spaCy/tests/matcher/test_matcher_bugfixes.py

import pytest

from spacy.matcher import Matcher
from spacy.attrs import LOWER


def test_overlap_issue118(EN):
    '''Test a bug that arose from having overlapping matches'''
    doc = EN.tokenizer(u'how many points did lebron james score against the boston celtics last night')
    ORG = doc.vocab.strings['ORG']
    matcher = Matcher(EN.vocab,
        {'BostonCeltics':
            ('ORG', {},
                [
                    [{LOWER: 'celtics'}],
                    [{LOWER: 'boston'}, {LOWER: 'celtics'}],
                ]
            )
        }
    )
    
    assert len(list(doc.ents)) == 0
    matches = matcher(doc)
    assert matches == [(ORG, 9, 11), (ORG, 10, 11)]
    ents = list(doc.ents)
    assert len(ents) == 1
    assert ents[0].label == ORG
    assert ents[0].start == 9
    assert ents[0].end == 11


def test_overlap_reorder(EN):
    '''Test order dependence'''
    doc = EN.tokenizer(u'how many points did lebron james score against the boston celtics last night')
    ORG = doc.vocab.strings['ORG']
    matcher = Matcher(EN.vocab,
        {'BostonCeltics':
            ('ORG', {},
                [
                    [{LOWER: 'boston'}, {LOWER: 'celtics'}],
                    [{LOWER: 'celtics'}],
                ]
            )
        }
    )
    
    assert len(list(doc.ents)) == 0
    matches = matcher(doc)
    assert matches == [(ORG, 9, 11), (ORG, 10, 11)]
    ents = list(doc.ents)
    assert len(ents) == 1
    assert ents[0].label == ORG
    assert ents[0].start == 9
    assert ents[0].end == 11


def test_overlap_prefix(EN):
    '''Test order dependence'''
    doc = EN.tokenizer(u'how many points did lebron james score against the boston celtics last night')
    ORG = doc.vocab.strings['ORG']
    matcher = Matcher(EN.vocab,
        {'BostonCeltics':
            ('ORG', {},
                [
                    [{LOWER: 'boston'}],
                    [{LOWER: 'boston'}, {LOWER: 'celtics'}],
                ]
            )
        }
    )
    
    assert len(list(doc.ents)) == 0
    matches = matcher(doc)
    assert matches == [(ORG, 9, 10), (ORG, 9, 11)]
    ents = list(doc.ents)
    assert len(ents) == 1
    assert ents[0].label == ORG
    assert ents[0].start == 9
    assert ents[0].end == 11


def test_overlap_prefix_reorder(EN):
    '''Test order dependence'''
    doc = EN.tokenizer(u'how many points did lebron james score against the boston celtics last night')
    ORG = doc.vocab.strings['ORG']
    matcher = Matcher(EN.vocab,
        {'BostonCeltics':
            ('ORG', {},
                [
                    [{LOWER: 'boston'}, {LOWER: 'celtics'}],
                    [{LOWER: 'boston'}],
                ]
            )
        }
    )
    
    assert len(list(doc.ents)) == 0
    matches = matcher(doc)
    assert matches == [(ORG, 9, 10), (ORG, 9, 11)]
    ents = list(doc.ents)
    assert len(ents) == 1
    assert ents[0].label == ORG
    assert ents[0].start == 9
    assert ents[0].end == 11