remorhaz /
php-unilex
| Total Complexity | 93 |
| Total Lines | 609 |
| Duplicated Lines | 0 % |
| Changes | 10 | ||
| Bugs | 1 | Features | 0 |
Complex classes like TokenMatcherGenerator 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 TokenMatcherGenerator, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 32 | class TokenMatcherGenerator |
||
| 33 | { |
||
| 34 | private $spec; |
||
| 35 | |||
| 36 | private $output; |
||
| 37 | |||
| 38 | private $dfa; |
||
| 39 | |||
| 40 | private $regExpFinishMap = []; |
||
| 41 | |||
| 42 | private $conditionFunctions = []; |
||
| 43 | |||
| 44 | public function __construct(TokenMatcherSpec $spec) |
||
| 47 | } |
||
| 48 | |||
| 49 | /** |
||
| 50 | * @return string |
||
| 51 | * @throws Exception |
||
| 52 | * @throws ReflectionException |
||
| 53 | */ |
||
| 54 | private function buildOutput(): string |
||
| 55 | { |
||
| 56 | $this->conditionFunctions = []; |
||
| 57 | |||
| 58 | return |
||
| 59 | "{$this->buildFileComment()}\ndeclare(strict_types=1);\n\n" . |
||
| 60 | "{$this->buildHeader()}\n" . |
||
| 61 | "class {$this->spec->getTargetShortName()} extends {$this->spec->getTemplateClass()->getShortName()}\n" . |
||
| 62 | "{\n" . |
||
| 63 | " public function match({$this->buildMatchParameters()}): bool\n" . |
||
| 64 | " {\n{$this->buildMatchBody()}" . |
||
| 65 | " }\n" . |
||
| 66 | $this->buildConditionFunctions() . |
||
| 67 | "}\n"; |
||
| 68 | } |
||
| 69 | |||
| 70 | /** |
||
| 71 | * @return TokenMatcherInterface |
||
| 72 | * @throws Exception |
||
| 73 | */ |
||
| 74 | public function load(): TokenMatcherInterface |
||
| 75 | { |
||
| 76 | $targetClass = $this->spec->getTargetClassName(); |
||
| 77 | if (!class_exists($targetClass)) { |
||
| 78 | try { |
||
| 79 | $source = $this->getOutput(false); |
||
| 80 | eval($source); |
||
|
introduced
by
|
|||
| 81 | } catch (Throwable $e) { |
||
| 82 | throw new Exception("Invalid PHP code generated", 0, $e); |
||
| 83 | } |
||
| 84 | if (!class_exists($targetClass)) { |
||
| 85 | throw new Exception("Failed to generate target class"); |
||
| 86 | } |
||
| 87 | } |
||
| 88 | |||
| 89 | return new $targetClass(); |
||
| 90 | } |
||
| 91 | |||
| 92 | /** |
||
| 93 | * @param bool $asFile |
||
| 94 | * @return string |
||
| 95 | * @throws Exception |
||
| 96 | * @throws ReflectionException |
||
| 97 | */ |
||
| 98 | public function getOutput(bool $asFile = true): string |
||
| 99 | { |
||
| 100 | if (!isset($this->output)) { |
||
| 101 | $this->output = $this->buildOutput(); |
||
| 102 | } |
||
| 103 | |||
| 104 | return $asFile ? "<?php\n\n{$this->output}" : $this->output; |
||
| 105 | } |
||
| 106 | |||
| 107 | private function buildFileComment(): string |
||
| 108 | { |
||
| 109 | $content = $this->spec->getFileComment(); |
||
| 110 | if ('' == $content) { |
||
| 111 | return ''; |
||
| 112 | } |
||
| 113 | $comment = "/**\n"; |
||
| 114 | $commentLineList = explode("\n", $content); |
||
| 115 | foreach ($commentLineList as $commentLine) { |
||
| 116 | $comment .= rtrim(" * {$commentLine}") . "\n"; |
||
| 117 | } |
||
| 118 | $comment .= " */\n"; |
||
| 119 | |||
| 120 | return $comment; |
||
| 121 | } |
||
| 122 | |||
| 123 | /** |
||
| 124 | * @return string |
||
| 125 | * @throws ReflectionException |
||
