includes/api/ApiFormatJson_json.php

   1 <?php
   2 /* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
   3
   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 * @ingroup API
  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: ApiFormatJson_json.php 45765 2009-01-15 10:18:44Z catrope $
  54 * @license http://www.opensource.org/licenses/bsd-license.php
  55 * @see 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  * Converts to and from JSON format.
  95  *
  96  * Brief example of use:
  97  *
  98  * <code>
  99  * // create a new instance of Services_JSON
 100  * $json = new Services_JSON();
 101  *
 102  * // convert a complexe value to JSON notation, and send it to the browser
 103  * $value = array('foo', 'bar', array(1, 2, 'baz'), array(3, array(4)));
 104  * $output = $json->encode($value);
 105  *
 106  * print($output);
 107  * // prints: ["foo","bar",[1,2,"baz"],[3,[4]]]
 108  *
 109  * // accept incoming POST data, assumed to be in JSON notation
 110  * $input = file_get_contents('php://input', 1000000);
 111  * $value = $json->decode($input);
 112  * </code>
 113  *
 114  * @ingroup API
 115  */
 116 class Services_JSON
 117 {
 118         /**
 119          * constructs a new JSON instance
 120          *
 121          * @param int $use object behavior flags; combine with boolean-OR
 122          *
 123          *      possible values:
 124          *      - SERVICES_JSON_LOOSE_TYPE:  loose typing.
 125          *                      "{...}" syntax creates associative arrays
 126          *                      instead of objects in decode().
 127          *      - SERVICES_JSON_SUPPRESS_ERRORS:  error suppression.
 128          *                      Values which can't be encoded (e.g. resources)
 129          *                      appear as NULL instead of throwing errors.
 130          *                      By default, a deeply-nested resource will
 131          *                      bubble up with an error, so all return values
 132          *                      from encode() should be checked with isError()
 133          */
 134         function Services_JSON($use = 0)
 135         {
 136                 $this->use = $use;
 137         }
 138
 139         /**
 140          * convert a string from one UTF-16 char to one UTF-8 char
 141          *
 142          * Normally should be handled by mb_convert_encoding, but
 143          * provides a slower PHP-only method for installations
 144          * that lack the multibye string extension.
 145          *
 146          * @param string  $utf16  UTF-16 character
 147          * @return string  UTF-8 character
 148          * @access private
 149          */
 150         function utf162utf8($utf16)
 151         {
 152                 // oh please oh please oh please oh please oh please
 153                 if(function_exists('mb_convert_encoding')) {
 154                         return mb_convert_encoding($utf16, 'UTF-8', 'UTF-16');
 155                 }
 156
 157                 $bytes = (ord($utf16{0}) << 8) | ord($utf16{1});
 158
 159                 switch(true) {
 160                         case ((0x7F & $bytes) == $bytes):
 161                                 // this case should never be reached, because we are in ASCII range
 162                                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 163                                 return chr(0x7F & $bytes);
 164
 165                         case (0x07FF & $bytes) == $bytes:
 166                                 // return a 2-byte UTF-8 character
 167                                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 168                                 return chr(0xC0 | (($bytes >> 6) & 0x1F))
 169                                          . chr(0x80 | ($bytes & 0x3F));
 170
 171                         case (0xFC00 & $bytes) == 0xD800 && strlen($utf16) >= 4 && (0xFC & ord($utf16{2})) == 0xDC:
 172                                 // return a 4-byte UTF-8 character
 173                                 $char = ((($bytes & 0x03FF) << 10)
 174                                            | ((ord($utf16{2}) & 0x03) << 8)
 175                                            | ord($utf16{3}));
 176                                 $char += 0x10000;
 177                                 return chr(0xF0 | (($char >> 18) & 0x07))
 178                                          . chr(0x80 | (($char >> 12) & 0x3F))
 179                                          . chr(0x80 | (($char >> 6) & 0x3F))
 180                                          . chr(0x80 | ($char & 0x3F));
 181
 182                         case (0xFFFF & $bytes) == $bytes:
 183                                 // return a 3-byte UTF-8 character
 184                                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 185                                 return chr(0xE0 | (($bytes >> 12) & 0x0F))
 186                                          . chr(0x80 | (($bytes >> 6) & 0x3F))
 187                                          . chr(0x80 | ($bytes & 0x3F));
 188                 }
 189
 190                 // ignoring UTF-32 for now, sorry
 191                 return '';
 192         }
 193
 194         /**
 195          * convert a string from one UTF-8 char to one UTF-16 char
 196          *
 197          * Normally should be handled by mb_convert_encoding, but
 198          * provides a slower PHP-only method for installations
 199          * that lack the multibye string extension.
