jvm.py
847 lines
| 39.4 KiB
| text/x-python
|
PythonLexer
Jeandet Alexis
|
r0 | # -*- coding: utf-8 -*- | ||
""" | ||||
pygments.lexers.jvm | ||||
~~~~~~~~~~~~~~~~~~~ | ||||
Pygments lexers for JVM languages. | ||||
:copyright: Copyright 2006-2012 by the Pygments team, see AUTHORS. | ||||
:license: BSD, see LICENSE for details. | ||||
""" | ||||
import re | ||||
from pygments.lexer import Lexer, RegexLexer, include, bygroups, using, \ | ||||
this | ||||
from pygments.token import Text, Comment, Operator, Keyword, Name, String, \ | ||||
Number, Punctuation | ||||
from pygments.util import get_choice_opt | ||||
from pygments import unistring as uni | ||||
__all__ = ['JavaLexer', 'ScalaLexer', 'GosuLexer', 'GosuTemplateLexer', | ||||
'GroovyLexer', 'IokeLexer', 'ClojureLexer', 'KotlinLexer'] | ||||
class JavaLexer(RegexLexer): | ||||
""" | ||||
For `Java <http://www.sun.com/java/>`_ source code. | ||||
""" | ||||
name = 'Java' | ||||
aliases = ['java'] | ||||
filenames = ['*.java'] | ||||
mimetypes = ['text/x-java'] | ||||
flags = re.MULTILINE | re.DOTALL | ||||
#: optional Comment or Whitespace | ||||
_ws = r'(?:\s|//.*?\n|/[*].*?[*]/)+' | ||||
tokens = { | ||||
'root': [ | ||||
# method names | ||||
(r'^(\s*(?:[a-zA-Z_][a-zA-Z0-9_\.\[\]]*\s+)+?)' # return arguments | ||||
r'([a-zA-Z_][a-zA-Z0-9_]*)' # method name | ||||
r'(\s*)(\()', # signature start | ||||
bygroups(using(this), Name.Function, Text, Operator)), | ||||
(r'[^\S\n]+', Text), | ||||
(r'//.*?\n', Comment.Single), | ||||
(r'/\*.*?\*/', Comment.Multiline), | ||||
(r'@[a-zA-Z_][a-zA-Z0-9_\.]*', Name.Decorator), | ||||
(r'(assert|break|case|catch|continue|default|do|else|finally|for|' | ||||
r'if|goto|instanceof|new|return|switch|this|throw|try|while)\b', | ||||
Keyword), | ||||
(r'(abstract|const|enum|extends|final|implements|native|private|' | ||||
r'protected|public|static|strictfp|super|synchronized|throws|' | ||||
r'transient|volatile)\b', Keyword.Declaration), | ||||
(r'(boolean|byte|char|double|float|int|long|short|void)\b', | ||||
Keyword.Type), | ||||
(r'(package)(\s+)', bygroups(Keyword.Namespace, Text)), | ||||
(r'(true|false|null)\b', Keyword.Constant), | ||||
(r'(class|interface)(\s+)', bygroups(Keyword.Declaration, Text), 'class'), | ||||
(r'(import)(\s+)', bygroups(Keyword.Namespace, Text), 'import'), | ||||
(r'"(\\\\|\\"|[^"])*"', String), | ||||
(r"'\\.'|'[^\\]'|'\\u[0-9a-f]{4}'", String.Char), | ||||
(r'(\.)([a-zA-Z_][a-zA-Z0-9_]*)', bygroups(Operator, Name.Attribute)), | ||||
(r'[a-zA-Z_][a-zA-Z0-9_]*:', Name.Label), | ||||
(r'[a-zA-Z_\$][a-zA-Z0-9_]*', Name), | ||||
(r'[~\^\*!%&\[\]\(\)\{\}<>\|+=:;,./?-]', Operator), | ||||
(r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float), | ||||
(r'0x[0-9a-f]+', Number.Hex), | ||||
(r'[0-9]+L?', Number.Integer), | ||||
(r'\n', Text) | ||||
], | ||||
'class': [ | ||||
(r'[a-zA-Z_][a-zA-Z0-9_]*', Name.Class, '#pop') | ||||
], | ||||
'import': [ | ||||
(r'[a-zA-Z0-9_.]+\*?', Name.Namespace, '#pop') | ||||
], | ||||
} | ||||
class ScalaLexer(RegexLexer): | ||||
""" | ||||
For `Scala <http://www.scala-lang.org>`_ source code. | ||||
""" | ||||
name = 'Scala' | ||||
aliases = ['scala'] | ||||
filenames = ['*.scala'] | ||||
mimetypes = ['text/x-scala'] | ||||
flags = re.MULTILINE | re.DOTALL | ||||
#: optional Comment or Whitespace | ||||
_ws = r'(?:\s|//.*?\n|/[*].*?[*]/)+' | ||||
# don't use raw unicode strings! | ||||
op = u'[-~\\^\\*!%&\\\\<>\\|+=:/?@\u00a6-\u00a7\u00a9\u00ac\u00ae\u00b0-\u00b1\u00b6\u00d7\u00f7\u03f6\u0482\u0606-\u0608\u060e-\u060f\u06e9\u06fd-\u06fe\u07f6\u09fa\u0b70\u0bf3-\u0bf8\u0bfa\u0c7f\u0cf1-\u0cf2\u0d79\u0f01-\u0f03\u0f13-\u0f17\u0f1a-\u0f1f\u0f34\u0f36\u0f38\u0fbe-\u0fc5\u0fc7-\u0fcf\u109e-\u109f\u1360\u1390-\u1399\u1940\u19e0-\u19ff\u1b61-\u1b6a\u1b74-\u1b7c\u2044\u2052\u207a-\u207c\u208a-\u208c\u2100-\u2101\u2103-\u2106\u2108-\u2109\u2114\u2116-\u2118\u211e-\u2123\u2125\u2127\u2129\u212e\u213a-\u213b\u2140-\u2144\u214a-\u214d\u214f\u2190-\u2328\u232b-\u244a\u249c-\u24e9\u2500-\u2767\u2794-\u27c4\u27c7-\u27e5\u27f0-\u2982\u2999-\u29d7\u29dc-\u29fb\u29fe-\u2b54\u2ce5-\u2cea\u2e80-\u2ffb\u3004\u3012-\u3013\u3020\u3036-\u3037\u303e-\u303f\u3190-\u3191\u3196-\u319f\u31c0-\u31e3\u3200-\u321e\u322a-\u3250\u3260-\u327f\u328a-\u32b0\u32c0-\u33ff\u4dc0-\u4dff\ua490-\ua4c6\ua828-\ua82b\ufb29\ufdfd\ufe62\ufe64-\ufe66\uff0b\uff1c-\uff1e\uff5c\uff5e\uffe2\uffe4\uffe8-\uffee\ufffc-\ufffd]+' | ||||
