Question

In this part of the Python programming tutorial, we will talk about functions.

A function is a piece of code in a program. The function performs a specific task. The advantages of using functions are:

• Reducing duplication of code
• Decomposing complex problems into simpler pieces
• Improving clarity of the code
• Reuse of code
• Information hiding

Functions in Python are first-class citizens. It means that functions have equal status with other objects in Python. Functions can be assigned to variables, stored in collections or passed as arguments. This brings additional flexibility to the language.

There are two basic types of functions. Built-in functions and user defined ones. The built-in functions are part of the Python language. Examples are: dir() , len() or abs() . The user defined functions are functions created with the def keyword.

Defining functions

A function is created with the def keyword. The statements in the block of the function must be indented.

def function():
  pass

The def keyword is followed by the function name with round brackets and a colon. The indented statements form a body of the function.

The function is later executed when needed. We say that we call the function. If we call a function, the statements inside the function body are executed. They are not executed until the function is called.

myfunc()

To call a function, we specify the function name with the round brackets.

#!/usr/bin/python

"""
The ret.py script shows how to work with
functions in Python. 
author: Jan Bodnar
ZetCode, 2011
"""


def showModuleName():
  print __doc__

def getModuleFile():
   return __file__

a = showModuleName()
b = getModuleFile()

print a, b

The string at the top of the script is called the documentation string. It documents the current script. The file in which we put Python code is called a module. We define two functions. The first function will print the module doc string. The second will return the path of our module. Function may or may not return a value. If they explicitly do not return a value, they implicitly return None . The __doc__ and __file__ are special state attributes. Note that there are two underscores on both sides of the attribute.

$ ./ret.py 

The ret.py script shows how to work with
functions in Python. 
author: Jan Bodnar
ZetCode, 2011

None ./ret.py

Definitions of functions must precede their usage. Otherwise the interpreter will complain with a NameError .

#!/usr/bin/python

def f1():
  print "f1()"
  
f1()
#f2()

def f2():
  print "f2()"

In the above example we have two definitions of functions. One line is commented. Function call cannot be ahead of its definition.

#f2()

def f2():
  print "f2()"

We can call the f2() only after its definition. Uncommenting the line we get a NameError .

Where to define functions

Functions can be defined inside a module, a class or another function. Function defined inside a class is called a method.

#!/usr/bin/python

class Some:

  @staticmethod  
  def f():
    print "f() method"

def f():
  print "f() function"
  
def g():
  def f():
    print "f() inner function"
  f()    
    
Some.f()
f()
g()

In this example, we define an f() function in all possible places.

class Some:

  @staticmethod  
  def f():
    print "f() method"

A static method is defined with a decorator in a Some class.

def f():
  print "f() function"

The function is defined in a module.

def g():
  def f():
    print "f() inner function"
  f()   

Here the f() function is defined inside another g() function. It is an inner function.

Some.f()
f()
g() 

The static method is called by specifying the class name, the dot operator and the function name with square brackets. Other functions are called using their names and square brackets.

$ ./defining.py
f() method
f() function
f() inner function

This is the output.

Functions are objects

Functions in Python are objects. They can be manipulated like other objects in Python. Therefore functions are called first-class citizens. This is not true in other OOP languages like Java or C#.

#!/usr/bin/python

def f():
  """This function prints a message """
  print "Today it is a cloudy day"
  
print isinstance(f, object)
print id(f)
  
print f.func_doc
print f.func_name

In this script we show that our function is an object too.

def f():
  """This function prints a message """
  print "Today it is a cloudy day"

We define an f() function. It prints a message to the console. It has a documentation string.

print isinstance(f, object)

The isinstance() function checks whether the f() function is an instance of the object . All objects in Python inherit from this base entity.

print id(f)

Each object in Python has a unique id. The id() function returns the object's id.

print f.func_doc
print f.func_name

Objects may have attributes. Here we print two attributes of the function.

$ ./fobj.py
True
3077407212
This function prints a message 
f

Output of the program.

Objects can be stored in collections and passed to functions.

#!/usr/bin/python

def f():
  pass

def g():
  pass
  
def h(f):
  print id(f)
  
a = (f, g, h)

for i in a:
  print i
  
h(f)
h(g)

We define three functions. We place them in a tuple and pass them to a function.

a = (f, g, h)

for i in a:
  print i

We place three function objects in a tuple and traverse it with a for loop.

h(f)
h(g)

We pass the f() and g() functions to the h() function.

$ ./fobj2.py
<function f at 0xb7664fb4>
<function g at 0xb766c1b4>
<function h at 0xb766c3ac>
3076935604
3076964788

Output of the fobj.py script.

Three kinds of functions

Looking from a particular point of view, we can discern three kinds of functions. Functions that are always available for usage, functions that are contained within external modules, which must be imported and functions defined by a programmer with the def keyword.