 200          *
 201          * @param string  $utf8   UTF-8 character
 202          * @return string  UTF-16 character
 203          * @access private
 204          */
 205         function utf82utf16($utf8)
 206         {
 207                 // oh please oh please oh please oh please oh please
 208                 if(function_exists('mb_convert_encoding')) {
 209                         return mb_convert_encoding($utf8, 'UTF-16', 'UTF-8');
 210                 }
 211
 212                 switch(strlen($utf8)) {
 213                         case 1:
 214                                 // this case should never be reached, because we are in ASCII range
 215                                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 216                                 return $utf8;
 217
 218                         case 2:
 219                                 // return a UTF-16 character from a 2-byte UTF-8 char
 220                                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 221                                 return chr(0x07 & (ord($utf8{0}) >> 2))
 222                                          . chr((0xC0 & (ord($utf8{0}) << 6))
 223                                                  | (0x3F & ord($utf8{1})));
 224
 225                         case 3:
 226                                 // return a UTF-16 character from a 3-byte UTF-8 char
 227                                 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 228                                 return chr((0xF0 & (ord($utf8{0}) << 4))
 229                                                  | (0x0F & (ord($utf8{1}) >> 2)))
 230                                          . chr((0xC0 & (ord($utf8{1}) << 6))
 231                                                  | (0x7F & ord($utf8{2})));
 232
 233                         case 4:
 234                                 // return a UTF-16 surrogate pair from a 4-byte UTF-8 char
 235                                 if(ord($utf8{0}) > 0xF4) return ''; # invalid
 236                                 $char = ((0x1C0000 & (ord($utf8{0}) << 18))
 237                                            | (0x03F000 & (ord($utf8{1}) << 12))
 238                                            | (0x000FC0 & (ord($utf8{2}) << 6))
 239                                            | (0x00003F & ord($utf8{3})));
 240                                 if($char > 0x10FFFF) return ''; # invalid
 241                                 $char -= 0x10000;
 242                                 return chr(0xD8 | (($char >> 18) & 0x03))
 243                                          . chr(($char >> 10) & 0xFF)
 244                                          . chr(0xDC | (($char >> 8) & 0x03))
 245                                          . chr($char & 0xFF);
 246                 }
 247
 248                 // ignoring UTF-32 for now, sorry
 249                 return '';
 250         }
 251
 252         /**
 253          * encodes an arbitrary variable into JSON format
 254          *
 255          * @param mixed $var any number, boolean, string, array, or object to be encoded.
 256          *                      see argument 1 to Services_JSON() above for array-parsing behavior.
 257          *                      if var is a strng, note that encode() always expects it
 258          *                      to be in ASCII or UTF-8 format!
 259          * @param bool $pretty pretty-print output with indents and newlines
 260          *
 261          * @return mixed JSON string representation of input var or an error if a problem occurs
 262          * @access public
 263          */
 264         function encode($var, $pretty=false)
 265         {
 266                 $this->indent = 0;
 267                 $this->pretty = $pretty;
 268                 $this->nameValSeparator = $pretty ? ': ' : ':';
 269                 return $this->encode2($var);
 270         }
 271
 272         /**
 273          * encodes an arbitrary variable into JSON format
 274          *
 275          * @param mixed $var any number, boolean, string, array, or object to be encoded.
 276          *                      see argument 1 to Services_JSON() above for array-parsing behavior.
