jsminplus.php
2092 lines
| 1 | <?php |
| 2 | |
| 3 | namespace MailPoetVendor\pQuery; |
| 4 | |
| 5 | if (!defined('ABSPATH')) exit; |
| 6 | |
| 7 | |
| 8 | /** |
| 9 | * JSMinPlus version 1.4 |
| 10 | * |
| 11 | * Minifies a javascript file using a javascript parser |
| 12 | * |
| 13 | * This implements a PHP port of Brendan Eich's Narcissus open source javascript engine (in javascript) |
| 14 | * References: http://en.wikipedia.org/wiki/Narcissus_(JavaScript_engine) |
| 15 | * Narcissus sourcecode: http://mxr.mozilla.org/mozilla/source/js/narcissus/ |
| 16 | * JSMinPlus weblog: http://crisp.tweakblogs.net/blog/cat/716 |
| 17 | * |
| 18 | * Tino Zijdel <crisp@tweakers.net> |
| 19 | * |
| 20 | * Usage: $minified = JSMinPlus::minify($script [, $filename]) |
| 21 | * |
| 22 | * Versionlog (see also changelog.txt): |
| 23 | * 23-07-2011 - remove dynamic creation of OP_* and KEYWORD_* defines and declare them on top |
| 24 | * reduce memory footprint by minifying by block-scope |
| 25 | * some small byte-saving and performance improvements |
| 26 | * 12-05-2009 - fixed hook:colon precedence, fixed empty body in loop and if-constructs |
| 27 | * 18-04-2009 - fixed crashbug in PHP 5.2.9 and several other bugfixes |
| 28 | * 12-04-2009 - some small bugfixes and performance improvements |
| 29 | * 09-04-2009 - initial open sourced version 1.0 |
| 30 | * |
| 31 | * Latest version of this script: http://files.tweakers.net/jsminplus/jsminplus.zip |
| 32 | * |
| 33 | */ |
| 34 | |
| 35 | /* ***** BEGIN LICENSE BLOCK ***** |
| 36 | * Version: MPL 1.1/GPL 2.0/LGPL 2.1 |
| 37 | * |
| 38 | * The contents of this file are subject to the Mozilla Public License Version |
| 39 | * 1.1 (the "License"); you may not use this file except in compliance with |
| 40 | * the License. You may obtain a copy of the License at |
| 41 | * http://www.mozilla.org/MPL/ |
| 42 | * |
| 43 | * Software distributed under the License is distributed on an "AS IS" basis, |
| 44 | * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License |
| 45 | * for the specific language governing rights and limitations under the |
| 46 | * License. |
| 47 | * |
| 48 | * The Original Code is the Narcissus JavaScript engine. |
| 49 | * |
| 50 | * The Initial Developer of the Original Code is |
| 51 | * Brendan Eich <brendan@mozilla.org>. |
| 52 | * Portions created by the Initial Developer are Copyright (C) 2004 |
| 53 | * the Initial Developer. All Rights Reserved. |
| 54 | * |
| 55 | * Contributor(s): Tino Zijdel <crisp@tweakers.net> |
| 56 | * PHP port, modifications and minifier routine are (C) 2009-2011 |
| 57 | * |
| 58 | * Alternatively, the contents of this file may be used under the terms of |
| 59 | * either the GNU General Public License Version 2 or later (the "GPL"), or |
| 60 | * the GNU Lesser General Public License Version 2.1 or later (the "LGPL"), |
| 61 | * in which case the provisions of the GPL or the LGPL are applicable instead |
| 62 | * of those above. If you wish to allow use of your version of this file only |
| 63 | * under the terms of either the GPL or the LGPL, and not to allow others to |
| 64 | * use your version of this file under the terms of the MPL, indicate your |
| 65 | * decision by deleting the provisions above and replace them with the notice |
| 66 | * and other provisions required by the GPL or the LGPL. If you do not delete |
| 67 | * the provisions above, a recipient may use your version of this file under |
| 68 | * the terms of any one of the MPL, the GPL or the LGPL. |
| 69 | * |
| 70 | * ***** END LICENSE BLOCK ***** */ |
| 71 | |
| 72 | define('TOKEN_END', 1); |
| 73 | define('TOKEN_NUMBER', 2); |
| 74 | define('TOKEN_IDENTIFIER', 3); |
| 75 | define('TOKEN_STRING', 4); |
| 76 | define('TOKEN_REGEXP', 5); |
| 77 | define('TOKEN_NEWLINE', 6); |
| 78 | define('TOKEN_CONDCOMMENT_START', 7); |
| 79 | define('TOKEN_CONDCOMMENT_END', 8); |
| 80 | |
| 81 | define('JS_SCRIPT', 100); |
| 82 | define('JS_BLOCK', 101); |
| 83 | define('JS_LABEL', 102); |
| 84 | define('JS_FOR_IN', 103); |
| 85 | define('JS_CALL', 104); |
| 86 | define('JS_NEW_WITH_ARGS', 105); |
| 87 | define('JS_INDEX', 106); |
| 88 | define('JS_ARRAY_INIT', 107); |
| 89 | define('JS_OBJECT_INIT', 108); |
| 90 | define('JS_PROPERTY_INIT', 109); |
| 91 | define('JS_GETTER', 110); |
| 92 | define('JS_SETTER', 111); |
| 93 | define('JS_GROUP', 112); |
| 94 | define('JS_LIST', 113); |
| 95 | |
| 96 | define('JS_MINIFIED', 999); |
| 97 | |
| 98 | define('DECLARED_FORM', 0); |
| 99 | define('EXPRESSED_FORM', 1); |
| 100 | define('STATEMENT_FORM', 2); |
| 101 | |
| 102 | /* Operators */ |
| 103 | define('OP_SEMICOLON', ';'); |
| 104 | define('OP_COMMA', ','); |
| 105 | define('OP_HOOK', '?'); |
| 106 | define('OP_COLON', ':'); |
| 107 | define('OP_OR', '||'); |
| 108 | define('OP_AND', '&&'); |
| 109 | define('OP_BITWISE_OR', '|'); |
| 110 | define('OP_BITWISE_XOR', '^'); |
| 111 | define('OP_BITWISE_AND', '&'); |
| 112 | define('OP_STRICT_EQ', '==='); |
| 113 | define('OP_EQ', '=='); |
| 114 | define('OP_ASSIGN', '='); |
| 115 | define('OP_STRICT_NE', '!=='); |
| 116 | define('OP_NE', '!='); |
| 117 | define('OP_LSH', '<<'); |
| 118 | define('OP_LE', '<='); |
| 119 | define('OP_LT', '<'); |
| 120 | define('OP_URSH', '>>>'); |
| 121 | define('OP_RSH', '>>'); |
| 122 | define('OP_GE', '>='); |
| 123 | define('OP_GT', '>'); |
| 124 | define('OP_INCREMENT', '++'); |
| 125 | define('OP_DECREMENT', '--'); |
| 126 | define('OP_PLUS', '+'); |
| 127 | define('OP_MINUS', '-'); |
| 128 | define('OP_MUL', '*'); |
| 129 | define('OP_DIV', '/'); |
| 130 | define('OP_MOD', '%'); |
| 131 | define('OP_NOT', '!'); |
| 132 | define('OP_BITWISE_NOT', '~'); |
| 133 | define('OP_DOT', '.'); |
| 134 | define('OP_LEFT_BRACKET', '['); |
| 135 | define('OP_RIGHT_BRACKET', ']'); |
| 136 | define('OP_LEFT_CURLY', '{'); |
| 137 | define('OP_RIGHT_CURLY', '}'); |
| 138 | define('OP_LEFT_PAREN', '('); |
| 139 | define('OP_RIGHT_PAREN', ')'); |
| 140 | define('OP_CONDCOMMENT_END', '@*/'); |
| 141 | |
| 142 | define('OP_UNARY_PLUS', 'U+'); |
| 143 | define('OP_UNARY_MINUS', 'U-'); |
| 144 | |
| 145 | /* Keywords */ |
| 146 | define('KEYWORD_BREAK', 'break'); |
| 147 | define('KEYWORD_CASE', 'case'); |
| 148 | define('KEYWORD_CATCH', 'catch'); |
| 149 | define('KEYWORD_CONST', 'const'); |
| 150 | define('KEYWORD_CONTINUE', 'continue'); |
| 151 | define('KEYWORD_DEBUGGER', 'debugger'); |
| 152 | define('KEYWORD_DEFAULT', 'default'); |
| 153 | define('KEYWORD_DELETE', 'delete'); |
| 154 | define('KEYWORD_DO', 'do'); |
| 155 | define('KEYWORD_ELSE', 'else'); |
| 156 | define('KEYWORD_ENUM', 'enum'); |
| 157 | define('KEYWORD_FALSE', 'false'); |
| 158 | define('KEYWORD_FINALLY', 'finally'); |
| 159 | define('KEYWORD_FOR', 'for'); |
| 160 | define('KEYWORD_FUNCTION', 'function'); |
| 161 | define('KEYWORD_IF', 'if'); |
| 162 | define('KEYWORD_IN', 'in'); |
| 163 | define('KEYWORD_INSTANCEOF', 'instanceof'); |
| 164 | define('KEYWORD_NEW', 'new'); |
| 165 | define('KEYWORD_NULL', 'null'); |
| 166 | define('KEYWORD_RETURN', 'return'); |
| 167 | define('KEYWORD_SWITCH', 'switch'); |
| 168 | define('KEYWORD_THIS', 'this'); |
| 169 | define('KEYWORD_THROW', 'throw'); |
| 170 | define('KEYWORD_TRUE', 'true'); |
| 171 | define('KEYWORD_TRY', 'try'); |
| 172 | define('KEYWORD_TYPEOF', 'typeof'); |
| 173 | define('KEYWORD_VAR', 'var'); |
| 174 | define('KEYWORD_VOID', 'void'); |
| 175 | define('KEYWORD_WHILE', 'while'); |
| 176 | define('KEYWORD_WITH', 'with'); |
| 177 | |
| 178 | |
| 179 | class JSMinPlus |
| 180 | { |
| 181 | private $parser; |
| 182 | private $reserved = array( |
| 183 | 'break', 'case', 'catch', 'continue', 'default', 'delete', 'do', |
| 184 | 'else', 'finally', 'for', 'function', 'if', 'in', 'instanceof', |
| 185 | 'new', 'return', 'switch', 'this', 'throw', 'try', 'typeof', 'var', |
| 186 | 'void', 'while', 'with', |
| 187 | // Words reserved for future use |
| 188 | 'abstract', 'boolean', 'byte', 'char', 'class', 'const', 'debugger', |
| 189 | 'double', 'enum', 'export', 'extends', 'final', 'float', 'goto', |
| 190 | 'implements', 'import', 'int', 'interface', 'long', 'native', |
| 191 | 'package', 'private', 'protected', 'public', 'short', 'static', |
| 192 | 'super', 'synchronized', 'throws', 'transient', 'volatile', |
| 193 | // These are not reserved, but should be taken into account |
| 194 | // in isValidIdentifier (See jslint source code) |
