def test_extractor_1():
context = Context(
'Tom and Bob declined cancer treatment.',
['Tom', 'and', 'Bob', 'declined', 'cancer', 'treatment', '.'])
context.add('entities', ['B-A', '', 'B-A', '', 'B-B', 'B-C', ''])
spacy = SpacyIntegration('en_core_web_lg')
extract = StrictDependencyRelationshipExtract(spacy=spacy,
attribute='entities',
relationships=[
('A', 'has_B', 'B'),
('B', 'has_C', 'C'),
])
relationships = extract.evaluate(context)
assert len(relationships) == 2
kb_triples = relationships[0]
assert kb_triples[0].subj == 'Tom'
assert kb_triples[0].obj == 'cancer'
assert kb_triples[1].subj == 'cancer'
assert kb_triples[1].obj == 'treatment'
kb_triples = relationships[1]
assert kb_triples[0].subj == 'Bob'
assert kb_triples[0].obj == 'cancer'
assert kb_triples[1].subj == 'cancer'
assert kb_triples[1].obj == 'treatment'
def test_extractor_2():
context = Context('Tom declined cancer treatment.',
['Tom', 'declined', 'cancer', 'treatment', '.'])
context.add('scope1', ['', 'S1', 'S1', '', ''])
context.add('scope2', ['', '', 'S2', 'S2', ''])
relationships = ScopeOverlapExtract('scope1', 'scope2').evaluate(context)
assert relationships == ['', 'REL', 'REL', 'REL', '']
def test_chunker_single_term_phrase():
chunker = TreeChunker("NP: {<DT>?<JJ>*<NN>}", attribute='pos')
context = Context(
'learn php from guru99',
['learn','php', 'from','guru99']
)
context.add('pos', ['JJ', 'NN', 'IN', 'NN'])
chunks = chunker.tag(context)
assert chunks == ['B-NP', 'I-NP', '', 'B-NP']
def test_chunker_single_term_phrase_not_pos():
chunker = TreeChunker("AB: {<A><B>}", attribute='outer')
context = Context(
'learn php from guru99',
['learn','php', 'from','guru99']
)
context.add('outer', ['A', 'B', '', ''])
chunks = chunker.tag(context)
assert chunks == ['B-AB', 'I-AB', '', '']
def test_extractor():
context = Context('We are not going to New York.',
['We', 'are', 'not', 'going', 'to', 'New', 'York', '.'])
context.add('negative', ['', '', 'NEG', 'NEG', '', '', '', ''])
context.add('negative_scope',
['', '', 'NEG', 'NEG', 'NEG', 'NEG', 'NEG', 'NEG'])
context.add('entities', ['', '', '', '', '', 'B-LOC', 'I-LOC', ''])
negative_extracts = TagExtract('negative', ['NEG']).evaluate(context)
negative_scope_extracts = TagExtract('negative_scope',
['NEG']).evaluate(context)
location_extracts = TagExtract('entities', ['LOC']).evaluate(context)
assert negative_extracts == {
'NEG': [
[(2, 'not'), (3, 'going')],
]
}
assert negative_scope_extracts == {
'NEG': [
[(2, 'not'), (3, 'going'), (4, 'to'), (5, 'New'), (6, 'York'),
(7, '.')],
]
}
assert location_extracts == {'LOC': [[(5, 'New'), (6, 'York')]]}
def test_context_cache():
context = Context('We are not going to New York.',
['We', 'are', 'not', 'going', 'to', 'New', 'York', '.'],
actor='Tom',
line=4,
sent_number=3)
assert context.cache['actor'] == 'Tom'
assert context.cache['line'] == 4
assert context.cache['sent_number'] == 3
def test_combine():
context = Context(
'We are not going to New York.',
['We', 'are', 'not', 'going', 'to', 'New', 'York', '.']
)
context.add('entities', ['', '', '', '', '', 'LOC', 'LOC', ''])
context.add('negatives', ['', '', 'NEG', '', '', '', '', ''])
data = Merger.combine(
context,
['tokens', 'entities', 'negatives']
)
assert data == [
{ 'tokens': 'We' },
{ 'tokens': 'are' },
{ 'tokens': 'not', 'negatives': 'NEG' },
{ 'tokens': 'going' },
{ 'tokens': 'to' },
{ 'tokens': 'New', 'entities': 'LOC' },
{ 'tokens': 'York', 'entities': 'LOC' },
{ 'tokens': '.' },
]
def test_extracting_multiple_pipes():
pipeline = ContextPipeline(pipes=[
TokenAttributesPipe(),
ChunkerPipe('regex_ner',
PhraseChunker(tags=[
Tag('LOC', phrase=RegexPhrase(r'\b(New York)\b')),
]),
order=2),
ChunkerPipe('neg',
PhraseChunker(
tags=[
Tag('FRW-NEG', phrase=RegexPhrase(r'\b(not)\b')),
],
apply_iob2=False,
),
order=3),
TokenTaggerPipe(
'neg_scope',
focus='neg',
tagger=Scoper(scopes=[
Scope(
applied_tag='FRW-NEG',
scope_direction=ScopeDirection.RIGHT,
order=1,
)
], ),
order=4,
)
])
context = Context('We are not going to New York.',
['We', 'are', 'not', 'going', 'to', 'New', 'York', '.'])
context = pipeline.execute(context)
assert len(context.keys) == 8
assert context.get('tokens') == [
'We', 'are', 'not', 'going', 'to', 'New', 'York', '.'
]
assert context.get('pos') == [
'PRP', 'VBP', 'RB', 'VBG', 'TO', 'NNP', 'NNP', '.'
]
assert context.get('spacings') == [
'yes', 'yes', 'yes', 'yes', 'yes', 'yes', 'no', 'no'
]
assert context.get('start_positions') == [
'0', '3', '7', '11', '17', '20', '24', '28'
]
assert context.get('end_positions') == [
'1', '5', '9', '15', '18', '22', '27', '28'
]
assert context.get('regex_ner') == [
'', '', '', '', '', 'B-LOC', 'I-LOC', ''
]
assert context.get('neg') == ['', '', 'FRW-NEG', '', '', '', '', '']
assert context.get('neg_scope') == [
'',
'',
'FRW-NEG',
'FRW-NEG',
'FRW-NEG',
'FRW-NEG',
'FRW-NEG',
'FRW-NEG',
]
def test_context_add_and_get():
context = Context('We are not going to New York.',
['We', 'are', 'not', 'going', 'to', 'New', 'York', '.'])
context.add('entities', ['', '', '', '', '', 'LOC', 'LOC', ''])
context.add('negatives', ['', '', 'NEG', '', '', '', '', ''])
context.add('negative_scope',
['', '', 'NEG', 'NEG', 'NEG', 'NEG', 'NEG', 'NEG'])
assert len(context.keys) == 4
assert len(context.get('tokens')) == 8
assert len(context.get('entities')) == 8
assert len(context.get('negatives')) == 8
assert len(context.get('negative_scope')) == 8