2 /* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
5 * Converts to and from JSON format.
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.
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
25 * All strings should be in ASCII or UTF-8 format!
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
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
49 * @package Services_JSON
50 * @author Michal Migurski <mike-json@teczno.com>
51 * @author Matt Knapp <mdknapp[at]gmail[dot]com>
52 * @author Brett Stimmerman <brettstimmerman[at]gmail[dot]com>
53 * @copyright 2005 Michal Migurski
54 * @license http://www.opensource.org/licenses/bsd-license.php
55 * @link http://pear.php.net/pepr/pepr-proposal-show.php?id=198
59 * Marker constant for Services_JSON::decode(), used to flag stack state
61 define('SERVICES_JSON_SLICE', 1);
64 * Marker constant for Services_JSON::decode(), used to flag stack state
66 define('SERVICES_JSON_IN_STR', 2);
69 * Marker constant for Services_JSON::decode(), used to flag stack state
71 define('SERVICES_JSON_IN_ARR', 4);
74 * Marker constant for Services_JSON::decode(), used to flag stack state
76 define('SERVICES_JSON_IN_OBJ', 8);
79 * Marker constant for Services_JSON::decode(), used to flag stack state
81 define('SERVICES_JSON_IN_CMT', 16);
84 * Behavior switch for Services_JSON::decode()
86 define('SERVICES_JSON_LOOSE_TYPE', 10);
89 * Behavior switch for Services_JSON::decode()
91 define('SERVICES_JSON_STRICT_TYPE', 11);
94 * Converts to and from JSON format.
96 * Brief example of use:
99 * // create a new instance of Services_JSON
100 * $json = new Services_JSON();
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);
107 * // prints: ["foo","bar",[1,2,"baz"],[3,[4]]]
109 * // accept incoming POST data, assumed to be in JSON notation
110 * $input = file_get_contents('php://input', 1000000);
111 * $value = $json->decode($input);
117 * constructs a new JSON instance
119 * @param int $use object behavior: when encoding or decoding,
120 * be loose or strict about object/array usage
123 * - SERVICES_JSON_STRICT_TYPE: strict typing, default.
124 * "{...}" syntax creates objects in decode().
125 * - SERVICES_JSON_LOOSE_TYPE: loose typing.
126 * "{...}" syntax creates associative arrays in decode().
128 function Services_JSON($use = SERVICES_JSON_STRICT_TYPE)
134 * convert a string from one UTF-16 char to one UTF-8 char
136 * Normally should be handled by mb_convert_encoding, but
137 * provides a slower PHP-only method for installations
138 * that lack the multibye string extension.
140 * @param string $utf16 UTF-16 character
141 * @return string UTF-8 character
144 function utf162utf8($utf16)
146 // oh please oh please oh please oh please oh please
147 if(function_exists('mb_convert_encoding'))
148 return mb_convert_encoding($utf16, 'UTF-8', 'UTF-16');
150 $bytes = (ord($utf16{0}) << 8) | ord($utf16{1});
153 case ((0x7F & $bytes) == $bytes):
154 // this case should never be reached, because we are in ASCII range
155 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
156 return chr(0x7F & $bytes);
158 case (0x07FF & $bytes) == $bytes:
159 // return a 2-byte UTF-8 character
160 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
161 return chr(0xC0 | (($bytes >> 6) & 0x1F))
162 . chr(0x80 | ($bytes & 0x3F));
164 case (0xFFFF & $bytes) == $bytes:
165 // return a 3-byte UTF-8 character
166 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
167 return chr(0xE0 | (($bytes >> 12) & 0x0F))
168 . chr(0x80 | (($bytes >> 6) & 0x3F))
169 . chr(0x80 | ($bytes & 0x3F));
172 // ignoring UTF-32 for now, sorry
177 * convert a string from one UTF-8 char to one UTF-16 char
179 * Normally should be handled by mb_convert_encoding, but
180 * provides a slower PHP-only method for installations
181 * that lack the multibye string extension.
183 * @param string $utf8 UTF-8 character
184 * @return string UTF-16 character
187 function utf82utf16($utf8)
189 // oh please oh please oh please oh please oh please
190 if(function_exists('mb_convert_encoding'))
191 return mb_convert_encoding($utf8, 'UTF-16', 'UTF-8');
193 switch(strlen($utf8)) {
195 // this case should never be reached, because we are in ASCII range
196 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
200 // return a UTF-16 character from a 2-byte UTF-8 char
201 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
202 return chr(0x07 & (ord($utf8{0}) >> 2))
203 . chr((0xC0 & (ord($utf8{0}) << 6))
204 | (0x3F & ord($utf8{1})));
207 // return a UTF-16 character from a 3-byte UTF-8 char
208 // see: http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
209 return chr((0xF0 & (ord($utf8{0}) << 4))
210 | (0x0F & (ord($utf8{1}) >> 2)))
211 . chr((0xC0 & (ord($utf8{1}) << 6))
212 | (0x7F & ord($utf8{2})));
215 // ignoring UTF-32 for now, sorry
220 * encodes an arbitrary variable into JSON format
222 * @param mixed $var any number, boolean, string, array, or object to be encoded.
