Thinking in Patterns (Bruce Eckel)

Thinking in Patterns by Bruce Eckel
Revision 0.9, 5-20-2003 (This version contains the material that will be
used in the Crested Butte seminar; see http://www.mindview.net/Seminars
/ThinkingInPatterns/)
Please note this document is under development and incomplete.
Updates to this document can be found at http://www.Mindview.net
Best viewed with Mozilla! (free at www.Mozilla.org) (Even though this
document was created with MS Word, IE6 garbles lines with superscripts
on them. Mozilla seems to do a much better job).
___________________________________________
Note: This document requires the installation of the fonts Georgia, Verdana and
Andale Mono (code font) for proper viewing. These can be found at:
http://sourceforge.net/project/showles.php?group_id=34153&release_id=105355
Modications in Revision 0.9:
Prose has still had little/no work. My current goal is to get the structure and
examples worked out so that the seminar works well. Once it has been proven
in the seminars, then I will spend time on the prose.
Added proxy:PoolManager.java to make a more generic/customizeable Pool

Manager, and modied proxy:ConnectionPoolProxyDemo.java accordingly [[
Still need to decide what to return when you run out of objects in the pool ]]
Changed PoolManager.java to use an ArrayList (and thus does not require a

xed size at initialization)
Added KissingPrincess.java to State description, as a motivational example

for the pattern
Added a simple Flyweight example
Simplied the enumeration in PaperScissorsRock.java
Changed Bridge example to improve clarity.
Removed superscripts for better viewing with IE (see note above)
NOTE primary changes have been made to structure of book and code
examples, but not to prose. Prose can be considered to be mostly a
mess in this revision.
Complete reorganization under headings that attempt to describe the

problems you are trying to solve with a pattern.
Addition of placeholders for the remainder of the GoF patterns
Addition of Simplifying Idioms section and examples
Addition of Builder section and examples
Removed unit-testing chapter; replaced with reference to new JUnit (which

uses reection)
(4-30-2003) Added Ant build.xml les, and support les from TIJ necessary to
do a full standalone build. You should be able to type ant from the code root
directory and get a successful build.
Dramatically simplied chainofresponsibility:FindMinima.java
Added object pool/connection pool examples
Refactored small things in many examples
Some exercises may have been left behind when patterns were moved.
For simplicity, saved from Word into a single HTML document, using
ltered version to remove Oce stu. Seems to work pretty well; checked
it with both IE and Mozilla (actually seems to work better on Mozilla than on
IE!).
TODO:
Recongure for new backtalk system
Replace references to TIJ2 with TIJ3
Thinking
Patterns
in
Problem-Solving Techniques
using Java
Bruce Eckel
President, MindView, Inc.
Contents
Preface
Introduction
The Y2K syndrome

Context and composition
A word about checked exceptions
The pattern concept
What is a pattern?
Pattern taxonomy
Design principles
Classifying patterns
The development challenge
Unit testing
Location of test code
Simplifying Idioms
Messenger
Collecting Parameter
Object quantity
Singleton
Exercises
Object pool
Exercises
Object decoupling
Proxy: fronting for another object

The PoolManager using Proxy
Dynamic Proxies
State: changing object behavior

Iterators: decoupling algorithms from containers
Type-safe iterators
Exercises
Factoring commonality
Strategy: choosing the algorithm at run-time

Policy: generalized strategy
Template method
Exercises
Encapsulating creation
Simple Factory method

Polymorphic factories
Abstract factories
Exercises
Specialized creation
Prototype
Builder
Exercises
Too many
Flyweight: too many objects

Decorator: too many classes
Basic decorator structure
A coee example
Class for each combination
The decorator approach
Compromise
Other considerations
Exercises
Connecting dierent types

Adapter
Bridge
Exercises
Flexible structure
Composite
System decoupling
Observer
Observing owers
A visual example of observers
Mediator
Exercises
Reducing interface complexity

Faade
Package as a variation of Faade
Algorithmic partitioning
Command: choosing the operation at run-time

Exercises
Chain of responsibility
Exercises
Externalizing object state

Memento
Complex interactions
Multiple dispatching
Visitor, a type of multiple dispatching
Exercises
Multiple languages
Interpreter motivation
Python overview
Built-in containers
Functions
Strings
Classes
Creating a language
Controlling the interpreter
Putting data in
Getting data out
Multiple interpreters
Controlling Java from Jython

Inner Classes
Using Java libraries

Inheriting from Java library classes
Creating Java classes with Jython

Building the Java classes from the Python code
The Java-Python Extension (JPE)

Summary
Exercises
Complex system states

StateMachine
Exercises
Table-Driven State Machine

The State class
Conditions for transition
Transition actions
The table
The basic machine
Simple vending machine
Testing the machine
Tools
Table-driven code: conguration exibility
Table-driven code using anonymous inner classes
Exercises
Pattern refactoring
Simulating the trash recycler

Improving the design
Make more objects
A pattern for prototyping creation

Trash subclasses
Parsing Trash from an external le
Recycling with prototyping
Abstracting usage
Implementing the double dispatch
The Visitor pattern

A Reective Decorator
More coupling?
RTTI considered harmful?

Summary
Exercises
Projects
Rats & Mazes

Other maze resources
XML Decorator
A: Tools
Ant extensions
Array utilities
Preface
The material in this book has been developed in conjunction with
a seminar that I have given for several years, mostly with Bill
Venners. Bill and I have given many iterations of this seminar and
weve changed it many times over the years as we both have
learned more about patterns and about giving the seminar.
In the process weve both produced more than enough information for us each to
have our own seminars, an urge that weve both strongly resisted because we have
so much fun giving the seminar together. Weve given the seminar in numerous
places in the US, as well as in Prague (where we try to have a mini-conference every
Spring together with a number of other seminars). Weve also given it as an on-site
seminar.
A great deal of appreciation goes to the people who have participated in these
seminars over the years, as they have helped me work through these ideas and to
rene them. I hope to be able to continue to form and develop these kinds of ideas
through this book and seminar for many years to come.
This book will not stop here, either. After much pondering, Ive realized that I want
Thinking in Python to be, initially, a translation of this book rather than an
introduction to Python (there are already plenty of ne introductions to that superb
language). I nd this prospect to be much more exciting than the idea of struggling
through another language tutorial (my apologies to those who were hoping for that).
Introduction
This is a book about design that I have been working on for years,
basically ever since I rst started trying to read Design Patterns
(Gamma, Helm, Johnson & Vlissides, Addison-Wesley, 1995),
[1]
or just GoF).
commonly referred to as the Gang of Four
There is a chapter on design patterns in the rst edition of Thinking in C++, which
has evolved in Volume 2 of the second edition of Thinking in C++, and youll also
nd a chapter on patterns in the rst edition of Thinking in Java (I took it out of the
second edition because that book was getting too big, and also because I had
decided to write this book).
This is not an introductory book. I am assuming that you have worked your way
through Thinking in Java or an equivalent text before coming to this book.
In addition, I assume you have more than just a grasp of the syntax of Java. You
should have a good understanding of objects and what theyre about, including
polymorphism. Again, these are topics covered in Thinking in Java.
On the other hand, by going through this book youre going to learn a lot about
object-oriented programming by seeing objects used in many dierent situations. If
your knowledge of objects is rudimentary, it will get much stronger in the process of
understanding the designs in this book.
The Y2K syndrome

In a book that has problem-solving techniques in its subtitle, its worth mentioning
one of the biggest pitfalls in programming: premature optimization. Every time I
bring this concept forward, virtually everyone agrees to it. Also, everyone seems to
reserve in their own mind a special case except for this thing that I happen to
know is a particular problem.
The reason I call this the Y2K syndrome has to do with that special knowledge.
Computers are a mystery to most people, so when someone announced that those
silly computer programmers had forgotten to put in enough digits to hold dates past
the year 1999, then suddenly everyone became a computer expert these things
arent so dicult after all, if I can see such an obvious problem. For example, my
background was originally in computer engineering, and I started out by
programming embedded systems. As a result, I know that many embedded systems
have no idea what the date or time is, and even if they do that data often isnt used
in any important calculations. And yet I was told in no uncertain terms that all the
embedded systems were going to crash on January 1, 2000. As far as I can tell the
only memory that was lost on that particular date was that of the people who were
predicting doom its as if they had never said any of that stu.
The point is that its very easy to fall into a habit of thinking that the particular
algorithm or piece of code that you happen to partly or thoroughly understand is
naturally going to be the bottleneck in your system, simply because you can imagine
whats going on in that piece of code and so you think that it must somehow be
much less ecient than all the other pieces of code that you dont know about. But
unless youve run actual tests, typically with a proler, you cant really know whats
going on. And even if you are right, that a piece of code is very inecient,
remember that most programs spend something like 90% of their time in less than
10% of the code in the program, so unless the piece of code youre thinking about
happens to fall into that 10% it isnt going to be important.
We should forget about small eciencies, say about 97% of the time: Premature
optimization is the root of all evil.Donald Knuth
Context and composition

One of the terms you will see used over and over in design patterns literature is
context. In fact, one common denition of a design pattern is a solution to a
problem in a context. The GoF patterns often have a context object that the
client programmer interacts with. At one point it occurred to me that such objects
seemed to dominate the landscape of many patterns, and so I began asking what
they were about.
The context object often acts as a little faade to hide the complexity of the rest of
the pattern, and in addition it will often be the controller that manages the
operation of the pattern. Initially, it seemed to me that these were not really
essential to the implementation, use and understanding of the pattern. However, I
remembered one of the more dramatic statements made in the GoF: prefer
composition to inheritance. The context object allows you to use the pattern in a
composition, and that may be its primary value.
A word about checked exceptions

1) The great value of exceptions is the unication of error reporting: a standard
mechanism by which to report errors, rather than the popourri of ignorable
approaches that we had in C (and thus, C++, which only adds exceptions to the mix,
and doesn't make it the exclusive approach). The big advantage Java has over C++
is that exceptions are the only way to report errors.
2) "Ignorable" in the previous paragraph is the other issue. The theory is that if the
compiler forces the programmer to either handle the exception or pass it on in an
exception specication, then the programmer's attention will always be brought
back to the possibility of errors and they will thus properly take care of them. I
think the problem is that this is an untested assumption we're making as language
designers that falls into the eld of psychology. My theory is that when someone is
trying to do something and you are constantly prodding them with annoyances, they
will use the quickest device available to make those annoyances go away so they
can get their thing done, perhaps assuming they'll go back and take out the device
later. I discovered I had done this in the rst edition of Thinking in Java:
...
} catch (SomeKindOfException e) {}
And then more or less forgot it until the rewrite. How many people thought this was
a good example and followed it? Martin Fowler began seeing the same kind of code,
and realized people were stubbing out exceptions and then they were disappearing.
The overhead of checked exceptions was having the opposite eect of what was
intended, something that can happen when you experiment (and I now believe that
checked exceptions were an experiment based on what someone thought was a
good idea, and which I believed was a good idea until recently).
When I started using Python, all the exceptions appeared, none were accidentally
"disappeared." If you *want* to catch an exception, you can, but you aren't forced to
write reams of code all the time just to be passing the exceptions around. They go
up to where you want to catch them, or they go all the way out if you forget (and
thus they remind you) but they don't vanish, which is the worst of all possible cases.
I now believe that checked exceptions encourage people to make them vanish. Plus
they make much less readable code.
In the end, I think we must realize the experimental nature of exceptions and look
at them carefully before assuming that everything about exceptions in Java are
good. I believe that having a single mechanism for handling errors is excellent, and
I believe that using a separate channel (the exception handling mechanism) for
moving the exceptions around is good. But I do remember one of the early
arguments for exception handling in C++ was that it would allow the programmer
to separate the sections of code where you just wanted to get work done from the
sections where you handled errors, and it seems to me that checked exceptions do
not do this; instead, they tend to intrude (a lot) into your "normal working code" and
thus are a step backwards. My experience with Python exceptions supports this,
and unless I get turned around on this issue I intend to put a lot more
RuntimeExceptions into my Java code.
The pattern concept

Design patterns help you learn from others' successes instead of
[2]
your own failures .
Probably the most important step forward in object-oriented design is the design
[3]
patterns movement, chronicled in Design Patterns (ibid) . That book shows 23
dierent solutions to particular classes of problems. In this book, the basic concepts
of design patterns will be introduced along with examples. This should whet your
appetite to read Design Patterns by Gamma, et. al., a source of what has now
become an essential, almost mandatory, vocabulary for OOP programmers.
The latter part of this book contains an example of the design evolution process,
starting with an initial solution and moving through the logic and process of
evolving the solution to more appropriate designs. The program shown (a trash
sorting simulation) has evolved over time, and you can look at that evolution as a
prototype for the way your own design can start as an adequate solution to a
particular problem and evolve into a exible approach to a class of problems.
What is a pattern?
Initially, you can think of a pattern as an especially clever and insightful way of
solving a particular class of problems. That is, it looks like a lot of people have
worked out all the angles of a problem and have come up with the most general,
exible solution for it. The problem could be one you have seen and solved before,
but your solution probably didnt have the kind of completeness youll see embodied
in a pattern.
Although theyre called design patterns, they really arent tied to the realm of
design. A pattern seems to stand apart from the traditional way of thinking about
analysis, design, and implementation. Instead, a pattern embodies a complete idea
within a program, and thus it can sometimes appear at the analysis phase or
high-level design phase. This is interesting because a pattern has a direct
implementation in code and so you might not expect it to show up before low-level
design or implementation (and in fact you might not realize that you need a
particular pattern until you get to those phases).
The basic concept of a pattern can also be seen as the basic concept of program
design: adding a layer of abstraction. Whenever you abstract something youre
isolating particular details, and one of the most compelling motivations behind this
is to separate things that change from things that stay the same. Another way to put
this is that once you nd some part of your program thats likely to change for one
reason or another, youll want to keep those changes from propagating other
changes throughout your code. Not only does this make the code much cheaper to
maintain, but it also turns out that it is usually simpler to understand (which results
in lowered costs).
Often, the most dicult part of developing an elegant and cheap-to-maintain design
is in discovering what I call the vector of change. (Here, vector refers to the
maximum gradient and not a container class.) This means nding the most
important thing that changes in your system, or put another way, discovering where
your greatest cost is. Once you discover the vector of change, you have the focal
point around which to structure your design.
So the goal of design patterns is to isolate changes in your code. If you look at it this
way, youve been seeing some design patterns already in this book. For example,
inheritance can be thought of as a design pattern (albeit one implemented by the
compiler). It allows you to express dierences in behavior (thats the thing that
changes) in objects that all have the same interface (thats what stays the same).
Composition can also be considered a pattern, since it allows you to change
dynamically or staticallythe objects that implement your class, and thus the way
that class works.
Youve also already seen another pattern that appears in Design Patterns: the
iterator (Java 1.0 and 1.1 capriciously calls it the Enumeration; Java 2 containers
use iterator). This hides the particular implementation of the container as youre
stepping through and selecting the elements one by one. The iterator allows you to
write generic code that performs an operation on all of the elements in a sequence
without regard to the way that sequence is built. Thus your generic code can be
used with any container that can produce an iterator.
Pattern taxonomy
One of the events thats occurred with the rise of design patterns is what could be
thought of as the pollution of the term people have begun to use the term to
mean just about anything synonymous with good. After some pondering, Ive come
up with a sort of hierarchy describing a succession of dierent types of categories:
1. Idiom: how we write code in a particular language to do this particular type
of thing. This could be something as common as the way that you code the
process of stepping through an array in C (and not running o the end).
2. Specic Design: the solution that we came up with to solve this particular
problem. This might be a clever design, but it makes no attempt to be
general.
3. Standard Design: a way to solve this kind of problem. A design that has
become more general, typically through reuse.
4. Design Pattern: how to solve an entire class of similar problem. This usually
only appears after applying a standard design a number of times, and then
seeing a common pattern throughout these applications.
I feel this helps put things in perspective, and to show where something might t.
However, it doesnt say that one is better than another. It doesnt make sense to try
to take every problem solution and generalize it to a design pattern its not a good
use of your time, and you cant force the discovery of patterns that way; they tend
to be subtle and appear over time.
One could also argue for the inclusion of Analysis Pattern and Architectural Pattern
in this taxonomy.
Design principles
(Update from slides to here)
[4]
When I put out a call for ideas in my newsletter , a number of suggestions came
back which turned out to be very useful, but dierent than the above classication,
and I realized that a list of design principles is at least as important as design
structures, but for a dierent reason: these allow you to ask questions about your
proposed design, to apply tests for quality.
Principle of least astonishment (dont be astonishing).
Make common things easy, and rare things possible
Consistency. One thing has become very clear to me, especially because of
Python: the more random rules you pile onto the programmer, rules that have
nothing to do with solving the problem at hand, the slower the programmer
can produce. And this does not appear to be a linear factor, but an
exponential one.
Law of Demeter: a.k.a. Dont talk to strangers. An object should only

reference itself, its attributes, and the arguments of its methods.
Subtraction: a design is nished when you cannot take anything else away.
[5]
Simplicity before generality
. (A variation of Occams Razor, which says
the simplest solution is the best). A common problem we nd in frameworks
is that they are designed to be general purpose without reference to actual
systems. This leads to a dizzying array of options that are often unused,
misused or just not useful. However, most developers work on specic
systems, and the quest for generality does not always serve them well. The
best route to generality is through understanding well-dened specic
examples. So, this principle acts as the tie breaker between otherwise equally
viable design alternatives. Of course, it is entirely possible that the simpler
solution is the more general one.
Reexivity (my suggested term). One abstraction per class, one class per
abstraction. Might also be called Isomorphism.
Independence or Orthogonality. Express independent ideas independently.

This complements Separation, Encapsulation and Variation, and is part of the
Low-Coupling-High-Cohesion message.
Once and once only: Avoid duplication of logic and structure where the
duplication is not accidental, ie where both pieces of code express the same
intent for the same reason.
In the process of brainstorming this idea, I hope to come up with a small handful of
fundamental ideas that can be held in your head while you analyze a problem.
However, other ideas that come from this list may end up being useful as a checklist
while walking through and analyzing your design.
Classifying patterns
The Design Patterns book discusses 23 dierent patterns, classied under three
purposes (all of which revolve around the particular aspect that can vary). The three
purposes are:
1.
Creational: how an object can be created. This often involves isolating the
details of object creation so your code isnt dependent on what types of
objects there are and thus doesnt have to be changed when you add a new
type of object. The aforementioned Singleton is classied as a creational
pattern, and later in this book youll see examples of Factory Method and
Prototype.
2.
Structural: designing objects to satisfy particular project constraints. These

work with the way objects are connected with other objects to ensure that
changes in the system dont require changes to those connections.
3.
Behavioral: objects that handle particular types of actions within a program.

These encapsulate processes that you want to perform, such as interpreting a
language, fullling a request, moving through a sequence (as in an iterator),
or implementing an algorithm. This book contains examples of the Observer
and the Visitor patterns.
The Design Patterns book has a section on each of its 23 patterns along with one or
more examples for each, typically in C++ (rather restricted C++, at that) but
sometimes in Smalltalk. (Youll nd that this doesnt matter too much since you can
easily translate the concepts from either language into Java.) This book will revisit
many of the patterns shown in Design Patterns but with a Java orientation, since the
language changes the expression and understanding of the patterns. However, the
GoF examples will not be repeated here, since I believe that its possible to produce
more illuminating examples given some eort. The goal is to provide you with a
decent feel for what patterns are about and why they are so important.
After years of looking at these things, it began to occur to me that the patterns
themselves use basic principles of organization, other than (and more fundamental
than) those described in Design Patterns. These principles are based on the
structure of the implementations, which is where I have seen great similarities
between patterns (more than those expressed in Design Patterns). Although we
generally try to avoid implementation in favor of interface, for awhile I thought that
it was easier to understand the patterns in terms of these structural principles, and
tried reorganizing the book around the patterns based on their structure instead of
the categories presented in Design Patterns.
However, a later insight made me realize that its more useful to organize the
patterns in terms of the problems they solve. I believe this is a subtle but important
distinction from the way Metsker organizes the patterns by intent in Design
Patterns Java Workshop (Addison-Wesley 2002), because I hope that you will then be
able to recognize your problem and search for a solution, if the patterns are
organized this way.
In the process of doing all this book refactoring I realized that if I changed it
once, I would probably change it again (theres denitely a design maxim in there),
so I removed all references to chapter numbers in order to facilitate this change
(the little-known numberless chapter pattern J).
The development challenge

Issues of development, the UML process, Extreme Programming.
Is evaluation valuable? The Capability Immaturity Model:
Wiki Page: http://c2.com/cgi-bin/wiki?CapabilityImMaturityModel
Article: http://www.embedded.com/98/9807br.htm
Pair programming research:
http://collaboration.csc.ncsu.edu/laurie/
Unit testing
In an earlier version of this book I decided that unit testing was essential (for all of
my books) and that JUnit was too verbose and clunky to consider. At that time I
wrote my own unit testing framework using Java reection to simplify the syntax
necessary to achieve unit testing. For the third edition of Thinking in Java, we
developed another unit testing framework for that book which would test the output
of examples.
In the meantime, JUnit has changed to add a syntax remarkably similar to the one
that I used in an earlier version of this book. I dont know how much inuence I may
have had on that change, but Im simply happy that it has happened, because I no
longer feel the need to support my own system (which you can still nd <some URL
here>) and can simply recommend the defacto standard.
I have introduced and described the style of JUnit coding that I consider a best
practice (primarily because of simplicity), in Thinking in Java, 3 rd edition, chapter
15. That section provides an adequate introduction to any of the unit testing you
will see associated with this book (however, the unit testing code will not normally
be included in the text of this book). When you download the code for this book, you
will nd (4/9/2003: Eventually, not yet) unit tests along with the code examples
whenever possible.
Location of test code

(From Bill):
Public: in test subdirectory; dierent package (dont include in jar).
Package access: same package, subdirectory path underneath library code (dont
include in jar)
Private access: (white box testing). Nested class, strip out, or Junit addons.
Simplifying Idioms
Before getting into more complex techniques, its helpful to look at some basic ways
to keep code simple and straightforward.
Messenger
The most trivial of these is the messenger, which simply packages information into
an object to be passed around, instead of passing all the pieces around separately.
Note that without the messenger, the code for translate() would be much more
confusing to read:
//: simplifying:MessengerDemo.java
package simplifying;
import junit.framework.*;
class Point { // A messenger
public int x, y, z; // Since it's just a carrier
public Point(int x, int y, int z) {
this.x = x;
this.y = y;
this.z = z;
}
public Point(Point p) { // Copy-constructor
this.x = p.x;
this.y = p.y;
this.z = p.z;
}
public String toString() {
return "x: " + x + " y: " + y + " z: " + z;
}
}
class Vector {
public int magnitude, direction;
public Vector(int magnitude, int direction) {
this.magnitude = magnitude;
this.direction = direction;
}
}
class Space {
public static Point translate(Point p, Vector v) {
p = new Point(p); // Don't modify the original
// Perform calculation using v. Dummy calculation:
p.x = p.x + 1;
p.y = p.y + 1;
p.z = p.z + 1;
return p;
}
}
public class MessengerDemo extends TestCase {
public void test() {
Point p1 = new Point(1, 2, 3);
Point p2 = Space.translate(p1, new Vector(11, 47));
String result = "p1: " + p1 + " p2: " + p2;
System.out.println(result);
assertEquals(result,
"p1: x: 1 y: 2 z: 3 p2: x: 2 y: 3 z: 4");

}
public static void main(String[] args) {
junit.textui.TestRunner.run(MessengerDemo.class);
}
} ///:~
Since the goal of a messenger is only to carry data, that data is made public for
easy access. However, you may also have reasons to make the elds private.
Collecting Parameter
Messengers big brother is the collecting parameter, whose job is to capture
information from the method to which it is passed. Generally, this is used when the
collecting parameter is passed to multiple methods, so its like a bee collecting
pollen.
A container makes an especially useful collecting parameter, since it is already set
up to dynamically add objects:
//: simplifying:CollectingParameterDemo.java
package simplifying;
import java.util.*;
class CollectingParameter extends ArrayList {}
class Filler {
public void f(CollectingParameter cp) {
cp.add("accumulating");
}
public void g(CollectingParameter cp) {
cp.add("items");
}
public void h(CollectingParameter cp) {
cp.add("as we go");
}
}
public class CollectingParameterDemo extends TestCase {
Filler ller = new Filler();
CollectingParameter cp = new CollectingParameter();
ller.f(cp);
ller.g(cp);
ller.h(cp);
String result = "" + cp;
System.out.println(cp);
assertEquals(result,"[accumulating, items, as we go]");
}
junit.textui.TestRunner.run(
CollectingParameterDemo.class);
}
} ///:~
The collecting parameter must have some way to set or insert values. Note that by
this denition, a messenger could be used as a collecting parameter. The key is that
a collecting parameter is passed about and modied by the methods it is passed to.
Object quantity
The two patterns described here are solely used to control the
quantity of objects.
Singleton could actually be thought of as a special case of Object Pool, but the
applications of the Object Pool tend to be uniqe enough from Singleton that its
worth treating the two separately.
Singleton
Possibly the simplest design pattern is the singleton, which is a way to provide one
and only one object of a particular type. An important aspect of Singleton is that
you provide a global access point, so singletons are often a solution for what you
would have used a global variable for in C. In addition, a singleton often has the
characteristics of a registry or lookup service its a place you go to nd references
to other objects.
Singletons can be found in the Java libraries, but heres a more direct example:
//: singleton:SingletonPattern.java
// The Singleton design pattern: you can
// never instantiate more than one.
package singleton;
//
//
//
//
Since this isn't inherited from a Cloneable

base class and cloneability isn't added,
making it nal prevents cloneability from
being added through inheritance:
nal class Singleton {

private static Singleton s = new Singleton(47);
private int i;
private Singleton(int x) { i = x; }
public static Singleton getReference() {
return s;
}
public int getValue() { return i; }
public void setValue(int x) { i = x; }
}
public class SingletonPattern extends TestCase {

Singleton s = Singleton.getReference();
String result = "" + s.getValue();
assertEquals(result, "47");
Singleton s2 = Singleton.getReference();
s2.setValue(9);
result = "" + s.getValue();
assertEquals(result, "9");
try {
// Can't do this: compile-time error.
// Singleton s3 = (Singleton)s2.clone();
} catch(Exception e) {
throw new RuntimeException(e);
}
}
junit.textui.TestRunner.run(SingletonPattern.class);
}
} ///:~
The key to creating a singleton is to prevent the client programmer from having any
way to create an object except the ways you provide. You must make all
constructors private, and you must create at least one constructor to prevent the
compiler from synthesizing a default constructor for you (which it will create using
package access).
At this point, you decide how youre going to create your object. Here, its created
statically, but you can also wait until the client programmer asks for one and create
it on demand. In any case, the object should be stored privately. You provide access
through public methods. Here, getReference( ) produces the reference to the
Singleton object. The rest of the interface (getValue( ) and setValue( )) is the
regular class interface.
Java also allows the creation of objects through cloning. In this example, making the
class nal prevents cloning. Since Singleton is inherited directly from Object, the
clone( ) method remains protected so it cannot be used (doing so produces a
compile-time error). However, if youre inheriting from a class hierarchy that has
already overridden clone( ) as public and implemented Cloneable, the way to
prevent cloning is to override clone( ) and throw a CloneNotSupportedException
as described in Appendix A of Thinking in Java, 2nd edition. (You could also override
clone( ) and simply return this, but that would be deceiving since the client
programmer would think they were cloning the object, but would instead still be
dealing with the original.) Actually, this isnt precisely true, because even in the
above situation someone could still use reection to invoke clone( ) [[is this true?
clone( ) is still protected so Im not so sure. If it is true, youd have to throw
CloneNotSupportedException as the only way to guarantee un-cloneability ]]
Exercises
1.
SingletonPattern.java always creates an object, even if its never used.

Modify this program to use lazy initialization, so the singleton object is only
created the rst time that it is needed.
2.
Create a registry/lookup service that accepts a Java interface and produces a

reference to an object that implements that interface.
Object pool
Note that you arent restricted to creating only one object. This is also a technique
to create a limited pool of objects. In that situation, however, you can be confronted
with the problem of sharing objects in the pool. If this is an issue, you can create a
solution involving a check-out and check-in of the shared objects.
As an example, consider a database. Commercial databases often restrict the
number of connections that you can use at any one time. Here is an implementation
that uses an object pool to manage the connections. First, the basic concept of
managing a pool of objects is implemented as a separate class:
//: singleton:PoolManager.java
package singleton;
import java.util.*;
public class PoolManager {
private static class PoolItem {
boolean inUse = false;
Object item;
PoolItem(Object item) { this.item = item; }
}
private ArrayList items = new ArrayList();
public void add(Object item) {
items.add(new PoolItem(item));
}
static class EmptyPoolException extends Exception {}
public Object get() throws EmptyPoolException {
for(int i = 0; i < items.size(); i++) {
PoolItem pitem = (PoolItem)items.get(i);
if(pitem.inUse == false) {
pitem.inUse = true;
return pitem.item;
}
}
// Fail early:
throw new EmptyPoolException();
// return null; // Delayed failure
}
public void release(Object item) {
if(item == pitem.item) {
pitem.inUse = false;
return;
}
}
throw new RuntimeException(item + " not found");
}
} ///:~
//: singleton:ConnectionPoolDemo.java
package singleton;
interface Connection {
Object get();
void set(Object x);
}
class ConnectionImplementation implements Connection {
public Object get() { return null; }
public void set(Object s) {}
}
class ConnectionPool { // A singleton
private static PoolManager pool = new PoolManager();
public static void addConnections(int number) {
for(int i = 0; i < number; i++)
pool.add(new ConnectionImplementation());
}
public static Connection getConnection()
throws PoolManager.EmptyPoolException {
return (Connection)pool.get();
}
public static void releaseConnection(Connection c) {
pool.release(c);
}
}
public class ConnectionPoolDemo extends TestCase {
static {
ConnectionPool.addConnections(5);
}
Connection c = null;
try {
c = ConnectionPool.getConnection();
} catch (PoolManager.EmptyPoolException e) {
}
c.set(new Object());
c.get();
ConnectionPool.releaseConnection(c);
}
public void test2() {
Connection c = null;
try {
c = ConnectionPool.getConnection();
} catch (PoolManager.EmptyPoolException e) {
}
c.get();
ConnectionPool.releaseConnection(c);
}
public static void main(String args[]) {
junit.textui.TestRunner.run(ConnectionPoolDemo.class);
}
} ///:~
Exercises
1.
Add unit tests to ConnectionPoolDemo.java to demonstrate the problem that

the client may release the connection but still continue to use it.
Object decoupling
Both Proxy and State provide a surrogate class that you use in your code; the real
class that does the work is hidden behind this surrogate class. When you call a
method in the surrogate, it simply turns around and calls the method in the
implementing class. These two patterns are so similar that the Proxy is simply a
special case of State. One is tempted to just lump the two together into a pattern
called Surrogate, but the term proxy has a long-standing and specialized meaning,
which probably explains the reason for the two dierent patterns.
The basic idea is simple: from a base class, the surrogate is derived along with the
class or classes that provide the actual implementation:
When a surrogate object is created, it is given an implementation to which to send

all of the method calls.
Structurally, the dierence between Proxy and State is simple: a Proxy has only one
implementation, while State has more than one. The application of the patterns is
considered (in Design Patterns) to be distinct: Proxy is used to control access to its
implementation, while State allows you to change the implementation dynamically.
However, if you expand your notion of controlling access to implementation then
the two t neatly together.
Proxy: fronting for another object

If we implement Proxy by following the above diagram, it looks like this:
//: proxy:ProxyDemo.java
// Simple demonstration of the Proxy pattern.
package proxy;
interface ProxyBase {
void f();
void g();
void h();
}
class Proxy implements ProxyBase {
private ProxyBase implementation;
public Proxy() {
implementation = new Implementation();
}
// Pass method calls to the implementation:
public void f() { implementation.f(); }
public void g() { implementation.g(); }
public void h() { implementation.h(); }
}
class Implementation implements ProxyBase {
public void f() {
System.out.println("Implementation.f()");
}
public void g() {
System.out.println("Implementation.g()");
}
public void h() {
System.out.println("Implementation.h()");
}
}
public class ProxyDemo extends TestCase {
Proxy p = new Proxy();
// This just makes sure it will complete
// without throwing an exception.
p.f();
p.g();
p.h();
}
junit.textui.TestRunner.run(ProxyDemo.class);
}
} ///:~
Of course, it isnt necessary that Implementation have the same interface as

Proxy; as long as Proxy is somehow speaking for the class that it is referring
method calls to then the basic idea is satised (note that this statement is at odds
with the denition for Proxy in GoF). However, it is convenient to have a common
interface so that Implementation is forced to fulll all the methods that Proxy
needs to call.
The PoolManager using Proxy

