Table of Contents

Functional Approach
1. Function scope
2. Default Arguments
3. Keyword Arguments
4. Parameter Packing and Unpacking
5. Nested Functions and Scopes
6. map, reduce and filter functions
6.1. List Comprehensions

Functional Approach


Functions allow us to enclose a set of statements and call the function again and again instead of repeating the group of statements everytime. Functions also allow us to isolate a piece of code from all the other code and provides the convenience of not polluting the global variables.

Function in python is defined with the keyword def followed by the name of the function, in turn followed by a pair of parenthesis which encloses the list of parameters to the function. The definition line ends with a ':'. The definition line is followed by the body of the function intended by one block. The Function must return a value:

def factorial(n):
  fact = 1
  for i in range(2, n):
    fact *= i

  return fact

The code snippet above defines a function with the name factorial, takes the number for which the factorial must be computed, computes the factorial and returns the value.

A Function once defined can be used or called anywhere else in the program. We call a fucntion with its name followed by a pair of parenthesis which encloses the arguments to the function.

The value that function returns can be assigned to a variable. Let's call the above function and store the factorial in a variable:

fact5 = factorial(5)

The value of fact5 will now be 120, which is the factorial of 5. Note that we passed 5 as the argument to the function.

It may be necessary to document what the function does, for each of the function to help the person who reads our code to understand it better. In order to do this Python allows the first line of the function body to be a string. This string is called as Documentation String or docstring. docstrings prove to be very handy since there are number of tools which can pull out all the docstrings from Python functions and generate the documentation automatically from it. docstrings for functions can be written as follows:

def factorial(n):
  'Returns the factorial for the number n.'
  fact = 1
  for i in range(2, n):
    fact *= i

  return fact

An important point to note at this point is that, a function can return any Python value or a Python object, which also includes a Tuple. A Tuple is just a collection of values and those values themselves can be of any other valid Python datatypes, including Lists, Tuples, Dictionaries among other things. So effectively, if a function can return a tuple, it can return any number of values through a tuple

Let us write a small function to swap two values:

def swap(a, b):
  return b, a

c, d = swap(a, b)

1.Function scope

The variables used inside the function are confined to the function's scope and doesn't pollute the variables of the same name outside the scope of the function. Also the arguments passed to the function are passed by-value if it is of basic Python data type:

def cant_change(n):
  n = 10

n = 5
cant_change(n)

Upon running this code, what do you think would have happened to value of n which was assigned 5 before the function call? If you have already tried out that snippet on the interpreter you already know that the value of n is not changed. This is true of any immutable types of Python like Numbers, Strings and Tuples. But when you pass mutable objects like Lists and Dictionaries the values are manipulated even outside the function:

>>> def can_change(n):
...   n[1] = James
...

>>> name = ['Mr.', 'Steve', 'Gosling']
>>> can_change(name)
>>> name
['Mr.', 'James', 'Gosling']

If nothing is returned by the function explicitly, Python takes care to return None when the funnction is called.

2.Default Arguments

There may be situations where we need to allow the functions to take the arguments optionally. Python allows us to define function this way by providing a facility called Default Arguments. For example, we need to write a function that returns a list of fibonacci numbers. Since our function cannot generate an infinite list of fibonacci numbers, we need to specify the number of elements that the fibonacci sequence must contain. Suppose, additionally, we want to the function to return 10 numbers in the sequence if no option is specified we can define the function as follows:

def fib(n=10):
  fib_list = [0, 1]
  for i in range(n - 2):
    next = fib_list[-2] + fib_list[-1]
    fib_list.append(next)
  return fib_list

When we call this function, we can optionally specify the value for the parameter n, during the call as an argument. Calling with no argument and argument with n=5 returns the following fibonacci sequences:

fib()
[0, 1, 1, 2, 3, 5, 8, 13, 21, 34]
fib(5)
[0, 1, 1, 2, 3]

3.Keyword Arguments

When a function takes a large number of arguments, it may be difficult to remember the order of the parameters in the function definition or it may be necessary to pass values to only certain parameters since others take the default value. In either of these cases, Python provides the facility of passing arguments by specifying the name of the parameter as defined in the function definition. This is known as Keyword Arguments.

In a function call, Keyword arguments can be used for each argument, in the following fashion:

argument_name=argument_value
Also denoted as: keyword=argument

def wish(name='World', greetings='Hello'):
  print "%s, %s!" % (greetings, name)

This function can be called in one of the following ways. It is important to note that no restriction is imposed in the order in which Keyword arguments can be specified. Also note, that we have combined Keyword arguments with Default arguments in this example, however it is not necessary:

wish(name='Guido', greetings='Hey')
wish(greetings='Hey', name='Guido')

Calling functions by specifying arguments in the order of parameters specified in the function definition is called as Positional arguments, as opposed to Keyword arguments. It is possible to use both Positional arguments and Keyword arguments in a single function call. But Python doesn't allow us to bungle up both of them. The arguments to the function, in the call, must always start with Positional arguments which is in turn followed by Keyword arguments:

def my_func(x, y, z, u, v, w):
  # initialize variables.
  ...
  # do some stuff
  ...
  # return the value