| 126 | */ |
||
| 127 | public function buildHeader(): string |
||
| 128 | { |
||
| 129 | $headerParts = []; |
||
| 130 | $namespace = $this->spec->getTargetNamespaceName(); |
||
| 131 | if ($namespace != '') { |
||
| 132 | $headerParts[] = $this->buildMethodPart("namespace {$namespace};", 0); |
||
| 133 | } |
||
| 134 | $useList = $this->buildUseList(); |
||
| 135 | if ('' != $useList) { |
||
| 136 | $headerParts[] = $useList; |
||
| 137 | } |
||
| 138 | $header = $this->buildMethodPart($this->spec->getHeader(), 0); |
||
| 139 | if ('' != $header) { |
||
| 140 | $headerParts[] = $header; |
||
| 141 | } |
||
| 142 | |||
| 143 | return implode("\n", $headerParts); |
||
| 144 | } |
||
| 145 | |||
| 146 | /** |
||
| 147 | * @return string |
||
| 148 | * @throws ReflectionException |
||
| 149 | */ |
||
| 150 | private function buildUseList(): string |
||
| 151 | { |
||
| 152 | $result = ''; |
||
| 153 | foreach ($this->spec->getUsedClassList() as $alias => $className) { |
||
| 154 | $classWithAlias = is_string($alias) ? "{$className} {$alias}" : $className; |
||
| 155 | $result .= $this->buildMethodPart("use {$classWithAlias};", 0); |
||
| 156 | } |
||
| 157 | |||
| 158 | return $result; |
||
| 159 | } |
||
| 160 | |||
| 161 | /** |
||
| 162 | * @return string |
||
| 163 | * @throws ReflectionException |
||
| 164 | */ |
||
| 165 | private function buildMatchParameters(): string |
||
| 166 | { |
||
| 167 | $paramList = []; |
||
| 168 | foreach ($this->spec->getMatchMethod()->getParameters() as $matchParameter) { |
||
| 169 | if ($matchParameter->hasType()) { |
||
| 170 | $param = $matchParameter->getType()->isBuiltin() |
||
| 171 | ? $matchParameter->getType()->getName() |
||
| 172 | : (new ReflectionClass($matchParameter->getType()->getName()))->getShortName(); |
||
| 173 | $param .= " "; |
||
| 174 | } else { |
||
| 175 | $param = ""; |
||
| 176 | } |
||
| 177 | $param .= "\${$matchParameter->getName()}"; |
||
| 178 | $paramList[] = $param; |
||
| 179 | } |
||
| 180 | |||
| 181 | return implode(", ", $paramList); |
||
| 182 | } |
||
| 183 | |||
| 184 | /** |
||
| 185 | * @return string |
||
| 186 | * @throws Exception |
||
| 187 | */ |
||
| 188 | private function buildMatchBody(): string |
||
| 189 | { |
||
| 190 | $result = $this->buildBeforeMatch(); |
||
| 191 | |||
| 192 | foreach ($this->spec->getModeList() as $mode) { |
||
| 193 | if (TokenMatcherInterface::DEFAULT_MODE == $mode) { |
||
| 194 | continue; |
||
| 195 | } |
||
| 196 | $result .= |
||
| 197 | $this->buildMethodPart("if (\$context->getMode() == '{$mode}') {") . |
||
| 198 | $this->buildFsmEntry($mode, 3) . |
||
| 199 | $this->buildMethodPart("}"); |
||
| 200 | } |
||
| 201 | foreach ($this->spec->getModeList() as $mode) { |
||
| 202 | if (TokenMatcherInterface::DEFAULT_MODE == $mode) { |
||
| 203 | $result .= $this->buildFsmEntry(TokenMatcherInterface::DEFAULT_MODE) . "\n"; |
||
| 204 | } |
||
| 205 | $result .= $this->buildFsmMoves($mode); |
||
| 206 | } |
||
| 207 | |||
| 208 | $result .= $this->buildErrorState(); |
||
| 209 | |||
| 210 | return $result; |
||
| 211 | } |
||
| 212 | |||
| 213 | private function buildBeforeMatch(): string |
||
| 214 | { |
||
| 215 | return |
||
| 216 | $this->buildMethodPart("\$context = \$this->createContext(\$buffer, \$tokenFactory);") . |
