Localisation updates from https://translatewiki.net.
[mediawiki.git] / includes / libs / jsminplus.php
blob99cf399bd94d5ef67e7f33bf1eb379bcacf15aba
1 <?php
2 // @codingStandardsIgnoreFile File external to MediaWiki. Ignore coding conventions checks.
3 /**
4 * JSMinPlus version 1.4
6 * Minifies a javascript file using a javascript parser
8 * This implements a PHP port of Brendan Eich's Narcissus open source javascript engine (in javascript)
9 * References: http://en.wikipedia.org/wiki/Narcissus_(JavaScript_engine)
10 * Narcissus sourcecode: http://mxr.mozilla.org/mozilla/source/js/narcissus/
11 * JSMinPlus weblog: http://crisp.tweakblogs.net/blog/cat/716
13 * Tino Zijdel <crisp@tweakers.net>
15 * Usage: $minified = JSMinPlus::minify($script [, $filename])
17 * Versionlog (see also changelog.txt):
18 * 23-07-2011 - remove dynamic creation of OP_* and KEYWORD_* defines and declare them on top
19 * reduce memory footprint by minifying by block-scope
20 * some small byte-saving and performance improvements
21 * 12-05-2009 - fixed hook:colon precedence, fixed empty body in loop and if-constructs
22 * 18-04-2009 - fixed crashbug in PHP 5.2.9 and several other bugfixes
23 * 12-04-2009 - some small bugfixes and performance improvements
24 * 09-04-2009 - initial open sourced version 1.0
26 * Latest version of this script: http://files.tweakers.net/jsminplus/jsminplus.zip
28 * @file
31 /* ***** BEGIN LICENSE BLOCK *****
32 * Version: MPL 1.1/GPL 2.0/LGPL 2.1
34 * The contents of this file are subject to the Mozilla Public License Version
35 * 1.1 (the "License"); you may not use this file except in compliance with
36 * the License. You may obtain a copy of the License at
37 * http://www.mozilla.org/MPL/
39 * Software distributed under the License is distributed on an "AS IS" basis,
40 * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
41 * for the specific language governing rights and limitations under the
42 * License.
44 * The Original Code is the Narcissus JavaScript engine.
46 * The Initial Developer of the Original Code is
47 * Brendan Eich <brendan@mozilla.org>.
48 * Portions created by the Initial Developer are Copyright (C) 2004
49 * the Initial Developer. All Rights Reserved.
51 * Contributor(s): Tino Zijdel <crisp@tweakers.net>
52 * PHP port, modifications and minifier routine are (C) 2009-2011
54 * Alternatively, the contents of this file may be used under the terms of
55 * either the GNU General Public License Version 2 or later (the "GPL"), or
56 * the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
57 * in which case the provisions of the GPL or the LGPL are applicable instead
58 * of those above. If you wish to allow use of your version of this file only
59 * under the terms of either the GPL or the LGPL, and not to allow others to
60 * use your version of this file under the terms of the MPL, indicate your
61 * decision by deleting the provisions above and replace them with the notice
62 * and other provisions required by the GPL or the LGPL. If you do not delete
63 * the provisions above, a recipient may use your version of this file under
64 * the terms of any one of the MPL, the GPL or the LGPL.
66 * ***** END LICENSE BLOCK ***** */
68 define('TOKEN_END', 1);
69 define('TOKEN_NUMBER', 2);
70 define('TOKEN_IDENTIFIER', 3);
71 define('TOKEN_STRING', 4);
72 define('TOKEN_REGEXP', 5);
73 define('TOKEN_NEWLINE', 6);
74 define('TOKEN_CONDCOMMENT_START', 7);
75 define('TOKEN_CONDCOMMENT_END', 8);
77 define('JS_SCRIPT', 100);
78 define('JS_BLOCK', 101);
79 define('JS_LABEL', 102);
80 define('JS_FOR_IN', 103);
81 define('JS_CALL', 104);
82 define('JS_NEW_WITH_ARGS', 105);
83 define('JS_INDEX', 106);
84 define('JS_ARRAY_INIT', 107);
85 define('JS_OBJECT_INIT', 108);
86 define('JS_PROPERTY_INIT', 109);
87 define('JS_GETTER', 110);
88 define('JS_SETTER', 111);
89 define('JS_GROUP', 112);
90 define('JS_LIST', 113);
92 define('JS_MINIFIED', 999);
94 define('DECLARED_FORM', 0);
95 define('EXPRESSED_FORM', 1);
96 define('STATEMENT_FORM', 2);
98 /* Operators */
99 define('OP_SEMICOLON', ';');
100 define('OP_COMMA', ',');
101 define('OP_HOOK', '?');
102 define('OP_COLON', ':');
103 define('OP_OR', '||');
104 define('OP_AND', '&&');
105 define('OP_BITWISE_OR', '|');
106 define('OP_BITWISE_XOR', '^');
107 define('OP_BITWISE_AND', '&');
108 define('OP_STRICT_EQ', '===');
109 define('OP_EQ', '==');
110 define('OP_ASSIGN', '=');
111 define('OP_STRICT_NE', '!==');
112 define('OP_NE', '!=');
113 define('OP_LSH', '<<');
114 define('OP_LE', '<=');
115 define('OP_LT', '<');
116 define('OP_URSH', '>>>');
117 define('OP_RSH', '>>');
118 define('OP_GE', '>=');
119 define('OP_GT', '>');
120 define('OP_INCREMENT', '++');
121 define('OP_DECREMENT', '--');
122 define('OP_PLUS', '+');
123 define('OP_MINUS', '-');
124 define('OP_MUL', '*');
125 define('OP_DIV', '/');
126 define('OP_MOD', '%');
127 define('OP_NOT', '!');
128 define('OP_BITWISE_NOT', '~');
129 define('OP_DOT', '.');
130 define('OP_LEFT_BRACKET', '[');
131 define('OP_RIGHT_BRACKET', ']');
132 define('OP_LEFT_CURLY', '{');
133 define('OP_RIGHT_CURLY', '}');
134 define('OP_LEFT_PAREN', '(');
135 define('OP_RIGHT_PAREN', ')');
136 define('OP_CONDCOMMENT_END', '@*/');
138 define('OP_UNARY_PLUS', 'U+');
139 define('OP_UNARY_MINUS', 'U-');
141 /* Keywords */
142 define('KEYWORD_BREAK', 'break');
143 define('KEYWORD_CASE', 'case');
144 define('KEYWORD_CATCH', 'catch');
145 define('KEYWORD_CONST', 'const');
146 define('KEYWORD_CONTINUE', 'continue');
147 define('KEYWORD_DEBUGGER', 'debugger');
148 define('KEYWORD_DEFAULT', 'default');
149 define('KEYWORD_DELETE', 'delete');
150 define('KEYWORD_DO', 'do');
151 define('KEYWORD_ELSE', 'else');
152 define('KEYWORD_ENUM', 'enum');
153 define('KEYWORD_FALSE', 'false');
154 define('KEYWORD_FINALLY', 'finally');
155 define('KEYWORD_FOR', 'for');
156 define('KEYWORD_FUNCTION', 'function');
157 define('KEYWORD_IF', 'if');
158 define('KEYWORD_IN', 'in');
159 define('KEYWORD_INSTANCEOF', 'instanceof');
160 define('KEYWORD_NEW', 'new');
161 define('KEYWORD_NULL', 'null');
162 define('KEYWORD_RETURN', 'return');
163 define('KEYWORD_SWITCH', 'switch');
164 define('KEYWORD_THIS', 'this');
165 define('KEYWORD_THROW', 'throw');
166 define('KEYWORD_TRUE', 'true');
167 define('KEYWORD_TRY', 'try');
168 define('KEYWORD_TYPEOF', 'typeof');
169 define('KEYWORD_VAR', 'var');
170 define('KEYWORD_VOID', 'void');
171 define('KEYWORD_WHILE', 'while');
172 define('KEYWORD_WITH', 'with');
175 class JSMinPlus
177 private $parser;
178 private $reserved = array(
179 'break', 'case', 'catch', 'continue', 'default', 'delete', 'do',
180 'else', 'finally', 'for', 'function', 'if', 'in', 'instanceof',
181 'new', 'return', 'switch', 'this', 'throw', 'try', 'typeof', 'var',
182 'void', 'while', 'with',
183 // Words reserved for future use
184 'abstract', 'boolean', 'byte', 'char', 'class', 'const', 'debugger',
185 'double', 'enum', 'export', 'extends', 'final', 'float', 'goto',
186 'implements', 'import', 'int', 'interface', 'long', 'native',
187 'package', 'private', 'protected', 'public', 'short', 'static',
188 'super', 'synchronized', 'throws', 'transient', 'volatile',
189 // These are not reserved, but should be taken into account