 277          *                      if var is a strng, note that encode() always expects it
 278          *                      to be in ASCII or UTF-8 format!
 279          *
 280          * @return mixed JSON string representation of input var or an error if a problem occurs
 281          * @access private
 282          */
 283         function encode2($var)
 284         {
 285                 if ($this->pretty) {
 286                         $close = "\n" . str_repeat("\t", $this->indent);
 287                         $open = $close . "\t";
 288                         $mid = ',' . $open;
 289                 }
 290                 else {
 291                         $open = $close = '';
 292                         $mid = ',';
 293                 }
 294
 295                 switch (gettype($var)) {
 296                         case 'boolean':
 297                                 return $var ? 'true' : 'false';
 298
 299                         case 'NULL':
 300                                 return 'null';
 301
 302                         case 'integer':
 303                                 return (int) $var;
 304
 305                         case 'double':
 306                         case 'float':
 307                                 return (float) $var;
 308
 309                         case 'string':
 310                                 // STRINGS ARE EXPECTED TO BE IN ASCII OR UTF-8 FORMAT
 311                                 $ascii = '';
 312                                 $strlen_var = strlen($var);
 313
 314                            /*
 315                                 * Iterate over every character in the string,
 316                                 * escaping with a slash or encoding to UTF-8 where necessary
 317                                 */
 318                                 for ($c = 0; $c < $strlen_var; ++$c) {
 319
 320                                         $ord_var_c = ord($var{$c});
 321
 322                                         switch (true) {
 323                                                 case $ord_var_c == 0x08:
 324                                                         $ascii .= '\b';
 325                                                         break;
 326                                                 case $ord_var_c == 0x09:
 327                                                         $ascii .= '\t';
 328                                                         break;
 329                                                 case $ord_var_c == 0x0A:
 330                                                         $ascii .= '\n';
 331                                                         break;
 332                                                 case $ord_var_c == 0x0C:
 333                                                         $ascii .= '\f';
 334                                                         break;
 335                                                 case $ord_var_c == 0x0D:
 336                                                         $ascii .= '\r';
 337                                                         break;
 338
 339                                                 case $ord_var_c == 0x22:
 340                                                 case $ord_var_c == 0x2F:
 341                                                 case $ord_var_c == 0x5C:
 342                                                         // double quote, slash, slosh
 343                                                         $ascii .= '\\'.$var{$c};
 344                                                         break;
 345
 346                                                 case (($ord_var_c >= 0x20) && ($ord_var_c <= 0x7F)):
 347                                                         // characters U-00000000 - U-0000007F (same as ASCII)
 348                                                         $ascii .= $var{$c};
 349                                                         break;
 350
 351                                                 case (($ord_var_c & 0xE0) == 0xC0):
 352                                                         // characters U-00000080 - U-000007FF, mask 110XXXXX
 353                                                         // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 354                                                         $char = pack('C*', $ord_var_c, ord($var{$c + 1}));
 355                                                         $c += 1;
 356                                                         $utf16 = $this->utf82utf16($char);
 357                                                         $ascii .= sprintf('\u%04s', bin2hex($utf16));
 358                                                         break;
 359
 360                                                 case (($ord_var_c & 0xF0) == 0xE0):
 361                                                         // characters U-00000800 - U-0000FFFF, mask 1110XXXX
 362                                                         // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 363                                                         $char = pack('C*', $ord_var_c,
 364                                                                         ord($var{$c + 1}),
 365                                                                         ord($var{$c + 2}));
 366                                                         $c += 2;
 367                                                         $utf16 = $this->utf82utf16($char);
 368                                                         $ascii .= sprintf('\u%04s', bin2hex($utf16));
 369                                                         break;
 370
 371                                                 case (($ord_var_c & 0xF8) == 0xF0):
 372                                                         // characters U-00010000 - U-001FFFFF, mask 11110XXX
 373                                                         // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 374                                                         // These will always return a surrogate pair
 375                                                         $char = pack('C*', $ord_var_c,
 376                                                                         ord($var{$c + 1}),
 377                                                                         ord($var{$c + 2}),
 378                                                                         ord($var{$c + 3}));
 379                                                         $c += 3;
 380                                                         $utf16 = $this->utf82utf16($char);
 381                                                         if($utf16 == '') {
 382                                                                 $ascii .= '\ufffd';
 383                                                         } else {
 384                                                                 $utf16 = str_split($utf16, 2);
 385                                                                 $ascii .= sprintf('\u%04s\u%04s', bin2hex($utf16[0]), bin2hex($utf16[1]));
 386                                                         }
 387                                                         break;
 388                                         }
 389                                 }
 390
 391                                 return '"'.$ascii.'"';
 392
 393                         case 'array':
 394                            /*
 395                                 * As per JSON spec if any array key is not an integer
 396                                 * we must treat the the whole array as an object. We
 397                                 * also try to catch a sparsely populated associative
 398                                 * array with numeric keys here because some JS engines
 399                                 * will create an array with empty indexes up to
 400                                 * max_index which can cause memory issues and because
 401                                 * the keys, which may be relevant, will be remapped
 402                                 * otherwise.