letter = u'[a-zA-Z\\$_\u00aa\u00b5\u00ba\u00c0-\u00d6\u00d8-\u00f6\u00f8-\u02af\u0370-\u0373\u0376-\u0377\u037b-\u037d\u0386\u0388-\u03f5\u03f7-\u0481\u048a-\u0556\u0561-\u0587\u05d0-\u05f2\u0621-\u063f\u0641-\u064a\u066e-\u066f\u0671-\u06d3\u06d5\u06ee-\u06ef\u06fa-\u06fc\u06ff\u0710\u0712-\u072f\u074d-\u07a5\u07b1\u07ca-\u07ea\u0904-\u0939\u093d\u0950\u0958-\u0961\u0972-\u097f\u0985-\u09b9\u09bd\u09ce\u09dc-\u09e1\u09f0-\u09f1\u0a05-\u0a39\u0a59-\u0a5e\u0a72-\u0a74\u0a85-\u0ab9\u0abd\u0ad0-\u0ae1\u0b05-\u0b39\u0b3d\u0b5c-\u0b61\u0b71\u0b83-\u0bb9\u0bd0\u0c05-\u0c3d\u0c58-\u0c61\u0c85-\u0cb9\u0cbd\u0cde-\u0ce1\u0d05-\u0d3d\u0d60-\u0d61\u0d7a-\u0d7f\u0d85-\u0dc6\u0e01-\u0e30\u0e32-\u0e33\u0e40-\u0e45\u0e81-\u0eb0\u0eb2-\u0eb3\u0ebd-\u0ec4\u0edc-\u0f00\u0f40-\u0f6c\u0f88-\u0f8b\u1000-\u102a\u103f\u1050-\u1055\u105a-\u105d\u1061\u1065-\u1066\u106e-\u1070\u1075-\u1081\u108e\u10a0-\u10fa\u1100-\u135a\u1380-\u138f\u13a0-\u166c\u166f-\u1676\u1681-\u169a\u16a0-\u16ea\u16ee-\u1711\u1720-\u1731\u1740-\u1751\u1760-\u1770\u1780-\u17b3\u17dc\u1820-\u1842\u1844-\u18a8\u18aa-\u191c\u1950-\u19a9\u19c1-\u19c7\u1a00-\u1a16\u1b05-\u1b33\u1b45-\u1b4b\u1b83-\u1ba0\u1bae-\u1baf\u1c00-\u1c23\u1c4d-\u1c4f\u1c5a-\u1c77\u1d00-\u1d2b\u1d62-\u1d77\u1d79-\u1d9a\u1e00-\u1fbc\u1fbe\u1fc2-\u1fcc\u1fd0-\u1fdb\u1fe0-\u1fec\u1ff2-\u1ffc\u2071\u207f\u2102\u2107\u210a-\u2113\u2115\u2119-\u211d\u2124\u2126\u2128\u212a-\u212d\u212f-\u2139\u213c-\u213f\u2145-\u2149\u214e\u2160-\u2188\u2c00-\u2c7c\u2c80-\u2ce4\u2d00-\u2d65\u2d80-\u2dde\u3006-\u3007\u3021-\u3029\u3038-\u303a\u303c\u3041-\u3096\u309f\u30a1-\u30fa\u30ff-\u318e\u31a0-\u31b7\u31f0-\u31ff\u3400-\u4db5\u4e00-\ua014\ua016-\ua48c\ua500-\ua60b\ua610-\ua61f\ua62a-\ua66e\ua680-\ua697\ua722-\ua76f\ua771-\ua787\ua78b-\ua801\ua803-\ua805\ua807-\ua80a\ua80c-\ua822\ua840-\ua873\ua882-\ua8b3\ua90a-\ua925\ua930-\ua946\uaa00-\uaa28\uaa40-\uaa42\uaa44-\uaa4b\uac00-\ud7a3\uf900-\ufb1d\ufb1f-\ufb28\ufb2a-\ufd3d\ufd50-\ufdfb\ufe70-\ufefc\uff21-\uff3a\uff41-\uff5a\uff66-\uff6f\uff71-\uff9d\uffa0-\uffdc]' | ||||
upper = u'[A-Z\\$_\u00c0-\u00d6\u00d8-\u00de\u0100\u0102\u0104\u0106\u0108\u010a\u010c\u010e\u0110\u0112\u0114\u0116\u0118\u011a\u011c\u011e\u0120\u0122\u0124\u0126\u0128\u012a\u012c\u012e\u0130\u0132\u0134\u0136\u0139\u013b\u013d\u013f\u0141\u0143\u0145\u0147\u014a\u014c\u014e\u0150\u0152\u0154\u0156\u0158\u015a\u015c\u015e\u0160\u0162\u0164\u0166\u0168\u016a\u016c\u016e\u0170\u0172\u0174\u0176\u0178-\u0179\u017b\u017d\u0181-\u0182\u0184\u0186-\u0187\u0189-\u018b\u018e-\u0191\u0193-\u0194\u0196-\u0198\u019c-\u019d\u019f-\u01a0\u01a2\u01a4\u01a6-\u01a7\u01a9\u01ac\u01ae-\u01af\u01b1-\u01b3\u01b5\u01b7-\u01b8\u01bc\u01c4\u01c7\u01ca\u01cd\u01cf\u01d1\u01d3\u01d5\u01d7\u01d9\u01db\u01de\u01e0\u01e2\u01e4\u01e6\u01e8\u01ea\u01ec\u01ee\u01f1\u01f4\u01f6-\u01f8\u01fa\u01fc\u01fe\u0200\u0202\u0204\u0206\u0208\u020a\u020c\u020e\u0210\u0212\u0214\u0216\u0218\u021a\u021c\u021e\u0220\u0222\u0224\u0226\u0228\u022a\u022c\u022e\u0230\u0232\u023a-\u023b\u023d-\u023e\u0241\u0243-\u0246\u0248\u024a\u024c\u024e\u0370\u0372\u0376\u0386\u0388-\u038f\u0391-\u03ab\u03cf\u03d2-\u03d4\u03d8\u03da\u03dc\u03de\u03e0\u03e2\u03e4\u03e6\u03e8\u03ea\u03ec\u03ee\u03f4\u03f7\u03f9-\u03fa\u03fd-\u042f\u0460\u0462\u0464\u0466\u0468\u046a\u046c\u046e\u0470\u0472\u0474\u0476\u0478\u047a\u047c\u047e\u0480\u048a\u048c\u048e\u0490\u0492\u0494\u0496\u0498\u049a\u049c\u049e\u04a0\u04a2\u04a4\u04a6\u04a8\u04aa\u04ac\u04ae\u04b0\u04b2\u04b4\u04b6\u04b8\u04ba\u04bc\u04be\u04c0-\u04c1\u04c3\u04c5\u04c7\u04c9\u04cb\u04cd\u04d0\u04d2\u04d4\u04d6\u04d8\u04da\u04dc\u04de\u04e0\u04e2\u04e4\u04e6\u04e8\u04ea\u04ec\u04ee\u04f0\u04f2\u04f4\u04f6\u04f8\u04fa\u04fc\u04fe\u0500\u0502\u0504\u0506\u0508\u050a\u050c\u050e\u0510\u0512\u0514\u0516\u0518\u051a\u051c\u051e\u0520\u0522\u0531-\u0556\u10a0-\u10c5\u1e00\u1e02\u1e04\u1e06\u1e08\u1e0a\u1e0c\u1e0e\u1e10\u1e12\u1e14\u1e16\u1e18\u1e1a\u1e1c\u1e1e\u1e20\u1e22\u1e24\u1e26\u1e28\u1e2a\u1e2c\u1e2e\u1e30\u1e32\u1e34\u1e36\u1e38\u1e3a\u1e3c\u1e3e\u1e40\u1e42\u1e44\u1e46\u1e48\u1e4a\u1e4c\u1e4e\u1e50\u1e52\u1e54\u1e56\u1e58\u1e5a\u1e5c\u1e5e\u1e60\u1e62\u1e64\u1e66\u1e68\u1e6a\u1e6c\u1e6e\u1e70\u1e72\u1e74\u1e76\u1e78\u1e7a\u1e7c\u1e7e\u1e80\u1e82\u1e84\u1e86\u1e88\u1e8a\u1e8c\u1e8e\u1e90\u1e92\u1e94\u1e9e\u1ea0\u1ea2\u1ea4\u1ea6\u1ea8\u1eaa\u1eac\u1eae\u1eb0\u1eb2\u1eb4\u1eb6\u1eb8\u1eba\u1ebc\u1ebe\u1ec0\u1ec2\u1ec4\u1ec6\u1ec8\u1eca\u1ecc\u1ece\u1ed0\u1ed2\u1ed4\u1ed6\u1ed8\u1eda\u1edc\u1ede\u1ee0\u1ee2\u1ee4\u1ee6\u1ee8\u1eea\u1eec\u1eee\u1ef0\u1ef2\u1ef4\u1ef6\u1ef8\u1efa\u1efc\u1efe\u1f08-\u1f0f\u1f18-\u1f1d\u1f28-\u1f2f\u1f38-\u1f3f\u1f48-\u1f4d\u1f59-\u1f5f\u1f68-\u1f6f\u1fb8-\u1fbb\u1fc8-\u1fcb\u1fd8-\u1fdb\u1fe8-\u1fec\u1ff8-\u1ffb\u2102\u2107\u210b-\u210d\u2110-\u2112\u2115\u2119-\u211d\u2124\u2126\u2128\u212a-\u212d\u2130-\u2133\u213e-\u213f\u2145\u2183\u2c00-\u2c2e\u2c60\u2c62-\u2c64\u2c67\u2c69\u2c6b\u2c6d-\u2c6f\u2c72\u2c75\u2c80\u2c82\u2c84\u2c86\u2c88\u2c8a\u2c8c\u2c8e\u2c90\u2c92\u2c94\u2c96\u2c98\u2c9a\u2c9c\u2c9e\u2ca0\u2ca2\u2ca4\u2ca6\u2ca8\u2caa\u2cac\u2cae\u2cb0\u2cb2\u2cb4\u2cb6\u2cb8\u2cba\u2cbc\u2cbe\u2cc0\u2cc2\u2cc4\u2cc6\u2cc8\u2cca\u2ccc\u2cce\u2cd0\u2cd2\u2cd4\u2cd6\u2cd8\u2cda\u2cdc\u2cde\u2ce0\u2ce2\ua640\ua642\ua644\ua646\ua648\ua64a\ua64c\ua64e\ua650\ua652\ua654\ua656\ua658\ua65a\ua65c\ua65e\ua662\ua664\ua666\ua668\ua66a\ua66c\ua680\ua682\ua684\ua686\ua688\ua68a\ua68c\ua68e\ua690\ua692\ua694\ua696\ua722\ua724\ua726\ua728\ua72a\ua72c\ua72e\ua732\ua734\ua736\ua738\ua73a\ua73c\ua73e\ua740\ua742\ua744\ua746\ua748\ua74a\ua74c\ua74e\ua750\ua752\ua754\ua756\ua758\ua75a\ua75c\ua75e\ua760\ua762\ua764\ua766\ua768\ua76a\ua76c\ua76e\ua779\ua77b\ua77d-\ua77e\ua780\ua782\ua784\ua786\ua78b\uff21-\uff3a]' | ||||
