def extract_text(self, text, languages):
for attempt in service_retries():
try:
service = EntityExtractStub(self.channel)
req = Text(text=text, languages=languages)
for res in service.Extract(req):
clazz = self.TYPES.get(res.type)
yield (res.text, clazz, res.start, res.end)
break
except self.Error as e:
if e.code() not in self.TEMPORARY_ERRORS:
return
self.reset_channel()
log.warning("gRPC [%s]: %s", e.code(), e.details())
backoff(failures=attempt)
def extract(self, collector, document):
DocumentTagCollector(document, 'polyglot').save()
DocumentTagCollector(document, 'spacy').save()
try:
service = EntityExtractStub(self.channel)
texts = self.text_iterator(document)
entities = service.Extract(texts)
for entity in entities.entities:
type_ = self.TYPES.get(entity.type)
if type_ is None:
continue
collector.emit(entity.label, type_, weight=entity.weight)
log.info('Extracted %s entities.', len(collector))
except self.Error as e:
log.warning("gRPC [%s]: %s", e.code(), e.details())
def extract(self, collector, document):
DocumentTagCollector(document, 'polyglot').save()
DocumentTagCollector(document, 'spacy').save()
try:
service = EntityExtractStub(self.channel)
texts = self.text_iterator(document)
entities = service.Extract(texts)
for entity in entities.entities:
type_ = self.TYPES.get(entity.type)
if type_ is None:
continue
collector.emit(entity.label, type_, weight=entity.weight)
log.info('Extracted %s entities.', len(collector))
except self.Error as exc:
log.exception("gRPC Error: %s", self.SERVICE)
self.reset_channel()
def extract(self, text, languages):
if text is None or len(text) < self.MIN_LENGTH:
return
texts = textwrap.wrap(text, self.MAX_LENGTH)
for text in texts:
for attempt in range(10):
try:
service = EntityExtractStub(self.channel)
req = Text(text=text, languages=languages)
for res in service.Extract(req):
clazz = self.TYPES.get(res.type)
yield (res.text, clazz, res.start, res.end)
break
except self.Error as e:
if e.code() == self.Status.RESOURCE_EXHAUSTED:
continue
log.warning("gRPC [%s]: %s", e.code(), e.details())
backoff(failures=attempt)
self.reset_channel()
def extract(self, collector, document):
DocumentTagCollector(document, 'polyglot').save()
DocumentTagCollector(document, 'spacy').save()
try:
service = EntityExtractStub(self.channel)
texts = self.text_iterator(document)
entities = service.Extract(texts)
for entity in entities.entities:
if entity.type == ExtractedEntity.COUNTRY:
document.add_country(entity.label)
if entity.type == ExtractedEntity.LANGUAGE:
document.add_language(entity.label)
type_ = self.TYPES.get(entity.type)
# log.info('%s: %s', entity.label, type_)
if type_ is not None:
collector.emit(entity.label, type_, weight=entity.weight)
log.info('Extracted %s entities.', len(collector))
except self.Error as e:
log.warning("gRPC [%s]: %s", e.code(), e.details())
def extract(self, collector, document):
languages = list(document.languages)
if not len(languages):
languages = [settings.DEFAULT_LANGUAGE]
try:
channel = grpc.insecure_channel(self.SERVICE)
service = EntityExtractStub(channel)
for text in document.texts:
if len(text) <= self.MIN_LENGTH:
continue
text = Text(text=text, languages=languages)
for entity in service.Extract(text):
type_ = self.TYPES.get(entity.type)
if type_ is None:
continue
collector.emit(entity.label, type_)
log.info('%s Extracted %s entities.', self.SERVICE, len(collector))
except grpc.RpcError as exc:
log.warning("gRPC Error: %s", exc)
import statistics
import grpc
import time
from alephclient.services.entityextract_pb2_grpc import EntityExtractStub
from alephclient.services.common_pb2 import Text
URL = 'localhost:50000'
channel = grpc.insecure_channel(URL)
service = EntityExtractStub(channel)
def generate():
with open('tests/fixtures/pace.txt', 'r', encoding='utf-8') as fh:
for line in fh:
yield Text(text=line, languages=['en'])
times = []
for i in range(1):
start = time.time()
entities = service.Extract(generate())
for entity in entities.entities:
print((entity.label, entity.weight, entity.type))
pass
end = time.time()
times.append(end - start)
print(statistics.mean(times))
import grpc
import time
import statistics
# from threading import Thread
from alephclient.services.common_pb2 import Text
from alephclient.services.entityextract_pb2_grpc import EntityExtractStub
URL = 'localhost:50000'
TEXT = 'There was Joseph Stalin working at the Kremlin in Moscow'
channel = grpc.insecure_channel(URL)
service = EntityExtractStub(channel)
times = []
for i in range(100):
start = time.time()
image = Text(text=TEXT, languages=['en'])
for ent in service.Extract(image):
print(ent.text)
end = time.time()
times.append(end - start)
print(statistics.mean(times))
# def target():
# channel = grpc.insecure_channel(URL)
# service = EntityExtractStub(channel)
# for i in range(300):
# image = Text(text=TEXT, languages=['en'])
# for ent in service.Extract(image):
# # print(ent.text)
# pass