An introduction to Python

for absolute beginners

Bob Dowling
University Computing Service
[email protected]

http://www.ucs.cam.ac.uk/docs/course-notes/unix-courses/PythonAB 1

Welcome to the Computing Service's course “Introduction to Python”.

This course is designed for people with absolutely no experience of programming.
If you have any experience in programming other languages you are going to find
this course extremely boring and you would be better off attending our course
"Python for Programmers" where we teach you how to convert what you know from
other programming languages to Python.
This course is based around Python version 3. Python has recently undergone a
change from Python 2 to Python 3 and there are some incompatibilities between
the two versions. The older versions of this course were based around Python 2
but this course is built on Python 3.
Python is named after Monty Python and its famous flying circus, not the snake. It
is a trademark of the Python Software Foundation.

1
Course outline ― 1
Who uses Python & what for
What sort of language it is

How to launch Python

Python scripts

Text
Names for values
Reading in user data
Numbers
Conversions
Comparisons
Truth & Falsehood
2

2
Course outline ― 2
Assignment
Names

Our first “real” program

Loops
if… else…

Indentation

Comments

3
Course outline ― 3
Lists
Indices
Lengths
Changing items
Extending lists
Methods
Creating lists
Testing lists
Removing from lists
for… loop
Iterables
Slices
4

4
Course outline ― 4
Files
Reading & writing

Writing our own functions

Tuples

Modules
System modules
External modules

Dictionaries

Formatted text
5

5
 Who uses Python?
On-line games

Web services

Applications

Science

Instrument control

Embedded systems

en.wikipedia.org/wiki/List_of_Python_software
6

So who uses Python and what for?

Python is used for everything! For example:
“massively multiplayer online role-playing games” like Eve Online, science
fiction’s answer to World of Warcraft,
web applications written in a framework built on Python called “Django”,
desktop applications like Blender, the 3-d animation suite which makes
considerable use of Python scripts,
the Scientific Python libraries (“SciPy”),
instrument control and
embedded systems.

6
 What sort of language is Python?

Compiled Interpreted

Explicitly Explicitly Implicitly Purely

compiled compiled compiled interpreted
to machine to byte to byte
code code code

C, C++, Java, C# Python Shell,

Fortran Perl

What sort of language is Python? The naïve view of computer languages is

that they come as either compiled languages or interpreted languages.
At the strictly compiled end languages like C, C++ or Fortran are "compiled"
(converted) into raw machine code for your computer. You point your CPU at
that code and it runs.
Slightly separate from the strictly compiled languages are languages like
Java and C# (or anything running in the .net framework). You do need to
explicitly compile these programming languages but they are compiled to
machine code for a fake CPU which is then emulated on whichever system
you run on.
Then there is Python. Python does not have to be explicitly compiled but
behind the scenes there is a system that compiles Python into an
intermediate code which is stashed away to make things faster in future.
But it does this without you having to do anything explicit yourself. So from
the point of view of how you use it you can treat it as a purely interpreted
language like the shell or Perl.

7
 Running Python ― 1

We are going to use Python from the command line either directly or
indirectly.
So, first I need a Unix command line. I will get that from the GUI by clicking
on the terminal icon in the desktop application bar.

8
 Running Python ― 2
Unix prompt

Unix command

Introductory blurb
$ python3
Python 3.2.3 (default, May 3 2012, 15:54:42)
[GCC 4.6.3] on linux2

>>> Python version

Python prompt

Now, the Unix interpreter prompts you to give it a Unix command with a
short bit of text that ends with a dollar. In the slides this will be represented
simply as a dollar.
This is a Unix prompt asking for a Unix command.
The Unix command we are going to give is “python3”. Please note that
trailing “3”. The command “python” gives you either Python 2 or Python 3
depending on what system you are on. With this command we are insisting
on getting a version of Python 3.
The Python interpreter then runs, starting with a couple of lines of blurb. In
particular it identifies the specific version of Python it is running. (3.2.3 in this
slide.)
Then it gives a prompt of its own, three “greater than” characters. The
Python 3 program is now running and it is prompting us to give a Python
command.
You cannot give a Unix command at a Python prompt (or vice versa).

9
 Quitting Python

>>> exit()

Any one
>>> quit()
of these

>>> Ctrl + D

10

There are various ways to quit interactive Python. There are two commands
which are equivalent for our purposes: quit() and exit(), but the
simplest is the key sequence [Ctrl]+[D].

10
 A first Python command
Python prompt

Python command

>>> print('Hello, world!')

Hello, world! Output

>>> Python prompt

11

There is a tradition that the first program you ever run in any language
generates the output “Hello, world!”.
I see no reason to buck tradition. Welcome to your first Python command;
we are going to output “Hello, world!”.
We type this command at the Python prompt. The convention in these slides
is that the typewriter text in bold face is what you type and the text in regular
face is what the computer prints.
We type “print” followed by an opening round brackets and the text
“Hello, world!” surrounded by single quotes, ending with a closing
round bracket and hitting the Return key, [ ↲], to indicate that we are done
with that line of instruction.
The computer responds by outputting “Hello, world!” without the
quotes.
Once it has done that it prompts us again asking for another Python
command with another Python prompt, “>>>”.

11
 Python commands
Python “function”

Round brackets
― “parentheses”
print('Hello, world!')

Function’s “argument”

print ≠ PRINT “Case sensitive”

12

This is our first Python “function”. A function takes some input, does
something with it and (optionally) returns a value. The nomenclature derives
from the mathematics of functions, but we don’t need to fixate on the
mathematical underpinnings of computer science in this course.
Our function in this case is “print” and the command necessarily starts
with the name of the function.
The inputs to the function are called its “arguments” and follow the function
inside round brackets (“parentheses”).
In this case there is a single argument, the text to print.

Note that Python, as with many but not all programming languages, is “case
sensitive”. The word “print” is not the same as “Print” or “PRINT”.

12
 Python text
Quotation marks

'Hello, world!'

The body
of the text

The quotes are not

!
part of the text itself.

13

The text itself is presented within single quotation marks. (We will discuss
the choice of quotation marks later.)
The body of the text comes within the quotes.
The quotes are not part of the text; they merely indicate to the Python
interpreter that “hey, this is text!”
Recall that the the printed output does not have quotes.

13
 Quotes?

print Command

'print' Text

14

So what do the quotes “do”?

If there are no quotes then Python will try to interpret the letters as
something it should know about. With the quotes Python simply interprets it
as literal text.
For example, without quotes the string of characters p-r-i-n-t are a
command; with quotes they are the text to be printed.

14
 Python scripts
File in home directory print('Hello, world!')

Run from Unix prompt

hello1.py

Unix prompt

Unix command
to run Python

$ python3 hello1.py Python script

Hello, world! Python script’s output

$ Unix prompt 15

So we understand the “hello, world” command and how to run it from an

interactive Python. But serious Python programs can’t be typed in live; they
need to be kept in a file and Python needs to be directed to run the
commands from that file.
These files are called “scripts” and we are now going to look at the Python
script version of “hello, world”.
In your home directories we have put a file called “hello1.py”. It is
conventional that Python scripts have file names ending with a “.py” suffix.
Some tools actually require it. We will follow this convention and you should
too.
This contains exactly the same as we were typing manually: a single line
with the print command on it.
We are going to make Python run the instructions out of the script. We call
this “running the script”.
Scripts are run from the Unix command line. We issue the Unix command
“python3” to execute Python again, but this time we add an extra word: the
name of the script, “hello1.py”.
When it runs commands from a script, python doesn’t bother with the lines
of blurb and as soon as it has run the commands (hence the output) it exists
immediately, returning control to the Unix environment, so we get a Unix
prompt back.

15
 Editing Python scripts ― 1

16

To edit scripts we will need a plain text editor. For the purposes of this
course we will use an editor called “gedit”. You are welcome to use any
text editor you are comfortable with (e.g. vi or emacs).
Unfortunately the route to launch the editor the first time is a bit clunky.
Actually, it’s a lot clunky.
1. Click on the “Dash Home” icon at the top of the icon list.
This launches a selection tool that starts blank. If you have been using some
other files then these may show as “recent files”.
2. At the bottom of the widget you will see the “house” icon highlighted.
Click on the “three library books” icon next to it.
This switches the selector to the library of applications.

16
 Editing Python scripts ― 2

17

3. Click on the “see more results” text to expose the complete set of
supported applications.
4. Scroll down until you see the “Text Editor” application. (The scroll
mouse tends to work better than dragging the rather thin scroll bar.)
5. Click the “Text Editor” icon.

17
 Editing Python scripts ― 3

18

This will launch the text editor, gedit.

18
 Editing Python scripts ― 4

19

Future launches won’t be anything like as painful. In future the text editor will
be immediately available in “Recent Apps”.

19
Progress
Interactive Python

Python scripts

print() command

Simple Python text

20

20
 Exercise 1

1. Print “Goodbye, cruel world!” from interactive Python.

2. Edit exercise1.py to print the same text.

3. Run the modified exercise1.py script.

❢ Please ask if you have questions.

2 minutes
21

During this course there will be some “lightning exercises”. These are very
quick exercises just to check that you have understood what’s been covered
in the course up to that point.
Here is your first.
First, make sure you can print text from interactive Python and quit it
afterwards.
Second, edit the exercise1.py script and run the edited version with the
different output.
This is really a test of whether you can get the basic tools running. Please
ask if you have any problems!

21
 A little more text

Full “Unicode” support print('ℏэłᏓዐ, ω☺ⲗր∂‼')

www.unicode.org/charts/ hello2.py

22

Now let’s look at a slightly different script just to see what Python can do.
Python 3 has excellent support for fully international text. (So did Python 2
but it was concealed.)
Python 3 supports what is called the “Unicode” standard, a standard
designed to allow for characters from almost every language in the world. If
you are interested in international text you need to know about the Unicode
standard. The URL shown will introduce you to the wide range of characters
supported.
The example in the slide contains the following characters:
ℏ PLANCK’S CONSTANT DIVIDED BY TWO PI
э CYRILLIC SMALL LETTER E
ł LATIN SMALL LETTER L WITH BAR
Ꮣ CHEROKEE LETTER DA
ዐ ETHIOPIC SYLLABLE PHARYNGEAL A
ω GREEK SMALL LETTER OMEGA
☺ WHITE SMILING FACE
ր ARMENIAN SMALL LETTER REH
ⲗ COPTIC SMALL LETTER LAUDA
∂ PARTIAL DIFFERENTIAL
‼ DOUBLE EXCLAMATION MARK

22
 Getting characters
˘

AltGr + Shift + # g Linux

ğ Character Selector

“LATIN SMALL
LETTER G \u011f
WITH BREVE”

23

I don’t want to get too distracted by international characters, but I ought to

mention that the hardest part of using them in Python is typically getting
them into Python in the first place.
There are three “easy” ways.
There are key combinations that generate special characters. On Linux, for
example, the combination of the three keys [AltGr], [Shift], and [#] set up the
breve accent to be applied to the next key pressed.
Perhaps easier is the “Character Selector” application. This runs like a free-
standing “insert special character” function from a word processor. You can
select a character from it, copy it to the clipboard and paste it into any
document you want.
Finally, Python supports the idea of “Unicode codes”. The two characters
“\u” followed by the hexadecimal (base 16) code for the character in the
Unicode tables will represent that character. You have all memorized your
code tables, haven’t you?

23
 Text: a “string” of characters

>>> type('Hello, world!')

<class 'str'> A string of characters

Class: string

Length: 13

Letters

str 13 H e l l o , ␣ w o r l d !
24

We will quickly look at how Python stores text, because it will give us an
introduction to how Python stores everything.
Every object in Python has a “type” (also known as a “class”).
The type for text is called “str”. This is short for “string of characters” and is
the conventional computing name for text. We typically call them “strings”.
Internally, Python allocates a chunk of computer memory to store our text. It
stores certain items together to do this. First it records that the object is a
string, because that will determine how memory is allocated subsequently.
Then it records how long the string is. Then it records the text itself.

24
 Text: “behind the scenes”
str 13 72 101 108 108 111 44 32 … 100 33

>>> '\u011f' 011f16

'ğ'

>>> ord('ğ')

287
28710

>>> chr(287)

'ğ'
ğ 25

In these slides I’m going to represent the stored text as characters because
that’s easier to read. In reality, all computers can store are numbers. Every
character has a number associated with it. You can get the number
corresponding to any character by using the ord() function and you can
get the character corresponding to any number with the chr() function.
Mathematical note:
The subscript 10 and 16 indicate the “base” of the numbers.

25
 Adding strings together: +

“Concatenation” print('Hello, ' + 'world!')

hello3.py
>>> 'Hello, ' + 'world!'

'Hello, world!'

>>>

26

Now let’s do something with strings.

If we ‘add’ two strings together Python joins them together to form a longer
string.
Python actually permits you to omit the “+”. Don’t do this.

26
 Pure concatenation
>>> 'Hello,␣' + 'world!'

'Hello, world!'

>>> 'Hello,' + '␣world!' Only simple

concatenation
'Hello, world!'

>>> 'Hello,' + 'world!' No spaces added

automatically.
'Hello,world!'
27

This joining together is very simple. If you want words split by a space you
have to put the space in.

27
 Single & double quotes

>>> 'Hello, world!' Single quotes

'Hello, world!' Single quotes

>>> "Hello, world!" Double quotes

'Hello, world!' Single quotes

28

It doesn’t matter whether we write our strings with single or double quotes
(so long as they match at the two ends). Python simply notes that we are
defining a string.

28
 Python strings: input & output

'Hello, world!'
Single or double
quotes on input.
"Hello, world!"

Create same
string object.

'Hello, world!' Single quotes on output.

29

Internally there are no quotes, just a record that the object is text.
When Python comes to display the string and declares “this is text” itself it
uses single quotes.

29
 Uses of single & double quotes

>>> print('He said "hello" to her.')

He said "hello" to her.

>>> print("He said 'hello' to her.")

He said 'hello' to her.

30

Having two sorts of quotes can be useful in certain circumstances. If you

want the text itself to include quotes of one type you can define it
surrounded by the other type.

30
 Why we need different quotes

>>> print('He said 'hello' to her.')

✘
File "<stdin>", line 1
print('He said 'hello' to her.')
^
SyntaxError: invalid syntax

31

You must mix the quotes like that. If you do not then Python will be unable to
make sense of the command.
We will look at Python’s error messages in more detail later.

31
 Adding arbitrary quotes

>>> print('He said \'hello\' to her.')

He said 'hello' to her.

\' ' Just an ordinary

character.

\" " “Escaping”

32

There is a more general solution to the “quotes within quotes” problem.

Preceding each quote within the body of the text signals to Python that this
is just an ordinary quote character and should not be treated specially.
Note that what is encoded in the string is a single character. The backslash
is a signal to the Python interpreter as its constructs the string. Once the
string is constructed, with quotes in it, the backslash’s work is done.
This process of flagging a character to be treated differently than normal is
called “escaping” the character.

32
 Putting line breaks in text
Hello,
world! What we want

>>> print('Hello, ↵ Try this

world')

>>> print('Hello, ↵

✘
File "<stdin>", line 1
print('Hello,
^
SyntaxError: EOL while
scanning string literal “EOL”: End Of Line
33

We will follow the theme of “inserting awkward characters into strings” by

looking at line breaks.
We cannot insert a line break by hitting the [ ↵] key. This signals to Python
that it should process the line so far and Python cannot; it is incomplete.

33
 Inserting “special” characters
>>> print('Hello,\nworld!')
Hello, Treated as
world! a new line.

\n Converted into a
single character.

str 13 H e l l o , ↵ w o r l d !

>>> len('Hello,\nworld!')
13 len() function: gives
the length of the object
34

Again, the backslash character comes to our rescue.

If we create a string with the sequence “\n” then Python interprets this as
the single character ↵.
Python can tell us exactly how many characters there are in a string. The
len() function tells us the length of the string in characters. There are 13
characters in the string created by 'Hello,\nworld!'. The quotes are not
part of the text and the \n becomes a single character.

34
 The backslash
Special Ordinary \' '

\" "

Ordinary Special \n ↵

\t ⇥

35

We have used backslash again, this time for a slightly different result.
Backslash before a character with special significance, such as the quote
character, makes the character “ordinary”. Used before an ordinary
character, such as “n”, it produces something “special”.
Only a few ordinary characters have special characters associated with
them but the two most commonly useful are these:
\n ↵ new line
\t ⇥ tab stop

35
 \n: unwieldy for long text
'SQUIRE TRELAWNEY, Dr. Livesey, and the\n
rest of these gentlemen having asked me\n
to write down the whole particulars\nabou
t Treasure Island, from the\nbeginning to
the end, keeping nothing\nback but the b
earings of the island,\nand that only bec
ause there is still\ntreasure not yet lif
ted, I take up my\npen in the year of gra
ce 17__ and go\nback to the time when my
father kept\nthe Admiral Benbow inn and t
he brown\nold seaman with the sabre cut f
irst\ntook up his lodging under our roof.'

Single
line 36

The “\n” trick is useful for the occasional new line. It is no use for long texts
where we want to control the formatting ourselves.

36
 Special input method for long text
'''SQUIRE TRELAWNEY, Dr. Livesey, and the
rest of these gentlemen having asked me
to write down the whole particulars
about Treasure Island, from the
beginning to the end, keeping nothing
back but the bearings of the island,
and that only because there is still
treasure not yet lifted, I take up my
pen in the year of grace 17__ and go
back to the time when my father kept
the Admiral Benbow inn and the brown
old seaman with the sabre cut first
took up his lodging under our roof. '''
Triple Multiple
quotes lines 37

Python has a special trick precisely for convenient definition of long, multi-
line text.
If you start the text with a “triple quote” then the special treatment of hitting
the [↵] key is turned off. This lets you enter text “free form” with natural line
breaks.
The triple quote is three quote characters with no spaces between them.
The quote character used can be either one but the triple use at one end
must match the one used at the other end.

37
 Python’s “secondary” prompt

>>> '''Hello,
... world'''
Python asking for more
of the same command.

38

The triple quote lets us see another Python feature. If we type a long string
raw then after we hit ↵ we see Python’s “secondary prompt”. The three dots
indicate that Python is expecting more input before it will process what it has
in hand.

38
 It’s still just text!

>>> 'Hello,\nworld!'

'Hello\nworld'
Python uses \n to represent
line breaks in strings.

>>> '''Hello,
... world!'''

'Hello\nworld' Exactly the same!

39

It is also important to note that triple quotes are just a trick for input. The text
object created is still a standard Python string. It has no memory of how it
was created.
Also note that when Python is representing the content of a string object (as
opposed to printing it) it displays new lines as “\n”.

39
 Your choice of input quotes:
Four inputs:

'Hello,\nworld!' "Hello,\nworld!"

'''Hello, """Hello,
world!''' world!"""

Same result:

str 13 H e l l o , ↵ w o r l d !
40

We have now seen four different ways to create a string with an embedded
new line. They all produce the same string object.

40
Progress
International text

print()

Concatenation of strings

Special characters

Long strings
41

41
 Exercise 2

1. Replace XXXX in exercise2.py so it prints the following

text (with the line breaks) and then run the script.

coffee
café
caffè
Kaffee

è \u00e8 AltGr + ; e

é \u00e9 AltGr + # e 3 minutes

42

There is more than one way to do this.

You can get the line breaks with \n in a single-quoted string or with literal line
breaks in a triple-quoted string. An alternative, but not in keeping with the
exercise, is to have four print() statements.
You can get the accented characters by using the \u sequences or you can
type them in literally with the keyboard combinations shown. (Linux only)

42
 Attaching names to values
“variables” message = 'Hello, world!'
print(message)
>>> message='Hello, world!'

>>> message hello4.py

'Hello, world!'

>>> type(message)

<class 'str'>

message str 13 H e l l o , ␣ w o r l d !
43

Now we will move on to a serious issue in learning any computing language:

how to handle names for values.
Compare the two scripts hello1.py and hello4.py. They both do exactly
the same thing.
We can enter the text of hello4.py manually if we want using interactive
Python, it will work equally well there.
The first line of hello4.py creates the string ‘Hello, world!’ but instead of
printing it out directly the way that hello1.py does, it attaches a name,
“message”, to it.
The second line runs the print() function, but instead of a literal string as
its argument it has this name instead. Now the name has no quotes around
it, and as I said earlier this means that Python tries to interpret it as
something it should do something with. What it does is to look up the name
and substitute in the attached value.
Whenever the name is used, Python will look it up and substitute in the
attached value.

43
 Attaching names to values
message = 'Hello, world!'
print(message)
>>> type(print)

<class 'builtin_function_or_method'> hello4.py

print function

message str 13 H e l l o , ␣ w o r l d !
44

Both “print” and “message” are the same this way. Both are names
attached to Python objects. “print” is attached to a chunk of memory
containing the definition of a function and “message” is attached to a chunk
of memory containing the text.

