ENH: Add support for s3 in read_excel #11447

[parsleyt]

closes #11447

Updated excel.py to use pandas.io.common.get_filepath_or_buffer (handles s3 urls) instead of urlopen, only when _is_s3_url returns true. This was the minimally invasive change.

[jreback]

tests pls

[jreback]

pls add a test for this. you can post an .xls file and I will put it up on the pandas s3 bucket.

[jreback]

pls add a whatsnew note as well (enhancements)

[parsleyt]

Will do. What's the path to the s3 bucket?
Will use test1.xls and test1.xlsx for testing if you can put them in the bucket for me.

[jreback]

ok I uploaded those 2 files to the standard bucket https://s3.amazonaws.com/pandas-test/test1.xls and .xlsx. lmk if you cannot access them.

[parsleyt]

Can you go ahead and add test1.xlsm to the s3 bucket as well? Looks like the existing excel tests use all three extensions and I want to use the same base class.

[jreback]

ok, should be good to go.

[parsleyt]

ok, PR updated with test case and whatsnew note.

[jreback]

perfect. needs a rebase (prob a doc conflict), then ping when green.

[jreback]

also can you extend the doc-string for read_excel (similar to how read_csv is done) for the s3

[parsleyt]

Looks like the doc for read_excel already mentions s3 in the same way as read_csv. Does this work?
...Valid URL schemes include http, ftp, s3, and file....

[parsleyt]

Hmm, also looks like I broke the build. The unit test is failing when trying to import boto to handle the s3 url: ImportError: No module named 'boto'

Is there a config I need to update for the CI process to install boto?

[parsleyt]

Ahh, I see I can just implement a nose.SkipTest. Will add a _skip_if_no_boto()

[jreback]

yep - builds have different deps / versions of things to give code a workout

[parsleyt]

Good stuff! Ok, I think everything is good to go now. Thanks for walking me through my first contribution.

[jreback]

@parsleyt thanks for the contribution!