Functions are beautiful when they are small!

They are small when they do only one thing. A sort of single responsibility principle for functions, if you will.

One way to achieve this is by using decorators.

A decorator can be used to add some extra functionality to a function without modifying the actual function definition.

For example if we want to log the calling of a function, adding that functionality to a function definition violates the single responsibility principle. Besides, adding it to each and every function is ugly, and not very DRY.

Instead, we can define a decorator which does this and decorate all the functions with it.

Log calling of a function with a decorator

def log_call(func):

    def inner(*args, **kwargs):
        print "calling {}".format(func.__name__)
        if args:
            print "with arguments: {}".format(str(args))
        if kwargs:
            print "with keyword arguments: {}".format(str(kwargs))
        return func(*args, **kwargs)

    return inner

def add(a, b=5):
    return a + b

def sub(a, b):
    return a - b

def printer(string):
    print string

print add(2, 3)
print sub(2, 3)
printer("Hello world!")
print add(2, b=3)

 calling add
 with arguments: (2, 3)
 calling sub
 with arguments: (2, 3)
 calling printer
 with arguments: ('Hello world!',)
 Hello world!
 calling add
 with arguments: (2,)
 with keyword arguments: {'b': 3}

Calling any function decorated with log_call will log the call and execute the called function. This decorator might be useful while debugging your code.

How does it work?

Whenever a function is decorated with log_call we are baking a new function with the additional functionality and assigning the decorated function’s name to it.

When we call add, we are actually calling the decorated add.

def add(a, b=5)

is a syntactig sugar for:

add = log_call(add) 
print add(2, 3) # calling decorated 'add'