Using pandas.DataFrame.loc with a tuple[str, str] multi-index seems to elicit mypy errors

[goxberry]

Describe the bug

Using pandas.DataFrame.loc with a multi-index -- expressed as a tuple[str, str], per the syntax described in https://pandas.pydata.org/docs/user_guide/advanced.html#advanced-indexing-with-hierarchical-index -- seems to elicit mypy errors.

To Reproduce

import itertools

import pandas


def main() -> None:
    rows = pandas.Index(["project A", "project B", "project C"])
    years: tuple[str, ...] = ("Year 1", "Year 2", "Year 3")
    quarters: tuple[str, ...] = ("Q1", "Q2", "Q3", "Q4")
    index_tuples: list[tuple[str, ...]] = list(itertools.product(years, quarters))
    cols = pandas.MultiIndex.from_tuples(index_tuples)
    budget = pandas.DataFrame(index=rows, columns=cols)
    multi_index: tuple[str, str] = ("Year 1", "Q1")
    budget.loc["project A", multi_index] = 4700


if __name__ == "__main__":
    main()

Running mypy version 0.991 on that source code yields the following error:

Invalid index type "Tuple[str, Tuple[str, str]]" for "_LocIndexerFrame"; expected type "Union[Series[bool], ndarray[Any, dtype[bool_]], List[bool], str, Tuple[Union[Index, Union[Series[bool], ndarray[Any, dtype[bool_]], List[bool]], Union[str, bytes, date, datetime, timedelta, bool, int, float, complex, Timestamp, Timedelta], List[Any], slice], ...], List[<nothing>]]"  [index]

Please complete the following information:

• Linux
• Ubuntu 22.04.1
• 3.9.9
• mypy 0.991
• pandas-stubs 1.5.2.221124

Additional context

I tried changing

pandas-stubs/pandas-stubs/core/frame.pyi

Line 161 in 58416f2

list[HashableT] | slice | Series[bool] | Callable,

by appending an ellipsis, but it had no effect. I suspect part of the issue I saw no effect is that I neglected to modify the overloads of _LocIndexerFrame.__getitem__ as well, so I may try that, and if it works, I'll submit a PR.

[Dr-Irv]

Thanks for the report. There are 2 changes to make here. One is support __getitem__(), the other is to support __setitem__:
In

pandas-stubs/pandas-stubs/core/frame.pyi

Line 167 in 58416f2

idx: tuple[int | StrLike | tuple[ScalarT, ...], int | StrLike],

add tuple[ScalarT, ...] to the second part of the main tuple in the union.

In

pandas-stubs/pandas-stubs/core/indexing.pyi

Line 18 in 58416f2

Union[Index, MaskType, Scalar, list[ScalarT], slice], ...

add tuple[ScalarT, ...] after slice in the Union

Then create your test, testing both x = budget.loc["project A", multi_index] and the line you showed.