def test_patdict_op(self, doc):
matches = list(extract.token_matches(doc, [{
"POS": "NOUN",
"OP": "+"
}]))[:5]
assert matches
assert all(len(span) >= 1 for span in matches)
assert all(tok.pos_ == "NOUN" for span in matches for tok in span)
def test_patdict_int(self, doc):
matches = list(extract.token_matches(doc, [{"LENGTH": 5}]))[:5]
assert matches
assert all(len(span[0]) == 5 for span in matches)
def test_patdict_bool(self, doc):
matches = list(extract.token_matches(doc, [{"IS_PUNCT": True}]))[:5]
assert matches
assert all(span[0].is_punct is True for span in matches)
def test_patdict(self, doc):
matches = list(extract.token_matches(doc, [{"POS": "NOUN"}]))[:5]
assert matches
assert all(len(span) == 1 for span in matches)
assert all(span[0].pos_ == "NOUN" for span in matches)
def test_patstr_int(self, doc):
matches = list(extract.token_matches(doc, "LENGTH:int(5)"))[:5]
assert matches
assert all(len(span[0]) == 5 for span in matches)
def test_patstr_bool(self, doc):
matches = list(extract.token_matches(doc, "IS_PUNCT:bool(True)"))[:5]
assert matches
assert all(span[0].is_punct is True for span in matches)
def test_patstr_op(self, doc):
matches = list(extract.token_matches(doc, "POS:NOUN:+"))[:5]
assert matches
assert all(len(span) >= 1 for span in matches)
assert all(tok.pos_ == "NOUN" for span in matches for tok in span)
def test_patstr(self, doc):
matches = list(extract.token_matches(doc, "POS:NOUN"))[:5]
assert matches
assert all(len(span) == 1 for span in matches)
assert all(span[0].pos_ == "NOUN" for span in matches)
def test_pattern_types(self, doc, patterns):
matches = list(extract.token_matches(doc, patterns))[:5]
assert matches
assert all(isinstance(span, Span) for span in matches)