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     $use    object behavior flags; combine with boolean-OR
 126     *
 127     *                           possible values:
 128     *                           - SERVICES_JSON_LOOSE_TYPE:  loose typing.
 129     *                                   "{...}" syntax creates associative arrays
 130     *                                   instead of objects in decode().
 131     *                           - SERVICES_JSON_SUPPRESS_ERRORS:  error suppression.
 132     *                                   Values which can't be encoded (e.g. resources)
 133     *                                   appear as NULL instead of throwing errors.
 134     *                                   By default, a deeply-nested resource will
 135     *                                   bubble up with an error, so all return values
 136     *                                   from encode() should be checked with isError()
 137     *                           - SERVICES_JSON_USE_TO_JSON:  call toJSON when serializing objects
 138     *                                   It serializes the return value from the toJSON call rather
 139     *                                   than the object it'self,  toJSON can return associative arrays,
 140     *                                   strings or numbers, if you return an object, make sure it does
 141     *                                   not have a toJSON method, otherwise an error will occur.
 142     */
 143     function Services_JSON($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     // private - cache the mbstring lookup results..
 151     var $_mb_strlen = false;
 152     var $_mb_substr = false;
 153     var $_mb_convert_encoding = false;
 154
 155    /**
 156     * convert a string from one UTF-16 char to one UTF-8 char
 157     *
 158     * Normally should be handled by mb_convert_encoding, but
 159     * provides a slower PHP-only method for installations
 160     * that lack the multibye string extension.
 161     *
 162     * @param    string  $utf16  UTF-16 character
 163     * @return   string  UTF-8 character
 164     * @access   private
 165     */
 166     function utf162utf8($utf16)
 167     {
 168         // oh please oh please oh please oh please oh please
 169         if($this->_mb_convert_encoding) {
 170             return mb_convert_encoding($utf16, 'UTF-8', 'UTF-16');
 171         }
 172
 173         $bytes = (ord($utf16{0}) << 8) | ord($utf16{1});
 174
 175         switch(true) {
 176             case ((0x7F & $bytes) == $bytes):
 177                 // this case should never be reached, because we are in ASCII range
 178                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 179                 return chr(0x7F & $bytes);
 180
 181             case (0x07FF & $bytes) == $bytes:
 182                 // return a 2-byte UTF-8 character
 183                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 184                 return chr(0xC0 | (($bytes >> 6) & 0x1F))
 185                      . chr(0x80 | ($bytes & 0x3F));
 186
 187             case (0xFFFF & $bytes) == $bytes:
 188                 // return a 3-byte UTF-8 character
 189                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 190                 return chr(0xE0 | (($bytes >> 12) & 0x0F))
 191                      . chr(0x80 | (($bytes >> 6) & 0x3F))
 192                      . chr(0x80 | ($bytes & 0x3F));
 193         }
 194
 195         // ignoring UTF-32 for now, sorry
 196         return '';
 197     }
 198
 199    /**
 200     * convert a string from one UTF-8 char to one UTF-16 char
 201     *
 202     * Normally should be handled by mb_convert_encoding, but
 203     * provides a slower PHP-only method for installations
 204     * that lack the multibye string extension.
 205     *
 206     * @param    string  $utf8   UTF-8 character
 207     * @return   string  UTF-16 character
 208     * @access   private
 209     */
 210     function utf82utf16($utf8)
 211     {
 212         // oh please oh please oh please oh please oh please
 213         if($this->_mb_convert_encoding) {
 214             return mb_convert_encoding($utf8, 'UTF-16', 'UTF-8');
 215         }
 216
 217         switch($this->strlen8($utf8)) {
 218             case 1:
 219                 // this case should never be reached, because we are in ASCII range
 220                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 221                 return $utf8;
 222
 223             case 2:
 224                 // return a UTF-16 character from a 2-byte UTF-8 char
 225                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 226                 return chr(0x07 & (ord($utf8{0}) >> 2))
 227                      . chr((0xC0 & (ord($utf8{0}) << 6))
 228                          | (0x3F & ord($utf8{1})));
 229
 230             case 3:
 231                 // return a UTF-16 character from a 3-byte UTF-8 char
 232                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 233                 return chr((0xF0 & (ord($utf8{0}) << 4))
 234                          | (0x0F & (ord($utf8{1}) >> 2)))
 235                      . chr((0xC0 & (ord($utf8{1}) << 6))
 236                          | (0x7F & ord($utf8{2})));
 237         }
 238
 239         // ignoring UTF-32 for now, sorry
 240         return '';
 241     }
 242
 243    /**
 244     * encodes an arbitrary variable into JSON format (and sends JSON Header)
 245     *
 246     * @param    mixed   $var    any number, boolean, string, array, or object to be encoded.