It is valid to call the above functions in the following ways:

my_func(10, 20, 30, u=1.0, v=2.0, w=3.0)
my_func(10, 20, 30, 1.0, 2.0, w=3.0)
my_func(10, 20, z=30, u=1.0, v=2.0, w=3.0)
my_func(x=10, y=20, z=30, u=1.0, v=2.0, w=3.0)

Following lists some of the invalid calls:

my_func(10, 20, z=30, 1.0, 2.0, 3.0)
my_func(x=10, 20, z=30, 1.0, 2.0, 3.0)
my_func(x=10, y=20, z=30, u=1.0, v=2.0, 3.0)

4.Parameter Packing and Unpacking

The positional arguments passed to a function can be collected in a tuple parameter and keyword arguments can be collected in a dictionary. Since keyword arguments must always be the last set of arguments passed to a function, the keyword dictionary parameter must be the last parameter. The function definition must include a list explicit parameters, followed by tuple paramter collecting parameter, whose name is preceded by a *, for collecting positional parameters, in turn followed by the dictionary collecting parameter, whose name is preceded by a **

def print_report(title, *args, **name):
  """Structure of *args*
  (age, email-id)
  Structure of *name*
  {
      'first': First Name
      'middle': Middle Name
      'last': Last Name
  }
  """

  print "Title: %s" % (title)
  print "Full name: %(first)s %(middle)s %(last)s" % name
  print "Age: %d nEmail-ID: %s" % args

The above function can be called as. Note, the order of keyword parameters can be interchanged:

>>> print_report('Employee Report', 29, 'johny@example.com', first='Johny',
                 last='Charles', middle='Douglas')
Title: Employee Report
Full name: Johny Douglas Charles
Age: 29
Email-ID: johny@example.com

The reverse of this can also be achieved by using a very identical syntax while calling the function. A tuple or a dictionary can be passed as arguments in place of a list of Positional arguments or Keyword arguments respectively using * or **

def print_report(title, age, email, first, middle, last):
  print "Title: %s" % (title)
  print "Full name: %s %s %s" % (first, middle, last)
  print "Age: %d nEmail-ID: %s" % (age, email)

>>> args = (29, 'johny@example.com')
>>> name = {
        'first': 'Johny',
        'middle': 'Charles',
        'last': 'Douglas'
        }
>>> print_report('Employee Report', *args, **name)
Title: Employee Report
Full name: Johny Charles Douglas
Age: 29
Email-ID: johny@example.com

5.Nested Functions and Scopes

Python allows nesting one function inside another. This style of programming turns out to be extremely flexible and powerful features when we use Python decorators. We will not talk about decorators is beyond the scope of this course. If you are interested in knowing more about decorator programming in Python you are suggested to read:


http://avinashv.net/2008/04/python-decorators-syntactic-sugar/

http://personalpages.tds.net/~kent37/kk/00001.html

However, the following is an example for nested functions in Python:

def outer():
  print "Outer..."
  def inner():
    print "Inner..."
  print "Outer..."
  inner()

>>> outer()

6.map, reduce and filter functions

Python provides several built-in functions for convenience. The map(), reduce() and filter() functions prove to be very useful with sequences like Lists.

The map (function, sequence) function takes two arguments: function and a sequence argument. The function argument must be the name of the function which in turn takes a single argument, the individual element of the sequence. The map function calls function(item), for each item in the sequence and returns a list of values, where each value is the value returned by each call to function(item). map() function allows to pass more than one sequence. In this case, the first argument, function must take as many arguments as the number of sequences passed. This function is called with each corresponding element in the each of the sequences, or None if one of the sequence is exhausted:

def square(x):
  return x*x

>>> map(square, [1, 2, 3, 4])
[1, 4, 9, 16]

def mul(x, y):
  return x*y

>>> map(mul, [1, 2, 3, 4], [6, 7, 8, 9])

The filter (function, sequence) function takes two arguments, similar to the map() function. The filter function calls function(item), for each item in the sequence and returns all the elements in the sequence for which function(item) returned True:

def even(x):
  if x % 2:
    return True
  else:
    return False

>>> filter(even, range(1, 10))
[1, 3, 5, 7, 9]

The reduce (function, sequence) function takes two arguments, similar to map function, however multiple sequences are not allowed. The reduce function calls function with first two consecutive elements in the sequence, obtains the result, calls function with the result and the subsequent element in the sequence and so on until the end of the list and returns the final result:

def mul(x, y):
  return x*y

>>> reduce(mul, [1, 2, 3, 4])
24

6.1.List Comprehensions

List Comprehension is a convenvience utility provided by Python. It is a syntatic sugar to create Lists. Using List Comprehensions one can create Lists from other type of sequential data structures or other Lists itself. The syntax of List Comprehensions consists of a square brackets to indicate the result is a List within which we include at least one for clause and multiple if clauses. It will be more clear with an example:

>>> num = [1, 2, 3]
>>> sq = [x*x for x in num]
>>> sq
[1, 4, 9]
>>> all_num = [1, 2, 3, 4, 5, 6, 7, 8, 9]
>>> even = [x for x in all_num if x%2 == 0]

The syntax used here is very clear from the way it is written. It can be translated into english as, "for each element x in the list all_num, if remainder of x divided by 2 is 0, add x to the list."