190 // in isValidIdentifier (See jslint source code)
191 'arguments', 'eval', 'true', 'false', 'Infinity', 'NaN', 'null', 'undefined'
194 private function __construct()
196 $this->parser = new JSParser($this);
199 public static function minify($js, $filename='')
201 static $instance;
203 // this is a singleton
204 if(!$instance)
205 $instance = new JSMinPlus();
207 return $instance->min($js, $filename);
210 private function min($js, $filename)
214 $n = $this->parser->parse($js, $filename, 1);
215 return $this->parseTree($n);
217 catch(Exception $e)
219 echo $e->getMessage() . "\n";
222 return false;
225 public function parseTree($n, $noBlockGrouping = false)
227 $s = '';
229 switch ($n->type)
231 case JS_MINIFIED:
232 $s = $n->value;
233 break;
235 case JS_SCRIPT:
236 // we do nothing yet with funDecls or varDecls
237 $noBlockGrouping = true;
238 // FALL THROUGH
240 case JS_BLOCK:
241 $childs = $n->treeNodes;
242 $lastType = 0;
243 for ($c = 0, $i = 0, $j = count($childs); $i < $j; $i++)
245 $type = $childs[$i]->type;
246 $t = $this->parseTree($childs[$i]);
247 if (strlen($t))
249 if ($c)
251 $s = rtrim($s, ';');
253 if ($type == KEYWORD_FUNCTION && $childs[$i]->functionForm == DECLARED_FORM)
255 // put declared functions on a new line
256 $s .= "\n";
258 elseif ($type == KEYWORD_VAR && $type == $lastType)
260 // multiple var-statements can go into one
261 $t = ',' . substr($t, 4);
263 else
265 // add terminator
266 $s .= ';';
270 $s .= $t;
272 $c++;
273 $lastType = $type;
277 if ($c > 1 && !$noBlockGrouping)
279 $s = '{' . $s . '}';
281 break;
283 case KEYWORD_FUNCTION:
284 $s .= 'function' . ($n->name ? ' ' . $n->name : '') . '(';
285 $params = $n->params;
286 for ($i = 0, $j = count($params); $i < $j; $i++)
287 $s .= ($i ? ',' : '') . $params[$i];
288 $s .= '){' . $this->parseTree($n->body, true) . '}';
289 break;
291 case KEYWORD_IF:
292 $s = 'if(' . $this->parseTree($n->condition) . ')';
293 $thenPart = $this->parseTree($n->thenPart);
294 $elsePart = $n->elsePart ? $this->parseTree($n->elsePart) : null;
296 // empty if-statement
297 if ($thenPart == '')
298 $thenPart = ';';
300 if ($elsePart)
302 // be careful and always make a block out of the thenPart; could be more optimized but is a lot of trouble
303 if ($thenPart != ';' && $thenPart[0] != '{')
304 $thenPart = '{' . $thenPart . '}';
306 $s .= $thenPart . 'else';
308 // we could check for more, but that hardly ever applies so go for performance
309 if ($elsePart[0] != '{')
310 $s .= ' ';
312 $s .= $elsePart;
314 else
316 $s .= $thenPart;
318 break;
320 case KEYWORD_SWITCH:
321 $s = 'switch(' . $this->parseTree($n->discriminant) . '){';
322 $cases = $n->cases;
323 for ($i = 0, $j = count($cases); $i < $j; $i++)
325 $case = $cases[$i];
326 if ($case->type == KEYWORD_CASE)
327 $s .= 'case' . ($case->caseLabel->type != TOKEN_STRING ? ' ' : '') . $this->parseTree($case->caseLabel) . ':';
328 else
329 $s .= 'default:';
331 $statement = $this->parseTree($case->statements, true);
332 if ($statement)
334 $s .= $statement;
335 // no terminator for last statement
336 if ($i + 1 < $j)
337 $s .= ';';
340 $s .= '}';
341 break;
343 case KEYWORD_FOR:
344 $s = 'for(' . ($n->setup ? $this->parseTree($n->setup) : '')
345 . ';' . ($n->condition ? $this->parseTree($n->condition) : '')
346 . ';' . ($n->update ? $this->parseTree($n->update) : '') . ')';
348 $body = $this->parseTree($n->body);
349 if ($body == '')
350 $body = ';';
352 $s .= $body;
353 break;
355 case KEYWORD_WHILE:
356 $s = 'while(' . $this->parseTree($n->condition) . ')';
358 $body = $this->parseTree($n->body);
359 if ($body == '')
360 $body = ';';
362 $s .= $body;
363 break;
365 case JS_FOR_IN:
366 $s = 'for(' . ($n->varDecl ? $this->parseTree($n->varDecl) : $this->parseTree($n->iterator)) . ' in ' . $this->parseTree($n->object) . ')';
368 $body = $this->parseTree($n->body);
369 if ($body == '')
370 $body = ';';
372 $s .= $body;
373 break;
375 case KEYWORD_DO:
376 $s = 'do{' . $this->parseTree($n->body, true) . '}while(' . $this->parseTree($n->condition) . ')';
377 break;
379 case KEYWORD_BREAK:
380 case KEYWORD_CONTINUE:
381 $s = $n->value . ($n->label ? ' ' . $n->label : '');
382 break;
384 case KEYWORD_TRY:
385 $s = 'try{' . $this->parseTree($n->tryBlock, true) . '}';
386 $catchClauses = $n->catchClauses;
387 for ($i = 0, $j = count($catchClauses); $i < $j; $i++)
389 $t = $catchClauses[$i];
390 $s .= 'catch(' . $t->varName . ($t->guard ? ' if ' . $this->parseTree($t->guard) : '') . '){' . $this->parseTree($t->block, true) . '}';
392 if ($n->finallyBlock)
393 $s .= 'finally{' . $this->parseTree($n->finallyBlock, true) . '}';
394 break;
396 case KEYWORD_THROW:
397 case KEYWORD_RETURN:
398 $s = $n->type;
399 if ($n->value)
401 $t = $this->parseTree($n->value);
402 if (strlen($t))
404 if ($this->isWordChar($t[0]) || $t[0] == '\\')
405 $s .= ' ';
407 $s .= $t;
410 break;
412 case KEYWORD_WITH:
413 $s = 'with(' . $this->parseTree($n->object) . ')' . $this->parseTree($n->body);
414 break;
416 case KEYWORD_VAR:
417 case KEYWORD_CONST:
418 $s = $n->value . ' ';
419 $childs = $n->treeNodes;
420 for ($i = 0, $j = count($childs); $i < $j; $i++)
422 $t = $childs[$i];
423 $s .= ($i ? ',' : '') . $t->name;
424 $u = $t->initializer;
425 if ($u)
426 $s .= '=' . $this->parseTree($u);
428 break;
430 case KEYWORD_IN:
431 case KEYWORD_INSTANCEOF:
432 $left = $this->parseTree($n->treeNodes[0]);
433 $right = $this->parseTree($n->treeNodes[1]);
435 $s = $left;
437 if ($this->isWordChar(substr($left, -1)))
438 $s .= ' ';
440 $s .= $n->type;
442 if ($this->isWordChar($right[0]) || $right[0] == '\\')
443 $s .= ' ';
445 $s .= $right;
446 break;
448 case KEYWORD_DELETE:
449 case KEYWORD_TYPEOF:
450 $right = $this->parseTree($n->treeNodes[0]);
452 $s = $n->type;
454 if ($this->isWordChar($right[0]) || $right[0] == '\\')
455 $s .= ' ';
457 $s .= $right;
458 break;
460 case KEYWORD_VOID:
461 $s = 'void(' . $this->parseTree($n->treeNodes[0]) . ')';
462 break;
464 case KEYWORD_DEBUGGER:
465 throw new Exception('NOT IMPLEMENTED: DEBUGGER');
466 break;
468 case TOKEN_CONDCOMMENT_START:
469 case TOKEN_CONDCOMMENT_END:
470 $s = $n->value . ($n->type == TOKEN_CONDCOMMENT_START ? ' ' : '');
471 $childs = $n->treeNodes;
472 for ($i = 0, $j = count($childs); $i < $j; $i++)
473 $s .= $this->parseTree($childs[$i]);
474 break;
476 case OP_SEMICOLON:
477 if ($expression = $n->expression)
478 $s = $this->parseTree($expression);
479 break;
481 case JS_LABEL:
482 $s = $n->label . ':' . $this->parseTree($n->statement);
483 break;
485 case OP_COMMA:
486 $childs = $n->treeNodes;
487 for ($i = 0, $j = count($childs); $i < $j; $i++)
488 $s .= ($i ? ',' : '') . $this->parseTree($childs[$i]);
489 break;
491 case OP_ASSIGN:
492 $s = $this->parseTree($n->treeNodes[0]) . $n->value . $this->parseTree($n->treeNodes[1]);
493 break;
495 case OP_HOOK:
496 $s = $this->parseTree($n->treeNodes[0]) . '?' . $this->parseTree($n->treeNodes[1]) . ':' . $this->parseTree($n->treeNodes[2]);
497 break;
499 case OP_OR: case OP_AND:
500 case OP_BITWISE_OR: case OP_BITWISE_XOR: case OP_BITWISE_AND:
501 case OP_EQ: case OP_NE: case OP_STRICT_EQ: case OP_STRICT_NE:
502 case OP_LT: case OP_LE: case OP_GE: case OP_GT:
