wp-includes/class-json.php

   1 <?php
   2 if ( ! class_exists( 'Services_JSON' ) ) :
   3 /* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
   4 /**
   5  * Converts to and from JSON format.
   6  *
   7  * JSON (JavaScript Object Notation) is a lightweight data-interchange
   8  * format. It is easy for humans to read and write. It is easy for machines
   9  * to parse and generate. It is based on a subset of the JavaScript
  10  * Programming Language, Standard ECMA-262 3rd Edition - December 1999.
  11  * This feature can also be found in  Python. JSON is a text format that is
  12  * completely language independent but uses conventions that are familiar
  13  * to programmers of the C-family of languages, including C, C++, C#, Java,
  14  * JavaScript, Perl, TCL, and many others. These properties make JSON an
  15  * ideal data-interchange language.
  16  *
  17  * This package provides a simple encoder and decoder for JSON notation. It
  18  * is intended for use with client-side Javascript applications that make
  19  * use of HTTPRequest to perform server communication functions - data can
  20  * be encoded into JSON notation for use in a client-side javascript, or
  21  * decoded from incoming Javascript requests. JSON format is native to
  22  * Javascript, and can be directly eval()'ed with no further parsing
  23  * overhead
  24  *
  25  * All strings should be in ASCII or UTF-8 format!
  26  *
  27  * LICENSE: Redistribution and use in source and binary forms, with or
  28  * without modification, are permitted provided that the following
  29  * conditions are met: Redistributions of source code must retain the
  30  * above copyright notice, this list of conditions and the following
  31  * disclaimer. Redistributions in binary form must reproduce the above
  32  * copyright notice, this list of conditions and the following disclaimer
  33  * in the documentation and/or other materials provided with the
  34  * distribution.
  35  *
  36  * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
  37  * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
  38  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN
  39  * NO EVENT SHALL CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
  40  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
  41  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
  42  * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
  43  * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR
  44  * TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
  45  * USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
  46  * DAMAGE.
  47  *
  48  * @category
  49  * @package     Services_JSON
  50  * @author      Michal Migurski <mike-json@teczno.com>
  51  * @author      Matt Knapp <mdknapp[at]gmail[dot]com>
  52  * @author      Brett Stimmerman <brettstimmerman[at]gmail[dot]com>
  53  * @copyright   2005 Michal Migurski
  54  * @version     CVS: $Id: JSON.php 305040 2010-11-02 23:19:03Z alan_k $
  55  * @license     http://www.opensource.org/licenses/bsd-license.php
  56  * @link        http://pear.php.net/pepr/pepr-proposal-show.php?id=198
  57  */
  58
  59 /**
  60  * Marker constant for Services_JSON::decode(), used to flag stack state
  61  */
  62 define('SERVICES_JSON_SLICE',   1);
  63
  64 /**
  65  * Marker constant for Services_JSON::decode(), used to flag stack state
  66  */
  67 define('SERVICES_JSON_IN_STR',  2);
  68
  69 /**
  70  * Marker constant for Services_JSON::decode(), used to flag stack state
  71  */
  72 define('SERVICES_JSON_IN_ARR',  3);
  73
  74 /**
  75  * Marker constant for Services_JSON::decode(), used to flag stack state
  76  */
  77 define('SERVICES_JSON_IN_OBJ',  4);
  78
  79 /**
  80  * Marker constant for Services_JSON::decode(), used to flag stack state
  81  */
  82 define('SERVICES_JSON_IN_CMT', 5);
  83
  84 /**
  85  * Behavior switch for Services_JSON::decode()
  86  */
  87 define('SERVICES_JSON_LOOSE_TYPE', 16);
  88
  89 /**
  90  * Behavior switch for Services_JSON::decode()
  91  */
  92 define('SERVICES_JSON_SUPPRESS_ERRORS', 32);
  93
  94 /**
  95  * Behavior switch for Services_JSON::decode()
  96  */
  97 define('SERVICES_JSON_USE_TO_JSON', 64);
  98
  99 /**
 100  * Converts to and from JSON format.
 101  *
 102  * Brief example of use:
 103  *
 104  * <code>
 105  * // create a new instance of Services_JSON
 106  * $json = new Services_JSON();
 107  *
 108  * // convert a complexe value to JSON notation, and send it to the browser
 109  * $value = array('foo', 'bar', array(1, 2, 'baz'), array(3, array(4)));
 110  * $output = $json->encode($value);
 111  *
 112  * print($output);
 113  * // prints: ["foo","bar",[1,2,"baz"],[3,[4]]]
 114  *
 115  * // accept incoming POST data, assumed to be in JSON notation
 116  * $input = file_get_contents('php://input', 1000000);
 117  * $value = $json->decode($input);
 118  * </code>
 119  */
 120 class Services_JSON
 121 {
 122    /**
 123     * constructs a new JSON instance
 124     *
 125     * @param    int     $use    object behavior flags; combine with boolean-OR
 126     *
 127     *                           possible values:
 128     *                           - SERVICES_JSON_LOOSE_TYPE:  loose typing.