#!/usr/bin/python

from math import sqrt

def cube(x):
  return x * x * x  
  
print abs(-1)
print cube(9)
print sqrt(81)

Three kinds of functions are present in the above code.

from math import sqrt

The sqrt() function is imported from the math module.

def cube(x):
  return x * x * x  

The cube() function is a custom defined function.

print abs(-1)

The abs() function is a built-in function readily accessible. It is part of the core of the language.

The return keyword

A function is created to do a specific task. Often there is a result from such a task. The return keyword is used to return values from a function. A function may or may not return a value. If a function does not have a return keyword, it will send a None value.

#!/usr/bin/python

def showMessage(msg):
  print msg

def cube(x):
  return x * x * x   
   
x = cube(3)  
print x

showMessage("Computation finished.")
print showMessage("Ready.")

We have two functions defined. One uses the return keyword, one does not.

def showMessage(msg):
  print msg

The showMessage() function does not return explicitly a value. It shows a message on the console.

def cube(x):
  return x * x * x

The cube() functions computes an expression and returns its result with the return keyword.

x = cube(3) 

In this line we call the cube() function. The result of the computation of the cube() function is returned and assigned to the x variable. It holds the result value now.

showMessage("Computation finished.")

We call the showMessage() function with a message as a parameter. The message is printed to the console. We do not expect a value from this function.

print showMessage("Ready.")

This code produces two lines. One is a message printed by the showMessage() function. The other is the None value, which is implicitly sent by functions without the return statement.

$ ./return.py
27
Computation finished.
Ready.
None

Example output.

We can send more that one value from a function. The objects after the return keyword are separated by commas.

#!/usr/bin/python

n = [1, 2, 3, 4, 5]

def stats(x):
  mx = max(x)
  mn = min(x)
  ln = len(x)
  sm = sum(x)
  
  return mx, mn, ln, sm  
  
mx, mn, ln, sm = stats(n)
print stats(n)

print mx, mn, ln, sm

There is a definition of a stats() function. This function returns four values.

return mx, mn, ln, sm

The return keyword sends back four numbers. The numbers are separated by a comma character. In fact, we have sent a tuple containing these four values. We could also return a list instead of a tuple.

mx, mn, ln, sm = stats(n)

The returned values are assigned to local variables.

$ ./return2.py 
(5, 1, 5, 15)
5 1 5 15

Output.

Function redefinition

Python is dynamic in nature. It is possible to redefine an already defined function.

#!/usr/bin/python

from time import gmtime, strftime


def showMessage(msg):
  print msg
    
showMessage("Ready.")

def showMessage(msg):
  print strftime("%H:%M:%S", gmtime()),
  print msg
  
showMessage("Processing.")  

We define a showMessage() function. Later we provide a new definition of the same function.

from time import gmtime, strftime

From the time module we import two functions which are used to compute the current time.

def showMessage(msg):
  print msg    

This is the first definition of a function. It only prints a message to the console.

def showMessage(msg):
  print strftime("%H:%M:%S", gmtime()),
  print msg

Later in the source code, we set up a new definition of the showMessage() function. The message is preceded with a timestamp.

$ ./redefinition.py
Ready.
23:49:33 Processing.

Ouput of the script.

Function arguments

Most functions accept arguments. Arguments are values that are sent to the function. The functions process the values and optionally return some value back.

#!/usr/bin/python

def C2F(c):
  return c * 9/5 + 32

print C2F(100)
print C2F(0)
print C2F(30)

In our example, we convert Celsius temperature to Fahrenheit. The C2F() function accepts one argument c, which is the Celsius temperature.

$ ./fahrenheit.py 
212
32
86

The arguments in Python functions may have implicit values. An implicit value is used if no value is provided.

#!/usr/bin/python

def power(x, y=2):
  r = 1
  for i in range(y):
     r = r * x
  return r

print power(3)
print power(3, 3)
print power(5, 5)

Here we created a power function. The function has one argument with an implicit value. We can call the function with one or two arguments.

$ ./power.py 
9
27
3125

Python functions can specify their arguments with a keyword. This means that when calling a function, we specify both a keyword and a value. When we have multiple arguments and they are used without keywords, the order in which we pass those arguments is crucial. If we expect a name, age, or sex in a function without keywords, we cannot change their order. If we use keywords, we can.

#!/usr/bin/python

def display(name, age, sex):
   print "Name: ", name
   print "Age: ", age
   print "Sex: ", sex


display("Lary", 43, "M")
display("Joan", 24, "F")

In this example, the order in which we specify the arguments is important. Otherwise, we get incorrect results.

$ ./persons.py 
Name:  Lary
Age:  43
Sex:  M
Name:  Joan
Age:  24
Sex:  F

#!/usr/bin/python

# person2.py

def display(name, age, sex):
   print "Name: ", name
   print "Age: ", age
   print "Sex: ", sex


display(age=43, name="Lary", sex="M")
display(name="Joan", age=24, sex="F")

Now we call the functions with their keywords. The order may be changed, although it is not recommended to do so. Note that we cannot use a non-keyword argument after a keyword argument. This would end in a syntax error.

display("Joan", sex="F", age=24)

This is a legal construct. A non-keyword argument may be followed by keyword arguments.

display(age=24, name="Joan", "F")

This will end in a syntax error. A non-keyword argument may not follow keyword arguments.

