Fix #849: Update converters type in read_excel for better Pyright compatibility

[Sachi-Parashar]

Summary

This PR addresses issue #849, where pyright fails to type-check the converters argument in read_excel when using functools.partial or lambda functions.

Fix

Updated all read_excel and parse overloads in pandas/io/excel/_base.pyi:

- converters: Mapping[int | str, Callable[[object], object]] | None
+ converters: Mapping[int | str, Callable[[Any], Any]] | None

This relaxes the type signature to accept functools.partial and similar callables.

Test

Added a test tests/test_excel_converters.py with assert_type to confirm the fix works with pyright and mypy.

from functools import partial
import pandas as pd
from typing_extensions import assert_type

conv = partial(pd.to_datetime, errors="coerce")
df = pd.read_excel("foo.xlsx", converters={"col1": conv})
assert_type(df, pd.DataFrame)

Notes

• Verified with both mypy and pyright. - Worked✅
• If needed, I'm open to updating ConvertersArg in _typing.pyi for broader consistency in a follow-up PR.

Closes #849

[Dr-Irv]

This should go in test_io.py and be in a test_converters_partial() function

[Sachi-Parashar]

Thanks for the suggestion @Dr-Irv — I’ve moved the test to test_io.py as test_converters_partial() and deleted the separate test file.

[Dr-Irv]

move the first import to top of file.
the second 2 imports are already there at the top, so remove from here.

[Dr-Irv]

Use check(assert_type(df, pd.DataFrame), pd.DataFrame)

[Dr-Irv]

This will need to work at runtime. If you look at test_read_excel(), you will see the use of ensure_clean() and that the test needs to first write out a file, and then read back the file. So you need to change to use that pattern.