88 lines
2.1 KiB
Python
88 lines
2.1 KiB
Python
|
#!/bin/env python
|
||
|
|
|
||
|
|
# 2.3 imports
|
||
|
|
from optparse import OptionParser
|
||
|
|
|
||
|
|
# our imports
|
||
|
|
import lex
|
||
|
|
|
||
|
|
class XMLGrammar(lex.Grammar):
|
||
|
|
GRAMMAR_LIST = [
|
||
|
|
{'name': 'comment',
|
||
|
|
'expr': r'''<!--(?:.| |\n)+?(?:-->|$)''',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'template',
|
||
|
|
'expr': r'''\[%(?:.| |\n)*?%\]''',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'ltb',
|
||
|
|
'expr': r'<![^>]*>',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'ltq',
|
||
|
|
'expr': r'<\?',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'gtq',
|
||
|
|
'expr': r'\?>',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'ltc',
|
||
|
|
'expr': r'</',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'gtc',
|
||
|
|
'expr': r'/>',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'lt',
|
||
|
|
'expr': r'<',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'gt',
|
||
|
|
'expr': r'>',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'nodevalue',
|
||
|
|
'expr': r'''(?<=>)(?:[^<]|\n)+?(?=<)''',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'whitespace',
|
||
|
|
'expr': r'''(?: |\n)+''',
|
||
|
|
'action': lex.silent},
|
||
|
|
|
||
|
|
{'name': 'namespace',
|
||
|
|
'expr': r'[a-zA-Z_]+:',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
#{'name': 'xi',
|
||
|
|
# 'expr': r'xi:',
|
||
|
|
# 'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'opentag',
|
||
|
|
'expr': r'(?:(?<=<)|(?<=xi:))[^ >\n/]+',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'attrvalue',
|
||
|
|
'expr': r'''(?<==)"(?:\\.|[^"\\])*(?:"|\\?$)|(?<==)'(?:\\.|[^'\\])*(?:'|\\?$)''',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'attrname',
|
||
|
|
'expr': r'[^ \n=>]+(?:(?==)|$)',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'closetag',
|
||
|
|
'expr': r'[^ =\n<>/]+',
|
||
|
|
'action': lex.make_token},
|
||
|
|
|
||
|
|
{'name': 'default',
|
||
|
|
'expr': r""".|\n""",
|
||
|
|
'action': lex.silent}
|
||
|
|
]
|
||
|
|
|
||
|
|
def _default_rules(self):
|
||
|
|
"""subclasses can override this to define defaults for a grammar"""
|
||
|
|
for rdir in XMLGrammar.GRAMMAR_LIST:
|
||
|
|
self.add_rule(**rdir)
|