Functions in Python can even accept arbitrary number of arguments.

#!/usr/bin/python

def sum(*args):
   '''Function returns the sum 
   of all values'''
   r = 0
   for i in args:
    r += i
   return r


print sum.__doc__
print sum(1, 2, 3)
print sum(1, 2, 3, 4, 5)

We use the * operator to indicate that the function will accept arbitrary number of arguments. The sum() function will return the sum of all arguments. The first string in the function body is called the function documentation string. It is used to document the function. The string must be in triple quotes.

$ ./summation.py 
Function returns the sum 
   of all values
6
15

We can also use the ** construct in our functions. In such a case, the function will accept a dictionary. The dictionary has arbitrary length. We can then normally parse the dictionary, as usual.

#!/usr/bin/python

def display(**details):

   for i in details:
    print "%s: %s" % (i, details[i])


display(name="Lary", age=43, sex="M")

This example demonstrates such a case. We can provide arbitrary number of key-value arguments. The function will handle them all.

$ ./person.py 
age: 43
name: Lary
sex: M

Passing by reference

Parameters to functions are passed by reference. Some languages pass copies of the objects to functions. Passing objects by reference has two important conclusions. The process is faster than if copies of objects were passed. Mutable objects that are modified in functions are permanently changed.

#!/usr/bin/python

n = [1, 2, 3, 4, 5]

print "Original list:", n

def f(x):
  x.pop()
  x.pop()
  x.insert(0, 0)
  print "Inside f():", x  
  
f(n)  

print "After function call:", n

In our example, we pass a list of integers to a function. The object is modified inside the body of the function. After calling the function, the original object, the list of integers is modified.

def f(x):
  x.pop()
  x.pop()
  x.insert(0, 0)
  print "Inside f():", x  

In the body of the function we work with the original object. Not with a copy of the object. In many programming languages we woud receive a copy of an object by default.

$ ./byreference.py
Original list: [1, 2, 3, 4, 5]
Inside f(): [0, 1, 2, 3]
After function call: [0, 1, 2, 3]

Once the list was modified it was modified for good.

Global and local variables

Next we will talk about how variables are used in Python functions.

#!/usr/bin/python

name = "Jack"

def f():
   name = "Robert"
   print "Within function", name

print "Outside function", name
f()

A variable defined in a function body has a local scope. It is valid only within the body of the function.

$ ./local.py 
Outside function Jack
Within function Robert

Output.

#!/usr/bin/python

name = "Jack"

def f():
   print "Within function", name

print "Outside function", name
f()

By default, we can get the contents of a global variable inside the body of a function. But if we want to change a global variable in a function, we must use the global keyword.

$ ./global.py 
Outside function Jack
Within function Jack

#!/usr/bin/python

name = "Jack"

def f():
   global name 
   name = "Robert"
   print "Within function", name


print "Outside function", name
f()
print "Outside function", name

Now, we will change the contents of a global name variable inside a function.

global name 
name = "Robert"

Using the global keyword, we reference the variable defined outside the body of the function. The variable is given a new value.

$ ./global2.py 
Outside function Jack
Within function Robert
Outside function Robert

Anonymous functions

It is possible to create anonymous functions in Python. Anonymous functions do not have a name. With the lambda keyword, little anonymous functions can be created. Anonymous functions are also called lambda functions by Python programmers. They are part of the functional paradigm incorporated in Python.

Lambda functions are restricted to a single expression. They can be used wherever normal functions can be used.

#!/usr/bin/python

y = 6

z = lambda x: x * y
print z(8)

This is a small example of the lambda function.

z = lambda x: x * y

The lambda keyword creates an anonymous function. The x is a parameter that is passed to the lambda function. The parameter is followed by a colon character. The code next to the colon is the expression that is executed, when the lambda function is called. The lambda function is assigned to the z variable.

print z(8)

The lambda function is executed. The number 8 is passed to the anonymous function and it returns 48 as the result. Note that z is not a name for this function. It is only a variable to which the anonymous function was assigned.

$ ./lambda.py
48

Output of the example.

The lambda function can be used elegantly with other functional parts of the Python language, like map() or filter() functions.

#!/usr/bin/python

cs = [-10, 0, 15, 30, 40]

ft = map(lambda t: (9.0/5)*t + 32, cs)
print ft

In the example we have a list of celsius temperatures. We create a new list containing temperatures in fahrenheit.

ft = map(lambda t: (9.0/5)*t + 32, cs)

The map() function applies the anonymous function to each element of the cs list. It creates a new ft list containing the computed Fahrenheit temperatures.

$ ./lambda2.py
[14.0, 32.0, 59.0, 86.0, 104.0]

Example output.

This chapter was about functions in Python.