter is less than the second parameter
, the value can be any negative value .
is greater than the second parameter, the
Likewise, if the first parameter , such as
value can be any positive number. This differs from
other situations
EOF, where we can rely on one given value being returned .
S . Wide character functions are available

- . Their function declarations arc found in Appendix l \
The string compare operation is shown in Figure 1 1 - 17 .
strcmp
The function declaration for the string compare function is shown below.
int strcmp (const char* strl , const char* str2 );
While the string compare functions return integer values, ( < = l , o, >= i )
we can code a logical expression that returns true or false by using the com-
pare function and relational operators. For example, to compare two strings
for equal, we must write the statement as shown below.
if (strcmp(strl , str2 ) == 0)
// strings are equal
else
// strings are not equal
Stops at end
of string -
s1 | AlB |C| | C |O|M| P | A |N | Y |\0|

tttttttttttt
U l t t H t U t t
s 2 | A|B|C | |C|0|M|P| A|N Y \0
s1 equals s 2
Stops at Stops at
not equal not equal
s 1 | A | B | C | | ClO |M| P | AlNlYl\Q | s1 | A | B | C | 1 C|C|

> R|P \0
t t =t =t <t ft tt tt t
M i l l I M I M I
s2 | A | B |Cl | I | N|C|\0 s2 A [ B |C | TCTO|M| P | A N | Y |\O
' C‘ < T s1 less s2 R ' > ' M' s1 greater s2 |
strcmp ( s1 , s2 )
FIGURE 1 1 - 1 7 String Compares
I he I allowing statement tests whether the first string is less than the sec-
ond string.
if (strcmp (stringl , string2) < 0)

/ / stringl is less than string 2
1
99
Chapter 11 Strings
^
Io test for s t r i n g l greater than s t r i n g 2 , use the following code:
i f ( strcmp ( s t r i n g l , s t r i n g 2 ) > 0 )
/ / s t r i n g l i s g r e a t e r than s t r i n g 2
We can also test for greater than or equal to with the following statement :
i f ( strcmp ( s t r i n g l , s t r i n g 2 ) > = 0 )
/ / stringl is greater than or equal to string2
strncmp
I he string number compare function , strncmp, tests two strings for a speci-
fied maximum number of characters ( s i z e ). The function declaration s are
shown below.
i n t strncmp ( const c h a r * s t r l , c o n s t c h a r * s t r 2 ,
int size ) ;
In this function , s i z e specifies the maximum number of characters to be

compared in the first string. The strncmp compare logic is the same as seen in
strcmp , except for the length limit . Table 11 - 1 shows the results of comparing
two strings using the strncmp function for various sizes.
stringl
"ABC123"
string2
" ABC123"
1 8 equal
Results Returns
0
" ABC 123" "ABC456 " 3 equal 0
"ABC123" "ABC456" 4 stringl < string2 < 0
" ABC 123" " ABC" 3 equal 0
" ABC 123 " " ABC" 4 stringl > string 2 >0
" ABC " "ABC 123 " 3 equal 0

" ABC123" " 123ABC " -1 equal 0
TABLE 11 - 1 Results for String Compare
String Concatenate
to the end of another.
The string concatenate functions append one string
They return the address pointers to the destination string . The size of the des -
is assumed to he large enough to hold the resulting
tination string array
the data at the end of the string array are destroyed . As we
string. If it isn ’t ,
ble if the strings overlap .
saw with string copy, the results are unpredicta
Basic String Concatenation

The function declaration lor string concatenation, strccit , is shown below.
char * streat ( char * strl, const char * str 2 ) ;
The function copies str 2 to the end of strl, beginning with strl's
delimiter. That is, the delimiter is replaced with the first character of str 2 .
The delimiter from str 2 is copied to the resulting siring to ensure that a
valid string results. The length of the resulting string is the sum of the length
of strl plus the length of str 2. Figure 1 1 - 18a shows the operation.
Siring Concatenation — Length Controlled

The function declaration for length-controlled string concatenate function,
shown below.
strncat , is
char * strncat ( char * strl, const char * str 2 , int size ) ;
ClOlN \C C A T E N A T I O N \0
s 1- before s 2 - before
(a) strcat( s1, s2 ) ;
C |Q | N | C | A | T | E | N | A | T | I IOTNI 'O
| C | A | T | E | N | A | T | I | O | N | \0
s1 - after s2 - after
String Concatenate
C | Q | N \C C A T E N A T I O N \0
s 1 - before s2 - before
(b ) strncat ( s 1 , s2 , 3 ) ;
C|O|N|CTATT W C A T E N A T I O N S
s1 - after s 2 - after
String N Concatenate
FIGURE 11 - 1 8 String Concatenation
Figure 1 1- 18b demonstrates the operation of the strncat function. If

the length ol string2 is less than size , then the call works the same as the
basic string concatenation described above. However, i! the length of
string 2 is greater than size, then only the number of characters specified
b > size are copied, and a null character is appended at the end.
II the value ol size is zero or less than zero, then both strings are treated
as null, and no characters are moved. The variable string1 is unchanged.
Chapter 11 Strings / 01 I
Character in String
Sometimes we need to find the location of a character in a string. Two string
I unctions search lor a character in a string. The first function is called string
character strchr; it searches lor the first occurrence of a character from the
beginning of the string. The second string rear character called strrchr ,
searches lor the first occurrence beginning at the rear and working toward
the front .
In either case, il the character is located , the function returns a pointer
to it . If the character is not in the string, the function returns a null pointer.
I he declarations lor these functions are shown below.
char* strchr (const char* string , int ch );

char* strrchr (const char* string , int ch );
Note that , as often is the case in the library functions, the character in the
ASC 11 functions is typed as an integer. We don ’t need to worry about this.
I he compiler will implicitly cast any character types to integer types before the
call, figure 1 1 - 19 shows three examples of character- in -string calls.
si |C|O| N C A | T | E N | A | T | I Q]N \Q
1 1 1 1 1 1
P1
a* p1 = strchr ( s1, ’N’ ) ;
si C O N C A T E N A T I O N S
P1
v P2
p2 = strrchr ( s1, ’N’ ) ;
s1 |C|0|N|C| A |T |E|N| A | T | I |O|N|\0|

t [ t
p1 p3
^ ^
[ p2
p3 = strchr ( (p1 + 1 ) , ’N’ ) ;
FIGURE 11 - 19 Character in String (Strchr )
N at the third position in the

In the first example , we locate the first
string ( si [ 2 ] ). The pointer is then
assigned to pi . In the second example , we
] . In the third example, we want to
locate the last N in the string at si [ 12
to start the search after the first N .
locate the second N . To do this, we need
N , this is easily done with a pointer
Since we saved the location of the first
and pointer arithmetic as shown .
Search for a Substring

If we can locate a character in a string, we should be able to locate a string in
a string. We can, but only from the beginning of the string. There is no func-
tion to locate a substring starting at the rear. The function declaration for
strstr is shown below.
char * strstr (const char* string ,

const char* sub string );
As indicated by the function declaration, this function also returns a

pointer to a character. This pointer identifies the beginning of the substring
in the string. If the substring does not exist , then they return a null pointer.
Figure 1 1 - 20 demonstrates the operation of string in string .
Desired
Substrinc
s1 q O N C A T E j N A T I O N \0
t j p1 = strstr ( s1, "CAT" ) ;
FIGURE 11 - 20 String in String
Search for Character in Set

Very often, we need to locate one of a set of characters in a string. C provides
two string functions to do this. The first strspn, locates the first character that
does not match the string set. The second strcspn , locates the first character
that is in the set .
Basic String Span

I he basic string span function, strspn, searches the string, spanning charac-
ters that are in the set and stopping at the first character that is not in the set.
I hey return the number ol characters that matched those in the set. If no
characters match those in the set, they return zero. The function declaration
is seen below.
int strspn(const char* str , const char * set );
An example of string span is shown in Figure 11 21. We use strspn to

-
determine the number of characters that match the characters in the string
set . In this example, len is set to
five, since the first five characters match
the set.
len 151 len = strspn (s1, " AEIOUCN");
s 1 |C|Q|N|C| A|T | E |N| A | T | I |Q|N|\Q
len |5| len = strcspn (s1, "TEIBX");
IcioNC lAlTlE N A T I O N S
FIGURE 11 - 2 1 String Span
Complemented String Span

The second function , strcspn , is string complement span ; its functions stop at
the first character that matches one of the characters in the set . It none of the
characters in the string match the set , they return the length ot the string.
Figure 11 - 21 also contains an example of strcspn . The function declarations
are shown below.
int strcspn( const char* str , const char* set );
String Span — Pointer

Very often , we need a pointer to a substring rather than its index location . C
provides a set of functions that return pointers. They operate just like the
string complement functions except that they return a pointer to the first
character that matches the set . The p in the function names stands for
no match -
pointer brk is short for break . As with the complement spans if
— ,
ing characters are found , they return null.

Similar to strcspn , the strpbrk function returns a pointer to the first char-
acter found in the set . Its function declaration is shown below ’
/
char* strpbrk (const char* str, const char* set);
String Token
, called tokens ,
The string token function , strtoky is used to locate substrings
string into tokens , much as a
in a string. Its most common use is to parse a locates
how it is called it either
com piler parses lines of code. Depending on
,
string. Its function declaration statement is

the first or the next token in a
shown below.
char* delimiters ),
char* strtok (char* str , const
that returns a pointer to the last character found

6. Traditional C has a function called strrpbrk / C function library.
of the ANSI ISO
in the set. This function is not part
The first parameter is the string that is being parsed ; the second parame-
ter is a set of delimiters that will he used to parse the first string. If the first
parameter contains an address , then strtok starts at that address, which is
assumed to be the beginning of the string. It first skips over all leading delim -
iter characters. If all the characters in the string are delimiters, then it termi-
nates and returns a null pointer. When it finds a nondelimiter character, it
changes its search and skips over all characters that are not in the set ; that is
it searches until it finds a delimiter. When a delimiter is found , it is changed
to a null character ( ' \ 0 • ) , which turns the token just parsed into a string.
If the first parameter is not a string, strtok assumes that it has already
parsed part of the string and begins looking at the end of the previous string
token for the next delimiter. When a delimiter is located, it again changes the
delimiter to a null character, marking the end of the token , and returns a
pointer to the new token string.
Let ’s look at a simple example of a string containing words separated by
spaces. We begin hv calling string token with the address of the f ull string. It
returns the address of the first character ol the first string that was just
parsed . The second execution of the string token function parses the second
string and returns its address, and so on until the complete string has been
parsed. I bis design is shown in Figure 1 1 - 22 .
pToken = strtok (strng, " ");

pToken pToken Q—
j 0[ N | EnTIwlb | | TlH | R | E | Epl |0 [ N | E |\O| T |W|Q| | T | H | R | ETEI O
strng
( a ) Before Parsing
strng
^
( b ) After First Parsing
pToken = strtok (NULL, " "); pToken = strtok (NULL, " ");
pToken Q pToken ££)
strng | Q 1 N | E | \0| T |W|0|\0| T | H | R | E | E |\0' strng OlN E 0 Tlwlo! 0|T|H|R|E| EbQ
( c ) After Second Parsing ( d ) After Last Parsing
FIGURE 1 1 - 22 Parse a Simple String
String to Number
I he standard library ( stcllib.lt ) library provides several functions that convert
a string to anumber. I he two most important are string to long and string to
double . A complete list is found at the end of this section .
String to Long
I he string- to- long function , strtol , converts a string to a long integer. It skips
leading whitespace characters and stops with the first non - numeric character,
7. To convert «i string to ; m unsigned long integer, use sIrloul .

’'
1
which is considered to be the start of a trailing string. The address of the

trailing string is stored in the second parameter, unless it is a null pointer. A
third parameter determines the base of the string number. The function dec -
laration is shown below.
long strtol (char* str, char** ptr , int base);
I he base is determined by the following rules:

1 . The base may be 0 or 2 ... 35 .
2 . The letters a . . . z or A . . . Z represent the values 10 . . . 35. Only the numeric
and alphabetic characters less than the base are permitted in any string.
3. If the base is 0, the format is determined by the string as follows:
a . If the number begins Ox or OX the number is a hexadecim al constant.
.
b. If the first digit is 0 and the second digit is not x or X , the number is an
octal constant .
c . If the first digit is a nonzero, the number is a decimal constant.
If the string does not begin with a valid number, zero is returned and the
trailing string pointer is set to the beginning of the string. Program 1 1 - 13
demonstrat es the use of strtol .
PROGRAM 11 - 13 Demonstrate String to Long

1 / * Demonstrate string to long function.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
11 long num ;
12 char* ptr ;
13
14 // Statements
" , & ptr , 0 );
15 num = strtol ( " 12345 Decimal constant:
16 printf( " %s % ld \n " , ptr , num );
17
: " , &ptr , 2);
18 num = strtol ( " 11001 Binary constant
20
( " 13572 Octal constant : " , &ptr , 8 );
21 num = strtol
22 printf( " %s %ld\n" , ptr , num);
continual
/ 06 Section 11.5 String Manipulation Functions
PROGRAM 1 1 - 1 3 Demonstrate String to Long (continued )

23
24 num = strtol ( " 7AbC Hex constant : % & ptr, 16);
25 printf( " %s %ld\n " , ptr , num);
26
27 num = strtol ("11001 Base 0-Decimal : "f *Ptr, 0);
28 printf( " %s % ld\n " , ptr , num );
29
30 num = strtol ( "01101 Base 0-Octal : " , &ptr , 0 );
31 printf( "%s % ld\n " , ptr , num );
32
33 num = strtol ( "0x7AbC Base 0-Hex : " , & ptr, 0);
34 printf("%s %ld\n" , ptr, num);
35
36 num = strtol ("Invalid input : ", &ptr, 0);
38
39 return 0;
40 > // main
Results:
Decimal constant 12345
Binary constant 25
Octal constant 6010
Hex constant 31420
Base 0-Decimal 11001
Base 0-Octal 577
-
Base 0 Hex 31420
Invalid input 0
j Program 11 - 1 3 Analysis The first thing to note in this program is that the address of
the character pointer must
be passed to strtol. If this surprises you, go back and look at the strtol function decla -
ration statement again. Note that the declaration of ptr has
two asterisks.
Observe that the pointer is set to the first character after the last valid digit. In our
example, this character was always a space. If there were no characters after the last
digit, the pointer would point to the string delimiter, resulting iin a null string.
String to Double
I he string- to- double functions—strtocl and wcstod —
are similar to the string-
to- long functions, except that they do not have a hase. The function declara-
tion is seen below.
double strtod (char* str, char * ptr

* );
I lie numeric string must he a floating-
point in either decimal or scien-
t lic notation. I he valid numeric values
' in the string are the digits , the plus
^ Ij
T
’
and minus signs, and the letter e or E in the exponent . After using the longest
possible sequence ol digits and characters to convert the string to a number,
strtod sets the pointer to the address of the next character in the string. If no
characters are present , the pointer is set to the string delimiter. As with
string- to- Iong, if the number is invalid , zero is returned and the pointer is set
to the beginning of the string.
Other Number Functions

Table 1 1 - 2 lists the complete set of string- to - number functions.
Numeric Format ASCII Function Wide -character Function
double strtod wcstod

float strtof wcstof
long double strtold wcstold
long int strtol wcstol
long long int strtolI wcstoll

unsigned long int strtoul wcstoul
unsigned long long int strtoull wcstoull
TABLE 11 - 2 String - to -Number Functions
String Examples
In this section , we discuss two functions that use string functions.
The first
. The second uses string input /
uses strtok to parse an algebraic expression
and several string manipulatio n functions to compare
output functions
strings.
EXAMPLE 1 1 - 1 1 Parsing Tokens

Assume we want to parse a string containing a simple
algebraic expression as
shown below.
sum = sum + 10 ;
by whitespace. Our
Each token in this simple expression is separated
it out . The step- by-step parsing is
program will identify each token and print
Program 11 - 14 .
shown in Figure 1 1 -23. The code is shown
in
strng | S |U|M| |= S |U |M| | + | 11 |Q I ; |\0|

Step 1 pToken = strtok (strng, "
strng | S|u|M|\0|= | | S | U |M| [ + | 1110 | ; |\0
pToken (First Token)
In loop pToken = strtok ( NULL, " ;"); |

strng | S | U|M|\0| = |\0 | S |U |M| | + | 110 ; |\0
pToken (Second Token)
^
strng | S |U|M|\0|= |\0| S | U|M| | -t | 1110 ; |\0
pToken ( Third Token)
strng | S |U|M|\Q = |\0 S | U |M \Q + \0| 110 | ; \0
pToken ( FourthToken)
strng | S | U |M|\0|= |\Q S |U |M \Q + \0 0 \0 \0
pToken ( Last Token)
FIGURE 11 - 23 Parsing with String Token
PROGRAM 11 -14 Parsing a String with String Token

1 / * Parse a simple algebraic expression .
2 Written by:
3 Date:
4 */
5 #include <stdio.h >
6 tinclude <string.h>
7
8 int main (void )
9 {
11 char strng [16] = "sum = sum + 10 ;" ;
12 char * pToken ;
13 int tokenCount;
14
15 // Statements
16 tokenCount = 0 ;
17 pToken = strtok (strng, "
/ ) r
18
contin net
'
II

^
PROGRAM 11 - 1 4 Parsing a String with String Token ( continued )

19 while ( pToken )
20 {
21 tokenCount++;
22 printf("Token %2d contains %s\n" ,
23 tokenCount, pToken );
24 pToken = strtok ( NULL, "
25 > // while
26
27 printf("\nEnd of tokens\n" );
28 return 0 ;
29 > // main
Results:
Token 1 contains sum
Token 2 contains =
Token 3 contains sum
Token 4 contains +
Token 5 contains 10
End of tokens
Program 11 - 1 4 Analysis Since the first call to strtok must contain the address of the string, it is coded before
the loop. If the string contains at least one token, the first call will return a valid
address and the while loop will print it and parse out the remaining tokens. Note that
the delimiter set includes the semicolon as well as the blank between tokens. The
semicolon serves as the last token in the string.
Compare Packed Strings

When working with strings, we often find that two strings are logically the
same but physically different to the computer. For example , consider a pro -
gram that generates mailing labels . Often a name is put into a mailing list
with an extra space or other character that prevents it from being matched to
an existing name . One way to eliminate such errors is to compare only
the
letters of the names by removing everything except alphabetic characters .
that compares two strings after packing the
Program 11 - 15 shows a function
data so that only letters are left .
PROGRAM 11 - 1 5 Compare Packed String Function

1 / * This program packs and compares a string
.
2 Written by:
3 Date:
4 */
6 #include <string.h >
continuec
%
I
PROGRAM 1 1 - 1 5 Compare Packed String Function ( continued )

7
8 # define ALPHA \
9 " ABCDEFGHIJKLMNOPQRSTUVWXYZabcdef ghi jklmnopqrstuvwxyz"
10
12 int strCmpPk (char* SI, char* S2);
13 void strPk ( char* si , char* s2);
14
16 {
18 int cmpResult ;
19 char s i [ 8 0 ] ;
20 char s2[80 ];
21
22 // Statements
23 printf("Please enter first string:\n" );
24 fgets ( si , 80 , stdin );
25 si[ strlen( si ) 1 ] = '\0';
26
27 printf( "Please enter second string:\n " );
28 fgets ( s2 , 80, stdin );
29
30
s2[strlen(s2) - 1 ] = \0 ' ;
31 cmpResult = strCmpPk ( si , s2 );
32 if ( cmpResult < 0 )
33 printf("stringl < string2\n " );
34 else if (cmpResult > 0 )
35 printf( "stringl > string2 \n " );
li 36 else
37 printf( "stringl == string2\n " );
38
39 return 0;
40 } // main
41
42 ===== strCmpPk =
43 Packs two strings and then compares them.
44 Pre si and s2 contain strings
45 Post returns result of strcmp of packed strings
46 */
47 int strCmpPk ( char* si , char* s2)
48 {
50 char slln [ 80);
continue
'
\ n
PROGRAM 11 - 1 5 Compare Packed String Function ( continued )
51 char slOut[ 81 ];
52 char s2In [ 80 ];
53 char s20ut[81];
54
55 // Statements
56 strncpy (slln , si , sizeof(slln ) i );
57 strncpy (s2In , s2 , sizeof(s2In ) 1 );
58 strPk ( slln , slOut);
59 strPk (s2In , s20ut );
60 return (strcmp ( slOut , s20ut ));
61 > // strCmpPk
62
63 :
=====
64 -
Deletes all non alpha characters from si and
65 copies to s2.
66 Pre si is a string
67 Post packed string in s2
68 si destroyed
69 */
70 void strPk (char* si , char* s2)
71 {
73 int strSize;
74
75 // Statements
76 *s2 = '\01 ;
77 while ( *sl != \0 * )
78 {
79 // Find non-alpha character & replace
80 strSize = strspn( sl , ALPHA );
81 si[strSize ] = ’ \01 ;
82 strncat (s2 , si , 79 -
strlen( s2 ));
83 si += strSize + 1 ;
84 } // while
85 return ;
86 > // strPk
Results:
Please enter first string:
a b!c 234d
Please enter second string:
abed
stringl == string 2
c continued
712 Section 11.6 String/Dato Conversion
PROGRAM 1 1 - 1 5 Compare Packed String Function ( continued )

Please enter first string:
abed
Please enter second string:
aabb
stringl > string 2
Program 1 1 - 1 5 Analysis To test the string compare, we need to write a test driver. Since the test driver is throw -
away code, we have coded the test logic in main.
Statement 8 contains a C language construct that we have not used before: the
statement continuation. When a statement does not fit on a line, it can be continued by
putting an escape character ( \ ) immediately before the end of the line. This is consistent
with the meaning of the escape character; in this case, it means escape the end of line
that follows; it is not really an end of line. This allows us to continue the define statement
on the next line.
Since we are passing addresses to strings, we need to be careful that we don't
destroy the original data. The first thing we do in strCmpPk, therefore, is to use
strnepy to copy the input string to a work area . Once we have protected the original
data, we pack the two strings to be compared and then use the stremp function to
determine if the reformatted data are equal.
The strPk function merits a little discussion. It uses a while loop to process the
input string. Each iteration of the loop scans the valid characters, as defined in ALPHA,
looking for invalid data. When strPk finds an invalid character — and there will
always be at least one, the input string delimiter — it ( 1 ) replaces that character with a
string delimiter, ( 2 ) concatenates the input string to the output string, and then
( 3 ) adjusts the input string pointer to one past the last scanned character. This is neces-
sary to point it to the data beyond the string just copied.
But what if the pointer adjustment puts it beyond the end of the input data? In that
case, we will have a delimiter, because we used strnepy to copy the data from the orig-
.
inal input, and it pads the output string with delimiters. Therefore, the character after
the original string delimiter is another delimiter. Note that we made the output area one
character larger than the input area to provide for this extra delimiter even with a
I full string.
At least three test cases are required to test the program. We show two of them in
the result. In addition to an equal and greater - than result, you should conduct at least
one less - than test case.
11.6 String/Data Conversion

A common set ol applications format data by either converting a sequence o
characters into corresponding data types or vice versa , Two such application
are parsing and telecommunications.
C already has an extensive set of data conversion functions created to
scanj and print / . The C standard also includes two sets of memory formattin
functions, one for ASCII characters and one for wide characters, that use
these same formatting functions. Rather than reading and write Files, how
ever, they read and w' rite strings in memory.
If
1
String to Data Conversion
The string scan function is called sscanf . This function scans a string as
though the data were coming from a file. Just like fscanf , it requires a for -
mat string to provide the formatting parameters for the data. All fscanj format
codes are valid lor the scan memory functions. Like fscanf , the scan memory
functions also return the number of variables successfully formatted. If they
reach the end of a string before all the format string conversion codes have
been used—that is, if they attempt to “read” beyond the end of string—they
return an end-of-file flag. The basic concept is shown in Figure 1 1 - 24.
sscanf
string variables
FIGURE 11 - 2 4 ssconf Operation
int sscanf (char* str, const char* frmt str , ...); _

The first parameter specifies the string holding the data to be scanned.
The ellipsis (...) indicates that a variable number of pointers identify the fields
into which the formatted data are to he placed.
Let ’s demonstrate the use of sscanf with an example. Assume that we
have a string that contains a name terminated with a semicolon, a four-digit
student number, an exam score, and a character grade. Each field is sepa -
rated from the rest by at least one whitespace character. A data sample
is
shown below.
Einstein , Albert; 1234 97 A

number as
In this problem, we want to format the name and student
strings, the score as an integer, and the grade
as a character. We therefore
f
5 construct the following format string:
r "%49[
A
;] %*c %4s %d %*[" ABCDF] %c"
l
s Let ’s examine each format code individually
.
First, we are expecting a
string of up to 49 characters terminated
by a semicolon. Note that we use the
scan set with a terminating token
of a semicolon .
Following the scan set that
14 Section 11.6 Stiing/Data Conversion
reads to a semicolon is a format code that reads and discards one character.
The asterisk is a flag indicating that the character is not to be stored. This
code is necessary because the terminating semicolon I mm the string remains
in the input string. We need to read and discard it .
The third format code is a simple four-character string. Any leading
whitespace is discarded with the string token ( ). Likewise, the third field,
the score, is converted to an integer format after the leading whitespace is
discarded.
Parsing the grade is a little more difficult. We can have one or more
whitespaces between the numeric grade and the alphabetic grade, but we
must dispose of them ourselves, because the character token ( % c ) does not
discard whitespace. Our solution is to use a scan set with the suppress (lag
( * ) to discard all characters up to the grade. Again, we use the negation to set
the terminating character. Having discarded all the leading whitespace, we
can now format the grade with a character token.
Using our example above, we can code the sscanj call as shown below.
sscanf(strln , "%25[ ; ]%*c%4s%d %*[ AABCDF]%c" ,

A
name, stuNo , &score, &grade);
Note that this code has six format codes hut only four variables because
the second and fifth fields have their formatting suppressed.
sscanf is a one-to- many function.

It splits one string into many variables.
Data to String Conversion

The string print function, sprint / , follows the rules ol J print/ . Rather than
'
sending the data to a file, however, it simply ‘‘ writes” them to a string. When
all data have been formatted to the string, a terminating null character is
added to make the result a valid string. If an error is detected, sprintf returns
any negative value, traditionally EOF. If the formatting is successful, it returns
the number of characters formatted, not counting the terminating null char-
acter. The string print operation is shown in Figure 1 1 - 25 .
sprintf
string variables
FIGURE 11 - 25 sprintf Operation

I he junction declaration for formatting memory strings is shown below .

int sprintf (
_
char* out string ,
const char* format string , ...);
The first parameter is a pointer to a string that will contain the formatted
.
output The format string is the same as printf and follows all of the same
rules. The ellipse (...) contains the fields that correspond to the format codes
in the format string.
It is our responsibility to ensure that the output string is large enough to
hold all the formatted data. If it is not large enough, sprint / destroys memory
.
contents after the output string variable The wide-character function,
swprintf , on the other hand limits the output. The additional parameter,
max _out , specifies the maximum wide characters.
sprintf is a many-to-one function.

It joins many pieces of data into one string.
EXAMPLE 11 - 1 2 Demonstrate Memory Formatting

To demonstrate the memory formatting functions, we first format a string to
variables and print them. We then format them back to a string and print the
string. The data use the format described in "String to Data Conversion. I he
"
code is seen in Program 11- 16.
PROGRAM 11 - 1 6 Demonstrate Memory Formatting

1 /* Demonstrate memory formatting.
2 Written by:
3 Date:
4 */
include <stdio.h >
5
6 ^
7 int main ( void )
8 {
10 char strng[80] = "Einstein , Albert; 1234 97 A" ;
11 char strngOut[ 80];
12 char name[ 50 ];
13 char id[5];
14 int score ;
15 char grade;
16
17 // Statements
18 printf("String contains: \"%s\" \n" , strng);
continues
/ 16 Section 11.6 String/ Data Conversion
PROGRAM 1 1 - 1 6 Demonstrate Memory Formatting ( continued )

19
sscanf(strng , •• %49[ ;] % *c %4s %d %c " ,
A
20
21 name, id , &score, &grade);
22
23 printf( " Reformatted data: \n" );
24 printf( " Name: \"%s\"\n" , name );
25 printf( " id: \ " %s\"\n " , id );
26 printf( " score: %d\n" , score );
27 printf(" grade: %c\n " , grade);
28
29 sprintf(strngOut, "%s %4s %3d %c " ,
30 name , id , score , grade);
31 printf( " New string: \ " %s\"\n " , strngOut);
32 return 0 ;
33 } // main
Results:
String contains: "Einstein , Albert; 1234 97 A"
Reformatted data:
Name: "Einstein , Albert "
id: " 1234"
score: 97 I
grade: A
New string: " Einstein , Albert 1234 97 A"
EXAMPLE 1 1 - 1 3 Range of Internet Addresses

As a practical example, let ’s write a program that , given two Internet
addresses, determines all of the addresses in their range. An Internet address
is a 4 - byte Dotted Decimal Notation address in the form X.Y.Z.T. Each byte
can have a decimal value between 0 and 255.
Io determine the range , we multiply each portion of the address by 256
and add it to a counter. The second address is then subtracted from the first ,
which gives the number of addresses in the range.
I his program could be used by system administrators when their organi-
zation is given a range of addresses. For example , if an organization is given
the range of address 193.78.64.0 to 193.78.66.255, the system administrator
needs to know how' many addresses are available for computers to be con-
nected to the Internet . Program 11 - 17 answers this question .
PROGRAM 1 1 - 1 7 Determine Internet Addresses in a Range

1 /* Given two Internet addresses, determine the number
2 of unique addresses in their range.
continued
r n
Chapter 11 Strings 71 /
PROGRAM 11 - 1 7 Determine Internet Addresses in a Range (continued )

3 Written by:
4 Date:
5 */
8
9 int main ( void )
10 {
12 unsigned int strt[4];
13 unsigned int end[4 ];
14 unsigned long addl = 0 ;
15 unsigned long add 2 = 0 ;
16 unsigned long range;
17
18 char addrl[ 15 ];
19 char addr2[ 15 ];
20
21 // Statements
22 printf ( "Enter first address: " );
23 fgets ( addrl , sizeof (addrl ) , stdin);
24
25 printf ( "Enter second address: " );
26 fgets ( addr2 , sizeof ( addr2) , stdin );
27
28 sscanf (addrl , " %d %*c %d %*c %d %*c %d \n " ,
29 &strt[3], &strt( 2], &strt[ l ], &strt( 0]);
30 sscanf ( addr2 , " %d %*c %d %*c %d %*c %d\n " ,
31 &end[ 3 ], &end[ 2 ], &end[ l ], &end[0 ]);
32
33
34
for (int i
{
= 3 ; i >= 0 ; i — )
35 addl = addl * 256 + strt[ i];

36 add 2 = add2 * 256 + end[ i ];
37 > // for
38 range = abs ( addl -
add 2 ) + 1 ;
39
40 printf ( "\nFirst Address: %s" addrl );
/
41 printf ( "Second Address %s" addr2);

/
42 printf ( " \nThe range: %ld \n " , range);

43
44 return 0;
45 } // main
continued
718 Section 11.7 A Progromming Example — Morse Code
PROGRAM 11 - 1 7 Determine Internet Addresses in a Range ( continued

)
Results:
Enter first address: 23.56.34.0
Enter second address: 23.56.32.255
First Address: 23.56.34.0

Second Address 23.56.32.255
The range: 258
11.7 A Programming Example — Morse Code

Morse code, patented by Samuel F. B. Morse in 1837, is the language that
was used to send messages by telegraph from the middle of the nineteenth
century until the advent of the modern telephone and today’s computer-
controlled communications systems. In Morse code, each letter in the alpha-
.
bet is represented by a series of dots and dashes, as shown in Table 11 - 3
TABLE 1 1 - 3 Morse Code
Program 11 - 1 8 encodes (converts) a line of text to Morse code and

decodes (converts) Morse code to a line of text. We use a two-dimensional
array of pointers in which each row has two string pointers, one to a string
containing English characters and one to a string containing the corre -
sponding Morse code. Note that the pointers are stored in the programs
memory, while the strings are stored in the heap. The array structure is
shown in Figure 1 1 - 26.
T
Chapter 11 Strings / 19
Pointers to Pointers to
characters Morse code
A
[ A \O # \0
A
rBW # \0
A
z \o # \0
A
\0 $ | $ 1 # | \0
FIGURE 1 1 - 26 Character to Morse Code Structure
.
Each column has 27 pointers Each pointer in the first column points to
a string ol length one, which contains one English letter ( uppercase) Each .
pointer in the second column points to a string of varying size, a ragged array,
which contains the corresponding Morse code for the English letter.
The program is menu driven. The menu, which is shown in Figure 1 1 - 27,
has three options: encode English to Morse code, decode Morse code to
English, and quit.
MENU
E encode
D decode
Q quit
Enter option: press return key:
FIGURE 11 - 27 Morse Code Menu
The program begins by initializing the conversion table. It then displays

the menu and loops until the user quits. Each loop either encodes a line of
English text or decodes a line of Morse code. The only allowable characters
are the alphabetic characters and spaces. When Morse code is entered,
each coded character is terminated by a pound sign ( # ). Morse code space
is represented by two dollar signs ( $ $ ) . The complete design is shown in
Figure 1 1 - 28.
720 Section 11.7 A Programming Example — Morse Code
Morse
code
(+)
print getlnput
menu getlnput encode decode
Output/]
convert convert
A
FIGURE 11 - 28 Morse Code Program Design
If the encode or decode function detects an error —such as an invalid

character or an invalid Morse code sequence—it returns an error code, and
the calling program prints an error message.
The solution, as shown in Program 1 1 - 18, uses three arrays. The first is
the encode/decode (encDec) array. As seen in Figure 1 1 - 26, this is a two-
dimensional array of pointers to English and Morse code values The second .
array is an array of 81 characters to hold the input line. The third is an array
of 8 1 characters to hold the output line. To ensure that we do not overrun the
output line, we limit the input string to 16 characters when we are reading
English text . When appropriate, we provide some analysis to points in the
individual functions.
PROGRAM 11 - 1 8 Morse Code: mam

1 / * Convert English to Morse or Morse to English.
2 Written by:
3 Date Written:
4 */
5 # include < stdio.h >
6 # include < stdlib.h >
7 # include < string.h >
8 # include <ctype.h >
9 # include < stdbool h >.
10
11 # define FLUSH while( getchar( ) 1 = '\ n' )
12 # define STR LEN 81
13
15 char menu ( void );
16 void getlnput ( char * inStr );
continued
7 in
PROGRAM 1 1 - 1 8 Morse Code: main ( continued )

17 void printOutput (char* inStr , char* outSt);
18 bool encode (char* (*encDec)[ 2 ],
19 char* inStr ,
20 char* outStr );
21 bool decode (char* ( *encDec )[ 2],
22 char* inStr,
23 char* outStr );
24 int convert (char* (*encDec)[ 2 ],
25 char* si ,
26 int col,
27 char* s2);
28
30 {
32 char* encDec [ 27][ 2 ] =
33 {
34 { " A ", >,
35 { "B ” ,
36 { "C " ,
37 { "D " , -..#" > /
38 { "E", >,
39 { "F" , >r
40 { "G" , —.# "
41 { " H", # " },
42 { II T II
± 9 }r
43 { "J ", . #" >,
44 { "K " , >,
45 { "L" , >,
46 { "M " ,
47 { "N " ,
48 { "0" , #”
49 { "P" ,
50 { "Q " f
51 { "R " , .-.# M
52 { "S",
53 { "T",
54 { "U" ,
55 { " V" , > #
56 { "W " , >,

57 { "X",
58
59
{
{
"Y",
"Z " ,
-. — #M > /
60 { $$# M >,
continuec
PROGRAM 11 - 18 Morse Code: main ( continued )

61 }; // Encode / Decode array
62 char inStr [ STR LEN ]; _
63 char outStr [ STR LEN ]; _
64 char option;
65 bool done = false;
66
67 // Statements
68 while (!done)
69 {
70 option = menu ( );
71 switch ( option)
72 {
73 case E' :
74 getlnput ( inStr);
75 if ( lencode ( encDec , inStr , outStr))
76 {
77 printf( " Error! Try again " );
78 break ;
79 } // if
80 printOutput ( inStr , outStr );
81 break ;
82 case 'D' : getlnput ( inStr);
83 if ( idecode ( encDec , inStr , outStr))
84 {
85 printf( " Error! Try again " );
86 break;
87 } // if
88 printOutput ( inStr , outStr );
89 break ;
i,
90 default :
91 done = true ;
92 printf( " \nEnd of Morse Code.Nn " );
93 break ;
94 } // switch
95 > // while
96 return 0;
97 } // main
Program 1 H 8 Anolysu
.oteworthy P int is the default condition . We test for only two

ODtions
be ou e w
proaram
"
th,"
j
6
Ptl
^ ^' * ’
|
f °is neither'
n Can
menu
one or
^ assume quit. We can do this
Program 11 - 191 At this point in the
,ly three values . We test for two and
default the third ^ ° °
T u
PROGRAM 1 1 - 1 9 Morse Code: Menu

l
2 Display menu of choices; return selected character ,
3 Pre nothing
4 Post returns validated option code
5 */
6 char menu ( void )
7 {
9 char option;
10 bool validData ;
11
12 // Statements
13 printf("\t\t\tM E N U \n");
14 printf( " \t\tE) encode \n" );
15 printf( "\t\tD ) decode \nM );
16 printf( " \t \tQ ) quit \n" );
17
18 do
19 {
20 printf ( " \nEnter option: press return key: " );
21 option = toupper (getchar( ));
22 FLUSH ;
23 if (option == ' E' | | option == 'D' ||
24 option == ' Q' )
25 validData = true;
26 else
27 {
28 validData = false;
29 printf("\aEnter only one option\n " );
30 printf(" \tE, D , or Q\n " );
31 > // else
32 > while (!validData );
33 return option;
34 > // menu
Program 11 - 19 Analysis The menu function in Program 11 - 19 displays the options and reads the user's
choice. It then validates the choice and if invalid, displays an error message and asks
for the option again. Although it is always good design to validate the user input, it
must be validated in the correct place. Since menu is communicating with the user,
this is the logical place to do the validation. One more point: As the option is read, it
is converted to uppercase. Not only does this make the validation simpler, but it also
simplifies the switch statement in main. Whenever you have a single character code,
convert it to uppercase or lowercase for processing in the program .
The input string is read in Program 1 1 - 20.

PROGRAM 11 - 20 Morse Code: Get Input

1 /* ===== ===== getlnput ===== =====
2 Reads input string to be encoded or decoded.
3 Pre inStr is a pointer to the input area
4 Post string read into input area
5 */
6 void getlnput ( char* inStr)
7 {
8 // Statements
9 printf ( " \nEnter line of text to be coded: \n");
10 fgets _
( inStr , STR LEN , stdin );
11
12 // Eliminate newline in input string
13 *( inStr 1 + strlen( inStr )) = * \0' ;
-
14
15 if (isalpha( *inStr) && strlen( inStr) > 16 )
16 {
17 // Exceeds English input length
18 printf( " \n***WARNING: Input length exceeded: " );
19 printf( "Only 16 chars will be encoded.\a\a\n");
20 *( inStr + 16 ) = * \0' ;
21 >
// if
22 return ;
23 } // getlnput
Program 1 1 -20 Analysis Note that to prevent a runaway string in getlnput we use the fgets function and
specify the maximum number of characters in the string area. This function creates a
minor problem, however; it places the newline in the input string . We must therefore
overlay it with a null character.
The code to print the output of this program is shown in Program 11- 21.
PROGRAM 1 1 - 2 1 Morse Code: Print Output

i
1 ===== printOutput =====
2 Print the input and the transformed output
3 Pre inStr contains the input data
4 outStr contains the transformed string
5 Post output printed
6 */
7 void printOutput (char* inStr , char* outStr )
8 {
9 // Statements
10 printf( " \ nThe information entered was: \n " );
11 puts( inStr );
12 printf( " \nThe transformed information is: \n " );
continned
V
PROGRAM 11 - 2 1 Morse Code: Print Output ( continued )

13 puts(outStr );
14 return ;
15 > // printOutput
Program 11 - 22 transforms character data to Morse code .
PROGRAM 11 - 22 Morse Code: Encode to Morse

l
2 Transforms character data to Morse code
3 Pre encDec is the conversion table
4 inStr contains data to be put into Morse
5 Post data have been encoded in outStr
6 Return true if all valid characters ;
7 false if invalid character found
8 */
9 bool encode (char* ( *encDec )[ 2 ],
10 char* inStr , char* outStr )
11 {
13 char si(2];
14 char s2[6 ];
15 int error = 0 ;
16
17 // Statements
18 outStr[ 0] = ’ \0' ;
19 while ( *inStr != ’ \0' && lerror )
20 {
21 sl[ 0 ] = toupper( *inStr );
22 si[ 1 ] = ' \0 * ;
23 error = Iconvert (encDec, si, 0, s2);
24 strcat ( outStr , s2 );
25 inStr++ ;
26 } // while
27 return ( lerror );
28 } // encode
is declared in the formal parameter list in state

-
Program 1 1 - 22 Analysis Note how the encode/decode table
To understand it, let's use right left
- analysis ( see
ment 9 . It uses a complex type.
Chapter 10 ).
char * ( * encDec ) [2]

Reading this declaration, at 0 we have encDec . Moving to the right, we find the
1 , which is ignored. Moving left, at 2 we find a pointer. We therefore have encDec as
a pointer. Moving right to 3, we see that it is a pointer to
an array of two . Moving to the
left to 4, we see that it is an array of pointers. PI
At this point, we see that encDec is a pointer to an array of two pointers. Now,
moving back to the right, we see that 5 is empty, so we go to the left and find char at 6.
The final type is therefore a pointer to an array of two pointers to char. Referring back
to Figure 11 - 26, we see that this is exactly what we have.
The code to decode Morse code to English is shown in Program 11 - 23.
PROGRAM 11 -23 Morse code: Decode to English

1 ==== decode
2 Transforms Morse code data to character string
3 Pre encDec is the conversion table
4 inStr contains data to transform to string
5 Post data encoded and placed in outStr
6 Return true if all valid characters;
7 false if invalid character found
8 */
9 bool decode (char * ( *encDec )[ 2 ],
10 char* inStr , char* outStr )
11 {
13 char si[6];
14 char s2[ 2];
15 bool error = false;
16 int i;
17
18 // Statements
19 outStr[ 0 ] ='\0' ;
20 while (*inStr != '\0 && lerror )
21 {
22 for ( i = 0 ; i < 5 && *inStr != i++, inStr++)
23 si[ i ] = * inStr ;
Pi-ogr
24
25 sl[i] = *inStr ;
26 si[ ++i ] = • \0';
27
28 error
- !convert ( encDec , si , 1 , s2);
29 strcat (outStr , s2 );
30 inStr++ ;
31 > // while
32 return ( lerror);
33 > I I decode
r
Program 11 - 24 converts characters to Morse code and Morse code to

characters.
PROGRAM 11 - 24 Morse Code: Convert Codes

l
2 Looks up code and converts to opposite format
3 Pre encDec is a pointer decoding table
4 si is string being converted
5 s2 is output string
6 col is code: 0 for character to Morse
7 1 for Morse to character
8 Post converted output s2
9 */
10 int convert (char* (*encDec)[2 ],
11 char* si , int col , char* s2 )
12 {
14 bool found = false;
15 int i;
16
17 // Statements
18 for ( i = 0; i < 27 && !found ; i++)
19 found = !strcmp(sl , encDec[ i ][col ]);
20
21 if (found )
22 strcpy ( s2 , encDec [ i - l ][(col + 1 ) % 2 ]);
23 else
24 *s2 = ' \0 * ;
25
26 return found ;
27 } // convert
28 / / === === End of Program =====
rogram 1 1 - 24 Analysis The convert function does the actual conversion. Note, however, how it is designed
to handle the conversion both from English to Morse and Morse to English. The only dif -
ference between the two conversions is which pointer we want to use. We pass the col-
umn to be used for the search as a parameter. Once we have located the correct string,
we use the formula shown below to pick up the matching string ( see statement 22) .
(col + 1) % 2
If we are searching on the English letters in column 0, then the modulus of the column
(0 + 1 ) is column 1, which contains the pointer to the matching Morse code string. Con-
versely, if we are searching Morse code using the pointers in column 1, then the modulus
of the column ( 1 + 1 ) is column 0, which contains the pointer to the English letter.

In this section , weve formalized some ol the principles of good programming
that we’ve discussed throughout the text. Although you will find little in this
discussion of software engineering that relates directly to the subject of
strings, all of the string functions have been written using the principles dis-
cussed on the following pages.
Program Design Concepts

You will study many different analysis and design tools as you advance in
computer science. Since this text deals primarily with program analysis and
design , we will discuss the primary tool we have used throughout the text: the
structure chart .
The overriding premise of good design is that the program is modular;
that is , it is well structured. This is the sine qua non of good programming. A
program ’s degree of good design can he measured by two principles: Its mod -
—
ules are independent that is, their implementation is hidden from the
—
user and they have a single purpose.
Information Hiding
Information hiding is the principle of program design in which the data
structure and functional implementation are screened from the user. Mod -
ules are independent when their communication is only through well-defined
parameters and their implementation is hidden from the user. The purpose of
the function should be defined in its inputs and outputs, and the user should
not need to know how it is implemented or how its data are structured . Press-
man states it well when he says:
Hiding implies that effective modularity can be achieved by

defining a set of independent modules that communicate
with one another only that information necessary to achieve
software function . ... Hiding defines and enforces access
constraints to both procedural detail within a module and
any local data structure used by the module.8
The concept of information hiding is the basis of the object -oriented

design and programming. When you study data structures, you will see
another technique used for information hiding, the abstract data type.
Cohesion
The most common weakness of function design is combining related pro-
cesses into one primitive function. In Chapter 4 we discussed the concept
- Hill
8. Roger S . Pressman , Software Engineering : A Practitioner’s Approach , 2 nd ed , McGraw
Series in Software Engineering and Technology ( New York: McGraw- Hill, 1982 ), p. 228.
that each module ( function ) should do only one thing. This principle ol
structured programming is known as cohesion . Cohesion , first discussed by
Larry Constantine in the 1960s,9 is a measure of how closely the processes in
a function are related .
We are concerned with cohesion for three primary reasons. The first and
most important is accuracy. The more cohesive a function is, the simpler it is.
I he simpler it is, the easier it is to write , and the fewer errors it will have.
This is closely related to the second reason for high cohesion, maintainability.
II a function is easy to understand , it is easy to change . This means that we
will get the job done faster and with fewer errors.
Finally, cohesive modules are more reusable . Reusability is also closely
related to the concepts of accuracy and ease of use. Existing functions have
stood the test of time and have been tempered in the heat of use . They are
more likely to be error free , and they certainly arc easier and faster to develop.
Cohesion is most applicable to the primitive functions in a program ,
those that are at the bottom of the structure chart and less applicable to the
controlling functions that appear above the lowest level. This does not mean ,
however, that cohesion can be ignored at the higher levels. To make the point
with an absurd example, you wouldn ’t write a program to manage your check -
book and maintain your daily calendar. Even though both of these processes
are related to things you do, they are so unrelated that you wouldn ’t put them
in the same program . The same concept applies in a program . For example , at
the lower levels of your program , you shouldn ’t combine functions that read
data with functions that print a report.
The seven levels of cohesion are shown in Figure 1 1 - 29 .
low RANGE OF COHESION high
unacceptable acceptable good best
FIGURE 11 - 29 Types of Cohesion
Functional Cohesion
Functional modules that contain only one process demonstrate functional
cohesion. This is the highest level of cohesion and the level that we should
hold up as a model. Using the example of printing a report , the report func-
tion should call three lower-level functions , one to get the data , one to format
9 . E. N . Yourdon and L. L. Constantine, Structured Design ( Upper Saddle River, N .J.: Prentice
Hall , 1978 ).
an d print the report header, and one to format and print the data . This design
.
is shown in Figure 11 -30 The print report heading function is optional
because it is called only when a page is lull .
Sequential Cohesion
A sequential module contains two or more related tasks that are closely tied
together, usually with the output of one flowing as input to the other. An
exa mple of sequential cohesion is shown in the calculations for a sale. The
design for this function might well he
1 . Extend item prices
2. Sum items
.
3 Calculate sales tax
4. Calculate total
print
report
o
getData
I print report
heading
printLine
FIGURE 11 - 30 Example of Functional Cohesion
In this example , the first process multiplies the quantity purchased by the
price. I he extended prices are used by the process that calculates the sum of
the purchased items. This sum is then used to calculate the sales tax, which
is finally added to the sum to get the sale total. In each case , the output of
one process was used as the input of the next process.
Although it is quite common to find the detail code for these processes
combined into a single function , it does make the f unction more complex and
less reusable. On the other hand , reusability would he a concern if the same
or similar calculations were being made in different parts of one program .
Communicational Cohesion
Communicational cohesion combines processes that work on the same
data. It is natural to have communicational cohesion in the higher modules
in a program , but you should never find it at the primitive level . For example,
consider a function that reads an inventory file, prints the current status of
the parts , and then checks to see if any parts need to he ordered . The
pseudocode for this process is shown in Algorithm 11 - 1.
V
ALGORITHM 11 - 1 Process Inventory Pseudocode

Algorithm Process Inventory
1 while not end of file
1 read a record
2 print report
3 check reorder point
2 end while
end Process Inventory
All three ol these processes use the same data. II they are calls to lower -
level iunctions, they are acceptable. If the detail code is found in the func -
tion, however, the design is unacceptable.
The first three levels of cohesion are all considered to be good structured
programming principles. Beyond this point, however, ease of understanding
and implementation, maintainability, and accuracy begin to drop off rapidly.
The next two levels should be used only at the higher levels of a structure
chart, and then only rarely.
Well-structured programe are highly cohesive and loosely coupled.
Procedural Cohesion
The fourth level of cohesion, procedural cohesion, combines unrelated pro-
cesses that arc linked hv a control flow. (This differs from sequential cohe-
sion, where data flows from one process to the next.) As an example, consider
the high- level logic main line of a program that builds and processes a list. A
procedural How could look like Algorithm 1 1 - 2.
ALGORITHM 11 - 2 Process List Pseudocode

Algorithm Process List
1 open files
2 initialize work areas
3 create list
4 print menu
5 while not stop
1 get users response
2 if locate ...
3 if insert ...
4 if delete ...
5 print menu
6 end while
7 clean up
8 close files
end Process List
idion 11.8 Software Engineering
A much better approach would be to have only three function calls in

main , initialize, process , and end . Not only is this easier to understand , but it
also simplifies the communication .
Temporal Cohesion
The fifth level, temporal cohesion , is acceptable only over a limited range of
processes. It combines unrelated processes that always occur together. Two
temporally cohesive functions are initialization and end of job. They are
acceptable because they are used only once in the program and because thcv
are never portable . Recognize , however, that they should still contain calls to
functionally cohesive primitive functions whenever practical.
Logical and Coincidental Cohesion

The last two levels are seldom found in programs today. Logical cohesion
combines processes that are related only by the entity that controls them. A
function that conditionally opened different sets of files based on a flag
passed as a parameter would be logically cohesive . Finally, coincidental
cohesion combines processes that are unrelated. Coincidental cohesion
exists only in theory. We have never seen a productional program that con -
tained coincidental cohesion .
Summary
We have discussed two design concepts in this chapter: information hiding
and cohesion . Cohesion describes the relationship among processes within a
function . Keep functions as highly cohesive as possible. When designing a
program, pay attention to the levels of cohesion . It is much easier to design
high cohesion into a program than it is to program it in . For more information
about these concepts, refer to The Practical Guide to Structured Systems
Design by Meilir Page -Jones. 10
.
10. Meilir Page -Jones The Practical Guide to Structured Systems Design , Yourdon Press Com -
puting Series ( Upper Saddle River, N .J .: Prentice- Hall , 1988 ).
1

1. Do not conluse characters and string constants: The character constant
is enclosed in single quotes, and the string constant is enclosed in double
quotes.
2. Remember to allocate memory space for the string delimiter when
declaring and defining an array of char to hold a string.
.
3 Strings are manipulated with string functions, not operators.
4. I he header hie string .h is required when using string functions.

5. I he standard string functions require a delimited string. You cannot use
them on an array of char that is not terminated with a delimiter.
.
6 Do not conluse string arrays and string pointers. In the following exam -
ple, the first definition is a string ( array of characters), and the second is
a pointer to a character ( string). Each is shown with a typical initializa -
tion statement.
char str[20 ];
strcpy ( str, " Now is the time" );
char* str;
str = " Now is the time";
.
7 Passing a character to a function when a string is required is another
common error. This is most likely to occur with the formatted input/
output functions, in which case, it is a logic error. Passing a character in
place of a string when a function header is declared is a compile error .
.
8 Using the address operator for a parameter in the scanJ function with a
string is a coding error. The following is an invalid call and will most
likely cause the program to fail:
scanf( " %s" , &string ); // Coding Error
.
9 It is a compile error to assign a string to a character, even when the char -
acter is a part of a string.
10. Using the assignment operator with strings instead of a function call iis a
compile error.
char string [20];

string = "This is an error "; // Compile Error
I 1 Since strings are built in an array structure, they may he accessed with
.
indexes and pointers. When accessing individual bytes, it is a logic error
to access beyond the end of the data structure array .
( )
II Summary
11.10 Key Terms

cohesion information hiding
coincidental cohesion length -controlled string
communicational cohesion logical cohesion
delimited string procedural cohesion
delimiter sequential cohesion
scan set temporal cohesion
fixed -length string
11.11 Summary
Strings can be fixed length or variable length .
A variable- length string can he controlled by a length or a delimiter.
The C language uses a null - terminated ( delimited ) variable - length string.
A string constant in C is a sequence of characters enclosed in double
quotes. A character constant is enclosed in single quotes.
When string constants are used in a program , C automatically creates an
array of characters, initializes it to a null - terminated string, and stores it,
remembering its address.
To store a string, we need an array of characters whose size is one more
than the length of the string.
There is a difference between an array of characters and a string. An array
of characters is a derived data type ( built on the type character ) and is an
intrinsic structure in C . A string is a data structure that uses the array as
its basic type. It requires a delimiter for all of its operations.
We can initialize a string:
With an initializer when we define it .
By using a string copy function
By reading characters into it .
A string identifier is a pointer constant to the first element in the string.
An array of strings is a very efficient way of storing strings of different sizes.
The sccmf and printf functions can he used to read and write strings using
the % s format .
The scan set format , %[ . . . ], can also be used in the sccinf function to
read a siring.
The functions gets , fgets , puts , and / puts are used for reading and writing
strings. I hey transform a line to a string or a string to a line.
m
J The functions sscanf and sprintf are used for reading and writing strings
—
Irom memory that is, memory- to- memory formatting.
J I he functions strcpy and strncpy are used to copy one string into another.
—I I he functions strcmp and strncmp are used to compare two strings.

1 he function strlen is used to determine the length of a string.
The functions strcat and stmcat are used to concatenate one string to the
end of the other.
The functions strchr and strrchr are used to look for a character in a string.
The function strstr is used to look for a substring in a string.
The functions strspn and strcspn are used to locate the position of a set ol
characters in a string.
Information hiding is the principle of programming in which the data
structure and function ’s implementation are screened from the user.
The cohesion principle dictates that each module must do only a single job.
The sine qua non of good programming is modularity.
Program design has seven layers of cohesion ; only the first three ( func -
tional . sequential , and communicational ) should he used lor lower- level
functions. The last two (logical and coincidental ) should never he used .
11.12 Practice Sets

Review Questions
-
1. The two basic techniques for storing variable length data are fixed - length
strings and delimiters.
a. True
b. False
2. To initialize a string when it is defined , it is necessary to put the delimiter
character before the terminating double quote, as in hello \ 0 ".
"
a . True
.
b False
3. Both the gets and fgets functions change the newline to a delimiter.
a . True
b. False
4. The function that can be used to format several variables into
one string
in memory is string token ( strtok ) .
a . True
b. False
12 Practice Sets
5. A ( n ) is a series of characters treated as a unit .
.
a array
.
b character
c. field
d. record
e. string
6. The delimiter in a C string is
a . newline
b. defined by the programmer
c. a del character
d . a null character
e. not specified in ANSI C
7. Which of the following statements about string variables is false?
a. The assignment operator copies the value of one string variable to
another.
b. The array bracket operator is not needed if a string is defined as a
pointer to character.
c. The string name is a pointer.
d . When a string is initialized at definition time, C automatically inserts
the delimiter.
e. When a string is read with scarify it automatically inserts the delimiter.
8. Which of the following printf statements will not print a string? ( Assume
that name is properly defined and formatted . )
.
a printf ( " % s " , name ) ;
b. printf ( " % 30 s " , name ) ;
c . printf ( " % - 30 s " , name ) ;
d . printf ( " % 30 [ \ 0 ] " , name ) ;
A
9. The function reads a string from the standard input

device.
a . fgets
b. fputs
.
c gets
d . getstr
e. puts
10. The siring manipulation function returns the num-
ber ol characters in the string excluding the null character.
a . strernp
I ). strepy
c. strlen
d . strsize
e. strtok
1
1 1 . Which of the following statements determines if the contents of s t r i n g 1

are the same as the contents of s t r i n g 2 ? Assume that s t r i n g 1 and
s t r i n g 2 are properly defined and formatted strings.
a if. ( stringl == s t r i n g2 )
b if. ( strcmp ( s t r i n g l , s t r i n g 2 ) )
c if. ( strcmp ( s t r i n g l , s t r i n g 2 ) = = 0 )
.
d if ( strcmp ( s t r i n g l , s t r i n g 2 ) < 0 )
.
e if ( strcmp ( s t r i n g l , s t r i n g 2 ) > 0 )
.
12 The string manipulati on function adds a string to
the end of another string.
.
a stradd
b. strcat
c . strcmp
d . strepy
.
e strtok
.
13 The string manipulat ion function determines il a
character is contained in a string.
a . strehr
b. strehar
c. strcmp
d. strepy
e. strtok
.
14 The function is used to format a string into variables
i n memorv
*
.
a . scanf
b. sscanf
c. strfrmt
d . strscan
.
e strtok
15. The best level of cohesion is
.
a coinciden tal
b. communi cational
.
c procedura l
d . sequentia l
e. functiona l
Exercises
), a n d * ( x + 4 ) for the following declaration
:
16. Find the value of *x, * ( x + l
";
char* x = "The life is beautiful
Practice Sets
( y + 4 ) lor the following program

17 Find the value ot * y, * ( y + l , and *
)
.
segment:
char x [ ] = "Life is beautiful" ;

char* y = &x [3 J ;
. block?
18 What is the error in the following program
{
char* x;
scanf( " %s" , x);
>
19. What would be printed from the following program
block?
{
char si[50 ] = "xyzt" ;
char* s2 = " xyAt" ;
int dif ;
dif = strcmp (si , s2 );
printf( " \n%d " , dif );
>
20 . What would be printed from the following program block?
{
char si[ 50] = " xyzt" ;
char* s2 = " uabefgnpanm" ;
char* s3;
char* s4 ;
char* s5 ;
char* s6;
s3 = si ;
s4 = s2 ;
strcat (si , s2 );
s5 = strchr( si , 'y ' );
s6 = strrchr(s2 , 'n ' );
printf ( " \n%s" , s3 );
printf ("\n%s" , s4);
printf ( "\n%s" , s5);
printf ( "\n%s" , s6 );
>
.
21 W hat would be printed from the following program block?
{
char si[50] = " uabefgnpanm" ;
char* s2 = "ab" ;
continued
char* s3 = " pan " ;

char* s4 = " bef " ;
char* s5 = " panam" ;
char* s6;
char* s7;
char* s8 ;
char* s9 ;
s6 = strstr (si, s2 )
s8 = strstr (si , s4 )
printf ( " \n%s" , s6 )
printf ( " \n%s" , s7 )
printf ( " \n%s" , s8 )
printf ("\n%s" , s9 )
>
22. \\ hat would be printed from the following program block?
{
char* si = "abefgnpanm" ;
char* s2 = "ab";
char* s3 = " pan";
char* s4 = " bef";
char* s5 = " panam " ;
int dl;
int d 2 ;
int d 3;
int d4 ;
dl = strspn (si , s2);
d 2 = strspn (si , s3);
d3 = strcspn ( si , s4);
d4 = strcspn (si , s5 );
printf ("\n%d " , dl );
printf ("\n%d " , d 2);
printf ("\n %d " , d 3);
printf ("\n%d " , d4);
}
23. What would he printed from the following program block ?
{
char* w = " BOOBOO" ;
printf ( "%s\n " , "DOO" );
printf ( "%s\n " , "DICK " + 2 );
printf ( "%s\n", "DOOBOO" +3 );
printf ("%c\n " , w( 4]);
printf ( "%s\n" , w+4);
continued
w++;
w++;
printf ( " %s\n " ,w );
printf ("%c\n" , *(w+l));
>
24. What would be printed from the following program
block?
{
char* a[ 5 ] =
{"GOOD" , " BAD " , " UGLY " , " WICKED" , "NICE"};
printf ("%s\n" , a[0]);
printf ( " %s\n " , *(a + 2 ));
printf (" %c\n" , *(a[ 2] + 2));
printf ( " %s\n" , a[3]);
printf ( "%s\n" , a[ 2 ]);
printf ("%s\n" , a[4]);
printf ( "%c\n " , *( a[ 3] +2));
printf ( "%c\n " , *( *( a+4)+3 ));
>
25. What would he printed from the following program block?
{
char c[ ] = " programming " ;
char* p ;
int i?
for ( p = &c[5]; p > &c [0]; P )
printf("%c" , * p);
= —
printf ( "\n");
for ( p = c+5 , i=0 ; p >= c ; p ,i++ )
printf( " %c " , *( p i )); - —
>
Problems
a unction t iat accepts a string ( a pointer to a character) and
^
^
1
I - h, . 1 C 3St 1 araCttr ky moving the null character one position to
the left
27’
deletes the
'
^ l
°“ ^
a character) and
^ ?Cti°n that accePts a string (a pointer to

UI
deleted
deletes |
all the trailing spaces at the end of the string.
and
Ses llThe 'leadl
^ spaceT15 “^8 (“ P°inter *°
29' 3
^
1
.
30 Write a ( unction that returns the number of times the character is found
in a string. I he ( unction has two parameters . The first parameter is a
pointer to a string. The second parameter is the character to he counted .
31 . Write a function that inserts a string into another string at a specified

position . It returns a positive number if it is successful or zero if it has
any problems, such as an insertion location greater than the length of the
receiving string. The first parameter is the receiving string , the second
parameter is the string to be inserted , and the third parameter is the
insertion ( index ) position in the first string.
.
32 Write a function that , given a string, a width , and an empty string for
output , centers the string in the output area . The function is to return 1
il the formatting is successful and 0 if any errors, such as string length
greater than width , are found .
33. Write a ( unction called newStrCpy that does the same job as strep} . The
declaration ( or your function is to be the same as the library function .
34 . Write a function called newStrCat that does the same job as strccit . The
declaration for your function is to be the same as the library' function .
35. Write a function called newStrCmp that does the same job as stremp . The
declaration for your function is to he the same as the library function .
36. A string is a palindrome if it can be read forward and backward with the
same meaning. Capitalization and spacing are ignored , l or example,
anna and go clog are palindromes. Write a function that accepts a string
and returns true if the string is a palindrome and false il it is not. Test
your function with the following two palindromes and at least one case
that is not a palindrome:
Madam, I ’ m Adam
Able was I ere I saw Elba
.
37 Todays spelling checkers do much more than simply test for correctly
spelled words. They also verify common punctuation . For example, a
period must be followed by only one space. W rite a program that reads a
text file and removes any extra spaces after a period , comma, semicolon ,
or colon . Write the corrected text to a new file.
Your program is to read the data using fgets and parse the strings
using strtok . Print your test data before and after they are run through
your program .
Practice Sets
Projects
.
38 Write a C program that converts
a string representing a number in
Roman numeral form to decimal form . The
symbols used in the Roman
numeral system and their equivalent s are given below.
I 1
V 5
X 10
L 50
C 100
D 500
M 1000
For example , the following are Roman numbers: XII 12 CII 102 ; XL
( ) ( )
( 40 ). The rules for converting a Roman number to a decimal number are
as follows:
a . Set the value of the decimal number to zero.
b. Scan the string containing the Roman character from left to right. If
the character is not one of the symbols in the numeral symbol set , the
.
program must print an error message and terminate Otherwise, con-
tinue with the following steps. ( Note that there is no equivalent
to
zero in the Roman numerals.)
If the next character is a null character ( if the current character is
the last character ), add the value of the current character to the
decimal value.
II the value of the current character is greater than or equal to the
value of the next character, add the value of the current character
to the decimal value.
,
II the value ol the current character is less than the next character
decimal value .
subtract the value ol the current character from the
Solve this project using parallel arrays. Do not solve it using a switch
statement.
39. Rework Project 38 to convert a decimal number to a Roman numeral.
the search-and-replace operation in a text
ftC
e» l i or.
S
e program is to
âtesonly
' muhave three function calls in main . The first
unction prompts the user to type a string of less than 80 characters
. It
un prompts t e user to type the search substring of 10 or fewer charac -
ters Finally, it prompts the user to type the replace substring ot
. 10 or
fewer characters.
° ° .'
ttUrrtnLes are found t returns the
original
characters long
string. Theoreticall y
y’ » h ' new stnng could be 800
( 80
.
identical characters replaced by 10 characters each ) Your function
must he able to handle overflow by using the realloc function to extend
the new string when necessary. ( Start with an output string of 80 charac-
ters and extend by 80 as necessary. ) The search -and - replace function
returns the address ot the new string.
Alter the search -and - replace function returns, a print function prints
the resulting string as a series of 80 character lines. It performs word -
wrap. I hat is , a line can end only at a space. If there is no space in 80
characters, then print 79 characters and a hyphen , and continue on the
next line.
Write each called function using good structured programming tech -
niques. It is expected to call subfunctions as necessary.
Run the program at least three times:
a . First , run it with no substitutions in the original input.
b. Second , run it with two or more substitutions.
c. Finally, run it with substitutions that cause the output to be at least
three lines, one of which requires a hyphen .
41 . Write a program that "speaks " pig- latin . Words in pig- latin are taken
from English. To form a word in pig- latin , the first letter of the English
word beginning with a consonant is removed and added at the end of the
word , adding the letters ay after the moved consonant . Words that begin
with a vowel are simply appended with ay. Thus, in pig- latin , pig - latin is
igpay-atinlay.
Your program will read a sentence at a time using fgets. It is to parse
the words into strings. As words are parsed, they are to be converted to
pig- latin and printed .
42. Write a program that provides antonyms to common words. An antonym
is a word with the opposite meaning. For example, the antonym of happy
is sad .
The data structure is to he patterned after the Morse code problem
in Section 1 1.7. Given a word , the program is to look for the word on
both sides of the structure and , if found , report the antonym . If the word
is not found , the program is to display an appropriate message.
The program is to use an interactive user interface. The words are to
he read from a dictionary file. Use your dictionary to provide at least 30
antonyms, including at least one from each letter in the alphabet.
Test your program by finding the antonyms for the first word , last
word , and a word somewhere in the middle on both sides of the struc-
ture. Include in your test data at least three words that are not found ,
one less than the first word on the left side , one in the middle some-
where, and one greater than the last word on the right side.
43. The Morse code program in this chapter is a public example of a crypto -
graphic system. We can apply the same techniques to encoding and
decoding any message. For example, we can substitute the letter Z for the
letter A , the letter Y for the letter B , and

so forth , to create the following
simple encoded message:
NZWZN , R ZN ZWZN
MADAM , I AM ADAM
Write a program that encodes and decodes messages using any user-
, include spaces and the
su pnlied code . To make it more difficult to read
common punctuation characters in the code . The code is to be read from
a text file to build the encode/decode array. The user is then given a menu
of choices to encode, decode , or enter a new code from the keyboard .
Test your program with the following code and message and with the
com plete alphabet entered in its encoded sequence so that it prints out
in alphabetical order:
ABCDEFGHIJKLMNOPQRSTUVWXYZ .,?!;
?Q.W ,EMRNTBXYUV!ICO PZA ;SDLFKGJH
Message
WNWLSVPLM , LN A
44. Write a program that parses a text file into words and counts the
number
of occurrences of each word . Allow for up to 100 different words . .After
the list has been built , sort and print it .
45 Modify the program you wrote in Project 44 to eliminate common words
.
such as the , a , an , and , for, and of . You may add other common words to
the list .
Chapter 12
Enumerated, Structure,
and Union Types
We have already discussed three of the six derived types: functions in
Chapter 4, arrays in Chapter 8, and pointers in Chapter 9. In this chapter, we
discuss the three remaining derived types: enumerated, structure, and union.
We also discuss a very useful construct, the type definition. The derived types
are shown in Figure 12 - 1.
Derived
Types i
Function Array Pointer Structure
Type
Chapter 4
Type
Chapter 8
Type
Chapter 9
Type i Union
Type i Type i
Enumerated
FIGURE 12 - 1 Derived Types
Objectives
To introduce the structure, union, and enumerated types
To use the type definition statement in programs
To use enumerated types, including anonymous types, in programs
To create and use structures in programs
(J To be able to use unions in programs
To understand the software engineering concept of coupling and to be able
to evaluate coupling between functions.
746 Section 12.2 Enumerated Types
12.1 The Type Definition ( typedef )

a C declaration that
Before discussing these derived types, lets discuss
applies to all of them —the type definition .
A type definition , typedef , gives a name to a data type
by creating a new
type that can then he used anywhere a type
is permitted . The format for the
type definition is shown in Figure 12 - 2 .
We can use the type definition with any type. For example, we can rede
,
line int to INTEGER with the statement shown below although we \voul <
never recommend this.
typedef int INTEGER ;
Note that the typedef identifier is traditionally coded in uppercase. Thi

alerts the reader that there is something unusual about the type. We sawth
previously with defined constants in Chapter 2 . Simple style standards sue
as this make it much easier to follow a program .
Any standard Traditionally

or derived type uppercase
typedef type IDENTIFIER;
keyword
FIGURE 12 - 2 Type- definition Format
One of the more common uses of the type definition is with complex d
Iarations. To demonstrate the concept , let ’s see how we can declare a rat
simple construct, an array of pointers to strings ( a ragged array ). Withoi
type definition , we must define the array as shown below.
char* stringPtrAry[ 20 ];
We can simplify the declaration by using a type definition to creal

string type and then defining the array using the new type as shown belov
typedef char* STRING;

STRING stringPtrAry[ 20 ];
12.2 Enumerated Types

I he enumerated type is a user- defined type based on the standard ini
type. In an enumerated type , each integer value is given an identifier c
Chapter 12 Enumerated, Structure, and Union Types / 47
an enumeration constant. We can thus use the enumerated constants as

symbolic names, which makes our programs much more readable.
Recall from Chapter 2 that a type is a set of values and a set of operations
that can be applied on those values. Each type also has an identifier or name.
l or example , the standard type int has an identifier ( int ) , a set of values
( -oo. . . +°° ) , and a set ol operations ( such as add and multiply ) . While the sys -
tem defines the names, values, and operations for standard types, we must
define them for types we create. l
Declaring an Enumerated Type

Io declare an enumerated type, we must declare its identifier and its values.
Because it is derived from the integer type , its operations are the same as lor
s integers. The syntax lor declaring an enumerated type is:
s
h enum typeName {identifier list};
The keyword , enum , is followed by an identifier and a set ol enumeration con -

stants enclosed in a set of braces. The statement is terminated with a semico-
lon . The enumeration identifiers arc also known as an enumeration list .
Each enumeration identifier is assigned an integer value. If we do not
explicitly assign the values, the compiler assigns the first identifier the value
0 , the second identifier the value 1 , the third identifier the value 2 , and so on
until all of the identifiers have a value. For example, consider an enumerated
type for colors as defined in the next statement. Note that for enumeration
identifiers, we use uppercase alphabetic characters.
enum color {RED, BLUE, GREEN , WHITE};
ec- The color type has four and only four possible values. The range ol the val -
ler ues is 0 . . 3, with the identifier red representing the value 0 , blue the
ta value 1 , green the value 2 , and white the value 3.
Once we have declared an enumerated type, we can create variables from
it just as we can create variables from the standard types. In fact , C allows
the enumerated constants, or variables that hold enumerated constants, to he
used anywhere that integers can he used . The following example defines
ea three variables for our color type.
/.
enum color productColor ;
enum color skyColor ;
enum color flagColor ;
eger 1 . When the defined type directly translates into a standard type as
, with the enumerated type,
died the standard types may be automatically defined .
8 Section 12.2 Enumerated Types
Operations on Enumerated Types

in enumerated types and manipulat-
In this section we discuss storing values
ing them .
Assigning Values to Enumerated Types

, we can store values in it .
After an enumerated variable lias been declared
, that an enumera ted variable can hold only declared val-
Remember, however
defines a color variable and uses it in
ues for the type. The following example
several statements.
enum color x ;
enum color y ;
enum color z;
x = BLUE ;
y = WHITE;
/ / Error. There is no purple.
z = PURPLE;
Similarly, once a variable has been defined and assigned a value, we can
store its value in another variable of the same type. Given
the previous exam-
ple , the following statements are valid .
x = y;
z y;
Comparing Enumerated Types

Enumerated types can be compared using the comparison operators.
For
example, given two color variables, we can compare them for equal , greater
.
than , or less than . We can also compare them to enumeration identiliers
I hese comparisons are shown in the following example.
if ( colorl == color2 )
if (colorl == BLUE )
Another natural use ol enumerated types is with the switch statement

,
Because enumerated types are derived from integer types, they may be used
the
in the case expressions. For example, consider an enumerated type lor
months of the year.
enum months
{JAN ,FEB,MAR ,APR ,MAY,JUN,JUL,AUG,SEP,OCT,NOV ,DEC}
;
enum months dateMonth ;

T
Chapter 12 Enumerated, Structure, and Union Types / 49
( » iven the months type , we

can use it in a switch statement as shown in
the next example.
switch (dateMonth )
{
case JAN: .
break ;
case FEB: .
break ;
> // switch
Other Operations on Enumerated Types

In general , any operation defined for integers can he used with enumerated
types. We can add , subtract , multiply, and divide enumerated types. We can
also pass them to standard and application functions.
Be aware, however, that the result may not be in the defined set , which
may cause the program to fail. C does not ensure that the values stored in an
enumerated variable are valid .
Enumeration Type Conversion

Enumerated types can he implicitly and explicitly cast. The compiler implic -
itly casts an enumerated type to an integer as required. However, when we
implicitly cast an integer to an enumerated type we get either a warning or an
error depending on the compiler. The following example demonstrates
implicit casts.
int x;
enum color y ;
x = BLUE; // Valid , x contains 1
y = 2; // Compiler warning
We can explicitly cast an integer to an enumerated type without problems.

To assign y the value blue in the previous example, we could use the
following code.
enum color y ;
y = ( enum color)2 ; // Valid , y contains blue
Initializing Enumerated Constants

While the compiler automatically assigns values to enumerated types starting
with 0 , we can override it and assign our own values. For example, to set up
an enumerated type for the months of the year, we could use the following
declaration .
750 Section 12.2 Enumeroted Types
enum months
{JAN ,FEB , MAR , APR ,MAY ,JUN ,JUL ,AUG ,SEP ,OCT,NOV,DEC};
enum months dateMonth ;
While this declaration works, it could be confusing because JAN is

igned the value 0. FEB the value 1 , and so forth until DEC , which is 11. To
assi
make JAN start with 1 , we could use the following declaration.
enum months {JAN = 1 , FEB , MAR , APR , MAY , JUN ,

JUL , AUG , SEP , OCT, NOV , DEC};
Note that we don ’t have to assign initializers to every value. If we omit the
initializers, the compiler assigns the next value by adding 1. To initialize the
months, therefore, we simply assign the value for JAN . The rest will be auto-
matically assigned by the compiler.
C also allows us to assign duplicate values to identifiers. For example, we
could assign similar colors identical values as shown in the next example.
enum mycolor {RED , ROSE = 0, CRIMSON = 0, SCARLET = 0,

BLUE , AQUA = 1 , NAVY = 1,
GREEN , JADE = 2, WHITE};
To emphasize the point , even though RED in Color ( see page 747) has
the same value as RED in myColor, the two types are different and cannot be
used together unless one of them is cast.
Anonymous Enumeration: Constants

II we create an enumerated type without a name, it is an anonymous enumer-
ated type . Because the identifiers in enumerated types are constants, enumer-
ated types are a convenient way to declare constants. For example , to assign
names to common punctuation characters, we would use the following code.
enum {space = ', -' ,', colon

comma = ':' ...>?
As another example, to declare the constants ON and OFF, we could use the
lollowing code. Anonymous enumeration cannot he used to declare a variable.
enum {OFF , ON};
I he identifier OFF is a constant with a value of 0; ON is a value ol 1 . As an

aside , we coded OFF first because we wanted it to have a connotation ol false.
Similarly, ON has a connotation of true.
II you have cable TV» you,
) know that the cable supplier is always changer
the channels. Fo help us keep track of our favorite channels, therefore, we
'
'
wrote a program that prints out the cable television channels for our local
cable provider. Since the channels are changing continually, using their
names in an enumerated type makes it easy to change them and to follow the
code. The implementation is seen in Program 12- 1 .
PROGRAM 12 - 1 Print Cable TV Stations

1 /* Print selected TV stations for our cable TV system.
2 Written by:
3 Date:
4 */
6
7 int main (void )
8 {
10 enum TV {fox = 2, nbc = 4, cbs = 5,
11 abc = 11 , hbo = 15, show = 17 ,
12 max = 31, espn = 39 , cnn = 51};
13
14 // Statements
15 printf( " Here are my favorite cable stations:\n " );
16 printf( " ABC: \t%2d\n" , abc);
17 printf(" CBS: \t% 2d\n " , cbs );
18 printf( " CNN: \t%2d \n ", cnn );
19 printf(" ESPN:\t%2d\n" , espn);
20 printf( " Fox: \t%2d \n " , fox );
21 printf( " HBO: \t%2d \n " , hbo);
22 printf( " Max: \t%2d\n" , max);
23 printf( M NBC: \t%2d \n " , nbc );
24 printf( " Show:\t%2d \nM , show );
25 printf( "End of my favorite stations. \n" );
26 return 0;
27 } // main
Results:
Here are my favorite cable stations:
ABC: 11
CBS: 5
CNN: 51
ESPN: 39
Fox: 2
HBO: 15
Max: 31
NBC: 4
Show: 17
End of my favorite stations.
752 Section 12.3 Structure
Input/Output Operations
Because enumerated types are derived types, they cannot be read and written
using the formatted input/output functions. When we read an enumerated
type , we must read it as an integer. When we write it , it displays as an integer.
When necessary, C implicitly casts variables as described in " Enumeration
Type Conversion" earlier in the chapter.
The following example shows how we would read and write two months.
enum months {JAN = 1 , FEB , MAR , APR , MAY , JUN ,

JUL , AUG , SEP , OCT, NOV , DEC};
enum months monthl ;
enum months month2 ;
scant ( "%d %d " , &monthl, &month2); // Input 1 - 12

printf( "%d %d " , monthl , month2 ); // Prints 1
- 12
.
Don’t be confused about strings and enumerated types "Jan" is a string
made of three characters; JAN as defined in the previous code example, is an
enumerated type ( identifier) which has the integer value 1 .
12.3 Structure
Today’s modern applications require complex data structures to support
them. For example, the structures needed to support the graphical user inter-
lace found in computer workstation windows obviously are very complex. So
complex, in fact, that they could not be built using the relatively primitive
data types we have seen so far. A much more powerful capability is needed.
I his is the role of the structure.
A structure is a collection of related elements, possibly of different types,
having a single name. For example, consider the file structure you have been
using ever since you wrote your first program that used files . The file table is
a type-defined structure, FILE.
Each element in a structure is called a field . A field is the smallest ele-
ment ol named data that has meaning. It has many of the characteristics of
the variables you have been using in your programs . It has a type , and it exists
in memory. It can he assigned values, which in turn can he accessed lor selec -
tion or manipulation . A field clillers Irom a variable primarily in that it is part
ol a structure.
We have studied another derived data type that can hold multiple pieces
ol data , the array. I he
difference between an array and a structure is that all
elements in an array must be of the same type, while the elements in a struc-
ture can be of the same or different types .
Figure 12 - 3 contains two examples of structures. T he first example,
fraction, has two fields, both of which are integers . The second example,
student , has three fields, made up of two different types.
numerator denominator
fraction
structure structure
id name gradePoints
student
FIGURE 12 - 3 Structure Examples
Another way to look at a structure is as a pattern or outline that can be

applied to data to extract individual parts. It allows us to refer to a collection
of data using a single name and , at the same time , to refer to the individual
components through their names. By collecting all the attributes of an object
in one structure , we simplify our programs and make them more readable.
One design caution , however. The data in a structure should all be
related to one object . In Figure 12-3, the integers in the fraction both belong
to the same fraction , and the data in the second example all relate to one stu -
dent . Do not combine unrelated data for programming expediency. That
would not be good structured programming.
Elements in a structure can be of the same or different types. However all

,
elements in the structure should be logically related .
Structure Type Declaration

Like all data types, structures must be declared and defined C has
. two ways
to declare a structure : tagged structures and type- defined structures.