| 195 | 'arguments', 'eval', 'true', 'false', 'Infinity', 'NaN', 'null', 'undefined' |
| 196 | ); |
| 197 | |
| 198 | private function __construct() |
| 199 | { |
| 200 | $this->parser = new JSParser($this); |
| 201 | } |
| 202 | |
| 203 | public static function minify($js, $filename='') |
| 204 | { |
| 205 | static $instance; |
| 206 | |
| 207 | // this is a singleton |
| 208 | if(!$instance) |
| 209 | $instance = new JSMinPlus(); |
| 210 | |
| 211 | return $instance->min($js, $filename); |
| 212 | } |
| 213 | |
| 214 | private function min($js, $filename) |
| 215 | { |
| 216 | try |
| 217 | { |
| 218 | $n = $this->parser->parse($js, $filename, 1); |
| 219 | return $this->parseTree($n); |
| 220 | } |
| 221 | catch(Exception $e) |
| 222 | { |
| 223 | echo $e->getMessage() . "\n"; |
| 224 | } |
| 225 | |
| 226 | return false; |
| 227 | } |
| 228 | |
| 229 | public function parseTree($n, $noBlockGrouping = false) |
| 230 | { |
| 231 | $s = ''; |
| 232 | |
| 233 | switch ($n->type) |
| 234 | { |
| 235 | case JS_MINIFIED: |
| 236 | $s = $n->value; |
| 237 | break; |
| 238 | |
| 239 | case JS_SCRIPT: |
| 240 | // we do nothing yet with funDecls or varDecls |
| 241 | $noBlockGrouping = true; |
| 242 | // FALL THROUGH |
| 243 | |
| 244 | case JS_BLOCK: |
| 245 | $childs = $n->treeNodes; |
| 246 | $lastType = 0; |
| 247 | for ($c = 0, $i = 0, $j = count($childs); $i < $j; $i++) |
| 248 | { |
| 249 | $type = $childs[$i]->type; |
| 250 | $t = $this->parseTree($childs[$i]); |
| 251 | if (strlen($t)) |
| 252 | { |
| 253 | if ($c) |
| 254 | { |
| 255 | $s = rtrim($s, ';'); |
| 256 | |
| 257 | if ($type == KEYWORD_FUNCTION && $childs[$i]->functionForm == DECLARED_FORM) |
| 258 | { |
| 259 | // put declared functions on a new line |
| 260 | $s .= "\n"; |
| 261 | } |
| 262 | elseif ($type == KEYWORD_VAR && $type == $lastType) |
| 263 | { |
| 264 | // mutiple var-statements can go into one |
| 265 | $t = ',' . substr($t, 4); |
| 266 | } |
| 267 | else |
| 268 | { |
| 269 | // add terminator |
| 270 | $s .= ';'; |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | $s .= $t; |
| 275 | |
| 276 | $c++; |
| 277 | $lastType = $type; |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | if ($c > 1 && !$noBlockGrouping) |
| 282 | { |
| 283 | $s = '{' . $s . '}'; |
| 284 | } |
| 285 | break; |
| 286 | |
| 287 | case KEYWORD_FUNCTION: |
| 288 | $s .= 'function' . ($n->name ? ' ' . $n->name : '') . '('; |
| 289 | $params = $n->params; |
| 290 | for ($i = 0, $j = count($params); $i < $j; $i++) |
| 291 | $s .= ($i ? ',' : '') . $params[$i]; |
| 292 | $s .= '){' . $this->parseTree($n->body, true) . '}'; |
| 293 | break; |
| 294 | |
| 295 | case KEYWORD_IF: |
| 296 | $s = 'if(' . $this->parseTree($n->condition) . ')'; |
| 297 | $thenPart = $this->parseTree($n->thenPart); |
| 298 | $elsePart = $n->elsePart ? $this->parseTree($n->elsePart) : null; |
| 299 | |
| 300 | // empty if-statement |
| 301 | if ($thenPart == '') |
| 302 | $thenPart = ';'; |
| 303 | |
| 304 | if ($elsePart) |
| 305 | { |
| 306 | // be carefull and always make a block out of the thenPart; could be more optimized but is a lot of trouble |
| 307 | if ($thenPart != ';' && $thenPart[0] != '{') |
| 308 | $thenPart = '{' . $thenPart . '}'; |
| 309 | |
| 310 | $s .= $thenPart . 'else'; |
| 311 | |
| 312 | // we could check for more, but that hardly ever applies so go for performance |
| 313 | if ($elsePart[0] != '{') |
| 314 | $s .= ' '; |
| 315 | |
| 316 | $s .= $elsePart; |
| 317 | } |
| 318 | else |
| 319 | { |
| 320 | $s .= $thenPart; |
| 321 | } |
| 322 | break; |
| 323 | |
| 324 | case KEYWORD_SWITCH: |
| 325 | $s = 'switch(' . $this->parseTree($n->discriminant) . '){'; |
| 326 | $cases = $n->cases; |
| 327 | for ($i = 0, $j = count($cases); $i < $j; $i++) |
| 328 | { |
| 329 | $case = $cases[$i]; |
| 330 | if ($case->type == KEYWORD_CASE) |
| 331 | $s .= 'case' . ($case->caseLabel->type != TOKEN_STRING ? ' ' : '') . $this->parseTree($case->caseLabel) . ':'; |
| 332 | else |
| 333 | $s .= 'default:'; |
| 334 | |
| 335 | $statement = $this->parseTree($case->statements, true); |
| 336 | if ($statement) |
| 337 | { |
| 338 | $s .= $statement; |
| 339 | // no terminator for last statement |
| 340 | if ($i + 1 < $j) |
| 341 | $s .= ';'; |
| 342 | } |
| 343 | } |
| 344 | $s .= '}'; |
| 345 | break; |
| 346 | |
| 347 | case KEYWORD_FOR: |
| 348 | $s = 'for(' . ($n->setup ? $this->parseTree($n->setup) : '') |
| 349 | . ';' . ($n->condition ? $this->parseTree($n->condition) : '') |
| 350 | . ';' . ($n->update ? $this->parseTree($n->update) : '') . ')'; |
| 351 | |
| 352 | $body = $this->parseTree($n->body); |
| 353 | if ($body == '') |
| 354 | $body = ';'; |
| 355 | |
| 356 | $s .= $body; |
| 357 | break; |
| 358 | |
| 359 | case KEYWORD_WHILE: |
| 360 | $s = 'while(' . $this->parseTree($n->condition) . ')'; |
| 361 | |
| 362 | $body = $this->parseTree($n->body); |
| 363 | if ($body == '') |
| 364 | $body = ';'; |
| 365 | |
| 366 | $s .= $body; |
| 367 | break; |
| 368 | |
| 369 | case JS_FOR_IN: |
| 370 | $s = 'for(' . ($n->varDecl ? $this->parseTree($n->varDecl) : $this->parseTree($n->iterator)) . ' in ' . $this->parseTree($n->object) . ')'; |
| 371 | |
| 372 | $body = $this->parseTree($n->body); |
| 373 | if ($body == '') |
| 374 | $body = ';'; |
| 375 | |
| 376 | $s .= $body; |
| 377 | break; |
| 378 | |
| 379 | case KEYWORD_DO: |
| 380 | $s = 'do{' . $this->parseTree($n->body, true) . '}while(' . $this->parseTree($n->condition) . ')'; |
| 381 | break; |
| 382 | |
| 383 | case KEYWORD_BREAK: |
| 384 | case KEYWORD_CONTINUE: |
| 385 | $s = $n->value . ($n->label ? ' ' . $n->label : ''); |
| 386 | break; |
| 387 | |
| 388 | case KEYWORD_TRY: |
| 389 | $s = 'try{' . $this->parseTree($n->tryBlock, true) . '}'; |
| 390 | $catchClauses = $n->catchClauses; |
| 391 | for ($i = 0, $j = count($catchClauses); $i < $j; $i++) |
| 392 | { |
| 393 | $t = $catchClauses[$i]; |
| 394 | $s .= 'catch(' . $t->varName . ($t->guard ? ' if ' . $this->parseTree($t->guard) : '') . '){' . $this->parseTree($t->block, true) . '}'; |
| 395 | } |
| 396 | if ($n->finallyBlock) |
| 397 | $s .= 'finally{' . $this->parseTree($n->finallyBlock, true) . '}'; |
| 398 | break; |
| 399 | |
| 400 | case KEYWORD_THROW: |
| 401 | case KEYWORD_RETURN: |
| 402 | $s = $n->type; |
| 403 | if ($n->value) |
| 404 | { |
| 405 | $t = $this->parseTree($n->value); |
| 406 | if (strlen($t)) |
| 407 | { |
| 408 | if ($this->isWordChar($t[0]) || $t[0] == '\\') |
| 409 | $s .= ' '; |
| 410 | |
| 411 | $s .= $t; |
| 412 | } |
| 413 | } |
| 414 | break; |
| 415 | |
| 416 | case KEYWORD_WITH: |
| 417 | $s = 'with(' . $this->parseTree($n->object) . ')' . $this->parseTree($n->body); |
| 418 | break; |
| 419 | |
| 420 | case KEYWORD_VAR: |
| 421 | case KEYWORD_CONST: |
| 422 | $s = $n->value . ' '; |
| 423 | $childs = $n->treeNodes; |
| 424 | for ($i = 0, $j = count($childs); $i < $j; $i++) |
| 425 | { |
| 426 | $t = $childs[$i]; |
| 427 | $s .= ($i ? ',' : '') . $t->name; |
| 428 | $u = $t->initializer; |
| 429 | if ($u) |
| 430 | $s .= '=' . $this->parseTree($u); |
| 431 | } |
| 432 | break; |
| 433 | |
| 434 | case KEYWORD_IN: |
| 435 | case KEYWORD_INSTANCEOF: |
| 436 | $left = $this->parseTree($n->treeNodes[0]); |
| 437 | $right = $this->parseTree($n->treeNodes[1]); |
| 438 | |
| 439 | $s = $left; |
| 440 | |
| 441 | if ($this->isWordChar(substr($left, -1))) |
| 442 | $s .= ' '; |
| 443 | |
| 444 | $s .= $n->type; |
| 445 | |
| 446 | if ($this->isWordChar($right[0]) || $right[0] == '\\') |
| 447 | $s .= ' '; |
| 448 | |
| 449 | $s .= $right; |
| 450 | break; |
| 451 | |
| 452 | case KEYWORD_DELETE: |
| 453 | case KEYWORD_TYPEOF: |
| 454 | $right = $this->parseTree($n->treeNodes[0]); |
| 455 | |
| 456 | $s = $n->type; |
| 457 | |
| 458 | if ($this->isWordChar($right[0]) || $right[0] == '\\') |
| 459 | $s .= ' '; |
| 460 | |
| 461 | $s .= $right; |
| 462 | break; |
| 463 | |
| 464 | case KEYWORD_VOID: |
| 465 | $s = 'void(' . $this->parseTree($n->treeNodes[0]) . ')'; |
| 466 | break; |
| 467 | |
| 468 | case KEYWORD_DEBUGGER: |
| 469 | throw new Exception('NOT IMPLEMENTED: DEBUGGER'); |
| 470 | break; |
| 471 | |
| 472 | case TOKEN_CONDCOMMENT_START: |
| 473 | case TOKEN_CONDCOMMENT_END: |
| 474 | $s = $n->value . ($n->type == TOKEN_CONDCOMMENT_START ? ' ' : ''); |
| 475 | $childs = $n->treeNodes; |
| 476 | for ($i = 0, $j = count($childs); $i < $j; $i++) |
| 477 | $s .= $this->parseTree($childs[$i]); |
| 478 | break; |
| 479 | |
| 480 | case OP_SEMICOLON: |