 247     *                           see argument 1 to Services_JSON() above for array-parsing behavior.
 248     *                           if var is a strng, note that encode() always expects it
 249     *                           to be in ASCII or UTF-8 format!
 250     *
 251     * @return   mixed   JSON string representation of input var or an error if a problem occurs
 252     * @access   public
 253     */
 254     function encode($var)
 255     {
 256         header('Content-type: application/json');
 257         return $this->encodeUnsafe($var);
 258     }
 259     /**
 260     * encodes an arbitrary variable into JSON format without JSON Header - warning - may allow XSS!!!!)
 261     *
 262     * @param    mixed   $var    any number, boolean, string, array, or object to be encoded.
 263     *                           see argument 1 to Services_JSON() above for array-parsing behavior.
 264     *                           if var is a strng, note that encode() always expects it
 265     *                           to be in ASCII or UTF-8 format!
 266     *
 267     * @return   mixed   JSON string representation of input var or an error if a problem occurs
 268     * @access   public
 269     */
 270     function encodeUnsafe($var)
 271     {
 272         // see bug #16908 - regarding numeric locale printing
 273         $lc = setlocale(LC_NUMERIC, 0);
 274         setlocale(LC_NUMERIC, 'C');
 275         $ret = $this->_encode($var);
 276         setlocale(LC_NUMERIC, $lc);
 277         return $ret;
 278
 279     }
 280     /**
 281     * PRIVATE CODE that does the work of encodes an arbitrary variable into JSON format
 282     *
 283     * @param    mixed   $var    any number, boolean, string, array, or object to be encoded.
 284     *                           see argument 1 to Services_JSON() above for array-parsing behavior.
 285     *                           if var is a strng, note that encode() always expects it
 286     *                           to be in ASCII or UTF-8 format!
 287     *
 288     * @return   mixed   JSON string representation of input var or an error if a problem occurs
 289     * @access   public
 290     */
 291     function _encode($var)
 292     {
 293
 294         switch (gettype($var)) {
 295             case 'boolean':
 296                 return $var ? 'true' : 'false';
 297
 298             case 'NULL':
 299                 return 'null';
 300
 301             case 'integer':
 302                 return (int) $var;
 303
 304             case 'double':
 305             case 'float':
 306                 return  (float) $var;
 307
 308             case 'string':
 309                 // STRINGS ARE EXPECTED TO BE IN ASCII OR UTF-8 FORMAT
 310                 $ascii = '';
 311                 $strlen_var = $this->strlen8($var);
 312
 313                /*
 314                 * Iterate over every character in the string,
 315                 * escaping with a slash or encoding to UTF-8 where necessary
 316                 */
 317                 for ($c = 0; $c < $strlen_var; ++$c) {
 318
 319                     $ord_var_c = ord($var{$c});
 320
 321                     switch (true) {
 322                         case $ord_var_c == 0x08:
 323                             $ascii .= '\b';
 324                             break;
 325                         case $ord_var_c == 0x09:
 326                             $ascii .= '\t';
 327                             break;
 328                         case $ord_var_c == 0x0A:
 329                             $ascii .= '\n';
 330                             break;
 331                         case $ord_var_c == 0x0C:
 332                             $ascii .= '\f';
 333                             break;
 334                         case $ord_var_c == 0x0D:
 335                             $ascii .= '\r';
 336                             break;
 337
 338                         case $ord_var_c == 0x22:
 339                         case $ord_var_c == 0x2F:
 340                         case $ord_var_c == 0x5C:
 341                             // double quote, slash, slosh
 342                             $ascii .= '\\'.$var{$c};
 343                             break;
 344
 345                         case (($ord_var_c >= 0x20) && ($ord_var_c <= 0x7F)):
 346                             // characters U-00000000 - U-0000007F (same as ASCII)
 347                             $ascii .= $var{$c};
 348                             break;
 349
 350                         case (($ord_var_c & 0xE0) == 0xC0):
 351                             // characters U-00000080 - U-000007FF, mask 110XXXXX
 352                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 353                             if ($c+1 >= $strlen_var) {
 354                                 $c += 1;
 355                                 $ascii .= '?';
 356                                 break;
 357                             }
 358
 359                             $char = pack('C*', $ord_var_c, ord($var{$c + 1}));
 360                             $c += 1;
 361                             $utf16 = $this->utf82utf16($char);
 362                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 363                             break;
 364
 365                         case (($ord_var_c & 0xF0) == 0xE0):
 366                             if ($c+2 >= $strlen_var) {
 367                                 $c += 2;
 368                                 $ascii .= '?';
 369                                 break;
 370                             }
 371                             // characters U-00000800 - U-0000FFFF, mask 1110XXXX
 372                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 373                             $char = pack('C*', $ord_var_c,