Tagged Structure
structure . A tagged
The first way to define a structure is to use a tagged
, , and return types. The
structure can be used to define variables parameters
format is shown in Figure 12 - 4 .
struct TAG struct STUDENT

{ {
char id[ 10 ];
field list char name[ 26];
int gradePts ;
} ; > ; // STUDENT
Format Example
FIGURE 12 - 4 Tagged Structure Format
A tagged structure starts with the keyword struct. The second element in
the declaration is the tag. The tag is the identifier for the structure, and it
allows us to use it for other purposes, such as variables and parameters. If we
conclude the structure with a semicolon after the closing brace, no variables
are defined. In this case, the structure is simply a type template with no asso-
ciated storage.
Type Declaration with typedef

The more powerful way to declare a structure is to use a type definition,
typedef . The typedef format is shown in Figure 12 - S.
typedef struct typedef struct

{ {
char id[ 10 ];
field list char name[ 26 ];
int gradePts;
} TYPE; } STUDENT;
Format Example
FIGURE 1 2 - 5 Structure Declaration with typedef
The type-defined structure differs from the tagged declaration in two

ways. I irst, the keyword, typedef, is added to the beginning of the definition.
Second, an identifier is required at the end of the block; the identifier is the
type definition name.
Variable Declaration
After a structure has been declared, vve can declare variables using it. Gener-
ally, we declare the type in the global area of a program to make it visible to all
functions. The variables, on the other hand, are usually declared in the tunc *
lions, either in the header or in the local declarations section. Figure 12-6
demonstrates the declaration ot a structure type and variables that use it.
// Global Type Declarations // Global Type Declarations

struct STUDENT typedef struct
{ {
char id[ 10 ]; char id[ 10];
char name[ 26 ]; char name[ 26 ];
int gradePts ; int gradePts;
} ? > STUDENT;
struct STUDENT aStudent ; STUDENT aStudent ;
FIGURE 1 2 - 6 Structure Declaration Format and Example
Initialization
We can initialize a structure. The rules for structure initialization are similar
to the rules for array initialization. The initializers are enclosed in braces and
separated by commas. They must match their corresponding types in the
structure definition. Finally, when we use a nested structure ( see " Nested
Structures" later in this section), the nested initializers must be enclosed in
their own set ol braces.
Figure 12 - 7 shows two examples of structure initialization. The first
example shows an initializer for each field. Note how they are mapped to the
structure in sequence. The second example demonstrates what happens
when not all fields are initialized. As we saw with arrays, when one or more
initializers are missing, the structure elements will be assigned null values,
zero for integers and floating-point numbers, and null ( ' \ 0 ' ) for characters
and strings.
typedef struct
{
int x;
int y;
float t;
char u;
} SAMPLE;
SAMPLE sam1 = { 2, 5, 3.2, 'A' } ; SAMPLE sam2 = { 7 3 } ;. filled with

- nun 0 l
°'
2 5 3.2 im 7
t
0.0 \0 |
u
X y t u X y
filled with
float zero
FIGURE 12 - 7 Initializing Structures

Accessing Structures
Now that wc have discussed how to declare and initialize structures, it’s time
to see how we can use them in our programs. We will first describe how to
access individual components of a structure and then examine the assign-
ment of whole structures . After looking at how pointers are used with struc-
tures , we conclude with a program that uses structures.
Referencing Individual Fields

Each field in a structure can he accessed and manipulated using expressions
and operators . Anything we can do with an individual variable can be clone
with a structure field . The only problem is to identify the individual fields we
are interested in .
To refer to a field in a structure we need to refer to both the structure and
the field . C uses a hierarchical naming convention that first uses the struc-
ture variable- identifier and then the field identifier. The structure variable-
identifier is separated from the field identifier by a dot . The dot is the direct
selection operator, which is a postfix operator at precedence 16 in the prece-
dence table.
Using the structure student in “ Type Declaration with typcdef ," we
would refer to the individual components as shown below.
aStudent.id
aStudent.name
aStudent.gradePoints
Figure 12 -8 contains another example using the structure SAMPLE,

defined in Figure 12 - 7. With this structure , we can use a selection expression
to evaluate the sam2 field , u , and if it is an A , add the two integer fields and
store the result in the first . I bis code is shown below.
if (sam2.u = = 'A ' )

sam 2.x += sam2.y ;
FIGURE 1 2 - 8 Structure Direct Selection Operator
We can also read data into and write data from structure members just as
we can from individual variables. For exam pie , the value for the fields of tin
sample structure can be read from the keyboard and placed in saml using the
following scanf statement. Note that the address operator is at the beginning
of the variable structure identifier.
scanf( " %d %d %f %c " ,

&saml.x , &saml.y , &saml.t, &saml.u);
Precedence of Direct Selection Operator

As we saw in the previous section, the direct -selection operator ( .) is a postfix
expression that has a priority of 16. Let 's look at several examples to see how
it works.
hirst, let's look at examples in which all operators have a priority of 16. In
this case, the expressions are evaluated using left to right associativity as
shown in the following examples.
typedef struct
{
int total;
int ary[ 10];
} DATA ;
DATA sales;
sales.total++; // interpreted as (sales.total)++
fun( sales.total ); // sales.total passed to fun
fun( sales.ary[3 )); // ( sales.ary )[ 3 ] passed to fun
Now let 's examine cases in which the direct selection operator is used
.
with an operator of lesser precedence In these cases, the direct selection is
applied first.
++sales.total ; // interpreted as ++(sales.total )
fun(&sales.total ); // &( sales.total ) passed to fun
EXAMPLE 12- 1 Multiply Fractions

As an example of a function that uses structures and direct selection let's
,
look at a program that multiplies two fractions and prints the result . The
structure for this example was shown in Figure 12 - 3 . The code is shown in
Program 12 - 2.
PROGRAM 1 2 - 2 Multiply Fractions

1 /* This program uses structures to
simulate the
2 multiplication of fractions.
3 Written by:
4 Date:
5 */
continuei
r 58 Section 12.3 Structure
PROGRAM 12- 2 Multiply Fractions ( continued )

6 iinclude <stdio.h >
7
8 // Global Declarations
9 typedef struct
10 {
11 int numerator ;
12 int denominator ;
13 > FRACTION;
14
16 {
18 FRACTION frl ;
19 FRACTION fr2 ;
20 FRACTION res;
21
22 // Statements
23 printf( "Key first fraction in the form of x/y: ");
24 scanf ( " %d /%d " , & frl.numerator , &frl.denominator );
25 printf( " Key second fraction in the form of x/y: ");
26 scanf ("%d /%d " , &fr2.numerator , &fr2.denominator);
27
28 res.numerator = frl.numerator * fr2.numerator;
29 res.denominator = fr1.denominator * fr2.denominator;
30
31 printf( " \nThe result of %d / %d * %d /%d is %d /%d " ,
32 frl.numerator, fr 1.denominator,
33 fr2.numerator , fr2.denominator ,
34 res.numerator , res.denominator );
35 return 0 ;
36 } // main
Results:
Key first fraction in the form of x/y: 2/6
Key second fraction in the form of x/y: 7/4
The result of 2/6 * 7/4 is 14/24
Program 1 2 - 2 Analysis Although this iis a very simple program, several points are of interest.
.
1 Note that we have coded the typedef in the global area before main. This is
the first time that
— we have included statements here other than preprocessor
'
directives. Even though this program has no subfunction, it is customary to put

~
1
the typed definition statements there so that they are in scope for the entire
compilation unit.
2. Note that the name of the typed definition is all UPPERCASE. This is another
C style tradition. It warns the reader that there is something special about this
type, in this case that it is a typed definition.
3. Now examine the scanf statements. Since we need a pointer for scanf, we must
use the address operator as well as the direct selection operator. As you can see,
we have not used the parentheses around the field name. The reason is that the
direct selection operator ( . ) has a higher priority ( 16) than the address operator
( 1 5) . In other words, the expression
&frl.numerator
is interpreted by the compiler as
&(fr.numerator)
which is exactly what we need.
Operations on Structures
I he entity that can be treated as a whole. However, only one
structure is an
operation, assignment is allowed on the structure itself. That is, a structure
can only he copied to another structure of the same type using the assign-
ment operator.
Rather than assign individual members when we want to copy one
structure to another, as we did earlier, we can simply assign one to the other.
Figure 12- 9 copies saml to sam2
Before
| A
7 3 MU [El t
3 -2
u
X y t u
sam2 saml
sam2 saml ; j
2 5 32
~
|[ Al 2 5 3.2 im
y t u X y t u
X
sam2 saml
After
FIGURE 12- 9 Copying a Structure

'60 Section 12.3 Structure
Slack Bytes
It is interesting to examine why we cannot compare two structures. Sometimes hard-
ware requires that certain data, such as integers and floating- point numbers, be
aligned on a word boundary in memory. When we group data in a structure, the
arrangement of the data may require that slack bytes be inserted to maintain these
boundary requirements. For example, consider the structure shown below. In this .
structure we assume that a floating-point number is stored in a word that requires

6 bytes and must be on an address evenly divisible by 6, such as 24 or 30. We also
assume that integers are stored in 2- byte words that require an address evenly divisi-
ble by 2. The 25-byte string at the beginning of the structure forces 5 slack bytes
between the string and the float. Then the character after the float forces a slack byte
to align the integer at the end of the structure.
Vord Boundary Vord Boundary

Divisible by 6 . Divisible by 2
string ; slack 1 float ch jslack ; int

" ’
0 24 25 29 30 35 36 37 38 39
Since these extra bytes are beyond the control of the program, we cannot guarantee
what their values will be. Therefore, if we were to compare two structures and their
first components were equal, the inserted slack bytes could cause an erroneous com-
pare result, either high or low, if they weren' t equal. C prevents this problem by not
allowing selection statements with structures. Of course, if we really need to compare
structures, we can simply write a structure compare function of our own that would
compare the individual fields in the structure.
Pointer To Structures
Structures, like other types, can also be accessed through pointers. In fact,
this is one of the most common methods used to reference structures. For
example, lets use our SAMPLE structure with pointers (Figure 12 - 10).
( *ptr) . x j ( * ptr).u
typedef struct sam 1
{
int x;
V
int y; X t U
float t; ( * ptr). )
t
char u; ( *ptr ).y
} SAMPLE;
SAMPLE saml ;
SAMPLE* ptr ;
u
ptr
ptr = fitsaml ;
saml . x
Two Ways to Reference x
FIGURE 12 - 10 Pointers to Structures

I he first thing we need to do is define a pointer for the structure as

shown below.
SAMPLE* ptr ;
We now assign the address of saml to the pointer using the address oper-
would with any other pointer.
ator ( & ) as we
ptr = &saml ;
Now we can access the structure itself and all the members using the
pointer, ptr. I he structure itself can be accessed like any object using the
indirection operator ( * ) .
* ptr // Refers to whole structure
Since the pointer contains the address ol the beginning of the structure,
we no longer need to use the structure name with the direct selection opera-
tor. I he pointer takes its place. Phe reference to each of the sample mem-
hers is shown below and in Figure 12 - 10.
(* ptr).x (* ptr).y (* ptr).t (*ptr).u
.
Note the parentheses in the above expressions They are absolutely
necessary', and to omit them is a very common mistake. The reason they are
needed is that the precedence priority of the direct selection operator ( . )
is higher than the indirection operator ( * ) . If we forget to put the paren-
theses, C applies the dot operator first and the asterisk operator next . In
other words,
* ptr.x is interpreted as *( ptr.x )
which is wrong. The expression * ( p t r . x ) means that we have a completely

, x,
different (and undefined) structure called ptr that contains a member
which must be a pointer. Since this is not the case you , will get a compile
error. Figure 1 2 - 1 1 shows how this error is interpreted.
The correct notation, ( * p t r ) . x , first resolves the primary expression
( * p t r ) and then uses the pointer value to access the member
, x.
Indirect Selection Operator

Fortunately, another operator— indirect selection—eliminates
the problems
structures. The indirect selection operator is at the same level
w i t h pointers to
) as the direct selection operator.
in the Precedence Table ( see inside front cover
(*pointerName).fieldName -
pointerName >fie1dName
t u
The Correct Reference
The Wrong Way to Reference the Component
FIGURE 12 - 11 Interpretation of Invalid Pointer Use
The token for the indirect selection operator is an arrow formed by the minus
.
sign and the greater than symbol (->) It is placed immediately after the pointer
identifier and before the member to he referenced. We use this operator to refer
to the members of our previously defined structure, saml, in Figure 12 - 12.
-
typedef struct saml
int x;
int y; L y t u
float t; ptr
char u;
} SAMPLE;
SAMPLE saml; saml .X ptr -> x
SAMPLE* ptr ;
Direct Selection Indirection Indirect Selection
ptr = Srsaml ;
Three Ways to Reference the Field x
FIGURE I 2- 12 Indirect Selection Operator
Example: Clock Structure

Program 12 - 3 simulates a digital clock that shows time. A structure is defined
to represent the three components of time: hour, minute, and second. Two
functions are used. I he first function, called increment, simulates the passage
of the time. T he second function, called show, shows the time at any moment.
PROGRAM 12 - 3 Clock Simulation with Pointers

1 /* This program uses a structure to simulate the time.
2 Written by:
continue
PROGRAM 1 2 - 3 Clock Simulation with Pointers ( continued )

3 Date:
4 */
6
7 typedef struct
8 {
9 int hr;
10 int min ;
11 int sec;
12 > CLOCK ;
13
15 void increment ( CLOCK * clock );
16 void show ( CLOCK * clock );
17
19 {
21 CLOCK clock = {14 , 38 , 56};
22
23 // Statements
24 for(int i = 0; i < 6; ++i)
25 {
26 increment (&clock );
27 show ( &clock );
28 } // for
29 return 0;
30 } // main
31
32 /* ===== =====: increment =============
33 This function accepts a pointer to clock and
34 increments the time by one second ,
35 Pre previous clock setting
36 Post clock incremented by one second.
37 */
38 void increment ( CLOCK* clock )
39 {
40 // Statements
41
42
-
(clock >sec )++ ;
-
if ( clock >sec = = 60 )
43 {
44 -
clock >sec = 0;
45 -
(clock >min )++ ;
46 if (clock >min = = 60)
-
continued
PROGRAM 12 - 3 Clock Simulation with Pointers ( continued )

47 {
48 -
clock >min = 0;
49
50
-
( clock >hr)++ ;
-
if ( clock >hr = = 24 )
51 -
clock >hr = 0 ;
52 > // if 60 min
53 > // if 60 sec
54 return ;
55 > // increment
56
57 :
====
58 Show the current time in military form ,
59 Pre clock time
60 Post clock time displayed
61 */
62 void show ( CLOCK* clock)
63 {
64 // Statements
65 printf( "%02d:%02d:%02d\n " /
66 - -
clock >hr , clock >min , clock->sec );
67 return ;
68 } // show
Results:
14:38:57
14:38:58
14:38:59
14:39:00
14:39:01
14:39:02
Complex Structures
As mentioned earlier, structures were designed to deal with complex problems.
I he limitations on structures are not on the structures themselves but on the
imagination ol the soitware engineers who solve the problems. Structures
within structures (nested structures ) , arrays within structures, and arrays ol
structures are all common . We deal with the first two here and arrays of struc-
tures in the next section.
Nested Structures
We can have structures as members of a structure. When a structure includes
another structure, it is a nested structure . There is no limit to the number of
structures that can be nested , but we seldom go beyond three.
For example , we can have a structure called stamp that stores the date
and the time . The date is in turn a structure that stores the month , day, and
year. I he time is also a structure that stores the hour, minute, and second .
This structure design is shown in Figure 12 - 13.
stamp.date.month stamp.time.sec
stamp
month day year hour min sec
date time
stamp.date stamp.time
FIGURE 1 2 - 1 3 Nested Structure
There are two concerns with nested structures: declaring them and refer-
encing them .
Declaring Nested Structures

Although it is possible to declare a nested structure with one declaration , it is
not recommended. It is far simpler and much easier to follow the structure if
each structure is declared separately and then grouped in the high -level
structure.
When we declare the structures separately, the most important point we
must remember is that nesting must be done from inside out that is — , from
the lowest level to the most inclusive level . In other words , we must declare
the innermost structure first , then the next level , working upward toward the
outer, most inclusive structure.
Consider the time stamp structure seen in Figure 12 - 13. The inner two
structure,
structures, date and time, must be declared before the outside
stamp , is declared. We show the declaration of stamp and a variable that uses
it in the following example .
// Type Declaration
typedef struct
{
int month;
int day ;
int year ;
> DATE;
typedef struct
{
int hour ;
int min;
int sec ;
> TIME;
typedef struct
{
DATE date ;
TIME time;
> STAMP;
// Variable Declaration
STAMP stamp;
It is possible to nest the same structure type more than once in a dec-
laration . For example, consider a structure that contains start and end
.
times for a job Using the STAMP structure, we create a new declaration as
shown below.
// Type Declaration
typedef struct
{
STAMP startTime;
STAMP endTime;
> JOB;
// Variable Declaration
JOB job ;
^^ nL 7dWtr
!„felC olZ7ructuZ
^ ytZl
a
Jtif ’ T ,,
se
* flcxil lty > n working with them . For example, with DATE
.
)
declared St' l> tlrate tyPe definition , it is possible to pass the date structure
to a fun c ' w thout having to pass the rest of the stamp structure.
funetton
Referencing Nested Structures

lo access a nested structure, we include each level from the highest ( stamp
? ” “*
of r e f r e n c
'
stamp
stamp.date
stamp.date.month
stamp.date.day
stamp.date.year
stamp.time
stamp.time.hour
stamp.time.min
stamp.time.sec
job.startTime.time.hour
job.endTime.time.hour
Nested Structure Initialization

Initialization follows the rules mentioned for a simple structure. Each struc -
ture must he initialized completely before proceeding to the next member.
.
Each structure is enclosed in a set of braces For example, to initialize stamp,
first we initialize date , then time , separated by a comma. To initialize date,
we provide values lor month , day , and year , each separated by commas. We
can then initialize the members of time . A definition and initialization for
stamp are shown below.
STAMP stamp = {{05, 10, 1936}, {23, 45 , 00}};
Structures Containing Arrays

Structures can have one or more arrays as members. The arrays can be
accessed either through indexing or through pointers, as long as they are
properly qualified with the direct selection operator.
Defining Arrays for Structures

or
As with nested structures, an array may be included within the structure
. If declared separately , the
mav he declared separately and then included it is
declaration must be complete before it can be used in the structure
. For
the that contains the student name , three mid -
exam pie , consider structure
term scores , and the final exam score , as shown in
Figure 12 - 14 .
Referencing Arrays in Structures

will have the
Regardless of how we declared the structure, each element
. First we refer to the structure , then to the array component .
same reference
notation . Let
When we refer to the array, we can use either index pointer
or
us look at each in turn.
// Global Declarations
student.name[ student.midterm
^])
typedef struct
{
char name[ 26 ];
int midterm[3 ]; I II 111 -..
name midterm final
int final ;
} STUDENT ; student.final
student
STUDENT student ;
FIGURE 12- 14 Arrays in Structures
I hc index applies only to elements within an array. Therefore, it must fol-

low the identifier of an array. Our student example has two arrays, one of
characters ( a string) and the other of midterm scores. Each of these arrays
can be referenced with an index as shown below.
student
student.name
student.name[ i ]
student.midterm
student.midterm[ j]
student.final
We have already seen how to refer to fields in a structure using the indi-
rect .
selection operator (->) When one structure contains an array, we can
.
use a pointer to refer directly to the array elements For example, given a
pointer to integer, pScores, we could refer to the scores in student as
shown below.
pScores = student.midterm ;
totalScores = * pScores + *( pScores + 1 ) + *( pScores + 2);
Array Initialization in Structures

I he initialization ol a structure containing an array is simply an extension of
the rules lor structure initialization to include the initialization of the array
Since the array is a separate member, its values must he included in a sepa-
.
rate set of braces For example, the student structure can be initialized as
shown below.
STUDENT student = {"John Marcus" , {92, 80, 70 > , 87};
Note that the name is initialized as a string and the midterm scores art
simply enclosed in a set of braces.
Structure Containing Pointers

It should not be surprising that a structure can have pointers as members. In
fact, we will see that pointers are very common in structures.
The use ol pointers can save memory. For example, suppose that we
wanted to use the alphabetic month in our stamp structure rather than an
integer month. We could add an array of nine characters to each structure
element, but it would be much more memory efficient to use a 4 - byte
pointer it we had to store a lot of these structures. Given the months of the
year defined as strings, as shown below, we coidd then use a pointer to the
correct month.
char jan[] = "January " ;

char feb[ ] = "February " ;
char dec[ ] = "December " ;
lo assign the month May to the structure, we would use the following
statement. Note that we are copying a pointer to a string, not the string
itself .
stamp.date.month = may ;
The modified structure is shown in Figure 12 - 15 . The structure code for

the date structure is shown below. It is important to remember that declaring
a structure does not allocate memory for it . Even when the structure is
defined, however, there is still no memory provided for month until we allo-
cate it using malloc or calloc .
typedef struct
{
char* month;
int day ;
int year ;
} DATE;
rrrrrrrrrrrrrrrrrq
i i i i i i i i i i i i i i i
i t
• ' I
i i i i i i i i
stamp
day year hour min sec

month
date time
FIGURE 1 2- 1 5 Pointers in Structures

Section 12.3 Structure
Array of Structures
In many situations, we need to create an array of structures. To name just
one example, we would use an array of students to work w ith a group of stu -
dents stored in a structure. By putting the data in an array, we can quickly
and easily work with the data, to calculate averages , lor instance.
Let ’s create an array to handle the scores for up to 50 students in a class.
Figure 12 - 16 shows how such an array might look.
stuAry [ 0 ]
stuAry [ 1 ]
stuAry [ 2 ]
stuAry [ 49 ]
stuAry
FIGURE 1 2 - 1 6 Array of Structures
Since a structure is a type, we can create the array just as we would cre-
ate an array of integers. The code is shown below .
STUDENT stuAry[ 50];
Study this array carefully. Note that it is an array that contains two other
arrays , name and midterm. 1 his is not a multidimensional array. To be a mul-
tidimensional array, each level must have the same data type. In this case,
each type is different: the stuAry is STUDENT , while name is character, and
midterm is integer.
lo access the data lor one student , we refer only to the structure name
with an index or a pointer as shown below.
stuAry[i] *pStu
lor example, lets write a short segment of code to compute the average
loi the final exam . We use a for loop since wre know the number of students in
the array.
int totScore = 0;
float average ;
STUDENT* pStu ;
STUDENT* pLastStu ;
continued
Chapter 12 Enumerated, Structure, and Union Types 77 j
pLastStu = stuAry + 49;

for ( pStu = stuAry ; pStu <= pLastStu ; pStu++ )
average = totScore
- 50.0;
totScore += pStu >final;
/
However, to access an individual element in one of the students arrays,

such as the second midterm for the filth student , we need to use an index or
pointer for each field as shown below.
stuAry[4 ].midterm[ 1 ]
Io access students ’ midterms with pointers, we need one index or pointer

for the array. We also need a second index or pointer for the midterms . The
code to compute the average for each midterm is shown helow. We use a sep-
arate array, midTermAvrg, to store the average for the midterms. In this
example, we use indexes to access the midterms and pointers to access the
students.
float midTermAvrg[ 3];

int sum ;
STUDENT* pStu ;
STUDENT* pLastStu ;
pLastStu = stuAry + 49 ;
for ( int i = 0; i < 3; i++ )
{
sum = 0;
for ( pStu stuAry ; pStu <= pLastStu; pStu++)
=
-
sum += pStu >midterm[ i ];
midTermAvrg[i] = sum / 50.0;
> // for i
index oper-
The Precedence Table ( see inside front cover ) shows that the
and the indirect selection operator all have
ator, the direct selection operator,
y is from left to right . So we do
the same precedence and that the associativit
not need parentheses to total the scores for a midterm .
Insertion Sort Revisited

, let ’s sort an array of students. We
To demonstrate using structures in arrays
will use the student structure seen in Figure -
12 14. Whenever we sort a
define the field that controls the sort . This control field
structure, we need to
key . In our student structure , the key field is the name.
is usually called a
Program 12 - 4 show' s the code.
Ik
'i
PROGRAM 12 - 4 Sort Array of Student Structures

1 /* This program sorts an array of student structures
2 Written by:
3 Date:
4 */
6 #include <string.h>
8
9 #define NUM STU 5
10
11 // Global Type Declaration
12 typedef struct
13 {
14 char name[ 261 ;
15 int midterm[3];
16 int final;
17 > STUDENT ;
18
20 void insertionSort ( STUDENT list[ ], int last);
21
22 int main (void )
23 {
24
25
STUDENT _
stuAry[ NUM STU] =
26 {
27 {"Charles, George" , {85, 94, 79}, 93},
28 {"Adams, Karin", {75 , 91, 89}, 89},
29 {" Nguyen , Tuan " , {87, 88 , 89}, 90},
30 {"Oh , Bill" , {78, 96, 88}, 91},
31 {"Chavez, Maria", {83 , 79 , 93}, 91}
32 } ; // stuAry
33
34 // Statements
35 printf( "Unsorted data:\n" );
36
37
38
for (STUDENT* pStuPtr = stuAry ;
pStuPtr < stuAry + NUM STU ; _
pStuPtr++)
39
40
-
printf( " % 26s %4d %4d %4d %4d\n",
41
-
pStuPtr >name,
42
-
pStuPtr > midterm[0],
-
pStuPtr >midterm[ 1 ],
43
44
-
pStuPtr >midterm[ 2],
-
pStuPtr >final);
continue
PROGRAM 12- 4 Sort Array of Student Structures (continued )

45 printf( "\n " );
46
47 insertionSort (stuAry , NUM STU _1 ); -
48
49 printf( "Sorted data:\n " );
50 for (STUDENT* pStuPtr = stuAry ;
51 pStuPtr < stuAry + NUM STU ; _
52 pStuPtr++)
53 -
printf("% 26s %4d %4d %4d %4d\n" ,
54 -
pStuPtr >name,
55 -
56 -
pStuPtr >midterm[ 1 ],
57 -
58 -
pStuPtr >final );
59 return 0;
60 > // main
61
62 /* ================== insertionSort ==================
63 Sort list using Insertion Sort. The list is divided
64 into sorted and unsorted lists. With each pass, the
65 first element in unsorted list is inserted into
66 sorted list.
68 last is index to last element in list
69 Post list has been rearranged.
70 */
71 void insertionSort ( STUDENT list[ ], int last)
72 {
74 bool located;
75 STUDENT temp;
76 STUDENT* pCurrent ;
77 STUDENT* pWalker;
78 STUDENT* pLast ;
79
80 // Statements
81 for ( pCurrent = list + 1 , pLast = list + last;
82 pCurrent < = pLast;
83 pCurrent ++)
84 {
85 located = false ;
86 temp = * pCurrent ;
87
88 for ( pWalker = pCurrent 1; -
continuec
PROGRAM 12- 4 Sort Array of Student Structures (continued )

89 pWalker >= list && !located ;
90 )
91 if (strcmp(temp.name , pWalker >name ) < 0) -
92 {
93 *( pWalker + 1 ) = * pWalker;
94
95
pWalker ;
> // if
—
96 else
97 located = true;
98 *( pWalker + 1 ) = temp;
99 > // for pCurrent
100 return;
101 > // insertionSort
102 / / ==== ===== ===== End of Program =====
Results:
Unsorted data:
Charles , George 85 94 79 93
Adams, Karin 75 91 89 89
Nguyen , Tuan 87 88 89 90
Oh , Bill 78 96 88 91
Chavez , Maria 83 79 93 91
Sorted data:
Adams, Karin 75 91 89 89
Charles, George 85 94 79 93
Chavez , Maria 83 79 93 91
Nguyen , Tuan 87 88 89 90
Oh , Bill 78 96 88 91
Program 1 2 4 Analysis This program follows the insertion sort design we studied in Chapter 8 (Program 8- 11).
-
There are only two significant changes: First, all the array handling is done with
pointers, and second, the compare logic was changed to use the strcmp function.
To test the sort, we wrote a small driver that uses an initialized array for its data. To
verify that the sort works, we printed the data before and after the sort process.
Structures and Functions

For structures to be fully useful, we must be able to pass them to function
and return them. A function can access the members ol a structure m
three ways:
1. Individual members can be passed to the function.
2. The whole structure can be passed and the function can access the mem-
bers using pass by value.
.
3 I he address of a structure or member can be passed, and the function
can access the members through indirection and indirect selection
operators.
Sending Individual Members

Sending individual members to a function is no different from what we have
done since Chapter 4. lo demonstrate the calling sequence, we will use a
simple example to multiply two fractions. The flow is shown in Figure 12 - 17.
res.numerator =
k fr 1
multiply( frl.numerator , fr2.numerator ) fr2
res.denominator =
multiply( frl .denominator , fr2.denominator ); numerator denominator
res
// ============= = =
multiply =====
int multiply ( int x, int y )
{
return x * y ; X y
> // multiply
FIGURE 1 2 - 1 7 Passing Structure Members to Functions
As you can sec from Figure 12 - 17, the only difference is that we must
for the
use the direct selection operator to refer to the individual members
actual parameters The .
called program doesn ’t know if the two integers are
simple variables or structure members.
Sending The Whole Structure

logic is split
The problem with the above solution is that the multiplication
calling and called programs . This is not considered good struc -
between the
. A much better solution is to pass the entire structure and
tured programming
let multiply complete its job in one call.
individual ele-
Passing a structure is really no different from passing
, simply specify the in the formal
ments. Since the structure is a type we type
a struc -
parameters of the called function.
Similarly, the function can return
is to specify the structure as the return type
ture. Again, all that is necessary
in the called function .
The same pass-by-value rules apply, however;
when we pass a structure to
to the local just as it does for vari-
a function, C will copy the values
structure
when large structures are used.
ables. This may lead to some inefficiencies
We address this problem in the next section.
7 / 6 Section 12.3 Structure
Let us rework the multiply fractions program to pass structures In this case, .
we have written the complete program. The design is shown in Figure 12- 18,
and the code is shown in Program 12 - 5.
main
fr 1 res
i
numerator denominator numerator denominator
i
A
fr 2
numerator denominator —i
i
I ( returned)
r
i
multFr (copied) i
i
i
frl I
I res
fr2
FIGURE 1 2 - 1 8 Passing and returning structures
PROGRAM 1 2 - 5 Passing and Returning Structures

1 /* This program uses structures to multiply fractions.
2 Written by:
3 Date:
4 */
6
8 typedef struct
9 {
10 int numerator;
11 int denominator ;
12 > FRACTION ;
13
15 FRACTION getFr ( void );
16 FRACTION multFr ( FRACTION frl, FRACTION fr2);
17 void printFr (FRACTION frl , FRACTION fr2 ,
18 FRACTION result );
19
21 {
continued
PROGRAM 1 2 - 5 Passing and Returning Structures (continued )

23 FRACTION frl;
24 FRACTION fr2;
25 FRACTION res;
26
27 // Statements
28 frl =
getFr ();
29 fr2 = getFr ();
30 res = multFr ( frl , fr2);
31 printFr ( frl , fr2 , res );
32 return 0;
33 > // main
34
35 / * ===== :
36 Get two integers from the keyboard , make & return

37 a fraction to the main program ,
38 Pre nothing
39 Post returns a fraction
40 */
41 FRACTION getFr ( void )
42 {
44 FRACTION fr;
45
46 // Statements
47 printf( " Write a fraction in the form of x/y: ");
48 scanf ( " %d/%d" , &fr.numerator, &fr.denominator );
49 return fr ; I
50 } // getFraction
51
52 /* ==
Multiply two fractions and return the result
,
53
54 Pre frl and fr2 are fractions
55 Post returns the product
56 */
57 FRACTION multFr ( FRACTION frl , FRACTION fr2 )
58 {
60 FRACTION res;
61
62 // Statements
res.numerator = frl.numerator * fr2.numerator;
63
res.denominator = frl.denominator * fr2.denominator ;
64
65 return res ;
continued
n
PROGRAM 12 - 5 Passing and Returning Structures ( continued )

66 I } // multFr
67
68 ===== printFr ===
69 Prints the value of the fields in three fractions.
70 Pre two original fractions and the product
71 Post fractions printed
72 */
73 void printFr ( FRACTION frl , FRACTION fr2,
74 FRACTION res)
75 {
76 // Statements
77 printf( " \nThe result of %d /%d * %d /%d is %d /%d\n",
78 frl.numerator , frl.denominator ,
79 fr2.numerator , fr2.denominator ,
80 res.numerator , res.denominator);
81 return;
82 } // printFractions
83 / / ===== ===== End of Program =====
Results:
Write a fraction in the form of x/y: 4 /3
Write a fraction in the form of x/y: 6/7
The result of 4/3 * 6 /7 is 24/21
Program 1 2 - 5 Analysis There are five points you should study in this program.
1. The fraction structure is declared in global memory before main. This makes it
visible to all the functions.
2. In getFr, the data are read by using the address operator and the direct selec -
tion operator. Since the direct selection operator (. ) has a higher priority, they
can be coded without parentheses.
.
3 In getFr we pass back two values in the structure without using pointers. This is
another advantage of structures: We can return more than one piece of data
when we put the data in a structure.
.
4 The structure type is passed using the FRACTION type definition.
5. The references to the data in multFr and printFr must use the direct selec -
tion operator to get to the individual members.
Passing Structures Through Pointers

As mentioned, passing structures is still pass by value. For the multiply
lIons Pr gr« , this is the correct way to write the program. It provides
. ° jm
ie necessary data protection
and data encapsulation for good structured
f
programming while at the same time being efficient ( see “Software Engi -
neering" at the end of the chapter) .
W hen the structure is very large, however, efficiency can suffer, espe-
.
cially with a heavily used function I herefore, we often use pointers to pass
.
structures It is also common to pass structures through pointers when the
structure is in dynamic memory. In these cases, all we have to pass is the
pointer.
We modify the multiply fractions program once more to pass the struc -
tures using pointers. The memory flow is shown in Figure 12 - 19 and the cod-
ing in Program 12 -6.
main multFr ( & fr 1, & fr2, & res )
fr1 res
fr2
void multFr ( FRACTION * pFrl ,

FRACTION * pFr2,
FRACTION *pRes) pFr2 pFr 1 pRes
{
-
pRes >numerator = pFrl->numerator * pFr2->numerator;
-
pRes >denominator = pFrl >denominator * pFr2 >denominator ;
return ;
- -
} // multFr
FIGURE 12 - 19 Passing Structures Through Pointers
PROGRAM 12- 6 Passing Structures through Pointers

1 /* This program uses structures to multiply fractions.
2 Written by:
3 Date:
4 */
6
8 typedef struct
9 {
10 int numerator;
11 int denominator;
12 } FRACTION ;
13
continuec
PROGRAM 12 - 6 Passing Structures through Pointers (continued )

15 void getFr ( FRACTION* pFr );
16 void multFr (FRACTION* pFrl, FRACTION * pFr2,
17 FRACTION * pRes2 );
18 void printFr ( FRACTION* pFrl , FRACTION* pFr2,
19 FRACTION* pRes);
20
22 {
24 FRACTION frl ;
25 FRACTION fr 2 ;
26 FRACTION res;
27
28 // Statements
29 getFr (&frl );
30 getFr (&fr2 );
31 multFr ( & fr 1 , &fr 2 , &res);
32 printFr (&frl, & fr2, &res);
33 return 0 ;
34 I > // main
35
36 / *
37 Get two integers from the keyboard , make & return a
38 fraction to the main program.
39 Pre pFr is pointer to fraction structure
40 Post fraction stored at pFr.
41 */
42 void getFr (FRACTION * pFr )
43 {
44 // Statements
45 printf("Write a fraction in the form of x /y: ' )?
46 scanf ( "%d/%d" , &pFr >numerator ,
47
-
&(*pFr).denominator);
48 return ;
49 > // getFr
50
51
52
53
54
55 */
Pre r
^^
= == = = == = =
= = multFr = = *== = = = =
=
iply two fractions and return the result.
pRes are pointers to fractions
Post product stored at pRes
—
56 void multFr ( FRACTION* pFrl FRACTION pFr2,
, *
continued
_
PROGRAM 1 2 - 6 Passing Structures through Pointers (continued )

57 FRACTION * pRes )
58 {
59 // Statements
60 -
pRes >numerator
61 - -
pFrl > numerator * pFr 2 >numerator ;
62 -
pRes >denominator =
-
63 -
pFrl >denominator * pFr2 >denominator ;
64 return;
65 } // multFr
66
67 /* =====
68 Prints the value of the fields in three fractions.
69 Pre pointers to two fractions and their product
70 Post fractions printed
71 */
72 void printFr ( FRACTION * pFrl , FRACTION * pFr2 ,
73 FRACTION * pRes )
74 {
75 // Statements
76 printf( "\ nThe result of %d / %d * %d /%d is %d /%d \n" ,
77 pFr 1 Enumerator, pFrl->denominator,
78 pFr2 enumerator , pFr2 >denominator , -
79 pRes >numerator , pRes >denominator);
- -
80 return;
81 } // printFr
82 ===== End of Program =====
. Note
Program 1 2 -6 Analysis In this version of the program, the structure is passed and returned asWe reproduce it
a pointer
the syntactical notation for reading the data in the getFr function.
below for your convenience.
.denominator );
scanf( " %d/%d " , & pFr->numerator , &(*pFr )
Even though we are using pointers, we still

need to pass scanf addresses. We
in this example . ( This is not good coding style, but
have used two different notations
both techniqu es in one statement . ) In both cases, since we are read-
it demonstrates
, we need to pass the address of the individual field,
ing a field within the structure
not the structure . In the first example , we use the indirect selection operator ( >).-
Since it has a higher precedence than the address operator ( & ), it can be coded with-
out parentheses.
second example, we use the direct selection operator . .precedence ( 16)
( ) In this case we
In the
the direct selection operator has a higher
need parentheses because
operator, both 15. However, since the
than the indirection operator ( * ) and the address level, we need to use the paren-
selection operator are at the same
address and direct
theses only around the pointer dereference .
782 Section 12.4 Unions
12.4 Unions
The union is a construct that allows memory to he shared by different types
of data. This redefinition can be as simple as redeclaring an integer as four
characters or as complex as redeclaring an entire structure. For example, we know
that a short integer is 2 bytes and that each hvte is a character. Therefore, we could
process a short integer as a number or as two characters.
The union follows the same format syntax as the structure In fact, with .
the exception of the keywords struct and union , the formats are the same.
Figure 12 - 20 shows how we declare a union that can he used as either a
short integer or as two characters.
»
i chAry [ 0] chAry[1]
union shareData
i
'A'
i
I B‘ '
i
I
i
{
char chAry[ 2 ];
short num; l
16706 l
i i
}; i num i
Both num and chAry start at the same memory address. chAry[0]
occupies the same memory as the most significant byte of num.
FIGURE 12 - 20 Unions
Referencing Unions
I he rules for referencing a union are identical to those for structures. To ref *
erence individual fields within the union, we use the direct selection ( dot
operator. Each reference must he fully qualified from the beginning of the
'
structure to the element being referenced. This includes the name of the
union itself. When a union is being referenced through a pointer, the selec *
lion operator ( arrow ) can be used. The following are both valid references to
the union shown in Figure 12- 20.
shareData.num
shareData.chAry[ 0 ]
Initializers
\ \ bile C permits a union to be initialized, only
the first type declared in UH
union can be initialized when the variable is defined. The other types can
only be initialized by assigning values or reading values into the union. When
initializing a union, we must enclose the values iin a set of braces, even if
there is only one value.
Program 12- 7 demonstrates unions in a program. It uses the structure in
Figure 12 - 20 to print a variable, first as a number and then as two characters.
PROGRAM 1 2 - 7 Demonstrate Effect of Union

1 /* Demonstrate union of short int and two char.
2 Written by:
3 Date:
4 */
6
8 typedef union
9 {
10 short num ;
11 char chAry[ 2];
12 > SH CH 2 ;
13
15 {
17 SH CH 2 data;
18
19 // Statements
20 data.num = 16706 ;
21
22 printf( "Short: % hd\n" , data.num );
23 printf("Ch[0 ]: %c\n" / data.chAry[0]);
24 printf( "Ch(l]: %c\n " , data.chAry[ 1 ]);
25
26 return 0;
27 } // main
Results:
Short: 16706
Ch[0]: A
Ch[1]: B
rs Aand B, will need to ana -

Program 1 2 - 7 Analysis To prove to yourself that 16706 creates the characte
you
|yze its bit pattern. Don ' t be surprise d, however, if you get a different number on your
If do, it means that you have a "little - endian" computer ( see the box
computer . you
).
"Big and Little Endian," later in this section
Unions and Structures

at the same memory location and occu-
In a union, each piece of data starts
memory. Thus, when a union is shared by two
pies at least a part of the same
or more different data types,
only one piece of data can he in memory at
one time.
that contains both company and
Consider, for example, an address hook name, the name has only one
individual names. When we store a company
field . On the other hand , when we store an individuals name, the name has
at least —
three parts first name, middle initial , and last name. If we want to
have only one name field in our address hook, we need to use a union to store
the name. This design is shown in Figure 12 - 21 . Note that in the figure, unis
the member name to access the union structure.
typedef struct
{ NAME
char first[ 20]; type un
char init;
char last[ 30 ]; company
} PERSON ;
typedef struct
{
char type ; person
union first init last
{
char company[40];
PERSON person ;
} un ;
> NAME ;
FIGURE 12- 21 A Name Union
Because two different types of data can he stored in the same union, we
need to know which type currently is stored . This is done with a type flag that
is created when the data are stored. If the name is in the company format , the
flag is C, and if it is in the person format it is a P.
\ \ hen a union is defined , C reserves enough room to store the largest
data object in the construct . In Figure 12- 21 , the size of the company name
is 40 bytes and the size of the person name is 5 1 bytes. The size of the union
structure is therefore 52 bytes ( the size of the person name plus the type !
regardless of what type of data are being stored. When a company name is
stored , the last 1 1 bytes are present hut unused .
Program 12-8 demonstrates the use of a union in a structure by printing
each format .
PROGRAM 1 2 - 8 Demonstrate Unions in Structures

1 /* Demonstrate use of unions in structures.
2 Written by:
3 Date:
4 */
6 ffinclude <string.h>
7
8 / / Global Structures
continue
PROGRAM 1 2-8 Demonstrate Unions in Structures (continued )

9 typedef struct
10 {
11 char first[ 20 ];
12 char init;
13 char last[ 30 ];
14 > PERSON;
15
16 typedef struct
17 {
18
19
char type;
union
// C
— company: P — person
20 {
21 char company!40 ];
22 PERSON person ;
23 > un;
24 > NAME;
25
26 int main (void)
27 {
29 NAME business = {'C', " ABC Company "};
30 NAME friend ;
31 NAME names[ 2 ];
32
33 // Statements
34 friend.type = 'P';
strcpy ( friend.un.person.first, " Martha );
"
35
( friend .un.person.last , "Washington " );
36 strcpy
37 friend.un.person.init = 'C';
38
39 names[ 0] = business ;
40 names[1] = friend ;
41
42 for (int i= 0; i < 2 ; i++)
43 switch (names[ i ].type )
44 {
45 case C': printf( "Company: %s\n" ,
names( i ].un.company );
46
break ;
47
48 case 'P': printf("Friend: %s %c %s\n" ,
names[ i ].un.person.first,
49
50 namesli ].un.person.init,
names Ii].un.person.last );
51
continued
PROGRAM 12 - 8 Demonstrate Unions in Structures ( continued )

52 break ;
53 default : printf ( " Error in type \ a \ n " );
54 break;
55 } / / switch
56 return 0 ;
57 > / / main
Results:
Company: ABC Company
Friend: Martha C Washington
Program 1 2 - 8 Analysis We begin by creating two union structures, one for a business and one for a friend.
To initialize the business union, we used initializers ( see below ). To initialize the friend
union, we used string copy and assignment. Once the unions were created, we cop-
ied them to an array to demonstrate one common processing concept.
Unions are often processed with a switch statement that uses the type to determine
what data format is currently being processed. Note that it is the programmer's respon-
sibility to process the data correctly. If a business identifier is used while person data
are present, the program will run but garbage will be produced.
Once the correct union type has been determined, the rest of the processing is rou-
tine . We simply print the data in the union using the appropriate identifiers for its con-
tent. Note that we provide a default statement even though there is no possibility of an
error in this simple program.
Big and Little Endian

Computer hardware can be physically designed to store the most significant bvteofa
number at either the left end or the number (big endian) or at the right end or the
number (little endian), as shown below.
short 16706
0100 0001 0100 0010
data[0] data[0]
( A)1
0100 0001 0100 0010 (B )

1
0100 0010 MS byte I LS byte 0100 0001

data( 1 ] data(1]
( B) (•A )
1
Big Endian Little Endian
es then a computer that uses big endian

the above exlT_"
stores the most siqnificant
lt k l hat°-
ft
'
array ^^
Portio
Tb,
the left. Note that data rt nu 1 A and data [ 1 ] is- 'B \
dta I is
" of
s sbown in the
IS
<16'000onin
example
St ,gn,bcant
-
e is stored on the right. This exam
pie is seen on theTiqhMn the L °
-
6
data [ 1 contains A If ft °bovf **amPlef

S
^ timempUter
This |
, data [ 0 ] contains ' '
B and
' °^
]
checkV iWslltT^ ° is b 9 r
^
C
then run Program 12-7 and t
Internet Addresses
As a final example of union processing, let ’s consider how Internet addresses
are lormatted and processed. We are most familiar with the Internet domain
names such as www.deanza.edu. These names are easy to remember and
use. I he real Internet Protocol (IP) address, however, is defined as dotted-
decimal notation, as shown below.
153.18.8.105
I he decimal-dot notation is used primarily by system administrators. To

conserve transmission time and facilitate processing, however, the IP address
must be converted to a 4 - bvte integer for transmission down the internet .
Each portion of the IP address has a maximum value of 255 . This means
that it can be stored in 1 byte. To store the address in an integer, we simply
convert each portion of the address to 1 byte and store it in the appropriate
portion of the integer. Thus, in our example, we convert 153 to its corre-
sponding (extended ) ASCII character and store it in the first byte of the inte-
ger. Then we convert the 18 and store it in the second byte, 8 into the third
byte, and finally 105 into the fourth byte. To process the integer address as
both an integer and an array of four characters requires a union.
Given that the IP address is a string, we use the string token operator
and a loop for the conversion The code is shown in Program 12 -9.
.
PROGRAM 1 2 - 9 Convert IP Address to long
1 / * Reformat IP string address to long integer.
2 Written by:
3 Date:
4 */
8
9 // Global Structures
10 typedef union
11 {
12 unsigned char chAddr[4];
13 unsigned long numAddr ;
14 _
> IP ADDR ;
15
17 {
19 IP ADDR addr;
20 char* parser ;
continue*
Section 12.4 Unions
PROGRAM 1 2- 9 Convert IP Address to long (continued )

21 char strAddr[16] = "153.18.8.105";
22
23 // Statements
24 // Convert first address segment
25 parser = strtok ( strAddr ,
26 addr.chAddr[3] = strtol ( parser , ( char* *)NULL , 10 ) ;
27
28
29
for ( int i
{
= 2; i>=0; i
— )
30 // Convert decimal dot positions 2, 3, 4

31 parser = strtok (NULL,
32 addr.chAddr[i] =
33 strtol ( parser , (char** )NULL, 10);
34 > // for
35
36 // Now print results
37 printf ( " IP decimal dot : %d.%d.%d.%d\n",
38 addr.chAddr[3], addr.chAddr[2],
39 addr.chAddr[ 1 ], addr.chAddr[ 0]);
40 printf ( " IP binary : %lu\n" , addr.numAddr);
41 return 0 ;
42 } // main
Results:
IP decimal dot : 153.18.8.105
IP binary address: 2568095849
jgram 1 2 - 9 Analysis Before you study thiIS

in Chanter 11 l
Pr 9ra
° . you may want to review the parsing concepts discussed
JsnTr
0
up
„• ' "‘
( see statements 25 and 3 l ) *" ^ ^ ^ °the ne that P r 5es the rest of the sfrm9
°
Within the loop, we parse the rest of the string, convert it to a decimal value, and
store the address elements in
address. After the I^ ? second, third, and fourth positions of the binary
oop, we print the converted address, first in its character formatand
then as a long integer.
This
s t a t e m e n t Wô r ^. ^ and
little endian computers. For
wi - big endian
computers
the for in statement 28 would start
at 1 and move up to 3 Th°
reversed.
^ °
exes n s a ements 38 and 39 would also need to be
6 10
' ^^
12.5 Programming Application

In this section , we develop a program that simulates the operation of an ele -
vator. The elevator serves floors from zero ( the basement ) to the top floor .
I he elevator is very old and is not fully automatic. When people enter the ele -
vator, they enter their desired floor number. Several numbers can be
requested at a time. After all numbers have been entered , the passengers
close the door by pressing the close door button ( the return key ).
Each time the door closes, the elevator checks to see if any floors in the
current direction ( up or down ) need to be serviced . If they do, then it services
these floors first , starting with the closest one to the current floor. If no floors
in the current direction need service, it checks the opposite direction , again
servicing the one closest to the current floor.
Each time the elevator arrives at a floor, new passengers can get on and
request their floor. The new requests are added to the ones still pending, and
the elevator again evaluates which floor will be processed next .
The structure for this program is portrayed in Figure 12- 22. The elevator
is represented as a structure with two fields: the current floor and a pointer to
an array ol buttons. The button values are IN. meaning the floor has been
requested , and OUT , meaning the floor has not been requested . After a floor
has been serviced , the button is reset .
loor buttons
(IN / O U T ) ,
currentFloor
buttons 0 1 2 TOP _ FLOOR
FIGURE 1 2 - 22 Elevator Structure
The elevator design is shown two different ways . Figure 12- 23 is a struc -
ture chart for the program .
Figure 12 - 24 is a state diagram for the elevator. A state diagram
is a
design technique that is often used with realtime systems to show how a sys -
state to another . For an elevator , it can be in one of three
tern moves from one
states: moving up, moving down , or stopped .
Each of these states is repre -
sented by a circle in the diagram. To move from one state to another, a
change must occur in the elevator environment . For example, to change from
the stop state to the up state , a button must be pressed . This is reflected on
the line between stop and up as anyUp.
790 Section 12.5 Programming Application
elevator
run
initialize Elevator terminate
move
I
anyllp anyDown
request request
moveUp move
Down
timePass j timePass j
FIGURE 1 2 - 23 Elevator Structure Chart
anyllp lanyUp & & anyDown anyDown
up down
lanyDown & & anyllp
lanyUp && lanyDown

lanyDown & & lanyUp
anyDown
anyUp && lanyUp
stop
lanyUp & & lanyDown
FIGURE 1 2 - 24 Elevator State Diagram
is seen in Program 12- 10.

I he main function for the elevator program
It calls three other functions : initialize (
Program 12 - 11 ), run elevator
-
(Program 12 12 ) , and terminate (
Program 12 - 17 ) . Of the three , run eleva -
tor is the function ol primary
interest ; it simulates the actual running 0
the elevator.
PROGRAM 12- 10 Elevator : mam

1 /* This program simulates the operation of an elevator.
2 Written by:
3 Date:
4 */
9
_
10 #define TOP FLOOR 10
11 #define DELAY FACTOR 10000
12
13 // Global Type Declarations
_
14 typedef enum {OUT, IN} BUTTON STATUS ;
15 typedef enum {DOWN , STOP , UP} DIRECTION STATUS ; _
16 typedef struct
17 {
18 int currentFloor ;
19 BUTTON STATUS* buttons;
20 } ELEVATOR ;
21
23 void initialize ( ELEVATOR * elev );
24 void runElevator ( ELEVATOR * elev );
25 void terminate ( ELEVATOR * elev );
26 void move ( ELEVATOR * elev );
27 bool anyUpRequest ( ELEVATOR * elev );
28 bool anyDownRequest ( ELEVATOR * elev );
29 void moveUp ( ELEVATOR * elev );
30 void moveDown ( ELEVATOR * elev );
31 void timePass ( int m );
32
34 {
36 ELEVATOR elevator;
37
38 // Statements
39 initialize ( elevator );
^
40 runElevator (&elevator );
41 terminate (&elevator);
continued
PROGRAM 12 - 10 Elevator : main ( continued )

42 return 0 ;
43 > / / main
Program 12- 11 initializes the state of the elevator.
PROGRAM 12 - 11 Elevator: Initialize

1 /* = initialize ==
2 This function dynamically allocates memory locations
3 for buttons & initializes current floor to 1 to show
4 that the elevator is parked in the first floor ,
5 Pre nothing
6 Post elevator created , all buttons reset , and
7 parked at first floor ( not basement)
8 */
9 void initialize ( ELEVATOR * elev )
10 {
11 // Statements
12 -
elev >buttons = _
calloc(TOP FLOOR + 1,
13 sizeof ( BUTTON STATUS ));
14
15 _
for ( int i = 0; i <= TOP FLOOR ; i++ )
16 elev->buttons [ i] = OUT ;
17 -
elev >currentFloor = 1 ;
18
19 return;
20 } / / initialize
Program 12- 12 simulates the operation of the elevator.
PROGRAM 12 - 12 Elevator: Run Elevator

l /* ===== ==== runElevator ========= ==== =====
2 Simulate the operation of the elevator.
3 Pre elevator structure has been initialized
4 Post simulation is complete
5 */
6 void runElevator ( ELEVATOR
* elev )
7 {
9 char buffer (81];
10 int floor ;
11 char* pStrln ;
12
PROGRAM 12 - 12 Elevator: Run Elevator (continued )

13 // Statements
14 printf( "\n\nThis elevator goes from basement ( 0) ");
15 _
printf( "to floor %d " , TOP FLOOR );
16 printf( " \n\nType floors & press return to start" );
17 printf( " \nlf no new floors , press return key." );
18 printf( "\nTo quit, key EOF " );
19 printf( " \n\nPlease enter floors: " );
20
21 while ( fgets( buffer , 81 , stdin))
22 {
23 pStrln = buffer ;
24 while ( * pStr
! n != \n')
25 {
26 // Locate next floor digit & convert to int
27 while ( * pStrIn == )
28 pStrIn++;
29 if ( lisdigit ( * pStr
! n ))
30 {
31 printf("\alnvalid floor %c\n" , * pStrIn );
32 pStrIn++;
33 > // if
34 else
35 {
36 sscanf ( pStrln , "%d " , & floor);
37 -
if ( floor == elev >currentFloor )
38 printf( " \n\aAlready on floor %d." ,
39 elev->currentFloor);
40 else
41 if (floor < 0 || floor > TOP FLOOR ) _
42 printf( "\n\a%d invalid floor " ,
43 floor );
44 else
45 -
elev >buttons [ floor ] = IN;
46
47 // Synchronize sscanf & * pStrIn
48 while ( isdigit ( * pStrIn ))
49 pStrIn++ ;
50 > // else
51 } // while
52
53 move ( elev );
54 printf( "\n\nPlease enter floors: " > ?
55 } // while
56 return ;
57 > // runElevator
Program 1 2-12 Analysis The simulation is controlled by runElevator. The technique we use to prevent
invalid user input is interesting . After reading the input stream , we use a pointer to
parse the string after skipping whitespace . If the current character is a digit, we use
sscanf to format the digit(s) as an integer. If it is not a digit we print an error mes-
sage and skip it. (Of course, in a real elevator, it is impossible to get an invalid floor.)
Then , after converting the digit string to an integer, we synchronize the buffer and our
pointer with a small while loop .
As we parse the integers in the input stream , we " push" their corresponding but-
tons in our data structure. This code is shown at statement 45 in Program 12-12.
Program 12- 13 moves the elevator to a requested floor.
PROGRAM 1 2- 1 3 Elevator: Move

l /* =================== move ===================
2 Moves the elevator to a requested floor. It stops
3 the elevator after responding to one request ,
4 Pre given elevator
5
6 —
Post elevator has been moved while it is
moving , the floors are called out
7 */
8 void move ( ELEVATOR* elev )
9 {
11 static DIRECTION STATUS direction = STOP ;
12
13 bool anyUp;
14 bool anyDown ;
15
16 / / Statements
17 anyUp = anyUpRequest (elev );
18 anyDown = anyDownRequest (elev );
19
20 if (direction == UP)
21 {
22 if (!anyUp && anyDown )
23 direction = DOWN ;
24 else
25 if (!anyUp && !anyDown)
26 direction = STOP;
27 > // UP
28
29 else if ( direction == DOWN )
30 {
31 if ( lanyDown && anyUp)
32 direction = UP;
33 else
continue
PROGRAM 12- 13 Elevator: Move (continued )

34 if (!anyDown && lanyUp )
35 direction = STOP;
36 > // DOWN
37
38 else if (direction == STOP)
39 {
40 if (anyUp)
41 direction = UP ;
42 else
43 if ( anyDown )
44 direction = DOWN ;
45 > // else if stop
46
47 if (direction = = UP)
48 moveUp ( elev );
49 else
50 if ( direction == DOWN )
51 moveDown (elev );
52 else
53 printf( " \ n* * * ** NO BUTTON PRESSED * **** " );
54 return;
55 > I I move
Program 1 2- 1 3 Analysis The move function is one of the more interesting functions in the simulation. At any
given point in the operation, the elevator can be in one of three states: moving up,
moving down, or stopped. We use an enumerated type to track the state.
If the elevator is moving up, we continue up as long as there are requests for
higher floors. If there are none, we test for down requests. If there are any down
requests, we change the state to down and proceed. If there are no up requests
and
no down requests , we change the state to stopped. This logic is seen in statements 21
through 47 of Program 12- 13. Note the use of the blocks to ensure that the if and
else statements are properly paired.
Program 12- 14 simulates the elevator movement as it moves up and
down.
PROGRAM 12 - 14 Elevator : Move Up and Move Down

1 / * === == moveUp =====
2 This function simulate s the movement of the elevator
3 when it is going up.
4 Pre given moving up elevator
5 Post up simulation is displayed on the screen
6 */
7 void moveUp ( ELEVATOR * elev )
8 <
continued
PROGRAM 12- 14 Elevator: Move Up and Move Down (continued )

9 // Statements
10 printf ( " \nThe door is being closed
11 printf ( " \nWe are going up." );
12 -
(elev >currentFloor )++;
13 -
while ( elev > buttons[elev->currentFloor ] != IN)
14 {
15 printf( " \n " );
16 timePass (2);
17 -
printf( "\nPassing floor %d " , elev >currentFloor );
18 printf( " \n " );
19 timePass ( 2 );
20 -
( elev >currentFloor )++ ;
21 > // while
22
23 - -
elev >buttons [elev >currentFloor] = OUT;
24 printf( " \nThe door is being opened ..." );
25 printf( "\n " );
26 printf( " \n ** * FLOOR %d ***** " ,
•
27 elev->currentFloor );
28 printf("\n " );
29 timePass(4 );
30 return;
31 > // moveUp
32
33 /* ===== =============== moveDown
34 This function simulates the movement of the elevator
35 when it is going down.
36 Pre given moving down elevator
37 Post down simulation is displayed on the screen
38 */
39 void moveDown (ELEVATOR * elev )
40 {
41 // Statements
42 printf( " \nThe door is being closed ..." );
43 printf("\nWe are going down" );
44
45
-
(elev >currentFloor) ;
—
46
-
while (elev >buttons [elev >currentFloor]
- W
{
47 printf( " \n" );
48 timePass ( 2);
49 printf("\nPassing floor %d", elev->currentFloor
50 printf("\n");
51 timePass ( 2 );
52
-
( elev >currentFloor ) ;
— contititw
PROGRAM 1 2 - 1 4 Elevator : Move Up and Move Down ( continued )

53 > // while
54 -
elev->buttons ( elev >currentFloor ] = OUT;
55 printf( "\nThe door is being opened
56 printf( "\n " );
57 printf( " \n * * *** FLOOR %d ***** ",
58 -
elev >currentFloor );
59 printf( " \n ” );
60 timePass ( 4 );
61
62 return;
63 } // moveDown
Program 1 2 - 1 4 Analysis Both moveUp and moveDown operate similarly. They move past the current floor in
the correct direction and "call out" the floors as we pass them. A timing loop is
included to simulate the time it takes the elevator to reach the next floor.
Program 1 2 - 1 5 contains two functions: anyUpRequest checks for floors
above the current position; anyDownRequest checks for floors below the cur-
rent position.
PROGRAM 1 2 - 1 5 Elevator : Any Up and Any Down Request

1 === anyUpRequest ===================
2 This function checks to see if any request is for a
3 floor above the current floor ,
5 Post return true if button above current floor
6 pushed; return false otherwise
7 */
8 bool anyUpRequest ( ELEVATOR* elev )
9 {
11 bool isAny = false;
12
13 // Statements
14
15
_
check <= TOP FLOOR && !isAny;
-
for ( int check = elev >currentFloor ;
16 check++)
17 isAny = (elev > buttons[check ] = = IN );
-
18 return isAny ;
19 } // anyUpRequest
20
anyDownRequest =================
=====
21
This function checks to see if any request is for a
22
continued
PROGRAM 12 - 15 Elevator: Any Up and Any Down Request (continued )

23 floor below the current floor ,
25 Post return true if button below current floor
26 pushed ; return false otherwise
27 */
28 bool anyDownRequest ( ELEVATOR * elev )
29 {
31 bool isAny = false ;
32
33 // Statements
34 for (int check -
= elev >currentFloor;
35 check >= 0 ;
36 check ) --
37
38
isAny = isAny |
return isAny ;
-
| ( elev > buttons[check ] == IN );
39 > // anyDownRequest
Program 12- 15 Analysis Compare the for loops in anyDownRequest and anyUpRequest. One is much
more efficient than the other. Do you see why? In anyUpRequest, we stop when we
find the first request for a floor above the current one. In anyDownRequest, we
check all floors below the current one, even if the first one has been requested. This
means we always examine all lower floors.
Note the logic to set isAny at statement 37 in Program 12- 15. The current setting
is or ed with a button. If either are true, the result is true. Thus, once we find a floor has
been requested, isAny will remain set regardless of the settings of the other buttons.
This is a common technique to analyze a series of logical values.
Program 12 - 16 contains a loop to simulate time passing. We use a simple
for loop .
PROGRAM 12 - 16 Elevator: Time Pass

1 / * ===== ==== timePass ===
2 This function simulates the concept of passing time by
3
4
executing an empty for loop.
Pre
-
time to be passed ( number of moments)
5 Post time has passed
6 */
7 void timePass (int time )
8 {
9 // Statements
10 for ( long i = 0; i < (time* DELAY FACTOR ); i++)
11
continueA
PROGRAM 12 - 16 Elevator: Time Pass (continued )

12 return;
13 > // timePass
Program 12 - 16 Analysis The factor in the timing loop depends on the speed of your computer. In this case, we
use 10,000. On a faster computer, you need to make the factor larger. On a slower
computer, you need to make it smaller.
Note how we code the null statement in the for loop . Putting the null statement on a
line by itself makes the function more readable. If the null statement were put at the end
of the for statement, it would be too easy to think that the next statement, the return in
this function, were part of the for loop.
I he termination code is shown in Program 12 - 17 .
PROGRAM 12 - 17 Elevator: Terminate
1 ==== terminate ===
2 Release the memory occupied by buttons ,
4 Post elevator memory is released
5 */
6 void terminate ( ELEVATOR * elev )
7 {
8 // Statements
9 free (elev->buttons);
10 return;
11 } // terminate
a program, we
Program 12 - 17 Analysis Although we do not need to release memory when terminating
include the logic to do so for completeness.

In this section we discuss two important aspects of program design : function
coupling and data hiding.
Coupling
In Chapter 1 1 , we discussed a concept known as functional cohesion , a mea -
sure of how closely related the processes are within a function. A related
topic , coupling, is a measure of how tightly two functions are hound to each
other. The more tightly coupled they are, the less independent they are . Since
our objective is to make the modules as independent as possible, we want
them to he loosely coupled .
Loose coupling is desirable for several reasons:
1 . Independent —that is, loosely coupled —functions are more likely to be
reusable.
2. Loosely coupled functions are less likely to create errors in related func-
tions; conversely, the tighter the coupling, the higher the probability that
an error in one function will generate an error in a related function .
——
3. Maintenance modifications that is, modifications required to imple-
ment new user requirements are easier and less apt to create errors
w ith loosely coupled functions .
In his book on designing structured systems , Page-Jones describes five
types of coupling. We review them here. For an in -depth discussion of the
concept, reler to Chapter S in Page-Jones s book , The Practical Guide to
Structured Systems Design . 2
Data Coupling
Data coupling passes only the minimum required data from the calling func-
tion to the called function . All required data are passed as parameters, and no
extra data are passed . This is the best form of coupling and should be used
whenever possible.
When you write simple functions that work on on y one task the cou ,
.
*
. couPPn8* C ns der, for example, the’ function

pling naturally tend* k i33 ,
exchange in the sele f ° Fund
exchange » ,
iniegd I
° '7
IIM
, -
tWO ln egers that it will exchange and
nothing else. It moke, erences to any data outside the function, except
through the paramo °PO^,ntcrs bis function uses data coupling and is
highly reusable. ^ *
^
Well - structured functions are highly
cohesive and loosely coupled.
2 . Meilir Page Jones , The Practical Guide

- to Structured Systems Design , Yourdon Press Comp
ing Series ( Upper Saddle River, N . .:
J Prentice Hall , 1988 ) .
We could have fallen into the trap of passing extra parameters by passing
the function , the array, and the index locations of the two integers to be
exchanged . The function would have worked just as well, but the coupling
woidd not have been as loose. Now it requires an array of integers instead of
just integers, furthermore , we could have made the coupling even tighter
had we referred to the maximum size of the array using the precompiler dec -
laration S I Z E. At this point , it is highly questionable whether the function
could be used in another program .
Stamp Coupling
functions are stamp coupled if the parameters are composite objects such as
arrays or structures. Most ol the functions in the selection sort in Chapter 10
.
use stamp coupling because they pass the array ( Although it could be argued
that we are passing only a pointer to the array, the intent is to modify the
array. We are, therefore , passing the array for the purposes of this discussion . )
You should now be arguing, “ But we have to pass the array! ” Yes, that is
true. Stamp coupling is not bad and is often necessary. The danger with
stamp coupling is that often it is just too easy to send a structure when all the
data in the structure are not required . When extra data are sent , we begin to
open the door for errors and undesired side effects.
Consider the time stamp described in Figure 12- 15 . This structure con -
tains two nested structures, date and time . If we were to use these data , for
example , to print the date in a report heading, and passed the whole struc-
ture , we would send too much data ! In addition , if we were to pass the struc-
ture by address rather than by value , we risk the possibility of an error in one
function accidentally changing the data in the structure and causing a second
error. The correct solution is to pass only the data that are needed and then
only by value when possible.
Stamp coupling should pass only the data needed.
A common practice to reduce the number of parameters required for a

function is to create a structure that contains all the data the function needs
and pass it. Page-Jones refers to this as bundling .* It is a common practice
,
but it is not a good practice for three reasons.
is more difficult to trace
1 . Maintenance is made more difficult because it
data through a program .
2 Extra data can be passed . For example , a
. bundled structure is created for
a series of related functions, but not all
of them use all the data . The
temptation is just too great to pass the structure even though only one or
two of the members are needed .
.
3 The semantics of the structure are
often artificial , making the program
more difficult to read and understand .
3 Meilir Page-Jones, The Practical

Guide to Structured Systems Design.
Don’t bundle unrelated data to reduce the number of parameters .

Control Coupling
Control coupling is the passing of flags that may be used to direct the logic
flow of a function. It closely resembles data coupling except that a flag is
being passed rather than data.
In C, flags are often returned from a function rather than being passed as
parameters, hut the intent and usage are the same. For example, consider the
return values from scanf. It returns either EOF , a definite flag, or the number
of values successfully read, which can also be used as a flag for success. An
example of a flag being passed in a function you might write is the user-
selected option in the menu function of an interactive program. This flag
directs the entire flow of the program. The option is a special type of flag
known as a data flag. It is data entered by the user and at the same time itisa
flag intended to direct the flow of the program.
Properly used, control coupling is a necessary and valid method of com-
municating between two functions. Like stamp coupling, however, it can be
misused. Properly used, it communicates status: The end of the file has been
reached. The search value was found.
Poor flag usage is usually an indication of poor program design, for exam-
ple, when a process is divided between two or more independent functions.
Flags used to communicate horizontally across several functions in the struc -
ture chart are often an indication of poor design. Action flags, as opposed to
status flags, that require the receiving function to perform some special pro-
cessing are also highly suspect . An example of an action flag is a flag that
directs a customer’s purchase not he approved rather than simply reporting that
the credit limit has been exceeded or that no payment was received last month.
Control coupling should be used only to pass status .

Global Coupling
mIre° funafonseSp? bilJ‘ ‘ab,,
.,
8l0 Var es to communicate between two, orusu-
ally 8c .
>nes ca ls il common coupling With all that we
aj Ms „„” Z r* floW niclue
'
have said about i

not come as sor «
* In fact, it should never
be used,
° 7^ ^
J C tec
UP
You
should nev g| coupIing for several reasons. We cite only the
big three. ° USt
°^
Gl0bal
imPossible to determine which tomodbe-
ules
' ^
are communTcttf “
made to a program therT^ .CfC ot er* When a change needs
^ ^ the
** Poss,ble to evaluate and isolate
^
0 IS not
impact of the change ru - S tCn not changed
to suddenly fail. * ° causes functions that were
££££?l, z«z:;Zzsram "

that
'
“ n
“ ""'
Chapter J 2 [numerated, Structure, and Union Types 803
3. Global coupling leads to multiple flag meanings . This problem is often

made worse by using generic flag names , such as f 1, f 2 , f 21. ( Find -
ing 21 flags in a single program is not an exaggeration . We know of one
assembly program that bad more flags. In fact , it had one flag that was
used solely to indicate that another flag had been set but was now turned
off ; in other words, a flag that returned the status of a flag! )
The danger here should he obvious. If a flag can he used globally to
communicate between two functions, it is highly probable that at some
point this flag could he erroneously changed by a third f unction that used
it for another purpose .
Avoid global coupling within a program .
Content Coupling
The last type of coupling is very difficult , hut not impossible , to use in C.
Content coupling occurs when one function refers directly to the data or
statements in another function . Obviously, this concept breaks all the tenets
of structured programming.
Never use content coupling .

Referring to the data in another function requires that the data be made
externally visible outside the function . This is impossible in C. The only thing
that comes even remotely close is global variables. Since we have stressed the
dangers of global variables before, we will simply state here that they should
not he used for communication within one compile unit .
Data Hiding
We have previously discussed the concept of global and local variables. In
the
, we pointed out that anything placed before main was said to be
discussion
in the global part of the program . With the exception
of data that need to be
other compile units, no data need to be placed in this
visible to functions in
section .
that the internal
One of the principles of structured programming states
data structure should he hidden from the user’s view . The two terms you
usu ally hear are data hiding and data encapsulation . Both of these principles
objective protecting data from accidental destruction by parts
have as their
of your program that don ’t need access to the data . In other words , if a part
data do its job , it shouldn ’t be able to see
of your program doesn’t need to
the data .
Programming Standard:
Vo not place any variables in the global area
of a program .
mam
—
Any variables placed in the global area of your program that is , before
an be used and changed by every part of your program . This is in
direct conflict with the structured programming principles of data hiding and
data encapsulation .
Summary
We have described five different ways that two functions can communicate.
The first three are all valid and useful , although not without some dangers.
These communication techniques also provide data hiding. Data coupling is
universally accepted and provides the loosest communication between two
functions. Stamp and control coupling present some dangers that must be
recognized . When using stamp coupling, do not pass more data than are
—
required . Keep control coupling narrow that is , between only two functions.
You should always avoid the last two , global and content coupling. They do
not protect the data .