503 case OP_LSH: case OP_RSH: case OP_URSH:
504 case OP_MUL: case OP_DIV: case OP_MOD:
505 $s = $this->parseTree($n->treeNodes[0]) . $n->type . $this->parseTree($n->treeNodes[1]);
506 break;
508 case OP_PLUS:
509 case OP_MINUS:
510 $left = $this->parseTree($n->treeNodes[0]);
511 $right = $this->parseTree($n->treeNodes[1]);
513 switch ($n->treeNodes[1]->type)
515 case OP_PLUS:
516 case OP_MINUS:
517 case OP_INCREMENT:
518 case OP_DECREMENT:
519 case OP_UNARY_PLUS:
520 case OP_UNARY_MINUS:
521 $s = $left . $n->type . ' ' . $right;
522 break;
524 case TOKEN_STRING:
525 //combine concatenated strings with same quote style
526 if ($n->type == OP_PLUS && substr($left, -1) == $right[0])
528 $s = substr($left, 0, -1) . substr($right, 1);
529 break;
531 // FALL THROUGH
533 default:
534 $s = $left . $n->type . $right;
536 break;
538 case OP_NOT:
539 case OP_BITWISE_NOT:
540 case OP_UNARY_PLUS:
541 case OP_UNARY_MINUS:
542 $s = $n->value . $this->parseTree($n->treeNodes[0]);
543 break;
545 case OP_INCREMENT:
546 case OP_DECREMENT:
547 if ($n->postfix)
548 $s = $this->parseTree($n->treeNodes[0]) . $n->value;
549 else
550 $s = $n->value . $this->parseTree($n->treeNodes[0]);
551 break;
553 case OP_DOT:
554 $s = $this->parseTree($n->treeNodes[0]) . '.' . $this->parseTree($n->treeNodes[1]);
555 break;
557 case JS_INDEX:
558 $s = $this->parseTree($n->treeNodes[0]);
559 // See if we can replace named index with a dot saving 3 bytes
560 if ( $n->treeNodes[0]->type == TOKEN_IDENTIFIER &&
561 $n->treeNodes[1]->type == TOKEN_STRING &&
562 $this->isValidIdentifier(substr($n->treeNodes[1]->value, 1, -1))
564 $s .= '.' . substr($n->treeNodes[1]->value, 1, -1);
565 else
566 $s .= '[' . $this->parseTree($n->treeNodes[1]) . ']';
567 break;
569 case JS_LIST:
570 $childs = $n->treeNodes;
571 for ($i = 0, $j = count($childs); $i < $j; $i++)
572 $s .= ($i ? ',' : '') . $this->parseTree($childs[$i]);
573 break;
575 case JS_CALL:
576 $s = $this->parseTree($n->treeNodes[0]) . '(' . $this->parseTree($n->treeNodes[1]) . ')';
577 break;
579 case KEYWORD_NEW:
580 case JS_NEW_WITH_ARGS:
581 $s = 'new ' . $this->parseTree($n->treeNodes[0]) . '(' . ($n->type == JS_NEW_WITH_ARGS ? $this->parseTree($n->treeNodes[1]) : '') . ')';
582 break;
584 case JS_ARRAY_INIT:
585 $s = '[';
586 $childs = $n->treeNodes;
587 for ($i = 0, $j = count($childs); $i < $j; $i++)
589 $s .= ($i ? ',' : '') . $this->parseTree($childs[$i]);
591 $s .= ']';
592 break;
594 case JS_OBJECT_INIT:
595 $s = '{';
596 $childs = $n->treeNodes;
597 for ($i = 0, $j = count($childs); $i < $j; $i++)
599 $t = $childs[$i];
600 if ($i)
601 $s .= ',';
602 if ($t->type == JS_PROPERTY_INIT)
604 // Ditch the quotes when the index is a valid identifier
605 if ( $t->treeNodes[0]->type == TOKEN_STRING &&
606 $this->isValidIdentifier(substr($t->treeNodes[0]->value, 1, -1))
608 $s .= substr($t->treeNodes[0]->value, 1, -1);
609 else
610 $s .= $t->treeNodes[0]->value;
612 $s .= ':' . $this->parseTree($t->treeNodes[1]);
614 else
616 $s .= $t->type == JS_GETTER ? 'get' : 'set';
617 $s .= ' ' . $t->name . '(';
618 $params = $t->params;
619 for ($i = 0, $j = count($params); $i < $j; $i++)
620 $s .= ($i ? ',' : '') . $params[$i];
621 $s .= '){' . $this->parseTree($t->body, true) . '}';
624 $s .= '}';
625 break;
627 case TOKEN_NUMBER:
628 $s = $n->value;
629 if (preg_match('/^([1-9]+)(0{3,})$/', $s, $m))
630 $s = $m[1] . 'e' . strlen($m[2]);
631 break;
633 case KEYWORD_NULL: case KEYWORD_THIS: case KEYWORD_TRUE: case KEYWORD_FALSE:
634 case TOKEN_IDENTIFIER: case TOKEN_STRING: case TOKEN_REGEXP:
635 $s = $n->value;
636 break;
638 case JS_GROUP:
639 if (in_array(
640 $n->treeNodes[0]->type,
641 array(
642 JS_ARRAY_INIT, JS_OBJECT_INIT, JS_GROUP,
643 TOKEN_NUMBER, TOKEN_STRING, TOKEN_REGEXP, TOKEN_IDENTIFIER,
644 KEYWORD_NULL, KEYWORD_THIS, KEYWORD_TRUE, KEYWORD_FALSE
648 $s = $this->parseTree($n->treeNodes[0]);
650 else
652 $s = '(' . $this->parseTree($n->treeNodes[0]) . ')';
654 break;
656 default:
657 throw new Exception('UNKNOWN TOKEN TYPE: ' . $n->type);
660 return $s;
663 private function isValidIdentifier($string)
665 return preg_match('/^[a-zA-Z_][a-zA-Z0-9_]*$/', $string) && !in_array($string, $this->reserved);
668 private function isWordChar($char)
670 return $char == '_' || $char == '$' || ctype_alnum($char);
674 class JSParser
676 private $t;
677 private $minifier;
679 private $opPrecedence = array(
680 ';' => 0,
681 ',' => 1,
682 '=' => 2, '?' => 2, ':' => 2,
683 // The above all have to have the same precedence, see bug 330975
684 '||' => 4,
685 '&&' => 5,
686 '|' => 6,
687 '^' => 7,
688 '&' => 8,
689 '==' => 9, '!=' => 9, '===' => 9, '!==' => 9,
690 '<' => 10, '<=' => 10, '>=' => 10, '>' => 10, 'in' => 10, 'instanceof' => 10,
691 '<<' => 11, '>>' => 11, '>>>' => 11,
692 '+' => 12, '-' => 12,
693 '*' => 13, '/' => 13, '%' => 13,
694 'delete' => 14, 'void' => 14, 'typeof' => 14,
695 '!' => 14, '~' => 14, 'U+' => 14, 'U-' => 14,
696 '++' => 15, '--' => 15,
697 'new' => 16,
698 '.' => 17,
699 JS_NEW_WITH_ARGS => 0, JS_INDEX => 0, JS_CALL => 0,
700 JS_ARRAY_INIT => 0, JS_OBJECT_INIT => 0, JS_GROUP => 0
703 private $opArity = array(
704 ',' => -2,
705 '=' => 2,
706 '?' => 3,
707 '||' => 2,
708 '&&' => 2,
709 '|' => 2,
710 '^' => 2,
711 '&' => 2,
712 '==' => 2, '!=' => 2, '===' => 2, '!==' => 2,
713 '<' => 2, '<=' => 2, '>=' => 2, '>' => 2, 'in' => 2, 'instanceof' => 2,
714 '<<' => 2, '>>' => 2, '>>>' => 2,
715 '+' => 2, '-' => 2,
716 '*' => 2, '/' => 2, '%' => 2,
717 'delete' => 1, 'void' => 1, 'typeof' => 1,
718 '!' => 1, '~' => 1, 'U+' => 1, 'U-' => 1,
719 '++' => 1, '--' => 1,
720 'new' => 1,
721 '.' => 2,
722 JS_NEW_WITH_ARGS => 2, JS_INDEX => 2, JS_CALL => 2,
723 JS_ARRAY_INIT => 1, JS_OBJECT_INIT => 1, JS_GROUP => 1,
724 TOKEN_CONDCOMMENT_START => 1, TOKEN_CONDCOMMENT_END => 1
727 public function __construct($minifier=null)
729 $this->minifier = $minifier;
730 $this->t = new JSTokenizer();
733 public function parse($s, $f, $l)
735 // initialize tokenizer
736 $this->t->init($s, $f, $l);
738 $x = new JSCompilerContext(false);
739 $n = $this->Script($x);
740 if (!$this->t->isDone())
741 throw $this->t->newSyntaxError('Syntax error');
743 return $n;
746 private function Script($x)
748 $n = $this->Statements($x);
749 $n->type = JS_SCRIPT;
750 $n->funDecls = $x->funDecls;
751 $n->varDecls = $x->varDecls;
753 // minify by scope
754 if ($this->minifier)
756 $n->value = $this->minifier->parseTree($n);
758 // clear tree from node to save memory
759 $n->treeNodes = null;
760 $n->funDecls = null;
761 $n->varDecls = null;
763 $n->type = JS_MINIFIED;
766 return $n;
769 private function Statements($x)
771 $n = new JSNode($this->t, JS_BLOCK);
772 array_push($x->stmtStack, $n);
774 while (!$this->t->isDone() && $this->t->peek() != OP_RIGHT_CURLY)
775 $n->addNode($this->Statement($x));
777 array_pop($x->stmtStack);
779 return $n;
782 private function Block($x)
784 $this->t->mustMatch(OP_LEFT_CURLY);
785 $n = $this->Statements($x);
786 $this->t->mustMatch(OP_RIGHT_CURLY);
788 return $n;
791 private function Statement($x)
793 $tt = $this->t->get();
794 $n2 = null;
796 // Cases for statements ending in a right curly return early, avoiding the
797 // common semicolon insertion magic after this switch.
