def test_extract_patterns(
lookup_tables: Dict[Text, List[Text]],
regex_features: Dict[Text, Text],
expected_patterns: Dict[Text, Text],
):
training_data = TrainingData()
if lookup_tables:
training_data.lookup_tables = [lookup_tables]
if regex_features:
training_data.regex_features = [regex_features]
actual_patterns = pattern_utils.extract_patterns(training_data)
assert actual_patterns == expected_patterns
def test_extract_patterns_use_only_lookup_tables_or_regex_features(
lookup_tables: Dict[Text, List[Text]],
regex_features: Dict[Text, Text],
use_lookup_tables: bool,
use_regex_features: bool,
expected_patterns: Dict[Text, Text],
):
training_data = TrainingData()
if lookup_tables:
training_data.lookup_tables = [lookup_tables]
if regex_features:
training_data.regex_features = [regex_features]
actual_patterns = pattern_utils.extract_patterns(
training_data,
use_lookup_tables=use_lookup_tables,
use_regexes=use_regex_features,
)
assert actual_patterns == expected_patterns
def test_extract_patterns_use_only_entities_regexes(
entity: Text, regex_features: Dict[Text, Text],
expected_patterns: Dict[Text, Text]):
training_data = TrainingData()
if entity:
training_data.training_examples = [
Message(data={
"text": "text",
"entities": [{
"entity": entity,
"value": "text"
}]
})
]
if regex_features:
training_data.regex_features = [regex_features]
actual_patterns = pattern_utils.extract_patterns(training_data,
use_only_entities=True)
assert actual_patterns == expected_patterns
def test_regex_validation(
lookup_tables: Dict[Text, List[Text]],
regex_features: Dict[Text, Text],
use_lookup_tables: bool,
use_regex_features: bool,
):
"""Tests if exception is raised when regex patterns are invalid."""
training_data = TrainingData()
if lookup_tables:
training_data.lookup_tables = [lookup_tables]
if regex_features:
training_data.regex_features = [regex_features]
with pytest.raises(Exception) as e:
pattern_utils.extract_patterns(
training_data,
use_lookup_tables=use_lookup_tables,
use_regexes=use_regex_features,
)
assert "Model training failed." in str(e.value)
assert "not a valid regex." in str(e.value)
assert "Please update your nlu training data configuration" in str(e.value)
