aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/python/Pygments/py3/pygments/lexers/javascript.py
diff options
context:
space:
mode:
authorarcadia-devtools <arcadia-devtools@yandex-team.ru>2022-06-09 14:39:19 +0300
committerarcadia-devtools <arcadia-devtools@yandex-team.ru>2022-06-09 14:39:19 +0300
commitc04b663c7bb4b750deeb8f48f620497ec13da8fa (patch)
tree151ebc8bfdd2ad918caf5e6e2d8013e14272ddf8 /contrib/python/Pygments/py3/pygments/lexers/javascript.py
parent0d55ca22c507d18c2f35718687e0b06d9915397b (diff)
downloadydb-c04b663c7bb4b750deeb8f48f620497ec13da8fa.tar.gz
intermediate changes
ref:2d4f292087954c9344efdabb7b2a67f466263c65
Diffstat (limited to 'contrib/python/Pygments/py3/pygments/lexers/javascript.py')
-rw-r--r--contrib/python/Pygments/py3/pygments/lexers/javascript.py327
1 files changed, 181 insertions, 146 deletions
diff --git a/contrib/python/Pygments/py3/pygments/lexers/javascript.py b/contrib/python/Pygments/py3/pygments/lexers/javascript.py
index 7ddd1148e6f..1896fb46f6a 100644
--- a/contrib/python/Pygments/py3/pygments/lexers/javascript.py
+++ b/contrib/python/Pygments/py3/pygments/lexers/javascript.py
@@ -4,7 +4,7 @@
Lexers for JavaScript and related languages.
- :copyright: Copyright 2006-2021 by the Pygments team, see AUTHORS.
+ :copyright: Copyright 2006-2022 by the Pygments team, see AUTHORS.
:license: BSD, see LICENSE for details.
"""
@@ -13,7 +13,7 @@ import re
from pygments.lexer import bygroups, combined, default, do_insertions, include, \
inherit, Lexer, RegexLexer, this, using, words
from pygments.token import Text, Comment, Operator, Keyword, Name, String, \
- Number, Punctuation, Other, Generic
+ Number, Punctuation, Other, Generic, Whitespace
from pygments.util import get_bool_opt
import pygments.unistring as uni
@@ -37,18 +37,19 @@ class JavascriptLexer(RegexLexer):
"""
name = 'JavaScript'
+ url = 'https://www.ecma-international.org/publications-and-standards/standards/ecma-262/'
aliases = ['javascript', 'js']
filenames = ['*.js', '*.jsm', '*.mjs', '*.cjs']
mimetypes = ['application/javascript', 'application/x-javascript',
'text/x-javascript', 'text/javascript']
- flags = re.DOTALL | re.UNICODE | re.MULTILINE
+ flags = re.DOTALL | re.MULTILINE
tokens = {
'commentsandwhitespace': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'<!--', Comment),
- (r'//.*?\n', Comment.Single),
+ (r'//.*?$', Comment.Single),
(r'/\*.*?\*/', Comment.Multiline)
],
'slashstartsregex': [
@@ -59,10 +60,10 @@ class JavascriptLexer(RegexLexer):
default('#pop')
],
'badregex': [
- (r'\n', Text, '#pop')
+ (r'\n', Whitespace, '#pop')
],
'root': [
- (r'\A#! ?/.*?\n', Comment.Hashbang), # recognized by node.js
+ (r'\A#! ?/.*?$', Comment.Hashbang), # recognized by node.js
(r'^(?=\s|/|<!--)', Text, 'slashstartsregex'),
include('commentsandwhitespace'),
@@ -110,7 +111,7 @@ class JavascriptLexer(RegexLexer):
# Match stuff like: super(argument, list)
(r'(super)(\s*)(\([\w,?.$\s]+\s*\))',
- bygroups(Keyword, Text), 'slashstartsregex'),
+ bygroups(Keyword, Whitespace), 'slashstartsregex'),
# Match stuff like: function() {...}
(r'([a-zA-Z_?.$][\w?.$]*)(?=\(\) \{)', Name.Other, 'slashstartsregex'),
@@ -136,12 +137,13 @@ class JavascriptLexer(RegexLexer):
class TypeScriptLexer(JavascriptLexer):
"""
- For `TypeScript <http://typescriptlang.org/>`_ source code.
+ For TypeScript source code.