 129     *                                   "{...}" syntax creates associative arrays
 130     *                                   instead of objects in decode().
 131     *                           - SERVICES_JSON_SUPPRESS_ERRORS:  error suppression.
 132     *                                   Values which can't be encoded (e.g. resources)
 133     *                                   appear as NULL instead of throwing errors.
 134     *                                   By default, a deeply-nested resource will
 135     *                                   bubble up with an error, so all return values
 136     *                                   from encode() should be checked with isError()
 137     *                           - SERVICES_JSON_USE_TO_JSON:  call toJSON when serializing objects
 138     *                                   It serializes the return value from the toJSON call rather
 139     *                                   than the object it'self,  toJSON can return associative arrays,
 140     *                                   strings or numbers, if you return an object, make sure it does
 141     *                                   not have a toJSON method, otherwise an error will occur.
 142     */
 143     function __construct( $use = 0 )
 144     {
 145         $this->use = $use;
 146         $this->_mb_strlen            = function_exists('mb_strlen');
 147         $this->_mb_convert_encoding  = function_exists('mb_convert_encoding');
 148         $this->_mb_substr            = function_exists('mb_substr');
 149     }
 150
 151         /**
 152          * PHP4 constructor.
 153          */
 154         public function Services_JSON( $use = 0 ) {
 155                 self::__construct( $use );
 156         }
 157     // private - cache the mbstring lookup results..
 158     var $_mb_strlen = false;
 159     var $_mb_substr = false;
 160     var $_mb_convert_encoding = false;
 161
 162    /**
 163     * convert a string from one UTF-16 char to one UTF-8 char
 164     *
 165     * Normally should be handled by mb_convert_encoding, but
 166     * provides a slower PHP-only method for installations
 167     * that lack the multibye string extension.
 168     *
 169     * @param    string  $utf16  UTF-16 character
 170     * @return   string  UTF-8 character
 171     * @access   private
 172     */
 173     function utf162utf8($utf16)
 174     {
 175         // oh please oh please oh please oh please oh please
 176         if($this->_mb_convert_encoding) {
 177             return mb_convert_encoding($utf16, 'UTF-8', 'UTF-16');
 178         }
 179
 180         $bytes = (ord($utf16{0}) << 8) | ord($utf16{1});
 181
 182         switch(true) {
 183             case ((0x7F & $bytes) == $bytes):
 184                 // this case should never be reached, because we are in ASCII range
 185                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 186                 return chr(0x7F & $bytes);
 187
 188             case (0x07FF & $bytes) == $bytes:
 189                 // return a 2-byte UTF-8 character
 190                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 191                 return chr(0xC0 | (($bytes >> 6) & 0x1F))
 192                      . chr(0x80 | ($bytes & 0x3F));
 193
 194             case (0xFFFF & $bytes) == $bytes:
 195                 // return a 3-byte UTF-8 character
 196                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 197                 return chr(0xE0 | (($bytes >> 12) & 0x0F))
 198                      . chr(0x80 | (($bytes >> 6) & 0x3F))
 199                      . chr(0x80 | ($bytes & 0x3F));
 200         }
 201
 202         // ignoring UTF-32 for now, sorry
 203         return '';
 204     }
 205
 206    /**
 207     * convert a string from one UTF-8 char to one UTF-16 char
 208     *
 209     * Normally should be handled by mb_convert_encoding, but
 210     * provides a slower PHP-only method for installations
 211     * that lack the multibye string extension.
