source: trunk/phpgwapi/inc/class.JavaScriptPacker.php @ 2537

<?php
/* 9 April 2008. version 1.1
 * 
 * This is the php version of the Dean Edwards JavaScript's Packer,
 * Based on :
 * 
 * ParseMaster, version 1.0.2 (2005-08-19) Copyright 2005, Dean Edwards
 * a multi-pattern parser.
 * KNOWN BUG: erroneous behavior when using escapeChar with a replacement
 * value that is a function
 * 
 * packer, version 2.0.2 (2005-08-19) Copyright 2004-2005, Dean Edwards
 * 
 * License: http://creativecommons.org/licenses/LGPL/2.1/
 * 
 * Ported to PHP by Nicolas Martin.
 * 
 * ----------------------------------------------------------------------
 * changelog:
 * 1.1 : correct a bug, '\0' packed then unpacked becomes '\'.
 * ----------------------------------------------------------------------
 * 
 * examples of usage :
 * $myPacker = new JavaScriptPacker($script, 62, true, false);
 * $packed = $myPacker->pack();
 * 
 * or
 * 
 * $myPacker = new JavaScriptPacker($script, 'Normal', true, false);
 * $packed = $myPacker->pack();
 * 
 * or (default values)
 * 
 * $myPacker = new JavaScriptPacker($script);
 * $packed = $myPacker->pack();
 * 
 * 
 * params of the constructor :
 * $script:       the JavaScript to pack, string.
 * $encoding:     level of encoding, int or string :
 *                0,10,62,95 or 'None', 'Numeric', 'Normal', 'High ASCII'.
 *                default: 62.
 * $fastDecode:   include the fast decoder in the packed result, boolean.
 *                default : true.
 * $specialChars: if you are flagged your private and local variables
 *                in the script, boolean.
 *                default: false.
 * 
 * The pack() method return the compressed JavasScript, as a string.
 * 
 * see http://dean.edwards.name/packer/usage/ for more information.
 * 
 * Notes :
 * # need PHP 5 . Tested with PHP 5.1.2, 5.1.3, 5.1.4, 5.2.3
 * 
 * # The packed result may be different than with the Dean Edwards
 *   version, but with the same length. The reason is that the PHP
 *   function usort to sort array don't necessarily preserve the
 *   original order of two equal member. The Javascript sort function
 *   in fact preserve this order (but that's not require by the
 *   ECMAScript standard). So the encoded keywords order can be
 *   different in the two results.
 * 
 * # Be careful with the 'High ASCII' Level encoding if you use
 *   UTF-8 in your files... 
 */


class JavaScriptPacker {
        // constants
        const IGNORE = '$1';

        // validate parameters
        private $_script = '';
        private $_encoding = 62;
        private $_fastDecode = true;
        private $_specialChars = false;
        
        private $LITERAL_ENCODING = array(
                'None' => 0,
                'Numeric' => 10,
                'Normal' => 62,
                'High ASCII' => 95
        );
        
        public function __construct($_script, $_encoding = 62, $_fastDecode = true, $_specialChars = false)
        {
                $this->_script = $_script . "\n";
                if (array_key_exists($_encoding, $this->LITERAL_ENCODING))
                        $_encoding = $this->LITERAL_ENCODING[$_encoding];
                $this->_encoding = min((int)$_encoding, 95);
                $this->_fastDecode = $_fastDecode;      
                $this->_specialChars = $_specialChars;
        }
        
        public function pack() {
                $this->_addParser('_basicCompression');
                if ($this->_specialChars)
                        $this->_addParser('_encodeSpecialChars');
                if ($this->_encoding)
                        $this->_addParser('_encodeKeywords');
                
                // go!
                return $this->_pack($this->_script);
        }
        
        // apply all parsing routines
        private function _pack($script) {
                for ($i = 0; isset($this->_parsers[$i]); $i++) {
                        $script = call_user_func(array(&$this,$this->_parsers[$i]), $script);
                }
                return $script;
        }
        
        // keep a list of parsing functions, they'll be executed all at once
        private $_parsers = array();
        private function _addParser($parser) {
                $this->_parsers[] = $parser;
        }
        