1. Don t forget the semicolon at the end of the declaration of structures and
unions. This is one ot the locations where you see a semicolon after a
closing brace ( > ) in C.
2. Because the direct selection operator has a higher precedence than the
indirection operator, parentheses are required to reference a member
with a pointer.
( * ptr ).mem
3. The indirect selection operator ( - > ) is one token . Do not put a space
between its symbols ( between - and > ).
4. Io access a member in an array structure, you need the index operator.
For example, the correct expression to access a member named mem in an
array of structure named ary is
ary[i].mem
5 . The type name in a tvpedef comes after the closing brackets and before
the semicolon :
typedef struct
{
... r
... ,
_
} TYPE NAME;
6 . You cannot define a variable at the same time that you declare a type def -
inition . In other words, you are not allowed to do the following:
typedef struct
{
... ,
... ,
} TYPE NAME variable name;
_ // ERROR
7. You may not use the same structure name with a tag inside structure
a .
The following declaration is not valid :
typedef struct TAG NAME

_
{
... ,
... ,
_
struct TAG NAME field name;
_ I I ERROR
> ID;
Section 12.8 Key Terms
8. A union can store only one of its members at a time. You must always
keep track of the available member. In other words, storing one datatype
in a union and accessing another data type is a logic error and may be a
serious run - time error.
9. Although structures can be assigned to each other, it is a compile error to
assign a structure of one type to a structure ol a different type.
10. It is a compile error to use a structure type with only its tag; that is, with-
out the keyword struct, as shown below.
struct stu
{
>;
stu aStudent ; // Compile Error
struct stu aStudent; // Correct Code
.
11 It is a compile error to compare two structures or unions, even if they arc
ol the same type .
.
12 It is a compile error to refer to a structure member without qualification,
such as id rather than student ,id.
.
13 It is a compile error to omit a level of qualification when structures arc
.
nested For example, in Figure 12 - 13, it is an error to omit the time qual-
ifier in the following reference.
stamp.time.min
14. It is a compile error to forget the address operator with scanf when refer-
ring to anonstring member. The pointer is the address ol the structure,
not .
the address of the member
13. Referencing an identifier in a union when a different type is active is a
logic error and may cause the program to fail.
16. It is a compile error to initialize a union w ith data that do not match the
type of the first member.
12.8 Key Terms

content coupling slack bytes
control coupling stamp coupling
data coupling structure
enumeration constant structure variable
global coupling tagged structure
nested structures union
Chapter 12 Enumerated, Structure, and Union Types 80 /
12.9 Summary
An enumerated type is built on the standard type, integer.
In an enumerated type, each identifier is given an integer value.
A structure is a collection of related elements, possibly of different types,
having a single name.
Each element in a structure is called a field.
One difference between an array and a structure is that all elements in an
array must he of the same type while the elements in a structure can be of
the same or different types.
We have discussed two different ways to declare and/or define a structure:
tagged structure and type-defined structure.
A structure can be initialized when it is defined. The rule for structure ini-
tialization is the same as for array initialization.
LI We can access the members of a structure using the direct selection
operator ( )..
Structures can he accessed through a pointer. The best way to do this is to
use theindirect selection operator ( - > ).
The following two expressions are the same if ptr is a pointer to a
structure:
.
( *p ) x p- >x
j The information in a structure can he sent to a function using one of the

following methods:
1. Sending individual members
2. Sending the whole structure
3. Sending a pointer to the structure
A union is a construct that allows a portion of memory to he used by differ-
ent types of data.
two func -
In software engineering, coupling is the measure of how tightly
lions are hound to each other.
: data, stamp, con-
Computer science has identified five types of coupling
trol , global, and content.
Functions in a well-structured program are loosely coupled
.
Avoid global coupling and never use content coupling.
: Modules must
Good program design can he measured by three principles
he independent, modules must he loosely coupled, and each module must
do a single job.
12.10 Practice Sets

Review Questions
1 . Variables are created with a type definition ( typedef ) .
a . True
h . False
2 . An integer value can he assigned to only one enumeration constant in an
enumerated type.
a . True
b. False
3. A structure variable is used to declare a type containing multiple fields.
a . True
b. False
4. The indirect selection operator is used with a pointer to access individual
fields in a structure.
a . True
b. False
3. The structured programming concept known as coupling describes how
data are passed to functions.
a . True
b. False
6. Which ot the following is not a derived type?
a . Arrays
h . Enumerated
c. Float
d . Pointers
.
e Union
.
7 The can he used to create a new type that can be
used anywhere a type is permitted.
a . array
h . record type
c. structure (struct )
d . type definition
e. both a structure and a type definition
S. The enumerated type ( enum ) is derived from
the type.
a. character
b . boolean
c. floating- point
d . integer
e. structured
.
9 Which of the following statements about enumerated types is true?
a. Declaring an enumerated type automatically creates a variable.
b. Declaring an enumerated variable without a tag creates an enumer-
ated type.
c. Enumerated types cannot be used in a type definition.
d. I he enumerated values are automatically assigned constant values
unless otherwise directed.
e. I he identifiers in an enumerated type are enumeration variables.
10. A ( n) is the smallest element of named data that has
meaning.
a. array
b. field
c. record type
d. structure ( struct )
.
e type
I 1 . Which of the following statements about structures ( struct ) is true ?
.
a A structure without a tag can be used as a derived type.
b. Structures are derived types based on the integer standard type.
.
c The fields in a structure must be of the same type.
.
d C normally uses two different structure declarations: tagged and
type-defined.
.
12 The structure creates a type that can be used later to
define variables.
.
a array
b. record- typed
.
c tagged
d. type-defined
e . variable
1 3. Given a structure variable named stu. whose type is STU, which contains
a field named major , which of the following statements correctly
refers
to major?
a. major
b. stu major-
c. stu.major
d. STU-major
e. STU.major
containing a field name, which of
14. Given a pointer, ptr, to a structure stu
the following statements correctly references name?
a. ptr.name
b. ptr->name
c. ptr.stu.name
d. ptr >stu >name
-
e. ptr >stu.name
-
-
.
1 5 A(n) _ is a construct that allows a portion of memory
to be shared by different types of data.
.
a array
.
b field
.
c struct
d. union
e. variable
.
16 Determine which of the following statements are true and which are false:
.
a A structure cannot have two Reids with the same name.
.
b A structure cannot have two fields of the same type .
.
c A structure must have at least one field.
.
d A field in a structure can itself be a structure.
.
17 Determine which of the following statements are true and which are false:
a. A union can have another union as one of the fields .
b. A structure cannot have a union as one of its fields.
c. A structure cannot have an array as one of its elements.
d. W hen accessing the fields of a structure through a pointer p, the fol-
lowing two expressions are the same:
( *p ) . field_name p- > field_name
Exercises
18. Declare a tagged structure for a student record consisting of five fields:
student ID ( integer ) , first name ( a dynamically allocated string), last
name (a dynamically allocated string), total credits completed ( integer ) ,
and accumulated grade point average ( floating- point ). A graphic repre -
sentation of the structure is shown in Figure 12 - 25 .
i
t
i
i
H [IJEIlllLiJ
id irst last totCr gpa
FIGURE 1 2 - 25 Data for Exercise 18
f 30 inventory item consisting of six

.* ' ^
fields: part number Onteeert ârt cj ° ,
( integer), unit measmvm ' ’ » num
string), reorder point ( ’ Cger
'
,
escnPt on (a dynamically allocated
)er of items currently on
hand
string, maximum size 8), and unit price
*
( floating- point ) . A graphic representation of the structure is shown in

Figure 12 - 26.
_
'
_ _ . _ _ _ [_ T. _ :
fTTTTTlTt
i i i i i l. i LJ
-
partNo descr
i reOrder onHand unitMeas price
20. Declare an array of 12 elements. Each element is a structure with

three fields. The first field shows the month in numeric form ( 1 to 12 ).
rhe second field shows the name of the month ( a dynamically allo-
cated string ) . The third field shows the number of days in the month .
Initialize the array. A graphic representation of the structure is shown
-
in Figure 12 27 .
<
l
__
i
i
I
i
i
i
i
i
i
i
i
I l
_ II _ |L i
i
i
i
[0 ]
month alphaMonth days
7T Tir T7 1
U l - - -J i
in month alphaMonth days
i i i
r
i i
i i i
7i
LI
[ in
month alphaMonth days
21 . Declare a calendar as an array of 366 elements

. Each element of the array
is a structure having three fields . The first
field is the name of the month
( a dynamically allocated string ) . The second field is
the day in the month
( an integer ) . The third field is the description
of activities for a particular
) . A graphic representation of the
day ( a dynamically allocated string
structure is shown in Figure 12 28 . -
_ ••_ _
l
i
i
l LV
_ _ _•_
J
»
i
l l
'
*: : : :;;.7 TTi
month days activity
i I
l
I
I
I
l
_V
l
l -J
I i
i . _ - -- LI
I
'i «
i
m ' '
' -i
_ _ L.V.: Ji l'.lIJ.
I
I
i
i
.
month days activity
i
«
i
_ _ _ __ _
•«
l
i
i
L
•L..." J
i
i i
i
i
l t J
i
month days activity

« i
i
.
i
«
— .*
j
i
i. i. .
i
22 . Imagine we have declared the following structure:
typedef struct
{
char x;
char* y;
int z[ 20 ];
> FUN ;
Determine which of the following definitions are valid and which are
invalid. II invalid , explain the error.
a. struct FUN fl;
I). struct FUN f 5[ 23];
c. FUN f 3 ;
d. FUN f4 [ 20 ];
23. Imagine we have the following declaration and definitions:
struct FUN
{
char x ;
char* y ;
int z[20 ];
> ;
struct FUN fnl ;

struct FUN fn2 ;
struct FUN fn3 [ 10];
struct FUN fn4 [50];
Determine which of the following assignment statements are va

and which are invalid. If invalid , explain the error.
^
a. fnl.x = ' b';

b. f n2 . y = ' b ' ;
c. fn 3[4 ].z[ 5 ] = 234 ;
d. fn4 [ 23 ] . y = "1234";
e. fn4[23] = fn3[5 ];
.
24 Imagine we have the following declaration:
typedef enum {ONE = 1, TWO = 2} CHOICE ;

typedef union
{
charchoicel ;
int choice2;
> U TYPE;
typedef struct
{
float fixedBefore;
CHOICE choice;
_
U TYPE flexible;
float fixedAfter ;
} S TYPE;
Draw a schematic diagram for S TYPE. _

_
25. Using the declaration of S TYPE ( declared in Exercise 24), show what
will he printed from the following program segment. (Assume that the
S TYPE declaration is global.)
_
# include <stdio.h>
int main (void)
{
_
S TYPE s;
S TYPE* ps ;
s.fixedBefore = 23.34 ;
s.choice = ONE;
s.flexible.choicel = 'B ;
1
s.fixedAfter = 12.45 ;
ps = &s;
--
printf ( "\n%f " , ps >fixedAfter );
printf (" \n%d " , ps >flexible.choicel );
printf ( "\n% f " , s.fixedBefore);
return 0;
} // main
Problems
. me with two parameters, the start time
26 Write a function called elapsedTi
three fields showing
and the end time. Each parameter a structure with
is
the hours, minutes, and seconds of a specific time ( see Figure 12- 13).
The function is to return a time structure containing the time elapsed
between the two parameters. You must handle the situation when the
start time is in the previous day.
.
27 Write a function called increment that accepts a date structure with
three fields. The first field contains the month (a pointer to a string). The
second field is an integer showing the day in the month. The third field is
.
an integer showing the year The function increments the date by I day
and returns the new date. II the date is the last day in the month, the
month field must also he changed. II the month is December, the value
of the year field must also he changed when the day is 3 1 . A year is a leap
year if
.
a It is evenly divisible by 4 but not by 100, or
b. It is evenly divisible by 400.
.
28 Write a function called futureDate. The function is to use two parame-
ters. The first parameter is a structure containing today s date ( as defined
in Problem 27 ) . I he second parameter is an integer showing the number
ol days after today. The function returns a structure showing the next
date, which may he in a future year .
29. Write a function called later that receives two date parameters, com-
pares the two dates, and returns true ( 1 ) if the first date is earlier than
the second date and false (0) if the first date is later. Each parameter isa
structure containing a date ( as defined in Problem 27 ) .
.
30 Write a function that accepts an integer representing money in dollars
( no fraction) and returns a structure with 6 fields. The fields represent,
respectively, the minimum number of $ 100, $ 30, $ 20, $ 10, $5, and
S I hills needed to total the money in the parameter.
.
31 \ \ rite a function that compares two fraction structures (see Figure 12 - 3 ».
II the Iractions are equal, it returns zero. If the fraction in the first
parameter is less than the fraction in the second parameter, it returns a
.
negative number. Otherwise, it returns a positive number Hint: Convert
the fraction to a floating-point number.
32. A point in a plane can be represented by its two coordinates, x and y -
I hereIore, we can represent a point in a plane by a structure having two
fields as shown below.
typedef struct
{
int x;
int y;
> POINT;
W rite a function that accepts the structure representing a point and

returns an integer ( 1 , 2, 3, or 4) that indicates the
in which quadrant
point is located, as shown in Figure 12 - .
29 Zero
is positive.
x y
I positive positive
II negative positive
III negative negative
IV positive negative
FIGURE 1 2- 29 Quartile Coordinates for Problem 32
.
33 A straight line is an object connecting two points. Therefore, a line can
be represented by a nested structure having two structures of the type
POINT, as defined in Problem 32.
typedef struct
{
POINT beg ;
POINT end ;
> LINE;
Write a function that accepts two parameters of type POINT and

returns a structure of type LINE representing the line connecting the
two points.
34. Write a function that accepts a structure of type LINE ( see Problem 33)
and returns an integer ( 1, 2, 3), where 1 means vertical, 2 means hori-
zontal, and 3 means oblique. A vertical line is a line whose x coordinates
are the same. A horizontal line is a line whose y coordinates are the
same. An oblique line is a line that is not vertical or horizontal.
35 Write a function that shuffles a deck of cards. The deck of cards is repre-
.
.
sented by an array of 52 elements Each element in the array is a struc -
ture for one card, as shown below.
typedef struct
{
char* suit ; // Clubs, Diamonds, Hearts , Spades
int value; // Ace, 2..9, Jack , Queen , King
} CARD;
typedef CARD DECK [ 52 ];

)
must use a random number ( see Chapter 4 to ensure
The function
that each shuffle results in a different card sequence . Hint : Generate a
the current card
random number in the range 0... 51, and then exchange
with the card in the random position .
.
36 Program 12 -8 creates a union
of company and person names. Write a
program that reads names from the keyboard and places them into an
the user to enter a code that indi-
array of names. For each entry, prompt
if the name is a company or person name. After all names have
cates
been entered , print the array. Your program is to contain separate Func-
tions lor reading names and printing the array.
37. Modify the program in Problem 36 to add a function to sort the rnames
after they have been entered . Person names arc to he sorted on last
name , first name , and initial.
Projects
38. Write a program to keep records and perform statistical analysis for a
class of students. For each student , we need a name of up to 20 charac-
ters, an ID for four digits, four quizzes, and one examination The stu . -
.
dent data will he stored in an array of student structures Provide for up
to 50 students.
.
The input is read from a text file Each line in the file contains a stu-
dent ’s name, four quiz scores, and one examination score in order. If a
quiz or examination was not taken , the score is zero. The students name ,
the quiz scores, and the examination score are all separated from each
,
other by one or more spaces. A new line ends the data for one student.
I he number ol lines in this file is the same as the number of students.
I he output consists of a listing of the students in the order they are
.
read from the File; no sorting is required Print each student on a separate
line with an appropriate caption for each column. After the last student,
print the highest , lowest , and average score for each quiz and the exami-
nation . In determining the lowest score, do not consider zero scores. A
suggested report layout is shown in Table 12 - 1 .
DATA
Name ID Quiz Quiz Quiz Quiz Exam
1 2 3 4
Student 1 1234 23 19 22 23 89
Student 2 4321 0 23 21 18 76
Student n 1717 21 22 18 19 91
STATISTICS
Highest scores 23 25 23 25 96
Lowest scores 17 15 12 18 53
Average scores 21.3 20.1 19.8 21.1 81.3
TABLE 12- 1 Sample output for Project 38

The data tor the project are shown in Table 12- 2.
Name
Julie Adams 1234

i Quiz
52
1
7
Quiz
2
Quiz
100
3
Quiz
78
4
Exam
34
Harry Smith 2134 90 36 90 77 30
Tuan Nguyen 3124 100 45 20 90 70
Jorge Gonzales 4532 11 17 81 32 77
Amanda Trapp 5678 20 12 45 78 34
Lou Mason 6134 34 80 55 78 45
Sarah Black 7874 60 100 56 78 78
Bryan Devaux 8026 70 10 66 78 56
Ling Wong 9893 34 09 77 78 20

Bud Johnson 1947 45 40 88 78 55
Joe Giles 2877 55 50 99 78 80
Jim Nelson 3189 82 80 100 78 77
Paula Hung 4602 89 50 91 78 60
Ted Turner 5405 11 11 0 78 10
Evelyn Gilley 6999 0 98 89 78 20
TABLE 12 - 2 Data for Project 38

39. Rework Project 38 to report the average quiz score, total quizzes score
,
and total score for each student. Then assign a grade based on an abso -
lute scale of 90% for A, 80% for B, 70% for C , and 60% for D. Any score
below 60% is an F. A total of 500 points are available. Print the student
data to the right of the input data. At the end of the report , print
the
number of students who earned each grade , A to F.
structures to answer
40. Write a program that uses an array of student
. Using a menu - driven user interface , provide inquiries that
inquiries
based absolute scale
report a students scores, average, or grade on an
( 90% A. 80% B , etc. A ). fourth menu option provides all data for a
requested student , and a fifth prints a list of student IDs and names . To
create the array, load the data from Project 38
.
41 . Using a sort of your choice, modify
Project 40 to sort the data on stu -
dent ID.
.
42 A standard deck of playing cards consists of 52 cards as shown in
I’able 12 - 3 . Create an array of 52 structures to match a new deck of
cards as shown in Table 12 - 3.
Suit Rank
Clubs Ace...King
Diamonds Ace . ..King
Hearts Ace . .. King
Spades Ace ...King
TABLE 1 2 - 3 The Order of a New Deck of Playing Cards

Then simulate shuffling the deck using Algorithm 12 - 1.
ALGORITHM 12- 1 Shuffle Deck of Cards

Algorithm shuffle
1 loop through each card in the deck
1 get a random number in range 0...51
2 Swap current card with card at random position
2 end loop
end shuffle
Alter the cards have been shuffled, print them.

.
43 \\ rite a function that calculates the area of one of the geometric figures
shown in Iable 12 - 4. I he function receives one parameter, a structure
that contains the type of figure and the size of the components needed
.
lor the calculation structured as a union The format of the structure is
shown in Table 12 - 4. Use an enumerated type for the figure type .
Figure Type Components
Rectangle length width
Circle width
Triangle sidel side2 side3
TABLE 12- 4 Geometric Figure Components
I he formulas lor the figures are shown below.
RectanglcArea = length x width

CircleArea = rcr “
TriangleArea = Jt ( t - s i d e l ) x t ( t - s i d e 2 ) x t ( t - s i d e 3 )
1
where t = -( s i d e l + s i d e 2 + s i d e3 )
Then write an interactive test driver that prompts the user for the
type of figure to be entered and the appropriate components. It then calls
the area function and prints the area before looping for the next input .
I lint : Use a switch statement to process the different codes.
44 . Modify Program 12 - 5 to include functions to add , subtract, divide , and
compare fractions. Also change the print function to print the fraction in
—
its simplified format that is, to reduce the numerator and denominator
by the largest common factor. Thus, the fraction
20
8
would be printed as
5
2
The compare function is to compare the functions algebraically.

Thus, if the following two fractions were compared , it would return equal:
,
—
20 and 5
8
- 2
Then modify main to include a loop that allows the user to enter two
the
fractions and an option to add , subtract, multiply, divide, or compare
fractions.
Binary Input/Output
In Chapter 7, we discussed text input /output . We introduced text streams
and formatting text input and output. We also discussed some text input and
output facilities in Chapter 11.
In this chapter, we discuss binary input/output. We first show the differ -
ence between a text stream and a binary stream. We then discuss the C input
and output I unctions, which are mainly designed to be used with binary
streams. We conclude with a discussion of converting text files to binary files
and binary files to text files.
Objectives
To understand the differences between text and binary files
To write programs that read, write, and/ or append binary files
To be able to detect and take appropriate action when file errors occur
To be able to process files randomly
To be able to create text files from binary files and vice versa
To understand and be able to implement file merges
To understand and be able to implement the classic sequential file update
821
822 Section 13.1 Text versus Binary Streams
13.1 Text versus Binary Streams

In Chapter 7 , we described inputting data from a source and outputting data to
a destination: text input/output and binary input/output. For text input/output,
we convert a data type to a sequence of characters . For binary input/output, we
transfer data without changing their memory representations.
We also discussed that some data sources and destinations are capable of
only reading and writing text . For example, a keyboard and a monitor can be a
destination for only text ; a keyboard is not capable of handling binary input
and output . On the other hand , a file can be used to read or write any type of
data , both text and binary.
Text and Binary Files

Although a file records data only in a binary format, we can distinguish
between a text file and a binary file.
Text Files
A text file is a file in which data are stored using only characters; a text file is
written using a text stream . Non -character data types are converted into a
sequence of characters before they are stored in the file. When data are input
from a text file , they are read as a sequence of characters and converted into
the correct internal formats before being stored in memory. We read and
write text files using the input/output functions that convert the characters to
data types as shown in Figure 13 1 . -
Destination Character Converting Data

Function
Standard
Data Type
Source Character Converting Data

Function
FIGURE 1 3 - 1 Reading and Writing Text Files
of threc categories: formatting
^ ^ ,? ° nsis*String
1
(Chapter 7), character Ch m ~
/ * "
lunctions, such as scanf an,* t7 rnitJr reformat

r
-
Chapter 1 D The formatting
a series of input characters into
'
standard types or reforir dn d r l Pc data into characters for output. 1
he
Sj ^^ *
character input and out n . . ns suc 1 as ge char and putchar, input or
output one character at a i
input and output strings of
^ Unti . * such as ets
!. ° ’ functions
’ &
Chapter 13 Binary Input/Output 823
Formatted input/output, character input /output, and string input/output

functions can be used only with text files.
Binary Files
A binary file is a collection of data stored in the internal format of the computer.
Unlike text files, the data do not need to he reformatted as they are read and writ -
ten ; rather, the data are stored in the file in the same format that they are stored
in memory. Binary files are read and written using binary streams known as block
input/output lunctions. Figure 13- 2 depicts the operation of block read and write
functions.
Block Write
Destination
Function
Data
Block Read
Source
Function
Data
FIGURE 1 3 - 2 Block Input and Output
Differences between Text and Binary Files

Let’s review the major differences between text files and binary files.
The major characteristics of text files are:
1 . All data in a text file are human - readable graphic characters.
2 . Each line of data ends with a newline character.
3 There may be a special character called end -of -file ( EOF) marker at the
.
end of the file.
Binary files, on the other hand , store data in their internal computer for-
mat This means that an int in C is stored in its binary
format , usually 4 bytes
in a PC ; a character is stored in its character format , usually 1 byte; and so
forth . There are no lines in a binary file but there is an end -of - file marker.
The major characteristics of binary files are :
in memory.
I Data are stored in the same format as they are stored
2 . There are no lines or newline characters
.
3 There may be an end -of -file marker.
.
in a text file ( which
Figure 13 3 shows how two data items are stored
-
uses 8- bit ASCII code) and in a binary file.
824 Section 13.1 Text versus Binary Streams
short int 768 char PA
7 6 8 A
00110111 00110110 00111000 01000001 00000011 100000000 1010000Q1
768 H — A - 768
* *
Text File Binary File
FIGURE l 3 - 3 Binary and Text Files
In the text file, the number 768 is stored as three numeric characters. The
character A is stored as one character. At the end, there is an end-of-Hle marker
that is not shown because its representation depends on the implementation.
In the binary file, the number 768 can be stored in 2 bytes, which
assumes that the short int type has a size of 2 bytes. The character A is stored
as I byte. At the end ol the file there is an end-of- file ( EOF) marker, again
its format is implementation dependent.
Text files store data as a sequence of characters; binary files store data as
they are stored in primary memory.
State of a File
An opened file is either in a read state, a write state, or an error state. If we
want to read from a file, it must be in the read state. If we want to write toa
(ile, it must be in the write
state. The error state results when an error
occurs when the file is opened or when it is read or written. II we try to read
from a ( ile in the write state, an error occurs. Likewise, if we try to write toa
( ile in the read state, an error
occurs. When a file is in an error state, we can-
not read from it or write to it.
Io open a (ile in the read state, we specify the read mode in the open
statement. A file opened in the read state must already exist. II it doesntthen
the open fails.
A (ile can he opened in the write state using either the write or append
mode. In write mode, the writing starts at the beginning of the file. If the file
already exists, its data are lost . In append mode, the data are added at the end
ol the file. II the ( ile exists, the writing begins after the existing data. If lhc
file does not exist, a new file is created.
In addition to read, write, and append, files can he opened in the updaU
mode. Updating allows a file to he both read and written. Even when the
is opened lor updating, however, it can
still be in only one file state at a time
1 Pen a file lor updating, we add a plus sign ( ) to the basic mode. T e
°°
initial state ol a file opened for updating is
+
+
determined by the basic mode: r
opens the file in the read state, while w -f and the write
a+ open the file in
state. Iable 13 - 1 lists the six file
modes with their features.
Mode
Open State read write write read write write
Read Allowed yes no no yes yes yes
Write Allowed no yes yes yes yes yes
Append Allowed no no yes no no yes
File Must Exist yes no no yes no no
Contents of Existing File Lost no yes no no yes no
TABLE 13 - 1 File Modes i

I he file states and their potential error conditions are shown in Figure 13- 4.
Study this figure carefully. II a state is shaded gray, then that state is not avail-
able lor the indicated file mode. As shown in the read mode (r ) , only two
states arc possible, read and error. The file will stay in the read state as long
as we use only read functions. T hat is the meaning of the looping arrow in
the read - state circle. However, if we try to write when the file is in the read
state , the state changes to the error state. Once the file is in an error state ,
any subsequent attempt to read it will result in an error.
read write write

positioning
, functions
write read / write read write
state state /
* —estate state state
write read read

/
error error error
state state state
read mode ( r ) read update mode ( r+) write mode ( w)
write write write

read positioning M
positioning
read
. functions / write read write read — . / write
>Junctions
state /
state /
* —Istate state state
* state
write read write read

/
error error
state
state
append mode (a) append update mode (a+)

write update mode ( w +)
FIGURE 13 - 4 File States

826 Section 13.1 Text versus Binory Streams
Now look at the read update mode, r + . In this mode, the file can be in
either the read or the write state. To move from one to the other, we must use
one of the positioning functions, which are discussed later. That is the mean-
ing of the arrows between the read state and write state. If we try to write
after a read without repositioning, the file will he in the error state. The error
can be cleared by clearerr , which is discussed in the section " File Status
Functions” later in the chapter.
Opening Binary Files

The basic open operation is unchanged lor binary files only the mode —
changes. The function declaration for /open is repeated here for convenience.
FILE* fopen ( const char* filename , const char* mode);
Recall that the file name is the name used for the file by the operating
system . The name we use internally in our program is a file pointer whose
address is filled by the fopen return value. Here , we are most interested in the
mode , which is a string that contains a code defining how the file is to he
—
opened —for reading, writing, or both —and its format text or binary Text
files use just the basic modes discussed above. To indicate that a file is binary,
.
we add a binary indicator ( b ) to the mode. The six binary file modes are: read
binary ( rb) , write binary ( wb ), append binary ( ab ), read and update binary
( r + b ) , write and update binary ( w + b ) , and append and update binary ( a + bl.
I he following examples contain examples of open statements lor binary files.
spReadBin = " myFile.bin " , "rb" ); // Read Binary

spWriteUp = "myFile.bin" , "w+b"); // Write with Update
spApndBin = " myFile.bin" , "ab" ); // Append Binary
I iguu I 3- 5 shows the basic file modes and the file state they create. The
-
marker ( s ) represents the end -of file marker.
m a ]0
File File File
Marker Marker Marker
Read Mode ( r, r+) Write Mode ( w, w+) Append Mode (a, a+)
FIGURE 1 3 - 5 File -Opening Modes
Closing Binary Files

text files , binary files must be closed when they are not needed
Just like
anymore. Closing destroys the file table and erases the logical file name . The
close function was covered in Chapter 7. Its function declaration is repeated

here for your convenience.
int fclose (FILE* sp);
13.2 Standard Library Functions for Files

C has eight categories of standard file library functions ( see Figure 13 - 6 ). We
have already discussed the first four in Chapter 7 and Chapter 11; open and
close, character input and output, formatted input and output, and line input
and output. We discuss the other lour categories, which are more related to
binary files, in this section.
File
Open/Close
Character
Input/Output
Formatted
Input/Output
Line
Input/Output
Categories of
I/O Functions
Block
Input/Output
File
Positioning
System
File Operations
File
Status
FIGURE 13 -6 Types of Standard Input/ Output Functions
Block Input/Output Functions

to read and write
The C language uses the block input and output functions
data to binary files. As we discussed previously , when we read and write
transferred just as they are found in memory. There
binary files, the data are
are no format conversions . This means that, with the exception of character
828 Section 13.2 Standard Library Functions for Files
data, we cannot “see ” the data in a binary file; it looks like hieroglyphics. If
you have ever accidentally opened a binary file in a text editor, you have seen
these strange results. The block read function is file read ( / read ). The block
write function is file write ( fwrite ).
File Read: fread

The function fread , whose declaration is shown below, reads a specified num-
her of bytes from a binary file and places them into memory at the specified
location .
i n t fread ( void * plnArea ,

int elementSize ,
int count ,
FILE * sp ) ;
The first parameter, p l n A r e a , is a pointer to the input area in memory.

Note that a generic (void ) pointer is used. This allows any pointer type to lx?
passed to the function .
file read expects a pointer to the input area, which is usually a structure.
I his is because binary files are most often used to store structures.1 However,
C gives us the flexibility to read any type of data , from a character to a com-
plex structure or even a multidimensional array.
I he next two elements, elementSize and count , are multiplied to deter -
mine how much data are to he transferred. The size is normally specified using
the sizeof operator and the count is normally one when reading structures.
-
I he last parameter is the associated stream . Figure 13 7 is an example of
a file read that reads data into an array of integers. When fread is called. it
transfers the next three integers from the file to the array, inArea .
ED ED 0
3*4
— 12 bytes before
read
T after
read
- __T
inArea
^ HI
fread ( inArea , sizeof (int ) , 3, spData ) ;
FIGURE 1 3-7 File Read Operation
l. r programming languages, structures

stored in tiles are known as records.
Chapter 13 Binary lnput/Output 829
_
I he code to read the file is shown in Program 13 - 1 .
PROGRAM 1 3 - 1 Read File of Integers

1 // Read a file of integers, three integers at a time.
2 {
3
5 FILE* spIntFile;
6 int itemsRead ;
7 int intAry[3];
8
9 // Statements
10 spIntFile = fopen( " int file.dat" , "rb" );
11
12 while (( itemsRead = fread( intAry ,
13 sizeof( int), 3, spIntFile)) != 0)
14 {
15 // process array
16
17 } // while
18
19 > // block
File read returns the number of items read. In Figure 13 - 7, it will range
between 0 and 3 because we are reading three integers at a time. For exam-
ple, assume that when we try to read the file, only two integers are left to be
read. In this case, / read will return 2. If we return and try to read the file
againffread will then return 0.
Note that freiul does not return end of file—it returns the number of ele-
ments read. End of file is detected in the above situation when we called
fread with fewer than three integers left in the file. The question is, how can
we tell that we are at the end of file? C provides another input /output func
-
tion. feof to test for end of
}
file. We discuss this function in “ File Status Func -
tions ” later in this section.
Now let ’s look at a more common use of fread : reading structures
(records). Assume that we have defined a structure that stores data
about stu -
dents. Given the type of data that will he stored , we would expect the struc -
some string data and other data , such as integers or real
ture to contain
that they
numbers. One advantage of block input output / functions is can
transfer these data one structure ( record ) at a time . A second advantage is
that the data do not need to be formatted . Figure 13 - 8 shows the operation of
fread when a structure is being read.
830 Section 13.2 Standord Library Functions for Files
Before Read
I[HDD irnmin rrmo

?
t
before
oneStudent read
imrrnn
oneStudent
mn after
read
After Read
FIGURE 1 3 - 8 Reading a Structure
The code shown in Program 13- 2 reads the file.
PROGRAM 1 3 - 2 Read Student File

1 /* Reads one student 's data from a file
2 Pre spStuFile is opened for reading
3 Post stu data structure filled
4 ioResults returned
5 */
6 int readStudent (STU * oneStudent , FILE* spStuFile )
7 {
9 int ioResults ;
10
11 // Statements
12 ioResults = fread(oneStudent,
13 sizeof(STU ), 1 , spStuFile);
14 return ioResults ;
15 } // readStudent
Program 1 3- 2 Analysis Different companies have different standards. One

company with which we are
familiar has a standard that programs shall have only one read and one write state
ment for each file . The standard was
created to make it easier to make changes to the
programs. Program 1 3 - 2 is a typical implementation of this standard. One difficulty
with this type of function, however, is that it is impossible to generalize the steps that
are to be taken for various input results
, such as error handling and end of fib
Therefore, we pass the input/output result back to the calling function for analysis
and action.
Chapter 13 Binary Input/Output 83]
File Write: fwrite

I he function fwrite , whose declaration is shown below, writes a specified
number of items to a binary file.
int fwrite ( void* pOutArea ,

int elementSize ,
int count ,
FILE* sp);
1 he parameters for file write correspond exactly to the parameters for the
file read function.
4 * 3 ^ 1 2 bytes
/ /
IP
..
/ / before
write
/
/
/
/
/
/
E Al
/
t after
write
outArea
- TT
// /
/
/
/
fwrite (outArea , sizeof ( int ), 3 , spOut );

/
FIGURE 1 3 - 9 File Write Operation
Functionally, fivrite copies elementSize x count bytes from the address

specified by pOutArea to the file. It returns the number of items written. For
exampie, if it writes three integers, it returns 3. We can use the return value,
therefore, to test the write operation. If the number of items written is fewer
than count, then an error has occurred. Depending on the device we are
working with, it may be possible to repeat the write, but generally the pro-
gram should be aborted when we get a write error. Figure 13 - 9 shows
the
write operation that parallels the read in Figure 13 - 7.
Assuming that we are writing a file of student structures, Figure 13 - 10
function
shows the write operation. Program 13 - 3 contains the code for the
that w rites the data.
Before Write
itznn nurmn A
s
file
marker
aStudent
mn iirmn 11
file
marker
aStudent
After Write
FIGURE 13 - 10 Writing a Structure
PROGRAM 1 3 - 3 Write Structured Data

1 /* Writes one student ’ s record to a binary file.
2 Pre aStudent has been filled
3 spOut is open for writing
4 Post aStudent written to spOut
5 */
6 void writeStudent (STU * aStudent , FILE* spOut )
7
8 {
10 int ioResult;
11
12 // Statements
13 ioResult = fwrite( aStudent,
14 sizeof(STU ), 1 , spOut );
15 if ( ioResult != 1 )
16 {
17 printf( " \a Error writing student file \a\n");
18 exit (100);
19 > // if
20 return ;
21 } // writeStudent
Program 1 3 - 3 Analysis Contrast this write structured data

function with the one that we wrote to read data
.
Although it is not possible to generalize on the action to be taken if data are not read
It IS possible to do so with write
errors. If the program cannot write data, it must be
aborted. Therefore, we put the error checking and action in the write function itself.
File Status Functions

C provides three functions to handle file status questions: test end of file
, error (/error ) , and clear error ( clearerr ) 2
( feoj ) test .
Test End of File: feof
I he feof function is used to check ii the end of file has been reached. If the
— —
file is at the end that is, if all data have been read the function returns
nonzero ( true ). If end of file has not been reached , zero ( false ) is returned .
int feof (FILE* stream );
In general , two different techniques can be used to detect end of file.

Some languages have a look-ahead function . When look- ahead logic is being
used , the system transfers the current data to the program and then reads the
next data. Under this design , we can detect the end of file at the same time
that we read ( transfer data back to our work area ) the last data from the file.
The second technique , the one used by C , detects end of file when we
attempt to read and there is nothing left on the file . Even if all the data have
been read from the file, feof does not return true until we attempt to read
beyond the last data .
Test Error: ferror

Test error ( /error ) is used to check the error status of the file . Errors can be
created for many reasons , ranging from bad physical media ( disk or tape ) to
illogical operations, such as trying to read a file in the write state. The /error
function returns true ( nonzero) if an error has occurred . It returns false
( zero ) if no error has occurred . The function declaration is shown below.
int ferror ( FILE* stream);
Note, however, that testing for an error does not reset the error condi-
tion. Once a file enters the error state ( see Figure 13-
4 ) it can only return to
a read or write state by calling clear error ( see below ) .
Clear Error : clearerr

return nonzero, until the
When an error occurs, the subsequent calls to ferror
file is reset . The function clearerr is used for this purpose .
error status of the
Its function declaration is shown in the next
example .
void clearerr (FILE* stream);
the Boolean type , most standard functions continue

2 . Although the C99 Standard introduced .
nonzero for true and zero for false return status
to use the traditional
834 SectionJ 3L 2 Standard Library Functions for Files
Note, however, that even though we have cleared the error, we have not
necessarily cured the problem . We may find that the next read or write
returns to the error state .
Positioning Functions
Positioning functions have two uses. First , for randomly processing data in
disk files (we cannot process tape files randomly ) , we need to position the File
to read the desired data . Second , we can use the positioning functions to
change a file’s state. Thus, il we have been writing a file, we can change to a
read state after we use one of the positioning functions. It is not necessary to
change states after positioning a file, but it is allowed .
We will discuss three file position functions: rewind , tell location, and
file seek .
Rewind File: rewind

I lie rewind function simply sets the file position indicator to the beginning of
the file ( Figure 13- 11 ). I he function declaration is shown below.
void rewind(FILE* stream );
A common use of the rewind function is to change a work file from a

write state to a read state. Often it is necessary to place data in a file tempo-
rarily lor later processing. When all the data have been written and we are
ready to begin reading, we rewind the file and simply start reading. Remem-
ber, however, that to read and write a file with only one open, we must open it
in update mode, in this case , w+ or w+ b.
Before Rewind
mem EDD
file
marker
inj|nrmn
After Rewind
file
marker
FIGURE 13 - 11 Rewind File
I he same effect tan be accomplished by closing the output file and open
mg K aS inPut However, the rewind
* is a faster operation .
Chapter 13 Binary Input/ Output 835
Current Location: ftell

I he ftell function reports the current position of the file marker in the file,
relative to the beginning of the file. Recall that C considers files as streams of
data. It measures the position in the file by the number of bytes, relative to
zero from the beginning of the file. Thus, when the file position indicator is
at the beginning of the fitey ftell returns zero. If the file position indicator is at
the second byte of the filey ftell returns I , representing the position 1 byte off-
set from the beginning of the file. The function declaration for ftell is
shown below.
long int ftell ( FILE* stream );
Note that ftell returns a long integer. This is necessary because many files
have more than 32,767 bytes, which is the maximum integer value on many
computers. The operation of ftell is graphically shown in Figure 13 - 12.
Beginning
^
of File ^
mm
Current
Number of Bytes location (16)
FIGURE 13 - 12 Current Location ( ftell) Operation
of
Another important factor to consider is that ftell returns the number
bytes from the beginning of the file. This is true even when we read or write
structures. If we need to know the structure

number relative to the first
. This he done by dividing the ftell
structure, then we must calculate it can
return value by the size of the structure , as shown below.
numChar =
numStruct = numChar
ftell ( sp);
/ sizeof(STRUCTURE TYPE);
_
In Figure 13 - 12, ftell returns 16. Since
each structure is 4 bytes, the
above is 4 , which means that there are four
result of the calculation shown
the current location. Another way to look at it is that the file
structures before
is positioned at the fourth integer
relative to zero .
U' ftell encounters an error, it returns I We -. know of only two conditions
, ftell with a device that cannot store
that can cause an error. The first using
.
, program logic or design error The second
data, such as the keyboard is a
when the position is larger than can be represented in a long
error occurs
836 Section 13.2 Standard Library Functions (or Files
integer. Obviously, this could occur only with very large files, but files of more
than a million records are common in industry.
The primary purpose offtell is to provide a data address (offset ) that can
be used in a file seek . It is especially useful when we are dealing with text files
for which we cannot calculate the position of data .
Set Position: fseek

The fseek function positions the file location indicator to a specified byte
position in a file . It gets its name from the disk-positioning operation , seek.
Seek moves the access arm on a disk to a position in the file for reading or
writing. Since this is exactly the purpose of seek file , it is an appropriate
name. Its function declaration is shown below.
int fseek(FILE* stream, long offset, int wherefrom);
The first parameter is a pointer to an open file. Since the seek is used
with both reading and writing files , the file state can he either read or write.
The second parameter is a signed integer that specifies the number of bytes
the position indicator must move absolutely or relatively. To understand what
we mean by absolutely or relatively, we must first discuss the third parameter.
wherefrom .
C provides three named constants that can he used to specify the starting
point ( wherefrom ) of the seek. They are shown below.
tdefine SEEK SET 0

tdefine SEEK CUR 1 __
# define SEEK END 2
_
When wherefrom is SEEK SET or 0 , then the offset is measured abso-
lutely from the beginning of the file. This is the most common use of file
seek. Thus , to set the file indicator to byte 100 on a file , we would code the
following statement:
_
f s e e k ( sp , 99 L , SEEK SET ) ;
_ .
We can use zero in place of SEEK SET If you are puzzling over the sec
ond parameter in the above statement , remember that the file position is rel-
ative to zero and must be a long integer. Actually, the compiler is smart
enough to convert an integer value to long integer, but it is more efficient if
...
we specify the correct type , especially with
< > U Cts
OI
° k at
°
literals.
_
wherefrom option , SEEK CUR . If wherefrom
displacement is calculated relatively from the cur-
. - displacement is negative , the file’s position moves
i
P <
JSILOIL ^
ack toward the beginning of the file. If
it is positive, it moves forward toward
VC e. It is an error to move beyond the beginning of the file* 1
we move beyond the end of the file, the of
file is extended , but the conten ts
nc extended bytes are unknown . Whenever
we extend the file , there is aM *
the possibility of running out of space , which is an error. To position the file
marker to the next record in a structured file , we execute the following
statement:
_
fseek( sp, sizeof( STRUCTURE TYPE), SEEK CUR ); _
To position the student file described earlier at the structure indicated by
the integer variable stuLoc , we use the following statement:
fseek(stuFile, ( stuLoc - _
1)* sizeof( STU ), SEEK SET);
It is necessary to adjust the integer location , stuLoc, by subtractin g 1 to

.
convert the ordinal structure number to a zero base That is, if stuLoc con -
tains 55 , indicating we want to read the 55 th student in the file , we position
the file to the location of the 54 th student relative to zero.
_
Finally, il wherefrom is SEEK END or 2, the file location indicator is posi-
tioned relative to the end of the file. If the offset is negative, the file position
marker is moved backward toward the beginning of the file; if it is positive, it
extends the file. This technique can he used to write a new' record at the end
_
of the file We simply position the file at the end with a SEEK END and a dis-
.
placement of zero as shown below and then write the new record .
fseek(stuFile , OL, SEEK END);
The seek function returns zero if the positioning is successfu l . It returns

nonzero if the positioning is not successfu l. Figure 13- 13 shows the effect of
fseek in different situations.
am am *
ED nan QID
_
fseek (sp, 4 * sizeof(STRUCTURE TYPE), SEEK SET );
an urn A
fseek ( sp,
"
1
4 * sizeof(STRUCTURE_TYPE), SEEKEND);
TED ran CUED DOED *

_ _
fseek (sp, 2 * sizeof(STRUCTURE TYPE), SEEK CUR );
FIGURE 13 - 13 File Seek Operation

The file seek is intended primarily for binary files. It does, however, have
limited functionality with text files. We can position a text file to the begin -
ning using / seek with a zero offset from the beginning of the file (SEEK SET).
However, rewind provides the same functionality and is more appropriate for
text files. To position a text file at the end , we can use fseek with a zero offset
and a wherefrom SEEK_END, as discussed earlier.
We cannot use file seek to position ourself in the middle of a text file
unless we have used ftell to record the location . The reasons for this have to
do with control codes, newlines, vertical tabs, and other nuisances of text
files. However, if we have saved a location using ftell and want to go hack to
that position , we can use fseek , as shown helow.
_
fseek( sp , ftell location , SEEK SET); _
Note that since ftell returns a position relative to the beginning ol the
file , we must use SEEK _SET when we reposition the file.
EXAMPLE 13- 1 Block Input/output: Append Files

Let ’s look at a program that reads and writes binary files. Suppose , lor exam -
ple , that we had two copies of files with integer data . Perhaps one file repre-
sents data from one week and the other file represents data for a second
week. We want to combine both files into a single file. The most efficient way
to do this is to append the data from one file to the end of the other file. This
logic is shown in Program 13-4.
PROGRAM 1 3 - 4 Append Two Binary Files

1 / * This program appends two binary files of integers.
2 Written by:
3 Date:
4 */
7
8 int main ( void )
9 {
11 FILE* spl ;
12 FILE* sp2;
13 int data ;
14 long dataCount ;
15 char filelD[ 13];
16
cotitiIIM#
PROGRAM 1 3- 4 Append Two Binary Files (continued )

17 // Statements
18 printf( "This program appends two files.\n " );
19 printf( " Please enter file ID of the primary file: " );
20 scanf( "%12s", filelD);
21 if (!( spl = fopen ( filelD, " ab " )))
22 printf("\aCan't open %s\n", filelD), exit (100);
23
24 if (!( dataCount = ( ftell ( spl ))))
25 printf( "\a%s has no data \n " , filelD), exit ( 101);
26 dataCount /= sizeof( int );
27
28 printf( "Please enter file ID of the second file: " );
29 scanf(" %12s" , filelD);
30 if (!( sp2 = fopen ( filelD , "rb " )))
31 printf( " XaCan ’ t open %s\n ", filelD ), exit (110 );
32
33 while ( fread (&data , sizeof( int), 1 , sp2 ) == 1)
34 {
35 fwrite (&data , sizeof( int ), 1 , spl );
36 dataCount++;
37 } // while
38
39 if (! feof(sp2 ))
40 printf( "\aRead Error. No output.\n "), exit ( 120);
41
42 fclose ( spl );
43 fclose ( sp2 );
44
45 printf( " Append complete: %ld records in file\n" ,
46 dataCount );
47 return 0;
48 } // main
Program 1 3 - 4 Analysis First, note the way the files are opened in Program 13 4. Because it appends
- the data
from the second file to the end of the first file, the first file is opened in append mode
and the second file is opened in read mode. Both files are opened as binary.
The open is suc-
Opening the first file in append mode presents a minor problem:
the file marker
cessful even if the file doesn't exist. Recall that the append mode places
file , if there is no existing file , at the beginning of a new file.
at the end of an existing or
To ensure that an existing file was opened successfully
, therefore, we use the ftell func-
, dataCount will be nonzero.
tion in statement 24. If an existing file was opened
file are handled. Under the assumption
Now look at the way the external names
that this program will be used to append different files at different times, we asked the
user to enter the file names. This technique provides maximum flexibility for generalized
programs.
Statement 22 contains one of the few valid uses of multiple statements on one
—
line in this case, multiple expressions separated by the comma operator. Why is this
statement considered valid when we have so strongly emphasized one statement per
line? The answer is that we are handling an error situation that should never occur. In
good human engineering style, we want to deemphasize the error logic so that it
doesn't distract the reader. Therefore, we place the print error message and exit in
one expression .
The heart of the program is contained in statements 33 through 37. As long as
the read is successful, we keep going. When we reach the end of file, the read
returns zero and the loop terminates. We now have another problem, however. We
don' t know if the read terminated at end of file or because of a read error. We there-
fore use the feof function to make sure we read to the end of the file.
The program concludes with a printed message that contains the number of
records on the appended file. Since we didn' t read all the data on the first file, how-
ever, we must calculate the number of integers in it. To do this, we used the result of
the ftell that verified that the first file existed. Recall that ftell returns the number of
bytes to the current location in the file — in this case, the end of the file. To get the
number of integers, we simply divide by the size of one integer. (See statement 26.)
Then, each time we write, we add 1 to the record count. This gives us the number of
integers in the combined file.
System File Operations

A few functions operate on the whole file instead of the contents. These
functions generally use operating system calls to perform operations such as
remove a file, rename a file, or create a temporary binary file.
Remove File: remove

1 he remove function removes or deletes the file using its external name. The
parameter is a pointer to the name of the file. Its function is shown below.
int remove (char* filename );
It returns zero it the deletion is successful . It returns nonzero if there is

an error, such as the file can ’t be found . For example , if we want to delete a
.
file named filel dat , we execute the following statement .
if ( remove ( " filel.dat" ))

printf( " Error, file cannot be deleted ");
Any attempt to access a file after it has been removed will result in
an error.
Chapter 13 Binary Input/Qutput 841
Rename File: rename

\\ hen we create a new version ol a file and want to keep the same name, we
need to rename the old version ol the file. I he rename function declaration is
shown below.
int rename ( const char* oldFilename ,

const char* newFilename );
Both the old name and the new name must he given as parameters. The
rename function returns zero il renaming is successful; it returns nonzero if
there is an error.
For example, in a DOS system, if we want to rename a student file and
designate it a backup, we could use the rename function as shown below.
if ( rename ( "STUFILE.DAT" , "STUFILE.BAK " ))

printf( " Error, the file cannot be renamed ");
Create Temporary File: tmpfile

The tmpfile function creates a new temporary output file. Although we could
do the same thing with an fopen in the w+b mode, the difference is that the
file is available only while the program is running. It will be closed and erased
when the execution of the program is finished. It is a temporary file, not a
permanent one. Its function declaration is
FILE* tmpfile (void);
To create a temporary file, we first define a file pointer and then open it,
as shown below.
FILE* sp ;
sp = tmpfile ( );
Now we can write to the file as we would to any file. Because the file is
opened in the w+b mode, if we want to read from the file we must reposition
or fseek .
it using one of the repositioning functions such as rewind
13.3 Converting File Type

to convert a text file to a
A rather common but somewhat trivial problem is
binary file and vice versa. C has no standard functions for these tasks. We
must write a program to make the conversion. We describe the file conversion
logic in this section.
12 Section 13.3 Converting File Type
Creating a Binary File from a Text File

To create a binary file, we usually start with data provided by the user. Since
—
the user is providing the data , it will be in human readable form that is, in
text form. If only a small amount of initial data are required , they are often
read from a keyboard . With a lot of data , however, it is easier for the user to
enter the data with a text editor and then read the text file and create the
binary file .
When we read the text file , we can use either the fscanf function to con -
vert the data as they are being read , or the /gets and sscanf functions to read
the data a line at a time and then convert the data to the proper internal for-
mat . As the data are being converted , they are placed in a structure. At the
end of each line , the structure is written to the binary file. This process is
repeated until the text file has been completely converted to the binary struc-
ture . The structure chart for this program is shown in Figure 13- 14.
createFile
getData write
BinaryFile
FIGURE 13 - 14 Create Binary File Structure Chart
Let ’s assume that we want to convert student data to a binary file. The
data consist of a student s name, ID , three exams , eight problems, and a final
grade. In the text file version , each field is separated by one or more
whitespace characters and each student ’s data are stored on a separate line.
The create file program is shown in Program 13- 5.
PROGRAM 13 - 5 Text to Binary Student File

1 /* Reads text file of student data & creates binary file‘
2 Written by:
3 Date:
4 */
8
9 // Type Declarations
cotitittw
PROGRAM 1 3 - 5 Text to Binary Student File ( continued )

10 typedef struct stuData
11 {
12 char name[ 26 J ;
13 char id[ 5 ];
14 int exams[ 3 ];
15 int problems[8];
16 char grade;
17 > STU DATA;
18
20 bool getData ( FILE* textFile,
21 STU DATA *_ aStudent);
22 void writeBinaryFile ( STU_DATA * aStudent ,
23 FILE* binFile);
24
26 {
28 char * -
textFilelD = "P 13 stu.txt " ;
29 char* -
binFilelD = "P13 stu.bin" ;
30
31 STU DATA aStudent;
32
33 FILE* textFile ;
34 FILE* binFile;
35
36 // Statements
37 printf( "\nBegin Student Binary File CreationXn ");
38 if (!( textFile = fopen( textFilelD , " r " )))
39 {
40 printf( " \nCannot open %s\n " , textFilelD);
41 exit ( 100 );
42 >// if textFile
43 if (!( binFile = fopen( binFilelD, "wb " )))
44 {
45 printf( " \nCannot open %s\n" , binFilelD);
46 exit ( 200);
47 } // if binFile
48
49 while ( getData ( textFile, &aStudent ))
50 writeBinaryFile (&aStudent , binFile );
51
52 fclose( textFile );
53 fclose( binFile);
continued
844 Section 13.3 Converting Pile Type
PROGRAM 1 3 - 5 Text to Binary Student File ( continued )

54 printf( " \n\nFile creation complete\n" );
55 return 0;
56 > / / main
57
58 ===== getData ===== ========
59 This function reads the text file.
60 Pre textFile is opened for reading
61 Post data read and returned
62 */
63 _
bool getData ( FILE* textFile , STU DATA* aStu)
64 {
66 char buffer[ 100 ];
67
68 / / Statements
69 fgets( buffer , sizeof( buffer ), textFile );
70 if (!feof( textFile))
71 {
72 sscanf( buffer , " %s %s %d %d %d %d %d %d %d %d %d %d %d Ic" /
73
74
-
aStu >name, aStu >id , -
-
&aStu >exams[ 0 ],
75 - -
& aStu >exams( 1 ],&aStu >exams[ 2 ],
76
77
-
&aStu >problems[0], &aStu >problems[ 1],
-
78
-
&aStu > problems[ 2 ], & aStu > problems[ 3 ],
-
79
-
& aStu >problems[ 4 ], &aStu >problems[ 5 ],
-
& aStu->problems[6], &aStu >problems[7],
80
-
81
-
&aStu >grade );
return true ;
82 } / / if
83 return false ;
84 } // getData
85
86 /* = writeBinaryFile =
87 Write student data to a binary file.
88 Pre binFile is opened as a binary output fUe
89 aStudent is complete
90 Post Record written
91 */
92 void writeBinaryFile ( STU DATA _
* aStudent ,
93 FILE* binFile)
94 {
96 int amtWritten;
97
continue
PROGRAM 13 - 5 Text to Binary Student File (continued )

98 // Statements
99 amtWritten = fwrite ( aStudent ,
100 _
sizeof(STU DATA), 1 , binFile);
101 if ( amtWritten != 1 )
102 {
103 printf( "Can 't write student file. ExitingXn " );
104 exit ( 201);
105 > // if
106 return;
107 > // writeBinaryFile
Program 1 3 - 5 Analysis Note how we specified the external file names as strings. This makes it easy to
change the file names when necessary. It also allows us to identify the file by name if
the open fails.
The program starts and ends with a message that identifies what program is run-
ning and that it has successfully completed . This is a good programming technique and
a standard in many organizations.
The while loop is controlled by the results of the getData function call. To under-
stand how it works, therefore, first look at getData. When you need to read a lot of
text data, especially if it is coming from a file, the preferred technique is to use the get
string [ fgets ] function and then use the string scan ( sscanf ] function to convert it into
internal binary formats. The get string function reads text data until it finds a newline
character, an end of file, or the maximum number of characters has been read. It
returns the address of the input string, in this case buffer. When end of file is
detected, it returns a null pointer.
To make this function robust, we would break the sscanf into several calls, one for
the first two strings, one for the exams, one for the problems, and one for the grade at
the end. This would allow us to easily verify that the correct data were read by examin-
ing the returned value from sscanf .
Now that we understand that the getData function returns either an address or
NULL, we understand how the while statement in main works. It simply loops until an
end of file is detected- that is, until NULL is returned.
Creating a Text File from a Binary File

. The first is
Programmers convert a binary file to a text file in two situations
the data for people read ; we discuss this below.
when we need to display to
when it is necessary to export the data to another system ,
The second is
, if the word sizes for
which can 't read the binary file. This occurs, for example
the different hardware systems. As
integers and floats are different on two
formatted the same and they use the same character
long as all lines are
alphabet , text files are portable .
An interesting problem is to create a report
of the data in the binary file.
read the binary file and write the data as a text file, but
Obviously, we need to
it than that . First , the report needs a name , so each
there is much more to
page must have a title. The title

should include the report date and a page
meaningful , each column should have a column
number. To make the report
caption. Finally, the report should have an end of report message as the last
line of the report so that the user knows that all data have been reported.
Since we put a title on the top of each page, we need to know
when a
page is full. This is generally done by counting the number of lines on a
page
and when the count exceeds the maximum, skipping to the next page. Good
structured programming requires that the heading logic he in a separate
function. The design for printing the student data is shown in Figure 13 15
- .
printStudentsI
getData
I write
Report
pageHeaders
FIGURE 13 - 15 Design for Print Student Data
We write the data to a file so that it can be sent to the printer when we
want a hard copy. You should he aware of one final point before looking at the
program. In the print functions, we use a type modifier known as the s /«tic
storage class ( see statements 97 and 128). We discuss the
concept of storage
classes in Appendix K. l or now, you only need to know that static keeps the
contents ol a variable available between calls. The
first time we call a function
with a static variable, the variable is initialized. After that, its value is retained
between calls. I his is a great C feature that reduces the need to pass parame-
ters just to retain a local variable. The
program is shown in Program 13- 6.
PROGRAM 1 3 - 6 Print Student Data
1 /* Reads a binary file
of student data , and prints it.
2 Written by:
3 Date:
4 */
7
8
9
_
# define MAX LINES PER
_
#define BUFFER SIZE
_ _
PAGE 50
133
10 # define FORM FEED • \f 1
11
cotiti ,,,ie‘
PROGRAM 1 3 -6 Print Student Data (continued )

12 // Type Declarations
13 typedef struct stuData
14 {
15 char name[ 26 ];
16 char id[ 5 ];
17 int exams[3];
18 int problems!8 ];
19 char grade;
20 > STU DATA ;
21
23 _
STU DATA getData ( FILE* binFile );
24 void _
writeReport (STU DATA aStudent,
25 FILE* prtFile );
26 void pageHeaders (FILE* prtFile);
27
29 {
31
32
char
char
stuFileID[ ]
prtFilelD[ ]
-
= "P13 stu.bin";
= "P 13-stu.prt" ;
33 STU DATA aStudent;
34 FILE* stuFile ;
35 FILE* prtFile;
36
37 // Statements
38 printf( " \nBegin Student Report CreationNn ");
39
40 if(!(stuFile = fopen(stuFilelD , "rb" )))
41 {
42 printf( " \nCannot open %s\n " , stuFilelD );
43 exit ( 100 );
44 > // if stuFile
45 if (!( prtFile = fopen( prtFilelD, "w " )))
46
47 printf( \nCannot open %s\n" , prtFilelD);
M
48 exit ( 200);
49 } // if prtFile
50
51 aStudent = getData (stuFile );
52 while (!feof( stuFile ))
53 <
54 writeReport (aStudent , prtFile);
aStudent = getData (stuFile );
55
continued
PROGRAM 1 3 - 6 Print Student Data (continued )

56 } // while
57
58 fprintf( prtFile , "\nEnd of Report\n" );
59 fclose(stuFile);
60 fclose( prtFile);
61 printf( "\n \nEnd Student Report Creation\n " );
62 return 0;
63 > // main
64
65 /* === == getData == ==============
66 This function reads the student binary file.
67 Pre stuFile is opened for reading
68 Post one student record read and returned
69 */
70 _
STU DATA getData ( FILE* stuFile)
71 {
73 int ioResult ;
74 STU DATA aStu ;
75
76 // Statements
77 ioResult = fread(&aStu,
78 _
sizeof( STU DATA ), 1 , stuFile);
79 if (!ioResult)
80 if (!feof(stuFile))
81 {
82 printf( " \ n \nError reading student file\n ");
83 exit ( 100);
84 > // if !feof
85 return aStu ;
86 > // getData
87
88 /* ===== writeReport =====
89 Write student report to a text file.
90 Pre prtFile is opened as a text output file
91 aStudent is complete
92 Post Report line written
93 */
94 void writeReport (STU DATA aStu, FILE* prtFile)
95 {
96 // Local
Declarations
97 static int lineCount
_= MAX LINES PER PAGE
+ 1?
98 char buffer[ BUFFER SIZE];
99
continued
PROGRAM 13 -6 Print Student Data (continued )

100
101
// Statements
_
if ( ++lineCount > MAX LINES PER PAGE ) _ _
102 {
103 pageHeaders ( prtFile);
104 lineCount = 1;
105 } // if
106
107 sprintf ( buffer ,
108 -
" % 25s %4s %4d % 4d % 4d %4d %4d % 4d %4d %4d %4d %4d % 4d %c\n " ,
109 aStu.name , aStu.id ,
110 aStu.exams[ 0 ], aStu.exams( 1 ], aStu.exams[ 2],
111 aStu.problems(0],
112 aStu.problems[ 1 ], aStu.problems[ 2 ],
113 aStu.problems( 3], aStu.problems(4 ],
aStu.problems[ 5 ],
114 aStu.problems[ 6 ], aStu.problems[ 7 ],
115 aStu.grade );
116 fputs ( buffer , prtFile );
117 return ;
118
119
120
} // writeReport
pageHeaders =================
=====
!
121 Writes the page headers for the student report.
122 Pre prtFile is opened as a text output file
123 Post Report headers and captions written
124 */
125 void pageHeaders (FILE * prtFile)
126 {
128 static int pageNo = 0;
129
130 // Statements
131
132
pageNo++ ;
fprintf( prtFile, "%c", FORM FEED);
_
133 fprintf( prtFile, "% 66s Page %4d \n ",
-
134 "Student Report " , pageNo );
135 fprintf ( -
prtFile , " %-25s %-6s % 10s % 27s Grade\n\n ", -
136 "Student Name" , "ID " , " Exams " , "Problems " );
137 return ;
138 } // pageHeaders
Even though this program is rather simple, you

should note the following points:
Program 1 3 - 6 Analysis of lines per page and the print
First, we have declared the maximum number
buffer size as preprocessor defined
- constants. This makes it easy to change them
be necessary. It also makes it easy to set the print logic so that it will print the
should it
850 Section 13.4 File Progrom Examples
header the first time through the function. This leads us to the second point you should
note. The logic for pageHeaders will cause the first page to be blank. That is, we
issue a page form feed before any data have been written. This is standard in produc-
tion programs, but you may want to change it so that the first page is not wasted. In this
case, you will have to call the pageHeaders function before you start the file reading
to write the first headings and move the form feed write to just before statement 103.
The file is read using look -ahead logic . At statement 51, we read the first record of
the file. This allows us to use the feof function to control the while loop. At the end of the
loop, we read the file again. Although this loop closely resembles a posttest loop, there
is one difference . If the file is empty, a post - test loop would fail. By reading the first
record from the file before the loop, we ensure that we can process any file condition
using the simplest possible logic.
Note the way we handle the report title and line captions in pageHeaders.
Many programmers simply try to code them in the format string. This works, but it takes
a lot of trial and error to get it right. Our technique simply adds the widths from the
data write and uses them for the widths in the caption prints. The program may still
need a little manual adjustment, but it is a much simpler approach .
Finally, note how we formatted the student output. We used sprintf to format a line
of output. There is no significant advantage to using the string format function rather
than the print function, but we wanted to demonstrate its use. After the data have been
formatted, we use fputs to write them . Note also the way we aligned the data for read-
ability. Since it takes several lines, we group the common data on lines by themselves.
This makes it much easier to read the code. We could have used for statements to print
the array data, but that would have been less efficient.
13.4 File Program Examples

I his section contains two common file applications. The first uses the file
positioning functions to randomly process the data in a file. The second
merges two files .
EXAMPLE 13 - 2 Random File Accessing

Program 1 3- 7 demonstrates the concept of randomly accessing data in a file *
We begin by creating a binary file of integers. Each integer is the square of

the data ’s position in the file, relative to 1 . After the file has been created, we
print it in sequence, starting at the beginning of the file. We then print it in a
random sequence usingfseek and a random number generator.
PROGRAM 1 3 - 7 Random File Application

1 /* Shows application of
some functions we have studied
2 in this chapter. The program first creates a binary
3 file of integers. It then prints the file , first
4 sequentially and then randomly using rand( ).
5 Written by:
6 Date:
continue
PROGRAM 1 3 - 7 Random File Application ( continued )

7 */
10
12 void buildFile (FILE** sp);
13 void printFile ( FILE* sp );
14 void randomPrint (FILE* s p ) ;
15
17 {
19 FILE* fpData ;
20
21 // Statements
22 buildFile (&fpData );
23 printFile ( fpData );
24 randomPrint (fpData);
25 return 0;
26 > // main
Program 1 3 - 7 Analysis The main function simply calls three functions in order. The first function receives a
pointer to the file pointer, which is the only variable declared in main . It is necessary
to use double dereferencing here because buildFile needs to pass back the
file
pointer to main . The other two calls do not change the file pointer, they just use it.
Therefore, it can be passed to them by value.
EXAMPLE 13 - 3 Build Random File

Program 13 - 8 builds a binary file.
PROGRAM 13 - 8 Random File: Build File

1 ========== buildFile =====
Creates a disk file that we can process randomly
,
2
3 Pre nothing
4 Post file has been built
5 */
6 void buildFile ( FILE** spData )
7
9 int data;
10
11 // Statements
continuec
852 Section 13.4 File Program Examples
PROGRAM 1 3- 8 Random File: Build File (continued )

12 if (!( *spData = fopen( " SAMPLE.DAT" , "w+ b")))
13 {
14 printf( " \aError opening file.\n " );
15 exit ( 100);
16 } // if open
17 for (int i = 1; i <= 10; i++)
18 {
19 data = i * i ;
20 fwrite( &data , sizeof( int), 1 , *spData );
21 } // for
22 return ;
23 > // buildFile
Program 13 - 8 Analysis The buildFile function in Program 13 - 8 simply creates a file with ten records.
Each record consists of a single integer, which is the square of the ordinal record
number relative to 1 (not 0) . The file is opened with write plus so that we can first
write to it and then later in the program read it. Note that all references to the file use
the dereference operator to update the file pointer, which exists in main .
EXAMPLE 13 - 4 Sequentially Print a Random File

Program 13-9 prints the file sequentially.
PROGRAM 1 3 - 9 Random File: Sequential Print

1 /* === == printFile ==== :
========
2 Prints the file starting at the first record.
3 Pre spData is an open file
4 Post file has been printed
5 */
6 void printFile ( FILE* spData )
7 {
9 int data;
10 int recNum ;
11
12 // Statements
13 recNum = 0 ;
14 rewind(spData );
15 fread( &data , sizeof( int), 1 , spData );
16 while (!feof(spData ))
17 {
18 printf("Record %2d: %3d\n", recNum++, data);
19 fread(&data , sizeof( int ), 1, spData );
continue
PROGRAM 13- 9 Random File: Sequential Print (continued )

20 > // while
21 return;
22 } // printFile
Results:
Record 0: 1
Record 1: 4
Record 2: 9
Record 3: 16
Record 4: 25
Record 5: 36
Record 6: 49
Record 7: 64
Record 8: 81
Record 9: 100
Program 1 3 - 9 Analysis The printFile function in Program 13 - 9 simply reads the file sequentially starting
at the beginning (Record 0). Study the while loop . Note that we have coded it with
the first read before the loop. We use this technique so that we can detect end of file
in the while statement. The while loop prints the current record and then reads the
next record. When all records have been processed, the read will detect end of file,
and the loop will terminate with all records processed.
EXAMPLE 13- 5 Randomly Print Random File

Program 13 - 10 prints the file randomly.
PROGRAM 13 - 10 Random File: Random Print

1 / * =: ===== randomPrint ===
2 This function randomly prints the file. Some data
3 may be printed twice , depending on the random
4 numbers generated.
5 Pre spData is an open file
6 Post Ten records have been printed
7 */
8 void randomPrint ( FILE* spData)
9 {
11 int data ;
12 int randomSeek ;
13
14 // Statements
continued
PROGRAM 13 -10 Random File: Random Print ( continued )

15 printf( " \nFile contents in random sequence.\n " );
16 for (int i = 0; i < 10; i++)
17 {
18 randomSeek = ( rand( ) % 10 );
19 fseek(spData,
20 sizeof(int) * randomSeek , SEEK SET); _
21 fread(&data, sizeof(int), 1 , spData );
22 printf("Record %3d ==> %3d\n",
23 randomSeek , data);
24 } // for
25 return ;
26 > // randomPrint
Results:
File contents in random sequence.
Record 8 ==> 81
Record 8 ==> 81
Record 3 ==> 16
Record 5 ==> 36
Record 1 ==> 4
Record 7 ==> 64
Record 0 ==> 1
Record 9 = > 100
Record 2 = => 9
Record 6 ==> 49
Program 13- 10 Analysis The randomPrint functi
from *
zero to 0106 TU IS
^ * ^
° mo< u ° ten' which gives us potential record numbers
record nndf 0 ° ^
corresP nc s exactly with the file on disk, which occupies
^**^ » *‘
*
on output created using our computer is
,e uncstarts
shown in Pro9
s all oth" "" °* ^ ^ Piling Record 8 twice and then
prmT
Program 1 3 - 9
Merge Files
AS L
°
^
Qn Recorb
bX
Compare the output to the file print in
,
In “ B ock Input/Output Function • Se .
. 13 2 we discussed the con-
” c ,CtIon >
of
cePl combining two files hv aPPenc ng the data. In this section , we discuss
.
another wav to romKin A " '
,
, i“, When w merge data from two
fife . he | i, one C
requires that we comnl '| Wlj
Ibis
^
1 l e ata ol dered in key sequence. This
^ ^ *
create a new output flic

.
Chapter 13 Binary Input/ Output 855
/
28 / :
•/ «
2 7T
23 19
21 16
Merge Merge
File 1 18 15 File 2
17 12
14 10
10 12 14 15 16 17 18 19 21 23 25 27 28 31
Output ( Merged) File
FIGURE 13 - 16 File Merge Concept
The merge files pseudocode is shown in Algorithm 1 3- 1 . The design is rather

simple. We start by reading data from each merge file. We then ( 1 ) compare the
data in the two files, ( 2 ) write the smaller to the merge output file, ( 3) read the
next record from the file whose record was written , and ( 4 ) continue the loop.
ALGORITHM 13 - 1 Pseudocode for Merging Two Files

Algorithm MergeTwoFiles
This program merges two files
1 input (Filel , Reel )
3 HighSentinel
-
2 input (File2, Rec2 )
high value-
4 loop ( not eof(Filel)) OR ( not eof(File2 ))
1 if Reel.Key <= Rec2.Key then
1 output (File3, Reel )
2 input (Filel , Reel )
3 if eof(Filel)
1 Reel.Key = HighSentinel
4 end if
2 else
1 output (File3, Rec2 )
2 input ( File2, Rec2)
3 if eof(File2 )
1 Rec2.Key = HighSentinel
4 end if
continued
ALGORITHM 13- 1 Pseudocode for Merging Two Files (continued )
3 end if
5 end loop
End MergeTwoFiles
The difficult part of the merge design is the end - of - file logic. One of
the
merge files will end first , hut we never know which one . To simplify the
end
of file processing, this design introduces a concept known as a sentinel.
A
sentinel is a guard ; in our merge algorithm , the sentinel guards the end of
file. The sentinel has the property that its value is larger than any possible
key. For the sentinel value, we use INT_MAX , which is found in the limits.
!I
header file . The code is seen in Program 13- 1 1 .
PROGRAM 13 -11 Merge Two Files

1 /* This program merges two files
2 Written by:
3 Date:
4 */
6 #include <stdlib.h>
7 # include <limits.h >
8
_
9 #define READ MODE "rb”
10 #define WRITE MODE " wb "
11
12 int main (void )
13 {
15 FILE* spMl;
16 FILE* spM 2 ;
17 FILE* spOut;
18
19 int recMl;
20 int recM 2 ;
21
22
int sentinel = INT MAX ; _
int mergeCnt = 0;
23
24 char filelID[] = "P13Mrgl.bin "
;
25 char file2ID[] = "P 13 Mrg2.bin " ;
26 char fileOutID[ ] = "P13Mrg3.bin"
;
27
28 // Statements
29 printf("Begin File Merge:\
n" );
30 if (!( spMl = fopen (filellD, READMOD )))
E
contimid
PROGRAM 1 3 - 1 1 Merge Two Files ( continued )

31 printf( " \aError on %s\n " , filellD ), exit ( 100);
32
33 if (!( spM 2 = fopen ( file2ID, READ MODE ))) _
34 printf( "\aError on %s\n" r file2ID), exit (200);
35
36 if (!(spOut = fopen ( fileOutID , WRITE MODE ))) _
37 printf( "\aError on %s\n", fileOutID), exit ( 300);
38
39 fread ( fcrecMl , sizeof( int), 1 , spMl );
40 if ( feof(spMl ))
41 recMl = sentinel ;
42 fread (&recM 2, sizeof(int), 1 , spM2);
43 if ( feof( spM 2))
44 recM 2 = sentinel ;
45
46 while ( Ifeof(spMl ) !feof( spM 2 ))
47 {
48 if ( recMl <= recM 2 )
49 {
50 fwrite ( &recMl , sizeof( int), 1 , spOut);
51 mergeCnt++ ;
52 fread (&recMl , sizeof( int ), 1, spMl);
53 if ( feof(spMl ))
54 recMl = sentinel;
55 > // if
56 else
57 {
58 fwrite (&recM 2 , sizeof( int), 1 , spOut );
59 mergeCnt++;
60 fread (&recM 2 , sizeof( int ), 1 , spM 2 );
61 if (feof( spM 2 ))
62 recM 2 = sentinel ;
63 > // else
64 } // while
65 fclose ( spMl );
66 fclose ( spM 2);
67 fclose (spOut );
68 printf( "End File Merge. %d items merged.Nn " ,
69 mergeCnt);
70 return 0;
71 > // main
Program 1 3 - 1 1 Analysis We have written this simple program without any

subfunctions. Once again we have
used the multiple - statement error message format for the error message and exit after
each open statement.
858 Section 13.4 Pile Program Examples
Programs that involve the comparison of data from two files require that the first
record from both files be read before any comparisons can be made. This is sometimes
called "priming the files." The reads that prime the files are coded before the main
while loop. Since duplicate read statements are required in the loop, you might wonder
why we didn't use a do...while. The reason is that the program would fail if both files
were empty. As we coded it, the program works if either or both merge files are
empty.
Study the logic at statements 54 and 62 carefully. These statements implement the
sentinel concept. When a file reaches its end, we set the key in the record area for the
file to the sentinel value. This ensures that all the data on the other file will compare low
and be written to the output file.
The most difficult statement in this simple program is the while statement. We need
to keep looping as long as either of the files is not at the end of file. The most straight-
forward code is as we coded it in statement 46. Another common way to code it is
shown below.
(!(feof(spMl ) & & feof( spM2)))
Use De Morgan's ( see Chapter 5) rule to prove to yourself that these two statements
are identical.
Chapter 13 Binary Input/Output J159

Any file environment requires some means of keeping the file current. Data
are not static ; they are constantly changing, and these changes need to be
reflected in their files. I he function that keeps files current is known as
updating. Io complete our discussion of files , we discuss some of the soft -
ware engineering design considerations for file updating. For this discussion ,
we assume a student binary file similar to the ones we have discussed in the
chapter.
Update Files
I hree specific files arc associated with an update program . First is the perma -
nent data file , called the master file. The master file contains the most cur-
rent computer data for an application.
I he second file is the transaction file . It contains changes to be applied
to the master file. I hree basic types of changes occur in all file updates ; oth -
ers may also occur, depending on the application . Add transactions contain
data about new students to be added in the master file ; delete transactions
identify students who will he deleted from the file; and change transactions
contain revisions to specific student records in the file .
To process any of these transactions , we need a key. A key is one or more
fields that uniquely identify the data in the file. For example, in the student
file, the key would he student ID. In an employee file , the key would be Social
Security number.
The third file needed in an update program is an error report file. An
update process is very rarely error free. When an error occurs, we need to
report it to the user. The error report contains a listing of all errors discovered
during the update process and is presented to the user for corrective action .
File updates are of two types: hatch and online. In a batch update,
changes are collected over time and then all changes are applied to the file at
once. In an online update , the user is directly connected to the computer
—
and the changes are processed one at a time often as the change occurs.
Sequential File Update

For our discussion, we will assume a hatch , sequential file environment
.A
file is a file that must he processed serially starting at the begin -
sequential
ning. It does not have any random processing
capabilities. The sequential
attribute that it is ordered on the key.
master file has the additional
file update actually has two copies of the master file , the old
A sequential
master and the new master . This is because whenever a sequential file is
changed, it must be entirely re -created ven if only one student ’s score on
one exam is changed .

for a sequential file update.
Figure 13- 17 contains an environment chart
In this chart , we see the four files we discussed
above . We use the tape symbol
classic symbol for sequential files. Sequential files
for the files because it is the
could just as easily be stored on a disk. Note that after the update program
completes, the new master file is sent to off - line storage, where it is kept
until
it is needed again. When the file is to he updated , the master file is
retrieved
from the off - line storage and used as the old master.
Generally, at least three copies of a master file are retained in off line
storage , in case it becomes necessary to regenerate an unreadable file.
-
This
retention cycle is known as the grandparent system because three genera
tions of the file are always available: the grandparent , the parent ,
-
and
the child .
transactions old master
\ Sequential File Update
transaction record old master record
off-line
storage
imjn
new master record
error
report
new master
FIGURE 13 - 17 Sequential File Update Environment
The Update Program Design

SC Cntlst named Barry Dwyer publishe an update algo-
runm
rithm in r '
. d
. ...
-
in the Commun tcalions of the ACM that was so elegant that it Has
-
—
me * C SiC 3 w »» a Pttd I algorithm fo, our Wo , ,
° '“ ' *
Since a sequential master file is ordered on a
^ «.
S recIu , res that we match
1 b ... .53riri
the keys on the transaction and
master file and , assuming that there are no
errors, take one of the following
three actions:
1. II the transaction file key is less
than the master file key, add the trans-
action to the new master.
3* Barfy Dwyer> “One More Time How

24, no. 1 (January 1981 ): 3-8.
— to Update a Master File,” Communications of the ACM -
2. If the transaction file key is equal to the master file key, either
a. C hange the contents of the master file data if the transaction is a
revise transaction , or
b. Remove the data from the master file if the transaction is a delete .
3. If the transaction file key is greater than the master file key, write the old
master file record to the new master file.
This updating process is shown in Figure 13- 18 . In the transaction file ,

the transaction codes are A for add , D for delete, and R for revise. The process
begins by matching the keys for the first record on each file, in this case,
14 > 10
Thus, Rule 3 is used , and we write the master record to the new master
record . We then match 14 and 13, which results in 13 being written to the
new master. In the next match , we have
==
i
14 14
Thus, according to Rule 2a , we use the data in the transaction file to change
the data in the master file. However, we do not write the new master file at
this time. More transactions may match the master file, and we need to pro-
cess them , too.
'
\ A 31
R 25 21
A 23 20
D 21 16
Transaction Old
File A 18 14 Master File
A 17 13
R 14 10
New Master File
FIGURE 13 - 18 File Updating Example

After writing 16 to the new master, we have the following situation:
17 < 20
According to Rule 1 , we must add 1 7 to the new master file. We do this

by
copying the transaction to the new master file, hut again, we don ’t
write it yet.
This newly added record may have some revision transactions, and we
need to
be able to process them. For example, this capability is needed when a
new
student registers and adds classes on the same day. The computer has to
be
able to add the new student and then process the class registrations in
the
.
same hatch run We write the new master for 1 7 when we read
transaction 18.
The processing continues until we read the delete transaction, at which
time we have the following situation:
2 1 == 2 1
and since the transaction is a delete, according to Rule 2b, we need to drop
2 1 from the new master file. To do this, we simply read the next
master record
and transaction record without writing the new master. The
processing con-
tinues in a similar fashion until all records on both files have been
processed.
Update Errors
Two general classes of errors can occur in an update program. The user can
submit bad data, such as a grade that is not A, B, C, D, F. For our discus-
or
sion, \ \ e v \ i assume that no
.
data errors are present Detecting data errors is
the subject of data validation and has been
discussed.
I lie second class ol errors is file errors. File
errors occur when the data
on
n ,- lransiac ‘
' t on arc n t in synchronization with the data on the master
h e 1 hree different situations can °
occur:
.
on ansactIon matches a key on the master file, therefore, we
reject
reject the* transact
rhr f ,
on as invalid and report it on the error report .
• A rev se transaction's key does not
match a record on the master file. In
case, we are trvin u ,-
tJlat d<) not ex st This is also a file
error and must be renf °
» I on the
1 led error report .
3
n match a record on the master file In .
^^ ' '
this ase we are trvin8
be reported as an error.
() IT °‘
e etc data that do not exist, and this must also
Update Structure Chart

I he structure chart lor the
sequential file update is shown in Figure 13 19*
In this structure chart, process
contains the updating function.
Sequential
Update
Initialization Process End of Job
Not Sentinel
£
Read Update
Write
Master NewMaster
Key Change
Read
Add (+) Change (+) Delete Transaction
FIGURE 13 -19 Update Structure Chart
Update Logic
I n i t i a l i z a t i o n is a function that opens the files and otherwise prepares
the environment for processing. Endof Job is a function that closes the files
and displays any end of job messages. The mainline processing is done in
P r o c e s s.
Although it is beyond our scope to develop the complete set of update
functions, it is important that you at least understand the mainline logic
found in P r o c e s s. Its pseudocode is shown in .Algorithm 13- 2 .
ALGORITHM 13 - 2 Pseudocode for File Update

Algorithm Sequential Update
1 read f i r s t record from transaction f i l e
2 read f i r s t record from o l d master f i l e
3 s e l e c t next e n t i t y t o be processed
4 l o o p current e n t i t y not s e n t i n e l
1 i f current e n t i t y equals old
master e n t i t y
1 copy old master t o new master work area
2 read old master f i l e
2 end i f
entity
3 i f current e n t i t y equals transaction
1 update new master work area
4 end i f
master e n t i t y
5 i f current e n t i t y e q u a l s new
1 write new master f i l e
continued
ALGORITHM 1 3 - 2 Pseudocode for File Update

