From 455ca8bc1406ec2f6554fce9be7488bb3cca75dd Mon Sep 17 00:00:00 2001 From: Victor Chahuneau Date: Fri, 27 Jul 2012 23:33:45 -0400 Subject: [python] Move python files to avoid pythonpath conflicts --- python/pkg/cdec/sa/extract.py | 31 +++++++++++++++++++++++++++++++ 1 file changed, 31 insertions(+) create mode 100644 python/pkg/cdec/sa/extract.py (limited to 'python/pkg/cdec/sa/extract.py') diff --git a/python/pkg/cdec/sa/extract.py b/python/pkg/cdec/sa/extract.py new file mode 100644 index 00000000..918aa3bb --- /dev/null +++ b/python/pkg/cdec/sa/extract.py @@ -0,0 +1,31 @@ +#!/usr/bin/env python +import sys +import os +import argparse +import logging +import cdec.sa + +def main(): + logging.basicConfig(level=logging.INFO) + parser = argparse.ArgumentParser(description='Extract grammars from a compiled corpus.') + parser.add_argument('-c', '--config', required=True, + help='Extractor configuration') + parser.add_argument('-g', '--grammars', required=True, + help='Grammar output path') + args = parser.parse_args() + + if not os.path.exists(args.grammars): + os.mkdir(args.grammars) + + extractor = cdec.sa.GrammarExtractor(args.config) + for i, sentence in enumerate(sys.stdin): + sentence = sentence[:-1] + grammar_file = os.path.join(args.grammars, 'grammar.{0}'.format(i)) + with open(grammar_file, 'w') as output: + for rule in extractor.grammar(sentence): + output.write(str(rule)+'\n') + grammar_file = os.path.abspath(grammar_file) + print('{1}'.format(grammar_file, sentence)) + +if __name__ == '__main__': + main() -- cgit v1.2.3