pandas MultiIndex.levels return type not correct

[Dr-Irv]

Environment data

• Language Server version: 2020.9.6
• OS and version: Windows 10
• Python version (& distribution if applicable, e.g. Anaconda): Anaconda Python 3.7.5

Expected behaviour

No error

Actual behaviour

from typing import cast
import pandas as pd

df = pd.DataFrame.from_records([[1, 2, 3], [4, 5, 6]], columns=["a", "b", "c"])
dfmi = df.set_index(["a", "b"])
print(type(dfmi.index))
ind = cast(pd.MultiIndex, dfmi.index)
print(ind.levels[1])

reports

Object of type "() -> Unknown" cannot be subscripted

Issue here is that MultiIndex.levels acts like a python property with the internal decorator of @cache_readonly, except you are not allowed to set it. Not sure what to do about that part, but below is a fix that works for the most common usage. It just would not report a typing error on a line such as ind.levels = [3, 4], which is not allowed.

Possible fix:

diff core/indexes/multi.pyi.ORIG core/indexes/multi.pyi
3c3,4
< from typing import Hashable, Sequence, Union
---
> from typing import Hashable, Sequence, Union, List
20c21,22
<     def levels(self): ...
---
>     @property  # is cache_readonly in pandas source
>     def levels(self) -> List[Index]: ...

[jakebailey]

This issue has been fixed in version 2020.9.7, which we've just released. You can find the changelog here: https://github.com/microsoft/pylance-release/blob/master/CHANGELOG.md#202097-30-september-2020