Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.11/site-packages/pygments/lexers/perl.py: 32%

1""" 

2 pygments.lexers.perl 

3 ~~~~~~~~~~~~~~~~~~~~ 

4 

5 Lexers for Perl, Raku and related languages. 

6 

7 :copyright: Copyright 2006-2025 by the Pygments team, see AUTHORS. 

8 :license: BSD, see LICENSE for details. 

9""" 

10 

11import re 

12 

13from pygments.lexer import RegexLexer, ExtendedRegexLexer, include, bygroups, \ 

14 using, this, default, words 

15from pygments.token import Text, Comment, Operator, Keyword, Name, String, \ 

16 Number, Punctuation, Whitespace 

17from pygments.util import shebang_matches 

18 

19__all__ = ['PerlLexer', 'Perl6Lexer'] 

20 

21 

22class PerlLexer(RegexLexer): 

23 """ 

24 For Perl source code. 

25 """ 

26 

27 name = 'Perl' 

28 url = 'https://www.perl.org' 

29 aliases = ['perl', 'pl'] 

30 filenames = ['*.pl', '*.pm', '*.t', '*.perl'] 

31 mimetypes = ['text/x-perl', 'application/x-perl'] 

32 version_added = '' 

33 

34 flags = re.DOTALL | re.MULTILINE 

35 # TODO: give this to a perl guy who knows how to parse perl... 

36 tokens = { 

37 'balanced-regex': [ 

38 (r'/(\\\\|\\[^\\]|[^\\/])*/[egimosx]*', String.Regex, '#pop'), 

39 (r'!(\\\\|\\[^\\]|[^\\!])*![egimosx]*', String.Regex, '#pop'), 

40 (r'\\(\\\\|[^\\])*\\[egimosx]*', String.Regex, '#pop'), 

41 (r'\{(\\\\|\\[^\\]|[^\\}])*\}[egimosx]*', String.Regex, '#pop'), 

42 (r'<(\\\\|\\[^\\]|[^\\>])*>[egimosx]*', String.Regex, '#pop'), 

43 (r'\[(\\\\|\\[^\\]|[^\\\]])*\][egimosx]*', String.Regex, '#pop'), 

44 (r'\((\\\\|\\[^\\]|[^\\)])*\)[egimosx]*', String.Regex, '#pop'), 

45 (r'@(\\\\|\\[^\\]|[^\\@])*@[egimosx]*', String.Regex, '#pop'), 

46 (r'%(\\\\|\\[^\\]|[^\\%])*%[egimosx]*', String.Regex, '#pop'), 

47 (r'\$(\\\\|\\[^\\]|[^\\$])*\$[egimosx]*', String.Regex, '#pop'), 

48 ], 

49 'root': [ 

50 (r'\A\#!.+?$', Comment.Hashbang), 

51 (r'\#.*?$', Comment.Single), 

52 (r'^=[a-zA-Z0-9]+\s+.*?\n=cut', Comment.Multiline), 

53 (words(( 

54 'case', 'continue', 'do', 'else', 'elsif', 'for', 'foreach', 

55 'if', 'last', 'my', 'next', 'our', 'redo', 'reset', 'then', 

56 'unless', 'until', 'while', 'print', 'new', 'BEGIN', 

57 'CHECK', 'INIT', 'END', 'return'), suffix=r'\b'), 

58 Keyword), 

59 (r'(format)(\s+)(\w+)(\s*)(=)(\s*\n)', 

60 bygroups(Keyword, Whitespace, Name, Whitespace, Punctuation, Whitespace), 'format'), 

61 (r'(eq|lt|gt|le|ge|ne|not|and|or|cmp)\b', Operator.Word), 

62 # common delimiters 

63 (r's/(\\\\|\\[^\\]|[^\\/])*/(\\\\|\\[^\\]|[^\\/])*/[egimosx]*', 

64 String.Regex), 

65 (r's!(\\\\|\\!|[^!])*!(\\\\|\\!|[^!])*![egimosx]*', String.Regex), 

66 (r's\\(\\\\|[^\\])*\\(\\\\|[^\\])*\\[egimosx]*', String.Regex), 

67 (r's@(\\\\|\\[^\\]|[^\\@])*@(\\\\|\\[^\\]|[^\\@])*@[egimosx]*', 

68 String.Regex), 

69 (r's%(\\\\|\\[^\\]|[^\\%])*%(\\\\|\\[^\\]|[^\\%])*%[egimosx]*', 

70 String.Regex), 

71 # balanced delimiters 

72 (r's\{(\\\\|\\[^\\]|[^\\}])*\}\s*', String.Regex, 'balanced-regex'), 

73 (r's<(\\\\|\\[^\\]|[^\\>])*>\s*', String.Regex, 'balanced-regex'), 

74 (r's\[(\\\\|\\[^\\]|[^\\\]])*\]\s*', String.Regex, 

75 'balanced-regex'), 

76 (r's\((\\\\|\\[^\\]|[^\\)])*\)\s*', String.Regex, 

77 'balanced-regex'), 

78 

79 (r'm?/(\\\\|\\[^\\]|[^\\/\n])*/[gcimosx]*', String.Regex), 

80 (r'm(?=[/!\\{<\[(@%$])', String.Regex, 'balanced-regex'), 

81 (r'((?<==~)|(?<=\())\s*/(\\\\|\\[^\\]|[^\\/])*/[gcimosx]*', 

82 String.Regex), 

83 (r'\s+', Whitespace), 

84 (words(( 

85 'abs', 'accept', 'alarm', 'atan2', 'bind', 'binmode', 'bless', 'caller', 'chdir', 

