perl.js 8.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471
  1. /*
  2. Language: Perl
  3. Author: Peter Leonov <gojpeg@yandex.ru>
  4. Website: https://www.perl.org
  5. Category: common
  6. */
  7. /** @type LanguageFn */
  8. function perl(hljs) {
  9. const regex = hljs.regex;
  10. const KEYWORDS = [
  11. 'abs',
  12. 'accept',
  13. 'alarm',
  14. 'and',
  15. 'atan2',
  16. 'bind',
  17. 'binmode',
  18. 'bless',
  19. 'break',
  20. 'caller',
  21. 'chdir',
  22. 'chmod',
  23. 'chomp',
  24. 'chop',
  25. 'chown',
  26. 'chr',
  27. 'chroot',
  28. 'close',
  29. 'closedir',
  30. 'connect',
  31. 'continue',
  32. 'cos',
  33. 'crypt',
  34. 'dbmclose',
  35. 'dbmopen',
  36. 'defined',
  37. 'delete',
  38. 'die',
  39. 'do',
  40. 'dump',
  41. 'each',
  42. 'else',
  43. 'elsif',
  44. 'endgrent',
  45. 'endhostent',
  46. 'endnetent',
  47. 'endprotoent',
  48. 'endpwent',
  49. 'endservent',
  50. 'eof',
  51. 'eval',
  52. 'exec',
  53. 'exists',
  54. 'exit',
  55. 'exp',
  56. 'fcntl',
  57. 'fileno',
  58. 'flock',
  59. 'for',
  60. 'foreach',
  61. 'fork',
  62. 'format',
  63. 'formline',
  64. 'getc',
  65. 'getgrent',
  66. 'getgrgid',
  67. 'getgrnam',
  68. 'gethostbyaddr',
  69. 'gethostbyname',
  70. 'gethostent',
  71. 'getlogin',
  72. 'getnetbyaddr',
  73. 'getnetbyname',
  74. 'getnetent',
  75. 'getpeername',
  76. 'getpgrp',
  77. 'getpriority',
  78. 'getprotobyname',
  79. 'getprotobynumber',
  80. 'getprotoent',
  81. 'getpwent',
  82. 'getpwnam',
  83. 'getpwuid',
  84. 'getservbyname',
  85. 'getservbyport',
  86. 'getservent',
  87. 'getsockname',
  88. 'getsockopt',
  89. 'given',
  90. 'glob',
  91. 'gmtime',
  92. 'goto',
  93. 'grep',
  94. 'gt',
  95. 'hex',
  96. 'if',
  97. 'index',
  98. 'int',
  99. 'ioctl',
  100. 'join',
  101. 'keys',
  102. 'kill',
  103. 'last',
  104. 'lc',
  105. 'lcfirst',
  106. 'length',
  107. 'link',
  108. 'listen',
  109. 'local',
  110. 'localtime',
  111. 'log',
  112. 'lstat',
  113. 'lt',
  114. 'ma',
  115. 'map',
  116. 'mkdir',
  117. 'msgctl',
  118. 'msgget',
  119. 'msgrcv',
  120. 'msgsnd',
  121. 'my',
  122. 'ne',
  123. 'next',
  124. 'no',
  125. 'not',
  126. 'oct',
  127. 'open',
  128. 'opendir',
  129. 'or',
  130. 'ord',
  131. 'our',
  132. 'pack',
  133. 'package',
  134. 'pipe',
  135. 'pop',
  136. 'pos',
  137. 'print',
  138. 'printf',
  139. 'prototype',
  140. 'push',
  141. 'q|0',
  142. 'qq',
  143. 'quotemeta',
  144. 'qw',
  145. 'qx',
  146. 'rand',
  147. 'read',
  148. 'readdir',
  149. 'readline',
  150. 'readlink',
  151. 'readpipe',
  152. 'recv',
  153. 'redo',
  154. 'ref',
  155. 'rename',
  156. 'require',
  157. 'reset',
  158. 'return',
  159. 'reverse',
  160. 'rewinddir',
  161. 'rindex',
  162. 'rmdir',
  163. 'say',
  164. 'scalar',
  165. 'seek',
  166. 'seekdir',
  167. 'select',
  168. 'semctl',
  169. 'semget',
  170. 'semop',
  171. 'send',
  172. 'setgrent',
  173. 'sethostent',
  174. 'setnetent',
  175. 'setpgrp',
  176. 'setpriority',
  177. 'setprotoent',
  178. 'setpwent',
  179. 'setservent',
  180. 'setsockopt',
  181. 'shift',
  182. 'shmctl',
  183. 'shmget',
  184. 'shmread',
  185. 'shmwrite',
  186. 'shutdown',