.. versionadded:: 1.6
"""
name = 'TypeScript'
+ url = 'https://www.typescriptlang.org/'
aliases = ['typescript', 'ts']
filenames = ['*.ts']
mimetypes = ['application/x-typescript', 'text/x-typescript']
@@ -159,11 +161,11 @@ class TypeScriptLexer(JavascriptLexer):
# Match variable type keywords
(r'\b(string|boolean|number)\b', Keyword.Type),
# Match stuff like: module name {...}
- (r'\b(module)(\s*)(\s*[\w?.$][\w?.$]*)(\s*)',
- bygroups(Keyword.Reserved, Text, Name.Other, Text), 'slashstartsregex'),
+ (r'\b(module)(\s*)([\w?.$]+)(\s*)',
+ bygroups(Keyword.Reserved, Whitespace, Name.Other, Whitespace), 'slashstartsregex'),
# Match stuff like: (function: return type)
- (r'([\w?.$][\w?.$]*)(\s*:\s*)([\w?.$][\w?.$]*)',
- bygroups(Name.Other, Text, Keyword.Type)),
+ (r'([\w?.$]+)(\s*)(:)(\s*)([\w?.$]+)',
+ bygroups(Name.Other, Whitespace, Operator, Whitespace, Keyword.Type)),
# Match stuff like: Decorators
(r'@' + JS_IDENT, Keyword.Declaration),
inherit,
@@ -173,15 +175,13 @@ class TypeScriptLexer(JavascriptLexer):
class KalLexer(RegexLexer):
"""
- For `Kal`_ source code.
-
- .. _Kal: http://rzimmerman.github.io/kal
-
+ For Kal source code.
.. versionadded:: 2.0
"""
name = 'Kal'
+ url = 'http://rzimmerman.github.io/kal'
aliases = ['kal']
filenames = ['*.kal']
mimetypes = ['text/kal', 'application/kal']
@@ -189,18 +189,20 @@ class KalLexer(RegexLexer):
flags = re.DOTALL
tokens = {
'commentsandwhitespace': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'###[^#].*?###', Comment.Multiline),
- (r'#(?!##[^#]).*?\n', Comment.Single),
+ (r'(#(?!##[^#]).*?)(\n)', bygroups(Comment.Single, Whitespace)),
],
'functiondef': [
- (r'[$a-zA-Z_][\w$]*\s*', Name.Function, '#pop'),
+ (r'([$a-zA-Z_][\w$]*)(\s*)', bygroups(Name.Function, Whitespace),
+ '#pop'),
include('commentsandwhitespace'),
],
'classdef': [
- (r'\binherits\s+from\b', Keyword),
- (r'[$a-zA-Z_][\w$]*\s*\n', Name.Class, '#pop'),
- (r'[$a-zA-Z_][\w$]*\s*', Name.Class),
+ (r'\b(inherits)(\s+)(from)\b',
+ bygroups(Keyword, Whitespace, Keyword)),
+ (r'([$a-zA-Z_][\w$]*)(?=\s*\n)', Name.Class, '#pop'),
+ (r'[$a-zA-Z_][\w$]*\b', Name.Class),
include('commentsandwhitespace'),
],
'listcomprehension': [
@@ -209,7 +211,7 @@ class KalLexer(RegexLexer):
include('root'),
],
'waitfor': [
- (r'\n', Punctuation, '#pop'),
+ (r'\n', Whitespace, '#pop'),
(r'\bfrom\b', Keyword),
include('root'),
],
@@ -221,29 +223,48 @@ class KalLexer(RegexLexer):
Operator),
(r'\b(and|or|isnt|is|not|but|bitwise|mod|\^|xor|exists|'
r'doesnt\s+exist)\b', Operator.Word),
- (r'(?:\([^()]+\))?\s*>', Name.Function),
+ (r'(\([^()]+\))?(\s*)(>)',
+ bygroups(Name.Function, Whitespace, Punctuation)),
(r'[{(]', Punctuation),
(r'\[', Punctuation, 'listcomprehension'),
(r'[})\].,]', Punctuation),
(r'\b(function|method|task)\b', Keyword.Declaration, 'functiondef'),
(r'\bclass\b', Keyword.Declaration, 'classdef'),
- (r'\b(safe\s+)?wait\s+for\b', Keyword, 'waitfor'),
+ (r'\b(safe(?=\s))?(\s*)(wait(?=\s))(\s+)(for)\b',
+ bygroups(Keyword, Whitespace, Keyword, Whitespace,
+ Keyword), 'waitfor'),
(r'\b(me|this)(\.[$a-zA-Z_][\w.$]*)?\b', Name.Variable.Instance),
- (r'(?<![.$])(for(\s+(parallel|series))?|in|of|while|until|'
- r'break|return|continue|'
- r'when|if|unless|else|otherwise|except\s+when|'
- r'throw|raise|fail\s+with|try|catch|finally|new|delete|'
- r'typeof|instanceof|super|run\s+in\s+parallel|'
- r'inherits\s+from)\b', Keyword),
- (r'(?<![.$])(true|false|yes|no|on|off|null|nothing|none|'
- r'NaN|Infinity|undefined)\b',
- Keyword.Constant),
- (r'(Array|Boolean|Date|Error|Function|Math|'