86 'chmod', 'chomp', 'chop', 'chown', 'chr', 'chroot', 'close', 'closedir', 'connect', 

87 'continue', 'cos', 'crypt', 'dbmclose', 'dbmopen', 'defined', 'delete', 'die', 

88 'dump', 'each', 'endgrent', 'endhostent', 'endnetent', 'endprotoent', 

89 'endpwent', 'endservent', 'eof', 'eval', 'exec', 'exists', 'exit', 'exp', 'fcntl', 

90 'fileno', 'flock', 'fork', 'format', 'formline', 'getc', 'getgrent', 'getgrgid', 

91 'getgrnam', 'gethostbyaddr', 'gethostbyname', 'gethostent', 'getlogin', 

92 'getnetbyaddr', 'getnetbyname', 'getnetent', 'getpeername', 'getpgrp', 

93 'getppid', 'getpriority', 'getprotobyname', 'getprotobynumber', 

94 'getprotoent', 'getpwent', 'getpwnam', 'getpwuid', 'getservbyname', 

95 'getservbyport', 'getservent', 'getsockname', 'getsockopt', 'glob', 'gmtime', 

96 'goto', 'grep', 'hex', 'import', 'index', 'int', 'ioctl', 'join', 'keys', 'kill', 'last', 

97 'lc', 'lcfirst', 'length', 'link', 'listen', 'local', 'localtime', 'log', 'lstat', 

98 'map', 'mkdir', 'msgctl', 'msgget', 'msgrcv', 'msgsnd', 'my', 'next', 'oct', 'open', 

99 'opendir', 'ord', 'our', 'pack', 'pipe', 'pop', 'pos', 'printf', 

100 'prototype', 'push', 'quotemeta', 'rand', 'read', 'readdir', 

101 'readline', 'readlink', 'readpipe', 'recv', 'redo', 'ref', 'rename', 

102 'reverse', 'rewinddir', 'rindex', 'rmdir', 'scalar', 'seek', 'seekdir', 

103 'select', 'semctl', 'semget', 'semop', 'send', 'setgrent', 'sethostent', 'setnetent', 

104 'setpgrp', 'setpriority', 'setprotoent', 'setpwent', 'setservent', 

105 'setsockopt', 'shift', 'shmctl', 'shmget', 'shmread', 'shmwrite', 'shutdown', 

106 'sin', 'sleep', 'socket', 'socketpair', 'sort', 'splice', 'split', 'sprintf', 'sqrt', 

107 'srand', 'stat', 'study', 'substr', 'symlink', 'syscall', 'sysopen', 'sysread', 

108 'sysseek', 'system', 'syswrite', 'tell', 'telldir', 'tie', 'tied', 'time', 'times', 'tr', 

109 'truncate', 'uc', 'ucfirst', 'umask', 'undef', 'unlink', 'unpack', 'unshift', 'untie', 

110 'utime', 'values', 'vec', 'wait', 'waitpid', 'wantarray', 'warn', 'write'), suffix=r'\b'), 

111 Name.Builtin), 

112 (r'((__(DATA|DIE|WARN)__)|(STD(IN|OUT|ERR)))\b', Name.Builtin.Pseudo), 

113 (r'(<<)([\'"]?)([a-zA-Z_]\w*)(\2;?\n.*?\n)(\3)(\n)', 

114 bygroups(String, String, String.Delimiter, String, String.Delimiter, Whitespace)), 

