GH984 Add overload for DataFrame.clip and update those for Series.clip

[loicdiridollou]

• Closes Create overloads for DataFrame.clip() #984
• Tests added: Please use assert_type() to assert the type of any return value

Turns out there is a nasty edge case when you pass everything as None for upper and lower and ask for inplace=True, it will return Self and not None.

[Dr-Irv]

can you add tests where either lower or upper is not specified?

I think that might lead to some changes in how you wrote the overloads.

[loicdiridollou]

Turns out no changes were needed but agreed that this time we check all the cases.

[Dr-Irv]

I'm a little confused here, as you closed the PR and then reopened it. Should I review?

[loicdiridollou]

Yes it is open for review, I have accidentally pressed the close button, my mistake.

…

On Wed, Apr 30, 2025 at 9:46 AM Irv Lustig ***@***.***> wrote: *Dr-Irv* left a comment (pandas-dev/pandas-stubs#1206) <#1206 (comment)> I'm a little confused here, as you closed the PR and then reopened it. Should I review? — Reply to this email directly, view it on GitHub <#1206 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMQNSQBKOZWMMGWLDPGFMV324DH4LAVCNFSM6AAAAAB4BXDTWOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDQNBSGA2DOOJRHA> . You are receiving this because you modified the open/close state.Message ID: ***@***.***>

[Dr-Irv]

I think you may also need similar overloads for clip in series.pyi and then similar tests that omit the lower or upper arguments.

[loicdiridollou]

Since clip for series does not use the axis when passing a list like argument to lower or upper the overloads are already there. Let me clean up the tests and add for those without upper or lower passed and we should be in a good place, the big work was on the data frame side.

…

On Wed, Apr 30, 2025 at 11:50 AM Irv Lustig ***@***.***> wrote: ***@***.**** requested changes on this pull request. I think you may also need similar overloads for clip in series.pyi and then similar tests that omit the lower or upper arguments. ------------------------------ In tests/test_frame.py <#1206 (comment)> : > + check(assert_type(df.clip(upper=None, axis="index"), pd.DataFrame), pd.DataFrame) + check(assert_type(df.clip(upper=15, axis="index"), pd.DataFrame), pd.DataFrame) + check(assert_type(df.clip(upper=None, axis="index"), pd.DataFrame), pd.DataFrame) lines 808 and 810 are the same ------------------------------ In tests/test_frame.py <#1206 (comment)> : > + assert_type(df.clip(lower=None, axis="index"), pd.DataFrame), + pd.DataFrame, + ) + check( + assert_type(df.clip(lower=5, axis="index"), pd.DataFrame), + pd.DataFrame, + ) + check( + assert_type(df.clip(lower=None, axis="index"), pd.DataFrame), + pd.DataFrame, + ) + check( + assert_type(df.clip(lower=pd.Series([1, 2]), axis="index"), pd.DataFrame), + pd.DataFrame, + ) + check( + assert_type(df.clip(lower=None, axis="index"), pd.DataFrame), + pd.DataFrame, + ) lines 863 and 871 and 879 are the same test ------------------------------ In tests/test_frame.py <#1206 (comment)> : > + assert_type(df.clip(lower=None, axis="index", inplace=True), pd.DataFrame), + pd.DataFrame, + ) + check( + assert_type(df.clip(lower=5, axis="index", inplace=True), None), + type(None), + ) + check( + assert_type(df.clip(lower=pd.Series([1, 2]), axis="index", inplace=True), None), + type(None), + ) + check( + assert_type(df.clip(lower=None, axis="index", inplace=True), pd.DataFrame), lines 847 and 859 are the same test ------------------------------ In tests/test_frame.py <#1206 (comment)> : > + assert_type(df.clip(lower=None, axis=0, inplace=True), pd.DataFrame), + pd.DataFrame, + ) + check(assert_type(df.clip(lower=5, axis=0, inplace=True), None), type(None)) + check( + assert_type(df.clip(lower=None, axis=0, inplace=True), pd.DataFrame), lines 883 and 888 are the same test ------------------------------ In tests/test_frame.py <#1206 (comment)> : > + assert_type(df.clip(lower=None, axis=None, inplace=True), pd.DataFrame), + pd.DataFrame, + ) + check( + assert_type(df.clip(lower=5, axis=None, inplace=True), None), + type(None), + ) + check( + assert_type(df.clip(lower=None, axis=None, inplace=True), pd.DataFrame), + pd.DataFrame, lines 828 and 836 are the same test — Reply to this email directly, view it on GitHub <#1206 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMQNSQB5TEGEXPERXRNI5T324DWOBAVCNFSM6AAAAAB4BXDTWOVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZDQMBXGUYTGNBUGM> . You are receiving this because you modified the open/close state.Message ID: ***@***.***>

[loicdiridollou]

Turns out the axis parameter does not need anything specific (to split the overloads) since it is not used (unless you pass something not allowed).
So the overload did not need anything additional yet I have added more tests to lock in the behavior.

[Dr-Irv]

thanks @loicdiridollou