        // zero encoding - just removal of white space and comments
        private function _basicCompression($script) {
                $parser = new ParseMaster();
                // make safe
                $parser->escapeChar = '\\';
                // protect strings
                $parser->add('/\'[^\'\\n\\r]*\'/', self::IGNORE);
                $parser->add('/"[^"\\n\\r]*"/', self::IGNORE);
                // remove comments
                $parser->add('/\\/\\/[^\\n\\r]*[\\n\\r]/', ' ');
                $parser->add('/\\/\\*[^*]*\\*+([^\\/][^*]*\\*+)*\\//', ' ');
                // protect regular expressions
                $parser->add('/\\s+(\\/[^\\/\\n\\r\\*][^\\/\\n\\r]*\\/g?i?)/', '$2'); // IGNORE
                $parser->add('/[^\\w\\x24\\/\'"*)\\?:]\\/[^\\/\\n\\r\\*][^\\/\\n\\r]*\\/g?i?/', self::IGNORE);
                // remove: ;;; doSomething();
                if ($this->_specialChars) $parser->add('/;;;[^\\n\\r]+[\\n\\r]/');
                // remove redundant semi-colons
                $parser->add('/\\(;;\\)/', self::IGNORE); // protect for (;;) loops
                $parser->add('/;+\\s*([};])/', '$2');
                // apply the above
                $script = $parser->exec($script);

                // remove white-space
                $parser->add('/(\\b|\\x24)\\s+(\\b|\\x24)/', '$2 $3');
                $parser->add('/([+\\-])\\s+([+\\-])/', '$2 $3');
                $parser->add('/\\s+/', '');
                // done
                return $parser->exec($script);
        }
        
        private function _encodeSpecialChars($script) {
                $parser = new ParseMaster();
                // replace: $name -> n, $$name -> na
                $parser->add('/((\\x24+)([a-zA-Z$_]+))(\\d*)/',
                                         array('fn' => '_replace_name')
                );
                // replace: _name -> _0, double-underscore (__name) is ignored
                $regexp = '/\\b_[A-Za-z\\d]\\w*/';
                // build the word list
                $keywords = $this->_analyze($script, $regexp, '_encodePrivate');
                // quick ref
                $encoded = $keywords['encoded'];
                
                $parser->add($regexp,
                        array(
                                'fn' => '_replace_encoded',
                                'data' => $encoded
                        )
                );
                return $parser->exec($script);
        }
        
        private function _encodeKeywords($script) {
                // escape high-ascii values already in the script (i.e. in strings)
                if ($this->_encoding > 62)
                        $script = $this->_escape95($script);
                // create the parser
                $parser = new ParseMaster();
                $encode = $this->_getEncoder($this->_encoding);
                // for high-ascii, don't encode single character low-ascii
                $regexp = ($this->_encoding > 62) ? '/\\w\\w+/' : '/\\w+/';
                // build the word list
                $keywords = $this->_analyze($script, $regexp, $encode);
                $encoded = $keywords['encoded'];
                
                // encode
                $parser->add($regexp,
                        array(
                                'fn' => '_replace_encoded',
                                'data' => $encoded
                        )
                );
                if (empty($script)) return $script;
                else {
                        //$res = $parser->exec($script);
                        //$res = $this->_bootStrap($res, $keywords);
                        //return $res;
                        return $this->_bootStrap($parser->exec($script), $keywords);
                }
        }
        
        private function _analyze($script, $regexp, $encode) {
                // analyse
                // retreive all words in the script
                $all = array();
                preg_match_all($regexp, $script, $all);
                $_sorted = array(); // list of words sorted by frequency
                $_encoded = array(); // dictionary of word->encoding
                $_protected = array(); // instances of "protected" words
                $all = $all[0]; // simulate the javascript comportement of global match
                if (!empty($all)) {
                        $unsorted = array(); // same list, not sorted
                        $protected = array(); // "protected" words (dictionary of word->"word")
                        $value = array(); // dictionary of charCode->encoding (eg. 256->ff)
                        $this->_count = array(); // word->count
                        $i = count($all); $j = 0; //$word = null;
                        // count the occurrences - used for sorting later
                        do {
                                --$i;
                                $word = '$' . $all[$i];
                                if (!isset($this->_count[$word])) {
                                        $this->_count[$word] = 0;
                                        $unsorted[$j] = $word;
                                        // make a dictionary of all of the protected words in this script
                                        //  these are words that might be mistaken for encoding
                                        //if (is_string($encode) && method_exists($this, $encode))
                                        $values[$j] = call_user_func(array(&$this, $encode), $j);
                                        $protected['$' . $values[$j]] = $j++;
                                }
                                // increment the word counter
                                $this->_count[$word]++;
                        } while ($i > 0);
                        // prepare to sort the word list, first we must protect
                        //  words that are also used as codes. we assign them a code
                        //  equivalent to the word itself.
                        // e.g. if "do" falls within our encoding range
                        //      then we store keywords["do"] = "do";
                        // this avoids problems when decoding
                        $i = count($unsorted);
                        do {
                                $word = $unsorted[--$i];
                                if (isset($protected[$word]) /*!= null*/) {
                                        $_sorted[$protected[$word]] = substr($word, 1);
                                        $_protected[$protected[$word]] = true;
                                        $this->_count[$word] = 0;
                                }
                        } while ($i);
                        
