Problem using `terms.train-vectors` with patterns library

1

It seems like train-vectors recipe internally calls from pattern.text.en import wordnet. Problem is that if NLTK’s wordnet_ic, sentiwordnet are not downloaded (which is the case with the initial installation), the module emits the following error without giving useful information.

...
BadZipFile: File is not a zip file

I just figured out and solved the issue, and want to report this to request for more helpful error message (e.g. “Please download the following NLTK dataset …”) or make the dataset download happen automatically.

2

Hmm, I’ve never hit that error! We call into Gensim, which if I understand correctly prints a runtime warning if that pattern library is not installed. But pattern isn’t used for anything we’re doing, so the runtime warning should be safe to ignore.