 403                                 *
 404                                 * As per the ECMA and JSON specification an object may
 405                                 * have any string as a property. Unfortunately due to
 406                                 * a hole in the ECMA specification if the key is a
 407                                 * ECMA reserved word or starts with a digit the
 408                                 * parameter is only accessible using ECMAScript's
 409                                 * bracket notation.
 410                                 */
 411
 412                                 // treat as a JSON object
 413                                 if (is_array($var) && count($var) && (array_keys($var) !== range(0, sizeof($var) - 1))) {
 414                                         $this->indent++;
 415                                         $properties = array_map(array($this, 'name_value'),
 416                                                                 array_keys($var),
 417                                                                 array_values($var));
 418                                         $this->indent--;
 419
 420                                         foreach($properties as $property) {
 421                                                 if(Services_JSON::isError($property)) {
 422                                                         return $property;
 423                                                 }
 424                                         }
 425
 426                                         return '{' . $open . join($mid, $properties) . $close . '}';
 427                                 }
 428
 429                                 // treat it like a regular array
 430                                 $this->indent++;
 431                                 $elements = array_map(array($this, 'encode2'), $var);
 432                                 $this->indent--;
 433
 434                                 foreach($elements as $element) {
 435                                         if(Services_JSON::isError($element)) {
 436                                                 return $element;
 437                                         }
 438                                 }
 439
 440                                 return '[' . $open . join($mid, $elements) . $close . ']';
 441
 442                         case 'object':
 443                                 $vars = get_object_vars($var);
 444
 445                                 $this->indent++;
 446                                 $properties = array_map(array($this, 'name_value'),
 447                                                         array_keys($vars),
 448                                                         array_values($vars));
 449                                 $this->indent--;
 450
 451                                 foreach($properties as $property) {
 452                                         if(Services_JSON::isError($property)) {
 453                                                 return $property;
 454                                         }
 455                                 }
 456
 457                                 return '{' . $open . join($mid, $properties) . $close . '}';
 458
 459                         default:
 460                                 return ($this->use & SERVICES_JSON_SUPPRESS_ERRORS)
 461                                         ? 'null'
 462                                         : new Services_JSON_Error(gettype($var)." can not be encoded as JSON string");
 463                 }
 464         }
 465
 466         /**
 467          * array-walking function for use in generating JSON-formatted name-value pairs
 468          *
 469          * @param string $name name of key to use
 470          * @param mixed $value reference to an array element to be encoded
 471          *
 472          * @return string JSON-formatted name-value pair, like '"name":value'
 473          * @access private
 474          */
 475         function name_value($name, $value)
 476         {
 477                 $encoded_value = $this->encode2($value);
 478
 479                 if(Services_JSON::isError($encoded_value)) {
 480                         return $encoded_value;
 481                 }
 482
 483                 return $this->encode2(strval($name)) . $this->nameValSeparator . $encoded_value;
 484         }
 485
 486         /**
 487          * reduce a string by removing leading and trailing comments and whitespace
 488          *
 489          * @param $str string string value to strip of comments and whitespace
 490          *
 491          * @return string string value stripped of comments and whitespace
 492          * @access private
 493          */
 494         function reduce_string($str)
 495         {
 496                 $str = preg_replace(array(
 497
 498                                 // eliminate single line comments in '// ...' form
 499                                 '#^\s*//(.+)$#m',
 500
 501                                 // eliminate multi-line comments in '/* ... */' form, at start of string
 502                                 '#^\s*/\*(.+)\*/#Us',
 503
 504                                 // eliminate multi-line comments in '/* ... */' form, at end of string
 505                                 '#/\*(.+)\*/\s*$#Us'
 506
 507                         ), '', $str);
 508
 509                 // eliminate extraneous space
 510                 return trim($str);
 511         }
 512
 513         /**
 514          * decodes a JSON string into appropriate variable
 515          *
 516          * @param string $str JSON-formatted string
 517          *
 518          * @return mixed number, boolean, string, array, or object
 519          *                 corresponding to given JSON input string.