44
 Reading some text into a script
message = input('Yes?␣')
print(message)

$ python3 input1.py input1.py

input('Yes?␣')

Yes? Boo! message = …

Boo! print(message)

45

Now that we know how to attach names to values we can start receiving
input from the user of our script.
For this we will use the cunningly named “input()” function.
This function takes some (typically short) text as its argument. It prints this
text as a prompt and then waits for the user to type something back (and
press [↵]. It then returns whatever the user typed (without the [ ↵]) as its
value.
We can use this function on the right hand side of an assignment.
Recall that the assignment completely evaluates the right hand side first.
This means that it has to evaluate the input() function, so it prompts the
user for input and evaluates to whatever it was that the user typed.
Then the left hand side is processed and the name “message” is attached to
this value.
We can then print this input text by using the attached name.

45
 Can't read numbers directly!
$ python3 input2.py number = input('N?␣')
print(number + 1)
✘
N? 10
input2.py

Traceback (most recent call last):

File "input2.py", line 2, in <module>
print(number + 1 )
TypeError:
Can't convert 'int' object
to str implicitly

string integer 46

In the previous example script input1.py we simply took what we were

given by input() and printed it. The print() function is a flexible beast; it
can cope with almost anything.
The script hello2.py attempts to take what is given and do arithmetic with
it, namely add 1 to it. It fails, even though we type a number at input()’s
prompt.
This also gives us an error message and it’s time to investigate Python’s
error messages in more detail.
The first (the “trace back”) tells us where the error was. It was on line 2 of
the file input2.py. It also tells us what was on the line. Recall that with
syntax errors it also pointed out where in the line it realized something was
going wrong.
The second part tells us what the error was: we tried to add a string (text)
and an integer (a number). More precisely, Python couldn’t convert the
things we were adding together into things that we could add.

46
 input(): strings only
$ python3 input2.py number = input('N?␣')
print(number + 1)
✘
N? 10
input2.py

input('N?␣') str 2 1 0

≠ int 10

47

The problem is that the input() function always returns a string and the
string “character 1 followed by character 0” is not the same as the integer
ten.
We will need to convert from the string to the integer explicitly.

47
 Some more types

>>> type('Hello, world!')

<class 'str'> string of characters

>>> type(42)

<class 'int'> integer

>>> type(3.14159)

<class 'float'> floating point number

48

To date we have seen only two types: “str” and

“builtin_function_or_method”. Here are some more.
Integers (whole numbers) are a type called “int”.
Floating point numbers (how computers approximate real numbers) are a
type called “float”.
The input() function gave is a “str”. We want an “int”.

48
 Converting text to integers
>>> int('10')
str 2 1 0 int 10
10

>>> int('␣-100␣')
str 6 ␣ - 1 0 0 ␣
-100

int -100
>>> int('100-10')
✘
ValueError:
invalid literal for int() with base 10: '100-10'
49

There is a function — also called “int()” ― that converts the textual

representation of an integer into a genuine integer.
It copes with extraneous spaces and other junk around the integer but it
does not parse general expressions. It will take the textual form of a number,
but that's it.

49
 Converting text to floats

>>> float('10.0') '10.0' is a string

10.0 10.0 is a floating

point number

>>> float('␣10.␣')

10.0

50

There is a similar function called float() which creates floating point

numbers.

50
 Converting between ints and floats

>>> float(10)

10.0

>>> int(10.9)

10 Truncates
fractional part

>>> int(-10.9)

-10
51

The functions can take more than just strings, though. They can take other
numbers, for example. Note that the int() function truncates floating point
numbers.

51
 Converting into text

>>> str(10) integer string

'10'

>>> str(10.000) float string

'10.0'

52

There is also a str() function for turning things into strings.

52
 Converting between types

int() anything integer

float() anything float

str() anything string

Functions named after the type they convert into.

53

In general there is a function for each type that converts whatever it can into
that type.

53
 Reading numbers into a script
text = input('N?␣')
number = int(text)
print(number + 1)

$ python3 input3.py input3.py

N? 10

11

54

So finally we can see what we have to do to make our failing script work: we
need to add a type conversion line.

54
Progress
Names Values name = value

Types strings

integers

floating point numbers

Reading in text input(prompt)

Type conversions str() int() float()

55

55
Exercise 3

Replace the two XXXX in exercise3.py to do the following:

1. Prompt the user with the text “How much?␣”.

2. Convert the user’s answer to a floating point number.

3. Print 2.5 plus that number.

3 minutes
56

56
 Integers

ℤ {… -2, -1, 0,
1, 2, 3, 4 …}
57

Now that we have some rudimentary understanding of Python it’s time to

dive in a little deeper to the three types we have met so far.
We are going to start with the whole numbers, “integers” in technical
language.
Mathematical note:
The fancy Z is the mathematical symbol for the integers based on the
German word Zahlen.

57
 Integer addition & subtraction
>>> 20+5
25

>>> 20␣-␣5 Spaces around the

operator don’t matter.
15

“No surprises”
58

We can start our handling of integers with some very basic arithmetic.
Note that spaces around the plus and minus character are ignored.
Adding or subtracting two integers simply gives a third integer.

58
 Integer multiplication
There is no “×” on the keyboard. Linux:
AltGr + Shift + ,
Use “*” instead

>>> 20␣*␣5
100

Still no surprises
59

The arithmetical operations addition and subtraction have their usual

mathematical symbols reflected on the standard keyboard. We have a plus
sign and a minus sign (actually a “hyphen”) character and we use them.
There is no multiplication symbol on the standard keyboard. You can
generate it as one of the octopus-friendly key combinations, but it’s not a
simple key.
Instead, the computing profession has settled on using the asterisk (“*”) to
represent multiplication. On your keyboards this is [Shift]+[8].
Multiplying two integers gives a third integer.

59
 Integer division
There is no “÷” on the keyboard. Linux:
AltGr + Shift + .
Use “/” instead

>>> 20␣/␣5
4.0 This is a floating point number!

Surprise!
60

There is no division symbol on the keyboard without holding three keys

down at the same time. Again a convention has arisen to use the forward
slash character (strictly called a “solidus”) for division.
So far there have been no surprises in Python’s integer arithmetic. That
changes with division.
Not all integer division can be achieved precisely. You cannot divide 3 into 5
exactly. Because of this Python 3 always returns a type of number capable
of representing fractional values (a floating point number) even when the
division would have been exact.

60
 Integer division gives floats !
Fractions Floats sometimes
!
Consistency Floats always

>>> 20␣/␣40
0.5

>>> 20␣/␣30
0.6666666666666666

61

The designers of Python decided that consistency of output was important

and therefore because it might sometimes need to use a float it should
always use a float.
Note that even floating point numbers cannot exactly represent all fractions.
½ can be precisely represented but ⅔ cannot. We will return to the
imprecision of floating point numbers when we look at them in detail.
(If you really want to stick to integers then Python 3 offers the “//“ operator
which returns an integer answer, rounded strictly down in case of fractional
answers.)

61
 Integer powers
There is no “42” on the keyboard.

Use “**” instead

>>> 4␣**␣2 Spaces around the

operator don’t matter.
16

>>> 4*␣*2 Spaces in the

operator do!
SyntaxError: invalid syntax

62

Just as there is no mathematical symbol on the keyboard for multiplication

and division, there is no symbol at all for raising to powers. Mathematically
we represent it by superscripting the power after the number being raised.
We can’t do this on the keyboard so instead we cheat and invent our own
symbol for the operation.
Computing has split for this operation. Some languages use the circumflex
accent (“^”) and others, including Python, use a double asterisk, “**”.
Note that while spaces around the operator are ignored you can’t split the
two asterisks.

62
 Integer remainders
e.g. Is a number even or odd?

Use “%”

>>> 4␣%␣2
0

>>> 5␣%␣2
1

>>> -5␣%␣2
1 Remainder is always non-negative
63

There’s one last integer arithmetic operator we will use once in a while.
Another way to look at division is to ask what the remainder is after a
division. Python represents this concept by using the percent sign between
to numbers to represent the remainder when the first number is divided by
the second.
We will use it for one purpose only in this course: to determine if a number is
even or odd. If a number’s remainder when divided by 2 is 0 then it’s even
and if the remainder is 1 then it’s odd.

63
 How big can a Python integer be?
>>> 2**2
4

>>> 4**2
16

>>> 16**2
256

>>> 256**2
65536

>>> 65536**2
4294967296
64

Now we will look at the numbers themselves. We can ask the question “how
big can an integer be?” Mathematically, of course, there is no limit. In a
computer there are always limits. Each computer has only a finite amount of
memory to store information so there has to be a limit. We will see that
Python has no limits in principle and it is only the technical limit of the
computer that can restrict us. In practice this is never an issue for the size of
integers.
We will experiment with large integers by repeated squaring. We start with a
2 and keep squaring.

64
 How big can a Python integer be?
>>> 4294967296**2
18446744073709551616

>>> 18446744073709551616**2
340282366920938463463374607431768211456

>>> 340282366920938463463374607431768211456**2
1157920892373161954235709850086879078532699846
65640564039457584007913129639936

>>> 115792089237316195423570985008687907853269
984665640564039457584007913129639936**2
1340780792994259709957402499820584612747936582
0592393377723561443721764030073546976801874298
1669034276900318581864860508537538828119465699
65
46433649006084096

Python takes it in its stride and happily carries on.

65
 How big can a Python integer be?
10443888814131525066917527107166243825799642490473837803842334832839
53907971557456848826811934997558340890106714439262837987573438185793
60726323608785136527794595697654370999834036159013438371831442807001
18559462263763188393977127456723346843445866174968079087058037040712
84048740118609114467977783598029006686938976881787785946905630190260
94059957945343282346930302669644305902501597239986771421554169383555
98852914863182379144344967340878118726394964751001890413490084170616
There is no limit!
75093668333850551032972088269550769983616369411933015213796825837188
09183365675122131849284636812555022599830041234478486259567449219461
70238065059132456108257318353800876086221028342701976982023131690176
78006675195485079921636419370285375124784014907159135459982790513399
61155179427110683113409058427288427979155484978295432353451706522326
Except for
90613949059876930021229633956877828789484406160074129456749198230505
71642377154816321380631045902916136926708342856440730447899971901781
machine
46576347322385026725305989979599609079946920177462481771844986745565
memory
92501783290704731194331655508075682218465717463732968849128195203174
57002440926616910874148385078411929804522981857338977648103126085903
00130241346718972667321649151113160292078173803343609024380470834040
3154190336
66

The Python language has no limit on the size of integer.

66
 Big integers
2
C / C++
Fortran
4

16
int
INTEGER*4 256

long 65536
INTEGER*8
4294967296
long long
INTEGER*16 18446744073709551616

Out of the reach 3402823669209384634…

67
of C or Fortran! 63374607431768211456

This may sound rather trivial but, in fact, Python is quite exceptional in this
regard. The compiled languages have to allocate space for their integers in
advance and so place limits on how large they can grow.

67
 Floating point numbers

ℝ
✗ 1.0
0.33333333
3.14159265
2.71828182 68

And that’s it for whole numbers. Now we will look at floating point numbers,
a computer’s way of storing fractional values. This is a computer’s
approximation to the real numbers. As we will see it is a problematic
approximation.
The fancy ℝ is the mathematical symbol for the real numbers, from the
English word Real.

68
 Basic operations
>>> 20.0 + 5.0 >>> 20.0 - 5.0
25.0 15.0

>>> 20.0 * 5.0 >>> 20.0 / 5.0

100.0 4.0

>>> 20.0 ** 5.0 Equivalent to integer arithmetic

3200000.0

69

For our basic operations, floating point numbers behave just the same as
integers, using the same operators to achieve the same results. Floating
point division creates a floating point number.

69
 Floating point imprecision
>>> 1.0 / 3.0
0.3333333333333333

>>> 10.0 / 3.0

If you are relying on
3.3333333333333335 this last decimal place,
you are doing it wrong!

≈ 17 significant figures

70

So let’s see our first problem.

Floating point arithmetic is not exact, and cannot be.
Floating point numbers on modern hardware tends to give a precision of
17 significant figures. You do see the occasional issue as shown on the slide
but, frankly, if you are relying on the exact value of the final decimal place
you are doing it wroing.

70
 Hidden imprecision
>>> 0.1
!
0.1

>>> 0.1 + 0.1

0.2

>>> 0.1 + 0.1 + 0.1

Really: if you are relying on
0.30000000000000004 this last decimal place,
you are doing it wrong!
71

Not all imprecision is overt. Some of it can creep up on you.

Computers work in base 2. They can store numbers like ½ and ⅞ exactly.
But they cannot store numbers like 1/10 exactly, just like we can’t represent
⅓ exactly in a decimal expansion.
The errors in storing 1/10 are small enough, though, that they are invisible at
first. However, if they accumulate they become large enough to show
through.
Really: don’t depend on precise values.

71
 How big can a Python float be? ― 1
>>> 65536.0**2 So far, so good.
4294967296.0

>>> 4294967296.0**2 Switch to

1.8446744073709552e+19 “scientific notation”

1.8446744073709552 e+19
1.8446744073709552 ×1019

72

Let’s ask the same question about floats as we asked about integers: how
large can they be?
We will repeat our approach of repeated squaring. We fast-forward to start
at 65536.0 squared and notice that we soon get anomolous responses.
When we square 4,294,967,296 we get a number with the letter “e” in it.
User’s of pocket calculators at school may recognise this representation: it
indicates a number between 1.0 and 9.999… multiplied by a power of 10.
Floating point numbers can only hold roughly 17 significant figures of
accuracy. This means that when the integer needs more than 17 digits
something has to give.

72
 Floats are not exact
>>> 4294967296.0**2 Floating point
1.8446744073709552e+19

>>> 4294967296**2 Integer

18446744073709551616

1.8446744073709552×1019 18446744073709552000

- 18446744073709551616

384
73

The approximation isn’t bad. The error is 384 in 18446744073709551616, or

approximately 2×10-17.

73
 How big can a Python float be? ― 2
>>> 1.8446744073709552e+19**2
3.402823669209385e+38

>>> 3.402823669209385e+38**2
1.157920892373162e+77

>>> 1.157920892373162e+77**2 So far, so good.

1.3407807929942597e+154

>>> 1.3407807929942597e+154**2 Too big!

OverflowError: (34,
'Numerical result out of range')

74

If we accept that our answers are now only approximate we can keep
squaring. The “e-number” representation of scientific notation is accepted on
input by Python.
When we come to square 1.3407807929942597×10 154, though, we hit
another issue, this one fatal.
We get an “overflow error”. This means we have tried to create a floating
point number larger than Python can cope with. Under some circumstances
the “too big” problem gives rise to a sort-of-number called “inf” (standing
for “infinity”).

74
 Floating point limits

1.2345678901234567 × 10N

17 significant figures

-325 < N < 308

Positive values:
4.94065645841×10-324 < N < 8.98846567431×10307

75

Just for the record, floating point numbers have limits both in terms of the
largest and smallest numbers they can contain.

75
 Complex numbers
ℑ

ℂ z
z2

>>> (1.25+0.5j)**2
(1.3125+1.25j)

76

Python also supports complex numbers, using j for the square root of -1.
We will not use them in this course, but you ought to know they exist.

76
Progress

Arithmetic + - * / ** %

Integers No limits!

Floating point numbers Limited size

Limited precision

Complex numbers

77

77
Exercise 4

Replace the XXXX in exercise4.py to evaluate

and print out the following calculations:

1. 223 ÷ 71

2. (1 + 1/10)10

3. (1 + 1/100)100

4. (1 + 1/1000)1000
3 minutes
78

78
 Comparisons

5 < 10 ✔

5 > 10 ✘

79

We can do arithmetic on numbers. What else?

We need to be able to compare numbers.
Is 5 less than 10? Yes it is.
Is 5 greater than 10? No it isn’t.

79
 Comparisons

>>> 5 < 10 Asking the question

True
✔
>>> 5 > 10 Asking the question
False
✘

80

Now let’s see that in Python.

The “less than” character appears on the keyboard so we don’t need
anything special to express the concept like “**” for powers.
Python seems to answer the questions with “True” and “False”.

80
 True & False
>>> type(True)
<class 'bool'> “Booleans”

5 + 10 15 int

int int

5 < 10 True bool

81

The important thing to understand is that this is not just Python reporting on
a test but rather the value generated by the test. True and False are (the
only) two values of a special Python type called a “Boolean” used for
recording whether something is true or not.
Just as the “+” operator takes two integers and returns an integer value, the
“<” operator takes two integers and returns a Boolean.
Booleans are named after George Boole, whose work laid the ground for
modern algebraic logic. (His classic book’s full title is “An Investigation of the
Laws of Thought on Which are Founded the Mathematical Theories of Logic
and Probabilities”, in true Victorian style.)

81
 True & False

bool ✓ True
Only two values
bool ✗ False

82

The boolean type has precisely two values.

82
 Six comparisons
Maths Python

= == Double equals sign

≠ !=
< <
> >
≤ <=
≥ >=
83

There are six comparison operations in Python.

The equality comparison is defined in Python with a double equals sign,
“==”. The sign is doubled to distinguish comparison from assignment.
There is no “not equals” symbol on the standard keyboard. Instead, Python
uses the “!=” pair of characters. (As with “**” there must be no space
between the two characters.)
“Less than” and “greater than” we have already covered. These are
implemented directly by the “<” and “>” characters.
There are no “less than or equal to” or “greater than or equal to” keys,
though, so Python resorts to double character sequences again.

83
 Equality comparison & assignment

= name = value

Attach a name to a value.

== value1 == value2

Compare two values

84

If ever there was a “classic typo” in programming it is the confusion of “=”

and “==”. Be careful.

84
 Textual comparisons

>>> 'cat' < 'dog' Alphabetic ordering

True

>>> 'Cat' < 'cat' Uppercase before lowercase

True

>>> 'Dog' < 'cat' All uppercase before lowercase

True
85

Booleans typically arise from comparisons. We can compare more than

numbers (integers or floating point). We can also compare strings.
Text comparisons are based around the ordering of characters in the
Unicode character set. Note that all the uppercase letters in the Latin
alphabet precede all the lowercase letters. So any text that starts with an
uppercase letter counts as “less than“ any text that starts with a lowercase
letter.

85
 Ordering text is complicated
Python inequalities use Unicode character numbers.

This is over-simplistic for “real” use.

“Collation” is a whole field of computing in itself

Alphabetical order? Traditional German usage?

German: z<ö Dictionary: of < öf

Swedish: ö<z Phone book: öf < of

86

Please note, however, that this is just a comparison of strings. It is not a

general comparison of text. Ordering text is called “collation” and is a very
compicated field.
For example, different languages order characters differently. Some
countries have different orderings for different purposes.
If you want to learn more about this field, start with the Unicode page on
collation: http://www.unicode.org/reports/tr10/

86
 “Syntactic sugar”
0 < number

0 < number < 10 and

number < 10

>>> number = 5

>>> 0 < number < 10

True
87

A common requirement is to determine if a number lies in a particular range.

For this purpose, Python supports the mathematical notation a < b < c. The
inequalities can be any combination that make sense.

87
 Converting to booleans
float() Converts to floating point numbers
<class 'float'>
int() Converts to integers
<class 'int'>
str() Converts to strings
<class 'str'>

bool() Converts to booleans

<class 'bool'>
88

As with all Python types there is a function named after the type that tries to
convert arbitrary inputs into Booleans. Given that there are only two Boolean
values this tends to be a very simple function.

88
 Useful conversions

'' False Empty string

'Fred' True Non-empty string

0 False Zero

1 True Non-zero

12 True
89
-1 True

The empty string is mapped to False. Every other string is mapped to

True.
For integers, 0 is mapped to False and every other value to True.
For floating point numbers, 0.0 is mapped to False and every other value to
True.

89
 Boolean operations
bool
? bool
bool

Numbers have +, –, * … bool ✓

What do booleans have? bool ✗

90

Boolean types have their own arithmetic just like ordinary numbers. It was
the algebra of these that George Boole developed.

90
 Boolean operations ― “and”
bool
and bool
bool

True and True True Both have

to be True
True and False False

False and True False

False and False False

91

The first operation on Booleans is the “and” operator.

The and of two booleans values is True if (and only if) both its inputs are
True. If either is False then its output is False.
>>> True and False
False
>>> True and True
True

91
 Boolean operations ― “and”

>>> 4 < 5 and 6 < 7 4 < 5 True

and True
True 6 < 7 True

>>> 4 < 5 and 6 > 7 4 < 5 True

and False
False 6 > 7 False

92

We are much more likely to encounter the input booleans as the results of
comparisons that as literal values.

92
 Boolean operations ― “or”
bool
or bool
bool

True or True True At least

one has
True or False True to be True

False or True True

False or False False

93

The next boolean operation to look at is “or”. The results of this operation is
True if either of its inputs are True and False only if both its inputs are
False.

