Functional Programming using itertools and functools modules in Python

By Aditya Vilas Dorge

In this article we will take a look on two important modules of python’s standard library and various functions that helps dealing with iterables and higher-order function .

Itertools

Before talking about itertools module lets take a quick look at what are iterators.

What are iterators ?
Iterator is an objects which contains countable number of objects . The most common iterator in python is List. 

fruits=['apple','banana','orange','strawberry']

Coming back to itertools ,
Itertools is the module in python’s standard library that is used for efficiently looping over iterators.

Types of iterator :
1. Infinite iterator : continue to run indefinitely if no stopping condition is placed
2. Finite iterator : Terminates on shortest input sequence
3. Combinatoric iterator : These iterators deals with counting that is enumeration, permutation and combination.

Infinite Iterators

There are multiple ways to iterate over a iterator. For simplicity purposes, we will be working with list data structure.
Traditional way : Using a for loop

 

>> fruits = ['apple', 'banana' , 'orange' , 'strawberry' ] 
>> for fruit in fruits:
       print(fruit)
>> apple
   banana
   orange
   strawberry

(1) count( ) : Returns evenly spaced values starting with number start .
syntax : itertools.count(start,step)
• start : determines the starting (initial) value
• step : determines the increment in each step

>> import itertools
# find all the odd numbers from 0 to 10
>> numbers = itertools.count(start=1 , step=2) 
>> for number in numbers :
       if number < 10 :
          print(number ,  end=" ")
       else :
          break 
>> 1 3 5 7 9

Similar result can be achieved using a for loop :

>> for number in range(1 , 10 , 2):
       print( number , end=” “ )
>> 1 3 5 7 9 

(2) cycle( ) : Iterates over the iterable’s element one at a time indefinitely until a termination condition is not met .
syntax : itertools.cycle(iterable)

import itertools
>> alphabets = ['A' ,'B' ,'C' ,'D']
>> letters = itertools.cycle(alphabets) 
>> count = 0 
>> for letter in letters :
       if count < 10 :
          print(letter ,  end=" " )
          count += 1
       else :
          break 
>> A B C D A B C D A B

(3) repeat( ) : Returns object over and over again indefinitely unless the time keyword argument is specified .
syntax : itertools.repeat(object[,times])
• repeat takes an optional parameter times to specify number of times to repeat the object .

>> import itertools
# print 10 three times 
>> numbers = itertools.repeat(object=10 , times=3) 
>> for number in numbers :
       print(number ,  end=" " )
>> 10 10 10

 

>> import itertools
>> alphabets= itertools.repeat("Hello world !" , 3)
>> for alphabet in alphabets:
       print(alphabet)
>> Hello world !
   Hello world !
   Hello world !

Finite Iterators

(1) accumulate( ) : Returns the accumulated sum based on the function provided.
syntax : itertools.accumulate(iterable[,func,*,initial=None])
• if the initial keyword argument is provided then the accumulation leads with the initial value resulting in an extra element in output.

>> import itertools
>> import operator
>> foo = [ 1 , 2 , 3 , 4 , 5 ] 
>> numbers = itertools.accumulate( foo , operator.add)
>> for number in numbers :
       print(number , end=," ")
>> 1 3 6 10 15

How did it calculate ?
Initially it took the first value that is 1 and then ,
1 + 2 = 3
3 + 3 = 6
6 + 4 = 10
10 + 5 = 15
Roughly equivalent to (((((1+2)+3)+4)+5)+6)

With keyword argument initial :

>> import itertools
>> import operator
>> foo = [ 1 , 2 , 3 , 4 , 5 ] 
>> numbers = itertools.accumulate( foo , operator.add , initial = 100)
>> for number in numbers :
       print(number , end=" ")
>> 100 101 103 106 110 115

Note : 1 more element is added to the output which is the initial element 100. 