//: proxy:PoolManager.java
package proxy;
import java.util.*;
public class PoolManager {
private static class PoolItem {
boolean inUse = false;
Object item;
PoolItem(Object item) { this.item = item; }
}
public class ReleasableReference { // Used to build the proxy
private PoolItem reference;
private boolean released = false;
public ReleasableReference(PoolItem reference) {
this.reference = reference;
}
public Object getReference() {
if(released)
throw new RuntimeException(
"Tried to use reference after it was released");
return reference.item;
}
public void release() {
released = true;
reference.inUse = false;
}
}
private ArrayList items = new ArrayList();
public void add(Object item) {
items.add(new PoolItem(item));
}
// Dierent (better?) approach to running out of items:
public static class EmptyPoolItem {}
public ReleasableReference get() {
if(pitem.inUse == false) {
pitem.inUse = true;
return new ReleasableReference(pitem);
}
}
// Fail as soon as you try to cast it:
// return new EmptyPoolItem();
return null; // temporary
}
} ///:~
//: proxy:ConnectionPoolProxyDemo.java
package proxy;
interface Connection {
Object get();
void set(Object x);
void release();
}
class ConnectionImplementation implements Connection {
public Object get() { return null; }
public void set(Object s) {}
public void release() {} // Never called directly
}
class ConnectionPool { // A singleton
private static PoolManager pool = new PoolManager();
private ConnectionPool() {} // Prevent synthesized constructor
public static void addConnections(int number) {
for(int i = 0; i < number; i++)
pool.add(new ConnectionImplementation());
}
public static Connection getConnection() {
PoolManager.ReleasableReference rr =
(PoolManager.ReleasableReference)pool.get();
if(rr == null) return null;
return new ConnectionProxy(rr);
}
// The proxy as a nested class:
private static
class ConnectionProxy implements Connection {
private PoolManager.ReleasableReference implementation;
public
ConnectionProxy(PoolManager.ReleasableReference rr) {
implementation = rr;
}
public Object get() {
return
((Connection)implementation.getReference()).get();
}
public void set(Object x) {
((Connection)implementation.getReference()).set(x);
}
public void release() { implementation.release(); }
}
}
public class ConnectionPoolProxyDemo extends TestCase {
static {
ConnectionPool.addConnections(5);
}
Connection c = ConnectionPool.getConnection();
c.get();
c.release();
}
public void testDisable() {
Connection c = ConnectionPool.getConnection();
String s = null;
c.get();
c.release();
try {
c.get();
s = e.getMessage();
System.out.println(s);
}
assertEquals(s,
"Tried to use reference after it was released");
}
ConnectionPoolProxyDemo.class);
}
} ///:~
Dynamic Proxies
In JDK 1.3, the Dynamic Proxy was introduced. Although a little complex at rst,
this is an intruiging tool.
Here's an interesting little starting example, which works and proves that yes,
indeed, the invocation handler is being called so the proxying etc. is actually
working. So it's pretty cool, and it's in my head now, but I still have to gure out
something reasonable to do with the invocation handler to come up with a useful
example...
// proxy:DynamicProxyDemo.java
// Broken in JDK 1.4.1_01
package proxy;
import java.lang.reect.*;
interface Foo {
void f(String s);
void g(int i);
String h(int i, String s);
}
public class DynamicProxyDemo {
public static void main(String[] clargs) {
Foo prox = (Foo)Proxy.newProxyInstance(
Foo.class.getClassLoader(),
new Class[]{ Foo.class },
new InvocationHandler() {
public Object invoke(

Object proxy, Method method,
Object[] args) {
System.out.println(
"InvocationHandler called:" +
"\n\tMethod = " + method);
if (args != null) {
System.out.println("\targs = ");
for (int i = 0; i < args.length; i++)
System.out.println("\t\t" + args[i]);
}
return null;
}
});
prox.f("hello");
prox.g(47);
prox.h(47, "hello");
}
} ///:~
Exercise: Use the Java dynamic proxy to create an object that acts as a front end
for a simple conguration le. For example, in good_stu.txt you can have entries
like this:
a=1
b=2
c="Hello World"
A client programmer of this NeatPropertyBundle could then write:

NeatPropertyBundle p =
new NeatPropertyBundle("good_stu");
System.out.println(p.a);
System.out.println(p.b);
System.out.println(p.c);
The contents of the conguration le can contain anything, with any variable
names. The dynamic proxy will either map to the name or tell you it doesnt exist
somehow (probably by returning null). If you set a property and it doesnt already
exist, the dynamic proxy will create the new entry. The toString( ) method should
display all the current entries.
Exercise: similar to the previous exercise, use the Java dynamic proxy to make a
connection to the DOS Autoexec.bat le.
Exercise: Accept an SQL query which returns data, then read the DB metadata.
Now, for each record, provide an object which has attributes corresponding to the
column names and of appropriate data types.
Exercise: Create a simple server and client that uses XML-RPC. Each object the
client returns should use the dynamic proxy concept to exercise the remote
methods.
Andrea writes:
I'm not sure about the exercises you suggest, except for the last one. The thing is
that I like to think at invocation handler as somthing providing features that are
orthogonal to the ones provided by the object being "proxied".
In other words: the implementation of the invocation handler is completely
independent from the interface(s) of the object that the dynamically-generated
proxy represent. Which means that once you have implemented an invocation
handler, you can use for any class that exposes interfaces, even for classes and
interfaces that were not present when the handler was implemented.
That's why I say the the handler provides services that are orthogonal to the
ones provided by the proxied object. Rickard has a few handlers in his
SmartWorld example, and they one I like the best is a call-retry handler. It
basically makes a call into the actual object, and if the call generates an
exception if waits for a while, then makes the same call again for a total of three
times. If all three calls fails, it returns an exception. And you can use such a
handler on _any_ class.
The implementation is way too complex for what you are trying to demonstrate.
I'm using this example just to explain what I mean by orthogonal services.
In your list of exercises, the only one that, in my opinion, makes sense to
implement using dynamic proxies is the last one, the one using XML-RPC to
communicate with an object. And that's because the mechanism you use to
dispatch the message (XML-RPC) is orthogonal to the services provided by the
object you want to reach.
State: changing object behavior

An object that appears to change its class.
Indications: conditional code in most or all methods.
The State pattern switches from one implementation to another during the lifetime
of the surrogate, in order to produce dierent behavior from the same method
call(s). Its a way to improve the implementation of your code when you seem to be
doing a lot of testing inside each of your methods before deciding what to do for
that method. For example, the fairy tale of the frog-prince contains an object (the
creature) that behaves dierently depending on what state its in. You could
implement this using a boolean that you test:
//: state:KissingPrincess.java
package state;
class Creature {
private boolean isFrog = true;
public void greet() {
if(isFrog)
System.out.println("Ribbet!");
else
System.out.println("Darling!");
}
public void kiss() { isFrog = false; }
}
public class KissingPrincess extends TestCase {

Creature creature = new Creature();
creature.greet();
creature.kiss();
creature.greet();
}
junit.textui.TestRunner.run(KissingPrincess.class);
}
} ///:~
However, the greet() method, and any other methods that must test isFrog before
they perform their operations, ends up with awkward code. By delegating the
operations to a State object that can be changed, this code is simplied.
//: state:KissingPrincess2.java
package state;
class Creature {
private interface State {
String response();
}
private class Frog implements State {
public String response() { return "Ribbet!"; }
}
private class Prince implements State {
public String response() { return "Darling!"; }
}
private State state = new Frog();
public void greet() {
System.out.println(state.response());
}
public void kiss() { state = new Prince(); }
}
public class KissingPrincess2 extends TestCase {
Creature creature = new Creature();
creature.greet();
creature.kiss();
creature.greet();
}
junit.textui.TestRunner.run(KissingPrincess2.class);
}
} ///:~
In addition, changes to the State are automatically propagated throughout, rather

than requiring an edit across the class methods in order to eect changes.
Heres the basic structure of State:
//: state:StateDemo.java
// Simple demonstration of the State pattern.

package state;
interface State {
void operation1();
void operation2();
void operation3();
}
class ServiceProvider {
private State state;
public ServiceProvider(State state) {
this.state = state;
}
public void changeState(State newState) {
state = newState;
}
// Pass method calls to the implementation:
public void service1() {
// ...
state.operation1();
// ...
state.operation3();
}
// ...
state.operation1();
// ...
state.operation2();
}
// ...
state.operation3();
// ...
state.operation2();
}
}
class Implementation1 implements State {
public void operation1() {
System.out.println("Implementation1.operation1()");
}
}
}
}
class Implementation2 implements State {
}

}
}
}
public class StateDemo extends TestCase {
static void run(ServiceProvider sp) {
sp.service1();
sp.service2();
sp.service3();
}
ServiceProvider sp =
new ServiceProvider(new Implementation1());
run(sp);
sp.changeState(new Implementation2());
run(sp);
}
junit.textui.TestRunner.run(StateDemo.class);
}
} ///:~
In main( ), you can see that the rst implementation is used for a bit, then the
second implementation is swapped in and that is used.
There are a number of details that are choices that you must make according to the
needs of your own implementation, such as whether the fact that you are using
State is exposed to the client, and how the changes to State are made. Sometimes
(as in the Swing LayoutManager) the client may pass in the object directly, but in
KissingPrincess2.java the fact that State is used is invisible to the client. In addition,
the mechanism for changing state may be simple or complex in State Machine,
described later in this book, larger sets of states and dierent mechanisms for
changing are explored.
The Swing LayoutManager example mentioned above is an interesting example
because it show behavior of both Strategy and State.
The dierence between Proxy and State is in the problems that are solved. The
common uses for Proxy as described in Design Patterns are:
1.
Remote proxy. This proxies for an object in a dierent address space. A

remote proxy is created for you automatically by the RMI compiler rmic as it
creates stubs and skeletons.
2.
Virtual proxy. This provides lazy initialization to create expensive objects

on demand.
3.
Protection proxy. Used when you dont want the client programmer to have
full access to the proxied object.
4.
Smart reference. To add additional actions when the proxied object is

accessed. For example, or to keep track of the number of references that are
held for a particular object, in order to implement the copy-on-write idiom and
prevent object aliasing. A simpler example is keeping track of the number of
calls to a particular method.
You could look at a Java reference as a kind of protection proxy, since it controls
access to the actual object on the heap (and ensures, for example, that you dont
use a null reference).
[[ Rewrite this: In Design Patterns, Proxy and State are not seen as related to each
other because the two are given (what I consider arbitrarily) dierent structures.
State, in particular, uses a separate implementation hierarchy but this seems to me
to be unnecessary unless you have decided that the implementation is not under
your control (certainly a possibility, but if you own all the code there seems to be no
reason not to benet from the elegance and helpfulness of the single base class). In
addition, Proxy need not use the same base class for its implementation, as long as
the proxy object is controlling access to the object it fronting for. Regardless of
the specics, in both Proxy and State a surrogate is passing method calls through to
an implementation object.]]]
State can be found everywhere because its such a fundamental idea. For example,
in Builder, the Director uses a backend Builder object to produce dierent
behaviors.
Iterators: decoupling algorithms

from containers
Alexander Stepanov thought for years about the problem of
generic programming techniques before creating the STL (along
with Dave Musser). He came to the conclusion that all algorithms
are dened on algebraic structures what we would call
containers.
In the process, he realized that iterators are central to the use of algorithms,
because they decouple the algorithms from the specic type of container that the
algorithm might currently be working with. This means that you can describe the
algorithm without worrying about the particular sequence it is operating on. More
generally, any code that you write using iterators is decoupled from the data
structure that the code is manipulating, and thus your code is more general and
reusable.
The use of iterators also extends your code into the realm of functional
programming, whose objective is to describe what a program is doing at every step
rather than how it is doing it. That is, you say sort rather than describing the sort.
The objective of the C++ STL was to provide this generic programming approach
for C++ (how successful this approach will actually be remains to be seen).
If youve used containers in Java (and its hard to write code without using them),
youve used iterators in the form of the Enumeration in Java 1.0/1.1 and the
Iterator in Java 2. So you should already be familiar with their general use. If not,
see Chapter 9, Holding Your Objects, under Iterators in Thinking in Java, 2 nd edition
(freely downloadable from www.BruceEckel.com).
Because the Java 2 containers rely heavily on iterators they become excellent
candidates for generic/functional programming techniques. This chapter will
explore these techniques by converting the STL algorithms to Java, for use with the
Java 2 container library.
Type-safe iterators
In Thinking in Java, 2nd edition, I show the creation of a type-safe container that will
only accept a particular type of object. A reader, Linda Pazzaglia, asked for the
other obvious type-safe component, an iterator that would work with the basic
java.util containers, but impose the constraint that the type of objects that it
iterates over be of a particular type.
If Java ever includes a template mechanism, this kind of iterator will have the added
advantage of being able to return a specic type of object, but without templates
you are forced to return generic Objects, or to require a bit of hand-coding for
every type that you want to iterate through. I will take the former approach.
A second design decision involves the time that the type of object is determined.
One approach is to take the type of the rst object that the iterator encounters, but
this is problematic because the containers may rearrange the objects according to
an internal ordering mechanism (such as a hash table) and thus you may get
dierent results from one iteration to the next. The safe approach is to require the
user to establish the type during construction of the iterator.
Lastly, how do we build the iterator? We cannot rewrite the existing Java library
classes that already produce Enumerations and Iterators. However, we can use
the Decorator design pattern, and create a class that simply wraps the
Enumeration or Iterator that is produced, generating a new object that has the
iteration behavior that we want (which is, in this case, to throw a
RuntimeException if an incorrect type is encountered) but with the same interface
as the original Enumeration or Iterator, so that it can be used in the same places
(you may argue that this is actually a Proxy pattern, but its more likely Decorator
because of its intent). Here is the code:
//: com:bruceeckel:util:TypedIterator.java
package com.bruceeckel.util;
import java.util.*;
public class TypedIterator implements Iterator {
private Iterator imp;
private Class type;
public TypedIterator(Iterator it, Class type) {
imp = it;
this.type = type;
}
public boolean hasNext() {
return imp.hasNext();
}
public void remove() { imp.remove(); }
public Object next() {
Object obj = imp.next();
if(!type.isInstance(obj))
throw new ClassCastException(
"TypedIterator for type " + type +
" encountered type: " + obj.getClass());
return obj;
}
} ///:~
Exercises
1.
Create an example of the virtual proxy.
2.
Create an example of the Smart reference proxy where you keep count of
the number of method calls to a particular object.
3.
Create a program similar to certain DBMS systems that only allow a certain
number of connections at any time. To implement this, use a singleton-like
system that controls the number of connection objects that it creates. When
a user is nished with a connection, the system must be informed so that it
can check that connection back in to be reused. To guarantee this, provide a
proxy object instead of a reference to the actual connection, and design the
proxy so that it will cause the connection to be released back to the system.
4.
Using the State pattern, make a class called UnpredictablePerson which

changes the kind of response to its hello( ) method depending on what kind
of Mood its in. Add an additional kind of Mood called Prozac.
5.
Create a simple copy-on write implementation.
6.
The java.util.Map has no way to automatically load a two-dimensional array

of objects into a Map as key-value pairs. Create an adapter class that does
this.
7.
Create an Adapter Factory that dynamically nds and produces the adapter
that you need to connect a given object to a desired interface.
8.
Solve the above exercise using the dynamic proxy thats part of the Java
standard library.
9.
Modify the Object Pool solution so that the objects are returned to the pool
automatically after a certain amount of time.
10.
11.
Modify the above solution to use leasing so that the client can renew the
lease on the object to prevent it from being automatically released by the
timer.
Modify the Object Pool system to take threading issues into account.
Factoring commonality
Applying the once and only once principle produces the most
basic pattern of putting code that changes into a method.
This can be expressed two ways:
Strategy: choosing the algorithm

at run-time
Strategy also adds a Context which can be a surrogate class that controls the
selection and use of the particular strategy objectjust like State! Heres what it
looks like:
//: strategy:StrategyPattern.java
package strategy;
import com.bruceeckel.util.*; // Arrays2.toString()
// The strategy interface:
interface FindMinima {
// Line is a sequence of points:
double[] algorithm(double[] line);
}
// The various strategies:
class LeastSquares implements FindMinima {
public double[] algorithm(double[] line) {
return new double[] { 1.1, 2.2 }; // Dummy
}
}
class NewtonsMethod implements FindMinima {
}
}
class Bisection implements FindMinima {
}
}
class ConjugateGradient implements FindMinima {
}
}
// The "Context" controls the strategy:
class MinimaSolver {
private FindMinima strategy;
public MinimaSolver(FindMinima strat) {
strategy = strat;
}
double[] minima(double[] line) {
return strategy.algorithm(line);
}
void changeAlgorithm(FindMinima newAlgorithm) {
strategy = newAlgorithm;
}
}
public class StrategyPattern extends TestCase {
MinimaSolver solver =
new MinimaSolver(new LeastSquares());
double[] line = {
1.0, 2.0, 1.0, 2.0, -1.0,
3.0, 4.0, 5.0, 4.0 };
System.out.println(
Arrays2.toString(solver.minima(line)));
solver.changeAlgorithm(new Bisection());
System.out.println(
Arrays2.toString(solver.minima(line)));
}
junit.textui.TestRunner.run(StrategyPattern.class);
}
} ///:~
Note similarity with template method TM claims distinction that it has more than
one method to call, does things piecewise. However, its not unlikely that strategy
object would have more than one method call; consider Shalloways order
fulfullment system with country information in each strategy.
Strategy example from JDK: comparator objects.
Policy: generalized strategy

Although GoF says that Policy is just another name for strategy, their use of
Strategy implicitly assumes a single method in the strategy object that youve
broken out your changing algorithm as a single piece of code.
[6]
Others
use Policy to mean an object that has multiple methods that may vary
independently from class to class. This gives more exibility than being restricted to
a single method.
For example, a shipping policy for a product can be used to describe shipping issues
for sending a package to various dierent countries. This may include the available
methods of shipping, how to calculate postage or shipping cost, customs
requirements and fees, and special handling costs. All these things may vary
independently of each other, and more importantly you may need the information
from each at dierent points in the shipping process.
It also seems generally useful to distinguish Strategies with single methods from
Policies with multiple methods.
Template method
An application framework allows you to inherit from a class or set of classes and
create a new application, reusing most of the code in the existing classes and
overriding one or more methods in order to customize the application to your needs.
A fundamental concept in the application framework is the Template Method which
is typically hidden beneath the covers and drives the application by calling the
various methods in the base class (some of which you have overridden in order to
create the application).
For example, whenever you create an applet youre using an application framework:
you inherit from JApplet and then override init( ). The applet mechanism (which is
a Template Method) does the rest by drawing the screen, handling the event loop,
resizing, etc.
An important characteristic of the Template Method is that it is dened in the base
class and cannot be changed. Its sometimes a private method but its virtually
always nal. It calls other base-class methods (the ones you override) in order to do
its job, but it is usually called only as part of an initialization process (and thus the
client programmer isnt necessarily able to call it directly).
//: templatemethod:TemplateMethod.java
// Simple demonstration of Template Method.
package templatemethod;
abstract class ApplicationFramework {
public ApplicationFramework() {
templateMethod(); // Dangerous!
}
abstract void customize1();
abstract void customize2();
nal void templateMethod() {
for(int i = 0; i < 5; i++) {
customize1();
customize2();
}
}
}
// Create a new "application":
class MyApp extends ApplicationFramework {
void customize1() {
System.out.print("Hello ");
}
void customize2() {
System.out.println("World!");
}
}
public class TemplateMethod extends TestCase {

MyApp app = new MyApp();
// The MyApp constructor does all the work.
}
junit.textui.TestRunner.run(TemplateMethod.class);
}
} ///:~
The base-class constructor is responsible for performing the necessary initialization

and then starting the engine (the template method) that runs the application (in a
GUI application, this engine would be the main event loop). The client
programmer simply provides denitions for customize1( ) and customize2( ) and
the application is ready to run.
Exercises
1.
Create a framework that takes a list of le names on the command line. It

opens each le except the last for reading, and the last for writing. The
framework will process each input le using an undetermined policy and
write the output to the last le. Inherit to customize this framework to create
two separate applications:
1) Converts all the letters in each le to uppercase.
2) Searches the les for words given in the rst le.
Encapsulating creation
When you discover that you need to add new types to a system, the most sensible
rst step is to use polymorphism to create a common interface to those new types.
This separates the rest of the code in your system from the knowledge of the
specic types that you are adding. New types may be added without disturbing
existing code or so it seems. At rst it would appear that the only place you need
to change the code in such a design is the place where you inherit a new type, but
this is not quite true. You must still create an object of your new type, and at the
point of creation you must specify the exact constructor to use. Thus, if the code
that creates objects is distributed throughout your application, you have the same
problem when adding new typesyou must still chase down all the points of your
code where type matters. It happens to be the creation of the type that matters in
this case rather than the use of the type (which is taken care of by polymorphism),
but the eect is the same: adding a new type can cause problems.
The solution is to force the creation of objects to occur through a common factory
rather than to allow the creational code to be spread throughout your system. If all
the code in your program must go through this factory whenever it needs to create
one of your objects, then all you must do when you add a new object is to modify the
factory.
Since every object-oriented program creates objects, and since its very likely you
will extend your program by adding new types, I suspect that factories may be the
most universally useful kinds of design patterns.
Although only the Simple Factory Method is a true singleton, youll nd that each
specify factory class in the more general types of factories will only have a single
instance.
Simple Factory method

As an example, lets revisit the Shape system.
One approach is to make the factory a static method of the base class:
//: factory:shapefact1:ShapeFactory1.java
// A simple static factory method.
package factory.shapefact1;
import java.util.*;
abstract class Shape {
public abstract void draw();
public abstract void erase();
public static Shape factory(String type) {
if(type.equals("Circle")) return new Circle();
if(type.equals("Square")) return new Square();
"Bad shape creation: " + type);
}
}
class Circle extends Shape {
Circle() {} // Package-access constructor
public void draw() {
System.out.println("Circle.draw");
}
public void erase() {
System.out.println("Circle.erase");
}
}
class Square extends Shape {
Square() {} // Package-access constructor
System.out.println("Square.draw");
}
System.out.println("Square.erase");
}
}
public class ShapeFactory1 extends TestCase {
String shlist[] = { "Circle", "Square",
"Square", "Circle", "Circle", "Square" };
List shapes = new ArrayList();

Iterator it = Arrays.asList(shlist).iterator();
while(it.hasNext())
shapes.add(Shape.factory((String)it.next()));
it = shapes.iterator();
while(it.hasNext()) {
Shape s = (Shape)it.next();
s.draw();
s.erase();
}
}
junit.textui.TestRunner.run(ShapeFactory1.class);
}
} ///:~
The factory( ) takes an argument that allows it to determine what type of Shape to
create; it happens to be a String in this case but it could be any set of data. The
factory( ) is now the only other code in the system that needs to be changed when
a new type of Shape is added (the initialization data for the objects will presumably
come from somewhere outside the system, and not be a hard-coded array as in the
above example).
To encourage creation to only happen in the factory( ), the constructors for the
specic types of Shape are give package access, so factory( ) has access to the
constructors but they are not available outside the package.
Polymorphic factories
The static factory( ) method in the previous example forces all the creation
operations to be focused in one spot, so thats the only place you need to change the
code. This is certainly a reasonable solution, as it throws a box around the process
of creating objects. However, the Design Patterns book emphasizes that the reason
for the Factory Method pattern is so that dierent types of factories can be
subclassed from the basic factory (the above design is mentioned as a special case).
However, the book does not provide an example, but instead just repeats the
example used for the Abstract Factory (youll see an example of this in the next
section). Here is ShapeFactory1.java modied so the factory methods are in a
separate class as virtual functions. Notice also that the specic Shape classes are
dynamically loaded on demand:
//: factory:shapefact2:ShapeFactory2.java
// Polymorphic factory methods.
package factory.shapefact2;
import java.util.*;
interface Shape {
void draw();
void erase();
}
abstract class ShapeFactory {

protected abstract Shape create();
private static Map factories = new HashMap();
public static void
addFactory(String id, ShapeFactory f) {
factories.put(id, f);
}
// A Template Method:
public static nal
Shape createShape(String id) {
if(!factories.containsKey(id)) {
try {
// Load dynamically
Class.forName("factory.shapefact2." + id);
} catch(ClassNotFoundException e) {
"Bad shape creation: " + id);
}
// See if it was put in:
if(!factories.containsKey(id))
"Bad shape creation: " + id);
}
return
((ShapeFactory)factories.get(id)).create();
}
}
class Circle implements Shape {
private Circle() {}
System.out.println("Circle.draw");
}
System.out.println("Circle.erase");
}
private static class Factory
extends ShapeFactory {
protected Shape create() {
return new Circle();
}
}
static {
ShapeFactory.addFactory(
"Circle", new Factory());
}
}
class Square implements Shape {
private Square() {}
System.out.println("Square.draw");
}
System.out.println("Square.erase");
}
private static class Factory
extends ShapeFactory {
protected Shape create() {
return new Square();
}
}
static {
ShapeFactory.addFactory(
"Square", new Factory());
}
}
public class ShapeFactory2 extends TestCase {
String shlist[] = { "Circle", "Square",
"Square", "Circle", "Circle", "Square" };
List shapes = new ArrayList();
Iterator it = Arrays.asList(shlist).iterator();
while(it.hasNext())
shapes.add(
ShapeFactory.createShape((String)it.next()));
it = shapes.iterator();
Shape s = (Shape)it.next();
s.draw();
s.erase();
}
}
junit.textui.TestRunner.run(ShapeFactory2.class);
}
} ///:~
Now the factory method appears in its own class, ShapeFactory, as the create( )
method. This is a protected method which means it cannot be called directly, but it
can be overridden. The subclasses of Shape must each create their own subclasses
of ShapeFactory and override the create( ) method to create an object of their
own type. The actual creation of shapes is performed by calling
ShapeFactory.createShape( ), which is a static method that uses the Map in
ShapeFactory to nd the appropriate factory object based on an identier that you
pass it. The factory is immediately used to create the shape object, but you could
imagine a more complex problem where the appropriate factory object is returned
and then used by the caller to create an object in a more sophisticated way.
However, it seems that much of the time you dont need the intricacies of the
polymorphic factory method, and a single static method in the base class (as shown
in ShapeFactory1.java) will work ne.
Notice that the ShapeFactory must be initialized by loading its Map with factory
objects, which takes place in the static initialization clause of each of the Shape
implementations. So to add a new type to this design you must inherit the type,
create a factory, and add the static initialization clause to load the Map. This extra
complexity again suggests the use of a static factory method if you dont need to
create individual factory objects.
Abstract factories
The Abstract Factory pattern looks like the factory objects weve seen previously,
with not one but several factory methods. Each of the factory methods creates a
dierent kind of object. The idea is that at the point of creation of the factory object,
you decide how all the objects created by that factory will be used. The example
given in Design Patterns implements portability across various graphical user
interfaces (GUIs): you create a factory object appropriate to the GUI that youre
working with, and from then on when you ask it for a menu, button, slider, etc. it
will automatically create the appropriate version of that item for the GUI. Thus
youre able to isolate, in one place, the eect of changing from one GUI to another.
As another example suppose you are creating a general-purpose gaming
environment and you want to be able to support dierent types of games. Heres
how it might look using an abstract factory:
//: factory:Games.java
// An example of the Abstract Factory pattern.
package factory;
interface Obstacle {
void action();
}
interface Player {
void interactWith(Obstacle o);
}
class Kitty implements Player {
public void interactWith(Obstacle ob) {
System.out.print("Kitty has encountered a ");
ob.action();
}
}
class KungFuGuy implements Player {
public void interactWith(Obstacle ob) {
System.out.print("KungFuGuy now battles a ");
ob.action();
}
}
class Puzzle implements Obstacle {
public void action() {
System.out.println("Puzzle");
}
}
class NastyWeapon implements Obstacle {
public void action() {
System.out.println("NastyWeapon");
}
}
// The Abstract Factory:
interface GameElementFactory {
Player makePlayer();
Obstacle makeObstacle();
}
// Concrete factories:
class KittiesAndPuzzles
implements GameElementFactory {
public Player makePlayer() {
return new Kitty();
}
public Obstacle makeObstacle() {
return new Puzzle();
}
}
class KillAndDismember
implements GameElementFactory {
public Player makePlayer() {
return new KungFuGuy();
}
public Obstacle makeObstacle() {
return new NastyWeapon();
}
}
class GameEnvironment {
private GameElementFactory gef;
private Player p;
private Obstacle ob;
public GameEnvironment(
GameElementFactory factory) {
gef = factory;
p = factory.makePlayer();
ob = factory.makeObstacle();
}
public void play() { p.interactWith(ob); }
}
public class Games extends TestCase {
GameElementFactory
kp = new KittiesAndPuzzles(),
kd = new KillAndDismember();
GameEnvironment
g1 = new GameEnvironment(kp),
g2 = new GameEnvironment(kd);
// These just ensure no exceptions are thrown:
public void test1() { g1.play(); }
public void test2() { g2.play(); }
junit.textui.TestRunner.run(Games.class);
}
} ///:~
In this environment, Player objects interact with Obstacle objects, but there are
dierent types of players and obstacles depending on what kind of game youre
playing. You determine the kind of game by choosing a particular
GameElementFactory, and then the GameEnvironment controls the setup and
play of the game. In this example, the setup and play is very simple, but those
activities (the initial conditions and the state change) can determine much of the
games outcome. Here, GameEnvironment is not designed to be inherited,
although it could very possibly make sense to do that.
This also contains examples of Double Dispatching and the Factory Method, both of
which will be explained later.
Exercises
1.
Add a class Triangle to ShapeFactory1.java
2.
Add a class Triangle to ShapeFactory2.java
3.
Add a new type of GameEnvironment called GnomesAndFairies to

Games.java
4.
Modify ShapeFactory2.java so that it uses an Abstract Factory to create

dierent sets of shapes (for example, one particular type of factory object
creates thick shapes, another creates thin shapes, but each factory object
can create all the shapes: circles, squares, triangles etc.).
Specialized creation
Prototype
Objects are created by cloning a prototypical instance. An example of this appears
in the Pattern Refactoring chapter.
Builder
The goal of builder is to separate the construction from the representation, to
allow multiple dierent representations. The construction process stays the same,
but the resulting object has dierent possible representations. GoF points out that
the main dierence with Abstract Factory is that a Builder creates the object
step-by-step, so the fact that the creation process is spread out in time seems to be
important. In addition, it seems that the director gets a stream of pieces that it
passes to the Builder, and each piece is used to perform one of the steps in the build
process.
One example given in GoF is that of a text format converter. The incoming format is
RTF, and once it is parsed the directives are passed to the text converter, which may
be implemented in dierent ways depending on whether the resulting format is
ASCII, TeX, or a GUI Text Widget. Although the resulting object (the entire
converted text le) is created over time, if you consider the conversion of each RTF
directive to be an object, this feels to me a little more like Bridge, because the
specic types of converters extend the interface of the base class. Also, the general
solution to the problem would allow multiple readers on the front end and
multiple converters on the back end, which is a primary characteristic of Bridge.
To me, the fact that Builder has multiple steps in creating an object, and those steps
are accessed externally to the Builder object, is the essence of what distinguishes it
(structurally, anyway) from a regular factory. However, GoF emphasizes that youre
able to create dierent representations using the same process. They never dene
exactly what they mean by representation. (Does the representation involve an
object that is too large? Would the need for Builder vanish if the representation was
broken into smaller objects?)
The other example in GoF creates a maze object and adds rooms within the maze
and doors within the rooms. Thus it is a multistep process, but alas, the dierent
representations are the Standard and Complex mazes not really dierent
kinds of mazes, but instead dierent complexity. I think I would have tried to create
one maze builder that could handle arbitrarily complex mazes. The nal variation of
the maze builder is something that doesnt create mazes at all, but instead counts
the rooms in an existing maze.
Neither the RTF converter nor the Mazebuilder example makes an overwhelmingly
compelling case for Builder. Readers have suggested that the output of the Sax XML
parser, and standard compiler parsers, might naturally be fed into a Builder.
Heres an example that may be a little more compelling, or at least give more of an
idea of what Builder is trying to do. Media may be constructed into dierent
representations, in this case books, magazines and web sites. The example argues
that the steps involved are the same, and so can be abstracted into the director
class.
//: builder:BuildMedia.java
// Example of the Builder pattern
package builder;
import java.util.*;
// Dierent "representations" of media:
class Media extends ArrayList {}
class Book extends Media {}
class Magazine extends Media {}
class WebSite extends Media {}
// ... contain dierent kinds of media items:
class MediaItem {
private String s;
public MediaItem(String s) { this.s = s; }
public String toString() { return s; }
}
class Chapter extends MediaItem {
public Chapter(String s) { super(s); }
}
class Article extends MediaItem {

public Article(String s) { super(s); }
}
class WebItem extends MediaItem {
public WebItem(String s) { super(s); }
}
// ... but use the same basic construction steps:
class MediaBuilder {
public void buildBase() {}
public void addMediaItem(MediaItem item) {}
public Media getFinishedMedia() { return null; }
}
class BookBuilder extends MediaBuilder {
private Book b;
public void buildBase() {
System.out.println("Building book framework");
b = new Book();
}
public void addMediaItem(MediaItem chapter) {
System.out.println("Adding chapter " + chapter);
b.add(chapter);
}
public Media getFinishedMedia() { return b; }
}
class MagazineBuilder extends MediaBuilder {
private Magazine m;
System.out.println("Building magazine framework");
m = new Magazine();
}
public void addMediaItem(MediaItem article) {
System.out.println("Adding article " + article);
m.add(article);
}
public Media getFinishedMedia() { return m; }
}
class WebSiteBuilder extends MediaBuilder {
private WebSite w;
System.out.println("Building web site framework");
w = new WebSite();
}
public void addMediaItem(MediaItem webItem) {
System.out.println("Adding web item " + webItem);
w.add(webItem);
}
public Media getFinishedMedia() { return w; }
}
class MediaDirector { // a.k.a. "Context"
private MediaBuilder mb;
public MediaDirector(MediaBuilder mb) {

this.mb = mb; // Strategy-ish
}
public Media produceMedia(List input) {
mb.buildBase();
for(Iterator it = input.iterator(); it.hasNext();)
mb.addMediaItem((MediaItem)it.next());
return mb.getFinishedMedia();
}
};
public class BuildMedia extends TestCase {
private List input = Arrays.asList(new MediaItem[] {
new MediaItem("item1"), new MediaItem("item2"),
new MediaItem("item3"), new MediaItem("item4"),
});
public void testBook() {
MediaDirector buildBook =
new MediaDirector(new BookBuilder());
Media book = buildBook.produceMedia(input);
String result = "book: " + book;
"book: [item1, item2, item3, item4]");
}
public void testMagazine() {
MediaDirector buildMagazine =
new MediaDirector(new MagazineBuilder());
Media magazine = buildMagazine.produceMedia(input);
String result = "magazine: " + magazine;
"magazine: [item1, item2, item3, item4]");
}
public void testWebSite() {
MediaDirector buildWebSite =
new MediaDirector(new WebSiteBuilder());
Media webSite = buildWebSite.produceMedia(input);
String result = "web site: " + webSite;
"web site: [item1, item2, item3, item4]");
}
junit.textui.TestRunner.run(BuildMedia.class);
}
} ///:~
Note that in some ways this could be seen as a more complicated State pattern,
since the behavior of the director depends on what type of builder you use. Instead
of simply forwarding the requests through to the underlying State object, however,
the director has a sequence of operations to perform, and it uses the State object as
a Policy to fulll its job. Thus, Builder could be described as using a Policy to create
objects.
Exercises
1.
Break a text le up into an input stream of words (consider using regular

expressions for this). Create one Builder that puts the words into a
java.util.TreeSet, and another that produces a java.util.HashMap
containing words and occurrences of those words (that is, it does a word
count).
Too many
Flyweight: too many objects
The odd thing about yweight, in the company of the other design patterns, is that
its a performance hack. Its generally ideal to simply make an object for every item
in your system, but some problems generate a prohibitive number of objects, which
may result in excessive slowness or running out of memory.
Flyweight solves this problem by reducing the number of objects. To do this, you
externalize some of the data in an object, so that you can pretend that you have
more objects than you really do. However, this adds complexity to the interface for
using such objects, because you must pass in additional information to method calls
in order to tell the method how to nd the externalized information.
As a very simple example, consider a DataPoint object that holds an int, a oat,
and an id that carries the object number. Suppose you need to create a million of
these objects, and then manipulate them, like so:
//: yweight:ManyObjects.java
class DataPoint {
private static int count = 0;
private int id = count++;
private int i;
private oat f;
public int getI() { return i; }
public void setI(int i) { this.i = i; }
public oat getF() { return f; }
public void setF(oat f) { this.f = f; }
public String toString() {
return "id: " + id + ", i = " + i + ", f = " + f;
}
}
public class ManyObjects {
static nal int size = 1000000;
DataPoint[] array = new DataPoint[size];
for(int i = 0; i < array.length; i++)
array[i] = new DataPoint();
for(int i = 0; i < array.length; i++) {