93
 Boolean operations ― “or”

>>> 4 < 5 or 6 < 7 4 < 5 True

or True
True 6 < 7 True

>>> 4 < 5 or 6 > 7 4 < 5 True

or True
True 6 > 7 False

94

Again, we tend to encounter it more often with other tests than with literal
booleans.

94
 Boolean operations ― “not”

bool not bool

not True False

not False True

95

The final operation is “not”. This takes only one input and “flips” it. True
becomes False and vice versa.

95
Boolean operations ― “not”

>>> not 6 < 7 6 < 7 True not False

False

>>> not 6 > 7 6 > 7 False not True

True

96

96
Ambiguity in expressions

3+6/3
✘ ✔
(3 + 6) / 3 3 + (6 / 3)

3 5
97

97
Division before addition

3+6/3
Division first

3+ 2
Addition second

5
98

98
“Order of precedence”
First

x**y -x +x x%y x/y x*y x-y x+y

x==y x!=y x>=y x>y x<=y x<y

not x x and y x or y
99
Last

99
Progress
Comparisons == != < > <= >=

Numerical comparison 5 < 7

Alphabetical ordering 'dig' < 'dug'

Booleans True False

Boolean operators and or not

Order of precedence

100

100
Exercise 5
Predict whether these expressions will evaluate to True or False.
Then try them.

1. 'sparrow' > 'eagle'

2. 'dog' < 'Cat' or 45 % 3 == 15

3. 60 - 45 / 5 + 10 == 1

3 minutes
101

101
 Names and values: “assignment”

>>> alpha = 100

1. alpha = 100 Python creates

int 100 an “integer 100”
in memory.

2. alpha = 100 Python attaches

the name “alpha”
to the value.
alpha int 100
102

Now let’s go back to the attaching of names to values that we saw with our
hello3.py script.
Consider the simple Python instruction shown.
Python does two things, strictly in this order:
First, it notices the literal value 100 (an integer). So Python allocates a
chunk of memory large enough and creates a Python object in it that
contains a Python integer with value 100.
Second, it creates the name “alpha” and attaches it to the integer.

102
 Assignment: right to left

alpha = 100

“RHS”: right hand side

Evaluated first

“LHS”: left hand side

Processed second

103

The key thing to note is that the processing happens right to left. Everything
to the right hand side is processed first. Only after that processing is done is
the left hand side considered.
In this example it’s trivial. It will become less trivial very soon so remember
that the right hand side is evaluated before the left hand side is even looked
at.
ps: Computing uses the phrases “left hand side” and “right hand side” so
often that they are typically written as “LHS” and “RHS”.

103
 Simple evaluations
>>> beta = 100 + 20

1. 100 + 20 RHS

int 100 function + int 20

2. 120
int

3. LHS
beta int 120
104

We can see a slightly more involved example if we put some arithmetic on

the RHS.
Again, the RHS is evaluated first.
First, Python notices three “tokens”: the 100, the name “+” and the 20. It
creates two integer objects just as it did with the previous example and it
looks up a pre-existing function object that does addition of integers.
Second, Python triggers the addition function to generate a third integer with
the value 120.
This completes the evaluation of the RHS.
Third, Python creates a name “beta” and attaches it to the freshly created
integer 120.

104
 Names on the RHS — 1
>>> gamma = alpha + 40

alpha + 40 RHS
1.

alpha
2.

int 100 function + int 40

3.
int 140
105

Now we will consider a more significantly involved example, one with a

name on the RHS.
First, Python recognizes the three tokens on the RHS. These are the name
“alpha” the “+” and the literal integer 40.
Second, it looks up the names. The “alpha” is replaced by the integer 100
and the name “+” is replaced by the actual function that does addition. The
token 40 is replaced by the actual integer 40 in memory.
Third, it runs the function to give an integer 140 object in memory.

105
 Names on the RHS — 2
>>> gamma = alpha + 40

LHS
4. gamma 140
int
106

Only after all that is the LHS considered, and the name “gamma” is created
and attached to the newly minted integer.

106
 Same name on both sides — 0

Starting
gamma int 140 position

>>> print(gamma)

140

107

Now (finally!) we get to the interesting case.

We start with the name gamma being attached to the value 140.

107
 Same name on both sides — 1
RHS
>>> gamma = gamma + 10

1. gamma + 10

2. gamma

3. 140
int function + int 10

4. 150
int
108

Then we run an assignment that has the name gamma on both the left and
right hand sides.
Again, first of all Python focuses exclusively on the RHS.
The expression “gamma + 10” is evaluated to give rise to an integer 150 in
Python memory.

108
 Same name on both sides — 2
LHS
>>> gamma = gamma + 10

int 150 RHS

5.
gamma int 140

int 150 RHS

6. gamma int 140

109

Only once that evaluation is complete does Python turn its attention to the
LHS.
The name gamma is going to be attached to the integer 150 in Python
memory. No attention is paid to where the integer 150 came from.
The name gamma is already in use and is attached to the integer 140. Its
attachment is changed to the new integer 150.

109
 Same name on both sides — 3
LHS
>>> gamma = gamma + 10

int 150

✗
7.
gamma 140 No longer
int
used.

8. gamma int 150

110

Once that is done there are no remaining references to the old integer 140.
Python automatically cleans it up, freeing the space for re-use.
This is a process called “garbage collection”. In some languages you have
to free up unused space yourself; in Python the system does it for you
automatically.

110
 “Syntactic sugar”

thing += 10 thing = thing + 10

thing -= 10 thing = thing - 10

thing *= 10 thing = thing * 10

thing /= 10 thing = thing / 10

thing **= 10 thing = thing ** 10

thing %= 10 thing = thing % 10

111

The operation of modifying a value is so common that Python, and some

other languages, have short-cuts in their syntax to make the operations
shorter to write. These operations are called “augmented assignments”.
This sort of short-cut for an operation which could already be written in the
language is sometimes called “syntactic sugar”.

111
 Deleting a name ― 1

>>> print(thing)

Traceback (most recent call last):

File "<stdin>", line 1, in <module>
NameError: name 'thing' is not defined
Unknown
variable
>>> thing = 1

>>> print(thing)

1
112

There’s one last aspect of attaching names to values hat we need to

consider. How do we delete the attachment?
First of all, let’s see what it looks like when we refer to a name that isn’t
known to the system. The error message is quite straightforward:
name 'thing' is not defined
If we then create the name and attach it to a value, the integer 1 in this
example, we can then use the name without error message.

112
 Deleting a name ― 2

>>> print(thing)

1 Known
variable
>>> del thing

>>> print(thing)

Traceback (most recent call last):

File "<stdin>", line 1, in <module>
NameError: name 'thing' is not defined
Unknown
113
variable

To delete the attachment we use the Python command “del”. The command
del thing
returns us to the state where the name is no longer known.
You can delete multiple names with the slightly extended syntax
del thing1, thing2, thing3
This is equivalent to
del thing1
del thing2
del thing3
but slightly shorter.

113
 Common mistake
a = 10
!
b = 7

a = a + b a = 17 a has now changed!

b = a - b b = a - b b ≠ 10 - 7 = 3
= 17 - 7
= 10

Later in the course: “tuples”

(a,b) = (a+b, a-b)
114

While we are looking at attaching names to values and changing those

values, we will take the time to review a common “rookie mistake” especially
among people who do linear transformations.
Suppose we want to encode the mapping

(ab)→( a−b
a+b
)
we need to be careful not to use a “half-transformed” state in the second
half of the transformation. If we calculate the new value for one coordinate
we can’t (trivially) use it in the calculation of the new value of the second
coordinate.
Later in this course when we look at “tuples” we will see a slick Python way
to fix this problem.

114
Progress

Assignment thing = thing + 10

Deletion del thing

Strictly right to left thing = thing + 10

2nd 1st

+= etc. “syntactic sugar” thing += 10

115

115
 Our first “real” program

$ python3 sqrt.py We have to write sqrt.py

Number? 2.0
1.414213562373095

First, the maths.

Then, the Python.

116

We now have enough to make a start on “real programming”. We will need

some more Python elements but we can meet them as we need them rather
than up front.
We are going to write a program that prompts for a number and then
calculates and prints out its square root.
First we will review the maths of calculating square roots so that we know
what we are coding. Then we will do it in Python.

116
 Square root of 2.0 by “bisection”

0.0 too small for √2

2.0 too large for √2

“Interval of uncertainty”

0.0 < √2 < 2.0

117

The technique we are going to use is called “bisection”. If you know this
technique you can relax for the next few slides. Please don’t snore. ☺
We are going to go through it by hand for a few iterations because when we
come to implement it in Python it is important that any confusion is due to
the Python, and not the maths the Python is implementing.

The trick is to identify a range of values that must contain the actual value
of √2. That is, we identify a lower bound that is less than √2 and an upper
bound that is greater than √2.
We then have some way (which we will explain in the following slides) to
improve that estimate by reducing the length of that interval of uncertainty.
To be precise, we will cut the uncertainty in half which is why the process is
called “bisection”.
We start by taking a lower bound of x=0, which is definitely lower than x=√2
because y=02=0<2, and an upper bound of x=2, which is definitely higher
than x=√2 because y=22=4>2.

117
 Square root of 2.0 by bisection — 1

(0.0 + 2.0) / 2.0

1.0

1.0**2
1.0

Mid-point: 1.0
118

So, what's the trick for halving the interval of uncertainty?

We find the midpoint of the interval. In this case it’s obvious: the half-way
point between x=0 and x=2 is x=1.
Then we square it to find its corresponding value of y. In this case y=12=1.

118
 Square root of 2.0 by bisection — 2

1.0**2 < 2.0

so change lower bound

1.0 < √2 < 2.0

119

So what?
Well, y=1 is less than y=2 so the corresponding x-value, x=1 makes an
acceptable lower bound for the interval of uncertainty. And if we change our
lower bound to this value then our interval only runs from x=1 to x=2 with
total length 1, rather than its original length 2.
We have halved our uncertainty.
If we can do this trick multiple times then we will reduce the interval of
uncertainty very quickly to a length so small as to be irrelevant.

119
 Square root of 2.0 by bisection — 3

(1.0 + 2.0) / 2.0

1.5

1.5**2
2.25

Mid-point: 1.5
120

So we do it again. The new mid-point lies at x=1.5. This has a corresponding

y-value of y=2.25.

120
 Square root of 2.0 by bisection — 4

1.5**2 > 2.0

so change upper bound

1.0 < √2 < 1.5

121

y=2.25 is greater than y=2 so we can use the corresponding x-value of

x=1.5 as our new upper bound. Now the interval of uncertainty is halved in
length again to be ½.

121
 Square root of 2.0 by bisection — 5

(1.0 + 1.5) / 2.0

1.25

1.25**2
1.5625

Mid-point: 1.25
122

We find the new mid-point again, x=1.25. Squaring this gives the
corresponding y-value y=1.5625.

122
 Square root of 2.0 by bisection — 6

1.25**2 < 2.0

so change lower bound

1.25 < √2 < 1.5

123

y=1.5625 is less than y=2 so we change the lower bound. Our interval of
uncertainty now has length ¼.

123
 Square root of 2.0 by bisection — 7

(1.25 + 1.5) / 2.0

1.375

1.375**2
1.890625

Mid-point: 1.375
124

And again…

124
 Square root of 2.0 by bisection — 8

1.375**2 < 2.0

so change lower bound

1.375 < √2 < 1.5

125

…to give an interval of length ⅛.

125
 Exercise 6

One more iteration.

Find the mid-point.

Square it.
Compare the square to 2.0.

Do you change the

lower or upper bound?

2 minutes
126
1.375 < √2 < 1.5

Please make sure that you understand the principle and do one more
iteration by hand.

126
 Understanding before Programming

127

Apologies for spending so long on the maths but this is a general situation.
You must understand the situation before you can program it.

127
 And now using Python!

lower = 0.0
upper = 2.0

middle = (lower+upper)/2

128

So now let's start the implementation of this process in Python.

We will do exactly the same maths, but this time with Python syntax.
First we set the end points for the interval of uncertainty and attach names
to the two x-values.
The names lower and upper are attached to the end points:
lower = 0.0
upper = 2.0
We establish the x-value of the mid-point and attach the name middle to it:
middle = (lower+upper)/2
This is exactly where we started last time, but we have attached Python
names to the values.

128
 And now using Python — 2

middle**2 < 2.0

True

lower = middle

print(lower, upper)

1.0 2.0

129

Next, we find the y‑value corresponding to the mid-point (by squaring the
x‑value 1.0) and ask if it is less than 2.0, the number whose square root we
are looking for.
middle**2 < 2.0
Recall that this will return a Python boolean value: True or False.
The squared value is 1.0 which is less than 2.0 (i.e. we get True) so we
raise the lower limit of the interval to the mid-point.
lower = middle
In this example we print the x‑value at each end of the interval to track our
progress.

129
 And now using Python — 3

middle = (lower+upper)/2

print(middle, middle**2)

1.5 2.25

130

So we do it again.
We re-calculate the x‑value for the mid-point. Note that because we
changed the value the name lower was attached to the Python instruction is
identical to the one we gave first time round:
middle = (lower+upper)/2
We do some additional printing to track progress.

130
 And now using Python — 4

middle**2 < 2.0

False

upper = middle

print(lower, upper)

1.0 1.5

131

Again, we ask if the mid-point’s y‑value (i.e. its x‑value squared) is above or
below our target of 2.0:
middle**2 < 2.0
and this time get a boolean False. Because the value is greater than 2.0
(our test evaluates to False) we change the value of the upper bound of the
interval by attaching the name upper to the x‑value of the mid-point:
upper = middle
The values being handled are exactly the same as they were when we did it
as “raw maths” but this time they have names.

131
 And now using Python — 5

middle = (lower+upper)/2

print(middle, middle**2)

1.25 1.5625

132

We now do a third iteration.

132
 And now using Python — 6

middle**2 < 2.0

True

lower = middle

print(lower, upper)

1.25 1.5

133

This time the test evaluates to True so we change lower.

133
 And now using Python — 7

middle = (lower+upper)/2

print(middle, middle**2)

1.375 1.890625

134

Fourth iteration.

134
 And now using Python — 8

middle**2 < 2.0

True

lower = middle

print(lower, upper)

1.375 1.5

135

And another True so we change lower again.

And that's enough of stepping through it manually.

135
 Looking at the Python code
lower = 0.0
upper = 2.0

middle = (lower+upper)/2
print(middle, middle**2)

middle**2 < 2.0

✔ ? ✘
lower = middle upper = middle

print(lower, upper) 136

Let’s look at the Python code we have used.

We started by initializing our interval of uncertainty:
lower = 0.0
upper = 2.0
Then we started the operations we would repeat by calculating the x‑value
of the mid-point:
middle = (lower+upper)/2
We squared this and compared the squared y‑value with 2.0, our target
value:
middle**2 < 2.0
and, based on whether this evaluated to True or False we ran either:
lower = middle
or:
upper = middle
Then we ran the iteration again.

136
 Looking at the Python structures
lower = 0.0 Set up
upper = 2.0
Loop

middle = (lower+upper)/2
print(middle, middle**2)

Choice
middle**2 < 2.0

✔ ? ✘
lower = middle upper = middle

print(lower, upper) 137

Structurally, we need to be able to do two things beyond our current

knowledge of Python. We need to be able to run certain instructions time
and time again (“looping”) and we need to be able to choose one of two
different actions depending on whether a boolean value is True or False.

137
 Looping

Before

Should the
Loop test
loop run
✘ ✔ (again)?

What is
Loop body run each
loop?

After
138

We will address looping first.

A loop has a number of components.
Strictly not part of the loop are the “before” and “after” sections but these
give context and may use values needed by the loop.
The loop itself must have some sort of test to indicate whether the loop
should run again or whether the looping can stop and control can pass to
the “after” code.
Then there must be the set of instructions that will be run each time the loop
repeats. We call this the “body” of the loop.

138
 Loop example: Count from 1 to 10

Before number = 1

Loop test number <= 10

✘ ✔ ✘ ✔

print(number)
Loop body number += 1

After print('Done!')
139

Let’s consider an example that’s simpler than our square root loop: counting
from 1 to 10.
Our “before” block initializes the attachment of a name number to a value 1:
number = 1
Our test sees if number is attached to a value less than or equal to 10 (our
final value):
number <= 10
Recall that this evaluates to a boolean value.
If the test evaluates to True then we run the loop body. This has two lines,
the first to print the value of number and the second to increase it by one:
print(number)
number += 1
If the test evaluates to False then we don’t loop and exit the structure. We
have a pointless print statement as a place-holder for more substantive
code in serious scripts:
print('Done!')
This is what we want to encode in Python.

139
 Loop example: Count from 1 to 10
number = 1 number = 1

while number <= 10 : number <= 10

✘ ✔

␣␣␣␣ print(number) print(number)

␣␣␣␣ number += 1 number += 1

print('Done!') print('Done!')
140

This is how we encode the structure in Python. We will examine it element

by element, but at first glance we observe a “while” keyword and a colon
on either wise of the test and the loop body being indented four spaces.

140
 Loop test: Count from 1 to 10
number = 1
“while” keyword
loop test
while number <= 10 : colon

␣␣␣␣ print(number)
␣␣␣␣ number += 1

print('Done!')
141

We will start by looking at what we have done to the test. The test itself is
number <= 1
which is a Python expression that evaluates to a boolean, True or False.
We precede the test expression with the Python keyword “while”. This is
what tells Python that there’s a loop coming. It must be directly followed by
an expression that evaluates to a Boolean.
We follow the test expression with a colon. This is the marker that the
expression is over and must be the last element on the line.
Note that the test evaluates to True for the loop to be run and False for the
loop to quit. We are testing for “shall the loop keep going” not “shall the loop
stop”. Python tests for while, not until.

141
 Loop body: Count from 1 to 10
number = 1

while number <= 10 :

␣␣␣␣ print(number)
␣␣␣␣ number += 1 loop body

indentation
print('Done!')
142

The loop body, the code that is repeated, appears on the lines following the
“while line”. Both its lines are indented by four spaces each.
Note that the “after” section is not indented.

142
 Loop example: Count from 1 to 10
$ python3 while1.py
number = 1
1
while number <= 10 : 2
3
print(number)
4
number += 1
5
6
print('Done!') 7
8
9
while1.py 10
Done!
$
143

First let’s check that this really works. In the file while1.py in your home
directories you will find exactly the code shown in the slide. Run it and watch
Python count from 1 to 10.

143
 Python’s use of indentation

number = 1

while number <= 10 :

Four spaces’ indentation
indicate a “block” of code.
␣␣␣␣ print(number)
␣␣␣␣ number += 1
The block forms
print('Done!') the repeated lines.

The first unindented line

marks the end of the block.
144

The four spaces of indentation are not cosmetic. A sequence of lines that
have the same indentation form blocks in Python and can be thought of as a
single unit of code. In this case both lines get run (again) or neither of them
do.
The indented block is ended by the first line that follows it with no
indentation.

144
 c.f. “legalese”

1
1(a)
1(b)
1(b)(i)

1(b)(ii)
1(b)(iii)
1(c)
2
3

145

If this seems a little alien consider the “legalese” of complex documents.

They have paragraphs, sub-paragraphs and sub-sub-paragraphs etc., each
indented relative to the one containing them.

145
 Other languages

Shell while ...

do do ... done Syntax
␣␣␣␣...
␣␣␣␣... ␣␣␣␣... Clarity
done

C while ...
{ { ... } Syntax
␣␣␣␣...
␣␣␣␣... ␣␣␣␣... Clarity
}
146

Marking blocks of code is one of the places where computing languages

differ from one another.
Some have special words that appear at the start and end of blocks like “do”
and “done”. Others use various forms of brackets like “{” and “}”.
Interestingly, programmers in these languages typically also indent code
within the blocks for visual clarity. Python simply uses the indentation for its
core syntax rather than just for ease of reading.
Purely for interest, the Shell and C versions of while1.py are also in your
home directory as while1.sh and while1.c.

146
Progress
while ... : before

while test :
test to keep looping ␣␣␣␣action1
␣␣␣␣action2
␣␣␣␣action3
code blocks
afterwards
␣␣␣␣indentation

147

147
Exercise 7

For each script: while2.py

␣␣␣␣ Predict what it will do. while3.py

␣␣␣␣ Run the script. while4.py

␣␣␣␣ Were you right? while5.py

while6.py

To kill a running script: Ctrl + C

5 minutes
148

148
 Back to our square root example
uncertainty tolerance
0.0 2.0
2.0 1.0×10–15

×½
1.0 2.0
1.0
What we want
×½
1.0 1.5
0.5

×½
1.25 1.5
0.25

×½
1.375 1.5
0.125 What we get 149

Now let’s return to our square root example. We have a loop the body of
which halves the length of the interval of uncertainty. We need to put this
into a Python loop so we need a corresponding loop test.
One typical approach is to test to see if the interval is longer than some
acceptable value. In this case we will demand that the interval have length
no longer than 10-15. (It will take 51 halvings to get from an initial length of
2.0 to something less than 10-15.)
A common name for an “acceptable uncertainty” is a “tolerance”: the amount
of uncertainty we are prepared to tolerate.

