Please Improve Type Inference for DataFrame.loc[lambda df: ...]

[lvlh2]

Problem Description:

In VS Code with the Pylance extension, type inference works correctly for lambda parameters in methods like data.assign(rank=lambda df: df["n"].rank()), where df is properly recognized as a DataFrame. However, in data.loc[lambda df: df["n"].rank() > 5], the lambda parameter df is inferred as Unknown, which breaks autocomplete and type hints.

Works: df in assign() is correctly inferred as DataFrame

Fails: df in loc[] is inferred as Unknown

Why This Matters:

Using lambda inside .loc[] is a common pandas pattern for conditional indexing. Proper type inference would:

• Improve developer experience with autocomplete
• Enable static error detection
• Make code easier to maintain

Please enhance Pylance's type inference to recognize the lambda parameter in .loc[] as the same type as the calling DataFrame/Series.

[rchiodo]

This is likely caused by the type stubs that Panda ships (which we bundle with Pylance).
https://github.com/pandas-dev/pandas-stubs

It would be the one to return the type information for the df.loc. It looks to me like the _LocIndexerFrame in the frame.pyi isn't matching on an overload for the lambda.