The other day I spent far too much time trying to figure out how to mock a module-level function that was being used inside of my class's method in Python. My googling didn't lead to obvious answers, so I figured it'd be good to document here for future reference.

Imagine we have some module-level function like the following:

# file: project/some_module/functions.py

def fetch_thing():
    # query some database
    return data

And that we use it inside of a class within a different module:

# file: project/other_module/thing.py

from some_module.functions import fetch_thing

class Thing:
    def run(self):
        try:
            data = fetch_thing()
        except:
            self.fail_gracefully()

In this example, I want to test that a failure fetching from the db will fail gracefully, so I need to mock fetch_thing and have it raise an exception.

I kept trying to mock the function at its module path, like so:

from other_module.thing import Thing

thing = Thing()

with patch.object('some_module.functions.fetch_thing') as mocked:
    mocked.side_effect = Exception('mocked error')
    data = thing.run()

But this isn't right. It turns out that you need to mock/patch the function within the module it's being imported into.

from other_module.thing import Thing

thing = Thing()

with patch.object('other_module.fetch_thing') as mocked:
    mocked.side_effect = Exception('mocked error')
    data = thing.run()

Note the very subtle difference in the string path we are passing to patch.object. Because we are importing the function into other_module where our class uses it, that is what we need to mock.