def __init__(self,
conf_path: Union[str, Path],
output_directory: Union[str, Path],
*,
events_address: Optional[str] = None,
events_client: EventsClient = None,
serializer: Optional[str] = None,
include_label_text: bool = False):
if events_address == 'None' or events_address == 'none' or events_address == 'null' or events_address == '':
events_address = None
if events_client is not None:
self.close_client = False
self.events_client = events_client
else:
self.close_client = True
self.events_client = EventsClient(address=events_address)
self.pipeline = Pipeline.from_yaml_file(conf_path)
if serializer == 'None':
serializer = None
if serializer is not None:
serialization_proc = SerializationProcessor(
get_serializer(serializer),
output_directory,
include_label_text=include_label_text)
ser_comp = LocalProcessor(serialization_proc,
component_id='serializer',
client=self.events_client)
self.pipeline.append(ser_comp)
def __init__(self,
conf_path: Union[str, Path],
output_directory: Union[str, Path],
*,
events_addresses: Optional[str] = None,
serializer: Optional[str] = None,
include_label_text: bool = False):
self.pipeline = Pipeline.from_yaml_file(conf_path)
if events_addresses is not None:
self.pipeline.events_address = events_addresses
if serializer == 'None':
serializer = None
if serializer is not None:
serialization_proc = SerializationProcessor(
get_serializer(serializer),
output_directory,
include_label_text=include_label_text)
ser_comp = LocalProcessor(serialization_proc,
component_id='serializer')
self.pipeline.append(ser_comp)
def __init__(self,
conf: PipelineConf,
*,
events_client: EventsClient = None):
conf.populate_addresses()
if events_client is not None:
self.close_client = False
self.events_client = events_client
elif conf.events_address is not None:
self.close_client = True
self.events_client = EventsClient(address=conf.events_address)
else:
raise ValueError("Events client or address not specified.")
pipeline = [(conf.sentences_id, conf.sentences_address),
(conf.section_headers_id, conf.section_headers_address),
(conf.tagger_id, conf.tagger_address),
(conf.acronyms_id, conf.acronyms_address),
(conf.concepts_id, conf.concepts_address),
(conf.negation_id, conf.negation_address),
(conf.selective_dependencies_id,
conf.selective_dependencies_address),
(conf.deepen_id, conf.deepen_address)]
if conf.use_discovery:
self.pipeline = Pipeline(
*[RemoteProcessor(identifier) for identifier, _ in pipeline])
else:
self.pipeline = Pipeline(*[
RemoteProcessor(identifier, address=addr)
for identifier, addr in pipeline
])
if conf.serializer is not None:
serialization_proc = SerializationProcessor(
get_serializer(conf.serializer),
conf.output_directory,
include_label_text=conf.include_label_text)
ser_comp = LocalProcessor(serialization_proc,
component_id='serializer',
client=self.events_client)
self.pipeline.append(ser_comp)