- r'Number|Object|RegExp|String|decodeURI|'
- r'decodeURIComponent|encodeURI|encodeURIComponent|'
- r'eval|isFinite|isNaN|isSafeInteger|parseFloat|parseInt|document|'
- r'window|globalThis|Symbol|print)\b', Name.Builtin),
- (r'[$a-zA-Z_][\w.$]*\s*(:|[+\-*/]?\=)?\b', Name.Variable),
+ (r'(?<![.$])(run)(\s+)(in)(\s+)(parallel)\b',
+ bygroups(Keyword, Whitespace, Keyword, Whitespace, Keyword)),
+ (r'(?<![.$])(for)(\s+)(parallel|series)?\b',
+ bygroups(Keyword, Whitespace, Keyword)),
+ (r'(?<![.$])(except)(\s+)(when)?\b',
+ bygroups(Keyword, Whitespace, Keyword)),
+ (r'(?<![.$])(fail)(\s+)(with)?\b',
+ bygroups(Keyword, Whitespace, Keyword)),
+ (r'(?<![.$])(inherits)(\s+)(from)?\b',
+ bygroups(Keyword, Whitespace, Keyword)),
+ (r'(?<![.$])(for)(\s+)(parallel|series)?\b',
+ bygroups(Keyword, Whitespace, Keyword)),
+ (words((
+ 'in', 'of', 'while', 'until', 'break', 'return', 'continue',
+ 'when', 'if', 'unless', 'else', 'otherwise', 'throw', 'raise',
+ 'try', 'catch', 'finally', 'new', 'delete', 'typeof',
+ 'instanceof', 'super'), prefix=r'(?<![.$])', suffix=r'\b'),
+ Keyword),
+ (words((
+ 'true', 'false', 'yes', 'no', 'on', 'off', 'null', 'nothing',
+ 'none', 'NaN', 'Infinity', 'undefined'), prefix=r'(?<![.$])',
+ suffix=r'\b'), Keyword.Constant),
+ (words((
+ 'Array', 'Boolean', 'Date', 'Error', 'Function', 'Math',
+ 'Number', 'Object', 'RegExp', 'String', 'decodeURI',
+ 'decodeURIComponent', 'encodeURI', 'encodeURIComponent', 'eval',
+ 'isFinite', 'isNaN', 'isSafeInteger', 'parseFloat', 'parseInt',
+ 'document', 'window', 'globalThis', 'Symbol', 'print'),
+ suffix=r'\b'), Name.Builtin),
+ (r'([$a-zA-Z_][\w.$]*)(\s*)(:|[+\-*/]?\=)?\b',
+ bygroups(Name.Variable, Whitespace, Operator)),
(r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float),
(r'0x[0-9a-fA-F]+', Number.Hex),
(r'[0-9]+', Number.Integer),
@@ -288,14 +309,13 @@ class KalLexer(RegexLexer):
class LiveScriptLexer(RegexLexer):
"""
- For `LiveScript`_ source code.
-
- .. _LiveScript: https://livescript.net/
+ For LiveScript source code.
.. versionadded:: 1.6
"""
name = 'LiveScript'
+ url = 'https://livescript.net/'
aliases = ['livescript', 'live-script']
filenames = ['*.ls']
mimetypes = ['text/livescript']
@@ -303,9 +323,9 @@ class LiveScriptLexer(RegexLexer):
flags = re.DOTALL
tokens = {
'commentsandwhitespace': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'/\*.*?\*/', Comment.Multiline),
- (r'#.*?\n', Comment.Single),
+ (r'(#.*?)(\n)', bygroups(Comment.Single, Whitespace)),
],
'multilineregex': [
include('commentsandwhitespace'),
@@ -346,10 +366,13 @@ class LiveScriptLexer(RegexLexer):
r'decodeURIComponent|encodeURI|encodeURIComponent|'
r'eval|isFinite|isNaN|parseFloat|parseInt|document|window|'
r'globalThis|Symbol|Symbol|BigInt)\b', Name.Builtin),
- (r'[$a-zA-Z_][\w.\-:$]*\s*[:=]\s', Name.Variable,
- 'slashstartsregex'),
- (r'@[$a-zA-Z_][\w.\-:$]*\s*[:=]\s', Name.Variable.Instance,
- 'slashstartsregex'),
+ (r'([$a-zA-Z_][\w.\-:$]*)(\s*)([:=])(\s+)',
+ bygroups(Name.Variable, Whitespace, Operator, Whitespace),
+ 'slashstartsregex'),
+ (r'(@[$a-zA-Z_][\w.\-:$]*)(\s*)([:=])(\s+)',
+ bygroups(Name.Variable.Instance, Whitespace, Operator,
+ Whitespace),
+ 'slashstartsregex'),
(r'@', Name.Other, 'slashstartsregex'),
(r'@?[$a-zA-Z_][\w-]*', Name.Other, 'slashstartsregex'),
(r'[0-9]+\.[0-9]+([eE][0-9]+)?[fd]?(?:[a-zA-Z_]+)?', Number.Float),
@@ -399,12 +422,13 @@ class LiveScriptLexer(RegexLexer):
class DartLexer(RegexLexer):
"""
- For `Dart <http://dart.dev/>`_ source code.
+ For Dart source code.
