def _digit_number_parse(self, ext_result: ExtractResult) -> ParseResult:
result = ParseResult()
result.start = ext_result.start
result.length = ext_result.length
result.text = ext_result.text
result.type = ext_result.type
result.meta_data = MetaData(
) if not result.meta_data else result.meta_data
# [1] 24
# [2] 12 32/33
# [3] 1,000,000
# [4] 234.567
# [5] 44/55
# [6] 2 hundred
# dot occurred.
power = 1
tmp_index = -1
start_index = 0
handle = ext_result.text.lower()
matches = list(regex.finditer(self.config.digital_number_regex,
handle))
if matches:
for match in matches:
rep = self.config.round_number_map.get(match.group())
# \\s+ for filter the spaces.
power *= rep
tmp_index = handle.find(match.group(), start_index)
while tmp_index >= 0:
front = handle[0:tmp_index].rstrip()
start_index = len(front)
handle = front + handle[tmp_index + len(match):]
tmp_index = handle.find(match.group(), start_index)
# Scale used in the calculate of double
result.value = self._get_digital_value(handle, power)
return result
def dou_parse(self, source: ExtractResult) -> ParseResult:
result = ParseResult(source)
source_text = self.replace_unit(source.text)
if (regex.search(self.config.double_and_round_regex, source.text)) is not None:
power = self.config.round_number_map_char[source_text[-1:]]
result.value = self.get_digit_value(source_text[:-1], power)
else:
split_result = regex.split(self.config.point_regex, source_text)
if split_result[0] == '':
split_result[0] = '零'
if regex.search(self.config.negative_number_sign_regex, split_result[0]) is not None:
result.value = self.get_int_value(split_result[0]) - self.get_point_value(split_result[1])
else:
result.value = self.get_int_value(split_result[0]) + self.get_point_value(split_result[1])
result.resolution_str = self.__format(result.value)
return result
def per_parse_chs(self, source: ExtractResult) -> ParseResult:
result = ParseResult(source)
source_text = source.text
power = 1
if 'Spe' in source.data:
source_text = self.replace_full_with_half(source_text)
source_text = self.replace_unit(source_text)
if source_text == '半折':
result.value = 50
elif source_text == '10成':
result.value = 100
else:
matches = list(regex.finditer(self.config.spe_get_number_regex, source_text))
int_number: int
if len(matches) == 2:
int_number_char = matches[0].group()[0]
if int_number_char == '对':
int_number = 5
elif int_number_char == '十' or int_number_char == '拾':
int_number = 10
else:
int_number = self.config.zero_to_nine_map_chs[int_number_char]
point_number_char = matches[1].group()[0]
point_number: float
if point_number_char == '半':
point_number = 0.5
else:
point_number = self.config.zero_to_nine_map_chs[point_number_char] * 0.1
result.value = (int_number + point_number) * 10
else:
int_number_char = matches[0].group()[0]
if int_number_char == '对':
int_number = 5
elif int_number_char == '十' or int_number_char == '拾':
int_number = 10
else:
int_number = self.config.zero_to_nine_map_chs[int_number_char]
result.value = int_number * 10
elif 'Num' in source.data:
double_match = regex.search(self.config.percentage_regex, source_text)
double_text = double_match.group()
if any(x for x in ['k', 'K', 'ｋ', 'Ｋ'] if x in double_text):
power = 1000
elif any(x for x in ['M', 'Ｍ'] if x in double_text):
power = 1000000
elif any(x for x in ['G', 'Ｇ'] if x in double_text):
power = 1000000000
elif any(x for x in ['T', 'Ｔ'] if x in double_text):
power = 1000000000000
result.value = self.get_digit_value_chs(double_text, power)
else:
double_match = regex.search(self.config.percentage_regex, source_text)
double_text = self.replace_unit(double_match.group())
split_result = regex.split(self.config.point_regex_chs, double_text)
if split_result[0] == '':
split_result[0] = '零'
double_value = self.get_int_value_chs(split_result[0])
if len(split_result) == 2:
if regex.search(self.config.negative_number_sign_regex, split_result[0]) is not None:
double_value -= self.get_point_value_chs(split_result[1])
else:
double_value += self.get_point_value_chs(split_result[1])
result.value = double_value
result.resolution_str = self.__format(result.value) + '%'
return result