idrest = ur'%s(?:%s|[0-9])*(?:(?<=_)%s)?' % (letter, letter, op) | ||||
tokens = { | ||||
'root': [ | ||||
# method names | ||||
(r'(class|trait|object)(\s+)', bygroups(Keyword, Text), 'class'), | ||||
(ur"'%s" % idrest, Text.Symbol), | ||||
(r'[^\S\n]+', Text), | ||||
(r'//.*?\n', Comment.Single), | ||||
(r'/\*', Comment.Multiline, 'comment'), | ||||
(ur'@%s' % idrest, Name.Decorator), | ||||
(ur'(abstract|ca(?:se|tch)|d(?:ef|o)|e(?:lse|xtends)|' | ||||
ur'f(?:inal(?:ly)?|or(?:Some)?)|i(?:f|mplicit)|' | ||||
ur'lazy|match|new|override|pr(?:ivate|otected)' | ||||
ur'|re(?:quires|turn)|s(?:ealed|uper)|' | ||||
ur't(?:h(?:is|row)|ry)|va[lr]|w(?:hile|ith)|yield)\b|' | ||||
u'(<[%:-]|=>|>:|[#=@_\u21D2\u2190])(\\b|(?=\\s)|$)', Keyword), | ||||
(ur':(?!%s)' % op, Keyword, 'type'), | ||||
(ur'%s%s\b' % (upper, idrest), Name.Class), | ||||
(r'(true|false|null)\b', Keyword.Constant), | ||||
(r'(import|package)(\s+)', bygroups(Keyword, Text), 'import'), | ||||
(r'(type)(\s+)', bygroups(Keyword, Text), 'type'), | ||||
(r'""".*?"""', String), | ||||
(r'"(\\\\|\\"|[^"])*"', String), | ||||
(r"'\\.'|'[^\\]'|'\\u[0-9a-f]{4}'", String.Char), | ||||
# (ur'(\.)(%s|%s|`[^`]+`)' % (idrest, op), bygroups(Operator, | ||||
# Name.Attribute)), | ||||
(idrest, Name), | ||||
(r'`[^`]+`', Name), | ||||
(r'\[', Operator, 'typeparam'), | ||||
(r'[\(\)\{\};,.#]', Operator), | ||||
(op, Operator), | ||||
(r'([0-9][0-9]*\.[0-9]*|\.[0-9]+)([eE][+-]?[0-9]+)?[fFdD]?', | ||||
Number.Float), | ||||
(r'0x[0-9a-f]+', Number.Hex), | ||||
(r'[0-9]+L?', Number.Integer), | ||||
(r'\n', Text) | ||||
], | ||||
'class': [ | ||||
(ur'(%s|%s|`[^`]+`)(\s*)(\[)' % (idrest, op), | ||||
bygroups(Name.Class, Text, Operator), 'typeparam'), | ||||
(r'\s+', Text), | ||||
(r'{', Operator, '#pop'), | ||||
(r'\(', Operator, '#pop'), | ||||
(r'//.*?\n', Comment.Single, '#pop'), | ||||
(ur'%s|%s|`[^`]+`' % (idrest, op), Name.Class, '#pop'), | ||||
], | ||||
'type': [ | ||||
(r'\s+', Text), | ||||
(u'<[%:]|>:|[#_\u21D2]|forSome|type', Keyword), | ||||
(r'([,\);}]|=>|=)(\s*)', bygroups(Operator, Text), '#pop'), | ||||
(r'[\(\{]', Operator, '#push'), | ||||
(ur'((?:%s|%s|`[^`]+`)(?:\.(?:%s|%s|`[^`]+`))*)(\s*)(\[)' % | ||||
(idrest, op, idrest, op), | ||||
bygroups(Keyword.Type, Text, Operator), ('#pop', 'typeparam')), | ||||
(ur'((?:%s|%s|`[^`]+`)(?:\.(?:%s|%s|`[^`]+`))*)(\s*)$' % | ||||
(idrest, op, idrest, op), | ||||
bygroups(Keyword.Type, Text), '#pop'), | ||||
(r'//.*?\n', Comment.Single, '#pop'), | ||||
(ur'\.|%s|%s|`[^`]+`' % (idrest, op), Keyword.Type) | ||||
], | ||||
'typeparam': [ | ||||
(r'[\s,]+', Text), | ||||
(u'<[%:]|=>|>:|[#_\u21D2]|forSome|type', Keyword), | ||||
(r'([\]\)\}])', Operator, '#pop'), | ||||
(r'[\(\[\{]', Operator, '#push'), | ||||
(ur'\.|%s|%s|`[^`]+`' % (idrest, op), Keyword.Type) | ||||
], | ||||
'comment': [ | ||||
(r'[^/\*]+', Comment.Multiline), | ||||
(r'/\*', Comment.Multiline, '#push'), | ||||
(r'\*/', Comment.Multiline, '#pop'), | ||||
(r'[*/]', Comment.Multiline) | ||||
], | ||||
'import': [ | ||||
(ur'(%s|\.)+' % idrest, Name.Namespace, '#pop') | ||||
], | ||||
} | ||||
class GosuLexer(RegexLexer): | ||||
""" | ||||
For Gosu source code. | ||||
*New in Pygments 1.5.* | ||||
""" | ||||
name = 'Gosu' | ||||
aliases = ['gosu'] | ||||
filenames = ['*.gs', '*.gsx', '*.gsp', '*.vark'] | ||||
mimetypes = ['text/x-gosu'] | ||||
flags = re.MULTILINE | re.DOTALL | ||||
#: optional Comment or Whitespace | ||||
_ws = r'(?:\s|//.*?\n|/[*].*?[*]/)+' | ||||
tokens = { | ||||
'root': [ | ||||
# method names | ||||
(r'^(\s*(?:[a-zA-Z_][a-zA-Z0-9_\.\[\]]*\s+)+?)' # modifiers etc. | ||||
r'([a-zA-Z_][a-zA-Z0-9_]*)' # method name | ||||
r'(\s*)(\()', # signature start | ||||
bygroups(using(this), Name.Function, Text, Operator)), | ||||
(r'[^\S\n]+', Text), | ||||
(r'//.*?\n', Comment.Single), | ||||
(r'/\*.*?\*/', Comment.Multiline), | ||||
(r'@[a-zA-Z_][a-zA-Z0-9_\.]*', Name.Decorator), | ||||
(r'(in|as|typeof|statictypeof|typeis|typeas|if|else|foreach|for|' | ||||
r'index|while|do|continue|break|return|try|catch|finally|this|' | ||||