| 481 | if ($expression = $n->expression) |
| 482 | $s = $this->parseTree($expression); |
| 483 | break; |
| 484 | |
| 485 | case JS_LABEL: |
| 486 | $s = $n->label . ':' . $this->parseTree($n->statement); |
| 487 | break; |
| 488 | |
| 489 | case OP_COMMA: |
| 490 | $childs = $n->treeNodes; |
| 491 | for ($i = 0, $j = count($childs); $i < $j; $i++) |
| 492 | $s .= ($i ? ',' : '') . $this->parseTree($childs[$i]); |
| 493 | break; |
| 494 | |
| 495 | case OP_ASSIGN: |
| 496 | $s = $this->parseTree($n->treeNodes[0]) . $n->value . $this->parseTree($n->treeNodes[1]); |
| 497 | break; |
| 498 | |
| 499 | case OP_HOOK: |
| 500 | $s = $this->parseTree($n->treeNodes[0]) . '?' . $this->parseTree($n->treeNodes[1]) . ':' . $this->parseTree($n->treeNodes[2]); |
| 501 | break; |
| 502 | |
| 503 | case OP_OR: case OP_AND: |
| 504 | case OP_BITWISE_OR: case OP_BITWISE_XOR: case OP_BITWISE_AND: |
| 505 | case OP_EQ: case OP_NE: case OP_STRICT_EQ: case OP_STRICT_NE: |
| 506 | case OP_LT: case OP_LE: case OP_GE: case OP_GT: |
| 507 | case OP_LSH: case OP_RSH: case OP_URSH: |
| 508 | case OP_MUL: case OP_DIV: case OP_MOD: |
| 509 | $s = $this->parseTree($n->treeNodes[0]) . $n->type . $this->parseTree($n->treeNodes[1]); |
| 510 | break; |
| 511 | |
| 512 | case OP_PLUS: |
| 513 | case OP_MINUS: |
| 514 | $left = $this->parseTree($n->treeNodes[0]); |
| 515 | $right = $this->parseTree($n->treeNodes[1]); |
| 516 | |
| 517 | switch ($n->treeNodes[1]->type) |
| 518 | { |
| 519 | case OP_PLUS: |
| 520 | case OP_MINUS: |
| 521 | case OP_INCREMENT: |
| 522 | case OP_DECREMENT: |
| 523 | case OP_UNARY_PLUS: |
| 524 | case OP_UNARY_MINUS: |
| 525 | $s = $left . $n->type . ' ' . $right; |
| 526 | break; |
| 527 | |
| 528 | case TOKEN_STRING: |
| 529 | //combine concatted strings with same quotestyle |
| 530 | if ($n->type == OP_PLUS && substr($left, -1) == $right[0]) |
| 531 | { |
| 532 | $s = substr($left, 0, -1) . substr($right, 1); |
| 533 | break; |
| 534 | } |
| 535 | // FALL THROUGH |
| 536 | |
| 537 | default: |
| 538 | $s = $left . $n->type . $right; |
| 539 | } |
| 540 | break; |
| 541 | |
| 542 | case OP_NOT: |
| 543 | case OP_BITWISE_NOT: |
| 544 | case OP_UNARY_PLUS: |
| 545 | case OP_UNARY_MINUS: |
| 546 | $s = $n->value . $this->parseTree($n->treeNodes[0]); |
| 547 | break; |
| 548 | |
| 549 | case OP_INCREMENT: |
| 550 | case OP_DECREMENT: |
| 551 | if ($n->postfix) |
| 552 | $s = $this->parseTree($n->treeNodes[0]) . $n->value; |
| 553 | else |
| 554 | $s = $n->value . $this->parseTree($n->treeNodes[0]); |
| 555 | break; |
| 556 | |
| 557 | case OP_DOT: |
| 558 | $s = $this->parseTree($n->treeNodes[0]) . '.' . $this->parseTree($n->treeNodes[1]); |
| 559 | break; |
| 560 | |
| 561 | case JS_INDEX: |
| 562 | $s = $this->parseTree($n->treeNodes[0]); |
| 563 | // See if we can replace named index with a dot saving 3 bytes |
| 564 | if ( $n->treeNodes[0]->type == TOKEN_IDENTIFIER && |
| 565 | $n->treeNodes[1]->type == TOKEN_STRING && |
| 566 | $this->isValidIdentifier(substr($n->treeNodes[1]->value, 1, -1)) |
| 567 | ) |
| 568 | $s .= '.' . substr($n->treeNodes[1]->value, 1, -1); |
| 569 | else |
| 570 | $s .= '[' . $this->parseTree($n->treeNodes[1]) . ']'; |
| 571 | break; |
| 572 | |
| 573 | case JS_LIST: |
| 574 | $childs = $n->treeNodes; |
| 575 | for ($i = 0, $j = count($childs); $i < $j; $i++) |
| 576 | $s .= ($i ? ',' : '') . $this->parseTree($childs[$i]); |
| 577 | break; |
| 578 | |
| 579 | case JS_CALL: |
| 580 | $s = $this->parseTree($n->treeNodes[0]) . '(' . $this->parseTree($n->treeNodes[1]) . ')'; |
| 581 | break; |
| 582 | |
| 583 | case KEYWORD_NEW: |
| 584 | case JS_NEW_WITH_ARGS: |
| 585 | $s = 'new ' . $this->parseTree($n->treeNodes[0]) . '(' . ($n->type == JS_NEW_WITH_ARGS ? $this->parseTree($n->treeNodes[1]) : '') . ')'; |
| 586 | break; |
| 587 | |
| 588 | case JS_ARRAY_INIT: |
| 589 | $s = '['; |
| 590 | $childs = $n->treeNodes; |
| 591 | for ($i = 0, $j = count($childs); $i < $j; $i++) |
| 592 | { |
| 593 | $s .= ($i ? ',' : '') . $this->parseTree($childs[$i]); |
| 594 | } |
| 595 | $s .= ']'; |
| 596 | break; |
| 597 | |
| 598 | case JS_OBJECT_INIT: |
| 599 | $s = '{'; |
| 600 | $childs = $n->treeNodes; |
| 601 | for ($i = 0, $j = count($childs); $i < $j; $i++) |
| 602 | { |
| 603 | $t = $childs[$i]; |
| 604 | if ($i) |
| 605 | $s .= ','; |
| 606 | if ($t->type == JS_PROPERTY_INIT) |
| 607 | { |
| 608 | // Ditch the quotes when the index is a valid identifier |
| 609 | if ( $t->treeNodes[0]->type == TOKEN_STRING && |
| 610 | $this->isValidIdentifier(substr($t->treeNodes[0]->value, 1, -1)) |
| 611 | ) |
| 612 | $s .= substr($t->treeNodes[0]->value, 1, -1); |
| 613 | else |
| 614 | $s .= $t->treeNodes[0]->value; |
| 615 | |
| 616 | $s .= ':' . $this->parseTree($t->treeNodes[1]); |
| 617 | } |
| 618 | else |
| 619 | { |
| 620 | $s .= $t->type == JS_GETTER ? 'get' : 'set'; |
| 621 | $s .= ' ' . $t->name . '('; |
| 622 | $params = $t->params; |
| 623 | for ($i = 0, $j = count($params); $i < $j; $i++) |
| 624 | $s .= ($i ? ',' : '') . $params[$i]; |
| 625 | $s .= '){' . $this->parseTree($t->body, true) . '}'; |
| 626 | } |
| 627 | } |
| 628 | $s .= '}'; |
| 629 | break; |
| 630 | |
| 631 | case TOKEN_NUMBER: |
| 632 | $s = $n->value; |
| 633 | if (preg_match('/^([1-9]+)(0{3,})$/', $s, $m)) |
| 634 | $s = $m[1] . 'e' . strlen($m[2]); |
| 635 | break; |
| 636 | |
| 637 | case KEYWORD_NULL: case KEYWORD_THIS: case KEYWORD_TRUE: case KEYWORD_FALSE: |
| 638 | case TOKEN_IDENTIFIER: case TOKEN_STRING: case TOKEN_REGEXP: |
| 639 | $s = $n->value; |
| 640 | break; |
| 641 | |
| 642 | case JS_GROUP: |
| 643 | if (in_array( |
| 644 | $n->treeNodes[0]->type, |
| 645 | array( |
| 646 | JS_ARRAY_INIT, JS_OBJECT_INIT, JS_GROUP, |
| 647 | TOKEN_NUMBER, TOKEN_STRING, TOKEN_REGEXP, TOKEN_IDENTIFIER, |
| 648 | KEYWORD_NULL, KEYWORD_THIS, KEYWORD_TRUE, KEYWORD_FALSE |
| 649 | ) |
| 650 | )) |
| 651 | { |
| 652 | $s = $this->parseTree($n->treeNodes[0]); |
| 653 | } |
| 654 | else |
| 655 | { |
| 656 | $s = '(' . $this->parseTree($n->treeNodes[0]) . ')'; |
| 657 | } |
| 658 | break; |
| 659 | |
| 660 | default: |
| 661 | throw new Exception('UNKNOWN TOKEN TYPE: ' . $n->type); |
| 662 | } |
| 663 | |
| 664 | return $s; |
| 665 | } |
| 666 | |
| 667 | private function isValidIdentifier($string) |
| 668 | { |
| 669 | return preg_match('/^[a-zA-Z_][a-zA-Z0-9_]*$/', $string) && !in_array($string, $this->reserved); |
| 670 | } |
| 671 | |
| 672 | private function isWordChar($char) |
| 673 | { |
| 674 | return $char == '_' || $char == '$' || ctype_alnum($char); |
| 675 | } |
| 676 | } |
| 677 | |
| 678 | class JSParser |
| 679 | { |
| 680 | private $t; |
| 681 | private $minifier; |
| 682 | |
| 683 | private $opPrecedence = array( |
| 684 | ';' => 0, |
| 685 | ',' => 1, |
| 686 | '=' => 2, '?' => 2, ':' => 2, |
| 687 | // The above all have to have the same precedence, see bug 330975 |
| 688 | '||' => 4, |
| 689 | '&&' => 5, |
| 690 | '|' => 6, |
| 691 | '^' => 7, |
| 692 | '&' => 8, |
| 693 | '==' => 9, '!=' => 9, '===' => 9, '!==' => 9, |
| 694 | '<' => 10, '<=' => 10, '>=' => 10, '>' => 10, 'in' => 10, 'instanceof' => 10, |
| 695 | '<<' => 11, '>>' => 11, '>>>' => 11, |
| 696 | '+' => 12, '-' => 12, |
| 697 | '*' => 13, '/' => 13, '%' => 13, |
| 698 | 'delete' => 14, 'void' => 14, 'typeof' => 14, |
| 699 | '!' => 14, '~' => 14, 'U+' => 14, 'U-' => 14, |
| 700 | '++' => 15, '--' => 15, |
| 701 | 'new' => 16, |
| 702 | '.' => 17, |
| 703 | JS_NEW_WITH_ARGS => 0, JS_INDEX => 0, JS_CALL => 0, |
| 704 | JS_ARRAY_INIT => 0, JS_OBJECT_INIT => 0, JS_GROUP => 0 |
| 705 | ); |
| 706 | |
| 707 | private $opArity = array( |
| 708 | ',' => -2, |
| 709 | '=' => 2, |
| 710 | '?' => 3, |
| 711 | '||' => 2, |
| 712 | '&&' => 2, |
| 713 | '|' => 2, |
| 714 | '^' => 2, |
| 715 | '&' => 2, |
| 716 | '==' => 2, '!=' => 2, '===' => 2, '!==' => 2, |
| 717 | '<' => 2, '<=' => 2, '>=' => 2, '>' => 2, 'in' => 2, 'instanceof' => 2, |
| 718 | '<<' => 2, '>>' => 2, '>>>' => 2, |
| 719 | '+' => 2, '-' => 2, |
| 720 | '*' => 2, '/' => 2, '%' => 2, |
| 721 | 'delete' => 1, 'void' => 1, 'typeof' => 1, |
| 722 | '!' => 1, '~' => 1, 'U+' => 1, 'U-' => 1, |
| 723 | '++' => 1, '--' => 1, |
| 724 | 'new' => 1, |
| 725 | '.' => 2, |
| 726 | JS_NEW_WITH_ARGS => 2, JS_INDEX => 2, JS_CALL => 2, |
| 727 | JS_ARRAY_INIT => 1, JS_OBJECT_INIT => 1, JS_GROUP => 1, |
| 728 | TOKEN_CONDCOMMENT_START => 1, TOKEN_CONDCOMMENT_END => 1 |
| 729 | ); |
| 730 | |
| 731 | public function __construct($minifier=null) |