DataPoint dp = array[i];
dp.setI(dp.getI() + 1);
dp.setF(47.0f);
}
System.out.println(array[size -1]);
}
} ///:~
Depending on your computer, this program may take several seconds to run. More
complex objects and more involved operations may cause the overhead to become
untenable. To solve the problem the DataPoint can be reduced from a million
objects to one object by externalizing the data held in the DataPoint:
//: yweight:FlyWeightObjects.java
class ExternalizedData {
static nal int size = 5000000;
static int[] id = new int[size];
static int[] i = new int[size];
static oat[] f = new oat[size];
static {
for(int i = 0; i < size; i++)
id[i] = i;
}
}
class FlyPoint {
private FlyPoint() {}
public static int getI(int obnum) {
return ExternalizedData.i[obnum];
}
public static void setI(int obnum, int i) {
ExternalizedData.i[obnum] = i;
}
public static oat getF(int obnum) {
return ExternalizedData.f[obnum];
}
public static void setF(int obnum, oat f) {
ExternalizedData.f[obnum] = f;
}
public static String str(int obnum) {
return "id: " +
ExternalizedData.id[obnum] +
", i = " +
ExternalizedData.i[obnum] +
", f = " +
ExternalizedData.f[obnum];
}
}
public class FlyWeightObjects {
for(int i = 0; i < ExternalizedData.size; i++) {
FlyPoint.setI(i, FlyPoint.getI(i) + 1);
FlyPoint.setF(i, 47.0f);
}
System.out.println(
FlyPoint.str(ExternalizedData.size -1));
}
} ///:~
Since all the data is now in ExternalizedData, each call to a FlyPoint method must
include the index into ExternalizedData. For consistency, and to remind the reader
of the similarity with the implicit this pointer in method calls, the this index is
passed in as the rst argument.
Naturally, its worth repeating admonishments against premature optimization.
First make it work, then make it fast if you have to. Also, a proler is the tool to
use for discovering performance bottlenecks, not guesswork.
Decorator: too many classes

The use of layered objects to dynamically and transparently add
responsibilities to individual objects is referred to as the
decorator pattern.
Used when subclassing creates too many (& inexible) classes
All decorators that wrap around the original object must have the same basic
interface
Dynamic proxy/surrogate?
This accounts for the odd inheritance structure
Tradeo: coding is more complicated when using decorators
Basic decorator structure
A coee example
Consider going down to the local coee shop, BeanMeUp, for a coee. There are
typically many dierent drinks on oer -- espressos, lattes, teas, iced coees, hot
chocolate to name a few, as well as a number of extras (which cost extra too) such
as whipped cream or an extra shot of espresso. You can also make certain changes
to your drink at no extra cost, such as asking for decaf coee instead of regular
coee.
Quite clearly if we are going to model all these drinks and combinations, there will
be sizeable class diagrams. So for clarity we will only consider a subset of the
coees: Espresso, Espresso Con Panna, Caf Late, Cappuccino and Caf Mocha.
We'll include 2 extras - whipped cream ("whipped") and an extra shot of espresso;
and three changes - decaf, steamed milk ("wet") and foamed milk ("dry").
Class for each combination

One solution is to create an individual class for every combination. Each class
describes the drink and is responsible for the cost etc. The resulting menu is huge,
and a part of the class diagram would look something like this:
Here is one of the combinations, a simple implementation of a Cappuccino:

class Cappuccino {
private oat cost = 1;
private String description = "Cappucino";
public oat getCost() {
return cost;
}
public String getDescription() {
return description;
}
}
The key to using this method is to nd the particular combination you want. So,
once you've found the drink you would like, here is how you would use it, as shown
in the CoeeShop class in the following code:
//: decorator:nodecorators:CoeeShop.java
// Coee example with no decorators
package decorator.nodecorators;
class Espresso {}
class DoubleEspresso {}
class EspressoConPanna {}
class Cappuccino {
private String description = "Cappucino";
return cost;
}
return description;
}
}
class CappuccinoDecaf {}
class CappuccinoDecafWhipped {}
class CappuccinoDry {}
class CappuccinoDryWhipped {}
class CappuccinoExtraEspresso {}
class CappuccinoExtraEspressoWhipped {}
class CappuccinoWhipped {}
class CafeMocha {}
class CafeMochaDecaf {}
class CafeMochaDecafWhipped {
private oat cost = 1.25f;
private String description =
"Cafe Mocha decaf whipped cream";
return cost;
}
return description;
}
}
class CafeMochaExtraEspresso {}
class CafeMochaExtraEspressoWhipped {}
class CafeMochaWet {}
class CafeMochaWetWhipped {}
class CafeMochaWhipped {}
class
class
class
class
class
class
class
class
CafeLatte {}
CafeLatteDecaf {}
CafeLatteDecafWhipped {}
CafeLatteExtraEspresso {}
CafeLatteExtraEspressoWhipped {}
CafeLatteWet {}
CafeLatteWetWhipped {}
CafeLatteWhipped {}
public class CoeeShop extends TestCase {

public void testCappuccino() {
// Create a plain cappuccino

Cappuccino cappuccino = new Cappuccino();
System.out.println(cappuccino.getDescription()
+ ": $" + cappuccino.getCost());
}
public void testCafeMocha() {
// Create a decaf cafe mocha with whipped
// cream
CafeMochaDecafWhipped cafeMocha =
new CafeMochaDecafWhipped();
System.out.println(cafeMocha.getDescription()
+ ": $" + cafeMocha.getCost());
}
junit.textui.TestRunner.run(CoeeShop.class);
}
} ///:~
And here is the corresponding output:

Cappucino: $1.0
Cafe Mocha decaf whipped cream: $1.25
You can see that creating the particular combination you want is easy, since you are
just creating an instance of a class. However, there are a number of problems with
this approach. Firstly, the combinations are xed statically so that any combination
a customer may wish to order needs to be created up front. Secondly, the resulting
menu is so huge that nding your particular combination is dicult and time
consuming.
The decorator approach

Another approach would be to break the drinks down into the various components
such as espresso and foamed milk, and then let the customer combine the
components to describe a particular coee.
In order to do this programmatically, we use the Decorator pattern. A Decorator
adds responsibility to a component by wrapping it, but the Decorator conforms to
the interface of the component it encloses, so the wrapping is transparent.
Decorators can also be nested without the loss of this transparency.
Methods invoked on the Decorator can in turn invoke methods in the component,
and can of course perform processing before or after the invocation.

So if we added getTotalCost() and getDescription() methods to the
DrinkComponent interface, an Espresso looks like this:
class Espresso extends Decorator {
private String description = " espresso";
public Espresso(DrinkComponent component) {
super(component);
}
public oat getTotalCost() {
return component.getTotalCost() + cost;
}
return component.getDescription() +
description;
}
}
You combine the components to create a drink as follows, as shown in the code
below:
//: decorator:alldecorators:CoeeShop2.java
// Coee example using decorators
package decorator.alldecorators;
interface DrinkComponent {
String getDescription();
oat getTotalCost();
}
class Mug implements DrinkComponent {
return "mug";
}
return 0;
}
}
abstract class Decorator implements DrinkComponent
{
protected DrinkComponent component;
Decorator(DrinkComponent component) {
this.component = component;
}
return component.getTotalCost();
}
public abstract String getDescription();
}
class Espresso extends Decorator {
private String description = " espresso";

public Espresso(DrinkComponent component) {
super(component);
}
}
description;
}
}
class Decaf extends Decorator {
private String description = " decaf";
public Decaf(DrinkComponent component) {
super(component);
}
description;
}
}
class FoamedMilk extends Decorator {
private String description = " foamed milk";
public FoamedMilk(DrinkComponent component) {
super(component);
}
}
description;
}
}
class SteamedMilk extends Decorator {
private String description = " steamed milk";
public SteamedMilk(DrinkComponent component) {
super(component);
}
}
description;
}
}
class Whipped extends Decorator {

private String description = " whipped cream";
public Whipped(DrinkComponent component) {
super(component);
}
}
description;
}
}
class Chocolate extends Decorator {
private String description = " chocolate";
public Chocolate(DrinkComponent component) {
super(component);
}
}
description;
}
}
public class CoeeShop2 extends TestCase {
// Create a plain cappucino
DrinkComponent cappuccino = new Espresso(
new FoamedMilk(new Mug()));
System.out.println(cappuccino.
getDescription().trim() + ": $" +
cappuccino.getTotalCost());
}
// cream
DrinkComponent cafeMocha = new Espresso(
new SteamedMilk(new Chocolate(new Whipped(
new Decaf(new Mug())))));
System.out.println(cafeMocha.getDescription().
trim() + ": $" + cafeMocha.getTotalCost());
}
junit.textui.TestRunner.run(CoeeShop2.class);
}
} ///:~
This approach would certainly provide the most exibility and the smallest menu.
You have a small number of components to choose from, but assembling the
description of the coee then becomes rather arduous.
If you want to describe a plain cappuccino, you create it with
new Espresso(new FoamedMilk(new Mug()))
Creating a decaf Caf Mocha with whipped cream requires an even longer
description.
Compromise
The previous approach takes too long to describe a coee. There will also be certain
combinations that you will describe regularly, and it would be convenient to have a
quick way of describing them.
The 3rd approach is a mixture of the rst 2 approaches, and combines exibility
with ease of use. This compromise is achieved by creating a reasonably sized menu
of basic selections, which would often work exactly as they are, but if you wanted to
decorate them (whipped cream, decaf etc.) then you would use decorators to make
the modications. This is the type of menu you are presented with in most coee
shops.
Here is how to create a basic selection, as well as a decorated selection:

//: decorator:compromise:CoeeShop3.java
// Coee example with a compromise of basic
// combinations and decorators
package decorator.compromise;
interface DrinkComponent {
oat getTotalCost();
String getDescription();
}
class Espresso implements DrinkComponent {
private String description = "Espresso";
return cost;
}
return description;
}
}
class EspressoConPanna implements DrinkComponent {
private String description = "EspressoConPare";
return cost;
}
return description;
}
}
class Cappuccino implements DrinkComponent {
private String description = "Cappuccino";
return cost;
}
return description;
}
}
class CafeLatte implements DrinkComponent {
private String description = "Cafe Late";
return cost;
}
return description;
}
}
class CafeMocha implements DrinkComponent {
private String description = "Cafe Mocha";
return cost;
}
return description;
}
}
abstract class Decorator implements DrinkComponent {
protected DrinkComponent component;
public Decorator(DrinkComponent component) {
this.component = component;
}
return component.getTotalCost();
}
return component.getDescription();
}
}
class ExtraEspresso extends Decorator {
public ExtraEspresso(DrinkComponent component) {
super(component);
}
" extra espresso";
}
return cost + component.getTotalCost();
}
}
class Whipped extends Decorator {
public Whipped(DrinkComponent component) {
super(component);
}
return cost + component.getTotalCost();
}
" whipped cream";
}
}
class Decaf extends Decorator{
public Decaf(DrinkComponent component) {
super(component);
}
return component.getDescription() + " decaf";
}
}
class Dry extends Decorator {
public Dry(DrinkComponent component) {
super(component);
}
" extra foamed milk";
}
}
class Wet extends Decorator {
public Wet(DrinkComponent component) {
super(component);
}
" extra steamed milk";
}
}
public class CoeeShop3 extends TestCase {
// Create a plain cappucino
DrinkComponent cappuccino = new Cappuccino();
System.out.println(cappuccino.getDescription()
+ ": $" + cappuccino.getTotalCost());
}
// cream
DrinkComponent cafeMocha = new Whipped(
new Decaf(new CafeMocha()));
System.out.println(cafeMocha.getDescription()
+ ": $" + cafeMocha.getTotalCost());
}
junit.textui.TestRunner.run(CoeeShop3.class);
}
} ///:~
You can see that creating a basic selection is quick and easy, which makes sense
since they will be described regularly. Describing a decorated drink is more work
than when using a class per combination, but clearly less work than when only
using decorators.
The nal result is not too many classes, but not too many decorators either. Most of
the time it's possible to get away without using any decorators at all, so we have the
benets of both approaches.
Other considerations
What happens if we decide to change the menu at a later stage, such as by adding a
new type of drink? If we had used the class per combination approach, the eect of
adding an extra such as syrup would be an exponential growth in the number of
classes. However, the implications to the all decorator or compromise approaches
are the same - one extra class is created.
How about the eect of changing the cost of steamed milk and foamed milk, when
the price of milk goes up? Having a class for each combination means that you need
to change a method in each class, and thus maintain many classes. By using
decorators, maintenance is reduced by dening the logic in one place.
Exercises
1.
2.
Add a Syrup class to the decorator approach described above. Then create a
Caf Latte (you'll need to use steamed milk with an espresso) with syrup.
Repeat Exercise 1 for the compromise approach.
3.
Create a simple decorator system that models the fact that some birds y
and some dont, some swim and some dont, and some do both.
4.
Implement the decorator pattern to create a Pizza restaurant, which has a

set menu of choices as well as the option to design your own pizza. Follow
the compromise approach to create a menu consisting of a Margherita,
Hawaiian, Regina, and Vegetarian pizzas, with toppings (decorators) of
Garlic, Olives, Spinach, Avocado, Feta and Pepperdews. Create a Hawaiian
pizza, as well as a Margherita decorated with Spinach, Feta, Pepperdews and
Olives.
5.
About Decorator, the Design Patterns book states: With decorators,

responsibilities can be added and removed at run-time simply by attaching
and detaching them. Implement the coee decoration system to allow this
simple detaching of a responsibility from the middle of the list of decorators
of a complex coee beverage.
Connecting dierent
types
Adapter
Adapter takes one type and produces an interface to some other type. When youve
got this, and you need that, Adapter solves the problem. The only requirement is to
produce a that, and there are a number of ways you can accomplish this adaptation.
//: adapter:SimpleAdapter.java
// "Object Adapter" from GoF diagram
package adapter;
class Target {
public void request() {}
}
class Adaptee {
public void specicRequest() {
System.out.println("Adaptee: SpecicRequest");
}
}
class Adapter extends Target {

private Adaptee adaptee;
public Adapter(Adaptee a) {
adaptee = a;
}
public void request() {
adaptee.specicRequest();
}
}
public class SimpleAdapter extends TestCase {
Adaptee a = new Adaptee();
Target t = new Adapter(a);
t.request();
}
junit.textui.TestRunner.run(SimpleAdapter.class);
}
} ///:~
//: adapter:AdapterVariations.java
// Variations on the Adapter pattern.
package adapter;
class WhatIHave {
public void g() {}
public void h() {}
}
interface WhatIWant {
void f();
}
class SurrogateAdapter implements WhatIWant {
WhatIHave whatIHave;
public SurrogateAdapter(WhatIHave wih) {
whatIHave = wih;
}
public void f() {
// Implement behavior using
// methods in WhatIHave:
whatIHave.g();
whatIHave.h();
}
}
class WhatIUse {
public void op(WhatIWant wiw) {
wiw.f();
}
}
// Approach 2: build adapter use into op():

class WhatIUse2 extends WhatIUse {
public void op(WhatIHave wih) {
new SurrogateAdapter(wih).f();
}
}
// Approach 3: build adapter into WhatIHave:
class WhatIHave2 extends WhatIHave
implements WhatIWant {
public void f() {
g();
h();
}
}
// Approach 4: use an inner class:
class WhatIHave3 extends WhatIHave {
private class InnerAdapter implements WhatIWant{
public void f() {
g();
h();
}
}
public WhatIWant whatIWant() {
return new InnerAdapter();
}
}
public class AdapterVariations extends TestCase {
WhatIUse whatIUse = new WhatIUse();
WhatIHave whatIHave = new WhatIHave();
WhatIWant adapt= new SurrogateAdapter(whatIHave);
WhatIUse2 whatIUse2 = new WhatIUse2();
WhatIHave2 whatIHave2 = new WhatIHave2();
WhatIHave3 whatIHave3 = new WhatIHave3();
whatIUse.op(adapt);
// Approach 2:
whatIUse2.op(whatIHave);
// Approach 3:
whatIUse.op(whatIHave2);
// Approach 4:
whatIUse.op(whatIHave3.whatIWant());
}
junit.textui.TestRunner.run(AdapterVariations.class);
}
} ///:~
Im taking liberties with the term proxy here, because in Design Patterns they
assert that a proxy must have an identical interface with the object that it is a
surrogate for. However, if you have the two words together: proxy adapter, it is
perhaps more reasonable.
Bridge
While researching Bridge, I discovered that it appears to be the most poorlydescribed pattern in the GoF. I began to come to this conclusion when reading Alan
Shalloways chapter on Bridge in his book Design Patterns Explained he begins by
pointing out that the description in GoF left him quite unenlightened.
At a conference, I talked to two people who had written about and were giving talks
on design patterns, including Bridge. In two separate discussions I got completely
dierent perspectives on the structure of Bridge.
Armed with misinformation, I delved back into the GoF and realized that neither of
the above perspectives agreed with the book. I also found that the book did a
miserable job of describing Bridge, except in one place not the general structure
chart describing the pattern, which wasnt helpful, but in the structure chart
describing their specic example. Only if you stare at that for a bit does Bridge
begin to make sense.
An important feature to understand when looking at Bridge is that it is often a
construct that is used to help you write code. You may choose the objects you use
for a particular situation at compile-time or runtime, but the goal of Bridge is to
allow you to structure your code so that you can easily add new kinds of front-end
objects which are implemented with functionality in new kinds of back-end objects.
Thus, both front-end and back-end can vary independently of each other.
The front-end classes can have completely dierent interfaces from each other, and
typically do. What they have in common is that they can implement their
functionality using facilities from any number of dierent back-end objects. The
back-end objects also dont have the same interface. The only thing the back-end
objects must have in common is that they implement the same kind of functionality
for example, a group of dierent ways to implement a graphics library or a set of
dierent data-storage solutions.
Bridge is really a code-organization tool that allows you to add in any number of
new front-end services that implement their operations by delegating to any number
of back-end options. Using Bridge, you can accomplish this without the normal
combinatorial explosion of possibilities that would otherwise occur. But keep in
mind that the vector of change with Bridge is typically happening at coding time: it
keeps your code organized when you are dealing with an increasing number of
options for implementing functionality.
Heres an example whose sole purpose is to demonstrate the structure of Bridge (it
implements the above diagram):
//: bridge:BridgeStructure.java
// A demonstration of the structure and operation
// of the Bridge Pattern.
package bridge;
class Abstraction {
private Implementation implementation;
public Abstraction(Implementation imp) {
implementation = imp;
}
// Abstraction used by the various front-end
// objects in order to implement their
// dierent interfaces.
// Implement this feature using some
// combination of back-end implementation:
implementation.facility1();
}
// Implement this feature using some other
}
// Implement this feature using some other
}
// For use by subclasses:
protected Implementation getImplementation() {

return implementation;
}
}
class ClientService1 extends Abstraction {
public ClientService1(Implementation imp) { super(imp); }
public void serviceA() {
service1();
service2();
}
public void serviceB() {
service3();
}
}
class ClientService2 extends Abstraction {
public ClientService2(Implementation imp) { super(imp); }
public void serviceC() {
service2();
service3();
}
public void serviceD() {
service1();
service3();
}
public void serviceE() {
getImplementation().facility3();
}
}
interface Implementation {
// The common implementation provided by the
// back-end objects, each in their own way.
void facility1();
void facility2();
void facility3();
void facility4();
}
class Library1 {
public void method1() {
System.out.println("Library1.method1()");
}
public void method2() {
System.out.println("Library1.method2()");
}
}
class Library2 {
System.out.println("Library2.operation1()");
}
}
}
}
class Implementation1 implements Implementation {
// Each facility delegates to a dierent library
// in order to fulll the obligations.
private Library1 delegate = new Library1();
public void facility1() {
System.out.println("Implementation1.facility1");
delegate.method1();
}
delegate.method2();
}
delegate.method2();
delegate.method1();
}
delegate.method1();
}
}
class Implementation2 implements Implementation {
private Library2 delegate = new Library2();
delegate.operation1();
}
}
}
}
}
public class BridgeStructure extends TestCase {
// Here, the implementation is determined by
// the client at creation time:
ClientService1 cs1 =
new ClientService1(new Implementation1());
cs1.serviceA();
cs1.serviceB();
}
cs1.serviceA();
cs1.serviceB();
}
cs2.serviceC();
cs2.serviceD();
cs2.serviceE();
}
cs2.serviceC();
cs2.serviceD();
cs2.serviceE();
}
junit.textui.TestRunner.run(BridgeStructure.class);
}
} ///:~
The front-end base class provides the operations used to fulll the front-end derived
classes, in terms of the methods of the back-end base class. Thus, any back-end
derived class can be used to perform the operations needed by the front-end class.
Notice that the bridging happens in this sequence of steps, each of which is
providing a layer of abstraction. Here, Implementation is dened as an interface
to emphasize that all the functionality is implemented in the back-end derived
classes, and none in the back-end base.
The back-end derived classes perform the operations dened in the base class by
delegating to objects (of class Library1 and Library2, in this case) that typically
have radically dierent interfaces, but somehow provide the same functionality (this
is one of the important objectives of Bridge). Eectively, each of the back-end
implementations is an adapter to a dierent library or tool used to implement the
desired functionality in a dierent way.
Exercises
1.
Modify BridgeStructure.java so that the implementations are chosen using

a factory.
2.
Modify BridgeStructure.java to use delegation instead of inheritance on

the front end. What benets and drawbacks do you see by using delegation
rather than inheritance?
3.
Create an example of Bridge with an abstraction that is an associative array.

This allows you to fetch elements by passing in a key Object. The constructor
provides an initial set of key-value pairs which are placed in an array. As long
as you only fetch elements, the array is used, but as soon as you set a new
key-value pair, the implementation is switched to a map.
4.
Use a bridge along with the collections in java.util.collections to create

stack and queue classes using an ArrayList. After you get the system
working, add a double-ended queue class. Now add a LinkedList as an
implementation. These steps will demonstrate how Bridge allows you to add
new front-end classes and new back-end classes in your code, with minimal
impact.
5.
Create a Bridge that provides a connection between various kinds of

bookkeeping programs (along with their interfaces and data formats) and
dierent banks (which provide dierent kinds of services and interfaces).
Flexible structure
Composite
The important thing here is that all elements in the part-whole have operations, and
that performing an operation on a node/composite also performs that operation on
any children of that node/composite. GoF includes implementation details of
containment and visitation of children in the interface of the base class, but this
doesnt seem necessary. In the following example, the Composite class simply
inherits ArrayList in order to gain its containment abilities.
//: composite:CompositeStructure.java
package composite;
import java.util.*;
interface Component {
void operation();
}
class Leaf implements Component {
private String name;
public Leaf(String name) { this.name = name; }
public String toString() { return name; }
public void operation() {
System.out.println(this);
}
}
class Node extends ArrayList implements Component {
public Node(String name) { this.name = name; }
public void operation() {
System.out.println(this);
for(Iterator it = iterator(); it.hasNext(); )
((Component)it.next()).operation();
}
}
public class CompositeStructure extends TestCase {
Node root = new Node("root");
root.add(new Leaf("Leaf1"));
Node c2 = new Node("Node1");
c2.add(new Leaf("Leaf2"));
root.add(c2);
c2 = new Node("Node2");
root.add(c2);
root.operation();
}
junit.textui.TestRunner.run(CompositeStructure.class);
}
} ///:~
While this approach seems to be the simplest thing that could possibly work, its
possible that in a larger system problems could arise. However, its probably best to
start with the simplest approach and change it only if the situation demands.
System decoupling
Observer
Like the other forms of callback, this contains a hook point where you can change
code. The dierence is in the observers completely dynamic nature. It is often used
for the specic case of changes based on other objects change of state, but is also
the basis of event management. Anytime you want to decouple the source of the call
from the called code in a completely dynamic way.
The observer pattern solves a fairly common problem: What if a group of objects
needs to update themselves when some object changes state? This can be seen in
the model-view aspect of Smalltalks MVC (model-view-controller), or the almostequivalent Document-View Architecture. Suppose that you have some data (the
document) and more than one view, say a plot and a textual view. When you
change the data, the two views must know to update themselves, and thats what
the observer facilitates. Its a common enough problem that its solution has been
made a part of the standard java.util library.
There are two types of objects used to implement the observer pattern in Java. The
Observable class keeps track of everybody who wants to be informed when a
change happens, whether the state has changed or not. When someone says OK,
everybody should check and potentially update themselves, the Observable class
performs this task by calling the notifyObservers( ) method for each one on the
list. The notifyObservers( ) method is part of the base class Observable.
There are actually two things that change in the observer pattern: the quantity of
observing objects and the way an update occurs. That is, the observer pattern
allows you to modify both of these without aecting the surrounding code.
------------Observer is an interface class that only has one member function, update( ).
This function is called by the object thats being observed, when that object decides
its time to update all its observers. The arguments are optional; you could have an
update( ) with no arguments and that would still t the observer pattern; however
this is more generalit allows the observed object to pass the object that caused
the update (since an Observer may be registered with more than one observed
object) and any extra information if thats helpful, rather than forcing the Observer
object to hunt around to see who is updating and to fetch any other information it
needs.
The observed object that decides when and how to do the updating will be called
the Observable.
Observable has a ag to indicate whether its been changed. In a simpler design,
there would be no ag; if something happened, everyone would be notied. The ag
allows you to wait, and only notify the Observers when you decide the time is right.
Notice, however, that the control of the ags state is protected, so that only an
inheritor can decide what constitutes a change, and not the end user of the
resulting derived Observer class.
Most of the work is done in notifyObservers( ). If the changed ag has not been
set, this does nothing. Otherwise, it rst clears the changed ag so repeated calls
to notifyObservers( ) wont waste time. This is done before notifying the observers
in case the calls to update( ) do anything that causes a change back to this
Observable object. Then it moves through the set and calls back to the update( )
member function of each Observer.
At rst it may appear that you can use an ordinary Observable object to manage
the updates. But this doesnt work; to get an eect, you must inherit from
Observable and somewhere in your derived-class code call setChanged( ). This is
the member function that sets the changed ag, which means that when you call
notifyObservers( ) all of the observers will, in fact, get notied. Where you call
setChanged( ) depends on the logic of your program.
Observing owers
Here is an example of the observer pattern:
//: observer:ObservedFlower.java
// Demonstration of "observer" pattern.
package observer;
import java.util.*;
class Flower {
private boolean isOpen;
private OpenNotier oNotify =
new OpenNotier();
private CloseNotier cNotify =
new CloseNotier();
public Flower() { isOpen = false; }
public void open() { // Opens its petals
isOpen = true;
oNotify.notifyObservers();
cNotify.open();
}
public void close() { // Closes its petals
isOpen = false;
cNotify.notifyObservers();
oNotify.close();
}
public Observable opening() { return oNotify; }
public Observable closing() { return cNotify; }
private class OpenNotier extends Observable {
private boolean alreadyOpen = false;
public void notifyObservers() {
if(isOpen && !alreadyOpen) {
setChanged();
super.notifyObservers();
alreadyOpen = true;
}
}
public void close() { alreadyOpen = false; }
}
private class CloseNotier extends Observable{
private boolean alreadyClosed = false;
public void notifyObservers() {
if(!isOpen && !alreadyClosed) {
setChanged();
super.notifyObservers();
alreadyClosed = true;
}
}
public void open() { alreadyClosed = false; }
}
}
class Bee {
private OpenObserver openObsrv =
new OpenObserver();
private CloseObserver closeObsrv =
new CloseObserver();
public Bee(String nm) { name = nm; }
// An inner class for observing openings:
private class OpenObserver implements Observer{
public void update(Observable ob, Object a) {
System.out.println("Bee " + name
+ "'s breakfast time!");
}
}
// Another inner class for closings:
private class CloseObserver implements Observer{

System.out.println("Bee " + name
+ "'s bed time!");
}
}
public Observer openObserver() {
return openObsrv;
}
public Observer closeObserver() {
return closeObsrv;
}
}
class Hummingbird {
private OpenObserver openObsrv =
new OpenObserver();
private CloseObserver closeObsrv =
new CloseObserver();
public Hummingbird(String nm) { name = nm; }
private class OpenObserver implements Observer{
System.out.println("Hummingbird " + name
+ "'s breakfast time!");
}
}
private class CloseObserver implements Observer{
System.out.println("Hummingbird " + name
+ "'s bed time!");
}
}
public Observer openObserver() {
return openObsrv;
}
public Observer closeObserver() {
return closeObsrv;
}
}
public class ObservedFlower extends TestCase {
Flower f = new Flower();
Bee
ba = new Bee("A"),
bb = new Bee("B");
Hummingbird
ha = new Hummingbird("A"),
hb = new Hummingbird("B");
f.opening().addObserver(ha.openObserver());
f.opening().addObserver(hb.openObserver());
f.opening().addObserver(ba.openObserver());
f.opening().addObserver(bb.openObserver());
f.closing().addObserver(ha.closeObserver());
f.closing().addObserver(hb.closeObserver());
f.closing().addObserver(ba.closeObserver());
f.closing().addObserver(bb.closeObserver());
// Hummingbird B decides to sleep in:
f.opening().deleteObserver(
hb.openObserver());
// A change that interests observers:
f.open();
f.open(); // It's already open, no change.
// Bee A doesn't want to go to bed:
f.closing().deleteObserver(
ba.closeObserver());
f.close();
f.close(); // It's already closed; no change
f.opening().deleteObservers();
f.open();
f.close();
}
junit.textui.TestRunner.run(ObservedFlower.class);
}
} ///:~
The events of interest are that a Flower can open or close. Because of the use of
the inner class idiom, both these events can be separately observable phenomena.
OpenNotier and CloseNotier both inherit Observable, so they have access to
setChanged( ) and can be handed to anything that needs an Observable.
The inner class idiom also comes in handy to dene more than one kind of
Observer, in Bee and Hummingbird, since both those classes may want to
independently observe Flower openings and closings. Notice how the inner class
idiom provides something that has most of the benets of inheritance (the ability to
access the private data in the outer class, for example) without the same
restrictions.
In main( ), you can see one of the prime benets of the observer pattern: the ability
to change behavior at run time by dynamically registering and un-registering
Observers with Observables.
If you study the code above youll see that OpenNotier and CloseNotier use the
basic Observable interface. This means that you could inherit other completely
dierent Observer classes; the only connection the Observers have with Flowers
is the Observer interface.
A visual example of observers