115 (r'__END__', Comment.Preproc, 'end-part'), 

116 (r'\$\^[ADEFHILMOPSTWX]', Name.Variable.Global), 

117 (r"\$[\\\"\[\]'&`+*.,;=%~?@$!<>(^|/-](?!\w)", Name.Variable.Global), 

118 (r'[$@%#]+', Name.Variable, 'varname'), 

119 (r'0_?[0-7]+(_[0-7]+)*', Number.Oct), 

120 (r'0x[0-9A-Fa-f]+(_[0-9A-Fa-f]+)*', Number.Hex), 

121 (r'0b[01]+(_[01]+)*', Number.Bin), 

122 (r'(?i)(\d*(_\d*)*\.\d+(_\d*)*|\d+(_\d*)*\.\d+(_\d*)*)(e[+-]?\d+)?', 

123 Number.Float), 

124 (r'(?i)\d+(_\d*)*e[+-]?\d+(_\d*)*', Number.Float), 

125 (r'\d+(_\d+)*', Number.Integer), 

126 (r"'(\\\\|\\[^\\]|[^'\\])*'", String), 

127 (r'"(\\\\|\\[^\\]|[^"\\])*"', String), 

128 (r'`(\\\\|\\[^\\]|[^`\\])*`', String.Backtick), 

129 (r'<([^\s>]+)>', String.Regex), 

130 (r'(q|qq|qw|qr|qx)\{', String.Other, 'cb-string'), 

131 (r'(q|qq|qw|qr|qx)\(', String.Other, 'rb-string'), 

132 (r'(q|qq|qw|qr|qx)\[', String.Other, 'sb-string'), 

133 (r'(q|qq|qw|qr|qx)\<', String.Other, 'lt-string'), 

134 (r'(q|qq|qw|qr|qx)([\W_])(.|\n)*?\2', String.Other), 

135 (r'(package)(\s+)([a-zA-Z_]\w*(?:::[a-zA-Z_]\w*)*)', 

136 bygroups(Keyword, Whitespace, Name.Namespace)), 

137 (r'(use|require|no)(\s+)([a-zA-Z_]\w*(?:::[a-zA-Z_]\w*)*)', 

138 bygroups(Keyword, Whitespace, Name.Namespace)), 

139 (r'(sub)(\s+)', bygroups(Keyword, Whitespace), 'funcname'), 

140 (words(( 

141 'no', 'package', 'require', 'use'), suffix=r'\b'), 

142 Keyword), 

143 (r'(\[\]|\*\*|::|<<|>>|>=|<=>|<=|={3}|!=|=~|' 

144 r'!~|&&?|\|\||\.{1,3})', Operator), 

145 (r'[-+/*%=<>&^|!\\~]=?', Operator), 

146 (r'[()\[\]:;,<>/?{}]', Punctuation), # yes, there's no shortage 

147 # of punctuation in Perl! 

148 (r'(?=\w)', Name, 'name'), 

149 ], 

150 'format': [ 

151 (r'\.\n', String.Interpol, '#pop'), 

152 (r'[^\n]*\n', String.Interpol), 

153 ], 

154 'varname': [ 

155 (r'\s+', Whitespace), 

156 (r'\{', Punctuation, '#pop'), # hash syntax? 

157 (r'\)|,', Punctuation, '#pop'), # argument specifier 

158 (r'\w+::', Name.Namespace), 

159 (r'[\w:]+', Name.Variable, '#pop'), 

160 ], 

161 'name': [ 

162 (r'[a-zA-Z_]\w*(::[a-zA-Z_]\w*)*(::)?(?=\s*->)', Name.Namespace, '#pop'), 

163 (r'[a-zA-Z_]\w*(::[a-zA-Z_]\w*)*::', Name.Namespace, '#pop'), 

164 (r'[\w:]+', Name, '#pop'), 

165 (r'[A-Z_]+(?=\W)', Name.Constant, '#pop'), 

166 (r'(?=\W)', Text, '#pop'), 

167 ], 

168 'funcname': [ 

169 (r'[a-zA-Z_]\w*[!?]?', Name.Function), 

170 (r'\s+', Whitespace), 

171 # argument declaration 

172 (r'(\([$@%]*\))(\s*)', bygroups(Punctuation, Whitespace)), 

173 (r';', Punctuation, '#pop'), 

174 (r'.*?\{', Punctuation, '#pop'), 

175 ], 

176 'cb-string': [ 

177 (r'\\[{}\\]', String.Other), 

178 (r'\\', String.Other), 

179 (r'\{', String.Other, 'cb-string'), 

180 (r'\}', String.Other, '#pop'), 

181 (r'[^{}\\]+', String.Other) 

182 ], 

183 'rb-string': [ 

184 (r'\\[()\\]', String.Other), 

185 (r'\\', String.Other), 

186 (r'\(', String.Other, 'rb-string'), 

187 (r'\)', String.Other, '#pop'), 

188 (r'[^()]+', String.Other) 

189 ], 

190 'sb-string': [ 

191 (r'\\[\[\]\\]', String.Other), 

192 (r'\\', String.Other), 

193 (r'\[', String.Other, 'sb-string'), 

194 (r'\]', String.Other, '#pop'), 

195 (r'[^\[\]]+', String.Other) 

196 ], 

197 'lt-string': [ 

198 (r'\\[<>\\]', String.Other), 

199 (r'\\', String.Other), 

200 (r'\<', String.Other, 'lt-string'), 

201 (r'\>', String.Other, '#pop'), 

202 (r'[^<>]+', String.Other) 

203 ], 

204 'end-part': [ 

205 (r'.+', Comment.Preproc, '#pop') 

206 ] 

207 } 

208 

209 def analyse_text(text): 

210 if shebang_matches(text, r'perl'): 

211 return True 

212 

213 result = 0 

214 

215 if re.search(r'(?:my|our)\s+[$@%(]', text): 

216 result += 0.9 

217 

218 if ':=' in text: 

219 # := is not valid Perl, but it appears in unicon, so we should 

220 # become less confident if we think we found Perl with := 

221 result /= 2 

222 

223 return result 

224 

225 

226class Perl6Lexer(ExtendedRegexLexer): 

227 """ 

228 For Raku (a.k.a. Perl 6) source code. 

229 """ 

230 

231 name = 'Perl6' 

232 url = 'https://www.raku.org' 

233 aliases = ['perl6', 'pl6', 'raku'] 

