What is the best practice for setting custom extensions and force

[SamEdwardes]

spaCy has some great documentation on creating custom extensions.

I am confused about how I should use the force parameter though in the context of building a custom pipeline component that I plan to turn into a python package.

Currently I do set force=True so I avoid the AttributeError (from spaCyTextBlob):

class SpacyTextBlob(object):
    """spaCy pipeline component for adding TextBlob sentiment analysis."""


    name = "text_blob_sentiment"


    def __init__(self):
        Token.set_extension("sentiment", getter=self.get_sentiment, force=True)
        Span.set_extension("sentiment", getter=self.get_sentiment, force=True)
        Doc.set_extension("sentiment", getter=self.get_sentiment, force=True)


    def __call__(self, doc):
        return doc


    def get_sentiment(self, tokens):
        blob = TextBlob(tokens.text)
        sentiment = blob.sentiment_assessments
        return sentiment

In reviewing a few other packages I see there are other patterns. What is the "best" pattern? Should I check if the extension is already set? Should I let the error occur? Should I force True?

From negspacy:

    def __init__(
        self,
        nlp: Language,
        name: str,
        neg_termset: dict,
        ent_types: list,
        extension_name: str,
        chunk_prefix: list,
    ):
        # if not termset_lang in LANGUAGES:
        #     raise KeyError(
        #         f"{termset_lang} not found in languages termset. "
        #         "Ensure this is a supported termset or specify "
        #         "your own termsets when initializing Negex."
        #     )
        # termsets = LANGUAGES[termset_lang]
        if not Span.has_extension(extension_name):
            Span.set_extension(extension_name, default=False, force=True)

From LemmInflect (they handle it in init.py):

if 'spacy' in sys.modules:
    min_version = '2.0'
    mv = min_version.split('.')
    sv = spacy.__version__.split('.')
    if sv[0] > mv[0] or (sv[0] == mv[0] and sv[1] >= mv[1]):
        spacy.tokens.Token.set_extension('lemma',   method=Lemmatizer().spacyGetLemma)
        spacy.tokens.Token.set_extension('inflect', method=Inflections().spacyGetInfl)
    else:
        logging.warning('Spacy extensions are disabled.  Spacy version is %s.  '
                        'A minimum of %s is required', spacy.__version__, min_version)

From my pending update for spaCyTextBlob that works with spacy 3. In this approach I just check if one of my extensions has been set. If not set then I set them all. The reason being is it seems like a lot of extra lines of code to check each one.

@Language.factory("spacytextblob")
class SpacyTextBlob(object):
    """A spacy pipeline object for sentiment analysis."""
    
    def __init__(self, nlp, name):
        # Register custom extensions
        if not Doc.has_extension("polarity"):            
            Doc.set_extension("polarity", default=None)
            Doc.set_extension("subjectivity", default=None)
            Doc.set_extension("assessments", default=None)
            Span.set_extension("polarity", getter=self.get_polarity)
            Span.set_extension("subjectivity", getter=self.get_subjectivity)
            Span.set_extension("assessments", getter=self.get_assessments)
            Token.set_extension("polarity", getter=self.get_polarity)
            Token.set_extension("subjectivity", getter=self.get_subjectivity)
            Token.set_extension("assessments", getter=self.get_assessments)

Thank you for your help!