2020-09-01 19:57:52 +00:00
|
|
|
from spacy.lang.en import English
|
2020-09-04 12:05:55 +00:00
|
|
|
from spacy.pipeline import merge_entities
|
2020-09-21 18:43:54 +00:00
|
|
|
import pytest
|
2020-09-01 19:57:52 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
def test_issue5918():
|
|
|
|
|
# Test edge case when merging entities.
|
|
|
|
|
nlp = English()
|
2020-09-04 12:05:55 +00:00
|
|
|
ruler = nlp.add_pipe("entity_ruler")
|
2020-09-01 19:57:52 +00:00
|
|
|
patterns = [
|
|
|
|
|
{"label": "ORG", "pattern": "Digicon Inc"},
|
|
|
|
|
{"label": "ORG", "pattern": "Rotan Mosle Inc's"},
|
|
|
|
|
{"label": "ORG", "pattern": "Rotan Mosle Technology Partners Ltd"},
|
|
|
|
|
]
|
|
|
|
|
ruler.add_patterns(patterns)
|
|
|
|
|
|
|
|
|
|
text = """
|
|
|
|
|
Digicon Inc said it has completed the previously-announced disposition
|
|
|
|
|
of its computer systems division to an investment group led by
|
|
|
|
|
Rotan Mosle Inc's Rotan Mosle Technology Partners Ltd affiliate.
|
|
|
|
|
"""
|
|
|
|
|
doc = nlp(text)
|
|
|
|
|
assert len(doc.ents) == 3
|
|
|
|
|
# make it so that the third span's head is within the entity (ent_iob=I)
|
|
|
|
|
# bug #5918 would wrongly transfer that I to the full entity, resulting in 2 instead of 3 final ents.
|
2020-09-21 18:43:54 +00:00
|
|
|
with pytest.warns(UserWarning):
|
|
|
|
|
doc[29].head = doc[33]
|
2020-09-01 19:57:52 +00:00
|
|
|
doc = merge_entities(doc)
|
|
|
|
|
assert len(doc.ents) == 3
|