                        // sort the words by frequency
                        // Note: the javascript and php version of sort can be different :
                        // in php manual, usort :
                        // " If two members compare as equal,
                        // their order in the sorted array is undefined."
                        // so the final packed script is different of the Dean's javascript version
                        // but equivalent.
                        // the ECMAscript standard does not guarantee this behaviour,
                        // and thus not all browsers (e.g. Mozilla versions dating back to at
                        // least 2003) respect this. 
                        usort($unsorted, array(&$this, '_sortWords'));
                        $j = 0;
                        // because there are "protected" words in the list
                        //  we must add the sorted words around them
                        do {
                                if (!isset($_sorted[$i]))
                                        $_sorted[$i] = substr($unsorted[$j++], 1);
                                $_encoded[$_sorted[$i]] = $values[$i];
                        } while (++$i < count($unsorted));
                }
                return array(
                        'sorted'  => $_sorted,
                        'encoded' => $_encoded,
                        'protected' => $_protected);
        }
        
        private $_count = array();
        private function _sortWords($match1, $match2) {
                return $this->_count[$match2] - $this->_count[$match1];
        }
        
        // build the boot function used for loading and decoding
        private function _bootStrap($packed, $keywords) {
                $ENCODE = $this->_safeRegExp('$encode\\($count\\)');

                // $packed: the packed script
                $packed = "'" . $this->_escape($packed) . "'";

                // $ascii: base for encoding
                $ascii = min(count($keywords['sorted']), $this->_encoding);
                if ($ascii == 0) $ascii = 1;

                // $count: number of words contained in the script
                $count = count($keywords['sorted']);

                // $keywords: list of words contained in the script
                foreach ($keywords['protected'] as $i=>$value) {
                        $keywords['sorted'][$i] = '';
                }
                // convert from a string to an array
                ksort($keywords['sorted']);
                $keywords = "'" . implode('|',$keywords['sorted']) . "'.split('|')";

                $encode = ($this->_encoding > 62) ? '_encode95' : $this->_getEncoder($ascii);
                $encode = $this->_getJSFunction($encode);
                $encode = preg_replace('/_encoding/','$ascii', $encode);
                $encode = preg_replace('/arguments\\.callee/','$encode', $encode);
                $inline = '\\$count' . ($ascii > 10 ? '.toString(\\$ascii)' : '');

                // $decode: code snippet to speed up decoding
                if ($this->_fastDecode) {
                        // create the decoder
                        $decode = $this->_getJSFunction('_decodeBody');
                        if ($this->_encoding > 62)
                                $decode = preg_replace('/\\\\w/', '[\\xa1-\\xff]', $decode);
                        // perform the encoding inline for lower ascii values
                        elseif ($ascii < 36)
                                $decode = preg_replace($ENCODE, $inline, $decode);
                        // special case: when $count==0 there are no keywords. I want to keep
                        //  the basic shape of the unpacking funcion so i'll frig the code...
                        if ($count == 0)
                                $decode = preg_replace($this->_safeRegExp('($count)\\s*=\\s*1'), '$1=0', $decode, 1);
                }

                // boot function
                $unpack = $this->_getJSFunction('_unpack');
                if ($this->_fastDecode) {
                        // insert the decoder
                        $this->buffer = $decode;
                        $unpack = preg_replace_callback('/\\{/', array(&$this, '_insertFastDecode'), $unpack, 1);
                }
                $unpack = preg_replace('/"/', "'", $unpack);
                if ($this->_encoding > 62) { // high-ascii
                        // get rid of the word-boundaries for regexp matches
                        $unpack = preg_replace('/\'\\\\\\\\b\'\s*\\+|\\+\s*\'\\\\\\\\b\'/', '', $unpack);
                }
                if ($ascii > 36 || $this->_encoding > 62 || $this->_fastDecode) {
                        // insert the encode function
                        $this->buffer = $encode;
                        $unpack = preg_replace_callback('/\\{/', array(&$this, '_insertFastEncode'), $unpack, 1);
                } else {
                        // perform the encoding inline
                        $unpack = preg_replace($ENCODE, $inline, $unpack);
                }
                // pack the boot function too
                $unpackPacker = new JavaScriptPacker($unpack, 0, false, true);
                $unpack = $unpackPacker->pack();
                