r'throw|new|switch|case|default|eval|super|outer|classpath|' | ||||
r'using)\b', Keyword), | ||||
(r'(var|delegate|construct|function|private|internal|protected|' | ||||
r'public|abstract|override|final|static|extends|transient|' | ||||
r'implements|represents|readonly)\b', Keyword.Declaration), | ||||
(r'(property\s+)(get|set|)', Keyword.Declaration), | ||||
(r'(boolean|byte|char|double|float|int|long|short|void|block)\b', | ||||
Keyword.Type), | ||||
(r'(package)(\s+)', bygroups(Keyword.Namespace, Text)), | ||||
(r'(true|false|null|NaN|Infinity)\b', Keyword.Constant), | ||||
(r'(class|interface|enhancement|enum)(\s+)([a-zA-Z_][a-zA-Z0-9_]*)', | ||||
bygroups(Keyword.Declaration, Text, Name.Class)), | ||||
(r'(uses)(\s+)([a-zA-Z0-9_.]+\*?)', | ||||
bygroups(Keyword.Namespace, Text, Name.Namespace)), | ||||
(r'"', String, 'string'), | ||||
(r'(\??[\.#])([a-zA-Z_][a-zA-Z0-9_]*)', | ||||
bygroups(Operator, Name.Attribute)), | ||||
(r'(:)([a-zA-Z_][a-zA-Z0-9_]*)', | ||||
bygroups(Operator, Name.Attribute)), | ||||
(r'[a-zA-Z_\$][a-zA-Z0-9_]*', Name), | ||||
(r'and|or|not|[\\~\^\*!%&\[\]\(\)\{\}<>\|+=:;,./?-]', Operator), | ||||
(r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float), | ||||
(r'[0-9]+', Number.Integer), | ||||
(r'\n', Text) | ||||
], | ||||
'templateText': [ | ||||
(r'(\\<)|(\\\$)', String), | ||||
(r'(<%@\s+)(extends|params)', | ||||
bygroups(Operator, Name.Decorator), 'stringTemplate'), | ||||
(r'<%!--.*?--%>', Comment.Multiline), | ||||
(r'(<%)|(<%=)', Operator, 'stringTemplate'), | ||||
(r'\$\{', Operator, 'stringTemplateShorthand'), | ||||
(r'.', String) | ||||
], | ||||
'string': [ | ||||
(r'"', String, '#pop'), | ||||
include('templateText') | ||||
], | ||||
'stringTemplate': [ | ||||
(r'"', String, 'string'), | ||||
(r'%>', Operator, '#pop'), | ||||
include('root') | ||||
], | ||||
'stringTemplateShorthand': [ | ||||
(r'"', String, 'string'), | ||||
(r'\{', Operator, 'stringTemplateShorthand'), | ||||
(r'\}', Operator, '#pop'), | ||||
include('root') | ||||
], | ||||
} | ||||
class GosuTemplateLexer(Lexer): | ||||
""" | ||||
For Gosu templates. | ||||
*New in Pygments 1.5.* | ||||
""" | ||||
name = 'Gosu Template' | ||||
aliases = ['gst'] | ||||
filenames = ['*.gst'] | ||||
mimetypes = ['text/x-gosu-template'] | ||||
lexer = GosuLexer() | ||||
def get_tokens_unprocessed(self, text): | ||||
stack = ['templateText'] | ||||
for item in self.lexer.get_tokens_unprocessed(text, stack): | ||||
yield item | ||||
class GroovyLexer(RegexLexer): | ||||
""" | ||||
For `Groovy <http://groovy.codehaus.org/>`_ source code. | ||||
*New in Pygments 1.5.* | ||||
""" | ||||
name = 'Groovy' | ||||
aliases = ['groovy'] | ||||
filenames = ['*.groovy'] | ||||
mimetypes = ['text/x-groovy'] | ||||
flags = re.MULTILINE | re.DOTALL | ||||
#: optional Comment or Whitespace | ||||
_ws = r'(?:\s|//.*?\n|/[*].*?[*]/)+' | ||||
tokens = { | ||||
'root': [ | ||||
# method names | ||||
(r'^(\s*(?:[a-zA-Z_][a-zA-Z0-9_\.\[\]]*\s+)+?)' # return arguments | ||||
r'([a-zA-Z_][a-zA-Z0-9_]*)' # method name | ||||
r'(\s*)(\()', # signature start | ||||
bygroups(using(this), Name.Function, Text, Operator)), | ||||
(r'[^\S\n]+', Text), | ||||
(r'//.*?\n', Comment.Single), | ||||
(r'/\*.*?\*/', Comment.Multiline), | ||||
(r'@[a-zA-Z_][a-zA-Z0-9_\.]*', Name.Decorator), | ||||
(r'(assert|break|case|catch|continue|default|do|else|finally|for|' | ||||
r'if|goto|instanceof|new|return|switch|this|throw|try|while|in|as)\b', | ||||
Keyword), | ||||
(r'(abstract|const|enum|extends|final|implements|native|private|' | ||||
r'protected|public|static|strictfp|super|synchronized|throws|' | ||||
r'transient|volatile)\b', Keyword.Declaration), | ||||
(r'(def|boolean|byte|char|double|float|int|long|short|void)\b', | ||||
Keyword.Type), | ||||
(r'(package)(\s+)', bygroups(Keyword.Namespace, Text)), | ||||
(r'(true|false|null)\b', Keyword.Constant), | ||||
(r'(class|interface)(\s+)', bygroups(Keyword.Declaration, Text), | ||||
'class'), | ||||
(r'(import)(\s+)', bygroups(Keyword.Namespace, Text), 'import'), | ||||
(r'"(\\\\|\\"|[^"])*"', String.Double), | ||||
(r"'(\\\\|\\'|[^'])*'", String.Single), | ||||
(r'\$/((?!/\$).)*/\$', String), | ||||
(r'/(\\\\|\\"|[^/])*/', String), | ||||
(r"'\\.'|'[^\\]'|'\\u[0-9a-f]{4}'", String.Char), | ||||
(r'(\.)([a-zA-Z_][a-zA-Z0-9_]*)', bygroups(Operator, Name.Attribute)), | ||||
(r'[a-zA-Z_][a-zA-Z0-9_]*:', Name.Label), | ||||
(r'[a-zA-Z_\$][a-zA-Z0-9_]*', Name), | ||||
(r'[~\^\*!%&\[\]\(\)\{\}<>\|+=:;,./?-]', Operator), | ||||
(r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float), | ||||
(r'0x[0-9a-f]+', Number.Hex), | ||||
(r'[0-9]+L?', Number.Integer), | ||||
(r'\n', Text) | ||||
], | ||||
'class': [ | ||||
(r'[a-zA-Z_][a-zA-Z0-9_]*', Name.Class, '#pop') | ||||