798 switch ($tt)
800 case KEYWORD_FUNCTION:
801 return $this->FunctionDefinition(
803 true,
804 count($x->stmtStack) > 1 ? STATEMENT_FORM : DECLARED_FORM
806 break;
808 case OP_LEFT_CURLY:
809 $n = $this->Statements($x);
810 $this->t->mustMatch(OP_RIGHT_CURLY);
811 return $n;
813 case KEYWORD_IF:
814 $n = new JSNode($this->t);
815 $n->condition = $this->ParenExpression($x);
816 array_push($x->stmtStack, $n);
817 $n->thenPart = $this->Statement($x);
818 $n->elsePart = $this->t->match(KEYWORD_ELSE) ? $this->Statement($x) : null;
819 array_pop($x->stmtStack);
820 return $n;
822 case KEYWORD_SWITCH:
823 $n = new JSNode($this->t);
824 $this->t->mustMatch(OP_LEFT_PAREN);
825 $n->discriminant = $this->Expression($x);
826 $this->t->mustMatch(OP_RIGHT_PAREN);
827 $n->cases = array();
828 $n->defaultIndex = -1;
830 array_push($x->stmtStack, $n);
832 $this->t->mustMatch(OP_LEFT_CURLY);
834 while (($tt = $this->t->get()) != OP_RIGHT_CURLY)
836 switch ($tt)
838 case KEYWORD_DEFAULT:
839 if ($n->defaultIndex >= 0)
840 throw $this->t->newSyntaxError('More than one switch default');
841 // FALL THROUGH
842 case KEYWORD_CASE:
843 $n2 = new JSNode($this->t);
844 if ($tt == KEYWORD_DEFAULT)
845 $n->defaultIndex = count($n->cases);
846 else
847 $n2->caseLabel = $this->Expression($x, OP_COLON);
848 break;
849 default:
850 throw $this->t->newSyntaxError('Invalid switch case');
853 $this->t->mustMatch(OP_COLON);
854 $n2->statements = new JSNode($this->t, JS_BLOCK);
855 while (($tt = $this->t->peek()) != KEYWORD_CASE && $tt != KEYWORD_DEFAULT && $tt != OP_RIGHT_CURLY)
856 $n2->statements->addNode($this->Statement($x));
858 array_push($n->cases, $n2);
861 array_pop($x->stmtStack);
862 return $n;
864 case KEYWORD_FOR:
865 $n = new JSNode($this->t);
866 $n->isLoop = true;
867 $this->t->mustMatch(OP_LEFT_PAREN);
869 if (($tt = $this->t->peek()) != OP_SEMICOLON)
871 $x->inForLoopInit = true;
872 if ($tt == KEYWORD_VAR || $tt == KEYWORD_CONST)
874 $this->t->get();
875 $n2 = $this->Variables($x);
877 else
879 $n2 = $this->Expression($x);
881 $x->inForLoopInit = false;
884 if ($n2 && $this->t->match(KEYWORD_IN))
886 $n->type = JS_FOR_IN;
887 if ($n2->type == KEYWORD_VAR)
889 if (count($n2->treeNodes) != 1)
891 throw $this->t->SyntaxError(
892 'Invalid for..in left-hand side',
893 $this->t->filename,
894 $n2->lineno
898 // NB: n2[0].type == IDENTIFIER and n2[0].value == n2[0].name.
899 $n->iterator = $n2->treeNodes[0];
900 $n->varDecl = $n2;
902 else
904 $n->iterator = $n2;
905 $n->varDecl = null;
908 $n->object = $this->Expression($x);
910 else
912 $n->setup = $n2 ? $n2 : null;
913 $this->t->mustMatch(OP_SEMICOLON);
914 $n->condition = $this->t->peek() == OP_SEMICOLON ? null : $this->Expression($x);
915 $this->t->mustMatch(OP_SEMICOLON);
916 $n->update = $this->t->peek() == OP_RIGHT_PAREN ? null : $this->Expression($x);
919 $this->t->mustMatch(OP_RIGHT_PAREN);
920 $n->body = $this->nest($x, $n);
921 return $n;
923 case KEYWORD_WHILE:
924 $n = new JSNode($this->t);
925 $n->isLoop = true;
926 $n->condition = $this->ParenExpression($x);
927 $n->body = $this->nest($x, $n);
928 return $n;
930 case KEYWORD_DO:
931 $n = new JSNode($this->t);
932 $n->isLoop = true;
933 $n->body = $this->nest($x, $n, KEYWORD_WHILE);
934 $n->condition = $this->ParenExpression($x);
935 if (!$x->ecmaStrictMode)
937 // <script language="JavaScript"> (without version hints) may need
938 // automatic semicolon insertion without a newline after do-while.
939 // See http://bugzilla.mozilla.org/show_bug.cgi?id=238945.