 212     *
 213     * @param    string  $utf8   UTF-8 character
 214     * @return   string  UTF-16 character
 215     * @access   private
 216     */
 217     function utf82utf16($utf8)
 218     {
 219         // oh please oh please oh please oh please oh please
 220         if($this->_mb_convert_encoding) {
 221             return mb_convert_encoding($utf8, 'UTF-16', 'UTF-8');
 222         }
 223
 224         switch($this->strlen8($utf8)) {
 225             case 1:
 226                 // this case should never be reached, because we are in ASCII range
 227                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 228                 return $utf8;
 229
 230             case 2:
 231                 // return a UTF-16 character from a 2-byte UTF-8 char
 232                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 233                 return chr(0x07 & (ord($utf8{0}) >> 2))
 234                      . chr((0xC0 & (ord($utf8{0}) << 6))
 235                          | (0x3F & ord($utf8{1})));
 236
 237             case 3:
 238                 // return a UTF-16 character from a 3-byte UTF-8 char
 239                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 240                 return chr((0xF0 & (ord($utf8{0}) << 4))
 241                          | (0x0F & (ord($utf8{1}) >> 2)))
 242                      . chr((0xC0 & (ord($utf8{1}) << 6))
 243                          | (0x7F & ord($utf8{2})));
 244         }
 245
 246         // ignoring UTF-32 for now, sorry
 247         return '';
 248     }
 249
 250    /**
 251     * encodes an arbitrary variable into JSON format (and sends JSON Header)
 252     *
 253     * @param    mixed   $var    any number, boolean, string, array, or object to be encoded.
 254     *                           see argument 1 to Services_JSON() above for array-parsing behavior.
 255     *                           if var is a strng, note that encode() always expects it
 256     *                           to be in ASCII or UTF-8 format!
 257     *
 258     * @return   mixed   JSON string representation of input var or an error if a problem occurs
 259     * @access   public
 260     */
 261     function encode($var)
 262     {
 263         header('Content-type: application/json');
 264         return $this->encodeUnsafe($var);
 265     }
 266     /**
 267     * encodes an arbitrary variable into JSON format without JSON Header - warning - may allow XSS!!!!)
 268     *
 269     * @param    mixed   $var    any number, boolean, string, array, or object to be encoded.
 270     *                           see argument 1 to Services_JSON() above for array-parsing behavior.
 271     *                           if var is a strng, note that encode() always expects it
 272     *                           to be in ASCII or UTF-8 format!
 273     *
 274     * @return   mixed   JSON string representation of input var or an error if a problem occurs
 275     * @access   public
 276     */
 277     function encodeUnsafe($var)
 278     {
 279         // see bug #16908 - regarding numeric locale printing
 280         $lc = setlocale(LC_NUMERIC, 0);
 281         setlocale(LC_NUMERIC, 'C');
 282         $ret = $this->_encode($var);
 283         setlocale(LC_NUMERIC, $lc);
 284         return $ret;
 285
 286     }
 287     /**
 288     * PRIVATE CODE that does the work of encodes an arbitrary variable into JSON format
 289     *
 290     * @param    mixed   $var    any number, boolean, string, array, or object to be encoded.
 291     *                           see argument 1 to Services_JSON() above for array-parsing behavior.
 292     *                           if var is a strng, note that encode() always expects it
 293     *                           to be in ASCII or UTF-8 format!
 294     *
 295     * @return   mixed   JSON string representation of input var or an error if a problem occurs
 296     * @access   public
 297     */
 298     function _encode($var)
 299     {
 300
 301         switch (gettype($var)) {
 302             case 'boolean':
 303                 return $var ? 'true' : 'false';
 304
 305             case 'NULL':
 306                 return 'null';
 307
 308             case 'integer':
 309                 return (int) $var;
 310
 311             case 'double':
 312             case 'float':
 313                 return  (float) $var;
 314
 315             case 'string':
 316                 // STRINGS ARE EXPECTED TO BE IN ASCII OR UTF-8 FORMAT
 317                 $ascii = '';
 318                 $strlen_var = $this->strlen8($var);
 319
 320                /*
 321                 * Iterate over every character in the string,
 322                 * escaping with a slash or encoding to UTF-8 where necessary
 323                 */
 324                 for ($c = 0; $c < $strlen_var; ++$c) {
 325
 326                     $ord_var_c = ord($var{$c});
 327
 328                     switch (true) {
 329                         case $ord_var_c == 0x08:
 330                             $ascii .= '\b';
 331                             break;
 332                         case $ord_var_c == 0x09:
 333                             $ascii .= '\t';
 334                             break;
 335                         case $ord_var_c == 0x0A:
 336                             $ascii .= '\n';
 337                             break;
 338                         case $ord_var_c == 0x0C:
 339                             $ascii .= '\f';
 340                             break;
 341                         case $ord_var_c == 0x0D:
 342                             $ascii .= '\r';
 343                             break;
 344
 345                         case $ord_var_c == 0x22:
 346                         case $ord_var_c == 0x2F:
 347                         case $ord_var_c == 0x5C:
 348                             // double quote, slash, slosh
 349                             $ascii .= '\\'.$var{$c};
 350                             break;
 351
 352                         case (($ord_var_c >= 0x20) && ($ord_var_c <= 0x7F)):
