Unit-test/Travis CI: Cleaner Log, Features; Closes #137

[henrifroese]

No description provided.

[henrifroese]

from #137 :

Also, sometimes the Travis CI does not fail even if some errors are raised, see example below at line 576 for instance (`raise IOError(Errors.E050.format(name=name))). This should also investigate further. In this case, we may want the job to fail rather than pass?

We solved this by loading the spacy en_core_web_sm model in a different way. The log looks much nicer now, no more IOErrors etc. 🪂

[henrifroese]

The only things that are still annoying in the log are both warnings when testing (we were able to fix all other warnings etc.)

1. warning when importing plotly.express in visualization; we could filter this with with warnings.filter.... but that's not really nice when importing - might make sense though
2. Warnings from Gtk.Widget...; probably somewhere from visualization.py, but we were not able to find these 😠 😡 👿

[jbesomi]

Changes in .travis.yml are due to a previous PR?

[henrifroese]

Changes in .travis.yml are due to a previous PR?

Yes I think so. Will test some more stuff and incorporate the changes from above soon.

[henrifroese]

Now ready to merge from my side 🥈

[jbesomi]

Amazing, it looks much more beautiful yet!

There is still a warning:

There are NaNs (missing values) in the given input series. They were replaced with appropriate values before the function was applied. Consider using hero.fillna to replace those NaNs yourself or hero.drop_no_content to remove them.

Do you know where it came from? Can we remove it?

Also, I noticed how we don't have a clear position regarding the CR/CRLF issue (you changed the kind of ending newline in the first two files). I will take care of this by bulk-formating all files and will update the CONTRIBUTION.md accordingly.

[henrifroese]

There is still a warning:

Aaaah, now it all makes sense: that's why we were skipping the doctest in _helper.py with the replace_b_with_c. That makes sense as a doctest with warnings.filterwarnings is really ugly 🤢 and we're testing the function in test_helpers.py anyways. So I'll skip this doctest again and we're good 🌛

Also, I noticed how we don't have a clear position regarding the CR/CRLF issue (you changed the kind of ending newline in the first two files). I will take care of this by bulk-formating all files and will update the CONTRIBUTION.md accordingly.

Yes, makes sense 🙆‍♂️

[jbesomi]

Thanks, merged! 🎉 (had to fix some conflicts, but should be good!)