 374                                          @ord($var{$c + 1}),
 375                                          @ord($var{$c + 2}));
 376                             $c += 2;
 377                             $utf16 = $this->utf82utf16($char);
 378                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 379                             break;
 380
 381                         case (($ord_var_c & 0xF8) == 0xF0):
 382                             if ($c+3 >= $strlen_var) {
 383                                 $c += 3;
 384                                 $ascii .= '?';
 385                                 break;
 386                             }
 387                             // characters U-00010000 - U-001FFFFF, mask 11110XXX
 388                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 389                             $char = pack('C*', $ord_var_c,
 390                                          ord($var{$c + 1}),
 391                                          ord($var{$c + 2}),
 392                                          ord($var{$c + 3}));
 393                             $c += 3;
 394                             $utf16 = $this->utf82utf16($char);
 395                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 396                             break;
 397
 398                         case (($ord_var_c & 0xFC) == 0xF8):
 399                             // characters U-00200000 - U-03FFFFFF, mask 111110XX
 400                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 401                             if ($c+4 >= $strlen_var) {
 402                                 $c += 4;
 403                                 $ascii .= '?';
 404                                 break;
 405                             }
 406                             $char = pack('C*', $ord_var_c,
 407                                          ord($var{$c + 1}),
 408                                          ord($var{$c + 2}),
 409                                          ord($var{$c + 3}),
 410                                          ord($var{$c + 4}));
 411                             $c += 4;
 412                             $utf16 = $this->utf82utf16($char);
 413                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 414                             break;
 415
 416                         case (($ord_var_c & 0xFE) == 0xFC):
 417                         if ($c+5 >= $strlen_var) {
 418                                 $c += 5;
 419                                 $ascii .= '?';
 420                                 break;
 421                             }
 422                             // characters U-04000000 - U-7FFFFFFF, mask 1111110X
 423                             // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 424                             $char = pack('C*', $ord_var_c,
 425                                          ord($var{$c + 1}),
 426                                          ord($var{$c + 2}),
 427                                          ord($var{$c + 3}),
 428                                          ord($var{$c + 4}),
 429                                          ord($var{$c + 5}));
 430                             $c += 5;
 431                             $utf16 = $this->utf82utf16($char);
 432                             $ascii .= sprintf('\u%04s', bin2hex($utf16));
 433                             break;
 434                     }
 435                 }
 436                 return  '"'.$ascii.'"';
 437
 438             case 'array':
 439                /*
 440                 * As per JSON spec if any array key is not an integer
 441                 * we must treat the the whole array as an object. We
 442                 * also try to catch a sparsely populated associative
 443                 * array with numeric keys here because some JS engines
 444                 * will create an array with empty indexes up to
 445                 * max_index which can cause memory issues and because
 446                 * the keys, which may be relevant, will be remapped
 447                 * otherwise.
 448                 *
 449                 * As per the ECMA and JSON specification an object may
 450                 * have any string as a property. Unfortunately due to
 451                 * a hole in the ECMA specification if the key is a
 452                 * ECMA reserved word or starts with a digit the
 453                 * parameter is only accessible using ECMAScript's
 454                 * bracket notation.
 455                 */
 456
 457                 // treat as a JSON object
 458                 if (is_array($var) && count($var) && (array_keys($var) !== range(0, sizeof($var) - 1))) {
 459                     $properties = array_map(array($this, 'name_value'),
 460                                             array_keys($var),
 461                                             array_values($var));
 462
 463                     foreach($properties as $property) {
 464                         if(Services_JSON::isError($property)) {
 465                             return $property;
 466                         }
 467                     }
 468
 469                     return '{' . join(',', $properties) . '}';
 470                 }
 471
 472                 // treat it like a regular array
 473                 $elements = array_map(array($this, '_encode'), $var);
 474
 475                 foreach($elements as $element) {
 476                     if(Services_JSON::isError($element)) {
 477                         return $element;
 478                     }
 479                 }
 480
 481                 return '[' . join(',', $elements) . ']';
 482
 483             case 'object':
 484
 485                 // support toJSON methods.
