Python：测试函数/方法签名是否已更改的正确方法

Question

I am currently developing an application ( my_app ) that uses logic from another python package ( internal_package ), that is being developed in parallel.

Lets say I use a function func1 from internal_package in my code. In internal_package , it is defined like this:

def func1(a, b):
    # do something
    ...

However, If the function signature of func1 (ie the parameters with which the function is called) changes, lets say to

def func1(a, b, c):
    # do something
    ...

My code would of course raise an Exception, since I did not pass parameter c .

To become aware of that immediately, I want to write a test in pytest. In my understanding, this kind of test is not a unit test, but an integration test.

My approach would be something like this:

import inspect
import internal_package

def test_func1_signature():
    expected_signature = ?  # not sure how to correctly provide the expected signature here
    actual_signature = inspect.signature(internal_package.func1)
    
    assert actual_signature == expected_signature

I would expect this to be a common issue, but I was not able to find anything related. What would be the most pythonic way of performing this sort of test?

Answer 1

In case anyone is interested, I ended up using getfullargspec from the inbuilt inspect library. In the documentation it says:

Get the names and default values of a Python function's parameters. A named tuple is returned: FullArgSpec(args, varargs, varkw, defaults, kwonlyargs, kwonlydefaults, annotations)

As I am interested in in the args part, I retrieve them from func1 as follows:

inspect.getfullargspec(func1)[0]

Which then returns ['a', 'b'] . I can then write a simple test in pytest like this:

import pytest
import inspect

def test_func1_signature:
    expected_signature = ['a', 'b']  # I specify what i expect

    actual_signature = inspect.getfullargspec(func1)[0]

    assert actual_signature == expected_signature 

If anyone knows a better/more pythonic/more correct way, please let me know.