| 732 | { |
| 733 | $this->minifier = $minifier; |
| 734 | $this->t = new JSTokenizer(); |
| 735 | } |
| 736 | |
| 737 | public function parse($s, $f, $l) |
| 738 | { |
| 739 | // initialize tokenizer |
| 740 | $this->t->init($s, $f, $l); |
| 741 | |
| 742 | $x = new JSCompilerContext(false); |
| 743 | $n = $this->Script($x); |
| 744 | if (!$this->t->isDone()) |
| 745 | throw $this->t->newSyntaxError('Syntax error'); |
| 746 | |
| 747 | return $n; |
| 748 | } |
| 749 | |
| 750 | private function Script($x) |
| 751 | { |
| 752 | $n = $this->Statements($x); |
| 753 | $n->type = JS_SCRIPT; |
| 754 | $n->funDecls = $x->funDecls; |
| 755 | $n->varDecls = $x->varDecls; |
| 756 | |
| 757 | // minify by scope |
| 758 | if ($this->minifier) |
| 759 | { |
| 760 | $n->value = $this->minifier->parseTree($n); |
| 761 | |
| 762 | // clear tree from node to save memory |
| 763 | $n->treeNodes = null; |
| 764 | $n->funDecls = null; |
| 765 | $n->varDecls = null; |
| 766 | |
| 767 | $n->type = JS_MINIFIED; |
| 768 | } |
| 769 | |
| 770 | return $n; |
| 771 | } |
| 772 | |
| 773 | private function Statements($x) |
| 774 | { |
| 775 | $n = new JSNode($this->t, JS_BLOCK); |
| 776 | array_push($x->stmtStack, $n); |
| 777 | |
| 778 | while (!$this->t->isDone() && $this->t->peek() != OP_RIGHT_CURLY) |
| 779 | $n->addNode($this->Statement($x)); |
| 780 | |
| 781 | array_pop($x->stmtStack); |
| 782 | |
| 783 | return $n; |
| 784 | } |
| 785 | |
| 786 | private function Block($x) |
| 787 | { |
| 788 | $this->t->mustMatch(OP_LEFT_CURLY); |
| 789 | $n = $this->Statements($x); |
| 790 | $this->t->mustMatch(OP_RIGHT_CURLY); |
| 791 | |
| 792 | return $n; |
| 793 | } |
| 794 | |
| 795 | private function Statement($x) |
| 796 | { |
| 797 | $tt = $this->t->get(); |
| 798 | $n2 = null; |
| 799 | |
| 800 | // Cases for statements ending in a right curly return early, avoiding the |
| 801 | // common semicolon insertion magic after this switch. |
| 802 | switch ($tt) |
| 803 | { |
| 804 | case KEYWORD_FUNCTION: |
| 805 | return $this->FunctionDefinition( |
| 806 | $x, |
| 807 | true, |
| 808 | count($x->stmtStack) > 1 ? STATEMENT_FORM : DECLARED_FORM |
| 809 | ); |
| 810 | break; |
| 811 | |
| 812 | case OP_LEFT_CURLY: |
| 813 | $n = $this->Statements($x); |
| 814 | $this->t->mustMatch(OP_RIGHT_CURLY); |
| 815 | return $n; |
| 816 | |
| 817 | case KEYWORD_IF: |
| 818 | $n = new JSNode($this->t); |
| 819 | $n->condition = $this->ParenExpression($x); |
| 820 | array_push($x->stmtStack, $n); |
| 821 | $n->thenPart = $this->Statement($x); |
| 822 | $n->elsePart = $this->t->match(KEYWORD_ELSE) ? $this->Statement($x) : null; |
| 823 | array_pop($x->stmtStack); |
| 824 | return $n; |
| 825 | |
| 826 | case KEYWORD_SWITCH: |
| 827 | $n = new JSNode($this->t); |
| 828 | $this->t->mustMatch(OP_LEFT_PAREN); |
| 829 | $n->discriminant = $this->Expression($x); |
| 830 | $this->t->mustMatch(OP_RIGHT_PAREN); |
| 831 | $n->cases = array(); |
| 832 | $n->defaultIndex = -1; |
| 833 | |
| 834 | array_push($x->stmtStack, $n); |
| 835 | |
| 836 | $this->t->mustMatch(OP_LEFT_CURLY); |
| 837 | |
| 838 | while (($tt = $this->t->get()) != OP_RIGHT_CURLY) |
| 839 | { |
| 840 | switch ($tt) |
| 841 | { |
| 842 | case KEYWORD_DEFAULT: |
| 843 | if ($n->defaultIndex >= 0) |
| 844 | throw $this->t->newSyntaxError('More than one switch default'); |
| 845 | // FALL THROUGH |
| 846 | case KEYWORD_CASE: |
| 847 | $n2 = new JSNode($this->t); |
| 848 | if ($tt == KEYWORD_DEFAULT) |
| 849 | $n->defaultIndex = count($n->cases); |
| 850 | else |
| 851 | $n2->caseLabel = $this->Expression($x, OP_COLON); |
| 852 | break; |
| 853 | default: |
| 854 | throw $this->t->newSyntaxError('Invalid switch case'); |
| 855 | } |
| 856 | |
| 857 | $this->t->mustMatch(OP_COLON); |
| 858 | $n2->statements = new JSNode($this->t, JS_BLOCK); |
| 859 | while (($tt = $this->t->peek()) != KEYWORD_CASE && $tt != KEYWORD_DEFAULT && $tt != OP_RIGHT_CURLY) |
| 860 | $n2->statements->addNode($this->Statement($x)); |
| 861 | |
| 862 | array_push($n->cases, $n2); |
| 863 | } |
| 864 | |
| 865 | array_pop($x->stmtStack); |
| 866 | return $n; |
| 867 | |
| 868 | case KEYWORD_FOR: |
| 869 | $n = new JSNode($this->t); |
| 870 | $n->isLoop = true; |
| 871 | $this->t->mustMatch(OP_LEFT_PAREN); |
| 872 | |
| 873 | if (($tt = $this->t->peek()) != OP_SEMICOLON) |
| 874 | { |
| 875 | $x->inForLoopInit = true; |
| 876 | if ($tt == KEYWORD_VAR || $tt == KEYWORD_CONST) |
| 877 | { |
| 878 | $this->t->get(); |
| 879 | $n2 = $this->Variables($x); |
| 880 | } |
| 881 | else |
| 882 | { |
| 883 | $n2 = $this->Expression($x); |
| 884 | } |
| 885 | $x->inForLoopInit = false; |
| 886 | } |
| 887 | |
| 888 | if ($n2 && $this->t->match(KEYWORD_IN)) |
| 889 | { |
| 890 | $n->type = JS_FOR_IN; |
| 891 | if ($n2->type == KEYWORD_VAR) |
| 892 | { |
| 893 | if (count($n2->treeNodes) != 1) |
| 894 | { |
| 895 | throw $this->t->newSyntaxError( |
| 896 | 'Invalid for..in left-hand side', |
| 897 | $this->t->filename, |
| 898 | $n2->lineno |
| 899 | ); |
| 900 | } |
| 901 | |
| 902 | // NB: n2[0].type == IDENTIFIER and n2[0].value == n2[0].name. |
| 903 | $n->iterator = $n2->treeNodes[0]; |
| 904 | $n->varDecl = $n2; |
| 905 | } |
| 906 | else |
| 907 | { |
| 908 | $n->iterator = $n2; |
| 909 | $n->varDecl = null; |
| 910 | } |
| 911 | |
| 912 | $n->object = $this->Expression($x); |
| 913 | } |
| 914 | else |
| 915 | { |
| 916 | $n->setup = $n2 ? $n2 : null; |
| 917 | $this->t->mustMatch(OP_SEMICOLON); |
| 918 | $n->condition = $this->t->peek() == OP_SEMICOLON ? null : $this->Expression($x); |
| 919 | $this->t->mustMatch(OP_SEMICOLON); |
| 920 | $n->update = $this->t->peek() == OP_RIGHT_PAREN ? null : $this->Expression($x); |
| 921 | } |
| 922 | |
| 923 | $this->t->mustMatch(OP_RIGHT_PAREN); |
| 924 | $n->body = $this->nest($x, $n); |
| 925 | return $n; |
| 926 | |
| 927 | case KEYWORD_WHILE: |
| 928 | $n = new JSNode($this->t); |
| 929 | $n->isLoop = true; |
| 930 | $n->condition = $this->ParenExpression($x); |
| 931 | $n->body = $this->nest($x, $n); |
| 932 | return $n; |
| 933 | |
| 934 | case KEYWORD_DO: |
| 935 | $n = new JSNode($this->t); |
| 936 | $n->isLoop = true; |
| 937 | $n->body = $this->nest($x, $n, KEYWORD_WHILE); |
| 938 | $n->condition = $this->ParenExpression($x); |
| 939 | if (!$x->ecmaStrictMode) |
| 940 | { |
| 941 | // <script language="JavaScript"> (without version hints) may need |
| 942 | // automatic semicolon insertion without a newline after do-while. |
| 943 | // See http://bugzilla.mozilla.org/show_bug.cgi?id=238945. |
| 944 | $this->t->match(OP_SEMICOLON); |
| 945 | return $n; |
| 946 | } |
| 947 | break; |
| 948 | |
| 949 | case KEYWORD_BREAK: |
| 950 | case KEYWORD_CONTINUE: |
| 951 | $n = new JSNode($this->t); |
| 952 | |
| 953 | if ($this->t->peekOnSameLine() == TOKEN_IDENTIFIER) |
| 954 | { |
| 955 | $this->t->get(); |
| 956 | $n->label = $this->t->currentToken()->value; |
| 957 | } |
| 958 | |
| 959 | $ss = $x->stmtStack; |
| 960 | $i = count($ss); |
| 961 | $label = $n->label; |
| 962 | if ($label) |
| 963 | { |
| 964 | do |
| 965 | { |
| 966 | if (--$i < 0) |
| 967 | throw $this->t->newSyntaxError('Label not found'); |
| 968 | } |
| 969 | while ($ss[$i]->label != $label); |
| 970 | } |
| 971 | else |
| 972 | { |
| 973 | do |
| 974 | { |
| 975 | if (--$i < 0) |
| 976 | throw $this->t->newSyntaxError('Invalid ' . $tt); |
| 977 | } |
| 978 | while (!$ss[$i]->isLoop && ($tt != KEYWORD_BREAK || $ss[$i]->type != KEYWORD_SWITCH)); |
| 979 | } |
| 980 | |
| 981 | $n->target = $ss[$i]; |
| 982 | break; |
| 983 | |
| 984 | case KEYWORD_TRY: |
| 985 | $n = new JSNode($this->t); |
| 986 | $n->tryBlock = $this->Block($x); |
| 987 | $n->catchClauses = array(); |
| 988 | |
| 989 | while ($this->t->match(KEYWORD_CATCH)) |
| 990 | { |
| 991 | $n2 = new JSNode($this->t); |
| 992 | $this->t->mustMatch(OP_LEFT_PAREN); |
| 993 | $n2->varName = $this->t->mustMatch(TOKEN_IDENTIFIER)->value; |
| 994 | |
| 995 | if ($this->t->match(KEYWORD_IF)) |
| 996 | { |
| 997 | if ($x->ecmaStrictMode) |
| 998 | throw $this->t->newSyntaxError('Illegal catch guard'); |
| 999 | |
| 1000 | if (count($n->catchClauses) && !end($n->catchClauses)->guard) |
| 1001 | throw $this->t->newSyntaxError('Guarded catch after unguarded'); |
| 1002 | |
| 1003 | $n2->guard = $this->Expression($x); |
| 1004 | } |
| 1005 | else |
| 1006 | { |
| 1007 | $n2->guard = null; |
| 1008 | } |
| 1009 | |
| 1010 | $this->t->mustMatch(OP_RIGHT_PAREN); |