234 filenames = ['*.pl', '*.pm', '*.nqp', '*.p6', '*.6pl', '*.p6l', '*.pl6', 

235 '*.6pm', '*.p6m', '*.pm6', '*.t', '*.raku', '*.rakumod', 

236 '*.rakutest', '*.rakudoc'] 

237 mimetypes = ['text/x-perl6', 'application/x-perl6'] 

238 version_added = '2.0' 

239 flags = re.MULTILINE | re.DOTALL 

240 

241 PERL6_IDENTIFIER_RANGE = r"['\w:-]" 

242 

243 PERL6_KEYWORDS = ( 

244 #Phasers 

245 'BEGIN','CATCH','CHECK','CLOSE','CONTROL','DOC','END','ENTER','FIRST', 

246 'INIT','KEEP','LAST','LEAVE','NEXT','POST','PRE','QUIT','UNDO', 

247 #Keywords 

248 'anon','augment','but','class','constant','default','does','else', 

249 'elsif','enum','for','gather','given','grammar','has','if','import', 

250 'is','let','loop','made','make','method','module','multi','my','need', 

251 'orwith','our','proceed','proto','repeat','require','return', 

252 'return-rw','returns','role','rule','state','sub','submethod','subset', 

253 'succeed','supersede','token','try','unit','unless','until','use', 

254 'when','while','with','without', 

255 #Traits 

256 'export','native','repr','required','rw','symbol', 

257 ) 

