Stopwords do not load properly

[jenniferjiangkells]

The stopwords do not load in MedCAT - we've been passing in a set of stopwords from the MedCAT config and this set of stopwords is entirely ignored as MedCAT still uses the spacy defaults. This is because to load a set of stopwords in spacy, you need to initialise this in the Language.Defaults class BEFORE the spacy model is loaded (see explosion/spaCy#10942)

I believe all you need to do is change the order of initialisation in medcat/pipe.py to:

from spacy.language import Language

if config.preprocessing.stopwords is not None:
   Language.Defaults.stop_words = set(config.preprocessing.stopwords)
self._nlp = spacy.load(config.general.spacy_model, disable=config.general.spacy_disabled_components)

[mart-r]

Thank you for the report along with a proposed solution.

Would you mind submitting the proposed change along with some automated tests as a pull request?

[jenniferjiangkells]

@mart-r Done - I've referenced this issue in the PR

[mart-r]

Thanks for the PR. It's much appreciated!

[mart-r]

Addressed and merged in PR #383