| 1011 | $n2->block = $this->Block($x); |
| 1012 | array_push($n->catchClauses, $n2); |
| 1013 | } |
| 1014 | |
| 1015 | if ($this->t->match(KEYWORD_FINALLY)) |
| 1016 | $n->finallyBlock = $this->Block($x); |
| 1017 | |
| 1018 | if (!count($n->catchClauses) && !$n->finallyBlock) |
| 1019 | throw $this->t->newSyntaxError('Invalid try statement'); |
| 1020 | return $n; |
| 1021 | |
| 1022 | case KEYWORD_CATCH: |
| 1023 | case KEYWORD_FINALLY: |
| 1024 | throw $this->t->newSyntaxError($tt + ' without preceding try'); |
| 1025 | |
| 1026 | case KEYWORD_THROW: |
| 1027 | $n = new JSNode($this->t); |
| 1028 | $n->value = $this->Expression($x); |
| 1029 | break; |
| 1030 | |
| 1031 | case KEYWORD_RETURN: |
| 1032 | if (!$x->inFunction) |
| 1033 | throw $this->t->newSyntaxError('Invalid return'); |
| 1034 | |
| 1035 | $n = new JSNode($this->t); |
| 1036 | $tt = $this->t->peekOnSameLine(); |
| 1037 | if ($tt != TOKEN_END && $tt != TOKEN_NEWLINE && $tt != OP_SEMICOLON && $tt != OP_RIGHT_CURLY) |
| 1038 | $n->value = $this->Expression($x); |
| 1039 | else |
| 1040 | $n->value = null; |
| 1041 | break; |
| 1042 | |
| 1043 | case KEYWORD_WITH: |
| 1044 | $n = new JSNode($this->t); |
| 1045 | $n->object = $this->ParenExpression($x); |
| 1046 | $n->body = $this->nest($x, $n); |
| 1047 | return $n; |
| 1048 | |
| 1049 | case KEYWORD_VAR: |
| 1050 | case KEYWORD_CONST: |
| 1051 | $n = $this->Variables($x); |
| 1052 | break; |
| 1053 | |
| 1054 | case TOKEN_CONDCOMMENT_START: |
| 1055 | case TOKEN_CONDCOMMENT_END: |
| 1056 | $n = new JSNode($this->t); |
| 1057 | return $n; |
| 1058 | |
| 1059 | case KEYWORD_DEBUGGER: |
| 1060 | $n = new JSNode($this->t); |
| 1061 | break; |
| 1062 | |
| 1063 | case TOKEN_NEWLINE: |
| 1064 | case OP_SEMICOLON: |
| 1065 | $n = new JSNode($this->t, OP_SEMICOLON); |
| 1066 | $n->expression = null; |
| 1067 | return $n; |
| 1068 | |
| 1069 | default: |
| 1070 | if ($tt == TOKEN_IDENTIFIER) |
| 1071 | { |
| 1072 | $this->t->scanOperand = false; |
| 1073 | $tt = $this->t->peek(); |
| 1074 | $this->t->scanOperand = true; |
| 1075 | if ($tt == OP_COLON) |
| 1076 | { |
| 1077 | $label = $this->t->currentToken()->value; |
| 1078 | $ss = $x->stmtStack; |
| 1079 | for ($i = count($ss) - 1; $i >= 0; --$i) |
| 1080 | { |
| 1081 | if ($ss[$i]->label == $label) |
| 1082 | throw $this->t->newSyntaxError('Duplicate label'); |
| 1083 | } |
| 1084 | |
| 1085 | $this->t->get(); |
| 1086 | $n = new JSNode($this->t, JS_LABEL); |
| 1087 | $n->label = $label; |
| 1088 | $n->statement = $this->nest($x, $n); |
| 1089 | |
| 1090 | return $n; |
| 1091 | } |
| 1092 | } |
| 1093 | |
| 1094 | $n = new JSNode($this->t, OP_SEMICOLON); |
| 1095 | $this->t->unget(); |
| 1096 | $n->expression = $this->Expression($x); |
| 1097 | $n->end = $n->expression->end; |
| 1098 | break; |
| 1099 | } |
| 1100 | |
| 1101 | if ($this->t->lineno == $this->t->currentToken()->lineno) |
| 1102 | { |
| 1103 | $tt = $this->t->peekOnSameLine(); |
| 1104 | if ($tt != TOKEN_END && $tt != TOKEN_NEWLINE && $tt != OP_SEMICOLON && $tt != OP_RIGHT_CURLY) |
| 1105 | throw $this->t->newSyntaxError('Missing ; before statement'); |
| 1106 | } |
| 1107 | |
| 1108 | $this->t->match(OP_SEMICOLON); |
| 1109 | |
| 1110 | return $n; |
| 1111 | } |
| 1112 | |
| 1113 | private function FunctionDefinition($x, $requireName, $functionForm) |
| 1114 | { |
| 1115 | $f = new JSNode($this->t); |
| 1116 | |
| 1117 | if ($f->type != KEYWORD_FUNCTION) |
| 1118 | $f->type = ($f->value == 'get') ? JS_GETTER : JS_SETTER; |
| 1119 | |
| 1120 | if ($this->t->match(TOKEN_IDENTIFIER)) |
| 1121 | $f->name = $this->t->currentToken()->value; |
| 1122 | elseif ($requireName) |
| 1123 | throw $this->t->newSyntaxError('Missing function identifier'); |
| 1124 | |
| 1125 | $this->t->mustMatch(OP_LEFT_PAREN); |
| 1126 | $f->params = array(); |
| 1127 | |
| 1128 | while (($tt = $this->t->get()) != OP_RIGHT_PAREN) |
| 1129 | { |
| 1130 | if ($tt != TOKEN_IDENTIFIER) |
| 1131 | throw $this->t->newSyntaxError('Missing formal parameter'); |
| 1132 | |
| 1133 | array_push($f->params, $this->t->currentToken()->value); |
| 1134 | |
| 1135 | if ($this->t->peek() != OP_RIGHT_PAREN) |
| 1136 | $this->t->mustMatch(OP_COMMA); |
| 1137 | } |
| 1138 | |
| 1139 | $this->t->mustMatch(OP_LEFT_CURLY); |
| 1140 | |
| 1141 | $x2 = new JSCompilerContext(true); |
| 1142 | $f->body = $this->Script($x2); |
| 1143 | |
| 1144 | $this->t->mustMatch(OP_RIGHT_CURLY); |
| 1145 | $f->end = $this->t->currentToken()->end; |
| 1146 | |
| 1147 | $f->functionForm = $functionForm; |
| 1148 | if ($functionForm == DECLARED_FORM) |
| 1149 | array_push($x->funDecls, $f); |
| 1150 | |
| 1151 | return $f; |
| 1152 | } |
| 1153 | |
| 1154 | private function Variables($x) |
| 1155 | { |
| 1156 | $n = new JSNode($this->t); |
| 1157 | |
| 1158 | do |
| 1159 | { |
| 1160 | $this->t->mustMatch(TOKEN_IDENTIFIER); |
| 1161 | |
| 1162 | $n2 = new JSNode($this->t); |
| 1163 | $n2->name = $n2->value; |
| 1164 | |
| 1165 | if ($this->t->match(OP_ASSIGN)) |
| 1166 | { |
| 1167 | if ($this->t->currentToken()->assignOp) |
| 1168 | throw $this->t->newSyntaxError('Invalid variable initialization'); |
| 1169 | |
| 1170 | $n2->initializer = $this->Expression($x, OP_COMMA); |
| 1171 | } |
| 1172 | |
| 1173 | $n2->readOnly = $n->type == KEYWORD_CONST; |
| 1174 | |
| 1175 | $n->addNode($n2); |
| 1176 | array_push($x->varDecls, $n2); |
| 1177 | } |
| 1178 | while ($this->t->match(OP_COMMA)); |
| 1179 | |
| 1180 | return $n; |
| 1181 | } |
| 1182 | |
| 1183 | private function Expression($x, $stop=false) |
| 1184 | { |
| 1185 | $operators = array(); |
| 1186 | $operands = array(); |
| 1187 | $n = false; |
| 1188 | |
| 1189 | $bl = $x->bracketLevel; |
| 1190 | $cl = $x->curlyLevel; |
| 1191 | $pl = $x->parenLevel; |
| 1192 | $hl = $x->hookLevel; |
| 1193 | |
| 1194 | while (($tt = $this->t->get()) != TOKEN_END) |
| 1195 | { |
| 1196 | if ($tt == $stop && |
| 1197 | $x->bracketLevel == $bl && |
| 1198 | $x->curlyLevel == $cl && |
| 1199 | $x->parenLevel == $pl && |
| 1200 | $x->hookLevel == $hl |
| 1201 | ) |
| 1202 | { |
| 1203 | // Stop only if tt matches the optional stop parameter, and that |
| 1204 | // token is not quoted by some kind of bracket. |
| 1205 | break; |
| 1206 | } |
| 1207 | |
| 1208 | switch ($tt) |
| 1209 | { |
| 1210 | case OP_SEMICOLON: |
| 1211 | // NB: cannot be empty, Statement handled that. |
| 1212 | break 2; |
| 1213 | |
| 1214 | case OP_HOOK: |
| 1215 | if ($this->t->scanOperand) |
| 1216 | break 2; |
| 1217 | |
| 1218 | while ( !empty($operators) && |
| 1219 | $this->opPrecedence[end($operators)->type] > $this->opPrecedence[$tt] |
| 1220 | ) |
| 1221 | $this->reduce($operators, $operands); |
| 1222 | |
| 1223 | array_push($operators, new JSNode($this->t)); |
| 1224 | |
| 1225 | ++$x->hookLevel; |
| 1226 | $this->t->scanOperand = true; |
| 1227 | $n = $this->Expression($x); |
| 1228 | |
| 1229 | if (!$this->t->match(OP_COLON)) |
| 1230 | break 2; |
| 1231 | |
| 1232 | --$x->hookLevel; |
| 1233 | array_push($operands, $n); |
| 1234 | break; |
| 1235 | |
| 1236 | case OP_COLON: |
| 1237 | if ($x->hookLevel) |
| 1238 | break 2; |
| 1239 | |
| 1240 | throw $this->t->newSyntaxError('Invalid label'); |
| 1241 | break; |
| 1242 | |
| 1243 | case OP_ASSIGN: |
| 1244 | if ($this->t->scanOperand) |
| 1245 | break 2; |
| 1246 | |
| 1247 | // Use >, not >=, for right-associative ASSIGN |
| 1248 | while ( !empty($operators) && |
| 1249 | $this->opPrecedence[end($operators)->type] > $this->opPrecedence[$tt] |
| 1250 | ) |
| 1251 | $this->reduce($operators, $operands); |
| 1252 | |
| 1253 | array_push($operators, new JSNode($this->t)); |
| 1254 | end($operands)->assignOp = $this->t->currentToken()->assignOp; |
| 1255 | $this->t->scanOperand = true; |
| 1256 | break; |
| 1257 | |
| 1258 | case KEYWORD_IN: |
| 1259 | // An in operator should not be parsed if we're parsing the head of |
| 1260 | // a for (...) loop, unless it is in the then part of a conditional |
| 1261 | // expression, or parenthesized somehow. |
| 1262 | if ($x->inForLoopInit && !$x->hookLevel && |
| 1263 | !$x->bracketLevel && !$x->curlyLevel && |
| 1264 | !$x->parenLevel |
| 1265 | ) |
| 1266 | break 2; |
| 1267 | // FALL THROUGH |
| 1268 | case OP_COMMA: |
| 1269 | // A comma operator should not be parsed if we're parsing the then part |
| 1270 | // of a conditional expression unless it's parenthesized somehow. |
| 1271 | if ($tt == OP_COMMA && $x->hookLevel && |