149
 Keep looping while … ?
uncertainty > tolerance

while uncertainty > tolerance :

␣␣␣␣
Do stuff.
␣␣␣␣
␣␣␣␣
␣␣␣␣

150

We need a Python test for this. Recall that Python needs a test that
evaluates to True for the loop body to run. Our test then is ”is the current
uncertainty larger than the acceptable tolerance?”
We will set a name, tolerance, to have the value 1.0e-15, calculate an
uncertainty each loop and perform the test
uncertainty > tolerance
If this is True then we need to keep going.
If it is False then we can stop.

150
 Square root: the loop
tolerance = 1.0e-15 Set up
lower = 0.0
upper = 2.0
uncertainty = upper - lower
Loop
while uncertainty > tolerance :

middle = (lower + upper)/2

Choice

print(lower, upper)
151
uncertainty = upper - lower

So, if we return to our basic structure we can now see how Python’s while
syntax fits in.
We establish a tolerance.
We establish an initial uncertainty.
We test for
uncertainty > tolerance
as the loop test.
We recalculate the uncertainty at the end of the loop block for use in the
next round of the test.

All we have to do now is to add in the choice block.

151
 Choosing
Choice
middle**2 < 2.0

✔ ? ✘
lower = middle upper = middle

middle**2 < 2.0 True or False

True lower = middle

False upper = middle 152

Once again we have a test followed by some actions. This time, however,
the test doesn’t decide whether or not to run a block of code again, but
rather which of two blocks of code to run once.
Our test ― a Python expression that evaluates to a boolean ― is simply:
middle**2 < 2.0
and if this evaluates to True then we change the lower bound:
lower = middle
and if it evaluates to False then we change the upper bound:
upper = middle
Either way, once one or the other has run we move on and do not return to
the test.

152
 Simple example
text = input('Number? ') $ python3 ifthenelse1.py
number = int(text)
Number? 8
if number % 2 == 0: Even number
print('Even number') That was fun
else:
print('Odd number')

print('That was fun!')

$ python3 ifthenelse1.py
Number? 7
Odd number
ifthenelse1.py That was fun

153

Again, we will look at an example that demonstrates just the structure.

There is a script in your home directories called ifthenelse1.py which
illustrates the structure on its own.
Mathematical note:
The script tests for a number being even by using the “remainder” operator
“%” to calculate the remainder if we divide by 2 and testing for that remainder
being 0.

153
 if…then… else… ― 1
if keyword

Test

if number % 2 == 0 : Colon

␣␣␣␣ print('Even number')

else :

␣␣␣␣ upper = middle

print('That was fun!')

154

The first line of the test looks very similar to the while syntax we have
already seen. In this case, however, it uses a new keyword: “if”.
The if keyword is followed by the test: a Python expression that evaluates
to a boolean.
The line ends with a colon.

154
 if…then… else… ― 2

if number % 2 == 0 :

␣␣␣␣ print('Even number') Run if test is True

Indentation
else :

␣␣␣␣ upper = middle

print('That was fun!')

155

The test line is immediately followed by the block of code that is run if the
test evaluates as True.
Because it is a block of code it is indented by four spaces to mark it as a
block. This example has a single line, but the block can be as long as you
want.
This block is sometimes called the “then-block” because “if the test is True
then run this block”.

155
 if…then… else… ― 3

if number % 2 == 0 :

␣␣␣␣ print('Even number')

else : else: keyword

␣␣␣␣ upper = middle Run if test is False

Indentation
print('That was fun!')

156

After the then-block comes another new keyword, “else:”. This is not
indented and is level with the “if” to indicate that it is not part of the then-
block.
It is then followed by a second block of code, know as the “else-block”. This
is the code that is run if the test evaluates as False.
Again, because it is a block of code it is indented.
The else keyword and its corresponding block are optional. You can do
nothing if the test returns False. The then-block is compulsory.

156
 if…then… else… ― 4

if number % 2 == 0 :

␣␣␣␣ print('Even number')

else :

␣␣␣␣ upper = middle

Run afterwards
print('That was fun!')
regardless of test

157

After the else-block the script continues. The print line is unindented so is
not part of the else-block. This line is run regardless of the result of the test.

157
 Our square root example
middle = (lower + upper)/2 Before

if middle**2 < 2.0 :

␣␣␣␣ lower = middle

if… block
else :

␣␣␣␣ upper = middle

print(lower, upper) After

158

Let’s return to our square root example.

Here we have the creation of a mid-point x-value followed by an if-test on it:
middle = (lower+upper)/2
if middle**2 < 2.0:
This switches between two single-line code blocks. If the test evaluates to
True then the then-block is run:
lower = middle
and if it evaluates to False then the else-block is run:
else:
upper = middle
After one or other is run the print statement is always run:
print(lower, upper)
All we have to do now is to fit it inside our while loop.

158
Progress
if ... : before

else: if test :
␣␣␣␣action1
␣␣␣␣action2
choice of two else:
code blocks ␣␣␣␣action3

afterwards
␣␣␣␣indentation

159

159
Exercise 8

For each script: ifthenelse2.py

␣␣␣␣ Predict what it will do. ifthenelse3.py

␣␣␣␣ Run the script. ifthenelse4.py

␣␣␣␣ Were you right? ifthenelse5.py

5 minutes
160

160
 Back to our example
tolerance = 1.0e-15
lower = 0.0
upper = 2.0
uncertainty = upper - lower

while uncertainty > tolerance :

middle = (lower + upper)/2
if starts
if middle**2 < 2.0 :
indented
lower = middle
else : Doubly
indented
upper = middle
print(lower, upper)
161
uncertainty = upper - lower

So how do we embed an if-test with its two code blocks inside a while-loop
as the loop’s body?
The body of the while-loop is indented four spaces. So we start the if-test
indented four spaces and make its indented blocks doubly indented.

161
 Levels of indentation
tolerance = 1.0e-15
lower = 0.0
upper = 2.0
uncertainty = upper - lower

while uncertainty > tolerance :

␣␣␣␣middle = (lower + upper)/2 4 spaces
␣␣␣␣if middle**2 < 2.0 :
␣␣␣␣␣␣␣␣lower = middle 8 spaces
␣␣␣␣else :
␣␣␣␣␣␣␣␣upper = middle
␣␣␣␣print(lower, upper)
162
␣␣␣␣uncertainty = upper - lower

So if our standard indentation is four spaces then the doubly indented

sections are indented eight spaces.
This is a simple example with only two levels of indentation. Python can
‘nest’ blocks much further than this.

162
 Trying it out
tolerance = 1.0e-15 $ python3 sqrt1.py
lower = 0.0
upper = 2.0 1.0 2.0
uncertainty = upper - lower 1.0 1.5
1.25 1.5
while uncertainty > tolerance :
1.375 1.5
middle = (lower + upper)/2
1.375 1.4375
if middle**2 < 2.0: 1.40625 1.4375
lower = middle 1.40625 1.421875
else:
upper = middle
...
print(lower, upper)
uncertainty = upper - lower 1.414213... 1.414213...

sqrt1.py
☺ 163

The file sqrt1.py in your home directories contains the code as described
in the slide. It produces a very nice approximation to the square root of 2.

163
 Script for the square root of 2.0
tolerance = 1.0e-15
lower = 0.0
upper = 2.0 √2.0
uncertainty = upper - lower

while uncertainty > tolerance :

␣␣␣␣middle = (lower + upper)/2
␣␣␣␣if middle**2 < 2.0 : √2.0
␣␣␣␣␣␣␣␣lower = middle
␣␣␣␣else :
␣␣␣␣␣␣␣␣upper = middle
␣␣␣␣print(lower, upper)
164
␣␣␣␣uncertainty = upper - lower

So now we have the script for the square root of 2. The next thing to do is to
generalize it to produce square roots of any number.

164
 Input target

text = input('Number? ')

number = float(text)

if middle**2 < number :

165

Obviously we have to input the number whose square root we want. We

have already seen how to do this and to convert it from a string into a
floating point number:
text = input('Number? ')
number = float(text)
Once we have the number the test
middle**2 < 2.0
is straightforwardly extended to
middle**2 < number

165
 Initial bounds?

lower = ?
upper = ?

x > 1.0 1.0 < √x < x

1.0 > x 1.0 > √x > x

if...then...else...

166

We have to set initial bounds for our interval of uncertainty.

This is where it is important that you think about the problem before coding.
If the number whose square root is sought is less than 1 then the square
root is bigger than the number and less than 1. If it is larger than 1 then its
square root is less than the number and bigger than 1.
In Python terms this means we can test for the number being less than 1
and set the bounds accordingly.

166
 Initial bounds

if number < 1.0 :

␣␣␣␣lower = number
␣␣␣␣upper = 1.0
else :
␣␣␣␣lower = 1.0
␣␣␣␣upper = number

167

It looks like this.

167
 Generic square root script?
text = input('Number?␣')
number = float(text)
User input

if number < 1.0:

␣␣␣␣lower = number Initialization
␣␣␣␣upper = 1.0
else:
␣␣␣␣lower = 1.0
␣␣␣␣upper = number

tolerance = 1.0e-15
uncertainty = upper - lower

while uncertainty > tolerance:

␣␣␣␣middle = (lower+upper)/2.0
␣␣␣␣if middle**2 < number: Processing
␣␣␣␣␣␣␣␣lower = middle
␣␣␣␣else:
␣␣␣␣␣␣␣␣upper = middle

␣␣␣␣uncertainty = upper - lower

168
print(lower, upper) sqrt2.py
Output

This gives us enough of a script to see the overarching structure of a script:

We start with getting the data we need from the outside world. (“input”)
Then we set up any initial state we need based on that. (“initialization”)
Then we do our processing.
Finally we reveal our results. (“output”)
Typically the processing phase takes longest to run, but note that, as here, it
is often not the majority of the lines of code.

168
 Negative numbers?

Need to catch negative numbers

if number < 0.0:

␣␣␣␣print('Number must be positive!')
␣␣␣␣exit() Quit immediately

"else" is optional

169

We can improve our code. A step missing from the previous script is “input
validation” where we check that the input makes sense. We ought to check
that we have not been asked to generate the square root of a negative
number.

169
 “Chained” tests

text = input('Number?␣') User input

number = float(text)

if number < 0.0: Input validation

␣␣␣␣print('Number must be positive!')
␣␣␣␣exit()

if number < 1.0: Initialization

␣␣␣␣lower = number
␣␣␣␣upper = 1.0
else:
␣␣␣␣lower = 1.0
␣␣␣␣upper = number

...

170

The input validation phase comes straight after the input itself.

170
 “Chained” tests ― syntactic sugar

text = input('Number?␣')
number = float(text)

if number < 0.0:

␣␣␣␣print('Number must be positive!')
␣␣␣␣exit()

elif number < 1.0: elif: “else if”

␣␣␣␣lower = number
␣␣␣␣upper = 1.0
else:
␣␣␣␣lower = 1.0
␣␣␣␣upper = number

...

171
sqrt3.py

However, it can be integrated with the initialization phase, and often is. After
all, if you can’t initialize from the input then the input isn’t valid.
This leads us to a multi-stage test of the number whose square root we
want:
Is it less than 0?
If not, is it less than 1?
If not then…
Python has an extension to the simple if…else… test to allow for the “if
not then is it…” situation.
“elif” introduces a test and a corresponding block of code. The code is
called only if the previous if… test failed and its own test passes.

171
 Without elif…

text = input('Number?␣')
number = float(text)

if number < 0.0:

␣␣␣␣print('Number is negative.')
else:
␣␣␣␣if number < 1.0:
␣␣␣␣␣␣␣␣print('Number is between zero and one.')
␣␣␣␣else:
␣␣␣␣␣␣␣␣if number < 2.0:
␣␣␣␣␣␣␣␣␣␣␣␣print('Number is between one and two.')
␣␣␣␣␣␣␣␣else:
␣␣␣␣␣␣␣␣␣␣␣␣if number < 3.0:
␣␣␣␣␣␣␣␣␣␣␣␣␣␣␣␣print('Number is between two and three.')
␣␣␣␣␣␣␣␣␣␣␣␣else:
␣␣␣␣␣␣␣␣␣␣␣␣␣␣␣␣print('Number is three or more.')
172
Stacked clauses get unwieldy

To take an extreme example, consider this multi-level test. The continual

nesting inside the else clauses causes the whole script to drift to the right.

172
 With elif…

text = input('Number?␣')
number = float(text)

if number < 0.0:

␣␣␣␣print('Number is negative.')
elif number < 1.0:
␣␣␣␣print('Number is between zero and one.')
elif number < 2.0:
␣␣␣␣print('Number is between one and two.')
elif number < 3.0:
␣␣␣␣print('Number is between two and three.')
else:
␣␣␣␣print('Number is three or more.')

173

Applying elif causes everything to slide back into place.

173
Progress
Nested structures while … :
if … :

Chained tests if … :
…
elif … :
…
Testing inputs to scripts elif … :
…
else:
…

exit() 174

174
Exercise 9
Only do the second part after exercise9.py
you have the first part working!

1. Edit the square root script to catch negative numbers.

2. Edit the square root script to ask for the tolerance.

The tolerance must be bigger than 5×10–16.
Test for that.

5 minutes
175

175
 Comments

We have written our first real Python script

What did it do?

Why did it do it?

Need to annotate the script

176

sqrt3.py is a real program.

Now imagine you pass it to someone else or put it away for twelve months
and come back to it forgetting how you wrote it in the first place.
Chances are that the reader of your script might like some hints as to what it
is doing and why.
“Comments” in computer programs are pieces of text that describe what ios
going on without getting in the way of the lines of code that are executed.

176
 Python comment character

The “hash” character

#
a.k.a. “pound”, “number”, “sharp”

Lines starting with “#” are ignored

Partial lines starting “#” are ignored

Used for annotating scripts

177

Python, in common with most other scripting languages, uses the hash
character to introduce comments. The hash character and everything
beyond it on the line is ignored.
(Strictly speaking the musical sharp character “ ♯” is not the same as “#” but
people get very sloppy with similar characters these days. This isn’t at all
relevant to Python but the author is an annoying pedant on the correct use
of characters. And don’t get him started on people who use a hyphen when
they should use an en-dash or em-dash.)

177
 Python commenting example

# Script to calculate square roots by bisection

# (c) Bob Dowling 2012. Licensed under GPL v3.0
text = input('Number?␣')
number = float(text) # Need a real number

# Test number for validity,

# set initial bounds if OK.
if number < 0.0:
␣␣␣␣print('Number must be non-negative!')
␣␣␣␣exit()
elif number < 1.0:
␣␣␣␣lower = number
␣␣␣␣upper = 1.0
else:
␣␣␣␣lower = 1.0
␣␣␣␣upper = number
178

This is what a commented script looks like.

178
 On a real Unix system…

#!/usr/bin/python3

# Script to calculate square roots by bisection

# (c) Bob Dowling 2012. Licensed under GPL v3.0
text = input('Number?␣')
number = float(text) # Need a real number

Magic line for executable files

$ fubar.py
instead of
$ python3 fubar.py
179

You may encounter a “hash pling“ first line in many imported Python scripts.
This is part of some “Unix magic” that lets us simplify our command lines.
We can’t demonstrate it here because the MCS file server doesn’t support
Unix semantics.

179
Progress

Comments

“#” character
#
180

180
Exercise 10

Comment your square root script from exercise 9.

2 minutes
181

181
 Recap: Python types so far

Whole numbers -127

Floating point numbers 3.141592653589793

Complex numbers (1.0 + 2.0j)

Text 'The cat sat on the mat.'

Booleans True False

182

We are about to introduce a new Python type, so we will take a moment to

remind ourseoves of the various Python types we have met already.

182
 Lists

[ 'hydrogen', 'helium', 'lithium', 'beryllium',

'boron', …, 'thorium', 'protactinium', 'uranium' ]

[ -3.141592653589793, -1.5707963267948966,
0.0, 1.5707963267948966, 3.141592653589793 ]

[ 2, 3, 5, 7, 11, 13, 17, 19 ]

183

The new Python type we are going to meet is called a “list”.

183
 What is a list?
hydrogen, helium, lithium, beryllium, …, protactinium, uranium

A sequence of values The names of the elements

Values stored in order Atomic number order

Individual value identified “helium” is the name of the

by position in the sequence second element
184

So what is a list?
A list is simply a sequence of values stored in a specific order with each
value identified by its position in that order.
So for an example consider the list of names of the elements up to uranium.

184
 What is a list?
2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59

A sequence of values The prime numbers

less than sixty

Values stored in order Numerical order

Individual value identified 7 is the fourth prime

by position in the sequence
185

Or the list of primes up to 60.

Note that a list must be finite.

185
 Creating a list in Python

>>> primes = [ 2, 3, 5, 7, 11, 13, 17, 19]

A literal list

>>> primes
[2, 3, 5, 7, 11, 13, 17, 19] The whole list

>>> type(primes)
<class 'list'> A Python type

186

So how might we do this in Python?

We will create a list in Python of the primes less than 20. We can do this in a
single line as shown.
A list in Python is a single Python object, albeit with multiple contents, and
has its own type, unsurprisingly called “list”.

186
 How Python presents lists

Square brackets
at the ends

[2, 3, 5, 7, 11, 13, 17, 19]

Commas between items

187

Python presents (and accepts) lists as a series of values separated by

commas, surrounded by square brackets.

187
 Square brackets

primes = [2, 3, 5, 7, 11] Literal list

188

We are going to meet square brackets used fr a lot of things so I will build
up a summary slide of their various uses. Here is use 1.

188
 Python counts from zero

“index” 0 1 2 3 4 5 6 7

“value” [2, 3, 5, 7, 11, 13, 17, 19]

189

We still need to get at individual items in the list. Each is identified by its
position in the list.
Python, in common with many programming languages (but not all) starts its
count from zero. The leading element in the list is “item number zero”. The
one that follows it is “item number one” and so on. This number, the position
in the list counting from zero, is called the “index” into the list. The plural of
“index” is “indices”.
To keep yourself sane, we strongly recommend the language “item
number 3” instead of “the fourth item”.

189
 Looking things up in a list
>>> primes = [ 2, 3, 5, 7, 11, 13, 17, 19]

0 1 2 3 4 5 6 7

[2, 3, 5, 7, 11, 13, 17, 19]

index
>>> primes[0]
2
square brackets
>>> primes[6]
17 190

So, how do we get “item number 5” from a list?

We can follow the list (or, more usually, a name attached to the list) with the
index in square brackets.

190
 Square brackets

primes = [2, 3, 5, 7, 11] Literal list

primes[3] Index into list

191

And this is the second use of square brackets.

191
 Counting from the end
>>> primes = [ 2, 3, 5, 7, 11, 13, 17, 19]

0 1 2 3 4 5 6 7

[2, 3, 5, 7, 11, 13, 17, 19]

-8 -7 -6 -5 -4 -3 -2 -1

getting at the last item

>>> primes[-1]
19 192

Python has a trick for getting at the last element of the list. Negative indices
are also valid and count backwards from the end of the list. Typically the
only case of this that is used in paractice is that the index -1 gets the last
element of the list.

192
 Inside view of a list
list
8 int 2 primes[0]

int 3 primes[1]
primes
…

int 17 primes[6]

int 19 primes[7]
193

We’ve seen these box diagrams for simple Python types already. The
structures for lists are a little more complicated but only a little. The list type
records how long it is and an ordered set of references to the actual objects
it contains.

193
 Length of a list
list
>>> len(primes)
8 8

0 len() function:
length of list
primes

Maximum
7
index is 7

194

Note that the length of a list is the number of items it contains. The largest
legitimate index is one less than that because indices count from zero.

194
 Changing a value in a list
>>> data = ['alpha', 'beta', 'gamma'] The list

>>> data[2] Initial value

'gamma'

>>> data[2] = 'G' Change value

>>> data[2] Check change

'G'
>>> data Changed list
['alpha', 'beta', 'G']
195

So far we have created lists all in one go by quoting a literal list. We can use
the indexing notation (square brackets) to change items in a list too.

195
 Changing a value in a list ― 1
Right to left
>>> data = ['alpha', 'beta', 'gamma']

list
3 str 5 a l p h a

str 4 b e t a

str 5 g a m m a

196

We can track what happens in that example in some detail. We will start with
the first line, defining the initial list.
As ever, Python assignment is done right to left. The right hand side is
evaluated as a list of three items, all of themstrings.

196
 Changing a value in a list ― 2
Right to left
>>> data = ['alpha', 'beta', 'gamma']

list
3 str 5 a l p h a
data
str 4 b e t a

str 5 g a m m a

197

This then has the name “data” attached to it.

