pc.py

# http://theorangeduck.com/page/you-could-have-invented-parser-combinators

stream = list('abcdefghijklmnopqrstuvwxyz')


def literal(char):
    """
    Parser which matches a character.
    """
    def parser(stream):
        if stream[0] == char:
            return stream.pop(0)
        raise Exception()
    return parser


def or_(parser1, parser2):
    """
    Boolean or statement.
    """
    def parser(stream):
        try:
            return parser1(stream)
        except:
            return parser2(stream)
    return parser


def and_(parser1, parser2):
    """
    Boolean and statement.
    """
    def parser(stream):
        return [parser1(stream), parser2(stream)]
    return parser


def apply_(func, parser):
    """
    Parser which applies a func to the result
    of another parser.
    """
    def _parser(stream):
        return func(parser(stream))
    return _parser


def literal_to_int(char):
    """
    Parser which converts characters to ints.
    """
    return apply_(ord, literal(char))


if __name__ == "__main__":
    # We want to match tokens that begin with either 'ab'
    # or 'cd'.
    parser = or_(
        and_(literal_to_int('a'), literal_to_int('b')),
        and_(literal_to_int('c'), literal_to_int('d')))

    print parser(stream)
    print stream
    print parser(stream)
    print stream
    print parser(['c', 'd'])
    try:
        parser(stream)
    except Exception:
        # All good.
        pass

    next_parser = literal_to_int('e')
    print next_parser(stream)