 353                             // characters U-00000000 - U-0000007F (same as ASCII)
 354                             $ascii .= $var{$c};
 355                             break;
 356
 357                         case (($ord_var_c & 0xE0) == 0xC0):
 358                             // characters U-00000080 - U-000007FF, mask 110XXXXX
 359                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 360                             if ($c+1 >= $strlen_var) {
 361                                 $c += 1;
 362                                 $ascii .= '?';
 363                                 break;
 364                             }
 365
 366                             $char = pack('C*', $ord_var_c, ord($var{$c + 1}));
 367                             $c += 1;
 368                             $utf16 = $this->utf82utf16($char);
 369                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 370                             break;
 371
 372                         case (($ord_var_c & 0xF0) == 0xE0):
 373                             if ($c+2 >= $strlen_var) {
 374                                 $c += 2;
 375                                 $ascii .= '?';
 376                                 break;
 377                             }
 378                             // characters U-00000800 - U-0000FFFF, mask 1110XXXX
 379                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 380                             $char = pack('C*', $ord_var_c,
 381                                          @ord($var{$c + 1}),
 382                                          @ord($var{$c + 2}));
 383                             $c += 2;
 384                             $utf16 = $this->utf82utf16($char);
 385                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 386                             break;
 387
 388                         case (($ord_var_c & 0xF8) == 0xF0):
 389                             if ($c+3 >= $strlen_var) {
 390                                 $c += 3;
 391                                 $ascii .= '?';
 392                                 break;
 393                             }
 394                             // characters U-00010000 - U-001FFFFF, mask 11110XXX
 395                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 396                             $char = pack('C*', $ord_var_c,
 397                                          ord($var{$c + 1}),
 398                                          ord($var{$c + 2}),
 399                                          ord($var{$c + 3}));
 400                             $c += 3;
 401                             $utf16 = $this->utf82utf16($char);
 402                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 403                             break;
 404
 405                         case (($ord_var_c & 0xFC) == 0xF8):
 406                             // characters U-00200000 - U-03FFFFFF, mask 111110XX
 407                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 408                             if ($c+4 >= $strlen_var) {
 409                                 $c += 4;
 410                                 $ascii .= '?';
 411                                 break;
 412                             }
 413                             $char = pack('C*', $ord_var_c,
 414                                          ord($var{$c + 1}),
 415                                          ord($var{$c + 2}),
 416                                          ord($var{$c + 3}),
 417                                          ord($var{$c + 4}));
 418                             $c += 4;
 419                             $utf16 = $this->utf82utf16($char);
 420                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 421                             break;
 422
 423                         case (($ord_var_c & 0xFE) == 0xFC):
 424                         if ($c+5 >= $strlen_var) {
 425                                 $c += 5;
 426                                 $ascii .= '?';
 427                                 break;
 428                             }
 429                             // characters U-04000000 - U-7FFFFFFF, mask 1111110X
 430                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 431                             $char = pack('C*', $ord_var_c,
 432                                          ord($var{$c + 1}),
 433                                          ord($var{$c + 2}),
 434                                          ord($var{$c + 3}),
 435                                          ord($var{$c + 4}),
 436                                          ord($var{$c + 5}));
 437                             $c += 5;
 438                             $utf16 = $this->utf82utf16($char);
 439                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 440                             break;
 441                     }
 442                 }
 443                 return  '"'.$ascii.'"';
 444
 445             case 'array':
 446                /*
 447                 * As per JSON spec if any array key is not an integer
 448                 * we must treat the the whole array as an object. We
 449                 * also try to catch a sparsely populated associative
 450                 * array with numeric keys here because some JS engines
 451                 * will create an array with empty indexes up to
 452                 * max_index which can cause memory issues and because
 453                 * the keys, which may be relevant, will be remapped
 454                 * otherwise.
 455                 *
 456                 * As per the ECMA and JSON specification an object may
 457                 * have any string as a property. Unfortunately due to
 458                 * a hole in the ECMA specification if the key is a
 459                 * ECMA reserved word or starts with a digit the
 460                 * parameter is only accessible using ECMAScript's
 461                 * bracket notation.
