ylvarw /
module
| Total Complexity | 138 |
| Total Lines | 936 |
| Duplicated Lines | 0 % |
| Changes | 0 | ||
Complex classes like Highlighter often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use Highlighter, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 41 | class Highlighter |
||
| 42 | { |
||
| 43 | /** |
||
| 44 | * @since 9.12.0.4 |
||
| 45 | */ |
||
| 46 | const SPAN_END_TAG = "</span>"; |
||
| 47 | |||
| 48 | /** @var bool */ |
||
| 49 | private $safeMode = true; |
||
| 50 | |||
| 51 | // @TODO In v10.x, this value should be static to match highlight.js behavior |
||
| 52 | /** @var array<string, mixed> */ |
||
| 53 | private $options; |
||
| 54 | |||
| 55 | /** @var string */ |
||
| 56 | private $modeBuffer = ""; |
||
| 57 | |||
| 58 | /** @var string */ |
||
| 59 | private $result = ""; |
||
| 60 | |||
| 61 | /** @var Mode|null */ |
||
| 62 | private $top = null; |
||
| 63 | |||
| 64 | /** @var Language|null */ |
||
| 65 | private $language = null; |
||
| 66 | |||
| 67 | /** @var int */ |
||
| 68 | private $relevance = 0; |
||
| 69 | |||
| 70 | /** @var bool */ |
||
| 71 | private $ignoreIllegals = false; |
||
| 72 | |||
| 73 | /** @var array<string, Mode> */ |
||
| 74 | private $continuations = array(); |
||
| 75 | |||
| 76 | /** @var RegExMatch */ |
||
| 77 | private $lastMatch; |
||
| 78 | |||
| 79 | /** @var string The current code we are highlighting */ |
||
| 80 | private $codeToHighlight; |
||
| 81 | |||
| 82 | /** @var string[] A list of all the bundled languages */ |
||
| 83 | private static $bundledLanguages = array(); |
||
| 84 | |||
| 85 | /** @var array<string, Language> A mapping of a language ID to a Language definition */ |
||
| 86 | private static $classMap = array(); |
||
| 87 | |||
| 88 | /** @var string[] A list of registered language IDs */ |
||
| 89 | private static $languages = array(); |
||
| 90 | |||
| 91 | /** @var array<string, string> A mapping from alias (key) to main language ID (value) */ |
||
| 92 | private static $aliases = array(); |
||
| 93 | |||
| 94 | /** |
||
| 95 | * @param bool $loadAllLanguages If true, will automatically register all languages distributed with this library. |
||
| 96 | * If false, user must explicitly register languages by calling `registerLanguage()`. |
||
| 97 | * |
||
| 98 | * @since 9.18.1.4 added `$loadAllLanguages` parameter |
||
| 99 | * @see Highlighter::registerLanguage() |
||
| 100 | */ |
||
| 101 | public function __construct($loadAllLanguages = true) |
||
| 102 | { |
||
| 103 | $this->lastMatch = new RegExMatch(array()); |
||
| 104 | $this->lastMatch->type = ""; |
||
|
Bug
introduced
by
|
|||
| 105 | $this->lastMatch->rule = null; |
||
| 106 | |||
| 107 | // @TODO In v10.x, remove the default value for the `languages` value to follow highlight.js behavior |
||
| 108 | $this->options = array( |
||
| 109 | 'classPrefix' => 'hljs-', |
||
| 110 | 'tabReplace' => null, |
||
| 111 | 'useBR' => false, |
||
| 112 | 'languages' => array( |
||
| 113 | "xml", "json", "javascript", "css", "php", "http", |
||
| 114 | ), |
||
| 115 | ); |
||
| 116 | |||
| 117 | if ($loadAllLanguages) { |
||
| 118 | self::registerAllLanguages(); |
||
| 119 | } |
||
| 120 | } |
||
| 121 | |||
| 122 | /** |
||
| 123 | * Return a list of all available languages bundled with this library. |
||
| 124 | * |
||
| 125 | * @since 9.18.1.4 |
||
| 126 | * |
||
| 127 | * @return string[] An array of language names |
||
| 128 | */ |
||
| 129 | public static function listBundledLanguages() |
||
| 130 | { |
||
| 131 | if (!empty(self::$bundledLanguages)) { |
||
| 132 | return self::$bundledLanguages; |
||
| 133 | } |
||
| 134 | |||
| 135 | // Languages that take precedence in the classMap array. (I don't know why...) |
||
| 136 | $bundledLanguages = array( |
||
| 137 | "xml" => true, |
||
| 138 | "django" => true, |
||
| 139 | "javascript" => true, |
||
| 140 | "matlab" => true, |
||
| 141 | "cpp" => true, |
||
| 142 | ); |
||
| 143 | |||
| 144 | $languagePath = __DIR__ . '/languages/'; |
||
| 145 | $d = @dir($languagePath); |
||
| 146 | |||
| 147 | if (!$d) { |
||
| 148 | throw new \RuntimeException('Could not read bundled language definition directory.'); |
||
| 149 | } |
||
| 150 | |||
