Fix E1102 / not-callable false positive for property that returns a lambda function conditionally

[tushar-deepsource]

Type of Changes

	Type
✓	🐛 Bug fix

Description

Update E1102 to only be raised when no inferred value is callable.

Resolves #5931

[coveralls]

Pull Request Test Coverage Report for Build 2081161263

• 6 of 6 (100.0%) changed or added relevant lines in 1 file are covered.
• 16 unchanged lines in 5 files lost coverage.
• Overall coverage decreased (-0.03%) to 94.273%

---

Files with Coverage Reduction	New Missed Lines	%
pylint/checkers/base/docstring_checker.py	1	96.67%
pylint/checkers/typecheck.py	1	95.2%
pylint/config/arguments_manager.py	1	97.5%
pylint/config/config_initialization.py	3	93.18%
pylint/config/init.py	10	70.77%

Totals
Change from base Build 2079086702:	-0.03%
Covered Lines:	15505
Relevant Lines:	16447

---

💛 - Coveralls

[srijan-paul]

zamn :^)

[jacobtylerwalls]

nice! one suggestion to loosen this even more.

[tushar-deepsource]

@jacobtylerwalls loosening it to decline any instance of Uninferable is a bit too strict.

Here's an example from the test case:

class PropertyTest(object):
    """ class """

    def __init__(self):
        self.attr = 4

    @property
    def test(self):
        """ Get the attribute """
        return self.attr

    @test.setter
    def test(self, value):
        """ Set the attribute """
        self.attr = value

PROP = PropertyTest()
PROP.test(40) # [not-callable]

This is the result of infer_call_result:

[<Const.int l.44 at 0x10eb29b10>, Uninferable]

Since we inferred one Const.int, I think that's enough evidence to raise this issue here.

[jacobtylerwalls]

Thanks for looking into it. Darn.

I still think long-term we could consider making that change, though. There's no restriction on what you can set with the property. You can set something that's callable, and then not-callable becomes sort of wrong. But no one is asking for that change right now :-)

[DanielNoord]

@tushar-deepsource Coming into this pretty late but I was taking a look at stale PRs and found #5593. Could you quickly check if this solves that issue?

[tushar-deepsource]

@DanielNoord This change only affects properties, so no it doesn't solve that issue.

[DanielNoord]

Only needs a changelog entry!

Didn't see this was only for properties. But that makes sense :)

[tushar-deepsource]

@DanielNoord hopefully this is correct?

[DanielNoord]

Almost 😄

But thanks @tushar-deepsource! Wonderful fix!

[jacobtylerwalls]

Are we moving the changelog entry to 2.13.5?