 462                 */
 463
 464                 // treat as a JSON object
 465                 if (is_array($var) && count($var) && (array_keys($var) !== range(0, sizeof($var) - 1))) {
 466                     $properties = array_map(array($this, 'name_value'),
 467                                             array_keys($var),
 468                                             array_values($var));
 469
 470                     foreach($properties as $property) {
 471                         if(Services_JSON::isError($property)) {
 472                             return $property;
 473                         }
 474                     }
 475
 476                     return '{' . join(',', $properties) . '}';
 477                 }
 478
 479                 // treat it like a regular array
 480                 $elements = array_map(array($this, '_encode'), $var);
 481
 482                 foreach($elements as $element) {
 483                     if(Services_JSON::isError($element)) {
 484                         return $element;
 485                     }
 486                 }
 487
 488                 return '[' . join(',', $elements) . ']';
 489
 490             case 'object':
 491
 492                 // support toJSON methods.
 493                 if (($this->use & SERVICES_JSON_USE_TO_JSON) && method_exists($var, 'toJSON')) {
 494                     // this may end up allowing unlimited recursion
 495                     // so we check the return value to make sure it's not got the same method.
 496                     $recode = $var->toJSON();
 497
 498                     if (method_exists($recode, 'toJSON')) {
 499
 500                         return ($this->use & SERVICES_JSON_SUPPRESS_ERRORS)
 501                         ? 'null'
 502                         : new Services_JSON_Error(get_class($var).
 503                             " toJSON returned an object with a toJSON method.");
 504
 505                     }
 506
 507                     return $this->_encode( $recode );
 508                 }
 509
 510                 $vars = get_object_vars($var);
 511
 512                 $properties = array_map(array($this, 'name_value'),
 513                                         array_keys($vars),
 514                                         array_values($vars));
 515
 516                 foreach($properties as $property) {
 517                     if(Services_JSON::isError($property)) {
 518                         return $property;
 519                     }
 520                 }
 521
 522                 return '{' . join(',', $properties) . '}';
 523
 524             default:
 525                 return ($this->use & SERVICES_JSON_SUPPRESS_ERRORS)
 526                     ? 'null'
 527                     : new Services_JSON_Error(gettype($var)." can not be encoded as JSON string");
 528         }
 529     }
 530
 531    /**
 532     * array-walking function for use in generating JSON-formatted name-value pairs
 533     *
 534     * @param    string  $name   name of key to use
 535     * @param    mixed   $value  reference to an array element to be encoded
 536     *
 537     * @return   string  JSON-formatted name-value pair, like '"name":value'
 538     * @access   private
 539     */
 540     function name_value($name, $value)
 541     {
 542         $encoded_value = $this->_encode($value);
 543
 544         if(Services_JSON::isError($encoded_value)) {
 545             return $encoded_value;
 546         }
 547
 548         return $this->_encode(strval($name)) . ':' . $encoded_value;
 549     }
 550
 551    /**
 552     * reduce a string by removing leading and trailing comments and whitespace
 553     *
 554     * @param    $str    string      string value to strip of comments and whitespace
 555     *
 556     * @return   string  string value stripped of comments and whitespace
 557     * @access   private
 558     */
 559     function reduce_string($str)
 560     {
 561         $str = preg_replace(array(
 562
 563                 // eliminate single line comments in '// ...' form
 564                 '#^\s*//(.+)$#m',
 565
 566                 // eliminate multi-line comments in '/* ... */' form, at start of string
 567                 '#^\s*/\*(.+)\*/#Us',
 568
 569                 // eliminate multi-line comments in '/* ... */' form, at end of string
 570                 '#/\*(.+)\*/\s*$#Us'
 571
 572             ), '', $str);
 573
 574         // eliminate extraneous space
 575         return trim($str);
 576     }
 577
 578    /**
 579     * decodes a JSON string into appropriate variable
 580     *
 581     * @param    string  $str    JSON-formatted string
 582     *
 583     * @return   mixed   number, boolean, string, array, or object
 584     *                   corresponding to given JSON input string.
 585     *                   See argument 1 to Services_JSON() above for object-output behavior.
 586     *                   Note that decode() always returns strings
 587     *                   in ASCII or UTF-8 format!