.. versionadded:: 1.5
"""
name = 'Dart'
+ url = 'http://dart.dev/'
aliases = ['dart']
filenames = ['*.dart']
mimetypes = ['text/x-dart']
@@ -417,11 +441,11 @@ class DartLexer(RegexLexer):
(r'#!(.*?)$', Comment.Preproc),
(r'\b(import|export)\b', Keyword, 'import_decl'),
(r'\b(library|source|part of|part)\b', Keyword),
- (r'[^\S\n]+', Text),
- (r'//.*?\n', Comment.Single),
+ (r'[^\S\n]+', Whitespace),
+ (r'(//.*?)(\n)', bygroups(Comment.Single, Whitespace)),
(r'/\*.*?\*/', Comment.Multiline),
(r'\b(class|extension|mixin)\b(\s+)',
- bygroups(Keyword.Declaration, Text), 'class'),
+ bygroups(Keyword.Declaration, Whitespace), 'class'),
(r'\b(as|assert|break|case|catch|const|continue|default|do|else|finally|'
r'for|if|in|is|new|rethrow|return|super|switch|this|throw|try|while)\b',
Keyword),
@@ -440,7 +464,7 @@ class DartLexer(RegexLexer):
# DIGIT+ (‘.’ DIGIT*)? EXPONENT?
(r'\d+(\.\d*)?([eE][+-]?\d+)?', Number),
(r'\.\d+([eE][+-]?\d+)?', Number), # ‘.’ DIGIT+ EXPONENT?
- (r'\n', Text)
+ (r'\n', Whitespace)
# pseudo-keyword negate intentionally left out
],
'class': [
@@ -448,7 +472,7 @@ class DartLexer(RegexLexer):
],
'import_decl': [
include('string_literal'),
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'\b(as|deferred|show|hide)\b', Keyword),
(r'[a-zA-Z_$]\w*', Name),
(r'\,', Punctuation),
@@ -502,7 +526,7 @@ class DartLexer(RegexLexer):
class LassoLexer(RegexLexer):
"""
- For `Lasso <http://www.lassosoft.com/>`_ source code, covering both Lasso 9
+ For Lasso source code, covering both Lasso 9
syntax and LassoScript for Lasso 8.6 and earlier. For Lasso embedded in
HTML, use the `LassoHtmlLexer`.
@@ -529,7 +553,7 @@ class LassoLexer(RegexLexer):
'root': [
(r'^#![ \S]+lasso9\b', Comment.Preproc, 'lasso'),
(r'(?=\[|<)', Other, 'delimiters'),
- (r'\s+', Other),
+ (r'\s+', Whitespace),
default(('delimiters', 'lassofile')),
],
'delimiters': [
@@ -565,8 +589,8 @@ class LassoLexer(RegexLexer):
include('lasso'),
],
'whitespacecomments': [
- (r'\s+', Text),
- (r'//.*?\n', Comment.Single),
+ (r'\s+', Whitespace),
+ (r'(//.*?)(\s*)$', bygroups(Comment.Single, Whitespace)),
(r'/\*\*!.*?\*/', String.Doc),
(r'/\*.*?\*/', Comment.Multiline),
],
@@ -586,18 +610,20 @@ class LassoLexer(RegexLexer):
# names
(r'\$[a-z_][\w.]*', Name.Variable),
(r'#([a-z_][\w.]*|\d+\b)', Name.Variable.Instance),
- (r"(\.\s*)('[a-z_][\w.]*')",
- bygroups(Name.Builtin.Pseudo, Name.Variable.Class)),
- (r"(self)(\s*->\s*)('[a-z_][\w.]*')",
- bygroups(Name.Builtin.Pseudo, Operator, Name.Variable.Class)),
- (r'(\.\.?\s*)([a-z_][\w.]*(=(?!=))?)',
- bygroups(Name.Builtin.Pseudo, Name.Other.Member)),
- (r'(->\\?\s*|&\s*)([a-z_][\w.]*(=(?!=))?)',
- bygroups(Operator, Name.Other.Member)),
+ (r"(\.)(\s*)('[a-z_][\w.]*')",
+ bygroups(Name.Builtin.Pseudo, Whitespace, Name.Variable.Class)),
+ (r"(self)(\s*)(->)(\s*)('[a-z_][\w.]*')",
+ bygroups(Name.Builtin.Pseudo, Whitespace, Operator, Whitespace,
+ Name.Variable.Class)),
+ (r'(\.\.?)(\s*)([a-z_][\w.]*(=(?!=))?)',
+ bygroups(Name.Builtin.Pseudo, Whitespace, Name.Other.Member)),
+ (r'(->\\?|&)(\s*)([a-z_][\w.]*(=(?!=))?)',
+ bygroups(Operator, Whitespace, Name.Other.Member)),
(r'(?<!->)(self|inherited|currentcapture|givenblock)\b',
Name.Builtin.Pseudo),
(r'-(?!infinity)[a-z_][\w.]*', Name.Attribute),
- (r'::\s*[a-z_][\w.]*', Name.Label),
+ (r'(::)(\s*)([a-z_][\w.]*)',
+ bygroups(Punctuation, Whitespace, Name.Label)),
(r'(error_(code|msg)_\w+|Error_AddError|Error_ColumnRestriction|'
