Mercurial Repositories > eric / file comparison
comparison: eric6/ThirdParty/Pygments/pygments/lexers/rdf.py
eric6/ThirdParty/Pygments/pygments/lexers/rdf.py
- changeset 7547
- 21b0534faebc
- parent 6942
- 2602857055c5
- child 7701
- 25f42e208e08
equal
deleted
inserted
replaced
| 7546:bf5f777260a6 | 7547:21b0534faebc |
|---|---|
| 3 pygments.lexers.rdf | 3 pygments.lexers.rdf |
| 4 ~~~~~~~~~~~~~~~~~~~ | 4 ~~~~~~~~~~~~~~~~~~~ |
| 5 | 5 |
| 6 Lexers for semantic web and RDF query languages and markup. | 6 Lexers for semantic web and RDF query languages and markup. |
| 7 | 7 |
| 8 :copyright: Copyright 2006-2017 by the Pygments team, see AUTHORS. | 8 :copyright: Copyright 2006-2019 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 | 13 |
| 14 from pygments.lexer import RegexLexer, bygroups, default | 14 from pygments.lexer import RegexLexer, bygroups, default |
| 15 from pygments.token import Keyword, Punctuation, String, Number, Operator, Generic, \ | 15 from pygments.token import Keyword, Punctuation, String, Number, Operator, Generic, \ |
| 16 Whitespace, Name, Literal, Comment, Text | 16 Whitespace, Name, Literal, Comment, Text |
| 17 | 17 |
| 18 __all__ = ['SparqlLexer', 'TurtleLexer'] | 18 __all__ = ['SparqlLexer', 'TurtleLexer', 'ShExCLexer'] |
| 19 | 19 |
| 20 | 20 |
| 21 class SparqlLexer(RegexLexer): | 21 class SparqlLexer(RegexLexer): |
| 22 """ | 22 """ |
| 23 Lexer for `SPARQL <http://www.w3.org/TR/rdf-sparql-query/>`_ query language. | 23 Lexer for `SPARQL <http://www.w3.org/TR/rdf-sparql-query/>`_ query language. |
| 266 | 266 |
| 267 default('#pop:2'), | 267 default('#pop:2'), |
| 268 | 268 |
| 269 ], | 269 ], |
| 270 } | 270 } |
| 271 | |
| 272 # Turtle and Tera Term macro files share the same file extension | |
| 273 # but each has a recognizable and distinct syntax. | |
| 274 def analyse_text(text): | |
| 275 for t in ('@base ', 'BASE ', '@prefix ', 'PREFIX '): | |
| 276 if re.search(r'^\s*%s' % t, text): | |
| 277 return 0.80 | |
| 278 | |
| 279 | |
| 280 class ShExCLexer(RegexLexer): | |
| 281 """ | |
| 282 Lexer for `ShExC <https://shex.io/shex-semantics/#shexc>`_ shape expressions language syntax. | |
| 283 """ | |
| 284 name = 'ShExC' | |
| 285 aliases = ['shexc', 'shex'] | |
| 286 filenames = ['*.shex'] | |
| 287 mimetypes = ['text/shex'] | |
| 288 | |
| 289 # character group definitions :: | |
| 290 | |
| 291 PN_CHARS_BASE_GRP = (u'a-zA-Z' | |
| 292 u'\u00c0-\u00d6' | |
| 293 u'\u00d8-\u00f6' | |
| 294 u'\u00f8-\u02ff' | |
| 295 u'\u0370-\u037d' | |
| 296 u'\u037f-\u1fff' | |
| 297 u'\u200c-\u200d' | |
| 298 u'\u2070-\u218f' | |
| 299 u'\u2c00-\u2fef' | |
| 300 u'\u3001-\ud7ff' | |
| 301 u'\uf900-\ufdcf' | |
| 302 u'\ufdf0-\ufffd') | |
| 303 | |
| 304 PN_CHARS_U_GRP = (PN_CHARS_BASE_GRP + '_') | |
| 305 | |
| 306 PN_CHARS_GRP = (PN_CHARS_U_GRP + | |
| 307 r'\-' + | |
| 308 r'0-9' + | |
| 309 u'\u00b7' + | |
| 310 u'\u0300-\u036f' + | |
| 311 u'\u203f-\u2040') | |
| 312 | |
| 313 HEX_GRP = '0-9A-Fa-f' | |
| 314 | |
| 315 PN_LOCAL_ESC_CHARS_GRP = r"_~.\-!$&'()*+,;=/?#@%" | |
| 316 | |
| 317 # terminal productions :: | |
| 318 | |
| 319 PN_CHARS_BASE = '[' + PN_CHARS_BASE_GRP + ']' | |
| 320 | |
| 321 PN_CHARS_U = '[' + PN_CHARS_U_GRP + ']' | |
| 322 | |
| 323 PN_CHARS = '[' + PN_CHARS_GRP + ']' | |
| 324 | |
| 325 HEX = '[' + HEX_GRP + ']' | |
| 326 | |
| 327 PN_LOCAL_ESC_CHARS = '[' + PN_LOCAL_ESC_CHARS_GRP + ']' | |
| 328 | |
| 329 UCHAR_NO_BACKSLASH = '(?:u' + HEX + '{4}|U' + HEX + '{8})' | |
| 330 | |
