incorrect behavior of spacy.tests.util.assert_docs_equal

[tamuhey]

import spacy
from spacy.tests.util import assert_docs_equal

nlp = spacy.load("en_core_web_sm")
text = "This is a sentence"
doc1 = nlp(text)
doc2 = nlp(text)
assert_docs_equal(doc1, doc2) # raise error

spacy version: master

[adrianeboyd]

Thanks for the report! Span comparisons were modified in #5005, so now spans from different documents are never equal. I think this test could be updated to check the Token.is_sent_start values instead.

[tamuhey]

Or simply compare sent.text instead of sent?

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.