def test_should_fail_generating_entity_with_wrong_file_name(self):
# Given
examples_path = PACKAGE_PATH / "cli" / "dataset" / "examples"
entity_file = examples_path / "location.txt"
# When / Then
with self.assertRaises(AssertionError):
CustomEntity.from_file(entity_file)
def test_should_generate_entity_from_file(self):
# Given
examples_path = PACKAGE_PATH / "cli" / "dataset" / "examples"
entity_file = examples_path / "entity_location.txt"
# When
entity_dataset = CustomEntity.from_file(entity_file)
entity_dict = entity_dataset.json
# Then
expected_entity_dict = {
"automatically_extensible":
True,
"data": [{
"synonyms": ["big apple"],
"value": "new york"
}, {
"synonyms": ["city of lights"],
"value": "paris"
}, {
"synonyms": [],
"value": "london"
}],
"use_synonyms":
True
}
self.assertDictEqual(expected_entity_dict, entity_dict)
def from_files(cls, language, filenames):
"""Creates an :class:`.AssistantDataset` from a language and a list of
intent and entity files
Args:
language (str): language of the assistant
filenames (list of str): Intent and entity files.
The assistant will associate each intent file to an intent,
and each entity file to an entity. For instance, the intent
file 'intent_setTemperature.txt' will correspond to the intent
'setTemperature', and the entity file 'entity_room.txt' will
correspond to the entity 'room'.
"""
intent_filepaths = set()
entity_filepaths = set()
for filename in filenames:
filepath = Path(filename)
stem = filepath.stem
if stem.startswith("intent_"):
intent_filepaths.add(filepath)
elif stem.startswith("entity_"):
entity_filepaths.add(filepath)
else:
raise AssertionError("Filename should start either with "
"'intent_' or 'entity_' but found: %s"
% stem)
intents_datasets = [IntentDataset.from_file(f)
for f in intent_filepaths]
entities = [CustomEntity.from_file(f) for f in entity_filepaths]
entity_names = set(e.name for e in entities)
# Add entities appearing only in the intents data
for intent_data in intents_datasets:
for entity_name in intent_data.entities_names:
if entity_name not in entity_names:
entity_names.add(entity_name)
entities.append(create_entity(entity_name))
return cls(language, intents_datasets, entities)