 486                 if (($this->use & SERVICES_JSON_USE_TO_JSON) && method_exists($var, 'toJSON')) {
 487                     // this may end up allowing unlimited recursion
 488                     // so we check the return value to make sure it's not got the same method.
 489                     $recode = $var->toJSON();
 490
 491                     if (method_exists($recode, 'toJSON')) {
 492
 493                         return ($this->use & SERVICES_JSON_SUPPRESS_ERRORS)
 494                         ? 'null'
 495                         : new Services_JSON_Error(get_class($var).
 496                             " toJSON returned an object with a toJSON method.");
 497
 498                     }
 499
 500                     return $this->_encode( $recode );
 501                 }
 502
 503                 $vars = get_object_vars($var);
 504
 505                 $properties = array_map(array($this, 'name_value'),
 506                                         array_keys($vars),
 507                                         array_values($vars));
 508
 509                 foreach($properties as $property) {
 510                     if(Services_JSON::isError($property)) {
 511                         return $property;
 512                     }
 513                 }
 514
 515                 return '{' . join(',', $properties) . '}';
 516
 517             default:
 518                 return ($this->use & SERVICES_JSON_SUPPRESS_ERRORS)
 519                     ? 'null'
 520                     : new Services_JSON_Error(gettype($var)." can not be encoded as JSON string");
 521         }
 522     }
 523
 524    /**
 525     * array-walking function for use in generating JSON-formatted name-value pairs
 526     *
 527     * @param    string  $name   name of key to use
 528     * @param    mixed   $value  reference to an array element to be encoded
 529     *
 530     * @return   string  JSON-formatted name-value pair, like '"name":value'
 531     * @access   private
 532     */
 533     function name_value($name, $value)
 534     {
 535         $encoded_value = $this->_encode($value);
 536
 537         if(Services_JSON::isError($encoded_value)) {
 538             return $encoded_value;
 539         }
 540
 541         return $this->_encode(strval($name)) . ':' . $encoded_value;
 542     }
 543
 544    /**
 545     * reduce a string by removing leading and trailing comments and whitespace
 546     *
 547     * @param    $str    string      string value to strip of comments and whitespace
 548     *
 549     * @return   string  string value stripped of comments and whitespace
 550     * @access   private
 551     */
 552     function reduce_string($str)
 553     {
 554         $str = preg_replace(array(
 555
 556                 // eliminate single line comments in '// ...' form
 557                 '#^\s*//(.+)$#m',
 558
 559                 // eliminate multi-line comments in '/* ... */' form, at start of string
 560                 '#^\s*/\*(.+)\*/#Us',
 561
 562                 // eliminate multi-line comments in '/* ... */' form, at end of string
 563                 '#/\*(.+)\*/\s*$#Us'
 564
 565             ), '', $str);
 566
 567         // eliminate extraneous space
 568         return trim($str);
 569     }
 570
 571    /**
 572     * decodes a JSON string into appropriate variable
 573     *
 574     * @param    string  $str    JSON-formatted string
 575     *
 576     * @return   mixed   number, boolean, string, array, or object
 577     *                   corresponding to given JSON input string.
 578     *                   See argument 1 to Services_JSON() above for object-output behavior.
 579     *                   Note that decode() always returns strings
 580     *                   in ASCII or UTF-8 format!
