def test_multi_match_will_not_match_if_regex_starts_ends_with_space(
searcher: RegexSearcher, nlp: Language) -> None:
"""Regexes that match on spaces will not produce match."""
doc = nlp(
"We want to identify US but will fail because regex includes whitespaces."
)
matches = searcher.match(doc, "\\s[Uu](nited|\\.?) ?[Ss](tates|\\.?)\\s")
assert matches == []
def test_multi_match_will_expand_on_partial_match_if_partials(
searcher: RegexSearcher, nlp: Language) -> None:
"""It extends partial matches to span boundaries."""
doc = nlp(
"We want to identify 'USA' even though only first two letters will matched."
)
matches = searcher.match(doc, "[Uu](nited|\\.?) ?[Ss](tates|\\.?)")
assert matches == [(5, 6)]
def test_multi_match_will_not_expand_if_not_partials(searcher: RegexSearcher,
nlp: Language) -> None:
"""It will not extend partial matches to span boundaries if not partial."""
doc = nlp(
"We want to identify 'USA' even though only first two letters will matched."
)
matches = searcher.match(doc,
"[Uu](nited|\\.?) ?[Ss](tates|\\.?)",
partial=False)
assert matches == []
def test_multi_match_raises_error_if_regex_str_not_str(searcher: RegexSearcher,
nlp: Language) -> None:
"""It raises a type error if regex_str is not a string."""
doc = nlp("My phone number is (555) 555-5555.")
with pytest.raises(TypeError):
searcher.match(doc, 1, predef=True)
def test_multi_match(searcher: RegexSearcher, nlp: Language) -> None:
"""It produces matches."""
doc = nlp("My phone number is (555) 555-5555, not (554) 554-5554.")
matches = searcher.match(doc, "phones", predef=True)
assert matches == [(4, 10), (12, 18)]