223 * see argument 1 to Services_JSON() above for array-parsing behavior.
224 * if var is a strng, note that encode() always expects it
225 * to be in ASCII or UTF-8 format!
227 * @return string JSON string representation of input var
230 function encode($var)
232 switch (gettype($var)) {
234 return $var ? 'true' : 'false';
247 // STRINGS ARE EXPECTED TO BE IN ASCII OR UTF-8 FORMAT
249 $strlen_var = strlen($var);
252 * Iterate over every character in the string,
253 * escaping with a slash or encoding to UTF-8 where necessary
255 for ($c = 0; $c < $strlen_var; ++$c) {
257 $ord_var_c = ord($var{$c});
260 case $ord_var_c == 0x08:
263 case $ord_var_c == 0x09:
266 case $ord_var_c == 0x0A:
269 case $ord_var_c == 0x0C:
272 case $ord_var_c == 0x0D:
276 case $ord_var_c == 0x22:
277 case $ord_var_c == 0x2F:
278 case $ord_var_c == 0x5C:
279 // double quote, slash, slosh
280 $ascii .= '\\'.$var{$c};
283 case (($ord_var_c >= 0x20) && ($ord_var_c <= 0x7F)):
284 // characters U-00000000 - U-0000007F (same as ASCII)
288 case (($ord_var_c & 0xE0) == 0xC0):
289 // characters U-00000080 - U-000007FF, mask 110XXXXX
290 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
291 $char = pack('C*', $ord_var_c, ord($var{$c + 1}));
293 $utf16 = $this->utf82utf16($char);
294 $ascii .= sprintf('\u%04s', bin2hex($utf16));
297 case (($ord_var_c & 0xF0) == 0xE0):
298 // characters U-00000800 - U-0000FFFF, mask 1110XXXX
299 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
300 $char = pack('C*', $ord_var_c,
304 $utf16 = $this->utf82utf16($char);
305 $ascii .= sprintf('\u%04s', bin2hex($utf16));
308 case (($ord_var_c & 0xF8) == 0xF0):
309 // characters U-00010000 - U-001FFFFF, mask 11110XXX
310 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
311 $char = pack('C*', $ord_var_c,
316 $utf16 = $this->utf82utf16($char);
317 $ascii .= sprintf('\u%04s', bin2hex($utf16));
320 case (($ord_var_c & 0xFC) == 0xF8):
321 // characters U-00200000 - U-03FFFFFF, mask 111110XX
322 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
323 $char = pack('C*', $ord_var_c,
329 $utf16 = $this->utf82utf16($char);
330 $ascii .= sprintf('\u%04s', bin2hex($utf16));
333 case (($ord_var_c & 0xFE) == 0xFC):
334 // characters U-04000000 - U-7FFFFFFF, mask 1111110X
335 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
336 $char = pack('C*', $ord_var_c,
343 $utf16 = $this->utf82utf16($char);
344 $ascii .= sprintf('\u%04s', bin2hex($utf16));
349 return '"'.$ascii.'"';
353 * As per JSON spec if any array key is not an integer
354 * we must treat the the whole array as an object. We
355 * also try to catch a sparsely populated associative
356 * array with numeric keys here because some JS engines
357 * will create an array with empty indexes up to
358 * max_index which can cause memory issues and because
359 * the keys, which may be relevant, will be remapped
362 * As per the ECMA and JSON specification an object may
363 * have any string as a property. Unfortunately due to
364 * a hole in the ECMA specification if the key is a
365 * ECMA reserved word or starts with a digit the
366 * parameter is only accessible using ECMAScript's
370 // treat as a JSON object
371 if (is_array($var) && count($var) && (array_keys($var) !== range(0, sizeof($var) - 1))) {
373 join(',', array_map(array($this, 'name_value'),
379 // treat it like a regular array
380 return '[' . join(',', array_map(array($this, 'encode'), $var)) . ']';
383 $vars = get_object_vars($var);
385 join(',', array_map(array($this, 'name_value'),
387 array_values($vars)))
396 * array-walking function for use in generating JSON-formatted name-value pairs
398 * @param string $name name of key to use
399 * @param mixed $value reference to an array element to be encoded
401 * @return string JSON-formatted name-value pair, like '"name":value'
404 function name_value($name, $value)
406 return $this->encode(strval($name)) . ':' . $this->encode($value);
410 * reduce a string by removing leading and trailing comments and whitespace
412 * @param $str string string value to strip of comments and whitespace
414 * @return string string value stripped of comments and whitespace
417 function reduce_string($str)
419 $str = preg_replace(array(
421 // eliminate single line comments in '// ...' form
424 // eliminate multi-line comments in '/* ... */' form, at start of string
425 '#^\s*/\*(.+)\*/#Us',
427 // eliminate multi-line comments in '/* ... */' form, at end of string
432 // eliminate extraneous space
437 * decodes a JSON string into appropriate variable
439 * @param string $str JSON-formatted string
441 * @return mixed number, boolean, string, array, or object
442 * corresponding to given JSON input string.
