Mercurial Repositories > eric / file comparison
comparison: ThirdParty/Pygments/pygments/util.py
ThirdParty/Pygments/pygments/util.py
- changeset 2426
- da76c71624de
- parent 1705
- b0fbc9300f2b
- child 2525
- 8b507a9a2d40
equal
deleted
inserted
replaced
| 2425:ace8a08028f3 | 2426:da76c71624de |
|---|---|
| 3 pygments.util | 3 pygments.util |
| 4 ~~~~~~~~~~~~~ | 4 ~~~~~~~~~~~~~ |
| 5 | 5 |
| 6 Utility functions. | 6 Utility functions. |
| 7 | 7 |
| 8 :copyright: Copyright 2006-2012 by the Pygments team, see AUTHORS. | 8 :copyright: Copyright 2006-2013 by the Pygments team, see AUTHORS. |
| 9 :license: BSD, see LICENSE for details. | 9 :license: BSD, see LICENSE for details. |
| 10 """ | 10 """ |
| 11 | 11 |
| 12 import re | 12 import re |
| 13 import sys | 13 import sys |
| 203 if m is not None: | 203 if m is not None: |
| 204 return True | 204 return True |
| 205 rv = tag_re.search(text[:1000]) is not None | 205 rv = tag_re.search(text[:1000]) is not None |
| 206 _looks_like_xml_cache[key] = rv | 206 _looks_like_xml_cache[key] = rv |
| 207 return rv | 207 return rv |
| 208 | |
| 209 # Python narrow build compatibility | |
| 210 | |
| 211 def _surrogatepair(c): | |
| 212 return (0xd7c0 + (c >> 10), (0xdc00 + (c & 0x3ff))) | |
| 213 | |
| 214 def unirange(a, b): | |
| 215 """ | |
| 216 Returns a regular expression string to match the given non-BMP range. | |
| 217 """ | |
| 218 if b < a: | |
| 219 raise ValueError("Bad character range") | |
| 220 if a < 0x10000 or b < 0x10000: | |
| 221 raise ValueError("unirange is only defined for non-BMP ranges") | |
| 222 | |
| 223 if sys.maxunicode > 0xffff: | |
| 224 # wide build | |
| 225 return '[%s-%s]' % (chr(a), chr(b)) | |
| 226 else: | |
| 227 # narrow build stores surrogates, and the 're' module handles them | |
| 228 # (incorrectly) as characters. Since there is still ordering among | |
| 229 # these characters, expand the range to one that it understands. Some | |
| 230 # background in http://bugs.python.org/issue3665 and | |
| 231 # http://bugs.python.org/issue12749 | |
| 232 # | |
| 233 # Additionally, the lower constants are using unichr rather than | |
| 234 # literals because jython [which uses the wide path] can't load this | |
| 235 # file if they are literals. | |
| 236 ah, al = _surrogatepair(a) | |
| 237 bh, bl = _surrogatepair(b) | |
| 238 if ah == bh: | |
| 239 return '(?:%s[%s-%s])' % (chr(ah), chr(al), chr(bl)) | |
| 240 else: | |
| 241 buf = [] | |
| 242 buf.append('%s[%s-%s]' % | |
| 243 (chr(ah), chr(al), | |
| 244 ah == bh and chr(bl) or chr(0xdfff))) | |
| 245 if ah - bh > 1: | |
| 246 buf.append('[%s-%s][%s-%s]' % | |
| 247 chr(ah+1), chr(bh-1), chr(0xdc00), chr(0xdfff)) | |
| 248 if ah != bh: | |
| 249 buf.append('%s[%s-%s]' % | |
| 250 (chr(bh), chr(0xdc00), chr(bl))) | |
| 251 | |
| 252 return '(?:' + '|'.join(buf) + ')' | |
| 208 | 253 |
| 209 # Python 2/3 compatibility | 254 # Python 2/3 compatibility |
| 210 | 255 |
| 211 if sys.version_info < (3,0): | 256 if sys.version_info < (3,0): |
| 212 b = bytes = str | 257 b = bytes = str |