258 

259 PERL6_BUILTINS = ( 

260 'ACCEPTS','abs','abs2rel','absolute','accept','accessed','acos', 

261 'acosec','acosech','acosh','acotan','acotanh','acquire','act','action', 

262 'actions','add','add_attribute','add_enum_value','add_fallback', 

263 'add_method','add_parent','add_private_method','add_role','add_trustee', 

264 'adverb','after','all','allocate','allof','allowed','alternative-names', 

265 'annotations','antipair','antipairs','any','anyof','app_lifetime', 

266 'append','arch','archname','args','arity','Array','asec','asech','asin', 

267 'asinh','ASSIGN-KEY','ASSIGN-POS','assuming','ast','at','atan','atan2', 

268 'atanh','AT-KEY','atomic-assign','atomic-dec-fetch','atomic-fetch', 

269 'atomic-fetch-add','atomic-fetch-dec','atomic-fetch-inc', 

270 'atomic-fetch-sub','atomic-inc-fetch','AT-POS','attributes','auth', 

271 'await','backtrace','Bag','BagHash','bail-out','base','basename', 

272 'base-repeating','batch','BIND-KEY','BIND-POS','bind-stderr', 

273 'bind-stdin','bind-stdout','bind-udp','bits','bless','block','Bool', 

274 'bool-only','bounds','break','Bridge','broken','BUILD','build-date', 

275 'bytes','cache','callframe','calling-package','CALL-ME','callsame', 

276 'callwith','can','cancel','candidates','cando','can-ok','canonpath', 

277 'caps','caption','Capture','cas','catdir','categorize','categorize-list', 

278 'catfile','catpath','cause','ceiling','cglobal','changed','Channel', 

279 'chars','chdir','child','child-name','child-typename','chmod','chomp', 

280 'chop','chr','chrs','chunks','cis','classify','classify-list','cleanup', 

281 'clone','close','closed','close-stdin','cmp-ok','code','codes','collate', 

282 'column','comb','combinations','command','comment','compiler','Complex', 

283 'compose','compose_type','composer','condition','config', 

284 'configure_destroy','configure_type_checking','conj','connect', 

285 'constraints','construct','contains','contents','copy','cos','cosec', 

286 'cosech','cosh','cotan','cotanh','count','count-only','cpu-cores', 

287 'cpu-usage','CREATE','create_type','cross','cue','curdir','curupdir','d', 

288 'Date','DateTime','day','daycount','day-of-month','day-of-week', 

289 'day-of-year','days-in-month','declaration','decode','decoder','deepmap', 

290 'default','defined','DEFINITE','delayed','DELETE-KEY','DELETE-POS', 

291 'denominator','desc','DESTROY','destroyers','devnull','diag', 

292 'did-you-mean','die','dies-ok','dir','dirname','dir-sep','DISTROnames', 

293 'do','does','does-ok','done','done-testing','duckmap','dynamic','e', 

294 'eager','earlier','elems','emit','enclosing','encode','encoder', 

295 'encoding','end','ends-with','enum_from_value','enum_value_list', 

296 'enum_values','enums','eof','EVAL','eval-dies-ok','EVALFILE', 

297 'eval-lives-ok','exception','excludes-max','excludes-min','EXISTS-KEY', 

298 'EXISTS-POS','exit','exitcode','exp','expected','explicitly-manage', 

299 'expmod','extension','f','fail','fails-like','fc','feature','file', 

300 'filename','find_method','find_method_qualified','finish','first','flat', 

301 'flatmap','flip','floor','flunk','flush','fmt','format','formatter', 

302 'freeze','from','from-list','from-loop','from-posix','full', 

303 'full-barrier','get','get_value','getc','gist','got','grab','grabpairs', 

304 'grep','handle','handled','handles','hardware','has_accessor','Hash', 

305 'head','headers','hh-mm-ss','hidden','hides','hour','how','hyper','id', 

306 'illegal','im','in','indent','index','indices','indir','infinite', 

307 'infix','infix:<+>','infix:<->','install_method_cache','Instant', 

308 'instead','Int','int-bounds','interval','in-timezone','invalid-str', 

309 'invert','invocant','IO','IO::Notification.watch-path','is_trusted', 

310 'is_type','isa','is-absolute','isa-ok','is-approx','is-deeply', 

311 'is-hidden','is-initial-thread','is-int','is-lazy','is-leap-year', 

312 'isNaN','isnt','is-prime','is-relative','is-routine','is-setting', 

313 'is-win','item','iterator','join','keep','kept','KERNELnames','key', 

314 'keyof','keys','kill','kv','kxxv','l','lang','last','lastcall','later', 

315 'lazy','lc','leading','level','like','line','lines','link','List', 

316 'listen','live','lives-ok','local','lock','log','log10','lookup','lsb', 

317 'made','MAIN','make','Map','match','max','maxpairs','merge','message', 

318 'method','method_table','methods','migrate','min','minmax','minpairs', 

319 'minute','misplaced','Mix','MixHash','mkdir','mode','modified','month', 

320 'move','mro','msb','multi','multiness','my','name','named','named_names', 

321 'narrow','nativecast','native-descriptor','nativesizeof','new','new_type', 

322 'new-from-daycount','new-from-pairs','next','nextcallee','next-handle', 

323 'nextsame','nextwith','NFC','NFD','NFKC','NFKD','nl-in','nl-out', 

324 'nodemap','nok','none','norm','not','note','now','nude','Num', 

325 'numerator','Numeric','of','offset','offset-in-hours','offset-in-minutes', 

326 'ok','old','on-close','one','on-switch','open','opened','operation', 

327 'optional','ord','ords','orig','os-error','osname','out-buffer','pack', 

328 'package','package-kind','package-name','packages','pair','pairs', 

329 'pairup','parameter','params','parent','parent-name','parents','parse', 

330 'parse-base','parsefile','parse-names','parts','pass','path','path-sep', 

331 'payload','peer-host','peer-port','periods','perl','permutations','phaser', 

332 'pick','pickpairs','pid','placeholder','plan','plus','polar','poll', 

333 'polymod','pop','pos','positional','posix','postfix','postmatch', 

334 'precomp-ext','precomp-target','pred','prefix','prematch','prepend', 

335 'print','printf','print-nl','print-to','private','private_method_table', 

336 'proc','produce','Promise','prompt','protect','pull-one','push', 

337 'push-all','push-at-least','push-exactly','push-until-lazy','put', 

338 'qualifier-type','quit','r','race','radix','rand','range','Rat','raw', 

339 're','read','readchars','readonly','ready','Real','reallocate','reals', 

340 'reason','rebless','receive','recv','redispatcher','redo','reduce', 

341 'rel2abs','relative','release','rename','repeated','replacement', 

342 'report','reserved','resolve','restore','result','resume','rethrow', 

343 'reverse','right','rindex','rmdir','role','roles_to_compose','rolish', 

344 'roll','rootdir','roots','rotate','rotor','round','roundrobin', 

345 'routine-type','run','rwx','s','samecase','samemark','samewith','say', 

346 'schedule-on','scheduler','scope','sec','sech','second','seek','self', 

347 'send','Set','set_hidden','set_name','set_package','set_rw','set_value', 

348 'SetHash','set-instruments','setup_finalization','shape','share','shell', 

349 'shift','sibling','sigil','sign','signal','signals','signature','sin', 

350 'sinh','sink','sink-all','skip','skip-at-least','skip-at-least-pull-one', 

351 'skip-one','skip-rest','sleep','sleep-timer','sleep-until','Slip','slurp', 

352 'slurp-rest','slurpy','snap','snapper','so','socket-host','socket-port', 

353 'sort','source','source-package','spawn','SPEC','splice','split', 

354 'splitdir','splitpath','sprintf','spurt','sqrt','squish','srand','stable', 

355 'start','started','starts-with','status','stderr','stdout','Str', 

356 'sub_signature','subbuf','subbuf-rw','subname','subparse','subst', 

357 'subst-mutate','substr','substr-eq','substr-rw','subtest','succ','sum', 

358 'Supply','symlink','t','tail','take','take-rw','tan','tanh','tap', 

359 'target','target-name','tc','tclc','tell','then','throttle','throw', 

360 'throws-like','timezone','tmpdir','to','today','todo','toggle','to-posix', 

361 'total','trailing','trans','tree','trim','trim-leading','trim-trailing', 

362 'truncate','truncated-to','trusts','try_acquire','trying','twigil','type', 

363 'type_captures','typename','uc','udp','uncaught_handler','unimatch', 

364 'uniname','uninames','uniparse','uniprop','uniprops','unique','unival', 

365 'univals','unlike','unlink','unlock','unpack','unpolar','unshift', 

366 'unwrap','updir','USAGE','use-ok','utc','val','value','values','VAR', 

367 'variable','verbose-config','version','VMnames','volume','vow','w','wait', 

368 'warn','watch','watch-path','week','weekday-of-month','week-number', 

369 'week-year','WHAT','when','WHERE','WHEREFORE','WHICH','WHO', 

370 'whole-second','WHY','wordcase','words','workaround','wrap','write', 

371 'write-to','x','yada','year','yield','yyyy-mm-dd','z','zip','zip-latest', 

372 

373 ) 