r'Error_DatabaseConnectionUnavailable|Error_DatabaseTimeout|'
r'Error_DeleteError|Error_FieldRestriction|Error_FileNotFound|'
@@ -609,18 +635,20 @@ class LassoLexer(RegexLexer):
r'Error_UpdateError)\b', Name.Exception),
# definitions
- (r'(define)(\s+)([a-z_][\w.]*)(\s*=>\s*)(type|trait|thread)\b',
- bygroups(Keyword.Declaration, Text, Name.Class, Operator, Keyword)),
- (r'(define)(\s+)([a-z_][\w.]*)(\s*->\s*)([a-z_][\w.]*=?|[-+*/%])',
- bygroups(Keyword.Declaration, Text, Name.Class, Operator,
- Name.Function), 'signature'),
+ (r'(define)(\s+)([a-z_][\w.]*)(\s*)(=>)(\s*)(type|trait|thread)\b',
+ bygroups(Keyword.Declaration, Whitespace, Name.Class,
+ Whitespace, Operator, Whitespace, Keyword)),
+ (r'(define)(\s+)([a-z_][\w.]*)(\s*)(->)(\s*)([a-z_][\w.]*=?|[-+*/%])',
+ bygroups(Keyword.Declaration, Whitespace, Name.Class,
+ Whitespace, Operator, Whitespace, Name.Function),
+ 'signature'),
(r'(define)(\s+)([a-z_][\w.]*)',
- bygroups(Keyword.Declaration, Text, Name.Function), 'signature'),
+ bygroups(Keyword.Declaration, Whitespace, Name.Function), 'signature'),
(r'(public|protected|private|provide)(\s+)(([a-z_][\w.]*=?|[-+*/%])'
- r'(?=\s*\())', bygroups(Keyword, Text, Name.Function),
+ r'(?=\s*\())', bygroups(Keyword, Whitespace, Name.Function),
'signature'),
(r'(public|protected|private|provide)(\s+)([a-z_][\w.]*)',
- bygroups(Keyword, Text, Name.Function)),
+ bygroups(Keyword, Whitespace, Name.Function)),
# keywords
(r'(true|false|none|minimal|full|all|void)\b', Keyword.Constant),
@@ -628,8 +656,8 @@ class LassoLexer(RegexLexer):
(r'(array|date|decimal|duration|integer|map|pair|string|tag|xml|'
r'null|boolean|bytes|keyword|list|locale|queue|set|stack|'
r'staticarray)\b', Keyword.Type),
- (r'([a-z_][\w.]*)(\s+)(in)\b', bygroups(Name, Text, Keyword)),
- (r'(let|into)(\s+)([a-z_][\w.]*)', bygroups(Keyword, Text, Name)),
+ (r'([a-z_][\w.]*)(\s+)(in)\b', bygroups(Name, Whitespace, Keyword)),
+ (r'(let|into)(\s+)([a-z_][\w.]*)', bygroups(Keyword, Whitespace, Name)),
(r'require\b', Keyword, 'requiresection'),
(r'(/?)(Namespace_Using)\b', bygroups(Punctuation, Keyword.Namespace)),
(r'(/?)(Cache|Database_Names|Database_SchemaNames|'
@@ -655,8 +683,9 @@ class LassoLexer(RegexLexer):
# other
(r',', Punctuation, 'commamember'),
(r'(and|or|not)\b', Operator.Word),
- (r'([a-z_][\w.]*)(\s*::\s*[a-z_][\w.]*)?(\s*=(?!=))',
- bygroups(Name, Name.Label, Operator)),
+ (r'([a-z_][\w.]*)(\s*)(::)(\s*)([a-z_][\w.]*)?(\s*=(?!=))',
+ bygroups(Name, Whitespace, Punctuation, Whitespace, Name.Label,
+ Operator)),
(r'(/?)([\w.]+)', bygroups(Punctuation, Name.Other)),
(r'(=)(n?bw|n?ew|n?cn|lte?|gte?|n?eq|n?rx|ft)\b',
bygroups(Operator, Operator.Word)),
@@ -695,7 +724,8 @@ class LassoLexer(RegexLexer):
(r'(([a-z_][\w.]*=?|[-+*/%])(?=\s*\())', Name, 'requiresignature'),
(r'(([a-z_][\w.]*=?|[-+*/%])(?=(\s*::\s*[\w.]+)?\s*,))', Name),
(r'[a-z_][\w.]*=?|[-+*/%]', Name, '#pop'),
- (r'::\s*[a-z_][\w.]*', Name.Label),
+ (r'(::)(\s*)([a-z_][\w.]*)',
+ bygroups(Punctuation, Whitespace, Name.Label)),
(r',', Punctuation),
include('whitespacecomments'),
],
@@ -703,7 +733,8 @@ class LassoLexer(RegexLexer):
(r'(\)(?=(\s*::\s*[\w.]+)?\s*,))', Punctuation, '#pop'),
(r'\)', Punctuation, '#pop:2'),
(r'-?[a-z_][\w.]*', Name.Attribute),
- (r'::\s*[a-z_][\w.]*', Name.Label),
+ (r'(::)(\s*)([a-z_][\w.]*)',
+ bygroups(Punctuation, Whitespace, Name.Label)),