6 end if
7 select next entity to be processed
5 end loop
End Sequential Update
Let ’s look at the update logic in a little more detail . The first three state-
.
ments contain initialization logic for Process The driving force behind the
update logic is that in each while loop, we process all the data for one student
.
( entity) To determine which student we need to process next (statements 3
and 4.7 ) , we determine the current entry by comparing the current transaction
key to the current master key. The current key is the smaller .
Before we can compare the keys, however, we must read the first record in
.
each ( lie This is known as " priming the files” and is seen in statements 1 and 2.
The loop statement in Algorithm 13- 2 contains the driving logic for the
entire program . It is built on a very simple principle: As long as data are
present in either the transaction file or the master file, we continue to loop.
When a file has been completely read , we set its key to a sentinel value.
W hen both files are at their end , therefore, both of their keys will be senti-
.
nels I hen , when we select the next student to be processed , it will be a sen-
tinel , which is the event that terminates the while loop.
1 hree major processing functions take place in the while loop. First we
determine if the student on the old master file needs to be processed. If it
does, we move it to the new master output area and read the next student
f rom the old master file. The key on the old master can match the current key
in two situations: a change or delete transaction exists for the current stu -
dent . This logic is seen in statement 4.1 .
1 he second major process handles transactions that match the current
student . It calls a function that determines the type of transaction being pro-
cessed ( add , change, or delete ) and handles it accordingly. If it is an add. it
moves the new student’s data to the new master area. If it is a change, it
updates the data in the new master area . And if it is a delete, it clears the key
in tlu* new master area so that the record will not be written . To handle mul -
tiple transactions in the update function , it reads the next transaction and
continues if its key matches the current student.
I he last major process writes the new master when appropriate. If die
current student matches the key in the new master file area , then the record
needs to be written to the file . This will he the case unless a delete transaction
was processed .
Summary
In this section we looked at a very important algorithm , the classic sequ tial
file update, and discussed its mainline logic flow.
The elegance of Dwyer’s algorithm , as seen in Algorithm 13- 2, ics
determination of the current student and the separation of the up a e
into three distinct functions: read the old master, update the current student ,
and write the new master. Study the concept of the current student carefully,
and make sure you understand how it controls the three major processes in
the loop. Then , with a little thought , you should be able to develop the other
( unctions in the update program .
Also note that Algorithm 13- 2 contains only one read transaction func -
tion . Its function is to read a valid transaction . Therefore , it contains all of
the simple data validation logic to determine if the data are correct. It does
not perform any file errors ; they are handled in the update function . If any
errors are found , the read transaction function writes the transaction to an
error report and reads the next transaction . When it returns to the calling
function , it has either read a valid transaction or has found the end of the file.
866 Section 13.6 Tips ond Common Programming Errors

Refer the “Tips and Common Programming Errors ” section in Chapter 7
to
.
Many of those tips apply to binary files as well .
1 . EOF is type integer, and its value is normally - 1 . Therefore , if
you want to
test the value of a variable to EOF , use an integer variable , not a
character.
In most systems , a character variable cannot store a negative
number
(- 1 here ) .
2 . Remember to open a file before using it .
3 . You can create a file for writing; but to read a file , it must exist .
4 . When you open a file for writing using w mode , you must close
it and
open it for reading ( r mode ) if you want to read from it . To avoid this
problem , you can open it in w + mode.
5 . An open file can be in one of the three states : read , write , or error. If you
want to switch from read to write or from write to read , you must use one
of the file - positioning functions .
6 . Do not open a file in w mode when you want to preserve the contents of
the file . Opening a file in w mode erases the contents of the file .
7 . Remember that in general you cannot print the
contents ol a binary' file.
It must he converted to a text file first .
8 . Unlike other input/output functions, the first paramet of the fread and
er
fwrite functions is a pointer to the input area , not a file pointer. The file
parameter is the last ( fourth ) parameter.
9 . The second parameter of the fread and
fwrite functions is the size of the
element , and the third parameter is the count of elements.
10 . The fread and fwrite functions return the number of elements read or
written , not the number of bytes read or written .
11 . Remember that feof does not look ahead
. It returns true only if
attempt is made to read the end of file .
12 . I he second parameter in the fseek function
is the number ol bytes , not
the number of elements .
pdrdmctcr m the fseek function is SEEK _END ( 2 ) , the
middle ,,
°U norma y he a negative long integer to access a
14
byte
V in
!"
C S
-' “
Bl' \ff -
the fil
‘
^ . , ,,
, . lo add an element at“ the it i referring o a hvtc „f e h end with
end of the file , you use the fseek lunction
7 .
the second parameter set to zero and the third
parameter to the value o
SEEK END .
5 Remember that every time U USC
*
fread or fwrite function , it auto *
matically advances the fil K°)sltlon

°
.^ Indicat or toward the end of the file
the number of bvtes Pn . 1 t0 the
y equa *Slze of the element times the numberof
elements read or written . '
16 . It is good practice to close
all files before terminating a program .
Chapter 13 Binary Input/Output 86 /
17 . It is a compile error to reter to a file with its external file name rather
than its file pointer.
18. It is a compile error to omit the file pointer when using a file function .
I his error most often occurs when using the file versions of the file for-
mat functions, such as fprintf .
1 9 . It is a logic error to refer to a file before it is opened .
20. It is a logic error to open a file for reading when it doesn ’t exist . This is
usually an error in the external file name in the open statement .
2 1 . It is a logic error to attempt to read from a file in the write state and vice
versa. This is true even when the file has been opened in the update mode.
22. Opening an output disk file will fail if the disk is full .
23. Opening an existing file in write mode deletes the file. If your input file
disappears , check your open modes.
24 . It is a logic error to usefseek to place the file marker before the first byte
of a file.
13.7 Key Terms

batch update online update
end of file read state
error report file seek file
error state sequential file
kev transaction file
master file update mode
merge write state
13.8 Summary
A file is a collection of related data stored in an auxiliary storage device.
A stored stream of Os and Is can he interpreted as either a text or a
binary file.
A text file is a file of characters.
A binary file is a collection of related data stored in the internal format of
the computer.
(ile is always in one of the following states: read , write , or error.
^
A binary file can be opened in one of the following modes:
rb, r + b, wb ,
w + b , ab, and a + b .
J The fread function reads the number of bytes specified
by the product of
the element size and the number of elements .
The fwrite function writes the number of bytes specified by the product
of
the element size and the number of elements.
The feof function checks for end of file.
The /error function is used to check the error status of a file.
The clearerr function is used to clear an error.
The rewind function sets the file position indicator to the beginnin
g of
the file.
I he /tel / function tells you the current position of the file position indicator
.
Thefseek function positions the file position indicator to the beginning of
a specified byte.
The remove function removes or deletes a file from the disk.
The rename function renames a file on the disk.
The tmpfile function creates a new temporary file .
Io be kept current , master files must be updated .
The master file update requires four files: the old master, a new master, a
transaction file, and an error report file.
J 1 he basic master file transactions are add , revise , and delete.
13.9 Practice Sets
Review Questions
1. A file can be read only if it is in the read
state.
a . True
b. False
2. A file opened in the r + b mode is a
binary file opened in the read state and
update mode.
a. True
b. False
3. Binary files are read using
the block input/output function fread .
a . True
b. False
4. Using fseek to position a file
beyond the current end of file places the
in an error state .
a . True
b. False
5 . Io merge two files, you can use the merge function found in the standard
input /output library.
a. True
b. False
.
6 Which of the following statements about files is true?
a. All files must be opened before they can be used.
b. Binary files arc more portable than text files.
c. Binary files are slower than text files.
d. Files must be closed.
e. Text fdes contain data stored in the internal format of the computer.
7. Which of the following statements about files is false?
a. Because they are more flexible, binary files are more portable.
b.Binary' files contain data stored in the internal format of the computer.
.
c Binary files usually contain records in the form of structured data.
d. Text files store only character data.
e. Some text file data need to be converted to internal storage formats for
processing.
.
8 The results when a failure occurs during an open or
during either a read or write operation.
.
a error state
b. fail state
c. read state
d. update state
e. write state
9. If a file in the read state is written to, then the following occurs:
a. The file is placed in an error state regardless of the file mode.
b. The file is placed in an error state unless it was opened in the update
write mode (w+b).
The file state is automatically switched if the file is opened for updating.
c.
-
d. The program is aborted regardless of how the file was opened.
e. The program is aborted unless the file was opened for updating.
10. Which of the following file modes opens a file in the write state for
updating?
a. ab
b. a+b
.
c rb
d. r +b
.
e wb
1 1. Which of the following C functions is used to output data to a binary file ?

a. fwrite
b. output
c . write
d . writef
e . fprint/
12 . Which of the following is a file status function?
a . feof
b . /report
c. / rewind
d. / seek
e . / tell
13 . The function may he used to position a file at the
beginning .
a . /close
h . / eo/
c . / report
d . /seek
e. / tell
14 . A is one or more fields that uniquely identify the
data in a file.
a . field
b . identifier
c . key
d . structure
e . transaction
1 3. \ \ hich ol the following statements about sequent
ial file updating is falser
a . Add , change , and delete transactions are used to update the file,
h . Sequential files are often updated in an online environm .
ent
c . Sequential files contain structures (records )
with a key to identif'
the data .
d . 1 he file must he processed starting at the beginnin .
g
e . The file must he entirely re -
created when it is updated .
Exercises
16 . 1 xplain the difference between the
following pairs of modes:
rb and r + b
wb and w + b
ab and a + b
.
17 Find the error( s ) in the followin
been properly declared . )
_
g code . ( Assume the PAY REC lyPe
char*
char* str = "Payroll ” ;
_
PAY REC payRec ;
FILE* sp;
sp = fopen (str , m );
fread( payRec, sizeof( payRec ), 1 ,
18. Given the following declarations and assuming that the file is open:
FILE * sp;
char s[ 20];
find any errors in each of the following statements:

a. fread( s , 20 , sp);
b. fread( s , 20, 1 sp);
/
c. fread( s , 1, 20 , sp);
d . fread(sp, 1 , 20, s);
e. fread( sp, 20, 1 , s);
19. Given the following declarations and assuming that the file is open:
FILE* sp;
char s[ 20];
find any errors in each of the following lines:

a. locn = ftell(sp);
b . locn = ftell( l , sp );
c. fseek(0, 20L, sp);
d . fseek(sp, 20L , 0 );
.
e fseek( sp, 20L, 1 );
20. What would he printed from the following program ? Draw a picture of
the file with the file marker to explain your answer.
# include <stdio.h>
int main (void )
{
long int pos;
FILE* sp;
sp = fopen( "SAMPLE.DAT" , "w+b " );

for (char c = ' A '; c <= E' ; C++ )
fwrite(&c , sizeof(char ), 1 , sp);
pos = ftell (sp);
printf( "The position of the file marker is : %ld " ,
pos);
return 0;
} // main
21 . What would be printed from the following program? Draw a picture

of
# include <stdio.h>
int main ( void )
{
char c;
FILE* sp ;

for (c = ' A ' ; c <= 'E' ; C++)
fwrite(&c , sizeof( char ), 1 , sp );
fseek(sp, 2, 0);
fread(&c , 1, 1 , sp );
printf( " \n\n%c " , c );
return 0;
} // main
22 . What would be printed from the following program? Draw a picture of

# include <stdio.h>
int main ( void )
{
char c ;
FILE* sp;
sp = fopen( "SAMPLE.DAT" , Mw+b" );

for (c = ' A ' ; c <= 'E'; C++ )
fwrite(&c , sizeof (char), 1, sp);
rewind(sp );
fread( &c , 1 , 1 / sp );
printf( ,,\n\n%c" , c);
return 0 ;
} // main
23 . What would be printed from

the following program? Draw a picture ot
# include <stdio.h>
int main ( void )
{
char c ;
long int pos;
FILE* sp ;
sp = fopen( "SAMPLE.DAT" ,
Mw+b");
for (c = ' A * ; c <
= E'; C++)
fwrite(&c , sizeof(char ), 1 , sp);
continue
pos = ftell (sp );

pos— ;
POS ;
—
fseek(sp, pos, 0);
fread(&c, 1, 1 sp); /
printf( "\n\n%c", c );
return 0;
> // main
24. What would be printed from the following program? Draw a picture of
iinclude <stdio.h >

int main ( void )
{
char c ;
long int pos;
FILE* sp;
sp = fopen( "SAMPLE.DAT", "w+b " );

for (c = 'A' ; c <= 'E'; C++ )
fwrite(&c , sizeof(char), 1 , sp);
pos = ftell(sp);
——
pos ;
pos ;
fseek(sp, pos, 1); -
fread(&c, 1 , 1 , sp );
printf( "\n\n %c" , c );
return 0;
} // main
.
25 What would be printed from the following program ? Draw a picture of
# include <stdio.h>
int main ( void )
{
char d;
long int pos;
FILE* sp;
sp = fopen( "SAMPLE.DAT" , "w+b");

for (char c = 'A ; c <= E'; C++)
1
fwrite(&c , sizeof(char), 1, sp );
pos = ftell(sp);
——
pos ;
pos ;
fseek(sp, 2 , 2 );
continued
fseek( sp , 1 , 1 );
fread( &d , 1 , 1 , sp );
printf( "\n\n%c " , d);
return 0 ;
} // main
26 . What would be printed from the following program ? Draw a picture of


int main ( void)
{
long int pos ;
FILE* sp;

for ( int i = 1 ; i < = 5; i++)
fwrite( & i , sizeof( int ), 1 , sp );
pos = ftell(sp);
printf( "The position of the file marker is : %ld"
pos );
return 0;
} // main
27. What would be printed from the following program ? Draw a picture ot
# include <stdio.h>
int main ( void )
{
int i;
FILE* sp;
sp =
fopen( " SAMPLE.DAT" , " w+b " );
for ( i = 1; i < = 5 ; i++ )
fwrite(& i , sizeof(int), 1, sp);
fseek( sp , sizeof ( int ) * 2 , 0 );
fread(&i, sizeof (int), 1, s p ) ;
printf( " %d \n" , i );
return 0;
> // main
.
28 \ \ hat would be printed from
the following program ? Draw a picture o!
the file with the file marker to explain your
answer.
int main (void )
{
contimid
int i;
FILE* sp;
sp = fopen ( "SAMPLE.DAT", " w+b" );

for ( i = 1; i <= 5; i++ )
fwrite(&i, sizeof(int), 1, s p ) ;
-
fseek(sp, sizeof(int) * 2, 1);
fread( &i, sizeof( int), 1 , sp );
printf("%d ", i);
fclose (sp);
return 0;
> // main
29. What would be printed from the following program ? Draw a picture of
the file with the file marker to explain your answer. Is the answer
strange? Why?
tinclude <stdio.h>
int main ( void )
{
int i;
FILE* sp;
sp = fopen( "SAMPLE.DAT" , "w+b" );

for ( i = 1; i <= 5; i++)
fwrite(&i, sizeof(int), 1, sp );
fseek(sp, 7, 0);
fread(&i, sizeof(int ), 1, sp);
printf( "%d ", i);
fclose (sp);
return 0;
> // main
30. What would be printed from the following program ? Draw a picture of
#include <stdio.h>
int main (void)
{
int i;
long int pos ;
FILE* sp;
sp = fopen("SAMPLE.DAT", " w+b");

for (i = 1; i <= 5; i++)
fwrite(&i, sizeof(int), 1 , sp);
_
pos = ftell(sp);
pos = 2 * sizeof( int );
continued
8 / 6 Section 13.9 Practice Sets
fseek( sp , pos, 0 );
fread(&i, sizeof(int), 1, sp);
printf( "%d " , i);
fclose (sp);
return 0;
} // main
Problems
.
31 Write a function that copies the contents of a binary file of integers toa
second file. The function must accept two file pointers and return an
integer ( zero representing a processing error and nonzero indicating suc-
cessful completion ).
32 . Write a function that prints a specified number of records from the
.
beginning of a file The function is to accept two parameters. The first is
a pointer to a binary file of structure type STR. The second is an integer
that specifies the number of records to he printed ( inclusive ). The struc-
ture type is shown below:
typedef struct
{
int 1?
float f ;
> STR ;
II any errors occur, such as fewer records in the file than specified , it
should return zero. Otherwise, it returns nonzero.
33. Write a lunction that compares two files and returns equal or not equal
based on the result of the comparison . The functions should receive file
pointers to two opened files and compare them byte by byte.
34. Write a function that returns the number of
items in a binary file.
35. Write a function that prints the
last integer in a binary file of integers.
36 . Write a function that physically removes all items with a specified value
(data) f r o m a binary file of structure STR. You may use a temporary file-
I IK file may contain more than one
record with the delete value. The ke>
value removed is to be entered from the keyboard .
typedef struct
{
int data;
char c;
> STR ;
37. Write a function that appends anothe

one binary file at the end of
the structure described in Problem 36.
.
38 Write a function that, given a binary file, copies the odd items ( items 1,
3, 5, n) to a second binary file and the even items (items 2, 4, 6, n)
to a third binary file. After all items have been copied, print the contents
of both output files.
.
39 Write a function that reads items from a binary file and copies them to a
dynamically allocated array. I he function must first find the size of the
hinarv file to allocate the array.
.
40 Write a function that takes a binary file ol long integers and appends a new
long integer at the end that is the sum of all integers in the original file .
Projects
. .
41 A company has two small warehouses The list of the products in each
warehouse is kept in a text file(invFilel and InvFile2), with each line
representing information about one product. The manager w'ants to have
only one list showing information about all products in both warehouses.
Therefore, the two text files must he combined into one single text file
(OutFile).
Write a program that will copy information from the two text files
(InvFilel and InvFile2)to two binary files(BinFilel and BinFile2).
After creating the binary files, the program is to merge the two binary
files to produce a combined binary file. After the combined binary file has
been created, create a report file that can he printed. The report file is to
contain page headers with an appropriate title and page numbers. The
structure for the files is shown below.
typedef struct inv rec _

{
char partNo[ 5];
char partName[ 15];
qtyOnHand ;
int
_
> INV REC;
42 A company keeps a list of parts that it purchases, with a line
. of informa-
the ’s unique code, name, and three
tion for each part that gives part
codes for three suppliers that supply that particular part. This list is kept
to the sup-
in a binary file and is sorted in ascending order according
plier ’s code.
The company also keeps a list of its suppliers, with a line of informa-
tion for each supplier, w hich gives the supplier’s unique
code, name, and
sorted in ascending
address. This list is also kept in a binary file, w hich is
order according to the supplier ’s code .
Write a program that enables the user to enter a part’s unique code
and to receive a list of three suppliers. If the code is found the program
,
prints the names and addresses of the three suppliers . If the code is not
prints message to tell the user that the code is not in the file.
found , it a
the option to quit.
After each inquiry, the program gives the user
Each record in the part file is made up of a part ’s code, name,

and
the codes for three suppliers. The part ’s code is an integer ; the
nameisa
string with a maximum length of ten characters; and each suppliers
code
is an integer. Note that not all parts have three suppliers. For
parts with
less than three suppliers , a special supplier code of 0000 is used to
indi-
cate no supplier.
Each record in the supplier file is made up of a supplier's code,
name, and address. The supplier’s code is an integer, the name is a string
with a maximum length of ten characters, and the address has a
maxi-
mum length of 20 characters .
The output is to he formatted with the first line showing the data for
the part and the following lines showing data for the suppliers, indented
one tab .
Sample data for the files are shown in Tables 13- 2 and 13- 3. You will
first need to write a file conversion program to create the binary files. We
suggest that you create a text file version of each file with your text editor
and then read it to create the binary version .
Port Code Port Nome Supplier 1 Supplier 2 Supplier 3

1000 Pen 5010 5007 5012
1001 Pencil 5006 5008 0000
1002 Paper 5001 5000 5003
1003 Ball Pen 5013 5009 5014
1004 Folder 5009 5007 5002
1005 Pointer 5012 5006 5005
1006 Mouse 5012 0000 0000
1007 Monitor 5000 5002 5007
TABLE 13- 2 Project 42 Part File
Supplier Supplier Name Supplier Address

Code
5000 John Marcus 2322 Glen Place
5001 Steve Chu 1435 Main Ave.
5002 David White 2345 Steve Drive
5003 Bryan Walljasper 780 Rose Mary Street
TABLE 13 -3 Project 42 Supplier File
continued
Supplier Supplier Name Supplier Address

Code
5004 Andrew Tse P. O. Box 7600
5005 Joanne Brown 1411 Donnybrook Square
5006 Lucy Nguyen 2345 Saint Mary Road
5007 Fred West 1 1 Duarte Rd.
5008 Dennis Andrews 14 California Ave.
5009 Leo Washington 134234 San Rosa Place
5010 Frankie South 12234 North Justin St.
5011 Amanda Trapp 1345 South Bush Circle
5012 Dave Lightfoot 222 George Territory Drive
5013 Danna Mayor 1 1 George Bush Street
5014 Robert Hurley 14 Republican Alley
TABLE 1 3 - 3 Project 42 Supplier File ( continued )
To read a record on the Ides, you will need to determine its position
on the file. This is easily done by subtracting 1000 from the PartCode
for the part file and 5000 from the SupplierCode for the supplier file.
Then use /seek to position the file for reading.
43. Write a program that builds a histogram ( see Chapter 8 ) using a ran-
domly accessed disk file. The program is to begin by creating a new file of
20 integers. Each integer will represent the accumulator for its relative
position; the first for the number 0, the second for the number 1, and so
forth until the last for the number 19. The program is to then use the ran-
dom number generator to create 100 random numbers in the range
0 . .. 19. As each random number is generated, it will be displayed in a 10 10
x
matrix ( 10 lines of 10 numbers each ) and 1 added to its accumulator on
the disk using the fseeli function . After the random numbers have been
generated, the file is to be read sequentially and a histogram displayed.
that allows you to check
44. Your stockbroker has an online inquiry system
the price of stocks using your personal computer . Simulate this system as
described .
below
Each stock is assigned a unique integral number in the range
that stock 1000 is stored in
1000... 5000. They are stored on the disk so
location 0. stock 1001 in location 1 , stock 2010 in location 1010, and so
the disk address for a requested stock, your program
forth. To calculate
1000 from the stock number and uses the result as the address
subtracts
in the file. (This is actually a simplified version of a

concept known as
“ hashing,” which you will learn when you study data structures.)
The data for each stock are described as follows:
structure
stock key short integer
stock name string[21]
stock symbol string!6 ]
current price
YTD High
-
floating point number
YTD Low
-
-
-
Price Earning Ratio short integer
( YTD: Year to Date)
Using data from your local newspaper, create a binary file of at

least
20 stocks. Then write a menu -driven system
that allows the user to
request data on any individual stock.
In addition , provide a capability to get a report of up to 20 stocks
at
one time. When this option is requested , open a temporary work ,
file and
write the requested stocks to the file. After the
last stock has been
entered , read the file ( without closing it ) , and prepare the
report .
Bilwise Operators •• to
\ \ hile the preceding chapters contain the material required for a traditional
first course in programming, we have not discussed all the basic capabilities
of the C language . This chapter discusses the C bitwise operators.
I he C language is well suited to system programming because it contains
operators that can manipulate data at the bit level. For example, the Internet
requires that hits be manipulated to create addresses for subnets.
C has two categories of bitwise operators that operate on data at the bit
level : logical bitwise operators and shift bitwise operators. Before we discuss
these operators, however, we need to introduce exact -size integer types.
To demonstrate these concepts we develop three programs. The first dis -
cusses a checksum calculation and develops an example that could be used to
validate Internet transmissions. The second uses masks to determine the first
or the last address in a block of Internet addresses. The third demonstrates a
hit manipulation technique that can be used in polynomial calculations.
In Software Engineering, we present a case study to demonstrate an
approach to designing structure charts.
Objectives
IJ To be able to use the bitwise logical operators in programs
To be able to use the bitwise shift operators in programs
To understand how to create and use masks to manipulate bits
882 Section 14.2 Logical Bitwise Operators
14.1 Exact Size Integer Types

The integer types, such as int and long , discussed in Chapter 2 , are machine
dependent. In one computer, the size of int may be four bytes; in another
computer it may he two bytes. While many bitwise applications work well on
machine- dependent integer types, other applications need to assure that the
size is fixed . For example, to manipulate an Internet address, we need to
define an unsigned integer of size 32 hits ( 4 bytes ). Beginning with C99, C
allows us to define integer types of sizes 8, 16, 32 , and 64 hits. They arc
defined in the stdint . h header file. Table 1 4 - 1 documents these types. As you
will see, most of the time we use unsigned integers.
Type Description
int 8 t 8-bit signed integer
int!6 t 16-bit signed integer
Signed
uint 8 t 8-bit unsigned integer
uint!6 t 16-bit unsigned integer
Unsigned
TABLE 1 4 - 1 Fixed - size Integer Types
14.2 Logical Bitwise Operators

,
indivic ual bits be manipulated. Four
^^ T
^ ^ndi' hits
ta 3S to
operators are bitwise “ nd ( ), bitwise inclusiveof
( I ), bitwise exclusive or '‘pUjate es comPlement
:
*
(-) The first three are
’ " °" •
binary> operators the,. one s complement
perators, th is a unary operator.
Bitwise and Operator

The bitwise and ( & —precedence 8 ) iis a binary operator that requires two
integral operands (charict • ,•nte8er)* It does a bit-by-bit comparison
between the two operand * e rcsi,b of the comparison is 1 only when
both hits are I ; it ,s n .
e wise. fable 14- 2 shows the result of bit-
*
by-bit
comparison. *
I
First Operand Bit Second Operand Bit Result

0 0 0
0 1 0
1 0 0
1 1 1
TABLE 14 - 2 And Truth Table
Program 14 - 1 demonstrates the bitwise and operator.
PROGRAM 14 - 1 Simple Bitwise And Demonstration

1 /* Demonstrate bitwise AND operator
2 Written by:
3 Date:
4 */
7 tinclude <stdint.h>
8
9 int main ( void )
10 {
_
12
13
_
uintl6 t numl = 0x0257;
uintl6 t num2 = 0xA463;
14 uintl6 t res;
15
16 // Statements
17 res = numl & num 2 ;
18
19 printf ( " Input and results in hexadecimal:\n " );
20 printf ( " numl: % # 06X \n" , numl );
21 printf ( " num 2: % #06 X \n " , num2 );
22 printf ( "result: % # 06 X \n "/ res);
23
24 return 0;
25 > // main
Results:
Input and results in hexadecimal:
numl: 0X0257
num2: 0XA463
result: 0X0043
Program 14 - 1 Analysis The result of an and operation is logically the intersection of the two values. We con
veriify the result by manually calculating it using binary values.
Bitwise Inclusive or Operator

The bitwise inclusive or (|—precedence 6 ) is a binary operator that requires
two integral operands (character or integer ) . It docs a bit - by-bit comparison
between the two operands. The result of the comparison is 0 if both operands
are 0; it is 1 otherwise. Table 14- 3 shows the result of bit-by-bit comparison
.
First Bit Second Bit Result

0 0 0
0 1 1
1 0 1
1 1 1
TABLE 14 - 3 Inclusive Or Truth Table
Program 14 - 2 demonstrates the basic operation of the inclusive or.
PROGRAM 14 - 2 Simple Inclusive or Demonstration

1 /* Demonstrate the inclusive OR operator
2 Written by:
3 Date:
4 */
5 Jfinclude <stdio.h>
7 # include <stdint.h >
8
9 int main ( void )
10 {
12 _
13 _
uintl6 t num2 = 0xA463 ;
14 uintl6 t res;
15
16 // Statements
17 res = numl|num2;
18 printf ( " Input and results in hexadecimal:\n" );
19 printf ( " numl: % #06X \n" , numl);
20 printf ( " num2: % #06 X\n " , num2 );
21 printf ( "res: % # 06X \n" , res );
coittin0
PROGRAM 14 - 2 Simple Inclusive or Demonstration (continued )

22
23 return 0;
24 } // main
Results:
numl: 0X 0257
num2: 0XA463
res: 0XA677
Bitwise Exclusive or Operator

I he bitwise exclusive or ( —precedence 7 ) is a binary operator that requires
A
two integral .
operands It does a bit -by-bit comparison between the two oper -
ands. I he result ol the comparison is 1 only if one of the operands is 1 and
the other is 0; it is 0 if both operands’ hits are 0 or 1 — that is, if they are both
the same. Table 14 - 4 shows the result of bit-by-bit comparison.
First Bit Second Bit Result

0 0 0
0 1 1
1 0 1
1 1 0
TABLE 14 - 4 Exclusive Or Truth Table
Program 14 - 4 demonstrates the basic operation of the bitwise exclusive or.
PROGRAM 14 - 3 Simple Exclusive or Demonstration

1 /* This program demonstrates the use of the
exclusive
or
2 Written by:
3 Date:
4 */
8
9 int main ( void )
10 {
continued
PROGRAM 14- 3 Simple Exclusive or Demonstration ( continued )

1 1 //
12
_
Local Declarations
13 _
uintl6 t num2 = 0xA463;
14 uint
! 6 t res;
15
16 // Statements
17 res = numl A
num2;
18
19 // Print results in hexadecimal
20 printf ( "Input and results in hexadecimal:\n");
21 printf (" numl: %#06X\n", numl);
22 printf ( "num2: % # 06 X\n" , num2);
23 printf ( " res: % #06 X\n" , res);
24
25 return 0;
26 > // main
Results:
numl: 0X0257
num2: 0XA463
res: 0XA634
One's Complement Operator

ones complement ( ~ precedence IS ) is a unary operator applied to an
.
. , ,
Ckr<l . % a ut 'c laratler () r integer). It complements the hits in the
operand;
S rt rCVerse the bit va ue The result is 1 when the original bit is 0; it
is 0
’
^
*
.
-
en t it origina hit is 1 . fable 14 5 shows the result of the one’s complement
Original Bit Result

0 1
1 0
TABLE 14 - 5 Ones Complement Truth Table
Program 14- 4 demonstrates the basic operation of the one’s complcmen
PROGRAM 14 - 4 One's Complement

1
2
I /* Demonstrate use of one s complement
Written by:
'
continue
PROGRAM 14- 4 Ones Complement ( continued )

3 Date:
4 */
8
9 int main ( void )
10 {
11
12
_
uintl6 t num = 0x0257 ;
13 uintl6 t res;
14
15 // Statements
16
17
res = -
num;
18 // Print results in hexadecimal

19 printf ( "Input and results in hexadecimal:\n");
20 printf ( "num: % # 06 X \n " , num );
21 printf ( "res: % #06 X \n " , res );
22
23 return 0;
24 > // main
Results:
num: 0X0257
res: 0XFDA8
Checksum
One bitwise manipulation application is the calculation of a checksum. A
checksum is a mathematical calculation used for error detection. For exam-
pie, credit cards use a checksum to verify that the card number is valid.
Another example involves the detection of a bad transmission through a com-
munication channel. When the data are sent, a checksum is calculated and
attached to the transmission. When the data are received, the checksum that
includes the sender ’s checksum is recalculated. If the newly calculated
checksum is not zero, an error has occurred.
The traditional Internet checksum uses a 16-hit calculation. To calculate
the checksum, the following steps are used.
.
1 Set the checksum to 0 .
using one’s complement
2. Add 16-bit sections of data to the checksum
arithmetic. ( See Appendix D.)
.
3 Complement the result .
We complement the result at the end because a number (T) and its
com-
plement (-T), when added always give - 0 in one's complement.
Figure 14- 1 shows the calculation of a checksum for a small
string.
“ABCDEFGHI.” Note that because the number of characters is nine, we
include the null character ( \ 0 ) at the end of the string to make live 16- bit val
ues . The string is logically divided into groups of 16 hits ( two characte
-
rs),
which are then added together using one’s complement.
1 Carry from 4th column

1 Carry from 3rd column
1 Carry from 2nd column
This column 1 Carry from 1st column
1
should be added
to the partial sum
* 3
4
4
2
4
Hexadecimal A & B
Hexadecimal C & D
4 5 4 6 Hexadecimal E & F
4 7 4 8 Hexadecimal G & H
4 9 0 0 Hexadecimal I
5 A 1 4 Partial sum
1 Carry from last column
5 A 1 5 Sum
A 5 E A Checksum
FIGURE l 4 - 1 Checksum Calculation
Program 14- 5 demonstrates the calculation of a checksum for the sender

site , lo properly weigh the first character, we
multiply it by 256. To get the
one’s complement , we add the carry hits after the sum is calculat as shown
ed
in Figure 14- 1 . Ibis is done using a
loop. The loop extracts the higher ( lett 1
16 hits and adds them to the lower ( right )
16 hits.
PROGRAM 14- 5 Demonstrate Checksum
1 / * Demonstrate the calcula
tion of a checksum using
one’s
2 complement arithmetic.
3 Written by:
4 Date:
5 */
8 #include <stdint.h>
9
11 {
contint ei
‘
PROGRAM 14 - 5 Demonstrate Checksum ( continued )

13 uint32 t sum = 0x00000000;
14 _
uintl6 t checksum = 0x0000;
15 char* str = " ABCDEFGHI" ;
16 int len ;
17
18 // Statements
19 len = strlen (str );
20 if ( len % 2 == 1)
21 / / Make the number of characters even
22 len++;
23
24 for ( int i = 0; i < len ; i += 2 )
25 sum = (sum + str[i] * 256 + str[i + 1 ]);
26
27 // Add carries into lower 16 bits
28 while (sum » 16 )
29 sum = (sum & Oxffff ) + (sum » 16 );
30
31 // Complement
32 -
checksum = sum ;
33
34 printf ( "str: %s\n " , str );
35 printf ( "checksum: %#06X\n" , checksum);
36
37 return 0;
38 } // main
Results:
str: ABCDEFGHI
checksum: 0XA5EA
14.3 Shift Operators

The shift operators move bits to the right or the left. When applied to
unsigned numbers, these operators are implementation independent . When
used with signed numbers, however, the implementation is left to the discre-
tion of the software engineer who designs the compiler. It is often predicated
on the hardware for which the compiler is being written . Therefore, they
must be used with caution with signed numbers. Because the C standard
leaves the implementation up to the compiler writer there is no standard
( )
code that shifts signed negative numbers may not he portable to other platforms .
890 Section 14.3 Shift Operators
Bitwise Shift - Right Operator

The bitwise shift right (»—precedence 1 1 ) is a binary operator that requires
two integral operands (character or integer). The first operand is the value
to
he shifted . The second operand specifies the number of hits to be shifted .
Shifting binary numbers is just like shifting decimal numbers. When bits
are shifted right , the hits at the rightmost end are deleted . What is shifted
in
on the left , however, depends on the type and the implementation. If the
type
is unsigned , then the standard calls for zero hits to be shifted in . If the type is
signed , however, the implementation may either shift in zeros or copies of the
leftmost hit. Since the implementation is the system programmers responsi-
bility, any function that shifts signed negative values may not be portable.
The shift - right operation is diagrammed in Figure 14 - 2.
0 0 0 ... 1 1 0 0 0 1
/'"O or 1 bits"^
Inserted on righi Rightmost bit
^
discarded .
... 1 1 1 0 0 1
FIGURE l 4 - 2 Shift - right Operation
Program 14 -6 demonstrates the shift - right operation , because C does

not provide any formatted conversion for binary numbers, we wrote a simple
lunction to print a 16 - bit fixed integer variable as a binary number.
PROGRAM 1 4 - 6 Simple Shift - right Demonstration

1 /* Demonstrate the bitwise shift right operator.
2 Written by:
-
3 Date:
4 */
6 tinclude <stdlib.h >
7 # include <stdint.h>
8
9 / / Function Declaration
10 void binl6 ( uint! _
6 t num , char * bitStr );
11
13 {
14 / / Local Definitions
15 _
uintl6 t num = 0x0040 ;
16 _
uintl6 t res;
coiitin nei
PROGRAM 14 - 6 Simple Shift - right Demonstration (continued )

17 char bitStr[ 17 ] = {0};
18
19 // Statements
20 binl6 ( num , bitStr );
21 printf( "Original value: %s ( % # 06x )\n " ,
22 bitStr , num );
23
24 res = num » 1;
25 binl6 (res , bitStr );
26 printf( "Shifted 1 right: %s ( % #06 x )\n " ,
27 bitStr , res);
28
29 res = num » 2 ;
30 binl6 ( res , bitStr );
31 printf( "Shifted 2 right: %s ( % #06 x )\n " ,
32 bitStr , res);
33
34 res = num » 4 ;
35 binl6 (res, bitStr );
36 printf( "Shifted 4 right: %s ( % #06x )\n " ,
37 bitStr , res );
38
39 return 0;
40 } // main
41
42 /* ===== ==== binl6 ====
43 Convert fixed 16 bit integer to binary digit string.
-
44 Pre num contains integral value to be converted
45 bitStr is pointer to variable for bit string
46 Post bit string stored in str
47 */
48 void binl6 ( uintl6_t num , char* bitStr )
49 {
50 // Statements
51 for ( int i = 0; i < 16 ; i++)
52 bitStr[i] = (char) ((num » 15 - i) &
53 0 X0001 ) + 48;
54 return;
55 } // binl6
Results:
Original value: 0000000001000000 (0x0040)
Shifted 1 right: 0000000000100000 ( 0x0020)
Shifted 2 right: 0000000000010000 (0x0010)
Shifted 4 right: 0000000000000100 (0x0004 )
Dividing by 2
Let’s start with something we know, decimal numbers When we shift a
deci- .
mal number one position to the right and insert a 0 on the left, we are
in
effect dividing by 10. If we shift it two places, we are dividing by 100. If we
.
shift it three places, we are dividing by 1000 But actually, we are dividing by
a power of 10—in our examples, 101, 102, and 103 .
Applying the same principle to binary numbers, the right-shift operator
divides by a power of 2. If we shift a binary number two places to the right,
we are dividing by 4 ( 22). If we shift it three places, we are dividing by ( }
8 2 ).
Table 14 - 6 shows the division pattern used w i t h bit shitting.
2 shift value Divides by Shift Operator

1 2 »1
2 4 »2
3 8 »3
4 16 »4
n 2n »n
TABLE 1 4 - 6 Divide by Shift
Bitwise Shift- Left Operator
rcmi
th< *
.bltVy.,
v|
S
Se Shlft
JCftPerantk
°
<< precedence 1 1 ) is a binary operator that
(
(character or integer). The first operand is
nd specifies the number of bits to
he shifted ° *
^ ** ^ S seconc 0Pe
' k is just like shifting decimal numbers. If we have

an • i'* > numbers
number
lar
we shift it three places to the left, then

the 'T i . are ostand CClnla
>
zero digits are added on the right.
/**operation is ^shown indiree
anci
S
The
1 hC bmary
h shlft L
Figure 14- 3
0 0 0 ... 1 1 0 0 0 1
Leftmost zero
vbit discarded.
1 Zero insert
on right
L
0 0 0 0 . .. 1 0 0 0 1 0
FIGURE 14 - 3 Shift-left Operation

Program 14- 7 demonstrates the basic operation of the shift - left operator.
Again, it uses the bin 16 1 unction , which we have now included as a library.
PROGRAM 14 - 7 Simple Shift-left Demonstration

1 /* Demonstrate the bitwise shift-left operator .
2 Written by:
3 Date:
4 */
7 # include <stdint.h>
8
9 # include " binl6.c"
10
12 {
14 uintl6 t num = 0x0031 ;
15 _
uintl6 t res;
16 char bitStr[ 17] = {"0"};
17
18 // Statements
19 binl6 ( num , bitStr );
20 printf( "Original value: %s ( % #06x )\n " , bitStr, num);
21
22 res = num « 1 ;
23 binl6 (res , bitStr);
24 printf( "Shifted 1 left: %s ( % #06x )\n", bitStr , res );
25
26 res = num « 2;
27 binl6 ( res, bitStr);
28 printf( "Shifted 2 left: %s ( % # 06x )\n " , bitStr, res );
29
30 res = num « 4 ;
31 binl6 ( res , bitStr);
32 printf( "Shifted 4 left: %s ( % # 06x )\n", bitStr , res);
33
34 return 0;
35 > // main
Results:
Original value: 0000000000110001 (0x0031 )
Shifted 1 left: 0000000001100010 (0x0062)
Shifted 2 left: 0000000011000100 (0x00c4)
Shifted 4 left: 0000001100010000 ( 0x0310)
Multiplying by 2
Shift left is the opposite of shift right ; multiply is the opposite of divide. It is
reasonable to assume , therefore , that shifting left has the effect of multiplying
by a power of 2. If you examine the results of Program 14-7 closely, you will
note that is exactly what happens. When we shift a binary number two places
to the left , we are multiplying by 4 ( 22 ) . If we shift it three places, we are mul
-
tiplying by 8 ( 23). Table 14-7 shows the multiplication pattern used with bit
shifting.
2 shift value Multiplies by Shift Operator

1 2 «1
2 4 «2
3 8 «3
4 16 «4
n 2n «n
TABLE 1 4 -7 Multiply by Shift
Rotation
In computer science , we often need to rotate bits . For example , in security
the creation of a hash digest of a message rotates and then scrambles bits to
create a hash digest of data.
Rotation requires that hits be taken off one end of a value and moved to
the other end. When we rotate hits left , we shift them off the left end of the
value and insert them on the right . When we rotate right , we shilt oil the
right end and insert them on the left . Figure 14 - 4 demonstrates the rotation
of a 16 - hit integer 4 bits to the right and then 4 bits to the left .
Rotate Right 0101 0010 0011 0100

right
Original 0010 0011 0100 0101
left
Rotate Left 0011 0100 0101 0010
FIGURE 14- 4 Right and Left Rotation
Note that the original number in hexadecimal is 0x 2345. The r()tale

right number is 0x 5234; the rotate left numbers is 0x 3452.
Although there are no explicit operands for rotating left or right , the
rotations can be coded using two shifts and an or. For example, to rotate the
original 16 - bit number in Figure 14 - 4 , we use the following steps .
Original: 0x2345
Shift right 4 bits: 0x0234
or'd results:
-
Shift left ( 16 4 ) bits: 0x5000
0x 5234
I he rotate left operation is similar. Program 14 -8 is a test driver for the

rotate left and right functions.
PROGRAM 14 - 8 Rotate Left and Right Test Driver

1 / * Test driver for rotate left and right .
2 Written by:
3 Date:
4 */
7
8
9 _
_
uintl6 t rotatel6Left (uintl6 t num , int n );
10 uint! _
6 t rotatel6Right ( uint! _
6 t num , int n);
11
13 {
15 uintl6 t num = 0X2345;
16
17 // Statements i
18 printf( "Original: % # 06 X \n " , num );
19 printf( " Rotated Left: % #06X\n " ,
20 rotatel6Left ( num, 4 ));
21 printf( "Rotated Right: % #06X\n " ,
22 rotate16Right(num, 4));
23 } // main
24
25 /* = ============ rotatel6Left ============
26 Rotate 16 bit fixed size integer left n bits.
27 Pre num is a fixed size 16-bit integer
28 Post num rotated n bits left
29
30
*/ _
uintl6 t rotatel6Left (uintl6 t num , int n)
31 {
32 return ( ( num « n ) ( num » 16 - n ) );
896 Section 14.4 Masks
PROGRAM 14- 8 Rotate Left and Right Test Driver (continued )

33 } // rotatel6Left
34
35 /* === === rotate 16Right === ==
36 Rotate 16 bit fixed size integer right n bits.
37 Pre num is a fixed size 16 bit integer -
38 Post num rotated n bits right
39 */
40 _
uintl6 t rotatel6Right ( uintl6 t num , int n)
41 {
42
43
return ( ( num » n )|( num « 16
} // rotatel6Right
- n ) )?
Results:
Original: 0X 2345
Rotated Left: 0X 3452
Rotated Right: 0X5234
14.4 Masks
In many programs, bits are used as binary ( lags: 0 is off, and 1 is on. To set
and test the Hags, we use a bit mask. A mask is a variable or constant, usu-
ally stored in a byte or short integer, that contains a bit configuration used to
control the setting or testing of bits in a bitwise operation. The bits are num-
bered Irom the least significant bit (rightmost), starting at 0, as shown in
figure 14 - 5.
Bit 15 Bit 7 BitO
1000000100000001
FIGURE l 4 - 5 Bit Mask in a 16-bit Integer
Creating Masks
Masks can 1 ,,r < .
niically Sialic masks are generally liter-
*
•Is. For esaie “o!" "'
oZm IT “
V» L
m k in <«•'« could code the constant:
operators
,
shift, and, or. In thi SCtl < )n > We examine * the basic manipulation
^
CTCalec us ng the bit
techniques for creating
dynamic masks, tb ,.
thata ;s masks that may change as the program executes.
One - Bit Masks

Given a byte flag in which hit 5 is an error flag, we need to create a mask to
test it. To test hit 5 , the mask must have hit 5 set to 1 and the rest of the bits
to 0. This is easily done , as shown below.
_
uint8 t mask ;
mask = 0x01 « 5;
00000001 « 5 -> 00100000
We start with a constant binary 1 . This places a one hit in the rightmost
position . We then shift the one bit to the left five positions. The result is
shown in the previous example .
An easier way to control the location of the hit flags is to name them .
This is easily done with a defined constant , as shown in the next example.
# define FLAG0 0
# define FLAG 1 1
# define FLAG5 5
mask = 0x01 « FLAG5 ;
By using named flag positions, we free ourselves from remembering their

locations and at the same time make the code more readable. Of course, in
an application , we would use meaningful names such as OVERFLOW FLAG,
not FLAG 1.
Two- Bit Masks

Creating a mask with two flags requires that we set 2 hits on . The easiest way
to do this is to set two individual flags and then or them .
uint8 t mask ;
mask = (0x01 « FLAG1) (0x01 « FLAG5);
The result of the above code would set the mask as show n below.
00100010
Setting Bit Range Masks

To set ,a range of bits requires a loop The . following example sets bits 7
through 3 .
uint8 _t mask _ = 0x00;
for ( i = 3; i < = 6; i++) // 01111000
mask = mask | 0x01 « i ;
Complement Masks
To create a mask that turns off a s pecific bit — that is, that complements it—
we must set the hit to 0 while setting the rest of the hits in the mask to I .
This_ requires two statements. In the first statement, we set only the desired
hit on. In the second step, we complement the mask, changing all zero bits
to I and all one hits to 0 The code is shown below. .
mask = 0x01 << FLAG 1 ;
mask = mask ;-
Using Masks
Now that we’ve seen how to create masks with various bit flags set, let ’s look
at some examples that use them.
Turning Bits Off

One of the applications of the bitwise and operator is turning bits off that —
rC n® SC , ,
h ts to zero For example, an Internet address uses the for-
’ ,° ; eCteLd
e seen before. A mask in the form / n, often called prefix,
\ aSp ' "
m lf ' ,uluor
.
i
it. 1 TI rC ,
I C
UJ of /2 means
j
^ or subnet address to which a computer

U
c nes lbe number of leftmost contiguous Is out of
)
|
^ ^ that the leftmost 20 hits are I s and the right-

X
is
32 It A Prefix
7 . \ |
most bits are Os. This°is demonstrated in the next

: S
example.
Computer address: 01111011010011100001100100001101
Mask: 11111111 11111111 11110000 00000000
Network address: 01111011010011100001000000000000
Program 14- 9 determines a network address given a computer host
address. To be flexible, it asks the user to enter the size of the prefix.
PROGRAM 14 -9 Determine Network Address
1 /* Given a host address and the size of the prefix,
2 determine its network address.
continue
PROGRAM 14 - 9 Determine Network Address (continued )

3 Written by:
4 Date:
5 */
9
11 {
13 unsigned int comAddr[ 4];
14 unsigned int mask[4];
15 unsigned int netAddr( 4 ];
16 _
uint32 t comAd = 0
_ mask 32 = 0
17
18 _
uint32 t
uint32 t netAd = 0
19 int prefix ;
20
21 // Statements
22 printf ( "Enter host address <x.y.z.t>: ")7
23 scanf ( " %d %*c %d %*c%d %*c %d " ,
24 &comAddr[ 3 ], &comAddr[ 2 ], &comAddr[ l ],
&comAddr(0]);
25
26 printf ( "Enter prefix: " );
27 scanf ("%d", &prefix);
28
-
// Convert address to a 32 bit computer address
29
30
31
32
for (int i
comAd
= 3; i >= 0; i )
—
= comAd * 256 + comAddr[ i];
33 -
// Create a 32 bit mask
34 for ( int i = 32
mask32 = mask 32
-
prefix ; i < 32; i++)
( 1 « i);
35
36
37
netAd = comAd & mask32 ;
-
// AND to get a 32 bit Network Address
38
39
40 // Change mask into the form x.y.z.t
41 for (int i = 0; i < 4 ; i++ )
42 {
43 mask[i] = mask32 % 256;
44 mask32 = mask 32 / 256 ;
continued
PROGRAM 14 -9 Determine Network Address (continued )

45 } // for
46
47 // Change IP address into the form x.y.z.t
48 for ( int i = 0; i < 4 ; i++ )
49 {
50 netAddr[i] = netAd % 256;
51 netAd = netAd / 256 ;
52 > // for
53
54 // Print Addresses
55 printf ( " \nAddresses:\n ” );
56 printf ( " Computer Address: " );
57 printf ( " %d.%d.%d.%d\n" ,
58 comAddr[ 3], comAddr[ 2],
comAddr[ 1 ],comAddr[ 0 ]);
59
60 printf ( " Mask:
61 printf ( " %d.%d.%d.%d\n" ,
62 mask[ 3], mask[ 2], mask[ l ], mask[0]);
63
64 printf ( " Net Address: " )?
65 printf ( "%d.%d.%d.%d\n",
66 netAddr[3], netAddr[2], netAddr[ l],
netAddr[ 0 ]);
67 return 0 ;
68 } // main
Results:
Enter host address <x.y.z.t>: 123.45.78.12
Enter prefix: 18
Addresses:
Computer Address: 123.45.78.12
Mask: 255.255.192.0
Net Address: 123.45.64.0
Program 14 - 9 Analysis Several parts of this program need to

be studied carefully. First, note that the
for the network address is an array of four integers. In stateme 23, we rea
address in dotted decimal format into the a r r a y.
H r ..
nt
components of the address into a 32 -bit

^
varJ
^Ta? ,!
COn /erts
address Each comtv P nen o the address is considered one byte in the final address.
Since each byte has a value of 256 ( 28), we com
pute the address as follows:
bytel 0] + byte [ 1]x 2561 + byte [ 2 ] x 2562 + byte [ 3]x 2563

Once we have the dotted- decimal address converted to a 32 - bit address, we must
create a prefix mask by setting a 32-bit mask to /n leading 1 bits. This is done in the
for loop in statement 34.
Finally, the network address and mask are converted back to the dotted-decimal
format. Each byte is determined by taking modulo 256 of part of the network address.
To get the next byte in the address, we then divide by 256 . This logic is seen in the for
loops in statement 41 and 48 .
Turning Bits On
—
One common application of bit inclusive or is turning bits on that is, forc -
ing selected bits in a field to one. Io force to one, all that we need to do is to
construct a mask with the desired one bit set on. This guarantees that the
result will have a one bit in that location .
A mask can be used to find the last address in a block of network
addresses. I he last address is used for limited broadcasts. If a computer wants
to send a message to all computers inside its network, it uses this address.
To find the last address, we or the complement of a prefixed mask with
any address in the block . Conversely, to find the first address in the network
block , we and the mask with any address. Program 14 - 10 determines the last
address in a network.
PROGRAM 14 - 10 Determine Last Address in a Network

1 /* Determine the last address in a broadcast network.
2 Written by:
3 Date:
4 */
6 # include <math.h >
8
9 int main ( void )
10 {
12 unsigned int comAddr[4];
13 unsigned int mask[4];
14 unsigned int broadAddr[ 4 ];
15
16
17 uint32 t__
uint32 t comAd
mask 32
= 0
= 0
18 uint32 t broadAd = 0
19 int prefix;
20
21 // Statements
22 printf ( "Enter host address <x.y.z.t>:
23 scanf ( " %d %*c%d %*c%d %*c%d " ,
continuec
J
PROGRAM 14 - 10 Determine Last Address in a Network (continued )

24 &comAddr[ 3 ], &comAddr[ 2 ],
25 &comAddr[ l ], &comAddr[ 0 ]);
26 printf( " Enter prefix: " );
27 scanf ("%d " , &prefix);
28
29 // Convert address to 32- bit computer address
30
31
32
for ( int i = 3 ; i >= 0 ; i )
—
comAd = comAd * 256 + comAddr[ ij;
33 // Create 32- bit prefix mask

34 for ( int i = 32-prefix ; i < 32; i++ )
35 mask 32 = mask32 | ( 0x0001 « i );
36
37 -
// And to get a 32 bit Network Address
38 broadAd = comAd |(~mask 32);
39
40 // Change the mask into the form x.y.z.t
41 for ( int i= 0 ; i < 4; i++ )
42 {
43 mask[ i]= mask32 % 256 ;
44 mask 32 = mask 32 / 256 ;
45 } // for
46
47 // Change the IP address to the form x.y.z.t
48 for ( int i= 0 ; i < 4 ; i++ )
49 {
50 broadAddr[ i] = broadAd % 256 ;
51 broadAd = broadAd / 256 ;
52 } // for
53
54 printf ( " \nPrinting Addresses\n" );
55 printf ( " Computer Address: " )?
56 printf ( " %d.%d.%d.%d \n " ,
57 comAddr[ 3], comAddr[ 2],
58 comAddr[ l ], comAddr[ 0]);
59 printf ( " Mask: ")?
60 printf ( " %d.%d.%d.%d\n " ,
61 mask[3], mask[2], mask[ l ], mask[0]);
62
63 printf ( " Broadcast Address: ” )?
64 printf ( "%d.%d.%d.%d \n " ,
65 broadAddr[3], broadAddr[2],
66 broadAddr[ 1 ], broadAddr[ 0]);
67 return 0 ;
PROGRAM 14 - 10 Determine Last Address in a Network ( continued )

67 } // main
Results:
Enter host address <x.y.z.t>: 123.45.78.12
Enter prefix: 18
Printing Addresses
Computer Address: 123.45.78.12
Mask: 255.255.192.0
Last Address: 123.45.127.255
Program 14 - 10 Analysis Let's compare Program 14- 9 and Program 14- 10. In Program 14 - 9, we calculate the
first address in a block . In Program 14- 10, we calculate the last address in a block. If
we subtract these addresses ( in base 256 ) and add 1, we get the total addresses in
the block. Looking at the program results, we see that the first address is 123.45.64.0
( Program 14 - 9 ) and the last address is 123.45. 127.255 (Program 14- 10) . The dif -
ference is
-
( 127 64) * 256 + ( 255 - 0) + 1 = 16 ,384
We have 16,384 addresses in this block . We could have found this result using the
prefix ( 18).
232-prefix
_ 232- 18 = 214 = 16 ,384
Flipping Bits
One of the applications of bitwise exclusive or is Hipping bits— that is, forc-
ing Os to Is and Is to 0s , called force to change. The resulting hit is 1 when 1
bit is a 0 and the other is a 1 ; if both hits are 0 or both are 1 , the result is 0.
To force a bit to change: therefore, the forcing bit in the mask is set to 1 ; bits
that are to he unchanged are set to 0.
For example , assume that we want to change the 5 leftmost bits of a
number stored as an 8- hit integer. We create the mask with 5 one bits and 3
zero hits, as shown below. ( Y indicates a changed hit
.)
operator number XXXXXXXX

mask 11111000
result YYYYYXXX
Dividing Polynomials
bits, let’s examine how we can use
As a more advanced example of flipping
fixed -size integers to represent polynomials. Polynomials with coefficients of
1 or 0 are used in computer science for error detection , encoding, and other
applications.
A fixed -size integer can represent a polynomial il we think of 0 and 1 as
coefficients of the polynomial. For example, the hit stream 01001001 can
represent a polynomial of degree 6 or x + + 1 . In this polynomial , the coel -
6
ficients of x , x \ x , x , and x are 0 and the coefficients x , x \ and x are 1 .

"
4 2 1 6
°
This relationship is shown in Figure 14 - 6 .
In general a hit stream of length n can represent a polynomial of length
_
n - 1 . So a data type of uintl 6 t can represent a polynomial of up to
degree 1 5 .
A common polynomial operation is division . The division of one polyno-
mial by another is shown in Figure 14- 7.
Positions 7 6 5 4 3 2 1 0
.
0 1 0 0 1 0 0 1
X6 + X3
'
+
.
x
°
FIGURE 14 - 6 Polynomial Coefficients
+ X + X2+1
6 3 Quotient
X
Divisor x3 + 1) x9 + x5 + X2 + X + 1 Dividend
x9 + x6
+ X6 + X 5 + X2 + X +1
Term Hex
+ x6 + X3
Dividend 0x0227
+ x5 + x3 + x2 + x + 1 Divisor 0 x 0009
Quotient 0 x004D
+ X5 + x2
Remainder 0 x 0002
+ X3 +X +1
+ X3 +1
x Remainder
FIGURE 14 -7 Polynomial Division
Although the division of polynomials is similar to the division of numbers

we learned in elementary school, there are two differences:
1 . Because we use only coefficients 0 and 1 , if we have two terms of the
same exponent, the result is zero ( 1 + 1 = 0 ). In other words, polynomial
division uses modulo 2 arithmetic on the coefficients.
2. Subtraction and addition of the coefficients are actually the same ( exclu -
sive or ) because we are interested in only coefficients of 0 and 1 .
"

\
Note that il we multiply the quotients by the divisor and add it with the
remainder, we get the dividend. However, if we get two identical coefficients
(such as x6), we cancel them because 1 + 1 = 0.
Program 1 4 - 1 1 demonstrates polynomial division .
PROGRAM 14 - 11 Polynomial Division

1 / * Demonstrate polynomial division .
2 Written by :
3 D a t e:
4 */
5 # include < stdio h > .
6 # include < stdint h > .
7
8 / / Function Declaration
9 i n t degree ( uintl 6 t ) ; _
10
12 I {
13 / / Local Declaration
u i n t l6 t dvdn = 0X 0227;
14
_ 0 X 0009 ;
15
16
uintl6 t
uintl6 t __ dvsr
qtnt
=
= 0 X 0000 ;
17 uintl6 t rmdr ;
18 uintl6 jt q;
int dgre ;
19
20
/ / Statements dvdn ) ;
: % # 06 X \ n " ,
21
22 printf ( " Dividend X \ n " , dvsr ) ;
% # 06
23 printf ( " Divisor :
24
) >= 0)
25
26
rmdr = dvdn ;
while ( ( d g r e = degree ( dvdn
) - degree ( dvsr )
27 {
28 q = 1 « dgre; < < degree (
q)) /
( dvsr A
29 rmdr = dvdn
qtnt | q
30 qtnt = i
31 dvdn = rmdr ;
32 > / / while
", qtnt ) ;
33
Quotient : %#06X\n
printf (" rmdr);
34
"Remaind er: % # 06X \ n " /
35 printf (
36 return 0 ;
37 > / / main by
represented
38 degree of
polynomial continued
39 /* Determine *.
T
PROGRAM 14- 11 Polynomial Division ( continued )

40 -
a fixed-length 16 bit variable.
41 Pre poly represents a polynomial
42 Post degree returned
43 */
44 int degree ( uintl6 t _ poly )
45 {
47 _
uintl6 t mask = 0X0001 ;
48 _
uintl6 t temp ;
49
50
int pos= 1;
-
51 // Statements
52 for ( int i = 0 ; i <16 ; i++)
53 {
54 temp = poly » i;
55 if (( temp & mask ) == 1 )
56 pos l?
57 } // for
58 return pos;
59 } // degree
Results:
Dividend: 0X 0227
Divisor: 0X0009
Quotient : 0 X 004D
Remainder: 0X0002
Program 1 4 - 1 1 Analysis We use a function to determine the degree of the polynomial. In main we use a while
loop to simulate the division. The variable poly is the term in the quotient that is
obtained in each step. Note that the multiplication of a term such as xn by the division
is done by shifting the divisor n bits.
ppj

In Chapter 12, we looked at what makes a good
function . In this section , we
look at how you design good programs.
Payroll Case Study

Io provide a discussion focus, we will use a payroll program . Although our
example is rather simple , it does contain all the elements involved in designing
a program. The description of the payroll program is shown in Figure 14 -8 .
Payroll Case Study

1 . Requirements:
Given employees and their hours worked , compute net pay and
record all payroll data for subsequent processing, such as W 2 state-
ments. Prepare paychecks and a payroll ledger.
Maintain data on a sequential payroll file.
2 . Provide for the following nonstatutory deductions:
a. Health plan
b. United Way
c. Union dues
3. The payroll data are:
a . Employee number
b. Pay rate
c. Union member flag
d . United Way contribution
e. Exemptions
4. Maintain the following year- to-date totals
:
a. Earnings
.
b FICA taxes
c. SD 1 taxes
d . Federal withholding
e. State withholding
f. Health plan fees
g. United Way donations
h . Union dues
5. Algorithms: * Rate * 1.5 )
^ ^J _
rs
^
a. Gross Pay = ( 8
,
« piCA Rate ) if less than Max
b. FICA Taxes = ( G «* ' >, D Ratc ) if ess than Max SDI
FICA
Pa } >
c. SDI Taxes = ( Gross GrossFay t _ jons • Exemption Rate ) )
d. Taxable Earnmgs
( = ^
Payroll Case Study
FIGURE 14- 8 Requiremen ts for
Payroll Case Study

e. Federal Taxes = (Taxable Earnings * Federal TaxRate)
f . State Taxes = (Taxable Earnings * State Fax Rate )
—
g. Net Pay = ( Gross Pay ( FICA Taxes + SDI Taxes +
Federal Taxes + State Taxes + Health Fee +
United Way Donation + Union Dues ) )
FIGURE 14 - 8 Requirements for Payroll Case Study ( continued )
Program Design Steps

As you know by now, a program is developed in seven steps:
.
1 Determine requirements.
2. Determine data structures.
.
3 Build structure charts.
4. Create test cases.
5. Write and unit test the programs.
.
6 Test system .
7 . Implement system in production .
Our interest here is only in the third step: Build the structure charts. A
few general comments are in order, however. The second and third steps are
olten reversed or done concurrently. Which one is done first is not of major
consequence , as long as they are done before the fourth step.
Many programmers think that test cases should be built after the pro-
gram is written . Good programmers know better. By creating
test cases based
on the requirements (Step 1 ) and your design ( Steps
2 and 3) , you will under-
stand the problem better ou will even find occasions when
,
^
design based on what you learned creating test cases.
you change your
I his does not mean that you will he done with
creating test cases at
Step 4 ; you just start there. You will develop more test cases while you are
writing the program , and you will create still more as
you conduct unit testing.
Structure Chart Design

A good program starts with a good design , as reflected
in the structure chart .
By now, you should have progressed to the point at
which you are designing
\ om programs before you start
coding; that is , you are creating your structure
chart first.
One tool to design the structure chart is known
as transform analysis.
Transform analysis is a design technique that identifies
the processes in a
program as input , process, and output and then
organizes them around one
or more processes that convert inputs to
outputs. These conversion processes
in
ait known as the central transforms. Having

determined the first -cut design ,
u repeat
the process, decomposing the identified modules into subtasks
\
°
using transform analysis.
This is a good design technique whenever a program
reads, processes,
and writes data , which covers the majority ot programs. Although
it is usually
used in conjunction with another tool known as a data flow diagram, you
can
use it independently.
Good programs start with a good structure chart design .

Recognize , however, that transform analysis is an approach to the design
of a program . It is not a cookbook that leads to the same results every time.
Different programmers using the same steps will arrive at different designs.
Programmers use six steps in designing a structure chart :
1 . Determine program modules.
2. Classify modules.
.
3 Construct preliminary structure chart .
4. Decompose modules.
5. Complete structure chart .
6. Validate design .
Determine Modules
The first step in program design is to identify the processes that the program
will use This is usually done by reviewing the program specification and
identifying the tasks it needs to accomplish . A review of our payroll case iden -
tifies the following tasks. (The references are to the case study descnpt on in
Figure 14-8.)
.
1 Read hours worked ( 1 ).
.
2 . Compute pay ( l ) -
.
3 Maintain payroll master file ( 1 ) .
4. Prepare paychecks ( 1 ).
5. Prepare payroll ledger ( 1 .
)
6. Calculate nonstatutory
deductions ( 2 ).
7 . Calculate year- to-date
totals ( 4 ) .
( S ).
8. Calculate gross pay
9. Calculate taxes ( 5 ).
Remember the rule ha ‘ w prepare output instead of prepare
miight be tempted
to start wi
£AnJ fact, you may well end up with
pa ychecks and prepare-
an intermediate PJ ' Jines
module ma
„ the preparation of all reports. But at
this point , we want to keep different things separate as much as possible. On

the other hand, experience indicates that to have separate modules for calcu -
lating FICA taxes , SD 1 taxes, federal taxes, and state taxes is unnecessary.
Classify Modules
—
In transform analysis , we look for the central transforms that is, the mod
ule( s ) that turns inputs into outputs. To identify these transforms , we classify
-
each identified task , which now represents a module in the structure chart,
as afferent , efferent , or transform. A module is an afferent module if its pro-
cessing is directed toward the central transform . Another way of looking at
this concept is to say that the module is a gatherer of data . Efferent modules
direct data away from the central processing or toward the outputs of the pro-
gram . ( Remember that input comes before output and a comes before e\
therefore , afferent is input and efferent is output . ) A transform module is
balanced ; that is, it has data flowing both in and out . The concepts of affer-
ent , efferent , and transform are shown in Figure 14 -9 .
t J JJJ
afferent efferent transform
FIGURE 14 - 9 Afferent, Efferent, and Transform Modules
I able 14-8 classifies each of the modules identified above as afferent

efferent , or transform .
Module Afferent Efferent Transform

Read hours worked /
Compute pay /
Maintain payroll master file / /
Prepare paychecks /
Prepare payroll ledger /
Calculate nonstatutory deductions /
Calculate year-to-date totals /
Calculate gross pay /
Calculate taxes /
TABLE 14- 8 Classification of Payroll Modules

11
We have two comments about this classification.

Note that "maintain payroll
filC \\r
if ^
Cii 1166 JSSt ^
1S C C 3S
nPut ^ ,
an output. This is because it is a master
' emP oyee personnel and history data, and
^
0 rCaC 1110 Ct
^
we will needJ to write the updated data after they have been
calculated. There-
fore it is really two modules: one to read the master and one to write
the master.
Second, there are many transform modules. This will result in a large fan
out, which is not desirable. Fan out is the number of submodules
emanating
1 mm a module. We will need to reduce the fan out later.
Construct Structure Chart

At this point, you are ready to construct the first -cut structure chart. Transform
analysis structure charts are organized with inputs on the left, transforms in the
center, and outputs on the right. This organization is shown in Figure 14- 10.
program
input process output
FIGURE 1 4 - 1 0 Basic Structure Chart Organization
For the first -cut structure chart, place all afferent modules below the
,
input block, all transform modules below the process block and all efferent
modules below the output block. In this process , analyzx each module to
be called before or after the other processes at the
determine if it needs to
placed the left, and those
same level. Those that need to be called first
are on
that need to he called last are placed on the right . The resulting structure
chart is shown in Figure 14- 11.
program
process output
input
write write payroll

read hours ] freadpayroilj write pay
checks ledger master
worked I
^nasteij
compute
pay
calc
gross pay
calc non-
stat dedns
calc
taxes ] calc
year to date )
FIGURE 1 4 - 1 1 First-cut Structure Chart

Up to this point, we have been almost algorithmic; that is, we have exer-
cised little judgment. We now need to analyze the preliminary structure chart
to see if it makes sense . This is done by asking a simple question , "What do
we mean by...?” For example, "What do we mean by compute pay? ” The
answer is that we must take hours worked and multiply them by the pay rate,
also considering overtime and so forth . Or, to put it another way, we compute
gross pay. But this is already a module in the structure chart , so these two
modules are the same. We will therefore delete compute pay since it is a less
specific description than calculate gross pay.
Decompose Modules
To decompose modules, we look at the cohesion of each module. For exam -
ple, when we ask what we mean by calculate taxes, the answer is calculate
federal taxes and calculate state taxes. Since we are dealing with two different
entities ( things ) , the cohesion of this module is communicational; it uses the
gross pay and payroll data to calculate the different taxes.
When we find that a module is doing more than one thing, or that it is
so complex that it is difficult to understand , then we need to consider break-
ing the module into submodules. This refinement of the modules was
named stepwise refinement by Niklaus Wirth , the creator of Pascal . 1 It
refines the processes in a module until each module is at its most basic,
primitive meaning.
Decomposition continues until the lowest levels in our structure chart
are all lunctionally cohesive and easily understood.
Complete Structure Chart

At this point we have the nucleus of the structure chart complete , and all we
have to do is add the finishing touches. These steps are almost mechanical.
1 . Identify any common processes with a cross- hatch in the lower right cor-
ner. In our payroll case, there are none.
2 . Consider adding intermediate ( middle- level ) modules if necessary to
reduce fan out. I his step should not he done arbitrarily, however. II the
modules next to each other have a common entity, then they can be com-
bined . 11 they don’t, they should he left separate. For example , in the pay-
roll case , we have combined calculate nonstatutory deductions and
calculate taxes into one module called calculate deductions. We would
not combine calculate gross pay with calculate nonstatutory deductions,
and we would not combine calculate taxes with calculate year- to-date
deductions.
3. Verify that the names are descriptive and meaningful
for their processes.
4. Add loops, conditional calls , and exclusive or designators .
I . Niklaus Wirth, Program Development by

Stepwise Refinement ," Communications of the
ACM , 14 , no . 4 ( April 1971 ).
”T1
i
5. Add input/output modules, if not already p resent .

6 . Add initialization and end of job modules.
/ . Add error routines ( if necessary ) .
8. Add data flows and flags as required .