374 

375 PERL6_BUILTIN_CLASSES = ( 

376 #Booleans 

377 'False','True', 

378 #Classes 

379 'Any','Array','Associative','AST','atomicint','Attribute','Backtrace', 

380 'Backtrace::Frame','Bag','Baggy','BagHash','Blob','Block','Bool','Buf', 

381 'Callable','CallFrame','Cancellation','Capture','CArray','Channel','Code', 

382 'compiler','Complex','ComplexStr','Cool','CurrentThreadScheduler', 

383 'Cursor','Date','Dateish','DateTime','Distro','Duration','Encoding', 

384 'Exception','Failure','FatRat','Grammar','Hash','HyperWhatever','Instant', 

385 'Int','int16','int32','int64','int8','IntStr','IO','IO::ArgFiles', 

386 'IO::CatHandle','IO::Handle','IO::Notification','IO::Path', 

387 'IO::Path::Cygwin','IO::Path::QNX','IO::Path::Unix','IO::Path::Win32', 

388 'IO::Pipe','IO::Socket','IO::Socket::Async','IO::Socket::INET','IO::Spec', 

389 'IO::Spec::Cygwin','IO::Spec::QNX','IO::Spec::Unix','IO::Spec::Win32', 

390 'IO::Special','Iterable','Iterator','Junction','Kernel','Label','List', 

391 'Lock','Lock::Async','long','longlong','Macro','Map','Match', 

392 'Metamodel::AttributeContainer','Metamodel::C3MRO','Metamodel::ClassHOW', 

393 'Metamodel::EnumHOW','Metamodel::Finalization','Metamodel::MethodContainer', 

394 'Metamodel::MROBasedMethodDispatch','Metamodel::MultipleInheritance', 

395 'Metamodel::Naming','Metamodel::Primitives','Metamodel::PrivateMethodContainer', 

396 'Metamodel::RoleContainer','Metamodel::Trusting','Method','Mix','MixHash', 

397 'Mixy','Mu','NFC','NFD','NFKC','NFKD','Nil','Num','num32','num64', 

398 'Numeric','NumStr','ObjAt','Order','Pair','Parameter','Perl','Pod::Block', 

399 'Pod::Block::Code','Pod::Block::Comment','Pod::Block::Declarator', 

400 'Pod::Block::Named','Pod::Block::Para','Pod::Block::Table','Pod::Heading', 

401 'Pod::Item','Pointer','Positional','PositionalBindFailover','Proc', 

402 'Proc::Async','Promise','Proxy','PseudoStash','QuantHash','Range','Rat', 

403 'Rational','RatStr','Real','Regex','Routine','Scalar','Scheduler', 

404 'Semaphore','Seq','Set','SetHash','Setty','Signature','size_t','Slip', 

405 'Stash','Str','StrDistance','Stringy','Sub','Submethod','Supplier', 

406 'Supplier::Preserving','Supply','Systemic','Tap','Telemetry', 

407 'Telemetry::Instrument::Thread','Telemetry::Instrument::Usage', 

408 'Telemetry::Period','Telemetry::Sampler','Thread','ThreadPoolScheduler', 

409 'UInt','uint16','uint32','uint64','uint8','Uni','utf8','Variable', 

410 'Version','VM','Whatever','WhateverCode','WrapHandle' 

411 ) 

412 

413 PERL6_OPERATORS = ( 

414 'X', 'Z', 'after', 'also', 'and', 'andthen', 'before', 'cmp', 'div', 

415 'eq', 'eqv', 'extra', 'ff', 'fff', 'ge', 'gt', 'le', 'leg', 'lt', 'm', 

416 'mm', 'mod', 'ne', 'or', 'orelse', 'rx', 's', 'tr', 'x', 'xor', 'xx', 

417 '++', '--', '**', '!', '+', '-', '~', '?', '|', '||', '+^', '~^', '?^', 

418 '^', '*', '/', '%', '%%', '+&', '+<', '+>', '~&', '~<', '~>', '?&', 

419 'gcd', 'lcm', '+', '-', '+|', '+^', '~|', '~^', '?|', '?^', 

420 '~', '&', '^', 'but', 'does', '<=>', '..', '..^', '^..', '^..^', 

421 '!=', '==', '<', '<=', '>', '>=', '~~', '===', '!eqv', 

422 '&&', '||', '^^', '//', 'min', 'max', '??', '!!', 'ff', 'fff', 'so', 

423 'not', '<==', '==>', '<<==', '==>>','unicmp', 

424 ) 

425 

426 # Perl 6 has a *lot* of possible bracketing characters 

427 # this list was lifted from STD.pm6 (https://github.com/perl6/std) 