940 $this->t->match(OP_SEMICOLON);
941 return $n;
943 break;
945 case KEYWORD_BREAK:
946 case KEYWORD_CONTINUE:
947 $n = new JSNode($this->t);
949 if ($this->t->peekOnSameLine() == TOKEN_IDENTIFIER)
951 $this->t->get();
952 $n->label = $this->t->currentToken()->value;
955 $ss = $x->stmtStack;
956 $i = count($ss);
957 $label = $n->label;
958 if ($label)
962 if (--$i < 0)
963 throw $this->t->newSyntaxError('Label not found');
965 while ($ss[$i]->label != $label);
967 else
971 if (--$i < 0)
972 throw $this->t->newSyntaxError('Invalid ' . $tt);
974 while (!$ss[$i]->isLoop && ($tt != KEYWORD_BREAK || $ss[$i]->type != KEYWORD_SWITCH));
977 $n->target = $ss[$i];
978 break;
980 case KEYWORD_TRY:
981 $n = new JSNode($this->t);
982 $n->tryBlock = $this->Block($x);
983 $n->catchClauses = array();
985 while ($this->t->match(KEYWORD_CATCH))
987 $n2 = new JSNode($this->t);
988 $this->t->mustMatch(OP_LEFT_PAREN);
989 $n2->varName = $this->t->mustMatch(TOKEN_IDENTIFIER)->value;
991 if ($this->t->match(KEYWORD_IF))
993 if ($x->ecmaStrictMode)
994 throw $this->t->newSyntaxError('Illegal catch guard');
996 if (count($n->catchClauses) && !end($n->catchClauses)->guard)
997 throw $this->t->newSyntaxError('Guarded catch after unguarded');
999 $n2->guard = $this->Expression($x);
1001 else
1003 $n2->guard = null;
1006 $this->t->mustMatch(OP_RIGHT_PAREN);
1007 $n2->block = $this->Block($x);
1008 array_push($n->catchClauses, $n2);
1011 if ($this->t->match(KEYWORD_FINALLY))
1012 $n->finallyBlock = $this->Block($x);
1014 if (!count($n->catchClauses) && !$n->finallyBlock)
1015 throw $this->t->newSyntaxError('Invalid try statement');
1016 return $n;
1018 case KEYWORD_CATCH:
1019 case KEYWORD_FINALLY:
1020 throw $this->t->newSyntaxError($tt . ' without preceding try');
1022 case KEYWORD_THROW:
1023 $n = new JSNode($this->t);
1024 $n->value = $this->Expression($x);
1025 break;
1027 case KEYWORD_RETURN:
1028 if (!$x->inFunction)
1029 throw $this->t->newSyntaxError('Invalid return');
1031 $n = new JSNode($this->t);
1032 $tt = $this->t->peekOnSameLine();
1033 if ($tt != TOKEN_END && $tt != TOKEN_NEWLINE && $tt != OP_SEMICOLON && $tt != OP_RIGHT_CURLY)
1034 $n->value = $this->Expression($x);
1035 else
1036 $n->value = null;
1037 break;
1039 case KEYWORD_WITH:
1040 $n = new JSNode($this->t);
1041 $n->object = $this->ParenExpression($x);
1042 $n->body = $this->nest($x, $n);
1043 return $n;
1045 case KEYWORD_VAR:
1046 case KEYWORD_CONST:
1047 $n = $this->Variables($x);
1048 break;
1050 case TOKEN_CONDCOMMENT_START:
1051 case TOKEN_CONDCOMMENT_END:
1052 $n = new JSNode($this->t);
1053 return $n;
1055 case KEYWORD_DEBUGGER:
1056 $n = new JSNode($this->t);
1057 break;
1059 case TOKEN_NEWLINE:
1060 case OP_SEMICOLON:
1061 $n = new JSNode($this->t, OP_SEMICOLON);
1062 $n->expression = null;
1063 return $n;
1065 default:
1066 if ($tt == TOKEN_IDENTIFIER)
1068 $this->t->scanOperand = false;
1069 $tt = $this->t->peek();
1070 $this->t->scanOperand = true;
1071 if ($tt == OP_COLON)
1073 $label = $this->t->currentToken()->value;
1074 $ss = $x->stmtStack;
1075 for ($i = count($ss) - 1; $i >= 0; --$i)
1077 if ($ss[$i]->label == $label)
1078 throw $this->t->newSyntaxError('Duplicate label');
1081 $this->t->get();
1082 $n = new JSNode($this->t, JS_LABEL);
1083 $n->label = $label;
1084 $n->statement = $this->nest($x, $n);
1086 return $n;
1090 $n = new JSNode($this->t, OP_SEMICOLON);
1091 $this->t->unget();
1092 $n->expression = $this->Expression($x);
1093 $n->end = $n->expression->end;
1094 break;
1097 if ($this->t->lineno == $this->t->currentToken()->lineno)
1099 $tt = $this->t->peekOnSameLine();
1100 if ($tt != TOKEN_END && $tt != TOKEN_NEWLINE && $tt != OP_SEMICOLON && $tt != OP_RIGHT_CURLY)
1101 throw $this->t->newSyntaxError('Missing ; before statement');
1104 $this->t->match(OP_SEMICOLON);
1106 return $n;
1109 private function FunctionDefinition($x, $requireName, $functionForm)
1111 $f = new JSNode($this->t);
1113 if ($f->type != KEYWORD_FUNCTION)
1114 $f->type = ($f->value == 'get') ? JS_GETTER : JS_SETTER;
1116 if ($this->t->match(TOKEN_IDENTIFIER))
1117 $f->name = $this->t->currentToken()->value;
1118 elseif ($requireName)
1119 throw $this->t->newSyntaxError('Missing function identifier');
1121 $this->t->mustMatch(OP_LEFT_PAREN);
1122 $f->params = array();
1124 while (($tt = $this->t->get()) != OP_RIGHT_PAREN)
1126 if ($tt != TOKEN_IDENTIFIER)
1127 throw $this->t->newSyntaxError('Missing formal parameter');
1129 array_push($f->params, $this->t->currentToken()->value);
1131 if ($this->t->peek() != OP_RIGHT_PAREN)
1132 $this->t->mustMatch(OP_COMMA);
1135 $this->t->mustMatch(OP_LEFT_CURLY);
1137 $x2 = new JSCompilerContext(true);
1138 $f->body = $this->Script($x2);
1140 $this->t->mustMatch(OP_RIGHT_CURLY);
1141 $f->end = $this->t->currentToken()->end;
1143 $f->functionForm = $functionForm;
1144 if ($functionForm == DECLARED_FORM)
1145 array_push($x->funDecls, $f);
1147 return $f;
1150 private function Variables($x)
1152 $n = new JSNode($this->t);
1156 $this->t->mustMatch(TOKEN_IDENTIFIER);
1158 $n2 = new JSNode($this->t);
1159 $n2->name = $n2->value;
1161 if ($this->t->match(OP_ASSIGN))
1163 if ($this->t->currentToken()->assignOp)
1164 throw $this->t->newSyntaxError('Invalid variable initialization');
1166 $n2->initializer = $this->Expression($x, OP_COMMA);
1169 $n2->readOnly = $n->type == KEYWORD_CONST;
1171 $n->addNode($n2);
1172 array_push($x->varDecls, $n2);
1174 while ($this->t->match(OP_COMMA));
1176 return $n;
1179 private function Expression($x, $stop=false)
1181 $operators = array();
1182 $operands = array();
1183 $n = false;
1185 $bl = $x->bracketLevel;
1186 $cl = $x->curlyLevel;
1187 $pl = $x->parenLevel;
1188 $hl = $x->hookLevel;
1190 while (($tt = $this->t->get()) != TOKEN_END)
1192 if ($tt == $stop &&
1193 $x->bracketLevel == $bl &&
1194 $x->curlyLevel == $cl &&
1195 $x->parenLevel == $pl &&
1196 $x->hookLevel == $hl
1199 // Stop only if tt matches the optional stop parameter, and that
1200 // token is not quoted by some kind of bracket.
1201 break;
1204 switch ($tt)
1206 case OP_SEMICOLON:
1207 // NB: cannot be empty, Statement handled that.
1208 break 2;
1210 case OP_HOOK:
1211 if ($this->t->scanOperand)
1212 break 2;
1214 while ( !empty($operators) &&
1215 $this->opPrecedence[end($operators)->type] > $this->opPrecedence[$tt]
1217 $this->reduce($operators, $operands);
1219 array_push($operators, new JSNode($this->t));
1221 ++$x->hookLevel;
1222 $this->t->scanOperand = true;
1223 $n = $this->Expression($x);
1225 if (!$this->t->match(OP_COLON))
1226 break 2;
1228 --$x->hookLevel;
1229 array_push($operands, $n);
1230 break;
1232 case OP_COLON:
1233 if ($x->hookLevel)
1234 break 2;
1236 throw $this->t->newSyntaxError('Invalid label');
1237 break;
1239 case OP_ASSIGN:
1240 if ($this->t->scanOperand)
1241 break 2;
1243 // Use >, not >=, for right-associative ASSIGN
1244 while ( !empty($operators) &&
1245 $this->opPrecedence[end($operators)->type] > $this->opPrecedence[$tt]
1247 $this->reduce($operators, $operands);
1249 array_push($operators, new JSNode($this->t));
1250 end($operands)->assignOp = $this->t->currentToken()->assignOp;
1251 $this->t->scanOperand = true;
1252 break;
1254 case KEYWORD_IN:
1255 // An in operator should not be parsed if we're parsing the head of
1256 // a for (...) loop, unless it is in the then part of a conditional
1257 // expression, or parenthesized somehow.
1258 if ($x->inForLoopInit && !$x->hookLevel &&
1259 !$x->bracketLevel && !$x->curlyLevel &&
1260 !$x->parenLevel
1262 break 2;
1263 // FALL THROUGH
1264 case OP_COMMA:
1265 // A comma operator should not be parsed if we're parsing the then part
1266 // of a conditional expression unless it's parenthesized somehow.