The completed design for the payroll case study is shown in Figure 14- 12.
payroll
for each
initialize employee end of job
get emp
data
calculate
pay i output
i
I
worked I master ^
read hours rea payroll write pay
checks
write
ledger
write payroll
master i
calc calc calc
gross pay deductions year to date
report
errors
calc non- calc
stat dedns taxes
calc federal calc

taxes state taxes
FIGURE 1 4 - 1 2 Final Payroll Structure Chart
Validate the Design
your design once more by repeating

all the design steps, espee ally the func- .
tional decomposition step.
: ‘
,K
Section 14.6 Tips and Common Programming Errors

1 . We recommend that bitwise operators be used only with unsigned fixed-
size operands.
2. To use fixed -size integers, we need to include the stdint .h file header file.
3. The bitwise and operator is only one ampersand ( & ) , not two.
4. The bitwise inclusive or operator is only one bar ( | ), not two.
3. There are two differences between the logical and operator (&&) and the
bitwise and operator ( & ).
. & & is a logical value ( true or false ) . The result of & is
a The result of
another number.
b. The evaluation of an expression that uses & & terminates if the first
operand is false. Expressions with the & operator are always completely
evaluated.
6. There are two differences between the logical or operator ( | | ) and the
bitwise inclusive or operator ( | ).
a. The result of the || is a logical value ( true or false ). The result of | is
another number.
b. The evaluation of an expression that uses the || operator terminates if
the first operand is true . Expressions with the | operator are always
completely evaluated .
7 . I here is a difference between the not operator ( ! ) and the one’s comple-
ment operator ( -). The result of ! is a logical value ( true or false ). The
result of - is another number.
8. I he bitwise and operator ( & ) turns off only the corresponding bits in a
data item that match Os in the mask.
9 . I he bitwise or operator ( | ) turns on only the
corresponding hits in a data
item that match Is in the mask.
10. I he bitwise exclusive or operator ( * ) flips only the correspondin bits in a
g
data item that match Is in the mask.
11. I he bitwise complement operator ( ) flips all the hits
- in a variable.
12 . Avoid using the shift operators with the signed number
.
13. I he shift- left operator (« ) multiplr les a number by a power of 2 ( 2 ) , not
simply by 2 . ”
14. I he shift- right operator ( >> ) divides a n
umber by a power of 2 ( 2 ” ) , not
simply by 2 .
1 3. Rotating bits requires a combination of
shift right and shift left operations.
16 . The difference between bitwise exclusive or ( * ) and the one’s comple-
-
ment ( ) is that the - operator flips all the hits; the operator flips only
the specific hits in the mask.
A
1 / . To add n - bit numbers in one’s complement ,

we use modulo 2 ” - 1 .
wn
i i
14.7 Key Terms

afferent module turning bits on
bitwise and operation turn bits off
bitwise exclusive or operation mask
bitwise inclusive or operation one’s complement
bitwise operation bitwise shift left operation
central transform bitwise shift right operation
efferent module stepwise refinement
fan out transform analysis
flipping bits transform modules
14.8 Summary
J C is a proper language for system programming because it contains opera-
tors that can manipulate data at the bit level . These operators are called
bitwise operators.
I here are two categories of bitw ise operators: logical bitw ise operators and
shift bitwise operators.
I he bitwise and ( & ) is a binary operator applied to two operands of integral
value. It does a bit - by- bit comparison of the two operands. The result of
each comparison is 1 if both bits are Is. It is 0 otherwise.
bits in an operand .
J The bitwise and operator ( & ) is used to turn off masked
to two integral
The bitwise inclusive or ( | ) is a binary operator applied The result of
operands. It does a bit - by-bit comparison of its two operands .
each comparison is 0 if both bits are Os. It is otherwise
1 .
) used to turn on mask bits in an
The bitwise inclusive or operator ( | is
operand.
binary operator applied to two integral
The bitwise exclusive or ( ) is a
A
of its two operands. The result of

operands. It docs a bit-by-bit comparison
two bits are different. If they are the same
,
each comparison is 1 only if the
the result is 0.
( ) is used to flip masked bits in an
A
The bitwise exclusive or operator

operand.
operator (- ) is a unary operator. It changes the
The bitwise complement 1 or from 1 to 0.
0 to
value of each bit from
inclusive or, and bitwise
second operand in the bitwise and , bitwise
The a mask
exclusive or is often called
,
to shift data to the right

are binary operators used
Bitwise shift operators operand in these operators is an integer that
or the left . The second to he shifted . ; •
defines the number of

hits •- i
The bitwise left-shift operator («) shifts the hits in the first operand to the
left as many hits as is defined by the second operand. This effectively mul-
tiplies a number by 2”, where n is the number of hits to be shifted.
The bitwise right -shift operator (») shifts the hits in the first operand to
the right as many hits as is defined by the second operand. This effectively
divides a number by 2”, where n is the number of bits to be shifted.
The C language has no operator to rotate hits in a data item. Rotation can
be accomplished by or ing a shift-left operation and a shift - right operation.
14.9 Practice Sets

Review Questions
1. Ihe result of bitwise and 'ing of two hits is 1 if
.
a both hits are Os .
.
b both hits are Is .
c. hits are different
2. I he result of bitwise inclusive or ’ing of two hits is 0 if

a . both bits are Os.
b. both hits are Is.
.
3 The result of bitwise exclusive or ’ing of two hits is 1 if
a. both hits are Os.
b. both bits are 1 s.
4. To change every bit in a variable from a 1 to a 0 or from a 0 to a 1, we
need to use the operator.
a. &
b. |
c . A
d. -
.
5 The operator can he used to turn masked hit on.
.
a &
b. |
c. A
d. ~
6. The operator can he used to turn masked hit off.
a. &
b. |
c. A
d. -
7. The operator can be used to flip masked bit .
a. &
I, . |
c. A
d. ~
S. I he bitwise - operator is used to divide a data item by
a power of 2.
a. &
b. |
c. A
d. »
e. «
9. In program design , a function that collects data that are directed toward
a central transform is classified as a ( n ) module.
a. afferent
b. efferent
c. input
d . output
e. transform
to use the
10. To find the first address in a block of address, we need
operator with the network mask.
a. &
b. |
c. A
d. »
e. «
i block of addresses
, we need to use the
I 1 . To find the last address in a of network mask.
operator with the complem ent
a. &
b. i
c.
d. »
e. «
Exercises in hexadecim al:

of the following binary numbers
12 . Show the value of each
a. 11011001
b. 11111111
c. 1111000100011111
d. 000000 001110 0001
.
13 Show the value of each of the following hexadecimal numbers in binary:
a. 0x37
I). OxAB
C. 0x0237
d. 0xA234
14. Determine the result of the following operations (operands are in binary ):
a . -11011001
-
b. 11111111
.
c -11110001
-
d. 00000000
1 3. Determine the result of the following operations (operands are in binary ):
a . 11111100 & 00111111
b. 11111111 & 10101010
c. 00000000 & 11111001
d. 10101010 & 11100001
16. Determine the result of the following operations ( operands are in binary ):
a . 11111100 | 00111111
b. 11111111|10101010
c. 00000000 j 11111001
d . 10101010 j 11100001
1 7. Determine the result of the follow ing operations (operands are in binary ):
a . 11111100 A
00111111
b. 11111111 A
10101010
c. 00000000 A
11111001
d . 10101010 A
11100001
18. When numbers are represented in hexadecimal, we can simplify the
result ol one’s complement ( ~ ) operation . The result of complementing
each digit is 13 minus the value of that digit ( 3 = ~ A ) . Using the above
short -cut rule, find the result of the follow ing operations:
a. 0x02Al -
b. 0XF305 -
C. ~0xE2 A 8
d. OxOFFA -
19. When numbers are represented in hexadecimal, we can use some rules
.
that simplify the bitwise and ( & ) operation in three cases If one of the
digits is 0, the result is 0 (0 = 5 & 0). If one of the digits is F, the result is
-
the other digit ( 3 = I & 3). II two digits are the same, the result is one o!
them. ( A = A & A ). Using the above short-cut rules, find the result ol the
following operations:
a. 0xC201 & 0x02AF
b. 0xA240 & 0xF005
C . 0 xE 205 & 0 xE 2 A 0
d . 0 xA 531 & OxOFF O
.
20 \ \ hen numbers are represented in hexadecimal , we can use some rules
that simplify the bitwise inclusive or ( |) operation in three cases. If one
of the digits is 0, the result is the other digit ( 5 = 5 0 ). If one of the dig-
its is F, the result is F ( F = F | 3). It two digits are the same , the result is
one ol them . ( A = A | A) . Using the above short-cut rules, find the result
of the following operations:
.
a 0 xC 201 | 0 x02AF
.
b 0 xA 240 | 0 xF 005
.
C 0 xE 205 | 0 xE 2 A 0
d. 0 xA531 | OxOFFO
.
21 When numbers are represented in hexadecimal , we can use some rules
that simplify the bitwise exclusive or ( ) operation in three cases. II one
A
of the digits is 0, the result is the other digit ( 5 = 5 0). II one of the dig -
its is F, the result is 15 minus the other digit ( 9 F 6 ). II two digits are
= A
-
the same, the result is 0 (0 = A A A). Using the above short cut rules, find
the result of the following operations:
a . 0 xC 201 0 x 02 AF
A
b. 0 xA 240 A
0 xF005
c. 0 xE 205 0 xE 2 A 0
A
d. 0 xA 531 A
OxOFFO
:
22. Find the value of the following expressions
a. 0 x13 « 2
b . 0 x 13 » 2
.
c 0 x l 3A 2 « 4
d. 0 xB23E » 2
that turn on the most significant bit of
.
2.3 Show the mask and the operator
, the hits are numbered from the right , start -
an 8- bit integer. ( Remember .
in binary an d hexadecimal form
ing with 0.) Give the answer
most significant bit of
24. Show the mask and
the operator that turn off the , start -
, the hits are numbered from the right
an 8-bit integer. ( Remember
ing with 0. )
first and third bits of
complements the values of the
25. Find the mask that , the hits are number ed from the right , start -
an 8- bit integer. ( Remem ber
should be used?
with zero. ) What operator
ing
operator sets the second and
that when used with the &
26. Find the mask
integer to zero. ( Remem
ber, the bits are numbered
fourth bits of an 8-hit with 0.)
starting
from the right ,
27. Find the mask that when used with the | operator sets the second and
fourth hits of an 8- hit integer to 1 . ( Remember, the hits are numbered
from the right , starting with 0. )
. -
28 What is the result of the following expressions if a is an 8 hit unsigned
integer?
a. a & -a
b. a | a -
c. a a
d. a &
A
-OxFF
e. a & 0x00
f. a | OxFF
g. a|0x00
h. a A
OxFF
i. a A
0x00
29. Show the binary and hexadecimal representation of following polynomial
using 16- hit unsigned integer: x 1 * + x6 + x 3 + x + 1
30. Given two polynomials x 4 + x 3 + 1 and x6 + x3 + x + 1
a . Add them using modulo 2 arithmetic for coefficients.
b. Use 8- bit unsigned integers ( in hexadecimal) to represent each poly-
nomial and apply addition using exclusive or operations.
c . Compare the result of part a and b .
31 . Given two polynomials x5 + x2and x6 + x4 + x2 + 1
+ 1
a . Subtract the first polynomial from the second using modulo 2 arith -
metic lor coefficients.
b. Use 8- hit unsigned integers ( in hexadecimal ) to represent each poly-
nomial and apply subtraction using exclusive or operations.
c. Compare the result of parts a and b.
32. Given two polynomials x 12 + x 4 + x3 + 1 and x6 + x 3 + x + 1
a . Divide the first polynomial by the second using modulo 2 arithmetic
for coefficients. Find the quotient and the remainder.
b . Use 16 - bit unsigned integers ( in hexadecimal ) to represent each poly-
nomial and use Program 14 - 11 to find the quotient and the remainder.
c. Compare the result of part a and b.
Problems
.
33 Write a function that sets a specified bit in a 16- bit unsigned integer to 1 .
I he 1 unction is to have two parameters: The first parameter is the integer
to he manipulated and the second is the bit location , relative to the least
.
significant bit , that is to he turned on Your test program should start with
0 and turn on several hits in turn .
34 . Rework Problem 33 to turn off ( set to zero ) the specified bit.
.
35 Rework Problem 33 to flip the specified bit.
71
W
.
36 Write a function that flips
the hits in an 16-hit unsigned integer.
37. VY rite a function that
accepts two polynomials ( with 0 and 1
in the form of two 16-hit unsigned coefficient)
integers and returns the result of add-
ing the two in the form of a 16-hit
unsigned integer. Use hexadecimal
notations. Can the same function be used for
subtracting the two polyno-
mials? If yes, does the order of two polynomials
make a difference?
38. Write a function that returns the first (
leftmost) hexadecimal digit from a
32-bit unsigned integer.
.
39 Rework Problem 38 to return the last (rightmost ) digit.
.
40 Write a function that complements the first (
leftmost) hexadecimal digit
in a 32 -bit unsigned integer. Note that the complement of a
hexadecimal
digit is the 15 minus that digit.
.
41 Rework Problem 40 to complement the last (rightmost) digit .
42. Write a function that changes the first ( leftmost) hexadecimal digit i in a
32-bit unsigned integer. The function is to have two parameters: The first
is the integer to be manipulated, the second the replacement digit.
43 . Rework Problem 42 to change the last (rightmost ) digit.
Projects
.
44 Modify Program 14- 5 for checksum calculation at the sender site to a
checksum checker at the receiver site. The program receives a message
of an arbitrary length and a 16-hit checksum calculated at the sender
site. It then calculates a new checksum out of both. If the new checksum
is 0, it prints a message that the data is valid; otherwise, it prints a mes-
sage that the data is corrupted.
a First test your program with the string and checksum calculated in
.
Program 14- 5. The new checksum should be 0 .
b. Then change one of the characters in the string and test the program
. - .
again The result should be non zero
c. Note that the checksum itself can
be corrupted during the transmis-
program by changing one of the hexadecimal
sion. Also check your
checksum calculated the sender site. The result should
digits of the at
be non-zero again.
with the traditional checksum calculation
.
45 There is one major problem
If 16 -bit items are transposed in transmission,
discussed in the text two
error. The reason is that the traditional
the checksum cannot catch this each data item equally. In other
checksum is not weighted immaterial to the calculation
; it treats
item is
.
w ords, the order of data a new checksum
To catch transposition errors, J. Fletcher created
was devised to weigh each data
algorithm. The Fletcher checksum
. The algorithm uses eight bits (known as
item according to its position
16-bit checksum.
systems) and rcreates a
octets in telecommunication
The calculation uses modulo 256 ( 2s ) , which means the intermediate

results are divided by 256 and the remainder is kept. The algorithm uses
two accumulators, a and b . The first is the sum ol the data ; the second is a
weighted sum. These two accumulators are then used to form the check-
sum , a being the first byte and b being the second byte. Figure 14 - 13
shows the logic diagram for the Fletcher checksum.
The following equations represent the calculations shown in the
flowchart .
// Calculations in modulo 256

a = dx + d 2 + •• • + dn
b = ndx + (n - l )d 2 + • • • + dn
If , for example , dj and d 2 are swapped during the transmission , the

calculation of b at the receiver is different from the one done at the
sender. Write a function to calculate a Fletcher checksum . Test it with an
appropriate test driver.
Start
a =b=0
More Data
a=(
r
a+ d( )
mod 256
>
i
b = ( b + a)
mod 256
f
checksum = B A
b x 256 + a
Return 16-bit Checksum
FIGURE 14 - 13 Fletcher Checksum Design
46. I he Adler checksum, designed by Mark Adler, is a 32-bit checksum. It is

similar to the Fletcher checksum with three differences. First , calculation
is done on two bytes at a time to produce a 32- bit checksum . Second , a is
initialized to 1 instead of 0. Third , the modulus is
i a prime number, 65 , 521 ,
which is the largest prime number less than 65 , 536 ( 232 ) . It has been
proved that a prime modulo has a better detecting capability in some com -
binations of data. Figure 14-14 shows a simple, though inefficient , algo-
rithm in flowchart form .
Start
a= 1
b=0
-V
/ More Data v\—
T
a = ( a + d, )
mod 65521
b = ( b + a)
mod 65521
checksum = B B A A
b x 65536 + a
32-bit Checksum
Return
FIGURE 1 4 - 1 4 Adler Checksum Process
B is the weighted sum of data

It can he proved that the accumulator
items. We have
// Calculations in modulo+ 65521

1
= d2 + d2 + • • • + dn
a
b = nd2 + ( n - 1) d2 +
... + d„ + n
an Adler checksum.
Write a program to calculate
to create a digest of a
file or a message to protect the
47 In security, we need . A sender can send
changes during transmission
message from malicious then uses the
digest to a receiver. The receiver
the message with the to create a new digest. If the received digest is
key
same process and the sent digest , the message has not been changed
the
exactly the same as ; probability that two mes-
are one- to-one functions secret key between the
because the digests is extremely low . The
sages create the same digest to create a phony
does not allow an intruder
sender and the receiver from the true sender. The
message and digest and
pretend that is coming
y simplified version of a pro-
8 an extremel
e PimirpIT - 15 shows
14
diagram in rigurc
.
cess that creates a
digest J A
Block
128-bit (16-character ) block
1 } 1
32-bit 32-bit 32-bit 32-bit
T l f T
- -
Compress +||« HI*
1 \ \ I
16-bit 16 - bit 16-bit 16-bit
r1
!
^ 64-bit
1
f
32-bit key
H-
64-bit key
Exclusive or A
H
\ Expand
64-bit
Swapping
£ t
64-bit
\
Compress
32-bit
Digest
FIGURE 1 4 - 1 5 Simplified Digest Process
a. Each 16-character block is divided into four 32-bit integers.

b. Each 32-bit integer is compressed into a 16-bit integer. The compres -
sion removes the first and the last 8 hits.
c. I he four 16-bit blocks are combined to make a 64-bit block.
d. A 32-hit key is expanded into a 64-hit key by adding 16 I s at the
beginning and 16 Is at the end.
e. I he 64 -bit key is exclusive or ed with the block from part c.
E I he resulting 64-bit is swapped from the middle. The right 32 hit goes
to the left and the right 32 bits goes to the right ,
g. I he result is compressed to create a 32-bit digest by removing the first
and the last 16 bits.
Your program needs to use a main function, two compress functions,
.
an expansions function, a swap function The main function accepts a
16-character string and one 32 - bit key from the keyboard and create
and print a 32-bit digest.
T1
4 S. C ombine Programs 14-9 and 14 - 10 and make them interactive. The new
program accepts an address in the form x.y.z. t , and an integer prefix
between 0 and 32 and creates the network address ( first address ) , broad -
cast address ( last address ), the mask. The program also prints the total
number of addresses in the block. This can be done by first changing the
first and last address to a 32 - bit integer and then finding the difference
plus 1.
49. Write a program that , given the first address in a block and the number
of addresses in the block , the program can find the last address and the
corresponding mask. Note that there is only one restriction to this prob -
lem : the first address must divide the number of addresses evenly. Your
program must check for this restriction .
Aside from some advanced C concepts
and
future studies, we have covered the basics of functions that we leave for your
we turn our attention to a concept the C language. In this chapter,
that is pervasive throughout computer
ence: lists. sci
A list is a collection of related data. We can
ries: linear lists and non -linear lists. Linear
divide lists into two catego-
lists are divided into general lists,
stacks , and queues. Non - linear lists are divided
into trees and graphs. We
define and discuss each of these lists briefly in this chapter . Figure 1 5- 1 con -
tains the breakdown of lists.
Lists
Linear Non-Linear
General
) Stacks | Queues Trees Graphs
FIGURE 15- 1 Lists
Objectives
To introduce the basic concepts of linked lists
To introduce the basic concepts of stacks
To introduce the basic concepts of queues
To introduce the basic concepts of tree structures
To introduce the basic concepts of graph structures
927
-
928 Section 15.1 List Implementotions
15.1 List Implementations

The C language does not provide any list structures or implementations.
When we need them , we must provide the structures and functions for them.
Traditionally, two data types , arrays and pointers, are used for their imple-
mentation . The array implementation uses static structures that are deter -
mined during the compilation or while the program is running. The pointer
implementation uses dynamically allocated structures known as linked lists.
Array Implementation
In an array, the sequentiality of a list is maintained by the order structure of
elements in the array ( indexes ) . Although searching an array for an individual
element can be very efficient , addition and deletion of elements are complex
and inefficient processes. For this reason arrays are seldom used , especially
when the list changes frequently. In addition , array implementations ol lists
can become excessively large, especially when there are several successors lor
each element .
Linked List Implementation

A linked list is an ordered collection of data in which each element contains
the location ol the next element or elements. In a linked list , each element con -
tains two parts: data and one or more links . The data part holds the application
data — the data to be processed. Links are used to chain the data together. They
contain pointers that identify the next element or elements in the list .
We can use a linked list to create linear and non -linear structures. In lin -
ear lists , each element has only zero or one successor. In non - linear lists,
each element can have zero, one , or more successors.
rhe major advantage of the linked list over the array is that data are eas-
ily inserted and deleted . It is not necessary to shift elements of a linked list to
make room for a new element or to delete an element. On the other hand ,
because the elements are no longer physically sequenced , we are limited to
sequential searches: 1 we cannot use a binary search . When we examine trees,
we discuss the binary search tree , which allows for easy updates and efficient
searches.
f igure 15- 2 ( a ) shows a linked list implementation of a linear list . The
link in each element , except the last , points to its unique successor; the link
in the last element contains a nidi pointer, indicating the end ol the list ,
figure 15- 2 ( b ) shows a linked list implementation of a non - linear list . An ele-
ment in a non -linear list can have two or more links. I lere each element con -
tains two links, each to one successor. Figure 1 5 - 2(c ) contains an example ol
an empty list , linear or non - linear. We define an empty list as a pointer with a
null value.
In this section , we discuss only the basic concepts for linked lists.
1 . Sequential and binary searches are discussed in Chapter 8.

HI
list data link data link data link data link
( a) Linear list
list
1zlink \
data link\
A
a list
(c ) Empty list
X
link data link link data link
(b) Non-linear list
FIGURE 15 - 2 Linked Lists
Nodes
In linked list implementations, the elements in a list are called nodes A node .
is a structure that has two parts: the data and one or more links. Figure I 5 -3
shows two different nodes: one for a linear list and the other for a non-
linear list.
The nodes in a linked list are called self-referential structures. In a self-
referential structure, each instance of the structure contains one or more
.
pointers to other instances of the same structural type In Figure 15 - 3, the
colored boxes with arrows are the pointers that make the linked list a self-
referential structure.
(a ) Node in a linear list
(b) Node in a non-linear list
FIGURE 15 - 3 Nodes
be a single field, multiple fields, or a struc -

The data part in a node can
fields , but it always acts as a single field. Figure 15 - 4 ,,
ture that contains several
of a linked hst. The upper-left node con a ns a
hows,,
,
three designs for a node The upper-right node is more common It
single. ntld,, number, and
a link.
..
an id and grade points ( grdPts —and a
)
, "““,,”, 7 , ,
*****
|,hei «
"„r
'Jetl
d
h «
which hen pu In n he
i
The field are defined
definition of » node rucu e. „ „
T?Tz
'
930 Section 15.2 General Linear Lists
Node with Node with

' Three Data
.One Data Field
Fields .
^
number name id grdPts
Structure
in a Node
name address phone
FIGURE 15 - 4 Linked List Node Structures
Pointers to Linked Lists

A linked list must always have a head pointer. Depending on how we use the
list , we may have several other pointers as well . For example , if we are going
to search a linear list , we will need an additional pointer to the location
where we found the data we were looking for. Furthermore, in many struc-
tures, programming is more efficient if there is a pointer to the last node in
the list as well as a head pointer.
15.2 General Linear Lists

A general linear list is a list in which operations , such as retrievals, inser-
tions , changes , and deletions, can be done anywhere in the list , that is , at
the beginning, in the middle, or at the end of the list . We use many differ-
ent types of general lists in our daily lives . We use lists of employees , stu -
dent lists , and lists of our favorite songs. When we process our song list ,
we need to be able to search for a song , add a new song, or delete one we
gave away. I his chapter describes how we can maintain and process gen -
eral lists .
lo work with a linear list , we need some basic operations that manipulate
the nodes. For example, we need lunctions to insert a node at the beginning or
end ol the list , or anywhere in the middle ol the list . We also need a function
to delete a node from the list . Finally, we need a function to find a requested
node in the list. Given these primitive functions we can build functions that
process any linear list .
Insert a Node
We use the following four steps to insert a node into a linear list :
.
I Allocate memory for the new node .
2 . Determine the insertion point that is , the position within the list where
—
the new data are to be placed . Fo identify the insertion position, we need
to know only the new nodes logical predecessor ( pPre ) .
MIS
3. Point the new node to its successor.

M
4. Point the predecessor to the new node.
As seen in step 2 above, to insert a node into a list ,

we need to know the
location ol the node that precedes the new node ( pPre ) . This pointer can
be
in one o* two states: ll can contain the address of a
node, or it can be null .
II the predecessor is null , then we are inserting to an empty list or at the
beginning ol the list . II it is not null , then we are inserting somewhere alter
the first node that is, in the middle ol the list —or at the end of the list . A
list and its pointers are shown in Figure I 5 - 5.
_
l j Hdata
pList
i
241link+Hdata
181link+I Hdata 39
i i
link
••• H 57 N
data link
pPre Add t0 empty list or add at be9innin9 list
pPreQ* Add in middle of list or add at end of list
FIGURE 1 5 - 5 Pointer Combinations for Insert
We discuss each of these situations in turn .
Insert into Empty List

When the head of the list is null , then the list is empty. This situation
is
necessary insert a node to an empty list is
shown in Figure 15 - 6 . All that is to
list head pointer to the address ol the new node and make sure
to point the
null to set the link field of
that its link field is null. We could use a constant
null contained in the list head pointer. The rea -
the new node, but we use the .
apparent in the next section
son for this will become
pNew ->link = pList ;

pList = pNew ;
After Add
Before Add ~
pNew [— 39
pNew I
pList
pList
pPre | |
^
pPre
^
FIGURE 15- 6 Insert Node to Empty List _ _
1
/
mT
^fy
932 Section 15.2 Generol Linear Lists
The* statements to insert a node into an empty list are shown below.
-
pNew >link = pList; // set link to NULL
pList = pNew ; // point list to first node
Note the order of these two statements. We must first point the new node
to NULL , then we can change the head pointer. If we reverse these state-
ments, we end up with the new node pointing to itself, which would put our
program into a never-ending loop when we process the list .
Insert at Beginning
We insert a node at the beginning of the list whenever it must be placed
before the first node of the list . We determine that we are inserting at the
beginning of the list by testing the predecessor pointer (pPre). If it is null,
then there is no predecessor, so we are at the beginning of the list.
To insert a node at the beginning of the list, we simply point the new
node to the first node of the list and then set the head pointer (pList) to
point to the new first node. We know the address of the new node. I low can
we find the address of the first node currently in the list so we can point the
new node to it ? The answer is simple: The first node’s address is stored in the
head pointer (pList).
If you compare these two statements to the statements to insert into an
empty list, you will see that they are the same. This is because, logically,
inserting into an empty list is the same as inserting at the beginning of a list.
We can therefore use the same logic to cover both situations. Inserting at the
beginning of the list is shown in Figure 1 5 - 7.
pNew ->link = pList ;

pList = pNew;
Before Add After Add

pNew 39 pNew 39
pLisl 75
“V J
124 pList 75 - 124
pPre | |
^
ppre
FIGURE 15 - 7 Insert Node at Beginning
Insert in Middle
When we insert a node anywhere in the middle of the list, the predecessor
( pPre) is not null. Io insert a node between two nodes, we must point the
new node to its successor and then point the predecessor to the new node.
Again, the address of the new node’s successor can be found in the predeces-
sors link field. I he statements to insert a node in the middle of the list are
shown in Figure 15 -8.
pNew->link = pPre->link;
oPre->link = pNew;
~ ~
pNew \\*\
~ 96 [ pNew
£
[ 75 | -) *fl24 -U-.
"
Tn n124
pPre pPre
Before Add After Add
FIGURE 1 5 - 8 Insert Node in Middle
Insert at End
When we insert at the end of the list , we only need to point the predecessor
to the new node. There is no successor to point to. It is necessary, however to
,
set the new node s link field
'
to NULL . The statements to insert a node at the
end of a list are shown below .
pNew-> link = NULL;
pPre-> link = pNew ;
for inserting at the end ,

Rather than have special logic in the function
linear list structure . We know
however we can take advantage of the existing
of NULL . If we use this pointer
that the last node in the list has a link pointer exactly the same as the code
rather than a constant , then the code becomes
The revised code is shown in F gure 15-9.
for inserting in the middle of the list . .
.
Compare it to the code in Figure -
15 8
pNew - >link = pPre ->link;

-
pPre >link = pNew;
\M \ \ pNew £ 134 X
pNew £ +
J 75 - J 124 ^
pPre
pPre After Add
Before Add
FIGURE 15 - 9 Insert Node at End
a node into
".
inserted ( item). We
m . J
the data to be
( malloc ) and adjust the link pointers appropriately. Since we might insert a
node before the beginning of the list , we also need to he able to change the
contents of the head pointer. We ensure that the head pointer is always cor-
rect by returning it . The function declaration for insertNode is shown below.
NODE* insertNode ( NODE* pList , NODE* pPre , DATA item );
The complete function is shown in Program 1 S - 1 .
PROGRAM 15 - 1 Insert a Node

1
2 This function inserts a single node into a linear list.
3 Pre pList is pointer to the list ; may be null
4 pPre points to new node's predecessor
5 item contains data to be inserted
6 Post returns the head pointer
7 */
8 NODE* insertNode ( NODE* pList , NODE* pPre , DATA item )
9 {
11 NODE* pNew ;
12
13 // Statements
14 if (!( pNew = ( NODE* )malloc(sizeof( NODE ))))
15 printf( " \aMemory overflow in insert\n " ),
16 exit ( 100 );
17
18 -
pNew >data = item ;
19 if ( pPre == NULL )
20 {
21 // Inserting before first node or to empty list
22 -
pNew >link = pList;
23 pList = pNew ;
24 > // if pPre
25 else
26 {
27 // Inserting in middle or at end
28 -
pNew >link = pPre->link ;
29 -
pPre >link = pNew ;
30 > // else
31 return pList;
32 > // insertNode
y
Program 1 5 - 1 Analysis We have discussed all the logic in this function except the memory allocation. Recall
'
that ma/ oc returns either on address in the heap or null when there is no more room
m the heap . Therefore, we need to test to see if memory is available.
When memo7 is exhausted, we have an overflow condition. The action taken
depends on the application being programmed; however, the general course of action
is to abort the program. We assume the general case in the algorithm; if null is returned
from malloc, we print a message to the user and exit the program.
With the idea in mind that all code needs to be tested, how can we test the over -
flow logic? We could use the brute- force method-insert data until the list overflows —
but that could take a long time. We tested it by inserting an array of 10,000 long dou-
ble variables to the data structure. On our system, we got an overflow on the fifth node
and the logic was tested.
The technique used to call insertNode is important, since it returns the head of
the list under all circumstances. The call should assign the return value to the list head
pointer. This ensures that when the head of the list changes, the change will be reflected
in the calling function. A typical call to insert data into a student list is coded below.
Note that the head pointer (stuList) is found both in the assignment and as an
actual parameter.
stuList = insertNode (stuList, pPre , stuData );
Delete a Node
from the linear list by
When we delete a node we logically remove the node
deleting it from the heap.
changing various link pointers and then physically the first node,
The delete situations parallel those for insert . We can delete
end node of a list . As you will see below, these
any node in the middle, or the first node, and
ons: Delete the
three situations reduce to only two combinati
delete any other node. (identified
must first locate the node itself
To logically delete a node, we
by pPre . We) discuss local ,
on con-
bv pcur)and its predecessor (identifiedthat we want to delete, we can s ,mpjy
cepts shortly. .After we
locate the node
to point to the deleted
node s successor We
change its predecessor s link field
’
. We need to be concerned, however, about
list so we
the only node in a list.
must be careful that in this case the head po
.
then recycle the node using /*Deleting the only node results man empty
deleting ,
nter s set to NULL .
Delete First Node
null, we
936 Section 15.2 Generol Linear Lists
pList = pCur ->link ;
~~
[ — I -(-
pList
~T ^|
^
75 - 124 pList ; (Recycled) 124 •••
^
ppre pCur pPre
Before Delete After Delete
FIGURE 15 -10 Delete First Node
I he statements to delete the first node are shown below.
pList = pCur-> link ;

free( pCur);
If you examine this logic carefully, you will note that it also handles the
situation when we are deleting the only node in the list . If the first node is
the only node, then its link field is null. Since we move its link field ( a null )
to the head pointer, the result is by definition an empty list .
General Delete Case

We call deleting any node other than the first a general case, since the same
logic handles deleting a node in the middle of the list and deleting a node at
the end of the list . For both of these cases , we simply point the predecessor
node, identified by pPre, to the successor of the node being deleted . I he
node being deleted is identified by the current node pointer, pCur. Its succes -
sor is therefore pCur > link. -
Deleting the last node is handled automatically. When the node being
deleted is the last node of the list , its null pointer is moved to the predeces-
sor 's link field , making the predecessor the new logical end of the list . After
the pointers have been adjusted , the current node is recycled . The delete gen -
eral case statements are shown below.
-
pPre > link =
free ( pCur);
-
pCur >link ;
I he general case is shown in Figure 15- 11 .
Function to Delete a Node

I he complete logic to delete a node is shown in Program 15 - 2. It is given a
pointer to the head ol the list , the node to be deleted , and the delete nodes
predecessor. After deleting and recycling the node, it returns the pointer to
the beginning of the list .
mn
Yl
Before Delete
pPre
f
75
pCur [J
- -
- 96 - 124
pPre- >link = pCur->link ;
75 ( Recycled ) 124
pPre
of pcuroT After Delete
FIGURE 15 - 11 Delete — General Case
PROGRAM 15 - 2 Delete a Node

1 j* ====== deleteNode
2 This function deletes a single node from the link list.
3 Pre pList is a pointer to the head of the list
4 pPre points to node before the delete node
5 pCur points to the node to be deleted
6 Post deletes and recycles pCur
7 returns the head pointer
8 */
9 NODE* deleteNode ( NODE* pList
, NODE* pPre, NODE* pCur )
10 {
11 // Statements
12 if (pPre = = NULL)
13 // Deleting first node
14 pList = pCur >link ; -
15 else
16 // Deleting other nodes
17 -
pPre >link = -
pCur >link;
18 free ( pCur );
19 return pList ;
20 } // deleteNode
. A'iscu ssion The first and most important is that the
fill&L .- . . * *
^. Unction
men '
Program 1 5 - 2 Analysis Three points in this functioniden |
is ca ec || that the
. It assumes
ef f they aren't, then the ,prog
deleted must be ram
node to be $
will b
|
, ,
ong.|l b b «r a pro
mo
gram we placed the recycle
ed it to the end of the I
.
* i
i
function. When the same statements appear in both the true and false blocks of a selec-
tion statement, they should be moved out of the selection logic . This is the same concept
as factoring common expressions in algebra . The result is a program that is smaller
and easier to maintain.
Finally, we return the list. This is necessary because it is possible for the first node of
the list to be deleted, which results in a new node being the head of the list.
Locating Data in Linear Lists

When we insert and delete data in a linear list , we must first search the list .
To insert a node , we search to identify the logical predecessor of the new
node in its key sequence. To delete a node , we search to identify the location
of the node to he deleted and its logical predecessor.
Search Linear List

Both the insert and delete for lists require that we search the list . For inserts,
we need to know the predecessor to the node to be inserted ; for deletes, we
need to know the predecessor to the node to he deleted . .Also , to locate a node
for processing , such as adding to a count or printing its contents , we need to
know its location . Although we could write separate search functions for all
three cases, the traditional solution is to write one search that satisfies all
requirements . This means that our search must return both the predecessor
and the current ( found ) locations .
Generally, lists are maintained in key sequence . A key is a Held within a
structure that identifies the data . Examples of keys would include student
number to identify data about students and part number to identify' data
about parts . In this text , we limit our discussion of lists to key - sequenced
lists , that is , lists in which each key in the list is less than or greater than the
key that follows it .
Io search a list on a key, we need a key Held . For simple lists , the key and the
data can he the same Held . More complex structures require a separate key field .
A generic linked- list node and key structure are seen in the following example.
typedef int KEY TYPE; // Application Dependent
typedef struct
{
_
KEY TYPE key ;
// Other Data Fields
> DATA ;
typedef struct nodeTag

{
DATA data;
struct nodeTag* link ;
> NODE;
n
~^
Given a target key, the search attempts to locate the requested node in
t le linear list. If a node in the list matches the target value , the search
.
returns true ; if no key matches , it returns false The predecessor and current
pointers are set according to the rules in Table 15 - 1 .
Condition
target < first node NULL
pPre
2
first node
Return
0
target == first node NULL first node 1
first < target < last largest node < target first node > target 0
target == middle node's predecessor equal node 1

node
target == last node last's predecessor last node 1
target > last node last node NULL 0
TABLE 1 5 - 1 Linear List Search Results
Each of these conditions is also shown in Figure 15 - 12

.
pList
BKOB * QBr DrQI

- "
o
pPre pCur
ho
pPre pCur
oh
pPre pCur
target == last
target == first target == middle
)
Successful Searches (Return true
pList
> pPre < pCur
HQ RRElh PHD! ~
o pPre pCur
a
pPre pCur
pPre pCur target > last
target > *pPre
target < first
target < *pCur
UnSUC
^ ^
; Tseareh (Return false )
FIGURE 15 - 12 Search Results

Since the list is in key sequence, we use a modified version of the

sequential search. Knuth 2 calls this search “ Sequential search in ordered
table.” We simply call it an ordered sequential search.
We start at the beginning and search the list sequentially until the target
value is no longer greater than the current node’s key. At this point , the target
value is either less than or equal to the current node’s key while the predeces-
sor is pointing to the node immediately before the current node. We now use
the current node pointer (pCur)to test for equality and set the return value to
true if the target value is equal to the list value or false if it is less ( it cannot be
greater ) and terminate the search . The eode lor this search is shown in
Program IS - 3 .
PROGRAM 15 - 3 Search Linear List

1 /* ==================== searchList ==================
2 Given key value , finds the location of a node
3 Pre pList points to a head node
4 pPre points to variable to receive pred
5 pCur points to variable for current node
6 target is key being sought
7 Post pCur points to first node with >= key
8
9
-
-or null if target > key of last node
pPre points to largest node < key
10 - -
or null if target < key of first node
11 function returns true if found
12 false if not found
13 */
14 bool searchList ( NODE* pList, NODE** pPre ,
15 NODE** pCur , KEY TYPE target )
16 {
18 bool found = false;
19
20 // Statements
21 * pPre = NULL;
22 *pCur = pList;
23
24 // start the search from beginning
25 while (* pCur != NULL && target > (*pCur) >data.key)
26 {
-
27 * pPre = * pCur ;
28 *pCur = (* pCur)->link;
con tin net
.
2 Donald L . Knuth , / he Art of Computer Programming
, Volume 3 Sort ing and Searching Second
Edition . ( Boston , Mass.: Adilison Wesley, 1979 ) , Algorithm
- T, p. 398.
PROGRAM 1 5 - 3 Search Linear List (continued)

29 } // while
30
31 if ( * pCur && target = = ( * pCur )-> data.key )
32 found = true;
33 return found ;
34 } // searchList
Program 1 5 - 3 Analysis Examine the while statement at statement 25 carefully. Note that there are two tests.
The first test protects us from running off the end of the list; the second test stops the
loop when we find the target or, if the target doesn't exist, when we find a node
larger than the target. It is important to make the null list test first. If the function is at
the end of the list, then pCur is no longer valid. Testing the key first gives unpredict-
able results.
We can make the search slightly more efficient if we use a rear pointer. With a rear
pointer, we can test the last node before the loop to make sure that the target isn't larger
than its key value. If the target is larger, we simply exit the search function after setting
the predecessor pointer to the last node and the current pointer to null. Once we know
that the target is not greater than the last node, the loop doesn't need to test for running
off the end of the list.
Traversing Linear Lists

and examine each node in
Algorithms that traverse a list start at the first node
. Several different types of
ion until the last node has been processed
succession
examples are printing the list , counting
functions use list traversal logic . Some Held in the node, or cal-
list, totalling a numeric
the number of nodes in the
, any application that requires process
-
culating the average of a Held . In fact 1 - 13 is a graphic representation of
. Figure 15
ing the entire list uses a traversal
a linear list traversal.
V’7 IX
~
-T* “* •" '
-
39
f /
LrJ .
..
pList
pWalker
FIGURE 15- 13 Linear List Traversal
in the pseudocode shown

a linear list is found
logic to traverse is used to guard
The basic
two concepts
. First , an event loop
incorporates processing the current
below. It . Second, after
overrunning the end of the list element .
against is advanced to
the next
node, the looping pointer i i ^
Algorithm traverse list

1 set pointer to the first node in list
2 while ( not end of the list )
1 process (current node )
2 set pointer to next node
3 end while
end traverse
Print Linear List

Program 15 - 4 uses the traversal logic to print a linear list .
PROGRAM 15 - 4 Print Linear List

1 /* Traverse and print a linear list.
2 Pre pList is a valid linear list
3 Post List has been printed
4 */
5 void printList ( NODE* pList)
6 {
8 NODE* pWalker;
9
10 // Statements
11 pWalker = pList ;
12 printf("List contains:\n");
13
14 while ( pWalker )
15 {
16
17 pWalker = pWalker->link ;
-
printf( "% 3d " , pWalker >data.key );
18 > // while
19 printf("\n " );
20 return ;
21 > / / printList
Program 1 5- 4 Analysis Study the while expression in statement 14 carefully. This is the most common form of
pointer evaluation for a linear list. C guarantees that the evaluation of a null pointer
is false, and all other pointer values evaluate to true. This expression is therefore
equivalent to
while ( pWalker ! = NULL )
where pWalker is a pointer to current node.

The other important piece of code in the traversal is the loop update in statement 17.
pWalker = -
pWalker >link;
TO
This statement advances us through the list. If we forget it, we are in a permanent
loop. Since pWalker points to the current node, we need to change it to advance to
the next node. This is done by simply assigning the link field to pWalker. When we.
finally arrive at the end of the list, the link is null, which, when assigned to pWalker,
terminates the while loop.
Linear List Average

Let s write one more linear list traversal algorithm , one that averages the val -
ues in a linear list . To pass a linear list to a function , we only need to pass the
value of the head pointer. Look at Program 15- 5. The actual data structure is
not shown , but the function assumes that data . key is an integer field .
PROGRAM 15 - 5 Average Linear List

1 /* This function averages the values in a linear list.
2 Pre pList is a pointer to a linear list
3 Post list average is returned
4 */
5 double averageList (NODE* pList)
6 {
8 NODE* pWalker ;
9 int total;
10 int count;
11
12 // Statements
13 total = count = 0;
14 pWalker = pList;
15 while (pWalker)
16 {
total += pWalker >data
.key;
-
17
18 count++;
19 pWalker = pWalker
>link;
-
20 } // while
/ count;
21 return (double)total
22 > // averageList
"walks" us through
i i
descriptive pointer, pWalker, that

Sram 15-5 Analysis To traverse the list, we used the against ever using the list
Although we could have used pList, we advise. Then , ,f you need to modify
the list . . Keep it in its
original state it's the small
is a small paint but
logic ,
pointer far any function its starting point. This
will still know
the function , you easy to maintain.
points that make a program
Building a Linear List

vve wrote three
low-level functions
the list and average. It
—
for a list insert ,
’s time
In the previous sections applicatio ns |> rint
delete, and search
to p ut them all
a;nd two —
together in a program
.
n
To build the list , we need to get the data, create the node, determine the
insertion position in the list, and then insert the new node I he design is .
shown in Figure 15 - 14 .
list pointer
m
buildList
list pointer
getData searchList
£
insertNode
I
FIGURE 15 - 14 Design for Inserting a Node in a List
Build List
Building a key-sequenced list uses the search function to determine the
insertion position. It uses the design shown in Figure 15 - 14. The implemen-
tation is shown in Program 15 - 6.
PROGRAM 15 - 6 Build List

l ===== buildList ====
2
3
This program builds a key sequenced linear list.
Pre
-
filelD is file that contains data for list
4 Post list built
5 returns pointer to head of list
6 */
7 NODE* buildList ( char* filelD )
8 {
10 DATA data ;
11 NODE* pList ;
12 NODE* pPre;
13 NODE* pCur ;
14 FILE* fpData ;
15
16 // Statements
17 pList = NULL ;
18
19 fpData = fopen( filelD , "r");
20 if (!fpData )
21 {
22 printf( " Error opening file %s\a\n " , filelD );
contimiec
PROGRAM 15 - 6 Build List (continued )
23 exit (210);
24 } // if open fail
25
26 while (getData (fpData, &data
))
27 {
28 // Determine insert
position
29 searchList (pList, &pPre, &pCur, data.key
30 );
pList = imsertNode(pList, pPre
, data);
31 } // while
32 return pList;
3 3 } // buildList
Program 15 - 6 Analysis Note that in statement 29, we do not test

for duplicate keys when search list returns. If
we want to prevent duplicates , we would test the return value and
print an error mes-
sage rather than insert the data when a duplicate key is detected .
Remove a Node
I IK* design to remove a node is also simple. After reading the key to he deleted ,
we search Lor its location and its predecessor location and then call the delete
node function to do the physical deletion . The design is shown in Figure 15- 15.
deleteKey
searchList
deleteNode
FIGURE 15 - 15 Design for Remove Node
The code is shown in Program 15 - / .
PROGRAM 15- 7 Delete Key

deleteKey ===
=
1 / * ===== === = === ====- ==
Delete node from a linear list ,
2
to the head of the list
3 Pre list is a pointer
Post node has been deleted
4 not found
5 or a warning message printed if
- - first node ( pList)
6 returns pointer to
7 */
PROGRAM 15-7 Delete Key ( continued )

8 NODE* deleteKey ( NODE* pList )
9 {
11 int key;
12 NODE* pHead ;
13 NODE* pCur ;
14 NODE* pPre ;
15
16 // Statements
17 printf( " Enter key of node to be deleted: " );
18 scanf ( " %d " , &key);
19
20 if (!searchList( pList , & pPre , &pCur , key ))
21 printf( "%d is an invalid key\a\n" , key );
22 else
23 pHead = deleteNode (pList, pPre, pCur );
24
25 return pHead ;
26 > // deleteKey
Linear List Test Driver

Now we need to test our functions. To test them , we write a test driver that
builds a list , then prints it , inserts more data into it , computes its average,
and finally deletes data from it . The design of our test driver is shown in
Figure 15 - 16.
testLinkList
I
§
buildList
I printList
I printList
J deleteKey printList
o
1
searchList
A insertNodeJ
averageList
I
\ >1
searchList insertNode searchList deleteNode
FIGURE 1 5 - 1 6 Link List Test Driver
The code for the test driver is shown in Program 1 5-8.

PROGRAM 1 5 - 8 Test Driver for Link List

1 /* Test driver for list
functions.
2 Written by:
3 Date:
4 */
8
9
10 _
typedef int KEY TYPE;
11 typedef struct
12
13
{
_
KEY TYPE key ;
14 } DATA ;
15 typedef struct nodeTag
16 {
17 DATA data ;
18 struct nodeTag* link;
19 } NODE;
20
22 NODE* insertNode ( NODE* pList, NODE* pPre , DATA item );
23 NODE* deleteNode ( NODE* List, NODE* pPre, NODE* pCur);
24 bool searchList (NODE* pList, NODE** pPre,
25 NODE** pCur, KEY TYPE target );
26 void printList ( NODE* pList);
27 NODE* buildList (char* filelD);
28 NODE* deleteKey ( NODE* pList);
29 bool getData ( FILE* fpData , DATA* pData );
30
,
31 double averageList ( NODE* pList )
32
34 {
36 NODE* pList;
37 NODE* pPre;
38 NODE* pCur;
39 DATA data;
40 double avrg ;
41 char more;
42
43 // Statements test driver\n\n " );
44 printf( " Begin list continued
'
PROGRAM 1 5 - 8 Test Driver Lor Link List ( continued )

45
46 // Build List
47 -
pList = buildList( "PI5 LIST.DAT" );
48 if (!pList)
49 {
50 printf( " Error building chron file\a\n " );
51 exit ( 100 );
52 > // if
53 printList ( pList );
54
55 printf( "\nlnsert data tests.Xn");
56 printf( " Enter key: ")?
57 scanf ( " %d " , &data.key );
58 do
59 {
60 if (searchList ( pList , & pPre, & pCur , data.key ))
61 printf( " Key already in list. Not inserted \n " );
62 else
63 pList = insertNode( pList , pPre , data );
64
65
-
printf( " Enter key < l > to stop: " );
scanf ("%d", &data.key );
66 > while ( data.key != 1 );
-
67 printList ( pList);
68
69 avrg = averageList( pList );
70 printf( " \ nData average: %.lf \n" , avrg );
71
72 printf( " \nDelete data tests.Xn " );
73 do
74 {
75 pList = deleteKey ( pList );
76 printf( "Delete another <Y /N>: " );
77 scanf (" %c" , &more );
78 > while (more == 'Y'| |more == 'y');
79
80 printList ( pList);
81 printf( " \nTests complete.\n ");
82 return 0;
83 > // main
Results:
Begin list test driver
List contains:
continued
rn
'
Chapter 15 Usts 949
PROGRAM 1 5 - 8 Test Driver for Link List (continued )

111 222 333 444 555 666 777
Insert data tests.

Enter key: 50
-
Enter key < l> to stop: 1
List contains:
-
50 111 222 333 444 555 666 777
Data average: 394.8
Delete data tests.

Enter key of node to be deleted: 50
Delete another <Y/N>: n
List contains:
111 222 333 444 555 666 777
Tests complete.
each of the functions carefully

Program 1 5 - 8 Analysis While the code is rather lengthy if you have studied interest are the test cases
you will find this program straightforward. Of greater for the inserts, we need to
Note that
needed to completely validate the programthe middle. Likewise, for the deletes, we
.
insert at the end , at the beginning , and in

in the middle. We also need to try to
delete the last node, the first node, and one
delete a node that doesn ' t exist .
15.3 Stacks restricted to

all additions and deletions are and then
\ stack is a linear list in which a stack
you insert a data series into
.
one end , called the top If . . Data input
..
as {5, 10, 15 , 20} is

data is reversed
remove it , the order of the attribute is why stacks are known
removed as {20 , 15 , 10 .
, 5} This reversing
.
( LIFO ) data structure
as the last in-first out our daily lives. We often talk of a
different types of stacks in
in which you can only add
We use many or
or a s tack of dishes. Any situation remove any object other
stack of coins .
is a stack If you want
to
objects above it. A graphic
remove an object at the,top you must first remove all
than the one at the top .
shown in Figure 15- 17
representation of a stack is us from designing a data structure that allows
Although nothing prevents , such as moving the item at the top of the
operations
us to perform other result would not be a stack.
stack to the bottom the
,
in which all insertions

A stack is a last in -
first out ( LIFO
restricted
) data structure
to one end, called
the top .
and deletions are
950 Section 15.3 Stacks
Top
~ ~
Stack of Coins Stack of Books Computer Stack
FIGURE 15 - 17 Stack
Stack Structures
Several data structures can be used to implement a stack. We implement the
stack as a linked list .
Data Structure
To implement the linked list stack , we need two different structures, a head
—
node and a data node. T he head structure contains metadata that is , data
—
about data and a pointer to the top of the stack. The data structure contains
data and a link pointer to the next node in the stack. The conceptual and
physical implementations of the stack are shown in Figure 15 - 18.
Head
Top
13-
count top
Data
nodes
(a) Conceptual (b) Physical
FIGURE 15 - 18 Conceptual and Physical Stack Implementations
Stock Head
Generally, the head lor a stack requires only two attributes: a top pointer and
a count of the number of elements in the stack. These two elements are
placed in a structure. Other stack attributes can he placed here also. A basic
head structure is shown in Figure 15- 19 .
Chapter 15 lists 951
typedef struct
{
count top int count;
Stack Head Structure struct node* top;
} STACK ;
typedef struct node
{
int data ;
data link struct node* link ;
Stack Node Structure } STACKNODE ;
FIGURE 1 5 - 1 9 Stack Data Structure
Stock Dota Node

I he rest of the data structure is a typical linked list data node . Although the
application determines the data that are stored in the stack, the stack data
node looks like any linked list node. For our example, we use numbers for
data. In addition to the data, the data node contains a link pointer to other
data nodes, making it a self-referential structure. In a self- referential struc-
ture, each instance of the structure contains a pointer to another instance of
the same structure. The stack data node is also shown in Figure 15 - 19 .
Stack Algorithms
The basic stack operations defined in this section should be sufficient to
solve basic stack problems. If an application requires additional stack opera
-
describe it and then

tions, they can be easily added. For each operation
we
develop it. , . ,
Although the implementation of a stack depends somewhat
on the imple-
with a stack head structure in
mentation language , it is usually implemented
15 - 19.
C. We use the design shown in Figure
Push Stack
the stack. The first thing we need to do
Push stack inserts an element into
^ .
en we push data into
lore »11«, : * attack .
is
..
simply assign
to the node cu
stack top pointer and add 1
^
push stack operation in which .
to
,
llack ,„1 . ,„d hen
„ K link
a new pointer ( pNew

) is used to identify the
stack
data to be inserted into the
7
_
pNew R I
^ data
123 I K pNew R
123 _
in < data link
stack stack
m n
count op
757 3
count op
757
data ink
data ink
„ I 357 I K r IWI K
data ink
data in <
( a ) Before ( b ) After
FIGURE 15 - 20 Push Stack Example
To develop the insertion algorithm using a linked list, we need to analyze

three different stack conditions: ( 1 ) insertion into an empty stack, ( 2 ) inser -
tion into a stack with data, and ( 3 ) insertion into a stack when the available
memory is exhausted.
When we insert into a stack that contains data, the new node’s link
pointer is set to point to the node currently at the top, and the stacks top
pointer is set to point to the new node. When we insert into an empty stack,
the new node ’s link pointer is set to null and the stack’s top pointer is set to
point to the new node. However, because the stack’s top pointer is null, we
can use it to set the new node ’s link pointer to null. Thus the logic for insert -
ing into a stack with data and the logic for inserting into an empty stack are
identical. Figure 15 - 20 shows the before and after conditions when we push
data into a stack.
As we saw with the linear list , a push can cause an overflow condition. II
it does, the function simply returns a false condition. The code is shown in
Program 15 - 9.
PROGRAM 15 - 9 Push Stack

1 / * ===
2 Inserts node into linked list stack.
3 Pre pStack is pointer to valid stack header
4 Post dataln inserted
5 Return true if successful
6 false if overflow
7 */
8 bool push (STACK* pStack , int dataln )
9 {
11 _
STACK NODE* pNew ;
12 bool success;
continues
TT
n
PROGRAM 15 9 -
Push Stack (continued)
13
14 // Statements
15 pNew = (STACK NODE _
*)malloc(sizeof (STACK
NODE)); _
16 if ( IpNew)
17 success = false;
18 else
19 {
20
21
-
pNew >data = dataln;
22
-
pNew >link - pStack >top;
-
23
-
pStack >top = pNew;
24
-
pStack >count++;
success = true;
25 } // else
26 return success;
27 > // push
Pop Stack
Pop stack sends the data in the node at the top of the stack hack to the call-
ing algorithm. It then adjusts the pointers to logically delete the node. After
the node has been logically deleted, it is physically deleted by recycling the
memory, that is returning it to dynamic memory. After the count is adjusted
by subtracting 1, the algorithm returns the status to the caller: if the pop was
successful, it returns true ; if the stack is empty when pop is called, it returns
.
false The operations for pop stack arc traced in Figure 15 - 21.
Recycled
-
r “i
dItPtr Q Hal E3 dItPtr Q > •-.I

^ data link data link
dT
stack stack
IDO top I 757 1 [3
count data link count top data link
c HZ
data linl
r ran KI
data link
(a) Before (b) After
FIGURE 15 - 21 Pop Stack Example
15 - 10.
The pop stack code is shown in Program
i
;
t
PROGRAM 15- 10 Pop Stack

l == pop ==
2 Delete node from linked list stack.
3 Pre pStackTop is pointer to valid stack
4 Post dataOut contains deleted data
5 Return true if successful
6 false if underflow
7 */
8 bool pop ( STACK * pStack , int* dataOut )
9 {
11 _
STACK NODE* pDlt;
12 bool success;
13
14 // Statements
15 -
if ( pStack >top ) // Test for Empty Stack
16 {
17 success =
true ;
18 *dataOut =
pStack->top->data ;
19 pDlt =
pStack >top ; -
20 -
pStack >top = ( pStack->top) >link ; -
21
22
-
pStack >count ;
free ( pDlt);
—
23 } // else
24 else
25 success = false ;
26 return success;
27 > // pop
Program 15 - 10 Analysis It is interesting to follow the logic when the last node is being deleted. In this case, the
result is an empty stack . No special logic is required; however, the empty stack is cre-
ated automatically because the last node has a null link pointer, which when moved
to top indicates that the stack is empty. A count of zero, which automatically occurs
when we decrement the count, is also an indication of an empty stack.
Stack Demonstration
lo demonstrate the push and pop stack algorithms, we write a program that
inserts random numbers into a stack . After the numbers are inserted, they are
popped and printed. When the stack is empty, the program terminates. I he
design for this program is shown in Figure 15 - 22.
I he node declarations, function declarations, and test driver are con-
tained in Program 15 - 1 1 .
Basic
Stack
insert
Data
push pop
FIGURE 15 - 22 Design for Basic Stack Program
PROGRAM 1 5 - 1 1 Simple Stack Application Program

e the
1 /* This program is a test driver to demonstrat
and pop functions.
2 basic operation of the stack push
3 Written by:
4 Date:
5 */
9
11 typedef struct node
12 {
int data ;
13
struct node* link ;
14
15
_
} STACK NODE;
16
17 typedef struct
18 {
count;
STACK_NODE
19 int
* top;
20
21 } STACK;
22
23 // Function Declarations );
(STACK * pStack
24 void insertData * pStack ;
)
(STACK
25 void print int dataln );
(STACK * pList r
26 bool push , int* dataOut );
(STACK* pList
27 bool pop
28
)
29 int main ( void
PROGRAM 15- 11 Simple Stack Application Program ( continuedI

30 {
32 STACK* pStack ;
33
34 // Statements
35 printf( " Beginning Simple Stack ProgramXn" );
36
37 pStack = malloc(sizeof( STACK ));
38 if (!pStack)
39 printf( " Error allocating stack" ), exit( 100 );
40
41 pStack->top = NULL ;
42 -
pStack >count = 0 ;
43 insertData ( pStack );
44 print ( pStack );
45
46 printf( " \nEnd Simple Stack ProgramXn " );
47 return 0;
48 } // main
Results:
Beginning Simple Stack Program
Creating numbers: 854 763 123 532 82 632 33 426 228 90
Stack contained: 90 228 426 33 632 82 532 123 763 854
End Simple Stack Program
Program 1 5 - 1 1 Analysis To verify that the program works correctly, we print the numbers as they are gener -
ated . This allows us to verify that the stack output was in fact correct. Note that this
simple program verifies the LIFO operation of a stack.
Insert Data
the insert data function loops while creating numbers and inserting them
into the stack. To create a random number, vve use the random number gen-
erator and scale the return value to a three- digit range . This code is devel -
oped in Program 1 5 - 1 2 .
PROGRAM 15 - 12 Insert Data

l /* == = insertData ============
2 This program creates random numbers and
3 inserts them into a linked list stack.
4 Pre pStack is a pointer to first node
5 Post Stack has been created
6 */
7 void insertData ( STACK* pStack )
continuec
Chapter 15 Lists
PROGRAM 15 - 12 Insert Data (
continued)
8 {
10 int numln;
11 bool success;
12
13 // Statements
14 printf( "Creating numbers
: ");
15 for (int nodeCount = 0;
16
nodeCount < 10; nodeCount++ )
{
17 // Generate random
number
18 numln = rand() % 999;
19 printf("%4d\ numln);
20 success = push(pStack , numln);
21 if (!success)
22 {
23 printf( Error 101: Out of Memory \n " )
;
24 exit ( 101);
25 } // if
26 } // for
27 printf( "\n");
28 return;
29 } // insertData
Print Stack
Once the stack has been built, we print it to verify the output. The print func -
tion calls the pop function until the stack is empty. The print code is found in
Program 15 - 13.
PROGRAM 15- 13 Print Stack

1 /* ====
2 This function prints a singly linked stack.
3 Pre pStack is pointer to valid stack
4 Post data in stack printed
5 */
6 void print (STACK * pStack )
7 {
9 int printData ;
10
11 // Statements
printf( "Stack contained: );
"
12
13 while (popfpStack, sprintData))
14 printf("%4d", printData);
continued
958 Section 15.4 Queues
PROGRAM 15 - 13 Print Stack ( continued )

15 return;
16 } // print
15.4 Queues
A queue is a linear list in which data can he inserted only at one end , called
the rear, and deleted from the other end , called the front. These restrictions
ensure that the data are processed through the queue in the order in which
—
they are received. In other words, a queue is a first in first out ( FIFO)
structure.
A queue is the same as a line . In fact, if you were in England , you would
not get into a line , you would get into a queue. A line of people waiting lor
the bus at a bus station is a queue, a list of calls put on hold to be answered
by a telephone operator is a queue, and a list of waiting jobs to he processed
by a computer is a queue .
A queue is a linear list in which data can be inserted at one end, called the
.
rear, and deleted from the other end, called the front It is a first in^first out
( FIFO) restricted data structure .
Figure 15 - 23 shows two representations of a queue: one a queue ol peo-
ple and the other a computer queue . Both people and data enter the queue at
the rear and progress through the queue until they arrive at the front. Once
they are at the front of the queue, they leave the queue and are served .
Banks'R 'Us
lIKt hi
(a ) A queue ( line) of people
Remove Insert
( dequeue) ( enqueue)
(b) A computer queue
FIGURE 15 - 23 Queue Concept
Queue Operations
There are two basic queue operations. Data can be inserted at the rear and
deleted from the front. Although there are many similarities between stacks and
queues, one significant structural difference is that the queue implementation
H
needs to keep track of the front and the rear of the queue, whereas the stack
only needs to worry about one end: the top.
Enqueue
I he queue insert operation is known as enqueue . After the data have been
inserted into the queue, the new element becomes the rear. As we saw with
stacks, the only potential problem with enqueue is running out of room for
the data. If there is not enough room for another element in the queue, the
queue is in an overflow state.
Enqueue inserts an element at the rear of the queue.
Figure 15 - 24 shows the enqueue operation .
Queue
Queue Operation
FIGURE 15 - 24 Enqueue
uequeue at the front of the

operation is known as dequeue. The data
The queue delete the queue. If there are no data
user and removed from
is in an underflow state.
queue are returned to the
when a dequeue is attempted , the queue
in the queue
the queue.
element a t the front of
Dequeue deletes an
25.
is
is shown in Figure 15 -
The dequeue operation
EH
data
357 757
Dequeue front rear
123
rear
front Queue After
Operation
Queue Before
FIGURE 15-25 Dequeue

Queue Linked List Design

As with a stack, wc implement our queue using a linked list .
Data Structure
We need two different structures to implement the queue: a queue head struc -
ture and a data node structure. After it is created, the queue will have one head
node and zero or more data nodes, depending on its current state. Figure 15 - 26
shows the conceptual and physical implementations for our queue structure.
I 123 357
(a ) Conceptual queue
757
front count rear
HD
c
-
123
front - I 357
( b ) Physical queue
j H 757
rear
DO
FIGURE 15 - 26 Conceptual and Physical Queue Implementations
Queue Head
.
I he queue requires two pointers and a count These fields are stored in the
.
queue head structure Other queue attributes, such as the maximum num-
ber of items that were ever present in the queue and the total number of
items that have been processed through the queue, can he stored in the head
node if such data are relevant to an application. The queue head structure is
shown in Figure 15 - 27.
typedef struct node

{
data next int data ;
struct node* next;
} QUEUE_ NODE;
Node Structure
typedef struct
{
QUEUENODE* front;
int count;
front count rear QUEUENODE* rear ;
} QUEUE;
Head Structure
FIGURE 15 - 27 Queue Data Structure

—
Chapter jj Lists 961
Queue Data Node The queue
data node contains the user data and
pointing to the next node, if any. These a link field
and are inserted and deleted as nodes are stored in dynamic memory
requested by the using program. Its
is also shown in Figure 15 27. structure
-
Queue Functions
I he two basic functions for a queue are
enqueue and dequeue.
Enqueue
I he enqueue is a little more complex than inserting data into
a stack. To
develop the insertion algorithm , we need to analyze two different queue con -
ditions: insertion into an empty queue and insertion into a queue with
data .
These operations are shown in Figure 15-28.
When we insert data into an empty queue , the queue’s front and rear
pointers must both be set to point to the new node. When we insert data into
a queue with data already in it , we must point both the link field in the last
node and the rear pointer to the new node. If the insert was successful , we
return a Boolean true ; if there is no memory left for the new node , we return
a Boolean false . The code for enqueue is show n in Program 15 - 14.
front count rear front count rear

Queue
H DO K
Queue
y on y
B
newPtr data next newPtr
rn%i
data next
Before After
(a ) Case 1: Insert into Null Queue
front count rear
Queue
front count rear
0 Queue
ra
?
^
ED newPtr newPtr
S B!
data next data next data next data next
Before After
( b) Case 2: Insert into Queue with Data
FIGURE 15 - 28 Enqueue Example
PROGRAM 1 5 - 1 4 Enqueue
1 /* == === enqueue ========
2 This algorithm inserts data into a queue.
3 Pre queue is valid
inserted
4 Post data have been
continued
PROGRAM 1 5 - 1 4 Enqueue ( continued )

5 Return true if successful , false if overflow
6 */
7 bool enqueue ( QUEUE* queue , int dataln )
8 {
10 QUEUE NODE* newPtr ;
11
12 // Statements
13 if (!(newPtr = malloc(sizeof(QUEUE NODE)))) _
14 return false;
15
16 -
newPtr >data = dataln ;
17
18
-
newPtr >next = NULL ;
19 -
if ( queue >count == 0 )
20 // Inserting into null queue
21 queue->front = newPtr ;
22 else
23 queue->rear->next = newPtr ;
24 -
(queue >count)++;
25 -
queue >rear = newPtr ;
26 return true;
27 } // enqueue
Program 1 5 - 1 4 Analysis Because we must maintain both a front and a rear pointer, we need to check to see if
we are inserting into a null queue . If we are, we must set both pointers to the data just
inserted. If there are already data in the queue, we need to set the next field of the
node at the rear of the queue and the rear pointer to the new node. In this case the
front pointer is unchanged . Because the rear pointer is updated in either case, we
changed it after the if statement ( see statement 25 ).
Dequeue
Although dequeue is also a little more complex than deleting data from a
stack , it starts out much the same . We must first ensure that the queue con -
tains data . If the queue is empty, we have underflow and we return false , indi -
cating that the dequeue was not successful .
Given that there are data to he dequeued , we pass the data back through
the parameter list and then set the front pointer to the next item in the
queue. If we have just dequeued the last item , the queue front pointer auto -
matically becomes a null pointer by assigning it the null pointer from the link
held of the last node . I lowever, if the queue is now empty, we must also set
the rear pointer to null . These cases are shown in Figure 15 - 29 .
The dequeue implementation is shown in Program 15 - 15 .

Queue
y cm p
—
Queue
[ E3 m E3
data next deleteLoc
Before After
( a ) Case 1: Delete only item in queue
Queue Queue
a
5[J3Q [XI Q B,E
data next
Before
-
SKI
I
data next
[B
—ii
deleteLoc
After
r*T\
data next
( b) Case 2: Delete item at front of queue
FIGURE 15 - 29 Dequeue Examples
PROGRAM 1 5 - 1 5 Dequeue
1 /* = = dequeue =====
2 This algorithm deletes a node from the queue.
3 Pre queue is pointer to queue head structure
4 dataOut is pointer to data being deleted
5 Post Data pointer to queue front returned and
6 front element deleted and recycled.
7 Return true if successful; false if underflow
8 */
9 bool dequeue (QUEUE* queue
, int* dataOut )
10 {
Local Declarations
11
12
//
_
QUEUE NODE* deleteLoc;
13
14 // Statements
15 if (!queue >count)
return false;
-
16
17
18 *dataOut = queue
>front >data;- -
19 deleteLoc = queue
=
>front;
= 1)
-
20 if (queue- >count
Deleting only item in
queue
/ /
21
22 queue-> rear = queue >front = NULL; -
else
23
24 queue >front = queue
- >front >next ;
- -
25 -
(queue >count)~
;
con ti ttueil
PROGRAM 15- 15 Dequeue ( continued )

26 free ( deleteLoc );
27
28 return true;
29 } // dequeue
Queue Demonstration
To demonstrate a queue, we write a simple program that creates a queue of
.
colors and then prints them The design closely parallels the design ol the
stack demonstration. The node declarations, Function declarations, and test
driver are contained in Program I 5 - 16.
PROGRAM 15- 16 Simple Queue Demonstration

1 /* This program is a test driver to demonstrate the
2 basic operation of the enqueue and dequeue functions.
3 Written by:
4 Date:
5 */
10
13 {
14 int data ;
15 struct node* next;
16 > QUEUE NODE;
17
18 typedef struct
19 {
20 _
QUEUE NODE* front ;
21 int count ;
22 _
QUEUE NODE* rear ;
23 } QUEUE;
24
26 void insertData (QUEUE* pQueue );
27 void print ( QUEUE* pQueue);
28 bool enqueue (QUEUE* pList , int dataln);
29 bool dequeue ( QUEUE* pList, int* dataOut);
30
continues
i
PROGRAM 1 5- 1 6 Simple Queue Demonstration (
continued )
31 int main ( void
)
32 {
33 // Local
Declarations
34 QUEUE* pQueue;
35
36 // Statements
37 printf("Beginning Simple
Queue Program \n ");
38
39 pQueue = malloc(sizeof(QUEUE
));
40 if (!pQueue)
41 printf(" Error allocating queue"
), exit(100 );
42
43
-
pQueue >front = NULL;
44 -
pQueue >count = 0;
45
46
-
pQueue >rear = NULL;
47 insertData (pQueue);
48 print ( pQueue);
49
50 printf( "\nEnd Simple Queue Program \n " );
51 return 0;
52 } // main
Results:
Beginning Simple Queue Program
Creating numbers: 854 763 123 532 82
Queue contained: 854 763 123 532 82
End Simple Queue Program
Insert Data
The insert data function creates a queue by inserting numbers into the
queue. The code is developed in Program 15 - 17.
PROGRAM 1 5 - 1 7 Insert Data

1 /* ======= ====== insertData =============
2 This program creates random number data and
3 inserts them into a linked list queue.
4 Pre pQueue is a pointer to first node
5 Post Queue created and filled
6 */
7 void insertData (QUEUE* pQueue)
8 {
continued
PROGRAM 15- 17 Insert Data (continued )

10 int numln ;
11 bool success;
12
13 // Statements
14 printf( "Creating numbers: " );
15 for ( int nodeCount = 0; nodeCount < 5; nodeCount++)
16 {
17 // Generate random number
18 numln = rand( ) % 999 ;
19 printf( "%4d " , numln);
20 success = enqueue( pQueue , numln );
21 if (!success)
22 {
23 printf( " Error 101: Out of Memory \n" );
24 exit ( 101 );
25 > // if
26 } // for
27 printf( "\n" );
28 return ;
29 } // insertData
Print Queue
Once the queue has been built, we print it to verify' that the queue was built
.
correctly The print function calls the dequeue function until all items have
been printed. I he print code is found in Program 15 - 18.
PROGRAM 15 - 18 Print Queue

1 /* ======
2 This function prints a singly linked queue.
3 Pre pQueue is pointer to valid queue
4 Post data in queue printed
5 */
6 void print ( QUEUE* pQueue )
7 {
9 int printData ;
10
11 // Statements
12 printf( "Queue contained: " );
13 while ( dequeue( pQueue, &printData ))
14 printf( " %4d " , printData );
15 return;
16 > // print
1
15.5 Trees
I he study of trees in mathematics
can be traced to Gustav Kirchhoff in the
middle nineteenth century and several years
later to Arthur Cayley, who used
trees to study the structure ol algebraic
formulas .
laid the framework for Grace Hopper’s use of trees Cayley s work undoubtedly
in 1951 to represent arith -
metic expressions. Hopper’s work hears a
strong resemblance to today’s
binary tree formats.3
Trees are used extensively in computer science to represent
formulas; as an efficient method for searching large, dynamic listsalgebraic
; and lor
such diverse applications as artificial intelligence systems and
encoding
algorithms.
Basic Tree Concepts

A tree consists of a finite set of elements, called nodes, and a finite set of
directed lines, called branches, that connect the nodes. The number of
branches associated with a node is the degree of the node. When the branch
is directed toward the node, it is an indegree branch ; when the branch is
directed away from the node, it is an outdegree branch. The sum of the inde-
gree and outdegree branches is the degree of the node.
A tree consists of a finite set of elements, called nodes, and a finite set of
directed lines, called branches, that connect the nodes.
II the tree is not empty, the first node is called the root . The indegree of
the root is, by definition , zero. With the exception of the root , all of the nodes
in a tree must have an indegree of exactly one; that is, they may have only one
predecessor. All nodes in the tree can have zero, one, or more branches leav-
(
ing them; that is, they may have an outdegree of zero, one, or more zero or
more successors ). Figure 15- 30 is a representation of a tree.
FIGURE 15- 30 Tree
.
3. Donald E. Knuth, op. cil ,
.
405, 458
968 Section 15.5 Trees
Terminology
In addition to root, many different terms are used to describe the attributes of
a tree. A leaf is any node with an outdegree of zero, that is, a node with no
.
successors A node that is not a root or a leaf is known as an internal node
because it is found in the middle portion of a tree.
A node is a parent if it has successor nodes— that is, if it has an outdegree
greater than zero. Conversely, a node with a predecessor is a child A child .
node has an indegree of one. Two or more nodes with the same parent are
siblings. Fortunately, we don’t have to worry about aunts, uncles, nieces, neph-
ews, and cousins. Although some literature uses the term grandparent , we do
not. We prefer the more general term ancestor. An ancestor is any node in the
path from the root to the node. A descendent is any node in the path below
the parent node; that is, all nodes in the paths from a given node to a leaf are
descendents of that node. Figure 15 -31 shows the usage of these terms .
Several terms drawn from mathematics or created by computer scien-
tists are used to describe attributes of trees and their nodes A path is a .
sequence of nodes in which each node is adjacent to the next one. Every
node in the tree can he reached hv following a unique path starting from the
root. In Figure 1 3 - 3 1 the path from the root to the leaf I is designated as
AFI. It includes two distinct branches, AF and FI.
Level 0
A
Branch
AF
Level 1
B
<i} F
Branch
FI
Level 2 C D G H
Root: A
Parents: A , B, F
.. . . . .. .
Siblings: { B, E ,F } {C, D} , {G H I}
Leaves: C D EG HI
.
Children: B, E, F, C. D, G, H I Internal nodes: B, F
FIGURE 1 5 - 3 1 Tree Nomenclature
The level of a node is its distance from the root. Because the root has a
zero distance from itself, the root is at level 0. The children of the root are at
level 1 , their children are at level 2, and so forth. Note the relationship
between levels and siblings in Figure 1 5 - 3 1. Siblings are always at the same
level, hut all nodes in a level are not necessarily siblings. For example, at
level 2 , C and D are siblings, as are G , H , and I . Flowever, D and G arc not sib-
lings because they have different parents.
I he height ol the tree is the level ol the leaf in the longest path from the
root plus 1. By definition the height of an empty tree is 1. Figure 1 5 - 3 1
-
I
Chapter 15 Lists 969 T!
contains nodes at three levels: 0, 1 , and 2.
Its height is 3. Because the tree is
drawn upside down, some texts reler to
the depth of a tree rather than
its height .
The level of a node is its distance from the root.

The height of a tree is the
level of the leaf in the longest path from the root
plus 1.
A tree may he divided into subtrees. A subtree is any connected

structure
below the root . The first node in a subtree is known as the root of the
subtree
and is used to name the subtree. Subtrees can also be further subdivided into
subtrees. In Figure 15 -32. BCD is a subtree, as arc E and FGHI . Note that by
this definition , a single node is a subtree. Thus, the subtree B can be divided
into two subtrees, C and D, and the subtree F contains the subtrees G , H
and I .
A
Subtree
B
Root of
Subtree I
FIGURE 1 5 - 32 Subtrees
The concept of subtrees leads us to a recursive definition of a tree A tree

:
) empty or ( 2 ) has a designated node , called
is a set of nodes that either: ( 1 is
the root , from which hierarchically descend zero or more subtrees, which are
also trees.
A tree is a set of nodes that either.

1. is empty, or
the root , from which hierarchically descend
2. Has a designated node, called
are also trees
zero or more subtrees, which
Binary Trees more than two subtrees;

w hich no node can have
A binary tree is a tree in ' node is two. In other words, a node can have
the maximum outdegree for a subtrees are designated as the left subtree
zero, one, or two subtrees
. These
and the right subtree .
Figure 15 - 33 contains a collection of eight binary trees, the (irst of

which is a null tree , that is, a tree with no nodes. As you study this figure,
note that symmetry is not a tree requirement.
CD
(a) (b) (c)

Left A Left A
Subtree Subtree
B C B C
D E
(e ) (f )
A Right A
Subtree
Right
B B Subtree
C C
(g) (h)
FIGURE 15 - 33 Collection of Binary Trees
Binary Tree Data Structure

A binary tree needs two separate data structures: one for the head and one for
the nodes. As with the linear list, the structure uses a simple head structure
that contains a count and a root pointer.
I he binary tree nodes contain the application data and two self-referential
pointers to the left and right subtrees. These data structures are shown in
Figure 1 5 - 34.
Binary Tree Traversals

Given that a binary tree consists of a root, a left subtree, and a right subtree,
we can define six different traversal sequences. Computer scientists have
assigned three ol these sequences standard names in the literature; the other
three are unnamed but are easily derived. The first two standard traversals are
shown in Figure 15 - 35. Note that we only discuss the preorder and inorder
traversals in this text . I he others will be covered in your data structures class.
The traditional terminology lor the traversals uses a designation ol node
( \ ) lor the root, left ( L ) for the
left subtree, and right ( R) for the right
""
subtree. Io demonstrate the different traversal sequences

for a binary tree,
we use Figure -
15 36.
BINJREE
count root
typedef struct
{
int count;
NODE* root;
to tree } BIN TREE;_
NODE typedef struct node
left data right <
int data;
¥ ¥
struct node* left;
struct node* right;
} NODE;
to left to right
subtree subtree
FIGURE 1 5 - 34 Binary Tree Data Structure
1 2
Left Right Left Right

subtree subtree subtree subtree
(a) Preorder Traversal (b) Inorder Traversal
FIGURE 1 5 - 35 Binary Tree Traversals
FIGURE 15 - 36 Binary Tree for Traversals
Preorder Jroversal (NLP)
L „. which means » „ . 8«« b f» « h

p
The fc ' ' ">
9 / 2 Section 15.5 Trees
In the preorder traversal, the root is processed first, before its subtrees.
Given the recursive characteristics of trees, it is only natural to implement

tree traversals recursively. First we process the root, then the left subtree, and
then the right subtree. The left subtree is in turn processed recursively, as is the
right subtree. The code for the preorder traversal is shown in Program 15 - 19.
PROGRAM 15 - 19 Preorder Traversal of a Binary Tree

1 /* Traverse a binary tree and print its data ( integers )
2 Pre root is entry node of a tree or subtree
3 Post each node has been printed
4 */
5 void preOrder ( NODE* root )
6 {
7 // Statements
8 if ( root)
9 {
10 -
printf( " %4d " , root >data );
11 preOrder ( root->left);
12 -
preOrder (root >right);
13 } // if
14 return ;
15 > // preOrder
Using the tree in figure 15- 36 the processing sequence for a preorder
traversal processes this tree as follows: First we process the root A. After
the root , we process the left subtree. To process the left subtree , we lirst
process its root , B , then its left subtree and right subtree in order. When B s
left and right subtrees have been processed in order, we are then ready to
process A s right subtree, E . To process the subtree E , we first process the
root and then the left subtree and the right subtree. Because there is no
left subtree, we continue immediately with the right subtree , which com -
pletes the tree.
Figure 15- 37 shows another way to visualize the traversal of the tree.
Imagine that we are walking around the tree, starting on the left ol the root
and keeping as close to the nodes as possible. In the preorder traversal we
process the node when we meet it for the first time ( on the left of the node ).
I his is shown as a black box on the leit ol the node. The path is shown as a
line following a route completely around the tree and back to the root .
figure 15 - 38 shows the recursive algorithmic traversal of the tree. The
first call processes the root of the tree, A. It then recursively calls itself to pro-
cess the root of the subtree B, as shown in Figure 15- 38( b ). The third call ,
shown in Figure 15 -38 ( c ) , processes node C , which is also subtree C. At this
point we call preorder with a null pointer, which results in i an immediate
return to subtree C to process its right subtree. Because C s right subtree is
also null , we return to node B so that we can process

its right tree, D , in
F igure 15 -38( d ) . After processing node D, we
make two more calls, one with
Ds null lelt pointer and one with its null right
pointer. Because subtree B has
now been completely processed , we return to the tree root and
process its
right subtree, E, in Figure 15-38(e ). After a call to Es null left subtree, we call
Es right subtree, F , in Figure 15- 38( f ). Although the tree is
completely pro-
cessed at this point, we still have two more calls to make: one to F s null left
subtree and one to its null right subtree. We can now hack out of the tree,
returning First to E and then to A, which concludes the traversal ol the tree.
( a ) Processing Order (b ) “ Walking ” Order
FIGURE 15 - 37 Preorder Traversal- A B C D E F
( b) Process Tree B
( a ) Process Tree A
(d) Process Tree D

(c ) Process Tree C
(f ) Process Tree F
(e) Process Tree E
Tree
FIGURE 15 - 38 Algorithmic Traversa of
| Binary
T
9 / 4 Section 15.5 Trees
Inorder Traversal ( LNR )

The inorder traversal processes the left subtree first , then the root , and finally
the right subtree. The meaning of the prefix in is that the root is processed in
between the subtrees. Once again we implement the algorithm recursively, as
shown in Program 15 20. -
PROGRAM 15 - 20 Inorder Traversal of a Binary Tree
1 /* Traverse a binary tree and print its data ( integers )
2 Pre root is entry node of a tree or subtree
3 Post each node has been printed
4 */
5 void inOrder ( NODE* root )
6 {
7 // Statements
8 if ( root)
9 {
10 -
inOrder ( root >left );
11 -
printf( " %4d " , root >data );
12 -
inOrder (root >right);
13 > // if
14 return ;
15 > // inOrder
Because the left subtree must be processed first , we trace from the root
to the far- left leaf node before processing any nodes. After processing the far-
left subtree, C, we process its parent node, B. We are now ready to process the
right subtree , D. Processing D completes the processing of the roots left sub-
tree , and we are now ready to process the root , A, followed by its right sub-
tree . Because the right subtree, E, has no left child , we can process its root
immediately followed by its right subtree, F. The complete sequence for inor-
der processing is shown in Figure 15 - 39 .
(a) Processing Order (b) “Walking” Order
FIGURE 1 5 - 39 Inorder Traversal — C B D A E F

To walk around the tree in
inorder
but process each node when we meet it sequence, we follow the same path
for
the node). This processing route is shown the second time ( the bottom of
in Figure 15-39( b).
In the inorder traversal, the root is

processed between its subtrees.
Binary Search Trees

In this section we define and discuss one of the most common
binary trees, the
binary search trees. The binary search tree is constructed so that when the tree
is
traversed using an inordcr traversal, the data are in ascending sequence.
Definition
A binary search tree ( BST) is a binary tree with the following properties:
• All items in the left subtree are less than the root.
• All items in the right subtree are greater than or equal to the root .
• Each subtree is itself a binary search tree.
In a binary search tree, the left subtree contains key values less than the
root, and the right subtree contains key values greater than or equal to
the root.
Generally, the information represented by each node is a record rather

than a single data element. When the binary search tree definition is applied to
a record , the sequencing properties refer to the key of the record. Figure 15 - 40
reflects the properties of a binary' tree in which K is the key.
All < K All > K
FIGURE 15 - 40 Binary Search Tree
Figure 15-41 contains five binary'

search trees.
that do not have the properties ol a
Now let s look at
’ some
binary search tree. Examine the
binary trees
binary trees in Figure 15 -42. The first tree .
rule: all items in the left subtree must be less
Figure 15 - 42(a ), breaks the first
.
root ( 17 ) The second tree
, Figure -
.
subtree ( 22 ) is greater than the key in the
than the root. The key in the left I 5 42( b), breaks the second rule: all terns
than or equal to the root. The key m the
in the right subtree must greater
be
right subtree ( 1 1 ) is less than the key in the root ( 1 7 ). Figure 1 5 -42 (c ) breaks
the third rule: each subtree must be a binary search tree. In this tree the left
subtree key ( 6) is less than the root ( 17 ) , and the right subtree key ( 19 ) is
greater than the root. However, the left subtree is not a valid binary search
tree because it breaks the first rule: its left subtree ( 11 ) is greater than the
root ( 6 ) . Figure 15- 42 ( d ) also breaks one of the three rules . Do you see
which one? ( H i n t : What is the largest key in the left subtree? )
FIGURE 15 - 41 Valid Binary Search Trees
FIGURE 15 - 42 Invalid Binary Search Trees
Insertion
The insert node function adds data to a BST. To insert data all we need to do
is follow the branches to an empty subtree and then insert the new node. In
—
other words, all inserts take place at a leaf or at a leaflike node a node that
has at least one null subtree.
All BST insertions take place at a leaf or a leaflike node .