                // arguments
                $params = array($packed, $ascii, $count, $keywords);
                if ($this->_fastDecode) {
                        $params[] = 0;
                        $params[] = '{}';
                }
                $params = implode(',', $params);
                
                // the whole thing
                return 'eval(' . $unpack . '(' . $params . "))\n";
        }
        
        private $buffer;
        private function _insertFastDecode($match) {
                return '{' . $this->buffer . ';';
        }
        private function _insertFastEncode($match) {
                return '{$encode=' . $this->buffer . ';';
        }
        
        // mmm.. ..which one do i need ??
        private function _getEncoder($ascii) {
                return $ascii > 10 ? $ascii > 36 ? $ascii > 62 ?
                       '_encode95' : '_encode62' : '_encode36' : '_encode10';
        }
        
        // zero encoding
        // characters: 0123456789
        private function _encode10($charCode) {
                return $charCode;
        }
        
        // inherent base36 support
        // characters: 0123456789abcdefghijklmnopqrstuvwxyz
        private function _encode36($charCode) {
                return base_convert($charCode, 10, 36);
        }
        
        // hitch a ride on base36 and add the upper case alpha characters
        // characters: 0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ
        private function _encode62($charCode) {
                $res = '';
                if ($charCode >= $this->_encoding) {
                        $res = $this->_encode62((int)($charCode / $this->_encoding));
                }
                $charCode = $charCode % $this->_encoding;
                
                if ($charCode > 35)
                        return $res . chr($charCode + 29);
                else
                        return $res . base_convert($charCode, 10, 36);
        }
        
        // use high-ascii values
        // characters: ¡¢£€¥Š§š©ª«¬­®¯°±²³Žµ¶·ž¹º»ŒœŸ¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßà áâãÀåÊçÚéêëìíîïðñòóÎõö÷ÞùúûÌÜß
        private function _encode95($charCode) {
                $res = '';
                if ($charCode >= $this->_encoding)
                        $res = $this->_encode95($charCode / $this->_encoding);
                
                return $res . chr(($charCode % $this->_encoding) + 161);
        }
        
        private function _safeRegExp($string) {
                return '/'.preg_replace('/\$/', '\\\$', $string).'/';
        }
        
        private function _encodePrivate($charCode) {
                return "_" . $charCode;
        }
        
        // protect characters used by the parser
        private function _escape($script) {
                return preg_replace('/([\\\\\'])/', '\\\$1', $script);
        }
        
        // protect high-ascii characters already in the script
        private function _escape95($script) {
                return preg_replace_callback(
                        '/[\\xa1-\\xff]/',
                        array(&$this, '_escape95Bis'),
                        $script
                );
        }
        private function _escape95Bis($match) {
                return '\x'.((string)dechex(ord($match)));
        }
        
        
        private function _getJSFunction($aName) {
                if (defined('self::JSFUNCTION'.$aName))
                        return constant('self::JSFUNCTION'.$aName);
                else 
                        return '';
        }
        
        // JavaScript Functions used.
        // Note : In Dean's version, these functions are converted
        // with 'String(aFunctionName);'.
        // This internal conversion complete the original code, ex :
        // 'while (aBool) anAction();' is converted to
        // 'while (aBool) { anAction(); }'.
        // The JavaScript functions below are corrected.
        
        // unpacking function - this is the boot strap function
        //  data extracted from this packing routine is passed to
        //  this function when decoded in the target
        // NOTE ! : without the ';' final.
        const JSFUNCTION_unpack =

'function($packed, $ascii, $count, $keywords, $encode, $decode) {
    while ($count--) {
        if ($keywords[$count]) {
            $packed = $packed.replace(new RegExp(\'\\\\b\' + $encode($count) + \'\\\\b\', \'g\'), $keywords[$count]);
        }
    }
    return $packed;
}';
/*
'function($packed, $ascii, $count, $keywords, $encode, $decode) {
    while ($count--)
        if ($keywords[$count])
            $packed = $packed.replace(new RegExp(\'\\\\b\' + $encode($count) + \'\\\\b\', \'g\'), $keywords[$count]);
    return $packed;
}';
*/
        
        // code-snippet inserted into the unpacker to speed up decoding
        const JSFUNCTION_decodeBody =
//_decode = function() {
// does the browser support String.replace where the
//  replacement value is a function?

