def tag_pattern(patterns, utterances):
final_utterances = []
for iter in utterances:
utterance = get(iter, "utterance")
case_converted_utterance = get(iter, "case_converted_utterance")
mapping = get(iter, "mapping")
mapping = json.loads(mapping)
tokens = get(mapping, "tokens")
text = get(mapping, "text")
tags = get(mapping, "tags")
intent = get(mapping, "intent")
pattern = regex_checker(utterance, patterns)
final_tags = tags
for pat in pattern:
final_tags.append(check(tags, pat))
final_tags = remove_duplicates(final_tags)
final_tags = sorted(final_tags, key=itemgetter('start'))
mappings = {
"tokens": tokens,
"text": text,
"tags": final_tags,
"intent": intent
}
data = {
"utterance": utterance,
"case_converted_utterance": case_converted_utterance,
"mapping": json.dumps(mappings)
}
final_utterances.append(data)
return final_utterances
def test_regex_checker_with_data1(mocker):
patterns = regex_checker(
'my birthdate is 27-06-1995',
[{
"entity": "custom_date",
"pattern": "\s\d{4}-\d{2}-\d{2}|\d{2}-\d{2}-\d{4}"
}])
assert patterns == [{
'end': 4,
'entity': '27-06-1995',
'start': 3,
'tag': 'custom_date'
}]
def get_entities_for_ds(self, serviceid, text, datasources_map):
predefined_tags = datasources_map["predefined_entities"] if 'predefined_entities' in datasources_map else []
pattern_response = []
phrase_response = []
pattern_entities = datasources_map[
"patterns"] if 'patterns' in datasources_map else []
phrase_entities = datasources_map[
"phrases"] if 'phrases' in datasources_map else []
if len(pattern_entities) != 0:
pattern_response = regex_checker(text, pattern_entities)
if len(phrase_entities) != 0:
phrase_response = phrase_checker(text, phrase_entities)
return pattern_response, phrase_response, predefined_tags
def test_regex_checker_with_null(mocker):
patterns = regex_checker('', [])
resp = mocker.patch('ice_commons.store.util.remove_duplicates', [])
assert patterns == resp