def parser_duration_with_ago_and_later(
self, source: str,
reference: datetime) -> DateTimeResolutionResult:
result = DateTimeResolutionResult()
duration_res = self.duration_extractor.extract(source, reference).pop()
if duration_res:
match = self.config._unit_regex.search(source)
if match:
suffix = source[duration_res.start + duration_res.length:]
src_unit = RegExpUtility.get_group(match, 'unit')
number_str = source[duration_res.start:match.lastindex -
duration_res.start + 1]
number = self.parse_chinese_written_number_to_value(number_str)
if src_unit in self.config.unit_map:
unit_str = self.config.unit_map.get(src_unit)
before_match = RegExpUtility.get_matches(
ChineseDateExtractor.before_regex, suffix)
if before_match and suffix.startswith(before_match[0]):
if unit_str == Constants.TIMEX_DAY:
date = reference + timedelta(days=-number)
elif unit_str == Constants.TIMEX_WEEK:
date = reference + timedelta(days=-7 * number)
elif unit_str == Constants.TIMEX_MONTH_FULL:
date = reference.replace(month=reference.month - 1)
elif unit_str == Constants.TIMEX_YEAR:
date = reference.replace(year=reference.year - 1)
else:
return result
result.timex = DateTimeFormatUtil.luis_date_from_datetime(
date)
result.future_value = result.past_value = date
result.success = True
return result
after_match = RegExpUtility.get_matches(
ChineseDateExtractor.after_regex, suffix)
if after_match and suffix.startswith(after_match[0]):
if unit_str == Constants.TIMEX_DAY:
date = reference + timedelta(days=number)
elif unit_str == Constants.TIMEX_WEEK:
date = reference + timedelta(days=7 * number)
elif unit_str == Constants.TIMEX_MONTH_FULL:
date = reference.replace(month=reference.month + 1)
elif unit_str == Constants.TIMEX_YEAR:
date = reference.replace(year=reference.year + 1)
else:
return result
result.timex = DateTimeFormatUtil.luis_date_from_datetime(
date)
result.future_value = result.past_value = date
result.success = True
return result
return result
def _parser_duration_with_ago_and_later(
self, source: str,
reference: datetime) -> DateTimeResolutionResult:
result = DateTimeResolutionResult()
duration_res = self.duration_extractor.extract(source, reference).pop()
if duration_res:
match = ChineseDateTimeExtractor.date_time_period_unit_regex.search(
source)
if match:
suffix = source[duration_res.start + duration_res.length:]
src_unit = RegExpUtility.get_group(match, 'unit')
number_str = source[duration_res.start:match.lastindex -
duration_res.start + 1]
number = ChineseDateParser.parse_chinese_written_number_to_value(
ChineseDateParser(), number_str)
if src_unit in self.config.unit_map:
unit_str = self.config.unit_map.get(src_unit)
before_match = RegExpUtility.get_matches(
ChineseDateExtractor.before_regex, suffix)
if before_match and suffix.startswith(before_match[0]):
if unit_str == Constants.TIMEX_HOUR:
date = reference + timedelta(hours=-number)
elif unit_str == Constants.TIMEX_MINUTE:
date = reference + timedelta(minutes=-number)
elif unit_str == Constants.TIMEX_SECOND:
date = reference + timedelta(seconds=-number)
else:
return result
result.timex = DateTimeFormatUtil.luis_date_from_datetime(
date)
result.future_value = result.past_value = date
result.success = True
return result
after_match = RegExpUtility.get_matches(
ChineseDateExtractor.after_regex, suffix)
if after_match and suffix.startswith(after_match[0]):
if unit_str == Constants.TIMEX_HOUR:
date = reference + timedelta(hours=number)
elif unit_str == Constants.TIMEX_MINUTE:
date = reference + timedelta(minutes=number)
elif unit_str == Constants.TIMEX_SECOND:
date = reference + timedelta(seconds=number)
else:
return result
result.timex = DateTimeFormatUtil.luis_date_from_datetime(
date)
result.future_value = result.past_value = date
result.success = True
return result
return result
def relative_duration_date(self, source: str, reference: datetime) -> List[Token]:
ret: List[Token] = list()
duration_er = self.duration_extractor.extract(source, reference)
for er in duration_er:
if not regex.search(self.date_time_period_unit_regex, er.text):
pos = er.start + er.length
if pos < len(source):
suffix = source[pos]
before_match = RegExpUtility.get_matches(self.before_regex, suffix)
after_match = RegExpUtility.get_matches(self.after_regex, suffix)
if (before_match and suffix.startswith(before_match[0])) \
or (after_match and suffix.startswith(after_match[0])):
meta_data = MetaData()
meta_data.is_duration_with_ago_and_later = True
ret.append(Token(er.start, pos + 1, meta_data))
return ret
def extract(self, source: str):
results: List[ExtractResult] = list()
partial_results: List[ExtractResult] = list()
trimmed_source = source.lower()
if source is None or source.strip() == '':
return results
source_tokens = self.__tokenize(trimmed_source)
for (regexp, type_extracted) in self.config.regexes_map.items():
for match in RegExpUtility.get_matches(regexp, trimmed_source):
match_tokens = self.__tokenize(match)
top_score = 0.0
for i in range(len(source_tokens)):
score = self.match_value(source_tokens, match_tokens, i)
top_score = max(top_score, score)
if top_score > 0.0:
value = ExtractResult()
start = trimmed_source.index(match)
length = len(match)
text = source[start:start + length].strip()
value.start = start
value.length = length
value.text = text
value.type = type_extracted
value.data = ChoiceExtractDataResult(source, top_score)
partial_results.append(value)
if len(partial_results) == 0:
return results
partial_results = sorted(partial_results, key=lambda res: res.start)
if self.config.only_top_match:
top_score = 0.0
top_result_index = 0
for i in range(len(partial_results)):
data = ChoiceExtractDataResult(source,
partial_results[i].data.score)
if data.score > top_score:
top_score = data.score
top_result_index = i
top_result = ChoiceExtractDataResult(
partial_results[top_result_index].data.source,
partial_results[top_result_index].data.score)
top_result.other_matches = partial_results
results.append(partial_results[top_result_index])
else:
results = partial_results
return results