'    if (!\'\'.replace(/^/, String)) {
        // decode all the values we need
        while ($count--) {
            $decode[$encode($count)] = $keywords[$count] || $encode($count);
        }
        // global replacement function
        $keywords = [function ($encoded) {return $decode[$encoded]}];
        // generic match
        $encode = function () {return \'\\\\w+\'};
        // reset the loop counter -  we are now doing a global replace
        $count = 1;
    }
';
//};
/*
'       if (!\'\'.replace(/^/, String)) {
        // decode all the values we need
        while ($count--) $decode[$encode($count)] = $keywords[$count] || $encode($count);
        // global replacement function
        $keywords = [function ($encoded) {return $decode[$encoded]}];
        // generic match
        $encode = function () {return\'\\\\w+\'};
        // reset the loop counter -  we are now doing a global replace
        $count = 1;
    }';
*/
        
         // zero encoding
         // characters: 0123456789
         const JSFUNCTION_encode10 =
'function($charCode) {
    return $charCode;
}';//;';
        
         // inherent base36 support
         // characters: 0123456789abcdefghijklmnopqrstuvwxyz
         const JSFUNCTION_encode36 =
'function($charCode) {
    return $charCode.toString(36);
}';//;';
        
        // hitch a ride on base36 and add the upper case alpha characters
        // characters: 0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ
        const JSFUNCTION_encode62 =
'function($charCode) {
    return ($charCode < _encoding ? \'\' : arguments.callee(parseInt($charCode / _encoding))) +
    (($charCode = $charCode % _encoding) > 35 ? String.fromCharCode($charCode + 29) : $charCode.toString(36));
}';
        
        // use high-ascii values
        // characters: ¡¢£€¥Š§š©ª«¬­®¯°±²³Žµ¶·ž¹º»ŒœŸ¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßà áâãÀåÊçÚéêëìíîïðñòóÎõö÷ÞùúûÌÜß
        const JSFUNCTION_encode95 =
'function($charCode) {
    return ($charCode < _encoding ? \'\' : arguments.callee($charCode / _encoding)) +
        String.fromCharCode($charCode % _encoding + 161);
}'; 
        
}


class ParseMaster {
        public $ignoreCase = false;
        public $escapeChar = '';
        
        // constants
        const EXPRESSION = 0;
        const REPLACEMENT = 1;
        const LENGTH = 2;
        
        // used to determine nesting levels
        private $GROUPS = '/\\(/';//g
        private $SUB_REPLACE = '/\\$\\d/';
        private $INDEXED = '/^\\$\\d+$/';
        private $TRIM = '/([\'"])\\1\\.(.*)\\.\\1\\1$/';
        private $ESCAPE = '/\\\./';//g
        private $QUOTE = '/\'/';
        private $DELETED = '/\\x01[^\\x01]*\\x01/';//g
        
        public function add($expression, $replacement = '') {
                // count the number of sub-expressions
                //  - add one because each pattern is itself a sub-expression
                $length = 1 + preg_match_all($this->GROUPS, $this->_internalEscape((string)$expression), $out);
                
                // treat only strings $replacement
                if (is_string($replacement)) {
                        // does the pattern deal with sub-expressions?
                        if (preg_match($this->SUB_REPLACE, $replacement)) {
                                // a simple lookup? (e.g. "$2")
                                if (preg_match($this->INDEXED, $replacement)) {
                                        // store the index (used for fast retrieval of matched strings)
                                        $replacement = (int)(substr($replacement, 1)) - 1;
                                } else { // a complicated lookup (e.g. "Hello $2 $1")
                                        // build a function to do the lookup
                                        $quote = preg_match($this->QUOTE, $this->_internalEscape($replacement))
                                                 ? '"' : "'";
                                        $replacement = array(
                                                'fn' => '_backReferences',
                                                'data' => array(
                                                        'replacement' => $replacement,
                                                        'length' => $length,
                                                        'quote' => $quote
                                                )
                                        );
                                }
                        }
                }
                // pass the modified arguments
                if (!empty($expression)) $this->_add($expression, $replacement, $length);
                else $this->_add('/^$/', $replacement, $length);
        }
        
        public function exec($string) {
                // execute the global replacement
                $this->_escaped = array();
                