 520          *                 See argument 1 to Services_JSON() above for object-output behavior.
 521          *                 Note that decode() always returns strings
 522          *                 in ASCII or UTF-8 format!
 523          * @access public
 524          */
 525         function decode($str)
 526         {
 527                 $str = $this->reduce_string($str);
 528
 529                 switch (strtolower($str)) {
 530                         case 'true':
 531                                 return true;
 532
 533                         case 'false':
 534                                 return false;
 535
 536                         case 'null':
 537                                 return null;
 538
 539                         default:
 540                                 $m = array();
 541
 542                                 if (is_numeric($str)) {
 543                                         // Lookie-loo, it's a number
 544
 545                                         // This would work on its own, but I'm trying to be
 546                                         // good about returning integers where appropriate:
 547                                         // return (float)$str;
 548
 549                                         // Return float or int, as appropriate
 550                                         return ((float)$str == (integer)$str)
 551                                                 ? (integer)$str
 552                                                 : (float)$str;
 553
 554                                 } elseif (preg_match('/^("|\').*(\1)$/s', $str, $m) && $m[1] == $m[2]) {
 555                                         // STRINGS RETURNED IN UTF-8 FORMAT
 556                                         $delim = substr($str, 0, 1);
 557                                         $chrs = substr($str, 1, -1);
 558                                         $utf8 = '';
 559                                         $strlen_chrs = strlen($chrs);
 560
 561                                         for ($c = 0; $c < $strlen_chrs; ++$c) {
 562
 563                                                 $substr_chrs_c_2 = substr($chrs, $c, 2);
 564                                                 $ord_chrs_c = ord($chrs{$c});
 565
 566                                                 switch (true) {
 567                                                         case $substr_chrs_c_2 == '\b':
 568                                                                 $utf8 .= chr(0x08);
 569                                                                 ++$c;
 570                                                                 break;
 571                                                         case $substr_chrs_c_2 == '\t':
 572                                                                 $utf8 .= chr(0x09);
 573                                                                 ++$c;
 574                                                                 break;
 575                                                         case $substr_chrs_c_2 == '\n':
 576                                                                 $utf8 .= chr(0x0A);
 577                                                                 ++$c;
 578                                                                 break;
 579                                                         case $substr_chrs_c_2 == '\f':
 580                                                                 $utf8 .= chr(0x0C);
 581                                                                 ++$c;
 582                                                                 break;
 583                                                         case $substr_chrs_c_2 == '\r':
 584                                                                 $utf8 .= chr(0x0D);
 585                                                                 ++$c;
 586                                                                 break;
 587
 588                                                         case $substr_chrs_c_2 == '\\"':
 589                                                         case $substr_chrs_c_2 == '\\\'':
 590                                                         case $substr_chrs_c_2 == '\\\\':
 591                                                         case $substr_chrs_c_2 == '\\/':
 592                                                                 if (($delim == '"' && $substr_chrs_c_2 != '\\\'') ||
 593                                                                    ($delim == "'" && $substr_chrs_c_2 != '\\"')) {
 594                                                                         $utf8 .= $chrs{++$c};
 595                                                                 }
 596                                                                 break;
 597
 598                                                         case preg_match('/\\\uD[89AB][0-9A-F]{2}\\\uD[C-F][0-9A-F]{2}/i', substr($chrs, $c, 12)):
 599                                                                 // escaped unicode surrogate pair
 600                                                                 $utf16 = chr(hexdec(substr($chrs, ($c + 2), 2)))
 601                                                                            . chr(hexdec(substr($chrs, ($c + 4), 2)))
 602                                                                            . chr(hexdec(substr($chrs, ($c + 8), 2)))
 603                                                                            . chr(hexdec(substr($chrs, ($c + 10), 2)));
 604                                                                 $utf8 .= $this->utf162utf8($utf16);
 605                                                                 $c += 11;
 606                                                                 break;