||
| 217 | $this->buildMethodPart($this->spec->getBeforeMatch()); |
||
| 218 | } |
||
| 219 | |||
| 220 | /** |
||
| 221 | * @param string $mode |
||
| 222 | * @param int $indent |
||
| 223 | * @return string |
||
| 224 | * @throws Exception |
||
| 225 | */ |
||
| 226 | private function buildFsmEntry(string $mode, int $indent = 2): string |
||
| 227 | { |
||
| 228 | $state = $this->getDfa($mode)->getStateMap()->getStartState(); |
||
| 229 | |||
| 230 | return $this->buildMethodPart("goto {$this->buildStateLabel('state', $mode, $state)};", $indent); |
||
| 231 | } |
||
| 232 | |||
| 233 | private function buildStateLabel(string $prefix, string $mode, int $state): string |
||
| 240 | } |
||
| 241 | |||
| 242 | /** |
||
| 243 | * @param string $mode |
||
| 244 | * @return string |
||
| 245 | * @throws Exception |
||
| 246 | */ |
||
| 247 | private function buildFsmMoves(string $mode): string |
||
| 248 | { |
||
| 249 | $result = ''; |
||
| 250 | foreach ($this->getDfa($mode)->getStateMap()->getStateList() as $stateIn) { |
||
| 251 | if ($this->isFinishStateWithSingleEnteringTransition($mode, $stateIn)) { |
||
| 252 | continue; |
||
| 253 | } |
||
| 254 | $result .= |
||
| 255 | $this->buildStateEntry($mode, $stateIn) . |
||
| 256 | $this->buildStateTransitionList($mode, $stateIn) . |
||
| 257 | $this->buildStateFinish($mode, $stateIn); |
||
| 258 | } |
||
| 259 | |||
| 260 | return $result; |
||
| 261 | } |
||
| 262 | |||
| 263 | /** |
||
| 264 | * @param string $mode |
||
| 265 | * @param int $stateIn |
||
| 266 | * @return string |
||
| 267 | * @throws Exception |
||
| 268 | */ |
||
| 269 | private function buildStateEntry(string $mode, int $stateIn): string |
||
| 270 | { |
||
| 271 | $result = ''; |
||
| 272 | $result .= $this->buildMethodPart("{$this->buildStateLabel('state', $mode, $stateIn)}:"); |
||
| 273 | $moves = $this->getDfa($mode)->getTransitionMap()->getExitList($stateIn); |
||
| 274 | if (empty($moves)) { |
||
| 275 | return $result; |
||
| 276 | } |
||
| 277 | $result .= $this->buildMethodPart("if (\$context->getBuffer()->isEnd()) {"); |
||
| 278 | $result .= $this->getDfa($mode)->getStateMap()->isFinishState($stateIn) |
||
| 279 | ? $this->buildMethodPart("goto {$this->buildStateLabel('finish', $mode, $stateIn)};", 3) |
||
| 280 | : $this->buildMethodPart("goto error;", 3); |
||
| 281 | $result .= |
||
| 282 | $this->buildMethodPart("}") . |
||
| 283 | $this->buildMethodPart("\$char = \$context->getBuffer()->getSymbol();"); |
||
| 284 | |||
| 285 | return $result; |
||
| 286 | } |
||
| 287 | |||
| 288 | /** |
||
| 289 | * @param string $mode |
||
| 290 | * @param int $stateIn |
||
| 291 | * @return string |
||
| 292 | * @throws Exception |
||
| 293 | */ |
||
| 294 | private function buildStateTransitionList(string $mode, int $stateIn): string |
||
| 295 | { |
||
| 296 | $result = ''; |
||
| 297 | foreach ($this->getDfa($mode)->getTransitionMap()->getExitList($stateIn) as $stateOut => $symbolList) { |
||
| 298 | foreach ($symbolList as $symbol) { |
||
| 299 | $result .= |
||
| 300 | $this->buildMethodPart("if ({$this->buildRangeSetCondition($mode, $symbol)}) {") . |
||
| 301 | $this->buildOnTransition() . |
||
| 302 | $this->buildMethodPart("\$context->getBuffer()->nextSymbol();", 3); |
||
