functools in Python

Posted by chunyang on October 17, 2019
TAGS: #python


functools like itertool is very useful, but we pay little attention to it. It contains useful tools in everyday life:

  • wraps
  • cmp_to_key
  • partial
  • total_ordering
  • reduce


Python’s decorator is very useful in many scenarios. But a decorator may break docstring, signature of a function. wraps is used to fix this problem. By default, it will fix:

  • __module__
  • __name__
  • __qualname__
  • __doc__
  • __annotations__
  • __dict__: update this dict

Its implementation uses: partial and functools.update_wrapper. Usually we use it here:

import functools

def decorator(func):
    def wrapper(*args, **kwargs):
        return func(*args, **kwargs)
    return wrapper


cmp_to_key is used to solve the problem of API evovling. sorted method supports two keyword arguments:

  • cmp
  • key

The suggested way is to use the key function. How to covert a compare function which receives two arguments into a key function? cmp_to_key comes to help.

from functools import cmp_to_key

a = [1,7,4,3]

def cmp(a, b):
    """Cmpare a and b"""
    return a < b

sorted(a, key=cmp_to_key(cmp))


Similar to C++’s std::bind function. It binds certain arguments of a function and return it.

import functools as ft
from operator import add

a = 2
add_4 = ft.partial(add, 4)


This decorator dynamically generates all the order related methods. At least one of the following methods is defined:

  • __lt__
  • __le__
  • __gt__
  • __ge__

And the other ordering methods will be generated using one of the above methods and the sequence is the priority that total_ordering chooses to implement other methods.

import functools

class Student(object):
    def __init__(self, height):
        self._height = height

    def __lt__(self, other):
        return self._height > other._height


Similar to python’s keyword: reduce

import functools as ft
from operator import add

a = list(range(10))
print(ft.reduce(add, a))
print(ft.reduce(add, a, 10))