Figure 15- 43 shows our binary search tree after we have inserted two
nodes. We first added node 19. To locate its insertion point , we searched the
tree through the path 23, 18, and 20 to a null left branch . After locating the
insertion point , we inserted the new node as the left subtree of 20. We then
added 38. This time we searched the tree through 23, 44 , and 35 to a null
right subtree and inserted the new node.
T"1
Chapter IS Lists 977
( a ) Before inserting 19
( b ) After inserting 19
(c ) Before inserting 38 ( d ) After inserting 38
FIGURE 15 - 43 BST Insertion
Insertions of both 19 and 38 were made at a leal node. If we inserted a

duplicate of the root, 23, it would become the left subtree of 35. Remember
that in a binary search tree, nodes with equal values are inserted in the right
subtree. The path for its insertion would therefore be 23, 44, and 35. In this
case the insertion takes place at a leaflike node. Although 35 has a right sub-
tree, its left subtree is null. We would therefore place the new node, 23, as
the left subtree of 35.
We are now ready to develop the insert algorithm. We can write an ele-
gant algorithm that inserts the data into the tree using recursion. If the tree
or subtree is empty, we simply insert the data at the root. II we are not at an
empty tree, we determine which branch we need to
follow and call recursively
to determine w hether we are at a leaf yet. The code is show n in Program 15 -21.
Note that as we saw with other dynamic structures, the BST insert can raise
an overflow error .
PROGRAM 15- 21 Binary Tree Insert Function
BST Insert =====
==
1 /* ====
to insert the new data
2 This function uses recursion
in the BST tree.
3 into a leaf node
Pre Application has called _
BST Insert, which
4
passes root and data pointer
5
Post Data have been
inserted
6 ] new root
Return pointer to [ potentially
7
continued
PROGRAM 15 - 21 Binary Tree Insert Function ( continued )

8
9
*/
_ _
NODE* BST Insert ( BST TREE* tree ,
10 NODE* root, int dataln)
11 {
13 NODE* newPtr;
14
15 // Statements
16
17 if (!root )
18 {
19
20
// NULL tree
—
create new node
newPtr = malloc( sizeof ( NODE ));
21 if (!newPtr)
22 printf( " Overflow in Insert\n" ), exit ( 100 );
23 -
newPtr >data = dataln ;
24 - -
newPtr >left = newPtr >right = NULL;
25 return newPtr ;
26 > // if
27
28 // Locate null subtree for insertion
29 - _
if ( dataln < root >data)
30 root-> left = BST Insert(tree , root > left , -
31 dataln );
32 else
33 // new data >= root data
34 - _
root >right = BST Insert( tree , root >right ,
-
35 dataln );
36 return root ;
37 } // BST Insert
Program 1 5 - 21 Analysis This algorithm must be carefully studied to fully understand its logic. It begins with a
recursive search to locate the correct insertion point in a leaf node. A leaf node is
identified by a subtree pointer, either right or left, that is null. When we find a leaf
pointer, we create a new node and return its address so that it can be inserted into
the parent pointer (statement 25).
Because this is a recursive function, it must have a base case. Can you see it? The
base case occurs when we locate a leaf and return newPtr in statement 25. At this
point we begin to back out of the tree.
Binary Tree Example

Let ’s write a program that builds a binary search tree. We build the tree by ask-
ing the user to enter numbers. Once the tree is built, we print it in both preorder
and inordcr sequence to verily the insertions. To keep the program simple, we
will use integer data . The design is shown in Figure 15 - 44.
BST
i
preOrder inOrder ]
BST Insert
i
FIGURE 1 5 - 44 Binary Tree Program Design
PROGRAM 1 5 - 22 Binary Tree Example

1 /* Demonstrate the binary search tree insert and
2 traversals.
3 Written by:
4 Date:
5 */
8
11 {
12 int data ;
13 struct node* left;
14 struct node* right;
15 } NODE ;
16
17 typedef struct
18 {
19 int count ;
20 NODE* root;
21
__
} BST TREE;
22
2 3 // Function
Declarations );
preOrder (NODE* root
2 4 void
(NODE* root);
2 5 void inOrder * tree ,
2 6 NODE* BST Insert
( BSTJTREE
, int data);
NODE* root
27
28
2 9 int main (
void )
30 {
3 1 II Local
Declarations continuec
PROGRAM 15 - 22 Binary Tree Example ( continued )

32 int numln ;
33 BST TREE tree;
34
35 // Statements
36 printf( "Please enter a series of integers."
37 "\nEnter a negative number to stop\n" );
38
39 tree.count = 0 ;
40 tree.root = NULL;
41 do
42 {
43 printf( " Enter a number: " );
44 scanf( " %d " , & numln );
45 if (numln > 0)
46 {
47 _
tree.root = BST Insert
48 (&tree, tree.root, numln);
49 tree.count++ ;
50 > // if
51 } while ( numln > 0 );
52
53 printf( " \nData in preOrder: " );
54 preOrder ( tree.root );
55
56 printf( "\n\nData in inOrder:
57 inOrder ( tree.root );
58
59 printf("\n\nEnd of BST Demonstration
60
61
return 0 ; ^
" );
> // main
Results
Please enter a series of integers.
Enter a negative number to stop
Enter a number: 45
Enter a number: 54
Enter a number: 23
Enter a number: 32
Enter a number: 3
Enter a number: 1 -
Data in preOrder: 45 23 3 32 54
Data in inOrder: 3 23 32 45 54
End of BST Demonstration

15.6 Graphs
A graph is a collection of nodes, called vertices, and
merits, called lines, connecting pairs of vertices a collection of seg-
sists of two sets, a set of vertices . In other words, a graph con -
and a set of lines.
Graphs may be either directed or
digraph for short , is a graph in which eachundirected. A directed graph, or
to its successor. The lines in a line has a direction (arrow head)
directed graph are known as arcs. In a directed
graph, the flow along the arcs between two
vertices can follow only the indi-
cated direction. An undirected graph is a
graph in which there is no direc-
tion (arrow head ) on any of the lines,
which are
undirected graph , the flow between two vertices canknown as edges. In an
go in either direction .
Figure 15-45 contains an example of both a
directed graph (a ) and an undi-
rected graph ( b).
( a ) Directed Graph (b) Undirected Graph
FIGURE 1 5 - 4 5 Directed and Undirected Graphs
A path is a sequence of vertices in which each vertex is adjacent to the

.
next one. In Figure 15 - 45, {A, B C, E} is one path and {A, B, E , F) is
another. Note that both directed and undirected graphs have paths. In an
undirected graph , you may travel in either direction.
Graphs may be directed or undirected . In a directed graph , each line, called an

arc, has a direction indicating how it may be traversed. In an undirected graph ,
either direction.
the line is known as an edge, and it may be traversed in
( or neighbors) if
Two vertices in a graph are said to be adjacent vertices
connecting them . In Figure 15 45 ( a ), B is adjacent
there is a path of length 1 , D is adjacent to E . In
adjacent to D; on the other hand
to A, whereas E is not
D and I are not .
Figure I 5 45 ( b), E and D are adjacent hut
- ,
A graph is a collection nodes

of , calied vertices, and line segments, called
arcs or edges, that connect pairs
of nodes .
I
982 Section 15.6 Graphs
A cycle is a path consisting of at least three vertices that starts and ends
with the same vertex. In Figure 15 - 45 (b), B, C, D, E, B is a cycle. Note, how-
.
ever that the same vertices in Figure 15 - 45 (a ) do not constitute a cycle
because in a digraph a path can follow only the direction ol the arc, whereas
in an undirected graph a path can move in either direction along the edge. A
loop is a special case of a cycle in which a single arc begins and ends with the
same vertex. In a loop the end points of the line are the same. Figure 15 - 46
contains a loop .
© Cycle fB
(ABC) Loop
<S > (D)
FIGURE 1 5 - 46 Cycles and Loops
Two vertices are said to be connected if there is a path between them. A

graph is said to he connected if, ignoring direction, there is a path from any ver-
tex to any other vertex. Furthermore, a directed graph is strongly connected
it there is a path from each vertex to every' other vertex in the digraph. A
directed graph is weakly connected if at least two vertices are not con-
nected . ( A connected undirected graph would always he strongly connected,
so the concept is not normally used with undirected graphs.) A graph is a
disjoint graph ii it is not connected. Figure 15 - 47 contains a weakly con-
nected graph ( a ) , a strongly connected graph (b), and a disjoint graph (c ).
( a ) Weakly Connected (b ) Strongly Connected (c ) Disjoint Graph
FIGURE 1 5- 47 Connected and Disjoint Graphs
The degree of a vertex is the number of lines incident to it. In Figure 15 - 47(a)
the degree of vertex B is 3 and the degree of vertex E is 4. The outdegree of a
vertex in a digraph is the number of arcs leaving
the vertex; the indegree is
the number of arcs entering the vertex. Again, in Figure 1 5 - 47 ( a) the inde-
gree of vertex B is 1 and its outdegree is 2; in Figure 1 5 - 47 ( b ) the indegree of
vertex E is 3 and its outdegree is 1 .
F\
One final point: a tree is a graph in which each vertex has onlyy one
prede-
cessor ; however, a graph is not a tree. We will see latcr in the chapter that some
graphs have one or more trees in them that can he algorithmically determined .
Graph Traversal
A complete discussion of graph algorithms is beyond the scope of this text . As
« m example of a graph algorithm , we discuss two graph
traversals. The
remaining algorithms to maintain and traverse graphs are left for your data
structures course .
I here is always at least one application that requires that all vertices in a
given graph he visited; that is , there is at least one application that requires
that the graph be traversed. Because a vertex in a graph can have multiple
parents, the traversal of a graph presents some problems not found in the tra -
versal of linear lists and trees. Specifically, we must somehow ensure that we
process the data in each vertex only once. However, because there are multi -
ple paths to a vertex, we may arrive at it from more than one direction as we
traverse the graph . The traditional solution to this problem is to include a vis -
ited flag at each vertex . Before the traversal we set the visited flag in each ver -
tex to off . Then , as we traverse the graph , we set the visited flag to on to
indicate that the data have been processed.
There are two standard graph traversals: depth first and breadth first
.
Both use the visited flag .
Depth-first traversal
vertexs dcscendents before
In the depth -first traversal, we process all of a
. This concept is most easily seen when the
we move to an adjacent vertex
tree preorder traversal - processing
graph is a tree. In Figure 15-48 we show the
first traversals.
sequence, one of the standard depth -
Depth - first Iraversal. A BEFCDGHI j

Tree
FIGURE 15- 48 Depth -first Traversal of a
traversal of a graph starts byselectprocess -

, the depth - first
In a similar manner processing the first vertex , we any
of the graph . After each vertex , we
ing the first vertex
to the first vertex and process it. As we process adjacen t entries.
vertex adjacent with no
vertex until we reach a vertex of the structure ,
select an adjacent . We then back out
a leaf in a tree
This is similar to reaching
w
984 Section 15.6 Grophs
processing adjacent vertices as we go. It should be obvious that this logic

requires a stack ( or recursion ) to complete the traversal.
The order in which the adjacent vertices are processed depends on bow
the graph is physically stored . Because we are using a stack, however, the
traversal processes adjacent vertices in descending, or last in first out
( LI I - ( ) ) , order.
—
-
In the depth first traversal, all of a node’s descendents are processed before
moving to an adjacent node .
Let ’s trace a depth -first traversal through the graph in Figure 1 5- 49. The
number in the box next to a vertex indicates the processing order. I he stacks
below the graph show the stack contents as we work our way down the graph
and then as we hack out .
Depth- first Traversal

AXHPEYM JG
Y
UK E
0
0( G 00 0( M 0
( a ) Graph
p Y
H E E M M J
A X Q G G G G G G
1 2 3 4 5 6 7 8 9
( b) Stack Contents
FIGURE l 5 - 49 Depth - first Traversal of a Graph
1 . We begin by pushing the first vertex, A , into the stack.

2 . We then loop , pop the stack , and , after processing the vertex , push all ol
the adjacent vertices into the stack. To process X at step 2 , therefore, we
pop \ Irom the stack, process it , and then push G and H into the stack ,
giving the stack contents for step 3 as shown in Figure 1 5 - 49 ( h ) 11 G . —
3. When the stack is empty, the traversal is complete.
Breadth-first Jroversol
-
In the breadth first traversal ol a graph , we process all adjacent vertices ol a
.
vertex before going to the next level We saw the breadth -first traversal earlier
.
in the chapter Looking at the tree in Figure 15- 50 , we see that its breadth -
first traversal starts at level 0 and then processes all the vertices in level 1
before going on to process the vertices in level 2 .
T1
Breadth - first Traversal: A B C D E F G H I I
FIGURE 1 5 - 50 Breadth- first Traversal of a Tree
I he breadth-first traversal ol a graph follows the same concept . We begin

by picking a starting vertex ( A ); after processing it we process all of its adjacent
vertices ( BCD). After we process all of the first vertex’s adjacent vertices, we
pick its first adjacent vertex ( B ) and process all of its vertices, then the second
adjacent vertex (C ) and all of its vertices, and so forth until we are finished.
The breadth -first traversal uses a queue rather than a stack. As we pro-
cess each vertex, we place all of its adjacent vertices in the queue. Then
, to
select the next vertex to he processed , we delete a vertex from the queue and
51.
process it . Let ’s trace this logic through the graph in Figure 15 -
-
Breadth first Traversal
AXGHPEMYJ
W Y
UK E
[D( G 0TM J
(a) Graph
M Y Y J J
H P P E E
G H 7 8 9
4 5 6
1 2 3
( b) Queue contents
Graph
FIGURE 15 - 51 Breadth- first Traversal of a
vertex A in the queue.

1 We begin by enqueuing the vertex from the
.
2. w ,hen loop.
front of the queue . At 2 in Figure 15 - 51 ( b
he , , “"
) , we dequeue
u We are
^
C H in
*“ 0 G'
to Jqueuefcr step 3, in
empty, the traversal
”
^complete. is
.
3 When the
is
is
before pro-
-first tra
tergal, all adjacent vertices are processed
In the breadth of a vertex .
cessing the descendents
J
n 15.7 Software Engineering

Because linear lists are useful structures, programmers use them in many
applications. Rather than rewrite their functions each time we need them , we
can write functions once and put them in a library. I hen when we need to
use a linear list , we simply include the library. The name given to a complete
set of functions built like this is abstract data type (ADT). To present the
concept , we need to define a few new terms.
Atomic and Composite Data

Atomic data are data that we choose to consider as a single, nondecompos -
ahle, entity. For example , the integer 4562 may he considered as a single inte-
ger value. Of course, you can decompose it into digits , hut the decomposed
digits do not have the same characteristics of the original integer ; they arc
four one -digit integers in the range 0 to 9.
An atomic data type is a set of atomic data having identical properties.
These properties distinguish one atomic data type from another. Atomic data
types are defined by a set of values and a set of operations that act on the
values.
Atomic Data Type

1. A set of values .
2. A set of operations on the values .
For example, we can define the following atomic data types:
int
VALUES: - oo , ... , -2, -1, 0 , 1 , 2 ... , oo
OPERATIONS:
float
*t +I %, I , ++ , —
, < / > , ...
VALUES: - oo , ... , 0.0 , ... OO

OPERATIONS:
char
* . +» I , < , > , ...
VALUES: \ 0 , ... , ' A' , 'B', ... / ' a 1

, * b \ ... , \ 127
OPERATIONS: < , > , ...
I he opposite ol atomic data is composite data . Composite data can he

broken into subfields that have meaning. As an example of a composite data
item , consider your telephone number. A telephone number actually has
three different parts. First is the area code. Then , what you consider to be
your phone number is actually two different data items , a prefix consisting ol
three digits and the number within the prefix , consisting of four digits. Years
ago , these prefixes were names such as DAvenport and CYp ress.
i,\
Data Structure
A data structure is a collection of elements and the relationships among
.
them. Data structures can be nested That is we can have a data structure
that consists of other data structures.
Data Structure
1. A combination of elements, each of which is either a data type or another
data structure .
2. A set of associations or relationships (structure) involving the combined
elements.
I or example, we can define the two structures array and struct , as shown
in Table 15 - 2.
array struct
A homogeneous combination of data A heterogeneous combination of

structures. data structures.
Position association. No association .
TABLE 1 5 - 2 Two Structures

data structures. In addi-
Most programming languages support several
tion, modern programming languages
allow programmers to create new data
7
in the language they are using. In C

, this is
structures that are not available
done with struct ( see C hapter 12
).
Abstract Data Type
analysts, your too s will

, file management, and systems to add to your tool kit.
data structures
.
. ,
data
/
rtypes ( AD another
1 are anoi
Aryn tool
languages had no ADT*. lo read
Abstrac t hie,
..
increase
The first computer programming . It did not take long to
to read the file device
programmers wrote the code writing the same code over and over again . So
. ,
irp
<
A rne wrote h cod o
P
^
read a file and placed Urn
This concept
library is an
"Jo ^
11
data.
a„d . standard inpat/oatpat
set of operations that
is done
can be
but
used to read and write rned with how the task
With an ADT , the user ADT consists of a set of
what it can do. In other uor
rather with
prototype definitions that allow the programmer to use the functions while
hiding the implementation . This generalization of operations with unspeci -
fied implementations is known as abstraction . We abstract the essence of the
process and leave the implementation details hidden .
In the concept of abstraction

We know what a data type can do .
How it is done is hidden.
Consider the concept of a list . At least three data structures can support
a list. We can use an array, a linear list , or a file. If we place the list in an
abstract data type , the user should not he aware of the structure we use. As
long as data can be inserted and retrieved , it should make no difference how
we store the data. Figure 15- 52 shows several structures that might be used
to hold a list .
OO-OO
A linked list
A matrix A tree A network
FIGURE 15- 52 Structures for Holding a List
An abstract data type is formally defined as a data declaration packaged

together with the operations that are allowed on the data type. In other
words , we encapsulate the data and the operations on data , and we hide their
implementation from the user.
Abstract Data Type

.
1 Declaration of data .
2. Declaration of operations .
The AD I definition implies two attributes for ADTs:
1 . The structures are opaque . We can use them without knowing how they
are implemented .
2. The operations are opaque . We know what they d o: we don’t know' how
they do it .
We cannot overstress the importance of hiding the implementation. For
example , the programmer should not have to know the data structure to use
the AD I . 1 his is a common fault in many implementations that keep the ADT
Irom being Lully portable to other
applications
capability gives us the tools to fully implement . Fortunately, C s rich library
any ADT.
A Model for an Abstract Data Type

1 be AD I model is shown in Figure 15
-53. The model is represented by the
blue area with an irregular outline. Inside are
two different aspects of the
model: the data structure and the operational
functions. Both are entirely
contained in the model and are not within the user’s
scope. However, the
data structure is available to all the ADT’s operations as
needed , and an oper-
ation may call on other functions to accomplish its
task. In other words, the
data structure and the functions are within scope of each other.
J Public
Interface Private
Functions Functions
Application I
Program Data Structures

= S=/ °=
( ii N
n
Q B
Array UnkedList
O
Record
Dynamic Memory
FIGURE 15 - 53 Abstract Data Type Model : .
Data flows in and out of the ADT through the operation headers repre-
sented by the rectangular " Interface ” pathway. The interface is also the path -
would have
way for the ADT functions. For instance, a linear list ADT
in this chapter , such as insert , remove, and search . In
operations that we saw
C, these operations are defined as prototype header
j
declarations that are visi-
only the ‘‘ Public " functions are available
ble to the user. Note, .however, that
" functions are totally contained with
to the application program; the Private
"
1»
within the ADT.
the ADT and can only be used by other functions
need to create a user header file.
To use the ADT in our program , we
list ADT, we give its header file a
When we create a header file for a linear our program , we would use a
name, such as Llinklist . h. To
include it in
preprocessor directive such as :
# include " linklist h

" .
i It
.
peering
ADT Data Structure

When the list is controlled entirely by the program , it is often implemented
using simple structures such as those shown in this chapter. Since the ADT
must hide the implementation from the user, however, all data about the
structure must be maintained inside the ADT. But just encapsulating
the structure in the ADT is not sufficient. In addition , multiple versions
of the structure must he able to coexist . This means that we must hide the
implementation from the user while storing data about the structure in
the user ’s program .
You have seen this concept before. When you create a file, you use the
predefined structure FILE. Defining a file in your program creates a file struc-
ture that becomes a part of your program . We can do the same thing with the
ADT. Each ADT must have a defined type that the users can define in their
programs. Just like the file type, the ADI type is a pointer to a structure that
contains attributes about the structure. When the ADT attribute structure is
created , it is stored in the heap. The only structural element in the user’s pro-
gram is a pointer to the structure .
This short description of ADTs just begins to introduce the topic. When
you take a data structure class , you may have the opportunity to create and
use some in your programs.
Chapter 15 lists 991

1 . I he link held in the last node
of a linked list must have a null
value.
2. Memory must be allocated for a
node before you add the node
linked list . to a
3. Be sure to tree memory after you

delete a node Irom a linked list .
•4. You must
create an em pty linked list ( by assigning NULL
to the header
pointer ) before using the functions introduced
in this chapter.
5. Remember that a null link means there is no
pointer; therefore , you can -
not use it to dereference another node (or
any other object ) . For example,
the following code creates a run - time error because when the loop
termi-
nates, the value of pCur is NULL .
while ( pCur != NULL)

{
-
pCur = pCur >link ;
- _
printf( " %d ", pCur >data.member name); // ERROR
} // while
6. It is a logic error to allocate a node in the heap and not test for overflow.
7. It is a logic error to refer to a node after its memory has been released
with free.
8. It is a logic error to set the only pointer to a node to NULL before the
node has been freed . The node is irretrievably lost .
9. It is a logic error to delete a node from a linked list without verifying that
the node contains the target of the delete.
.
10 It is a logic error to fail to set the head pointer to the new node when a
node is added before the first node in a linked fist . The new node is irre-
trievably lost .
to a new node
11 . It is a logic error to update the link field in the predecessor
before pointing the new node to its logical successor. This error results in
a never-ending loop next time the list is traversed.
in the last node to NULL . This
12. It is a logic error to fail to set the link field
the end of the list .
causes the next traversal to run off
node pointer in a linked list search
.
13 It is a potential logic error to use the
following statement , the compares
before testing for a null pointer. In the memory access.
need to be reversed to prevent an invalid
while ((*pCur) >data.key

< target
-
&& *pCur I = NULL)
15.10 Summary
15.9 Key Terms

abstract data type (ADT) —
last in first out ( LII ( ) )
ancestor line
arc linear list
atomic data link
binary search tree ( BST) linked list
binary tree list
-
breadth first traversal loop
child metadata
chronological list node
composite data nonlinear list
connected outdegree
cycle parent
data structure pop
degree push
depth queue
depth -first traversal rear
dequeue root
descendant se 11 - referential structure
direct graph stack
edge strongly connected
enqueue top
first in —first out ( FIFO) traversal
front tree
graph underflow
indegree vertex
inorder traversal weakly connected
key-sequence list weighted graph
15.10 Summary
U List can be divided into linear lists and non - linear lists.
Linear lists can be divided into general lists , stacks , and queues.
Non - linear lists can be divided into trees and graphs.
In a general linear list , data can be inserted anywhere, and there are no
restrictions on the operations that can be used to process the list .
Four common operations are associated with general linear lists: insertion ,
deletion , retrieval , and traversal .
When we want to insert into a general linear list we must consider
four cases:
a . adding to the empty list
b. adding at the beginning
Chapter 15 Lists 993 TH
c. adding at the middle
d. adding at the end
When we want to delete a node from a general
sider two cases: delete the first node or linear list , we must con -
delete any other node.
Traversing a general linear list means going through the
and processing each item. list, item by item,
J A stack is a linear list in which all

additions are restricted to one end ,
called the top. A stack is also called a LIFO list .
- I We discuss two basic operations for stack: push
and pop.
Push stack adds an item to the top ol the stack. .After the push ,
the new-
item becomes the top.
Pop stack removes the item at the top of the stack. After the top the next
item , if any, becomes the top.
A queue is a linear list in which data can be inserted at one end called the
rear, and deleted from the other end, called the front.
We discussed only two main operations for queues in this chapter:
enqueue and dequeue.
The enqueue operation inserts an element at the rear of the queue.
The dequeue operation deletes the element at the front of the queue.
A tree consists of a finite set of elements called nodes and a finite set of
directed lines called branches that connect the nodes.
A node in a tree can be a parent , a child , or both . Two or more nodes with
the same parents are called siblings.
J A binary tree is a tree in which no node can have more than two children .
J
-J In a preorder traversal of a tree, we process the left subtree first , followed

by the root and then the right subtree.
the
In the inorder traversal of a tree , we process the root first, followed by
left subtree and then followed by the right subtree .
A binary search tree is a binary tree with the following properties
:
root.
a . All items in the left subtree are less than the
or equal to the root .
b. All items in the right subtree are greater than
c. Each subtree is itself a binary search
tree
vertices, and a collection of line
A graph is a collection of nodes , called
connecting a pair of nodes.
segments, called edges or arcs,
. In a directed graph , or digraph ,
Graphs may be directed or u ndirected graph , there is no direction on
each line has a direction . In an undirected
is called an arc.
the lines. A line in a directed graph
There are two standard graph traversals: depth first and breadth first .
994 Section 15.11 Practice Set
15.11 Practice Set

Review Questions
1 . In a linked list implementation , each node must contain data and a link
field .
a. True
b. False
2. In a linked list implementation , there is a need lor a head pointer to iden -
tify the beginning of the list .
a. True
b . False
3. In a linked list implementation , the first step in adding a node to a list is
to allocate memory for the new node.
a . True
.
b False
4 . The C language provides a list structure .
a . True
b. False
3. In an array, the sequentiality of a list is maintained by
a . the order structure of elements
h . a pointer to the next element
c. either a or b
d . neither a nor b
6. is an ordered collection of data in which each ele-
ment contains the location of the next element or elements.
a . An array
b. A structure
c. A linked list
d . None of the above
7. A stack is a structure.
a . first in-last out
b. last in-first out
—
c. first in first out
d. last in-last out
.
8 A queue is a structure .
—
a . first in last out
.
b last in-first out
c. first in-first out
. —
d last in last out
c>. A general linear list is a list in which operations, such as retrievals, inser-
tions, changes, and deletions, can he done
a. anywhere in the list
I ). only at the beginning
c. only at the end
cl . only at the middle
10. A stack is a list in which operations, such as retrievals, insertions,
changes, and deletions, can be done
a. anywhere in the list
b. only at the top
c. only at the base ( bottom )
.
d only at the middle
1 1 . A queue is a list in which operations, such as retrievals , insertions,
changes, and deletions, can be done
a. only at the beginning
h . only at the end
c. only at the middle
d . none ol the above
in
1 2. A is a last in-first out ( LIFO ) data structure
are restricted to one end , called the top .
which insertion and deletions
a. stack
b. queue
c. tree
d. binary tree
13. A
_ is a first in-first out ( FIFO) data, andstructure in
deletions
end , called the rear
which insertions are restricted to one front .
are restricted to another end
. called the
a. stack
b. queue
c. tree
.
d binary tree operation.
a stack, we use the
14. To add an element to
a. pop
h. push
.
c enqueue
d. dequeue use the
from a stack , we
15. To delete an element
operation .
a . pop
b. push
c . enqueue
d . dequeue
16. To add an clement to a queue, we use the operation .

.
a pop
b. push
c. enqueue
d . dequeue
17. To delete an element Irom a queue, we use the
operation .
a . pop
b. push
c. enqueue
d. dequeue
18. Data that consist of a single , nondecomp osable entity arc known as
a . atomic data
b. composite data
c. derived data
d . standard data
e. structured data
.
19 A( n ) is a collection of elements and the relationship
among them .
a. abstract data type
b. array
c. data structure
d . standard type
e. type definition
Exercises
20. Imagine we have the general list shown in Figure 15 - 54 . Show what hap -
pens if we apply the following statement to this general list:
pList = pList -> link;
pList
FIGURE 15- 54 Figure for Exercise 20
What is the problem with using this kind of statement ? Does it jus -
tify the need for two walking pointers (pPre and pCur) that we intro-
duced in the text ?
"1
Lcur .
' • haVe the 8eneral list shown in Figure 15 - .
WC
55 As discussed in
W w
,
'
rent (peur ) node K
rcn )
' 1
lh’' Th ^“ ' b bk >» P
* of the Sack
(pPre ) and the location ~
cur-
based on search criteria. A typical search design is
shown in Figure 15 - 55 .
FIGURE 15- 55 Figure for Exercise 21
I he following code to set pPre and pCur contains a common error.

\ \ hat is it, and how should it he corrected? ( Hint : What are the contents
of these pointers at the beginning of the search?)
pCur = -
pCur > link ;
pPre = -
pPre > link ;
.
22 Imagine we have a dummy node at the beginning of a general list The .
dummy node does not carry any data. It is not the first data node; it is an
empty node. Figure 15 - 56 shows a general list with a dummy node. Write
the code to delete the first node ( the node after the dummy node) in the
general list.
Dummy a
list
pPre pCur
FIGURE 15 - 56 Figure for Exercise 22
,w*,
2 ke
.
on a general list Ho
node simplify the operation
with a dummy node. Write the
an empty general list
24’ Figure 15 - 57 shows general list .
code to add a node to this
empty
] Section 15.11 Practice Set
Dummy
pPre
FIGURE 1 5 - 57 Figure for Exercise 24
25. Write the statements to ad cl a node in the middle of a general list with
the dummy node ( see Exercise 22 ) . Compare your answer with the
answer to Exercise 24. Arc they the same? What do you conclude? Does
the dummy node simplify the operation on a general list ? I low?
.
26 Imagine we have the two general lists shown in Figure 15 - 58. What
would happen if we apply the following statement to these two lists?
listl = list2 ;
—
listl
H —
Iist 2
.
27 \ \ hat would happen if we apply the following statements to the two lists
in Exercise 26?
temp = listl ;
-
while ( temp >link 1 = NULL)
-
temp = temp >link ;
-
temp > link = list2 ;
28. Imagine we have the general list shown in Figure 15 - 59. What would
happen if we apply the following statements to this list?
temp = list;
-
while (temp > link != NULL )
-
temp = temp >link ;
-
temp >link = list;
Dummy
list
29. Imagine we have two empty stacks of integers , si and s2. Draw a picture
of each stack after the following operations:
Algorithm Exercise 29
1 pushStack (si, 3)
2 pushStack (si , 5 )
3 pushStack (si , 7)
4 pushStack (si, 9 )
7 loop not emptyStack (si)
1 popStack (si , x)
2 pushStack (s2 , x )
8 end loop
End Exercise 33
a queue of integers , Q . Draw
30. Imagine you have a stack of integers, S and
,
following operations:
a picture of S and Q after the
Algorithm Exercise 30
1 pushStack (S, 3)
2 pushStack (S, 12)
3 enqueue ( Q , 5 )
4 enqueue ( Q , 8 )
5 popStack (S, x )
6 pushStack ( S, 2 )
7 (Q / x )
enqueue
8 dequeue (Q/ y)
9 pushStack (S, x )
10 pushStack ( S, y )
End Exercise 30
of queue Q1 and queue Q2 are as shown . ?What
31. Imagine that the contents code is executed The
, „f 03 after the following
>">» <« > »“ r
31
Algorithmi Exercise
1 Q3 = createQueue
2 count continual
3 y = o
4 loop ( not empty Q 1 or not empty Q2)
1 dequeue (Ql , x )
1 if ( count > y )
1 dequeue (Q2, y)
2 end if
3 if (y equal count )
1 enqueue (Q3, x)
4 end if
5 count = count + 1
5 end loop
End Exercise 31
Ql : 42 30 41 31 19 20 25 14 10 1112 15
Q 2: 4 5 6 10 13
32. Draw all possible non similar binary trees with three nodes ( A , B , C ).
-
33. Draw all possible binary search trees for the data elements 5 , 9 , and 12.
34. Create a binary search tree using the following data entered as a sequen -
tial set :
14 , 23 , 7, 10 , 33 , 56 , 80 , 66 , 70
35. Create a binary search tree using the following data entered as a sequen -
tial set :
7 , 10 , 14 , 23 , 33, 56 , 66 , 70 , 80
36. Insert 44 and 50 into the tree created in Exercise 34 .

37 . Insert 44 and 50 into the tree created in Exercise 35 .
Problems
38. Write a program that reads a list of integers from the keyboard , creates a
general list from them, and prints the result .
39. Write a function that accepts a general list , traverses it , and returns the
key ol the node with the minimum key value .
40. Write a function that traverses a general list and deletes all nodes whose
keys are negative.
41 . Write a function that traverses a general list and deletes all nodes that
are after a node with a negative key.
42. W rite a function that traverses a general list and deletes all nodes that
are before a node with a negative key.
43. Rewrite the function
deleteNode (see Program 15-2) using a general list
with a dummy node.
44. Rewrite the function searchList (
see Program 15-3) using a general list
with a dummy node.
45. Write a function that returns a pointer
to the last node in a general list.
.
46 Write a function that appends two general
lists together.
47. \ \ rite a I unction that appends a general list to
itself .
48. One of the applications of a stack is to backtrack
—
that is, to retrace its
steps. As an example, imagine we want to read a list of items, and each
time we read a negative number we must backtrack and print the five
numbers that come before the negative number and then discard the
negative number. Use a stack to solve this problem . Read the numbers
and push them into the stack ( without printing them ) until a negative
number is read . At this time , stop reading and pop five items from the
stack and print them. If there are fewer than five items in the stack, print
an error message and stop the program . After printing the five items,
resume reading data and placing them in the stack. When the end of the
file is detected , print a message and the items remaining in the stack.
Test your program with the following data:
- -
1 2 3 4 5 1 1 2 3 4 5 6 7 8 9 10 2 11 12 3 1 2 3 4 5 -
49. \ \ rite a function called copyStack that copies the contents of one stack
into another. The algorithm passes two stacks, the source stack and the
destination stack. The order of the stacks must be identical. ( Hint: Use a
temporary stack to preserve the order.)
50. Write a function , catStack , that concatenates the contents of one stack
on top of another.
identi -
function to check whether the contents of two stacks are
;
5 I . Write a
should be changed . You need to write affunotion that
cal. Neither stack works .
your function
prints the contents of a stack to verify that
copies the contents of one queue
52. Write a function called copyQueue that
to another .
that creates a queue from a stack.
53. Write a function called stackToQueue
top of the stack should be the front
After the queue has been created the
,
should be the rear of the queue. At
of the queue and the base of the stack be empty.
the end of the function, the stack
should
, write a function that calculates
and prints the
.
54 Given a queue of integers
integers in the queue without changing the
sum a:nd the average of the
,
contents of the queue.

Section 15.11 Practice Set
55. Given a queue of integers, write a function that deletes all negative inte-
gers without changing the order of the remaining elements in the queue.
56 . Write a function that calculates and passes up to the calling function the
sum and average of the nodes in a tree.
57. Write a function that counts the number of leaves in a binary tree .
58. Write a function to find the smallest node in a binary search tree .
Projects
59. Write a program that reads a file and builds a key-sequenced general list .
After the list is built , display it on the monitor. You may use any appropri -
ate data structure, but it must have a key field and data. Two possibilities
are a list of your favorite CDs or your friends' telephone numbers.
.
60 Write a program to read a list of students from a file and create a general
list . Each entry in the general list is to have the student ’s name, a pointer
to the next student , and a pointer to a general list of scores. You may have
up to four scores for each student! . be program initializes the student list
by reading the students’ names from the text file and creating null scores
lists. After building the student list , it loops through the list , prompting
the user to enter the scores for each student . The scores prompt is to
include the name of the student.
61 . W rite a stack and queue test driver. A test driver is a program created to
test functions that are to be placed in a library. Its primary purpose is to
completely lest functions; therefore, it has no application use . The func -
tions to be tested are push stack , pop stack, enqueue, and dequeue. You
may include other stack and queue functions as required . All data should
be integers. You need two stacks and two queues in the program , as
described below.
a . Input stack: used to store all user input
b. Input queue: used to store all user input
c. Output stack: used to store data deleted from input queue
d . Output queue: used to store data deleted from input stack
Use a menu - driven user interface that prompts the user to select
either insert or delete. If an insert is requested , the system should prompt
the user for the integer to be inserted. The data are then inserted into the
input stack and input queue. II a delete is requested , the data are deleted
Irom both structures: the data popped from the input stack are enqueued
in the output queue, and the data dequeued from the input queue are
pushed into the output stack.
Processing continues until the input structures are empty. At this
point print the contents ol the output stack while deleting all of its data.
Label this output “Output Stack,” then print all of the data in the output
TT1
queue while deleting all of its data. Label this output "Output Queue. ”
Your output should he formatted as shown below.
Output Stack: 18 9 13 7 5 1
Output Queue: 7 13 9 18 5 1
Test your program with the following operations:
1 input 1 5 delete 9 input 6 13 input 8

2 input 2 6 input 0 10 delete 14 delete
3 delete 7 input 5 11 input 7 15 delete
8 delete 12 delete 16 delete
4 input 3
test , write a short
In addition to the computer output from your
structural concepts were
report ( less than one page) describing what
demonstrated by your output .
L
Character Sets
Computers use numbers. They store characters by assigning a
number to
each one. The C language was originally designed using
the American Stan -
dard Code for Information Interchange (ASCII ), which was the
standard of
the time. Based on the Latin alphabet , ASCII uses 128 characters (valued
horn 0 to 1 2 ) stored as / - hit numbers. This is enough to handle the lower
/
and uppercase letters , digits, most common punctuation characters, and some
control characters. Later, an attempt was made to extend the ASCII character
set to 8 hits . The new code, which was called Extended ASCII , was never
internationally standardized.
To overcome the limitations inherent in ASCII and Extended ASCII , the
International Standard Organization ( ISO ) and the Unicode Consortium ( a
consortium of manufacturers of multilingual software) created a universal
encoding system to provide a comprehensive character set. The Unicode con -
sortium created Unicode. ISO created the Universal Character Set ( UCS ),
designated ISO 10646 standard.
The main UCS set, sometimes referred to as UCS-4, is a four- bvte code.
A subset of UCS - 4, called UCS-2 , is a two-byte code that is compatible with
UCS- 4 ; it contains the first 32 ,364 values in UCS-4.
Unicode was originally a two- byte character set. The early versions of
Unicode were only compatible with UCS-2. Unicode version 3 is a four- byte
code and is fully compatible with UCS- 4. The compatibility of Unicode with
UCS- 4 makes Unicode compatible with ASCII and Extended ASCII.
The ASCII set , which is now called Basic Latin, is included in UCS- 4
, which
and Unicode with the upper 25 bits set to zero. The Extended ASCII
is now called Latin - 1 , is also included w ith 24 upper bits set to zero. Figure A- 1
shows the compatibility between the different systems .
1006 Section A . 1 Unicode
Extended ASCII
Defining Planes ASCII
8 bits 8 bits 8 bits 8 bits
UCS- 2
*
UCS-4 and Unicode
FIGURE A - l USC and Unicode Compatibility
A. l Unicode
The prevalent code today is Unicode. Each character or symbol in this code
is defined by a 32- hit number. The code can define up to 2 ^ 2 ( 4, 294 ,967,296 )
characters or symbols . The presentation uses hexadecimal digits in the lol -
Iowing format :
u -xxxxxxxx
where each X is a hexadecimal digit . Therefore, the numbering goes from
U -00000000 to U - FFFFFFFF.
Planes
Unicode divides the whole space code into planes. The most significant 16
hits define the plane, which means we can have 65,536 planes. For plane 0,
the most significant 16 hits arc Os (0x0000 ) ; in plane 1 , the hits are 0x0001 ;
in plane 2 , they are 0x0002; and so on until in the plane 65 , 536 , they are
OxFFFF.
Each plane can define up to 65, 536 character or symbols. Figure A- 2
shows the structure of Unicode spaces and its planes.
Basic Multilingual Plane ( BMP )

Basic Multilingual Plane, Plane 0 , is designed to he com patible with the
previous 16 - hit Unicode version and UCS - 2 . The most significant 16 hits in
.
this plane are all zeros The codes are normally shown as U +XXXX with the
understanding that XXXX defines only the least significant 16 hits. This plane
mostly defines character sets in different languages with the exception of
some codes used for control or other special characters. Table A- l shows the
main classification of codes in Plane 0.
Reserved
Reserved
U-
u. 2 -o IL
o UJ
°oo
Q
o
u. o °o O
o o
o
Plane 0000: Basic Multilingual Plane (BMP)
Plane 0001: Supplementary Multilingual Plane (
SMP)
Plane 0002: Supplementary Ideographic Plane (
SIP)
Plane 000E: Supplementary Special Plane (SSP)
Plane 000F: Private Use Plane (PUP)
Plane 0010: Private Use Plane (PUP)
FIGURE A - 2 Unicode Planes
Range Description
A-Zone ( Alphabetical Characters and Symbols)
U+0000 to U+00FF Basic Latin and Latin- 1
U+0100 to U+01FF Latin Extended
U+0200 to U+02FF IPA Extension, and Space Modifier Letters
U+0300 to U+03 FF Combining Diacritical Marks, Greek
U+0400 to U+04FF Cyrillic
U+0500 to U+05 FF Armenian, Hebrew
U+0600 to U+06FF Arabic
U+0700 to U+08 FF Reserved
U+0900 to U+09FF Devanagari, Bengali
U+0A00 to U+OAFF Gurmukhi, Gujarati
U+0B00 to U+OBFF Oriya, Tamil
U+0C00 to U+OCFF Telugu, Kannda

U+0D00 to U+ODFF Malayalam
continued
TABLE A - 1 Unicode BMP

1008 Section A.l Unicode
Range Description
U+OEOO to U+OEFF Thai, Lao
U+OFOO to U+OFFF Reserved
U+ lOOO to U+ l OFF Georgian
U+ l 100 to U+ l IFF Hangul Jamo
U+ l 200 to U+ 1 DFF Reserved
U + 1 E00 to U+ lEFF Latin extended additional
U + l FOO to U+ 1 FFF Greek extended
U + 2000 to U+ 20FF Punctuation, sub / superscripts, currency, marks
U+ 2100 to U+21 FF Letter-like symbols, number forms, arrows
U+2200 to U+ 22FF Mathematical operations
U + 2300 to U+23 FF Miscellaneous technical symbols
U+2400 to U+24FF Control pictures, OCR, and enclosed alphanumeric
U+ 2500 to U+ 25 FF Box drawing, block drawing, and geometric shapes
U+ 2600 to U+ 26FF Miscellaneous symbols
U+ 2700 to U+ 27FF Dingbats and Braille patterns
U+2800 to U+ 2 FFF Reserved
U + 3000 to U+ 30FF CJK symbols and punctuation, Hiragana, Katakana
U+ 3100 to U+31 FF Bopomofo, Hangul Jamo, CJK Miscellaneous
U + 3200 to U+32 FF Enclosed CJK letters and months
U+ 3300 to U+33 FF CJK compatibility
U+3400 to U+4DFF Hangul
1-Zone (Ideographic Characters)
U+ 4E00 to U+9FFF CJK Unified Ideographic
O-Zone (Open)
U +A000 to U+DFFF Reserved
R -Zone (Restricted Use)
U+E000 to U+F 8FF Private Use
continued
TABLE A - l Unicode BMP (continued )
Range Description
U+F 900 to U+FAFF CJK Compatibility Ideographs
U+FBOO to U+FBFF Arabic Presentation Form-A
U+FCOO to U+FDFF Arabic Presentation Form-B
U +FEOO to U+FEFF Half marks, small forms
U +FFOO to U+FFFF Half-width and full-width forms
TABLE A - l Unicode BMP ( continued )
Supplementary Multilingual Plane (SMP)

Supplementary Multilingual Plane, Plane 1, is designed to provide more
code for those multilingual characters that are not included in the BMP plane.
Supplementary Ideographic Plane (SIP)

code for
Supplementary Ideographic Plane, Plane 2 , is designed to preside ( or meaning)
symbol that primarily denotes an idea
ideographic symbols any
in contrast to a sound ( or pronunciation ) .
Supplementary Special Plane (SSP )

) is used for special char-
Supplementary Special Plane , Plane 14 OxOOOE1 codes.
(
acters not found in the Basic Latin

or Basic Latin -
Private Use Planes ( PUP)

( OxOOOF ) and 16 ( OxOOlO ) arc
reserved for
.
Private Use Planes Planes 15
private use .
A.2 ASCII
The American Standard designed
Code f ,
code f() r 28 symbols. Today
,
seven - bit code, that was

ASCII or Basic Latm, is part
o £ code. It occupies the first 128 codes
Tabk' A- 2 contains the decimalif
,
to U 0
Unicode ( U-00000000 wjth an English interpretation >
’
in
integer .
when converted to an
1010 Section A . 2 ASCII
Decimal Hex Symbol Interpretation

0 00 null Null value
1 01 SOH Start of heading
2 02 STX Start of text
3 03 ETX End of text
4 04 EOT End of transmission

5 05 ENQ Enquiry
6 06 ACK Acknowledgment
7 07 BEL Ring bell
8 08 BS Backspace
9 09 HT Horizontal tab
10 0A LF Line feed
11 0B VT Vertical tab
12 0C FF Form feed
13 0D CR Carriage return
14 0E SO Shift out
15 OF SI Shift in
16 10 DLE Data link escape
17 11 DC 1 Device control 1
21 15 NAK Negative acknowledgment
22 16 SYN Synchronous idle
23 17 ETB End of transmission block
24 18 CAN Cancel
25 19 EM End of medium
26 1A SUB Substitute
continued
TABLE A - 2 ASCII Codes
Appendix A Charocfer Sets 1011
n
Decimal x Symbol Interpretation
27 IB ESC Escape
28 1C FS File separator
29 ID GS Group separator
30 IE RS Record separator
31 IF US Unit separator
32 20 SP Space
33 21
34 22 Double quote
35 23 #
36 24 $
37 25 %
38 26 &
39 27 Apostrophe
40 28
41 29
42 2A
43 2B +
44 2C Comma
45 2D Minus
46 2E
47 2F /
48 30 0
49 31
50 32 2
51 33 3
52 34 4
53 35 5
continued
TABLE A - 2 ASCII Codes (continued!

’I
1012 Section A.2 ASCII
54 36 6
55 37 7
56 38 8
57 39 9
58 3A Colon
59 3B Semicolon
60 3C <
61 3D
62 3E >
63 3F 2
64 40 @
65 41 A
66 42 B
67 43 C
68 44 D
69 45 E
70 46 F
71 47 G
72 48 H
73 49
74 4A J
75 4B K
76 4C L
77 4D M
78 4E N
79 4F O
80 50 P
continued
TABLE A - 2 ASCII Codes (continued )

81 51 Q
82 52 R
83 53 S
84 54 T
85 55 U
86 56 V
87 57 W
88 58 X
89 59 Y
90 5A Z
91 5B [ Open bracket
92 5C \ Backslash
93 5D ] Close bracket
94 5E A Caret
95 5F Underscore
96 60 Grave accent
97 61 a
98 62 b
99 63 c
TOO 64 d
101 65 e
102 66 f
103 67 9
104 68 h
105 69
106 6A
107 6B k
continued
TABLE A - 2 ASCII Codes (continued)

'1
1014 Section A.2 ASCII

108 6C I
109 6D m
110 6E n
111 6F o
112 70 P
113 71 q
114 72 r
115 73 s
116 74 t
117 75 u
118 76 v
119 77 w
120 78 x
121 79 y
122 7A z
123 7B { Open brace
124 7C Bar
125 7D } Close brace
126 7E Tilde
127 7F DEL Delete
TABLE A - 2 ASCII Codes ( continued )
Some Properties of ASCII

ASCII has some interesting properties that we need to briefly mention here:
1 . I he lirst code (0), is the null character, which represents the absence ol
.
any character
. .
2 I he lirst 32 codes, 0 to 3 I , and the last code, 2 5 5 , are control characters
They can he checked with the iscontrol function .
.
3 I he space character, which is a printable character, is at
position 32 .
4. The uppercase letters start from 65 ( A ). The lowercase

T ( a ). When numerically compared, uppercase letters are
letters start from
ues the uppercase letters show before the lowercase letters .

smaller than
lowercase ones. This means that when we sort a list based on ASCII val-
5 . I he uppercase and lowercase letters differ by only one bit in the seven
-
n
hit code. For example , character A is 1000001 (0x 41 ) and character a is
1 100001 ( 0x61 ). The difference is hit 6, which is 0 in uppercase letters
and 1 in lowercase letters. II we know the code lor one case, we can find
easily the code for the other by adding or subtracting 32 in decimal , 0x20
hexadecimal , or Hipping the sixth hit .
in
6. I he uppercase letters are not immediately followed by lowercase letters.
I here are some punctuation characters in between.
7. Digits ( 0 to 9 ) begin at 48 ( 0x30 ). This means that if we want to change a
numeric character to its face value as an integer, we need to subtract 48
from it .
A.3 Universal Encoding in C

The C language allows the use of the UCS- 2 and UCS-4 codes in character
constants, string constants, and identifiers. C uses two formats , one with
four
hexadecimal digits for
hexadecimal digits for UCS- 2 and another with 8
compatible \ U 000 OHK «H
UCS- 4 as shown below. Note that the two formats are ;
is the same asWmw.
\unnnn
// For UCS 2 -
\Unnnnnnnn
// For UCS 4 -
hical character using the uni-
Program A- l shows how we can print a grap
versal encoding in C.
PROGRAM A - 1 Example of Universal Encoding

1 /* Show the use of universal encoding
2 Written by:
3 Date:
4 */
# include <stdio.h
>
5 i
|
6
7 int main ( void )
8 {
9 // Declarations
10 char chi = \ u 0 0A2 ' ;
char ch2 = \ U 000000 A 3 ' ;
11 con tinned
71
1016 Section A.3 Universal Encoding in C
PROGRAM A - l Example of Universal Encoding ( continued )

12
13 // Statements
14 printf ( "Character chi: %c\ n " , chi );
15 printf ( "Character ch 2: %c\n" / ch 2 );
16
17 return 0 ;
18 > // main
Result:
Character chi: C
Character ch2: £
r
Keywords
rhe C language contains 37 keywords, also known as reserved words, that
cannot be used as identifiers for functions, variables, or named constants.
They are shown in Table B-l.
auto extern short
Bool float signed
break for sizeof

goto static
case
if struct
char
Jmaginary switch
_Complex
inline typedef
const
union
continue int
long unsigned
default
register
void
do
restrict
volatile
double
return
while
else
enum
TABLE B - l C Keywords
1018 Section
In addition to these keywords, we recommend that the macros and C + +

.
keywords in Table B - 2 also be treated as keywords
and compl or
and_eq complex OR EQ
bitand imaginary struct
bitor not xor _eq
bool not_eq
TABLE B - 2 C Macros and C++ Keywords

H
A flowchart is a tool to show the logic flow of a program. Although

it is gener-
ally considered a computer programming tool , it has been used for
many
other purposes. Progra mmers use flowcharts to design a complete program or
just part of a program . Depending on the programming language, the parts
can be called such things as procedures ( Pascal ), functions ( C ), or para
-
graphs ( COBOL ). We will use the general term algorithm to indicate any part
of a program that needs design .
ITie primary purpose of a flow chart is to show the design of an algorithm .
I he flowchart also frees the programmer from the syntax and details of a pro-
gramming language w hile allow ing him or her to concentrate on the details of
the problem to be solved .
—
A flowchart gives a pictorial representation of an algorithm in contrast
to another programming design tool , pseudocode, which provides a textual
design solution . Both tools have their advantages, but a flowchart has the pic-
torial power that other tools lack. A new student of computer science must
learn how to think about an algorithm before writing it . A flowchart is a tool
for this pictorial thinking.
Cl Auxiliary Symbols
the read -
A flowchart is a combination of symbols. Some symbols enhance
they do not directly show instruc-
ability or functionality of the flowchart;
'
tions or commands. They show the start and stop points . the order and
how one part of a flowchart is connected to
sequence of actions, and
another. These auxiliary symbols are shown in
Figure -
C l .
1020 Section C.l Auxiliary Symbols
SYMBOL NAME APPLICATION
Shows the beginning or

( ) Terminal ending of an algorithm
Shows the action order

Flow Line in an algorithm
Q Connector Shows connecting points

in an algorithm
FIGURE C - l Auxiliary Symbols
An oval shows the beginning or ending of an algorithm. When it is used

to show the beginning of an algorithm, we write the word START inside the
oval. When it is used to show the ending of an algorithm, we write the word
STOP or RETURN in the oval.
One of the first rules of structured programming is that each algorithm
should have only one entry point and one exit point. Thus, a good structured
flowchart should have one and only one START and one and only one STOP.
Ovals should he aligned to show clearly the flow of the actions in an algo-
rithm. For example, a flowchart for a program that does nothing is shown in
Figure C - 2. This program starts and stops without doing anything.
( START )
( STOP }
FIGURE C - 2 Use of the Start - Stop Symbols
Flow Lines
Flow lines show the order or sequence of actions in a program. They connect
symbols. Usually a symbol has some entering and some exiting lines. I he
S FAR I oval has only one exiting line. The STOP oval has only one entering
line. We have already shown the use of flow lines in Figure C - 2. We will show
other flows in the examples that follow.
'
Appendix C Flowchorfing 1021
Connectors
Wc use only one connector symbol, a
circle with a number in it, to show con
nectivity. It is used in two situations. -
1• When
at
. . .
_ we reach the end of a column or
,
0 VV cont nues at the t0
page, but our chart is not fin-
shcd In this case, at the bottom of the flow
we use a connector to show
P the next column or page. This condi-
tion is shown in Figure C-3 ( a).
( START
.
)
o
(
.
STOP )
(a) Simple Flow Connectors (b) Insert Flow Connectors
FIGURE C - 3 Use of Connectors
2. When we need to show logic details that do not fit in the flow. This is
much like an inset in a map used to magnify a part of the map. In a
flowchart, it magnifies part of the logic flow. In this case, we use two
.
sets ol connectors I he first set connects the main flow to the entry
point in the magnified flow; the second set connects the end of the mag-
nified flow to the return point in the main flow. This condition is shown
in Figure C -3 (b) .
In either case, the number in the connector can be a simple
serial num-
and number in
her for an on-page flow, or it can be a combination of a page
the form page.number for an off page
- flow .
C. 2 Primary Symbols
to show the instructions or actions
^ ^ ^^ ^
needeXtc soLe hcproblt '
pre ented n the algorithm. With
these symbols,
"Ipossible represent all three structured
to
.
programmmg constructs:
sequence decision, and repetition

1022 Section C . 2 Primary Symbols
Sequence
Sequence statements simply represent a series of actions that must continue
.
in a linear order Although the actions represented in the sequence symbol
may be very complex, such as an input or output operation, the logic flow
must enter the symbol at the top and flow out at the bottom. Sequence sym -
bols do not allow any decisions or flow changes within the symbol .
Flowcharts use four sequence symbols: assignment, input/output, mod-
ule call, and compound statement. They are shown in Figure C - 4.
Assignment Statement Module Call
/
7 Input/Output Statement Compound Statement
FIGURE C - 4 Sequence Symbols
Null Statement
It is worth noting that do nothing is a valid statement . It is commonly referred
to as a null statement. The null statement is considered a sequence state-
ment , since it cannot change the flow direction of a program. There is no
symbol lor a nidi statement. It is simply a flow line. Figure C - 2 is an example
of a null statement .
Assignment Statement
The assignment statement is shown using a rectangle. Within the assignment
symbol, the assignment operator is shown as a left -pointing arrow. At the
right side of the arrow is an expression whose value must be stored in the
variable at the left side. Figure C - 5 shows an assignment statement.
variable expression
FIGURE C - 5 The Assignment Statement

Input/Output Statement
A parallelogram shows any
input or output, such as reading
or writing on the system from a keyboard
console. For example, an algorithm
value of two variables from the keyboard that reads the
and then writes their values on the
screen in reverse order is shown in
Figure C-6.
( START
^
j READ ( A )
j
l READ ( B )
J
j WRITE ( B )
j
j WRITE ( A )
j
l
STOP
FIGURE C - 6 Read and Write Statements
Module- Call Statement

I he symbol used for a module call is a rectangle with two vertical bars
.
inside 1he flowchart for the called module must be somewhere else. In other
words, each time you see a module call statement, look for another flowchart
with the module name.
In C programs, the module-call symbol is used primarily for void func -
tions. Functions that return values are shown as assignments or other expres-
sions in a sequence symbol.
To show how a module call is used in a program, let s design the flow-
'
chart for a program that calculates and prints the average of three numbers.
First , the flowchart for the called
This example has two noteworthy points :
module ( AVRG) does not begin with START . Rather , it shows the name of the
RE I URN , indicat -
module and the parameter list. Also, the exit oval contains
but , rather a return from a called
ing that it is not the end of the program
,
module. The resulting flowchart is shown in Figure C - /.

1024 Section C.2 Primory Symbols
c START (AVRG ( rslt, x, y, z ))
READ ( a )
7 sum
x+y+z
—
READ ( b )
i rslt
sum / 3
READ ( c )
I RETURN
AVRG ( ave, a, b, c )
WRITE ( ave )
I
/
STOP
FIGURE C - 7 Module Call Example
Compound Statement
Although flowcharts have no actual symbol to show a compound statement,
we encapsulate all statements that make a compound statement in a broken-
line rectangle. In a C program, compound statements represent a block ol
code, code that is enclosed in braces. In a flowchart , we use a dashed rectan-
gle to enclose a compound statement. An example of this is shown in Figure C -9.
Selection Statements
Unlike the sequence statements, conditional statements can cause the flow
of the program to change. They allow the execution of selected statements
and the skipping ol other statements. Structured programming has two selec -
tion statements: two - way and multiway selection.
Two - Way Selection

I lie two-way symbol is the diamond. When it is used to represent an if ...else
statement, the true condition logic is shown as the right leg of the logic flow,
and the false condition, if present, is shown on the left leg of the logic flow. With
the if ...else , there must always be two logic flows, although often one of them
is null. ( Remember that the null

statement is represented by a flow line; there
is no symbol lor null.) Finally, the
statement ends with a connector where the
true and false flows join. In this case,
the connector has nothing in it .
Although you will often see decisions drawn with the
tom of the diamond , this is not good style.
flow from the bot -
Even w hen one of the flows is null,
it still must flow from the left or right sides
of the diamond .
I igure C -8 shows the use ol the decision symbol in
the if...else state-
ment . As we pointed out, a decision always has
two branches. On each
branch, we are allowed to have one and only one statement. Of course, the
statement in each branch can be a null or a compound statement. But only
one statement is allowed in each branch ; not less, not more. Also remember
that the w hole figure is only one statement, not two or three; it is one if ...else
statement .
false action true action

(one statement) (one statement)
T
FIGURE C - 8 i f ... else Statement
Let ’s design an algorithm that reads an integer. If the integer

’s value is
the original number and the result.
greater than 10, it subtracts 10 and writes
is
does nothing . The flowchart for this program
If the value is less than 10, it
seen in Figure C-9.
Multiway Selection
symbol used with structured pro-
The second application of the selection selection is
s implementation of multiway
gramming is multiway selection. C selection statement is nothing
multiway
the switch statement. Actually, the does
for the if...else statement . If a
language
more than a shorthand notation nested
the same logic is implemented
using
not have a switch statement
if ...else statements or the else- if construct .
1026 Section C. 2 Primary Symbols
START
READ ( aNum )
/
\aNum > 10y
N/ IR
newNum
aNum - 10
—
WRITE (newNum)
/
STOP
FIGURE C - 9 Example: Read and Subtract 10
Figure C - 10 shows the use o! the switch statement. As you can see, we
can have as many branches as we need. On each branch, we are allowed to
have one, and only one, statement. Of course, the statement in each branch
can he a null or a compound statement. But remember that only one state-
ment in each branch is allowed; not less, not more. Also remember that the
whole Figure is only one statement, not two or three; it is one switch state-
ment . ( II these rules sound familiar, remember that the switch statement is
nothing but a shorthand form of if . .. else , so the rules are the same for both.
Let ’s design an algorithm for a program that reads one character repre-
senting a letter grade and prints the corresponding grade point average
(GPA). Figure C - l 1 shows this design.
Looping Statements
C has three looping statements: for , while, and do ...while .
for Statement
.
I he for statement is a counter-controlled loop It is actually a complex state-
ment that has three parts, any of which can he null: ( 1 ) the loop initialization,
which normally sets the loop counter; ( 2 ) the limit test; and (3 ) the end-of-
loop action statements, which usually increment a counter. Since the for
.
statement is a pretest loop, the loop might not be executed If the terminating
condition is true at the start, the body of the for statement is skipped.
I
Appendix C Flowcharting 102 /
T1
expression
m n o P
m Action n Action o Action p Action
?
FIGURE C- 10 Multiway Selection Statement
( START )
READ ( grade )
1
grade
A B jc \D
gpa 4 0 gpa 3.0l gpa
^ ôj ^^^j
-2 gpa 9Pa ^ 0.0
"
WRITE ( gpa )
STOP
for Calculate Grade Point Average

FIGURE C - 1 1 Example: Design
As is the case in
loop can contain one
anJ 0n
structs, this one statement

can
- " e nuH or a compound statement. F gure C-12 .
shows the for construct
.
1028 Section C. 2 Primary Symbols
/ Initializationp
(Update \-
\ Condition M
Body
(One Statement )
\
FIGURE C - 1 2 for Statement
The for loop uses three actions: initialization , test , and update. However,
only two of these actions are required in each iteration flow. In the first flow,
when we enter the loop , only initialization and loop are used . This flow is
shown in Figure C - 13 ( a ). In all iterations except the first , only update and
test are used. Figure C - 13 ( h ) shows these flows.
1
Initialization Initialization
Update Update
TestV - ^ estV
^-
T T
Iteration Exit From Iteration Iteration Exit
( a) Initial flow (b) 2 ... n flow
FIGURE C -l 3 for Statement Flow
Let 's design an algorithm that will read 20 numbers and print their sum .
Since the number of times is known in advance, for is an excellent choice lor
the looping construct. I be design for this program is shown in Figure C - l 4.
while Statement
The second looping construct is the while statement. The major difference
between the for and while loops is that the while loop is not a counting loop.
Both are pretest loops; therefore, like the for , the body of the while loop may
never be executed .
We use the same basic symbol for the while loop, but since there is only a
limit test , the internal divisions are not necessary. Figure C - l 5 shows the
basic format of the while statement .
START
sum
— o
READ ( num )
i ++
i — 0\ F
i < 20
T
sum 4<
sum + num —
o Note that t h e "'
body of the loop is a
compound statement
(
WRITE ( sum )
l
STOP
7
)
FIGURE C- 14 Design to Read 20 Numbers
^ —
condition
Body
(One Statement)
FIGURE C - 15 while Statement Format
Let 's design another program that reads numbers from the keyboard and
we may be
prints their total. This time, we don t know how many numbers
’
reading. All we know is that all the numbers are positive . We can therefore
signal the end of the numbers by having the user key - 1 .
Since we don ’t know how many times we will loop
, we need a different
well designed for this type of logic . The program flow is
construct ; the while is
first number is read before
- .
seen in Figure C 16 Note that in this design
loop
,
and
the
is common in pretest loops.
the loop. This is know n as priming the
) Section C. 2 Primary Symbols
( START
sum o
/ num > 0
\—
M
F
U
sum
sum + num
/
/READ ( num
/WRITE ( sunn f
( STOP
I
^
)
FIGURE C - l 6 Program Flow to while Read and Total
c/o... w/7 /Ve Statement

The third application of the loop symbol is the do...while statement .
Because of the inherent differences between the for and while loops and
do ... while loop, it must he used differently in a flowchart . There are two
major differences between the while and the do...while:
1 . Awhile loop is a pretest loop. The do . .. while loop is a post - test loop.
..
2. I he body of a while loop may never he executed . The body of a do. while
loop is executed at least once.
Figure C - l 7 shows the use of the do. . .while statement . Note that in this
statement the condition is tested at the end of the loop.
Let ’s design an algorithm that reads and processes a number. In this algo-
rithm , the number must be between 1 and 5 . To make the program robust , we
use a do...while loop that forces the user to enter a valid number. This is a
common technique lor validating user input . Note that as in previous looping
constructs, the do . ..while loop can have only one statement. Therefore, we
enclose the prompt and read in a composite symbol . The algorithm is shown
in Figure C - l 8.
'
"1
Body
(One Statement )
, rTA
— condition \
(
I7
FIGURE C- l 7 do...while Statement
( START
/ WRITE /
/ ( 'Enter number " )i
j READ ( num )
J
I
(num < 1)
T
or
( num > 5 )
process ( num )
(not shown)
T
( STOP
with do. . .while

FIGURE C- 1 8 Example: Algorithm for Input Validation
Numbering Systems
Today the whole world uses the decimal number
system developed
by Arabian
mathematicians in the eighth century. We acknowledge their contribution to
numbers when we refer to our decimal system as Arabic
numerals. But deci-
mal numbers were not always commonly used.
The first to use a decimal
numbering system were the ancient Egyptians. The Babylonians improved on
the Egyptian system by making the positions in the numbering
systems mean-
ing! ul. But the Babylonians also used a sexagesimal (base 60) numbering
sys-
tem. Whereas our decimal system has 10 values in its graphic
representations,
.
a sexagesimal system has 60 We still see remnants of the Babylonians’
sexage-
simal system in time, which is based on 60 minutes to an hour, and in the
division of circles, which contain 360 degrees.
D. 1 Computer Numbering System

C omputer science uses several numbering systems. The computer itself uses
binary (base 2). A binary system has only two values for each number posi-
tion, 0 and 1. Programmers use a shorthand notation to represent binary
numbers, hexadecimal (base 16 ). And of course, programmers also use the
decimal system (base 10). Occasionally, we also encounter applications that
require we use base 256. Since all these systems are used in C, we need to
have a basic understanding of each to fully understand the language.
All of the numbering systems examined here are positional, meaning that
the position of a symbol in relation to other symbols determines its value.
Each symbol in a number has a position. In integrals and the integral portion
of real numbers, the position starts from 0 and goes to n - 1, where n is the
number of symbols in the integral part. In the fraction part of real numbers
,
the number of symbols
the position starts from -I and goes to -w, where m is
in the fraction part . Each position is assigned a
weight ; the weights vary
according to the numbering system .
1033
1034 Section D. l Computer Numbering System
Decimal Numbers ( Base 10 )

We all readily understand the decimal numbers ( base 10 ) . In fact , we have
used it so much that it is intuitive. All of our terms for indicating countable
quantities are based on it , and , in fact , when we speak of other numbering
systems , we tend to refer to their quantities by their decimal equivalents.
The word decimal is derived from the Latin stem deci , meaning ten . The
decimal system uses 10 symbols to represent quantitative values: 0, 1, 2, 3, 4.
5 , 6 , 7 , 8, and 9.
Decimal numbers use 10 symbols: 0 , 1, 2, 3 , 4, 5, 6, 7 , 3, and 9.
For example, in Figure l )- l , the decimal number 14782.721 has eight

—
digits in positions 3 to 4 .
Decimal Number: 14782.721
1 4 7 8 2 7 2 1 Digits
4 3 2 1 0 -1 -2 -3 Positions
FIGURE D - l The Decimal Number 14782.721
Weights
In the decimal system , each weight is 10 raised to the power of its position .
The weight of the symbol at position - 1 is 10 ! ( 1 / 10 ) while the weight of the
~
symbol at position 0 is 10° ( 1 ) and the weight of the symbol at position 1 is

101 ( 10 ) ; and so on .
Binary Numbers ( Base 2 )

The binary number system ( base 2 ) provides the basis for all computer oper -
ations. The binary system uses two symbols, 0 and 1 . The word binary derives
from the Latin stem bi , meaning two .
Dinary numbers use two symbols: 0 and 1.
Weights
In the binary system , each weight equals 2 raised to the power of its position .
-
The weight of the symbol at position 1 is 2~ l ( 1 /2 ); the weight of the symbol at
position 0 is 2° ( 1 ); the weight ot the symbol at position 1 is 2 * ( 2 ) ; and so on.
Appendix D Numbering Systems
1035
Binory -to-Decimal Conversion
To convert a binary number to
digit by its weight and add all decimal we use the weights. We multiply each
of the weighted results. Figure
we can convert binary D-2 shows how
1001110.101 to its decimal equivalent 78. 625.
Binary Number: 1001110.101

1
6 5
1 oj . [
~
r o 1
4
2rt
2 2 2
3 2
2
1
2 2
0 -1
2
-2
2 23 Weights
64 0 0 8 4 2 0 0.5 0.0 0.125 Weighted Results
+
+
78 0.625
Decimal Number: 78.625
FIGURE D- 2 Binary- to-Decimal Conversiion
Decimal-to- Binary Conversion

I wo simple operations, divide and multiply, give us a convenient way to con
-
vert a decimal number to its binary equivalent as shown in Figure D- 3. To
convert the integral part we divide the number by 2 and write down the
remainder, which must be 0 or 1 . The first remainder becomes the least sig-
nificant binary digit . Now, we divide the quotient of that division by 2 and
write down the new remainder in the second position . We repeat this process
until the quotient becomes zero.
Decimal: 78.625
Divide Multiply
>
0 H H H 1 2 4 9 19 H 39 H 78 M 0- 625 H 0 -25 H 0 50 H 000
F T T T T To T T T l
1 0 0 1 1 1 1 0 1
Binary: 1001110.101
FIGURE D- 3 Decimal- fo -Binary Conversion
by
To convert the fractional part , we need to multiply the fractional part
either a 0 or a 1 ; it becomes the binary
two. The integral in the product is
by two to get the
digit . We then multiply the fractional part the product
of
is zero . For example, to con -
next binary digit and continue until the product
, resulting in 1.25 . We take the inte-
vert 0.625 to binary, we multiply it by 2 . In the next
(0.25 ) to the next step
gral part 1 , and move the fraction part 5
1036 Section D.l Computer Numbering System
step, after we multiply the 0.25 by two, we get 0.5. The integral part is 0,
.
which we keep The fractional part is 0.50 , which we move to the next step.
However, we need to limit the process because the product may never
become zero. If the resulting fraction becomes 0.0 , we stop because more
binary digits does not contribute to the precision of the number. 11 the prod -
uct does not become zero, we need to make a decision as to how many digits
we need at the righthand side of the binary number and stop when we
have them .
Hexadecimal Numbers ( Base 16 )

Another system used in this text is hexadecimal numbers ( base 16 ) . The
word hexadecimal is derived from the Greek word hexadec , meaning 16. The
hexadecimal number system is convenient for formatting a large binary num -
ber in a shorter form . It uses 16 symbols, 0, 1 , ..., 9, A , B , C , I ), E , and F. The
hexadecimal system uses the same first 10 symbols as the decimal system, hut
instead of using 10 , 11 , 12 , 13, 14 , and 15 , it uses A , B , C , I ). E , and F. This
prevents any confusion between two adjacent symbols. Note that the hexa -
decimal symbols A to F can be either upper- or lowercase.
Hexadecimal numbers use 16 symbols: 0, 1» •••» 9, A, 3 , C , 0, E, and F.
Weights
In the hexadecimal system , each weight equals 16 raised to the power ol its
position . The weight of the symbol at position 0 is 16° ( 1 ) ; the weight ol the
symbol at position 1 is 161 ( 16); and so on .
Hexadecimal- to- Decimal Conversion

Io convert a hexadecimal number to decimal, we use the weights. We multi-
ply each digit by its weight and add all of the weighted results. Figure D-4
shows how hexadecimal 3A73.A0C is transformed to its decimal equivalent
14963.628.
Hexadecimal Number: 3A 73 . A0C
3 A 7 3 A 0 C
163 1i 161 °
16 16
1
16
16
2 3
Weights
12 ,288 2 ,560 112 3 0.625 0.0 0.003 Weighted Results
+
14,963 0.628
Decimal Number: 14 ,963.628
FIGURE D - 4 Hexadecimal - to-Decimal Conversion

Appendix 0 Numbering Systems 1037
Decimal- to- Hexadecimal Conversion

We use the same process we used for changing decimal to binary to
transform
a decimal number to hexadecimal. The only difference
is that we divide the
number by 16 instead of 2 to get the integral part and we multiply the num -
her hv 16 to get the fractional part. Figure D-5 shows how 14963.628 in dec -
imal is converted to hexadecimal 3 A 73.AOC. Note that we stop after three
digits in the fractional part.
Decimal: 14963.628
Divide Multiply
4 >
0 3 58 935 — 14963 0.628 0.048 0.768 — 0.288
V T T T .X X 1c
3 A 7 3 A 0
Hexadecimal: 3A 73 . A0C
FIGURE D - 5 Decimal - to- Hexadecimal Conversion
Base 256
Another numbering system that is used in computer science is base 256. We
we need to make a
encounter this base normally in two situations: when
and when we are dealing with Internet
number from individual bytes
addresses.
bytes where each byte
In the first application, we normally have individual
255. We need to consider several
contains an unsigned number between 0 to
the situation with four bytes.
of these bytes as a number. Figure I -6 shows
)
A 32-bit Unsigned Integer
Byte 2 Byte 1 Byte 0

Byte 3
00 0 5_
o
2 1 0
3
Value of integer: 7 X 2563 + 14 x 2562 + 22

X 256 ' + 130 x 256°
FIGURE D - 6 Byte Conversion

;•
Internet addresses in ver-

.
is in Internet addresses decimal notation .
Another application dotted - in
represent an address using a base-256 number,
sion 4 use base 256
to
as, 131.32.7.8, we are
When we define an addressnumbers, a dot is used. For example, the address
Todistinguish between the
1038 Section D.l Computer Numbering System
131.32.7.8 is made of four numbers 8, 7, 32, and 131 at positions 0, 1 , 2 , 3,

respectively, as shown in Figure D- 7.
131 • 32 •7 • 8
3 2 1 0
Value of address as an integer : 131 x 2563 + 32 x 2562 + 7 x 2561 + 8 x 256°
FIGURE D - 7 Internet Dotted - Decimal Notation
Weights
In base 256, each weight equals 256 raised to the power of its position. The
weight of the symbol 1 at position 0 is 256 ( 1 ); the weight of the symbol at
°
position I is 2561 ( 256 ) ; and so on.
Base 256 - to- Decimal Conversion

To convert a base 256 number to its decimal equivalent, we follow the same
process we discussed for converting base 2 or base 16 to decimal; the
weights, however, are 256" where n is the position. Note, however, that in
this case, we are normally dealing with an integral number.
Decimal- to - Base 256 Conversion

Fo convert a decimal number to base 256, we follow the same method we dis-
cussed for converting decimal to base 2 or base 16, but we divide the number
by 256 and keep the remainder.
A Comparison
I able D- l shows how the three systems represent the decimal numbers 0
through 1 5 . As you can see, decimal 1 3 is equivalent to binary 1 101, which is
equivalent to hexadecimal D.
Decimal Binary Hexadecimal

0 0 0
1 1 1
2 10 2
continued
TABLE D- l Comparison of Decimal, Binary, and Hexadecimal Systems
.
1 In base 256, each symbol can be one to three digits.
I
Decimal Binary Hexadecimal

3 11 3
4 100 4
5 101 5
6 110 6
7 111 7
8 1000 8
9 1001 9
10 1010 A
11 1011 B
12 1100 C
13 1101 D
14 1110 E
15 mi F
TABLE D- 1 Comparison of Decimal, Binary, and Hexadecimal Systems continuedI

(
Other Conversions
some easy methods lor common conversions.
. .,
Binary- to- Hexadecimal Conversion
To change non be ta
from the right by fours. Then
we c
Sk
. we convert binary
^ D8
mal equivalent using Table ^ )
1010001 110 to hexadecimal.
TOTTOQ oJI 11 QJ Binafy

2 8 E
Hexadecimal
FIGURE D- 8 Rinnrv-to-Hexadecimal
Conversion
5
1040 Section D . 2 Storing Integers
Hexadecimal-to- Binary Conversion

To change a hexadecimal number to binary, we convert each hexadecimal
digit to its equivalent binary number using t able I)- 1 and concatenating the
results. In Figure l)- 9 we convert hexadecimal 28 E to binary.
Hexadecimal
2 8 E
10 1000 1 1 1 0
Binary
FIGURE D- 9 Hexadecimal - to - Binary Conversion
Base 256 - to - Binary Conversion

to convert a base 256 number to binary, we first need to convert the num-
ber in each position to an 8 - bit binary group and then concatenate the
groups.
Binory-to - Base 256 Conversion

to convert from binary to base 256, we need to divide the binary number into
groups of 8 bits, convert each group to decimal, and then insert separators
( dots ) between the decimal numbers.
D . 2 Storing Integers
We have discussed bow integers and real numbers are presented in different
bases. Although base 16 and base 256 are used in computer science, data are
stored in the computer in binary. Numbers must be changed to base 2 to be
stored in the computer. All of our discussions so far have ignored the sign of
the number. In computer science, we use both positive and negative num-
bers. We need some way to store the sign. In this section, we concentrate on
bow integers are stored in a computer; in the next section, we show bow real
numbers are stored.
Unsigned Integers
Storing unsigned integers is a straightforward process. The number is
changed to the corresponding binary form, and the binary representation is
stored, lor example, an unsigned integer can be stored as a number from 0 to
1 5 in a 4 - bit integer, as shown in Figure D- 10.
fa
o
( oooo )
7 8
( oin ) ( 10OP )
15
QTTT)
in
FIGURE D- 10 Unsigned Integers Format
Addition of unsigned integers is very straightforward as long as there is

no overflow ( see the section on overflow ). Subtraction of unsigned integers
must he done with caution . If the result is negative, the number is not an
unsigned number anymore. Normally computers promote the result to a
signed integer in this case.
Signed Integers
Storing signed integers is different from storing unsigned integers because
we must consider both positive and negative numbers. Four methods
are
designed to store signed integers in a computer: sign and magnitude , one’s
com plement, two s complement , and Excess system .
Sign and Magnitude

range between the positive
In sign and magnitude , we divide the available
range occupies the positive
and negative numbers. The lower part of the
numbers; the upper part occupies the negative
numbers . To do so, we con -
of the number and the rest of the
sider the leftmost bit to represent the sign
of the number. Let us see how a
bit to be the absolute value, the magnitude
and negative numbers. Figure D- l 1
4 bit integer can both hold the positive
- itive and negative numbers.
shows the partition of the range between
positive
..L - Complement
Complement
+7 -0 -7
+0
0111 1000
foooo )
FIGURE D- 1 1 Signed Integer Format
Properties magnitude properties:

i /.e the sign and
Let us summarize ; 0 for positive, 1 for
contains the sign of the number
1. The leftmost bit
negative.
thod : +0 (0000 ) and -
0 ( 1000 ) .
zeros in this me
2. There are two
1042 Section 0.2 Storing Integers
.
3 To change the sign of a number, we need to flip only the first hit We can -.
not use the complement operator (~) we learned in Chapter 14 because
this operator will flip all hits, not just the leftmost hit .
.
4 Addition and subtraction are very inefficient operations. We need to do
the following:
a. Separate the sign from the magnitude.
b. Compare two magnitudes to see if we need to add or subtract the
magnitude.
c . We have eight cases in total, four for addition and four for subtraction.
d. Insert the sign of the result after the absolute value of the result is
determined.
Applications
Properties 2, 3, and 4 make this method unsuitable for a general-purpose
computer in which we need to perform mathematical operations. The sign
and magnitude method, however, is used in computer science when we do
not need mathematical operations. One of these areas is storing analog and
digital signals.
One's Complement
One’s complement is similar to sign and magnitude; however, the partition
of the range between the positive and negative numbers is different. The
numbers are arranged symmetrically. A positive and a negative number are
.
symmetric with respect to the middle of the range Figure I) - 12 shows how
the positive and negative numbers are distributed.
Complements
+0 +3 +7 -7 -3 -0
/ \
( 0000 ) ( 0011 ) ( 0111 ) ( 1000 ) ( 1100 ) CmD
FIGURE D - l 2 One's Complement Format
Properties
Let us summarize the one’s complement properties:
1 . The leftmost hit contains the sign; 0 for positive, I for negative.
2. There are two zeros in this method: +0 ( ()()()()) and -0 ( 1 1 1 1 ).
3 . To flip the sign of a number, we flip each individual hit. We use the com-
plement operator (~) we learned in Chapter 14 to find -A from A as
shown in Figure D- l 3.
FIGURE D- 1 3 Flipping Signs in One's
-(-3)
-3
Complement
- (0 0 1 1 )
- (1 1 0 0)
>ndix D
1100
0011
Numbefingjysfems 1043
11
The figure also shows that if we complement
complement, we also can find the complement-A, we get back A. In ones
For example, -3 is stored as 15 -3 or 12, of A as -A = ( 2” -1 ) - A.
when n is 4 ( number of hits in
the integer ).
4 . If we add A + (-A) we get 0 (all
- hits are Is) . We saw this property in
Chapter 14 when we used checksum.
Addition and subtraction are very simple operations. To add (A )
+ B , we
just add the numbers hit by hit . To subtract A
- B, we just add A and the
complement of B. In other words, A - B = A + (- B ) . The only thing
we
need to consider about adding or subtracting is to add the carry
produced
at the last column to the result . Figure D- 14 shows how
numbers are
complemented, added , and subtracted.
Add
1 1111
( + 3) 0011 ( +3) 0011
+ (+2) +0010 + (-2) + 1101
( +5) 0101 (+ D 0000
> 1 -
0001
Subtract (add first with the complement of the second)

1 1 1
( + 3) 0011 (-3) 110 0
- ( +4) + 10 11 - (-2) + 0010
(- 1) 1110 HD 1110
FIGURE D- 14 Add and Subfracf One's Complement
Applications
one’s complement. The
Early general - purpose computers, such as VAX, used for arithmetic cal-
does not use this method
modern general-purpose computer
of two and the need lor
0 s)
culation because of the second property (existence , ones complement
. However
keeping track of the carry from the last column
1044 Section D.2 Storing Integers
.
arithmetic has its own place in computer science We saw the use of it in the
checksum calculation in Chapter 14. We simulated the use of ones comple-
ment with unsigned numbers. To simulate adding the carry from the last col-
umn to the partial result , we used modulo 2n — 1 in our calculations.
Two 's Complement

Like one’s complement, two’s complement also shares the range between
positive and negative numbers. However, the partition of the range between
the positive and negative numbers is different . The symmetry is an offset. For
example, the fourth number from the beginning of the range is the comple-
ment of the third number from the end. This is done for two purposes: to
eliminate the negative 0 and to avoid adding the carry from the last column to
the partial result. Figure D- 15 shows how the positive and negative numbers
are distributed.
Complements
0 +3 +7 -8 -3 -1
/ \
( 0000 ) ( 0011 ) ( 0111 ) ( 1000 ) ( 1101 ) ( 1111 )
FIGURE D- 15 Two's Complement Format
Note that the 0 in the positive range, which means that the negative
.
range can accommodate one more number In the figure, the first half of the
range accommodates 0 to 7; the second half accommodates - 1 to -8.
Properties
Let us summarize the two’s complement properties:
.
1 The leftmost bit contains the sign of the number; 0 for positive, I for
negative.
.
2 I here is only one zero in this method: 0 ( ()( ) ( )() ).
3. fo change the sign of a number, we need two operations. First we need to
flip each individual bit, then we need to add 1 to the previous result as
shown in Figure D- 16.
-3 ~ (0 0 1 1) + 1 1 1 0 1
-(-3) ~ (1 1 0 0) + 1 0 0 1 1
FIGURE D - 16 Complementing Two's Complement

104 5
The figure also shows at
^ ..
if we
complement, we also can .. comp!ement -A, we get hack A. In two’s
find
where ft is the number of hits the complement of A as -A = (2") - A,
in the integer. For
16 —3 or 13. example, -3 is stored as
4 . If we add A + (-A) we
get 0.
3. Addition and subtraction are
simpler. To
numbers hit hv hit. To subtract (A B) add (A + B), we just add the
ment ol B. We do not
- , we just add A
have to worry about the earn fromwith the comple-
we drop it. Figure D- 17 the last column;
shows this feature.
Subtract (add first with the complement of the second)

1 1 11
(+3 ) 0011 (-3) 1101
- ( +4) +1100 - (-2) + 0010
(-1) 1111 H) 1111
FIGURE D- 1 7 Adding and Subtracting Two's Complement
Applications
One single zero and the simplicity of adding and subtracting have made two’s
complement arithmetic the best candidate for modern computers. C uses the
hardware instruction set of the computers it runs on, which almost univer-
sally uses two ’s complement arithmetic.
Modern computers use two’s complement to store signed integers .
Storing and Retrieving Two's Complement

As we said before, computers today use twos complement numbers lor han-
Iunc -
dling mathematical problems. However, we may wonder how an input
complement ,
tion, such as scanf converts a signed decimal integer to twos
}
number
and an output function, such as printf , converts a twos complement
shows high- level algorithms that
to a signed decimal integer. Figure 18
D- two
can be used by these functions.