 588     * @access   public
 589     */
 590     function decode($str)
 591     {
 592         $str = $this->reduce_string($str);
 593
 594         switch (strtolower($str)) {
 595             case 'true':
 596                 return true;
 597
 598             case 'false':
 599                 return false;
 600
 601             case 'null':
 602                 return null;
 603
 604             default:
 605                 $m = array();
 606
 607                 if (is_numeric($str)) {
 608                     // Lookie-loo, it's a number
 609
 610                     // This would work on its own, but I'm trying to be
 611                     // good about returning integers where appropriate:
 612                     // return (float)$str;
 613
 614                     // Return float or int, as appropriate
 615                     return ((float)$str == (integer)$str)
 616                         ? (integer)$str
 617                         : (float)$str;
 618
 619                 } elseif (preg_match('/^("|\').*(\1)$/s', $str, $m) && $m[1] == $m[2]) {
 620                     // STRINGS RETURNED IN UTF-8 FORMAT
 621                     $delim = $this->substr8($str, 0, 1);
 622                     $chrs = $this->substr8($str, 1, -1);
 623                     $utf8 = '';
 624                     $strlen_chrs = $this->strlen8($chrs);
 625
 626                     for ($c = 0; $c < $strlen_chrs; ++$c) {
 627
 628                         $substr_chrs_c_2 = $this->substr8($chrs, $c, 2);
 629                         $ord_chrs_c = ord($chrs{$c});
 630
 631                         switch (true) {
 632                             case $substr_chrs_c_2 == '\b':
 633                                 $utf8 .= chr(0x08);
 634                                 ++$c;
 635                                 break;
 636                             case $substr_chrs_c_2 == '\t':
 637                                 $utf8 .= chr(0x09);
 638                                 ++$c;
 639                                 break;
 640                             case $substr_chrs_c_2 == '\n':
 641                                 $utf8 .= chr(0x0A);
 642                                 ++$c;
 643                                 break;
 644                             case $substr_chrs_c_2 == '\f':
 645                                 $utf8 .= chr(0x0C);
 646                                 ++$c;
 647                                 break;
 648                             case $substr_chrs_c_2 == '\r':
 649                                 $utf8 .= chr(0x0D);
 650                                 ++$c;
 651                                 break;
 652
 653                             case $substr_chrs_c_2 == '\\"':
 654                             case $substr_chrs_c_2 == '\\\'':
 655                             case $substr_chrs_c_2 == '\\\\':
 656                             case $substr_chrs_c_2 == '\\/':
 657                                 if (($delim == '"' && $substr_chrs_c_2 != '\\\'') ||
 658                                    ($delim == "'" && $substr_chrs_c_2 != '\\"')) {
 659                                     $utf8 .= $chrs{++$c};
 660                                 }
 661                                 break;
 662
 663                             case preg_match('/\\\u[0-9A-F]{4}/i', $this->substr8($chrs, $c, 6)):
 664                                 // single, escaped unicode character
 665                                 $utf16 = chr(hexdec($this->substr8($chrs, ($c + 2), 2)))
 666                                        . chr(hexdec($this->substr8($chrs, ($c + 4), 2)));
 667                                 $utf8 .= $this->utf162utf8($utf16);
 668                                 $c += 5;
 669                                 break;
 670
 671                             case ($ord_chrs_c >= 0x20) && ($ord_chrs_c <= 0x7F):
 672                                 $utf8 .= $chrs{$c};
 673                                 break;
 674
 675                             case ($ord_chrs_c & 0xE0) == 0xC0:
 676                                 // characters U-00000080 - U-000007FF, mask 110XXXXX
 677                                 //see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 678                                 $utf8 .= $this->substr8($chrs, $c, 2);
 679                                 ++$c;
 680                                 break;
 681
 682                             case ($ord_chrs_c & 0xF0) == 0xE0:
 683                                 // characters U-00000800 - U-0000FFFF, mask 1110XXXX
 684                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 685                                 $utf8 .= $this->substr8($chrs, $c, 3);
 686                                 $c += 2;
 687                                 break;
 688
 689                             case ($ord_chrs_c & 0xF8) == 0xF0:
 690                                 // characters U-00010000 - U-001FFFFF, mask 11110XXX
 691                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 692                                 $utf8 .= $this->substr8($chrs, $c, 4);
 693                                 $c += 3;
 694                                 break;
 695
 696                             case ($ord_chrs_c & 0xFC) == 0xF8:
 697                                 // characters U-00200000 - U-03FFFFFF, mask 111110XX
 698                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 699                                 $utf8 .= $this->substr8($chrs, $c, 5);
 700                                 $c += 4;
 701                                 break;
 702
 703                             case ($ord_chrs_c & 0xFE) == 0xFC:
 704                                 // characters U-04000000 - U-7FFFFFFF, mask 1111110X
 705                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 706                                 $utf8 .= $this->substr8($chrs, $c, 6);
 707                                 $c += 5;
 708                                 break;
 709