443 * See argument 1 to Services_JSON() above for object-output behavior.
444 * Note that decode() always returns strings
445 * in ASCII or UTF-8 format!
448 function decode($str)
450 $str = $this->reduce_string($str);
452 switch (strtolower($str)) {
463 if (is_numeric($str)) {
464 // Lookie-loo, it's a number
466 // This would work on its own, but I'm trying to be
467 // good about returning integers where appropriate:
468 // return (float)$str;
470 // Return float or int, as appropriate
471 return ((float)$str == (integer)$str)
475 } elseif (preg_match('/^("|\').*(\1)$/s', $str, $m) && $m[1] == $m[2]) {
476 // STRINGS RETURNED IN UTF-8 FORMAT
477 $delim = substr($str, 0, 1);
478 $chrs = substr($str, 1, -1);
480 $strlen_chrs = strlen($chrs);
482 for ($c = 0; $c < $strlen_chrs; ++$c) {
484 $substr_chrs_c_2 = substr($chrs, $c, 2);
485 $ord_chrs_c = ord($chrs{$c});
488 case $substr_chrs_c_2 == '\b':
492 case $substr_chrs_c_2 == '\t':
496 case $substr_chrs_c_2 == '\n':
500 case $substr_chrs_c_2 == '\f':
504 case $substr_chrs_c_2 == '\r':
509 case $substr_chrs_c_2 == '\\"':
510 case $substr_chrs_c_2 == '\\\'':
511 case $substr_chrs_c_2 == '\\\\':
512 case $substr_chrs_c_2 == '\\/':
513 if (($delim == '"' && $substr_chrs_c_2 != '\\\'') ||
514 ($delim == "'" && $substr_chrs_c_2 != '\\"')) {
515 $utf8 .= $chrs{++$c};
519 case preg_match('/\\\u[0-9A-F]{4}/i', substr($chrs, $c, 6)):
520 // single, escaped unicode character
521 $utf16 = chr(hexdec(substr($chrs, ($c + 2), 2)))
522 . chr(hexdec(substr($chrs, ($c + 4), 2)));
523 $utf8 .= $this->utf162utf8($utf16);
527 case ($ord_chrs_c >= 0x20) && ($ord_chrs_c <= 0x7F):
531 case ($ord_chrs_c & 0xE0) == 0xC0:
532 // characters U-00000080 - U-000007FF, mask 110XXXXX
533 //see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
534 $utf8 .= substr($chrs, $c, 2);
538 case ($ord_chrs_c & 0xF0) == 0xE0:
539 // characters U-00000800 - U-0000FFFF, mask 1110XXXX
540 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
541 $utf8 .= substr($chrs, $c, 3);
545 case ($ord_chrs_c & 0xF8) == 0xF0:
546 // characters U-00010000 - U-001FFFFF, mask 11110XXX
547 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
548 $utf8 .= substr($chrs, $c, 4);
552 case ($ord_chrs_c & 0xFC) == 0xF8:
553 // characters U-00200000 - U-03FFFFFF, mask 111110XX
554 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
555 $utf8 .= substr($chrs, $c, 5);
559 case ($ord_chrs_c & 0xFE) == 0xFC:
560 // characters U-04000000 - U-7FFFFFFF, mask 1111110X
561 // see http://www.cl.cam.ac.uk/~mgk25/unicode.html#utf-8
562 $utf8 .= substr($chrs, $c, 6);
572 } elseif (preg_match('/^\[.*\]$/s', $str) || preg_match('/^\{.*\}$/s', $str)) {
573 // array, or object notation
575 if ($str{0} == '[') {
576 $stk = array(SERVICES_JSON_IN_ARR);
579 if ($this->use == SERVICES_JSON_LOOSE_TYPE) {
580 $stk = array(SERVICES_JSON_IN_OBJ);
583 $stk = array(SERVICES_JSON_IN_OBJ);
584 $obj = new stdClass();
588 array_push($stk, array('what' => SERVICES_JSON_SLICE,
592 $chrs = substr($str, 1, -1);
593 $chrs = $this->reduce_string($chrs);
596 if (reset($stk) == SERVICES_JSON_IN_ARR) {
605 //print("\nparsing {$chrs}\n");
607 $strlen_chrs = strlen($chrs);
609 for ($c = 0; $c <= $strlen_chrs; ++$c) {
612 $substr_chrs_c_2 = substr($chrs, $c, 2);
614 if (($c == $strlen_chrs) || (($chrs{$c} == ',') && ($top['what'] == SERVICES_JSON_SLICE))) {
615 // found a comma that is not inside a string, array, etc.,
616 // OR we've reached the end of the character list