| 331 UCHAR = r'\\' + UCHAR_NO_BACKSLASH | |
| 332 | |
| 333 IRIREF = r'<(?:[^\x00-\x20<>"{}|^`\\]|' + UCHAR + ')*>' | |
| 334 | |
| 335 BLANK_NODE_LABEL = '_:[0-9' + PN_CHARS_U_GRP + '](?:[' + PN_CHARS_GRP + \ | |
| 336 '.]*' + PN_CHARS + ')?' | |
| 337 | |
| 338 PN_PREFIX = PN_CHARS_BASE + '(?:[' + PN_CHARS_GRP + '.]*' + PN_CHARS + ')?' | |
| 339 | |
| 340 PERCENT = '%' + HEX + HEX | |
| 341 | |
| 342 PN_LOCAL_ESC = r'\\' + PN_LOCAL_ESC_CHARS | |
| 343 | |
| 344 PLX = '(?:' + PERCENT + ')|(?:' + PN_LOCAL_ESC + ')' | |
| 345 | |
| 346 PN_LOCAL = ('(?:[' + PN_CHARS_U_GRP + ':0-9' + ']|' + PLX + ')' + | |
| 347 '(?:(?:[' + PN_CHARS_GRP + '.:]|' + PLX + ')*(?:[' + | |
| 348 PN_CHARS_GRP + ':]|' + PLX + '))?') | |
| 349 | |
| 350 EXPONENT = r'[eE][+-]?\d+' | |
| 351 | |
| 352 # Lexer token definitions :: | |
| 353 | |
| 354 tokens = { | |
| 355 'root': [ | |
| 356 (r'\s+', Text), | |
| 357 # keywords :: | |
| 358 (r'(?i)(base|prefix|start|external|' | |
| 359 r'literal|iri|bnode|nonliteral|length|minlength|maxlength|' | |
| 360 r'mininclusive|minexclusive|maxinclusive|maxexclusive|' | |
| 361 r'totaldigits|fractiondigits|' | |
| 362 r'closed|extra)\b', Keyword), | |
| 363 (r'(a)\b', Keyword), | |
| 364 # IRIs :: | |
| 365 ('(' + IRIREF + ')', Name.Label), | |
| 366 # blank nodes :: | |
| 367 ('(' + BLANK_NODE_LABEL + ')', Name.Label), | |
| 368 # prefixed names :: | |
| 369 (r'(' + PN_PREFIX + r')?(\:)(' + PN_LOCAL + ')?', | |
| 370 bygroups(Name.Namespace, Punctuation, Name.Tag)), | |
| 371 # boolean literals :: | |
| 372 (r'(true|false)', Keyword.Constant), | |
| 373 # double literals :: | |
| 374 (r'[+\-]?(\d+\.\d*' + EXPONENT + r'|\.?\d+' + EXPONENT + ')', Number.Float), | |
| 375 # decimal literals :: | |
| 376 (r'[+\-]?(\d+\.\d*|\.\d+)', Number.Float), | |
| 377 # integer literals :: | |
| 378 (r'[+\-]?\d+', Number.Integer), | |
| 379 # operators :: | |
| 380 (r'[@|$&=*+?^\-~]', Operator), | |
| 381 # operator keywords :: | |
| 382 (r'(?i)(and|or|not)\b', Operator.Word), | |
| 383 # punctuation characters :: | |
| 384 (r'[(){}.;,:^\[\]]', Punctuation), | |
| 385 # line comments :: | |
| 386 (r'#[^\n]*', Comment), | |
| 387 # strings :: | |
| 388 (r'"""', String, 'triple-double-quoted-string'), | |
| 389 (r'"', String, 'single-double-quoted-string'), | |
| 390 (r"'''", String, 'triple-single-quoted-string'), | |
| 391 (r"'", String, 'single-single-quoted-string'), | |
| 392 ], | |
| 393 'triple-double-quoted-string': [ | |
| 394 (r'"""', String, 'end-of-string'), | |
| 395 (r'[^\\]+', String), | |
| 396 (r'\\', String, 'string-escape'), | |
| 397 ], | |
| 398 'single-double-quoted-string': [ | |
| 399 (r'"', String, 'end-of-string'), | |
| 400 (r'[^"\\\n]+', String), | |
| 401 (r'\\', String, 'string-escape'), | |
| 402 ], | |
| 403 'triple-single-quoted-string': [ | |
| 404 (r"'''", String, 'end-of-string'), | |
| 405 (r'[^\\]+', String), | |
| 406 (r'\\', String.Escape, 'string-escape'), | |
| 407 ], | |
| 408 'single-single-quoted-string': [ | |
| 409 (r"'", String, 'end-of-string'), | |
| 410 (r"[^'\\\n]+", String), | |
| 411 (r'\\', String, 'string-escape'), | |
| 412 ], | |
| 413 'string-escape': [ | |
| 414 (UCHAR_NO_BACKSLASH, String.Escape, '#pop'), | |
| 415 (r'.', String.Escape, '#pop'), | |
| 416 ], | |
| 417 'end-of-string': [ | |
| 418 (r'(@)([a-zA-Z]+(?:-[a-zA-Z0-9]+)*)', | |
| 419 bygroups(Operator, Name.Function), '#pop:2'), | |
| 420 (r'\^\^', Operator, '#pop:2'), | |
| 421 default('#pop:2'), | |
| 422 ], | |
| 423 } |