197
 Changing a value in a list ― 3
Right to left
>>> data[2] = 'G'

list
3 str 5 a l p h a
data
str 4 b e t a

str 5 g a m m a

str 1 G New value

198

Now we come to the second line which changes one of these list items.
The right hand side is evaluated as a string containing a single characters.
That object gets created.

198
 Changing a value in a list ― 4
Right to left
>>> data[2] = 'G'

list
3 str 5 a l p h a
data
str 4 b e t a

str 5 g a m m a

No longer
str 1 G referenced
199

The assignment causes the reference within the string to be changed to

refer to the new string and to stop referring to the previous one, “gamma”.
In this case there are now no references to the “gamma” string.

199
 Changing a value in a list ― 5
Right to left
>>> data[2] = 'G'

list
3 str 5 a l p h a
data
str 4 b e t a

str 1 G
200

Python then clears out the memory used for that old string so that it can
reuse it for something else. This process is called “garbage collection” in
computing.

200
 Removing an entry from a list ― 1
>>> del data[1]

list
3 str 5 a l p h a data[0]
data

✗ str 4 b e t a data[1]

str 5 g a m m a data[2]

201

We can remove entries from the list too with the “del” keyword, just as we
removed names. The del keyword removes the reference from the list.

201
 Removing an entry from a list ― 2
>>> del data[1]

list
2 str 5 a l p h a data[0]
data
No longer
str 4 b e t a
referenced

str 5 g a m m a data[1]

202

This leaves the string “beta” no longer referenced by anything.

202
 Removing an entry from a list ― 3
>>> del data[1]

list
2 str 5 a l p h a data[0]
data

str 5 g a m m a data[1]

203

And garbage collection kicks in again.

203
 Running off the end
list
8 int 2

int 3
primes
…

int 17

int 19
204
primes[8]

We have remarked on a couple of occasions that the largest valid index is a

number one less than the length of the list.
So what happens if you ask for an index greater than the largest legal
value?

204
 Running off the end
>>> len(primes)
8

>>> primes[7]
19

>>> primes[8]
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
IndexError: list index out of range

205
Type of error Description of error

You get an error unsurprisingly.

The type of the error is an “IndexError” ― something went wrong with an
index.
The error message specifies that the index asked for was outside the valid
range.

205
 Running off the end
>>> primes[8] = 23
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
IndexError: list assignment index out of range

Same type Similar description of error

of error but with “assignment”

206

Note that we can’t use indicies beyond the limit to extend a list either.

206
Progress

Lists [2, 3, 5, 7, 11, 13, 17, 19]

index primes[4]

Count from zero primes[0]

Deletion del primes[6]

Length len(primes)

Over-running primes[8]
207

207
 Exercise 11

Track what is happening to this list at each stage.

>>> numbers = [5, 7, 11, 13, 17, 19, 29, 31]

>>> numbers[1] = 3

>>> del numbers[3]

>>> numbers[3] = 37

>>> numbers[4] = numbers[5]

5 minutes
208

Do this yb hand. The script exercise11.py will tell you if you were right.

208
 How can we add to a list?
list Same list list
8 New length 9

int 2 int 2

… ? …

int 17 int 17

int 19 int 19

Extra item int 23 209

So, how can we extend a list?

209
 Appending to a list
>>> primes
[2, 3, 5, 7, 11, 13, 17, 19]

A function built into a list

>>> primes.append(23)

>>> primes The list is now updated

[2, 3, 5, 7, 11, 13, 17, 19, 23] 210

This is the Python syntax for appending an item to the end of a list. You
won’t recognise the syntax; it is something new.

210
 primes.append() ?

The list

A connecting dot

>>> primes.append(23) append()

The value to append

All lists have this

function built in.

211

So we need to look at this new construction.

We have the list, “primes”, followed by a dot which acts as a connector.
This is followed by the name of a function, “append”. This function is not a
standard Python function like print() or len(). Instead it is a function
that is “built in” to the list itself. The list has its own function which appends
to that list.
Alternatively, think of “primes.append()” as a function that appends to
primes.

211
 “Methods”
Behaves just
like a function

object . function (arguments)

a function that has

special access to
the object’s data.
212

These built in functions are called “methods” or, more precisely, “methods of
the object” are used all over Python and are a general concept across an
entire type of programming called “object oriented programming”.

212
 Using the append() method

>>> print(primes)
[2, 3, 5, 7, 11, 13, 17, 19]

>>> primes.append(23)
The function doesn’t
return any value.
>>> primes.append(29)

>>> primes.append(31)
It modifies
>>> primes.append(37)
the list itself.
>>> print(primes)
[2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37] 213

We can use the append() method repeatedly to extend the list as far as we
want.

213
 Other methods on lists: reverse()

>>> numbers = [4, 7, 5, 1]

>>> numbers.reverse()
The function doesn’t
return any value.
>>> print(numbers)
It modifies
[1, 5, 7, 4]
the list itself.

214

append() is not the only method built into lists.

For example reverse() takes the list and reverses its contents.
Note that it doesn’t return a reversed list as a value; t doesn’t return
anything at all.
It silently reverses the content of the list itself.
Also note that it takes no argument; the brackets on the end of the function
are empty.

214
 Other methods on lists: sort()

>>> numbers = [4, 7, 5, 1]

>>> numbers.sort()
The function does not
return the sorted list.
>>> print(numbers)
It sorts the
[1, 4, 5, 7]
list itself.

Numerical order.

215

Similarly, the sort() method doesn’t return a sorted list but silently sorts
the list internally.

215
 Other methods on lists: sort()

>>> greek = ['alpha', 'beta', 'gamma', 'delta']

>>> greek.sort()

>>> print(greek)

['alpha', 'beta', 'delta', 'gamma']

Alphabetical order
of the words.
216

More or less any type can be sorted. Text sorting carries all the cautions
about the complexities of collation that we covered under comparisons.

216
 Other methods on lists: insert()
0 1 2

>>> greek = ['alpha', 'gamma', 'delta']

>>> greek.insert(1, 'beta' )

Where to insert What to insert

>>> greek

['alpha', 'beta', 'gamma', 'delta']

0 1
Displaced 217

The append() method sticks an item on the end of a list. If you want to
insert an item elsewhere in the list we have the insert() method.
The insert() method takes two arguments:
The first is the item to be inserted.
The second is in index where it should go. This does not replace the original
item but rather “shuffles up” all the items beyond it by one place to make
room.

217
 Other methods on lists: remove()

>>> numbers = [7, 4, 8, 7, 2, 5, 4]

>>> numbers.remove(8) Value to remove

>>> print(numbers)

[7, 4, 7, 2, 5, 4]

c.f. del numbers[2] Index to remove 218

There is a remove() method.

This is passed a value to remove from the list. It then removes that value
from the list, wherever it is in the list.
Contrast with with del where you had to know the index to remove.

218
 Other methods on lists: remove()

>>> print(numbers)

[7, 4, 7, 2, 5, 4]
There are two instances of 4.
>>> numbers.remove(4)

>>> print(numbers)

[7, 7, 2, 5, 4]

Only the first instance is removed 219

If the value appears more than once in a list then only the first instance is
removed.
Trying to remove something that isn’t there will lead to an error.

219
 What methods are there?
>>> help(numbers)

Help on list object:

class list(object)
...
| append(...)
| L.append(object) -- append object to end
...

Pagination: ␣ next page

B back one page

220
Q quit

That’s a lot of methods, and it’s only some of them. How can we know all of
them?
You can always ask for help on any Python object and you will be told all
about the methods it possesses. It is a very formal documentation but the
information is there.
Incidentally, Python uses a program to paginate its help output. press the
space bar to move on one page, “B” to move back a page and “Q” to quit.

220
 Help on a single method
>>> help(numbers.append)

Help on built-in function append:

append(...)
L.append(object) -- append object to end

221

You can also get help on a single method which is often simpler to deal with.

221
 Sorting a list redux

>>> greek = ['alpha', 'beta', 'gamma', 'delta']

>>> greek.sort() Recall: greek.sort()

sorts the list “in place”.

>>> print(greek)

['alpha', 'beta', 'delta', 'gamma']

222

We noted that the sort() method sorts the list itself. Experience shows that
sorting is one of those operations where people want a sorted copy of the
list quite often.

222
 Sorting a list redux: “sorted()”

>>> greek = ['alpha', 'beta', 'gamma', 'delta']

>>> print(sorted(greek)) sorted() function

returns a sorted list…
['alpha', 'beta', 'delta', 'gamma']

>>> print(greek) …and leaves the

list alone
['alpha', 'beta', 'gamma', 'delta']
223

To assist with this, Python 3 offers a standalone function called sorted()

which makes a copy of the list and sorts that copy, leaving the original
unchanged.

223
 Adding to a list redux: “+”

>>> primes

[2, 3, 5, 7, 11, 13, 17, 19]

Concatenation List to add

operator

>>> primes + [23, 29, 31]

[2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31]

224

The list method we saw first appended a single item to the end of a list.
What happens if we want to add a whole list of items at the end?
In this regard, lists are like strings. The “+” operator performs concatenation
and creates a new list which is one concatenated after the other.

224
 Concatenation
Create a new list

>>> newlist = primes + [23, 29, 31]

Update the list

>>> primes = primes + [23, 29, 31]

Augmented assignment

>>> primes += [23, 29, 31]

225

We can use this to update a list in place. Note that the augmented
assignment operator “+=” also works and is more than syntactic sugar this
time. It is actually more efficient than the long hand version because it
updates the list in place rather than creating a new one.

225
 Creating lists from text ― 1

>>> list('Hello') str 1 H

['H', 'e', 'l', 'l', 'o'] list

str 1 e
5

str 5 H e l l o str 1 l

str 1 l

str 1 o
226

We ought to look at a couple of ways to create lists from text.

The first is to simply convert a string into a list with the list() function.
(As with all Python types, there is a function of the same name as the type
that converts into the type.)
Applying list() to a string gives a list of the characters in the string.

226
 Creating lists from text ― 2
Built in method
>>> 'Hello, world!'.split()

['Hello,', 'world!'] Splits on spaces

str 13 H e l l o , ␣ w o r l d !

list
2 str 6 H e l l o ,

str 6 w o r l d !
227

The string type has methods of its own. One of these is split() which returns
a list of the components of the string as separated by white space.
The split() method can take an argument identifying other characters to
split on. If you want to get into more complex splitting of text we recommend
you investigate regular expressions or format-specific techniques (e.g. for
comma-separated values.

227
Progress
“Methods” object.method(arguments)
append(item)
reverse()
sort()
insert(index,item)
remove(item)

Help help(object)
help(object.method)

Sorting list.sort()
sorted(list)
Concatenation
+ [1,2,3] + [4,5,6] 228
+= primes += [29, 31]

228
Exercise 12

5 minutes
229

229
 Is an item in a list? ― 1

>>> odds = [3, 5, 7, 9] Does not include 2

>>> odds.remove(2) Try to remove 2

Traceback (most recent call last): Hard error

File "<stdin>", line 1, in <module>
ValueError: list.remove(x): x not in list

✗ x must be in the
list before it can
be removed
230

Recall that a list’s remove() method will give an error if the value to be
removed is not in the list to start with.
We need to be able to test for whether an item is in a list or not.

230
 Is an item in a list? ― 2
>>> odds = [3, 5, 7, 9]

>>> 2 in odds

False

>>> 3 in odds

True

>>> 2 not in odds

True 231

Python uses the keyword “in” for this purpose. It can be used on its own or
as part of “not in”. “value in list” evaluates to a boolean: True or
False.

231
 Precedence
First

x**y -x +x x%y x/y x*y x-y x+y

x==y x!=y x>=y x>y x<=y x<y

x not in y x in y

not x x and y x or y The list now contains

every operator we
Last meet in this course.
232

These operators fit naturally into the order of precedence. While Python
does contain other operators that belong in this list, we will not be meeting
them in this introductory course.

232
 Safe removal

if number in numbers :

numbers.remove(number)

What’s the
difference?

while number in numbers :

numbers.remove(number)

233

We now have a safe way to remove values from lists, testing before we
remove them.
Quick question: What’s the difference between the two code snippets in the
slide?

233
 Working through a list ― 1

e.g. Printing each element on a line

['The', 'cat', 'sat', 'on', 'the', 'mat.']

The
cat
sat
on
the
mat.
234

There is an obvious thing to want to do with a list, and that is to work

through each item in a list, in order, and perform some operation or set of
operations on each item.
In the most trivial case, we might want to print each item. The slide shows a
list of strings, probably from the split() of a string. How do we print each
item one after the other?

234
 Working through a list ― 2

e.g. Adding the elements of a list

[45, 76, -23, 90, 15]

203

What is the sum of an empty list? [] ?

235

Alternatively, we might want to accumulate the items in a list in some way.

For example, we might want to sum the numbers in a list. This is another
example of applying an operation to each item in a list. This time the
operation is folding the list items’ values into some final result.
In this case we would probably need an initial value of the result that it takes
before any items get folded into it. What is the sum of an empty list? Zero?
Is that an integer zero, a floating point zero, or a complex zero?

235
 Working through a list ― 3

e.g. Squaring every number in a list

[4, 7, -2, 9, 1]

[16, 49, 4, 81, 1]

236

Finally, we might want to convert one list into another where each item in the
output list is the result of some operation on the corresponding item in the
input list.

236
 The “for loop” ― 1

name of list list

words = ['The', 'cat', 'sat', 'on', 'the', 'mat.']

for word in words : A new Python

looping construct
␣␣␣␣print(word)
print: What we want
to do with the list items.
237

Python has a construct precisely for stepping through the elements of a list.
This is the third and final construct we will meet in this course. We have
already seen if… and while… in this course. Now we meet for….
This a looping construct, but rather than repeat while a test evaluates to
True, it loops once for each item in a list. Furthermore it defines a name
which it attaches to one item after another in the list as it repeats the loop.

237
 The “for loop” ― 2

words = ['The', 'cat', 'sat', 'on', 'the', 'mat.']

keywords

for word in words :

colon followed by
an indented block
␣␣␣␣print(word)

238

We'll look at the expression one step at a time.

The expression is introduced by the “for” keyword.
This is followed by the name of a variable. We will return to this in the next
slide.
After the name comes the keyword “in”. We have met this word in the
context of lists before when we tested for an item's presence in a list. This is
a different use. We aren't asking if a specific value is in a list but rather we
are asserting that we are going to be processing those values that are in it.
After this comes the list itself, or more often the name of a list. All it has to
be is an expression that evaluates to a list.
The line ends with a colon.
The lines following the colon are indented marking the block of code that is
to be run once for each item in the list.

238
 The “for loop” ― 3

words = ['The', 'cat', 'sat', 'on', 'the', 'mat.']

Defining the loop variable

for word in words :

␣␣␣␣print(word)

Using the loop variable

239

Now let's return to the name between “for” and “in”.

This is called the “loop variable”. Each time the loop block is run this name
is attached to on item of the list. Each tim the loop is run the name is
attached to the next item in the list. The looping stops after the name has
been attached to the last item I nthe list.

239
 The “for loop” for printing

words = ['The', 'cat', 'sat', 'on', 'the', 'mat.']

for word in words :

␣␣␣␣print(word)
for1.py

240

There is a simple example of this in the file for1.py in your home

directories.

$ python3 for1.py
The
cat
sat
on
the
mat.
$

240
 The “for loop” for adding

numbers = [45, 76, -23, 90, 15]

sum = 0 Set up before the loop

for number in numbers :

␣␣␣␣sum += number Processing in the loop

for2.py

241
print(sum) Results after the loop

Our second case was an “accumulator” adding the elements in a list.

Here we establish an initial start value for our total of 0 and give it the name
“sum”.
Then we loop through the elements in the list, adding their values to the
running total as we move through them.
The unindented line after the loop block marks the end of the loop block and
is only run after all the looping is completed. This prints out the value of the
total now that all the numbers in the list have been added into it.

$ python3 for2.py
203
$

241
 The “for loop” for creating a new list

numbers = [4, 7, -2, 9, 1]

squares = [ ] Set up before the loop

for number in numbers :

␣␣␣␣squares.append(number**2) Processing in the loop

for3.py

242
print(squares) Results after the loop

Our third example made a new list from the elements of an old list. For
example, we might want to take a list of numbers and produce the list of
their squares.
In this case the usual process is that rather than have an accumulator with
an initial value we start with an empty list and, as we loop through the input
values, we append() the corresponding output values.

$ python3 for3.py
[16, 49, 4, 81, 1]
$

242
 The loop variable persists!
numbers = [4, 7, -2, 9, 1]

squares = [ ]

for number in numbers : Loop variable only

meant for use in loop!
␣␣␣␣squares.append(number**2)

print(number) But it persists!

243

There is one nicety we should observe.

The loop variable was created for the purpose of running through the
elements in the list. But it is just a Python name, no different from the ones
we establish by direct assignment. While the for… loop creates the name it
does not clean it up afterwards.

243
 “for loop hygeine”
numbers = [4, 7, -2, 9, 1]

squares = [ ]

for number in numbers :

␣␣␣␣squares.append(number**2)

del number Delete it after use

244

It is good practice to delete the name after we have finished using it. So we
will follow our for… loops with a del statement.
This is not required by the Python language but we recommend it as good
practice.

244
Progress

Testing items in lists 3 in [1,2,3,4] True

for loops sum = 0

for number in [1,2,3,4]:
sum += number
del number

loop variables for number in [1,2,3,4]:

sum += number
del number
245

245
Exercise 13

What does this print?

numbers = [0, 1, 2, 3, 4, 5]

sum = 0
sum_so_far = []

for number in numbers:

sum += number
sum_so_far.append(sum)

print(sum_so_far) 5 minutes
246

246
 “Sort-of-lists”

Python “magic”:

Treat it like a list and it will

behave like a useful list

What can “it” be?

247

We have seen already that every Python type comes with a function that
attempts to convert other Python objects into that type. So the list type
has a list() function.
However, with lists Python goes further and puts a lot of work into making
this transparent and convenient. In very many cases in Python you can drop
an object into a list construct and it will act as if it was a list, and a
convenient list at that.

247
 Strings as lists

Recall:

list('Hello') ['H', 'e', 'l', 'l', 'o']

for letter in 'Hello' : H

print(letter) e
l
l
Gets turned o
for4.py
into a list.
248

For example, we know that if we apply the list() function to a string we

get the list of characters. But the Python “treat it like a list” magic means that
if we simply drop a string into the list slot in a for… loop then it is treated as
exactly that list of characters automatically.

$ python for4.py
H
e
l
l
o
$

248
 Creating lists of numbers

Built in to Python:

range(start,limit)

for number in range(3,8): 3

print(number) 4
5
6
7
8 not included
249

There are other Python objects which, while not lists exactly, can be treated
like lists in a for… loop. A very important case is a “range” object.
Note that the range defined by 3 and 8 starts at 3 but ends one short. This is
part of the whole “count from zero” business.

249
 ranges of numbers

Not actually lists: >>> range(0,5)

range(0,5)

But close enough: >>> list(range(0,5))

[0, 1, 2, 3, 4]

Treat it like a list and

it will behave like a list
250

Strictly speaking a range is not a list. But it is close enough to a list that
when you drop it into a for… loop, which is its most common use by far,
then it behaves like the list of numbers.

250
 Why not just a list?

Most common use: for number in range(0, 10000):

…

Inefficient to make a
huge list just for this

“iterables” : anything that can be treated like a list

list(iterable) Explicit list

251

So why does the range() function not just produce a list?

Well, its most common use is in a for… loop. Only one value is required at a
time. If the list was explicitly created t would waste computer memory for all
the items not in use at the time and computer time for creating them all at
once.
(Truth be told, for the purposes of this course you wouldn’t notice.)

251
 Ranges of numbers again
via list()

range(10) [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
Start at 0
range(3, 10) [3, 4, 5, 6, 7, 8, 9]

range(3, 10, 2) [3, 5, 7, 9] Every nth number

range(10, 3, -2) [10, 8, 6, 4] Negative steps

252

The range() function can be used with different numbers of arguments.

A single argument gives a list running from zero to the number.
Two arguments gives the lists we have already seen.
A third argument acts as a “stride” and can be negative.

252
 Indices of lists
>>> primes = [ 2, 3, 5, 7, 11, 13, 17, 19]

>>> len(primes)

>>> list(range(8))

[0, 1, 2, 3, 4, 5, 6, 7] valid indices

0 1 2 3 4 5 6 7

[2, 3, 5, 7, 11, 13, 17, 19] 253

Now that we have the range object we can move on to one of the most
important uses of it.
So far we have used a for… loop to step through the values in a list. From
time to time it is important to be able to step through the valid indices of the
list.
Observe that if we apply the range() function to a single number which is
the length of a list then we get a list of the valid indices for that list.

253
 Direct value or via the index?
primes = [2, 3, 5, 7, 11, 13, 17, 19]

for prime in primes: Simpler

print(prime)

for index in range(len(primes)): Equivalent

print(primes[index])

254

What good is a list of valid indices?