], | ||||
'import': [ | ||||
(r'[a-zA-Z0-9_.]+\*?', Name.Namespace, '#pop') | ||||
], | ||||
} | ||||
class IokeLexer(RegexLexer): | ||||
""" | ||||
For `Ioke <http://ioke.org/>`_ (a strongly typed, dynamic, | ||||
prototype based programming language) source. | ||||
*New in Pygments 1.4.* | ||||
""" | ||||
name = 'Ioke' | ||||
filenames = ['*.ik'] | ||||
aliases = ['ioke', 'ik'] | ||||
mimetypes = ['text/x-iokesrc'] | ||||
tokens = { | ||||
'interpolatableText': [ | ||||
(r'(\\b|\\e|\\t|\\n|\\f|\\r|\\"|\\\\|\\#|\\\Z|\\u[0-9a-fA-F]{1,4}' | ||||
r'|\\[0-3]?[0-7]?[0-7])', String.Escape), | ||||
(r'#{', Punctuation, 'textInterpolationRoot') | ||||
], | ||||
'text': [ | ||||
(r'(?<!\\)"', String, '#pop'), | ||||
include('interpolatableText'), | ||||
(r'[^"]', String) | ||||
], | ||||
'documentation': [ | ||||
(r'(?<!\\)"', String.Doc, '#pop'), | ||||
include('interpolatableText'), | ||||
(r'[^"]', String.Doc) | ||||
], | ||||
'textInterpolationRoot': [ | ||||
(r'}', Punctuation, '#pop'), | ||||
include('root') | ||||
], | ||||
'slashRegexp': [ | ||||
(r'(?<!\\)/[oxpniums]*', String.Regex, '#pop'), | ||||
include('interpolatableText'), | ||||
(r'\\/', String.Regex), | ||||
(r'[^/]', String.Regex) | ||||
], | ||||
'squareRegexp': [ | ||||
(r'(?<!\\)][oxpniums]*', String.Regex, '#pop'), | ||||
include('interpolatableText'), | ||||
(r'\\]', String.Regex), | ||||
(r'[^\]]', String.Regex) | ||||
], | ||||
'squareText': [ | ||||
(r'(?<!\\)]', String, '#pop'), | ||||
include('interpolatableText'), | ||||
(r'[^\]]', String) | ||||
], | ||||
'root': [ | ||||
(r'\n', Text), | ||||
(r'\s+', Text), | ||||
# Comments | ||||
(r';(.*?)\n', Comment), | ||||
(r'\A#!(.*?)\n', Comment), | ||||
#Regexps | ||||
(r'#/', String.Regex, 'slashRegexp'), | ||||
(r'#r\[', String.Regex, 'squareRegexp'), | ||||
#Symbols | ||||
(r':[a-zA-Z0-9_!:?]+', String.Symbol), | ||||
(r'[a-zA-Z0-9_!:?]+:(?![a-zA-Z0-9_!?])', String.Other), | ||||
(r':"(\\\\|\\"|[^"])*"', String.Symbol), | ||||
#Documentation | ||||
(r'((?<=fn\()|(?<=fnx\()|(?<=method\()|(?<=macro\()|(?<=lecro\()' | ||||
r'|(?<=syntax\()|(?<=dmacro\()|(?<=dlecro\()|(?<=dlecrox\()' | ||||
r'|(?<=dsyntax\())\s*"', String.Doc, 'documentation'), | ||||
#Text | ||||
(r'"', String, 'text'), | ||||
(r'#\[', String, 'squareText'), | ||||
#Mimic | ||||
(r'[a-zA-Z0-9_][a-zA-Z0-9!?_:]+(?=\s*=.*mimic\s)', Name.Entity), | ||||
#Assignment | ||||
(r'[a-zA-Z_][a-zA-Z0-9_!:?]*(?=[\s]*[+*/-]?=[^=].*($|\.))', | ||||
Name.Variable), | ||||
# keywords | ||||
(r'(break|cond|continue|do|ensure|for|for:dict|for:set|if|let|' | ||||
r'loop|p:for|p:for:dict|p:for:set|return|unless|until|while|' | ||||
r'with)(?![a-zA-Z0-9!:_?])', Keyword.Reserved), | ||||
# Origin | ||||
(r'(eval|mimic|print|println)(?![a-zA-Z0-9!:_?])', Keyword), | ||||
# Base | ||||
(r'(cell\?|cellNames|cellOwner\?|cellOwner|cells|cell|' | ||||
r'documentation|hash|identity|mimic|removeCell\!|undefineCell\!)' | ||||
r'(?![a-zA-Z0-9!:_?])', Keyword), | ||||
# Ground | ||||
(r'(stackTraceAsText)(?![a-zA-Z0-9!:_?])', Keyword), | ||||
#DefaultBehaviour Literals | ||||
(r'(dict|list|message|set)(?![a-zA-Z0-9!:_?])', Keyword.Reserved), | ||||
#DefaultBehaviour Case | ||||
(r'(case|case:and|case:else|case:nand|case:nor|case:not|case:or|' | ||||
r'case:otherwise|case:xor)(?![a-zA-Z0-9!:_?])', Keyword.Reserved), | ||||
#DefaultBehaviour Reflection | ||||
(r'(asText|become\!|derive|freeze\!|frozen\?|in\?|is\?|kind\?|' | ||||
r'mimic\!|mimics|mimics\?|prependMimic\!|removeAllMimics\!|' | ||||
r'removeMimic\!|same\?|send|thaw\!|uniqueHexId)' | ||||
r'(?![a-zA-Z0-9!:_?])', Keyword), | ||||
#DefaultBehaviour Aspects | ||||
(r'(after|around|before)(?![a-zA-Z0-9!:_?])', Keyword.Reserved), | ||||
# DefaultBehaviour | ||||
(r'(kind|cellDescriptionDict|cellSummary|genSym|inspect|notice)' | ||||
r'(?![a-zA-Z0-9!:_?])', Keyword), | ||||
(r'(use|destructuring)', Keyword.Reserved), | ||||
#DefaultBehavior BaseBehavior | ||||
(r'(cell\?|cellOwner\?|cellOwner|cellNames|cells|cell|' | ||||
r'documentation|identity|removeCell!|undefineCell)' | ||||
r'(?![a-zA-Z0-9!:_?])', Keyword), | ||||
#DefaultBehavior Internal | ||||
(r'(internal:compositeRegexp|internal:concatenateText|' | ||||
r'internal:createDecimal|internal:createNumber|' | ||||
r'internal:createRegexp|internal:createText)' | ||||
r'(?![a-zA-Z0-9!:_?])', Keyword.Reserved), | ||||
#DefaultBehaviour Conditions | ||||
(r'(availableRestarts|bind|error\!|findRestart|handle|' | ||||
r'invokeRestart|rescue|restart|signal\!|warn\!)' | ||||
r'(?![a-zA-Z0-9!:_?])', Keyword.Reserved), | ||||
# constants | ||||
(r'(nil|false|true)(?![a-zA-Z0-9!:_?])', Name.Constant), | ||||
# names | ||||
(r'(Arity|Base|Call|Condition|DateTime|Aspects|Pointcut|' | ||||
