class DefaultPipeline:
"""The biomedicus default pipeline for processing clinical documents.
Attributes
events_client (mtap.EventsClient): An MTAP events client used by the pipeline.
pipeline (mtap.Pipeline): An MTAP pipeline to use to process documents.
"""
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)
def process_text(self, text: str, *, event_id: str = None) -> ProcessingResult:
with Event(event_id=event_id, client=self.events_client) as event:
document = event.create_document('plaintext', text=text)
f = self.pipeline.run(document)
return f
def __enter__(self):
return self
def __exit__(self, exc_type, exc_val, exc_tb):
self.pipeline.close()
if self.close_client:
self.events_client.close()
class DefaultPipeline:
def __init__(self,
conf: DefaultPipelineConf,
*,
events_client: EventsClient = None):
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.tagger_id, conf.tagger_address),
(conf.acronyms_id, conf.acronyms_address),
(conf.concepts_id, conf.concepts_address),
(conf.negation_id, conf.negation_address)]
if conf.use_discovery:
self.pipeline = Pipeline(
*[RemoteProcessor(identifier) for identifier, _ in pipeline],
n_threads=conf.threads)
else:
self.pipeline = Pipeline(*[
RemoteProcessor(identifier, address=addr)
for identifier, addr in pipeline
],
n_threads=conf.threads)
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)
def process_text(self,
text: str,
*,
event_id: str = None) -> ProcessingResult:
with Event(event_id=event_id, client=self.events_client) as event:
document = event.create_document('plaintext', text=text)
f = self.pipeline.run(document)
return f
def __enter__(self):
return self
def __exit__(self, exc_type, exc_val, exc_tb):
self.pipeline.close()
if self.close_client:
self.events_client.close()
class DefaultPipeline:
"""The biomedicus default pipeline for processing clinical documents.
Attributes
events_client (mtap.EventsClient): An MTAP events client used by the pipeline.
pipeline (mtap.Pipeline): An MTAP pipeline to use to process documents.
"""
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.tagger_id, conf.tagger_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
])
def process_text(self,
text: str,
*,
event_id: str = None) -> ProcessingResult:
with Event(event_id=event_id, client=self.events_client) as event:
document = event.create_document('plaintext', text=text)
f = self.pipeline.run(document)
return f
def __enter__(self):
return self
def __exit__(self, exc_type, exc_val, exc_tb):
self.pipeline.close()
if self.close_client:
self.events_client.close()