diff options
Diffstat (limited to 'python/pkg/cdec/sa/extract.py')
| -rw-r--r-- | python/pkg/cdec/sa/extract.py | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/python/pkg/cdec/sa/extract.py b/python/pkg/cdec/sa/extract.py index 782bed8b..b6502c52 100644 --- a/python/pkg/cdec/sa/extract.py +++ b/python/pkg/cdec/sa/extract.py @@ -8,6 +8,7 @@ import logging import signal import multiprocessing as mp import cdec.sa +from cdec.sa._sa import monitor_cpu extractor, prefix = None, None online, compress = False, False @@ -62,6 +63,7 @@ def extract(inp): return '<seg grammar="{}" id="{}">{}</seg>{}'.format(grammar_file, i, sentence, suffix) def main(): + global online logging.basicConfig(level=logging.INFO) parser = argparse.ArgumentParser(description='Extract grammars from a compiled corpus.') parser.add_argument('-c', '--config', required=True, @@ -87,7 +89,10 @@ def main(): sys.stderr.write('Error: feature definition file <{}>' ' should be a python module\n'.format(featdef)) sys.exit(1) - + + online = args.online + + start_time = monitor_cpu() if args.jobs > 1: logging.info('Starting %d workers; chunk size: %d', args.jobs, args.chunksize) pool = mp.Pool(args.jobs, make_extractor, (args,)) @@ -101,5 +106,8 @@ def main(): for output in map(extract, enumerate(sys.stdin)): print(output) + stop_time = monitor_cpu() + logging.info("Overall extraction step took %f seconds", stop_time - start_time) + if __name__ == '__main__': main() |