(r'\.\.\.', Name.Builtin.Pseudo),
(r'[(,]', Punctuation),
include('whitespacecomments'),
@@ -784,11 +815,11 @@ class ObjectiveJLexer(RegexLexer):
using(this))),
# class definition
- (r'(@interface|@implementation)(\s+)', bygroups(Keyword, Text),
+ (r'(@interface|@implementation)(\s+)', bygroups(Keyword, Whitespace),
'classname'),
- (r'(@class|@protocol)(\s*)', bygroups(Keyword, Text),
+ (r'(@class|@protocol)(\s*)', bygroups(Keyword, Whitespace),
'forward_classname'),
- (r'(\s*)(@end)(\s*)', bygroups(Text, Keyword, Text)),
+ (r'(\s*)(@end)(\s*)', bygroups(Whitespace, Keyword, Whitespace)),
include('statements'),
('[{()}]', Punctuation),
@@ -796,20 +827,20 @@ class ObjectiveJLexer(RegexLexer):
],
'whitespace': [
(r'(@import)(\s+)("(?:\\\\|\\"|[^"])*")',
- bygroups(Comment.Preproc, Text, String.Double)),
+ bygroups(Comment.Preproc, Whitespace, String.Double)),
(r'(@import)(\s+)(<(?:\\\\|\\>|[^>])*>)',
- bygroups(Comment.Preproc, Text, String.Double)),
+ bygroups(Comment.Preproc, Whitespace, String.Double)),
(r'(#(?:include|import))(\s+)("(?:\\\\|\\"|[^"])*")',
- bygroups(Comment.Preproc, Text, String.Double)),
+ bygroups(Comment.Preproc, Whitespace, String.Double)),
(r'(#(?:include|import))(\s+)(<(?:\\\\|\\>|[^>])*>)',
- bygroups(Comment.Preproc, Text, String.Double)),
+ bygroups(Comment.Preproc, Whitespace, String.Double)),
(r'#if\s+0', Comment.Preproc, 'if0'),
(r'#', Comment.Preproc, 'macro'),
- (r'\n', Text),
- (r'\s+', Text),
- (r'\\\n', Text), # line continuation
+ (r'\s+', Whitespace),
+ (r'(\\)(\n)',
+ bygroups(String.Escape, Whitespace)), # line continuation
(r'//(\n|(.|\n)*?[^\\]\n)', Comment.Single),
(r'/(\\\n)?[*](.|\n)*?[*](\\\n)?/', Comment.Multiline),
(r'<!--', Comment),
@@ -822,7 +853,7 @@ class ObjectiveJLexer(RegexLexer):
default('#pop'),
],
'badregex': [
- (r'\n', Text, '#pop'),
+ (r'\n', Whitespace, '#pop'),
],
'statements': [
(r'(L|@)?"', String, 'string'),
@@ -889,10 +920,10 @@ class ObjectiveJLexer(RegexLexer):
(r'([a-zA-Z_]\w*)', Name.Class, '#pop'),
],
'forward_classname': [
- (r'([a-zA-Z_]\w*)(\s*,\s*)',
- bygroups(Name.Class, Text), '#push'),
- (r'([a-zA-Z_]\w*)(\s*;?)',
- bygroups(Name.Class, Text), '#pop'),
+ (r'([a-zA-Z_]\w*)(\s*)(,)(\s*)',
+ bygroups(Name.Class, Whitespace, Text, Whitespace), '#push'),
+ (r'([a-zA-Z_]\w*)(\s*)(;?)',
+ bygroups(Name.Class, Whitespace, Text), '#pop'),
],
'function_signature': [
include('whitespace'),
@@ -955,21 +986,21 @@ class ObjectiveJLexer(RegexLexer):
(r'"', String, '#pop'),
(r'\\([\\abfnrtv"\']|x[a-fA-F0-9]{2,4}|[0-7]{1,3})', String.Escape),
(r'[^\\"\n]+', String), # all other characters
- (r'\\\n', String), # line continuation
+ (r'(\\)(\n)', bygroups(String.Escape, Whitespace)), # line continuation
(r'\\', String), # stray backslash
],
'macro': [
(r'[^/\n]+', Comment.Preproc),
(r'/[*](.|\n)*?[*]/', Comment.Multiline),
- (r'//.*?\n', Comment.Single, '#pop'),
+ (r'(//.*?)(\n)', bygroups(Comment.Single, Whitespace), '#pop'),
(r'/', Comment.Preproc),
- (r'(?<=\\)\n', Comment.Preproc),
- (r'\n', Comment.Preproc, '#pop'),
+ (r'(?<=\\)\n', Whitespace),
+ (r'\n', Whitespace, '#pop'),
],
'if0': [
(r'^\s*#if.*?(?<!\\)\n', Comment.Preproc, '#push'),
(r'^\s*#endif.*?(?<!\\)\n', Comment.Preproc, '#pop'),
- (r'.*?\n', Comment),
+ (r'(.*?)(\n)', bygroups(Comment, Whitespace)),
]
}
@@ -982,14 +1013,13 @@ class ObjectiveJLexer(RegexLexer):
class CoffeeScriptLexer(RegexLexer):
"""
- For `CoffeeScript`_ source code.