| 303 | $result .= $this->isFinishStateWithSingleEnteringTransition($mode, $stateOut) |
||
| 304 | ? $this->buildToken($mode, $stateOut, 3) |
||
| 305 | : $this->buildStateTransition($mode, $stateOut, 3); |
||
| 306 | $result .= $this->buildMethodPart("}"); |
||
| 307 | } |
||
| 308 | } |
||
| 309 | |||
| 310 | return $result; |
||
| 311 | } |
||
| 312 | |||
| 313 | /** |
||
| 314 | * @param string $mode |
||
| 315 | * @param int $stateOut |
||
| 316 | * @param int $indent |
||
| 317 | * @return string |
||
| 318 | */ |
||
| 319 | private function buildStateTransition(string $mode, int $stateOut, int $indent = 3): string |
||
| 322 | } |
||
| 323 | |||
| 324 | /** |
||
| 325 | * @param string $mode |
||
| 326 | * @param int $stateOut |
||
| 327 | * @return bool |
||
| 328 | * @throws Exception |
||
| 329 | */ |
||
| 330 | private function isFinishStateWithSingleEnteringTransition(string $mode, int $stateOut): bool |
||
| 331 | { |
||
| 332 | if (!$this->getDfa($mode)->getStateMap()->isFinishState($stateOut)) { |
||
| 333 | return false; |
||
| 334 | } |
||
| 335 | $enters = $this->getDfa($mode)->getTransitionMap()->getEnterList($stateOut); |
||
| 336 | $exits = $this->getDfa($mode)->getTransitionMap()->getExitList($stateOut); |
||
| 337 | if (!(count($enters) == 1 && count($exits) == 0)) { |
||
| 338 | return false; |
||
| 339 | } |
||
| 340 | $symbolList = array_pop($enters); |
||
| 341 | |||
| 342 | return count($symbolList) == 1; |
||
| 343 | } |
||
| 344 | |||
| 345 | private function buildHex(int $char): string |
||
| 346 | { |
||
| 347 | $hexChar = strtoupper(dechex($char)); |
||
| 348 | if (strlen($hexChar) % 2 != 0) { |
||
| 349 | $hexChar = "0{$hexChar}"; |
||
| 350 | } |
||
| 351 | |||
| 352 | return "0x{$hexChar}"; |
||
| 353 | } |
||
| 354 | |||
| 355 | private function buildRangeCondition(RangeInterface $range): array |
||
| 356 | { |
||
| 357 | $startChar = $this->buildHex($range->getStart()); |
||
| 358 | if ($range->getStart() == $range->getFinish()) { |
||
| 359 | return ["{$startChar} == \$char"]; |
||
| 360 | } |
||
| 361 | $finishChar = $this->buildHex($range->getFinish()); |
||
| 362 | if ($range->getStart() + 1 == $range->getFinish()) { |
||
| 363 | return [ |
||
| 364 | "{$startChar} == \$char", |
||
| 365 | "{$finishChar} == \$char", |
||
| 366 | ]; |
||
| 367 | } |
||
| 368 | |||
| 369 | return ["{$startChar} <= \$char && \$char <= {$finishChar}"]; |
||
| 370 | } |
||
| 371 | |||
| 372 | /** |
||
| 373 | * @param string $mode |
||
| 374 | * @param int $symbol |
||
| 375 | * @return string |
||
| 376 | * @throws Exception |
||
| 377 | */ |
||
| 378 | private function buildRangeSetCondition(string $mode, int $symbol): string |
||
| 379 | { |
||
| 380 | $rangeSet = $this->getDfa($mode)->getSymbolTable()->getRangeSet($symbol); |
||
| 381 | |||
| 382 | $conditionList = []; |
||
| 383 | foreach ($rangeSet->getRanges() as $range) { |
||
| 384 | $conditionList = array_merge($conditionList, $this->buildRangeCondition($range)); |
||
| 385 | } |
||
| 386 | $result = implode(" || ", $conditionList); |
||
| 387 | if (strlen($result) + 15 <= 120 || count($conditionList) == 1) { |
||
| 388 | return ltrim($result); |
||
| 389 | } |
||
| 390 | $result = $this->buildMethodPart(implode(" ||\n", $conditionList), 1); |
||