(2) chain( ) : Returns element from the first iterable to the last iterable until all the iterables are exhausted.
loosely speaking,
combines the elements of all iterable one by one into one iterable.
syntax : itertools.chain(*iterables)

>> import itertools
>> foo =['Coderspacket', 'is']
>> bar =['awesome']
>> results = list(itertools.chain(foo, bar))
>> for result in results :
  print(result , end =" ")
>> Coderspacket is awesome

(3) pairwise( ) : Introduced in python 3.10 version, returns successive overlapping pairs taken from the input iterable.
In other words , starts from the element which ended in the last iteration (successive overlapping).
syntax : itertools.pairwise(iterable)

>> import itertools
>> fruits=['apple','banana', 'orange']
>> result = itertools.pairwise(fruits)
>> for fruit in result:
       print(fruit)
>>
  ('apple', 'banana')
  ('banana', 'orange')       

Combinatoric iterators

(1) permutations( ) : Returns successive r length permutations of elements in an iterable .
loosely speaking, it returns an arrangement of values in a definite order of length r.

syntax : itertools.permutations( iterable , r = None )
• If r is not specified or is None then r defaults automatically to the length of the iterable.

>> import itertools
>> fruits=['apple','banana','orange']
>> result= itertools.permutations(fruits , 2)
>> for fruit in result:
       print(fruit)
>> ('apple', 'banana')
   ('apple', 'orange')
   ('banana', 'apple')
   ('banana', 'orange')
   ('orange', 'apple')
   ('orange', 'banana')

(2) Combinations( ) : The function selects the elements from iterable and returns their combination with r members.
syntax : itertools.combinations( iterable , r )

>> import itertools
>> fruits=['apple','banana','orange']
>> result= itertools.combinations(fruits , 2)
>> for fruit in result:
       print(fruit)

>> ('apple', 'banana')
   ('apple', 'orange')
   ('banana', 'orange')

Note : permutations refer to the arrangement of the object, whereas combinations refer to the way of picking up elements (selection) from a set of objects.

 

functools

(1) cache : Helps reduce the complexity of a function using an optimization technique called memoization.

What is memoization?
Memoization is done by storing the result in a cache and using it again when needed instead of computing it again and again.
syntax : @functools.cache(user functions)

 >> from functools import cache 
 >> @cache
 >> def factorial(n):
        return n * factorial(n-1) if n else 1

>> start = time.time()	
>> print(factorial(100))
>> end = time.time()	
>> print("Time taken for function is : ",end-start)
>> 93326215443944152681699238856266700490715968264381621468592963895217599993229915608941463976156518286253697920827223758251185210916864000000000000000000000000
Time taken for function is :  0.0016701221466064453

Without @cache :

>> from functools import cache 
 >> def factorial(n):
  return n * factorial(n-1) if n else 1

>> start = time.time()	
>> print(factorial(100))
>> end = time.time()	
>> print("Time taken for function is : ",end-start)
>> 93326215443944152681699238856266700490715968264381621468592963895217599993229915608941463976156518286253697920827223758251185210916864000000000000000000000000
Time taken for function is :  0.0005016326904296875     

You can see the difference in the execution time of a function with and without @cache

Note : cache decorator is smaller and faster than the lru_cache( ) which is with a size limit.

(2) reduce : Takes a function of two arguments and applies it cumulatively from left to right of the elements of the iterable. syntax : functools.reduce(function ,iterable[,initializer])

>> from functools
>> def sum(a,b):
       return a + b 
>> result = functools.reduce(sum,[1,2,3,4,5])
>> print(result)
>> 15

Note : itertools.accumulate() Vs functools.reduce()
• Accumulate( ) keeps the existing value whereas reduce( ) returns only single value .
• Simply speaking , reduce( ) returns the last element of the result of accumulate( ) function .

(3) singledispatch : Suppose you want a single function to perform different operations based on the argument provided. @singledispatch decorator modifies a function behavior based on the type of a single argument provided to it.
syntax : @functools.singledispatch

>> from functools import singledispatch
>> @singledispatch
def fruits(arg:str):
  print(f'Do you like {arg} ?')

>> @fruits.register
def _(arg: int):
    print(f'would you like to buy {arg} kg of fruits ? ') 

>> @fruits.register
    def _(arg: list):
 	print("your fruit basket has : ")
 	for fruit in arg:
            print(fruit)
 	
>> fruits(‘apple’)
>> fruits(10)
>> fruits(['apple','banana','cherry']) 

>> Do you like apple ?
>> would you like to buy 10 kg of fruits ?
>> your fruit basket has :
   apple
   banana
   cherry

There are many more functions and decorators that will help you implement functional programming.
Keep experimenting with you code and don’t forget to check the Python documentation for more details.