 581     * @access   public
 582     */
 583     function decode($str)
 584     {
 585         $str = $this->reduce_string($str);
 586
 587         switch (strtolower($str)) {
 588             case 'true':
 589                 return true;
 590
 591             case 'false':
 592                 return false;
 593
 594             case 'null':
 595                 return null;
 596
 597             default:
 598                 $m = array();
 599
 600                 if (is_numeric($str)) {
 601                     // Lookie-loo, it's a number
 602
 603                     // This would work on its own, but I'm trying to be
 604                     // good about returning integers where appropriate:
 605                     // return (float)$str;
 606
 607                     // Return float or int, as appropriate
 608                     return ((float)$str == (integer)$str)
 609                         ? (integer)$str
 610                         : (float)$str;
 611
 612                 } elseif (preg_match('/^("|\').*(\1)$/s', $str, $m) && $m[1] == $m[2]) {
 613                     // STRINGS RETURNED IN UTF-8 FORMAT
 614                     $delim = $this->substr8($str, 0, 1);
 615                     $chrs = $this->substr8($str, 1, -1);
 616                     $utf8 = '';
 617                     $strlen_chrs = $this->strlen8($chrs);
 618
 619                     for ($c = 0; $c < $strlen_chrs; ++$c) {
 620
 621                         $substr_chrs_c_2 = $this->substr8($chrs, $c, 2);
 622                         $ord_chrs_c = ord($chrs{$c});
 623
 624                         switch (true) {
 625                             case $substr_chrs_c_2 == '\b':
 626                                 $utf8 .= chr(0x08);
 627                                 ++$c;
 628                                 break;
 629                             case $substr_chrs_c_2 == '\t':
 630                                 $utf8 .= chr(0x09);
 631                                 ++$c;
 632                                 break;
 633                             case $substr_chrs_c_2 == '\n':
 634                                 $utf8 .= chr(0x0A);
 635                                 ++$c;
 636                                 break;
 637                             case $substr_chrs_c_2 == '\f':
 638                                 $utf8 .= chr(0x0C);
 639                                 ++$c;
 640                                 break;
 641                             case $substr_chrs_c_2 == '\r':
 642                                 $utf8 .= chr(0x0D);
 643                                 ++$c;
 644                                 break;
 645
 646                             case $substr_chrs_c_2 == '\\"':
 647                             case $substr_chrs_c_2 == '\\\'':
 648                             case $substr_chrs_c_2 == '\\\\':
 649                             case $substr_chrs_c_2 == '\\/':
 650                                 if (($delim == '"' && $substr_chrs_c_2 != '\\\'') ||
 651                                    ($delim == "'" && $substr_chrs_c_2 != '\\"')) {
 652                                     $utf8 .= $chrs{++$c};
 653                                 }
 654                                 break;
 655
 656                             case preg_match('/\\\u[0-9A-F]{4}/i', $this->substr8($chrs, $c, 6)):
 657                                 // single, escaped unicode character
 658                                 $utf16 = chr(hexdec($this->substr8($chrs, ($c + 2), 2)))
 659                                        . chr(hexdec($this->substr8($chrs, ($c + 4), 2)));
 660                                 $utf8 .= $this->utf162utf8($utf16);
 661                                 $c += 5;
 662                                 break;
 663
 664                             case ($ord_chrs_c >= 0x20) && ($ord_chrs_c <= 0x7F):
 665                                 $utf8 .= $chrs{$c};
 666                                 break;
 667
 668                             case ($ord_chrs_c & 0xE0) == 0xC0:
 669                                 // characters U-00000080 - U-000007FF, mask 110XXXXX
 670                                 //see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 671                                 $utf8 .= $this->substr8($chrs, $c, 2);
 672                                 ++$c;
 673                                 break;
 674
 675                             case ($ord_chrs_c & 0xF0) == 0xE0:
 676                                 // characters U-00000800 - U-0000FFFF, mask 1110XXXX
 677                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 678                                 $utf8 .= $this->substr8($chrs, $c, 3);
 679                                 $c += 2;
 680                                 break;
 681
 682                             case ($ord_chrs_c & 0xF8) == 0xF0:
 683                                 // characters U-00010000 - U-001FFFFF, mask 11110XXX
 684                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 685                                 $utf8 .= $this->substr8($chrs, $c, 4);
 686                                 $c += 3;
 687                                 break;
 688
 689                             case ($ord_chrs_c & 0xFC) == 0xF8:
 690                                 // characters U-00200000 - U-03FFFFFF, mask 111110XX
 691                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 692                                 $utf8 .= $this->substr8($chrs, $c, 5);
 693                                 $c += 4;
 694                                 break;
 695
 696                             case ($ord_chrs_c & 0xFE) == 0xFC:
 697                                 // characters U-04000000 - U-7FFFFFFF, mask 1111110X