| 1272 | !$x->bracketLevel && !$x->curlyLevel && |
| 1273 | !$x->parenLevel |
| 1274 | ) |
| 1275 | break 2; |
| 1276 | // Treat comma as left-associative so reduce can fold left-heavy |
| 1277 | // COMMA trees into a single array. |
| 1278 | // FALL THROUGH |
| 1279 | case OP_OR: |
| 1280 | case OP_AND: |
| 1281 | case OP_BITWISE_OR: |
| 1282 | case OP_BITWISE_XOR: |
| 1283 | case OP_BITWISE_AND: |
| 1284 | case OP_EQ: case OP_NE: case OP_STRICT_EQ: case OP_STRICT_NE: |
| 1285 | case OP_LT: case OP_LE: case OP_GE: case OP_GT: |
| 1286 | case KEYWORD_INSTANCEOF: |
| 1287 | case OP_LSH: case OP_RSH: case OP_URSH: |
| 1288 | case OP_PLUS: case OP_MINUS: |
| 1289 | case OP_MUL: case OP_DIV: case OP_MOD: |
| 1290 | case OP_DOT: |
| 1291 | if ($this->t->scanOperand) |
| 1292 | break 2; |
| 1293 | |
| 1294 | while ( !empty($operators) && |
| 1295 | $this->opPrecedence[end($operators)->type] >= $this->opPrecedence[$tt] |
| 1296 | ) |
| 1297 | $this->reduce($operators, $operands); |
| 1298 | |
| 1299 | if ($tt == OP_DOT) |
| 1300 | { |
| 1301 | $this->t->mustMatch(TOKEN_IDENTIFIER); |
| 1302 | array_push($operands, new JSNode($this->t, OP_DOT, array_pop($operands), new JSNode($this->t))); |
| 1303 | } |
| 1304 | else |
| 1305 | { |
| 1306 | array_push($operators, new JSNode($this->t)); |
| 1307 | $this->t->scanOperand = true; |
| 1308 | } |
| 1309 | break; |
| 1310 | |
| 1311 | case KEYWORD_DELETE: case KEYWORD_VOID: case KEYWORD_TYPEOF: |
| 1312 | case OP_NOT: case OP_BITWISE_NOT: case OP_UNARY_PLUS: case OP_UNARY_MINUS: |
| 1313 | case KEYWORD_NEW: |
| 1314 | if (!$this->t->scanOperand) |
| 1315 | break 2; |
| 1316 | |
| 1317 | array_push($operators, new JSNode($this->t)); |
| 1318 | break; |
| 1319 | |
| 1320 | case OP_INCREMENT: case OP_DECREMENT: |
| 1321 | if ($this->t->scanOperand) |
| 1322 | { |
| 1323 | array_push($operators, new JSNode($this->t)); // prefix increment or decrement |
| 1324 | } |
| 1325 | else |
| 1326 | { |
| 1327 | // Don't cross a line boundary for postfix {in,de}crement. |
| 1328 | $t = $this->t->tokens[($this->t->tokenIndex + $this->t->lookahead - 1) & 3]; |
| 1329 | if ($t && $t->lineno != $this->t->lineno) |
| 1330 | break 2; |
| 1331 | |
| 1332 | if (!empty($operators)) |
| 1333 | { |
| 1334 | // Use >, not >=, so postfix has higher precedence than prefix. |
| 1335 | while ($this->opPrecedence[end($operators)->type] > $this->opPrecedence[$tt]) |
| 1336 | $this->reduce($operators, $operands); |
| 1337 | } |
| 1338 | |
| 1339 | $n = new JSNode($this->t, $tt, array_pop($operands)); |
| 1340 | $n->postfix = true; |
| 1341 | array_push($operands, $n); |
| 1342 | } |
| 1343 | break; |
| 1344 | |
| 1345 | case KEYWORD_FUNCTION: |
| 1346 | if (!$this->t->scanOperand) |
| 1347 | break 2; |
| 1348 | |
| 1349 | array_push($operands, $this->FunctionDefinition($x, false, EXPRESSED_FORM)); |
| 1350 | $this->t->scanOperand = false; |
| 1351 | break; |
| 1352 | |
| 1353 | case KEYWORD_NULL: case KEYWORD_THIS: case KEYWORD_TRUE: case KEYWORD_FALSE: |
| 1354 | case TOKEN_IDENTIFIER: case TOKEN_NUMBER: case TOKEN_STRING: case TOKEN_REGEXP: |
| 1355 | if (!$this->t->scanOperand) |
| 1356 | break 2; |
| 1357 | |
| 1358 | array_push($operands, new JSNode($this->t)); |
| 1359 | $this->t->scanOperand = false; |
| 1360 | break; |
| 1361 | |
| 1362 | case TOKEN_CONDCOMMENT_START: |
| 1363 | case TOKEN_CONDCOMMENT_END: |
| 1364 | if ($this->t->scanOperand) |
| 1365 | array_push($operators, new JSNode($this->t)); |
| 1366 | else |
| 1367 | array_push($operands, new JSNode($this->t)); |
| 1368 | break; |
| 1369 | |
| 1370 | case OP_LEFT_BRACKET: |
| 1371 | if ($this->t->scanOperand) |
| 1372 | { |
| 1373 | // Array initialiser. Parse using recursive descent, as the |
| 1374 | // sub-grammar here is not an operator grammar. |
| 1375 | $n = new JSNode($this->t, JS_ARRAY_INIT); |
| 1376 | while (($tt = $this->t->peek()) != OP_RIGHT_BRACKET) |
| 1377 | { |
| 1378 | if ($tt == OP_COMMA) |
| 1379 | { |
| 1380 | $this->t->get(); |
| 1381 | $n->addNode(null); |
| 1382 | continue; |
| 1383 | } |
| 1384 | |
| 1385 | $n->addNode($this->Expression($x, OP_COMMA)); |
| 1386 | if (!$this->t->match(OP_COMMA)) |
| 1387 | break; |
| 1388 | } |
| 1389 | |
| 1390 | $this->t->mustMatch(OP_RIGHT_BRACKET); |
| 1391 | array_push($operands, $n); |
| 1392 | $this->t->scanOperand = false; |
| 1393 | } |
| 1394 | else |
| 1395 | { |
| 1396 | // Property indexing operator. |
| 1397 | array_push($operators, new JSNode($this->t, JS_INDEX)); |
| 1398 | $this->t->scanOperand = true; |
| 1399 | ++$x->bracketLevel; |
| 1400 | } |
| 1401 | break; |
| 1402 | |
| 1403 | case OP_RIGHT_BRACKET: |
| 1404 | if ($this->t->scanOperand || $x->bracketLevel == $bl) |
| 1405 | break 2; |
| 1406 | |
| 1407 | while ($this->reduce($operators, $operands)->type != JS_INDEX) |
| 1408 | continue; |
| 1409 | |
| 1410 | --$x->bracketLevel; |
| 1411 | break; |
| 1412 | |
| 1413 | case OP_LEFT_CURLY: |
| 1414 | if (!$this->t->scanOperand) |
| 1415 | break 2; |
| 1416 | |
| 1417 | // Object initialiser. As for array initialisers (see above), |
| 1418 | // parse using recursive descent. |
| 1419 | ++$x->curlyLevel; |
| 1420 | $n = new JSNode($this->t, JS_OBJECT_INIT); |
| 1421 | while (!$this->t->match(OP_RIGHT_CURLY)) |
| 1422 | { |
| 1423 | do |
| 1424 | { |
| 1425 | $tt = $this->t->get(); |
| 1426 | $tv = $this->t->currentToken()->value; |
| 1427 | if (($tv == 'get' || $tv == 'set') && $this->t->peek() == TOKEN_IDENTIFIER) |
| 1428 | { |
| 1429 | if ($x->ecmaStrictMode) |
| 1430 | throw $this->t->newSyntaxError('Illegal property accessor'); |
| 1431 | |
| 1432 | $n->addNode($this->FunctionDefinition($x, true, EXPRESSED_FORM)); |
| 1433 | } |
| 1434 | else |
| 1435 | { |
| 1436 | switch ($tt) |
| 1437 | { |
| 1438 | case TOKEN_IDENTIFIER: |
| 1439 | case TOKEN_NUMBER: |
| 1440 | case TOKEN_STRING: |
| 1441 | $id = new JSNode($this->t); |
| 1442 | break; |
| 1443 | |
| 1444 | case OP_RIGHT_CURLY: |
| 1445 | if ($x->ecmaStrictMode) |
| 1446 | throw $this->t->newSyntaxError('Illegal trailing ,'); |
| 1447 | break 3; |
| 1448 | |
| 1449 | default: |
| 1450 | throw $this->t->newSyntaxError('Invalid property name'); |
| 1451 | } |
| 1452 | |
| 1453 | $this->t->mustMatch(OP_COLON); |
| 1454 | $n->addNode(new JSNode($this->t, JS_PROPERTY_INIT, $id, $this->Expression($x, OP_COMMA))); |
| 1455 | } |
| 1456 | } |
| 1457 | while ($this->t->match(OP_COMMA)); |
| 1458 | |
| 1459 | $this->t->mustMatch(OP_RIGHT_CURLY); |
| 1460 | break; |
| 1461 | } |
| 1462 | |
| 1463 | array_push($operands, $n); |
| 1464 | $this->t->scanOperand = false; |
| 1465 | --$x->curlyLevel; |
| 1466 | break; |
| 1467 | |
| 1468 | case OP_RIGHT_CURLY: |
| 1469 | if (!$this->t->scanOperand && $x->curlyLevel != $cl) |
| 1470 | throw new Exception('PANIC: right curly botch'); |
| 1471 | break 2; |
| 1472 | |
| 1473 | case OP_LEFT_PAREN: |
| 1474 | if ($this->t->scanOperand) |
| 1475 | { |
| 1476 | array_push($operators, new JSNode($this->t, JS_GROUP)); |
| 1477 | } |
| 1478 | else |
| 1479 | { |
| 1480 | while ( !empty($operators) && |
| 1481 | $this->opPrecedence[end($operators)->type] > $this->opPrecedence[KEYWORD_NEW] |
| 1482 | ) |
| 1483 | $this->reduce($operators, $operands); |
| 1484 | |
| 1485 | // Handle () now, to regularize the n-ary case for n > 0. |
| 1486 | // We must set scanOperand in case there are arguments and |
| 1487 | // the first one is a regexp or unary+/-. |
| 1488 | $n = end($operators); |
| 1489 | $this->t->scanOperand = true; |
| 1490 | if ($this->t->match(OP_RIGHT_PAREN)) |
| 1491 | { |
| 1492 | if ($n && $n->type == KEYWORD_NEW) |
| 1493 | { |
| 1494 | array_pop($operators); |
| 1495 | $n->addNode(array_pop($operands)); |
| 1496 | } |
| 1497 | else |
| 1498 | { |
| 1499 | $n = new JSNode($this->t, JS_CALL, array_pop($operands), new JSNode($this->t, JS_LIST)); |
| 1500 | } |
| 1501 | |
| 1502 | array_push($operands, $n); |
| 1503 | $this->t->scanOperand = false; |
| 1504 | break; |
| 1505 | } |
| 1506 | |
| 1507 | if ($n && $n->type == KEYWORD_NEW) |
| 1508 | $n->type = JS_NEW_WITH_ARGS; |
| 1509 | else |
| 1510 | array_push($operators, new JSNode($this->t, JS_CALL)); |
| 1511 | } |
| 1512 | |
| 1513 | ++$x->parenLevel; |
| 1514 | break; |
| 1515 | |
| 1516 | case OP_RIGHT_PAREN: |
| 1517 | if ($this->t->scanOperand || $x->parenLevel == $pl) |
| 1518 | break 2; |
| 1519 | |
| 1520 | while (($tt = $this->reduce($operators, $operands)->type) != JS_GROUP && |
| 1521 | $tt != JS_CALL && $tt != JS_NEW_WITH_ARGS |