| 391 | if (count($conditionList) > 10) { |
||
| 392 | $method = "isMode" . ucfirst($mode) . "Symbol{$symbol}"; |
||
| 393 | $this->conditionFunctions[$method] = $result; |
||
| 394 | |||
| 395 | return "\$this->{$method}(\$char)"; |
||
| 396 | } |
||
| 397 | |||
| 398 | return "\n " . ltrim($result); |
||
| 399 | } |
||
| 400 | |||
| 401 | private function buildConditionFunctions(): string |
||
| 402 | { |
||
| 403 | $result = ''; |
||
| 404 | |||
| 405 | foreach ($this->conditionFunctions as $method => $conditionList) { |
||
| 406 | $result .= |
||
| 407 | "\n private function {$method}(int \$char): bool\n {\n" . |
||
| 408 | $this->buildMethodPart("return") . |
||
| 409 | $this->buildMethodPart(rtrim($conditionList) . ';') . |
||
| 410 | " }\n"; |
||
| 411 | } |
||
| 412 | |||
| 413 | return $result; |
||
| 414 | } |
||
| 415 | |||
| 416 | /** |
||
| 417 | * @param string $mode |
||
| 418 | * @param int $stateIn |
||
| 419 | * @return string |
||
| 420 | * @throws Exception |
||
| 421 | */ |
||
| 422 | private function buildStateFinish(string $mode, int $stateIn): string |
||
| 423 | { |
||
| 424 | if (!$this->getDfa($mode)->getStateMap()->isFinishState($stateIn)) { |
||
| 425 | return $this->buildMethodPart("goto error;\n"); |
||
| 426 | } |
||
| 427 | $result = ''; |
||
| 428 | if (!empty($this->getDfa($mode)->getTransitionMap()->getExitList($stateIn))) { |
||
| 429 | $result .= $this->buildMethodPart("{$this->buildStateLabel('finish', $mode, $stateIn)}:"); |
||
| 430 | } |
||
| 431 | $result .= "{$this->buildToken($mode, $stateIn)}\n"; |
||
| 432 | |||
| 433 | return $result; |
||
| 434 | } |
||
| 435 | |||
| 436 | /** |
||
| 437 | * @param string $mode |
||
| 438 | * @param int $stateIn |
||
| 439 | * @param int $indent |
||
| 440 | * @return string |
||
| 441 | * @throws Exception |
||
| 442 | */ |
||
| 443 | private function buildToken(string $mode, int $stateIn, int $indent = 2): string |
||
| 444 | { |
||
| 445 | if (!isset($this->regExpFinishMap[$mode][$stateIn])) { |
||
| 446 | throw new Exception("No regular expressions found for state {$mode}:{$stateIn}"); |
||
| 447 | } |
||
| 448 | $tokenSpec = $this->spec->getTokenSpec($mode, $this->regExpFinishMap[$mode][$stateIn]); |
||
| 449 | |||
| 450 | return |
||
| 451 | $this->buildMethodPart("// {$tokenSpec->getRegExp()}", $indent) . |
||
| 452 | $this->buildSingleToken($tokenSpec, $indent); |
||
| 453 | } |
||
| 454 | |||
| 455 | private function buildSingleToken(TokenSpec $tokenSpec, int $indent): string |
||
| 456 | { |
||
| 457 | return |
||
| 458 | $this->buildMethodPart($tokenSpec->getCode(), $indent) . |
||
| 459 | $this->buildOnToken($indent) . "\n" . |
||
| 460 | $this->buildMethodPart("return true;", $indent); |
||
| 461 | } |
||
| 462 | |||
| 463 | private function buildErrorState(): string |
||
| 464 | { |
||
| 465 | $code = $this->spec->getOnError(); |
||
| 466 | |||
| 467 | return |
||
| 468 | $this->buildMethodPart("error:") . |
||
| 469 | $this->buildMethodPart('' == $code ? "return false;" : $code); |
||
| 470 | } |
||
| 471 | |||
| 472 | private function buildMethodPart(string $code, int $indent = 2): string |
||
| 473 | { |
||
| 474 | if ('' == $code) { |
||
| 475 | return ''; |
||
| 476 | } |
||
| 477 | $result = ''; |
||
| 478 | $codeLineList = explode("\n", $code); |