 698                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 699                                 $utf8 .= $this->substr8($chrs, $c, 6);
 700                                 $c += 5;
 701                                 break;
 702
 703                         }
 704
 705                     }
 706
 707                     return $utf8;
 708
 709                 } elseif (preg_match('/^\[.*\]$/s', $str) || preg_match('/^\{.*\}$/s', $str)) {
 710                     // array, or object notation
 711
 712                     if ($str{0} == '[') {
 713                         $stk = array(SERVICES_JSON_IN_ARR);
 714                         $arr = array();
 715                     } else {
 716                         if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 717                             $stk = array(SERVICES_JSON_IN_OBJ);
 718                             $obj = array();
 719                         } else {
 720                             $stk = array(SERVICES_JSON_IN_OBJ);
 721                             $obj = new stdClass();
 722                         }
 723                     }
 724
 725                     array_push($stk, array('what'  => SERVICES_JSON_SLICE,
 726                                            'where' => 0,
 727                                            'delim' => false));
 728
 729                     $chrs = $this->substr8($str, 1, -1);
 730                     $chrs = $this->reduce_string($chrs);
 731
 732                     if ($chrs == '') {
 733                         if (reset($stk) == SERVICES_JSON_IN_ARR) {
 734                             return $arr;
 735
 736                         } else {
 737                             return $obj;
 738
 739                         }
 740                     }
 741
 742                     //print("\nparsing {$chrs}\n");
 743
 744                     $strlen_chrs = $this->strlen8($chrs);
 745
 746                     for ($c = 0; $c <= $strlen_chrs; ++$c) {
 747
 748                         $top = end($stk);
 749                         $substr_chrs_c_2 = $this->substr8($chrs, $c, 2);
 750
 751                         if (($c == $strlen_chrs) || (($chrs{$c} == ',') && ($top['what'] == SERVICES_JSON_SLICE))) {
 752                             // found a comma that is not inside a string, array, etc.,
 753                             // OR we've reached the end of the character list
 754                             $slice = $this->substr8($chrs, $top['where'], ($c - $top['where']));
 755                             array_push($stk, array('what' => SERVICES_JSON_SLICE, 'where' => ($c + 1), 'delim' => false));
 756                             //print("Found split at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 757
 758                             if (reset($stk) == SERVICES_JSON_IN_ARR) {
 759                                 // we are in an array, so just push an element onto the stack
 760                                 array_push($arr, $this->decode($slice));
 761
 762                             } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {
 763                                 // we are in an object, so figure
 764                                 // out the property name and set an
 765                                 // element in an associative array,
 766                                 // for now
 767                                 $parts = array();
 768
 769                                if (preg_match('/^\s*(["\'].*[^\\\]["\'])\s*:/Uis', $slice, $parts)) {
 770                                       // "name":value pair
 771                                     $key = $this->decode($parts[1]);
 772                                     $val = $this->decode(trim(substr($slice, strlen($parts[0])), ", \t\n\r\0\x0B"));
 773                                     if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 774                                         $obj[$key] = $val;
 775                                     } else {
 776                                         $obj->$key = $val;
 777                                     }
 778                                 } elseif (preg_match('/^\s*(\w+)\s*:/Uis', $slice, $parts)) {
 779                                     // name:value pair, where name is unquoted
 780                                     $key = $parts[1];
 781                                     $val = $this->decode(trim(substr($slice, strlen($parts[0])), ", \t\n\r\0\x0B"));
 782
 783                                     if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 784                                         $obj[$key] = $val;
 785                                     } else {
 786                                         $obj->$key = $val;
 787                                     }
 788                                 }
 789
 790                             }
 791
 792                         } elseif ((($chrs{$c} == '"') || ($chrs{$c} == "'")) && ($top['what'] != SERVICES_JSON_IN_STR)) {
 793                             // found a quote, and we are not inside a string
 794                             array_push($stk, array('what' => SERVICES_JSON_IN_STR, 'where' => $c, 'delim' => $chrs{$c}));
 795                             //print("Found start of string at {$c}\n");
 796
 797                         } elseif (($chrs{$c} == $top['delim']) &&
 798                                  ($top['what'] == SERVICES_JSON_IN_STR) &&
 799                                  (($this->strlen8($this->substr8($chrs, 0, $c)) - $this->strlen8(rtrim($this->substr8($chrs, 0, $c), '\\'))) % 2 != 1)) {
 800                             // found a quote, we're in a string, and it's not escaped
