def make_entity(self, text, type_, start, end):
text = text.strip()
if not len(text):
return
entity = ExtractedEntity()
entity.text = text
entity.type = type_
entity.start = start
entity.end = end
return entity
def extract_spacy(self, text):
try:
doc = self.spacy(text)
for ent in doc.ents:
type_ = SPACY_TYPES.get(ent.label_)
label = ent.text.strip()
if type_ is not None and len(label):
entity = ExtractedEntity()
entity.text = label
entity.type = type_
entity.start = ent.start
entity.end = ent.end
yield entity
except Exception:
log.exception("spaCy failed")
def Extract(self, request, context):
try:
doc = self.nlp(request.text)
count = 0
for ent in doc.ents:
type_ = SPACY_TYPES.get(ent.label_)
label = ent.text.strip()
if type_ is not None and len(label):
count += 1
entity = ExtractedEntity()
entity.text = label
entity.type = type_
entity.start = ent.start
entity.end = ent.end
yield entity
log.info("[NER]: %d entities from %d chars", count,
len(request.text))
except Exception as exc:
log.exception("Failed to extract entities")
context.abort(grpc.StatusCode.INTERNAL, str(exc))