1267 if ($tt == OP_COMMA && $x->hookLevel &&
1268 !$x->bracketLevel && !$x->curlyLevel &&
1269 !$x->parenLevel
1271 break 2;
1272 // Treat comma as left-associative so reduce can fold left-heavy
1273 // COMMA trees into a single array.
1274 // FALL THROUGH
1275 case OP_OR:
1276 case OP_AND:
1277 case OP_BITWISE_OR:
1278 case OP_BITWISE_XOR:
1279 case OP_BITWISE_AND:
1280 case OP_EQ: case OP_NE: case OP_STRICT_EQ: case OP_STRICT_NE:
1281 case OP_LT: case OP_LE: case OP_GE: case OP_GT:
1282 case KEYWORD_INSTANCEOF:
1283 case OP_LSH: case OP_RSH: case OP_URSH:
1284 case OP_PLUS: case OP_MINUS:
1285 case OP_MUL: case OP_DIV: case OP_MOD:
1286 case OP_DOT:
1287 if ($this->t->scanOperand)
1288 break 2;
1290 while ( !empty($operators) &&
1291 $this->opPrecedence[end($operators)->type] >= $this->opPrecedence[$tt]
1293 $this->reduce($operators, $operands);
1295 if ($tt == OP_DOT)
1297 $this->t->mustMatch(TOKEN_IDENTIFIER);
1298 array_push($operands, new JSNode($this->t, OP_DOT, array_pop($operands), new JSNode($this->t)));
1300 else
1302 array_push($operators, new JSNode($this->t));
1303 $this->t->scanOperand = true;
1305 break;
1307 case KEYWORD_DELETE: case KEYWORD_VOID: case KEYWORD_TYPEOF:
1308 case OP_NOT: case OP_BITWISE_NOT: case OP_UNARY_PLUS: case OP_UNARY_MINUS:
1309 case KEYWORD_NEW:
1310 if (!$this->t->scanOperand)
1311 break 2;
1313 array_push($operators, new JSNode($this->t));
1314 break;
1316 case OP_INCREMENT: case OP_DECREMENT:
1317 if ($this->t->scanOperand)
1319 array_push($operators, new JSNode($this->t)); // prefix increment or decrement
1321 else
1323 // Don't cross a line boundary for postfix {in,de}crement.
1324 $t = $this->t->tokens[($this->t->tokenIndex + $this->t->lookahead - 1) & 3];
1325 if ($t && $t->lineno != $this->t->lineno)
1326 break 2;
1328 if (!empty($operators))
1330 // Use >, not >=, so postfix has higher precedence than prefix.
1331 while ($this->opPrecedence[end($operators)->type] > $this->opPrecedence[$tt])
1332 $this->reduce($operators, $operands);
1335 $n = new JSNode($this->t, $tt, array_pop($operands));
1336 $n->postfix = true;
1337 array_push($operands, $n);
1339 break;
1341 case KEYWORD_FUNCTION:
1342 if (!$this->t->scanOperand)
1343 break 2;
1345 array_push($operands, $this->FunctionDefinition($x, false, EXPRESSED_FORM));
1346 $this->t->scanOperand = false;
1347 break;
1349 case KEYWORD_NULL: case KEYWORD_THIS: case KEYWORD_TRUE: case KEYWORD_FALSE:
1350 case TOKEN_IDENTIFIER: case TOKEN_NUMBER: case TOKEN_STRING: case TOKEN_REGEXP:
1351 if (!$this->t->scanOperand)
1352 break 2;
1354 array_push($operands, new JSNode($this->t));
1355 $this->t->scanOperand = false;
1356 break;
1358 case TOKEN_CONDCOMMENT_START:
1359 case TOKEN_CONDCOMMENT_END:
1360 if ($this->t->scanOperand)
1361 array_push($operators, new JSNode($this->t));
1362 else
1363 array_push($operands, new JSNode($this->t));
1364 break;
1366 case OP_LEFT_BRACKET:
1367 if ($this->t->scanOperand)
1369 // Array initialiser. Parse using recursive descent, as the
1370 // sub-grammar here is not an operator grammar.
1371 $n = new JSNode($this->t, JS_ARRAY_INIT);
1372 while (($tt = $this->t->peek()) != OP_RIGHT_BRACKET)
1374 if ($tt == OP_COMMA)
1376 $this->t->get();
1377 $n->addNode(null);
1378 continue;
1381 $n->addNode($this->Expression($x, OP_COMMA));
1382 if (!$this->t->match(OP_COMMA))
1383 break;
1386 $this->t->mustMatch(OP_RIGHT_BRACKET);
1387 array_push($operands, $n);
1388 $this->t->scanOperand = false;
1390 else
1392 // Property indexing operator.
1393 array_push($operators, new JSNode($this->t, JS_INDEX));
1394 $this->t->scanOperand = true;
1395 ++$x->bracketLevel;
1397 break;
1399 case OP_RIGHT_BRACKET:
1400 if ($this->t->scanOperand || $x->bracketLevel == $bl)
1401 break 2;
1403 while ($this->reduce($operators, $operands)->type != JS_INDEX)
1404 continue;
1406 --$x->bracketLevel;
1407 break;
1409 case OP_LEFT_CURLY:
1410 if (!$this->t->scanOperand)
1411 break 2;
1413 // Object initialiser. As for array initialisers (see above),
1414 // parse using recursive descent.
1415 ++$x->curlyLevel;
1416 $n = new JSNode($this->t, JS_OBJECT_INIT);
1417 while (!$this->t->match(OP_RIGHT_CURLY))
1421 $tt = $this->t->get();
1422 $tv = $this->t->currentToken()->value;
1423 if (($tv == 'get' || $tv == 'set') && $this->t->peek() == TOKEN_IDENTIFIER)
1425 if ($x->ecmaStrictMode)
1426 throw $this->t->newSyntaxError('Illegal property accessor');
1428 $n->addNode($this->FunctionDefinition($x, true, EXPRESSED_FORM));
1430 else
1432 switch ($tt)
1434 case TOKEN_IDENTIFIER:
1435 case TOKEN_NUMBER:
1436 case TOKEN_STRING:
1437 $id = new JSNode($this->t);
1438 break;
1440 case OP_RIGHT_CURLY:
1441 if ($x->ecmaStrictMode)
1442 throw $this->t->newSyntaxError('Illegal trailing ,');
1443 break 3;
1445 default:
1446 throw $this->t->newSyntaxError('Invalid property name');
1449 $this->t->mustMatch(OP_COLON);
1450 $n->addNode(new JSNode($this->t, JS_PROPERTY_INIT, $id, $this->Expression($x, OP_COMMA)));
1453 while ($this->t->match(OP_COMMA));
1455 $this->t->mustMatch(OP_RIGHT_CURLY);
1456 break;
1459 array_push($operands, $n);
1460 $this->t->scanOperand = false;
1461 --$x->curlyLevel;
1462 break;
1464 case OP_RIGHT_CURLY:
1465 if (!$this->t->scanOperand && $x->curlyLevel != $cl)
1466 throw new Exception('PANIC: right curly botch');
1467 break 2;
1469 case OP_LEFT_PAREN:
1470 if ($this->t->scanOperand)
1472 array_push($operators, new JSNode($this->t, JS_GROUP));
1474 else
1476 while ( !empty($operators) &&
1477 $this->opPrecedence[end($operators)->type] > $this->opPrecedence[KEYWORD_NEW]
1479 $this->reduce($operators, $operands);
1481 // Handle () now, to regularize the n-ary case for n > 0.
1482 // We must set scanOperand in case there are arguments and
1483 // the first one is a regexp or unary+/-.
1484 $n = end($operators);
1485 $this->t->scanOperand = true;
1486 if ($this->t->match(OP_RIGHT_PAREN))
1488 if ($n && $n->type == KEYWORD_NEW)
1490 array_pop($operators);
1491 $n->addNode(array_pop($operands));
1493 else
1495 $n = new JSNode($this->t, JS_CALL, array_pop($operands), new JSNode($this->t, JS_LIST));
1498 array_push($operands, $n);
1499 $this->t->scanOperand = false;
1500 break;
1503 if ($n && $n->type == KEYWORD_NEW)
1504 $n->type = JS_NEW_WITH_ARGS;
1505 else
1506 array_push($operators, new JSNode($this->t, JS_CALL));
1509 ++$x->parenLevel;
1510 break;
1512 case OP_RIGHT_PAREN:
1513 if ($this->t->scanOperand || $x->parenLevel == $pl)
1514 break 2;
1516 while (($tt = $this->reduce($operators, $operands)->type) != JS_GROUP &&
1517 $tt != JS_CALL && $tt != JS_NEW_WITH_ARGS
1520 continue;
1523 if ($tt != JS_GROUP)
1525 $n = end($operands);
1526 if ($n->treeNodes[1]->type != OP_COMMA)
1527 $n->treeNodes[1] = new JSNode($this->t, JS_LIST, $n->treeNodes[1]);
1528 else
1529 $n->treeNodes[1]->type = JS_LIST;
1532 --$x->parenLevel;
1533 break;
1535 // Automatic semicolon insertion means we may scan across a newline
1536 // and into the beginning of another statement. If so, break out of
1537 // the while loop and let the t.scanOperand logic handle errors.