The following example is similar to the ColorBoxes example from Chapter 14 in
Thinking in Java, 2nd Edition. Boxes are placed in a grid on the screen and each one
is initialized to a random color. In addition, each box implements the Observer
interface and is registered with an Observable object. When you click on a box, all
of the other boxes are notied that a change has been made because the
Observable object automatically calls each Observer objects update( ) method.
Inside this method, the box checks to see if its adjacent to the one that was clicked,
and if so it changes its color to match the clicked box.
//: observer:BoxObserver.java
// Demonstration of Observer pattern using
// Java's built-in observer classes.
package observer;
import javax.swing.*;
import java.awt.*;
import java.awt.event.*;
import java.util.*;
// You must inherit a new type of Observable:
class BoxObservable extends Observable {
public void notifyObservers(Object b) {
// Otherwise it won't propagate changes:
setChanged();
super.notifyObservers(b);
}
}
public class BoxObserver extends JFrame {
Observable notier = new BoxObservable();
public BoxObserver(int grid) {
setTitle("Demonstrates Observer pattern");
Container cp = getContentPane();
cp.setLayout(new GridLayout(grid, grid));
for(int x = 0; x < grid; x++)
for(int y = 0; y < grid; y++)
cp.add(new OCBox(x, y, notier));
}
int grid = 8;
if(args.length > 0)
grid = Integer.parseInt(args[0]);
JFrame f = new BoxObserver(grid);
f.setSize(500, 400);
f.setVisible(true);
f.setDefaultCloseOperation(EXIT_ON_CLOSE);
}
}
class OCBox extends JPanel implements Observer {
Observable notier;
int x, y; // Locations in grid
Color cColor = newColor();
static nal Color[] colors = {
Color.BLACK, Color.BLUE, Color.CYAN,
Color.DARK_GRAY, Color.GRAY, Color.GREEN,
Color.LIGHT_GRAY, Color.MAGENTA,
Color.ORANGE, Color.PINK, Color.RED,
Color.WHITE, Color.YELLOW
};
static Random rand = new Random();
static nal Color newColor() {
return colors[rand.nextInt(colors.length)];
}
OCBox(int x, int y, Observable notier) {
this.x = x;
this.y = y;
notier.addObserver(this);
this.notier = notier;
addMouseListener(new ML());
}
public void paintComponent(Graphics g) {
super.paintComponent(g);
g.setColor(cColor);
Dimension s = getSize();
g.llRect(0, 0, s.width, s.height);
}
class ML extends MouseAdapter {
public void mousePressed(MouseEvent e) {
notier.notifyObservers(OCBox.this);
}
}
public void update(Observable o, Object arg) {
OCBox clicked = (OCBox)arg;
if(nextTo(clicked)) {
cColor = clicked.cColor;
repaint();
}
}
private nal boolean nextTo(OCBox b) {
return Math.abs(x - b.x) <= 1 &&
Math.abs(y - b.y) <= 1;
}
} ///:~
When you rst look at the online documentation for Observable, its a bit confusing
because it appears that you can use an ordinary Observable object to manage the
updates. But this doesnt work; try itinside BoxObserver, create an Observable
object instead of a BoxObservable object and see what happens: nothing. To get an
eect, you must inherit from Observable and somewhere in your derived-class code
call setChanged( ). This is the method that sets the changed ag, which means
that when you call notifyObservers( ) all of the observers will, in fact, get notied.
In the example above setChanged( ) is simply called within notifyObservers( ),
but you could use any criterion you want to decide when to call setChanged( ).
BoxObserver contains a single Observable object called notier, and every time
an OCBox object is created, it is tied to notier. In OCBox, whenever you click the
mouse the notifyObservers( ) method is called, passing the clicked object in as an
argument so that all the boxes receiving the message (in their update( ) method)
know who was clicked and can decide whether to change themselves or not. Using a
combination of code in notifyObservers( ) and update( ) you can work out some
fairly complex schemes.
It might appear that the way the observers are notied must be frozen at compile
time in the notifyObservers( ) method. However, if you look more closely at the
code above youll see that the only place in BoxObserver or OCBox where you're
aware that youre working with a BoxObservable is at the point of creation of the
Observable objectfrom then on everything uses the basic Observable interface.
This means that you could inherit other Observable classes and swap them at run
time if you want to change notication behavior then.
Mediator
Sweep coupling under the rug, how is this dierent from MVC?
MVC has distinct model and view; mediator could be anything. MVC a avor of
mediator
Exercises
1.
Create a minimal Observer-Observable design in two classes. Just create the

bare minimum in the two classes, then demonstrate your design by creating
one Observable and many Observers, and cause the Observable to update
the Observers.
2.
Create a minimal Observer system using java.util.Timer inside your

Observable, to generate events that are reported to the Observers. Create
several dierent Observers using anonymous inner classes, register these
with the Observable, and show that they are called when the Timer events
occur.
3.
Modify BoxObserver.java to turn it into a simple game. If any of the squares

surrounding the one you clicked is part of a contiguous patch of the same
color, then all the squares in that patch are changed to the color you clicked
on. You can congure the game for competition between players or to keep
track of the number of clicks that a single player uses to turn the eld into a
single color. You may also want to restrict a player's color to the rst one that
was chosen.
Reducing interface
complexity
Sometimes the problem that youre solving is as simple as I dont have the
interface that I want. Faade creates an interface to a set of classes, simply to
provide a more comfortable way to deal with a library or bundle of resources.
Faade
A general principle that I apply when Im casting about trying to mold requirements
into a rst-cut object is If something is ugly, hide it inside an object. This is
basically what Faade accomplishes. If you have a rather confusing collection of
classes and interactions that the client programmer doesnt really need to see, then
you can create an interface that is useful for the client programmer and that only
presents whats necessary.
Faade is often implemented as singleton abstract factory. Of course, you can easily
get this eect by creating a class containing static factory methods:
//: facade:Facade.java
package facade;
class A { public A(int x) {} }
class B { public B(long x) {} }
class C { public C(double x) {} }
// Other classes that aren't exposed
// by the facade go here ...
public class Facade extends TestCase {
static A makeA(int x) { return new A(x); }
static B makeB(long x) { return new B(x); }
static C makeC(double x) { return new C(x); }
// The client programmer gets the objects
// by calling the static methods:
A a = Facade.makeA(1);
B b = Facade.makeB(1);
C c = Facade.makeC(1.0);
}
junit.textui.TestRunner.run(Facade.class);
}
} ///:~
The example given in Design Patterns is just a class that uses the other classes.
A tax adviser is a Faade between you and the tax code, and a mediator between
you and the tax system.
Package as a variation of Faade

To me, the Faade has a rather procedural (non-object-oriented) feel to it: you are
just calling some functions to give you objects. And how dierent is it, really, from
Abstract Factory? The point of Faade is to hide part of a library of classes (and
their interactions) from the client programmer, to make the interface to that group
of classes more digestible and easier to understand.
However, this is precisely what the packaging features in Java accomplish: outside
of the library, you can only create and use public classes; all the non-public classes
are only accessible within the package. Its as if Faade is a built-in feature of Java.
To be fair, Design Patterns is written primarily for a C++ audience. Although C++
has namespaces to prevent clashes of globals and class names, this does not provide
the class hiding mechanism that you get with non-public classes in Java. The
majority of the time I think that Java packages will solve the Faade problem.
Algorithmic partitioning
Command: choosing the operation

at run-time
In Advanced C++:Programming Styles And Idioms (Addison-Wesley, 1992), Jim
Coplien coins the term functor which is an object whose sole purpose is to
encapsulate a function (since functor has a meaning in mathematics, in this book I
shall use the more explicit term function object). The point is to decouple the choice
of function to be called from the site where that function is called.
This term is mentioned but not used in Design Patterns. However, the theme of the
function object is repeated in a number of patterns in that book.
[7]
A Command is a function object in its purest sense: a method thats an object . By
wrapping a method in an object, you can pass it to other methods or objects as a
parameter, to tell them to perform this particular operation in the process of
fullling your request. You could say that a Command is a messenger (because its
intent and use is very straightforward) that carries behavior, rather than data.
//: command:CommandPattern.java
package command;
import java.util.*;
interface Command {
void execute();
}
class Hello implements Command {
public void execute() {
System.out.print("Hello ");
}
}
class World implements Command {
System.out.print("World! ");
}
}
class IAm implements Command {
System.out.print("I'm the command pattern!");
}
}
// An object that holds commands:
class Macro {
private List commands = new ArrayList();
public void add(Command c) { commands.add(c); }
public void run() {
Iterator it = commands.iterator();
while(it.hasNext())
((Command)it.next()).execute();
}
}
public class CommandPattern extends TestCase {
Macro macro = new Macro();
macro.add(new Hello());
macro.add(new World());
macro.add(new IAm());
macro.run();
}
junit.textui.TestRunner.run(CommandPattern.class);
}
} ///:~
The primary point of Command is to allow you to hand a desired action to a method
or object. In the above example, this provides a way to queue a set of actions to be
performed collectively. In this case, it allows you to dynamically create new
behavior, something you can normally only do by writing new code but in the above
example could be done by interpreting a script (see the Interpreter pattern if what
you need to do gets very complex).
Another example of Command is refactor:DirList.java [????]. The DirFilter class
is the command object which contains its action in the method accept( ) that is
passed to the list( ) method. The list( ) method determines what to include in its
result by calling accept( ).
Design Patterns says that Commands are an object-oriented replacement for
[8]
callbacks . However, I think that the word back is an essential part of the
concept of callbacks. That is, I think a callback actually reaches back to the creator
of the callback. On the other hand, with a Command object you typically just create
it and hand it to some method or object, and are not otherwise connected over time
to the Command object. Thats my take on it, anyway. Later in this book, I combine a
group of design patterns under the heading of callbacks.
Strategy appears to be a family of Command classes, all inherited from the same
base. But if you look at Command, youll see that it has the same structure: a
hierarchy of function objects. The dierence is in the way this hierarchy is used. As
seen in refactor:DirList.java, you use Command to solve a particular problemin
that case, selecting les from a list. The thing that stays the same is the body of
the method thats being called, and the part that varies is isolated in the function
object. I would hazard to say that Command provides exibility while youre writing
the program, whereas Strategys exibility is at run time. Nonetheless, it seems a
rather fragile distinction.
Exercises
4.
Use Command in Chapter 3, Exercise 1.
Chain of responsibility
Example: translation service (local->global->Babelsh).
Chain of Responsibility might be thought of as a dynamic generalization of
recursion using Strategy objects. You make a call, and each Strategy in a linked
sequence tries to satisfy the call. The process ends when one of the strategies is
successful or the chain ends. In recursion, one method calls itself over and over
until a termination condition is reached; with Chain of Responsibility, a method calls
itself, which (by moving down the chain of Strategies) calls a dierent
implementation of the method, etc., until a termination condition is reached. The
termination condition is either the bottom of the chain is reached (in which case a
default object is returned; you may or may not be able to provide a default result so
you must be able to determine the success or failure of the chain) or one of the
Strategies is successful.
Instead of calling a single method to satisfy a request, multiple methods in the chain
have a chance to satisfy the request, so it has the avor of an expert system. Since
the chain is eectively a linked list, it can be dynamically created, so you could also
think of it as a more general, dynamically-built switch statement.
In the GoF, theres a fair amount of discussion of how to create the chain of
responsibility as a linked list. However, when you look at the pattern it really
shouldnt matter how the chain is maintained; thats an implementation detail.
Since GoF was written before the Standard Template Library (STL) was
incorporated into most C++ compilers, the reason for this is most likely (1) there
was no list and thus they had to create one and (2) data structures are often taught
as a fundamental skill in academia, and the idea that data structures should be
standard tools available with the programming language may not have occurred to
the GoF authors. I maintain that the implementation of Chain of Responsibility as a
chain (specically, a linked list) adds nothing to the solution and can just as easily
be implemented using a standard Java List, as shown below. Furthermore, youll see
that Ive gone to some eort to separate the chain-management parts of the
implementation from the various Strategies, so that the code can be more easily
reused.
In StrategyPattern.java, above, what you probably want is to automatically nd a
solution. Chain of Responsibility provides a way to do this by chaining the Strategy
objects together and providing a mechanism for them to automatically recurse
through each one in the chain:
//: chainofresponsibility:FindMinima.java
package chainofresponsibility;
import com.bruceeckel.util.*; // Arrays2.toString()
// Carries the result data and
// whether the strategy was successful:
class LineData {
public double[] data;
public LineData(double[] data) { this.data = data; }
private boolean succeeded;
public boolean isSuccessful() { return succeeded; }
public void setSuccessful(boolean b) { succeeded = b; }
}
interface Strategy {
LineData strategy(LineData m);
}
class LeastSquares implements Strategy {
public LineData strategy(LineData m) {
System.out.println("Trying LeastSquares algorithm");
LineData ld = (LineData)m;
// [ Actual test/calculation here ]
LineData r = new LineData(
new double[] { 1.1, 2.2 }); // Dummy data
r.setSuccessful(false);
return r;
}
}
class NewtonsMethod implements Strategy {
System.out.println("Trying NewtonsMethod algorithm");
r.setSuccessful(false);
return r;
}
}
class Bisection implements Strategy {
System.out.println("Trying Bisection algorithm");
r.setSuccessful(true);
return r;
}
}
class ConjugateGradient implements Strategy {
System.out.println(
"Trying ConjugateGradient algorithm");
r.setSuccessful(true);
return r;
}
}
class MinimaFinder {
private static Strategy[] solutions = {

new LeastSquares(),
new NewtonsMethod(),
new Bisection(),
new ConjugateGradient(),
};
public static LineData solve(LineData line) {
LineData r = line;
for(int i = 0; i < solutions.length; i++) {
r = solutions[i].strategy(r);
if(r.isSuccessful())
return r;
}
throw new RuntimeException("unsolved: " + line);
}
}
public class FindMinima extends TestCase {
LineData line = new LineData(new double[]{
1.0, 2.0, 1.0, 2.0, -1.0, 3.0, 4.0, 5.0, 4.0
});
System.out.println(Arrays2.toString(
((LineData)MinimaFinder.solve(line)).data));
}
junit.textui.TestRunner.run(FindMinima.class);
}
} ///:~
Exercises
1.
Implement Chain of Responsibility to create an "expert system" that solves

problems by successively trying one solution after another until one matches.
You should be able to dynamically add solutions to the expert system. The test
for solution should just be a string match, but when a solution ts, the expert
system should return the appropriate type of ProblemSolver object. What
other pattern/patterns show up here?
2.
Implement Chain of Responsibility to create a language translator which

begins by searching for a local specialized translation system (which may
know specics about your problem domain), then a more global generalized
system, and nally falls back on BabelFish if it cant translate everything.
Note that each link in the chain may partially translate what its able to.
3.
Implement Chain of Responsibility to create an tool to help reformat Java

source code by trying multiple approaches to breaking lines. Note that
normal code and comments will probably need to be treated dierently,
leading to the possibility of implementing a Tree of Responsibility. Also note
the similarity between this approach and the Composite design pattern;
perhaps the more general description of this technique is a Composite of
Strategies.
Externalizing object
state
Memento
Use serialization to create an undo mechanism.
Complex interactions
When dealing with multiple types which are interacting, a program can get
particularly messy. For example, consider a system that parses and executes
mathematical expressions. You want to be able to say Number + Number,
Number * Number, etc., where Number is the base class for a family of numerical
objects. But when you say a + b, and you dont know the exact type of either a or b,
so how can you get them to interact properly?
The answer starts with something you probably dont think about: Java performs
only single dispatching. That is, if you are performing an operation on more than
one object whose type is unknown, Java can invoke the dynamic binding mechanism
on only one of those types. This doesnt solve the problem, so you end up detecting
some types manually and eectively producing your own dynamic binding behavior.
The solution is called multiple dispatching. Remember that polymorphism can occur
only via member function calls, so if you want double dispatching to occur, there
must be two member function calls: the rst to determine the rst unknown type,
and the second to determine the second unknown type. With multiple dispatching,
you must have a polymorphic method call to determine each of the types. Generally,
youll set up a conguration such that a single member function call produces more
than one dynamic member function call and thus determines more than one type in
the process. To get this eect, you need to work with more than one polymorphic
method call: youll need one call for each dispatch. The methods in the following
example are called compete( ) and eval( ), and are both members of the same
type. (In this case there will be only two dispatches, which is referred to as double
dispatching). If you are working with two dierent type hierarchies that are
interacting, then youll have to have a polymorphic method call in each hierarchy.
Heres an example of multiple dispatching:
//: multipledispatch:PaperScissorsRock.java
// Demonstration of multiple dispatching.
package multipledispatch;
import java.util.*;
// An enumeration type:
class Outcome {
private Outcome(String name) { this.name = name; }
public nal static Outcome
WIN = new Outcome("wins"),
LOSE = new Outcome("loses"),
DRAW = new Outcome("draws");
}
interface Item {
Outcome compete(Item it);
Outcome eval(Paper p);
Outcome eval(Scissors s);
Outcome eval(Rock r);
}
class Paper implements Item {
public Outcome compete(Item it) { return it.eval(this); }
public Outcome eval(Paper p) { return Outcome.DRAW; }
public Outcome eval(Scissors s) { return Outcome.WIN; }
public Outcome eval(Rock r) { return Outcome.LOSE; }
public String toString() { return "Paper"; }
}
class Scissors implements Item {
public Outcome eval(Paper p) { return Outcome.LOSE; }
public Outcome eval(Scissors s) { return Outcome.DRAW; }
public Outcome eval(Rock r) { return Outcome.WIN; }
public String toString() { return "Scissors"; }
}
class Rock implements Item {
public Outcome eval(Paper p) { return Outcome.WIN; }
public Outcome eval(Scissors s) { return Outcome.LOSE; }
public Outcome eval(Rock r) { return Outcome.DRAW; }
public String toString() { return "Rock"; }
}
class ItemGenerator {
private static Random rand = new Random();
public static Item newItem() {
switch(rand.nextInt(3)) {
default:
case 0: return new Scissors();
case 1: return new Paper();
case 2: return new Rock();
}
}
}
class Compete {
public static void match(Item a, Item b) {
System.out.println(
a + " " + a.compete(b) + " vs. " + b);
}
}
public class PaperScissorsRock extends TestCase {
static int SIZE = 20;
for(int i = 0; i < SIZE; i++)
Compete.match(ItemGenerator.newItem(),
ItemGenerator.newItem());
}
junit.textui.TestRunner.run(PaperScissorsRock.class);
}
} ///:~
Visitor, a type of multiple

dispatching
The assumption is that you have a primary class hierarchy that is xed; perhaps its
from another vendor and you cant make changes to that hierarchy. However, youd
like to add new polymorphic methods to that hierarchy, which means that normally
youd have to add something to the base class interface. So the dilemma is that you
need to add methods to the base class, but you cant touch the base class. How do
you get around this?
The design pattern that solves this kind of problem is called a visitor (the nal one
in the Design Patterns book), and it builds on the double dispatching scheme shown
in the last section.
The visitor pattern allows you to extend the interface of the primary type by
creating a separate class hierarchy of type Visitor to virtualize the operations
performed upon the primary type. The objects of the primary type simply accept
the visitor, then call the visitors dynamically-bound member function.
//: visitor:BeeAndFlowers.java
// Demonstration of "visitor" pattern.
package visitor;
import java.util.*;
interface Visitor {
void visit(Gladiolus g);
void visit(Runuculus r);

void visit(Chrysanthemum c);
}
// The Flower hierarchy cannot be changed:
interface Flower {
void accept(Visitor v);
}
class Gladiolus implements Flower {
public void accept(Visitor v) { v.visit(this);}
}
class Runuculus implements Flower {
}
class Chrysanthemum implements Flower {
}
// Add the ability to produce a string:
class StringVal implements Visitor {
String s;
public String toString() { return s; }
public void visit(Gladiolus g) {
s = "Gladiolus";
}
public void visit(Runuculus r) {
s = "Runuculus";
}
public void visit(Chrysanthemum c) {
s = "Chrysanthemum";
}
}
// Add the ability to do "Bee" activities:
class Bee implements Visitor {
public void visit(Gladiolus g) {
System.out.println("Bee and Gladiolus");
}
public void visit(Runuculus r) {
System.out.println("Bee and Runuculus");
}
public void visit(Chrysanthemum c) {
System.out.println("Bee and Chrysanthemum");
}
}
class FlowerGenerator {
public static Flower newFlower() {
default:
case 0: return new Gladiolus();
case 1: return new Runuculus();

case 2: return new Chrysanthemum();
}
}
}
public class BeeAndFlowers extends TestCase {
List owers = new ArrayList();
public BeeAndFlowers() {
for(int i = 0; i < 10; i++)
owers.add(FlowerGenerator.newFlower());
}
// It's almost as if I had a function to
// produce a Flower string representation:
StringVal sval = new StringVal();
Iterator it = owers.iterator();
((Flower)it.next()).accept(sval);
System.out.println(sval);
}
// Perform "Bee" operation on all Flowers:
Bee bee = new Bee();
it = owers.iterator();
while(it.hasNext())
((Flower)it.next()).accept(bee);
}
junit.textui.TestRunner.run(BeeAndFlowers.class);
}
} ///:~
Exercises
1.
Create a business-modeling environment with three types of Inhabitant:

Dwarf (for engineers), Elf (for marketers) and Troll (for managers). Now
create a class called Project that creates the dierent inhabitants and causes
them to interact( ) with each other using multiple dispatching.
2.
Modify the above example to make the interactions more detailed. Each
Inhabitant can randomly produce a Weapon using getWeapon( ): a Dwarf
uses Jargon or Play, an Elf uses InventFeature or SellImaginaryProduct,
and a Troll uses Edict and Schedule. You must decide which weapons win
and lose in each interaction (as in PaperScissorsRock.java). Add a
battle( ) member function to Project that takes two Inhabitants and
matches them against each other. Now create a meeting( ) member function
for Project that creates groups of Dwarf, Elf and Manager and battles the
groups against each other until only members of one group are left standing.
These are the winners.
3.
Modify PaperScissorsRock.java to replace the double dispatching with a
table lookup. The easiest way to do this is to create a Map of Maps, with the
key of each Map the class of each object. Then you can do the lookup by
saying:
((Map)map.get(o1.getClass())).get(o2.getClass())
Notice how much easier it is to recongure the system. When is it more
appropriate to use this approach vs. hard-coding the dynamic dispatches?
Can you create a system that has the syntactic simplicity of use of the
dynamic dispatch but uses a table lookup?
4.
Modify Exercise 2 to use the table lookup technique described in Exercise 3.
Multiple languages
This chapter looks at the value of crossing language boundaries. It is often very
advantageous to solve a problem using more than one programming language,
rather than being arbitrarily stuck using a single language. As youll see in this
chapter, a problem that is very dicult or tedious to solve in one language can often
be solved quickly and easily in another. If you can combine the use of languages,
you can create your product much more quickly and cheaply.
The most straightforward use of this idea is the Interpreter design pattern, which
adds an interpreted language to your program to allow the end user to easily
customize a solution. In Java, the easiest and most powerful way to do this is with
[9]
Jython , an implementation of the Python language in pure Java byte codes.
Interpreter solves a particular problem that of creating a scripting language for
the user. But sometimes its just easier and faster to temporarily step into another
language to solve a particular aspect of your problem. Youre not creating an
interpreter, youre just writing some code in another language. Again, Jython is a
good example of this, but CORBA also allows you to cross language boundaries.
Interpreter motivation
If the application user needs greater run time exibility, for example to create
scripts describing the desired behavior of the system, you can use the Interpreter
design pattern. Here, you create and embed a language interpreter into your
program.
Remember that each design pattern allows one or more factors to change, so its
important to rst be aware of which factor is changing. Sometimes the end users of
your application (rather than the programmers of that application) need complete
exibility in the way that they congure some aspect of the program. That is, they
need to do some kind of simple programming. The interpreter pattern provides this
exibility by adding a language interpreter.
The problem is that developing your own language and building an interpreter is a
time-consuming distraction from the process of developing your application. You
must ask whether you want to nish writing your application or create a new
language. The best solution is to reuse code: embed an interpreter thats already
been built and debugged for you. The Python language can be freely embedded into
your for-prot application without signing any license agreement, paying royalties,
or dealing with strings of any kind. There are basically no restrictions at all when
you're using Python.
Python is a language that is very easy to learn, very logical to read and write,
supports functions and objects, has a large set of available libraries, and runs on
virtually every platform. You can download Python and learn more about it by going
to www.Python.org.
For solving Java problems, we will look at a special version of Python called Jython.
This is generated entirely in Java byte codes, so incorporating it into your
application is quite simple, and its as portable as Java is. It has an extremely clean
interface with Java: Java can call Python classes, and Python can call Java classes.
Python is designed with classes from the ground up and is a truly pure object
oriented language (both C++ and Java violate purity in various ways). Python scales
up so that you can create very big programs without losing control of the code.
To install Python, go to www.Python.org and follow the links and instructions. To
install Jython, go to http://jython.sourceforge.net. The download is a .class le,
which will run an installer when you execute it with Java. You also need to add
jython.jar to your CLASSPATH. You can nd further installation instructions at
http://www.bruceeckel.com/TIPatterns/Building-Code.html.
Python overview
To get you started, here is a brief introduction for the experienced programmer
(which is what you should be if youre reading this book). You can refer to the full
documentation at www.Python.org (especially the incredibly useful HTML page A
Python Quick Reference), and also numerous books such as Learning Python by
Mark Lutz and David Ascher (OReilly, 1999).
Python is often referred to as a scripting language, but scripting languages tend to
be limiting, especially in the scope of the problems that they solve. Python, on the
other hand, is a programming language that also supports scripting. It is marvelous
for scripting, and you may nd yourself replacing all your batch les, shell scripts,
and simple programs with Python scripts. But it is far more than a scripting
language.
Python is designed to be very clean to write and especially to read. You will nd that
its quite easy to read your own code long after youve written it, and also to read
other peoples code. This is accomplished partially through clean, to-the-point
syntax, but a major factor in code readability is indentation scoping in Python is
determined by indentation. For example:
##interpreter:if.py
response = "yes"
if response == "yes":
print "armative"
val = 1
print "continuing..."
##~
The # denotes a comment that goes until the end of the line, just like C++ and
Java // comments.
First notice that the basic syntax of Python is C-ish; notice the if statement. But in a
C if, you would be required to use parentheses around the conditional, whereas
they are not necessary in Python (but it wont complain if you use them anyway).
The conditional clause ends with a colon, and this indicates that what follows will be
a group of indented statements, which are the then part of the if statement. In
this case, there is a print statement which sends the result to standard output,
followed by an assignment to a variable named val. The subsequent statement is not
indented so it is no longer part of the if. Indenting can nest to any level, just like
curly braces in C++ or Java, but unlike those languages there is no option (and no
argument) about where the braces are placed the compiler forces everyones code
to be formatted the same way, which is one of the main reasons for Pythons
consistent readability.
Python normally has only one statement per line (you can put more by separating
them with semicolons), thus no terminating semicolon is necessary. Even from the
brief example above you can see that the language is designed to be as simple as
possible, and yet still very readable.
Built-in containers
With languages like C++ and Java, containers are add-on libraries and not integral
to the language. In Python, the essential nature of containers for programming is
acknowledged by building them into the core of the language: both lists and
associative arrays (a.k.a. maps, dictionaries, hash tables) are fundamental data
types. This adds much to the elegance of the language.
In addition, the for statement automatically iterates through lists rather than just
counting through a sequence of numbers. This makes a lot of sense when you think
about it, since youre almost always using a for loop to step through an array or a
container. Python formalizes this by automatically making for use an iterator that
works through a sequence. Heres an example:
## interpreter:list.py
list = [ 1, 3, 5, 7, 9, 11 ]
print list
list.append(13)
for x in list:
print x
##~
The rst line creates a list. You can print the list and it will look exactly as you put it
in (in contrast, remember that I had to create a special Arrays2 class in Thinking in
Java, 2nd Edition in order to print arrays in Java). Lists are like Java containers you
can add new elements to them (here, append( ) is used) and they will automatically
resize themselves. The for statement creates an iterator x which takes on each
value in the list.
You can create a list of numbers with the range( ) function, so if you really need to
imitate Cs for, you can.
Notice that there arent any type declarations the object names simply appear, and
Python infers their type by the way that you use them. Its as if Python is designed
so that you only need to press the keys that absolutely must. Youll nd after youve
worked with Python for a short while that youve been using up a lot of brain cycles
parsing semicolons, curly braces, and all sorts of other extra verbiage that was
demanded by your non-Python programming language but didnt actually describe
what your program was supposed to do.
Functions
To create a function in Python, you use the def keyword, followed by the function
name and argument list, and a colon to begin the function body. Here is the rst
example turned into a function:
## interpreter:myFunction.py
def myFunction(response):
val = 0
if response == "yes":
print "armative"
val = 1
print "continuing..."
return val
print myFunction("no")
print myFunction("yes")
##~
Notice there is no type information in the function signature all it species is the
name of the function and the argument identiers, but no argument types or return
types. Python is a weakly-typed language, which means it puts the minimum
possible requirements on typing. For example, you could pass and return dierent
types from the same function:
## interpreter:dierentReturns.py
def dierentReturns(arg):
if arg == 1:
return "one"
if arg == "one":
return 1
print dierentReturns(1)
print dierentReturns("one")
##~
The only constraints on an object that is passed into the function are that the
function can apply its operations to that object, but other than that, it doesnt care.
Here, the same function applies the + operator to integers and strings:
## interpreter:sum.py
def sum(arg1, arg2):
return arg1 + arg2
print sum(42, 47)
print sum('spam ', "eggs")
##~
When the operator + is used with strings, it means concatenation (yes, Python
supports operator overloading, and it does a nice job of it).
Strings
The above example also shows a little bit about Python string handling, which is the
best of any language Ive seen. You can use single or double quotes to represent
strings, which is very nice because if you surround a string with double quotes, you
can embed single quotes and vice versa:
## interpreter:strings.py
print "That isn't a horse"
print 'You are not a "Viking"'
print """You're just pounding two
coconut halves together."""
print '''"Oh no!" He exclaimed.
"It's the blemange!"'''
print r'c:\python\lib\utils'
##~
Note that Python was not named after the snake, but rather the Monty Python
comedy troupe, and so examples are virtually required to include Python-esque
references.
The triple-quote syntax quotes everything, including newlines. This makes it
particularly useful for doing things like generating web pages (Python is an
especially good CGI language), since you can just triple-quote the entire page that
you want without any other editing.
The r right before a string means raw, which takes the backslashes literally so
you dont have to put in an extra backslash.
Substitution in strings is exceptionally easy, since Python uses Cs printf( )
substitution syntax, but for any string at all. You simply follow the string with a %
and the values to substitute:
## interpreter:stringFormatting.py
val = 47
print "The number is %d" % val
val2 = 63.4
s = "val: %d, val2: %f" % (val, val2)
print s
##~
As you can see in the second case, if you have more than one argument you
surround them in parentheses (this forms a tuple, which is a list that cannot be
modied).
All the formatting from printf( ) is available, including control over the number of
decimal places and alignment. Python also has very sophisticated regular
expressions.
Classes
Like everything else in Python, the denition of a class uses a minimum of
additional syntax. You use the class keyword, and inside the body you use def to
create methods. Heres a simple class:
## interpreter:SimpleClass.py
class Simple:
def __init__(self, str):
print "Inside the Simple constructor"
self.s = str
# Two methods:
def show(self):
print self.s
def showMsg(self, msg):
print msg + ':',
self.show() # Calling another method

if __name__ == "__main__":
# Create an object:
x = Simple("constructor argument")
x.show()
x.showMsg("A message")
##~
Both methods have self as their rst argument. C++ and Java both have a hidden
rst argument in their class methods, which points to the object that the method
was called for and can be accessed using the keyword this. Python methods also
use a reference to the current object, but when you are dening a method you must
explicitly specify the reference as the rst argument. Traditionally, the reference is
called self but you could use any identier you want (if you do not use self you will
probably confuse a lot of people, however). If you need to refer to elds in the
object or other methods in the object, you must use self in the expression. However,
when you call a method for an object as in x.show( ), you do not hand it the
reference to the object that is done for you.
Here, the rst method is special, as is any identier that begins and ends with
double underscores. In this case, it denes the constructor, which is automatically
called when the object is created, just like in C++ and Java. However, at the bottom
of the example you can see that the creation of an object looks just like a function
call using the class name. Pythons spare syntax makes you realize that the new
keyword isnt really necessary in C++ or Java, either.
All the code at the bottom is set o by an if clause, which checks to see if something
called __name__ is equivalent to __main__. Again, the double underscores indicate
special names. The reason for the if is that any le can also be used as a library
module within another program (modules are described shortly). In that case, you
just want the classes dened, but you dont want the code at the bottom of the le
to be executed. This particular if statement is only true when you are running this
le directly; that is, if you say on the command line:
Python SimpleClass.py
However, if this le is imported as a module into another program, the __main__