| 151 | // @TODO In 10.x, rewrite this as a generator yielding results |
||
| 152 | while (($entry = $d->read()) !== false) { |
||
| 153 | if (substr($entry, -5) === ".json") { |
||
| 154 | $languageId = substr($entry, 0, -5); |
||
| 155 | $filePath = $languagePath . $entry; |
||
| 156 | |||
| 157 | if (is_readable($filePath)) { |
||
| 158 | $bundledLanguages[$languageId] = true; |
||
| 159 | } |
||
| 160 | } |
||
| 161 | } |
||
| 162 | |||
| 163 | $d->close(); |
||
| 164 | |||
| 165 | return self::$bundledLanguages = array_keys($bundledLanguages); |
||
| 166 | } |
||
| 167 | |||
| 168 | /** |
||
| 169 | * Return a list of all the registered languages. Using this list in |
||
| 170 | * setAutodetectLanguages will turn on auto-detection for all supported |
||
| 171 | * languages. |
||
| 172 | * |
||
| 173 | * @since 9.18.1.4 |
||
| 174 | * |
||
| 175 | * @param bool $includeAliases Specify whether language aliases should be |
||
| 176 | * included as well |
||
| 177 | * |
||
| 178 | * @return string[] An array of language names |
||
| 179 | */ |
||
| 180 | public static function listRegisteredLanguages($includeAliases = false) |
||
| 181 | { |
||
| 182 | if ($includeAliases === true) { |
||
| 183 | return array_merge(self::$languages, array_keys(self::$aliases)); |
||
| 184 | } |
||
| 185 | |||
| 186 | return self::$languages; |
||
| 187 | } |
||
| 188 | |||
| 189 | /** |
||
| 190 | * Register all 185+ languages that are bundled in this library. |
||
| 191 | * |
||
| 192 | * To register languages individually, use `registerLanguage`. |
||
| 193 | * |
||
| 194 | * @since 9.18.1.4 Method is now public |
||
| 195 | * @since 8.3.0.0 |
||
| 196 | * @see Highlighter::registerLanguage |
||
| 197 | * |
||
| 198 | * @return void |
||
| 199 | */ |
||
| 200 | public static function registerAllLanguages() |
||
| 201 | { |
||
| 202 | // Languages that take precedence in the classMap array. |
||
| 203 | $languagePath = __DIR__ . DIRECTORY_SEPARATOR . "languages" . DIRECTORY_SEPARATOR; |
||
| 204 | foreach (array("xml", "django", "javascript", "matlab", "cpp") as $languageId) { |
||
| 205 | $filePath = $languagePath . $languageId . ".json"; |
||
| 206 | if (is_readable($filePath)) { |
||
| 207 | self::registerLanguage($languageId, $filePath); |
||
| 208 | } |
||
| 209 | } |
||
| 210 | |||
| 211 | // @TODO In 10.x, call `listBundledLanguages()` instead when it's a generator |
||
| 212 | $d = @dir($languagePath); |
||
| 213 | if ($d) { |
||
| 214 | while (($entry = $d->read()) !== false) { |
||
| 215 | if (substr($entry, -5) === ".json") { |
||
| 216 | $languageId = substr($entry, 0, -5); |
||
| 217 | $filePath = $languagePath . $entry; |
||
| 218 | if (is_readable($filePath)) { |
||
| 219 | self::registerLanguage($languageId, $filePath); |
||
| 220 | } |
||
| 221 | } |
||
| 222 | } |
||
| 223 | $d->close(); |
||
| 224 | } |
||
| 225 | } |
||
| 226 | |||
| 227 | /** |
||
| 228 | * Register a language definition with the Highlighter's internal language |
||
| 229 | * storage. Languages are stored in a static variable, so they'll be available |
||
| 230 | * across all instances. You only need to register a language once. |
||
| 231 | * |
||
| 232 | * @param string $languageId The unique name of a language |
||
| 233 | * @param string $filePath The file path to the language definition |
||
| 234 | * @param bool $overwrite Overwrite language if it already exists |
||
| 235 | * |
||
| 236 | * @return Language The object containing the definition for a language's markup |
||
| 237 | */ |
||
| 238 | public static function registerLanguage($languageId, $filePath, $overwrite = false) |
||
| 239 | { |
||
| 240 | if (!isset(self::$classMap[$languageId]) || $overwrite) { |
||
| 241 | $lang = new Language($languageId, $filePath); |
||
| 242 | self::$classMap[$languageId] = $lang; |
||
| 243 | |||
| 244 | self::$languages[] = $languageId; |
||
| 245 | self::$languages = array_unique(self::$languages); |
||
| 246 | |||
| 247 | if ($lang->aliases) { |
||
| 248 | foreach ($lang->aliases as $alias) { |
||
| 249 | self::$aliases[$alias] = $languageId; |
||
| 250 | } |
||
| 251 | } |
||
| 252 | } |
||
| 253 | |||
| 254 | return self::$classMap[$languageId]; |
||
| 255 | } |
||