r'Assignment|BaseBehavior|Boolean|Case|AndCombiner|Else|' | ||||
r'NAndCombiner|NOrCombiner|NotCombiner|OrCombiner|XOrCombiner|' | ||||
r'Conditions|Definitions|FlowControl|Internal|Literals|' | ||||
r'Reflection|DefaultMacro|DefaultMethod|DefaultSyntax|Dict|' | ||||
r'FileSystem|Ground|Handler|Hook|IO|IokeGround|Struct|' | ||||
r'LexicalBlock|LexicalMacro|List|Message|Method|Mixins|' | ||||
r'NativeMethod|Number|Origin|Pair|Range|Reflector|Regexp Match|' | ||||
r'Regexp|Rescue|Restart|Runtime|Sequence|Set|Symbol|' | ||||
r'System|Text|Tuple)(?![a-zA-Z0-9!:_?])', Name.Builtin), | ||||
# functions | ||||
(ur'(generateMatchMethod|aliasMethod|\u03bb|\u028E|fnx|fn|method|' | ||||
ur'dmacro|dlecro|syntax|macro|dlecrox|lecrox|lecro|syntax)' | ||||
ur'(?![a-zA-Z0-9!:_?])', Name.Function), | ||||
# Numbers | ||||
(r'-?0[xX][0-9a-fA-F]+', Number.Hex), | ||||
(r'-?(\d+\.?\d*|\d*\.\d+)([eE][+-]?[0-9]+)?', Number.Float), | ||||
(r'-?\d+', Number.Integer), | ||||
(r'#\(', Punctuation), | ||||
# Operators | ||||
(ur'(&&>>|\|\|>>|\*\*>>|:::|::|\.\.\.|===|\*\*>|\*\*=|&&>|&&=|' | ||||
ur'\|\|>|\|\|=|\->>|\+>>|!>>|<>>>|<>>|&>>|%>>|#>>|@>>|/>>|\*>>|' | ||||
ur'\?>>|\|>>|\^>>|~>>|\$>>|=>>|<<=|>>=|<=>|<\->|=~|!~|=>|\+\+|' | ||||
ur'\-\-|<=|>=|==|!=|&&|\.\.|\+=|\-=|\*=|\/=|%=|&=|\^=|\|=|<\-|' | ||||
ur'\+>|!>|<>|&>|%>|#>|\@>|\/>|\*>|\?>|\|>|\^>|~>|\$>|<\->|\->|' | ||||
ur'<<|>>|\*\*|\?\||\?&|\|\||>|<|\*|\/|%|\+|\-|&|\^|\||=|\$|!|~|' | ||||
ur'\?|#|\u2260|\u2218|\u2208|\u2209)', Operator), | ||||
(r'(and|nand|or|xor|nor|return|import)(?![a-zA-Z0-9_!?])', | ||||
Operator), | ||||
# Punctuation | ||||
(r'(\`\`|\`|\'\'|\'|\.|\,|@@|@|\[|\]|\(|\)|{|})', Punctuation), | ||||
#kinds | ||||
(r'[A-Z][a-zA-Z0-9_!:?]*', Name.Class), | ||||
#default cellnames | ||||
(r'[a-z_][a-zA-Z0-9_!:?]*', Name) | ||||
] | ||||
} | ||||
class ClojureLexer(RegexLexer): | ||||
""" | ||||
Lexer for `Clojure <http://clojure.org/>`_ source code. | ||||
*New in Pygments 0.11.* | ||||
""" | ||||
name = 'Clojure' | ||||
aliases = ['clojure', 'clj'] | ||||
filenames = ['*.clj'] | ||||
mimetypes = ['text/x-clojure', 'application/x-clojure'] | ||||
special_forms = [ | ||||
'.', 'def', 'do', 'fn', 'if', 'let', 'new', 'quote', 'var', 'loop' | ||||
] | ||||
# It's safe to consider 'ns' a declaration thing because it defines a new | ||||
# namespace. | ||||
declarations = [ | ||||
'def-', 'defn', 'defn-', 'defmacro', 'defmulti', 'defmethod', | ||||
'defstruct', 'defonce', 'declare', 'definline', 'definterface', | ||||
'defprotocol', 'defrecord', 'deftype', 'defproject', 'ns' | ||||
] | ||||
builtins = [ | ||||
'*', '+', '-', '->', '/', '<', '<=', '=', '==', '>', '>=', '..', | ||||
'accessor', 'agent', 'agent-errors', 'aget', 'alength', 'all-ns', | ||||
'alter', 'and', 'append-child', 'apply', 'array-map', 'aset', | ||||
'aset-boolean', 'aset-byte', 'aset-char', 'aset-double', 'aset-float', | ||||
'aset-int', 'aset-long', 'aset-short', 'assert', 'assoc', 'await', | ||||
'await-for', 'bean', 'binding', 'bit-and', 'bit-not', 'bit-or', | ||||
'bit-shift-left', 'bit-shift-right', 'bit-xor', 'boolean', 'branch?', | ||||
'butlast', 'byte', 'cast', 'char', 'children', 'class', | ||||
'clear-agent-errors', 'comment', 'commute', 'comp', 'comparator', | ||||
'complement', 'concat', 'conj', 'cons', 'constantly', 'cond', 'if-not', | ||||
'construct-proxy', 'contains?', 'count', 'create-ns', 'create-struct', | ||||
'cycle', 'dec', 'deref', 'difference', 'disj', 'dissoc', 'distinct', | ||||
'doall', 'doc', 'dorun', 'doseq', 'dosync', 'dotimes', 'doto', | ||||
'double', 'down', 'drop', 'drop-while', 'edit', 'end?', 'ensure', | ||||
'eval', 'every?', 'false?', 'ffirst', 'file-seq', 'filter', 'find', | ||||
'find-doc', 'find-ns', 'find-var', 'first', 'float', 'flush', 'for', | ||||
'fnseq', 'frest', 'gensym', 'get-proxy-class', 'get', | ||||
'hash-map', 'hash-set', 'identical?', 'identity', 'if-let', 'import', | ||||
'in-ns', 'inc', 'index', 'insert-child', 'insert-left', 'insert-right', | ||||
'inspect-table', 'inspect-tree', 'instance?', 'int', 'interleave', | ||||
'intersection', 'into', 'into-array', 'iterate', 'join', 'key', 'keys', | ||||
'keyword', 'keyword?', 'last', 'lazy-cat', 'lazy-cons', 'left', | ||||
'lefts', 'line-seq', 'list*', 'list', 'load', 'load-file', | ||||
'locking', 'long', 'loop', 'macroexpand', 'macroexpand-1', | ||||
'make-array', 'make-node', 'map', 'map-invert', 'map?', 'mapcat', | ||||
'max', 'max-key', 'memfn', 'merge', 'merge-with', 'meta', 'min', | ||||
'min-key', 'name', 'namespace', 'neg?', 'new', 'newline', 'next', | ||||
'nil?', 'node', 'not', 'not-any?', 'not-every?', 'not=', 'ns-imports', | ||||
'ns-interns', 'ns-map', 'ns-name', 'ns-publics', 'ns-refers', | ||||
