app/django/utils/simplejson/scanner.py
author Mario Ferraro <fadinlight@gmail.com>
Sun, 15 Nov 2009 22:12:20 +0100
changeset 3093 d1be59b6b627
parent 323 ff1a9aa48cfd
permissions -rw-r--r--
GMaps related JS changed to use new google namespace. Google is going to change permanently in the future the way to load its services, so better stay safe. Also this commit shows uses of the new melange.js module. Fixes Issue 634.

"""
Iterator based sre token scanner
"""
import re
from re import VERBOSE, MULTILINE, DOTALL
import sre_parse
import sre_compile
import sre_constants
from sre_constants import BRANCH, SUBPATTERN

__all__ = ['Scanner', 'pattern']

FLAGS = (VERBOSE | MULTILINE | DOTALL)

class Scanner(object):
    def __init__(self, lexicon, flags=FLAGS):
        self.actions = [None]
        # Combine phrases into a compound pattern
        s = sre_parse.Pattern()
        s.flags = flags
        p = []
        for idx, token in enumerate(lexicon):
            phrase = token.pattern
            try:
                subpattern = sre_parse.SubPattern(s,
                    [(SUBPATTERN, (idx + 1, sre_parse.parse(phrase, flags)))])
            except sre_constants.error:
                raise
            p.append(subpattern)
            self.actions.append(token)

        s.groups = len(p) + 1 # NOTE(guido): Added to make SRE validation work
        p = sre_parse.SubPattern(s, [(BRANCH, (None, p))])
        self.scanner = sre_compile.compile(p)

    def iterscan(self, string, idx=0, context=None):
        """
        Yield match, end_idx for each match
        """
        match = self.scanner.scanner(string, idx).match
        actions = self.actions
        lastend = idx
        end = len(string)
        while True:
            m = match()
            if m is None:
                break
            matchbegin, matchend = m.span()
            if lastend == matchend:
                break
            action = actions[m.lastindex]
            if action is not None:
                rval, next_pos = action(m, context)
                if next_pos is not None and next_pos != matchend:
                    # "fast forward" the scanner
                    matchend = next_pos
                    match = self.scanner.scanner(string, matchend).match
                yield rval, matchend
            lastend = matchend


def pattern(pattern, flags=FLAGS):
    def decorator(fn):
        fn.pattern = pattern
        fn.regex = re.compile(pattern, flags)
        return fn
    return decorator