diff options
Diffstat (limited to 'addon-sdk/source/python-lib/simplejson/scanner.py')
-rw-r--r-- | addon-sdk/source/python-lib/simplejson/scanner.py | 67 |
1 files changed, 67 insertions, 0 deletions
diff --git a/addon-sdk/source/python-lib/simplejson/scanner.py b/addon-sdk/source/python-lib/simplejson/scanner.py new file mode 100644 index 000000000..2a18390d0 --- /dev/null +++ b/addon-sdk/source/python-lib/simplejson/scanner.py @@ -0,0 +1,67 @@ +""" +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
\ No newline at end of file |