^
( Store Integer
Convert the absolute

j (Retrieve Integer
^
n: size of value to n-bit binary 1 leftmost 0
integer in bits . bit ? .
negative lositive Sign is negative
sign?
Complement Complement
the number Sign is positive
the number
Increment the Increment the

number by 1 number by 1
*
Change the n-bit
Store the number binary to decimal
and add sign
T T
Return Return
j /
FIGURE D- l 8 Storing and Retrieving Two's Complement
EXAMPLE D l Let s follow the storing algorithm to see how +76 is stored in a 16 - hit integer.
-
I he absolute value of the number ( 76 ) is changed to a 16 -bit binary number.

The sign is positive, so the number is stored in the memory.
Decimal number: +7 6
Convert absolute value: 0000000001001100
Store value: 0000000001001100
EXAMPLE D- 2 Let ’s follow the storing algorithm to see how -76 is stored in a 16 -bit integer.
I he absolute value ol the number ( 76 ) is changed to a 16 - bit binary number.
I he sign is negative, so we need to complement the number and add 1 to it as
shown below.
Decimal number: -7 6
Convert absolute value: 0000000001001100
Complement: 1111111110110011
Add 1: 1111111110110100
Store value: 1111111110110100
*
1047
EXAMPLE D- 3 Let ’s Follow the iretrieving
algorithm to see how a
ger is retrieved. The process is stored value in a 16- bit inte-
j
shown below.
Retrieved value:
Complement: 1111111111101011 ( sign 1
Add 1: 0000000000010100
Convert to decimal: 0000000000010101
21
Add sign: -21
The retrieved binary value is negative, so the

the end . I he number is complemented and 1 sign is stored to be added at
is added . The result is con -
verted to decimal and the sign is added .
Excess
I here are applications that require more
numeric comparisons than arith -
metic operations. In these cases we can use a simple strategy
named
by the IEEE, to store positive and negative integers. We just add a fixedExcess
value
(called the bias value ) to the negative number
to make them non - negative
(zero or positive) when we store the numbers. We
subtract the same bias
value when we retrieve the number. For example, in our hypothetical 4- bit
-
integer, we can store numbers From 7 to 8 as shown in Figure D- 19. In other
words, we store both positive and negative numbers in an unsigned Format.
-7 0 1 8
( oooo ) ( 0111 ) ( 1000 )

FIGURE D- l 9 Excess Format
When two numbers are stored using the Excess method , we can easily
compare them on the value; we do not have to worry about the sign of the
numbers. When adding two numbers in the Excess system , the bias value is
added twice ; therefore, the bias value must he subtracted when we store the
the
result. When subtracting two numbers, the bias value is cancelled during
be added hack when we store the result .
subtraction therefore
; , it must
Overflow
. Overflow can be the
Wc need to discuss a very important issue, overflow
source of much confusion for a programmer
. Sometimes when we print a
number, we get a surprising result . Often the
reason is that an overflow has
are stored on a limited size word
occurred . Overflow occurs because integers integer is only 4 bits, we can
of the
in the computer. For example, if the size
either store an unsigned integer between 0 and 15 or a signed integer

between -8 and + 7 ( using two’s complement ). Any number beyond this range
overflows the possible values. Most of the time , the system does not issue an
error or warning, it just drops the extra bit or hits that do not lit in the allo-
cated space. This creates an invalid result , which may be positive or negative.
Then when we print the results , we get a surprising number.
Overflow is better understood if we show the range of the integers that
can he stored in a number in a circle. Figure D-20 shows the range lor two
methods that are used in todays computers, unsigned integer and two’s
complement .
15 0 -1 0
14 1 -2 +1
13 2 -3 +2
12 3 -4 +3
Unsigned Two's
integer complement
11 4 -5 +4
10 5 -6 +5
9 6 -7 +6
8 7 -8 +7
FIGURE D - 20 Range of Integer Values
I he circle for the unsigned integer shows that 15 + 1 is 0. The circle lor
two’s complement shows that 7 + 1 is -8. If we increment an unsigned value
holding the maximum possible number, we get 0. If we increment an integer
value holding the maximum possible value, we get the minimum possible
value. We cannot test this concept with 4 - bit integers , but we can prove it
with the actual size of the integer in any computer as shown in Program D- l .
PROGRAM D-l Demonstrate Overflow

1 /* Demonstrate circular nature of unsigned and two ’s
2 complement integer numbers.
3 Written by:
4 Date:
5 */
7 # include <limits.h >
8
9 int main ( void )
10 {
continued
PROGRAM D - ] Demonstrate
Overflow ( continued )
11 // Local
12
Definitions
unsigned short x
13 = USHRT MAX;
short y = SHRT__MAX ;^
14
15 // Statements
16 printf( " Maximum
17 unsigned short value:
X+ +; % u \n",
18 printf( "Maximum
19
unsigned short value +
1: % u \n ",
21
short value:
y++; %d \n\
short value + l j
23 %d \n ",
24 return 0;
2 5 } // main
Results:
Maximum unsigned value: 65535
Maximum unsigned value + 1 : 0
Maximum short value: 32767
Maximum short value + 1 : - 32768
D. 3 Storing Real Numbers

We arc familiar with scientific notation to represent a real number. In this
-
notation, we can show a real number such as -314.625 as 3.14625 x 10+2.
We also know that a real number is made of three pieces of information: the
s * Mn (-), the precision (3.14625), and the power often ( + 2 ). Note that
we do
not have to show the multiplication operator or the base of the power ( 10)
because they are understood by the rules of scientific notation.
Computers use the scientific notation concepts to store a real number. II
we write all information in binary system , 314.625 is represented as shown
below with the power also in binary ( 4 ).
-10011.10101010 x 2 100
Normalization
We have one problem to solve before we can store this number in the com -
puter: the position of the binary point ( the point that separates the integral
part from the fractional part ). We can store only binary digits, not a point .
I he solution is normalization . We normalize the number so that the point is i
always at a fixed position. Tradition and the standard state that the number
should have only one binary digit to the left of the point . For non -zero values,
1050 Section D.3 Storing Real Numbers
the digit is a 1 . To normalize the number, therefore, we shift the point to the
left or to the right based on its original position . Shifting binary numbers
requires multiplying or dividing the number by two lor each shift. In other
words, if we move the point to the left , we need to add the number of digits
shifted to power ; if we move the point to the right , we need to subtract the
number of digits shifted from the power. The normalized version of the previ -
ous number is shown below. Note that we moved the point four positions to
the right , and we added 4 to the power, making it 8 ( the power is represented
in binary ) .
-1.001110101010 x 21000
After normalization , we are left with only the sign , the precision , and the
power to store; we do not have to store the integral part ol the precision or
the binary point .
Fortunately, C99 provides us with help to find how real numbers are
stored in the computer. We can use the % A or %a conversion code to print
the real number in a format that represents these values. Program D - 2 dis -
plays a positive and a negative real number. Note that the values are the same
except for the sign .
PROGRAM D - 2 Demonstrate Storage of Real Numbers

1 /* Demonstrate the storage of real numbers.
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
10 float x = -314.625;
11 float y = +314.625;
12
13 // Statements
14
15
-
printf ( " 314.625: % A \ n " , x ) ;
printf ( "+314.625: %A\n" , y ) ;
16 return 0 ;
17 } // main
Results:
-314.625: -0X 1.3AA0000000000P+8
+314.625: 0X 1.3AA0000000000P+8
The results show the sign , the mantissa ( 3AA0000000000 in hexadeci

mal ) , and the power ( 8 ) .
-
1051
Sign, Exponent, and Mantissa
Real numbers contain three parts,
nent, and mantissa ( m ) . which are referred
We can say that the original to as the sign (s) expo-
number ( N ) is
N = ( - l ) x l .m x 2e
Sign
The Sign of the number is
stored using one bit (0 for plus and I for minus).
Exponent
The exponent ( power of 2 )
defines the power. Note that the power can he
negative or positive. Excess is the
method used to store the exponent .
Mantissa
I he mantissa is the binary number to
the right of the binary point. It defines
the precision of the number. The mantissa is
stored as an unsigned integer.
IEEE Standards
I he Institute of Electrical and Electronics Engineers (
IEEE) has defined two
standards to store numbers in memory (single precision and double preci-
sion ) . These formats are shown in Figure D-21 . Note that the number inside
the boxes is the number of bits for each field .
Excess 127
23
Sign Exponent Mantissa
Single Precision
Excess 1023
II r
1 11 52
Sign Exponent Mantissa
Double Precision
FIGURE D- 21 IEEE Standards for Floating- Point Representation
Storing and Retrieving Algorithm

. These
Let ’s look at the algorithms for storing and retrieving real numbers
insight how the input and
algorithms, shown in Figure D-22, give us an into
.
or retrieve real numbers
output functions such as scan/ and print/ store
ion D.3 Storing Real Numbers
Store Real )
^
( Retrieve Real J
negative positive 1 0
sign? leftmost
bit ?
s 1 s 0 S S
* *
Convert absolute Convert power to

value to binary decimal
Normalize and Calc Add 1.0 to mantissa

mantissa (m) & power and
denormalize
Convert the power Convert previous

to Excess (e) result
to decimal (D)
Concatenate Concatenate
s e m S D
and store and present
T
Return
I
Return
V
FIGURE D- 22 IEEE Algorithms
EXAMPLE D - 4 We show how the real number 123.8125 is stored in the computer using the
storing algorithm in Figure D - 22 and IEEE for single precision.
1 . The sign is positive; s = 0.
2. T he absolute value in binary is 1111011.1101
3. We normalize it to 1.1110111 101. I he value ol in in IEEE single preci -
sion ( 23 hits ) is m = 1 I 101 1 I 1010000000000000 and the power is 6.
4 . The value of e in IEEE single precision is 6 + 127 or 10000101.
5. When we concatenate s, e, and m , we have the value shown below:
0 10000101 11101111010000000000000
EXAMPLE D-5 We show how to find the decimal value of the following 32- hit real number
stored using IEEE single precision.
1 10000010 00011000000000000000000
We use the retrieve algorithm in Figure D-22.
1 . The leftmost bit is 1 ; S =
2. The value of next 8 bits is 130. We subtract 127 from it to

.3. We add 1 and the binary point to m and shift
get 3.
the binary point 3 digits to
the right to get 1000.11 ( ignoring the trailing zeros ).
T
4. We convert the above number to decimal to get I) = 8.75 .
5. We concatenate S and D to get -8.75.
\
\
Integer and Float Ubraries

rhis appendix documents two of the more important C libraries. Note that the
values shown here are representative only and change from hardware to hard -
ware. Libraries contain unformatted numbers ( no commas ) , often expressed
in hexadecimal . We use decimally formatted numbers for readability.
E.l limits.h
Table E - l contains hardware- specific values for the integer types.
Identifier Meaning Minimum Value

8
CHAR BIT bits in a char
-127
SCHAR MIN short char minimum
127
SCHAR MAX short char maximum
255
UCHAR MAX unsigned char maximum
See SCHAR.MIN
CHAR MIN char minimum
See SCHAR.MAX
CHAR MAX char maximum
-32,767
SHRT_MIN short int minimum
32,767
SHRT.MAX short int maximum
65,535
USHRT_MAX unsigned short maximum
-32,767
INT_MIN int minimum
32,767
INT MAX int maximum continued
Library
TABLE E- 1 Partial Contents of Limits
1056 Section E.2 float.h
UINT.MAX unsigned int maximum 65,535

LONG MIN long minimum -2, 147,483 ,647
LONG MAX long maximum 2,147,483,647

ULONG.MAX unsigned long maximum 4,294,967,295
LLONG MIN long long int minimum -(263 - 1 )
LLONG MAX long long int maximum 263 - 1
ULLONG_MAX unsigned long long maximum 264 - 1
TABLE E - l Partial Contents of Limits Library ( continued )
E. 2 f l o a t . h
-
Table E - 2 contains hardware-specific values for the floating point types.

FLT DIG digits of precision 6
DBL DIG 10
LDBL DIG 10
DECIMAL DIG decimal digits needed to repre- 10
sent floating-point value
FLT_MANT _DIG size of mantissa none
DBL MANT DIG none
LDBL MANT DIG none
FLT MIN EXP smallest integer for negative none
exponent ( float radix)
DBL_MIN EXP none
LDBL_MIN EXP none
FLT_MIN_ 10 _ EXP smallest integer for negative -37
exponent (base 10)
DBL_MIN 10 EXP -37
LDBL_MIN 10 EXP -37
coni i lined
TABLE E - 2 Partial Contents of Limits Library
Appendix E Into and float I ihtnrins 1057
fl
FLT MAX EXP largest integer for positive expo- 37
nent ( float radix)
DBL_MAX EXP 37
LDBL MAX EXP 37
FLT MAX 10 EXP largest integer for positive expo- 37
nent (base 10)
DBL MAX 10 EXP 37
LDBL _MAX _ 10_EXP 37
FLT MAX largest possible floating-point 1037
number
DBL MAX 1037
LDBL MAX 1037
FLT MIN smallest possible floating-point io-37
number
DBL MIN io-37
LDBL MIN io-37
TABLE E - 2 Partial Contents of Limits Library Icontinuedl

i
. *
Function libraries
In this appendix we list most of the standard functions found in
the C language.
We have grouped them by library so that related functions are grouped
together. We have also listed them alphabetically in Section F.l for your con-
venience. Note that not all functions are covered in the text and there are
functions in the libraries that are not covered in this appendix.
F.l Function Index

Function Page Library Function Page Library Function Page Library
Exit 1068 sfdlib atanl 1063 math ceill 1063 math
abort 1068 sfdlib atan 2 1063 math dearerr 1066 stdio
abs 1067 sfdlib atan2f 1063 math dock 1069 time
acos 1063 math atan2l 1063 math cos 1063 math
aosf 1063 math atexit 1068 sfdlib cosf 1063 math
atof 1068 sfdlib cosl 1063 math

acosl 1063 math
atoi 1068 sfdlib cosh 1063 math
asctime 1069 time
atol 1068 sfdlib coshf 1063 math
asm 1063 math
1068 sfdlib coshI 1063 math
asinf 1063 math atoll
1068 sfdlib dime 1069 time
asinl 1063 math calloc
1063 math difftime 1069 time
atan 1063 math ceil
div 1067 sfdlib
atanf 1063 math ceilf 1063 math
continues
105
1 Function Index

atanl 1063 math isupper 1063 ctype scaleblnf 1065 math
atan2 1063 math isxdigit 1063 ctype scaleblnl 1065 math

atan 2 f 1063 math iswalnum 1063 wctype _ t scanf 1066 stdio
atan 2l 1063 math iswalpha 1063 wctypej sin 1065 math
atexit 1068 stdlib iswcntrl 1063 wctypej sinf 1065 math
atof 1068 stdlib iswdigit 1063 wctypej sinl 1065 math

atoi 1068 stdlib iswgraph 1063 wctypej sinh 1065 math
atol 1068 stdlib iswlower 1063 wctypej sinhf 1065 math
atoll 1068 stdlib iswprint 1063 wctypej sinhl 1065 math
calloc 1068 stdlib iswpunct 1063 wctypej snprintf 1066 stdio
ceil 1063 math iswspace 1063 wctypej sprintf 1066 stdio
ceilf 1063 math iswupper 1063 wctypej swprintf 1066 stdio
ceill 1063 math iswxdigit 1063 wctypej sqrt 1065 math
clearerr 1066 stdio labs 1067 stdlib sqrtf 1065 math
clock 1069 time llabs 1067 stdlib sqrtl 1065 math
cos 1063 math Idexp 1064 math srand 1067 stdlib
cosf 1063 math Idexpf 1064 math sscanf 1066 stdio
cosl 1063 math Idexpl 1064 math swscanf 1066 stdio
cosh 1063 math Idiv 1067 stdlib strcat 1069 string
coshf 1063 math lldiv 1067 stdlib strchr 1069 string
coshl 1063 math lint 1064 math strcmp 1069 string
ctime 1069 time lintf 1064 math strcpy 1069 string
difftime 1069 time lintl 1064 math strcspn 1069 string
div 1067 stdlib Irint 1064 math strftime 1069 time
exit 1068 stdlib Irintf 1064 math strlen 1069 string
Exit 1068 stdlib Irintl 1064 math strncat 1069 string
exp 1063 math llrint 1064 math strncmp 1069 string
expf 1064 math llrintf 1064 math strncpy 1069 string
expl 1064 math llrintl 1064 math strpbrk 1069 string
continued
Appendix F Funcfion libraries 1061

expml 1064 math localtime 1069 time strrchr 1069 string
expm 1 f 1064 math log 1064 math strspn 1069 string
expm 11 1064 math logf 1064 math strstr 1069 string
exp2 1064 math logl 1064 math strtod 1068 stdlib
exp2f 1064 math log2 1064 math strtof 1068 stdlib
exp 2l 1064 math log2f 1064 math strtok 1069 string
fabs 1064 math log2l 1064 math strtol 1068 stdlib
fabsf 1064 math log10 1064 math strtold 1068 stdlib
fabsl 1064 math logl Of 1064 math strtoll 1068 stdlib
fclose 1066 stdio logl 01 1064 math strtoul 1068 stdlib
feof 1066 stdio Iround 1064 math system 1068 stdlib
ferror 1066 stdio Iroundf 1064 math tan 1065 math
fgetc 1066 stdio Iroundl 1064 math tanf 1065 math
fgetwc 1066 stdio llround 1064 math tanl 1065 math
fgets 1067 stdio llroundf 1064 math tanh 1065 math
fgetws 1067 stdio llroundl 1064 math tanhf 1065 math
malloc 1068 stdlib tanhl 1065 math

floor 1064 math
memchr 1068 string time 1069 time
floorf 1064 math
memcmp 1068 string tmpfile 1067 stdio
floorl 1064 math
memcpy 1068 string tmpnam 1067 stdio
fmod 1064 math
1068 string trunc 1065 math
fmodf 1064 math memmove
truncf 1065 math
fmodl 1064 math mktime 1069 time
trund 1065 math
modf 1064 math
fopen 1066 stdio
tolower 1063 ctype
modff 1065 math
freopen 1066 stdio
toupper 1063 ctype
modfl 1065 math
fprintf 1066 stdio
towlower 1063 ctype
nearbyint 1065 math
fputc 1066 stdio
towupper 1063 ctype
nearbyintf 1065 math
fputwc 1066 stdio
ungetc 1066 stdio
nearbyintl 1065 math
fputs 1067 stdio
ungetwc 1066 stdio
pow 1065 math
fputws 1067 stdio continued
1062 Section F.2 Type Library

fread 1067 stdio powf 1065 math wcscat 1069 wchar
free 1068 stdlib powl 1065 math wcschr 1069 wchar
frexp 1064 math printf 1066 stdio wcsrchr 1069 wchar
frexpf 1064 math putc 1066 stdio wcscmp 1069 wchar
frexpl 1064 math putwc 1066 stdio wcscpy 1069 wchar
fscanf 1066 stdio putchar 1066 stdio wcscspn 1069 wchar
fwscanf 1066 stdio putwchar 1066 stdio wcslen 1069 string
fseek 1067 stdio puts 1067 stdio wcsncat 1069 wchar
ftell 1067 stdio rand 1067 stdlib wcsncmp 1069 wchar

fwprintf 1066 stdio realloc 1068 stdlib wcsncpy 1069 wchar
fwrite 1067 stdio remainder 1065 math wcspbrk 1069 wchar
getc 1066 stdio remainderf 1065 math wcsspn 1069 wchar

getchar 1066 stdio remainderl 1065 math wcsstr 1069 wchar
getwc 1066 stdio remquo 1065 math wcstok 1069 wchar

getwchar 1066 stdio remquof 1065 math wprintf 1066 stdio
gets 1067 stdio remquol 1065 math wscanf 1066 stdio
gmtime 1069 time remove 1067 stdio
ilogb 1064 math
F. 2 Type Library
I he following functions are found in ctype . lt .
isalnum int isalnum (int a_char );

isalpha int isalpha (int a_char);
isascii int isascii (int a_char ). 1
iscntrl int iscntrl (int a_char );
isdigit int isdigit (int a_char );
isgraph int isgraph (int a _char );
islower int islower _
(int a char );
isprint int isprint (int a_char );
1 . Not standard C. Traditional extension included in most implementations.

n
ispunct int ispunct _
( int a char ) ;
isspace int isspace _
( int a char ) ;
isnpper int isupper _
( int a char ) ;
isxdigit int isxdigit _
( int a char ) ;
to lower int tolower _
( int a char ) ;
tou pper int toupper _
( int a char ) ;
I he following functions are found in wctype . h .

iswalnum int iswalnum _t a_char)
_t a_char )
( wint
iswalpha
iswcntrl
int
int
iswalpha
iswcntrl
( wint
( wint _
_t a char)
_t a_char)
isw digit
_t a _char )
int iswdigit ( wint
iswgraph
iswlower
int
int
iswgraph
iswlower
( wint
( wint _
_t a char)
_t a _char)
_t a_char )
iswprint int iswprint ( wint
iswpunct
_t a _char )
int iswpunct ( wint
_t a _char)
iswspace int iswspace ( wint
_t a _char)
iswupper int iswupper ( wint
iswxdigit iswxdigit
_t a _char)
int ( wint
_t a _char)
towlower int towlower ( wint
towupper int towupper ( wint
F.3 Math Library

The following functions are found in math.h.
acos double acos (double number);
float acosf ( float number) ;
acosf
long double acosl (long double number ) ;
acosl
asin ( double number );
asin double
float asinf ( float number );
asinf
long double asinl ( long double number );
asinl
double atan (double number);
atan
atanf ( float number);
atanf float )
( long double number ;
long double atanl
numberI . double number2 ;
atanl )
atan 2 (double
atan 2 double , float Humbert );
atan 2f (float numberl
atan 2 f float
( long double number 1
, long double number2 ); i
atan 2 l long double atan 2 l
ceil (double number );
ceil double
ceilf ( float number);
ceilf float );
ceill ( long double number
ceil I long double
cos (double number);
cos double ( float number) ;
float cosl );
cosf
cosl ( long double number
long double
cosl ( double number );
double cosh
cosh ( float number);
float coshf );
coshf
cosbl ( long double number
cosh! long double (double number );
double exp
exp
1064 Section F.3 Math Library
expf float expf ( float number) ;

expl long double expl ( long double number);
expm 1 double expm 1 ( double number );
expm 11 float expm 11 ( float number);
expm 11 long double expm 11 ( long double number) ;
exp 2 double exp 2 ( double number) ;
exp 2f float exp 2 f ( float number ) ;
exp 2 l long double exp21 ( long double number);
fabs double fabs ( double number ) ;
fabsf float fabsf ( float number);
fabsl long double fabsl ( long double number ) ;
floor double floor ( double number);
floorf float floorf ( float number );
floorl long double floorl ( long double number ) ;
fmod double fmod ( double number 1 , double number2 );
fmodf float fmodf ( float number 1 , float number 2 ) ;
fmod I long double fmodl ( long double numberl , long double number2 );
frexp double frexp ( double number, int * exponent ) ;
frexpf float frexpf ( float number, int * exponent );
frexp 1 long double frexp1 ( long double number, int * exponent ) ;
ilogb double ilogb ( double number ) ;
ilogbf float ilogbf ( float number );
ilogbl long double ilogbl ( long double number) ;
Idexp double Idexp ( double number, int power ) ;
ldexpf float ldexpf ( float number, int power ) ;
Idexpl long double Idexpl ( long double number, int power );
lint double lint ( double number ) ;
lintf float lintf ( float number ) ;
I inti long double lintl ( long double number ) ;
lrint long lrint ( double number ) ;
Irintf long Irintf ( float number) ;
Irintl long Irintl ( long double number ) ;
llrint long long llrint ( double number ) ;
I Irintf long long llrintf ( float number );
llrintl long long llrintl ( long double number);
log double log ( double number);
logf float logf ( float number ) ;
logl long double logl ( long double number ) ;
log2 double log2 ( double number ) ;
log2 f float log2 f ( float number ) ;
log2 l long double log2 l ( long double number ) ;
log 10 double log 10 ( double number ) ;
loglOf float loglOf ( float number ) ;
log 101 long double log 101 ( long double number );
1 round long I round ( double number ) ;
lroundf long lroundf ( float number ) ;
lroundl long lroundl ( long double number ) ;
llround long long llround ( double number ) ;
I lroundf long long I lroundf ( float number ) ;
llroundl long long llroundl ( long double number ) ;
'
modi double modi’ ( double number, double * integral) ;
modff float modff ( float number, float 4 integral );

mod 11 long double mod 11 ( long double number, long double 4 integral );
nearbyint double nearbyint ( double number );
nearbyintf float nearbyint ( float number );
nearbyintl long double nearbyintl ( long double number ) ;
pow double pow (double base, double power );
povvf float powf ( float base, float power );
powl long double powl ( long double base, long double power);
remainder double remainder ( double numberl , double number2 );
remainderf float remainderf ( float numberl . float number2);
remainderl long double remainder! ( long double numberl , long double number2 );
remquo double remquo ( double numberl . double quotient );
remquof float remquof ( float numberl , float quotient );
remquol long double remquol ( long double numberl , long double quotient ) ;
rint double rint ( double number ) ;
rintf float rintf (float number );
rintl long double rintl ( long double number ) ;
round double round ( double number);
roundf float roundf ( float number );
round! long double roundl (long double number);
scalebn double scalebn ( double number, int factor) ;
scalebnf float scalebnf ( float number, int factor );
scalebnl long double scalebnl (long double number, int factor);
scalebln double scalebln ( double number, long factor );
scalcblnf float scalebn If ( float number, long factor ) ;
scaleblnl long double scalebnll ( long double number, long factor);
sin double sin ( double number );
si nf float sinf ( float number) ;
sinl long double sinl ( long double number);
si nil double sinh ( double number) ;
sinhf float sinhf ( float number);

long double sinhl ( long double number );
sinhl
double sqrt ( double number) ;
sqrt
float sqrtf ( float number);
sqrtf
long double sqrt I ( long double number );
sqrt I
double tan (double number );
tan
float tanf ( float number);
tanf
long double tanl ( long double number );
tan I
tanh ( double number);
tanh double
tanhl ( float number);
tanhf float
long double tanh I ( long double number );
tanhl ( double number);
trunc double trunc
( float number );
truncf float truncf );
truncl ( long double number
trunc I long double
F.4 Standard I/O Library ’ ( . h ) by the type of

input/output library stdio
We have divided the system
data being read or written .
Section F. 4 Standard I/O Library
General I /O
General input/output contains functions that apply to all files.
clearerr void clearerr ( FILE * sp);
fclose int fclose ( FILE * sp );
feof int feof ( FILE * sp );
terror
terror ini ( FILE * sp);
_
(const char * extn name , const char * filc mode ) ;
__
fopen
freopen
FILE *
FILE *
fopen
frcopen _
( const char * extn name, const char * file mode, FILE * stream );
Formatted I /O
Convert text data to/from internal memory formats.
fprintf int fprintf ( FILE* fileOut , const char * format _string, ... ) ;
_
fscanf
printf
int
int
fscanf
printf ( const char * format string, ... ) ; _
( FILE * fileln , const char * format string, ...) ;
scant int scanf ( const char* format _string, ...);

_ _
snprintf int snprintf (char * to loc, sizc t n , const char* format string, ... ); _
_
sprintf
sscanf
int sprintf
int sscanf
( char * to_ Ioc , const char * format string, ... ) ;
_
( const char * from loc , const char * format string, ...) ; _
I he following wide-character functions also require the wchar.h library tile.
_
fwprintf int fwprintf ( FILE* fileOut , const wchar t * format _string, ... ) ;
fwscanf int fuscanf ( FILE * fileln, const vvchar _ t * format string, ... ); _
_
svvprintf int swprintf ( vvchar_ t * to_ loc , const wchar t * format string, ... ) ; _
___ _
swscanf int swscanf ( const wchar t * from Ioc, const char * format string, ...);
wprintf int wprintf ( const wchar t * format _string, ... ) ;
_
_
vvscanf int wscanf ( const wchar t * format string, ... );
Character I /O
Read and write one character at a time .
fgetc
fputc
int
int
fgetc
fputc
( FILE * sp ) ;
_
( int char out, FILE * sp ) ;
getc int getc ( FILE * sp ) ;
getchar int getchar (void ) ;
putc int putc _
( int char out , FILE * sp ) ;
putchar int putchar _
( int char out ) ;
ungetc int ungetc _
( int char out , FILE * sp ) ;
I he following wide-character functions also require the wchar .h library file .

fgetwc wint t _ Igetwc ( FILE * sp ) ;
wint _ t _ _out , FILE * sp );
fputwc
getwc wint t _ fputwc
getwc
( wchar t char
( FILE * sp ) ;
getwchar wint t _ getwchar ( void ) ;
_ _ _out , FILE * sp ) ;
putwc
putwchar
wint t
wint t _ putwc
putwchar _
( wchar t char
( wchar t char _out ) ;
ungewtc wint t unget wc ( int char out, _ FILE * sp ) ;
File I/O
These functions work with binary files.
( read
( seek
size _t fread _ _ _
( void* in area , size t size, size t count ,
FILE* sp);
int fseek ( FILE* sp, long offset , int from
Joe );
ftell long ftell ( FILE * sp ) ;
( write size_ t fwrite ( const void * out_data , size_ t size , size_ t
count , FILE* sp ) -
rewind void rewind ( FILE * sp );
String I /O
Read and write strings.
fgets char 4 ( gets (char 4 string, int size, FILE 4 sp );
fputs int fputs ( const char 4 string, FILE 4 sp );
gets char 4 gets (char 4 string) ;
puts int puts ( const char 4 string ) ;
I he following wide-character functions also require the wchar.h library file.

fgetws wchar t 4 _ fgetws _
( wchar t 4 string, int size, FILE 4 sp );
fputws wchar t 4 _ fputws _
( const wchar t 4 string, FILE 4 sp );
System File Control

System commands that create and delete files on the disk.
remove int remove _
( const char * file name );
rename int rename ( const char * old _ name, const char * new_ name );
tmpfilc FILE* tmpfilc ( void );
tmpnam char * tmpnam
_
( char* file name );
F.5 Standard Library '
The following functions are found in stdlih.h.
Math Functions
.h .
The following math functions are found in stdlih
int abs ( int number );
ahs
div div_ t div ( int numerator, int divisor ) ;
long labs ( long number ) ;
labs
long long llabs ( long number ) ;
llabs
_ Idiv ( long numerator, long divisor ) ;
Idiv
lldiv
ldiv t
lldiv t _ lldiv ( long long numerator, long long divisor
( void ) ;
);
rand int rand

srand ( unsigned seed ) ;
srand void
F.6 String Library
Memory Functions
The Following are memory allocation Functions.
calloc void * calloc _ _
( size t num elements, size t element size ) ; _ _
Free
malloc
void
void
free
malloc
( void * );
_ __
( size t nilm bytes);
_ _
realloc void * realloc ( void * stge ptr, size t elemcnt size );
Program Control
The Following Functions control the program flow.
abort void abort (void );
_
atexit
exit
mt
void
at ex it
exit _
( void ( * ) function name (void ));
( int exit code );
Exit void Exit _
( int exit code ) ;
System Communication
The Following Function communicates with the operating system.
system int system _
(const char * system command ) ;
Conversion Functions
I he Following functions convert data From one type to another.
atof double' atof ( const char * string);
atoi int atoi (const char * string) ;
atol long atol ( const char * string) ;
atoll long long atoll ( const char * string) ;
strtod double strtod (const char* str, char** next str );_
strtof float strtof .
(const char* str char* * ncxt str ) ; _
strtol long strtol _
( const char * str, char** next str, int base);
_
strtold
strtoll
long double
long long
strtold
strtoll
(const char* str, char * * next str );
__
(const char * str, char * * next str, int base);
strtoul
strtoull
unsigned long
unsigned long long
strtoul
strtoull _
( const char* str, char** next str, int base );
( const char* str, char* * next str, int base);
F.6 String Library

I he following Functions are Found in string .h.
Memory Functions
The Following Functions manipulate block data in memory.
__
memchr void * memchr _ _
(const void mem , int a char, size t bytes );
1
mememp int mememp ( const void * mem I , const void * mem 2, size t bytes ) ;
memepy void * memepy _ _
( void * to mem , const void * Fr mem , size t bytes );
memmove void * memmove _
( void * to mem , const void * fr _ mem , size_ t bytes ) ;
AppendixP Function Libraries 1069
String Functions
I IK Following functions manipulate strings.
*
strcat char 4 strcat _

( char 4 to str, const char 4 fr _str ) ;
strchr char 4 strchr ( const char 4 str int a char ) ; . _
strcmp int strcmp ( const char 4 str 1 , const char 4 str2 );
strcpy char 4
_ strcpy ( char _ char fr_str);
to str, const 4
strlen size t
char to_str. const char fr_str, size_t bytes );
strlen ( const char 4 str ) ;
strncat char 4 strncat ( 4
strncmp int strncmp (const char strl const char str 2 , size_ t bytes );
4 4
( char to_str const char fr _str, size_ t bytes ) ;

,
strncpy char 4 strncpy 4
, 4
strpbrk char 4 strpbrk ( const char 4 strl , const char 4 str2 );

strcspn _
size t strcspn ( const char 4 strl , const char 4 str 2 );
strrchr char 4
strrchr ( const char 4 str, int a char ); _
strspn size_ t wstrspn .
( const char 4 strl const char 4 str 2 );
strstr char 4 strstr ( const char 4 strl , const char 4 str 2 );
strtok char 4 strtok ( char 4 strl , const char 4 str 2 ) ;
I he following functions manipulate wide-character strings. They require the

ivcluir.h library.
vchar t _ 4 _ _ , wchar_t fr_str)
( wchar t 4 to 4
_ a _char)
wcscat wcscat str const ;
\
_
wcschr
_ strl. vvcharjwcharj
(const \vchar t 4 str
4
\ vchar t wcschr ;
4
str ) ;
vvcscmp
vvcscpy
int
wcharj 4
vvcscmp
vvcscpy _ __ . const vvchar_t fr__str 2
( const \vchar
( vvchar t 4 to
t4
str
, const
4
);
t 4 str );
vvcscspn sizej vvcscspn
_t strl , wchar 2 _
( const \vchar t4 const
vvcslcn sizej vvcslcn
_ __ const vchar_t fr__ size bytes
(const wchar 4
str );
); 4
2 size_ t bytes
vvcsncat wcharj 4 strncat ( \vchar t 4 to str, \ str, t
4 );
_str, const vchar_ fr__ _ bytes)
t strl , const wchar t
4 str ,
( const \vchar
vchar _
wcsncmp int wstrncmp
wcsncpy wcharj4 wcsncpy (\ t 4 to \ t4 str, size t ;
4
const wcharj strl const wchar t 2 str );
wcharj a _char);
4
wcspbrk wcharj4 wcspbrk ( ,
4
wcsrchr wcharj4 wcsrchr (const wcharj str,
4
wcsspn sizej wcsspn ( const wcharj4 strl , const wcharj str 2 ) ;
wcharj 4
wcsstr (const wcharj 4
strl , const wcharj 4
str 2 );
wcsstr 4
( wcharj strl , const wcharj str 2 ;
4 )
vvcstok wcharj 4 vvcstok
r
F.7 Time Library
The following functions are found in time.h.
clock clockj clock
difftime
( void ) ;
( timej: time start , timej time end ;
) _ _
difftime double
(struct tm caljime );
4
mktime timej mktime

( timej numjime );
4
time timej time

(const struct tm 4 caljime);
asctime char 4 asctime 4 numjime ) ,
ctime (const timej
ctime char 4
(const timej numjime );
4
gmtime struct tm
4
gmtime
( const timej numjime );
4
local time struct tm 4 localtime 4

(char 4
str, sizej maxsize, const char format ,
strftime sizej strl time 4 ) ;
const struct tm timeptr
I
Preprocessor Commands
fhe C com piler is made of two functional parts: a
preprocessor and
tor. The preprocessor uses programmer-supplied commands to a transla-
prepare the
source program for compilation. The translator converts the C
statements
into machine code that it _places iin an object module.
Depending on the com -
piler design , the preprocessor and translator can work together, or
the prepro-
cessor can create a separate version of the source program , which is then
read by the translator. This is the design shown in Figure G - l .
Compilation
Source
Preprocessor
h
Translation
> Translator
•
H
*
Object
Program Module
Unit
FIGURE G- l Compiler Components
The preprocessor can he thought ol as a smart editor. Like a smart editor,

excludes, and replaces text based on commands supplied
it inserts, includes,
by the programmer. In this case, however, the commands are made a perma -
nent part of the source program .
All preprocessor commands start with a pound sign ( # ). Some of the
traditional compilers require the pound sign to be in the first column. ANSI /
ISO C specifies that it can be anywhere on the line. Preprocessor commands
can be placed anywhere in the source program. To distinguish
a preprocessor
command , the preprocessor commands are often
command and a program
called commands.
1071
1072 Section G . 2 Macro Definition
In this appendix, we first discuss three major tasks of a preprocessor: file

inclusion , macro definition , and conditional compilation . We then briefly dis-
cuss less common commands such as line, error, and pragma.
G.l File Inclusion

The first and most common job of a preprocessor is file inclusion , the copy-
ing of one or more files into programs. The files are usually header files that
contain function and data declarations for the program , hut they can contain
any valid C statement .
The preprocessor command is Hinclude , and it has two different formats.
The first format is used to direct the preprocessor to include header files from the
system library. In this format, the name of the header file is enclosed in
pointed brackets. The second format makes the preprocessor look for the
header files in the user-defined directory. In this format, the name of the file
pathname is enclosed in double quotes. The two formats are shown below:
# include <filename.h>
# include "filepath.h "
Figure G - 2 shows the situation of the source file before and after the pre-
processor as included the header files.
System Library User Directory

; stdio.h : • ilel .h i
#i. nclude <stdio.h> tv tk

#include " filel.h "
int main ( void ) int main ( void )
{ {
} // main } // main
Before After
FIGURE G- 2 File Inclusion
G. 2 Macro Definition
The second preprocessor task is expanding macro definition . A macro defini -
tion command associates a name with a sequence of tokens. The name is
the macro body. A

macro .
called the macro name and the tokens are referred
definition has the following form:
to as
# define name body
The body is the text that is used to specify how the name is replaced in
the program before it is translated. Before we discuss different applications of
the macro definition, let us clarify two important issues:
1. Macros must be coded on a single line. If the macro is too long to fit on a
line, we must use a continuation token. The continuation token is a
.
backslash ( \ ) followed immediately by a newline If any whitespace is
present between the backslash and the newline, then the backsla
sh is not
a continuationtoken, and the code will most likely generate an error. An
example of a macro continuation is shown below:
#define PREND \
printf ( " Normal end of program PA 5 01./n ); -
"
2 We need to be careful in coding the macro

. body. Whenever a macro call
( name of the macro in the program) is encoun
tered, the preprocessor
. body is not created carefully,
replaces the call with the macro body If the
it may create an error or undesired
result l or .
examp le, the following
macro definition:
# define ANS = 0
shown below:
compile error when it is used
as
creates a
num = ANS;
After preprocessing, the

result would he
// W e needed num - 0;
num =- 0;
wanted.
which is not what we
Coding Defined Constants
—
is to define a constan
t . As we saw in
of a macro following shows

rivmter 2 this
an Spl of
'
e
is
.
The simplest application ways we define a constant The
one of the
con
define n ,,
end for con definition. The name is SIZE
1074 Section G.2 Macro Definition
and the body is 9. Whenever in the program SIZE is encountered , it is replaced

with 9.
#define SIZE 9
Figure G -3 shows how we use constant definitions. As the figure shows,

the replacement can happen anywhere in the code, including the declaration
section .
#define ROWS 5
#define COLS 4
int main ( void ) int main ( void )
{ } // main
int ary[ ROWS][ COLS ]; int ary[ 5 ][ 4 ];
} // main } // main
Before After
FIGURE G- 3 Macro Definition to Define Constants
I he body of the macro definition can be any constant value including

integer, real , character, or string. However, character constants must be
enclosed in single quotes and string constants in double quotes.
A macro definition can simulate a constant definition.
Macros That Simulate Functions

I he preprocessor’s macro facility is very powerful . It can even be used to
simulate functions. In this section we discuss using macros in place ol func-
tions , first lor functions with no parameters and then for functions with
parameters.
Macros to Simulate Functions without Parameters

When we simulate a function with a macro, the macro definition replaces the
function definition . The macro name serves as the header and the macro
body serves as the function body. The name of the macro is used in the pro-
gram to replace the function call . Figure G - 4 shows how we can write a func -
tion to flush the standard input buffer.
void flush ( void ) ;

i n t main ( void )
void flush ( void )
{ {
while ( g e t c h a r ( ) 1 = ' \ n ' ) ;
flush ( ) ; return ;
} / / flush
} / / main
main flush
FIGURE G- 4 A Function for Flushing the Standard Input Buffer
\ \ hile functions are called when the program is executed , macro defini-
tions are inserted into the code during the preprocessing. The preprocessor
replaces the macro name in the program with the macro body. The macro
definition is simpler than the function , however, because we do not have to
write the function declaration , the function header, or the return statement.
It is also more efficient because the overhead associated with a function call
is not required ; the substituted code simply becomes statements in the pro-
gram . Figure G - 5 shows how we can use a macro definition to simulate the
function in Figure G - 4.
#d e f i n e FLUSH \
w h i l e ( getchar ( ) i = \n ' )
i n t main ( void ) i n t main ( void )
{ {
FLUSH ;
while ( getchar ( ) ! = \n ' ) ;
} / / main
} / / main
After
Before
FIGURE G- 5 Macro to Simulate a Function Call Without

Parameters
Macros to Simulate Functions with Parameters

to simulate a simple function
with parame-
We can also use macro definitions of the macro is done in two steps,
ters. In this case, however, the
replacement
as shown in Figure G -6.
1076 Section G.2 Macro Definition
#define \
PRODUCT( x , y ) x *
int main ( void ) int main ( void ) int main ( void )
{ { {
p = PRODUCT ( 4 , 5); p = x * y; p = 4 * 5;
} // main } // main } // main

Before After Step 1 After Step 2
FIGURE G- 6 A Macro to Simulate a Function Call with Parameters
1. The body of the macro replaces the macro call with the same actual
parameters used in the macro definition .
2 . The actual parameters are replaced with formal parameters.
One of the advantages of using macros instead of function is type inde-

pendence. We can define a PRODUCT macro that multiplies any two pairs of
data type. We do not have to write separate macros, one for each pair.
We need to give two warnings:
1 . To include parameters in the macro, the opening parenthesis must he
placed immediately at the end of the macro name; that is, there can he
no whitespace between the macro name and the opening parenthesis ol
the parameter list . It there is a space, then the opening parenthesis is
considered part of the token body and the macro is assumed to he simple.
2 . It is a strong recommendation that the formal parameters, in the body of
macro he placed inside parentheses. The reason is that macro uses text
replacement. It replaces the actual parameters with formal parameters.
Let us look at the same example , hut this time we pass a + 1 and b + 2
instead of 4 and 5 :
Alter the first and second step , we have:
p = x * y;
p = a + 1 * b + 2 ; // We need ( a +1 ) * ( b+2)
I he solution would he to include the formal parameters inside

parentheses. In other words , make the parentheses part ol body code.
Now we have the following:
# define PRODUCT( x , y ) ( x ) * ( y )
Appendix G Preprocessor Commends 1077
Alter the first and second step, we have the correct result :
P = (x) * (y); / / After f i r s t step

p = ( a + 1) * ( b + 2 ) ; / / After second step
EXAMPLE G- l Macros can replace many simple functions. One good example is the func-
tion we wrote in Chapter 14 to rotate a number n hits. This function can be
written as a macro as shown below. Note the use of backslash to continue the
definition on the next line.
_
# d e f i n e ROTATE LEFT ( x , n ) \
( ( (x) « (n) ) | ( (x) » ( 32 -x) ) )
EXAMPLE G - 2 We often need to use a power of 2 in a program. To calculate 2 \ we can easily

use the shift operator in a macro as shown below:
tdefine POWER 2 ( x ) 1 « ( x )
he empty. For example, most C

EXAMPLE G- 3 The parameter list in the macro definition can function from the getc
implementations use a macro to define the getchar
function.
)
tdefine getchar ( ) getc ( stdin
Nested Macros ga
nested macros by simply rescannin
It is possible to nest macros. c handles an expansion results in a new state-
, if
line after macro expansion. Therefore will be properly expanded . For exam -
macro
ment with a macro, the second
below.
pie, consider the macros shown
a) * ( b)
tdefine PRODUCT(a, b) (CT ( a , a )
tdefine SQUARE(a) PRODU
The expansion of
x = SQUARE ( 5 ) ;
n :
results in the following expansio
;
x = PRODUCT(5 , 5 )
1078 Section G. 2 Macro Definition
which after rescanning becomes
x = (5 ) * (5);
Undefining Macros
Once defined, a macro command cannot be redefined. Any attempt to rede-
fine it will result in a compilation error. I lowever, it is possible to redefine a
macro by first undefining it, using the ttundef command and defining it again
as shown below:
#define SIZE 10
#undef SIZE
#define SIZE 20
Predefined Macros
The C language provides several predefined macros, fable G- l lists some of
them. Note that these macros cannot be undefined using the undej command.
Command Meaning
DATE Provides a string constant in the form "Mmm dd yyyy"
containing the date of translation.
FILE Provides a string constant containing the name of the
source file.
LINE Provides an integer constant containing the current statement

number in the source file .
TIME Provides a string constant in the form "hh:mm:ss" containing
the time of the translation.
STDC Provides an integer constant with value 1 if and only if the
compiler confirms with ISO implementation.
TABLE G- l Predefined Macros
Program G - l demonstrates these macros.
PROGRAM G- l Demonstrate Pre - Defined Macros

1 / * Show the use of pre-defined macros
2 Written by:
3 Date:
continuec
PROGRAM G- 1 Demonstrate Pre -Defined Macros (continued)

4 */
6
7 int main ( void )
8 {
9 // Statements
10 printf (" line %d \n" , LINE )
11 printf ( "file %s\n" , FILE )
12 printf ( "date %s\n" , DATE )
13 printf ("time %s\n" , TIME )
14 printf ("ISO compliance %d\n", STDC );
15 printf (" line %d \n" , LINE );
16
17 return 0 ;
18 } // main
Results:
line 10
file ApG-Ol.c
date May 23 2005
time 20:36:28
ISO compliance 1
line 15
Note that we use the _ _LINE

_ _ macros two times to show that it prints
the current line number .
Operators Related to Macros directly or indirectly related
to macros.
that are
C provides several operators
We briefly discuss them here
.
String Converting Operator (#) that converts a formal

( # ) is a macro operation
t he string convertin g operato r
. examp le, the following
surrou nded by quotes For
parameter into a string by its value. It can he a very
of a variab le follow ed
macro prints the name .
m
helpful macro when debugging a progra
" , ( a))
#
_
define PRINT VAL (a) printf (
#a » contains: %d\n
following example
, as shown in the
When called in a program
PRINT VAL (amt);

1080 Section G . 3 Conditional Compilation
the preprocessor expands the macro to
printf ( " amt" "contains: %d \n" / amt );
Recall that the preprocessor automatically concatenates two string literals

into one string. After the automatic concatenation , the statement becomes
printf ( "amt contains: %d \n" , amt );
Merge Operator ( ##)

Occasionally, it may he necessary to write macros that generate new tokens.
With the merge command operator, two tokens are combined . For example,
imagine we want to create Al, B3, and Z8 in our program. This can be done
easily with the following macro definition :
# define FORM(T , N ) T# # N
Now ii we use the following code:
int FORM ( A , 1 ) = 1;
float FORM ( B , 3 ) = 1.1;
char FORM ( Z , 8 ) = ' A';
we get
int Al = 1;
float B3 = 1.1 ;
char Z8 = ' A' ;
The defined Operator

The defined operator can he used only in a conditional compilation ( see
Section G . 3 ) ; it cannot he used in macros. The value of defined ( macro- name )
is 0 il the name is not defined and 1 if it is defined . For example, after
#define PI 3.14
the value of defined ( PI ) is 1 and the value of !defined (PI ) i s 0.
G.3 Conditional Compilation

Ihe third use of the preprocessor commands is conditional compilation.
C onditional compilation allows us to control the compilation process by
including or excluding statements.
Two-Way Commands
I he two-way command tells the preprocessor to select
in Figure G - 7.
# if expression
between two choices.

This is similar to selection , which we studied in Chapter 5. The idea is
shown
1
Code to be included if
“expression" is true.
# else
“expression" is false.
#endif
FIGURE G-7 Two - Way Command
I he expression is a constant value that evaluates to zero or non -zero. If the

value is non -zero, it is interpreted as true and the code after # if is included . If
it is zero, it is interpreted as false and the code after # else is executed .
I here are two differences between this format and the one used
in
does not to have to be included in parenthe-
Chapter 5 . First , the expression
. Second , the codes
ses , although it helps readability when we use parentheses
). The commands
can not he included in a compound statement braces
(
# else and # endif serve as delimiters .

, in these cases, the
The # if part or the else part can be empty. However 8.
shown in Figure G-
alternative formats are normally used as
#if expression
# if expression
code for non-zero code for non-zero
expression expression
# else # endif
# endif
a. else Section Empty
# if Expression
# if expression
code for zero
#else expression
code for zero # endif
expression
# endif
b. // Section Empty
FIGURE G-8 Alternative Commands

1082 Section G . 3 Conditional Compilation
EXAMPLE G - 4 Let us see how a conditional command can help us in program development .
Imagine we need to include a very large file in our program . However, we
want to test the syntax of our program gradually, first without the included
file and then with it . One way we can do this is to include the file in our pro-
gram using conditional commands as shown below:
# if 0
# include " large.h "
# endif
When we are ready to test the syntax for the entire program , we change
the condition to true .
# if 1
# include " large.h "
# endif
As we see in a later example, there are easier and better ways to do this.
EXAMPLE G - 5 Imagine we need to include three files in our program : filel.h, file2.h,
file3.h. These files contain generalized code that are used in projects.
Each file must therefore be able to stand alone. When combined , the result
may he multiple inclusion of common macros. If the macros are duplicated ,
we get a compilation error because the preprocessor does not allow duplica -
tion of macros.
To prevent this error we have a project standard : whenever we need to
define a macro, we define it in a conditional command using the defined
operator we discussed in “The defined Operator ” in Section G . 2 . So.
instead of
#define GO 0
we use
# if Idefined ( GO)
#define GO 0
# endif
Now we are safe. Even if all three files have used the same conditional
definitions, only the first one appears in the program ; the rest are skipped by
the preprocessor. When the first definition is encountered , GO is not defined ,
so the expression is true and it becomes defined . In the second and third files ,
( > ( ) is already defined , so the conditional
compilation expressions are false,
and the define command is bypassed .
Appendix 6 Preprocessor Commonds 1083
To be safe, always use macro definitions in a conditional command .

Two Abbreviations
You may have guessed that the previous commands are used often iin source
liles. The C language provides two additional macro commands as shown in
-
Figure G 9.
#if defined name #ifdef name
#if !defined name > #ifndef name
FIGURE G- 9 Two Abbreviations
, it is common to
EXAMPLE G -6 When we decompose a large project into different programs
data types and struc-
create a library hie that contains all of the application
tures . Each program in the project then
includes the library file, ensuring
descriptions.
that everyone has exactly the same data names and
, when multiple files are compiled
This design creates a problem , however , which results in
included for the data
together. Each file has the same library
solution is to enclose the data
compile errors with the duplicate names. The the next example.
library file in a defined command as shown
in
_
# ifndef PROG LIB
__
—
# define PROG LIB
# include "proj lib.h
"
# endif
named pROG_LIBhas y
cuted only if
case , we
„ and
a constant
defined, the statements are more than once.

the library from being included
^ M
jf jt
tQ shovv the
^ usp version of
has been
Note that the name
the filename, without the

“. h ^
A good strategy t0
inclusions, always
use the
include command is used. " suffix, as a defined name.
uppercase
use file Inclusion

In a conditional
To prevent duplicate
command .
1084 Section G.3 Conditional Compilation
In most implementations , system header files such as stdio . h use the

above strategy to prevent multiple inclusions of that file in a source program .
For example , an implementation may use
# ifndef STDIO
# define STDIO
Rest of the code
# endif
EXAMPLE G - 7 Another application of conditional compilation is found in debugging. Most

GUI coinpliers have a debugger that can be used by the programmer. If
there is not such a debugger, however, we can insert printj statements to
check the value of the program during testing. If we just include them as part
of the code , however, when the program is ready for production we must
remove them . Later, when we need to change the program , we must re - insert
them as needed .
A simpler solution is to insert them conditionally. We enclose the
print / statement inside the # if . .. #endif commands to print the debug state -
ments only when we run in the “ debug" mode . For example , the following
program prints the value of x only when DEBUG is true . After testing is
done , we can just set the define command to false and the debug state -
ments are not included in the code . This debug concept is demonstrated in
Program G - 2 .
PROGRAM G - 2 Demonstrate Conditional Debugging

1 /* Use of conditional compilation for debugging
2 Written by:
3 Date:
4 */
5 // Macro to print integer values
6 _
#define PRINT VAL( a ) \
7
8
—
printf ( " At line %d " , LINE ) ; \
printf ( # a " contains: %d \n" , ( a ))
9
10 # define DEBUG 1
11
13
15 {
16 // Declarations
17 int x ;
18 // Statements
continue*
PROGRAM G- 2 Demonstrate Conditional Debugging (continued)

19 | x = 1023 ;
20
21 # if DEBUG
22 PRINT_VAL ( x );
23 # endif
24
25 // Later in program
26 for ( int i = 0; i < 2; i++)
27 {
28
29
x = x * x;
_ _
PRINT VAL ( i ); PRINT VAL( x );
// Square x
30 > // for
31 return 0;
32 } // main
Result during testing:

At
At
line 22 x
line 29 i
—— contains:
contains:
1023
0
At
At
line 29 x
line 29 i
—— contains:
contains:
1046529
1
At line 29 x — contains: 6287361
production pro-
Program G- 2 Analysis Study this proqram carefully; you will find it useful when you write two of the tech-
6-8 This uses
grams. First, study the macro definition in lines
. macro
chapte r We begin by display ing the source file
niques we discussed earlier in the .
. Becaus e we have
line number so that we know where we are within the program each Then we use the
unique ly identify one .
multiple debugging displays, we need to the variabl e identifi er and its
ion operator ( see Section G.2) to display
string convers
contents.
In the body of the program we display
the values of the variables x and i. In the
more displays to track the execution
of
debugging of a large program, we would use
the program.
Multi- Way Commands of the choices

be multiway; selecting one
Conditional commands can also 5 that multi-way selection is possible
among several. We saw in
Chapter , we use the ttelif
the use of else - if constru ct . In the preprocessor selection in the
with
the format of multiway
command. Figure G - IO shows
preprocessor . application that must run instal-
at mul
examp le, when we write a software . Each -
For code for each location
tiple location s, we need to define unique unique code . One way
headings and perhaps other
lation needs unique report
to do this is shown in Program G -3.
G.3 Conditional Compilation
#if expression!
"expression!” is true.
#elif expression2
"expression2" is true.
#else
both expressions are false.
#endif
FIGURE G- 10 Multi - Way Commands
ROGRAM G- 3 Conditional Multi - Way Selection

1 #define Denver 0
2 #define Phoenix 0
3 #define SanJose 1
4 #define Seattle 0
5
6 #if ( Denver )
7 # include "Denver.h "
8 #elif ( Phoenix )
9 # include "Phoenix.h "
10 #elif ( SanJose)
11 #include " SanJose.h "
12 #else
13 tinclude "Seattle.h "
14 #endif
m G- 3 Analysis First, we create a unique header file for each installation. Then, using conditional
commands, we select which include file to use based on defined flags for each site. In
this example, we have selected San Jose, so only its code is included in the program.
Summary of Conditional Commands

I able G- 2 shows the summary of commands used in this section.
Command Meaning
#if expression When expression is true, the code that follows is included
for compilation.
# endif Terminates the conditional command.
# else Specifies alternate code in two-way decision.
# elif Specifies alternate code in multi-way decision.
# ifdef name Abbreviation for #if defined name.
# ifndef name Abbreviation for #if /defined name.
TABLE G- 2 Conditional Compilation Commands
G.4 Other Commands

that we briefly discuss here.
The preprocessor uses some other commands
Line Command
in two formats, can set the
line number
The line command , which is used
and the filename lor a program. For
example ,
.... . .IZ*--;* ^
# line 100
“the program
nc >
The second
of ,K r
. For example,
«
»
«
c
“d — 3 i
# line 100 "MyProgram

.c "
..
sW he next line number
» 100 »d
be checked by the predebned
macro
_ “—“ for the program that can
the concept .
Program G - 4 demonstrates
1088 Section G.4 Other Commands
PROGRAM G- 4 Demonstrate Line Command

1 /* Demonstrate the use of line command
2 Written by:
3 Date:
4 */
5 # line 100 " myprogram.c "
7
8 int main ( void )
9 {
10
11 // Statements
12 printf ( " line %d \n" , LINE )
13 printf ( " file %s\n " , FILE )
14 printf ( " line %d \n " , LINE )
15 return 0 ;
16 > // main
Results:
line 106
file myprogram.c
line 108
Error Command
I he error command is of the form
terror message
It is used to print the message detected by the preprocessor. For example,

Program G - 5 demonstrates code that verifies that a program needs to define
both TRUE and FALSE or neither of them and prints an appropriate mes-
sage when an error is detected. Note that you may get a slightly different
message format depending on your compiler.
PROGRAM G- 5 Demonstrate Error Command

1 /* Show the use of error command
2 Written by:
3 Date:
4 */
5 #define TRUE 1
6
7 # if defined (TRUE ) && Idefined ( FALSE )
8 terror You need to defined FALSE too.
continual
Appendix G Preprocessor Commends 1089
PROGRAM G- 5 Demonstrate Error Command ( continued )

9 # e l i f defined ( FALSE ) & & Idefined ( TRUE )
10 terror You need to defined TRUE too .
11 tendif
12
13 tinclude < stdio . h >
15 {
16 / / Statements
17 p r i n t f ( " J u s t a test \ n " ) ;
18 return 0 ;
19 > / / main
Results :
Error : preprocessor terror command
- .
ApG 04 c l i n e 8 terror You need to defined FALSE too .
When this program is compiled , we get
terror You need t o defined FALSE too .
Pragma Command
The pragma command
tpragma tokens
actions. We do not dis-

causes the compiler to perform implementation -defined
used only in advanced environments.
cuss t his command in detail because it is
Null Command allows a null command. A com-

It is interesting to mention that C language
mand of the type
t
and does not generate a compilation error.
is considered a null command
Command-Line Arguments
In all the programs we have written , main has been
coded with no parame-
ters. But main is a function, and as a function, it may have parameters.
When
main has parameters, they are known as command - line
arguments.
Command - line arguments are parameters to main when the program
starts. I hey allow the user to specify additional information when the
pro-
gram is invoked. For instance, if you write a program to append two files,
rather than specif ) the names of the files as constants in the code, the user
could supply them when the program starts. Thus, a UNIX user in the Korn
Shell might execute the program with the command line shown below.
$appendFiles filel file2
H. 1 Defining Command- Line Arguments

As the programmer, you design the parameter lists for functions you write.
W hen vou use system functions, such as rand , you follow the parameter
design set up by the language specification. Command- line arguments are a
little like both . As the programmer, you have control over the names of the
parameters , but their type and format are predefined for the language.
I he function main can he defined either with no argument { void ) or with
two arguments: one an integer and the other an array of pointers to
char
(strings) that represent user-determined values to be passed to main. The
number of elements in the array is stored in the first argument . The pointers
to the user values are stored in the array. The two different
formats are shown
in Figure H - l .
1091
1092 Section H.l Defining Command-Line Arguments
int main ( void ) int main ( int argc ,

char *argv[ ])
{ {
// Statements // Statements
} // main } // main
Without command - line arguments With command-line arguments
FIGURE H - 1 Arguments to main
Although the names of the arguments are your choice, traditionally

they are called urge ( argument count ) and argv (argument vector). I his
data structure design, with six string pointers in the vector, is shown in
Figure H - 2.
program
name
\0
& \0
—
\0
6 —
\0
*
argc
hd
- * \0
& \0
ml
argv
FIGURE H - 2 orge and argv format
I he first argument defines the number of elements in the array identified

in the second argument . The value for this argument is not entered using the
keyboard; the system determines it from the arguments the user types .
The value of argc is determined from the user -typed values for arg s
^.
Appendix H Command-Line Arguments 1093
H.2 Using Command-Line Arguments

The urge array has several elements. The first element points to the name
of the program ( its filename ). It is provided automatically by the program .
The last element contains NULL and may he used to identify the end of the
list . The rest of the elements contain pointers to the user-entered string
values.
-
To fully demonstrate how command line arguments work, let’s write a
small nonsense program. It does nothing hut exercise the command - line
arguments. The code is shown in Program H - l .
PROGRAM H- 1 Display Command -Line Arguments

1 -
/* Demonstrate the use of command line arguments.
2 Written by:
3 Date:
4 */
6 #include <string.h>
8
9 int main (int arge, char* argv[ ])
10 {
11 // Statements
" arge );
12 printf ("The number of arguments:: %d\n ,
name of the program:: %s\n " , argv[0]);
13 printf ( " The
14
15 for (int i = 1 ; i < arge; i++)
\ ", i, argv[i]);
16 printf ("User Value No. %d: %s n
17
18 return 0;
19 } // main
, let’s run it with several different

Now that we’ve written the program arguments. Even when the user
no user
arguments. First , let s run it with
’
is still supplied by the
system . ( For
’ t supply values, the program name
is
doesn - run line environment .)
all of these runs, we assume a Windows
C:>cmdline
:1
The number of arguments
program: CMDLINE
The name of the
command .
' add hello to the run
For the second run , let s
1094 Section H.2 Using Command-Line Arguments
C:>cmdline hello
The number of arguments: 2
The name of the program: CMDLINE
User Value No. 1: hello
To make the exercise more interesting, let’s run the program with a
phrase on the command line.
C:>cmdline Now is the time

User Value No. 1: Now
User Value No. 2: is
User Value No. 3: the
User Value No. 4: time
But what if our intent is to read a phrase? C looks at the user values as
strings. This means that the spaces between the words separate each word
into a different element. If we want an element to contain more than one
.
word, we enclose it in quotes just like a string in code
C:>cmdline "To err is human " Pope

User Value No. 1: To err is human
User Value No. 2: Pope
The user elements cannot he numbers. If you enter a number, it is taken

as astring. I lowever, the string can he converted to a number using a stan -
dard function such as strtocl .
Pointers to void and to Functions
In this appendix we discuss pointer to void and pointer to function .
1.1 Pointer to void

Because C is strongly typed , operations such as assign and compare must use
compatible types or be cast to compatible types. The one exception is the
pointer to void , which can be assigned without a cast . In other words
,a
that can he used to represent any data type
pointer to void is a generic pointer
to void .
during compilation or run time. Figure 1- 1 shows the idea of a pointer data
to a generic
Note that a pointer to void is not a null pointer; it is pointing
type ( void ) .
FIGURE 1 - 1 Pointer to void
to void concept . It
demonstrate the pointer
EXAMPLE 1- 1 Let us write a simple program to , a floating- point number, and a void
contains three variables: an integer the pointer can he set
to the
in the program - shows
pointer. At times
different in 1 2
floating point value. Figure
-
of the integer value or ol the
address
the situation .
Program 1 - 1 uses a pointer
to ind that
.
we can use to pr int either
an inte -
ger or a real number.

1096 Section 1.1 Pointer to void
void* Pr -
K
int i;
float f;
P
P = &i;
P = &f;
*
P
FIGURE 1 - 2 Pointers for Program 1 - 1
PROGRAM 1 - 1 Demonstrate Pointer to void

1 /* Demonstrate pointer to void.
2 Written by:
3 Date:
4
5 */
7
8 int main ( void )
9 {
11 void * p ;
12 int i = 7;
13 float f = 23.5 ;
14
15 // Statements
16 p = &i;
17 printf ( " i contains: %d \n " , *(( int* )p) );
18
19 P = &f ;
20 printf ( " f contains: %f \n" , *(( float* )p ));
21 return 0;
22 > // main
Results:
i contains 7
f contains 23.500000
Program 1- 1 Analysis The program is trivial, but it demonstrates the point. The pointer p is declared as a
void pointer, but it can accept the address of an integer or floating-point number.
However, we must remember a very important point about pointers to void : a pointer
ApgendjxLPointers to Void and to functions 1097
to void cannot be dereferenced unless it

is cast. In other words, we cannot use
without casting. That is why we need to cast the pointer *p
in the print function before
we use it for printing .
A pointer to void cannot be dereferenced unlees

it is cast .
1.2 Pointer to Function

I unctions in a program occupy memory. The name of the function
is a
pointer constant to its first byte of memory. For example, imagine that we
have tour functions stored in memory: main, fun, pun, and sun. This rela-
tionship is shown graphically in Figure 1-3. The name of each function is a
pointer to its code in memory'.
main
code for:
int main (void )
fun — code for:
void fun (void)
pun — code for:
f2 int pun (int, int)
sun — code for:
f3 double sun (float)
pointer to Memory
function variables
FIGURE 1 - 3 Functions in Memory
Defining Pointers to Functions

to function vari-
Just with all other pointer types, we can define pointers . To declare a
as
ables and store the address of fun. pun, and sun in them
a prototype definition, with the
pointer to function, we code it as if it were
is shown in Figure 1- 4. The
function pointer in parentheses. This format the function return
C interprets
parentheses are important • without them
type as a pointer .
Using Pointers to Functions , lets write a
use pointers to functions
Now that you’ve seen how to create and any two pieces of data. The func-
larger of
generic function that returns the . While our
tion uses two pointers to void as
described in the previous section by the void
of the two values represented
function needs to determine which compare them because it doesnt know
directly
pointers is larger, it cannot program
what type casts to use with the
void pointers. Only the application
knows the data types.
1098 Section 1.2 Pointer to Function
f 1 is a pointer to a
function with no parameters
s. that returns void
^
void (*fl ) (void );
int ( f 2 ) ( int, int );
double ( *f 3 ) (float );
// Statements
fl = fun;
f2 pun;
f3 sun ;
FIGURE I- 4 Pointers to Functions
The solution is to write simple compare functions for each program that
.
uses our generic function Then, when we call the generic compare function,
we use a pointer to function to pass it the specific compare function that it
must use .
EXAMPLE I- 2 We place our generic function, which we call larger , in a header file so
that it can be easily used. The program interfaces and pointers are shown in
Figure I- 5 . The code is shown in Program 1- 2.
PROGRAM 1 - 2 Larger Compare Function

1 I* Generic function to determine the larger of two
2 values referenced as void pointers.
3 Pre dataPtrl and dataPtr2 are pointers to values
4 of an unknown type.
5 ptrToCmpFun is address of a function that
6 knows the data types
7 Post data compared and larger value returned
8 */
9 void * larger ( void * dataPtrl , void * dataPtr 2 ,
10 int ( *ptrToCmpFun)(void* , void*))
11 {
12 if (( * ptrToCmpFun) (dataPtrl , dataPtr 2) > 0)
13 return dataPtrl ;
14 else
15 return dataPtr2 ;
16 } // larger
Appendix I Pointers to Void and to Functions 1099
main larger
X X
i 7 j 8 lrg 8 ] |
1
TI | dataPtrl
I
—
i
l
compare
i
i
t
- 1
dataPtr2
I
ptrl
I
ptrl
a
compare
FIGURE 1- 5 Larger Compare Function
Program 1 - 3 contains an example of how to use our generic compare pro-

gram and pass it a specific compare function .
PROGRAM 1 3 Compare Two Integers

-
1 /* Demonstrate generic compare functions and pointer to

2 function.
3 Written by:
4 Date:
5 */
// Header file
7 -
tinclude " Apl 05.h"
8
ptr2 );
9 int compare ( void * ptrl , void *
10
12 {
14 int i = 7 ;
15 int j = 8;
16 int lrg;
17
18 // statements , compare));
larger (&i / j'
19
20
lrg = ( *(int*)
is
printf( "Larger value is
: % d \ n ^
" / lrg );
21 return 0;
22 } // main continued
ion 1.2 Pointer to Function
PROGRAM 1- 3 Compare Two Integers (continued )

23
24 /*
25 Integer specific compare function.
26 Pre ptr1 and ptr2 are pointers to integer values
27 Post returns +1 if ptrl >= ptr 2
28 -
returns 1 if ptrl < ptr2
29 */
30 int compare ( void * ptrl , void * ptr 2 )
31 {
32 if ( *( int*)ptrl >= *(int* )ptr 2 )
33 return 1 ;
34 else
35 return -1 ;
36 > // compare
Results:
Larger value is: 8
EXAMPLE 1 - 3 Now, let ’s write a program that compares two real ( float ) numbers. We can use
our larger function, but we need to write a new compare function. We repeat
Program 1 - 3, changing only the compare function and the data-specific state -
ments in main . The result is shown in Program 1- 4.
PROGRAM 1- 4 Compare Two Floating - Point Values

1 /* Demonstrate generic compare functions and pointer to
2 function.
3 Written by:
4 Date:
5 */
7 -
#include " Apl 02.c" // Header file
8
9 int compare ( void * ptrl , void * ptr2 );
10
12 {
14 float i = 73.4 ;
15 float j = 81.7;
16 float lrg ;
17
18 // Statements
19 lrg = (*( float* ) larger (&i, & j, compare ));
continuec
Appendix I PointerstoVoidondjo Functions 1101
PROGRAM 1- 4 Compare Two Floating-Point Values (continued)

20 printf( " Larger value is: %.if \n" , I r g )
;
21 return 0;
22 } // main
23
24 /* ===== ==== compare ===== ============
25 Float specific compare function.
26 Pre ptrl and ptr2 are pointers to float values
27 Post returns +1 if ptrl >= ptr2
28 returns -1 if ptrl < ptr2
29 */
30 int compare ( void * ptrl , void* ptr2)
31 {
32 if ( *( float* )ptrl >= *( float* )ptr2)
33 return 1;
34 else
35 return 1; -
36 } // compare
Results:
Larger value is: 81.7
Appendix J
Storage Classes and
Type Qualifiers
of data types, storage classes and
In this appendix, we discuss two attributes
type qualifiers. Storage classes
specify the scope of objects. Type qualifiers
.
specify processing limitations on objects
J . l Storage Classes : auto,

the storage class of an object using one of four specifiers ,
We define class specifiers
we can discuss the storage
register , static , and extent . Before they are used .
however, we mus t describe
the environment in which
Object Storage Attributes three attributes of an

object’s storage as
Storage class specifiers control
, extent , 'and linkage.
scope
show n in Figure J - l : its
Object
Attrilbutes
Linkage
Extent
Scope
— automatic — internal
U- block — external
— file
I — static
FIGURE J- l Object Storage
Attributes
1103
1104 Section J.l Storage Closses
Scope
Scope defines the visibility of an object; it defines where an object can be refer-
enced . In C , an object can have four levels of scope: block, file, function , and
function - prototype. In this text , we do not discuss the function and function
prototype scopes.
Block ( Local ) Scope

The body of a function is a block and a compound statement in a loop is a
nested block within its function block. When the scope of an object is block ,
it is visible only in the block in which it is defined . An object with a block
scope is sometimes referred to as a local object.
The scope of an object declared anywhere in a block has block scope. F or
example, a variable declared in the formal parameter list ol a Function has
block scope. A variable declared in the initialization section ol a for loop also
has block scope, but only within the for statement.
File (Global ) Scope

File scope includes the entire source file lor a program , including any files
included in it . An object with file scope has visibility through the whole
source file in which it is declared . However, objects within block scope are
excluded from file scope unless specifically declared to have file scope; in
other words , by default block scope bides objects from file scope. File scope
generally includes all declarations outside a function and all function headers.
An object with file scope is sometimes referred to as a global object.
Extent
The extent of an object defines the duration for which the computer allocates
memory lor it . In C , an object can have automatic , static extent , or dynamic
extent . The extent ol an object is also known as its storage duration .
Automatic Fxtent
An object with an automatic extent is created each time its declaration is
encountered and is destroyed each time its block is exited . For example, a
variable declared in the body ol a loop is created and destroyed in each itera -
tion . Note however, that declarations in a function are not destroyed until the
function is complete; when the function calls a function , they are out ol
scope but not destroyed .
Static Fxtent
A variable with a static extent is created when the program is loaded for exe -
.
cution and is destroyed when the execution stops This is true no matter bow
many times the declaration is encountered during the execution .
Appendix J Storage Classes ondjype Qualifiers 1105
Dynamic Extent
Dynamic extent is created by the program through the malloc and its related
library functions. We do not discuss static memory in this text.
Linka ge
A large application program may be broken into several modules, with each
module potentially written by a different programmer. Each module iis a sepa-
rate source file with its own objects. Different modules may be related when
the program is link edited .
We can define two types of linkage: internal and external. As we discuss
later, the linkage characteristic differs from scope characteristic, although
they look the same at the first glance.
Internal Linkage
An object with an internal linkage is declared and \isible only in one module.
Other modules cannot refer to this object.
External Linkage
An object with an external linkage is declared in one module but is visible
in
declare with a special keyword , extern , as we discuss
all other modules that it
shortly.
Storage Class Specifiers

We have defined three storage classes scope — , extent , and linkage with —
x 2 ). If we want to use a speci-
potentially eight different combinations ( 2 x 2
eight different specifiers. How-
fier lor each combination , we need a total of
ever, there are only four storage class
specifiers, as shown in Figure J -2:
factors have contributed to this
unto , register , static , and extern. Three
reduced set .
Storage Class
Specifiers
extern
register static
auto
Has two uses.
FIGURE J- 2 Storage Class Specifiers

1106 Section J .l Storage Classes
1. The scope is not defined , it is understood . The place where an object is

declared defines the scope.
2. Not all combinations are logically possible.
3. One specifier, static , is used for two different purposes. The scope shows
what it defines.
In the rest of the section , we discuss these specifiers only when applied
to variables. The application of these specifiers to other objects , such as func -
tions, is similar.
Auto Variables
A variable with an auto specification has the following storage characteristic :
Scope: block Extent: automatic Linkage: internal
In other words, the variable must be declared in a block. Each time the
declaration is encountered , it is re-created. It is visible only in the source file
in which it is declared . Most of the variables we have declared so lar are auto
storage class. T his type of specifier is so prevalent that C allows us to omit the
specifier.
Initialization
An auto variable can be initialized where it is defined or left uninitialized . II
initialized , it receives the same initialization value each time it is created . II it
is not initialized , its value will he undefined every time it is horn . When the
auto variable is not initialized , its value is unpredictable. Program J - l shows
the use ol the auto variables.
PROGRAM J-l Demonstration of outo Variables

1 /* Show the use of auto variables
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
9 // Statements
10 for ( int i = 1 ; i <= 3 ; i++)
11 {
12 int x = 1 ;
13 x+ + ;
14 printf ( " Value of x in iteration %d is: %d \n " ,
15 if x);
continues
Appendix J Sloioge Classes and Type Qualifiers ]
I 07
PROGRAM J- l Demonstration of auto Variables (
continued )
16 } / / for
17 return 0;
18 > // main
Result:
Value of x in iteration 1 is: 2
Both i and x are auto variables. The declaration of i is encountered only

once { for loop initialization ), but the declaration of x is encountered three
times. Each time the program encounters the declaration for x, it is initial-
ized to 1 again .
.
The keyword auto does not have to be explicitly coded Each time a variable of
block scope with no specifier is encountered, it defaults to auto .
Register Variables
A register storage class is the same as the auto class with only one difference.
I he declaration includes a recommendation to the compiler to use a central
processing unit ( CPU ) register for the variable instead of a memory loca -
tion . Ibis is done for efficiency. The time required to access a CPU register is
significantly less than the time required to access a memory location. The
compiler, however, may ignore the recommendation. Registers are scarce
resources, and the compiler often needs all of them for the program to run
efficiently.
vari-
There is one restriction on the use of a register variable; a register
- we can ’t use the
able address is Lnot available to the user. This means that
address operator and the indirection operator with a register. It also disallows
register with mixed
implicit conversions, such as might occur when using a
mode arithmetic.
Static Variables with Block Scope