||
| 479 | foreach ($codeLineList as $codeLine) { |
||
| 480 | $line = ''; |
||
| 481 | for ($i = 0; $i < $indent; $i++) { |
||
| 482 | $line .= " "; |
||
| 483 | } |
||
| 484 | $result .= rtrim($line . $codeLine) . "\n"; |
||
| 485 | } |
||
| 486 | |||
| 487 | return $result; |
||
| 488 | } |
||
| 489 | |||
| 490 | private function buildOnTransition(): string |
||
| 491 | { |
||
| 492 | return $this->buildMethodPart($this->spec->getOnTransition(), 3); |
||
| 493 | } |
||
| 494 | |||
| 495 | private function buildOnToken(int $indent = 2): string |
||
| 496 | { |
||
| 497 | return $this->buildMethodPart($this->spec->getOnToken(), $indent); |
||
| 498 | } |
||
| 499 | |||
| 500 | /** |
||
| 501 | * @param string $context |
||
| 502 | * @return Dfa |
||
| 503 | * @throws Exception |
||
| 504 | */ |
||
| 505 | private function getDfa(string $context): Dfa |
||
| 506 | { |
||
| 507 | if (!isset($this->dfa[$context])) { |
||
| 508 | $this->dfa[$context] = $this->buildDfa($context); |
||
| 509 | } |
||
| 510 | |||
| 511 | return $this->dfa[$context]; |
||
| 512 | } |
||
| 513 | |||
| 514 | /** |
||
| 515 | * @param string $mode |
||
| 516 | * @return Dfa |
||
| 517 | * @throws Exception |
||
| 518 | */ |
||
| 519 | private function buildDfa(string $mode): Dfa |
||
| 520 | { |
||
| 521 | $nfa = new Nfa(); |
||
| 522 | $startState = $nfa->getStateMap()->createState(); |
||
| 523 | $nfa->getStateMap()->addStartState($startState); |
||
| 524 | $nfaRegExpMap = []; |
||
| 525 | /** @var Dfa[] $dfaList */ |
||
| 526 | $dfaList = []; |
||
| 527 | foreach ($this->spec->getTokenSpecList($mode) as $tokenSpec) { |
||
| 528 | $existingStates = $nfa->getStateMap()->getStateList(); |
||
| 529 | $regExpEntryState = $nfa->getStateMap()->createState(); |
||
| 530 | $nfa |
||
| 531 | ->getEpsilonTransitionMap() |
||
| 532 | ->addTransition($startState, $regExpEntryState, true); |
||
| 533 | $this->buildRegExp($nfa, $regExpEntryState, $tokenSpec->getRegExp()); |
||
| 534 | $regExpStates = array_diff($nfa->getStateMap()->getStateList(), $existingStates); |
||
| 535 | $nfaRegExpMap[$tokenSpec->getRegExp()] = $regExpStates; |
||
| 536 | $dfaList[$tokenSpec->getRegExp()] = $this->buildIndependentRegExp($tokenSpec->getRegExp()); |
||
| 537 | } |
||
| 538 | |||
| 539 | $joinedNfa = new Nfa(); |
||
| 540 | $startState = $joinedNfa->getStateMap()->createState(); |
||
| 541 | $joinedNfa->getStateMap()->addStartState($startState); |
||
| 542 | $languageBuilder = LanguageBuilder::forNfa($joinedNfa); |
||
| 543 | $joinedNfaStates = []; |
||
| 544 | $nfaRegExpMap = []; |
||
| 545 | $regExpFinishMap = []; |
||
| 546 | foreach ($dfaList as $regExp => $dfa) { |
||
| 547 | $nfaRegExpMap[$regExp] = []; |
||
| 548 | foreach ($dfa->getStateMap()->getStateList() as $dfaState) { |
||
| 549 | $nfaState = $joinedNfa->getStateMap()->createState(); |
||
| 550 | $nfaRegExpMap[$regExp][] = $nfaState; |
||
| 551 | $joinedNfaStates[$dfaState] = $nfaState; |
||
| 552 | if ($dfa->getStateMap()->isStartState($dfaState)) { |
||
| 553 | $joinedNfa->getEpsilonTransitionMap()->addTransition($startState, $nfaState, true); |
||
| 554 | } |
||
| 555 | if ($dfa->getStateMap()->isFinishState($dfaState)) { |
||
| 556 | $regExpFinishMap[$regExp][] = $nfaState; |
||