'ns-resolve', 'ns-unmap', 'nth', 'nthrest', 'or', 'parse', 'partial', | ||||
'path', 'peek', 'pop', 'pos?', 'pr', 'pr-str', 'print', 'print-str', | ||||
'println', 'println-str', 'prn', 'prn-str', 'project', 'proxy', | ||||
'proxy-mappings', 'quot', 'rand', 'rand-int', 'range', 're-find', | ||||
're-groups', 're-matcher', 're-matches', 're-pattern', 're-seq', | ||||
'read', 'read-line', 'reduce', 'ref', 'ref-set', 'refer', 'rem', | ||||
'remove', 'remove-method', 'remove-ns', 'rename', 'rename-keys', | ||||
'repeat', 'replace', 'replicate', 'resolve', 'rest', 'resultset-seq', | ||||
'reverse', 'rfirst', 'right', 'rights', 'root', 'rrest', 'rseq', | ||||
'second', 'select', 'select-keys', 'send', 'send-off', 'seq', | ||||
'seq-zip', 'seq?', 'set', 'short', 'slurp', 'some', 'sort', | ||||
'sort-by', 'sorted-map', 'sorted-map-by', 'sorted-set', | ||||
'special-symbol?', 'split-at', 'split-with', 'str', 'string?', | ||||
'struct', 'struct-map', 'subs', 'subvec', 'symbol', 'symbol?', | ||||
'sync', 'take', 'take-nth', 'take-while', 'test', 'time', 'to-array', | ||||
'to-array-2d', 'tree-seq', 'true?', 'union', 'up', 'update-proxy', | ||||
'val', 'vals', 'var-get', 'var-set', 'var?', 'vector', 'vector-zip', | ||||
'vector?', 'when', 'when-first', 'when-let', 'when-not', | ||||
'with-local-vars', 'with-meta', 'with-open', 'with-out-str', | ||||
'xml-seq', 'xml-zip', 'zero?', 'zipmap', 'zipper'] | ||||
# valid names for identifiers | ||||
# well, names can only not consist fully of numbers | ||||
# but this should be good enough for now | ||||
# TODO / should divide keywords/symbols into namespace/rest | ||||
# but that's hard, so just pretend / is part of the name | ||||
valid_name = r'(?!#)[\w!$%*+<=>?/.#-]+' | ||||
def _multi_escape(entries): | ||||
return '(%s)' % ('|'.join(re.escape(entry) + ' ' for entry in entries)) | ||||
tokens = { | ||||
'root': [ | ||||
# the comments - always starting with semicolon | ||||
# and going to the end of the line | ||||
(r';.*$', Comment.Single), | ||||
# whitespaces - usually not relevant | ||||
(r'[,\s]+', Text), | ||||
# numbers | ||||
(r'-?\d+\.\d+', Number.Float), | ||||
(r'-?\d+', Number.Integer), | ||||
(r'0x-?[abcdef\d]+', Number.Hex), | ||||
# strings, symbols and characters | ||||
(r'"(\\\\|\\"|[^"])*"', String), | ||||
(r"'" + valid_name, String.Symbol), | ||||
(r"\\(.|[a-z]+)", String.Char), | ||||
# keywords | ||||
(r':' + valid_name, String.Symbol), | ||||
# special operators | ||||
(r'~@|[`\'#^~&]', Operator), | ||||
# highlight the special forms | ||||
(_multi_escape(special_forms), Keyword), | ||||
# Technically, only the special forms are 'keywords'. The problem | ||||
# is that only treating them as keywords means that things like | ||||
# 'defn' and 'ns' need to be highlighted as builtins. This is ugly | ||||
# and weird for most styles. So, as a compromise we're going to | ||||
# highlight them as Keyword.Declarations. | ||||
(_multi_escape(declarations), Keyword.Declaration), | ||||
# highlight the builtins | ||||
(_multi_escape(builtins), Name.Builtin), | ||||
# the remaining functions | ||||
(r'(?<=\()' + valid_name, Name.Function), | ||||
# find the remaining variables | ||||
(valid_name, Name.Variable), | ||||
# Clojure accepts vector notation | ||||
(r'(\[|\])', Punctuation), | ||||
# Clojure accepts map notation | ||||
(r'(\{|\})', Punctuation), | ||||
# the famous parentheses! | ||||
(r'(\(|\))', Punctuation), | ||||
], | ||||
} | ||||
class TeaLangLexer(RegexLexer): | ||||
""" | ||||
For `Tea <http://teatrove.org/>`_ source code. Only used within a | ||||
TeaTemplateLexer. | ||||
*New in Pygments 1.5.* | ||||
""" | ||||
flags = re.MULTILINE | re.DOTALL | ||||
#: optional Comment or Whitespace | ||||
_ws = r'(?:\s|//.*?\n|/[*].*?[*]/)+' | ||||
tokens = { | ||||
'root': [ | ||||
# method names | ||||
(r'^(\s*(?:[a-zA-Z_][a-zA-Z0-9_\.\[\]]*\s+)+?)' # return arguments | ||||
r'([a-zA-Z_][a-zA-Z0-9_]*)' # method name | ||||
r'(\s*)(\()', # signature start | ||||
bygroups(using(this), Name.Function, Text, Operator)), | ||||
(r'[^\S\n]+', Text), | ||||
(r'//.*?\n', Comment.Single), | ||||
(r'/\*.*?\*/', Comment.Multiline), | ||||
(r'@[a-zA-Z_][a-zA-Z0-9_\.]*', Name.Decorator), | ||||
(r'(and|break|else|foreach|if|in|not|or|reverse)\b', | ||||
Keyword), | ||||
(r'(as|call|define)\b', Keyword.Declaration), | ||||
(r'(true|false|null)\b', Keyword.Constant), | ||||
(r'(template)(\s+)', bygroups(Keyword.Declaration, Text), 'template'), | ||||
(r'(import)(\s+)', bygroups(Keyword.Namespace, Text), 'import'), | ||||
(r'"(\\\\|\\"|[^"])*"', String), | ||||
(r'\'(\\\\|\\\'|[^\'])*\'', String), | ||||
(r'(\.)([a-zA-Z_][a-zA-Z0-9_]*)', bygroups(Operator, Name.Attribute)), | ||||
(r'[a-zA-Z_][a-zA-Z0-9_]*:', Name.Label), | ||||