 607
 608                                                         case preg_match('/\\\u[0-9A-F]{4}/i', substr($chrs, $c, 6)):
 609                                                                 // single, escaped unicode character
 610                                                                 $utf16 = chr(hexdec(substr($chrs, ($c + 2), 2)))
 611                                                                            . chr(hexdec(substr($chrs, ($c + 4), 2)));
 612                                                                 $utf8 .= $this->utf162utf8($utf16);
 613                                                                 $c += 5;
 614                                                                 break;
 615
 616                                                         case ($ord_chrs_c >= 0x20) && ($ord_chrs_c <= 0x7F):
 617                                                                 $utf8 .= $chrs{$c};
 618                                                                 break;
 619
 620                                                         case ($ord_chrs_c & 0xE0) == 0xC0:
 621                                                                 // characters U-00000080 - U-000007FF, mask 110XXXXX
 622                                                                 //see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 623                                                                 $utf8 .= substr($chrs, $c, 2);
 624                                                                 ++$c;
 625                                                                 break;
 626
 627                                                         case ($ord_chrs_c & 0xF0) == 0xE0:
 628                                                                 // characters U-00000800 - U-0000FFFF, mask 1110XXXX
 629                                                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 630                                                                 $utf8 .= substr($chrs, $c, 3);
 631                                                                 $c += 2;
 632                                                                 break;
 633
 634                                                         case ($ord_chrs_c & 0xF8) == 0xF0:
 635                                                                 // characters U-00010000 - U-001FFFFF, mask 11110XXX
 636                                                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 637                                                                 $utf8 .= substr($chrs, $c, 4);
 638                                                                 $c += 3;
 639                                                                 break;
 640
 641                                                         case ($ord_chrs_c & 0xFC) == 0xF8:
 642                                                                 // characters U-00200000 - U-03FFFFFF, mask 111110XX
 643                                                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 644                                                                 $utf8 .= substr($chrs, $c, 5);
 645                                                                 $c += 4;
 646                                                                 break;
 647
 648                                                         case ($ord_chrs_c & 0xFE) == 0xFC:
 649                                                                 // characters U-04000000 - U-7FFFFFFF, mask 1111110X
 650                                                                 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
 651                                                                 $utf8 .= substr($chrs, $c, 6);
 652                                                                 $c += 5;
 653                                                                 break;
 654
 655                                                 }
 656
 657                                         }
 658
 659                                         return $utf8;
 660
 661                                 } elseif (preg_match('/^\[.*\]$/s', $str) || preg_match('/^\{.*\}$/s', $str)) {
 662                                         // array, or object notation
 663
 664                                         if ($str{0} == '[') {
 665                                                 $stk = array(SERVICES_JSON_IN_ARR);
 666                                                 $arr = array();
 667                                         } else {
 668                                                 if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 669                                                         $stk = array(SERVICES_JSON_IN_OBJ);
 670                                                         $obj = array();
 671                                                 } else {
 672                                                         $stk = array(SERVICES_JSON_IN_OBJ);
 673                                                         $obj = new stdClass();
 674                                                 }
 675                                         }
 676
 677                                         array_push($stk, array( 'what'  => SERVICES_JSON_SLICE,
 678                                                                 'where' => 0,
 679                                                                 'delim' => false));
 680
 681                                         $chrs = substr($str, 1, -1);
 682                                         $chrs = $this->reduce_string($chrs);
 683
 684                                         if ($chrs == '') {
 685                                                 if (reset($stk) == SERVICES_JSON_IN_ARR) {