code is not executed.
Something thats a little surprising at rst is that you dene elds inside methods,
and not outside of the methods like C++ or Java (if you create elds using the
C++/Java style, they implicitly become static elds). To create an object eld, you
just name it using self inside of one of the methods (usually in the constructor,
but not always), and space is created when that method is run. This seems a little
strange coming from C++ or Java where you must decide ahead of time how much
space your object is going to occupy, but it turns out to be a very exible way to
program.
Inheritance
Because Python is weakly typed, it doesnt really care about interfaces all it cares
about is applying operations to objects (in fact, Javas interface keyword would be
wasted in Python). This means that inheritance in Python is dierent from
inheritance in C++ or Java, where you often inherit simply to establish a common
interface. In Python, the only reason you inherit is to inherit an implementation to
re-use the code in the base class.
If youre going to inherit from a class, you must tell Python to bring that class into
your new le. Python controls its name spaces as aggressively as Java does, and in a
similar fashion (albeit with Pythons penchant for simplicity). Every time you create
a le, you implicitly create a module (which is like a package in Java) with the same
name as that le. Thus, no package keyword is needed in Python. When you want
to use a module, you just say import and give the name of the module. Python
searches the PYTHONPATH in the same way that Java searches the CLASSPATH
(but for some reason, Python doesnt have the same kinds of pitfalls as Java does)
and reads in the le. To refer to any of the functions or classes within a module, you
give the module name, a period, and the function or class name. If you dont want
the trouble of qualifying the name, you can say
from module import name(s)
Where name(s) can be a list of names separated by commas.
You inherit a class (or classes Python supports multiple inheritance) by listing the
name(s) of the class inside parentheses after the name of the inheriting class. Note
that the Simple class, which resides in the le (and thus, module) named
SimpleClass is brought into this new name space using an import statement:
## interpreter:Simple2.py
from SimpleClass import Simple
class Simple2(Simple):
def __init__(self, str):
print "Inside Simple2 constructor"
# You must explicitly call
# the base-class constructor:
Simple.__init__(self, str)
def display(self):
self.showMsg("Called from display()")
# Overriding a base-class method
def show(self):
print "Overridden show() method"
# Calling a base-class method from inside
# the overridden method:
Simple.show(self)
class Dierent:
def show(self):
print "Not derived from Simple"
if __name__ == "__main__":
x = Simple2("Simple2 constructor argument")
x.display()
x.show()
x.showMsg("Inside main")
def f(obj): obj.show() # One-line denition
f(x)
f(Dierent())
##~
Simple2 is inherited from Simple, and in the constructor, the base-class

constructor is called. In display( ), showMsg( ) can be called as a method of self,
but when calling the base-class version of the method you are overriding, you must
fully qualify the name and pass self in as the rst argument, as shown in the
base-class constructor call. This can also be seen in the overridden version of
show( ).
In __main__, you will see (when you run the program) that the base-class
constructor is called. You can also see that the showMsg( ) method is available in
the derived class, just as you would expect with inheritance.
The class Dierent also has a method named show( ), but this class is not derived
from Simple. The f( ) method dened in __main__ demonstrates weak typing: all it
cares about is that show( ) can be applied to obj, and it doesnt have any other type
requirements. You can see that f( ) can be applied equally to an object of a class
derived from Simple and one that isnt, without discrimination. If youre a C++
programmer, you should see that the objective of the C++ template feature is
exactly this: to provide weak typing in a strongly-typed language. Thus, in Python
you automatically get the equivalent of templates without having to learn that
particularly dicult syntax and semantics.
Creating a language
It turns out to be remarkably simple to use Jython to create an interpreted language
inside your application. Consider the greenhouse controller example from Chapter 8
of Thinking in Java, 2nd edition. This is a situation where you want the end user
the person managing the greenhouse to have conguration control over the
system, and so a simple scripting language is the ideal solution.
To create the language, well simply write a set of Python classes, and the
constructor of each will add itself to a (static) master list. The common data and
behavior will be factored into the base class Event. Each Event object will contain
an action string (for simplicity in reality, youd have some sort of functionality)
and a time when the event is supposed to run. The constructor initializes these
elds, and then adds the new Event object to a static list called events (dening it
in the class, but outside of any methods, is what makes it static):
#:interpreter:GreenHouseLanguage.py
class Event:
events = [] # static
def __init__(self, action, time):
self.action = action
self.time = time
Event.events.append(self)
# Used by sort(). This will cause
# comparisons to be based only on time:
def __cmp__ (self, other):
if self.time < other.time: return -1
if self.time > other.time: return 1
return 0
def run(self):
print "%.2f: %s" % (self.time, self.action)
class LightOn(Event):
def __init__(self, time):
Event.__init__(self, "Light on", time)
class LightO(Event):
Event.__init__(self, "Light o", time)
class WaterOn(Event):
Event.__init__(self, "Water on", time)
class WaterO(Event):
Event.__init__(self, "Water o", time)
class ThermostatNight(Event):
Event.__init__(self,"Thermostat night", time)
class ThermostatDay(Event):
Event.__init__(self, "Thermostat day", time)
class Bell(Event):
Event.__init__(self, "Ring bell", time)
def run():
Event.events.sort();
for e in Event.events:
e.run()
# To test, this will be run when you say:
# python GreenHouseLanguage.py
if __name__ == "__main__":
ThermostatNight(5.00)
LightO(2.00)
WaterOn(3.30)
WaterO(4.45)
LightOn(1.00)
ThermostatDay(6.00)
Bell(7.00)
run()
##~
The constructor of each derived class calls the base-class constructor, which adds
the new object to the list. The run( ) function sorts the list, which automatically
uses the __cmp__( ) method that was dened in Event to base comparisons on time
only. In this example, it only prints out the list, but in the real system it would wait
for the time of each event to come up and then run the event.
The __main__ section performs a simple test on the classes.
The above le is now a module that can be included in another Python program to
dene all the classes it contains. But instead of an ordinary Python program, lets
use Jython, inside of Java. This turns out to be remarkably simple: you import some
Jython classes, create a PythonInterpreter object, and cause the Python les to be
loaded:
//- interpreter:GreenHouseController.java
package interpreter;
import org.python.util.PythonInterpreter;
import org.python.core.*;
public class
GreenHouseController extends TestCase {
PythonInterpreter interp =
new PythonInterpreter();
public void test() throws PyException {
System.out.println(
"Loading GreenHouse Language");
interp.execle("GreenHouseLanguage.py");
System.out.println(
"Loading GreenHouse Script");
interp.execle("Schedule.ghs");
System.out.println(
"Executing GreenHouse Script");
interp.exec("run()");
}
public static void
main(String[] args) throws PyException {
junit.textui.TestRunner.run(GreenHouseController.class);
}
} ///:~
The PythonInterpreter object is a complete Python interpreter that accepts

commands from the Java program. One of these commands is execle( ), which
tells it to execute all the statements it nds in a particular le. By executing
GreenHouseLanguage.py, all the classes from that le are loaded into our
PythonInterpreter object, and so it now holds the greenhouse controller
language. The Schedule.ghs le is the one created by the end user to control the
greenhouse. Heres an example:
//:! interpreter:Schedule.ghs
Bell(7.00)
ThermostatDay(6.00)
WaterOn(3.30)
LightOn(1.00)
ThermostatNight(5.00)
LightO(2.00)
WaterO(4.45)
///:~
This is the goal of the interpreter design pattern: to make the conguration of your
program as simple as possible for the end user. With Jython you can achieve this
with almost no eort at all.
One of the other methods available to the PythonInterpreter is exec( ), which
allows you to send a command to the interpreter. Here, the run( ) function is called
using exec( ).
Remember, to run this program you must go to http://jython.sourceforge.net and
download and install Jython (actually, you only need jython.jar in your
CLASSPATH). Once thats in place, its just like running any other Java program.
Controlling the interpreter

The prior example only creates and runs the interpreter using external scripts. In
the rest of this chapter, we shall look at more sophisticated ways to interact with
Jython. The simplest way to exercise more control over the PythonInterpreter
object from within Java is to send data to the interpreter, and pull data back out.
Putting data in
To inject data into your Python program, the PythonInterpreter class has a
deceptively simple method: set( ). However, set( ) takes many dierent data types
and performs conversions upon them. The following example is a reasonably
thorough exercise of the various set( ) possibilities, along with comments that
should give a fairly complete explanation:
//- interpreter:PythonInterpreterSetting.java
// Passing data from Java to python when using
// the PythonInterpreter object.
import java.util.*;
import com.bruceeckel.python.*;
public class
PythonInterpreterSetting extends TestCase {
// It automatically converts Strings
// into native Python strings:
interp.set("a", "This is a test");
interp.exec("print a");
interp.exec("print a[5:]"); // A slice
// It also knows what to do with arrays:
String[] s = { "How", "Do", "You", "Do?" };
interp.set("b", s);
interp.exec("for x in b: print x[0], x");
// set() only takes Objects, so it can't
// gure out primitives. Instead,
// you have to use wrappers:
interp.set("c", new PyInteger(1));
interp.set("d", new PyFloat(2.2));
interp.exec("print c + d");
// You can also use Java's object wrappers:
interp.set("c", new Integer(9));
interp.set("d", new Float(3.14));
interp.exec("print c + d");
// Dene a Python function to print arrays:
interp.exec(
"def prt(x): \n" +
" print x \n" +
" for i in x: \n" +
" print i, \n" +

" print x.__class__\n");
// Arrays are Objects, so it has no trouble
// guring out the types contained in arrays:
Object[] types = {
new boolean[]{ true, false, false, true },
new char[]{ 'a', 'b', 'c', 'd' },
new byte[]{ 1, 2, 3, 4 },
new int[]{ 10, 20, 30, 40 },
new long[]{ 100, 200, 300, 400 },
new oat[]{ 1.1f, 2.2f, 3.3f, 4.4f },
new double[]{ 1.1, 2.2, 3.3, 4.4 },
};
for(int i = 0; i < types.length; i++) {
interp.set("e", types[i]);
interp.exec("prt(e)");
}
// It uses toString() to print Java objects:
interp.set("f", new Date());
interp.exec("print f");
// You can pass it a List
// and index into it...
List x = new ArrayList();
for(int i = 0; i < 10; i++)
x.add(new Integer(i * 10));
interp.set("g", x);
interp.exec("print g");
interp.exec("print g[1]");
// ... But it's not quite smart enough
// to treat it as a Python array:
interp.exec("print g.__class__");
// interp.exec("print g[5:]); // Fails
// If you want it to be a python array, you
// must extract the Java array:
System.out.println("ArrayList to array:");
interp.set("h", x.toArray());
interp.exec("print h.__class__");
interp.exec("print h[5:]");
// Passing in a Map:
Map m = new HashMap();
m.put(new Integer(1), new Character('a'));
m.put(new Integer(3), new Character('b'));
m.put(new Integer(5), new Character('c'));
m.put(new Integer(7), new Character('d'));
m.put(new Integer(11), new Character('e'));
System.out.println("m: " + m);
interp.set("m", m);
interp.exec("print m, m.__class__, " +
"m[1], m[1].__class__");
// Not a Python dictionary, so this fails:
//! interp.exec("for x in m.keys():" +
//! "print x, m[x]");
// To convert a Map to a Python dictionary,
// use com.bruceeckel.python.PyUtil:
interp.set("m", PyUtil.toPyDictionary(m));
interp.exec("print m, m.__class__, " +

"m[1], m[1].__class__");
interp.exec("for x in m.keys():print x,m[x]");
}
public static void
PythonInterpreterSetting.class);
}
} ///:~
As usual with Java, the distinction between real objects and primitive types causes
trouble. In general, if you pass a regular object to set( ), it knows what to do with it,
but if you want to pass in a primitive you must perform a conversion. One way to do
this is to create a Py type, such as PyInteger or PyFloat. but it turns out you can
also use Javas own object wrappers like Integer and Float, which is probably
going to be a lot easier to remember.
Early in the program youll see an exec( ) containing the Python statement:
print a[5:]
The colon inside the indexing statement indicates a Python slice, which produces a
range of elements from the original array. In this case, it produces an array
containing the elements from number 5 until the end of the array. You could also say
a[3:5] to produce elements 3 through 5, or a[:5] to produce the elements zero
through 5. The reason a slice is used in this statement is to make sure that the Java
String has really been converted to a Python string, which can also be treated as an
array of characters.
You can see that its possible, using exec( ), to create a Python function (although
its a bit awkward). The prt( ) function prints the whole array, and then (to make
sure its a real Python array), iterates through each element of the array and prints
it. Finally, it prints the class of the array, so we can see what conversion has taken
place (Python not only has run-time type information, it also has the equivalent of
Java reection). The prt( ) function is used to print arrays that come from each of
the Java primitive types.
Although a Java ArrayList does pass into the interpreter using set( ), and you can
index into it as if it were an array, trying to create a slice fails. To completely
convert it into an array, one approach is to simply extract a Java array using
toArray( ), and pass that in. The set( ) method converts it to a PyArray one of the
classes provided with Jython which can be treated as a Python array (you can also
explicitly create a PyArray, but this seems unnecessary).
Finally, a Map is created and passed directly into the interpreter. While it is
possible to do simple things like index into the resulting object, its not a real
Python dictionary so you cant (for example) call the keys( ) method. There is no
straightforward way to convert a Java Map into a Python dictionary, and so I wrote
a utility called toPyDictionary( ) and made it a static method of
com.bruceeckel.python.PyUtil. This also includes utilities to extract a Python
array into a Java List, and a Python dictionary into a Java Map:
//- com:bruceeckel:python:PyUtil.java
// PythonInterpreter utilities
package com.bruceeckel.python;
import java.util.*;
public class PyUtil {
/** Extract a Python tuple or array into a Java
List (which can be converted into other kinds
of lists and sets inside Java).
@param interp The Python interpreter object
@param pyName The id of the python list object
*/
public static List
toList(PythonInterpreter interp, String pyName){
return new ArrayList(Arrays.asList(
(Object[])interp.get(
pyName, Object[].class)));
}
/** Extract a Python dictionary into a Java Map
@param interp The Python interpreter object
@param pyName The id of the python dictionary
*/
public static Map
toMap(PythonInterpreter interp, String pyName){
PyList pa = ((PyDictionary)interp.get(
pyName)).items();
Map map = new HashMap();
while(pa.__len__() != 0) {
PyTuple po = (PyTuple)pa.pop();
Object rst = po.__nditem__(0)
.__tojava__(Object.class);
Object second = po.__nditem__(1)
.__tojava__(Object.class);
map.put(rst, second);
}
return map;
}
/** Turn a Java Map into a PyDictionary,
suitable for placing into a PythonInterpreter
@param map The Java Map object
*/
public static PyDictionary
toPyDictionary(Map map) {
Iterator it = map.entrySet().iterator();
Map.Entry e = (Map.Entry)it.next();
m.put(Py.java2py(e.getKey()),
Py.java2py(e.getValue()));
}
// PyDictionary constructor wants a Hashtable:
return new PyDictionary(new Hashtable(m));
}
} ///:~
Here is the (black-box) unit testing code:
//- com:bruceeckel:python:Test.java
package com.bruceeckel.python;
import java.util.*;
public class Test extends TestCase {
PythonInterpreter pi =
pi.exec("tup=('fee','','fo','fum','')");
List lst = PyUtil.toList(pi, "tup");
System.out.println(lst);
System.out.println(new HashSet(lst));
}
pi.exec("ints=[1,3,5,7,9,11,13,17,19]");
List lst = PyUtil.toList(pi, "ints");
System.out.println(lst);
}
pi.exec("dict = { 1 : 'a', 3 : 'b', " +
"5 : 'c', 9 : 'd', 11 : 'e'}");
Map mp = PyUtil.toMap(pi, "dict");
System.out.println(mp);
}
m.put("twas", new Integer(11));
m.put("brillig", new Integer(27));
m.put("and", new Integer(47));
m.put("the", new Integer(42));
m.put("slithy", new Integer(33));
m.put("toves", new Integer(55));
System.out.println(m);
pi.set("m", PyUtil.toPyDictionary(m));
pi.exec("print m");
pi.exec("print m['slithy']");
}
junit.textui.TestRunner.run(Test.class);
}
} ///:~
Well see the use of the extraction tools in the next section.
Getting data out

There are a number of dierent ways to extract data from the PythonInterpreter.
If you simply call the get( ) method, passing it the object identier as a string, it
returns a PyObject (part of the org.python.core support classes). Its possible to
cast it using the __tojava__( ) method, but there are better alternatives:
1.
The convenience methods in the Py class, such as py2int( ), take a PyObject

and convert it to a number of dierent types.
2.
An overloaded version of get( ) takes the desired Java Class object as a

second argument, and produces an object that has that run-time type (so you
still need to perform a cast on the result in your Java code).
Using the second approach, getting an array from the PythonInterpreter is quite
easy. This is especially useful because Python is exceptionally good at manipulating
strings and les, and so you will commonly want to extract the results as an array of
strings. For example, you can do a wildcard expansion of le names using Pythons
glob( ), as shown further down in the following code:
//- interpreter:PythonInterpreterGetting.java
// Getting data from the PythonInterpreter object.
import java.util.*;
public class
PythonInterpreterGetting extends TestCase {
interp.exec("a = 100");
// If you just use the ordinary get(),
// it returns a PyObject:
PyObject a = interp.get("a");
// There's not much you can do with a generic
// PyObject, but you can print it out:
System.out.println("a = " + a);
// If you know the type it's supposed to be,
// you can "cast" it using __tojava__() to
// that Java type and manipulate it in Java.
// To use 'a' as an int, you must use
// the Integer wrapper class:
int ai= ((Integer)a.__tojava__(Integer.class))
.intValue();
// There are also convenience functions:
ai = Py.py2int(a);
System.out.println("ai + 47 = " + (ai + 47));
// You can convert it to dierent types:
oat af = Py.py2oat(a);
System.out.println("af + 47 = " + (af + 47));
// If you try to cast it to an inappropriate
// type you'll get a runtime exception:
//! String as = (String)a.__tojava__(
//! String.class);
// If you know the type, a more useful method
// is the overloaded get() that takes the
// desired class as the 2nd argument:
interp.exec("x = 1 + 2");
int x = ((Integer)interp
.get("x", Integer.class)).intValue();
System.out.println("x = " + x);

// Since Python is so good at manipulating
// strings and les, you will often need to
// extract an array of Strings. Here, a le
// is read as a Python array:
interp.exec("lines = " +
"open('PythonInterpreterGetting.java')" +
".readlines()");
// Pull it in as a Java array of String:
String[] lines = (String[])
interp.get("lines", String[].class);
for(int i = 0; i < 10; i++)
System.out.print(lines[i]);
// As an example of useful string tools,
// global expansion of ambiguous le names
// using glob is very useful, but it's not
// part of the standard Jython package, so
// you'll have to make sure that your
// Python path is set to include these, or
// that you deliver the necessary Python
// les with your application.
interp.exec("from glob import glob");
interp.exec("les = glob('*.java')");
String[] les = (String[])
interp.get("les", String[].class);
for(int i = 0; i < les.length; i++)
System.out.println(les[i]);
// You can extract tuples and arrays into
// Java Lists with com.bruceeckel.PyUtil:
interp.exec(
"tup = ('fee', '', 'fo', 'fum', '')");
List tup = PyUtil.toList(interp, "tup");
System.out.println(tup);
// It really is a list of String objects:
System.out.println(tup.get(0).getClass());
// You can easily convert it to a Set:
Set tups = new HashSet(tup);
System.out.println(tups);
interp.exec("ints=[1,3,5,7,9,11,13,17,19]");
List ints = PyUtil.toList(interp, "ints");
System.out.println(ints);
// It really is a List of Integer objects:
System.out.println((ints.get(1)).getClass());
// If you have a Python dictionary, it can
// be extracted into a Java Map, again with
// com.bruceeckel.PyUtil:
interp.exec("dict = { 1 : 'a', 3 : 'b'," +
"5 : 'c', 9 : 'd', 11 : 'e' }");
Map map = PyUtil.toMap(interp, "dict");
System.out.println("map: " + map);
// It really is Java objects, not PyObjects:
Iterator it = map.entrySet().iterator();
Map.Entry e = (Map.Entry)it.next();
System.out.println(e.getKey().getClass());
System.out.println(e.getValue().getClass());
}
public static void
PythonInterpreterGetting.class);
}
} ///:~
The last two examples show the extraction of Python tuples and lists into Java Lists,
and Python dictionaries into Java Maps. Both of these cases require more
processing than is provided in the standard Jython library, so I have again created
utilities in com.bruceeckel.pyton.PyUtil: toList( ) to produce a List from a
Python sequence, and toMap( ) to produce a Map from a Python dictionary. The
PyUtil methods make it easier to take important data structures back and forth
between Java and Python.
Multiple interpreters
Its also worth noting that you can have multiple PythonInterpreter objects in a
program, and each one has its own name space:
//- interpreter:MultipleJythons.java
// You can run multiple interpreters, each
// with its own name space.
public class MultipleJythons extends TestCase {
PythonInterpreter
interp1 = new PythonInterpreter(),
interp2 = new PythonInterpreter();
interp1.set("a", new PyInteger(42));
interp2.set("a", new PyInteger(47));
interp1.exec("print a");
interp2.exec("print a");
PyObject x1 = interp1.get("a");
PyObject x2 = interp2.get("a");
System.out.println("a from interp1: " + x1);
System.out.println("a from interp2: " + x2);
}
public static void
junit.textui.TestRunner.run(MultipleJythons.class);
}
} ///:~
When you run the program youll see that the value of a is distinct within each
PythonInterpreter.
Controlling Java from Jython

Since you have the Java language at your disposal, and you can set and retrieve
values in the interpreter, theres a tremendous amount that you can accomplish
with the above approach (controlling Python from Java). But one of the amazing
things about Jython is that it makes Java classes almost transparently available from
within Jython. Basically, a Java class looks like a Python class. This is true for
standard Java library classes as well as classes that you create yourself, as you can
see here:
## interpreter:JavaClassInPython.py
#=M jython.bat JavaClassInPython.py
# Using Java classes within Jython
from java.util import Date, HashSet, HashMap
from interpreter.javaclass import JavaClass
from math import sin
d = Date() # Creating a Java Date object
print d # Calls toString()
# A "generator" to easily create data:
class ValGen:
def __init__(self, maxVal):
self.val = range(maxVal)
# Called during 'for' iteration:
def __getitem__(self, i):
# Returns a tuple of two elements:
return self.val[i], sin(self.val[i])
# Java standard containers:
map = HashMap()
set = HashSet()
for x, y in ValGen(10):
map.put(x, y)
set.add(y)
set.add(y)
print map
print set
# Iterating through a set:
for z in set:
print z, z.__class__
print map[3] # Uses Python dictionary indexing
for x in map.keySet(): # keySet() is a Map method
print x, map[x]
# Using a Java class that you create yourself is
# just as easy:
jc = JavaClass()
jc2 = JavaClass("Created within Jython")
print jc2.getVal()
jc.setVal("Using a Java class is trivial")

print jc.getVal()
print jc.getChars()
jc.val = "Using bean properties"
print jc.val
##~
The =M comment is recognized by the makele generator tool (that I created for
this book) as a replacement makele command. This will be used instead of the
commands that the extraction tool would normally place in the makele.
Note that the import statements map to the Java package structure exactly as you
would expect. In the rst example, a Date( ) object is created as if it were a native
Python class, and printing this object just calls toString( ).
ValGen implements the concept of a generator which is used a great deal in the
C++ STL (Standard Template Library, part of the Standard C++ Library). A
generator is an object that produces a new object every time its generation
method is called, and it is quite convenient for lling containers. Here, I wanted to
use it in a for iteration, and so I needed the generation method to be the one that is
called by the iteration process. This is a special method called __getitem__( ),
which is actually the overloaded operator for indexing, [ ]. A for loop calls this
method every time it wants to move the iteration forward, and when the elements
run out, __getitem__( ) throws an out-of-bounds exception and that signals the end
of the for loop (in other languages, you would never use an exception for ordinary
control ow, but in Python it seems to work quite well). This exception happens
automatically when self.val[i] runs out of elements, so the __getitem__( ) code
turns out to be simple. The only complexity is that __getitem__( ) appears to return
two objects instead of just one. What Python does is automatically package multiple
return values into a tuple, so you still only end up returning a single object (in C++
or Java you would have to create your own data structure to accomplish this). In
addition, in the for loop where ValGen is used, Python automatically unpacks the
tuple so that you can have multiple iterators in the for. These are the kinds of
syntax simplications that make Python so endearing.
The map and set objects are instances of Javas HashMap and HashSet, again
created as if those classes were just native Python components. In the for loop, the
put( ) and add( ) methods work just like they do in Java. Also, indexing into a Java
Map uses the same notation as for dictionaries, but note that to iterate through the
keys in a Map you must use the Map method keySet( ) rather than the Python
dictionary method keys( ).
The nal part of the example shows the use of a Java class that I created from
scratch, to demonstrate how trivial it is. Notice also that Jython intuitively
understands JavaBeans properties, since you can either use the getVal( ) and
setVal( ) methods, or assign to and read from the equivalent val property. Also,
getChars( ) returns a Character[] in Java, and this becomes an array in Python.
The easiest way to use Java classes that you create for use inside a Python program
is to put them inside a package. Although Jython can also import unpackaged java
classes (import JavaClass), all such unpackaged java classes will be treated as if
they were dened in dierent packages so they can only see each others public
methods.
Java packages translate into Python modules, and Python must import a module in
order to be able to use the Java class. Here is the Java code for JavaClass:
//- interpreter:javaclass:JavaClass.java
package interpreter.javaclass;
import com.bruceeckel.util.*;
public class JavaClass {
private String s = "";
public JavaClass() {
System.out.println("JavaClass()");
}
public JavaClass(String a) {
s = a;
System.out.println("JavaClass(String)");
}
public String getVal() {
System.out.println("getVal()");
return s;
}
public void setVal(String a) {
System.out.println("setVal()");
s = a;
}
public Character[] getChars() {
System.out.println("getChars()");
Character[] r = new Character[s.length()];
for(int i = 0; i < s.length(); i++)
r[i] = new Character(s.charAt(i));
return r;
}
public static class Test extends TestCase {
JavaClass
x1 = new JavaClass(),
x2 = new JavaClass("UnitTest");
System.out.println(x2.getVal());
x1.setVal("SpamEggsSausageAndSpam");
System.out.println(
Arrays2.toString(x1.getChars()));
}
}
junit.textui.TestRunner.run(Test.class);
}
} ///:~
You can see that this is just an ordinary Java class, without any awareness that it
will be used in a Jython program. For this reason, one of the important uses of
[10]
Jython is in testing Java code
. Because Python is such a powerful, exible,
dynamic language it is an ideal tool for automated test frameworks, without making
any changes to the Java code thats being tested.
Inner Classes
Inner classes becomes attributes on the class object. Instances of static inner
classes can be create by the usual call:

com.foo.JavaClass.StaticInnerClass()
Non-static inner classes must have an outer class instance supplied explicitly as the
rst argument:
com.foo.JavaClass.InnerClass(com.foo.JavaClass())
Using Java libraries

Jython wraps the Java libraries so that any of them can be used directly or via
inheritance. In addition, Python shorthand simplies coding.
As an example, consider the HTMLButton.java example from Chapter 9 of
Thinking in Java, 2nd edition (you presumably have already downloaded and
installed the source code for that book from www.BruceEckel.com, since a number
of examples in this book use libraries from that book). Here is its conversion to
Jython:
## interpreter:PythonSwing.py
# The HTMLButton.java example from
# "Thinking in Java, 2nd edition," Chapter 13,
# converted into Jython.
# Dont run this as part of the automatic make:
#=M @echo skipping PythonSwing.py
from javax.swing import JFrame, JButton, JLabel
from java.awt import FlowLayout
frame = JFrame("HTMLButton", visible=1,
defaultCloseOperation=JFrame.EXIT_ON_CLOSE)
def kapow(e):
frame.contentPane.add(JLabel("<html>"+
"Kapow!"))
# Force a re-layout to
# include the new label:
frame.validate()
button = JButton("<html>" +
"<center>Hello! Press me now!",
actionPerformed=kapow)
frame.contentPane.layout = FlowLayout()
frame.contentPane.add(button)
frame.pack()
frame.size=200, 500
##~
If you compare the Java version of the program to the above Jython implementation,
youll see that Jython is shorter and generally easier to understand. For example, in
the Java version to set up the frame you had to make several calls: the constructor
for JFrame( ), the setVisible( ) method and the setDefaultCloseOperation( )
method, whereas in the above code all three of these operations are performed with
a single constructor call.
Also notice that the JButton is congured with an actionListener( ) method inside
the constructor, with the assignment to kapow. In addition, Jythons JavaBean
awareness means that a call to any method with a name that begins with set can
be replaced with an assignment, as you can see above.
The only method that did not come over from Java is the pack( ) method, which
seems to be essential in order to force the layout to happen properly. Its also
important that the call to pack( ) appear before the size setting.
Inheriting from Java library classes

You can easily inherit from standard Java library classes in Jython. Heres the
Dialogs.java example from Chapter 13 of Thinking in Java, 2 nd edition, converted
into Jython:
## interpreter:PythonDialogs.py
# Dialogs.java from "Thinking in Java, 2nd
# edition," Chapter 13, converted into Jython.
# Dont run this as part of the automatic make:
#=M @echo skipping PythonDialogs.py
from java.awt import FlowLayout
from javax.swing import JFrame, JDialog, JLabel
from javax.swing import JButton
class MyDialog(JDialog):
def __init__(self, parent=None):
JDialog.__init__(self,
title="My dialog", modal=1)
self.contentPane.layout = FlowLayout()
self.contentPane.add(JLabel("A dialog!"))
self.contentPane.add(JButton("OK",
actionPerformed =
lambda e, t=self: t.dispose()))
self.pack()
frame = JFrame("Dialogs", visible=1,
defaultCloseOperation=JFrame.EXIT_ON_CLOSE)
dlg = MyDialog()
frame.contentPane.add(
JButton("Press here to get a Dialog Box",
actionPerformed = lambda e: dlg.show()))
frame.pack()
##~
MyDialog is inherited from JDialog, and you can see named arguments being used
in the call to the base-class constructor.
In the creation of the OK JButton, note that the actionPerformed method is set
right inside the constructor, and that the function is created using the Python
lambda keyword. This creates a nameless function with the arguments appearing
before the colon and the expression that generates the returned value after the
colon. As you should know, the Java prototype for the actionPerformed( ) method
only contains a single argument, but the lambda expression indicates two. However,
the second argument is provided with a default value, so the function can be called
with only one argument. The reason for the second argument is seen in the default
value, because this is a way to pass self into the lambda expression, so that it can
be used to dispose of the dialog.
Compare this code with the version thats published in Thinking in Java, 2 nd edition.
Youll nd that Python language features allow a much more succinct and direct
implementation.
Creating Java classes with Jython

Although it does not directly relate to the original problem of this chapter (creating
an interpreter), Jython has the additional ability to create Java classes directly from
your Jython code. This can produce very useful results, as you are then able to treat
the results as if they are native Java classes, albeit with Python power under the
hood.
To produce Java classes from Python code, Jython comes with a compiler called
jythonc.
The process of creating Python classes that will produce Java classes is a bit more
complex than when calling Java classes from Python, because the methods in Java
classes are strongly typed, while Python functions and methods are weakly typed.
Thus, you must somehow tell jythonc that a Python method is intended to have a
particular set of argument types and that its return value is a particular type. You
accomplish this with the @sig string, which is placed right after the beginning of
the Python method denition (this is the standard location for the Python
documentation string). For example:
def returnArray(self):
"@sig public java.lang.String[] returnArray()"
The Python denition doesnt specify any return type, but the @sig string gives the
full type information about what is being passed and returned. The jythonc
compiler uses this information to generate the correct Java code.
Theres one other set of rules you must follow in order to get a successful
compilation: you must inherit from a Java class or interface in your Python class
(you do not need to specify the @sig signature for methods dened in the
superclass/interface). If you do not do this, you wont get your desired methods
unfortunately, jythonc gives you no warnings or errors in this case, but you wont
get what you want. If you dont see whats missing, it can be very frustrating.
In addition, you must import the appropriate java class and give the correct
package specication. In the example below, java is imported so you must inherit
from java.lang.Object, but you could also say from java.lang import Object and
then youd just inherit from Object without the package specication.
Unfortunately, you dont get any warnings or errors if you get this wrong, so you
must be patient and keep trying.
Here is an example of a Python class created to produce a Java class. This also
introduces the =T directive for the makele builder tool, which species a
dierent target than the one that is normally used by the tool. In this case, the
Python le is used to build a Java .class le, so the class le is the desired makele
target. To accomplish this, the default makele command is replaced using the =M
directive (notice how you can break across lines using \):
## interpreter:PythonToJavaClass.py
#=T python\java\test\PythonToJavaClass.class
#=M jythonc.bat --package python.java.test \
#=M PythonToJavaClass.py
# A Python class created to produce a Java class

from jarray import array
import java
class PythonToJavaClass(java.lang.Object):
# The '@sig' signature string is used to create
# the proper signature in the resulting
# Java code:
def __init__(self):
"@sig public PythonToJavaClass()"
print "Constructor for PythonToJavaClass"
def simple(self):
"@sig public void simple()"
print "simple()"
# Returning values to Java:
def returnString(self):
"@sig public java.lang.String returnString()"
return "howdy"
# You must construct arrays to return along
# with the type of the array:
def returnArray(self):
"@sig public java.lang.String[] returnArray()"
test = [ "fee", "", "fo", "fum" ]
return array(test, java.lang.String)
def ints(self):
"@sig public java.lang.Integer[] ints()"
test = [ 1, 3, 5, 7, 11, 13, 17, 19, 23 ]
return array(test, java.lang.Integer)
def doubles(self):
"@sig public java.lang.Double[] doubles()"
test = [ 1, 3, 5, 7, 11, 13, 17, 19, 23 ]
return array(test, java.lang.Double)
# Passing arguments in from Java:
def argIn1(self, a):
"@sig public void argIn1(java.lang.String a)"
print "a: %s" % a
print "a.__class__", a.__class__
"@sig public void argIn1(java.lang.Integer a)"
print "a + 100: %d" % (a + 100)
print "a.__class__", a.__class__
"@sig public void argIn3(java.util.List a)"
print "received List:", a, a.__class__
print "element type:", a[0].__class__
print "a[3] + a[5]:", a[5] + a[7]
#! print "a[2:5]:", a[2:5] # Doesn't work

