def load_negation(version=None, path=None, skip=False, variation=0):
if skip:
return None
elif path: # requirement: must precede version as sometimes 'path' is always specified
return StatusMiner(path=path, rx_var=variation)
elif version:
return StatusMiner(tablename=f'status{version}', rx_var=variation)
else:
raise ValueError('No negation cues provided!')
def test_negation_bidirectional():
miner = StatusMiner(rules=[])
terms = [
Term('I', 0, 1, 'C01', 1),
Word('have', 2, 6),
Negation('no', 7, 9, 'negn', 3),
Term('beer', 10, 14, 'C02', 2),
]
miner.postprocess(terms)
assert terms[0].is_negated()
assert terms[-1].is_negated()
def test_negation_end_scope_conj():
miner = StatusMiner(rules=[])
terms = [
Term('I', 0, 1, 'C01', 1),
Word('have', 2, 6),
Negation('no', 7, 9, 'negn', 3),
Term('beer', 10, 14, 'C02', 2),
Negation('or', 15, 17, 'conj', 0),
Term('wine', 18, 22, 'C02', 3),
]
miner.postprocess(terms)
assert terms[0].is_negated()
assert terms[-3].is_negated() # negates 'beer'
assert not terms[-1].is_negated() # 'or' blocks 'wine' from being negated
def test_hypothetical_sentence(concept_file):
mc = MinerCollection()
mc.add(ConceptMiner([CsvDictionary(concept_file)]))
mc.add(StatusMiner())
concepts, sentence = mc.parse_sentence('if there is bad weather', offset=0)
for concept in concepts:
assert concept.is_hypothetical()
def test_negated_sentence(concept_file):
mc = MinerCollection()
mc.add(ConceptMiner([CsvDictionary(concept_file)]))
mc.add(StatusMiner())
concepts, sentence = mc.parse_sentence('no bad weather', offset=0)
for concept in concepts:
assert concept.is_negated()
def prepare(documents: List[Document], dictionaries: List[Dictionary],
outputs: List[Output], negation_dicts: List[Dictionary]):
"""
:param documents:
:param dictionaries:
:param outputs:
:param negation_dicts:
"""
mc = MinerCollection(ssplit=SentenceBoundary().ssplit)
mc.add(ConceptMiner(dictionaries))
mc.add(StatusMiner(negation_dicts))
for out in outputs:
out.create_output()
logging.info('Retrieving notes.')
for document in documents:
for num, doc in enumerate(document.read_next()):
if num % 100 == 0:
logging.info(f'Completed: {num:>5}.')
for sent_no, (sentence, cleaned_text) in enumerate(mc.parse(doc)):
for feat, new in sentence:
for out in outputs:
out.writerow(doc.get_metalist(),
feat,
text=doc.get_text())
logging.info('Completed: 100%')
def test_negation_with_punct3():
miner = StatusMiner(rules=[(r'c\s*/\s*o', 'negn', 3)])
assert len(miner.mine('pt c/o', 0)) == 1
assert len(miner.mine('pt c / o', 0)) == 1
def test_negation_with_punct2():
miner = StatusMiner(rules=[('c / o', 'negn', 3)])
assert len(miner.mine('pt c/o', 0)) == 0
assert len(miner.mine('pt c / o', 0)) == 1
def test_statusminer_hypothetical():
s = StatusMiner()
negations = s.mine('if he wants to', 0)
assert len(negations) == 1
def test_statusminer_negated():
s = StatusMiner()
negations = s.mine('no evidence of anything', 0)
assert len(negations) == 1