Type-checking in Python

Type checking function arguments is a common sanity check for function arguments in Python.

What do you mean "type-checking function arguments"?

Let's say you have a function add that takes in two arguments a and b. There is an implicit understanding that a and b must be either integer (type int) or float (type float).

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

add(1, 2) gives the output 3 which is what we intended the function for.
But add([2], [3]) seems to be returning [2, 3], a concatenated list.
Also add((2,), (3,)) seems to be returning (2, 3), a concatenated tuple.

To operate only on int and float arguments, we can add type or isinstance checks. Check out this StackOverflow post for understanding the difference between type and isinstance checks.

valid_add_types = (int, float)
def add(a, b):
    if isinstance(a, valid_add_types) and isinstance(b, valid_add_types):
        return a + b
    return None

add(1, 2) gives the output 3 which is what we intended the function for.
add([2], [3]) and add((2,), (3,)) return None because they're are neither integers nor floats.

Pros:

• It can help you, the developer, understand what types of arguments a function needs at a glance
• Useful for existing and new developers
• Bug-fixes and feature additions are less painful

Cons:

• It is manageable for small codebases
• Manual type checking makes large codebases convoluted

In the course of discussing how to improve the Developer Experience (DX) of a Python developer, my mentor, Sriram, and I came with up with an idea of a type checking decorator for type checking function arguments.

Introducing typy_checker, an easy-to-use library for type-checking your functions without losing your sanity over it.

Here is how we could use typy_checker in our example.

from typy_checker import type_checker

valid_add_types = (int, float)
@type_checker(a=valid_add_types, b=valid_add_types)
def add(a, b):
    return a + b

Calling add([2], [3]) will result in the following error -

SystemError: add(): Invalid type for variable 'a'. Expected int / float but got list.

Using typy_checker results in a cleaner codebase, from my experience. I have used it in two codebases and they are running in production at the moment.