Unit tests should not need the network #490
Description
Activity
I worked on this a bit and I found that only test_stripRawHTML requires an internet connection. If we remove that test, everything will still work fine and there will be no need for network connectivity.
Shouldn't we though keep the test to make sure we don't accidentally break something in the used functions?
If we are to keep it, I tried some different wikipedia pages to make the process faster. If instead of "https://en.wikipedia.org/wiki/Ethics" we use "https://en.wikipedia.org/wiki/Terminus_post_quem" (which is significantly shorter) we reduce the testing time by about 1 second on bad connections. I don't know if that one second is worth it, but we could try.
One way to deal with that is to mock the call to urllib.request.urlopen. I have done that in #495, if anyone wants to check it out and comment if that is a good approach, I will appreciate any feedback :)
Looks good, thanks!
test_nlp.py has a call to loadPageHTML -- I found out about it when running py.test seemed to hang up for minutes. (It eventually finished. I was downloading the Anaconda installer at the same time.)
Tests should try to use only local data -- maybe from aima-data? Or keep it cached after the first run. I'd rate this as a nice-to-have, not a big problem, currently.