| 1522 | ) |
| 1523 | { |
| 1524 | continue; |
| 1525 | } |
| 1526 | |
| 1527 | if ($tt != JS_GROUP) |
| 1528 | { |
| 1529 | $n = end($operands); |
| 1530 | if ($n->treeNodes[1]->type != OP_COMMA) |
| 1531 | $n->treeNodes[1] = new JSNode($this->t, JS_LIST, $n->treeNodes[1]); |
| 1532 | else |
| 1533 | $n->treeNodes[1]->type = JS_LIST; |
| 1534 | } |
| 1535 | |
| 1536 | --$x->parenLevel; |
| 1537 | break; |
| 1538 | |
| 1539 | // Automatic semicolon insertion means we may scan across a newline |
| 1540 | // and into the beginning of another statement. If so, break out of |
| 1541 | // the while loop and let the t.scanOperand logic handle errors. |
| 1542 | default: |
| 1543 | break 2; |
| 1544 | } |
| 1545 | } |
| 1546 | |
| 1547 | if ($x->hookLevel != $hl) |
| 1548 | throw $this->t->newSyntaxError('Missing : in conditional expression'); |
| 1549 | |
| 1550 | if ($x->parenLevel != $pl) |
| 1551 | throw $this->t->newSyntaxError('Missing ) in parenthetical'); |
| 1552 | |
| 1553 | if ($x->bracketLevel != $bl) |
| 1554 | throw $this->t->newSyntaxError('Missing ] in index expression'); |
| 1555 | |
| 1556 | if ($this->t->scanOperand) |
| 1557 | throw $this->t->newSyntaxError('Missing operand'); |
| 1558 | |
| 1559 | // Resume default mode, scanning for operands, not operators. |
| 1560 | $this->t->scanOperand = true; |
| 1561 | $this->t->unget(); |
| 1562 | |
| 1563 | while (count($operators)) |
| 1564 | $this->reduce($operators, $operands); |
| 1565 | |
| 1566 | return array_pop($operands); |
| 1567 | } |
| 1568 | |
| 1569 | private function ParenExpression($x) |
| 1570 | { |
| 1571 | $this->t->mustMatch(OP_LEFT_PAREN); |
| 1572 | $n = $this->Expression($x); |
| 1573 | $this->t->mustMatch(OP_RIGHT_PAREN); |
| 1574 | |
| 1575 | return $n; |
| 1576 | } |
| 1577 | |
| 1578 | // Statement stack and nested statement handler. |
| 1579 | private function nest($x, $node, $end = false) |
| 1580 | { |
| 1581 | array_push($x->stmtStack, $node); |
| 1582 | $n = $this->statement($x); |
| 1583 | array_pop($x->stmtStack); |
| 1584 | |
| 1585 | if ($end) |
| 1586 | $this->t->mustMatch($end); |
| 1587 | |
| 1588 | return $n; |
| 1589 | } |
| 1590 | |
| 1591 | private function reduce(&$operators, &$operands) |
| 1592 | { |
| 1593 | $n = array_pop($operators); |
| 1594 | $op = $n->type; |
| 1595 | $arity = $this->opArity[$op]; |
| 1596 | $c = count($operands); |
| 1597 | if ($arity == -2) |
| 1598 | { |
| 1599 | // Flatten left-associative trees |
| 1600 | if ($c >= 2) |
| 1601 | { |
| 1602 | $left = $operands[$c - 2]; |
| 1603 | if ($left->type == $op) |
| 1604 | { |
| 1605 | $right = array_pop($operands); |
| 1606 | $left->addNode($right); |
| 1607 | return $left; |
| 1608 | } |
| 1609 | } |
| 1610 | $arity = 2; |
| 1611 | } |
| 1612 | |
| 1613 | // Always use push to add operands to n, to update start and end |
| 1614 | $a = array_splice($operands, $c - $arity); |
| 1615 | for ($i = 0; $i < $arity; $i++) |
| 1616 | $n->addNode($a[$i]); |
| 1617 | |
| 1618 | // Include closing bracket or postfix operator in [start,end] |
| 1619 | $te = $this->t->currentToken()->end; |
| 1620 | if ($n->end < $te) |
| 1621 | $n->end = $te; |
| 1622 | |
| 1623 | array_push($operands, $n); |
| 1624 | |
| 1625 | return $n; |
| 1626 | } |
| 1627 | } |
| 1628 | |
| 1629 | class JSCompilerContext |
| 1630 | { |
| 1631 | public $inFunction = false; |
| 1632 | public $inForLoopInit = false; |
| 1633 | public $ecmaStrictMode = false; |
| 1634 | public $bracketLevel = 0; |
| 1635 | public $curlyLevel = 0; |
| 1636 | public $parenLevel = 0; |
| 1637 | public $hookLevel = 0; |
| 1638 | |
| 1639 | public $stmtStack = array(); |
| 1640 | public $funDecls = array(); |
| 1641 | public $varDecls = array(); |
| 1642 | |
| 1643 | public function __construct($inFunction) |
| 1644 | { |
| 1645 | $this->inFunction = $inFunction; |
| 1646 | } |
| 1647 | } |
| 1648 | |
| 1649 | class JSNode |
| 1650 | { |
| 1651 | private $type; |
| 1652 | private $value; |
| 1653 | private $lineno; |
| 1654 | private $start; |
| 1655 | private $end; |
| 1656 | |
| 1657 | public $treeNodes = array(); |
| 1658 | public $funDecls = array(); |
| 1659 | public $varDecls = array(); |
| 1660 | |
| 1661 | public function __construct($t, $type=0) |
| 1662 | { |
| 1663 | if ($token = $t->currentToken()) |
| 1664 | { |
| 1665 | $this->type = $type ? $type : $token->type; |
| 1666 | $this->value = $token->value; |
| 1667 | $this->lineno = $token->lineno; |
| 1668 | $this->start = $token->start; |
| 1669 | $this->end = $token->end; |
| 1670 | } |
| 1671 | else |
| 1672 | { |
| 1673 | $this->type = $type; |
| 1674 | $this->lineno = $t->lineno; |
| 1675 | } |
| 1676 | |
| 1677 | if (($numargs = func_num_args()) > 2) |
| 1678 | { |
| 1679 | $args = func_get_args(); |
| 1680 | for ($i = 2; $i < $numargs; $i++) |
| 1681 | $this->addNode($args[$i]); |
| 1682 | } |
| 1683 | } |
| 1684 | |
| 1685 | // we don't want to bloat our object with all kind of specific properties, so we use overloading |
| 1686 | public function __set($name, $value) |
| 1687 | { |
| 1688 | $this->$name = $value; |
| 1689 | } |
| 1690 | |
| 1691 | public function __get($name) |
| 1692 | { |
| 1693 | if (isset($this->$name)) |
| 1694 | return $this->$name; |
| 1695 | |
| 1696 | return null; |
| 1697 | } |
| 1698 | |
| 1699 | public function addNode($node) |
| 1700 | { |
| 1701 | if ($node !== null) |
| 1702 | { |
| 1703 | if ($node->start < $this->start) |
| 1704 | $this->start = $node->start; |
| 1705 | if ($this->end < $node->end) |
| 1706 | $this->end = $node->end; |
| 1707 | } |
| 1708 | |
| 1709 | $this->treeNodes[] = $node; |
| 1710 | } |
| 1711 | } |
| 1712 | |
| 1713 | class JSTokenizer |
| 1714 | { |
| 1715 | private $cursor = 0; |
| 1716 | private $source; |
| 1717 | |
| 1718 | public $tokens = array(); |
| 1719 | public $tokenIndex = 0; |
| 1720 | public $lookahead = 0; |
| 1721 | public $scanNewlines = false; |
| 1722 | public $scanOperand = true; |
| 1723 | |
| 1724 | public $filename; |
| 1725 | public $lineno; |
| 1726 | |
| 1727 | private $keywords = array( |
| 1728 | 'break', |
| 1729 | 'case', 'catch', 'const', 'continue', |
| 1730 | 'debugger', 'default', 'delete', 'do', |
| 1731 | 'else', 'enum', |
| 1732 | 'false', 'finally', 'for', 'function', |
| 1733 | 'if', 'in', 'instanceof', |
| 1734 | 'new', 'null', |
| 1735 | 'return', |
| 1736 | 'switch', |
| 1737 | 'this', 'throw', 'true', 'try', 'typeof', |
| 1738 | 'var', 'void', |
| 1739 | 'while', 'with' |
| 1740 | ); |
| 1741 | |
| 1742 | private $opTypeNames = array( |
| 1743 | ';', ',', '?', ':', '||', '&&', '|', '^', |
| 1744 | '&', '===', '==', '=', '!==', '!=', '<<', '<=', |
| 1745 | '<', '>>>', '>>', '>=', '>', '++', '--', '+', |
| 1746 | '-', '*', '/', '%', '!', '~', '.', '[', |
| 1747 | ']', '{', '}', '(', ')', '@*/' |
| 1748 | ); |
| 1749 | |
| 1750 | private $assignOps = array('|', '^', '&', '<<', '>>', '>>>', '+', '-', '*', '/', '%'); |
| 1751 | private $opRegExp; |
| 1752 | |
| 1753 | public function __construct() |
| 1754 | { |
| 1755 | $this->opRegExp = '#^(' . implode('|', array_map('preg_quote', $this->opTypeNames)) . ')#'; |
| 1756 | } |
| 1757 | |
| 1758 | public function init($source, $filename = '', $lineno = 1) |
| 1759 | { |
| 1760 | $this->source = $source; |
| 1761 | $this->filename = $filename ? $filename : '[inline]'; |
| 1762 | $this->lineno = $lineno; |
| 1763 | |
| 1764 | $this->cursor = 0; |
| 1765 | $this->tokens = array(); |
| 1766 | $this->tokenIndex = 0; |
| 1767 | $this->lookahead = 0; |
| 1768 | $this->scanNewlines = false; |
| 1769 | $this->scanOperand = true; |
| 1770 | } |
| 1771 | |
| 1772 | public function getInput($chunksize) |
| 1773 | { |
| 1774 | if ($chunksize) |
| 1775 | return substr($this->source, $this->cursor, $chunksize); |
| 1776 | |
| 1777 | return substr($this->source, $this->cursor); |
| 1778 | } |
| 1779 | |
| 1780 | public function isDone() |
| 1781 | { |
| 1782 | return $this->peek() == TOKEN_END; |
| 1783 | } |
| 1784 | |
| 1785 | public function match($tt) |
| 1786 | { |
| 1787 | return $this->get() == $tt || $this->unget(); |
| 1788 | } |
| 1789 | |
| 1790 | public function mustMatch($tt) |
| 1791 | { |
| 1792 | if (!$this->match($tt)) |
| 1793 | throw $this->newSyntaxError('Unexpected token; token ' . $tt . ' expected'); |
| 1794 | |
| 1795 | return $this->currentToken(); |
| 1796 | } |
| 1797 | |
| 1798 | public function peek() |
| 1799 | { |
| 1800 | if ($this->lookahead) |
| 1801 | { |
