False positive no-member for enum.Enum's _value2member_map_.

[jamesbraza]

Steps to reproduce

Here's a minimal repro:

"""Show that pylint doesn't see enum's _value2member_map_."""

from enum import Enum


class Foo(Enum):
    """Enum class whose values we care about."""

    BAR = 0
    BAZ = 1

    @classmethod
    def has_value(cls, val: int) -> bool:
        """Get if a value is in this enum.

        SEE: https://stackoverflow.com/a/43634746/11163122

        """
        return val in cls._value2member_map_  # Location of the false positive

Current behavior

Running pylint has the output:

has_value_enum.py:19:22: E1101: Class 'Foo' has no '_value2member_map_' member (no-member)

Expected behavior

As the _value2member_map_ attribute is present, I think pylint should not warn that it's not present.

pylint --version output

pylint 2.6.0
astroid 2.4.2
Python 3.8.6 (default, Oct 27 2020, 11:18:42)
[Clang 12.0.0 (clang-1200.0.32.27)]

[hippo91]

@jamesbraza thanks for the report. It is probably a lack in the astroid's brain_namedtuple_enum brain.

[mbyrnepr2]

Can be closed once https://github.com/PyCQA/astroid/milestone/58 is released.

[mbyrnepr2]

Can be closed once https://github.com/PyCQA/astroid/milestone/58 is released.

Closing!

[cdce8p]

Unfortunately, this wasn't actually included in astroid 2.12.11. I've pinned it for 2.12.13 now.

[jacobtylerwalls]

Unfortunately, this wasn't actually included in astroid 2.12.11.

@cdce8p can you help me understand what you're referring to? When I compare the tags for 2.11 and 2.10, I see this change included. I also downloaded the source for 2.11 from PyPI to be absolutely sure, and I see it there also.

We might need to undo part of pylint-dev/astroid#1863.

[cdce8p]

Unfortunately, this wasn't actually included in astroid 2.12.11.

@cdce8p can you help me understand what you're referring to? When I compare the tags for 2.11 and 2.10, I see this change included. I also downloaded the source for 2.11 from PyPI to be absolutely sure, and I see it there also.

You're right. Seems like I got confused with the different commits. Usually I open the commit from them merge message (in the PR). I.e. this one: pylint-dev/astroid@3d4dc50. Looking at the branches / tags it's included only shows main. However that's only because the backport is a different commit.

We might need to undo part of PyCQA/astroid#1863.

Only partially, but yes. Would you like to do it?

[jacobtylerwalls]

Sure thing! pylint-dev/astroid#1874