comparison g_chart_parser.py @ 1:fb617586f4b2 draft

planemo upload commit a81826fe44f09a3710a35c183aa88b745aeec064-dirty

0:e991d4e60c17

    args = parser.parse_args()
    return args


def chart_parse(in_file, grammar_file, out_file):
    text = unicode(open(in_file, 'r').read(), errors='ignore')
    output = open(out_file, 'w')
    grammar_string = unicode(open(grammar_file, 'r').read(), errors='ignore')
    try:
        grammar = nltk.parse_cfg(grammar_string)
        parser = nltk.ChartParser(grammar)
        sentences = nltk.sent_tokenize(text)
        for sentence in sentences:
            words = nltk.word_tokenize(sentence)
            tree = parser.parse(words)
            output.write(tree.pprint())
            output.write('\n')
    except Exception, e:
        message = "Error with parsing. Check the input files are correct and the grammar contains every word in the input sequence. \n----\n" + str(e)
        sys.stderr.write(message)
        sys.exit()
    output.close()

if __name__ == '__main__':
    args = arguments()
    chart_parse(args.input, args.grammar, args.output)

1:fb617586f4b2

    args = parser.parse_args()
    return args


def chart_parse(in_file, grammar_file, out_file):
    with open(in_file, 'r') as fd:
        text = fd.read()

    with open(grammar_file, 'r') as fd:
        grammar_string = fd.read()

    try:
        grammar = nltk.CFG.fromstring(grammar_string)
        parser = nltk.ChartParser(grammar)
        sentences = nltk.sent_tokenize(text)
        with open(out_file, 'w') as output:
            for sentence in sentences:
                words = nltk.word_tokenize(sentence)
                trees = parser.parse(words)
                for t in trees:
                    output.write(t.pformat())
                    output.write('\n')

    except Exception as e:
        message = "Error with parsing. Check the input files are correct and the grammar contains every word in the input sequence. \n----\n" + str(e) + "\n"
        sys.stderr.write(message)
        sys.exit()
    output.close()

if __name__ == '__main__':
    args = arguments()
    chart_parse(args.input, args.grammar, args.output)