 686                                                         return $arr;
 687
 688                                                 } else {
 689                                                         return $obj;
 690
 691                                                 }
 692                                         }
 693
 694                                         //print("\nparsing {$chrs}\n");
 695
 696                                         $strlen_chrs = strlen($chrs);
 697
 698                                         for ($c = 0; $c <= $strlen_chrs; ++$c) {
 699
 700                                                 $top = end($stk);
 701                                                 $substr_chrs_c_2 = substr($chrs, $c, 2);
 702
 703                                                 if (($c == $strlen_chrs) || (($chrs{$c} == ',') && ($top['what'] == SERVICES_JSON_SLICE))) {
 704                                                         // found a comma that is not inside a string, array, etc.,
 705                                                         // OR we've reached the end of the character list
 706                                                         $slice = substr($chrs, $top['where'], ($c - $top['where']));
 707                                                         array_push($stk, array('what' => SERVICES_JSON_SLICE, 'where' => ($c + 1), 'delim' => false));
 708                                                         //print("Found split at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 709
 710                                                         if (reset($stk) == SERVICES_JSON_IN_ARR) {
 711                                                                 // we are in an array, so just push an element onto the stack
 712                                                                 array_push($arr, $this->decode($slice));
 713
 714                                                         } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {
 715                                                                 // we are in an object, so figure
 716                                                                 // out the property name and set an
 717                                                                 // element in an associative array,
 718                                                                 // for now
 719                                                                 $parts = array();
 720
 721                                                                 if (preg_match('/^\s*(["\'].*[^\\\]["\'])\s*:\s*(\S.*),?$/Uis', $slice, $parts)) {
 722                                                                         // "name":value pair
 723                                                                         $key = $this->decode($parts[1]);
 724                                                                         $val = $this->decode($parts[2]);
 725
 726                                                                         if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 727                                                                                 $obj[$key] = $val;
 728                                                                         } else {
 729                                                                                 $obj->$key = $val;
 730                                                                         }
 731                                                                 } elseif (preg_match('/^\s*(\w+)\s*:\s*(\S.*),?$/Uis', $slice, $parts)) {
 732                                                                         // name:value pair, where name is unquoted
 733                                                                         $key = $parts[1];
 734                                                                         $val = $this->decode($parts[2]);
 735
 736                                                                         if ($this->use & SERVICES_JSON_LOOSE_TYPE) {
 737                                                                                 $obj[$key] = $val;
 738                                                                         } else {
 739                                                                                 $obj->$key = $val;
 740                                                                         }
 741                                                                 }
 742
 743                                                         }
 744
 745                                                 } elseif ((($chrs{$c} == '"') || ($chrs{$c} == "'")) && ($top['what'] != SERVICES_JSON_IN_STR)) {
 746                                                         // found a quote, and we are not inside a string
 747                                                         array_push($stk, array('what' => SERVICES_JSON_IN_STR, 'where' => $c, 'delim' => $chrs{$c}));
 748                                                         //print("Found start of string at {$c}\n");
 749
 750                                                 } elseif (($chrs{$c} == $top['delim']) &&
 751                                                                  ($top['what'] == SERVICES_JSON_IN_STR) &&
 752                                                                  (($chrs{$c - 1} != '\\') ||
 753                                                                  ($chrs{$c - 1} == '\\' && $chrs{$c - 2} == '\\'))) {
 754                                                         // found a quote, we're in a string, and it's not escaped
 755                                                         array_pop($stk);
 756                                                         //print("Found end of string at {$c}: ".substr($chrs, $top['where'], (1 + 1 + $c - $top['where']))."\n");