There are two ways to step through the values in a list. One is directly; this
is the method we have met already. The second is to step through the
indices and to look up the corresponding value in the list.
These are equivalent and the first method we have already seen is shorter
to write. So why bother with the second?

254
 Working with two lists: “dot product”

list1 = [ 0.3, 0.0, 0.4]

∙ × × ×
list2 = [ 0.2, 0.5, 0.6]

0.06 + 0.0 + 0.24 0.3

255

Consider operations on two lists.

A concrete example might be the “dot product” of two lists. This is the sum
of the products of matching elements.
So
[0.3, 0.0, 0.4] · [0.2, 0.5, 0.6]
= 0.3×0.2 + 0.0×0.5 + 0.4×0.6
= 0.06 + 0.0 + 0.24
= 0.6
How might we implement this in Python?

255
 Working with two lists: indices

0 1 2
list1 = [0.3, 0.0, 0.4]
indices
list2 = [0.2, 0.5, 0.6]

sum = 0.0

for index in range(len(list1)) :

sum += list1[index]*list2[index]
Dealing with
print(sum)
values from
both lists at
the same time.256

We can approach this problem by running through the valid indices and
looking up the corresponding values from each list in the body of the for…
loop.

256
 Iterables

range(from,to,stride)

Not a list but “close enough”

“Iterable”

257

The range object is one of the most commonly met examples of an iterable,
something that isn’t a list but is “close enough”.

257
 A little more about iterables ― 1
>>> greek = ['alpha', 'beta', 'gamma', 'delta']

>>> greek_i = iter(greek)

iter
str 5 a l p h a
>>> next(greek_i) 4

'alpha' Offset 2 str 4 b e t a

>>> next(greek_i) str 5 g a m m a

'beta'
str 5 d e l t a
258

We will look a little more closely at iterables so that we can recognise them
when we meet them later.
If we start with a list then we can turn it into an iterable with the iter()
function.
An iterable ceated from a list is essentially the same as the list with a note of
how far through the list we have read. This reference starts at zero,
obviously.
Python provides a next() function which can act on any iterable which
returns the next value (or the first if we’ve not started) and increments this
internal counter.

258
 A little more about iterables ― 2
iter
>>> next(greek_i) str 5 a l p h a
4
'gamma'
✗ str 4 b e t a

>>> next(greek_i)
str 5 g a m m a
'delta'

str 5 d e l t a
>>> next(greek_i)

Traceback (most recent call last):

File "<stdin>", line 1, in <module>
StopIteration ✗ 259

Note that next() complains vigorously if we try to run off the end. For this
course, where these errors are all fatal it means that we can’t use next()
directly. we don’t need to; we have the for… loop which handles the error
for us.

259
Progress
Non-lists as lists for letter in 'Hello':
...

range() range(limit)
range(start,limit)
range(start,limit,step)
range(3,7) [3,4,5,6]

“Iterables” greek_i = iter(greek)

next(greek_i)

Indices of lists for index in range(len(things)):

260
Parallel lists

260
 Exercise 14

Complete exercise14.py

list1 = [ 0.3, 0.0, 0.4]

list2 = [ 0.2, 0.5, 0.6]
Subtract
0.1 -0.5 -0.2
Square
0.01 + 0.25 + 0.04 0.3
Add
5 minutes
261

This exercise develops the Python to calculate the square of the distance
between two 3D points. We could use our square root Python from earlier to
calculate the distance itself but we will meet the “real” square root function
later in this course so we’ll hold back from that for now.

261
 List “slices”

>>> primes = [2, 3, 5, 7, 11, 13, 17, 19, 23, 29]

>>> primes
[2, 3, 5, 7, 11, 13, 17, 19, 23, 29] The list

>>> primes[3]
7 An item

>>> primes[3:9]
[7, 11, 13, 17, 19, 23] Part of the list
262

There is one last piece of list Pythonry we need to see.

Python has a syntax for making copies of parts of lists, which it calls “slices”.
If, instead of a simple index we put two indices separated by a colon then
we get the sub-list running from the first index up to but excluding the
seocnd index.

262
 Slices ― 1
from
Up to but not
to
including…
primes[3:9]

3 9

primes = [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31]

primes[3:9] [7, 11, 13, 17, 19, 23]

Item 9 not 263

included

The last index is omitted as part of the “count from zero” thing.

263
 Slices ― 2
primes [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31]

primes[3:9] [7, 11, 13, 17, 19, 23]

primes[:9] [2, 3, 5, 7, 11, 13, 17, 19, 23]

primes[3:] [7, 11, 13, 17, 19, 23, 29, 31]

primes[:] [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31]

264

We can omit either or both of the numbers. Missing the first number means
“from the start” and missing the second means “right up to the end”.

264
 Slices ― 3
primes [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31]

primes[3:9] [7, 11, 13, 17, 19, 23]

primes[3:9:2] [7, 13, 19 ]

primes[3:9:3] [7, 17 ]

265

We can also add a second colon which is followed by a stride, just as with
range().

265
 Copies and slices ― 1

>>> letters = ['a','b','c']

>>> alphabet = letters

list
str 1 a
letters 3

str 1 b
alphabet

str 1 c
266

Slices allow us to make copies of entire lists.

If we use simple name attachment then we just get two names for the same
list.

266
 Copies and slices ― 2
>>> letters[0] = 'A'

>>> print(alphabet)

['A', 'b', 'c']

list
str 1 A
letters 3

str 1 b
alphabet

str 1 c
267

Changing an item in the list via one name shows up via the other name.

267
 Copies and slices ― 3
>>> letters = ['a','b','c'] Slices are
copies.
>>> alphabet = letters[:]
letters alphabet

list list
str 1 a
3 3

str 1 b

str 1 c 268

Slices are copies, though, so if we attach a name to a slice from a list —

even if that slice is the entire list ― then we have two separate lists each
with their own name attached.

268
 Copies and slices ― 4
>>> letters[0] = 'A' Slices are
copies.
>>> print(alphabet)
['a', 'b', 'c'] alphabet

letters
list list
str 1 A str 1 a
3 3

str 1 b

str 1 c 269

So changes in one don’t show in the other because they are not the same
list.

269
Progress

Slices

End-limit excluded

Slices are copies

items[from:to] items[from:to:stride]

items[:to] items[:to:stride]

items[from:] items[from::stride]

items[:] items[::stride] 270

270
Exercise 15

Predict what this Python will do.

Then run it. exercise15.py
Were you right?

foo = [4, 6, 2, 7, 3, 1, 9, 4, 2, 7, 4, 6, 0, 2]

bar = foo[3:12:3]

bar[2] += foo[4]

foo[0] = bar[1]

print(bar)
5 minutes
271

271
 Files
Input Output

.txt

.dat

.csv

Reading Writing
272

Now we will look at something completely different that will turn out to be
just like a list: Files.
We want our Python scripts to be able to read in and write out files of text or
data.
We will consider reading files first and writing them second.

272
 Reading a text file

File name 'treasure.txt' string

“opening” the file

File object book file

reading from the file

File contents 'TREASURE ISLAND' string

“closing” the file

273
Finished with the file

Reading from a file involves four operations bracketing three phases.

We start with a file name. This is a string of characters.
I want to be pedantic about something in this course: a file name is not a
file. A file is a lump of data in the computer’s long-term store. A file name is a
short piece of text.
We link a file name to a file by a process called “opening the file”. This takes
the file name and creates a Python file object which will act as our conduit to
the file proper.
We will use this file object to read data from the file into the Python script.
When we are done reading data out of the file (via the file object) we will
signal to both python and the operating system that we are done with it by
“closing” it. This disconnects us from the file and we would have to re-open it
if we wanted more data.

273
 Opening a text file
Python function

File name

Read-only [text]
>>> book = open('treasure.txt', 'r') “mode”

Python file object

Read-only is
the default
book = open('treasure.txt' ) 274

We will start with opening a file.

We start with just the file name. This is passed into the open() function with
a second argument, 'r', indicating that we only want to read the file.
The function hooks into the operating system, which looks up the file by
name, checks that we have permission to read it, records the fact that we
are reading it, and hands us back a handle — the file object ― by which we
can access its contents.

274
 Reading from a file object

“next line, please”

File object
>>> line1 = next(book)

First line of file

>>> line1 Includes the

“end of line”
'TREASURE ISLAND\n'
275

Now that we have this hook into the file itself, how do we read the data from
it?
File objects are iterators, so we can apply the next() function to them to get
the next line, starting with the first.

275
 File object are iterable
>>> line2 = next(book)
Second line of file
>>> line2
'\n' A blank line

>>> line3 = next(book)

>>> line3 Third line of file

'PART ONE\n'
276

Note that a “blank” line actually contains the end of line character. It has
length 1, and is not an empty string, length 0.

276
 Closing the file

Method built in to file object

>>> book.close()

Frees the file for other

programs to write to it.

277

When we are done reading the data we need to signal that we are done with
it. Python file objects have a close() method built into them which does
precisely this.

277
 The file object ― 1

>>> book Text Input/Output

<_io.TextIOWrapper
name='treasure.txt' File name
encoding='UTF-8'>

Character encoding:
how to represent
letters as numbers.

278

UTF-8 means “UCS Transformation Format ― 8-bit”.

UCS means “”Universal Character Set. This is defined by International
Standard ISO/IEC 10646, Information technology — Universal multiple-octet
coded character set (UCS). For more information than you could possibly
want on this topic visit the Unicode web pages: www.unicode.org.

278
 The file object ― 2
File name

Text encoding

file 29 treasure.txt UTF-8

TREASURE ISLAND↵
“offset”: how far into ↵
the file we have read PART ONE↵

☞ ↵
The Old Buccaneer↵
↵
Pointer to the file
on the file system

279

The Python file object contains a lot of different bits of information. There are
also lots of different sorts of file object, but we are glossing over that in this
introductory course.
What they share is a reference into the operating system’s file system that
identifies the specific file and acts as the declaration to the operating system
that the Python process is using the file.
They also share an “offset”. This is a number that identifies how far into the
file the program has read so far. The next() function reads from the offset
to the next new line and increases the offset to be the distance into the file
of the new line character.
The file object also records the name it was opened from and the text
encoding it is using to convert bytes in the file to characters in the program.

279
 Reading through a file
Treat it like a list and
it will behave like a list

list(file_object) List of lines in the file

>>> book = open('treasure.txt', 'r')

>>> lines = list(book)

>>> print(lines)
280

Given that a file is an iterable, we can smply convert it into a list and we get
the list of lines.

280
 Reading a file moves the offset
>>> book = open('treasure.txt', 'r')
!
>>> lines_a = list(book)

>>> print(lines_a)

… Huge output

>>> lines_b = list(book)

>>> print(lines_b)

[] Empty list 281

Note, however, that the act of reading the file to get the lines reads through
the file, so doing it twice gives an empty result second time round.

281
 Reading a file moves the offset
>>> book = open('treasure.txt', 'r')
File object starts with offset at start.
>>> lines_a = list(book)
Operation reads entire file from offset.
>>> print(lines_a)
Offset changed to end of file.
…

>>> lines_b = list(book)

Operation reads entire file from offset.
>>> print(lines_b)
So there's nothing left to read.
[] 282

recall that the reading starts at the offset and reads forwards — to the end of
the file in this example. It also moves the offset to what was last read. So
the offset is changed to refer to the end of the file.

282
 Resetting the offset
>>> book = open('treasure.txt', 'r')

>>> lines_a = list(book)

>>> print(lines_a)

>>> book.seek(0) Set the offset explicitly

>>> lines_b = list(book)

>>> print(lines_b) 283

We can deliberately change the offset. For text files with lines of different
lengths this can be more complex that you would imagine. The only safe
value to change the offset to is zero which takes us back to the start of the
file.

283
 Older file object methods
>>> book = open('treasure.txt', 'r')

>>> book.readline()

'TREASURE ISLAND\n'

>>> book.readlines()

['\n', 'PART ONE\n' … ]

284

There are other ways to read in the data. Earlier versions of Python only
supported a couple of built-in methods. You may still see these in other
people’s scripts but we don’ recommend writing them in scripts that you
create.

284
 Typical way to process a file
book = open('treasure.txt', 'r')

for line in book : Treat it like a list…

…

285

Being able to read a single line of a file is all very well. Converting an entire
book into a list of its lines might prove to be unwieldy.
What is the typical way to do it?
Given that files can be treated like lists the easiest way to process each line
of a file is with a for… loop.

285
 Example: lines in a file
book = open('treasure.txt', 'r')

n_lines = 0 Line count

for line in book : Read each line of the file

n_lines += 1 Increment the count

print(n_lines) Print out the count

286
book.close()

For example, we could just increment a counter for each line to count the
number of lines.

286
 Example: characters in a file
book = open('treasure.txt', 'r')

n_chars = 0

for line in book : Number of characters on the line

n_chars += len(line)

Increase the count by that many

print(n_chars)

287
book.close()

We could measure the length of the line and increment the counter by the
number of characters in the line to count the total number of characters in
the file.

287
Progress

Reading files

Opening file book = open(filename, 'r')

Reading file for line in book:

...

Closing file book.close()

File offset book.seek(0)

288

288
 Exercise 16

Complete a script to count exercise16.py

lines, words and characters
of a file.

5 minutes
289

Count words & lines.

289
 Writing files

File name 'treasure.txt' string

“opening” the file

Data to write 'TREASURE ISLAND' string

writing to the file

File object book file

“closing” the file

290
Finished with the file

Enough of reading files Robert Louis Stephenson has prepared for us. Let’s
write our own.
This is again a three phase process. We will open a file for writing, write to it
and then close it.

290
 Opening a text file for writing

Open for
writing [text]

>>> output = open('output.txt' , 'w')

This will truncate

an existing file
! 291

Opening a file for writing is the same as opening it for reading except that
the mode is 'w' for writing.
If the file already exists then this overwrites it. The file gets truncated to zero
bytes long because you haven’t written to it yet.

291
 Writing to a file object ― 1

File object

Method built in
to file object
>>> output.write(line1)

Data being written

292

A writeable file object has a method write() which takes a text string
(typically but not necessarily a line) and writes it to the file.

292
 Writing to a file object ― 2

>>> output.write('alpha\n') Typical use: whole line.

Includes “end of line”

>>> output.write('be')
Doesn’t have
to be whole lines
>>> output.write('ta\n')

>>> output.write('gamma\ndelta\n')

Can be multiple lines 293

How the data is chopped up between writes is up to you.

293
 Closing the file object

>>> output.close() Vital for written files

294

It also has a close() method.

294
 Importance of closing
!
Data “flushed” to disc on closure.

write “flush”

Python script File object File system 295

Closing files is even more important for written files that read ones.
It is only when a file is closed that the data you have written to the Python
file object is definitely sent to the operating system’s file. This is a process
called “flushing the file to disc”.
Writing to disc is slow by computing standards so, for efficiency reasons,
systems like Python tend to wait until they have at least a certain amount of
data for a file before they flush it to disc. If your program exists before
flushing then the data may be lost. Closing a file signals that there will be no
more data coming for it so Python flushes whatever it has in hand to the disc
as part of the closure process.

295
 Importance of closing promptly !
Files locked for other access

open ['w']
✗ open ['r']

296
(More a problem for Windows than Unix)

There’s more to it than just flushing, though. Holding a file open signals to
the underlying computer operating system that you have an interest in the
file. How the operating system reacts to this varies from system to system,
but Microsoft Windows™ will lock a file so that if you have it open for writing
nobody else can open it for reading, even if you don’t plan to write any more
to it than you have already.

296
Progress

Writing files

Opening files for writing book = open(filename, 'w')

Writing data book.write(data)

Closing file important book.close()

Flushing book.flush()
297

297
Exercise 17

5 minutes
298

298
Functions

y = f(x)

299

299
 Functions we have met
input(prompt) bool(thing)

len(thing) float(thing)

open(filename, mode) int(thing)

print(line) list(thing)

range(from, to, stride) str(thing)

Not that many!
type(thing)
“The Python Way”:
ord(char)
If it is appropriate to
an object, make it a 300
chr(number) method of that object.

This is the complete set of Python functions that we have met to date.
Actually it’s surprising how few there are, not how many. Python’s
philosophy leads to functions that only make sense for a particular sort of
object being methods of that object, not free-standing functions.
We are now going to write our own functions.

300
 Why write our own functions?
Easier to …
… read

… write

… test

… fix

… improve

… add to
“Structured
programming”
… develop
301

Why?
Moving our scripts’ functionality into functions and then calling those
functions is going to make everything better. This is the first step towards
“structured programming” which is where programming goes when your
scripts get too long to write in one go without really thinking about it.

301
 Defining a function

(y1, y2, y3) = f(x1, x2, x3, x4, x5)

Identify the inputs

Identify the processing

Identify the outputs

302

So what do we need to do?

Well any functions starts by defining what inputs it needs, what it does with
those inputs to generate the results and exactly what results/outputs it
generates.

302
 A function to define: total()

Sum a list

[1, 2, 3] 6

[7, -4, 1, 6, 0] 10

[] 0 “Edge case”

303

To give ourselves a simple but concrete example to keep in mind we will set
ourselves the challenge of writing a function that sums the elements of a list.
This may sound trivial but immediately raises some interesting cases that
need to be considered.
What is the sum of an empty list? Zero? Is that an integer zero or a floating
point zero? (Or a complex zero?)
We will say it should sum to an integer zero.

303
 Defining a Python function ― 1
define a function called …
name of function
inputs
def total( … ): colon

304

We will plunge straight into the Python.

The Python keyword to define a function is “def”.
This is followed by the name of the function, “total” in this case.
This is followed by a pair of round brackets which will contain all the input
values for the function.
Finally there is a colon to mark the end of the line and the beginning of the
body of the function.

304
 Defining a Python function ― 2

name for the input

def total(numbers):
This name is
internal to
the function.

305

Our function takes a single input: the list of numbers to be summed.

What goes inside the brackets on the def line is the name that this list will
have inside the function’s definition. This is the “x” in maths. This internal
names is typically unrelated to the name the list has in the main body of the
script.
It is always a good idea to name your inputs (and other variables)
meaningfully. Please try to avoid calling them “x”, “y”, or “z”.
We will call ours “numbers”.

305
 Defining a Python function ― 3

def total(numbers): Colon followed by indentation

306

As ever with Python a colon at the end of a line is followed by an indented

block of code. This will be the body of the function where we write the
Python that defines what the function actually does with the input(s) it is
given.

306
 Defining a Python function ― 4

def total(numbers):

sum_so_far = 0

for number in numbers:

sum_so_far += number “Body” of function

307

For our function this is particularly simple.

307
 Defining a Python function ― 4

def total(numbers):
These variables exist only
sum_so_far = 0
within the function’s body.
for number in numbers:
sum_so_far += number

308

The numbers name we specified on the def line is visible to Python only
within the function definition. Similarly any names that get created within the
function body exist only within that function body and will not be visible
outside. Nor will they clash with any other uses of those names outside the
function body.
In our example code the name “numbers” is defined in the def line, the
“sum_so_far” name is defined explicitly in the function body and the
“number” name is defined by the for… loop as its loop variable.
None of these interact with the Python outside the function definition.

308
 Defining a Python function ― 5

def total(numbers):

sum_so_far = 0

for number in numbers:

sum_so_far += number

return sum_so_far This value

is returned

return this value 309

Finally we need to specify exactly what value the function is going to return.
We do this with another Python keyword, “return”. The value that follows
the return keyword is the value returned by the function.
When Python reaches the return statement in a function definition it hands
back the value and ends the execution of the function body itself.

309
 Defining a Python function ― 6

And that’s it!

def total(numbers):

sum_so_far = 0

for number in numbers:

sum_so_far += number

return sum_so_far

Unindented
after this 310

And that’s all that is involved in the creation of a Python function.

310
 Defining a Python function ― 7

And that’s it!

All internal names internal No need to avoid reusing names

All internal names cleaned up No need for del

311

Note that because of this isolation of names we don’t have to worry about
not using names that are used elsewhere in the script.
Also, as part of this isolation all these function-internal names are
automatically cleared when the function finishes. We do not need to worry
about deleting them.

311
 Using a Python function ― 1
def total(numbers):

sum_so_far = 0

for number in numbers:

sum_so_far += number

return sum_so_far

print(total([1, 2, 3]))

The list we
want to add up

312

We use this function we have defined in exactly the same way as we would
use a function provided by Python.

312
Using a Python function ― 2
def total(numbers):

sum_so_far = 0

for number in numbers:

sum_so_far += number

return sum_so_far

print(total([1, 2, 3]))

The function we
have just written

313

313
Using a Python function ― 3
def total(numbers):

sum_so_far = 0

for number in numbers:

sum_so_far += number

return sum_so_far

print(total([1, 2, 3]))

Printing out
the answer

314

314
 Using a Python function ― 4
def total(numbers):
nb: Unix prompt
sum_so_far = 0
$ python3 total1.py
for number in numbers:
sum_so_far += number 6

return sum_so_far

print(total([1, 2, 3]))

total1.py
315

The file total1.py in your home directories contains exactly the code you see
here.