| 256 | |||
| 257 | /** |
||
| 258 | * Clear all registered languages. |
||
| 259 | * |
||
| 260 | * @since 9.18.1.4 |
||
| 261 | * |
||
| 262 | * @return void |
||
| 263 | */ |
||
| 264 | public static function clearAllLanguages() |
||
| 265 | { |
||
| 266 | self::$classMap = array(); |
||
| 267 | self::$languages = array(); |
||
| 268 | self::$aliases = array(); |
||
| 269 | } |
||
| 270 | |||
| 271 | /** |
||
| 272 | * @param RegEx|null $re |
||
| 273 | * @param string $lexeme |
||
| 274 | * |
||
| 275 | * @return bool |
||
| 276 | */ |
||
| 277 | private function testRe($re, $lexeme) |
||
| 278 | { |
||
| 279 | if (!$re) { |
||
| 280 | return false; |
||
| 281 | } |
||
| 282 | |||
| 283 | $lastIndex = $re->lastIndex; |
||
| 284 | $result = $re->exec($lexeme); |
||
| 285 | $re->lastIndex = $lastIndex; |
||
| 286 | |||
| 287 | return $result && $result->index === 0; |
||
| 288 | } |
||
| 289 | |||
| 290 | /** |
||
| 291 | * @param string $value |
||
| 292 | * |
||
| 293 | * @return RegEx |
||
| 294 | */ |
||
| 295 | private function escapeRe($value) |
||
| 298 | } |
||
| 299 | |||
| 300 | /** |
||
| 301 | * @param Mode $mode |
||
| 302 | * @param string $lexeme |
||
| 303 | * |
||
| 304 | * @return Mode|null |
||
| 305 | */ |
||
| 306 | private function endOfMode($mode, $lexeme) |
||
| 307 | { |
||
| 308 | if ($this->testRe($mode->endRe, $lexeme)) { |
||
| 309 | while ($mode->endsParent && $mode->parent) { |
||
| 310 | $mode = $mode->parent; |
||
| 311 | } |
||
| 312 | |||
| 313 | return $mode; |
||
| 314 | } |
||
| 315 | |||
| 316 | if ($mode->endsWithParent) { |
||
| 317 | return $this->endOfMode($mode->parent, $lexeme); |
||
| 318 | } |
||
| 319 | |||
| 320 | return null; |
||
| 321 | } |
||
| 322 | |||
| 323 | /** |
||
| 324 | * @param Mode $mode |
||
| 325 | * @param RegExMatch $match |
||
| 326 | * |
||
| 327 | * @return mixed|null |
||
| 328 | */ |
||
| 329 | private function keywordMatch($mode, $match) |
||
| 330 | { |
||
| 331 | $kwd = $this->language->case_insensitive ? mb_strtolower($match[0], "UTF-8") : $match[0]; |
||
| 332 | |||
| 333 | return isset($mode->keywords[$kwd]) ? $mode->keywords[$kwd] : null; |
||
| 334 | } |
||
| 335 | |||
| 336 | /** |
||
| 337 | * @param string $className |
||
| 338 | * @param string $insideSpan |
||
| 339 | * @param bool $leaveOpen |
||
| 340 | * @param bool $noPrefix |
||
| 341 | * |
||
| 342 | * @return string |
||
| 343 | */ |
||
| 344 | private function buildSpan($className, $insideSpan, $leaveOpen = false, $noPrefix = false) |
||
| 345 | { |
||
| 346 | if (!$leaveOpen && $insideSpan === '') { |
||
| 347 | return ''; |
||
| 348 | } |
||
| 349 | |||
| 350 | if (!$className) { |
||
| 351 | return $insideSpan; |
||
| 352 | } |
||
| 353 | |||
| 354 | $classPrefix = $noPrefix ? "" : $this->options['classPrefix']; |
||
| 355 | $openSpan = "<span class=\"" . $classPrefix; |
||
| 356 | $closeSpan = $leaveOpen ? "" : self::SPAN_END_TAG; |
||
| 357 | |||
| 358 | $openSpan .= $className . "\">"; |
||
| 359 | |||
| 360 | return $openSpan . $insideSpan . $closeSpan; |
||
| 361 | } |
||
| 362 | |||
| 363 | /** |
||
| 364 | * @param string $value |
||
| 365 | * |
||
| 366 | * @return string |
||
| 367 | */ |
||
| 368 | private function escape($value) |
||
| 369 | { |
||
| 370 | return htmlspecialchars($value, ENT_NOQUOTES); |
||
| 371 | } |
||
| 372 | |||
| 373 | /** |
||
| 374 | * @return string |
||
| 375 | */ |
||
| 376 | private function processKeywords() |
||
| 377 | { |
||
| 378 | if (!$this->top->keywords) { |
||
| 379 | return $this->escape($this->modeBuffer); |
||
| 380 | } |
||
| 381 | |||
| 382 | $result = ""; |
||
| 383 | $lastIndex = 0; |
||
| 384 | $this->top->lexemesRe->lastIndex = 0; |
||
| 385 | $match = $this->top->lexemesRe->exec($this->modeBuffer); |
||
| 386 | |||
| 387 | while ($match) { |
||
| 388 | $result .= $this->escape(substr($this->modeBuffer, $lastIndex, $match->index - $lastIndex)); |
||
| 389 | $keyword_match = $this->keywordMatch($this->top, $match); |
||
| 390 | |||
| 391 | if ($keyword_match) { |
||
| 392 | $this->relevance += $keyword_match[1]; |
||
| 393 | $result .= $this->buildSpan($keyword_match[0], $this->escape($match[0])); |
||
| 394 | } else { |
||