1538 default:
1539 break 2;
1543 if ($x->hookLevel != $hl)
1544 throw $this->t->newSyntaxError('Missing : in conditional expression');
1546 if ($x->parenLevel != $pl)
1547 throw $this->t->newSyntaxError('Missing ) in parenthetical');
1549 if ($x->bracketLevel != $bl)
1550 throw $this->t->newSyntaxError('Missing ] in index expression');
1552 if ($this->t->scanOperand)
1553 throw $this->t->newSyntaxError('Missing operand');
1555 // Resume default mode, scanning for operands, not operators.
1556 $this->t->scanOperand = true;
1557 $this->t->unget();
1559 while (count($operators))
1560 $this->reduce($operators, $operands);
1562 return array_pop($operands);
1565 private function ParenExpression($x)
1567 $this->t->mustMatch(OP_LEFT_PAREN);
1568 $n = $this->Expression($x);
1569 $this->t->mustMatch(OP_RIGHT_PAREN);
1571 return $n;
1574 // Statement stack and nested statement handler.
1575 private function nest($x, $node, $end = false)
1577 array_push($x->stmtStack, $node);
1578 $n = $this->statement($x);
1579 array_pop($x->stmtStack);
1581 if ($end)
1582 $this->t->mustMatch($end);
1584 return $n;
1587 private function reduce(&$operators, &$operands)
1589 $n = array_pop($operators);
1590 $op = $n->type;
1591 $arity = $this->opArity[$op];
1592 $c = count($operands);
1593 if ($arity == -2)
1595 // Flatten left-associative trees
1596 if ($c >= 2)
1598 $left = $operands[$c - 2];
1599 if ($left->type == $op)
1601 $right = array_pop($operands);
1602 $left->addNode($right);
1603 return $left;
1606 $arity = 2;
1609 // Always use push to add operands to n, to update start and end
1610 $a = array_splice($operands, $c - $arity);
1611 for ($i = 0; $i < $arity; $i++)
1612 $n->addNode($a[$i]);
1614 // Include closing bracket or postfix operator in [start,end]
1615 $te = $this->t->currentToken()->end;
1616 if ($n->end < $te)
1617 $n->end = $te;
1619 array_push($operands, $n);
1621 return $n;
1625 class JSCompilerContext
1627 public $inFunction = false;
1628 public $inForLoopInit = false;
1629 public $ecmaStrictMode = false;
1630 public $bracketLevel = 0;
1631 public $curlyLevel = 0;
1632 public $parenLevel = 0;
1633 public $hookLevel = 0;
1635 public $stmtStack = array();
1636 public $funDecls = array();
1637 public $varDecls = array();
1639 public function __construct($inFunction)
1641 $this->inFunction = $inFunction;
1645 class JSNode
1647 private $type;
1648 private $value;
1649 private $lineno;
1650 private $start;
1651 private $end;
1653 public $treeNodes = array();
1654 public $funDecls = array();
1655 public $varDecls = array();
1657 public function __construct($t, $type=0)
1659 if ($token = $t->currentToken())
1661 $this->type = $type ? $type : $token->type;
1662 $this->value = $token->value;
1663 $this->lineno = $token->lineno;
1664 $this->start = $token->start;
1665 $this->end = $token->end;
1667 else
1669 $this->type = $type;
1670 $this->lineno = $t->lineno;
1673 if (($numargs = func_num_args()) > 2)
1675 $args = func_get_args();
1676 for ($i = 2; $i < $numargs; $i++)
1677 $this->addNode($args[$i]);
1681 // we don't want to bloat our object with all kind of specific properties, so we use overloading
1682 public function __set($name, $value)
1684 $this->$name = $value;
1687 public function __get($name)
1689 if (isset($this->$name))
1690 return $this->$name;
1692 return null;
1695 public function addNode($node)
1697 if ($node !== null)
1699 if ($node->start < $this->start)
1700 $this->start = $node->start;
1701 if ($this->end < $node->end)
1702 $this->end = $node->end;
1705 $this->treeNodes[] = $node;
1709 class JSTokenizer
1711 private $cursor = 0;
1712 private $source;
1714 public $tokens = array();
1715 public $tokenIndex = 0;
1716 public $lookahead = 0;
1717 public $scanNewlines = false;
1718 public $scanOperand = true;
1720 public $filename;
1721 public $lineno;
1723 private $keywords = array(
1724 'break',
1725 'case', 'catch', 'const', 'continue',
1726 'debugger', 'default', 'delete', 'do',
1727 'else', 'enum',
1728 'false', 'finally', 'for', 'function',
1729 'if', 'in', 'instanceof',
1730 'new', 'null',
1731 'return',
1732 'switch',
1733 'this', 'throw', 'true', 'try', 'typeof',
1734 'var', 'void',
1735 'while', 'with'
1738 private $opTypeNames = array(
1739 ';', ',', '?', ':', '||', '&&', '|', '^',
1740 '&', '===', '==', '=', '!==', '!=', '<<', '<=',
1741 '<', '>>>', '>>', '>=', '>', '++', '--', '+',
1742 '-', '*', '/', '%', '!', '~', '.', '[',
1743 ']', '{', '}', '(', ')', '@*/'
1746 private $assignOps = array('|', '^', '&', '<<', '>>', '>>>', '+', '-', '*', '/', '%');
1747 private $opRegExp;
1749 public function __construct()
1751 $this->opRegExp = '#^(' . implode('|', array_map('preg_quote', $this->opTypeNames)) . ')#';
1754 public function init($source, $filename = '', $lineno = 1)
1756 $this->source = $source;
1757 $this->filename = $filename ? $filename : '[inline]';
1758 $this->lineno = $lineno;
1760 $this->cursor = 0;
1761 $this->tokens = array();
1762 $this->tokenIndex = 0;
1763 $this->lookahead = 0;
1764 $this->scanNewlines = false;
1765 $this->scanOperand = true;
1768 public function getInput($chunksize)
1770 if ($chunksize)
1771 return substr($this->source, $this->cursor, $chunksize);
1773 return substr($this->source, $this->cursor);
1776 public function isDone()
1778 return $this->peek() == TOKEN_END;
1781 public function match($tt)
1783 return $this->get() == $tt || $this->unget();
1786 public function mustMatch($tt)
1788 if (!$this->match($tt))
1789 throw $this->newSyntaxError('Unexpected token; token ' . $tt . ' expected');
1791 return $this->currentToken();
1794 public function peek()
1796 if ($this->lookahead)
1798 $next = $this->tokens[($this->tokenIndex + $this->lookahead) & 3];
1799 if ($this->scanNewlines && $next->lineno != $this->lineno)
1800 $tt = TOKEN_NEWLINE;
1801 else
1802 $tt = $next->type;
1804 else
1806 $tt = $this->get();
1807 $this->unget();
1810 return $tt;
1813 public function peekOnSameLine()
1815 $this->scanNewlines = true;
1816 $tt = $this->peek();
1817 $this->scanNewlines = false;
1819 return $tt;
1822 public function currentToken()
1824 if (!empty($this->tokens))
1825 return $this->tokens[$this->tokenIndex];
1828 public function get($chunksize = 1000)
1830 while($this->lookahead)
1832 $this->lookahead--;
1833 $this->tokenIndex = ($this->tokenIndex + 1) & 3;
1834 $token = $this->tokens[$this->tokenIndex];
1835 if ($token->type != TOKEN_NEWLINE || $this->scanNewlines)
1836 return $token->type;
1839 $conditional_comment = false;
1841 // strip whitespace and comments
1842 while(true)
1844 $input = $this->getInput($chunksize);
1846 // whitespace handling; gobble up \r as well (effectively we don't have support for MAC newlines!)
