def get_all_annotations(
cls,
text: str,
float_digits: int = 4,
) -> List[DurationAnnotation]:
all_annotations: List[DurationAnnotation] = []
for match in cls.DURATION_PTN_RE.finditer(text.lower()):
source_text, number_text, duration_type = match.groups()
amount = list(get_amounts(number_text, float_digits=float_digits))
if len(amount) != 1:
continue
amount = amount[0]
_duration_fraction: Fraction = cls.DURATION_MAP[duration_type]
duration_days: Decimal = Decimal(
(_duration_fraction.numerator * amount) /
_duration_fraction.denominator)
if float_digits:
duration_days: Decimal = quantize_by_float_digit(
amount=duration_days, float_digits=float_digits)
if duration_type == 'anniversaries':
duration_type = 'anniversary'
ant: DurationAnnotation = DurationAnnotation(
coords=match.span(),
amount=amount,
duration_type=duration_type,
duration_days=duration_days,
text=source_text.strip())
all_annotations.append(ant)
return all_annotations
def parse_annotations(self,
text: str,
float_digits: int = 4,
return_sources: bool = True
) -> Generator[AmountAnnotation, None, None]:
"""
Find possible amount references in the text.
:param text: text
:param return_sources: return amount AND source text
:param extended_sources: return data around amount itself
:param float_digits: round float to N digits, don't round if None
:return: list of amounts
"""
for match in self.NUM_PTN_RE.finditer(text):
found_item = match.group()
if self.WRONG_FULLMATCH_RE.fullmatch(found_item):
continue
try:
amount = self.text2num(found_item)
except Exception as e:
print(e)
continue
if amount is None:
continue
if float_digits:
amount: Decimal = quantize_by_float_digit(
amount=amount, float_digits=float_digits)
ant = AmountAnnotation(coords=match.span(),
value=amount,
locale=self.language)
if return_sources:
unit = ''
next_text = text[match.span()[1]:]
if next_text:
for np in get_np(next_text):
if next_text.startswith(np):
unit = np
if unit:
found_item = ' '.join([found_item.strip(), unit])
if not unit:
prev_text = text[:match.span()[0]]
prev_text_tags = nltk.word_tokenize(prev_text)
if prev_text_tags and prev_text_tags[-1].lower(
) in allowed_prev_units:
sep = ' ' if text[match.span()[0] - 1] == ' ' else ''
found_item = sep.join(
[prev_text_tags[-1],
found_item.rstrip()])
ant.text = found_item.strip()
yield ant
def get_all_annotations(cls,
text: str,
float_digits: int = 4) -> List[DurationAnnotation]:
all_annotations: List[DurationAnnotation] = []
for match in cls.DURATION_PTN_RE.finditer(text):
capture = match.capturesdict()
amount_text = ''.join(capture.get('num_text', ''))
amounts = list(get_amounts(amount_text, float_digits=float_digits))
if len(amounts) != 1:
amount = Decimal('1.0')
else:
amount = amounts[0]
unit_name_local = ''.join(capture.get('unit_name', '')).lower()
unit_prefix = ''.join(capture.get('unit_prefix', '')).lower()
unit_name_local = cls.DURATION_MAP_RE.findall(unit_name_local)
if not unit_name_local:
continue
unit_name_local = unit_name_local[0]
unit_name_en = cls.DURATION_TRANSLATION_MAP.get(unit_name_local)
_duration_fraction: Fraction = cls.DURATION_MAP[unit_name_en]
amount_days: Decimal = Decimal(
(_duration_fraction.numerator * amount) /
_duration_fraction.denominator)
if float_digits:
amount_days: Decimal = quantize_by_float_digit(
amount=amount_days, float_digits=float_digits)
ant: DurationAnnotation = DurationAnnotation(
coords=match.span(),
text=''.join(capture.get('text', '')),
amount=amount,
duration_days=amount_days,
duration_type_en=unit_name_en,
duration_type=unit_name_local,
prefix=unit_prefix,
locale=cls.LOCALE)
all_annotations.append(ant)
return all_annotations