def zmq_coroutine(context, dealer_url, doc_class=None, automagic=False):
# FIXME: reduce overhead of reader/writer creation
ostream = io.BytesIO()
socket = context.socket(zmq.REP)
socket.connect(dealer_url)
while True:
msg = socket.recv()
istream = io.BytesIO(msg)
istream.seek(0)
reader = Reader(istream, doc_class, automagic)
writer = Writer(ostream, reader.doc_schema)
for doc in reader:
res = yield (doc)
writer.write(res or doc)
ostream.seek(0)
socket.send(ostream.getvalue())
ostream.truncate(0)
def zmq_coroutine(context, dealer_url, doc_class=None, automagic=False):
# FIXME: reduce overhead of reader/writer creation
ostream = io.BytesIO()
socket = context.socket(zmq.REP)
socket.connect(dealer_url)
while True:
msg = socket.recv()
istream = io.BytesIO(msg)
istream.seek(0)
reader = Reader(istream, doc_class, automagic)
writer = Writer(ostream, reader.doc_schema)
for doc in reader:
res = yield(doc)
writer.write(res or doc)
ostream.seek(0)
socket.send(ostream.getvalue())
ostream.truncate(0)
def stream_coroutine(istream, ostream, doc_class=None, automagic=False):
reader = Reader(istream, doc_class, automagic)
writer = Writer(ostream, reader.doc_schema)
for doc in reader:
res = yield (doc)
writer.write(res or doc)
def stream_coroutine(istream, ostream, doc_class=None, automagic=False):
reader = Reader(istream, doc_class, automagic)
writer = Writer(ostream, reader.doc_schema)
for doc in reader:
res = yield(doc)
writer.write(res or doc)
def stream_reader_writer(self):
docs, schema = self.get_reader_and_schema()
return docs, Writer(self.args.out_stream, schema)
ntokens += 1
prev_raw = raw
sentence = doc.sents.create(span=slice(sent_start, ntokens))
nsents += 1
paragraph = doc.paras.create(span=slice(paragraph_start, nsents))
return doc
if __name__ == '__main__':
# grep-like over HTTP response
import argparse
import sys
sys.path.append('.')
ap = argparse.ArgumentParser(description="Reads an input warc file, and converts it to docrep on stdout")
ap.add_argument('input_warc')
ap.add_argument('-a', '--annotations', default=None, help='FACC annotation file')
ap.add_argument('-d', '--debug', default=None, help='Print debug log information')
args = ap.parse_args()
sys.stdout = sys.stdout.detach() # make stdout binary (for py3)
writer = Writer(sys.stdout, Doc)
log.info("Processing {}".format(args.input_warc))
debug = None
if args.debug is not None:
debug = open(args.debug, "a")
try:
for doc in process(args.input_warc, args.annotations, debug):
writer.write(doc)
except Exception as e:
print(str(e), file=sys.stderr)
sys.exit(1)
log.info("Finished processing {}".format(args.input_warc))
