def get_parse_time_result(self, entity: str, match: Match, reference: datetime) -> TimeResult:
match_str: str = match.group()
extract_result = ExtractResult()
extract_result.start = match.start() + match_str.find(entity)
extract_result.length = len(entity)
extract_result.text = entity
extract_result.type = Constants.SYS_DATETIME_TIME
result = self.config.time_parser.parse(extract_result, reference)
return result.data
def extract(self,
source: str,
reference: datetime = None) -> List[ExtractResult]:
if reference is None:
reference = datetime.now()
result: List[ExtractResult] = list()
if not source:
return result
match_source: Dict[Match, any] = dict()
matched: List[bool] = [False] * len(source)
collections = list(
map(lambda x: (list(regex.finditer(x[0], source)), x[1]),
self._regex_dict.items()))
collections = list(filter(lambda x: len(x[0]) > 0, collections))
for collection in collections:
for match in collection[0]:
for j in range(len(match.group())):
matched[match.start() + j] = True
match_source[match] = collection[1]
last = -1
for i in range(len(source)):
if matched[i]:
if i + 1 == len(source) or not matched[i + 1]:
start = last + 1
length = i - last
text = source[start:start + length].strip()
src_match = next(
(x for x in iter(match_source)
if (x.start() == start and (x.end() -
x.start()) == length)),
None)
if src_match:
value = ExtractResult()
value.start = start
value.length = length
value.text = text
value.type = self.extractor_type_name
value.data = self.__get_data(match_source, src_match)
result.append(value)
else:
last = i
return result