TST: refactored Excel reader tests

[davidovitch]

This PR implements more flexible Excel (spreadsheet) reader tests. The work was originally part of the larger but unsuccessful PR #9070.

This PR includes:

• Excel reader base class tests to facilitate adding the similar tests for other readers
• not all Excel file formats (xls, xlsm, xlsx) had a corresponding test file. This PR adds the missing ones. Consequently, all Excel data files are now tested against all 3 Excel file formats.
• rename test to test1 (for xls, xlsm, xlsx extensions) for consistency with test2..4 test data files. This also means that before merging (but after Travis has ran the tests) the URL of Excel test file test1.* has to be changed from davidovitch/pandas to pydata/pandas in pandas/io/tests/test_excel.py at line 378.

[jreback]

can you report the number of tests on current master (for the excel stuff) and number on this branch. Trying to figure out if anything is left out.

[davidovitch]

With this PR I have the following:

# pandas/io/tests/test_excel.py
Ran 283 tests in 9.364s
OK (SKIP=10)

and with 9bdae60 (same as above minus the 2 commits of this PR) I get:

# pandas/io/tests/test_excel.py
Ran 242 tests in 6.969s
OK (SKIP=10)

I have 10 skipped tests with openpyxl=1.8.6 being indicated as incompatible (the same for both this PR and 9bdae60). With openpyxl=2.2.4 I get failures that I haven't look into yet, but I also get these failures with 9bdae60. These failures are most likely due to my system since Travis is not complaining about them, but I will have to investigate this further to figure out what exactly is going on.

[jreback]

@chris-b1 can you have a look here

[chris-b1]

Can this be more clear it returns 1 or 2 items? Part of me thinks it should just be two functions.

[chris-b1]

Added a couple comments, overall approach seems pretty reasonable.

[davidovitch]

@chris-b1 does this sufficiently address your comments?

[jreback]

@davidovitch pls squash.

[jreback]

ok, going to merge this, @chris-b1 you will prob need to rebase your excel pr

[jreback]

merged via 98b3a5d

thanks!

[davidovitch]

@jreback: Sorry, was too late for squashing and correcting one small thing: the URL of the renamed test file was still pointing my repo instead of pydata/pandas. See PR #11009.

[chris-b1]

I fixed this in my other PR when I rebased.
On Sep 5, 2015 3:14 PM, "David Verelst" [email protected] wrote:

@jreback https://github.com/jreback: Sorry, was too late for squashing
and correcting one small thing: the URL of the renamed test file was still
pointing my repo instead of pydata/pandas. See PR #11009
#11009.

—
Reply to this email directly or view it on GitHub
#10964 (comment).

[davidovitch]

ah ok, great! No need for PR #11009, closed now.