 710                         }
 711
 712                     }
 713
 714                     return $utf8;
 715
 716                 } elseif (preg_match('/^\[.*\]$/s', $str) || preg_match('/^\{.*\}$/s', $str)) {
 717                     // array, or object notation
 718
 719                     if ($str{0} == '[') {
 720                         $stk = array(SERVICES_JSON_IN_ARR);
 721                         $arr = array();
 722                     } else {
 723                         if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 724                             $stk = array(SERVICES_JSON_IN_OBJ);
 725                             $obj = array();
 726                         } else {
 727                             $stk = array(SERVICES_JSON_IN_OBJ);
 728                             $obj = new stdClass();
 729                         }
 730                     }
 731
 732                     array_push($stk, array('what'  => SERVICES_JSON_SLICE,
 733                                            'where' => 0,
 734                                            'delim' => false));
 735
 736                     $chrs = $this->substr8($str, 1, -1);
 737                     $chrs = $this->reduce_string($chrs);
 738
 739                     if ($chrs == '') {
 740                         if (reset($stk) == SERVICES_JSON_IN_ARR) {
 741                             return $arr;
 742
 743                         } else {
 744                             return $obj;
 745
 746                         }
 747                     }
 748
 749                     //print("\nparsing {$chrs}\n");
 750
 751                     $strlen_chrs = $this->strlen8($chrs);
 752
 753                     for ($c = 0; $c <= $strlen_chrs; ++$c) {
 754
 755                         $top = end($stk);
 756                         $substr_chrs_c_2 = $this->substr8($chrs, $c, 2);
 757
 758                         if (($c == $strlen_chrs) || (($chrs{$c} == ',') && ($top['what'] == SERVICES_JSON_SLICE))) {
 759                             // found a comma that is not inside a string, array, etc.,
 760                             // OR we've reached the end of the character list
 761                             $slice = $this->substr8($chrs, $top['where'], ($c - $top['where']));
 762                             array_push($stk, array('what' => SERVICES_JSON_SLICE, 'where' => ($c + 1), 'delim' => false));
 763                             //print("Found split at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 764
 765                             if (reset($stk) == SERVICES_JSON_IN_ARR) {
 766                                 // we are in an array, so just push an element onto the stack
 767                                 array_push($arr, $this->decode($slice));
 768
 769                             } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {
 770                                 // we are in an object, so figure
 771                                 // out the property name and set an
 772                                 // element in an associative array,
 773                                 // for now
 774                                 $parts = array();
 775
 776                                if (preg_match('/^\s*(["\'].*[^\\\]["\'])\s*:/Uis', $slice, $parts)) {
 777                                       // "name":value pair
 778                                     $key = $this->decode($parts[1]);
 779                                     $val = $this->decode(trim(substr($slice, strlen($parts[0])), ", \t\n\r\0\x0B"));
 780                                     if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 781                                         $obj[$key] = $val;
 782                                     } else {
 783                                         $obj->$key = $val;
 784                                     }
 785                                 } elseif (preg_match('/^\s*(\w+)\s*:/Uis', $slice, $parts)) {
 786                                     // name:value pair, where name is unquoted
 787                                     $key = $parts[1];
 788                                     $val = $this->decode(trim(substr($slice, strlen($parts[0])), ", \t\n\r\0\x0B"));
 789
 790                                     if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 791                                         $obj[$key] = $val;
 792                                     } else {
 793                                         $obj->$key = $val;
 794                                     }
 795                                 }
 796
 797                             }
 798
 799                         } elseif ((($chrs{$c} == '"') || ($chrs{$c} == "'")) && ($top['what'] != SERVICES_JSON_IN_STR)) {
 800                             // found a quote, and we are not inside a string
 801                             array_push($stk, array('what' => SERVICES_JSON_IN_STR, 'where' => $c, 'delim' => $chrs{$c}));
 802                             //print("Found start of string at {$c}\n");
 803
 804                         } elseif (($chrs{$c} == $top['delim']) &&
 805                                  ($top['what'] == SERVICES_JSON_IN_STR) &&
 806                                  (($this->strlen8($this->substr8($chrs, 0, $c)) - $this->strlen8(rtrim($this->substr8($chrs, 0, $c), '\\'))) % 2 != 1)) {
 807                             // found a quote, we're in a string, and it's not escaped
 808                             // we know that it's not escaped becase there is _not_ an
 809                             // odd number of backslashes at the end of the string so far
 810                             array_pop($stk);
 811                             //print("Found end of string at {$c}: ".$this->substr8($chrs, $top['where'], (1 + 1 + $c - $top['where']))."\n");
 812
 813                         } elseif (($chrs{$c} == '[') &&
 814                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 815                             // found a left-bracket, and we are in an array, object, or slice