-
- .. _CoffeeScript: http://coffeescript.org
+ For CoffeeScript source code.
.. versionadded:: 1.3
"""
name = 'CoffeeScript'
+ url = 'http://coffeescript.org'
aliases = ['coffeescript', 'coffee-script', 'coffee']
filenames = ['*.coffee']
mimetypes = ['text/coffeescript']
@@ -1002,9 +1032,9 @@ class CoffeeScriptLexer(RegexLexer):
flags = re.DOTALL
tokens = {
'commentsandwhitespace': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'###[^#].*?###', Comment.Multiline),
- (r'#(?!##[^#]).*?\n', Comment.Single),
+ (r'(#(?!##[^#]).*?)(\n)', bygroups(Comment.Single, Whitespace)),
],
'multilineregex': [
(r'[^/#]+', String.Regex),
@@ -1043,10 +1073,12 @@ class CoffeeScriptLexer(RegexLexer):
r'decodeURIComponent|encodeURI|encodeURIComponent|'
r'eval|isFinite|isNaN|parseFloat|parseInt|document|window|globalThis|Symbol)\b',
Name.Builtin),
- (r'[$a-zA-Z_][\w.:$]*\s*[:=]\s', Name.Variable,
- 'slashstartsregex'),
- (r'@[$a-zA-Z_][\w.:$]*\s*[:=]\s', Name.Variable.Instance,
- 'slashstartsregex'),
+ (r'([$a-zA-Z_][\w.:$]*)(\s*)([:=])(\s+)',
+ bygroups(Name.Variable, Whitespace, Operator, Whitespace),
+ 'slashstartsregex'),
+ (r'(@[$a-zA-Z_][\w.:$]*)(\s*)([:=])(\s+)',
+ bygroups(Name.Variable.Instance, Whitespace, Operator, Whitespace),
+ 'slashstartsregex'),
(r'@', Name.Other, 'slashstartsregex'),
(r'@?[$a-zA-Z_][\w$]*', Name.Other),
(r'[0-9][0-9]*\.[0-9]+([eE][0-9]+)?[fd]?', Number.Float),
@@ -1095,11 +1127,12 @@ class CoffeeScriptLexer(RegexLexer):
class MaskLexer(RegexLexer):
"""
- For `Mask <https://github.com/atmajs/MaskJS>`__ markup.
+ For Mask markup.
.. versionadded:: 2.0
"""
name = 'Mask'
+ url = 'https://github.com/atmajs/MaskJS'
aliases = ['mask']
filenames = ['*.mask']
mimetypes = ['text/x-mask']
@@ -1107,8 +1140,8 @@ class MaskLexer(RegexLexer):
flags = re.MULTILINE | re.IGNORECASE | re.DOTALL
tokens = {
'root': [
- (r'\s+', Text),
- (r'//.*?\n', Comment.Single),
+ (r'\s+', Whitespace),
+ (r'(//.*?)(\n)', bygroups(Comment.Single, Whitespace)),
(r'/\*.*?\*/', Comment.Multiline),
(r'[{};>]', Punctuation),
(r"'''", String, 'string-trpl-single'),
@@ -1143,19 +1176,23 @@ class MaskLexer(RegexLexer):
],
'interpolation': [
(r'\]', String.Interpol, '#pop'),
- (r'\s*:', String.Interpol, 'expression'),
- (r'\s*\w+:', Name.Other),
+ (r'(\s*)(:)', bygroups(Whitespace, String.Interpol), 'expression'),
+ (r'(\s*)(\w+)(:)', bygroups(Whitespace, Name.Other, Punctuation)),
(r'[^\]]+', String.Interpol)
],
'expression': [
(r'[^\]]+', using(JavascriptLexer), '#pop')
],
'node': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'\.', Name.Variable.Class, 'node-class'),
(r'\#', Name.Function, 'node-id'),
- (r'style[ \t]*=', Name.Attribute, 'node-attr-style-value'),
- (r'[\w:-]+[ \t]*=', Name.Attribute, 'node-attr-value'),
+ (r'(style)([ \t]*)(=)',
+ bygroups(Name.Attribute, Whitespace, Operator),
+ 'node-attr-style-value'),
+ (r'([\w:-]+)([ \t]*)(=)',
+ bygroups(Name.Attribute, Whitespace, Operator),
+ 'node-attr-value'),
(r'[\w:-]+', Name.Attribute),
(r'[>{;]', Punctuation, '#pop')
],
@@ -1170,20 +1207,20 @@ class MaskLexer(RegexLexer):
default('#pop')
],
'node-attr-value': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'\w+', Name.Variable, '#pop'),
(r"'", String, 'string-single-pop2'),
(r'"', String, 'string-double-pop2'),
default('#pop')
],
'node-attr-style-value': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r"'", String.Single, 'css-single-end'),
(r'"', String.Single, 'css-double-end'),
include('node-attr-value')
],
'css-base': [
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r";", Punctuation),
(r"[\w\-]+\s*:", Name.Builtin)
],
@@ -1210,9 +1247,7 @@ class MaskLexer(RegexLexer):
class EarlGreyLexer(RegexLexer):
"""
- For `Earl-Grey`_ source code.