I he specifier static has different usages
depending on its scope. When it is
, static defines the extent of the
used with a variable that is declared in a blockfollowing characteristic :
variable. In other words, the variable has the
Linkage: internal
Scope: block Extent: static
be referred to only in the block

it is
A static variable in this context can computer allocates storage for this
; the
defined. The extent , however, is static , which means that it is not visible
variable only once. 7 he linkage is internal
1108 Section J.l Storage Classes
in other modules. Note that the specifier static must he explicitly coded in
this case; it cannot he omitted .
Initialization
Astatic variable can be initialized where it is defined , or it can be left unini -
tialized . If initialized , it is initialized only once. If it is not initialized , its value
will be initialized to zero. Note however, that it is initialized only once in the
execution of the program .
In a static variable with block scope, static defines the extent of the variable.
Program J - 2 shows the use of static variable in the block scope.
PROGRAM J - 2 Static Variables in Block Scope

1 /* Show the use of static variable with block scope
2 Written by:
3 Date:
4 */
6
7 int main ( void )
8 {
9 // Statements
10 for ( int i = 1 ; i <= 3 ; i++)
11 {
12 static int x = 1;
13 x++;
14 printf ( " Value of x in iteration %d is: %d \n" , i, x );
15 > // for
16 return 0;
> // main
Result:
Note that the variable i is an auto variable. The variable x , however, is a

static variable. Il is only initialized once although the declaration is encoun -
tered three times. If we do not initialize x , it is initialized to 0 because a static
value needs an initialization value .
Static Variable with File Scope
When the static specifier is used with a variable that has file scope (global
scope ) and we want to keep its linkage internal, it is defined with the specifier
ing characteristic:
Scope: file
Appendix J Storage Gosses and Type Qualifiers 1109
static . Note that the extent is still static. In other words, we
Extent: static
have the follow
Linkage: internal
1
Program J -3 shows the use ol the static variable with file scope.
PROGRAM J - 3 Static Variables with File Scope

1 /* Show the use of static variable with file scope
2 Written by:
3 Date:
4 */
6
8 void fun ( void );
9
10 static int x = 1;
11
13 {
14 // Statements
15 for ( int i = 1; i <= 3; i++)
16 x++;
17 printf ( "Value of x in main is: %d \n" , x);
18 f u n( );
printf ( " Value of x in main is: %d \n x);
" /
19
20
21 return 0;
22 } // main
23
24 /* fun -
25 Increment and print global variable.
been defined
26 Pre Global variable x has
Post x incremented and printed
27
28 */
29 void fun ( void )
30 {
31 // Statements
32 x ++ ;
is: %d\n", x );
33 printf ( " Value of x in fun
34 return;
35 > // fun
Result:
1110 Section J.l Storage Glosses
PROGRAM J - 3 Static Variables with File Scope (continued )

Value of x in main is: 4
Value of x in fun is: 5
Value of x in main is: 5
The declaration of a static variable with file scope must be in the global
area of the file ( outside any function). If there is another declaration with the
same identifier in the global area, we get an error message. On the other
hand, if the identifier is used with another declaration inside a function, the
new variable is an auto variable or static variable with block scope. In other
words, in this case, we are defining a new variable that overrides the global
variable making it not in the current scope.
External Variables
External variables are used with separate compilations. It is common, on
large projects, to decompose the project into many source files. The decom-
posed source files are compiled separately and linked together to form
one unit .
A variable declared with a storage class of extern has a file scope; the
extent is static, but the linkage is external.
Scope: file Extent: static Linkage: external
An extern variable must be declared in all source files that reference it,
but it can be defined only in one of them. External variables are generally
declared at the global level. The keyword extern is optional.
I bis design can create a problem: if three source files use the extern
specifier for a variable, which one is the defining file and which are simply
declaring the variable ? The rules differ depending on whether the variable is
initialized or not .
.
1 II it is not initialized, then the first declaration seen by the linkage editor
is considered the definition and all other declarations reference it .
Figure J - 3 contains an example of an external reference used in three

programs.
2. If an initializer is used with a global definition, it is the defining declaration.
It is an error il two external variables are initialized. It is also an error to

use the same identifier lor external variables with incompatible types.
Appendix J Storage Classes and Type Qualifiers 111 ]
No ion
‘‘extern” if extern ;e
% K
/ /,/êxterr?
#
int a ;
#i
ired
extern int a ;
Ik
extern int a;
int main ( void ) int funl ( void) int fun2 ( void)
{ { {
} // main } // funl } // fun2

(a) Definition File (b) Reference Filel (c) Reference File2
FIGURE J - 3 Defining and Declaring External References
We recommend that each external definition used in a project be placed

.
in a common definitions source file These definitions should not use the
.
key word extern and should have an explicit initializer Each source file that
needs to relerence an externally defined variable should use the keyword
extern and must not use an initializer.
The keyword extern must be explicitly used in source files that use an external
.
variable It is omitted in the source file that defines and initializes the variable.
Summary
We have summarized what we discussed about storage classes in Table J - 1 -
Class Scope Extent Linkage Keyword
auto block automatic internal auto or none
register block automatic internal register
static (extent) block static internal static
file static internal static

static (linkage)
file static external extern or none
extern
TABLE J - 1 Summary of Storage Classes
J.2 Type Qualifiers .

special attributes to types: coast
The type < jualiKer (Figure j- 4) adds three
volatile , and restrict .
1112 Section J .2 Type Qualifiers
Type
Qualifiers
const volatile restrict
FIGURE J - 4 Type Qualifiers
When a storage class, such as static, and a type qualifier are both needed,
the storage class comes before the type qualifier.
Constants
The keyword for the constant type qualifier is const . A constant object is a
read-only object; that is, it can only be used as an rvalue. A constant object
must be initialized when it is declared because it cannot be changed later. A
simple constant is shown below.
const double PI = 3.1415926 ;
PI = 3.142 ; // Invalid
In addition to the simple types, arrays, structures, and unions can also be
defined as constant . In these cases, all their component elements are con-
stants. For example, in the string constant shown below, none of the individ -
ual elements can be changed.
const char str[ ] = "Hello" ;
Pointers and Constants

Pointers can also be defined as constants. Depending on how they are coded,
however, three different variations can occur.
.
1 The pointer itself is constant.
2. I he object being pointed to is constant.
3. Both the pointer and its object are constants.
Cose I: Pointer Constant

When the keyword const is associated with the identifier, that is, it is placed
after the type and before the identifier, the pointer itself is a constant. This
means that its contents cannot be changed after initialized. In other words,
the pointer can only point to the object that it was pointing to during initial-
ization; it cannot point to another object. In this case, it is an error to use the
pointer as an lvalue.
Appendix J Storage Classes ond ly Quolrfiers
1113
int a;
i n t b;
i n t * const p t r = &a;
ptr = & b;
/ / Error : p t r i s constant
Case II: Object Is Constant

When the keyword const is associated with the
type , that is, it is placed before
the type, then the object being referenced is a
constant , but the pointer itself
is not. So, while we can change the
value in the pointer (
dereference the operator as an lvalue. In other words, weaddress ), we cannot
cannot change the
value of the variable through this pointer. Consider the
following definitions:
i n t a = 5;
const i n t * p t r = & a;
* p t r = 2 1; / / Error : pointing to constant
Cose III: Both Pointer and Object Are Constant

lo indicate that both the pointer and the object that it points to are constant ,
use the keyword const twice.
i n t a = 5;
i n t b = 3;
const i n t * const p = &a;
*p = 5; / / Invalid : object of p i s a
constant
P * & b; / / Invalid : p i s a constant
Volatile
I he volatile qualifier tells the computer that an object value may be changed
by entities other than this program. Normally, objects used in a program
belong only to the C compiler. When the compiler owns an object, it can
store and handle it in any way necessary to optimize the program . As an
example, a C compiler may think that it is more efficient to remove an object
from HAM memory and put it in a register.
and
However, sometimes objects must be shared between your program
the C program , for example , some input /output
some other facilities outside
routines. To tell the compiler that the object is shared
, we declare it as type
that this object may be refer-
volatile. In this case, we are telling the compiler
enced or changed by other entities.
to an integer can be
The following shows how an integer or a pointer
declared volatile.
volatile int x;
volatile int* p t r;
1114 Section J .2 Type Qualifiers
Restricted
The restrict qualifier, which is used only with pointers, indicates that the
pointer is only the initial way to access the dereferenced data. It provides
more help to the compiler for optimization . Let us look at the following pro -
gram segment:
int* ptr ;
int a = 0;
ptr = & a ;
* ptr += 4 ;
* ptr += 5 ;
Here, the compiler cannot replace two statements * ptr + = 4 and * ptr
+= 5 by one statement * ptr + = 9 because it does not know il the variable a
can be accessed directly or through other pointers.
Now, let us look at the same program fragment using the restrict qualifier:
restrict int* ptr ;

int a = 0;
ptr = &a ;
* ptr += 4;
* ptr += 5 ;
I lere, the compiler can replace the two statements by one statement ptr
1
+ = 9 , because it is sure that variable a cannot be accessed through any other

resources.
As an application ot the restrict qualifier, let us look at two functions in
the C library that is used to copy bytes from one location to another:
void * memepy ( void* restrict destination ,

const void* restrict source , size t n);
void* memmove ( void* destination,

const void* source , size t n );
The memory copy function requires that the source and destination area
not overlap ; the memory move function does not . The restrict qualifier in the
first function guarantees that there is no overlap between the source and des-
tination area . The second function does not ; the user must be sure that there
is no overlap.
Appendix K
Program Development
In C hapter 2 , we briefly discussed how a source
file is changed to an execut-
able hie. In Appendix G we discussed the role of the
preprocessor in this pro-
cess. In this appendix, we show, with more details, how
a program , simple or
complex, is developed from one single source file, a source file and several
user header flies, or several source files, into one single executable program.
In the first section , we discuss the process of development , which is i com -
mon to all platforms. In the second section , we discuss how the process is
coded in some common platforms.
K. l Process
hirst , let us discuss the process, which creates a single executable program
from one or more source files. Although we use four different exhaustive
cases, the source program that we have chosen is a simple one. It is a simple
program that calculates
Y = 2X
where X is entered as input and is printed as output . In the first case, we

^
use just the main function . In the second , we use a function to do the calcu -
lation . In the third , we include the function as the header file. In the fourth ,
we use two separate source files to show the idea of separate compilations
.
First Case: A Simple File

seen outside an
The first case, as shown in Figure K- l , is trivial and seldom .
severa points
academic environment. Nevertheless it
, illustrates
1115
1116 Section K.l Process
Library Object Files

I N
scanf printf
întinclude <stdio.h> i main
<
>
main
// main
( void ’
Compiler Linker
K scant
printf
Source File User's Object File Executable File
FIGURE K - l Compilation and Linking of a Single Source File
We have only one source file that contains only one function, main. The
source file is first compiled to create an object file, the binary encoding ol
the main function. The linker adds two library functions, scanf and printf , to
the user ’s object file to create an executable file. The compiler instructs the
linker to do so when it encounters the function calls for these two functions.
I he scan ) and printj functions have already been compiled and stored as
.
object library modules in the standard library stdio h . No compilation is
needed lor these two functions. Note that the linker can only link ( add
together) object files. The executable file contains the binary code lor these
three functions.
Program K - l shows this situation. We have written only one main func -
tion to calculate the power of two of any unsigned integer.
PROGRAM K - l Our Simple Application Program

1 /* Demonstrate a simple compilation
2 Written by:
3 Date:
4 */
6 int main ( void )
7 {
9 unsigned int num;
10 unsigned int res ;
11
12 // Statements
13 printf ( " Enter the power:
14 scanf ( " %u" , &num);
15 res = 1 « num;
16 printf ( " %u\n" , res);
continue
\
PROGRAM K - 1 Our Simple Application Program (

continued )
17 return 0;
18 } // main
Results:
Enter the power: 14
16384
Second Case: Two Functions

In the second case> we stil1 have one source file, but two functions: main and
power 2. I he system flow is shown in Figure K- 2.

33
scanf IprintfI
5Fh
include <stdio.h>
^
int main (void]
{
main
} // main power2
main
unsigned power2 (...)
{
} // power2
Compiler
K power2
Linker scanf
printf
Source File User's Object File Executable File
FIGURE K - 2 Compilation and Linking of Source File Including a Function
We still have one user’s object file and two library module files
, but the
binary code for two functions: main and
user ’s object file now contains
Hie con-
power2. The linker still links only three object files. The executable
tains the code for four functions.
with two functions.
Program K - 2 contains the modified source program
PROGRAM K - 2 Source Code for main and power 2

functions.
1 /* Demonstrate compilation with two
2 Written by:
3 Date:
4 */
6
int exp);
8 unsigned int power2 (unsigned
9
10 int main ( void ) continued
PROGRAM K - 2 Source Code for main and pov / er 2 (continued )

11 {
13 unsigned int num ;
14 unsigned int res;
15
16 // Statements
17 printf ( " Enter the power: " );
18 scant ( " %u " , &num );
19 res = power2 ( num );
20 printf ( "%u\n " , res);
21 return 0;
22 } // main
23
24 unsigned int power2 ( unsigned int exp )
25 {
26 // Statements
27 return ( 1 << exp );
28 } // power 2
Results:
Enter the power: 14
16384
Third Case: Two Source Files

The third case is very common in writing structured programs, as you will see
in your data structures course. In this trivial situation, we can have one users
source file and one user 's header file. The two, however, are associated with
each other through the include directive in the source file. Figure K- 3 shows
the system How.

Source File
include <stdio .h >
^Pinclude "power 2. h"
printf
int main ( void]

(
\
main
} // main
power 2
unsigned power 2 (... J
{
] / / power 2
Compiler
h
main
power 2
Linker
H scanf
printf
Header File User’s Object File Executable File
FIGURE K - 3 Compilation and Linking of a Source File and a Header File

1\
Still there is one compilation . However, as we saw in Appendix G , the
preprocessor needs to act on the include directive and include the users
header file in the source file before translation . The rest of the process is the
same as the second case. One of the advantages of this case is that the
header file is an independent file that can be reused with any other program .
All we need to do to use it is to include its header file. Because the resulting
program is identical to the previous version , its results are also the same.
Program K- 3 shows the source file for main .
PROGRAM K - 3 Source Code for main Only

1 /* Demonstrate compilation with two source files.
2 Written by:
3 Date:
4 */
6 # include " power2.h "
7
8 int main (void )
9 {
1 3 / / Statements
printf (" Enter the power: ” )
;
14
15 scanf ( " %u " , &num );
16 res = power2 ( num );
17 printf ("%u\n" , res);
18 return 0 ;
1 9 } // main
totally independent Power 2 , h file.

Program K-4 contains the
2
PROGRAM K - 4 Source File for power power 2 function .
the
1 /* This is the
source file for
2 Written by:
3 Date:
4 */ unsigned int exp )
5 unsigned int power2 (
6 {
7 // statements
exp ) ;
8 return ( 1 «
9 > // power2
Case 4: Separate Compilation of Function

This case is more challenging. Not only do we want to make the functions
independent, hut we also want to compile them separately. Used primarily with
generalized software and large projects, it is known as separate compilations.
When a large project is designed, it is decomposed into modules that can
be developed and tested separately. It is much easier to write and debug a
small module than a large one. After each test file has been separately tested
and debugged, the object files can he linked together to form one executable
program. Figure K- 4 shows the (low for separate compilations.
Source File
#include <stdio.h>
#include "power2.h"
int main ( void! Compiler main
scanf
b
printf
.
<
} // main zr main
unsigned power 2 (...

I powers
<
} // power2
Compiler
j-^lpowerfj Linker |-^ scanf
printf
Header File User ' s Object File Executable File
FIGURE K - 4 Compilation and Linking of Two Independent Source Files
Although the resulting executable file has the same four binary-coded
modules, the process is different. Each source file is compiled separately and
two independent object files are made from them. The object files are only
related when they are linked together. Once again, because the code is the
same the output is identical.
Program K- 5 contains the source file for main .
PROGRAM K - 5 Separate Compilations — main

1 /* Demonstrate the separate compilations main —
2 Written by:
3 Date:
4 */
5 ffinclude <stdio.h >
6 int main ( void )
7 {
11
12 // Statements
continue*
Appendix K Progrom Development
1121
PROGRAM K -5 Separate Compilations
13
—
printf ( "Enter the
main (continued)
power: ");
14 scant ("%u", &num);
15 res = power2 (num);
16 printf ( "%u\n", res);
17 return 0;
18 > // main
Program k-6 contains the source file for

power2 .
PROGRAM K -6 Separate Compilations
—
1 /* Demonstrate the separate
power2
2 Written by:
compilations power2 -
3 Date:
4 */
5 unsigned int power2 (unsigned
int exp)
6 {
7 // Statements
8 return ( 1 « exp);
9 > // power2
K.2 Coding
Now that we have discussed the process, let 's look at how we code the differ-
ent steps in the process.
UNIX
UNIX, and its derivative Linux, are popular operating systems. They contain
several tools with many options. In this appendix, we briefly describe the
basic options and explain how to compile and link UNIX programs.
File Types
UNIX uses different extensions to distinguish between different file types
source files, object files, and executable files, as shown in Figure K- 5.
myprogram.c myprogram.o myprogram
Source File Object File Executable File
FIGURE K -5 File types in UNIX

onK .2 Coding
In Figure K - 5, the source file is myprogram. c, the object file is

myprogram. o and the executable lilt* is myprogram. In other words, a source
. .
file has the extension “ c ”, an object file has the extension ‘ o , and an exe-
'
cutable file lias no extension.
.
In UNIX, a source file in Standard C has the extension c, an object file has the
extension o, and an executable file has no extension.
.
Command and Options

UNIX uses onlyone command, c 9 9, to compile or link different files. The
command is used with different options to define compilation only, linking
only, or compilation and linking. Figure K- 6 shows the general format of the
command.
Executable Source or Object Files

Prompt C99 Option File
$ or % or ... -c or -o
FIGURE K - 6 Compilation and Linking Format in UNIX for C99
Note that the U N I X prompt depends on the shell being running. It can
he $ or % or any other customized prompt. The command c 99 invokes the C
compiler with the C99 standard implementation ( other versions are usually
available lor backward compatibility). The option can he either -c for compil-
ing only, or - o lor either linking or compiling and linking. The - o option indi -
cates that we need an executable module. The intent of the -o option is
conveyed through the parameters:
1 . II we want to compile only, we define only source files; the executable file
name is not used.
.
2 II we want to link only, we define only the object files; the source files are
not used.
Examples
Let ’s look at some examples.
EXAMPLE K - l Compile, Link, and Execute: We need to compile and link to create an executable file
Irom a source file named myprogram. c. The following code shows the com -
mands. In the first line, we create an executable file named myprogram; in
the second line we run that program. After these two lines of commands
have been executed, we have two files in our directory: myprogram. c and
myprogram.
Appendix K Program Development ] ] 23
$ c 99 o myprogram myprogram
$ myprogram
c .
EXAMPLE K - 2 Compile Jwo Source Files: We need only to

compile two files called first , c and
second . c. When the compiler
completes, and assuming no errors , we
four files in our directory: first , c, first . ( have
o object file), second . c, and
second o. .
$ c 99 -c first . c second . c
Of course, we could use two commands instead of one

and compile each
file separately.
EXAMPLE K - 3 Create Fxecutoble File: We need only to create an executable file from the first .o
file ( linking all library' files to it ). After the linker executes , we have a execut-
able new file in our directory, first .
$ c 99 -o first first o .
EXAMPLE K - 4 Create executable

from Jwo Object Files: We need to link two object files, first .o and
second o, .
together to create an executable file firstSecond. The following
shows the command .
$ c 99 -o firstSecond first o second o . .
EXAMPLE K -5 Separate Compilation: In separate compilations, we separately compile the source

files myprogram . c and power 2 .c and link them to create an executable file
mybigprogram. We then run the executable file.
$ c 99 -c myprogram . c
$ c 99 -c power 2 .c
$ c 99 -o mybigprogram myprogram . o power 2 . o
$ mybigprogram
that does not

MPLE K -6 Compile and execute Short Cut: Sometimes, programmers use a short cutexecutable file,
the name of the
save the executable file. If we do not define
, the following command
the system creates a file called a . out . For example .
file named temporary c.
shows how we can compile, link, and run a source
$ c 99 temporary . c
.
$ a out
1124 Section K.2 Coding
Note that because the name of the file is system defined , it remains in
the system until we created another executable file, which the system needs
to call a . out . This method is used only for temporary testing and is not rec-
ommended when we need to keep the executable file.
Other Platforms
In general, every hardware platform comes with its own application develop-
ment environment . In addition , there are numerous compilers and develop -
ment support tools available. To find the documentation or your local
(
support environment , ask your support staff or check with the help system
installed with your compiler.
Appendix L
Understanding Complex
Dedarations
I he declarations have become increasingly more complicated throughout
this text. Sometimes the declarations are difficult to interpret, even for some-
one experienced in the C language.
To help you read and understand complicated declarations, we provide a
rule that we call the right -left rule. Using this rule to interpret a declaration,
you start with the identifier in the center of a declaration and read the decla-
ration by alternatively going right and then left until you have read all enti-
ties. Figure L- 1 is a representation of the basic concept.
[ identifier ] 0
t t t
6 4 2
t
start here
t t t
1 3 5
FIGURE L- 1 Right-Left Rule Concept
We will begin with some simple examples and proceed to the more
complicated .
1 . Consider the simple declaration
int x
Section
This is read as " x is # an integer.” i
int x #
T T T
2 o 1
Since there is nothing on the right, we simply go leit .

2. Now consider the example of a pointer declaration. This example is read
as “p is # a pointer # to integer. ”
int P # #
T T T T T
4 2 0 1 3
Note that we keep going right even when there is nothing there until all
the entities on the left have been exhausted.
.3. In the next example, we have an equal number ol entities on the right
and the left.
int table [4]

T T t
2 o l
This declaration is read as “ table is an array of 4 integers.”

4. Regardless of how many dimensions are in an array, it is considered as
one element in the rule. Therefore, given the following declaration ol a
multidimensional array:
int table f 41 T 5 1
T T T
2 0 1
it is read as “ table is a [ 4 ][ 5 ] array of integers.”

5 . I he next example is quite difficult and is often misread In this declara - .
.
tion, we have an array of pointers to integers The structure is seen in
Figure L- 2a.
int * aryOfPtrs [5] #

T T T T T
4 2 o 1 3
It is read as “aryOfPtrs is an array of 5 pointers to # integer. ”
I. The pound sign ( # ) is just a placeholder to show that there is no entity to he considered. It is
ignored when read.
Appendix L Understanding Complex Declarations 112 /
ptrToAry
aryOfPtrs
(a) An array of pointers (b) A pointer to an array
FIGURE L - 2 Array of Pointers Versus Pointer to Array
6 . By using parentheses , we change the previous example to a pointer to an

array of live integers . In this case , the pointer is to the whole array,
not
just one element in it . ( See Figure L- 2 b . )
int ( *ptrToAry # ) [5 ]
T T T t T
4 2 0 1 3
to an array of 5 integers.
This declaration is read "ptrToAry is i a pointer
. Here , we see a simple
7 . This example deals with function declarations
an integer .
prototype for a function that returns
int dolt (...)

T T T
2 0 1
returning an integer .’
“ is ai function
This declaration is read as dolt .
returning a pointer to an integer
8 . The final example shows a Lfunction
int * dolt (int) #

T T T t T
4 2 0 1 3
integer.
a pointer to an *
a function returning
This example is read
"dolt is
I!
GIOSSOP
A application software: computer software devel-
oped to support a specific user requirement. Con -
absolute value: the magnitude of a number regard - trast with system software .
less of its sign . In C: abs. application -specific software: any application
abstract data type ( ADT ): a data declaration and software that can be used for only one purpose, sue
a set of operators that are allowed on the data , encap- as an accounting system .
sulated as a type. arc: a directed line in a graph ; contrast with edge .
accuracy: the quality factor that addresses the cor-
array: a seq uenced collection of elements of the
rectness of a system .
actual parameter: the parameter in the function
Stime
array
ôf
^ e
ala l
pointers
'
: an array of addresses often us

,
calling statement that contains the values to be passed
to represent a sparse array.
to the function . Contrast with formal parameter.
sequence: a list order in which eac
additive expression: in C , the binary addition and ascending list has a key greater than or equal
subtraction expressions. element in the
its predecessors. Contrast with
descending sequel
address list: in C : function parameter list that con - Code for Informat
sists ol variable addresses. ASCII: the American Standard defines con
Interchange. An encoding scheme that
address operator : in C : the ampersand ( & ). for the first 128
characters and graphic characters in Unicode.
afferent module: a module whose processing is first 128 characters
ues in a byte . Ihe
directed toward the central transform ; that is, a mod - assembler: system software that converts a soui
ule that gathers data to be transmitted toward the program into executable object code
; traditionally
central processing functions ol a module . program . See ;
associated with assembly language
algorithm : the logical steps necessary to solve a compiler.
problem in a computer; a function or a part of a assembly language: a programming
language ii
function . - correspondence
which there is a one- machine language and t
for one
I .* ,m
. .
lc rm crealecJ by Brassard and
rY S: between the computers
B 1 N 1 ,at re'ers the study of techniques used to symbolic instruction set of the language
taU c‘* hcient algorithms.
.
an expression contain
with assignment expression(: ) that results in the
flag used
i natc Hag: the data formatting the assignment operatorplaced into the left open
*
r C n8lnccr ng> hexadecimal, and
*
octal conversion of the expression, being ,. . > .
,
V
c C . ' e a t i n g that an alternative presentation . . . : the parsing direction used to eval•
.
n U 'd be' used . operators have an equal

associativity
° an expression when all right -to-left associativity.
ANSI / ISO C: the standard for the C language --
ity. See left to right
and
adopted by the American National Standards also precedence. be meaningf
Institute. atomic data: data that cannot
append mode: in file processing, the mode that
subdivided.
adds to the end ol a file .
1130 Glossary
atomic data type : a set of atomic data having iden - bitwise operator: any of the set of operators that
tical properties that distinguish one type Irom operate on individual hits.
another. Contrast with derived type . blackbox testing: testing based on the system
( into : the default storage class for a local variable . requirements rather than a knowledge ol the
auxiliary storage : any storage device outside main program . Contrast with whitebox testing .
memory; permanent data storage ; external storage. block: in C , a group of statements enclosed in
braces {...}.
B block comment: a comment that can span one or
more lines. Contrast with line comment .
base case: in recursion , the event that does not block scope: see scope .
involve a recursive call and therefore ends the recur-
sion . body ol loop: the code executed during each itera -
’
tion of a loop.
batch update: an update process in which transac -
tions are gathered over time for processing as a unit . Boolean: a variable or expression that can assume
Contrast with online update. only the values true or false .
bi - directional communication : communication braces: the { and > symbols.
between a calling and called function in which data breadth-first traversal : a graph traversal in which
flows in both direction , that is , from the calling func - nodes adjacent to the current node (siblings) are pro-
tion to the called function and then from the called cessed before their descendents. Contrast with
function to the calling function . depth - first traversed .
big- O analysis: the analysis of an algorithm that break: a C statement that causes a switch or loop
measures efficiency using big- O notation rather than
statement to terminate .
mathematical precision .
bubble sort : a sort algorithm in which each pass
big- O notation: a measure of the efficiency ol an
algorithm in which only the dominant factor is through the data moves ( bubbles ) the lowest element
to the beginning of the unsorted portion of the list .
considered .
binary: a numbering system with base 2. Its digits buffer: ( 1 ) hardware , usually memory, used to syn -
are 0 1 .
chronize the transfer of data to and from main mem -
ory. ( 2 ) memory used to hold data that have been
binary expression : any expression containing one read before they are processed , or data that are wait -
operator and two operands. ing to he written .
binary file : a collection of data stored in the inter- buffered input /output : input or output that are
nal format ol the computer. Contrast with text file .
temporarily stored in intermediate memory while
binary search : a search algorithm in which the being read or written.
search value is located by repeatedly dividing the list
in half .
buffered stream: C term for buffered input/output .
binary search tree : A binary tree in which : ( 1 ) the bug: a colloquial term used for any error in a piece
keys in the left subtree are all less than the root ol software.
key. ( 2 ) the keys in the right subtree are greater than byte: a unit of storage , 8 bits.
or equal to the root key, and ( 3 ) the subtrees are all
binary search trees.
C
binary stream: a stream in which data is repre -
sented as it is stored in the memory of a computer. C standard library: any of the libraries that con -
binary tree : a tree in which no node can have more tain predefined algorithms delivered with the system,
than two children . such as standard library, ( stdlib . h ) and standard
bit : acronym for Binary digIT. The basic storage unit input and output ( stdio . h ) .
in a computer with the capability of storing only the called function : in a function call , the function to
values 0 or 1 . which control is passed .
Glossory ] ) 31
calling function: in a function call , the
that invokes the call. function cohesion: the attribute of a module
how closely the processes within a that describes
cast: a C operator that changes the type of related to each other. module are
an
expression .
coincidental cohesion: the cohesion level in
ceiling: the smallest integral value that is larger which totally unrelated processes
than a floating- point value. a module. are combined into
central processing unit (CPU ): the part of column: in a two-dimensional table, the
computer that contains the control components
a
that is, the part that interprets instructions. In a
sonal computer, a microchip containing a controlper-
— representing vertical columns in the array.
withrow.
dimension
Contrast
comma expression: the last ( lowest precedence)
unit and an arithmetic-logical unit .
expression in C; used to combine multiple
expres-
central transform: the modules ol a program sions into one expression statement.
that
take input and convert it to output . See also afferent comma operator: in C, the operator that connects
module and efferent module. multiple statements in one expression.
changeability: the quality factor that addresses the command-line argument: in C , a parameter spec-
ease vvith which changes can be accurately Red on the run - time execute statement and passed
made to a
program. ‘
to the program for its use.
character: a member of the set of values that are comment: in a C program , a note to the program
used to represent data or control operations. reader that is ignored by the compiler.
See Unicode .
coniniunicational cohesion: the processes in a
character constant: a constant that defines one module are related only in that they share the
character enclosed in two single quotes same data .
character set: the set of values in a computer s ’ compilation: the process used to convert a source
alphabet . See Unicode . program into machine language.
child: a node in a tree or a graph that has a compiler: system software that converts a source
predecessor. program into executable object code; traditionally
associated with high-level languages. See also assembler.
chronological list : a list that is organized by time;
that is, in which the data are stored in the order in complement: in a set of values with two symmetric
which they were received . See also FIFO and LIFO. sections, the operation that gives the counterpart of
the current value, that is, the value in the other sec-
classifying function: a standard C character tion . For example, the complement of a number is a
function that classifies a character according to U s number with the same value and the sign reversed.
character taxonomy, such as printable or
“alphanumeric ” I he complement of a binary digit is the other binary
. digit. The complement of a logical value ( true or false )
't lear box
testing: see u hitebox testing . is the other value.
client: in a client/server network, the computer that com plex type: a data type consisting ol a real and-
provides the basic application computing; the com - an imaginary type and representing a complex num
puter residing in the user’s physical area. ber in mathematics.
client/server: a computer system design in which composite data: data that are built on other data
down into
two separate computers control the processing o t e structures; that is, data that can be broken
application , one providing the basic application com discrete atomic elements .
puting (the client) and the other providing auxi ,ar compound statement: a sequence of statements
services, such as database access ( the server .
> enclosed in braces. See also block .
compo-
close function: the function that concludes the computer hardware: the set of computer
consisting of at least an input device , an out -
writing of a file by writing any pending data to u nents,
and then making it unavailable lor processing. put device, a monitor, and a central
processing unit .
1132 Glossary
computer language: any ot the syntactical lan - coupling: a measure of the interdependence
guages used to write programs for computers , such between two separate functions. See also: content
as machine language , assembly language , C , cou fling , control coupling , data coupling , global
COBOL, and FORTRAN . coupling , and stamp coupling .
computer software : the set of programs, consist - CPU: see central processing unit.
ing of system and application programs, that provide
instructions for the computer. D
computer system: the set of computer hardware
and software that make it possible to use a computer. dangling else: a code sequence in which there is
no else statement for one of the if statements in a
conditional expression: in C , shorthand lorm of a
two-way selection .
nested if
connected graph : a graph is connected if , when data : the part of a structure that stores application
direction is suppressed , there is a path from any ver- information .
tex to any other vertex. data coupling: communication between modules
in which only the required data are passed . Consid -
const : a storage class that designates that a fields
contents cannot he changed during the execution of
ered the best form of coupling.
the program . data encapsulation: the software engineering con -
constant: a data value that cannot change during cept in which data are hidden from an application
the execution of the program . Contrast with variable . program’s access.
content coupling: the direct reference to the data data flow': in a structure chart , an indicator identi -
in one module by statements in another module. 1 he fying data input to or output from a module.
lowest lorm of coupling and one to be avoided . data hiding: the principle ol structured program -
continue : in C , a statement that causes the remain - ming in which data are available to a
function only il
it needs them to complete its processing; data not
ing code in a loop iteration to be skipped .
needed are hidden ’ from view. See also encapsida -
"
control character: a nonprintable character value tion and scope .

whose 1 unction is to perform some operation , such as
form -feed , or that is used to indicate status , such as data name: an identifier given to data in a program .
the start of a transmission . data structure: the syntactical representation ol
control coupling: communication between I unc - data organized to show the relationship among the
tions in which flags arc set by one module to control individual elements.
the actions of another. data type: a named set of values and operations
conversion code : in formatted input and output, defined to manipulate them , such as character and
the code in the format specification that identifies integer.
the data type . data validation : the process of verifying and vali -
conversion specification : the specification that dating data read from an external source .
defines how data are be
to represented in an input or declaration: in C , the association of a name w ith
output operation . an object , such as a type , variable, structure , or func -
converting function : a C character function that tion . See also definition .
converts lowercase alphabetic characters to uppercase, declaration section : the portion of a block in
or vice versa . which variables and constants may be defined . The
correctability: the quality factor that addresses the first section in a block. See also statement section .
ease with which errors in a module can he fixed . default: in C , the entry point to the code that is to
counter- controlled loop: a looping technique in be executed il none of the case values match the
which the number of iterations is controlled by a switch expression .
count ; in C , the for statement. Contrast with define : in C , a preprocessor command that names
event -controlled loop . and provides the code for a macro.
Glossary 1133
definition : in C , the process that reserves memory do...while loop: a sentinel -controlled,
for a named object , such as a variable or constant . post - test
loop in C.
See also declaration .
degree: the number ol lines incident to a node in double: the C type lor double- precision floating
point type.
a graph .
delimited string: a string terminated by a nondata downward communication: data flow from the
character, such as the null character in C . calling function toward the called function .
delimiter: the character or token that identifies the drive: an auxiliary storage device that can write and
end of a structure. read data , such as the internal hard disk, a floppy
De Morgan’s rule: a rule used to complement a disk, or a tape unit .
logical expression . dynamic allocation: allocation of memory for stor-
demotion: an implicit type conversion in which the ing data during the execution of a program. Contrast
rank of an expression is temporarily reduced to with static memory allocation .
match other elements ol the expression .
dynamic array: an array that has been allocated in
dependent statement : a program statement , such the heap during the execution of the program .
as the true action in an ij statement , whose execution
is controlled by another statement . dynamic memory: memory whose use can change
depth -first traversal: a traversal in which all ol a during the execution of the program (The heap ).
node’s descendents are processed before any adjacent
nodes (siblings). Contrast with breadth -first traversal . E
dequeue: delete an element from a queue.
dereference: access of a data variable through a edit set: a list of characters used to validate charac -
pointer containing its address. ters during the formatted reading of a string.
dereference operator: in C , the asterisk ( * ) . Used efferent module: a module whose processing is
to indicate that a pointer’s contents ( an address) are directed away from the central transform ; that is, a
to be used to access another piece ol data . module that predominately disposes of data by report-
derived type: a composite data type constructed ing or writing to a file.
from other types ( array, structure, union , pointer, and efficiency: the quality factor that addresses the
enumerated type). Contrast with atomic type . optimum use of computer hardware or responsive-
descendant: any node in a path from the current ness to a user.
node to a leaf . edge: a line in an undirected graph. Contrast
descending sequence: a list order in which each with arc.
element in a list has a key less than or equal to its
predecessor. else - if: a style (as opposed to syntax) convention nonin -
used to implement a multiway selection fora
directed graph : a graph in which direction is indi - tegral expression. Each expression in the series must
cated on the lines (arcs ). evaluate the same variable .
disk: an auxiliary storage medium used to store data but that
and programs required for a computer. empty list: .a. list that has been allocated list
contains no data . Also known as a null .
disk drive: the auxiliary storage hardware device design

used to read and write a disk . encapsulation: the software engineering are bun -
concept in which data and their
operations
diskette: a removable flexible disk, enclosed in a dled together and maintained
separately from the
application using them .
protective flexible or rigid cover, used to store data
lor a personal computer.
end of file: the condition after that occurs when a read
distributed environment: an environment that read it has processed the
provides a seamless integration ol computing IUIK - operation attempts to
tions between different servers and clients. last piece of data .
1134 Glossary
enumeration constant: the identifier associated extern : the storage class that specifies that an object
with a value in an enumeration declaration . has static extent and that its name is exportable to
enqueue: insert an element into a queue. the linker.
EOF: end of file. In C , a flag set to indicate that a
file is at the end . F
error report file : in a file update process , a report factorial efficiency: a measure of the efficiency of
of errors detected during the update. a module in which the run time is proportionate to
error state : one of three states that an open file the number of elements factorial .
may assume . An error state occurs when a program false: the Boolean value used to indicate that an
issues an impossible command , such as a read com -
expression is not true.
mand while the file is in a write state , or when a
physical device failure occurs. See also read state , fan out: an attribute of a module that describes the
write state . number of submodules it calls.
error stream : in C , the standard file used to display field : the smallest named unit of data that has
errors, often assigned to a printer ( stderr ). meaning in describing information . A field may be
escape character: in C , the backslash ( \ ) charac -
either a variable or a constant .
ter used to identify a special interpretation lor the field specification: a subcomponent of the format
character that follows it . string used to describe the formatting of data , in the
escape sequence : any group ol keystrokes or char- formatted input and output functions.
acters used to identify a control character. In C , the field width: in a field specification , the specifica -
backslash ( \ ) used to indicate that the character that tion of the maximum input width or minimum out -
follows represents a control character. put width lor formatted data .
event-controlled loop: a loop whose termination FIFO: first in - first out .
is predicated upon the occurrence ol a specified file : a named collection ol data stored on an auxil -
event . Contrast with counter-controlled loop . iary storage device . Compare with list .
exchanging values: the logic used to exchange the file mode : a designation of a file’s input and /or out -
values in two variables. put capability; files may be opened for reading, writ -
exclusive or : a binary' logical operation in which the ing, appending, or updating.
result is true only when one of the operands is true file state : the operating condition of a file: read
and the other is false . state , write , or error state .
executable file : a file that contains program code in file table: in C , the predefined standard structure ,
its executable form ; the result ol linking the source
FILE used to store the attributes of a file.
code object module with any required library modules.
,
explicit type conversion: the conversion of a filename: the operating system name ol a file on an
value from one type to another through the cast oper- auxiliary' storage device.
ator. Contrast w ith implicit type conversion . first in - first out: a data structure processing
exponential efficiency: a category of program / sequence in which data are processed in the order
module efficiency in which the run time is a function that they are received ; a queue .
of the power of the number of elements being fixed - length array: an array in which the num -
processed . ber of elements are predetermined during the
expression: a sequence of operators and operands declaration /definition .
that reduces to a single value. fixed- length string: a string whose size is constant
expression statement: an expression terminated regardless of the number of characters stored in it .
by a semicolon . Hag: an indicator used in a program to designate the
expression type: one of six expression attributes presence or absence of a condition; switch .
that broadly describes an expression s format: flag modifier: the input/output format specification
primary, postfix, prefix , unar)> binary, and ternary.
} that suppresses input or modifies output formatting.
Glossary 1135
flexibility: the quality factor that addresses the ease
with which a program can he changed to meet user function declaration: in C, statement
describes a functions return type that
requirements. , name, and formal
parameters.
flipping hits: changing a bit from a 0 to 1 or a 1 to 0.
function definition: in C, the implementation
-
floating point number: a number that contains
both an integral and a fraction .
function declaration. of a
function header: in a function definition , that
floor: the largest integral number smaller tha n a of the function that supplies the return part
floating- point value. type, func-
tion identifier, and formal parameters
with Junction body.
. Contrast
flowchart: a program design tool in which standard
graphical symbols are used to represent the logical functional cohesion: a module in which all of the
flow of data through a function . processing is related to a single task. The highest
for loop: a counter-controlled , pretest loop in C. level of cohesion.
force to change: the bit manipuation concept used
to force selected hits to change from 0 to I or from
1 G
to 0.
force to one: the bit manipulation concept used to general case: all the code in a recursive function
force selected bits to 1 . that is not a base case.
force to zero: the bit manipulation concept used to general - purpose software: software, such as a
force selected hits to zero. spreadsheet, that can be used in multiple
applications.
formal parameter: the parameter declaration m a
function prototype used to describe the type ol data global coupling: communication between differ-
to be processed by the function . Contrast with ent modules that uses data accessible to all modules
actual parameter. in a program. Considered to be a very poor communi-
cations technique for intraprogram communication .
formal parameter list: the declaration of |wra mo-
ters in a function header. global declaration section: the declaration and/or
definition of a variable or function outside the
fornnt C n r . s r n8: ln a formatted input /output
fun t i ' ii . °| ° *
i u siting ’that is used for formatting data.
—
boundaries of any function that is, before main or
between function definitions. Contrast with local
formatted input/output: in C , any of the standard declaration .
library functions that can reformat data to and from global variable: a variable defined in the global
ext bile they are being read or written .
"
frequency array: an array that contains the num - outside a function block.
—
declaration section of a program that is , defined
H 1 o occur rences ol a value or of a range of values.

graph: a non -linear list in which each element can
êe also histogram . have zero, one, or more predecessors and zero, one,
front: when used to refer to a list , a pointer or index or more successors.
that identifies the first element.
function : a named block of code that performs a H
process within a program; an executable unit o
consisting of a header, function name, and a hot \ ,
c'
. i
copv: any computer outpu
m;crofiche.
that is designed to perform a task within the program . or ot|ier readable mediums sue
.
—
Q) ntrast with soft copy
function body: the part ol a 1 unction that contains , j are; any ol the physical coml
the local definitions and statements; all of a funct on
hajrdwar
StlSSTw. -
^ printer,
he keyboard or a Pr
(
except the header declaration. Contrast with
, ration: that part of a function ia
(
function call: an expression that invok es t u exccu

tion of a function. pararneter declarations ,
. * *
Glossary
ider file : in C , a file consisting ot prototype infinite loop: a loop that does not terminate.
ements and other declarations and placed in a information hiding: a structured programming
ary' for shared use . concept in which the data structure and the imple-
id pointer: a pointer that identifies the first cle- mentation of its operations are not known hy the user.
at of a list .
initialization: the process of assigning values to a
ip: see heap memory. variable at the beginning of a program or a function .
ip memory: a pool ot memory that can he used
lynamically allocate space for data while the pro- initializer: in C , any value assigned to a constant or
111 is running . variable when it is defined.
tadecimal: a numbering svstem with base 16. Its inorder traversal : a binary tree traversal in which
its are 0 1 2 3 4 5 6 7 8 9 A B C D E F. the root is processed after the left subtree and before
;h - level language : a ( portable ) programming the right subtree. Contrast with preorder and
guage designed to allow the programmer to con - postorder traversal .
ltrate on the application rather than the structure input device: a device that provides data to he read
1 particular computer or operating system . by a program .
itogram : 1 graphical representation of a fre -
<
input stream : in C , the How of data from a hie to
ency distribution . See also frequency array. the program.
inquiry: a request lor information from a program .
I
inquiry all : an algorithm that determines if all data
jntifier: the name of an object . In C , identifiers items in a list meet some criteria .
a consist only of digits , letters, and the underscore.
inquiry any: an algorithm that determines il any
.. else : a decision statement in C . data item in a list meet some criteria .
laginary type : the data type in C that can repre-
nt and imaginary number in mathematics . insertion sort : a sort algorithm in which the first
element from the unsorted portion of the list is
iplicit type conversion: the automatic conversion inserted into its proper position relative to the data in
data from one type to another when required within
the sorted portion of the list .
2 program. Contrast with explicit type conversion .
elude : in C , a preprocessor command that speci - integer: an integral number; a number without a
es a library file to be inserted into the program . fractional part .
degree : in a tree or graph , the number ol lines integral type : a data type that can store only whole
itering a node. numbers.
dentation: a coding style in which statements intelligent data name : a software engineering
•pendent on a previous statement , such as / / or
principle that requires the identifier to convey the
Wile, are coded in an indented block to show their meaning or use of an object .
lationship to the controlling statement .
idex: the address of an element within an array. interoperability: the quality factor that addresses the
IQ also subscript .
ability of one system to exchange data with another.
id ex range checking: a feature available in some iteration : a single execution of the statements in
impilers that inserts code to ensure that all index a loop .
ferences are within the array.
idexing: a process used to identify one element in
J
l array.
idirect pointer: a pointer that locates the address justification : an output formatting parameter that
data through one or more other pointers ; pointer to controls the placement of a value when it is shorter
ainter. than the specified output width .
Glossary 1137
K linear loop: a loop whose execution is a function of

the number of elements being processed . See also
key: one or more fields that are used to identify a linear efficiency.
record (structure). line comment: a comment that spans only to the
keyboard: an input device used for text or control end of the current line. Contrast with block comment .
data , that consists of alphanumeric keys and linear search : see sequential search .
function keys.
link: in a list structure, the field that identifies the
-
key sequence list: a list in which the data items next element in the list .
arc ordered based on the value of a key.
linked list: a linear list structure in which the
keywords: see resen'ed words . ordering of the elements is determined by link fields.
KISS: in this text : Keep It Short and Simple. linked list traversal: processing in which every
element of a linked list is processed in order.
L linker: the program creation process in which an
object module is joined with precompiled functions
-
last in first out: a data structure processing to form an executable program.
sequence in which data are processed in the reverse list: an ordered set of data contained in main mem-
order that are received; a stack . on'. Compare with file .
Latin character set : theextended ASCII character list traversal : any logic that visits and processes all
set in Unicode . the data in a list .
leading zero Hag: the flag in the format string of a literal: an unnamed constant coded in an expression .
print statement indicating that numeric data are to loader: the operating system function that fetches
he printed with leading zeros. an exec utable program into memory for running
.
leading zero padding adding extra zeros at the

:
local declaration: a variable or type declaration
-
beginning of a data item . that is only visible to the block in which it is con
left justification: the orientation ol variable - length tained . Contrast with global declaration .
data in an output format such that trailing null values variaye; a variable defined with a block .
, ,
are inserted and the first data character is at the left rr
„ , ,„„ ,
e d or 1 0 p Com , ,
wilh
-
left to-right associativity: the evaluation ol an tionatc to the log of the number ol
elements being
expression that parses from the left to the right . Con - processeJ.
trast with right -to-left associativity. efficiency is a
logarithmic loop: a loop whose elements being
length -controlled string: a variable - length string log of the number of
function ol the
lunction in which the data are identified by a stiuc -
processed . See also logarithmic efficienc ’).
tural component containing the length ol the data . that describes
logical cohesion: a design attribute within the mod-
lexicographical: a textual data order based on the a module in which the processingtype ol processing
dictionarv. general
ule is related only by theunacceptable design in
being done . Considered
-
LIFO: last in first out .
structured programming ,
limit test expression: in a loop, the expression athat logical data: data whose values
can be only true or
determines if the loop will continue or stop . Sec so
false. See Boolean.
ternii nati ng condit ion . ( & & , ||,!) used in
of the efficiency of a logical operator:: a C operator
linear efficiency: a measure the a logical expression .
module in which the run time is proportionate to to repeat one or more
number of elements being processed. loop: the construct used while , for , and do...
program . The
, statements in a
linear list: a list structure in which each element while statements in C.
except the last , has a unique successor.
1 Glossary
p control expression : the expression in a loop nested if statement: an // statement coded as

ement that is used to determine if the body of the either the true or false statement within another if .
3 is to he executed . nested loop: a loop contained within another loop.
p update: the code within a loop statement or nested structure: a structure that contains other
ly that changes the environment such that the structures.
p will eventually terminate. node: in a data structure, an element that contains
lue : an expression that allows the contents oi a both data and structural elements used to process
iahle to he modified . the list .
nonlinear list : a list in which each element can
M have more than one successor.
null else: the absence of a false statement in an if
ichine language: the instructions native to the statement .
Ural processor of a computer and that are execut -
e without assembly or compilation .
tin memory: see memory. 0
isk: a variable or constant that contains a bit con - object module : the output of a compilation con -
uration used to control the setting ol hits in a bit - sisting of machine language instructions.
se operation .
octal : a numbering system with a base of 8 . The
aster file: a permanent file that contains the most octal digits are 0 I 2 3 4 5 6 7 .
rrent data regarding an application .
offset: in pointer arithmetic , a computed value that
emory : the main memory ol a computer consist - is used to determine another index location within
ed random access memory' ( HAM ) and read -only an array.
emory' ( ROM ) ; used to store data and program
structions. one -dimensional array: an array with only one
level of indexing.
emory constant: a C type defined using the type
lalilier const . one ’s complement : the bitwise operator that
reverses the value of the hits in a variable .
erge: to combine two or more sequential files into
le sequential file based on a common key and online update: an update process in which trans -
ructure format . actions are entered and processed by a user who has
letadata : data about a list or other data structure
direct access to the system . Contrast with hatch
update.
ored within the data structure itself .
lonitor: the visual display unit of a computer sys -
open function : the function that locates and pre -
pares a file for processing.
*m . usually a video display device .
lultidimensional array: an array whose elements operability: the quality factor that addresses the
ease with which a system can be used .
jnsist of one or more arrays.
uiltiplicative expression : an expression that con - operand: an object in a statement on which an
uns a multiply, divide, or modulus operator. operation is performed . Contrast with operator.
lultiway selection : a selection statement that is operating system: the software that controls the
apable of evaluating more than two alternatives. In computing environment and provides an interlace to
>, the switch statement. Contrast with two-way the user.
ilection . operator: the syntactical token representing an
action on data ( the operand ). Contrast with operand .
N ordered list: a list in which the elements are
arranged so that the key values are placed in ascend -
latural language: any spoken language . ing or descending sequence.
legative logic: an expression that begins with the outdegree: the number of lines leaving a node in a
legation operator ( ! ) . tree or a graph.
Glossary 1139
output device: a device that can be written but
not read . portability: the quality factor that addresses
ease with which a system can he moved the
output stream: in C, the flow of data fre> m a pro- to other
hardware environments.
gram to the file . postfix decrement: in C, the operator (
such a--)
overflow: the condition that results when an that subtracts 1 from a variable after its valueashas
attempt is made to insert data into a structure and been used in an expression.
there is no room . postfix expression: an expression in which the
operator follows the operand.
P postfix increment: in C , the operator (such as
a + + ) that adds I to a variable after its value has
been
padding: extra zeros or spaces added to the left or used in an expression.
right ol a data item . post-test loop: a loop in which the terminating
parameter: a value passed to a function . condition is tested only after the execution of the
parameter list: a list of values passed to a function .
loop statements. Contrast with pretest loop.
( Note: 1 he values may be data or addresses. ) precedence: the priority assigned to an operator or
group of operators that determines the order in
parent: a tree or graph node with an outdegree
which operators will he evaluated in an expression .
greater than 0; that is, with successor.
See also associativity.
pass by address: in C , the technique used when a precision: in the format string of a print function ,
pointer is passed to a function as a parameter. the maximum number of integral digits , the number
pass by reference: a parameter passing technique ol significant digits or fractional digits in a floating-
inwhich the called function refers to a passed point number, or the maximum number of characters
parameter using an alias name. in a string.
personal computer ( PC ): a computer designed precision modifier: the print format string modi-
lor individual use. fier that specifies the number of decimal places to be
printed in a floating- point value.
plane: the third dimension in a multidimensional
array. prefix decrement: in C. the operator (such as a )
that subtracts one from a variable before its value is
—
pointer: a constant or variable that contains an used in an expression. Also known as unary decrement.
address that can be used to access data. prefix expression: an expression in which the oper-
pointer arithmetic: addition or subtraction in ator precedes the operand.
which a pointers contents ( an address ) are changed prefix increment: in C , the operator (such as ++a )
by a multiple of the size ol the data to which it is that adds I to a variable before its value is used in an
'
pointing. expression. Also known as unary increment.

pointer constant : a pointer whose contents cannot preprocessor: the first phase ol a C compilation in
be changed . which the source statements are prepared for compi-
pointer indirection: the use of a pointer to lation and any necessary libraries are loaded.
access data . preprocessor directives: commands to the C
pointer to function: a pointer that identifies the precompiler,
con-
entry' point to a function. It is used to pass a func- pretest loop: a loop in which the terminating loop state-
tion’s address as a parameter. dition is. tested before the execution of the
pop: the stack delete operation. -
ments. Contrast with post test loop.
consisting of
primary expression: an expression
polyonymic efficiency: a measure ol the effi- only a single operator ; the highest priority expression .
ciency of a module in which the run time is propor ) memory ol a
primary storage: the volatile RAM
(
donate to the number of elements raised to the
highest factor im i polynomial.
<
computer.
40 Glossary
intable character: a character value that is asso- quadratic loop: a loop that consists of two or more
rted with a print graphic. loops , each of which have a linear efficiency, result -
ing in a loop with quadratic efficiency.
•
inter: an output device that displays the output
i paper. quality: see software quality.
rocedural cohesion: a module design in which query: see inquiry.
ie processing within the module is related by con - queue: a linear list in which data can he inserted
ol flows. Considered acceptable design only at the only at one end , called the rear, and deleted from the
gher levels of a program . .
other end called the front .
rocess-control loops : a continuous loop used to
rntrol operating equipment , such as a building air
Dnditioning/heating system . R
rocessing unit : see central processing unit . ragged array: a two-dimensional array in which the
rogram development : the system development number of elements in the rows is not equal .
ctivity in which requirement specifications are con - random number: a number selected from a set in
erted into executable programs. which all members have the same probability of
•rogram file : a file that contains an executable being selected .
• rogram . random number seed : the initial number passed
> rogram testing: the process that validates a pro- to a random-number generator.
gram ’s operation and verifies that it meets its design read mode: the attribute of a file that indicates that
equirements. it is opened for input only.
> rototype declaration : the function header that read state : one of three states that an open file may
Jeclares the return type, function name, and parame- assume. The state of a file during which only input
er number and types without defining the function operations may be performed. See also error state ,
tself . write state .
program documentation : comments placed real time : processing in which updating takes place
A ithin a program to help the reader understand the
at the time the event occurs .
purpose of the program or a portion ol its
implementation . real type: a data type representing a number with a
fraction.
promotion : an implicit type conversion in which
the rank of an expression is temporarily increased to rear : when used to refer to a list ; a pointer that
match other elements of the expression . identifies the last element .
prototype statement: in C , the declaration of a record : see structure .
function that provides the return type and formal recursion : a function design in which the function
parameter types. calls itself .
pseudocode: English - like statements that follow a register : the storage class that requests that , if pos -
loosely defined syntax and are used to convey the
design of an algorithm or function . sible, a variable be allocated to a CPU register.
push : the stack insert operation . reliability: the quality factor that addresses the
confidence or trust in a system s total operation.
reserved words: the set of words in C that have a
Q
predetermined interpretation and cannot be used in
the definition ol an object .
quadratic ef ficiency: a measure of the efficiency
of a module in which the run time is proportionate to return : the C statement that causes execution of a
the number of elements squared . Quadratic effi - function to terminate and control to be resumed by
ciency is one of the polyonymic factors. the calling function or the operating system .
Glossary 1141
1
reusability: the quality factor that addresses th e ease selected and placed at the end of the sorted portion
with which software can be used in other programs. of the list .
reusable code: code that can be used by more than selection statement: a statement that chooses
one process. between two or more alternatives. In C , the if ...else
rewind: the C function used to position the file or switch statements.
marker at the beginning of an open file . self - referential structure: a structure that con -
right-to-left associativity: the evaluation of an tains a pointer to itself .
expression that parses from the right to the left. Con - sentinel: a flag that guards the end of a list or a file.
-
trast with left -to righl associativity. The sentinel is usually the maximum value for a key
right justification : the orientation of variable length field and cannot he a valid data value.
data in an output format such that leading null values separate compilation: a capability within a pro-
are inserted and the last data character is at the right gramming language that allows parts of a program to
end of the print area . Contrast with left justification . be compiled independently and later assembled into
—
right left rule: a method of reading complex dec -
larations or definitions that starts with the identifier
an executable unit .
sequential cohesion: a module design in which
and alternately reads right and left until the declara - the processing within the module flows such that the
tion has been fullv read . data from one process are used in the next process.
row: in a two-dimensional array, elements along a sequential file: a file structure in which data must
horizontal axis . Contrast with column . be processed serially from the first entry in the file.
root: the first node of a tree; the node with no sequential search : a search technique used with a
predecessor. linear list in which the searching begins at the first
rvalue: an expression attribute indicating that the element and continues until the value of an element
expression can be used only to supply a value for equal to the value being sought is located , or until
another expression. the end of the list is reached .
serial search: any search technique that starts at the
beginning of the list and continues toward the end.
s server: in a client/server system , the centralized
scanset: the set of allowed or disallowed characters computer that provides auxiliary services.
when reading a string. short : the C type for short integer.
scope: an attribute of a variable that defines short integer: an integer format used for small
whether it is visible to or hidden from statements in a numbers, generally less than 32 ,768 .
program . See also temporal authority. side effect: a change in a variable that results from
search: the process that examines a list to locate the evaluation of an expression ; any input/output
one or more elements containing a designated value performed by a called function.
code
known as a search argument. sign flag: the flag value (+ ) in the conversion
of a print format string that indicates that .
positive
secondary storage: see aiexilian storage .
security: the quality factor that addresses the ease
numbers must be printed with a plus sign
a numeric
or difficulty with which an unauthorized user can signed: a type modifier indicating thative. Contrast with
value may be either positive or negative
access data .
unsigned
seed : the variable used in a random number genera conversion code of
tor to calculate the next number in the series . In , siz.e: an indicator ( h , 1, L ) in the
integer and float types.
the seed is set by the library function sranii . a format string that modifies
formatting token that modi -
seek file: in the processing of a disk file , the action Jze modifier: the data
that moves the index marker to a new position in the lies the conversion code.
memory locations added
file; in C , the fseek function , slnck bytes: inaccessible to force a hardware
between fields in a structure
selection sort: the sort algorithm in which the alignment .
requiired boundary
smallest value in the unsorted portion of a list > s
/
1142 Glossary
soft copy: computer output written to a nonperma - statement: a syntactical construct in C that repre -
nent display such as a monitor. Contrast with sents one operation in a function .
hard copy. statement section: the section in a program where
software: the application and system programs nec - the executable statements are written .
essary' for computer hardware to accomplish a task static : a storage class that designates ( a ) that a local
including their documentation and any required field must maintain its contents throughout the exe -
procedures. cution of a program ( contrast with auto ) , or ( b) that a
software quality: an attribute ol software that global variable’s name must not be exported to the
measures the user’s total satisfaction with a system . linker (contrast with extern ) .
sort : the process that orders a list or file. static memory: memory whose use (e.g., lor a vari -
sort pass : one sort loop during which all elements able ) does not change during the running ol a
are tested . program.
source file : the file that contains program state - static memory allocation: memory whose loca -
ments written hv a programmer before they are con - tion is determined by the compiler and therefore pre-
verted into machine language ; the input file to an set before run time . Contrast with dynamic mentor)'
assembler or compiler. allocation .
space flag: the flag value in the conversion code of stderr: see standard error stream .
a print format string indicating that positive numbers stdin: see standard input stream .
must be printed with a leading space.
stdout : see standard output stream .
-
sparse array: a two dimensional array in which the
stepwise refinement: a design methodology in
rows are only partially filled .
which a system or program is developed from the top
spatial authority: an attribute ol a field that deter- down ; starting with the most inclusive, each module
mines which functions within a program can refer to is decomposed and refined until the meaning of a
it or change its contents.
component is intrinsically understood .
square brackets: the [ and ] symbols.
storage class: an attribute of a field that deter-
stack: a restricted linear list in which data be mines its spatial and temporal usage . See also auto ,
inserted and deleted at only one end , called the top . extern , register , static .
stack memory: in C , the memory management stream : in C , the flow of data between a file and the
facility that is used to store local variables while they program ; a sequence of bytes in time.
are active .
string: in C , a variable- length sequence of charac -
stamp coupling: the communication technique ters delimited by a null character.
between modules in which data are passed as a
structure; often results in unrequired data being string constant: a constant comprised of a
passed . sequence characters enclosed in double quotes.
standard error stream: the text file automatically strongly connected graph: a graph in which eacf
opened by C lor display of error messages ( stderr ) . there is a path from every node to every other node ;
contrast with weakly connected graph .
standard input stream: the text file automatically
opened by C for input to a program ( stdin ) . structure: a named collection of fields grouped
standard library: any of a collection of libraries together lor processing; see record .
containing functions required by the C standard pro - structure chart: a design and documentation tool
vided by an implementation of the C language. that represents a program as a hierarchical flow ol
standard output stream: the text file automati - functions.
cally opened by C for output from a program ( stdout ) . structure variable : in C , a structure definition tha
standard type : one of the intrinsic C types that arc cannot be used as a type .
considered atomic ; that is , that cannot he broken stub: a temporary and incomplete function used tc
down ( void , char , ini , float ) . test function interfaces during program developmen
Glossary 1143
subscript: an ordinal number that indicates the
position of an element within an array. See also index. ternary expression: in C, an expression
tains three operands (the that con -
summation : an algorithm that adds a list of the only ternary expressionconditional
in C).
expression is
data items.
testability: an attribute of software that
suffix: a modifier to a numeric constant that indi - the ease with which the soltvvare can he measures
cates its type when used in an expression , operational system . tested as an
switch: see flag . testing: see program testing , blackbox testing , or
switch statement: the C implementation of the white box testing .
multiway selection . text editor: software used to create and
change text
symbolic language: a computer language, one lilcs, such as a word processor or a source program
level removed from machine language, that has a editor.
mnemonic identifier lor each machine instruction text file: a file in which all data are stored as
and has the capability of symbolic data names. characters. Contrast with binar)' file.
syntax: the grammatical rules of a language. In C, timeliness: an attribute of software that measures
the set of keywords and formatting rules that must be responsiveness of a system to a user’s time
followed when writing a program. requirements.
system development life cycle: a model of the time -sharing environment: an operating system
steps required to develop software that begins with concept in which more than one user has access to a
the need for the software and concludes with its computer at the same time.
implementation. token: in C, a syntactical construct that represents
system development software: any computer tool an operation or a flag, such as the assignment
used to develop software, such as but not limited to token ( = ).
compilers, debuggers, and documentation tools. top: in a stack , the next element to be removed .
system software: any software whose primary pur- top-down design: a program design concept in
pose is to support the operation of the computing which a design progresses through a decomposition
environment . Contrast w ith application software . of the f unctions beginning with the top of the struc-
system support software: software used for non - ture chart and working toward the lowest modules.
application processing, such as system utilities. See also stepwise refinement .
transaction file: a file containing relatively tran -
sient data that is used to change the contents of a
T master file .
tagged structure: a C structure in which the struc - transform analysis: an analytical process that cre-
ture is given an identifier that can he used to dec are ates a program design by classifying modules as
variables of the structure type. input , process, or output.
tape storage: an auxiliary storage medium that transform module: the function in a program that
stores data as a sequential file on a magnetic recorc takes input from the afferent modules and prepares it
ing surface . for processing by the efferent modules.
file
temporal authority: an attribute of a field that translation unit: in C, a temporary compilation
determines when a field is alive and active during tlle used to store modified source code.
language
execution of a program . translator: a generic term for any of theand compiler.
temporal cohesion: a module design in which pro- conversion programs. See also assembler
each ele-
t
cesses are combined because they all neet to c pro traversal : an algorithmic process in which
once and only once.
cessed in the same time sequence. ment in a structure is processed
that an
t. terminating condition: in a loop, the condition true: a boolean value used to indicate
that stops a loop. See also limit test expression. expression is true.
1144 Glossary
two-dimensional array : an array in which each ele - variable: a memory storage object whose value can
ment contains one array. See also multidimensional be changed during the execution ol a program . C on -
array. trast with constant .
-
two way selection: a selection statement that is variable- length array: an array in which the num -
ber of elements can be changed during the run time.
capable of evaluating only two alternatives. In C , the
if ...else statement . Contrast with multiway selection . Contrast with fixed - length array.
type: a set of values and a set of operations that can variable structure : a data structure in which two
be applied on these values. or more types of data may occupy the same positions
type qualifier: a modifier used in the definition ol within the structure . In C , a union .
an object. visibility: the temporal authority for a field that des -
ignates where in the program it can be accessed or
changed; see scope .
U
vertex: a node in a graph .
unary expression : an expression that consists of void : the absence ol data .
one operator and one operand .
void functions: functions whose return value
unarv minus: in C , the operator that complements
* is void .
the value of an expression .
void pointer: a generic pointer type that can hold
unbuffered input/output: input or output that is the address of any data type or structure.
transmitted directly to or from memory without the
use of a buffer. volatile : an attribute of a variable that indicates that
it may be accessed or changed by functions beyond
underflow: an event that occurs when an attempt is
made to delete data from a data structure and it the control of the compiler.
is empty.
Unicode : The internation character set used to W
define valid characters in computer science . Each
character is represented using a 32- bit unsigned waterfall model : a system development life cycle
in which each phase of development , such as
nu mber.
requirement development and design , is completed
union : C term for variable structure. before the next phase starts.
unsigned : a type modifier indicating that a numeric weakly connected graph : a graph in which there
value may be only positive. Contrast with signed . is at least one node with no path to any other node:
update: ( 1 ) in array processing, the process that contrast with strongly connected graph .
changes the contents of an element ; ( 2 ) in file pro- weighted graph: a graph whose lines are weighted .
cessing, a mode that allows a file to be both read and
written . while loop: a sentinel-controlled , pretest loop in C.
update mode : a file process mode that specifies whitebox testing: program testing in which the
that the file will be both read and written . internal design of the program is considered; clear
upward communication: data flow from the hox testing . Contrast with blackbox testing .
called function to the calling function . whitespace : in C , the space , vertical and horizontal
user-defined function : any function written by the tabs , newline, and form -feed characters.
programmer, as opposed to a standard library f unction . width modifier: in a format string, a conversion
user prompts: monitor messages to a user that code modifier that specifies the input maximum or
request the user input one or more values . output minimum size.
write mode: the attribute of a file indicating that it
is opened for output only.
V
w rite state : one of three states that an open file can
L
value error: an input value that does not satisfy assume. A file in the write state can only be used lor
specified criteria . output . See also error state , read state .
Symbols American Standard Code for Informa - assignment expression 104
lion Interchange. See ASCII compound 105
simple 104
^/ define 51
continuation 712
analysis 14
ancestor 968 assignment suppression flag 63
64, 408
<ctype. h> 264 , 265 and 233 associativity 94 , 106
<limits. h > 41 , 344 . 385 , 856 ANSI C 31
Icft - to- right 108
<math.h> 153, 18 /
append tile 42 -
right -to left 108, 109
<$tdbool. h > 34 application software 4
“
atan 2f 1063
<stddef . h> 433 , 562 arc 981 atan 2 l 1063
<stdin . h> 882 argc 1092 atanf 1063
<stdio. h > 33, 187 , 397, 402, 423, argv 1092 atanl 1063
562 , 629 array 460 atexit 1068
<stdlib. h> 187 , 268 , 628, 649 , 704 column 511 . 514, 520 atol 1068
<string. h> 691 , 733 constant 477 atoi 1068
<time.h> 193 declaration 463 atol 1068
caret 676 fixed - length 463
_ underscore
A
atomic
_bool 232 37 frequency 481
histogram 482
data 986
type 986
Exit 1068 in structures 767 auto 1106
index 464 automatic extent 1104
index as offset 465 auxiliary storage 394
A initialization 465 device 3
a _char 1062 1063 memory layout 514^ auxiliary symbol 1019
,
abort 268, 1068 multi-dimensional 519
abs 187 , 1067 of pointers 633 B
abstract data type 728, 986 , 987, 988 of structures 770
abstraction 988 passing 476 , 517 , 623 B Language 30
actual parameter 165 passing row 516 ^ base 2 1034
add 103 passing values 474 base 10 1034
additive expression 103 plane 520 base 16 1036
pointer 612 base 256 1037
address operator 559 to binary 1040
adjacent vertices 981 printing 469
ragged 639 weight and value 1038
Adler, Mark 922 base case 352
afferent 910 range checking 470
row 511 , 514 . 520 basic latin 1005 Plane 1006
ALGOL 30
size 463 Basic Multilingual
algorithm efficiency 365
string 688 batch update 859
algorithmics 366 BCPL 30 ^
big-0 notation 370 subscript 460 big endian 786
dimensional 509
exponential efficiency 372 two big-0 notation 370
factorial efficiency 372 variable-length 463 binary
linear efficiency 372 array of pointers 633 -
block i/o 8 3
linear logarithmic efficiency 372 ASCII 39 , 48 , 1005 file 823
logarithmic efficiency 372 character code table 1009 binary expression 101
polyonymic efficiency 372 extended 1005 addit Ve l ( 3R82
' !
quadratic efficiency 372 asctime 1069 bitwise and
—
alphabet ASCII 39, 48 assembler language
f
1156 Index
vertex 981
782
type qualifier 1111
const 52 , 477
union
imluh / er 82
referencing members /8
_ void 38
void function
with parameters 156
const pointer 1112 structure 83
"
without parameters 156
restrict 1114 switch statement 786 void pointer 581
volatile 1113 universal character set 1005 volatile 1113
typedef 746, 754 UNIX 30 , 1091 , 1121
identifier style 46
"
Korn Shell 1091

unsigned 40 w
storage 1040
u unsigned integers 1040 waterfall model 13
update wcscat 1069
UCS See Universal Character Set wcschr 1069
errors 862
unary program design 860 wcscmp 1069
minus 100 structure chart 862 wcscpy 1069
plus 100 updating 859 wcscspn 1069
unary expression user prompt 128 wcslen 1069
not 232 wcsncat 1069
one’s complement 886 wcsncmp 1069
prefix decrement 98 V wcsncpy 1069
prefix increment 98 vvcspbrk 1069
sizeof 100 variable 42 wcsrchr 1069
underflow declaration 43 wcsspn 1069
queue 959 , 962 definition 43 wcsstr 1069
undirected graph 981 initializer 44 wcstod 706
ungetc 435 , 1066 local 164 wcstok 1069
ungetwc 1066 pointer 561 wea kly connected graph 982
Unicode 1005 . 1006 string 671 while loop 310
Basic Multilingual Plane 1006 variable length string 666 whitebox testing 20 , 594
Private Use Planes 1009
variable- length array 463 whitespace
declaration 464 format string 404
Supplementary Ideographic function call 476 width 57 , 408 , 420 , 677
Plane 1009 Wirth , Niklaus 912
Supplementary Multilingual function prototype 476
initialization 465 write state 824
Plane 1009
two-dimensional 517 wscanl 1066
Supplementary Special Plane 1009
!•
3 SJCD 00358 198N
s
s ,
1
integral hh ( char ), h ( short ) , none ( int ), 1 ( long ). 11 ( long long ) l
integer h ( short ), none ( int ), 1 ( long ). 11 ( long long ) d
unsigned int hh ( char ), h ( short ), none ( int ), 1 ( long ), 11 ( long long ) u
character octal hh ( unsigned char ) o
integer hexadecimal h ( short ), none ( int ), 1 ( long ), 11 ( long long ) x
real .
none ( float ) 1 ( long double ), L (double ) f
real ( scientific ) none ( float ), 1 ( long double ), L ( double ) e
real ( scientific ) none ( float ), 1 ( long double ), L ( double ) g
real ( hexadecimal ) none ( float ), 1 ( long double ), L ( double ) a
character none ( char ), 1 ( wchar t ) _ c
string _
none ( char string ), 1 ( wchar t string ) s
pointer
P
integer ( for count ) none ( int ), hh ( char ), h ( short ), 1 ( long ), 11 ( long long n
set none ( char ), 1 ( wchar t ) _ [
Sizes, and Conversion Code for scanf Family
Argument TVpe Flag Size Specifier code

integer -, +, 0, space hh (char), h ( short ), none ( int ), d, i
1 ( long ), 11 ( long long )
unsigned integer -, +, 0, space hh ( char ), h ( short ), none ( int ), u
I ( long ), 11 ( long long )
integer ( octal ) -, +, 0, #, space hh ( char ), h ( short ), none ( int ), o
integer ( hex ) -, + , 0, #. space hh ( char ), h ( short ), none ( int ), x, X
real -, +, 0, # , space none ( float ), 1 ( long double ), f
L (double )
real ( scientific ) -, +, 0, #, space none ( float ), 1 ( long double ), e, E
L ( double )
real ( scientific ) -, + , 0, #. space none ( float ), 1 ( long double ), g. G
L (double )
real ( hexadecimal ) -, +, 0, #. space none ( float ), 1 ( long double ), a, A
L ( double )
character none (char ), 1 ( w -char ) c
string none ( char string ), 1 ( w-char string )
pointer P
integer ( for count ) none ( int ), h ( short ), 1 ( long ) n
to print % %
Flags, Sizes, and Conversion Codes for print/ Family

Hi
Behrouz A. Forouzan & Richard F. Gilberg
This third edition of Computer Science: A Structured Programming Approach Using C

offers introductory students a fully updated, comprehensive survey of computer science
theory and the C programming language. Now in full color, Forouzan and Gilberg
continue to use a highly visual approach to explain fundamental programming concepts.
RELATED TITLE
Changes to the new edition include full compliance with the C 99 standard, expanded
sections on inter - function communication and the basic data structures, and a thoroughly
revised Chapter 3.
FEATURES
DATA STRUCTURES
o Updated to comply with the C99 standard, o Uses a large number of examples,
including discussions of complex ranging from code snippets to complete
arithmetic and Unicode. implementations requiring several
functions, providing the student with a
o Features a new full-color page design that
range of techniques to study and practice.
enhances figures and presents programs
as they might be shown in the compiler. o Contains extensive end-of-chapter Data Structures:
pedagogy, including a summary, A Pseudocode
o Offers a groundbreaking visual approach,
keywords, tips, and a robust problem Approach with C
including a wide variety of figures, tables, Second Edition
set featuring review questions,
and programs.
exercises, problems, and projects. Richard F. Gilberg &
Behrouz A. Forouzan
0-534 - 39080-3
ABOUT THE AUTHORS
Behrouz A. Forouzan has more than 32 years of electronics and computer science experience
in industry and academia. His industry experience includes designing electronic systems . After
leaving industry, he joined De Anza College as a professor of computer science. In addition to
this text , he has authored and co-authored nine other textbooks, including Data Structures: A
Pseudocode Approach with C and Foundations of Computer Science .
Richard F. Gilberg has more than 40 years of computer science experience in industry and
academia. His industry experience includes the development of large application systems,
database administration , and system testing. After leaving industry, he joined De Anza College
as a professor of computer science. In addition to this text, he has also co-authored several
others , including Data Structures: A Pseudocode Approach with C .
;V COURSEE TECHNOLOGY -- - -
- -
ISBN 13: 176 0 534 41135 1
-
ISBN 10: 0 534 41135 4 - -
Learning 30 DO >
CENGAG '
F o r your lifelong learning solutions, visit course.cengage.com

Purchase any of our products at your local college store or at our
preferred online store www.ichapters.com
=1 7 fl 53 13E1

Computer Science A Structured Approach Using C by Behrouz A. Forouzan Richard F. Gilberg

Uploaded by

Copyright:

Available Formats

Computer Science A Structured Approach Using C by Behrouz A. Forouzan Richard F. Gilberg

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Computer Science A Structured Approach Using C by Behrouz A. Forouzan Richard F. Gilberg

Uploaded by

Copyright:

Available Formats

COMPUTE R SCIENCE

A Structured Programming Approach Using C

Singapore • Spam • United Kingdom •United States

SAN JACIN1U COLLEGE LIBHAHY- SOUTH

basic principles ol program -

Major Changes in Third Edition

Extended Chapter Material

• The concept of “ incremental program development ’ is presented through

students a head start into data structures

Features of the Book

Structure and Style

Principle Before Practice

and to emphasize the structure of the language, we label the sections in a

Appendices and Cover Material

Web Site Support Materials

Peter Gabrovsky, CSU Northridge

Course Technology Staff

Chapter 1 Introduction to Computers 1

Chapter 3 Structure of a C Program 93

Chapter 4 Functions 149

Chapter 6 Repetition 303

7.3 Standard Library Input/Output Functions 397

Chapter 8 Arrays 459

Chapter 9 Pointers 557

9.3 Pointers to Pointers 576

Chapter 10 Pointer Applications 611

Chapter 12 Enumerated, Structure, and Union Types 745

13.7 Key Terms 867

Chapter 14 Bitwise Operators 881

Chapter 15 Lists 927

Appendix A Character Sets 1005

Appendix B Keywords 1017

Appendix C Flowcharting 1019

Appendix D Numbering Systems 1033

Appendix E Integer and Float Libraries 1055

Appendix F Function Libraries 1059

Character I/O 1066

Appendix G Preprocessor Commands 1071

Appendix H Command-Line Arguments 1091

Appendix I Pointers to Void ond to Functions 1095

Appendix J Storage Classes and Type Qualifiers 1103

Appendix K Program Development 1115

Appendix L Understanding Complex Declarations 1125

1.1 Computer Systems

FIGURE 1 - 1 A Computer System

FIGURE 1 - 2 Basic Hardware Components

I he input device is usually a keyboard

Operating System System General Application

FIGURE 1 - 3 Types of Software

The relationship between system and application software is seen in

FIGURE 1 - 4 Relationship Between System and Application Software

1.2 Computing Environments

Personal Computing Environment

FIGURE 1 - 5 Personal Computing Environment

Time- Sharing Environment

systems, as opposed to Apple ’s Macintosh. We use it in the original , generic

FIGURE 1 - 6 Time- sharing Environment

In the time-sharing environment , all computing must he done

FIGURE 1 -7 The Client / Server Environment

FIGURE 1 - 8 Distributed Computing