def doc_for_text(self, text):
if self.component_config.get("crf_lexical"):
request = nlp_messages.NlTokenizerRequest(text=nlp_messages.NlText(
text=text))
response = self.nlp.Tokenizer(request)
return response
else:
request = nlp_messages.NlTokenizerRequest(text=nlp_messages.NlText(
text=text))
response = self.nlp.EntityExtractor(request)
return response
def process(self, message, **kwargs):
# type: (Message, **Any) -> None
# can't use the existing doc here (spacy_doc on the message)
# because tokens are lower cased which is bad for NER
hanlp = kwargs.get("hanlp", None)
request = nlp_messages.NlText(text=message.text)
doc = hanlp.ParseAmountTerms(request)
extracted = self.add_extractor_name(self.extract_entities(doc))
message.set("entities",
message.get("entities", []) + extracted,
add_to_output=True)
def run():
serv = ServiceClient(nlp_service, 'CabochaNlpProcsStub', 'localhost', 50051)
# Insert example metadata
metadata = [('ip', '127.0.0.1')]
response = serv.Tokenizer(
nlp_messages.NlText(text="お皿を二枚ください。"),
metadata=metadata
)
if response:
print("response:")
tokens=tokenize_msg("お皿を二枚ください。", response)
for t in tokens:
print(t.text, t.offset)
def doc_for_text(self, text):
request = nlp_messages.NlText(text=text)
response = self.nlp.Tokenizer(request)
return response
def extract(self, text):
request = nlp_messages.NlTokenizerRequest(text=nlp_messages.NlText(text=text))
response = self.client.EntityExtractor(request)
return response
def get_deps(self, text):
request = nlp_messages.NlTexts(texts=[nlp_messages.NlText(text=text)])
response = self.client.GetDependencyGraph(request)
return response
def tokenize(self, text):
request = nlp_messages.NlTokenizerRequest(text=nlp_messages.NlText(
text=text))
response = self.client.Tokenizer(request)
return response