 801                             // we know that it's not escaped becase there is _not_ an
 802                             // odd number of backslashes at the end of the string so far
 803                             array_pop($stk);
 804                             //print("Found end of string at {$c}: ".$this->substr8($chrs, $top['where'], (1 + 1 + $c - $top['where']))."\n");
 805
 806                         } elseif (($chrs{$c} == '[') &&
 807                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 808                             // found a left-bracket, and we are in an array, object, or slice
 809                             array_push($stk, array('what' => SERVICES_JSON_IN_ARR, 'where' => $c, 'delim' => false));
 810                             //print("Found start of array at {$c}\n");
 811
 812                         } elseif (($chrs{$c} == ']') && ($top['what'] == SERVICES_JSON_IN_ARR)) {
 813                             // found a right-bracket, and we're in an array
 814                             array_pop($stk);
 815                             //print("Found end of array at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 816
 817                         } elseif (($chrs{$c} == '{') &&
 818                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 819                             // found a left-brace, and we are in an array, object, or slice
 820                             array_push($stk, array('what' => SERVICES_JSON_IN_OBJ, 'where' => $c, 'delim' => false));
 821                             //print("Found start of object at {$c}\n");
 822
 823                         } elseif (($chrs{$c} == '}') && ($top['what'] == SERVICES_JSON_IN_OBJ)) {
 824                             // found a right-brace, and we're in an object
 825                             array_pop($stk);
 826                             //print("Found end of object at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 827
 828                         } elseif (($substr_chrs_c_2 == '/*') &&
 829                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 830                             // found a comment start, and we are in an array, object, or slice
 831                             array_push($stk, array('what' => SERVICES_JSON_IN_CMT, 'where' => $c, 'delim' => false));
 832                             $c++;
 833                             //print("Found start of comment at {$c}\n");
 834
 835                         } elseif (($substr_chrs_c_2 == '*/') && ($top['what'] == SERVICES_JSON_IN_CMT)) {
 836                             // found a comment end, and we're in one now
 837                             array_pop($stk);
 838                             $c++;
 839
 840                             for ($i = $top['where']; $i <= $c; ++$i)
 841                                 $chrs = substr_replace($chrs, ' ', $i, 1);
 842
 843                             //print("Found end of comment at {$c}: ".$this->substr8($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 844
 845                         }
 846
 847                     }
 848
 849                     if (reset($stk) == SERVICES_JSON_IN_ARR) {
 850                         return $arr;
 851
 852                     } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {
 853                         return $obj;
 854
 855                     }
 856
 857                 }
 858         }
 859     }
 860
 861     /**
 862      * @todo Ultimately, this should just call PEAR::isError()
 863      */
 864     function isError($data, $code = null)
 865     {
 866         if (class_exists('pear')) {
 867             return PEAR::isError($data, $code);
 868         } elseif (is_object($data) && (get_class($data) == 'services_json_error' ||
 869                                  is_subclass_of($data, 'services_json_error'))) {
 870             return true;
 871         }
 872
 873         return false;
 874     }
 875
 876     /**
 877     * Calculates length of string in bytes
 878     * @param string
 879     * @return integer length
 880     */
 881     function strlen8( $str )
 882     {
 883         if ( $this->_mb_strlen ) {
 884             return mb_strlen( $str, "8bit" );
 885         }
 886         return strlen( $str );
 887     }
 888
 889     /**
 890     * Returns part of a string, interpreting $start and $length as number of bytes.
 891     * @param string
 892     * @param integer start
 893     * @param integer length
 894     * @return integer length
 895     */
 896     function substr8( $string, $start, $length=false )
 897     {
 898         if ( $length === false ) {
 899             $length = $this->strlen8( $string ) - $start;
 900         }
 901         if ( $this->_mb_substr ) {
 902             return mb_substr( $string, $start, $length, "8bit" );
 903         }
 904         return substr( $string, $start, $length );
 905     }
 906
 907 }
 908
 909 if (class_exists('PEAR_Error')) {
 910
 911     class Services_JSON_Error extends PEAR_Error
 912     {
 913         function Services_JSON_Error($message = 'unknown error', $code = null,
 914                                      $mode = null, $options = null, $userinfo = null)
 915         {
 916             parent::PEAR_Error($message, $code, $mode, $options, $userinfo);
 917         }
 918     }
 919
 920 } else {
 921
 922     /**
 923      * @todo Ultimately, this class shall be descended from PEAR_Error
 924      */
 925     class Services_JSON_Error
 926     {
 927         function Services_JSON_Error($message = 'unknown error', $code = null,
 928                                      $mode = null, $options = null, $userinfo = null)
 929         {
 930
 931         }
 932     }
 933
 934 }
 935
 936 endif;