def test_exclude_types(self, spacy_doc):
ent_types = ["PERSON", "GPE"]
for exclude_types in ent_types:
result = extract.entities(spacy_doc, exclude_types=exclude_types)
assert all(span.label_ != exclude_types for span in result)
ent_types = [{"PERSON", "GPE"}, ("DATE", "ORG"), ["LOC"]]
for exclude_types in ent_types:
result = extract.entities(spacy_doc, exclude_types=exclude_types)
assert all(span.label_ not in exclude_types for span in result)
# special numeric cases!
ent_types = ["NUMERIC", ("NUMERIC",), {"PERSON", "NUMERIC"}]
for exclude_types in ent_types:
exclude_types_parsed = extract._parse_ent_types(exclude_types, "exclude")
result = extract.entities(spacy_doc, exclude_types=exclude_types)
assert all(span.label_ not in exclude_types_parsed for span in result)
def test_parse_ent_types_bad_type(self):
for bad_type in [1, 3.1415, True, b"PERSON"]:
with pytest.raises(TypeError):
_ = extract._parse_ent_types(bad_type, "include")