From 83ca181637dfc34804649e1d342e3cb3ee59b5df Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Thu, 16 Apr 2020 13:46:30 -0700 Subject: batch/multiprocess for ZipfilePusher --- python/grobid_tool.py | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'python/grobid_tool.py') diff --git a/python/grobid_tool.py b/python/grobid_tool.py index dc9780d..a2d74a1 100755 --- a/python/grobid_tool.py +++ b/python/grobid_tool.py @@ -55,8 +55,14 @@ def run_extract_cdx(args): def run_extract_zipfile(args): grobid_client = GrobidClient(host_url=args.grobid_host) - worker = GrobidBlobWorker(grobid_client, sink=args.sink) - pusher = ZipfilePusher(worker, args.zip_file) + if args.jobs > 1: + print("multi-processing: {}".format(args.jobs), file=sys.stderr) + worker = GrobidBlobWorker(grobid_client, sink=None) + multi_worker = MultiprocessWrapper(worker, args.sink, jobs=args.jobs) + pusher = ZipfilePusher(multi_worker, args.zip_file, batch_size=args.jobs) + else: + worker = GrobidBlobWorker(grobid_client, sink=args.sink) + pusher = ZipfilePusher(worker, args.zip_file) pusher.run() def run_transform(args): -- cgit v1.2.3