def _filter_overlapping_builtins(builtin_entities, tokens, tags,
tagging_scheme):
slots = tags_to_preslots(tokens, tags, tagging_scheme)
ents = []
for ent in builtin_entities:
if any(
ranges_overlap(ent[RES_MATCH_RANGE], s[RES_MATCH_RANGE])
for s in slots):
continue
ents.append(ent)
return ents
def test_ranges_overlap(self):
# Given
range1 = [4, 8]
range2 = [5, 7]
range3 = [3, 9]
range4 = [3, 4]
range5 = [8, 9]
range6 = [3, 6]
range7 = [4, 10]
# When / Then
self.assertTrue(ranges_overlap(range1, range2))
self.assertTrue(ranges_overlap(range1, range3))
self.assertFalse(ranges_overlap(range1, range4))
self.assertFalse(ranges_overlap(range1, range5))
self.assertTrue(ranges_overlap(range1, range6))
self.assertTrue(ranges_overlap(range1, range7))
def _disambiguate_builtin_entities(builtin_entities):
if not builtin_entities:
return []
builtin_entities = sorted(
builtin_entities,
key=lambda be: be[RES_MATCH_RANGE][END] - be[RES_MATCH_RANGE][START],
reverse=True)
disambiguated_entities = [builtin_entities[0]]
for entity in builtin_entities[1:]:
entity_rng = entity[RES_MATCH_RANGE]
conflict = False
for disambiguated_entity in disambiguated_entities:
disambiguated_entity_rng = disambiguated_entity[RES_MATCH_RANGE]
if ranges_overlap(entity_rng, disambiguated_entity_rng):
conflict = True
if entity_rng == disambiguated_entity_rng:
disambiguated_entities.append(entity)
break
if not conflict:
disambiguated_entities.append(entity)
return sorted(disambiguated_entities,
key=lambda be: be[RES_MATCH_RANGE][START])
def overlap(lhs_slot, rhs_slot):
return ranges_overlap(lhs_slot[RES_MATCH_RANGE],
rhs_slot[RES_MATCH_RANGE])