1847 $re = $this->scanNewlines ? '/^[ \r\t]+/' : '/^\s+/';
1848 if (preg_match($re, $input, $match))
1850 $spaces = $match[0];
1851 $spacelen = strlen($spaces);
1852 $this->cursor += $spacelen;
1853 if (!$this->scanNewlines)
1854 $this->lineno += substr_count($spaces, "\n");
1856 if ($spacelen == $chunksize)
1857 continue; // complete chunk contained whitespace
1859 $input = $this->getInput($chunksize);
1860 if ($input == '' || $input[0] != '/')
1861 break;
1864 // Comments
1865 if (!preg_match('/^\/(?:\*(@(?:cc_on|if|elif|else|end))?.*?\*\/|\/[^\n]*)/s', $input, $match))
1867 if (!$chunksize)
1868 break;
1870 // retry with a full chunk fetch; this also prevents breakage of long regular expressions (which will never match a comment)
1871 $chunksize = null;
1872 continue;
1875 // check if this is a conditional (JScript) comment
1876 if (!empty($match[1]))
1878 $match[0] = '/*' . $match[1];
1879 $conditional_comment = true;
1880 break;
1882 else
1884 $this->cursor += strlen($match[0]);
1885 $this->lineno += substr_count($match[0], "\n");
1889 if ($input == '')
1891 $tt = TOKEN_END;
1892 $match = array('');
1894 elseif ($conditional_comment)
1896 $tt = TOKEN_CONDCOMMENT_START;
1898 else
1900 switch ($input[0])
1902 case '0':
1903 // hexadecimal
1904 if (($input[1] == 'x' || $input[1] == 'X') && preg_match('/^0x[0-9a-f]+/i', $input, $match))
1906 $tt = TOKEN_NUMBER;
1907 break;
1909 // FALL THROUGH
1911 case '1': case '2': case '3': case '4': case '5':
1912 case '6': case '7': case '8': case '9':
1913 // should always match
1914 preg_match('/^\d+(?:\.\d*)?(?:[eE][-+]?\d+)?/', $input, $match);
1915 $tt = TOKEN_NUMBER;
1916 break;
1918 case "'":
1919 if (preg_match('/^\'(?:[^\\\\\'\r\n]++|\\\\(?:.|\r?\n))*\'/', $input, $match))
1921 $tt = TOKEN_STRING;
1923 else
1925 if ($chunksize)
1926 return $this->get(null); // retry with a full chunk fetch
1928 throw $this->newSyntaxError('Unterminated string literal');
1930 break;
1932 case '"':
1933 if (preg_match('/^"(?:[^\\\\"\r\n]++|\\\\(?:.|\r?\n))*"/', $input, $match))
1935 $tt = TOKEN_STRING;
1937 else
1939 if ($chunksize)
1940 return $this->get(null); // retry with a full chunk fetch
1942 throw $this->newSyntaxError('Unterminated string literal');
1944 break;
1946 case '/':
1947 if ($this->scanOperand && preg_match('/^\/((?:\\\\.|\[(?:\\\\.|[^\]])*\]|[^\/])+)\/([gimy]*)/', $input, $match))
1949 $tt = TOKEN_REGEXP;
1950 break;
1952 // FALL THROUGH
1954 case '|':
1955 case '^':
1956 case '&':
1957 case '<':
1958 case '>':
1959 case '+':
1960 case '-':
1961 case '*':
1962 case '%':
1963 case '=':
1964 case '!':
1965 // should always match
1966 preg_match($this->opRegExp, $input, $match);
1967 $op = $match[0];
1968 if (in_array($op, $this->assignOps) && $input[strlen($op)] == '=')
1970 $tt = OP_ASSIGN;
1971 $match[0] .= '=';
1973 else
1975 $tt = $op;
1976 if ($this->scanOperand)
1978 if ($op == OP_PLUS)
1979 $tt = OP_UNARY_PLUS;
1980 elseif ($op == OP_MINUS)
1981 $tt = OP_UNARY_MINUS;
1983 $op = null;
1985 break;
1987 case '.':
1988 if (preg_match('/^\.\d+(?:[eE][-+]?\d+)?/', $input, $match))
1990 $tt = TOKEN_NUMBER;
1991 break;
1993 // FALL THROUGH
1995 case ';':
1996 case ',':
1997 case '?':
1998 case ':':
1999 case '~':
2000 case '[':
2001 case ']':
2002 case '{':
2003 case '}':
2004 case '(':
2005 case ')':
2006 // these are all single
2007 $match = array($input[0]);
2008 $tt = $input[0];
2009 break;
2011 case '@':
2012 // check end of conditional comment
2013 if (substr($input, 0, 3) == '@*/')
2015 $match = array('@*/');
2016 $tt = TOKEN_CONDCOMMENT_END;
2018 else
2019 throw $this->newSyntaxError('Illegal token');
2020 break;
2022 case "\n":
2023 if ($this->scanNewlines)
2025 $match = array("\n");
2026 $tt = TOKEN_NEWLINE;
2028 else
2029 throw $this->newSyntaxError('Illegal token');
2030 break;
2032 default:
2033 // Fast path for identifiers: word chars followed by whitespace or various other tokens.
2034 // Note we don't need to exclude digits in the first char, as they've already been found
2035 // above.
2036 if (!preg_match('/^[$\w]+(?=[\s\/\|\^\&<>\+\-\*%=!.;,\?:~\[\]\{\}\(\)@])/', $input, $match))
2038 // Character classes per ECMA-262 edition 5.1 section 7.6
2039 // Per spec, must accept Unicode 3.0, *may* accept later versions.
2040 // We'll take whatever PCRE understands, which should be more recent.
2041 $identifierStartChars = "\\p{L}\\p{Nl}" . # UnicodeLetter
2042 "\$" .
2043 "_";
2044 $identifierPartChars = $identifierStartChars .
2045 "\\p{Mn}\\p{Mc}" . # UnicodeCombiningMark
2046 "\\p{Nd}" . # UnicodeDigit
2047 "\\p{Pc}"; # UnicodeConnectorPunctuation
2048 $unicodeEscape = "\\\\u[0-9A-F-a-f]{4}";
2049 $identifierRegex = "/^" .
2050 "(?:[$identifierStartChars]|$unicodeEscape)" .
2051 "(?:[$identifierPartChars]|$unicodeEscape)*" .
2052 "/uS";
2053 if (preg_match($identifierRegex, $input, $match))
2055 if (strpos($match[0], '\\') !== false) {
2056 // Per ECMA-262 edition 5.1, section 7.6 escape sequences should behave as if they were
2057 // the original chars, but only within the boundaries of the identifier.
2058 $decoded = preg_replace_callback('/\\\\u([0-9A-Fa-f]{4})/',
2059 array(__CLASS__, 'unicodeEscapeCallback'),
2060 $match[0]);
2062 // Since our original regex didn't de-escape the originals, we need to check for validity again.
2063 // No need to worry about token boundaries, as anything outside the identifier is illegal!
2064 if (!preg_match("/^[$identifierStartChars][$identifierPartChars]*$/u", $decoded)) {
2065 throw $this->newSyntaxError('Illegal token');
2068 // Per spec it _ought_ to work to use these escapes for keywords words as well...
2069 // but IE rejects them as invalid, while Firefox and Chrome treat them as identifiers
2070 // that don't match the keyword.
2071 if (in_array($decoded, $this->keywords)) {
2072 throw $this->newSyntaxError('Illegal token');
2075 // TODO: save the decoded form for output?
2078 else
2079 throw $this->newSyntaxError('Illegal token');
2081 $tt = in_array($match[0], $this->keywords) ? $match[0] : TOKEN_IDENTIFIER;
2085 $this->tokenIndex = ($this->tokenIndex + 1) & 3;
2087 if (!isset($this->tokens[$this->tokenIndex]))
2088 $this->tokens[$this->tokenIndex] = new JSToken();
2090 $token = $this->tokens[$this->tokenIndex];
2091 $token->type = $tt;
2093 if ($tt == OP_ASSIGN)
2094 $token->assignOp = $op;
2096 $token->start = $this->cursor;
2098 $token->value = $match[0];
2099 $this->cursor += strlen($match[0]);
2101 $token->end = $this->cursor;
2102 $token->lineno = $this->lineno;
2104 return $tt;
2107 public function unget()
2109 if (++$this->lookahead == 4)
2110 throw $this->newSyntaxError('PANIC: too much lookahead!');
2112 $this->tokenIndex = ($this->tokenIndex - 1) & 3;
2115 public function newSyntaxError($m)
2117 return new Exception('Parse error: ' . $m . ' in file \'' . $this->filename . '\' on line ' . $this->lineno);
2120 public static function unicodeEscapeCallback($m)
2122 return html_entity_decode('&#x' . $m[1]. ';', ENT_QUOTES, 'UTF-8');
2126 class JSToken
2128 public $type;
2129 public $value;
2130 public $start;
2131 public $end;
2132 public $lineno;
2133 public $assignOp;