(r'[a-zA-Z_\$][a-zA-Z0-9_]*', Name), | ||||
(r'(isa|[.]{3}|[.]{2}|[=#!<>+-/%&;,.\*\\\(\)\[\]\{\}])', Operator), | ||||
(r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float), | ||||
(r'0x[0-9a-f]+', Number.Hex), | ||||
(r'[0-9]+L?', Number.Integer), | ||||
(r'\n', Text) | ||||
], | ||||
'template': [ | ||||
(r'[a-zA-Z_][a-zA-Z0-9_]*', Name.Class, '#pop') | ||||
], | ||||
'import': [ | ||||
(r'[a-zA-Z0-9_.]+\*?', Name.Namespace, '#pop') | ||||
], | ||||
} | ||||
class KotlinLexer(RegexLexer): | ||||
""" | ||||
For `Kotlin <http://confluence.jetbrains.net/display/Kotlin/>`_ | ||||
source code. | ||||
Additional options accepted: | ||||
`unicodelevel` | ||||
Determines which Unicode characters this lexer allows for identifiers. | ||||
The possible values are: | ||||
* ``none`` -- only the ASCII letters and numbers are allowed. This | ||||
is the fastest selection. | ||||
* ``basic`` -- all Unicode characters from the specification except | ||||
category ``Lo`` are allowed. | ||||
* ``full`` -- all Unicode characters as specified in the C# specs | ||||
are allowed. Note that this means a considerable slowdown since the | ||||
``Lo`` category has more than 40,000 characters in it! | ||||
The default value is ``basic``. | ||||
*New in Pygments 1.5.* | ||||
""" | ||||
name = 'Kotlin' | ||||
aliases = ['kotlin'] | ||||
filenames = ['*.kt'] | ||||
mimetypes = ['text/x-kotlin'] # inferred | ||||
flags = re.MULTILINE | re.DOTALL | re.UNICODE | ||||
# for the range of allowed unicode characters in identifiers, | ||||
# see http://www.ecma-international.org/publications/files/ECMA-ST/Ecma-334.pdf | ||||
def _escape(st): | ||||
return st.replace(u'\\', ur'\\').replace(u'-', ur'\-').\ | ||||
replace(u'[', ur'\[').replace(u']', ur'\]') | ||||
levels = { | ||||
'none': '@?[_a-zA-Z][a-zA-Z0-9_]*', | ||||
'basic': ('@?[_' + uni.Lu + uni.Ll + uni.Lt + uni.Lm + uni.Nl + ']' + | ||||
'[' + uni.Lu + uni.Ll + uni.Lt + uni.Lm + uni.Nl + | ||||
uni.Nd + uni.Pc + uni.Cf + uni.Mn + uni.Mc + ']*'), | ||||
'full': ('@?(?:_|[^' + | ||||
_escape(uni.allexcept('Lu', 'Ll', 'Lt', 'Lm', 'Lo', 'Nl')) + '])' | ||||
+ '[^' + _escape(uni.allexcept('Lu', 'Ll', 'Lt', 'Lm', 'Lo', | ||||
'Nl', 'Nd', 'Pc', 'Cf', 'Mn', | ||||
'Mc')) + ']*'), | ||||
} | ||||
tokens = {} | ||||
token_variants = True | ||||
for levelname, cs_ident in levels.items(): | ||||
tokens[levelname] = { | ||||
'root': [ | ||||
# method names | ||||
(r'^([ \t]*(?:' + cs_ident + r'(?:\[\])?\s+)+?)' # return type | ||||
r'(' + cs_ident + ')' # method name | ||||
r'(\s*)(\()', # signature start | ||||
bygroups(using(this), Name.Function, Text, Punctuation)), | ||||
(r'^\s*\[.*?\]', Name.Attribute), | ||||
(r'[^\S\n]+', Text), | ||||
(r'\\\n', Text), # line continuation | ||||
(r'//.*?\n', Comment.Single), | ||||
(r'/[*](.|\n)*?[*]/', Comment.Multiline), | ||||
(r'\n', Text), | ||||
(r'[~!%^&*()+=|\[\]:;,.<>/?-]', Punctuation), | ||||
(r'[{}]', Punctuation), | ||||
(r'@"(""|[^"])*"', String), | ||||
(r'"(\\\\|\\"|[^"\n])*["\n]', String), | ||||
(r"'\\.'|'[^\\]'", String.Char), | ||||
(r"[0-9](\.[0-9]*)?([eE][+-][0-9]+)?" | ||||
r"[flFLdD]?|0[xX][0-9a-fA-F]+[Ll]?", Number), | ||||
(r'#[ \t]*(if|endif|else|elif|define|undef|' | ||||
r'line|error|warning|region|endregion|pragma)\b.*?\n', | ||||
Comment.Preproc), | ||||
(r'\b(extern)(\s+)(alias)\b', bygroups(Keyword, Text, | ||||
Keyword)), | ||||
(r'(abstract|as|break|catch|' | ||||
r'fun|continue|default|delegate|' | ||||
r'do|else|enum|extern|false|finally|' | ||||
r'fixed|for|goto|if|implicit|in|interface|' | ||||
r'internal|is|lock|null|' | ||||
r'out|override|private|protected|public|readonly|' | ||||
r'ref|return|sealed|sizeof|' | ||||
r'when|this|throw|true|try|typeof|' | ||||
r'unchecked|unsafe|virtual|void|while|' | ||||
r'get|set|new|partial|yield|val|var)\b', Keyword), | ||||
(r'(global)(::)', bygroups(Keyword, Punctuation)), | ||||
(r'(bool|byte|char|decimal|double|dynamic|float|int|long|' | ||||
r'short)\b\??', Keyword.Type), | ||||
(r'(class|struct)(\s+)', bygroups(Keyword, Text), 'class'), | ||||
(r'(package|using)(\s+)', bygroups(Keyword, Text), 'package'), | ||||
(cs_ident, Name), | ||||
], | ||||
'class': [ | ||||
(cs_ident, Name.Class, '#pop') | ||||
], | ||||
'package': [ | ||||
(r'(?=\()', Text, '#pop'), # using (resource) | ||||
('(' + cs_ident + r'|\.)+', Name.Namespace, '#pop') | ||||
] | ||||
} | ||||
def __init__(self, **options): | ||||
level = get_choice_opt(options, 'unicodelevel', self.tokens.keys(), | ||||
'basic') | ||||
if level not in self._all_tokens: | ||||
# compile the regexes now | ||||
self._tokens = self.__class__.process_tokendef(level) | ||||
else: | ||||
self._tokens = self._all_tokens[level] | ||||
RegexLexer.__init__(self, **options) | ||||