| 557 | $joinedNfa->getStateMap()->addFinishState($nfaState); |
||
| 558 | } |
||
| 559 | } |
||
| 560 | foreach ($dfa->getTransitionMap()->getTransitionList() as $dfaStateIn => $transitions) { |
||
| 561 | foreach ($transitions as $dfaStateOut => $symbols) { |
||
| 562 | foreach ($symbols as $symbol) { |
||
| 563 | $rangeSet = $dfa->getSymbolTable()->getRangeSet($symbol); |
||
| 564 | $newSymbols = $languageBuilder->getSymbolList(...$rangeSet->getRanges()); |
||
| 565 | $oldSymbols = $joinedNfa |
||
| 566 | ->getSymbolTransitionMap() |
||
| 567 | ->transitionExists($joinedNfaStates[$dfaStateIn], $joinedNfaStates[$dfaStateOut]) |
||
| 568 | ? $joinedNfa |
||
| 569 | ->getSymbolTransitionMap() |
||
| 570 | ->getTransition($joinedNfaStates[$dfaStateIn], $joinedNfaStates[$dfaStateOut]) |
||
| 571 | : []; |
||
| 572 | $joinedNfa->getSymbolTransitionMap()->replaceTransition( |
||
| 573 | $joinedNfaStates[$dfaStateIn], |
||
| 574 | $joinedNfaStates[$dfaStateOut], |
||
| 575 | array_unique(array_merge($oldSymbols, $newSymbols)) |
||
| 576 | ); |
||
| 577 | } |
||
| 578 | } |
||
| 579 | } |
||
| 580 | } |
||
| 581 | |||
| 582 | $dfa = new Dfa(); |
||
| 583 | (new DfaBuilder($dfa, $joinedNfa))->run(); |
||
| 584 | |||
| 585 | $dfaRegExpFinishMap = []; |
||
| 586 | foreach ($dfa->getStateMap()->getFinishStateList() as $dfaFinishState) { |
||
| 587 | $nfaFinishStates = array_intersect( |
||
| 588 | $dfa->getStateMap()->getStateValue($dfaFinishState), |
||
| 589 | $joinedNfa->getStateMap()->getFinishStateList() |
||
| 590 | ); |
||
| 591 | foreach ($regExpFinishMap as $regExp => $regExpFinishStates) { |
||
| 592 | foreach ($nfaFinishStates as $nfaFinishState) { |
||
| 593 | if (in_array($nfaFinishState, $regExpFinishStates)) { |
||
| 594 | $dfaRegExpFinishMap[$dfaFinishState] = (string) $regExp; |
||
| 595 | break 2; |
||
| 596 | } |
||
| 597 | } |
||
| 598 | } |
||
| 599 | } |
||
| 600 | foreach ($this->spec->getTokenSpecList($mode) as $tokenSpec) { |
||
| 601 | if (!in_array($tokenSpec->getRegExp(), $dfaRegExpFinishMap)) { |
||
| 602 | throw new Exception("Token not reachable for regular expression: {$tokenSpec->getRegExp()} "); |
||
| 603 | } |
||
| 604 | } |
||
| 605 | $this->regExpFinishMap[$mode] = $dfaRegExpFinishMap; |
||
| 606 | |||
| 607 | return $dfa; |
||
| 608 | } |
||
| 609 | |||
| 610 | /** |
||
| 611 | * @param Nfa $nfa |
||
| 612 | * @param int $entryState |
||
| 613 | * @param string $regExp |
||
| 614 | * @throws Exception |
||
| 615 | */ |
||
| 616 | private function buildRegExp(Nfa $nfa, int $entryState, string $regExp): void |
||
| 624 | } |
||
| 625 | |||
| 626 | /** |
||
| 627 | * @param string $regExp |
||
| 628 | * @return Dfa |
||
| 629 | * @throws Exception |
||
| 630 | */ |
||
| 631 | private function buildIndependentRegExp(string $regExp): Dfa |
||
| 632 | { |
||
| 633 | $buffer = CharBufferFactory::createFromString($regExp); |
||
| 634 | $tree = new Tree(); |
||
| 635 | ParserFactory::createFromBuffer($tree, $buffer)->run(); |
||
| 636 | $nfa = new Nfa(); |
||
| 637 | $nfaBuilder = new NfaBuilder($nfa, PropertyRangeLoader::create()); |
||
| 638 | (new Translator($tree, $nfaBuilder))->run(); |
||
| 639 | |||
| 640 | return DfaBuilder::fromNfa($nfa); |
||
| 641 | } |
||
| 642 | } |
||
| 643 |
Loading history...