GH-45175: [Python] Honor the strings_to_categorical keyword in to_pandas for string view type

[jorisvandenbossche]

Rationale for this change

Currently this keyword works for string or large string:

>>> table = pa.table({"col": pa.array(["a", "b", "a"], pa.string())})
>>> table.to_pandas(strings_to_categorical=True).dtypes
col    category
dtype: object
>>> table = pa.table({"col": pa.array(["a", "b", "a"], pa.large_string())})
>>> table.to_pandas(strings_to_categorical=True).dtypes
col    category
dtype: object

but not for string view:

>>> table = pa.table({"col": pa.array(["a", "b", "a"], pa.string_view())})
>>> table.to_pandas(strings_to_categorical=True).dtypes
col    object
dtype: object

For consistency we should make that keyword check for string view columns as well, I think

From https://github.com/apache/arrow/pull/44195/files#r1901831460

Are these changes tested?

Yes

Are there any user-facing changes?

Yes, when using the strings_to_categorical=True keyword and having a string_view type, this column will now be converted to a pandas Categorical

• GitHub Issue: [Python] Honor the strings_to_categorical keyword in to_pandas for string view type #45175

[github-actions]

⚠️ GitHub issue #45175 has been automatically assigned in GitHub to PR creator.

[raulcd]

Thanks @jorisvandenbossche
I've learnt today that Table.__getitem__ returns a ChunkedArray.
I am going to merge this. Where you expecting this to go on 19.0.0? cc @amoeba

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 2c5ae51.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 1 possible false positive for unstable benchmarks that are known to sometimes produce them.