315
Using a Python function ― 5
def total(numbers):

sum_so_far = 0
$ python3 total2.py
for number in numbers:
sum_so_far += number 6
10
return sum_so_far 0

print(total([1, 2, 3]))
print(total([7,-4,1,6,0])) Use the function
multiple times
print(total([]))

total2.py
316

316
Progress

Functions

“Structured programming”

Defining a function def function(input):

...

Returning a value return output

317

317
Exercise 18

5 minutes
318

318
 Reminder about indices
def total(numbers):
sum_so_far = 0
for number in numbers:
sum_so_far += number total2.py
return sum_so_far

Equivalent

def total(numbers):
sum_so_far = 0
for index in range(len(numbers)):
sum_so_far += numbers[index] total3.py
return sum_so_far
319

Let’s quickly remind ourselves about how we can uses indices for lists rather
than values from lists directly.
We found this particularly useful when we were traversing more than one list
at once.

319
 Example of multiple inputs
Want a function to add two lists of the same length term-by-term:

[1, 2, 3] & [5, 7, 4] [6, 9, 7]

[10, -5] & [15, 14] [25, 9]

[3, 7, 4, 1, 7] & [8, 4, -6, 1, 0] [11, 11, -2, 2, 7]

Two inputs
320

So how do we build functions that take in more than one input at once?

320
 Functions with multiple inputs
Multiple inputs are
separated by commas
def add_lists(a_list, b_list):

sum_list = []

for index in range(len(a_list)):

sum = a_list[index] + b_list[index]

sum_list.append(sum)

return sum_list

321

The Python syntax for multiple inputs is much the same as it is for a
mathemtical function: we separate the inputs by commas.

321
 Functions with multiple inputs
We have
two lists…
def add_lists(a_list, b_list):

sum_list = [] …so we have

to use indexing
for index in range(len(a_list)):

sum = a_list[index] + b_list[index]

sum_list.append(sum)

return sum_list

322

Note that functions that take in more than one list typically need to use
indices.

322
 Multiple outputs
Write a function to find minimum and maximum value in a list

[1, 2, 3] 1 & 3

[10, -5] -5 & 10

[3, 7, 4, 1, 7] 1 & 7

Two outputs
323

But what if we want to return multiple values?

We can write a function that determines the minimum value in a list, and we
can write a function that returns the maximum. What do we do if we want to
find both?

323
 Finding just the minimum
def min_list(a_list): minlist.py

min_so_far = a_list[0] List cannot be empty!

for a in a_list :

if a < min_so_far:
min_so_far = a

return min_so_far Returning a single value

324

So here’s the function that determines the minimum value in a list…

324
 Finding just the maximum
def max_list(a_list): maxlist.py

max_so_far = a_list[0]

for a in a_list :

if a > max_so_far: Only real change

max_so_far = a

return max_so_far Returning a single value

325

…and just the maximum.

325
 Finding both
def minmax_list(a_list):

min_so_far = a_list[0]
max_so_far = a_list[0]

for a in a_list :

if a < min_so_far:
min_so_far = a

if a > max_so_far:
max_so_far = a

326
return what? This is the real question

Combining the bodies of these two functions is quite straightforward.

But what do we return?

326
 Returning both
def minmax_list(a_list): minmaxlist.py

min_so_far = a_list[0]
max_so_far = a_list[0]

for a in a_list :

if a < min_so_far:
min_so_far = a

if a > max_so_far:
max_so_far = a

327
return min_so_far, max_so_far A pair of values

Two return two values we simply put them both after the return statement
separated by a comma, just as we would have done with the inputs.

327
 “Tuples”

e.g. min_value, max_value Pair

min_value, avg_value, max_value Triple

Commas

Often written with parentheses:

(min_value, max_value)

(min_value, avg_value, max_value)

328

These sets of values separated by commas (but not in square brackets to

make a list) are called “tuples” in Python. Sometimes they are written with
round brackets around them to make it clearer that they come together. But
it’s the comma that is the active ingredient making them a tuple, not the
brackets.

328
 Using tuples to return values
def … In the function definition

return (min_value, max_value)

Using the function

(minimum, maximum) = minmax_list(values)

329

If we return a pair of values in a tuple, we can also attach a pair of names to

them as a tuple too.

329
 Using tuples to attach names

alpha = 12
(alpha , beta) = ( 12, 56 )
beta = 56

330

We can do this outside the context of functions returning values, of course.

We can do it anywhere.

330
 Swapping values

>>> alpha = 12

>>> beta = 56

>>> (alpha, beta) = (beta, alpha) Swapping values

>>> print(alpha)

56

>>> print(beta)

12 331

Because the entire right hand side is evaluated before the left hand side is
considered this lets us use tuples for some particularly useful tricks. perhaps
the most useful is swapping two values.

331
 Assignment works right to left

alpha = 12

beta = 56

(alpha, beta) = (beta, alpha)

Stage 1: (beta, alpha) (56, 12)

Stage 2: (alpha, beta) = (56, 12)

332

The values associated with the names are evaluated first. Then the names
get reattached to those values, regardless of what names they might have
had before.

332
 Recall this “gotcha”:
!
a = 10

b = 7

a = a + b a = 17 a has now changed!

b = a - b b = a - b b ≠ 10 - 7 = 3
= 17 - 7
= 10

333

We can also use it to help us with our “change of coordinates” example.

333
 Again: assignment works right to left

a = 10

b = 7

(a, b) = (a + b, a - b)

Stage 1: (a+b, a-b) (10+7, 10-7) (17, 3)

Stage 2: (a, b) = (17, 3)

334

This works in exactly the same way.

334
Progress

Multiple inputs def thing(in1, in2, in3):

Multiple outputs return (out1, out2, out3)

“Tuples” (a, b, c)

Simultaneous assignment (a, b) = (a+b, a-b)

335

335
Exercise 19

Take the script from exercise 16 and turn it into:

1. the definition of a function file_stats() that takes a file

name and returns a triple (n_lines, n_words, n_chars)

2. a call to that function for file name treasure.txt

3. a print of that triple.

5 minutes
336

336
 One object or many?

two objects

“a pair of objects”
one object

337

Tuples tend to blur the boundary between multiple objects and a single
object.

337
 Tuples as single objects ― 1
>>> x = 20
>>> type(x)
<class 'int'>

>>> y = 3.14
>>> type(y)
<class 'float'>

>>> z = (20, 3.14) One name → Pair of values

>>> type(z)
<class 'tuple'> A single object 338

We can treat a tuple as a single object. It has a type called , “tuple”

unsurprisingly.

338
 Tuples as single objects ― 2

>>> z = (20, 3.14)

>>> print(z)

(20, 3.14)

>>> w = z Single name → Single tuple

>>> print(w)

(20, 3.14)

339

We can manipulate the tuple as a single object quite happily.

339
 Splitting up a tuple

>>> print(z)

(20, 3.14)

>>> (a,b) = z Two names → Single tuple

>>> print(a)

20

>>> print(b)

3.14
340

But a tuple is fundamentally made of separable pieces and can be split up.

340
 How tuples are like lists

>>> z = (20, 3.14)

>>> z[1] Indices

3.14

>>> len(z) Length

>>> z + (10, 2.17) Concatenation

(20, 3.14, 10, 2.17)

341

Tuples are a lot like lists at first glance.

341
 How tuples are not like lists

>>> z = (20, 3.14)

>>> z[0] = 10

Traceback (most recent call last):

File "<stdin>", line 1, in <module>
TypeError: 'tuple' object does not
support item assignment “Immutable”

342

They have one critical difference, though. A tuple is “immutable”. You cannot
change individual elements in a tuple. You get the whole tuple and you can’t
fiddle with it.

342
Progress

Tuples as single objects thing = (a, b, c)

Splitting up a tuple (x, y, z) = thing

Tuples as lists thing[0]

len(thing)
thing + thing

Immutability thing[0] = 10 ✗
343

343
 Functions we have written so far

total(list)

squares(N)

add_lists(list1,list2)

minmax_list(list)

344

To date we have written a small number of functions ourselves.

Once we become serious Python programmers using the computer for our
day job then we would expect to write many more.

344
 Reusing functions within a script
def square(limit):
... One definition

...

squares_a = square(34)

...

five_squares = squares(5) Multiple uses in

the same file
...

squares_b = squares(56)

...

Easy! 345

Within a script reusing a function is easy. We simply call the function

whenever we want it.

345
 Reusing functions between scripts?
def squares(limit):
... One definition

...
squares_a = squares(34)
Multiple uses in
...
multiple files
...
five_squares = squares(5)
... ...
squares_b = squares(56)
...

How? 346

But what happens if we want to use a function in more than one script?

346
 “Modules”

def square(limit): ...

...
five = squares(5)

def cubes(limit): ...

...

Definition Use

Module: a container of functions

347

Python has a mechanism to assist with this called “modules”. A module is a

collection of functions (and other material) which can then be imported into a
script and used within that script. If we can write our own module with our
own functions then we can import them into our own scripts.

347
 Modules: a worked example ― 1a
def squares(limit):
answer = []
utils.py for n in range(0,limit):
answer.append(n**2)
return answer

def total(numbers):
Starts empty sum_so_far = 0
for number in numbers:
sum_so_far += number
return sum_so_far

text = input('Number? ')

number = int(text)
squares_n = squares(number)
total_n = total(squares_n)
print(total_n)
348
sum_squares.py

We will start with a file called sum_squares.py which uses two functions to
add up the squares of numbers from zero to some limit. We want to transfer
those function definitions into a different file which we will call utils.py
(which starts empty) but still be able to use them in our original file.

348
 Modules: a worked example ― 1b

$ python3 sum_squares.py

Number? 5

30 = 0 + 1 + 4 + 9 + 16

$ python3 sum_squares.py

Number? 7

91 = 0 + 1 + 4 + 9 + 16 + 25 + 36

349

Just to prove I’m not fibbing, here it is working before we move anything
about.

349
 Modules: a worked example ― 2a
def squares(limit):
answer = []
for n in range(0,limit):
answer.append(n**2) text = input('Number? ')
return answer number = int(text)
squares_n = squares(number)
def total(numbers): total_n = total(squares_n)
sum_so_far = 0 print(total_n)
for number in numbers:
sum_so_far += number
return sum_so_far
sum_squares.py

utils.py
Move the definitions
into the other file.
350

Using the text editor we move the definitions from sum_squares.py to

utils.py.

350
 Modules: a worked example ― 2b

$ python3 sum_squares.py

Number? 5

Traceback (most recent call last):

File "sum_squares.py", line 4, in <module>
squares_n = squares(number)
NameError: name 'squares' is not defined

Because we have (re)moved its definition.

351

Unsurprisingly, this breaks sum_squares.py.

351
 Modules: a worked example ― 3a
def squares(limit): import utils
answer = []
for n in range(0,limit):
answer.append(n**2) text = input('Number? ')
return answer number = int(text)
squares_n = squares(number)
def total(numbers): total_n = total(squares_n)
sum_so_far = 0 print(total_n)
for number in numbers:
sum_so_far += number
return sum_so_far
sum_squares.py

utils.py
import: Make
import utils a reference to
and not the other file.
352
import utils.py

We need to import the functioned defined in utils.py into

sum_squares.py.
First, we add the instruction “import utils” to the top of the
stm_squares.py file.
Note that we import “utils”, not “utils.py”.

352
 Modules: a worked example ― 3b

$ python3 sum_squares.py

Number? 5

Traceback (most recent call last):

File "sum_squares.py", line 4, in <module>
squares_n = squares(number)
NameError: name 'squares' is not defined

Still can’t find the function(s).

353

On its own this is not sufficient.

353
 Modules: a worked example ― 4a
import utils

squares() text = input('Number? ')

utils.squares() number = int(text)
squares_n = utils.squares(number)
total_n = utils.total(squares_n)
total() print(total_n)

utils.total()
sum_squares.py

utils.…: Identify
the functions
as coming from
354
the module.

We have to indicate to Python that these functions it is looking for in

sum_squares.py come from the utils module. To do this we include
“utils.” at the start of their names.

354
 Modules: a worked example ― 4b

$ python3 sum_squares.py

Number? 5

30

$ python3 sum_squares.py

Number? 7

91
Working again!
☺
355

And now it works.

355
Progress

Sharing functions between scripts

“Modules”

Importing modules import module

Using functions from modules module.function()

356

356
Exercise 20

Move the function file_stats() from exercise18.py

into utils.py and edit exercise18.py so that it still works.

5 minutes
357

357
 The Python philosophy
A small core … plus lots
language … of modules

math
sys
“Batteries
included”
string 358

We have met the majority of the Python language already! But obviously
Python has facilities to do much more than we have seen so far. The trick is
that Python comes with a large number of modules of its own which have
functions for performing no end of useful things.
This philosophy is called “batteries included”. You probably already have the
module you need to do your specialist application.

358
 Example: the “math” module

>>> import math Load in the “math” module.

Run the sqrt() function…

>>> math.sqrt(2.0)
… from the math module.
1.4142135623730951

359

Let’s see an example of an “included battery”. At the very start of this course
we write ourselves a square root program. Now let’s see what Python offers
as an alternative.
We import the “math” module. (No trailing “s”; this is the American spelling.)
In the math module is a sqrt() function. We can access this as
math.sqrt().
Most of the fundamental mathematical operations can be found in the math
moWe will see how to find out exactly what is in a module in a few slides’
time.

359
 import module as alias

>>> import math

Too long to keep typing?
>>> math.sqrt(2.0)

1.4142135623730951

>>> import math as m “Alias”

>>> m.sqrt(2.0)

1.4142135623730951 360

There are those who object to typing. If “math” is too long then we can use
an aliasing trick to give the module a shorter name.
(The problem is rarely with math. There is a built-in module called
“multiprocessing” though which might get tiresome.)

360
 Don’t do these

>>> from math import sqrt

!
Much better
>>> sqrt(2.0) to track the
module.
1.4142135623730951

>>> from math import *

>>> sqrt(2.0)
!!
1.4142135623730951
361

python does permit you to do slightly more than that. You can suppress the
name of the module altogether.
You are beginners so please take it from an old hand on trust that this turns
out to be a very bad idea. You want to keep track of where your functions
came from!

361
 What system modules are there?
Python 3.2.3 comes with over 250 modules.

glob math argparse csv

io cmath datetime html

os random getpass json

signal colorsys logging re

subprocess email pickle string

sys http sqlite3 unicodedata

tempfile webbrowser unittest xml 362

There are many modules that come with Python.

362
 “Batteries included”

>>> help('modules')

Please wait a moment while I gather a list

of all available modules...

CDROM binascii inspect shlex

263 modules
bdb importlib shelve

Enter any module name to get more help. Or,

type "modules spam" to search for modules whose
descriptions contain the word "spam".
363
Not quite this simple

To find out exactly what modules come with your version of Python ask the
help system.
A word of warning, though. The text at the bottom “Enter any module
name…” is not quite right.
If you give the help() command with no argument then you are dropped
into an interactive help system. There you can type the name of a module or
type “modules spam”, etc.

>>> help()

Welcome to Python 3.1! This is the online help utility.

…
help> modules subprocess

Here is a list of matching modules. Enter any module

name to get more help.

subprocess - subprocess - Subprocesses with accessible

I/O streams

help> quit

>>>

363
 Additional downloadable modules
Numerical Databases

numpy pyodbc

scipy psycopg2 PostgreSQL

MySQLdb MySQL

cx_oracle Oracle

Graphics ibm_db DB2

matplotlib pymssql SQL Server

364

But, of course, there’s never the particular module you want. There are
modules provided by people who want Python to interoperate with whatever
it is they are offering.
There are three sets of additional modules that you may end up needing to
know about.
The numerical and scientific world has a collection of modules called
Numerical Python (“numpy”) and “scientific python” (”scipy”) which contain
enormous amounts of useful functions and types for numerical processing.
Every database under the sun offers a module to let Python access it.
Finally there is a module to offer very powerful 2D graphics for data
visualisation and presentation.

364
 An example system module: sys
import sys $ python3 argv.py one two three

print(sys.argv)
['argv.py', 'one', 'two', 'three']

index 0 1 2 3
argv.py

$ python3 argv.py 1 2 3

['argv.py', '1', '2', '3']

Always strings
365

We will take a brief look at another commonly used module to illustrate

some of the things Python has hidden away in its standard set.
The “sys” module contains many systems-y things. For example, it contains
a list called sys.argv which contains the argument values passed on the
command line when the script was launched.
Note two things:
1. that item zero in this list is always the name of the script itself,
2. the items are always strings

365
 sys.exit()

exit() What we have been using

sys.exit(rc) What we should use

“Return Code”: an integer

0: Everything was OK

≠0: Something went wrong

366

Also tucked away in the sys module is the sys.exit() function.

Up to this point we have been using the exit() function to quit our scripts
early. However this function is something of a filthy hack and sys.exit()
provides superior quitting and an extra facility we will be able to make use
of.
The sys.exit() function takes an integer argument. This is the program’s
“return code” which is a very short message back to the operating system to
indicate whether the program completed successfully or not. A return code
of 0 means “success”. A non-zero return code means failure. Some
programs use different non-zero codes for different failures but many
(most?) simply use a value of 1 to mean “something went wrong”.
If your script simply stops because it reached the end then Python does an
automatic sys.exit(0).

366
 An example system module: sys
But also…

sys.modules All modules currently imported

sys.path Directories Python searches for modules

sys.version Version of Python

sys.stdin Where input inputs from

sys.stdout Where print prints to

sys.stderr Where errors print to

sys.float_info All the floating point limits

367

…and there’s more!

And there’s plenty more…

367
 Modules in Python

“How do I do
X in Python?”

“What’s the Python

module to do X?”

“Where do I find
Python modules?”
368

So the Python philosophy places a lot of functionality into its modules.

This means that we have to be able to find modules and know what they can
do.

368
 Finding modules

Python: Built-in modules

SciPy: Scientific Python modules

PyPI: Python Package Index

Search: “Python3 module for X”

369

Some useful URLs:

http://docs.python.org/py3k/py-modindex.html
This contains the list of all the “batteries included” modules that come with
Python. For each module it links through to their documentation.

http://www.scipy.org/Topical_Software
http://numpy.scipy.org/
Scientific Python contains very many subject-specific modules for Python.
Most depend on the Numerical Python module numpy.

http://pypi.python.org/pypi (do check for Python3 packages)

This is the semi-official dumping ground for everything else.

http://www.google.co.uk/
And for everything else there’s Google (who are big Python users, by the
way).

369
 Help with modules

>>> import math

>>> help(math)

NAME
math

DESCRIPTION
This module is always available. It provides
access to the mathematical functions defined
by the C standard.

… 370

I promised information on how to find out what is in a module. Here it is.

Once a module has been imported you can ask it for help.

370
 Help with module functions
…

FUNCTIONS

acos(x)
Return the arc cosine (measured in
radians) of x.
…

>>> math.acos(1.0)

0.0

371

The help will always include information on every function in the module…

371
 Help with module constants
…

DATA
e = 2.718281828459045
pi = 3.141592653589793
…

>>> math.pi

3.141592653589793

372

…and every data item.

372
 Help for our own modules?
def squares(limit): >>> import utils
answer = []
for n in range(0,limit):
answer.append(n**2)
>>> help(utils)
return answer
NAME
def total(numbers): utils
sum_so_far = 0
for number in numbers:
sum_so_far += number FUNCTIONS
return sum_so_far squares(limit)

total(numbers)
utils.py
Basic help already FILE
provided by Python /home/y550/utils.py
373

What help can we get from our own module?

By now you should have a utils.py file with some functions of your own in
it. The help simply lists the functions the module contains and the file it is
defined in.

373
 Adding extra help text
"""Some utility functions >>> import utils Fresh start
from the Python for
Absolute Beginners course
"""
>>> help(utils)

def squares(limit): NAME

answer = [] utils
for n in range(0,limit):
answer.append(n**2)
return answer DESCRIPTION
Some utility functions
def total(numbers): from the Python for
sum_so_far = 0 Absolute Beginners course
for number in numbers:
sum_so_far += number
return sum_so_far FUNCTIONS
squares(limit)
374

utils.py total(numbers)

But we can do better than that.

If we simply put a Python string (typically in long text triple quotes) at the top
of the file before any used Python (but after comments is fine) then this
becomes the description text in the help.
Note: You need to restart Python and re-import the module to see changes.