"@sig public void \
argIn4(org.python.core.PyArray a)"
print "received type:", a.__class__
print "a: ", a
print "element type:", a[0].__class__
print "a[3] + a[5]:", a[5] + a[7]
print "a[2:5]:", a[2:5] # A real Python array
# A map must be passed in as a PyDictionary:
def argIn5(self, m):
"@sig public void \
argIn5(org.python.core.PyDictionary m)"
print "received Map: ", m, m.__class__
print "m['3']:", m['3']
for x in m.keys():
print x, m[x]
##~
First note that PythonToJavaClass is inherited from java.lang.Object; if you dont

do this you will quietly get a Java class without the right signatures. You are not
required to inherit from Object; any other Java class will do.
This class is designed to demonstrate dierent arguments and return values, to
provide you with enough examples that youll be able to easily create your own
signature strings. The rst three of these are fairly self-explanatory, but note the full
qualication of the Java name in the signature string.
In returnArray( ), a Python array must be returned as a Java array. To do this, the
Jython array( ) function (from the jarray module) must be used, along with the type
of the class for the resulting array. Any time you need to return an array to Java, you
must use array( ), as seen in the methods ints( ) and doubles( ).
The last methods show how to pass arguments in from Java. Basic types happen
automatically as long as you specify them in the @sig string, but you must use
objects and you cannot pass in primitives (that is, primitives must be ensconced in
wrapper objects, such as Integer).
In argIn3( ), you can see that a Java List is transparently converted to something
that behaves just like a Python array, but is not a true array because you cannot
take a slice from it. If you want a true Python array, then you must create and pass a
PyArray as in argIn4( ), where the slice is successful. Similarly, a Java Map must
come in as a PyDictionary in order to be treated as a Python dictionary.
Here is the Java program to exercise the Java classes produced by the above Python
code. This also introduces the =D directive for the makele builder tool, which
species a dependency in addition to those detected by the tool. Here, you cant
compile TestPythonToJavaClass.java until PythonToJavaClass.class is available:
//- interpreter:TestPythonToJavaClass.java
//+D python\java\test\PythonToJavaClass.class
import java.util.*;
// The package with the Python-generated classes:
import python.java.test.*;
public class
TestPythonToJavaClass extends TestCase {
PythonToJavaClass p2j = new PythonToJavaClass();
public void testDumpClassInfo() {
System.out.println(
Arrays2.toString(
p2j.getClass().getConstructors()));
Method[] methods =
p2j.getClass().getMethods();
for(int i = 0; i < methods.length; i++) {
String nm = methods[i].toString();
if(nm.indexOf("PythonToJavaClass") != -1)
System.out.println(nm);
}
}
p2j.simple();
System.out.println(p2j.returnString());
System.out.println(
Arrays2.toString(p2j.returnArray()));
System.out.println(
Arrays2.toString(p2j.ints());
System.out.println(
Arrays2.toString(p2j.doubles()));
p2j.argIn1("Testing argIn1()");
p2j.argIn2(new Integer(47));
ArrayList a = new ArrayList();
for(int i = 0; i < 10; i++)
a.add(new Integer(i));
p2j.argIn3(a);
p2j.argIn4(
new PyArray(Integer.class, a.toArray()));
for(int i = 0; i < 10; i++)
m.put("" + i, new Float(i));
p2j.argIn5(PyUtil.toPyDictionary(m));
}
TestPythonToJavaClass.class);
}
} ///:~
For Python support, youll usually only need to import the classes in
org.python.core. Everything else in the above example is fairly straightforward, as
PythonToJavaClass appears, from the Java side, to be just another Java class.
dumpClassInfo( ) uses reection to verify that the method signatures specied in
PythonToJavaClass.py have come through properly.
Building the Java classes from the Python

code
Part of the trick of creating Java classes from Python code is the @sig information in
the method documentation strings. But theres a second problem which stems from
the fact that Python has no package keyword the Python equivalent of packages
(modules) are implicitly created based on the le name. However, to bring the
resulting class les into the Java program, jythonc must be given information about
how to create the Java package for the Python code. This is done on the jythonc
command line using the --package ag, followed by the package name you wish to
produce (including the separation dots, just as you would give the package name
using the package keyword in a Java program). This will put the resulting .class
les in the appropriate subdirectory o of the current directory. Then you only need
to import the package in your Java program, as shown above (youll need . in your
CLASSPATH in order to run it from the code directory).
Here are the make dependency rules that I used to build the above example (the
backslashes at the ends of the lines are understood by make to be line
continuations). These rules are encoded into the above Java and Python les using
the comment syntax thats understood by my makele builder tool:
TestPythonToJavaClass.class: \
TestPythonToJavaClass.java \
python\java\test\PythonToJavaClass.class
javac TestPythonToJavaClass.java
python\java\test\PythonToJavaClass.class: \
PythonToJavaClass.py
jythonc.bat --package python.java.test \
PythonToJavaClass.py
The rst target, TestPythonToJavaClass.class, depends on both

TestPythonToJavaClass.java and the PythonToJavaClass.class, which is the
Python code thats converted to a class le. This latter, in turn, depends on the
Python source code. Note that its important that the directory where the target
lives be specied, so that the makele will create the Java program with the
minimum necessary amount of rebuilding.
The Java-Python Extension (JPE)

An alternative to Jython is the Java-Python Extension (JPE), which directly connects
with your native C-Python implementation.
Jython runs entirely within the JavaVM, which produces two fundamental
limitations: Jython cannot be called from CPython, and native Python extensions are
not accessible from JPython. JPE is linked with the Python C libraries, so JPE can be
called from C-Python, and native Python extensions can be called from Java through
JPE.
If you need to access the features on your native platform, JPE might be the easiest
solution. You can nd JPE at http://www.arakne.com/jpe.htm.
Summary
This chapter has arguably gone much deeper into Jython than required to use the
interpreter design pattern. Indeed, once you decide that you need to use interpreter
and that youre not going to get lost inventing your own language, the solution of
installing Jython is quite simple, and you can at least get started by following the
GreenHouseController example.
Of course, that example is often too simple and you may need something more
sophisticated, often requiring more interesting data to be passed back and forth.
When I encountered the limited documentation, I felt it necessary to come up with a
more thorough examination of Jython.
In the process, note that there could be another equally powerful design pattern
lurking in here, which could perhaps be called multiple languages. This is based on
the experience of having each language solve a certain class of problems better
than the other; by combining languages you can solve problems much faster than
with either language by itself. CORBA is another way to bridge across languages,
and at the same time bridging between computers and operating systems.
To me, Python and Java present a very potent combination for program development
because of Javas architecture and tool set, and Pythons extremely rapid
development (generally considered to be 5-10 times faster than C++ or Java).
Python is usually slower, however, but even if you end up re-coding parts of your
program for speed, the initial fast development will allow you to more quickly esh
out the system and uncover and solve the critical sections. And often, the execution
speed of Python is not a problem in those cases its an even bigger win. A number
of commercial products already use Java and Jython, and because of the terric
productivity leverage I expect to see this happen more in the future.
Exercises
1.
Modify GreenHouseLanguage.py so that it checks the times for the events

and runs those events at the appropriate times.
2.
Modify GreenHouseLanguage.py so that it calls a function for action

instead of just printing a string.
3.
Create a Swing application with a JTextField (where the user will enter
commands) and a JTextArea (where the command results will be displayed).
Connect to a PythonInterpreter object so that the output will be sent to the
JTextArea (which should scroll). Youll need to locate the PythonInterpreter
command that redirects the output to a Java stream.
4.
Modify GreenHouseLanguage.py to add a master controller class (instead

of the static array inside Event) and provide a run( ) method for each of the
subclasses. Each run( ) should create and use an object from the standard
Java library during its execution. Modify GreenHouseController.java to use
this new class.
5.
Modify the resulting GreenHouseLanguage.py from exercise two to

produce Java classes (add the @sig documentation strings to produce the
correct Java signatures, and create a makele to build the Java .class les).
Write a Java program that uses these classes.
Complex system states

StateMachine
While State has a way to allow the client programmer to change the
implementation, StateMachine imposes a structure to automatically change the
implementation from one object to the next. The current implementation represents
the state that a system is in, and the system behaves dierently from one state to
the next (because it uses State). Basically, this is a state machine using objects.
The code that moves the system from one state to the next is often a Template
Method, as seen in the following framework for a basic state machine. We start by
dening a tagging interface for input objects:
//: statemachine:Input.java
// Inputs to a state machine
package statemachine;
public interface Input {} ///:~
Each state can be run( ) to perform its behavior, and (in this design) you can also
pass it an Input object so it can tell you what new state to move to based on that
Input. The key distinction between this design and the next is that here, each State
object decides what other states it can move to, based on the Input, whereas in the
subsequent design all of the state transitions are held in a single table. Another way
to put it is that here, each State object has its own little State table, and in the
subsequent design there is a single master state transition table for the whole
system.
//: statemachine:State.java
// A State has an operation, and can be moved
// into the next State given an Input:
public interface State {
void run();
State next(Input i);
} ///:~
The StateMachine keeps track of the current state, which is initialized by the
constructor. The runAll( ) method takes an Iterator to a list of Input objects (an
Iterator is used here for convenience and simplicity; the important issue is that the
input information comes from somewhere). This method not only moves to the next
state, but it also calls run( ) for each state object thus you can see its an
expansion of the idea of the State pattern, since run( ) does something dierent
depending on the state that the system is in.
//: statemachine:StateMachine.java
// Takes an Iterator of Inputs to move from State
// to State using a template method.
import java.util.*;
public class StateMachine {

private State currentState;
public StateMachine(State initialState) {
currentState = initialState;
currentState.run();
}
// Template method:
public nal void runAll(Iterator inputs) {
while(inputs.hasNext()) {
Input i = (Input)inputs.next();
System.out.println(i);
currentState = currentState.next(i);
currentState.run();
}
}
} ///:~
Ive also treated runAll( ) as a template method. This is typical, but certainly not
required you could concievably want to override it, but typically the behavior
change will occur in States run( ) instead.
At this point the basic framework for this style of StateMachine (where each state
decides the next states) is complete. As an example, Ill use a fancy mousetrap that
[11]
can move through several states in the process of trapping a mouse
. The mouse
classes and information are stored in the mouse package, including a class
representing all the possible moves that a mouse can make, which will be the inputs
to the state machine:
//: statemachine:mouse:MouseAction.java
package statemachine.mouse;
import java.util.*;
import statemachine.*;
public class MouseAction implements Input {
private String action;
private static List instances = new ArrayList();
private MouseAction(String a) {
action = a;
instances.add(this);
}
public String toString() { return action; }
public int hashCode() {
return action.hashCode();
}
public boolean equals(Object o) {
return (o instanceof MouseAction)
&& action.equals(((MouseAction)o).action);
}
public static MouseAction forString(String description) {
Iterator it = instances.iterator();
MouseAction ma = (MouseAction)it.next();
if(ma.action.equals(description))
return ma;
}
throw new RuntimeException("not found: " + description);
}
public static MouseAction
appears = new MouseAction("mouse appears"),
runsAway = new MouseAction("mouse runs away"),
enters = new MouseAction("mouse enters trap"),
escapes = new MouseAction("mouse escapes"),
trapped = new MouseAction("mouse trapped"),
removed = new MouseAction("mouse removed");
} ///:~
Youll note that hashCode( ) and equals( ) have been overriden so that
MouseAction objects can be used as keys in a HashMap, but in the rst version of
the mousetrap we wont do this. Also, each possible move by a mouse is enumerated
as a static MouseAction object.
For creating test code, a sequence of mouse inputs is provided from a text le:
//:! statemachine:mouse:MouseMoves.txt
mouse appears
mouse runs away
mouse appears
mouse enters trap
mouse escapes
mouse appears
mouse enters trap
mouse trapped
mouse removed
mouse appears
mouse runs away
mouse appears
mouse enters trap
mouse trapped
mouse removed
///:~
To read this le in a generic fashion, here is a general-purpose tool called

StringList:
//: com:bruceeckel:util:StringList.java
// General-purpose tool that reads a le of text
// lines into a List, one line per list.
import java.io.*;
import java.util.*;
public class StringList extends ArrayList {
public StringList(String textFilePath) {
try {
BueredReader inputs = new BueredReader (
new FileReader(textFilePath));
String line;
while((line = inputs.readLine()) != null)
add(line.trim());
} catch(IOException e) {

}
}
} ///:~
This StringList only holds Objects, just as an ArrayList does, so we need an

adapter to turn the Strings into MouseActions:
//: statemachine:mouse:MouseMoveList.java
// A "transformer" to produce a
// List of MouseAction objects.
package statemachine.mouse;
import java.util.*;
public class MouseMoveList extends ArrayList {
public MouseMoveList(Iterator it) {
while(it.hasNext())
add(MouseAction.forString((String)it.next()));
}
} ///:~
The MouseMoveList looks a bit like a decorator, and acts a bit like an adapter.
However, an adapter changes one interface to another, and a decorator adds
functionality or data. MouseMoveList changes the contents of the container, so it
might be thought of as a Transformer.
With these tools in place, its now possible to create the rst version of the
mousetrap program. Each State subclass denes its run( ) behavior, and also
establishes its next state with an if-else clause:
//: statemachine:mousetrap1:MouseTrapTest.java
// State Machine pattern using 'if' statements
// to determine the next state.
package statemachine.mousetrap1;
import statemachine.mouse.*;
import java.util.*;
import java.io.*;
// A dierent subclass for each state:
class Waiting implements State {
public void run() {
System.out.println(
"Waiting: Broadcasting cheese smell");
}
public State next(Input i) {
MouseAction ma = (MouseAction)i;
if(ma.equals(MouseAction.appears))
return MouseTrap.luring;
return MouseTrap.waiting;
}
}
class Luring implements State {

public void run() {
System.out.println(
"Luring: Presenting Cheese, door open");
}
if(ma.equals(MouseAction.runsAway))
if(ma.equals(MouseAction.enters))
return MouseTrap.trapping;
return MouseTrap.luring;
}
}
class Trapping implements State {
public void run() {
System.out.println("Trapping: Closing door");
}
if(ma.equals(MouseAction.escapes))
if(ma.equals(MouseAction.trapped))
return MouseTrap.holding;
return MouseTrap.trapping;
}
}
class Holding implements State {
public void run() {
System.out.println("Holding: Mouse caught");
}
if(ma.equals(MouseAction.removed))
return MouseTrap.holding;
}
}
class MouseTrap extends StateMachine {
public static State
waiting = new Waiting(),
luring = new Luring(),
trapping = new Trapping(),
holding = new Holding();
public MouseTrap() {
super(waiting); // Initial state
}
}
public class MouseTrapTest extends TestCase {
MouseTrap trap = new MouseTrap();
MouseMoveList moves =
new MouseMoveList(
new StringList("../mouse/MouseMoves.txt")
.iterator());
trap.runAll(moves.iterator());
}
junit.textui.TestRunner.run(MouseTrapTest.class);
}
} ///:~
The StateMachine class simply denes all the possible states as static objects, and
also sets up the initial state. The UnitTest creates a MouseTrap and then tests it
with all the inputs from a MouseMoveList.
While the use of if-else statements inside the next( ) methods is perfectly
reasonable, managing a large number of these could become dicult. Another
approach is to create tables inside each State object dening the various next
states based on the input.
Initially, this seems like it ought to be quite simple. You should be able to dene a
static table in each State subclass that denes the transitions in terms of the other
State objects. However, it turns out that this approach generates cyclic
initialization dependencies. To solve the problem, Ive had to delay the initialization
of the tables until the rst time that the next( ) method is called for a particular
State object. Initially, the next( ) methods can appear a little strange because of
this.
The StateT class is an implementation of State (so that the same StateMachine
class can be used from the previous example) that adds a Map and a method to
initialize the map from a two-dimensional array. The next( ) method has a
base-class implementation which must be called from the overridden derived class
next( ) methods after they test for a null Map (and initialize it if its null):
//: statemachine:mousetrap2:MouseTrap2Test.java
// A better mousetrap using tables
package statemachine.mousetrap2;
import statemachine.mouse.*;
import java.util.*;
import java.io.*;
abstract class StateT implements State {
protected Map transitions = null;
protected void init(Object[][] states) {
transitions = new HashMap();
for(int i = 0; i < states.length; i++)
transitions.put(states[i][0], states[i][1]);
}
public abstract void run();
if(transitions.containsKey(i))
return (State)transitions.get(i);
else

"Input not supported for current state");
}
}
class MouseTrap extends StateMachine {
public static State
waiting = new Waiting(),
luring = new Luring(),
trapping = new Trapping(),
holding = new Holding();
public MouseTrap() {
super(waiting); // Initial state
}
}
class Waiting extends StateT {
public void run() {
System.out.println(
"Waiting: Broadcasting cheese smell");
}
// Delayed initialization:
if(transitions == null)
init(new Object[][] {
{ MouseAction.appears, MouseTrap.luring },
});
return super.next(i);
}
}
class Luring extends StateT {
public void run() {
System.out.println(
"Luring: Presenting Cheese, door open");
}
{ MouseAction.enters, MouseTrap.trapping },
{ MouseAction.runsAway, MouseTrap.waiting },
});
}
}
class Trapping extends StateT {
public void run() {
System.out.println("Trapping: Closing door");
}
{ MouseAction.escapes, MouseTrap.waiting },
{ MouseAction.trapped, MouseTrap.holding },
});
}
}
class Holding extends StateT {
public void run() {
System.out.println("Holding: Mouse caught");
}
{ MouseAction.removed, MouseTrap.waiting },
});
}
}
public class MouseTrap2Test extends TestCase {
MouseTrap trap = new MouseTrap();
MouseMoveList moves =
new MouseMoveList(
new StringList("../mouse/MouseMoves.txt")
.iterator());
trap.runAll(moves.iterator());
}
junit.textui.TestRunner.run(MouseTrap2Test.class);
}
} ///:~
The rest of the code is identical the dierence is in the next( ) methods and the
StateT class.
If you have to create and maintain a lot of State classes, this approach is an
improvement, since its easier to quickly read and understand the state transitions
from looking at the table.
Exercises
1.
Modify MouseTrap2Test.java so that the state table information is loaded

from an external text le, which only contains the state table information.
Table-Driven State Machine

The advantage of the previous design is that all the information about a state,
including the state transition information, is located within the state class itself.
This is generally a good design principle.
However, in a pure state machine, the machine can be completely represented by a
single state-transition table. This has the advantage of locating all the information
about the state machine in a single place, which means that you can more easily
create and maintain the table based on a classic state-transition diagram.
The classic state-transition diagram uses a circle to represent each state, and lines
from the state pointing to all states that state can transition into. Each transition
line is annotated with conditions for transition and an action during transition.
Heres what it looks like:
(Simple State Machine Diagram)
Goals:
Direct translation of state diagram
Vector of change: the state diagram representation
Reasonable implementation
No excess of states (you could represent every single change with a new
state)
Simplicity and exibility
Observations:
States are trivial no information or functions/data, just an identity
Not like the State pattern!
The machine governs the move from state to state
Similar to yweight
Each state may move to many others
Condition & action functions must also be external to states
Centralize description in a single table containing all variations, for ease of

conguration
Example:
State Machine & Table-Driven Code
Implements a vending machine
Uses several other patterns
Separates common state-machine code from specic application (like

template method)
Each input causes a seek for appropriate solution (like chain of

responsibility)
Tests and transitions are encapsulated in function objects (objects that hold
functions)
Java constraint: methods are not rst-class objects
The State class

The State class is distinctly dierent from before, since it is really just a
placeholder with a name. Thus it is not inherited from previous State classes:
//: statemachine2:State.java
package statemachine2;
public class State {
public State(String nm) { name = nm; }
} ///:~
Conditions for transition

In the state transition diagram, an input is tested to see if it meets the condition
necessary to transfer to the state under question. As before, the Input is just a
tagging interface:
//: statemachine2:Input.java
// Inputs to a state machine
public interface Input {} ///:~
The Condition evaluates the Input to decide whether this row in the table is the
correct transition:
//: statemachine2:Condition.java
// Condition function object for state machine
public interface Condition {
boolean condition(Input i);
} ///:~
Transition actions
If the Condition returns true, then the transition to a new state is made, and as
that transition is made some kind of action occurs (in the previous state machine
design, this was the run( ) method):
//: statemachine2:Transition.java
// Transition function object for state machine
public interface Transition {
void transition(Input i);
} ///:~
The table
With these classes in place, we can set up a 3-dimensional table where each row
completely describes a state. The rst element in the row is the current state, and
the rest of the elements are each a row indicating what the type of the input can be,
the condition that must be satised in order for this state change to be the correct
one, the action that happens during transition, and the new state to move into. Note
that the Input object is not just used for its type, it is also a Messenger object that
carries information to the Condition and Transition objects:
{ {CurrentState},
{Input, Condition(Input), Transition(Input), Next},
...
}
The basic machine

//: statemachine2:StateMachine.java
// A table-driven state machine
import java.util.*;
public class StateMachine {
private State state;
private Map map = new HashMap();
public StateMachine(State initial) {
state = initial;
}
public void buildTable(Object[][][] table) {
for(int i = 0; i < table.length; i++) {
Object[][] row = table[i];
Object currentState = row[0][0];
List transitions = new ArrayList();
for(int j = 1; j < row.length; j++)
transitions.add(row[j]);
map.put(currentState, transitions);
}
}
public void nextState(Input input) {
Iterator it=((List)map.get(state)).iterator();
Object[] tran = (Object[])it.next();
if(input == tran[0] ||
input.getClass() == tran[0]) {
if(tran[1] != null) {
Condition c = (Condition)tran[1];
if(!c.condition(input))
continue; // Failed test
}
if(tran[2] != null)
((Transition)tran[2]).transition(input);
state = (State)tran[3];
return;
}
}
"Input not supported for current state");
}
} ///:~
Simple vending machine

//: statemachine:vendingmachine:VendingMachine.java
// Demonstrates use of StateMachine.java
package statemachine.vendingmachine;
import statemachine2.*;
nal class VM extends State {
private VM(String nm) { super(nm); }
public nal static VM
quiescent = new VM("Quiesecent"),
collecting = new VM("Collecting"),
selecting = new VM("Selecting"),
unavailable = new VM("Unavailable"),
wantMore = new VM("Want More?"),
noChange = new VM("Use Exact Change Only"),
makesChange = new VM("Machine makes change");
}
nal class HasChange implements Input {
private HasChange(String nm) { name = nm; }
public nal static HasChange

yes = new HasChange("Has change"),
no = new HasChange("Cannot make change");
}
class ChangeAvailable extends StateMachine {
public ChangeAvailable() {
super(VM.makesChange);
buildTable(new Object[][][]{
{ {VM.makesChange}, // Current state
// Input, test, transition, next state:
{HasChange.no, null, null, VM.noChange}},
{ {VM.noChange}, // Current state
{HasChange.yes, null,
null, VM.makesChange}},
});
}
}
nal class Money implements Input {
private int value;
private Money(String nm, int val) {
name = nm;
value = val;
}
public int getValue() { return value; }
public nal static Money
quarter = new Money("Quarter", 25),
dollar = new Money("Dollar", 100);
}
nal class Quit implements Input {
private Quit() {}
public String toString() { return "Quit"; }
public nal static Quit quit = new Quit();
}
nal class FirstDigit implements Input {
private int value;
private FirstDigit(String nm, int val) {
name = nm;
value = val;
}
public nal static FirstDigit
A = new FirstDigit("A", 0),
B = new FirstDigit("B", 1),
C = new FirstDigit("C", 2),
D = new FirstDigit("D", 3);
}
nal class SecondDigit implements Input {
private int value;
private SecondDigit(String nm, int val) {
name = nm;
value = val;
}
public nal static SecondDigit
one = new SecondDigit("one", 0),
two = new SecondDigit("two", 1),
three = new SecondDigit("three", 2),
four = new SecondDigit("four", 3);
}
class ItemSlot {
int price;
int quantity;
static int counter = 0;
String id = Integer.toString(counter++);
public ItemSlot(int prc, int quant) {
price = prc;
quantity = quant;
}
public String toString() { return id; }
public int getPrice() { return price; }
public int getQuantity() { return quantity; }
public void decrQuantity() { quantity--; }
}
public class VendingMachine extends StateMachine{
StateMachine changeAvailable =
new ChangeAvailable();
int amount = 0;
FirstDigit rst = null;
ItemSlot[][] items = new ItemSlot[4][4];
Condition notEnough = new Condition() {
public boolean condition(Input input) {
int i1 = rst.getValue();
int i2 = ((SecondDigit)input).getValue();
return items[i1][i2].getPrice() > amount;
}
};
Condition itemAvailable = new Condition() {
return items[i1][i2].getQuantity() > 0;
}
};
Condition itemNotAvailable = new Condition() {
return !itemAvailable.condition(input);
}
};
Transition clearSelection = new Transition() {
public void transition(Input input) {
ItemSlot is = items[i1][i2];
System.out.println(
"Clearing selection: item " + is +
" costs " + is.getPrice() +
" and has quantity " + is.getQuantity());
rst = null;
}
};
Transition dispense = new Transition() {
ItemSlot is = items[i1][i2];
System.out.println("Dispensing item " +
is + " costs " + is.getPrice() +
" and has quantity " + is.getQuantity());
items[i1][i2].decrQuantity();
System.out.println("New Quantity " +
is.getQuantity());
amount -= is.getPrice();
System.out.println("Amount remaining " +
amount);
}
};
Transition showTotal = new Transition() {
amount += ((Money)input).getValue();
System.out.println("Total amount = " +
amount);
}
};
Transition returnChange = new Transition() {
System.out.println("Returning " + amount);
amount = 0;
}
};
Transition showDigit = new Transition() {
rst = (FirstDigit)input;
System.out.println("First Digit= "+ rst);
}
};
public VendingMachine() {
super(VM.quiescent); // Initial state
for(int i = 0; i < items.length; i++)
for(int j = 0; j < items[i].length; j++)
items[i][j] = new ItemSlot((j+1)*25, 5);
items[3][0] = new ItemSlot(25, 0);
buildTable(new Object[][][]{
{ {VM.quiescent}, // Current state
{Money.class, null,
showTotal, VM.collecting}},
{ {VM.collecting}, // Current state
{Quit.quit, null,
returnChange, VM.quiescent},
{Money.class, null,
showTotal, VM.collecting},
{FirstDigit.class, null,
showDigit, VM.selecting}},
{ {VM.selecting}, // Current state
{Quit.quit, null,
{SecondDigit.class, notEnough,
clearSelection, VM.collecting},
{SecondDigit.class, itemNotAvailable,
clearSelection, VM.unavailable},
{SecondDigit.class, itemAvailable,
dispense, VM.wantMore}},
{ {VM.unavailable}, // Current state
{Quit.quit, null,
{ {VM.wantMore}, // Current state
{Quit.quit, null,
});
}
} ///:~
Testing the machine

//: statemachine:vendingmachine:VendingMachineTest.java
// Demonstrates use of StateMachine.java
package statemachine.vendingmachine;
import statemachine2.*;
public class VendingMachineTest extends TestCase {
VendingMachine vm = new VendingMachine();
Input[] inputs = {
Money.quarter,
Money.quarter,
Money.dollar,
FirstDigit.A,
SecondDigit.two,
FirstDigit.A,
SecondDigit.two,
FirstDigit.C,
SecondDigit.three,
FirstDigit.D,
SecondDigit.one,
Quit.quit,
};
for(int i = 0; i < inputs.length; i++)
vm.nextState(inputs[i]);
}
junit.textui.TestRunner.run(VendingMachineTest.class);
}
} ///:~
Tools
Another approach, as your state machine gets bigger, is to use an automation tool
whereby you congure a table and let the tool generate the state machine code for
you. This can be created yourself using a language like Python, but there are also
free, open-source tools such as Libero, at http://www.imatix.com.
Table-driven code: conguration

exibility
Table-driven code using anonymous inner
classes
See ListPerformance.java example in TIJ from Chapter 9
Also GreenHouse.java
Exercises
1.
Apply TransitionTable.java to the Washer problem.
2.
Create a StateMachine system whereby the current state along with input
information determines the next state that the system will be in. To do this,
each state must store a reference back to the proxy object (the state
controller) so that it can request the state change. Use a HashMap to create
a table of states, where the key is a String naming the new state and the
value is the new state object. Inside each state subclass override a method
nextState( ) that has its own state-transition table. The input to nextState( )
should be a single word that comes from a text le containing one word per
line.
3.
Modify the previous exercise so that the state machine can be congured by
creating/modifying a single multi-dimensional array.
4.
Modify the mood exercise from the previous session so that it becomes a
state machine using StateMachine.java
5.
Create an elevator state machine system using StateMachine.java
6.
7.
Create a heating/air-conditioning system using StateMachine.java

A generator is an object that produces other objects, just like a factory,
except that the generator function doesnt require any arguments. Create a
MouseMoveGenerator which produces correct MouseMove actions as
outputs each time the generator function is called (that is, the mouse must
move in the proper sequence, thus the possible moves are based on the
previous move its another state machine). Add a method iterator( ) to
produce an iterator, but this method should take an int argument that
species the number of moves to produce before hasNext( ) returns false.
Pattern refactoring
This chapter will look at the process of solving a problem by applying design
patterns in an evolutionary fashion. That is, a rst cut design will be used for the
initial solution, and then this solution will be examined and various design patterns
will be applied to the problem (some of which will work, and some of which wont).
The key question that will always be asked in seeking improved solutions is what
will change?
This process is similar to what Martin Fowler talks about in his book Refactoring:
[12]
Improving the Design of Existing Code
(although he tends to talk about pieces
of code more than pattern-level designs). You start with a solution, and then when
you discover that it doesnt continue to meet your needs, you x it. Of course, this is
a natural tendency but in computer programming its been extremely dicult to
accomplish with procedural programs, and the acceptance of the idea that we can
refactor code and designs adds to the body of proof that object-oriented
programming is a good thing.
Simulating the trash recycler

The nature of this problem is that the trash is thrown unclassied into a single bin,
so the specic type information is lost. But later, the specic type information must
be recovered to properly sort the trash. In the initial solution, RTTI (described in
Chapter 12 of Thinking in Java, 2nd edition) is used.
This is not a trivial design because it has an added constraint. Thats what makes it
interestingits more like the messy problems youre likely to encounter in your
work. The extra constraint is that the trash arrives at the trash recycling plant all
mixed together. The program must model the sorting of that trash. This is where
RTTI comes in: you have a bunch of anonymous pieces of trash, and the program
gures out exactly what type they are.
//: refactor:recyclea:RecycleA.java
// Recycling with RTTI.
package refactor.recyclea;
import java.util.*;
import java.io.*;
abstract class Trash {

private double weight;
Trash(double wt) { weight = wt; }
abstract double getValue();
double getWeight() { return weight; }
// Sums the value of Trash in a bin:
static void sumValue(Iterator it) {
double val = 0.0f;
// One kind of RTTI:
// A dynamically-checked cast
Trash t = (Trash)it.next();
// Polymorphism in action:
val += t.getWeight() * t.getValue();
System.out.println(
"weight of " +
// Using RTTI to get type
// information about the class:
t.getClass().getName() +
" = " + t.getWeight());
}
System.out.println("Total value = " + val);
}
}
class Aluminum extends Trash {
static double val = 1.67f;
Aluminum(double wt) { super(wt); }
double getValue() { return val; }
static void setValue(double newval) {
val = newval;
}
}
class Paper extends Trash {
Paper(double wt) { super(wt); }
val = newval;
}
}
class Glass extends Trash {
Glass(double wt) { super(wt); }
val = newval;
}
}
public class RecycleA extends TestCase {
Collection
bin = new ArrayList(),

glassBin = new ArrayList(),
paperBin = new ArrayList(),
alBin = new ArrayList();
public RecycleA() {
// Fill up the Trash bin:
for(int i = 0; i < 30; i++)
case 0 :
bin.add(new
Aluminum(rand.nextDouble() * 100));
break;
case 1 :
bin.add(new
Paper(rand.nextDouble() * 100));
break;
case 2 :
bin.add(new
Glass(rand.nextDouble() * 100));
}
}
Iterator sorter = bin.iterator();
// Sort the Trash:
while(sorter.hasNext()) {
Object t = sorter.next();
// RTTI to show class membership:
if(t instanceof Aluminum)
alBin.add(t);
if(t instanceof Paper)
paperBin.add(t);
if(t instanceof Glass)
glassBin.add(t);
}
Trash.sumValue(alBin.iterator());
Trash.sumValue(paperBin.iterator());
Trash.sumValue(glassBin.iterator());
Trash.sumValue(bin.iterator());
}
junit.textui.TestRunner.run(RecycleA.class);
}
} ///:~
In the source code listings available for this book, this le will be placed in the
subdirectory recyclea that branches o from the subdirectory refactor. The
unpacking tool takes care of placing it into the correct subdirectory. The reason for
doing this is that this chapter rewrites this particular example a number of times
and by putting each version in its own directory (using the default package in each
directory so that invoking the program is easy), the class names will not clash.
Several ArrayList objects are created to hold Trash references. Of course,
ArrayLists actually hold Objects so theyll hold anything at all. The reason they
hold Trash (or something derived from Trash) is only because youve been careful
to not put in anything except Trash. If you do put something wrong into the
ArrayList, you wont get any compile-time warnings or errorsyoull nd out only
via an exception at run time.
When the Trash references are added, they lose their specic identities and become
simply Object references (they are upcast). However, because of polymorphism
the proper behavior still occurs when the dynamically-bound methods are called
through the Iterator sorter, once the resulting Object has been cast back to
Trash. sumValue( ) also takes an Iterator to perform operations on every object in
the ArrayList.
It looks silly to upcast the types of Trash into a container holding base type
references, and then turn around and downcast. Why not just put the trash into the
appropriate receptacle in the rst place? (Indeed, this is the whole enigma of
recycling). In this program it would be easy to repair, but sometimes a systems
structure and exibility can benet greatly from downcasting.
The program satises the design requirements: it works. This might be ne as long
as its a one-shot solution. However, a useful program tends to evolve over time, so
you must ask, What if the situation changes? For example, cardboard is now a
valuable recyclable commodity, so how will that be integrated into the system
(especially if the program is large and complicated). Since the above type-check
coding in the switch statement could be scattered throughout the program, you
must go nd all that code every time a new type is added, and if you miss one the
compiler wont give you any help by pointing out an error.
The key to the misuse of RTTI here is that every type is tested. If youre looking for
only a subset of types because that subset needs special treatment, thats probably
ne. But if youre hunting for every type inside a switch statement, then youre
probably missing an important point, and denitely making your code less
maintainable. In the next section well look at how this program evolved over
several stages to become much more exible. This should prove a valuable example
in program design.
Improving the design