  187. 'sin',
  188. 'sleep',
  189. 'socket',
  190. 'socketpair',
  191. 'sort',
  192. 'splice',
  193. 'split',
  194. 'sprintf',
  195. 'sqrt',
  196. 'srand',
  197. 'stat',
  198. 'state',
  199. 'study',
  200. 'sub',
  201. 'substr',
  202. 'symlink',
  203. 'syscall',
  204. 'sysopen',
  205. 'sysread',
  206. 'sysseek',
  207. 'system',
  208. 'syswrite',
  209. 'tell',
  210. 'telldir',
  211. 'tie',
  212. 'tied',
  213. 'time',
  214. 'times',
  215. 'tr',
  216. 'truncate',
  217. 'uc',
  218. 'ucfirst',
  219. 'umask',
  220. 'undef',
  221. 'unless',
  222. 'unlink',
  223. 'unpack',
  224. 'unshift',
  225. 'untie',
  226. 'until',
  227. 'use',
  228. 'utime',
  229. 'values',
  230. 'vec',
  231. 'wait',
  232. 'waitpid',
  233. 'wantarray',
  234. 'warn',
  235. 'when',
  236. 'while',
  237. 'write',
  238. 'x|0',
  239. 'xor',
  240. 'y|0'
  241. ];
  242. // https://perldoc.perl.org/perlre#Modifiers
  243. const REGEX_MODIFIERS = /[dualxmsipngr]{0,12}/; // aa and xx are valid, making max length 12
  244. const PERL_KEYWORDS = {
  245. $pattern: /[\w.]+/,
  246. keyword: KEYWORDS.join(" ")
  247. };
  248. const SUBST = {
  249. className: 'subst',
  250. begin: '[$@]\\{',
  251. end: '\\}',
  252. keywords: PERL_KEYWORDS
  253. };
  254. const METHOD = {
  255. begin: /->\{/,
  256. end: /\}/
  257. // contains defined later
  258. };
  259. const VAR = { variants: [
  260. { begin: /\$\d/ },
  261. { begin: regex.concat(
  262. /[$%@](\^\w\b|#\w+(::\w+)*|\{\w+\}|\w+(::\w*)*)/,
  263. // negative look-ahead tries to avoid matching patterns that are not
  264. // Perl at all like $ident$, @ident@, etc.
  265. `(?![A-Za-z])(?![@$%])`
  266. ) },
  267. {
  268. begin: /[$%@][^\s\w{]/,
  269. relevance: 0
  270. }
  271. ] };
  272. const STRING_CONTAINS = [
  273. hljs.BACKSLASH_ESCAPE,
  274. SUBST,
  275. VAR
  276. ];
  277. const REGEX_DELIMS = [
  278. /!/,
  279. /\//,
  280. /\|/,
  281. /\?/,
  282. /'/,
  283. /"/, // valid but infrequent and weird
  284. /#/ // valid but infrequent and weird
  285. ];
  286. /**
  287. * @param {string|RegExp} prefix
  288. * @param {string|RegExp} open
  289. * @param {string|RegExp} close
  290. */
  291. const PAIRED_DOUBLE_RE = (prefix, open, close = '\\1') => {
  292. const middle = (close === '\\1')
  293. ? close
  294. : regex.concat(close, open);
  295. return regex.concat(
  296. regex.concat("(?:", prefix, ")"),
  297. open,
  298. /(?:\\.|[^\\\/])*?/,
  299. middle,
  300. /(?:\\.|[^\\\/])*?/,
  301. close,
  302. REGEX_MODIFIERS
  303. );
  304. };
  305. /**
  306. * @param {string|RegExp} prefix
  307. * @param {string|RegExp} open
  308. * @param {string|RegExp} close
  309. */
  310. const PAIRED_RE = (prefix, open, close) => {
  311. return regex.concat(
  312. regex.concat("(?:", prefix, ")"),
  313. open,
  314. /(?:\\.|[^\\\/])*?/,
  315. close,
  316. REGEX_MODIFIERS
  317. );
  318. };
  319. const PERL_DEFAULT_CONTAINS = [
  320. VAR,
  321. hljs.HASH_COMMENT_MODE,
  322. hljs.COMMENT(
  323. /^=\w/,
  324. /=cut/,
  325. { endsWithParent: true }
  326. ),
  327. METHOD,
  328. {
  329. className: 'string',
  330. contains: STRING_CONTAINS,
  331. variants: [
  332. {
