def test_parse_complete_dates(self, text, day, month, year):
span = interaction_pb2.NumericValueSpan(
begin_index=0,
end_index=len(text),
values=[
interaction_pb2.NumericValue(
date=interaction_pb2.Date(year=year, month=month, day=day))
])
self.assertEqual([span], number_utils.parse_text(text))
def parse_text(text):
"""Extracts longest number and date spans.
Args:
text: text to annotate.
Returns:
List of longest numeric value spans.
"""
span_dict = collections.defaultdict(list)
for match in _NUMBER_PATTERN.finditer(text):
span_text = text[match.start():match.end()]
number = _parse_number(span_text)
if number is not None:
span_dict[match.span()].append(_get_numeric_value_from_float(number))
for begin_index, end_index in text_utils.get_all_spans(
text, max_ngram_length=1):
if (begin_index, end_index) in span_dict:
continue
span_text = text[begin_index:end_index]
number = _parse_number(span_text)
if number is not None:
span_dict[begin_index, end_index].append(
_get_numeric_value_from_float(number))
for number, word in enumerate(_NUMBER_WORDS):
if span_text == word:
span_dict[begin_index, end_index].append(
_get_numeric_value_from_float(float(number)))
break
for number, word in enumerate(_ORDINAL_WORDS):
if span_text == word:
span_dict[begin_index, end_index].append(
_get_numeric_value_from_float(float(number)))
break
for begin_index, end_index in text_utils.get_all_spans(
text, max_ngram_length=_MAX_DATE_NGRAM_SIZE):
span_text = text[begin_index:end_index]
date = _parse_date(span_text)
if date is not None:
span_dict[begin_index, end_index].append(date)
spans = sorted(
span_dict.items(),
key=lambda span_value: _get_span_length_key(span_value[0]),
reverse=True)
selected_spans = []
for span, value in spans:
for selected_span, _ in selected_spans:
if selected_span[0] <= span[0] and span[1] <= selected_span[1]:
break
else:
selected_spans.append((span, value))
selected_spans.sort(key=lambda span_value: span_value[0][0])
numeric_value_spans = []
for span, values in selected_spans:
numeric_value_spans.append(
interaction_pb2.NumericValueSpan(
begin_index=span[0], end_index=span[1], values=values))
return numeric_value_spans
def test_parse_numerals(self, text):
span = interaction_pb2.NumericValueSpan(
begin_index=0,
end_index=len(text),
values=[interaction_pb2.NumericValue(float_value=12)])
self.assertEqual([span], number_utils.parse_text(text))