374
 Adding extra help text to functions
"""Some utility functions >>> import utils Fresh start
from the Python for
Absolute Beginners course
"""
>>> help(utils)

def squares(limit): NAME

"""Returns a list of utils
squares from zero to
limit**2.
""" DESCRIPTION
answer = [] ...
for n in range(0,limit):
answer.append(n**2) FUNCTIONS
return answer squares(limit)
Returns a list of
squares from zero to
limit**2. 375

utils.py

If we put text immediately after a def line and before the body of the
function it becomes the help text for that function, both in the module-as-a-
whole help text…

375
 Adding extra help text to functions
"""Some utility functions >>> import utils Fresh start
from the Python for
Absolute Beginners course
"""
>>> help(utils.squares)

def squares(limit): squares(limit)

"""Returns a list of Returns a list of
squares from zero to squares from zero to
limit**2.
limit**2.
"""
answer = []
for n in range(0,limit):
answer.append(n**2)
return answer

376

utils.py

…and in the function-specific help text.

376
Progress
Python a small language… Functionality Module

…with many, many modules

System modules

Foreign modules

Modules provide help help(module)

Doc strings help(module.function)

377

377
Exercise 21
Add help text to your utils.py file.

5 minutes
378

378
 Recap: Lists & Indices

greek = ['alpha', 'beta', 'gamma']

list
str 5 a l p h a greek[0]
3

Position 0 str 4 b e t a greek[1]

Position 1
Position 2 str 5 g a m m a greek[2]

379

We have one last Python type to learn about. To give it some context, we will
recap the list type that we have spent so much time using.
A list is basically an ordered sequence of values. The position in that
sequence is known as the index.

379
 “Index in ― Value out”

index list value

Must be
a number

1 greek[1] 'beta'
380

If we now forget about the internals of a list, though, we can think of it as

“some sort of Python object” that takes in a number (the index) and spits out
a value.

380
 Other “indices”: Strings?

English dictionary Spanish

'cat' 'gato'

'dog' dictionary 'perro'

'mouse' 'ratón'
381

Can we generalise on this idea by moving away from the input (the index)
needing to be a number?
Can we model a dictionary where we take in a string (a word in English, say)
and give out a different string (the corresponding word in Spanish, say).
(Note: the author is fully aware that translation is not as simple as this. This
is just a toy example.)

381
 Other “indices”: Tuples?

(x, y) dictionary Objects

(2,5) 'Treasure'
dictionary
(4,5) 'Fortress'

382

Or, perhaps, pairs of numbers (x,y) in and items on a map out?

382
 Python “dictionaries”

>>> en_to_es = { 'cat':'gato' , 'dog':'perro' }

>>> en_to_es['cat']

'gato'

383

Python does have exactly such a general purpose mapper which it calls a
“dict”, short for “dictionary”.
Here is the Python for establishing a (very small) English to Spanish
dictionary that knows about two words.
We also see the Python for looking up a word in the dictionary.
We will review this syntax in some detail…

383
 Creating a dictionary — 1

Curly brackets
Comma

{ 'cat':'gato' , 'dog':'perro' }

'cat' → 'gato' 'dog' → 'perro'

384

First we will look at creating a dictionary. In the same way that we can create
a list with square brackets, we can create a dictionary with curly ones.
Each item in a dictionary is a pair of values separated by a colo
They are separated by commas.n.

384
 Creating a dictionary — 2

'cat' : 'gato'

{ 'cat':'gato' , 'dog':'perro' }

“key” : “value”

385

The pairs of items separated by colons are known as the “key” and “value”.
The key is what you put in (the English word in this example) that you look
up in the dictionary and the value is what you get out (the translation into
Spanish in this example).

385
 Using a dictionary — 1

>>> en_to_es = { 'cat':'gato' , 'dog':'perro' }

Creating the dictionary

>>> en_to_es['cat'] Using the dictionary

'gato'

386

Now we have seen how to create a (small) dictionary we should look at how
to use it.

386
 Using a dictionary — 2

Square brackets

en_to_es[ 'cat' ] 'gato'

dictionary key value

387

To look something up in a dictionary we pass it to the dictionary in exactly

the same way as we passed the index to a list: in square brackets. Curly
brackets are just for creating a dictionary; after that it’ square brackets
again.

387
 Missing keys

>>> en_to_es = { 'cat':'gato' , 'dog':'perro' }

✓
>>> en_to_es['dog']

'perro'

✗
>>> en_to_es['mouse']

Traceback (most recent call last):

File "<stdin>", line 1, in <module>
KeyError: 'mouse' Error message
388

The equivalent to shooting off the end of a list is asking for a key that’s not
in a dictionary.

388
 Dictionaries are one-way
!
>>> en_to_es = { 'cat':'gato' , 'dog':'perro' }

✓
>>> en_to_es['dog']

'perro'

✗
>>> en_to_es['perro']

Traceback (most recent call last):

File "<stdin>", line 1, in <module>
KeyError: 'perro'
389
Looking for a key

Also note that dictionaries are one-way.

389
 Adding to a dictionary

>>> en_to_es = { 'cat':'gato' , 'dog':'perro' }

Initial dictionary has no 'mouse'

>>> en_to_es['mouse'] = 'ratón'

Adding 'mouse' to the dictionary

>>> en_to_es['mouse']

'ratón'

390

Adding key-value pairs to a dictionary is a lot easier than it is with lists. With
lists we needed to append on the end of a list. With dictionaries, because
there is no inherent order, we can simply define them with a simple
expression on the left hand side.

390
 Removing from a dictionary

>>> print(en_to_es)

{'mouse': 'ratón', 'dog': 'perro', 'cat': 'gato'}

>>> del en_to_es['dog']

>>> print(en_to_es)

{'mouse': 'ratón', 'cat': 'gato'}

391

We can use del to remove from a dictionary just as we did with lists.

391
Progress

Dictionaries Key Value

{ key1:value1 , key2:value2, key3:value3 }

Looking up values dictionary[key] value

Setting values dictionary[key] = value

Removing keys del dictionary[key]

392

392
Exercise 22

Complete exercise22.py to create

an English to French dictionary.

cat chat

dog chien

mouse souris

snake serpent
5 minutes
393

393
 What’s in a dictionary? ― 1
>>> en_to_es
{'mouse': 'ratón', 'dog': 'perror', 'cat': 'gato'}

>>> en_to_es.keys()
dict_keys(['mouse', 'dog', 'cat'])

Orders
match
>>> en_to_es.values()
dict_values(['ratón', 'perro', 'gato'])

Just treat them like lists (or convert them to lists) 394

To date we have created our own dictionaries. If we are handed one how do
we find out what keys and values are in it?
Dictionaries support two methods which return the sort-of-lists of the keys
and values. We mention them here only for completeness.
Don’t forget that you can always convert a sort-of-list into a list with the
list() function.

394
 What’s in a dictionary? ― 2
>>> en_to_es.items() Most useful method
dict_items([('mouse', 'ratón'), ('dog', 'perro'),
('cat', 'gato')])
(Key,Value) pairs/tuples

>>> for (english, spanish) in en_to_es.items():

... print(spanish, english)
...
ratón mouse
perro dog
gato cat
395

By far the best way to get at the contents of a dictionary is to use the
items() method which generates a sort-of-list of the key-value pairs as
tuples. Running a for… loop over this list is the easiest way to process the
contents of a directory.

395
 What’s in a dictionary? ― 3
Common simplification

>>> list(en_to_es.items())
[('mouse','ratón'), ('dog','perro'),('cat','gato')]

396

Don’t be afraid to convert it explicitly into a list. Unless you dictionary is huge
you won’t see any problem with this.

396
 Getting the list of keys

list() list of
dictionary
keys
{'the': 2, 'cat': 1, 'sat': 1, 'on': 1, 'mat': 1}

['on', 'the', 'sat', 'mat', 'cat']

397

Unfortunately when you convert a dictionary directly into a list you get the list
of keys not the list of )(key,value) pairs. This is a shame but is a compromise
for back compatibility with previous versions.

397
 Is a key in a dictionary?

>>> en_to_es['snake']

Traceback (most recent call last):

File "<stdin>", line 1, in <module>
KeyError: 'snake' Want to avoid this

>>> 'snake' in en_to_es

False We can test for it

398

Because of this conversion to the list of keys we can ask if a key is in a

dictionary using the in keyword without having to know its corresponding
value.

398
 Example: Counting words ― 1

words = ['the','cat','sat','on','the','mat']

counts = {'the':2,'cat':1,'sat':1,'on':1,'mat':1}

399

Let’s have a serious worked example.

We might be given a list of words and want to count the words by how many
times they appear in the list.

399
 Example: Counting words ― 2

words = ['the','cat','sat','on','the','mat']

counts = {} Start with an empty dictionary

for word in words: Work through all the words

Do something

400

We start by creating an empty dictionary. It’s empty because we haven’t

read any words yet.
Then we loop through the list of words using a standard for… loop.
For each word we have to do something to increment the count in the
dictionary.

400
 Example: Counting words ― 3

words = ['the','cat','sat','on','the','mat']

counts = {}

for word in words:

✗ counts[word] += 1 This will not work

counter1.py
401

Unfortunately a simply increment of the value in the dictionary isn’t enough.

401
 Why doesn’t it work?
counts = {'the':1, 'cat':1}

The key must already

be in the dictionary.
✓ counts['the'] += 1

counts['the'] = counts['the'] + 1

Key is not in

✗ counts['sat'] += 1
the dictionary!

counts['sat'] = counts['sat'] + 1
402

We cannot increment a value that isn’t there. Until the program meets a
word for the first time it has no entry in the dictionary, and certainly not an
entry with numerical value 0.

402
 Example: Counting words ― 4

words = ['the','cat','sat','on','the','mat']

counts = {}

for word in words:

if word in counts:

counts[word] += 1

else:

Do something Need to add the key

403

So we have to test to see if the word is already in the dictionary to increment

it if it is there and to do something else if it is not. Note how we use the “if
key in dictionary” test.

403
 Example: Counting words ― 5

words = ['the','cat','sat','on','the','mat']

counts = {}

for word in words:

if word in counts:

counts[word] += 1

else:

counts[word] = 1 counter2.py
404

print(counts)

That something else is to create it with its initial value of 1 (because we have
met the word once now).

404
 Example: Counting words ― 6

$ python3 counter2.py

{'on': 1, 'the': 2, 'sat': 1, 'mat': 1, 'cat': 1}

You cannot predict the order of the

! keys when a dictionary prints out.

405

Dictionaries are unordered entities. You cannot predict the order that the
keys will appear when you print a dictionary or step through its keys.

405
 Example: Counting words ― 7

✗
print(counts) Too ugly

items = list(dictionary.items()) Better

items.sort()

for (key, value) in items:

print(key, value) counter3.py

406

Simply printing a dictionary gives ugly output.

We can pull out the (key,value) pairs and print them individually if we want.
Notice the use of pulling out the items, converting them into a list and then
sorting them.

406
Example: Counting words ― 8

$ python3 counter3.py

cat 1
mat 1
on 1
sat 1
the 2

407

407
Progress

Inspection methods dictionary.keys()

dictionary.values()
dictionary.items()

Testing keys in dictionaries if key in dictionary:

...

Creating a list of keys keys = list(dictionary)

408

408
Exercise 23

Complete exercise23.py to write

a function that reverses a dictionary.

{'mouse':'ráton', 'cat':'gato', 'dog':'perro'}

{'ráton':'mouse', 'gato':'cat', 'perro':'dog'}

10 minutes
409

409
 Formatted output
$ python3 counter3.py

cat 1
mat 1
on 1 Ugly!
sat 1
the 2

cat 1
mat 1
on 1 We want data
sat 1 nicely aligned
the 2
410

We have one last topic to cover. The output from our word counter is not as
pretty as it might be. The last topic we will cover is Python text formatting.
Full details of the formatting system can be found online at
docs.python.org/py3k/library/string.html#formatspec

410
 The format() method

>>> 'xxx{}yyy{}zzz'.format('A', 100)

'xxxAyyy100zzz'

'xxx{}yyy{}zzz'

'xxxAyyy100zzz'
411

In Python 3 the string type has a method called format() which takes
arguments and returns another string which is the original with the method’s
arguments inserted into it in certain places. Those places are marked with
curly brackets in the string.
Curly brackets are otherwise quite normal characters. It’s only the
format() method that cares about them.
In its simplest form, the format() method replaces each pair of curly
brackets with the corresponding argument.

411
 String formatting ― 1

>>> 'xxx{:5s}yyy'.format('A')

'xxxA␣␣␣␣yyy'

'xxx{:5s}yyy ' {:5s}

'xxxA␣␣␣␣yyy ' s ― substitute a string

5 ― pad to 5 spaces
(left aligns by default) 412

The real fun starts when we put something inside those curly brackets.
These are the formatting instructions.
The simplest examples start with a colon followed by some layout
instructions. (We will see what comes in front of the colon in a few slides
time.) The number indicates how many spaces to allocate for the insertion
and the letter that follows tells it what type ob object to expect. “s” stands for
string.

412
 String formatting ― 2

>>> 'xxx{:<5s}yyy'.format('A')

'xxxA␣␣␣␣yyy'

'xxx{:<5s}yyy ' {:<5s}

'xxxA␣␣␣␣yyy ' < ― align to the left (←)

413

By default strings align to the left of their space. We can be explicit about
this by inserting a left angle bracket, which you can think of as an arrow
head pointing the direction of the alignment.

413
 String formatting ― 3

>>> 'xxx{:>5s}yyy'.format('A')

'xxx␣␣␣␣Ayyy'

'xxx{:>5s}yyy ' {:>5s}

'xxx␣␣␣␣Ayyy ' > ― align to the right (→)

414

If we want right aligned strings then we have to use the alignment marker.

414
 Integer formatting ― 1

>>> 'xxx{:5d}yyy'.format(123)

'xxx␣␣123yyy'

'xxx{:5d}yyy ' {:5d}

d ― substitute an integer
'xxx␣␣123yyy ' (digits)
5 ― pad to 5 spaces
(right aligns by default)
415

If we change the letter to a “d” we are telling the format() method to insert
an integer (“digits”). These align to the right by default.

415
 Integer formatting ― 2

>>> 'xxx{:>5d}yyy'.format(123)

'xxx␣␣123yyy'

'xxx{:>5d}yyy ' {:>5d}

> ― align to the right (→)
'xxx␣␣123yyy '
416

We can be explicit about this if we want.

416
 Integer formatting ― 3

>>> 'xxx{:>5d}yyy'.format(123)

'xxx␣␣123yyy'

'xxx{:<5d}yyy ' {:<5d}

< ― align to the left (←)
'xxx123␣␣yyy '
417

And we have to be explicit to override the default.

417
 Integer formatting ― 4

>>> 'xxx{:05d}yyy'.format(123)

'xxx00123yyy'

'xxx{:05d}yyy ' {:05d}

0 ― pad with zeroes
'xxx00123yyy '
418

If we precede the width number with a zero then the number is padded with
zeroes rather than spaces and alignment is automatic.

418
 Integer formatting ― 5

>>> 'xxx{:+5d}yyy'.format(123)

'xxx␣+123yyy'

'xxx{:+5d}yyy ' {:05d}

+ ― always show sign
'xxx␣+123yyy '
419

If we put a plus sign in front of the number then its sign is always shown,
even if it is positive.

419
 Integer formatting ― 6

>>> 'xxx{:+05d}yyy'.format(123)

'xxx␣+0123yyy'

'xxx{:+05d}yyy ' {:05d}

+ ― always show sign
'xxx+0123yyy ' 0 ― pad with zeroes

420

And we can combine these.

420
 Integer formatting ― 7

>>> 'xxx{:5,d}yyy'.format(1234)

'xxx1,234yyy'

'xxx{:5,d}yyy ' {:5,d}

, ― 1,000s
'xxx1,234yyy '
421

Adding a comma between the width and the “d” adds comma breaks to
large numbers.

421
 Floating point formatting ― 1

>>> 'xxx{:5.2f}yyy'.format(1.2)

'xxx 1.20yyy'

'xxx{:5.2f}yyy ' {:5.2f}

f ― substitute a float
'xxx␣1.20yyy ' 5 ― 5 places in total

.2 ― 2 places after the point

422

Floating point numbers are slightly more complicated. The width parameter
has two parts, separated by a decimal point. The first number is the width of
the entire number (just as it is for strings and integers). The number after the
decimal point is the number of decimal points of precision that should be
included in the formatted output.

422
 Floating point formatting ― 2

>>> 'xxx{:f}yyy'.format(1.2)

'xxx1.200000yyy'

'xxx{:f}yyy ' {:f}

'xxx1.200000yyy ' {:.6f}

423

The default is to have six decimal points of precision and to make the field
as wide as it needs to be.

423
 Ordering and repeats ― 1
0 1 2
>>> 'X{:s}X{:d}X{:f}X'.format('abc', 123, 1.23)
0 1 2
'XabcX123X1.230000X'
0 1 2
Equivalent

>>> 'X{0:s}X{1:d}X{2:f}X'.format('abc', 123, 1.23)

0 1 2
'XabcX123X1.230000X'
0 1 2
424

What comes before the colon?

This is a selection parameter detailing what argument to insert.
The arguments to format() are given numbers starting at zero. We can
put these numbers in front of the colon.

424
 Ordering and repeats ― 2

>>> 'X{0:s}X {2:f}X{1:d} X'.format('abc', 123, 1.23)

0 1 2
'XabcX 1.230000X123X'
0 2 1

>>> 'X{0:s}X{1:d}X{1:d} X'.format('abc', 123, 1.23)

0 1 2
'XabcX123X123 X'
0 1 1
425

We can also use them to change the order that items appear in or to have
them appear more than once (or not at all).

425
 Formatting in practice
...
formatting = '{:3} {:1}'

for (word, number) in items:

print(formatting.format(word,number))

$ python3 counter4.py

cat 1
mat 1
on 1
sat 1
the 2 426

The script counter4.py is the same as counter3.py but with the new
formatting for its output.

426
Progress

Formatting string.format(args)

Numbered parameters {0} {1} {2}

Substitutions {:>6s}

{:+06d}

{:+012.7f}

427

427
Exercise 24

Complete exercise24.py
to format the output as shown:

[(Joe,9), (Samantha,45), (Methuselah,969)]

Joe 9
Samantha 45
Methuselah 969

5 minutes
428

428
 And that's it! (And “it” is a lot!)
Text “if” test Reading files
Prompting Indented blocks Writing files
Numbers Lists Functions
Arithmetic Indices “Structured programming”
Comparisons Object methods Tuples
Booleans Built-in help “for” loops
Variables “for” loops Modules
Deleting names “Treat it like a list…” Dictionaries
“while” loop Values direct/via index Formatting
429

And congratulations!
You have completed an introductory course on Python. Well done.
It is only an introductory course and there is more. But do not let that
dishearten you; just take a look at what you have accomplished. You now
have a firm grounding to go further with Python or to start learning other
programming languages. (But the author would like you to stick with
Python.)

429
 But wait! There’s more…

Advanced topics: Self-paced introductions to modules

Object-oriented programming in Python

430

If you do want more Python the UCS offers a selection of self-paced

courses on some additional language features and on various Python
modules to let you learn how to use Python for a specific purpose. We also
offer a taught course introducing you to the world of object-oriented
programming where you get to write your own methods and types.

430
 Congratulations!
Text “if” test Reading files
Prompting Indented blocks Writing files
Numbers Lists Functions
Arithmetic Indices “Structured programming”
Comparisons Object methods Tuples
Booleans Built-in help “for” loops
Variables “for” loops Modules
Deleting names “Treat it like a list…” Dictionaries
“while” loop Values direct/via index Formatting
431

So thank you and congratulations again.

431
Python 3 formatting codes
Use of the format() method:
>>> '{:4s}␣{:3d}'.format('Dave', 27)
'Dave␣␣27'
>>>
The symbol “␣” is used to represent a space.

Strings
hello
{:s} hello
{:10s} hello␣␣␣␣␣
{:<10s} hello␣␣␣␣␣
{:>10s} ␣␣␣␣␣hello

Integers
1,234 -1,234
{:d} 1234 -1234
{:10d} ␣␣␣␣␣␣1234 ␣␣␣␣␣-1234
{:<10d} 1234␣␣␣␣␣␣ -1234␣␣␣␣␣
{:>10d} ␣␣␣␣␣␣1234 ␣␣␣␣␣-1234
{:+10d} ␣␣␣␣␣+1234 ␣␣␣␣␣-1234
{:010d} 0000001234 -000001234
{:+010d} +000001234 -000001234
{:10,d} ␣␣␣␣␣1,234 ␣␣␣␣-1,234
{:+10,d} ␣␣␣␣+1,234 ␣␣␣␣-1,234

Floating point
3.141592653589793 -1.2
{:f} 3.141593 -1.200000
{:10f} ␣␣3.141593 ␣-1.200000
{:10.5f} ␣␣␣3.14159 ␣␣-1.20000
{:<10.5f} 3.14159␣␣␣ -1.20000␣␣
{:>10.5f} ␣␣␣3.14159 ␣␣-1.20000
{:+10.5f} ␣␣+3.14159 ␣␣-1.20000
{:010.5f} 0003.14159 -001.20000
{:+010.5f} +003.14159 -001.20000