 816                             array_push($stk, array('what' => SERVICES_JSON_IN_ARR, 'where' => $c, 'delim' => false));
 817                             //print("Found start of array at {$c}\n");
 818
 819                         } elseif (($chrs{$c} == ']') && ($top['what'] == SERVICES_JSON_IN_ARR)) {
 820                             // found a right-bracket, and we're in an array
 821                             array_pop($stk);
 822                             //print("Found end of array at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 823
 824                         } elseif (($chrs{$c} == '{') &&
 825                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 826                             // found a left-brace, and we are in an array, object, or slice
 827                             array_push($stk, array('what' => SERVICES_JSON_IN_OBJ, 'where' => $c, 'delim' => false));
 828                             //print("Found start of object at {$c}\n");
 829
 830                         } elseif (($chrs{$c} == '}') && ($top['what'] == SERVICES_JSON_IN_OBJ)) {
 831                             // found a right-brace, and we're in an object
 832                             array_pop($stk);
 833                             //print("Found end of object at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 834
 835                         } elseif (($substr_chrs_c_2 == '/*') &&
 836                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 837                             // found a comment start, and we are in an array, object, or slice
 838                             array_push($stk, array('what' => SERVICES_JSON_IN_CMT, 'where' => $c, 'delim' => false));
 839                             $c++;
 840                             //print("Found start of comment at {$c}\n");
 841
 842                         } elseif (($substr_chrs_c_2 == '*/') && ($top['what'] == SERVICES_JSON_IN_CMT)) {
 843                             // found a comment end, and we're in one now
 844                             array_pop($stk);
 845                             $c++;
 846
 847                             for ($i = $top['where']; $i <= $c; ++$i)
 848                                 $chrs = substr_replace($chrs, ' ', $i, 1);
 849
 850                             //print("Found end of comment at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 851
 852                         }
 853
 854                     }
 855
 856                     if (reset($stk) == SERVICES_JSON_IN_ARR) {
 857                         return $arr;
 858
 859                     } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {
 860                         return $obj;
 861
 862                     }
 863
 864                 }
 865         }
 866     }
 867
 868     /**
 869      * @todo Ultimately, this should just call PEAR::isError()
 870      */
 871     function isError($data, $code = null)
 872     {
 873         if (class_exists('pear')) {
 874             return PEAR::isError($data, $code);
 875         } elseif (is_object($data) && (get_class($data) == 'services_json_error' ||
 876                                  is_subclass_of($data, 'services_json_error'))) {
 877             return true;
 878         }
 879
 880         return false;
 881     }
 882
 883     /**
 884     * Calculates length of string in bytes
 885     * @param string
 886     * @return integer length
 887     */
 888     function strlen8( $str )
 889     {
 890         if ( $this->_mb_strlen ) {
 891             return mb_strlen( $str, "8bit" );
 892         }
 893         return strlen( $str );
 894     }
 895
 896     /**
 897     * Returns part of a string, interpreting $start and $length as number of bytes.
 898     * @param string
 899     * @param integer start
 900     * @param integer length
 901     * @return integer length
 902     */
 903     function substr8( $string, $start, $length=false )
 904     {
 905         if ( $length === false ) {
 906             $length = $this->strlen8( $string ) - $start;
 907         }
 908         if ( $this->_mb_substr ) {
 909             return mb_substr( $string, $start, $length, "8bit" );
 910         }
 911         return substr( $string, $start, $length );
 912     }
 913
 914 }
 915
 916 if (class_exists('PEAR_Error')) {
 917
 918     class Services_JSON_Error extends PEAR_Error
 919     {
 920         function __construct($message = 'unknown error', $code = null,
 921                                      $mode = null, $options = null, $userinfo = null)
 922         {
 923             parent::PEAR_Error($message, $code, $mode, $options, $userinfo);
 924         }
 925
 926         public function Services_JSON_Error($message = 'unknown error', $code = null,
 927                                      $mode = null, $options = null, $userinfo = null) {
 928                 self::__construct($message = 'unknown error', $code = null,
 929                                      $mode = null, $options = null, $userinfo = null);
 930         }
 931     }
 932
 933 } else {
 934
 935     /**
 936      * @todo Ultimately, this class shall be descended from PEAR_Error
 937      */
 938     class Services_JSON_Error
 939     {
 940             /**
 941              * PHP5 constructor.
 942              */
 943         function __construct( $message = 'unknown error', $code = null,
 944                                      $mode = null, $options = null, $userinfo = null )
 945         {
 946
 947         }
 948
 949             /**
 950              * PHP4 constructor.
 951              */
 952                 public function Services_JSON_Error( $message = 'unknown error', $code = null,
 953                                              $mode = null, $options = null, $userinfo = null ) {
 954                         self::__construct( $message, $code, $mode, $options, $userinfo );
 955                 }
 956     }
 957
 958 }
 959
 960 endif;