428 PERL6_BRACKETS = { 

429 '\u0028': '\u0029', '\u003c': '\u003e', '\u005b': '\u005d', 

430 '\u007b': '\u007d', '\u00ab': '\u00bb', '\u0f3a': '\u0f3b', 

431 '\u0f3c': '\u0f3d', '\u169b': '\u169c', '\u2018': '\u2019', 

432 '\u201a': '\u2019', '\u201b': '\u2019', '\u201c': '\u201d', 

433 '\u201e': '\u201d', '\u201f': '\u201d', '\u2039': '\u203a', 

434 '\u2045': '\u2046', '\u207d': '\u207e', '\u208d': '\u208e', 

435 '\u2208': '\u220b', '\u2209': '\u220c', '\u220a': '\u220d', 

436 '\u2215': '\u29f5', '\u223c': '\u223d', '\u2243': '\u22cd', 

437 '\u2252': '\u2253', '\u2254': '\u2255', '\u2264': '\u2265', 

438 '\u2266': '\u2267', '\u2268': '\u2269', '\u226a': '\u226b', 

439 '\u226e': '\u226f', '\u2270': '\u2271', '\u2272': '\u2273', 

440 '\u2274': '\u2275', '\u2276': '\u2277', '\u2278': '\u2279', 

441 '\u227a': '\u227b', '\u227c': '\u227d', '\u227e': '\u227f', 

442 '\u2280': '\u2281', '\u2282': '\u2283', '\u2284': '\u2285', 

443 '\u2286': '\u2287', '\u2288': '\u2289', '\u228a': '\u228b', 

444 '\u228f': '\u2290', '\u2291': '\u2292', '\u2298': '\u29b8', 

445 '\u22a2': '\u22a3', '\u22a6': '\u2ade', '\u22a8': '\u2ae4', 

446 '\u22a9': '\u2ae3', '\u22ab': '\u2ae5', '\u22b0': '\u22b1', 

447 '\u22b2': '\u22b3', '\u22b4': '\u22b5', '\u22b6': '\u22b7', 

448 '\u22c9': '\u22ca', '\u22cb': '\u22cc', '\u22d0': '\u22d1', 

449 '\u22d6': '\u22d7', '\u22d8': '\u22d9', '\u22da': '\u22db', 

450 '\u22dc': '\u22dd', '\u22de': '\u22df', '\u22e0': '\u22e1', 

451 '\u22e2': '\u22e3', '\u22e4': '\u22e5', '\u22e6': '\u22e7', 

452 '\u22e8': '\u22e9', '\u22ea': '\u22eb', '\u22ec': '\u22ed', 

453 '\u22f0': '\u22f1', '\u22f2': '\u22fa', '\u22f3': '\u22fb', 

454 '\u22f4': '\u22fc', '\u22f6': '\u22fd', '\u22f7': '\u22fe', 

455 '\u2308': '\u2309', '\u230a': '\u230b', '\u2329': '\u232a', 

456 '\u23b4': '\u23b5', '\u2768': '\u2769', '\u276a': '\u276b', 

457 '\u276c': '\u276d', '\u276e': '\u276f', '\u2770': '\u2771', 

458 '\u2772': '\u2773', '\u2774': '\u2775', '\u27c3': '\u27c4', 

459 '\u27c5': '\u27c6', '\u27d5': '\u27d6', '\u27dd': '\u27de', 

460 '\u27e2': '\u27e3', '\u27e4': '\u27e5', '\u27e6': '\u27e7', 

461 '\u27e8': '\u27e9', '\u27ea': '\u27eb', '\u2983': '\u2984', 

462 '\u2985': '\u2986', '\u2987': '\u2988', '\u2989': '\u298a', 

463 '\u298b': '\u298c', '\u298d': '\u298e', '\u298f': '\u2990', 

464 '\u2991': '\u2992', '\u2993': '\u2994', '\u2995': '\u2996', 

465 '\u2997': '\u2998', '\u29c0': '\u29c1', '\u29c4': '\u29c5', 

466 '\u29cf': '\u29d0', '\u29d1': '\u29d2', '\u29d4': '\u29d5', 

467 '\u29d8': '\u29d9', '\u29da': '\u29db', '\u29f8': '\u29f9', 

468 '\u29fc': '\u29fd', '\u2a2b': '\u2a2c', '\u2a2d': '\u2a2e', 

469 '\u2a34': '\u2a35', '\u2a3c': '\u2a3d', '\u2a64': '\u2a65', 

470 '\u2a79': '\u2a7a', '\u2a7d': '\u2a7e', '\u2a7f': '\u2a80', 

471 '\u2a81': '\u2a82', '\u2a83': '\u2a84', '\u2a8b': '\u2a8c', 

472 '\u2a91': '\u2a92', '\u2a93': '\u2a94', '\u2a95': '\u2a96', 

473 '\u2a97': '\u2a98', '\u2a99': '\u2a9a', '\u2a9b': '\u2a9c', 

474 '\u2aa1': '\u2aa2', '\u2aa6': '\u2aa7', '\u2aa8': '\u2aa9', 

475 '\u2aaa': '\u2aab', '\u2aac': '\u2aad', '\u2aaf': '\u2ab0', 

476 '\u2ab3': '\u2ab4', '\u2abb': '\u2abc', '\u2abd': '\u2abe', 

477 '\u2abf': '\u2ac0', '\u2ac1': '\u2ac2', '\u2ac3': '\u2ac4', 

478 '\u2ac5': '\u2ac6', '\u2acd': '\u2ace', '\u2acf': '\u2ad0', 

479 '\u2ad1': '\u2ad2', '\u2ad3': '\u2ad4', '\u2ad5': '\u2ad6', 

480 '\u2aec': '\u2aed', '\u2af7': '\u2af8', '\u2af9': '\u2afa', 

481 '\u2e02': '\u2e03', '\u2e04': '\u2e05', '\u2e09': '\u2e0a', 

