lib/pear/JSON.php

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