Skip to content

3. Functions

In python, all functions are actually just objects. Thus as a result, they can be passed as parameters just as any other type, stor__ed in other objects/data structures, or assinged to variables. This makes using functions in Python extremely versetile.

Functions are made callable as they define the special __call__ method. Any object that implements this method will become callable itself.

*args and **kwargs

The * operator in python is often overlooked, but can be extremely useful. It unpacks an iterable and essentially replaces it with a comma seperated list of its content inline. This can be seen in the following example.

>>> int.__eq__(*[1, 2])
False
int.__eq__ is the implementation of the binary equality function between integer objects, and thus accepts two integers. The * unpacks the arguments so the line is equivalent to int.__eq__(1, 2).

*args allows you to pass a variable number of arguments to a python function. For example, let's write an add function that accepts any number of objects to add. 

>>> def add(*args):
...     total = 0
...     for arg in args:
...         total += arg
...     return total 
>>> add(1, 4, -1, 7)
11
**kwargs unpacks keyword arguments. It is given in the form of a dictionary of string keys mapping from the name of the keyword argument to it's value. Here's an example to demonstrate this. 
>>> def print_kwargs(**kwargs):
...     for arg_name, arg_val in kwargs.items():
...         print(arg_name, arg_val)
... 
>>> print_kwargs(a=10, b=2)
a 10
b 2

If you are using *args and **kwargs, required parameters must be specified first followed by args then kwargs: def fn(req_arg1, req_arg2, *args, **kwargs)

You can use unpacking to call any function as well. Let's say we have a function with the following header: def fn(a1, a2, kw1=0, kw2=True, kw3=[]). We could define the variables args = ("a1", "a2") and kwargs = {"kw1" : 0}, then call fn(*args, **kwargs).

Return

By default, Python functions that don't explicitly return a value will return the None object.

Additionally, when returning multiple values, they are implicitly wrapped in a tuple that you can use as a tuple, or implicitly unpack. 

>>> def fn(): return 0, 1
... 
>>> fn()
(0, 1)
>>> a, b = fn()
>>> print(a, b)
0 1

Typing

Python 3.5 adding the typing module that allows for type hinting within functions. You can read the full documentation here. The typing module will perform checks on function arguments are return types to verify that they are behaving as desired. High quality code can leverage typing to improve readability and correctness.

Within function signatures, a colon : following an argument indicuates its expected type and a an arrow -> after the parenthetical indicates the expected return type. 

def append_and_return(items: list, item : int) -> list:
    items.append(item)
    return items

def get(map: dict, key: str) -> int:
    return map[key]

Type Aliases

The typing module includes support for aliases that can be leveraged for more complex type verification. Aliases for all of the common python data types already exist inclduing List, Dict, Tuple, and Set. Objects from the typing module have a bracket operator that allows you to indicate what type of items an object might hold. 

from typing import List, Dict, Tuple, Set
def dot(vec_1: List[float], vec_2: List[float]) -> float:
    # implementation
We can also rewrite the same signature using a type alias. We create a new typing object that further defines the signature. 
Vector = List[float]
def dot(vec_1: Vector, vec_2: Vector) -> float:
    # implementation
Both of the above examples are identical. Here's a more complex example taken from the Python documentation linked earlier. 
ConnectionOptions = Dict[str, str]
Address = Tuple[str, int]
Server = Tuple[Address, ConnectionOptions]

def broadcast_message(message: str, servers: Sequence[Server]) -> None:

Decorators

Function Caching

Lambdas