482 '\u2e0c': '\u2e0d', '\u2e1c': '\u2e1d', '\u2e20': '\u2e21', 

483 '\u3008': '\u3009', '\u300a': '\u300b', '\u300c': '\u300d', 

484 '\u300e': '\u300f', '\u3010': '\u3011', '\u3014': '\u3015', 

485 '\u3016': '\u3017', '\u3018': '\u3019', '\u301a': '\u301b', 

486 '\u301d': '\u301e', '\ufd3e': '\ufd3f', '\ufe17': '\ufe18', 

487 '\ufe35': '\ufe36', '\ufe37': '\ufe38', '\ufe39': '\ufe3a', 

488 '\ufe3b': '\ufe3c', '\ufe3d': '\ufe3e', '\ufe3f': '\ufe40', 

489 '\ufe41': '\ufe42', '\ufe43': '\ufe44', '\ufe47': '\ufe48', 

490 '\ufe59': '\ufe5a', '\ufe5b': '\ufe5c', '\ufe5d': '\ufe5e', 

491 '\uff08': '\uff09', '\uff1c': '\uff1e', '\uff3b': '\uff3d', 

492 '\uff5b': '\uff5d', '\uff5f': '\uff60', '\uff62': '\uff63', 

493 } 

494 

495 def _build_word_match(words, boundary_regex_fragment=None, prefix='', suffix=''): 

496 if boundary_regex_fragment is None: 

497 return r'\b(' + prefix + r'|'.join(re.escape(x) for x in words) + \ 

498 suffix + r')\b' 

499 else: 

500 return r'(?<!' + boundary_regex_fragment + r')' + prefix + r'(' + \ 

501 r'|'.join(re.escape(x) for x in words) + r')' + suffix + r'(?!' + \ 

502 boundary_regex_fragment + r')' 

503 

504 def brackets_callback(token_class): 

505 def callback(lexer, match, context): 

506 groups = match.groupdict() 

507 opening_chars = groups['delimiter'] 

508 n_chars = len(opening_chars) 

509 adverbs = groups.get('adverbs') 

510 

511 closer = Perl6Lexer.PERL6_BRACKETS.get(opening_chars[0]) 

512 text = context.text 

513 

514 if closer is None: # it's not a mirrored character, which means we 

515 # just need to look for the next occurrence 

516 

517 end_pos = text.find(opening_chars, match.start('delimiter') + n_chars) 

518 else: # we need to look for the corresponding closing character, 

519 # keep nesting in mind 

520 closing_chars = closer * n_chars 

521 nesting_level = 1 

522 

523 search_pos = match.start('delimiter') 

524 

525 while nesting_level > 0: 

526 next_open_pos = text.find(opening_chars, search_pos + n_chars) 

527 next_close_pos = text.find(closing_chars, search_pos + n_chars) 

528 

529 if next_close_pos == -1: 

530 next_close_pos = len(text) 

531 nesting_level = 0 

532 elif next_open_pos != -1 and next_open_pos < next_close_pos: 

533 nesting_level += 1 

534 search_pos = next_open_pos 

535 else: # next_close_pos < next_open_pos 

536 nesting_level -= 1 

537 search_pos = next_close_pos 

538 

539 end_pos = next_close_pos 

540 

541 if end_pos < 0: # if we didn't find a closer, just highlight the 

542 # rest of the text in this class 

543 end_pos = len(text) 

544 

545 if adverbs is not None and re.search(r':to\b', adverbs): 

546 heredoc_terminator = text[match.start('delimiter') + n_chars:end_pos] 

547 end_heredoc = re.search(r'^\s*' + re.escape(heredoc_terminator) + 

548 r'\s*$', text[end_pos:], re.MULTILINE) 

549 

550 if end_heredoc: 

551 end_pos += end_heredoc.end() 

552 else: 

553 end_pos = len(text) 

554 

555 yield match.start(), token_class, text[match.start():end_pos + n_chars] 

556 context.pos = end_pos + n_chars 

557 

558 return callback 

559 

560 def opening_brace_callback(lexer, match, context): 

561 stack = context.stack 

562 

563 yield match.start(), Text, context.text[match.start():match.end()] 

564 context.pos = match.end() 

565 

566 # if we encounter an opening brace and we're one level 

567 # below a token state, it means we need to increment 

568 # the nesting level for braces so we know later when