def test_definitions_in_one_sentence(self):
sentence = 'The "Pope": the head of the Catholic Church.'
definitions = list(get_definitions_in_sentence(sentence, return_sources=False))
self.assertEqual(1, len(definitions))
self.assertEqual('Pope', definitions[0].strip(' :"'))
definitions = list(get_definitions_in_sentence(sentence, return_sources=True))
self.assertEqual(1, len(definitions))
self.assertEqual('Pope', definitions[0][0].strip(' :"'))
def _matches_definition_words(self, sentence: str) -> bool:
if self._definition_words:
terms = get_definitions_in_sentence(sentence)
if not terms:
return False
terms = set([self._clean_def_words(t) for t in terms])
for w in self._definition_words:
if w in terms:
return True
return False
def parse(self, text, text_unit_id, _text_unit_lang,
**kwargs) -> ParseResults:
found = list(definitions.get_definitions_in_sentence(text))
if found:
unique = set(found)
return ParseResults({
DefinitionUsage: [
DefinitionUsage(text_unit_id=text_unit_id,
definition=item,
count=found.count(item)) for item in unique
]
})
def _matches_definition_words(self, text: str, text_is_sentence: bool) -> bool:
if not self.detector.detector_definition_words:
return False
try:
terms = get_definitions_in_sentence(text) \
if text_is_sentence else get_definitions(text)
except Exception as e:
msg = f'{self.get_detector_code()}: error in ' + \
f'_matches_definition_words("{text}"), ' + \
'in get_definitions_in_sentence' if text_is_sentence \
else 'if get_definitions'
e.detailed_error = msg
raise
if not terms:
return False
terms = set([self._clean_def_words(t) for t in terms])
for w in self.detector.detector_definition_words:
if w in terms:
return True
return False
def test_the(self):
text = '''The mean-\ning of our English term has shifted\nfrom character to actions-to
external\nacts, manner of life, conduct, or\nhabits.'''
definitions = list(get_definitions_in_sentence(text))
self.assertEqual(len(definitions), 0)
def test_trigger_word_fullmatches(self):
text = '''(i)\nThe meanings given to terms defined herein shall be equally applicable to both\n
the singular and plural forms of such terms.'''
definitions = list(get_definitions_in_sentence(text))
self.assertEqual(len(definitions), 0)
def test_dot_in_definition(self):
text = '''“U.S. Person” means any Person that is a “United States Person” as defined
in Section 7701(a)(30) of the Code.'''
definitions = list(get_definitions_in_sentence(text))
self.assertEqual(definitions[0], 'U.S. Person')