The solutions in Design Patterns are organized around the question What will
change as this program evolves? This is usually the most important question that
you can ask about any design. If you can build your system around the answer, the
results will be two-pronged: not only will your system allow easy (and inexpensive)
maintenance, but you might also produce components that are reusable, so that
other systems can be built more cheaply. This is the promise of object-oriented
programming, but it doesnt happen automatically; it requires thought and insight
on your part. In this section well see how this process can happen during the
renement of a system.
The answer to the question What will change? for the recycling system is a
common one: more types will be added to the system. The goal of the design, then,
is to make this addition of types as painless as possible. In the recycling program,
wed like to encapsulate all places where specic type information is mentioned, so
(if for no other reason) any changes can be localized to those encapsulations. It
turns out that this process also cleans up the rest of the code considerably.
Make more objects
This brings up a general object-oriented design principle that I rst heard spoken by
Grady Booch: If the design is too complicated, make more objects. This is
simultaneously counterintuitive and ludicrously simple, and yet its the most useful
guideline Ive found. (You might observe that making more objects is often
equivalent to add another level of indirection.) In general, if you nd a place with
messy code, consider what sort of class would clean that up. Often the side eect of
cleaning up the code will be a system that has better structure and is more exible.
Consider rst the place where Trash objects are created, which is a switch
statement inside main( ):
for(int i = 0; i < 30; i++)
switch((int)(rand.nextInt(3)) {
case 0 :
bin.add(new
Aluminum(rand.nextDouble() * 100));
break;
case 1 :
bin.add(new
Paper(rand.nextDouble() * 100));
break;
case 2 :
bin.add(new
Glass(rand.nextDouble() * 100));
}
This is denitely messy, and also a place where you must change code whenever a
new type is added. If new types are commonly added, a better solution is a single
method that takes all of the necessary information and produces a reference to an
object of the correct type, already upcast to a trash object. In Design Patterns this is
broadly referred to as a creational pattern (of which there are several). The specic
pattern that will be applied here is a variant of the Factory Method. Here, the
factory method is a static member of Trash, but more commonly it is a method that
is overridden in the derived class.
The idea of the factory method is that you pass it the essential information it needs
to know to create your object, then stand back and wait for the reference (already
upcast to the base type) to pop out as the return value. From then on, you treat the
object polymorphically. Thus, you never even need to know the exact type of object
thats created. In fact, the factory method hides it from you to prevent accidental
misuse. If you want to use the object without polymorphism, you must explicitly use
RTTI and casting.
But theres a little problem, especially when you use the more complicated
approach (not shown here) of making the factory method in the base class and
overriding it in the derived classes. What if the information required in the derived
class requires more or dierent arguments? Creating more objects solves this
problem. To implement the factory method, the Trash class gets a new method
called factory. To hide the creational data, theres a new class called Messenger
that carries all of the necessary information for the factory method to create the
appropriate Trash object (weve started referring to Messenger as a design pattern,
but its simple enough that you may not choose to elevate it to that status). Heres a
simple implementation of Messenger:
class Messenger {
int type;
// Must change this to add another type:

static nal int MAX_NUM = 4;
double data;
Messenger(int typeNum, double val) {
type = typeNum % MAX_NUM;
data = val;
}
}
A Messenger objects only job is to hold information for the factory( ) method.
Now, if theres a situation in which factory( ) needs more or dierent information
to create a new type of Trash object, the factory( ) interface doesnt need to be
changed. The Messenger class can be changed by adding new data and new
constructors, or in the more typical object-oriented fashion of subclassing.
The factory( ) method for this simple example looks like this:
static Trash factory(Messenger i) {
switch(i.type) {
default: // To quiet the compiler
case 0:
return new Aluminum(i.data);
case 1:
return new Paper(i.data);
case 2:
return new Glass(i.data);
// Two lines here:
case 3:
return new Cardboard(i.data);
}
}
Here, the determination of the exact type of object is simple, but you can imagine a
more complicated system in which factory( ) uses an elaborate algorithm. The
point is that its now hidden away in one place, and you know to come to this place
when you add new types.
The creation of new objects is now much simpler in main( ):
for(int i = 0; i < 30; i++)
bin.add(
Trash.factory(
new Messenger(
rand.nextInt(Messenger.MAX_NUM),
rand.nextDouble() * 100)));
A Messenger object is created to pass the data into factory( ), which in turn
produces some kind of Trash object on the heap and returns the reference thats
added to the ArrayList bin. Of course, if you change the quantity and type of
argument, this statement will still need to be modied, but that can be eliminated if
the creation of the Messenger object is automated. For example, an ArrayList of
arguments can be passed into the constructor of a Messenger object (or directly
into a factory( ) call, for that matter). This requires that the arguments be parsed
and checked at run time, but it does provide the greatest exibility.
You can see from this code what vector of change problem the factory is
responsible for solving: if you add new types to the system (the change), the only
code that must be modied is within the factory, so the factory isolates the eect of
that change.
A pattern for prototyping creation

A problem with the design above is that it still requires a central location where all
the types of the objects must be known: inside the factory( ) method. If new types
are regularly being added to the system, the factory( ) method must be changed for
each new type. When you discover something like this, it is useful to try to go one
step further and move all of the information about the typeincluding its
creationinto the class representing that type. This way, the only thing you need to
do to add a new type to the system is to inherit a single class.
To move the information concerning type creation into each specic type of Trash,
the prototype pattern (from the Design Patterns book) will be used. The general
idea is that you have a master sequence of objects, one of each type youre
interested in making. The objects in this sequence are used only for making new
objects, using an operation thats not unlike the clone( ) scheme built into Javas
root class Object. In this case, well name the cloning method tClone( ). When
youre ready to make a new object, presumably you have some sort of information
that establishes the type of object you want to create, then you move through the
master sequence comparing your information with whatever appropriate
information is in the prototype objects in the master sequence. When you nd one
that matches your needs, you clone it.
In this scheme there is no hard-coded information for creation. Each object knows
how to expose appropriate information and how to clone itself. Thus, the factory( )
method doesnt need to be changed when a new type is added to the system.
One approach to the problem of prototyping is to add a number of methods to
support the creation of new objects. However, in Java 1.1 theres already support for
creating new objects if you have a reference to the Class object. With Java 1.1
reection (introduced in Chapter 12 of Thinking in Java, 2nd edition) you can call a
constructor even if you have only a reference to the Class object. This is the perfect
solution for the prototyping problem.
The list of prototypes will be represented indirectly by a list of references to all the
Class objects you want to create. In addition, if the prototyping fails, the factory( )
method will assume that its because a particular Class object wasnt in the list, and
it will attempt to load it. By loading the prototypes dynamically like this, the Trash
class doesnt need to know what types it is working with, so it doesnt need any
modications when you add new types. This allows it to be easily reused throughout
the rest of the chapter.
//: refactor:trash:Trash.java
// Base class for Trash recycling examples.
package refactor.trash;
import java.util.*;
public abstract class Trash {
private double weight;
public Trash(double wt) { weight = wt; }
public Trash() {}
public abstract double getValue();
public double getWeight() { return weight; }

// Sums the value of Trash given an
// Iterator to any container of Trash:
public static void sumValue(Iterator it) {
double val = 0.0f;
// One kind of RTTI: A dynamically-checked cast
Trash t = (Trash)it.next();
val += t.getWeight() * t.getValue();
System.out.println(
"weight of " +
// Using RTTI to get type
// information about the class:
t.getClass().getName() + " = " + t.getWeight());
}
System.out.println("Total value = " + val);
}
public static class Messenger {
public String id;
public double data;
public Messenger(String name, double val) {
id = name;
data = val;
}
}
// Remainder of class provides
// support for prototyping:
private static List trashTypes = new ArrayList();
public static Trash factory(Messenger info) {
Iterator it = trashTypes.iterator();
// Somehow determine the new type
// to create, and create one:
Class tc = (Class)it.next();
if (tc.getName().indexOf(info.id) != -1) {
try {
// Get the dynamic constructor method
// that takes a double argument:
Constructor ctor = tc.getConstructor(
new Class[]{ double.class });
// Call the constructor
// to create a new object:
return (Trash)ctor.newInstance(
new Object[]{new Double(info.data)});
"Cannot Create Trash", e);
}
}
}
// Class was not in the list. Try to load it,
// but it must be in your class path!
try {
System.out.println("Loading " + info.id);
trashTypes.add(Class.forName(info.id));
throw new RuntimeException("Prototype not found", e);
}
// Loaded successfully.
// Recursive call should work:
return factory(info);
}
} ///:~
The basic Trash class and sumValue( ) remain as before. The rest of the class
supports the prototyping pattern. You rst see two inner classes (which are made
static, so they are inner classes only for code organization purposes) describing
exceptions that can occur. This is followed by an ArrayList called trashTypes,
which is used to hold the Class references.
In Trash.factory( ), the String inside the Messenger object id (a dierent version
of the Messenger class than that of the prior discussion) contains the type name of
the Trash to be created; this String is compared to the Class names in the list. If
theres a match, then thats the object to create. Of course, there are many ways to
determine what object you want to make. This one is used so that information read
in from a le can be turned into objects.
Once youve discovered which kind of Trash to create, then the reection methods
come into play. The getConstructor( ) method takes an argument thats an array of
Class references. This array represents the arguments, in their proper order, for
the constructor that youre looking for. Here, the array is dynamically created using
the Java 1.1 array-creation syntax:
new Class[] {double.class}
This code assumes that every Trash type has a constructor that takes a double
(and notice that double.class is distinct from Double.class). Its also possible, for
a more exible solution, to call getConstructors( ), which returns an array of the
possible constructors.
What comes back from getConstructor( ) is a reference to a Constructor object
(part of java.lang.reect). You call the constructor dynamically with the method
newInstance( ), which takes an array of Object containing the actual arguments.
This array is again created using the Java 1.1 syntax:
new Object[]{new Double(Messenger.data)}
In this case, however, the double must be placed inside a wrapper class so that it
can be part of this array of objects. The process of calling newInstance( ) extracts
the double, but you can see it is a bit confusingan argument might be a double
or a Double, but when you make the call you must always pass in a Double.
Fortunately, this issue exists only for the primitive types.
Once you understand how to do it, the process of creating a new object given only a
Class reference is remarkably simple. Reection also allows you to call methods in
this same dynamic fashion.
Of course, the appropriate Class reference might not be in the trashTypes list. In
this case, the return in the inner loop is never executed and youll drop out at the
end. Here, the program tries to rectify the situation by loading the Class object
dynamically and adding it to the trashTypes list. If it still cant be found something
is really wrong, but if the load is successful then the factory method is called
recursively to try again.

As youll see, the beauty of this design is that this code doesnt need to be changed,
regardless of the dierent situations it will be used in (assuming that all Trash
subclasses contain a constructor that takes a single double argument).
Trash subclasses
To t into the prototyping scheme, the only thing thats required of each new
subclass of Trash is that it contain a constructor that takes a double argument.
Java reection handles everything else.
Here are the dierent types of Trash, each in their own le but part of the Trash
package (again, to facilitate reuse within the chapter):
//: refactor:trash:Aluminum.java
// The Aluminum class with prototyping.
public class Aluminum extends Trash {
private static double val = 1.67f;
public Aluminum(double wt) { super(wt); }
public double getValue() { return val; }
public static void setValue(double newVal) {
val = newVal;
}
} ///:~
//: refactor:trash:Paper.java
// The Paper class with prototyping.
public class Paper extends Trash {
public Paper(double wt) { super(wt); }
val = newVal;
}
} ///:~
//: refactor:trash:Glass.java
// The Glass class with prototyping.
public class Glass extends Trash {
public Glass(double wt) { super(wt); }
val = newVal;
}
} ///:~
And heres a new type of Trash:
//: refactor:trash:Cardboard.java
// The Cardboard class with prototyping.
public class Cardboard extends Trash {
public Cardboard(double wt) { super(wt); }
val = newVal;
}
} ///:~
You can see that, other than the constructor, theres nothing special about any of
these classes.
Parsing Trash from an external le

The information about Trash objects will be read from an outside le. The le has
all of the necessary information about each piece of trash on a single line in the
form Trash:weight, such as:
//:! refactor:trash:Trash.dat
refactor.trash.Glass:54
refactor.trash.Paper:22
refactor.trash.Aluminum:89
refactor.trash.Cardboard:96
///:~
Note that the class path must be included when giving the class names, otherwise
the class will not be found.
This le is read using the previously-dened StringList tool, and each line is picked
aparat using the String method indexOf( ) to produce the index of the :. This is
rst used with the String method substring( ) to extract the name of the trash
type, and next to get the weight that is turned into a double with the static
Double.valueOf( ) method. The trim( ) method removes white space at both ends
of a string.
The Trash parser is placed in a separate le since it will be reused throughout this
chapter:
//: refactor:trash:ParseTrash.java
// Parse le contents into Trash objects,
// placing each into a Fillable holder.
import java.util.*;
import java.io.*;
import com.bruceeckel.util.StringList;
public class ParseTrash {
public static void
llBin(String lePath, Fillable bin) {
Iterator it = new StringList(lePath).iterator();
String line = (String)it.next();
String type = line.substring(0,
line.indexOf(':')).trim();
double weight = Double.valueOf(
line.substring(line.indexOf(':') + 1)
.trim()).doubleValue();
bin.addTrash(
Trash.factory(new Trash.Messenger(type, weight)));
}
}
// Special case to handle Collection:
public static void
llBin(String lePath, Collection bin) {
llBin(lePath, new FillableCollection(bin));
}
} ///:~
In RecycleA.java, an ArrayList was used to hold the Trash objects. However, other
types of containers can be used as well. To allow for this, the rst version of
llBin( ) takes a reference to a Fillable, which is simply an interface that
supports a method called addTrash( ):
//: refactor:trash:Fillable.java
// Any object that can be lled with Trash.
public interface Fillable {

void addTrash(Trash t);
} ///:~
Anything that supports this interface can be used with llBin. Of course,
Collection doesnt implement Fillable, so it wont work. Since Collection is used
in most of the examples, it makes sense to add a second overloaded llBin( )
method that takes a Collection. Any Collection can then be used as a Fillable
object using an adapter class:
//: refactor:trash:FillableCollection.java
// Adapter that makes a Collection Fillable.
import java.util.*;
public class FillableCollection implements Fillable {
private Collection c;
public FillableCollection(Collection cc) { c = cc; }
public void addTrash(Trash t) { c.add(t); }
} ///:~
You can see that the only job of this class is to connect Fillables addTrash( )
method to Collections add( ). With this class in hand, the overloaded llBin( )
method can be used with a Collection in ParseTrash.java:
public static void
llBin(String lePath, Collection bin) {
llBin(lePath, new FillableCollection(bin));
}
This approach works for any container class thats used frequently. Alternatively,
the container class can provide its own adapter that implements Fillable. (Youll see
this later, in DynaTrash.java.)
Recycling with prototyping

Now you can see the revised version of RecycleA.java using the prototyping
technique:
//: refactor:recycleap:RecycleAP.java
// Recycling with RTTI and Prototypes.
package refactor.recycleap;
import refactor.trash.*;
import java.util.*;
public class RecycleAP extends TestCase {
Collection
bin = new ArrayList(),
glassBin = new ArrayList(),
paperBin = new ArrayList(),
alBin = new ArrayList();
public RecycleAP() {
// Fill up the Trash bin:
ParseTrash.llBin("../trash/Trash.dat", bin);
}
Iterator sorter = bin.iterator();
// Sort the Trash:
while(sorter.hasNext()) {
Object t = sorter.next();
// RTTI to show class membership:
if(t instanceof Aluminum)
alBin.add(t);
else if(t instanceof Paper)
paperBin.add(t);
else if(t instanceof Glass)
glassBin.add(t);
else
System.err.println("Unknown type " + t);
}
Trash.sumValue(alBin.iterator());
Trash.sumValue(paperBin.iterator());
Trash.sumValue(glassBin.iterator());
}
junit.textui.TestRunner.run(RecycleAP.class);
}
} ///:~
All of the Trash objects, as well as the ParseTrash and support classes, are now
part of the package refactor.trash, so they are simply imported.
The process of opening the data le containing Trash descriptions and the parsing
of that le have been wrapped into the static method ParseTrash.llBin( ), so
now its no longer a part of our design focus. You will see that throughout the rest of
the chapter, no matter what new classes are added, ParseTrash.llBin( ) will
continue to work without change, which indicates a good design.
In terms of object creation, this design does indeed severely localize the changes
you need to make to add a new type to the system. However, theres a signicant
problem in the use of RTTI that shows up clearly here. The program seems to run
ne, and yet it never detects any cardboard, even though there is cardboard in the
list! This happens because of the use of RTTI, which looks for only the types that
you tell it to look for. The clue that RTTI is being misused is that every type in the
system is being tested, rather than a single type or subset of types. As you will see
later, there are ways to use polymorphism instead when youre testing for every
type. But if you use RTTI a lot in this fashion, and you add a new type to your
system, you can easily forget to make the necessary changes in your program and
produce a dicult-to-nd bug. So its worth trying to eliminate RTTI in this case,
not just for aesthetic reasonsit produces more maintainable code.
Abstracting usage
With creation out of the way, its time to tackle the remainder of the design: where
the classes are used. Since its the act of sorting into bins thats particularly ugly
and exposed, why not take that process and hide it inside a class? This is the
principle of If you must do something ugly, at least localize the ugliness inside a
class. It looks like this:
The TrashSorter object initialization must now be changed whenever a new type of
Trash is added to the model. You could imagine that the TrashSorter class might
look something like this:
class TrashSorter extends ArrayList {
void sort(Trash t) { /* ... */ }
}
That is, TrashSorter is an ArrayList of references to ArrayLists of Trash

references, and with add( ) you can install another one, like so:
TrashSorter ts = new TrashSorter();
ts.add(new ArrayList());
Now, however, sort( ) becomes a problem. How does the statically-coded method
deal with the fact that a new type has been added? To solve this, the type
information must be removed from sort( ) so that all it needs to do is call a generic
method that takes care of the details of type. This, of course, is another way to
describe a dynamically-bound method. So sort( ) will simply move through the
sequence and call a dynamically-bound method for each ArrayList. Since the job of
this method is to grab the pieces of trash it is interested in, its called grab(Trash).
The structure now looks like:
TrashSorter needs to call each grab( ) method and get a dierent result
depending on what type of Trash the current ArrayList is holding. That is, each
ArrayList must be aware of the type it holds. The classic approach to this problem
is to create a base Trash bin class and inherit a new derived class for each
dierent type you want to hold. If Java had a parameterized type mechanism that
would probably be the most straightforward approach. But rather than hand-coding
all the classes that such a mechanism should be building for us, further observation
can produce a better approach.
A basic OOP design principle is Use data members for variation in state, use
polymorphism for variation in behavior. Your rst thought might be that the
grab( ) method certainly behaves dierently for an ArrayList that holds Paper
than for one that holds Glass. But what it does is strictly dependent on the type,
and nothing else. This could be interpreted as a dierent state, and since Java has a
class to represent type (Class) this can be used to determine the type of Trash a
particular Tbin will hold.
The constructor for this Tbin requires that you hand it the Class of your choice.
This tells the ArrayList what type it is supposed to hold. Then the grab( ) method
uses Class BinType and RTTI to see if the Trash object youve handed it matches
the type its supposed to grab.
Here is the new version of the program:
//: refactor:recycleb:RecycleB.java
// Containers that grab objects of interest.
package refactor.recycleb;
import java.util.*;
// A container that admits only the right type
// of Trash (established in the constructor):
class Tbin {
private List list = new ArrayList();
private Class type;
public Tbin(Class binType) { type = binType; }
public boolean grab(Trash t) {
// Comparing class types:
if(t.getClass().equals(type)) {
list.add(t);
return true; // Object grabbed
}
return false; // Object not grabbed
}
public Iterator iterator() {
return list.iterator();
}
}
class TbinList extends ArrayList {
void sortTrashItem(Trash t) {
Iterator e = iterator(); // Iterate over self
while(e.hasNext())
if(((Tbin)e.next()).grab(t)) return;
// Need a new Tbin for this type:
add(new Tbin(t.getClass()));
sortTrashItem(t); // Recursive call
}
}
public class RecycleB extends TestCase {
Collection bin = new ArrayList();

TbinList trashBins = new TbinList();
public RecycleB() {
ParseTrash.llBin("../trash/Trash.dat",bin);
}
Iterator it = bin.iterator();
while(it.hasNext())
trashBins.sortTrashItem((Trash)it.next());
Iterator e = trashBins.iterator();
while(e.hasNext())
Trash.sumValue(((Tbin)e.next()).iterator());
}
junit.textui.TestRunner.run(RecycleB.class);
}
} ///:~
Tbin contains a Class reference type which establishes in the constructor what
what type it should grab. The grab() method checks this type against the object you
pass it. Note that in this design, grab() only accepts Trash objects so you get
compile-time type checking on the base type, but you could also just accept Object
and it would still work.
TbinList holds a set of Tbin references, so that sort( ) can iterate through the
Tbins when its looking for a match for the Trash object youve handed it. If it
doesnt nd a match, it creates a new Tbin for the type that hasnt been found, and
makes a recursive call to itself the next time around, the new bin will be found.
Notice the genericity of this code: it doesnt change at all if new types are added. If
the bulk of your code doesnt need changing when a new type is added (or some
other change occurs) then you have an easily extensible system.
The above design is certainly satisfactory. Adding new types to the system consists
of adding or modifying distinct classes without causing code changes to be
propagated throughout the system. In addition, RTTI is not misused as it was in
RecycleA.java. However, its possible to go one step further and take a purist
viewpoint about RTTI and say that it should be eliminated altogether from the
operation of sorting the trash into bins.
To accomplish this, you must rst take the perspective that all type-dependent
activitiessuch as detecting the type of a piece of trash and putting it into the
appropriate binshould be controlled through polymorphism and dynamic binding.
The previous examples rst sorted by type, then acted on sequences of elements
that were all of a particular type. But whenever you nd yourself picking out
particular types, stop and think. The whole idea of polymorphism
(dynamically-bound method calls) is to handle type-specic information for you. So
why are you hunting for types?
The answer is something you probably dont think about: Java performs only single
dispatching. That is, if you are performing an operation on more than one object
whose type is unknown, Java will invoke the dynamic binding mechanism on only
one of those types. This doesnt solve the problem, so you end up detecting some
types manually and eectively producing your own dynamic binding behavior.
The solution is called multiple dispatching, which means setting up a conguration
such that a single method call produces more than one dynamic method call and
thus determines more than one type in the process. To get this eect, you need to
work with more than one type hierarchy: youll need a type hierarchy for each
dispatch. The following example works with two hierarchies: the existing Trash
family and a hierarchy of the types of trash bins that the trash will be placed into.
This second hierarchy isnt always obvious and in this case it needed to be created
in order to produce multiple dispatching (in this case there will be only two
dispatches, which is referred to as double dispatching).
Implementing the double dispatch

Remember that polymorphism can occur only via method calls, so if you want
double dispatching to occur, there must be two method calls: one used to determine
the type within each hierarchy. In the Trash hierarchy there will be a new method
called addToBin( ), which takes an argument of an array of TypedBin. It uses this
array to step through and try to add itself to the appropriate bin, and this is where
you'll see the double dispatch.
The new hierarchy is TypedBin, and it contains its own method called add( ) that is
also used polymorphically. But here's an additional twist: add( ) is overloaded to
take arguments of the dierent types of trash. So an essential part of the double
dispatching scheme also involves overloading.
Redesigning the program produces a dilemma: its now necessary for the base class
Trash to contain an addToBin( ) method. One approach is to copy all of the code
and change the base class. Another approach, which you can take when you dont
have control of the source code, is to put the addToBin( ) method into an
interface, leave Trash alone, and inherit new specic types of Aluminum, Paper,
Glass, and Cardboard. This is the approach that will be taken here.
Most of the classes in this design must be public, so they are placed in their own
les. Heres the interface:
//: refactor:doubledispatch:TypedBinMember.java
// An interface for adding the double
// dispatching method to the trash hierarchy
// without modifying the original hierarchy.
package refactor.doubledispatch;
interface TypedBinMember {
// The new method:
boolean addToBin(TypedBin[] tb);
} ///:~
In each particular subtype of Aluminum, Paper, Glass, and Cardboard, the

addToBin( ) method in the interface TypedBinMember is implemented, but it
looks like the code is exactly the same in each case:
//: refactor:doubledispatch:DDAluminum.java
// Aluminum for double dispatching.
public class DDAluminum extends Aluminum
implements TypedBinMember {
public DDAluminum(double wt) { super(wt); }
public boolean addToBin(TypedBin[] tb) {
for(int i = 0; i < tb.length; i++)
if(tb[i].add(this))
return true;
return false;
}
} ///:~
//: refactor:doubledispatch:DDPaper.java
// Paper for double dispatching.
public class DDPaper extends Paper
public DDPaper(double wt) { super(wt); }
if(tb[i].add(this))
return true;
return false;
}
} ///:~
//: refactor:doubledispatch:DDGlass.java
// Glass for double dispatching.
public class DDGlass extends Glass
public DDGlass(double wt) { super(wt); }
if(tb[i].add(this))
return true;
return false;
}
} ///:~
//: refactor:doubledispatch:DDCardboard.java
// Cardboard for double dispatching.
public class DDCardboard extends Cardboard
public DDCardboard(double wt) { super(wt); }
if(tb[i].add(this))
return true;
return false;
}
} ///:~
The code in each addToBin( ) calls add( ) for each TypedBin object in the array.
But notice the argument: this. The type of this is dierent for each subclass of
Trash, so the code is dierent. (Although this code will benet if a parameterized
type mechanism is ever added to Java.) So this is the rst part of the double
dispatch, because once youre inside this method you know youre Aluminum, or
Paper, etc. During the call to add( ), this information is passed via the type of this.
The compiler resolves the call to the proper overloaded version of add( ). But since
tb[i] produces a reference to the base type TypedBin, this call will end up calling a
dierent method depending on the type of TypedBin thats currently selected. That
is the second dispatch.
Heres the base class for TypedBin:
//: refactor:doubledispatch:TypedBin.java
// A container for the second dispatch.
import java.util.*;
public abstract class TypedBin {
Collection c = new ArrayList();
protected boolean addIt(Trash t) {
c.add(t);
return true;
}
public Iterator iterator() {
return c.iterator();
}
public boolean add(DDAluminum a) {
return false;
}
public boolean add(DDPaper a) {
return false;
}
public boolean add(DDGlass a) {
return false;
}
public boolean add(DDCardboard a) {
return false;
}
} ///:~
You can see that the overloaded add( ) methods all return false. If the method is
not overloaded in a derived class, it will continue to return false, and the caller
(addToBin( ), in this case) will assume that the current Trash object has not been
added successfully to a container, and continue searching for the right container.
In each of the subclasses of TypedBin, only one overloaded method is overridden,
according to the type of bin thats being created. For example, CardboardBin
overrides add(DDCardboard). The overridden method adds the trash object to its
container and returns true, while all the rest of the add( ) methods in
CardboardBin continue to return false, since they havent been overridden. This is
another case in which a parameterized type mechanism in Java would allow
automatic generation of code. (With C++ templates, you wouldnt have to explicitly
write the subclasses or place the addToBin( ) method in Trash.)
Since for this example the trash types have been customized and placed in a
dierent directory, youll need a dierent trash data le to make it work. Heres a
possible DDTrash.dat:
//:! refactor:doubledispatch:DDTrash.dat
refactor.doubledispatch.DDGlass:54
refactor.doubledispatch.DDPaper:22
refactor.doubledispatch.DDAluminum:89
refactor.doubledispatch.DDCardboard:96
///:~
Heres the rest of the program:

//: refactor:doubledispatch:DoubleDispatch.java
// Using multiple dispatching to handle more
// than one unknown type during a method call.
import java.util.*;
class AluminumBin extends TypedBin {
public boolean add(DDAluminum a) {
return addIt(a);
}
}
class PaperBin extends TypedBin {
public boolean add(DDPaper a) {
return addIt(a);
}
}
class GlassBin extends TypedBin {
public boolean add(DDGlass a) {
return addIt(a);
}
}
class CardboardBin extends TypedBin {
public boolean add(DDCardboard a) {
return addIt(a);
}
}
class TrashBinSet {
private TypedBin[] binSet = {
new AluminumBin(),
new PaperBin(),
new GlassBin(),
new CardboardBin()
};
public void sortIntoBins(Iterator it) {
TypedBinMember t = (TypedBinMember)it.next();
if(!t.addToBin(binSet))
System.err.println("Couldn't add " + t);
}
}
public TypedBin[] binSet() { return binSet; }
}
public class DoubleDispatch extends TestCase {
TrashBinSet bins = new TrashBinSet();
public DoubleDispatch() {
// ParseTrash still works, without changes:
ParseTrash.llBin("DDTrash.dat", bin);
}
// Sort from the master bin into
// the individually-typed bins:
bins.sortIntoBins(bin.iterator());
TypedBin[] tb = bins.binSet();
// Perform sumValue for each bin...
Trash.sumValue(tb[i].c.iterator());
// ... and for the master bin
}
junit.textui.TestRunner.run(DoubleDispatch.class);
}
} ///:~
TrashBinSet encapsulates all of the dierent types of TypedBins, along with the
sortIntoBins( ) method, which is where all the double dispatching takes place. You
can see that once the structure is set up, sorting into the various TypedBins is
remarkably easy. In addition, the eciency of two dynamic method calls is probably
better than any other way you could sort.
Notice the ease of use of this system in main( ), as well as the complete
independence of any specic type information within main( ). All other methods
that talk only to the Trash base-class interface will be equally invulnerable to
changes in Trash types.
The changes necessary to add a new type are relatively isolated: you modify
TypedBin, inherit the new type of Trash with its addToBin( ) method, then inherit
a new TypedBin (this is really just a copy and simple edit), and nally add a new
type into the aggregate initialization for TrashBinSet.
The Visitor pattern

Now consider applying a design pattern that has an entirely dierent goal to the
trash sorting problem.
For this pattern, we are no longer concerned with optimizing the addition of new
types of Trash to the system. Indeed, this pattern makes adding a new type of
Trash more complicated. The assumption is that you have a primary class hierarchy
that is xed; perhaps its from another vendor and you cant make changes to that
hierarchy. However, youd like to add new polymorphic methods to that hierarchy,
which means that normally youd have to add something to the base class interface.
So the dilemma is that you need to add methods to the base class, but you cant
touch the base class. How do you get around this?
The design pattern that solves this kind of problem is called a visitor (the nal one
in the Design Patterns book), and it builds on the double dispatching scheme shown
in the last section.
The visitor pattern allows you to extend the interface of the primary type by
creating a separate class hierarchy of type Visitor to virtualize the operations
performed upon the primary type. The objects of the primary type simply accept
the visitor, then call the visitors dynamically-bound method. It looks like this:
Now, if v is a Visitable reference to an Aluminum object, the code:

PriceVisitor pv = new PriceVisitor();
v.accept(pv);
uses double dispatching to cause two polymorphic method calls: the rst one to
select Aluminums version of accept( ), and the second one within accept( ) when
the specic version of visit( ) is called dynamically using the base-class Visitor
reference v.
This conguration means that new functionality can be added to the system in the
form of new subclasses of Visitor. The Trash hierarchy doesnt need to be touched.
This is the prime benet of the visitor pattern: you can add new polymorphic
functionality to a class hierarchy without touching that hierarchy (once the
accept( ) methods have been installed). Note that the benet is helpful here but not
exactly what we started out to accomplish, so at rst blush you might decide that
this isnt the desired solution.
But look at one thing thats been accomplished: the visitor solution avoids sorting
from the master Trash sequence into individual typed sequences. Thus, you can
leave everything in the single master sequence and simply pass through that
sequence using the appropriate visitor to accomplish the goal. Although this
behavior seems to be a side eect of visitor, it does give us what we want (avoiding
RTTI).
The double dispatching in the visitor pattern takes care of determining both the
type of Trash and the type of Visitor. In the following example, there are two
implementations of Visitor: PriceVisitor to both determine and sum the price, and
WeightVisitor to keep track of the weights.
You can see all of this implemented in the new, improved version of the recycling
program.
As with DoubleDispatch.java, the Trash class is left alone and a new interface is
created to add the accept( ) method:
//: refactor:trashvisitor:Visitable.java
// An interface to add visitor functionality
// to the Trash hierarchy without
// modifying the base class.
package refactor.trashvisitor;
interface Visitable {
// The new method:
void accept(Visitor v);
} ///:~
Since theres nothing concrete in the Visitor base class, it can be created as an
interface:
//: refactor:trashvisitor:Visitor.java
// The base interface for visitors.
interface Visitor {
void visit(Aluminum a);
void visit(Paper p);
void visit(Glass g);
void visit(Cardboard c);
} ///:~
A Reective Decorator
At this point, you could follow the same approach that was used for double
dispatching and create new subtypes of Aluminum, Paper, Glass, and Cardboard
that implement the accept( ) method. For example, the new Visitable Aluminum
would look like this:
//: refactor:trashvisitor:VAluminum.java
// Taking the previous approach of creating a
// specialized Aluminum for the visitor pattern.
public class VAluminum extends Aluminum
implements Visitable {
public VAluminum(double wt) { super(wt); }
public void accept(Visitor v) {
v.visit(this);
}
} ///:~
However, we seem to be encountering an explosion of interfaces: basic Trash,

special versions for double dispatching, and now more special versions for visitor.
Of course, this explosion of interfaces is arbitraryone could simply put the
additional methods in the Trash class. If we ignore that we can instead see an
opportunity to use the Decorator pattern: it seems like it should be possible to
create a Decorator that can be wrapped around an ordinary Trash object and will
produce the same interface as Trash and add the extra accept( ) method. In fact,
its a perfect example of the value of Decorator.
The double dispatch creates a problem, however. Since it relies on overloading of
both accept( ) and visit( ), it would seem to require specialized code for each
dierent version of the accept( ) method. With C++ templates, this would be fairly
easy to accomplish (since templates automatically generate type-specialized code)
but Java has no such mechanismat least it does not appear to. However, reection
allows you to determine type information at run time, and it turns out to solve many
problems that would seem to require templates (albeit not as simply). Heres the
[13]
decorator that does the trick
:
//: refactor:trashvisitor:VisitableDecorator.java
// A decorator that adapts the generic Trash
// classes to the visitor pattern.
// [ Use a Dynamic Proxy here?? ]
public class VisitableDecorator
extends Trash implements Visitable {
private Trash delegate;
private Method dispatch;
public VisitableDecorator(Trash t) {
delegate = t;
try {
dispatch = Visitor.class.getMethod (
"visit", new Class[] { t.getClass() }

);
}
}
public double getValue() {
return delegate.getValue();
}
public double getWeight() {
return delegate.getWeight();
}
public void accept(Visitor v) {
try {
dispatch.invoke(v, new Object[]{delegate});
}
}
} ///:~
[[ Description of Reection use ]]

[[Note that a Dynamic Proxy might also be applied here.]]
The only other tool we need is a new type of Fillable adapter that automatically
decorates the objects as they are being created from the original Trash.dat le. But
this might as well be a decorator itself, decorating any kind of Fillable:
//: refactor:trashvisitor:FillableVisitor.java
// Adapter Decorator that adds the visitable
// decorator as the Trash objects are
// being created.
import java.util.*;
public class FillableVisitor
implements Fillable {
private Fillable f;
public FillableVisitor(Fillable ) { f = ; }
public void addTrash(Trash t) {
f.addTrash(new VisitableDecorator(t));
}
} ///:~
Now you can wrap it around any kind of existing Fillable, or any new ones that
havent yet been created.
The rest of the program creates specic Visitor types and sends them through a
single list of Trash objects:
//: refactor:trashvisitor:TrashVisitor.java
// The "visitor" pattern with VisitableDecorators.
import java.util.*;
// Specic group of algorithms packaged
// in each implementation of Visitor:
class PriceVisitor implements Visitor {
private double alSum; // Aluminum
private double pSum; // Paper
private double gSum; // Glass
private double cSum; // Cardboard
public void visit(Aluminum al) {
double v = al.getWeight() * al.getValue();
System.out.println(
"value of Aluminum= " + v);
alSum += v;
}
public void visit(Paper p) {
double v = p.getWeight() * p.getValue();
System.out.println(
"value of Paper= " + v);
pSum += v;
}
public void visit(Glass g) {
double v = g.getWeight() * g.getValue();
System.out.println(
"value of Glass= " + v);
gSum += v;
}
public void visit(Cardboard c) {
double v = c.getWeight() * c.getValue();
System.out.println(
"value of Cardboard = " + v);
cSum += v;
}
void total() {
System.out.println(
"Total Aluminum: $" + alSum +
"\n Total Paper: $" + pSum +
"\nTotal Glass: $" + gSum +
"\nTotal Cardboard: $" + cSum +
"\nTotal: $" +
(alSum + pSum + gSum + cSum));
}
}
class WeightVisitor implements Visitor {
private double alSum; // Aluminum
private double pSum; // Paper
private double gSum; // Glass
private double cSum; // Cardboard
public void visit(Aluminum al) {
alSum += al.getWeight();
System.out.println("weight of Aluminum = "
+ al.getWeight());
}
public void visit(Paper p) {

pSum += p.getWeight();
System.out.println("weight of Paper = "
+ p.getWeight());
}
public void visit(Glass g) {
gSum += g.getWeight();
System.out.println("weight of Glass = "
+ g.getWeight());
}
public void visit(Cardboard c) {
cSum += c.getWeight();
System.out.println("weight of Cardboard = "
+ c.getWeight());
}
void total() {
System.out.println(
"Total weight Aluminum: " + alSum +
"\nTotal weight Paper: " + pSum +
"\nTotal weight Glass: " + gSum +
"\nTotal weight Cardboard: " + cSum +
"\nTotal weight: " + (alSum + pSum + gSum + cSum));
}
}
public class TrashVisitor extends TestCase {
PriceVisitor pv = new PriceVisitor();
WeightVisitor wv = new WeightVisitor();
public TrashVisitor() {
ParseTrash.llBin("../trash/Trash.dat",
new FillableVisitor(
new FillableCollection(bin)));
}
Iterator it = bin.iterator();
Visitable v = (Visitable)it.next();
v.accept(pv);
v.accept(wv);
}
pv.total();
wv.total();
}
junit.textui.TestRunner.run(TrashVisitor.class);
}
} ///:~
In Test( ), note how visitability is added by simply creating a dierent kind of bin
using the decorator. Also notice that the FillableCollection adapter has the
appearance of being used as a decorator (for ArrayList) in this situation. However,
it completely changes the interface of the ArrayList, whereas the denition of
Decorator is that the interface of the decorated class must still be there after
decoration.
Note that the shape of the client code (shown in the Test class) has changed again,
from the original approaches to the problem. Now theres only a single Trash bin.
The two Visitor objects are accepted into every element in the sequence, and they
perform their operations. The visitors keep their own internal data to tally the total
weights and prices.
Finally, theres no run time type identication other than the inevitable cast to
Trash when pulling things out of the sequence. This, too, could be eliminated with
the implementation of parameterized types in Java.
One way you can distinguish this solution from the double dispatching solution
described previously is to note that, in the double dispatching solution, only one of
the overloaded methods, add( ), was overridden when each subclass was created,
while here each one of the overloaded visit( ) methods is overridden in every
subclass of Visitor.
More coupling?
Theres a lot more code here, and theres denite coupling between the Trash
hierarchy and the Visitor hierarchy. However, theres also high cohesion within the
respective sets of classes: they each do only one thing (Trash describes Trash,
while Visitor describes actions performed on Trash), which is an indicator of a
good design. Of course, in this case it works well only if youre adding new Visitors,
but it gets in the way when you add new types of Trash.
Low coupling between classes and high cohesion within a class is denitely an
important design goal. Applied mindlessly, though, it can prevent you from
achieving a more elegant design. It seems that some classes inevitably have a
certain intimacy with each other. These often occur in pairs that could perhaps be
called couplets; for example, containers and iterators. The Trash-Visitor pair above
appears to be another such couplet.
RTTI considered harmful?

Various designs in this chapter attempt to remove RTTI, which might give you the
impression that its considered harmful (the condemnation used for poor, ill-fated
goto, which was thus never put into Java). This isnt true; it is the misuse of RTTI
that is the problem. The reason our designs removed RTTI is because the
misapplication of that feature prevented extensibility, while the stated goal was to
be able to add a new type to the system with as little impact on surrounding code as
possible. Since RTTI is often misused by having it look for every single type in your
system, it causes code to be non-extensible: when you add a new type, you have to
go hunting for all the code in which RTTI is used, and if you miss any you wont get
help from the compiler.
However, RTTI doesnt automatically create non-extensible code. Lets revisit the
trash recycler once more. This time, a new tool will be introduced, which I call a
TypeMap. It contains a HashMap that holds ArrayLists, but the interface is
simple: you can add( ) a new object, and you can get( ) an ArrayList containing all
the objects of a particular type. The keys for the contained HashMap are the types
in the associated ArrayList. The beauty of this design is that the TypeMap
dynamically adds a new pair whenever it encounters a new type, so whenever you
add a new type to the system (even if you add the new type at run time), it adapts.
Our example will again build on the structure of the Trash types in package
refactor.Trash (and the Trash.dat le used there can be used here without
change):
//: refactor:dynatrash:DynaTrash.java
// Using a Map of Lists and RTTI to automatically sort
// trash into ArrayLists. This solution, despite the
// use of RTTI, is extensible.
package refactor.dynatrash;
import java.util.*;
// Generic TypeMap works in any situation:
class TypeMap {
private Map t = new HashMap();
public void add(Object o) {
Class type = o.getClass();
if(t.containsKey(type))
((List)t.get(type)).add(o);
else {
List v = new ArrayList();
v.add(o);
t.put(type,v);
}
}
public List get(Class type) {
return (List)t.get(type);
}
public Iterator keys() {
return t.keySet().iterator();
}
}
// Adapter class to allow callbacks
// from ParseTrash.llBin():
class TypeMapAdapter implements Fillable {
TypeMap map;
public TypeMapAdapter(TypeMap tm) { map = tm; }
public void addTrash(Trash t) { map.add(t); }
}
public class DynaTrash extends TestCase {
TypeMap bin = new TypeMap();
public DynaTrash() {
ParseTrash.llBin("../trash/Trash.dat",
new TypeMapAdapter(bin));
}
Iterator keys = bin.keys();
while(keys.hasNext())
Trash.sumValue(
bin.get((Class)keys.next()).iterator());
}
junit.textui.TestRunner.run(DynaTrash.class);
}
} ///:~
Although powerful, the denition for TypeMap is simple. It contains a HashMap,

and the add( ) method does most of the work. When you add( ) a new object, the
reference for the Class object for that type is extracted. This is used as a key to
determine whether an ArrayList holding objects of that type is already present in
the HashMap. If so, that ArrayList is extracted and the object is added to the
ArrayList. If not, the Class object and a new ArrayList are added as a key-value
pair.
You can get an Iterator of all the Class objects from keys( ), and use each Class
object to fetch the corresponding ArrayList with get( ). And thats all there is to it.
The ller( ) method is interesting because it takes advantage of the design of
ParseTrash.llBin( ), which doesnt just try to ll an ArrayList but instead
anything that implements the Fillable interface with its addTrash( ) method. All
ller( ) needs to do is to return a reference to an interface that implements
Fillable, and then this reference can be used as an argument to llBin( ) like this:
ParseTrash.llBin("Trash.dat", bin.ller());
To produce this reference, an anonymous inner class (described in Chapter 8 of

Thinking in Java, 2nd edition) is used. You never need a named class to implement
Fillable, you just need a reference to an object of that class, thus this is an
appropriate use of anonymous inner classes.
An interesting thing about this design is that even though it wasnt created to
handle the sorting, llBin( ) is performing a sort every time it inserts a Trash
object into bin.
Much of class DynaTrash should be familiar from the previous examples. This
time, instead of placing the new Trash objects into a bin of type ArrayList, the bin
is of type TypeMap, so when the trash is thrown into bin its immediately sorted by
TypeMaps internal sorting mechanism. Stepping through the TypeMap and
operating on each individual ArrayList becomes a simple matter.
As you can see, adding a new type to the system wont aect this code at all, and
the code in TypeMap is completely independent. This is certainly the smallest
solution to the problem, and arguably the most elegant as well. It does rely heavily
on RTTI, but notice that each key-value pair in the HashMap is looking for only one
type. In addition, theres no way you can forget to add the proper code to this
system when you add a new type, since there isnt any code you need to add.
Summary
Coming up with a design such as TrashVisitor.java that contains a larger amount
of code than the earlier designs can seem at rst to be counterproductive. It pays to
notice what youre trying to accomplish with various designs. Design patterns in
general strive to separate the things that change from the things that stay the same.
The things that change can refer to many dierent kinds of changes. Perhaps the
change occurs because the program is placed into a new environment or because
something in the current environment changes (this could be: The user wants to
add a new shape to the diagram currently on the screen). Or, as in this case, the
change could be the evolution of the code body. While previous versions of the trash
sorting example emphasized the addition of new types of Trash to the system,
TrashVisitor.java allows you to easily add new functionality without disturbing the
Trash hierarchy. Theres more code in TrashVisitor.java, but adding new
functionality to Visitor is cheap. If this is something that happens a lot, then its
worth the extra eort and code to make it happen more easily.
The discovery of the vector of change is no trivial matter; its not something that an
analyst can usually detect before the program sees its initial design. The necessary
information will probably not appear until later phases in the project: sometimes
only at the design or implementation phases do you discover a deeper or more
subtle need in your system. In the case of adding new types (which was the focus of
most of the recycle examples) you might realize that you need a particular
inheritance hierarchy only when you are in the maintenance phase and you begin
extending the system!
One of the most important things that youll learn by studying design patterns
seems to be an about-face from what has been promoted so far in this book. That is:
OOP is all about polymorphism. This statement can produce the two-year-old
with a hammer syndrome (everything looks like a nail). Put another way, its hard
enough to get polymorphism, and once you do, you try to cast all your designs into
that one particular mold.
What design patterns say is that OOP isnt just about polymorphism. Its about
separating the things that change from the things that stay the same.
Polymorphism is an especially important way to do this, and it turns out to be
helpful if the programming language directly supports polymorphism (so you dont
have to wire it in yourself, which would tend to make it prohibitively expensive). But
design patterns in general show other ways to accomplish the basic goal, and once
your eyes have been opened to this you will begin to search for more creative
designs.
Since the Design Patterns book came out and made such an impact, people have
been searching for other patterns. You can expect to see more of these appear as
time goes on. Here are some sites recommended by Jim Coplien, of C++ fame
(http://www.bell-labs.com/~cope), who is one of the main proponents of the patterns
movement:
http://st-www.cs.uiuc.edu/users/patterns
http://c2.com/cgi/wiki
http://c2.com/ppr
http://www.bell-labs.com/people/cope/Patterns/Process/index.html
http://www.bell-labs.com/cgi-user/OrgPatterns/OrgPatterns
http://st-www.cs.uiuc.edu/cgi-bin/wikic/wikic
http://www.cs.wustl.edu/~schmidt/patterns.html
http://www.espinc.com/patterns/overview.html
Also note there has been a yearly conference on design patterns, called PLOP, that
produces a published proceedings, the third of which came out in late 1997 (all
published by Addison-Wesley).
Exercises
1.
Add a class Plastic to TrashVisitor.java.
2.
Add a class Plastic to DynaTrash.java.
3.
Create a decorator like VisitableDecorator, but for the multiple

dispatching example, along with an adapter decorator class like the one
created for VisitableDecorator. Build the rest of the example and show that
it works.
Projects
A number of more challenging projects for you to solve. [[Some of
these may turn into examples in the book, and so at some point
might disappear from here]]
Rats & Mazes

First, create a Blackboard (cite reference) which is an object on which anyone may
record information. This particular blackboard draws a maze, and is used as
information comes back about the structure of a maze from the rats that are
investigating it.
Now create the maze itself. Like a real maze, this object reveals very little
information about itself given a coordinate, it will tell you whether there are walls
or spaces in the four directions immediately surrounding that coordinate, but no
more. For starters, read the maze in from a text le but consider hunting on the
internet for a maze-generating algorithm. In any event, the result should be an
object that, given a maze coordinate, will report walls and spaces around that
coordinate. Also, you must be able to ask it for an entry point to the maze.
Finally, create the maze-investigating Rat class. Each rat can communicate with
both the blackboard to give the current information and the maze to request new
information based on the current position of the rat. However, each time a rat
reaches a decision point where the maze branches, it creates a new rat to go down
each of the branches. Each rat is driven by its own thread. When a rat reaches a
dead end, it terminates itself after reporting the results of its nal investigation to
the blackboard.
The goal is to completely map the maze, but you must also determine whether the
end condition will be naturally found or whether the blackboard must be
responsible for the decision.
An example implementation by Jeremy Meyer:
//: projects:Maze.java
package projects;
import java.util.*;
import java.io.*;
import java.awt.*;
public class Maze extends Canvas {

private Vector lines; // a line is a char array
private int width = -1;
private int height = -1;
public static void main (String [] args)
throws IOException {
if (args.length < 1) {
System.out.println("Enter lename");
System.exit(0);
}
Maze m = new Maze();
m.load(args[0]);
Frame f = new Frame();
f.setSize(m.width*20, m.height*20);
f.add(m);
Rat r = new Rat(m, 0, 0);
f.setVisible(true);
}
public Maze() {
lines = new Vector();
setBackground(Color.lightGray);
}
synchronized public boolean
isEmptyXY(int x, int y) {
if (x < 0) x += width;
if (y < 0) y += height;
// Use mod arithmetic to bring rat in line:
byte[] by =
(byte[])(lines.elementAt(y%height));
return by[x%width]==' ';
}
synchronized public void
setXY(int x, int y, byte newByte) {
if (x < 0) x += width;
if (y < 0) y += height;
byte[] by =
(byte[])(lines.elementAt(y%height));
by[x%width] = newByte;
repaint();
}
public void
load(String lename) throws IOException {
String currentLine = null;
BueredReader br = new BueredReader(
new FileReader(lename));
for(currentLine = br.readLine();
currentLine != null;
currentLine = br.readLine()) {
lines.addElement(currentLine.getBytes());
if(width < 0 ||
currentLine.getBytes().length > width)
width = currentLine.getBytes().length;
}
height = lines.size();
br.close();
}
public void update(Graphics g) { paint(g); }
public void paint (Graphics g) {
int canvasHeight = this.getBounds().height;
int canvasWidth = this.getBounds().width;
if (height < 1 || width < 1)
return; // nothing to do
int width =
((byte[])(lines.elementAt(0))).length;
for (int y = 0; y < lines.size(); y++) {
byte[] b;
b = (byte[])(lines.elementAt(y));
for (int x = 0; x < width; x++) {
switch(b[x]) {
case ' ': // empty part of maze
g.setColor(Color.lightGray);
g.llRect(
x*(canvasWidth/width),
y*(canvasHeight/height),
canvasWidth/width,
canvasHeight/height);
break;
case '*':
// a wall
g.setColor(Color.darkGray);
g.llRect(
x*(canvasWidth/width),
(canvasWidth/width)-1,
(canvasHeight/height)-1);
break;
default:
// must be rat
g.setColor(Color.red);
g.llOval(x*(canvasWidth/width),
canvasWidth/width,
canvasHeight/height);
break;
}
}
}
}
} ///:~
//: projects:Rat.java
package projects;
public class Rat {
static int ratCount = 0;
private Maze prison;
private int vertDir = 0;
private int horizDir = 0;
private int x,y;
private int myRatNo = 0;
public Rat(Maze maze, int xStart, int yStart) {
myRatNo = ratCount++;
System.out.println("Rat no." + myRatNo +

" ready to scurry.");
prison = maze;
x = xStart;
y = yStart;
prison.setXY(x,y, (byte)'R');
new Thread() {
public void run(){ scurry(); }
}.start();
}
public void scurry() {
// Try and maintain direction if possible.
// Horizontal backward
boolean ratCanMove = true;
while(ratCanMove) {
ratCanMove = false;
// South
if (prison.isEmptyXY(x, y + 1)) {
vertDir = 1; horizDir = 0;
ratCanMove = true;
}
// North
if (prison.isEmptyXY(x, y - 1))
if (ratCanMove)
new Rat(prison, x, y-1);
// Rat can move already, so give
// this choice to the next rat.
else {
vertDir = -1; horizDir = 0;
ratCanMove = true;
}
// West
if (prison.isEmptyXY(x-1, y))
if (ratCanMove)
new Rat(prison, x-1, y);
else {
vertDir = 0; horizDir = -1;
ratCanMove = true;
}
// East
if (prison.isEmptyXY(x+1, y))
if (ratCanMove)
new Rat(prison, x+1, y);
else {
vertDir = 0; horizDir = 1;
ratCanMove = true;
}
if (ratCanMove) { // Move original rat.
x += horizDir;
y += vertDir;
prison.setXY(x,y,(byte)'R');
} // If not then the rat will die.

try {
Thread.sleep(2000);
} catch(InterruptedException e) {
}
}
System.out.println("Rat no." + myRatNo +
" can't move..dying..aarrgggh.");
}
} ///:~
The maze initialization le:

//:! projects:Amaze.txt
* **
* * **
*
*** * ******* * ****
***
***
***** ********** *****
* * * * ** ** * * * ** *
* * * * ** * * * * **
*
**
*
**
* ** * ** * ** * **
*** * *** ***** * *** **
*
* **
* *
* ** * *
* ** * *
///:~
Other maze resources

A discussion of algorithms to create mazes as well as Java source code to implement
them:
http://www.mazeworks.com/mazegen/mazegen.htm
A discussion of algorithms for collision detection and other individual/group moving
behavior for autonomous physical objects:
http://www.red3d.com/cwr/steer/
XML Decorator
Create a pair of decorators for I/O Readers and Writers that encode (for the Writer
decorator) and decode (for the reader decorator) XML
A: Tools
Contains tools needed to build the book etc. Some of these may be
temporary and disappear when the code base is moved to CVS.
Ant extensions
Ant comes with an extension API so that you can create your own tasks by writing
them in Java. You can nd full details in the ocial Ant documentation and in the
published books on Ant.
As an alternative, you can simply write a Java program and call it from Ant; this way,
you dont have to learn the extension API. For example, to compile the code in this
book, we need to verify that the version of Java that the user is running is JDK 1.3 or
greater, so we created the following program:
//: com:bruceeckel:tools:CheckVersion.java
// {RunByHand}
package com.bruceeckel.tools;
public class CheckVersion {
String version = System.getProperty("java.version");
char minor = version.charAt(2);
char point = version.charAt(4);
if(minor < '3' || point < '0')
throw new RuntimeException("JDK 1.3.0 or higher " +
"is required to run the examples in this book.");
System.out.println("JDK version "+ version + " found");
}
} ///:~
This simply uses System.getProperty( ) to discover the Java version, and throws
an exception if it isnt at least 1.3. When Ant sees the exception, it will halt. Now
you can include the following in any buildle where you want to check the version
number:
<java
taskname="CheckVersion"
classname="com.bruceeckel.tools.CheckVersion"
classpath="${basedir}"
fork="true"
failonerror="true"
/>
If you use this approach to adding tools, you can write them and test them quickly,
and if its justied, you can invest the extra eort and write an Ant extension.
Array utilities
Although useful, the Arrays class stops short of being fully functional. For example,
it would be nice to be able to easily print the elements of an array without having to
code a for loop by hand every time. And as youll see, the ll( ) method only takes a
single value and places it in the array, so if you wanted, for example, to ll an array
with randomly generated numbers, ll( ) is no help.
Thus it makes sense to supplement the Arrays class with some additional utilities,
which will be placed in the package com.bruceeckel.util for convenience. These
will print an array of any type and ll an array with values or objects that are
created by an object called a generator that you can dene.

Because code needs to be created for each primitive type as well as Object, theres
[14]
For example, a generator interface is required
a lot of nearly duplicated code.
for each type because the return type of next( ) must be dierent in each case:
//: com:bruceeckel:util:Generator.java
public interface Generator { Object next(); } ///:~
//: com:bruceeckel:util:BooleanGenerator.java
public interface BooleanGenerator { boolean next(); } ///:~
//: com:bruceeckel:util:ByteGenerator.java
public interface ByteGenerator { byte next(); } ///:~
//: com:bruceeckel:util:CharGenerator.java
public interface CharGenerator { char next(); } ///:~
//: com:bruceeckel:util:ShortGenerator.java
public interface ShortGenerator { short next(); } ///:~
//: com:bruceeckel:util:IntGenerator.java
public interface IntGenerator { int next(); } ///:~
//: com:bruceeckel:util:LongGenerator.java
public interface LongGenerator { long next(); } ///:~
//: com:bruceeckel:util:FloatGenerator.java
public interface FloatGenerator { oat next(); } ///:~
//: com:bruceeckel:util:DoubleGenerator.java
public interface DoubleGenerator { double next(); } ///:~
Arrays2 contains a variety of toString( ) methods, overloaded for each type. These
methods allow you to easily print an array. The toString( ) code introduces the use
of StringBuer instead of String objects. This is a nod to eciency; when youre
assembling a string in a method that might be called a lot, its wiser to use the more
ecient StringBuer rather than the more convenient String operations. Here,
the StringBuer is created with an initial value, and Strings are appended.
Finally, the result is converted to a String as the return value:
//: com:bruceeckel:util:Arrays2.java
// A supplement to java.util.Arrays, to provide additional
// useful functionality when working with arrays. Allows
// any array to be converted to a String, and to be lled
// via a user-dened "generator" object.
import java.util.*;
public class Arrays2 {

public static String toString(boolean[] a) {
StringBuer result = new StringBuer("[");
for(int i = 0; i < a.length; i++) {
result.append(a[i]);
if(i < a.length - 1)
result.append(", ");
}
result.append("]");
return result.toString();
}
public static String toString(byte[] a) {
}
result.append("]");
}
public static String toString(char[] a) {
}
result.append("]");
}
public static String toString(short[] a) {
}
result.append("]");
}
public static String toString(int[] a) {
}
result.append("]");
}
public static String toString(long[] a) {

}
result.append("]");
}
public static String toString(oat[] a) {
}
result.append("]");
}
public static String toString(double[] a) {
}
result.append("]");
}
// Fill an array using a generator:
public static void ll(Object[] a, Generator gen) {
ll(a, 0, a.length, gen);
}
public static void
ll(Object[] a, int from, int to, Generator gen) {
for(int i = from; i < to; i++)
a[i] = gen.next();
}
public static void
ll(boolean[] a, BooleanGenerator gen) {
}
public static void
ll(boolean[] a, int from, int to,BooleanGenerator gen){
a[i] = gen.next();
}
public static void ll(byte[] a, ByteGenerator gen) {
}
public static void
ll(byte[] a, int from, int to, ByteGenerator gen) {
a[i] = gen.next();
}
public static void ll(char[] a, CharGenerator gen) {
}
public static void

ll(char[] a, int from, int to, CharGenerator gen) {
a[i] = gen.next();
}
public static void ll(short[] a, ShortGenerator gen) {
}
public static void
ll(short[] a, int from, int to, ShortGenerator gen) {
a[i] = gen.next();
}
public static void ll(int[] a, IntGenerator gen) {
}
public static void
ll(int[] a, int from, int to, IntGenerator gen) {
a[i] = gen.next();
}
public static void ll(long[] a, LongGenerator gen) {
}
public static void
ll(long[] a, int from, int to, LongGenerator gen) {
a[i] = gen.next();
}
public static void ll(oat[] a, FloatGenerator gen) {
}
public static void
ll(oat[] a, int from, int to, FloatGenerator gen) {
a[i] = gen.next();
}
public static void ll(double[] a, DoubleGenerator gen){
}
public static void
ll(double[] a, int from, int to, DoubleGenerator gen) {
a[i] = gen.next();
}
private static Random r = new Random();
public static class
RandBooleanGenerator implements BooleanGenerator {
public boolean next() { return r.nextBoolean(); }
}
public static class
RandByteGenerator implements ByteGenerator {
public byte next() { return (byte)r.nextInt(); }
}
private static String ssource =
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
private static char[] src = ssource.toCharArray();
public static class
RandCharGenerator implements CharGenerator {
public char next() {
return src[r.nextInt(src.length)];
}
}
public static class
RandStringGenerator implements Generator {
private int len;
private RandCharGenerator cg = new RandCharGenerator();
public RandStringGenerator(int length) {
len = length;
}
public Object next() {
char[] buf = new char[len];
for(int i = 0; i < len; i++)
buf[i] = cg.next();
return new String(buf);
}
}
public static class
RandShortGenerator implements ShortGenerator {
public short next() { return (short)r.nextInt(); }
}
public static class
RandIntGenerator implements IntGenerator {
private int mod = 10000;
public RandIntGenerator() {}
public RandIntGenerator(int modulo) { mod = modulo; }
public int next() { return r.nextInt(mod); }
}
public static class
RandLongGenerator implements LongGenerator {
public long next() { return r.nextLong(); }
}
public static class
RandFloatGenerator implements FloatGenerator {
public oat next() { return r.nextFloat(); }
}
public static class
RandDoubleGenerator implements DoubleGenerator {
public double next() {return r.nextDouble();}
}
} ///:~
To ll an array of elements using a generator, the ll( ) method takes a reference to

an appropriate generator interface, which has a next( ) method that will somehow
produce an object of the right type (depending on how the interface is
implemented). The ll( ) method simply calls next( ) until the desired range has
been lled. Now you can create any generator by implementing the appropriate
interface and use your generator with ll( ).
Random data generators are useful for testing, so a set of inner classes is created to
implement all the primitive generator interfaces, as well as a String generator to

represent Object. You can see that RandStringGenerator uses
RandCharGenerator to ll an array of characters, which is then turned into a
String. The size of the array is determined by the constructor argument.
To generate numbers that arent too large, RandIntGenerator defaults to a
modulus of 10,000, but the overloaded constructor allows you to choose a smaller
value.
[1]
This is a tongue-in-cheek reference to an event in China after the death of Mao-Tze Tung,
when four persons including Maos widow made a power play, and were demonized by the
Chinese Communist Party under that name.
[2]
[3]
[4]
[5]
[6]
From Mark Johnson.

But be warned: the examples are in C++.
A free email publication. See www.BruceEckel.com to subscribe.
From an email from Kevlin Henney.
Shalloway, Design Patterns Explained, and Alexandrescu, Advanced C++ Design (??)
[7]
In the Python language, all functions are already objects and so the Command pattern is
often redundant.
[8]
Page 235.
[9]
The original version of this was called JPython, but the project changed and the name
was changed to emphasize the distinctness of the new version.
[10]
Changing the registry setting python.security.respectJavaAccessibility = true to

false makes testing even more powerful because it allows the test script to use *all*
methods, even protected and package-private.
[11]
[12]
No mice were harmed in the creation of this example.

Addison-Wesley, 1999.
[13]
This was a solution created by Jaroslav Tulach in a design patterns class that I gave in
Prague.
[14]
The C++ programmer will note how much the code could be collapsed with the use of
default arguments and templates. The Python programmer will note that this entire library
would be largely unnecessary in that language.

Thinking in Patterns (Bruce Eckel)

Uploaded by

Copyright:

Available Formats

Thinking in Patterns (Bruce Eckel)

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

Thinking in Patterns (Bruce Eckel)

Uploaded by

Copyright:

Available Formats

Thinking in Patterns by Bruce Eckel

Modications in Revision 0.9:

Added proxy:PoolManager.java to make a more generic/customizeable Pool

Changed PoolManager.java to use an ArrayList (and thus does not require a

Added KissingPrincess.java to State description, as a motivational example

Added a simple Flyweight example

Simplied the enumeration in PaperScissorsRock.java

Modications in Revision 0.8:

Changed Bridge example to improve clarity.

Removed superscripts for better viewing with IE (see note above)

Modications in Revision 0.7:

Complete reorganization under headings that attempt to describe the

Addition of placeholders for the remainder of the GoF patterns

Addition of Simplifying Idioms section and examples

Addition of Builder section and examples

Removed unit-testing chapter; replaced with reference to new JUnit (which

Dramatically simplied chainofresponsibility:FindMinima.java

Added object pool/connection pool examples

Refactored small things in many examples

Recongure for new backtalk system

Replace references to TIJ2 with TIJ3

The Y2K syndrome

The pattern concept

Proxy: fronting for another object

State: changing object behavior

Strategy: choosing the algorithm at run-time

Simple Factory method

Flyweight: too many objects

Connecting dierent types

Reducing interface complexity

Package as a variation of Faade

Command: choosing the operation at run-time

Externalizing object state

Controlling Java from Jython

Using Java libraries

Creating Java classes with Jython

The Java-Python Extension (JPE)

Complex system states

Table-Driven State Machine

Simulating the trash recycler

A pattern for prototyping creation

The Visitor pattern

RTTI considered harmful?

Rats & Mazes

The Y2K syndrome

Context and composition

A word about checked exceptions

The pattern concept

Principle of least astonishment (dont be astonishing).

Make common things easy, and rare things possible

Law of Demeter: a.k.a. Dont talk to strangers. An object should only

Independence or Orthogonality. Express independent ideas independently.

Structural: designing objects to satisfy particular project constraints. These

Behavioral: objects that handle particular types of actions within a program.

The development challenge

Location of test code

"p1: x: 1 y: 2 z: 3 p2: x: 2 y: 3 z: 4");

Since this isn't inherited from a Cloneable

nal class Singleton {

public void test() {