617 $slice = substr($chrs, $top['where'], ($c - $top['where']));
618 array_push($stk, array('what' => SERVICES_JSON_SLICE, 'where' => ($c + 1), 'delim' => false));
619 //print("Found split at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
621 if (reset($stk) == SERVICES_JSON_IN_ARR) {
622 // we are in an array, so just push an element onto the stack
623 array_push($arr, $this->decode($slice));
625 } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {
626 // we are in an object, so figure
627 // out the property name and set an
628 // element in an associative array,
630 if (preg_match('/^\s*(["\'].*[^\\\]["\'])\s*:\s*(\S.*),?$/Uis', $slice, $parts)) {
632 $key = $this->decode($parts[1]);
633 $val = $this->decode($parts[2]);
635 if ($this->use == SERVICES_JSON_LOOSE_TYPE) {
640 } elseif (preg_match('/^\s*(\w+)\s*:\s*(\S.*),?$/Uis', $slice, $parts)) {
641 // name:value pair, where name is unquoted
643 $val = $this->decode($parts[2]);
645 if ($this->use == SERVICES_JSON_LOOSE_TYPE) {
654 } elseif ((($chrs{$c} == '"') || ($chrs{$c} == "'")) && ($top['what'] != SERVICES_JSON_IN_STR)) {
655 // found a quote, and we are not inside a string
656 array_push($stk, array('what' => SERVICES_JSON_IN_STR, 'where' => $c, 'delim' => $chrs{$c}));
657 //print("Found start of string at {$c}\n");
659 } elseif (($chrs{$c} == $top['delim']) &&
660 ($top['what'] == SERVICES_JSON_IN_STR) &&
661 (($chrs{$c - 1} != '\\') ||
662 ($chrs{$c - 1} == '\\' && $chrs{$c - 2} == '\\'))) {
663 // found a quote, we're in a string, and it's not escaped
665 //print("Found end of string at {$c}: ".substr($chrs, $top['where'], (1 + 1 + $c - $top['where']))."\n");
667 } elseif (($chrs{$c} == '[') &&
668 in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
669 // found a left-bracket, and we are in an array, object, or slice
670 array_push($stk, array('what' => SERVICES_JSON_IN_ARR, 'where' => $c, 'delim' => false));
671 //print("Found start of array at {$c}\n");
673 } elseif (($chrs{$c} == ']') && ($top['what'] == SERVICES_JSON_IN_ARR)) {
674 // found a right-bracket, and we're in an array
676 //print("Found end of array at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
678 } elseif (($chrs{$c} == '{') &&
679 in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
680 // found a left-brace, and we are in an array, object, or slice
681 array_push($stk, array('what' => SERVICES_JSON_IN_OBJ, 'where' => $c, 'delim' => false));
682 //print("Found start of object at {$c}\n");
684 } elseif (($chrs{$c} == '}') && ($top['what'] == SERVICES_JSON_IN_OBJ)) {
685 // found a right-brace, and we're in an object
687 //print("Found end of object at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
689 } elseif (($substr_chrs_c_2 == '/*') &&
690 in_array($top['what'], array(SERVICES_JSON_SLICE, SERVICES_JSON_IN_ARR, SERVICES_JSON_IN_OBJ))) {
691 // found a comment start, and we are in an array, object, or slice
692 array_push($stk, array('what' => SERVICES_JSON_IN_CMT, 'where' => $c, 'delim' => false));
694 //print("Found start of comment at {$c}\n");
696 } elseif (($substr_chrs_c_2 == '*/') && ($top['what'] == SERVICES_JSON_IN_CMT)) {
697 // found a comment end, and we're in one now
701 for ($i = $top['where']; $i <= $c; ++$i)
702 $chrs = substr_replace($chrs, ' ', $i, 1);
704 //print("Found end of comment at {$c}: ".substr($chrs, $top['where'], (1 + $c - $top['where']))."\n");
710 if (reset($stk) == SERVICES_JSON_IN_ARR) {
713 } elseif (reset($stk) == SERVICES_JSON_IN_OBJ) {