Infer compression if file extension is uppercase

[willbowditch]

Inferring compression fails for files with uppercase extensions (e.g. x.zip works but y.ZIP does not)

• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[WillAyd]

Do we do this elsewhere? Outside of Windows most file systems I think are case sensitive so while relatively harmless I also am not sure this is worth doing

[willbowditch]

I think the path is retained with the original case and couldn't see any uses beyond identifying the compression method. Ran into this issue on OS X, tested locally with read_csv and capitalised zip files

[gfyoung]

Do we do this elsewhere?

Not that I'm aware of.

Outside of Windows most file systems I think are case sensitive so while relatively harmless I also am not sure this is worth doing

I think it would be okay to add this, but only after checking if we are on Windows.

cc @jreback

[jreback]

i think this is ok to do everywhere.

[jreback]

can you add this PR number here (use the issue format). Also no need to reference this internal function at all.

[willbowditch]

@jreback changed in a3558fd

[WillAyd]

Does pathlib offer any abstractions here? I don’t think it’s worthwhile for pandas to implement its own rules for file name equivalency.

…

Sent from my iPhone

On Jul 9, 2020, at 3:02 PM, Jeff Reback ***@***.***> wrote:  @jreback requested changes on this pull request. i think this is ok to do everywhere. In doc/source/whatsnew/v1.1.0.rst: > @@ -1050,6 +1050,7 @@ I/O - Bug in :meth:`~HDFStore.create_table` now raises an error when `column` argument was not specified in `data_columns` on input (:issue:`28156`) - :meth:`read_json` now could read line-delimited json file from a file url while `lines` and `chunksize` are set. - Bug in :meth:`DataFrame.to_sql` when reading DataFrames with ``-np.inf`` entries with MySQL now has a more explicit ``ValueError`` (:issue:`34431`) +- Bug in :meth:`io.common.infer_compression` where capitalised files extensions were not decompressed by read_* functions. can you add this PR number here (use the issue format). also add 'on windows platforms' — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

[jreback]

this is fine; we already have our own rules for when to decompress - this is just expanding to capitalized which is not a big deal

[jreback]

thanks @willbowditch