def test_matcher_warns_if_unknown_pattern_elements(nlp: Language) -> None:
"""Calling the matcher on a `Doc` object with no matches returns empty list."""
matcher = TokenMatcher(nlp.vocab)
matcher.add("TEST", [[{"TEXT": {"fuzzy": "test"}}]])
doc = nlp("test")
with pytest.warns(UserWarning):
matcher(doc)
def test_matcher_pipe_with_context(nlp: Language) -> None:
"""It returns a stream of Doc objects as tuples with context."""
doc_stream = (
(nlp("test doc 1: Corvold"), "Jund"),
(nlp("test doc 2: Prosh"), "Jund"),
)
matcher = TokenMatcher(nlp.vocab)
output = matcher.pipe(doc_stream, as_tuples=True)
assert list(output) == list(doc_stream)
def test_matcher_pipe(nlp: Language) -> None:
"""It returns a stream of Doc objects."""
doc_stream = (
nlp("test doc 1: Corvold"),
nlp("test doc 2: Prosh"),
)
matcher = TokenMatcher(nlp.vocab)
output = matcher.pipe(doc_stream)
assert list(output) == list(doc_stream)
def test_matcher_pipe_with_matches(nlp: Language) -> None:
"""It returns a stream of Doc objects and matches as tuples."""
doc_stream = (
nlp("test doc 1: Corvold"),
nlp("test doc 2: Prosh"),
)
matcher = TokenMatcher(nlp.vocab)
matcher.add(
"DRAGON", [[{"TEXT": {"FUZZY": "Korvold"}}], [{"TEXT": {"FUZZY": "Prossh"}}]]
)
output = matcher.pipe(doc_stream, return_matches=True)
matches = [entry[1] for entry in output]
assert matches == [[("DRAGON", 4, 5, None)], [("DRAGON", 4, 5, None)]]
def matcher(model: Language) -> TokenMatcher:
"""It returns a token matcher."""
matcher = TokenMatcher(vocab=model.vocab)
matcher.add(
"DATA",
[
[
{"TEXT": "SQL"},
{"LOWER": {"FREGEX": "(database){s<=1}"}},
{"LOWER": {"FUZZY": "access"}, "POS": "NOUN"},
],
[{"TEXT": {"FUZZY": "Sequel"}}, {"LOWER": "db"}],
],
)
matcher.add("NAME", [[{"TEXT": {"FUZZY": "Garfield"}}]])
return matcher
def test_add_with_zero_len_pattern(matcher: TokenMatcher) -> None:
"""Trying to add zero-length patterns raises a ValueError."""
with pytest.raises(ValueError):
matcher.add("TEST", [[]])
def test_add_without_sequence_of_patterns_raises_error(matcher: TokenMatcher,) -> None:
"""Trying to add non-sequences of patterns raises a TypeError."""
with pytest.raises(TypeError):
matcher.add("TEST", [{"TEXT": "error"}]) # type: ignore
def test_matcher_returns_empty_list_if_no_matches(nlp: Language) -> None:
"""Calling the matcher on a `Doc` object with no matches returns empty list."""
matcher = TokenMatcher(nlp.vocab)
matcher.add("TEST", [[{"TEXT": {"FUZZY": "blah"}}]])
doc = nlp("No matches here.")
assert matcher(doc) == []
def test_remove_label_raises_error_if_label_not_in_matcher(
matcher: TokenMatcher,
) -> None:
"""It raises a ValueError if trying to remove a label not present."""
with pytest.raises(ValueError):
matcher.remove("TEST")
def test_remove_label(matcher: TokenMatcher) -> None:
"""It removes a label from the matcher."""
matcher.add("TEST", [[{"TEXT": "test"}]])
assert "TEST" in matcher
matcher.remove("TEST")
assert "TEST" not in matcher