| 395 | $result .= $this->escape($match[0]); |
||
| 396 | } |
||
| 397 | |||
| 398 | $lastIndex = $this->top->lexemesRe->lastIndex; |
||
| 399 | $match = $this->top->lexemesRe->exec($this->modeBuffer); |
||
| 400 | } |
||
| 401 | |||
| 402 | return $result . $this->escape(substr($this->modeBuffer, $lastIndex)); |
||
| 403 | } |
||
| 404 | |||
| 405 | /** |
||
| 406 | * @return string |
||
| 407 | */ |
||
| 408 | private function processSubLanguage() |
||
| 409 | { |
||
| 410 | try { |
||
| 411 | $hl = new Highlighter(); |
||
| 412 | |||
| 413 | // @TODO in v10.x, this should no longer be necessary once `$options` is made static |
||
| 414 | $hl->setAutodetectLanguages($this->options['languages']); |
||
| 415 | $hl->setClassPrefix($this->options['classPrefix']); |
||
| 416 | $hl->setTabReplace($this->options['tabReplace']); |
||
| 417 | |||
| 418 | if (!$this->safeMode) { |
||
| 419 | $hl->disableSafeMode(); |
||
| 420 | } |
||
| 421 | |||
| 422 | $explicit = is_string($this->top->subLanguage); |
||
| 423 | if ($explicit && !in_array($this->top->subLanguage, self::$languages)) { |
||
| 424 | return $this->escape($this->modeBuffer); |
||
| 425 | } |
||
| 426 | |||
| 427 | if ($explicit) { |
||
| 428 | $res = $hl->highlight( |
||
| 429 | $this->top->subLanguage, |
||
| 430 | $this->modeBuffer, |
||
| 431 | true, |
||
| 432 | isset($this->continuations[$this->top->subLanguage]) ? $this->continuations[$this->top->subLanguage] : null |
||
| 433 | ); |
||
| 434 | } else { |
||
| 435 | $res = $hl->highlightAuto( |
||
| 436 | $this->modeBuffer, |
||
| 437 | count($this->top->subLanguage) ? $this->top->subLanguage : null |
||
| 438 | ); |
||
| 439 | } |
||
| 440 | |||
| 441 | // Counting embedded language score towards the host language may be disabled |
||
| 442 | // with zeroing the containing mode relevance. Use case in point is Markdown that |
||
| 443 | // allows XML everywhere and makes every XML snippet to have a much larger Markdown |
||
| 444 | // score. |
||
| 445 | if ($this->top->relevance > 0) { |
||
| 446 | $this->relevance += $res->relevance; |
||
| 447 | } |
||
| 448 | if ($explicit) { |
||
| 449 | $this->continuations[$this->top->subLanguage] = $res->top; |
||
| 450 | } |
||
| 451 | |||
| 452 | return $this->buildSpan($res->language, $res->value, false, true); |
||
| 453 | } catch (\Exception $e) { |
||
| 454 | return $this->escape($this->modeBuffer); |
||
| 455 | } |
||
| 456 | } |
||
| 457 | |||
| 458 | /** |
||
| 459 | * @return void |
||
| 460 | */ |
||
| 461 | private function processBuffer() |
||
| 462 | { |
||
| 463 | if (is_object($this->top) && $this->top->subLanguage) { |
||
| 464 | $this->result .= $this->processSubLanguage(); |
||
| 465 | } else { |
||
| 466 | $this->result .= $this->processKeywords(); |
||
| 467 | } |
||
| 468 | |||
| 469 | $this->modeBuffer = ''; |
||
| 470 | } |
||
| 471 | |||
| 472 | /** |
||
| 473 | * @param Mode $mode |
||
| 474 | * |
||
| 475 | * @return void |
||
| 476 | */ |
||
| 477 | private function startNewMode($mode) |
||
| 478 | { |
||
| 479 | $this->result .= $mode->className ? $this->buildSpan($mode->className, "", true) : ""; |
||
| 480 | |||
| 481 | $t = clone $mode; |
||
| 482 | $t->parent = $this->top; |
||
| 483 | $this->top = $t; |
||
| 484 | } |
||
| 485 | |||
| 486 | /** |
||
| 487 | * @param RegExMatch $match |
||
| 488 | * |
||
| 489 | * @return int |
||
| 490 | */ |
||
| 491 | private function doBeginMatch($match) |
||
| 492 | { |
||
| 493 | $lexeme = $match[0]; |
||
| 494 | $newMode = $match->rule; |
||
| 495 | |||
| 496 | if ($newMode && $newMode->endSameAsBegin) { |
||
| 497 | $newMode->endRe = $this->escapeRe($lexeme); |
||
| 498 | } |
||
| 499 | |||
| 500 | if ($newMode->skip) { |
||
| 501 | $this->modeBuffer .= $lexeme; |
||
| 502 | } else { |
||
| 503 | if ($newMode->excludeBegin) { |
||
| 504 | $this->modeBuffer .= $lexeme; |
||
| 505 | } |
||
| 506 | $this->processBuffer(); |
||
| 507 | if (!$newMode->returnBegin && !$newMode->excludeBegin) { |
||
| 508 | $this->modeBuffer = $lexeme; |
||
| 509 | } |
||
| 510 | } |
||
| 511 | $this->startNewMode($newMode); |
||
| 512 | |||
| 513 | return $newMode->returnBegin ? 0 : strlen($lexeme); |
||
| 514 | } |
||
| 515 | |||
| 516 | /** |
||
