bpo-32092: Consume self when autospeccing unbound instance methods

[claudiubelu]

bpo-32092: Fixes mock.patch autospec self / cls argument consumption issue

https://bugs.python.org/issue32092

#76273

Currently, when patching methods with autospec, their self arguments are not consumed, causing call asserts to fail (they
expect an instance reference as the first argument).

Example:

from unittest import mock

class Something(object):
    def foo(self, a):
        pass

patcher = mock.patch.object(Something, 'foo', autospec=True)
patcher.start()

s = Something()

s.foo("foo")

s.foo.assert_called_once_with("foo")

The assertion above will fail, expecting an instance as a first argument. This issue only occurs in this scenario, and not in any of the following scenarios:

• autospec=False
• if the autospecced method is a classmethod or a staticmethod
• if an instance is patched instead of a class
• if mock.create_autospec(Something) is used instead
• if mock.Mock(autospec=Something) is used instead (if the PR adding it gets accepted)

This issue affects all patch variants.

self is now consumed by the generated wrapper and excluded from call recording and signature checking, in both the sync and async paths.

wraps and side_effect still get self bound for the call if needed, since they need to proxy through to a potentially unbound callable (e.g.: wraps=getattr(cls, name).

Adds test coverage across instance methods, classmethods, staticmethods, sync and async, for both wraps and side_effect, including a recursive-wraps regression test.

[AraHaan]

@BCLAU mind editing the PR title to add bpo-32092: on the beginning to fix the CI-bot?

Also is bpo-32092 the replacement for bpo-30587?

[claudiubelu]

Hi @AraHaan , sure, will do.

And I agree. Yeah, I would have sent the fixes separately, but this depends on the other one, which I've sent separately: #1982

Once that one merges, I can simply rebase this one. Or, I could simply separate them, but they'll have a merge conflict.

[AraHaan]

Or, just merge the 2 and close out the older one?

[claudiubelu]

Hm, tbh, I'd like to keep it simple, since they're both addressing different issues.

Anyways, I guess I'll just separate this patch from the other. We'll deal with the merge conflict when it happens.

[claudiubelu]

Actually, nevermind, the unit tests are not passing without the 1st patch. :)

[AraHaan]

alright then.

[csabella]

@claudiubelu, please resolve the merge conflicts. Thank you!

[cjw296]

@claudiubelu - it looks like #1982 is with @voidspace and this PR depends on it? Would it be okay to close that one until #1982 is resolved?

[github-actions]

This PR is stale because it has been open for 30 days with no activity.

[github-actions]

This PR is stale because it has been open for 30 days with no activity.

[read-the-docs-community]

Documentation build overview

📚 cpython-previews | 🛠️ Build #33783563 | 📁 Comparing e5e25df against main (b8862ae)

  🔍 Preview build  

2 files changed

± library/unittest.mock-examples.html
± whatsnew/changelog.html