def result_entities():
try:
stdin = click.get_text_stream('stdin')
stdout = click.get_text_stream('stdout')
while True:
result = read_object(stdin)
if result is None:
break
for entity in result.entities:
write_object(stdout, entity)
except BrokenPipeError:
pass
def pretty():
stdin = click.get_text_stream('stdin')
stdout = click.get_text_stream('stdout')
try:
while True:
entity = read_object(stdin)
if entity is None:
break
data = json.dumps(entity.to_dict(), indent=2)
stdout.write(data + '\n')
except BrokenPipeError:
pass
def auto_match(threshold):
try:
stdin = click.get_text_stream('stdin')
stdout = click.get_text_stream('stdout')
while True:
result = read_object(stdin)
if result is None:
break
if result.score > threshold:
recon = Recon(result.subject, result.candidate, Recon.MATCH)
write_object(stdout, recon)
except BrokenPipeError:
pass
def apply_recon(recon):
try:
linker = EntityLinker()
for recon in Recon.from_file(recon):
if recon.judgement == Recon.MATCH:
linker.add(recon.subject, recon.canonical)
stdin = click.get_text_stream('stdin')
stdout = click.get_text_stream('stdout')
while True:
entity = read_object(stdin)
if entity is None:
break
entity = linker.apply(entity)
write_object(stdout, entity)
except BrokenPipeError:
pass
def expand(enricher):
enricher = load_enricher(enricher)
try:
stdin = click.get_text_stream('stdin')
stdout = click.get_text_stream('stdout')
while True:
entity = read_object(stdin)
if entity is None:
break
result = enricher.expand_entity(entity)
write_object(stdout, result)
except BrokenPipeError:
pass
except Exception:
log.exception("Error during enrichment expansion")
finally:
enricher.close()
def aggregate():
buffer = {}
try:
stdin = click.get_text_stream('stdin')
while True:
entity = read_object(stdin)
if entity is None:
break
if entity.id in buffer:
buffer[entity.id].merge(entity)
else:
buffer[entity.id] = entity
stdout = click.get_text_stream('stdout')
for entity in buffer.values():
write_object(stdout, entity)
except BrokenPipeError:
pass
def export_rdf():
stdin = click.get_text_stream('stdin')
stdout = click.get_text_stream('stdout')
try:
while True:
entity = read_object(stdin)
if entity is None:
break
graph = Graph()
for link in entity.links:
triple = link.rdf()
if triple is None:
continue
graph.add(triple)
nt = graph.serialize(format='nt').strip()
stdout.write(nt.decode('utf-8') + '\n')
except BrokenPipeError:
pass
def filter_results(recon):
try:
matches = set()
for recon in Recon.from_file(recon):
if recon.judgement == Recon.MATCH:
matches.add(recon.subject)
stdin = click.get_text_stream('stdin')
stdout = click.get_text_stream('stdout')
while True:
result = read_object(stdin)
if result is None:
break
if result.candidate is None:
continue
if result.candidate.id in matches:
write_object(stdout, result)
except BrokenPipeError:
pass