| 517 | * @param RegExMatch $match |
||
| 518 | * |
||
| 519 | * @return int|null |
||
| 520 | */ |
||
| 521 | private function doEndMatch($match) |
||
| 522 | { |
||
| 523 | $lexeme = $match[0]; |
||
| 524 | $matchPlusRemainder = substr($this->codeToHighlight, $match->index); |
||
| 525 | $endMode = $this->endOfMode($this->top, $matchPlusRemainder); |
||
| 526 | |||
| 527 | if (!$endMode) { |
||
| 528 | return null; |
||
| 529 | } |
||
| 530 | |||
| 531 | $origin = $this->top; |
||
| 532 | if ($origin->skip) { |
||
| 533 | $this->modeBuffer .= $lexeme; |
||
| 534 | } else { |
||
| 535 | if (!($origin->returnEnd || $origin->excludeEnd)) { |
||
| 536 | $this->modeBuffer .= $lexeme; |
||
| 537 | } |
||
| 538 | $this->processBuffer(); |
||
| 539 | if ($origin->excludeEnd) { |
||
| 540 | $this->modeBuffer = $lexeme; |
||
| 541 | } |
||
| 542 | } |
||
| 543 | |||
| 544 | do { |
||
| 545 | if ($this->top->className) { |
||
| 546 | $this->result .= self::SPAN_END_TAG; |
||
| 547 | } |
||
| 548 | if (!$this->top->skip && !$this->top->subLanguage) { |
||
| 549 | $this->relevance += $this->top->relevance; |
||
| 550 | } |
||
| 551 | $this->top = $this->top->parent; |
||
| 552 | } while ($this->top !== $endMode->parent); |
||
| 553 | |||
| 554 | if ($endMode->starts) { |
||
| 555 | if ($endMode->endSameAsBegin) { |
||
| 556 | $endMode->starts->endRe = $endMode->endRe; |
||
| 557 | } |
||
| 558 | |||
| 559 | $this->startNewMode($endMode->starts); |
||
| 560 | } |
||
| 561 | |||
| 562 | return $origin->returnEnd ? 0 : strlen($lexeme); |
||
| 563 | } |
||
| 564 | |||
| 565 | /** |
||
| 566 | * @param string $textBeforeMatch |
||
| 567 | * @param RegExMatch|null $match |
||
| 568 | * |
||
| 569 | * @return int |
||
| 570 | */ |
||
| 571 | private function processLexeme($textBeforeMatch, $match = null) |
||
| 572 | { |
||
| 573 | $lexeme = $match ? $match[0] : null; |
||
| 574 | |||
| 575 | // add non-matched text to the current mode buffer |
||
| 576 | $this->modeBuffer .= $textBeforeMatch; |
||
| 577 | |||
| 578 | if ($lexeme === null) { |
||
| 579 | $this->processBuffer(); |
||
| 580 | |||
| 581 | return 0; |
||
| 582 | } |
||
| 583 | |||
| 584 | // we've found a 0 width match and we're stuck, so we need to advance |
||
| 585 | // this happens when we have badly behaved rules that have optional matchers to the degree that |
||
| 586 | // sometimes they can end up matching nothing at all |
||
| 587 | // Ref: https://github.com/highlightjs/highlight.js/issues/2140 |
||
| 588 | if ($this->lastMatch->type === "begin" && $match->type === "end" && $this->lastMatch->index === $match->index && $lexeme === "") { |
||
| 589 | // spit the "skipped" character that our regex choked on back into the output sequence |
||
| 590 | $this->modeBuffer .= substr($this->codeToHighlight, $match->index, 1); |
||
| 591 | |||
| 592 | return 1; |
||
| 593 | } |
||
| 594 | $this->lastMatch = $match; |
||
| 595 | |||
| 596 | if ($match->type === "begin") { |
||
| 597 | return $this->doBeginMatch($match); |
||
| 598 | } elseif ($match->type === "illegal" && !$this->ignoreIllegals) { |
||
| 599 | // illegal match, we do not continue processing |
||
| 600 | $_modeRaw = isset($this->top->className) ? $this->top->className : "<unnamed>"; |
||
| 601 | |||
| 602 | throw new \UnexpectedValueException("Illegal lexeme \"$lexeme\" for mode \"$_modeRaw\""); |
||
| 603 | } elseif ($match->type === "end") { |
||
| 604 | $processed = $this->doEndMatch($match); |
||
| 605 | |||
| 606 | if ($processed !== null) { |
||
| 607 | return $processed; |
||
| 608 | } |
||
| 609 | } |
||
| 610 | |||
| 611 | // Why might be find ourselves here? Only one occasion now. An end match that was |
||
| 612 | // triggered but could not be completed. When might this happen? When an `endSameasBegin` |
||
| 613 | // rule sets the end rule to a specific match. Since the overall mode termination rule that's |
||
| 614 | // being used to scan the text isn't recompiled that means that any match that LOOKS like |
||
| 615 | // the end (but is not, because it is not an exact match to the beginning) will |
||
| 616 | // end up here. A definite end match, but when `doEndMatch` tries to "reapply" |
||
