ply-path.py

#!/usr/bin/env python

"""ply-path -- analysis of an path expression using PLY

Example: ply-path.py 'child::para[position()>1]'
"""

import sys
from ply import lex

def usage():
    print __doc__
    print 'usage: %s expression' % sys.argv[0]
    sys.exit(1)

def expression():
    """Subset of <http://www.w3.org/TR/xpath/#exprlex>."""

    tokens = ['BRACKET', 'ABBR', 'NUM', 'LITERAL', 'VAR', 'NAME']

    reserved = {
        'ancestor': 'AXIS',
        'attribute': 'AXIS',
        'child': 'AXIS',
        'descendant': 'AXIS',
        'parent': 'AXIS',

        'and': 'OP',
        'or': 'OP',
        'mod': 'OP',
        'div': 'OP',
    }

    tokens.extend(set(reserved.itervalues()))

    t_BRACKET = r'\(|\)|\[|\]'
    t_ABBR = r'\.{1,2}|@|,|::'
    t_LITERAL = r'"([^"]*)"|\'([^\']*)\''
    t_OP = r'\*|/|//|\||\+|\-|=|!=|<|<=|>|>='
    t_VAR = r'\$([a-zA-Z][\w\-]*)'

    def t_NUM(t):
        r'\d+(?:\.\d*)?|\.\d+'
        t.value = (float if '.' in t.value else int)(t.value)
        return t

    def t_NAME(t):
        ur'\*|[a-zA-Z][\w\-]*'
        t.type = reserved.get(t.value, 'NAME')
        return t

    t_ignore = ' \t\n\r'

    def t_error(t):
        print 'Illegal character %r at %r' % (t.value[0], t.value[0:15])

    return lex.lex()

def main(path):
    expr = expression()
    expr.input(path)
    while True:
        tok = expr.token()
        if not tok:
            break
        print tok


if __name__ == '__main__':
    if len(sys.argv) != 2:
        usage()
    main(sys.argv[1])