-
- .. _Earl-Grey: https://breuleux.github.io/earl-grey/
+ For Earl-Grey source code.
.. versionadded: 2.1
"""
@@ -1224,12 +1259,12 @@ class EarlGreyLexer(RegexLexer):
tokens = {
'root': [
- (r'\n', Text),
+ (r'\n', Whitespace),
include('control'),
(r'[^\S\n]+', Text),
- (r';;.*\n', Comment),
+ (r'(;;.*)(\n)', bygroups(Comment, Whitespace)),
(r'[\[\]{}:(),;]', Punctuation),
- (r'\\\n', Text),
+ (r'(\\)(\n)', bygroups(String.Escape, Whitespace)),
(r'\\', Text),
include('errors'),
(words((
@@ -1268,7 +1303,7 @@ class EarlGreyLexer(RegexLexer):
bygroups(Punctuation, Name.Class.DBS)),
(r'(\[)([\^#][a-zA-Z$_](?:[\w$\-]*[\w$])?)(\])',
bygroups(Punctuation, Name.Entity.DBS, Punctuation)),
- (r'\s+', Text),
+ (r'\s+', Whitespace),
(r'%', Operator.DBS, '#pop'),
],
'import_paths': [
@@ -1298,8 +1333,8 @@ class EarlGreyLexer(RegexLexer):
(?!and|as|each\*|each|in|is|mod|of|or|when|where|with)
(?=(?:[+\-*/~^<>%&|?!@#.])?[a-zA-Z$_](?:[\w$-]*[\w$])?)''',
Keyword.Control),
- (r'([a-zA-Z$_](?:[\w$-]*[\w$])?)(?!\n)\s+(?=[\'"\d{\[(])',
- Keyword.Control),
+ (r'([a-zA-Z$_](?:[\w$-]*[\w$])?)(?!\n)(\s+)(?=[\'"\d{\[(])',
+ bygroups(Keyword.Control, Whitespace)),
(r'''(?x)
(?:
(?<=[%=])|
@@ -1308,11 +1343,11 @@ class EarlGreyLexer(RegexLexer):
(?<=each\*|where)
)(\s+)
([a-zA-Z$_](?:[\w$-]*[\w$])?)(:)''',
- bygroups(Text, Keyword.Control, Punctuation)),
+ bygroups(Whitespace, Keyword.Control, Punctuation)),
(r'''(?x)
(?<![+\-*/~^<>%&|?!@#.])(\s+)
([a-zA-Z$_](?:[\w$-]*[\w$])?)(:)''',
- bygroups(Text, Keyword.Control, Punctuation)),
+ bygroups(Whitespace, Keyword.Control, Punctuation)),
],
'nested': [
(r'''(?x)
@@ -1426,25 +1461,24 @@ class EarlGreyLexer(RegexLexer):
class JuttleLexer(RegexLexer):
"""
- For `Juttle`_ source code.
-
- .. _Juttle: https://github.com/juttle/juttle
+ For Juttle source code.
.. versionadded:: 2.2
"""
name = 'Juttle'
+ url = 'http://juttle.github.io/'
aliases = ['juttle']
filenames = ['*.juttle']
mimetypes = ['application/juttle', 'application/x-juttle',
'text/x-juttle', 'text/juttle']
- flags = re.DOTALL | re.UNICODE | re.MULTILINE
+ flags = re.DOTALL | re.MULTILINE
tokens = {
'commentsandwhitespace': [
- (r'\s+', Text),
- (r'//.*?\n', Comment.Single),
+ (r'\s+', Whitespace),
+ (r'(//.*?)(\n)', bygroups(Comment.Single, Whitespace)),
(r'/\*.*?\*/', Comment.Multiline)
],
'slashstartsregex': [
@@ -1525,8 +1559,9 @@ class NodeConsoleLexer(Lexer):
for match in line_re.finditer(text):
line = match.group()
if line.startswith('> '):
- insertions.append((len(curcode),
- [(0, Generic.Prompt, line[:2])]))
+ insertions.append((len(curcode),
+ [(0, Generic.Prompt, line[:1]),
+ (1, Whitespace, line[1:2])]))
curcode += line[2:]
elif line.startswith('...'):
@@ -1534,21 +1569,21 @@ class NodeConsoleLexer(Lexer):
code = line.lstrip('.')
lead = len(line) - len(code)
- insertions.append((len(curcode),
+ insertions.append((len(curcode),
[(0, Generic.Prompt, line[:lead])]))
curcode += code
else:
if curcode:
- yield from do_insertions(insertions,
+ yield from do_insertions(insertions,
jslexer.get_tokens_unprocessed(curcode))
curcode = ''
insertions = []
- yield from do_insertions([],
+ yield from do_insertions([],
jslexer.get_tokens_unprocessed(line))
if curcode:
- yield from do_insertions(insertions,
+ yield from do_insertions(insertions,
jslexer.get_tokens_unprocessed(curcode))