| 617 | // the end rule and fails to match, we wind up here, and just silently ignore the end. |
||
| 618 | // |
||
| 619 | // This causes no real harm other than stopping a few times too many. |
||
| 620 | |||
| 621 | $this->modeBuffer .= $lexeme; |
||
| 622 | |||
| 623 | return strlen($lexeme); |
||
| 624 | } |
||
| 625 | |||
| 626 | /** |
||
| 627 | * Replace tabs for something more usable. |
||
| 628 | * |
||
| 629 | * @param string $code |
||
| 630 | * |
||
| 631 | * @return string |
||
| 632 | */ |
||
| 633 | private function replaceTabs($code) |
||
| 634 | { |
||
| 635 | if ($this->options['tabReplace'] !== null) { |
||
| 636 | return str_replace("\t", $this->options['tabReplace'], $code); |
||
| 637 | } |
||
| 638 | |||
| 639 | return $code; |
||
| 640 | } |
||
| 641 | |||
| 642 | /** |
||
| 643 | * Set the languages that will used for auto-detection. When using auto- |
||
| 644 | * detection the code to highlight will be probed for every language in this |
||
| 645 | * set. Limiting this set to only the languages you want to use will greatly |
||
| 646 | * improve highlighting speed. |
||
| 647 | * |
||
| 648 | * @param string[] $set An array of language games to use for autodetection. |
||
| 649 | * This defaults to a typical set Web development |
||
| 650 | * languages. |
||
| 651 | * |
||
| 652 | * @return void |
||
| 653 | */ |
||
| 654 | public function setAutodetectLanguages(array $set) |
||
| 655 | { |
||
| 656 | $this->options['languages'] = array_unique($set); |
||
| 657 | } |
||
| 658 | |||
| 659 | /** |
||
| 660 | * Get the tab replacement string. |
||
| 661 | * |
||
| 662 | * @return string The tab replacement string |
||
| 663 | */ |
||
| 664 | public function getTabReplace() |
||
| 665 | { |
||
| 666 | return $this->options['tabReplace']; |
||
| 667 | } |
||
| 668 | |||
| 669 | /** |
||
| 670 | * Set the tab replacement string. This defaults to NULL: no tabs |
||
| 671 | * will be replaced. |
||
| 672 | * |
||
| 673 | * @param string $tabReplace The tab replacement string |
||
| 674 | * |
||
| 675 | * @return void |
||
| 676 | */ |
||
| 677 | public function setTabReplace($tabReplace) |
||
| 678 | { |
||
| 679 | $this->options['tabReplace'] = $tabReplace; |
||
| 680 | } |
||
| 681 | |||
| 682 | /** |
||
| 683 | * Get the class prefix string. |
||
| 684 | * |
||
| 685 | * @return string The class prefix string |
||
| 686 | */ |
||
| 687 | public function getClassPrefix() |
||
| 688 | { |
||
| 689 | return $this->options['classPrefix']; |
||
| 690 | } |
||
| 691 | |||
| 692 | /** |
||
| 693 | * Set the class prefix string. |
||
| 694 | * |
||
| 695 | * @param string $classPrefix The class prefix string |
||
| 696 | * |
||
| 697 | * @return void |
||
| 698 | */ |
||
| 699 | public function setClassPrefix($classPrefix) |
||
| 700 | { |
||
| 701 | $this->options['classPrefix'] = $classPrefix; |
||
| 702 | } |
||
| 703 | |||
| 704 | /** |
||
| 705 | * @since 9.17.1.0 |
||
| 706 | * |
||
| 707 | * @return void |
||
| 708 | */ |
||
| 709 | public function enableSafeMode() |
||
| 710 | { |
||
| 711 | $this->safeMode = true; |
||
| 712 | } |
||
| 713 | |||
| 714 | /** |
||
| 715 | * @since 9.17.1.0 |
||
| 716 | * |
||
| 717 | * @return void |
||
| 718 | */ |
||
| 719 | public function disableSafeMode() |
||
| 722 | } |
||
| 723 | |||
| 724 | /** |
||
| 725 | * @param string $name |
||
| 726 | * |
||
| 727 | * @return Language|null |
||
| 728 | */ |
||
| 729 | private function getLanguage($name) |
||
| 730 | { |
||
| 731 | if (isset(self::$classMap[$name])) { |
||
| 732 | return self::$classMap[$name]; |
||
| 733 | } elseif (isset(self::$aliases[$name]) && isset(self::$classMap[self::$aliases[$name]])) { |
||
| 734 | return self::$classMap[self::$aliases[$name]]; |
||
| 735 | } |
||
| 736 | |||
| 737 | return null; |
||
| 738 | } |
||
| 739 | |||
| 740 | /** |
||
| 741 | * Determine whether or not a language definition supports auto detection. |
||
| 742 | * |
||
| 743 | * @param string $name Language name |
||
| 744 | * |
||
| 745 | * @return bool |
||
| 746 | */ |
||
| 747 | private function autoDetection($name) |
||
| 752 | } |
||
| 753 | |||
| 754 | /** |
||
| 755 | * Core highlighting function. Accepts a language name, or an alias, and a |
||