| 1802 | $next = $this->tokens[($this->tokenIndex + $this->lookahead) & 3]; |
| 1803 | if ($this->scanNewlines && $next->lineno != $this->lineno) |
| 1804 | $tt = TOKEN_NEWLINE; |
| 1805 | else |
| 1806 | $tt = $next->type; |
| 1807 | } |
| 1808 | else |
| 1809 | { |
| 1810 | $tt = $this->get(); |
| 1811 | $this->unget(); |
| 1812 | } |
| 1813 | |
| 1814 | return $tt; |
| 1815 | } |
| 1816 | |
| 1817 | public function peekOnSameLine() |
| 1818 | { |
| 1819 | $this->scanNewlines = true; |
| 1820 | $tt = $this->peek(); |
| 1821 | $this->scanNewlines = false; |
| 1822 | |
| 1823 | return $tt; |
| 1824 | } |
| 1825 | |
| 1826 | public function currentToken() |
| 1827 | { |
| 1828 | if (!empty($this->tokens)) |
| 1829 | return $this->tokens[$this->tokenIndex]; |
| 1830 | } |
| 1831 | |
| 1832 | public function get($chunksize = 1000) |
| 1833 | { |
| 1834 | while($this->lookahead) |
| 1835 | { |
| 1836 | $this->lookahead--; |
| 1837 | $this->tokenIndex = ($this->tokenIndex + 1) & 3; |
| 1838 | $token = $this->tokens[$this->tokenIndex]; |
| 1839 | if ($token->type != TOKEN_NEWLINE || $this->scanNewlines) |
| 1840 | return $token->type; |
| 1841 | } |
| 1842 | |
| 1843 | $conditional_comment = false; |
| 1844 | |
| 1845 | // strip whitespace and comments |
| 1846 | while(true) |
| 1847 | { |
| 1848 | $input = $this->getInput($chunksize); |
| 1849 | |
| 1850 | // whitespace handling; gobble up \r as well (effectively we don't have support for MAC newlines!) |
| 1851 | $re = $this->scanNewlines ? '/^[ \r\t]+/' : '/^\s+/'; |
| 1852 | if (preg_match($re, $input, $match)) |
| 1853 | { |
| 1854 | $spaces = $match[0]; |
| 1855 | $spacelen = strlen($spaces); |
| 1856 | $this->cursor += $spacelen; |
| 1857 | if (!$this->scanNewlines) |
| 1858 | $this->lineno += substr_count($spaces, "\n"); |
| 1859 | |
| 1860 | if ($spacelen == $chunksize) |
| 1861 | continue; // complete chunk contained whitespace |
| 1862 | |
| 1863 | $input = $this->getInput($chunksize); |
| 1864 | if ($input == '' || $input[0] != '/') |
| 1865 | break; |
| 1866 | } |
| 1867 | |
| 1868 | // Comments |
| 1869 | if (!preg_match('/^\/(?:\*(@(?:cc_on|if|elif|else|end))?.*?\*\/|\/[^\n]*)/s', $input, $match)) |
| 1870 | { |
| 1871 | if (!$chunksize) |
| 1872 | break; |
| 1873 | |
| 1874 | // retry with a full chunk fetch; this also prevents breakage of long regular expressions (which will never match a comment) |
| 1875 | $chunksize = null; |
| 1876 | continue; |
| 1877 | } |
| 1878 | |
| 1879 | // check if this is a conditional (JScript) comment |
| 1880 | if (!empty($match[1])) |
| 1881 | { |
| 1882 | $match[0] = '/*' . $match[1]; |
| 1883 | $conditional_comment = true; |
| 1884 | break; |
| 1885 | } |
| 1886 | else |
| 1887 | { |
| 1888 | $this->cursor += strlen($match[0]); |
| 1889 | $this->lineno += substr_count($match[0], "\n"); |
| 1890 | } |
| 1891 | } |
| 1892 | |
| 1893 | if ($input == '') |
| 1894 | { |
| 1895 | $tt = TOKEN_END; |
| 1896 | $match = array(''); |
| 1897 | } |
| 1898 | elseif ($conditional_comment) |
| 1899 | { |
| 1900 | $tt = TOKEN_CONDCOMMENT_START; |
| 1901 | } |
| 1902 | else |
| 1903 | { |
| 1904 | switch ($input[0]) |
| 1905 | { |
| 1906 | case '0': |
| 1907 | // hexadecimal |
| 1908 | if (($input[1] == 'x' || $input[1] == 'X') && preg_match('/^0x[0-9a-f]+/i', $input, $match)) |
| 1909 | { |
| 1910 | $tt = TOKEN_NUMBER; |
| 1911 | break; |
| 1912 | } |
| 1913 | // FALL THROUGH |
| 1914 | |
| 1915 | case '1': case '2': case '3': case '4': case '5': |
| 1916 | case '6': case '7': case '8': case '9': |
| 1917 | // should always match |
| 1918 | preg_match('/^\d+(?:\.\d*)?(?:[eE][-+]?\d+)?/', $input, $match); |
| 1919 | $tt = TOKEN_NUMBER; |
| 1920 | break; |
| 1921 | |
| 1922 | case "'": |
| 1923 | if (preg_match('/^\'(?:[^\\\\\'\r\n]++|\\\\(?:.|\r?\n))*\'/', $input, $match)) |
| 1924 | { |
| 1925 | $tt = TOKEN_STRING; |
| 1926 | } |
| 1927 | else |
| 1928 | { |
| 1929 | if ($chunksize) |
| 1930 | return $this->get(null); // retry with a full chunk fetch |
| 1931 | |
| 1932 | throw $this->newSyntaxError('Unterminated string literal'); |
| 1933 | } |
| 1934 | break; |
| 1935 | |
| 1936 | case '"': |
| 1937 | if (preg_match('/^"(?:[^\\\\"\r\n]++|\\\\(?:.|\r?\n))*"/', $input, $match)) |
| 1938 | { |
| 1939 | $tt = TOKEN_STRING; |
| 1940 | } |
| 1941 | else |
| 1942 | { |
| 1943 | if ($chunksize) |
| 1944 | return $this->get(null); // retry with a full chunk fetch |
| 1945 | |
| 1946 | throw $this->newSyntaxError('Unterminated string literal'); |
| 1947 | } |
| 1948 | break; |
| 1949 | |
| 1950 | case '/': |
| 1951 | if ($this->scanOperand && preg_match('/^\/((?:\\\\.|\[(?:\\\\.|[^\]])*\]|[^\/])+)\/([gimy]*)/', $input, $match)) |
| 1952 | { |
| 1953 | $tt = TOKEN_REGEXP; |
| 1954 | break; |
| 1955 | } |
| 1956 | // FALL THROUGH |
| 1957 | |
| 1958 | case '|': |
| 1959 | case '^': |
| 1960 | case '&': |
| 1961 | case '<': |
| 1962 | case '>': |
| 1963 | case '+': |
| 1964 | case '-': |
| 1965 | case '*': |
| 1966 | case '%': |
| 1967 | case '=': |
| 1968 | case '!': |
| 1969 | // should always match |
| 1970 | preg_match($this->opRegExp, $input, $match); |
| 1971 | $op = $match[0]; |
| 1972 | if (in_array($op, $this->assignOps) && $input[strlen($op)] == '=') |
| 1973 | { |
| 1974 | $tt = OP_ASSIGN; |
| 1975 | $match[0] .= '='; |
| 1976 | } |
| 1977 | else |
| 1978 | { |
| 1979 | $tt = $op; |
| 1980 | if ($this->scanOperand) |
| 1981 | { |
| 1982 | if ($op == OP_PLUS) |
| 1983 | $tt = OP_UNARY_PLUS; |
| 1984 | elseif ($op == OP_MINUS) |
| 1985 | $tt = OP_UNARY_MINUS; |
| 1986 | } |
| 1987 | $op = null; |
| 1988 | } |
| 1989 | break; |
| 1990 | |
| 1991 | case '.': |
| 1992 | if (preg_match('/^\.\d+(?:[eE][-+]?\d+)?/', $input, $match)) |
| 1993 | { |
| 1994 | $tt = TOKEN_NUMBER; |
| 1995 | break; |
| 1996 | } |
| 1997 | // FALL THROUGH |
| 1998 | |
| 1999 | case ';': |
| 2000 | case ',': |
| 2001 | case '?': |
| 2002 | case ':': |
| 2003 | case '~': |
| 2004 | case '[': |
| 2005 | case ']': |
| 2006 | case '{': |
| 2007 | case '}': |
| 2008 | case '(': |
| 2009 | case ')': |
| 2010 | // these are all single |
| 2011 | $match = array($input[0]); |
| 2012 | $tt = $input[0]; |
| 2013 | break; |
| 2014 | |
| 2015 | case '@': |
| 2016 | // check end of conditional comment |
| 2017 | if (substr($input, 0, 3) == '@*/') |
| 2018 | { |
| 2019 | $match = array('@*/'); |
| 2020 | $tt = TOKEN_CONDCOMMENT_END; |
| 2021 | } |
| 2022 | else |
| 2023 | throw $this->newSyntaxError('Illegal token'); |
| 2024 | break; |
| 2025 | |
| 2026 | case "\n": |
| 2027 | if ($this->scanNewlines) |
| 2028 | { |
| 2029 | $match = array("\n"); |
| 2030 | $tt = TOKEN_NEWLINE; |
| 2031 | } |
| 2032 | else |
| 2033 | throw $this->newSyntaxError('Illegal token'); |
| 2034 | break; |
| 2035 | |
| 2036 | default: |
| 2037 | // FIXME: add support for unicode and unicode escape sequence \uHHHH |
| 2038 | if (preg_match('/^[$\w]+/', $input, $match)) |
| 2039 | { |
| 2040 | $tt = in_array($match[0], $this->keywords) ? $match[0] : TOKEN_IDENTIFIER; |
| 2041 | } |
| 2042 | else |
| 2043 | throw $this->newSyntaxError('Illegal token'); |
| 2044 | } |
| 2045 | } |
| 2046 | |
| 2047 | $this->tokenIndex = ($this->tokenIndex + 1) & 3; |
| 2048 | |
| 2049 | if (!isset($this->tokens[$this->tokenIndex])) |
| 2050 | $this->tokens[$this->tokenIndex] = new JSToken(); |
| 2051 | |
| 2052 | $token = $this->tokens[$this->tokenIndex]; |
| 2053 | $token->type = $tt; |
| 2054 | |
| 2055 | if ($tt == OP_ASSIGN) |
| 2056 | $token->assignOp = $op; |
| 2057 | |
| 2058 | $token->start = $this->cursor; |
| 2059 | |
| 2060 | $token->value = $match[0]; |
| 2061 | $this->cursor += strlen($match[0]); |
| 2062 | |
| 2063 | $token->end = $this->cursor; |
| 2064 | $token->lineno = $this->lineno; |
| 2065 | |
| 2066 | return $tt; |
| 2067 | } |
| 2068 | |
| 2069 | public function unget() |
| 2070 | { |
| 2071 | if (++$this->lookahead == 4) |
| 2072 | throw $this->newSyntaxError('PANIC: too much lookahead!'); |
| 2073 | |
| 2074 | $this->tokenIndex = ($this->tokenIndex - 1) & 3; |
| 2075 | } |
| 2076 | |
| 2077 | public function newSyntaxError($m) |
| 2078 | { |
| 2079 | return new Exception('Parse error: ' . $m . ' in file \'' . $this->filename . '\' on line ' . $this->lineno); |
| 2080 | } |
| 2081 | } |
| 2082 | |
| 2083 | class JSToken |
| 2084 | { |
| 2085 | public $type; |
| 2086 | public $value; |
| 2087 | public $start; |
| 2088 | public $end; |
| 2089 | public $lineno; |
| 2090 | public $assignOp; |
| 2091 | } |
| 2092 |