  333. begin: 'q[qwxr]?\\s*\\(',
  334. end: '\\)',
  335. relevance: 5
  336. },
  337. {
  338. begin: 'q[qwxr]?\\s*\\[',
  339. end: '\\]',
  340. relevance: 5
  341. },
  342. {
  343. begin: 'q[qwxr]?\\s*\\{',
  344. end: '\\}',
  345. relevance: 5
  346. },
  347. {
  348. begin: 'q[qwxr]?\\s*\\|',
  349. end: '\\|',
  350. relevance: 5
  351. },
  352. {
  353. begin: 'q[qwxr]?\\s*<',
  354. end: '>',
  355. relevance: 5
  356. },
  357. {
  358. begin: 'qw\\s+q',
  359. end: 'q',
  360. relevance: 5
  361. },
  362. {
  363. begin: '\'',
  364. end: '\'',
  365. contains: [ hljs.BACKSLASH_ESCAPE ]
  366. },
  367. {
  368. begin: '"',
  369. end: '"'
  370. },
  371. {
  372. begin: '`',
  373. end: '`',
  374. contains: [ hljs.BACKSLASH_ESCAPE ]
  375. },
  376. {
  377. begin: /\{\w+\}/,
  378. relevance: 0
  379. },
  380. {
  381. begin: '-?\\w+\\s*=>',
  382. relevance: 0
  383. }
  384. ]
  385. },
  386. {
  387. className: 'number',
  388. begin: '(\\b0[0-7_]+)|(\\b0x[0-9a-fA-F_]+)|(\\b[1-9][0-9_]*(\\.[0-9_]+)?)|[0_]\\b',
  389. relevance: 0
  390. },
  391. { // regexp container
  392. begin: '(\\/\\/|' + hljs.RE_STARTERS_RE + '|\\b(split|return|print|reverse|grep)\\b)\\s*',
  393. keywords: 'split return print reverse grep',
  394. relevance: 0,
  395. contains: [
  396. hljs.HASH_COMMENT_MODE,
  397. {
  398. className: 'regexp',
  399. variants: [
  400. // allow matching common delimiters
  401. { begin: PAIRED_DOUBLE_RE("s|tr|y", regex.either(...REGEX_DELIMS, { capture: true })) },
  402. // and then paired delmis
  403. { begin: PAIRED_DOUBLE_RE("s|tr|y", "\\(", "\\)") },
  404. { begin: PAIRED_DOUBLE_RE("s|tr|y", "\\[", "\\]") },
  405. { begin: PAIRED_DOUBLE_RE("s|tr|y", "\\{", "\\}") }
  406. ],
  407. relevance: 2
  408. },
  409. {
  410. className: 'regexp',
  411. variants: [
  412. {
  413. // could be a comment in many languages so do not count
  414. // as relevant
  415. begin: /(m|qr)\/\//,
  416. relevance: 0
  417. },
  418. // prefix is optional with /regex/
  419. { begin: PAIRED_RE("(?:m|qr)?", /\//, /\//) },
  420. // allow matching common delimiters
  421. { begin: PAIRED_RE("m|qr", regex.either(...REGEX_DELIMS, { capture: true }), /\1/) },
  422. // allow common paired delmins
  423. { begin: PAIRED_RE("m|qr", /\(/, /\)/) },
  424. { begin: PAIRED_RE("m|qr", /\[/, /\]/) },
  425. { begin: PAIRED_RE("m|qr", /\{/, /\}/) }
  426. ]
  427. }
  428. ]
  429. },
  430. {
  431. className: 'function',
  432. beginKeywords: 'sub',
  433. end: '(\\s*\\(.*?\\))?[;{]',
  434. excludeEnd: true,
  435. relevance: 5,
  436. contains: [ hljs.TITLE_MODE ]
  437. },
  438. {
  439. begin: '-\\w\\b',
  440. relevance: 0
  441. },
  442. {
  443. begin: "^__DATA__$",
  444. end: "^__END__$",
  445. subLanguage: 'mojolicious',
  446. contains: [
  447. {
  448. begin: "^@@.*",
  449. end: "$",
  450. className: "comment"
  451. }
  452. ]
  453. }
  454. ];
  455. SUBST.contains = PERL_DEFAULT_CONTAINS;
  456. METHOD.contains = PERL_DEFAULT_CONTAINS;
  457. return {
  458. name: 'Perl',
  459. aliases: [
  460. 'pl',
  461. 'pm'
  462. ],
  463. keywords: PERL_KEYWORDS,
  464. contains: PERL_DEFAULT_CONTAINS
  465. };
  466. }
  467. module.exports = perl;