| 756 | * string with the code to highlight. Returns an object with the following |
||
| 757 | * properties: |
||
| 758 | * - relevance (int) |
||
| 759 | * - value (an HTML string with highlighting markup). |
||
| 760 | * |
||
| 761 | * @todo In v10.x, change the return type from \stdClass to HighlightResult |
||
| 762 | * |
||
| 763 | * @param string $languageName |
||
| 764 | * @param string $code |
||
| 765 | * @param bool $ignoreIllegals |
||
| 766 | * @param Mode|null $continuation |
||
| 767 | * |
||
| 768 | * @throws \DomainException if the requested language was not in this |
||
| 769 | * Highlighter's language set |
||
| 770 | * @throws \Exception if an invalid regex was given in a language file |
||
| 771 | * |
||
| 772 | * @return HighlightResult|\stdClass |
||
| 773 | */ |
||
| 774 | public function highlight($languageName, $code, $ignoreIllegals = true, $continuation = null) |
||
| 775 | { |
||
| 776 | $this->codeToHighlight = $code; |
||
| 777 | $this->language = $this->getLanguage($languageName); |
||
| 778 | |||
| 779 | if ($this->language === null) { |
||
| 780 | throw new \DomainException("Unknown language: \"$languageName\""); |
||
| 781 | } |
||
| 782 | |||
| 783 | $this->language->compile($this->safeMode); |
||
| 784 | $this->top = $continuation ? $continuation : $this->language; |
||
| 785 | $this->continuations = array(); |
||
| 786 | $this->result = ""; |
||
| 787 | |||
| 788 | for ($current = $this->top; $current !== $this->language; $current = $current->parent) { |
||
| 789 | if ($current->className) { |
||
| 790 | $this->result = $this->buildSpan($current->className, '', true) . $this->result; |
||
| 791 | } |
||
| 792 | } |
||
| 793 | |||
| 794 | $this->modeBuffer = ""; |
||
| 795 | $this->relevance = 0; |
||
| 796 | $this->ignoreIllegals = $ignoreIllegals; |
||
| 797 | |||
| 798 | /** @var HighlightResult $res */ |
||
| 799 | $res = new \stdClass(); |
||
| 800 | $res->relevance = 0; |
||
| 801 | $res->value = ""; |
||
| 802 | $res->language = ""; |
||
| 803 | $res->top = null; |
||
| 804 | $res->errorRaised = null; |
||
| 805 | |||
| 806 | try { |
||
| 807 | $match = null; |
||
| 808 | $count = 0; |
||
| 809 | $index = 0; |
||
| 810 | |||
| 811 | while ($this->top) { |
||
| 812 | $this->top->terminators->lastIndex = $index; |
||
| 813 | $match = $this->top->terminators->exec($this->codeToHighlight); |
||
| 814 | |||
| 815 | if (!$match) { |
||
| 816 | break; |
||
| 817 | } |
||
| 818 | |||
| 819 | $count = $this->processLexeme(substr($this->codeToHighlight, $index, $match->index - $index), $match); |
||
| 820 | $index = $match->index + $count; |
||
| 821 | } |
||
| 822 | |||
| 823 | $this->processLexeme(substr($this->codeToHighlight, $index)); |
||
| 824 | |||
| 825 | for ($current = $this->top; isset($current->parent); $current = $current->parent) { |
||
| 826 | if ($current->className) { |
||
| 827 | $this->result .= self::SPAN_END_TAG; |
||
| 828 | } |
||
| 829 | } |
||
| 830 | |||
| 831 | $res->relevance = $this->relevance; |
||
| 832 | $res->value = $this->replaceTabs($this->result); |
||
| 833 | $res->illegal = false; |
||
| 834 | $res->language = $this->language->name; |
||
| 835 | $res->top = $this->top; |
||
| 836 | |||
| 837 | return $res; |
||
| 838 | } catch (\Exception $e) { |
||
| 839 | if (strpos($e->getMessage(), "Illegal") !== false) { |
||
| 840 | $res->illegal = true; |
||
| 841 | $res->relevance = 0; |
||
| 842 | $res->value = $this->escape($this->codeToHighlight); |
||
| 843 | |||
| 844 | return $res; |
||
| 845 | } elseif ($this->safeMode) { |
||
| 846 | $res->relevance = 0; |
||
| 847 | $res->value = $this->escape($this->codeToHighlight); |
||
| 848 | $res->language = $languageName; |
||
| 849 | $res->top = $this->top; |
||
| 850 | $res->errorRaised = $e; |
||
| 851 | |||
| 852 | return $res; |
||
| 853 | } |
||
| 854 | |||
| 855 | throw $e; |
||
| 856 | } |
||
| 857 | } |
||
| 858 | |||
| 859 | /** |
||
| 860 | * Highlight the given code by highlighting the given code with each |
||
| 861 | * registered language and then finding the match with highest accuracy. |
||
| 862 | * |
||
| 863 | * @param string $code |
||
| 864 | * @param string[]|null $languageSubset When set to null, this method will attempt to highlight $text with each |
