diff options
Diffstat (limited to 'python/cdec/sa/extract.py')
-rw-r--r-- | python/cdec/sa/extract.py | 32 |
1 files changed, 0 insertions, 32 deletions
diff --git a/python/cdec/sa/extract.py b/python/cdec/sa/extract.py deleted file mode 100644 index c6da5e9d..00000000 --- a/python/cdec/sa/extract.py +++ /dev/null @@ -1,32 +0,0 @@ -#!/usr/bin/env python -import sys -import os -import argparse -import logging -import configobj -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(configobj.ConfigObj(args.config, unrepr=True)) - 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('<seg grammar="{0}">{1}</seg>'.format(grammar_file, sentence)) - -if __name__ == '__main__': - main() |