                // simulate the _patterns.toSTring of Dean
                $regexp = '/';
                foreach ($this->_patterns as $reg) {
                        $regexp .= '(' . substr($reg[self::EXPRESSION], 1, -1) . ')|';
                }
                $regexp = substr($regexp, 0, -1) . '/';
                $regexp .= ($this->ignoreCase) ? 'i' : '';
                
                $string = $this->_escape($string, $this->escapeChar);
                $string = preg_replace_callback(
                        $regexp,
                        array(
                                &$this,
                                '_replacement'
                        ),
                        $string
                );
                $string = $this->_unescape($string, $this->escapeChar);
                
                return preg_replace($this->DELETED, '', $string);
        }
                
        public function reset() {
                // clear the patterns collection so that this object may be re-used
                $this->_patterns = array();
        }

        // private
        private $_escaped = array();  // escaped characters
        private $_patterns = array(); // patterns stored by index
        
        // create and add a new pattern to the patterns collection
        private function _add() {
                $arguments = func_get_args();
                $this->_patterns[] = $arguments;
        }
        
        // this is the global replace function (it's quite complicated)
        private function _replacement($arguments) {
                if (empty($arguments)) return '';
                
                $i = 1; $j = 0;
                // loop through the patterns
                while (isset($this->_patterns[$j])) {
                        $pattern = $this->_patterns[$j++];
                        // do we have a result?
                        if (isset($arguments[$i]) && ($arguments[$i] != '')) {
                                $replacement = $pattern[self::REPLACEMENT];
                                
                                if (is_array($replacement) && isset($replacement['fn'])) {
                                        
                                        if (isset($replacement['data'])) $this->buffer = $replacement['data'];
                                        return call_user_func(array(&$this, $replacement['fn']), $arguments, $i);
                                        
                                } elseif (is_int($replacement)) {
                                        return $arguments[$replacement + $i];
                                
                                }
                                $delete = ($this->escapeChar == '' ||
                                           strpos($arguments[$i], $this->escapeChar) === false)
                                        ? '' : "\x01" . $arguments[$i] . "\x01";
                                return $delete . $replacement;
                        
                        // skip over references to sub-expressions
                        } else {
                                $i += $pattern[self::LENGTH];
                        }
                }
        }
        
        private function _backReferences($match, $offset) {
                $replacement = $this->buffer['replacement'];
                $quote = $this->buffer['quote'];
                $i = $this->buffer['length'];
                while ($i) {
                        $replacement = str_replace('$'.$i--, $match[$offset + $i], $replacement);
                }
                return $replacement;
        }
        
        private function _replace_name($match, $offset){
                $length = strlen($match[$offset + 2]);
                $start = $length - max($length - strlen($match[$offset + 3]), 0);
                return substr($match[$offset + 1], $start, $length) . $match[$offset + 4];
        }
        
        private function _replace_encoded($match, $offset) {
                return $this->buffer[$match[$offset]];
        }
        
        
        // php : we cannot pass additional data to preg_replace_callback,
        // and we cannot use &$this in create_function, so let's go to lower level
        private $buffer;
        
        // encode escaped characters
        private function _escape($string, $escapeChar) {
                if ($escapeChar) {
                        $this->buffer = $escapeChar;
                        return preg_replace_callback(
                                '/\\' . $escapeChar . '(.)' .'/',
                                array(&$this, '_escapeBis'),
                                $string
                        );
                        
                } else {
                        return $string;
                }
        }
        private function _escapeBis($match) {
                $this->_escaped[] = $match[1];
                return $this->buffer;
        }
        
        // decode escaped characters
        private function _unescape($string, $escapeChar) {
                if ($escapeChar) {
                        $regexp = '/'.'\\'.$escapeChar.'/';
                        $this->buffer = array('escapeChar'=> $escapeChar, 'i' => 0);
                        return preg_replace_callback
                        (
                                $regexp,
                                array(&$this, '_unescapeBis'),
                                $string
                        );
                        
                } else {
                        return $string;
                }
        }
        private function _unescapeBis() {
                if (isset($this->_escaped[$this->buffer['i']])
                        && $this->_escaped[$this->buffer['i']] != '')
                {
                         $temp = $this->_escaped[$this->buffer['i']];
                } else {
                        $temp = '';
                }
                $this->buffer['i']++;
                return $this->buffer['escapeChar'] . $temp;
        }
        
        private function _internalEscape($string) {
                return preg_replace($this->ESCAPE, '', $string);
        }
}
?>