 757
 758                                                 } elseif (($chrs{$c} == '[') &&
 759                                                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 760                                                         // found a left-bracket, and we are in an array, object, or slice
 761                                                         array_push($stk, array('what' => SERVICES_JSON_IN_ARR, 'where' => $c, 'delim' => false));
 762                                                         //print("Found start of array at {$c}\n");
 763
 764                                                 } elseif (($chrs{$c} == ']') && ($top['what'] == SERVICES_JSON_IN_ARR)) {
 765                                                         // found a right-bracket, and we're in an array
 766                                                         array_pop($stk);
 767                                                         //print("Found end of array at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 768
 769                                                 } elseif (($chrs{$c} == '{') &&
 770                                                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 771                                                         // found a left-brace, and we are in an array, object, or slice
 772                                                         array_push($stk, array('what' => SERVICES_JSON_IN_OBJ, 'where' => $c, 'delim' => false));
 773                                                         //print("Found start of object at {$c}\n");
 774
 775                                                 } elseif (($chrs{$c} == '}') && ($top['what'] == SERVICES_JSON_IN_OBJ)) {
 776                                                         // found a right-brace, and we're in an object
 777                                                         array_pop($stk);
 778                                                         //print("Found end of object at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 779
 780                                                 } elseif (($substr_chrs_c_2 == '/*') &&
 781                                                                  in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
 782                                                         // found a comment start, and we are in an array, object, or slice
 783                                                         array_push($stk, array('what' => SERVICES_JSON_IN_CMT, 'where' => $c, 'delim' => false));
 784                                                         $c++;
 785                                                         //print("Found start of comment at {$c}\n");
 786
 787                                                 } elseif (($substr_chrs_c_2 == '*/') && ($top['what'] == SERVICES_JSON_IN_CMT)) {
 788                                                         // found a comment end, and we're in one now
 789                                                         array_pop($stk);
 790                                                         $c++;
 791
 792                                                         for ($i = $top['where']; $i <= $c; ++$i)
 793                                                                 $chrs = substr_replace($chrs, ' ', $i, 1);
 794
 795                                                         //print("Found end of comment at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
 796
 797                                                 }
 798
 799                                         }
 800
 801                                         if (reset($stk) == SERVICES_JSON_IN_ARR) {
 802                                                 return $arr;
 803
 804                                         } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {
 805                                                 return $obj;
 806
 807                                         }
 808
 809                                 }
 810                 }
 811         }
 812
 813         /**
 814          * @todo Ultimately, this should just call PEAR::isError()
 815          */
 816         function isError($data, $code = null)
 817         {
 818                 if (class_exists('pear')) {
 819                         return PEAR::isError($data, $code);
 820                 } elseif (is_object($data) && (get_class($data) == 'services_json_error' ||
 821                                 is_subclass_of($data, 'services_json_error'))) {
 822                         return true;
 823                 }
 824
 825                 return false;
 826         }
 827 }
 828
 829
 830 // Hide the PEAR_Error variant from Doxygen
 831 /// @cond
 832 if (class_exists('PEAR_Error')) {
 833
 834         /**
 835          * @ingroup API
 836          */
 837         class Services_JSON_Error extends PEAR_Error
 838         {
 839                 function Services_JSON_Error($message = 'unknown error', $code = null,
 840                                                 $mode = null, $options = null, $userinfo = null)
 841                 {
 842                         parent::PEAR_Error($message, $code, $mode, $options, $userinfo);
 843                 }
 844         }
 845
 846 } else {
 847 /// @endcond
 848
 849         /**
 850          * @todo Ultimately, this class shall be descended from PEAR_Error
 851          * @ingroup API
 852          */
 853         class Services_JSON_Error
 854         {
 855                 function Services_JSON_Error($message = 'unknown error', $code = null,
 856                                                 $mode = null, $options = null, $userinfo = null)
 857                 {
 858
 859                 }
 860         }
 861 }