||
| 865 | * language. Set this to an array of languages of your choice to limit the |
||
| 866 | * amount of languages to try. |
||
| 867 | * |
||
| 868 | * @throws \Exception if an invalid regex was given in a language file |
||
| 869 | * @throws \DomainException if the attempted language to check does not exist |
||
| 870 | * |
||
| 871 | * @return HighlightResult|\stdClass |
||
| 872 | */ |
||
| 873 | public function highlightAuto($code, $languageSubset = null) |
||
| 874 | { |
||
| 875 | /** @var HighlightResult $result */ |
||
| 876 | $result = new \stdClass(); |
||
| 877 | $result->relevance = 0; |
||
| 878 | $result->value = $this->escape($code); |
||
| 879 | $result->language = ""; |
||
| 880 | $secondBest = clone $result; |
||
| 881 | |||
| 882 | if ($languageSubset === null) { |
||
| 883 | $optionsLanguages = $this->options['languages']; |
||
| 884 | |||
| 885 | if (is_array($optionsLanguages) && count($optionsLanguages) > 0) { |
||
| 886 | $languageSubset = $optionsLanguages; |
||
| 887 | } else { |
||
| 888 | $languageSubset = self::$languages; |
||
| 889 | } |
||
| 890 | } |
||
| 891 | |||
| 892 | foreach ($languageSubset as $name) { |
||
| 893 | if ($this->getLanguage($name) === null || !$this->autoDetection($name)) { |
||
| 894 | continue; |
||
| 895 | } |
||
| 896 | |||
| 897 | $current = $this->highlight($name, $code, false); |
||
| 898 | |||
| 899 | if ($current->relevance > $secondBest->relevance) { |
||
| 900 | $secondBest = $current; |
||
| 901 | } |
||
| 902 | |||
| 903 | if ($current->relevance > $result->relevance) { |
||
| 904 | $secondBest = $result; |
||
| 905 | $result = $current; |
||
| 906 | } |
||
| 907 | } |
||
| 908 | |||
| 909 | if ($secondBest->language) { |
||
| 910 | $result->secondBest = $secondBest; |
||
| 911 | } |
||
| 912 | |||
| 913 | return $result; |
||
| 914 | } |
||
| 915 | |||
| 916 | /** |
||
| 917 | * Return a list of all supported languages. Using this list in |
||
| 918 | * setAutodetectLanguages will turn on autodetection for all supported |
||
| 919 | * languages. |
||
| 920 | * |
||
| 921 | * @deprecated use `Highlighter::listRegisteredLanguages()` or `Highlighter::listBundledLanguages()` instead |
||
| 922 | * |
||
| 923 | * @param bool $include_aliases specify whether language aliases |
||
| 924 | * should be included as well |
||
| 925 | * |
||
| 926 | * @since 9.18.1.4 Deprecated in favor of `Highlighter::listRegisteredLanguages()` |
||
| 927 | * and `Highlighter::listBundledLanguages()`. |
||
| 928 | * @since 9.12.0.3 The `$include_aliases` parameter was added |
||
| 929 | * @since 8.3.0.0 |
||
| 930 | * |
||
| 931 | * @return string[] An array of language names |
||
| 932 | */ |
||
| 933 | public function listLanguages($include_aliases = false) |
||
| 934 | { |
||
| 935 | @trigger_error('This method is deprecated in favor `Highlighter::listRegisteredLanguages()` or `Highlighter::listBundledLanguages()`. This function will be removed in highlight.php 10.', E_USER_DEPRECATED); |
||
| 936 | |||
| 937 | if (empty(self::$languages)) { |
||
| 938 | trigger_error('No languages are registered, returning all bundled languages instead. You probably did not want this.', E_USER_WARNING); |
||
| 939 | |||
| 940 | return self::listBundledLanguages(); |
||
| 941 | } |
||
| 942 | |||
| 943 | if ($include_aliases === true) { |
||
| 944 | return array_merge(self::$languages, array_keys(self::$aliases)); |
||
| 945 | } |
||
| 946 | |||
| 947 | return self::$languages; |
||
| 948 | } |
||
| 949 | |||
| 950 | /** |
||
| 951 | * Returns list of all available aliases for given language name. |
||
| 952 | * |
||
| 953 | * @param string $name name or alias of language to look-up |
||
| 954 | * |
||
| 955 | * @throws \DomainException if the requested language was not in this |
||
| 956 | * Highlighter's language set |
||
| 957 | * |
||
| 958 | * @since 9.12.0.3 |
||
| 959 | * |
||
| 960 | * @return string[] An array of all aliases associated with the requested |
||
| 961 | * language name language. Passed-in name is included as |
||
| 962 | * well. |
||
| 963 | */ |
||
| 964 | public function getAliasesForLanguage($name) |
||
| 977 | } |
||
| 978 | } |
||
| 979 |
Loading history...