Beispiel #1
0
def main():
    parser = utils.arg_parser()
    parser.add_argument('-v',
                        '--version',
                        action='version',
                        version=('artext %s' % __version__))
    args = parser.parse_args()

    conf = config.Config()
    conf.error_rate = args.error_rate
    conf.path_protected_tokens = args.protected_tokens
    conf.samples = args.samples
    conf.separator = args.separator

    artxt = Artext(config=conf)

    with open(args.source, 'r') as fin, open(args.output, 'w') as fout:
        if args.level == 'sent':
            for line in fin:
                noises = artxt.noise_sentence(line)
                fout.write("{}\n".format(args.separator.join(noises)))
        elif args.level == 'doc':
            for line in fin:
                noises = artxt.noise_document(line)
                fout.write("{}\n".format(args.separator.join(noises)))
Beispiel #2
0
 def test_args(self):
     from artext import utils
     parser = utils.arg_parser()
     args = parser.parse_args('-src test -out test -n 5'.split())
     self.assertEqual(args.source, 'test')
     self.assertEqual(args.output, 'test')
     self.assertEqual(args.samples, 5)
Beispiel #3
0
def test_args():
    from artext import utils
    parser = utils.arg_parser()
    args = parser.parse_args('-src test -out test -n 5'.split())
    assert args.source == 'test'
    assert args.output == 'test'
    assert args.samples == 5
Beispiel #4
0
from artext import config, utils, Artext

if __name__ == "__main__":
    parser = utils.arg_parser()
    args = parser.parse_args()

    conf = config.Config()
    conf.error_rate = args.error_rate
    conf.path_protected_tokens = args.protected_tokens
    conf.samples = args.samples

    artxt = Artext(config=conf)

    with open(args.source, 'r') as fin, open(args.output, 'w') as fout:
        if args.level == 'sent':
            for line in fin:
                noises = artxt.noise_sentence(line)
                fout.write("{}\n".format(args.separator.join(noises)))
        elif args.level == 'doc':
            for line in fin:
                noises = artxt.noise_document(line)
                fout.write("{}\n".format(args.separator.join(noises)))