includes/json/FormatJson.php

   1 <?php
   2 /**
   3  * Wrapper for json_encode and json_decode.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 /**
  24  * JSON formatter wrapper class
  25  */
  26 class FormatJson {
  27         /**
  28          * Skip escaping most characters above U+007F for readability and compactness.
  29          * This encoding option saves 3 to 8 bytes (uncompressed) for each such character;
  30          * however, it could break compatibility with systems that incorrectly handle UTF-8.
  31          *
  32          * @since 1.22
  33          */
  34         const UTF8_OK = 1;
  35
  36         /**
  37          * Skip escaping the characters '<', '>', and '&', which have special meanings in
  38          * HTML and XML.
  39          *
  40          * @warning Do not use this option for JSON that could end up in inline scripts.
  41          * - HTML5, §4.3.1.2 Restrictions for contents of script elements
  42          * - XML 1.0 (5th Ed.), §2.4 Character Data and Markup
  43          *
  44          * @since 1.22
  45          */
  46         const XMLMETA_OK = 2;
  47
  48         /**
  49          * Skip escaping as many characters as reasonably possible.
  50          *
  51          * @warning When generating inline script blocks, use FormatJson::UTF8_OK instead.
  52          *
  53          * @since 1.22
  54          */
  55         const ALL_OK = 3;
  56
  57         /**
  58          * If set, treat json objects '{...}' as associative arrays. Without this option,
  59          * json objects will be converted to stdClass.
  60          * The value is set to 1 to be backward compatible with 'true' that was used before.
  61          *
  62          * @since 1.24
  63          */
  64         const FORCE_ASSOC = 0x100;
  65
  66         /**
  67          * If set, attempts to fix invalid json.
  68          *
  69          * @since 1.24
  70          */
  71         const TRY_FIXING = 0x200;
  72
  73         /**
  74          * If set, strip comments from input before parsing as JSON.
  75          *
  76          * @since 1.25
  77          */
  78         const STRIP_COMMENTS = 0x400;
  79
  80         /**
  81          * Regex that matches whitespace inside empty arrays and objects.
  82          *
  83          * This doesn't affect regular strings inside the JSON because those can't
  84          * have a real line break (\n) in them, at this point they are already escaped
  85          * as the string "\n" which this doesn't match.
  86          *
  87          * @private
  88          */
  89         const WS_CLEANUP_REGEX = '/(?<=[\[{])\n\s*+(?=[\]}])/';
  90
  91         /**
  92          * Characters problematic in JavaScript.
  93          *
  94          * @note These are listed in ECMA-262 (5.1 Ed.), §7.3 Line Terminators along with U+000A (LF)
  95          *       and U+000D (CR). However, PHP already escapes LF and CR according to RFC 4627.
  96          */
  97         private static $badChars = [
  98                 "\xe2\x80\xa8", // U+2028 LINE SEPARATOR
  99                 "\xe2\x80\xa9", // U+2029 PARAGRAPH SEPARATOR
 100         ];
 101
 102         /**
 103          * Escape sequences for characters listed in FormatJson::$badChars.
 104          */
 105         private static $badCharsEscaped = [
 106                 '\u2028', // U+2028 LINE SEPARATOR
 107                 '\u2029', // U+2029 PARAGRAPH SEPARATOR
 108         ];
 109
 110         /**
 111          * Returns the JSON representation of a value.
 112          *
 113          * @note Empty arrays are encoded as numeric arrays, not as objects, so cast any associative
 114          *       array that might be empty to an object before encoding it.
 115          *
 116          * @note In pre-1.22 versions of MediaWiki, using this function for generating inline script
 117          *       blocks may result in an XSS vulnerability, and quite likely will in XML documents
 118          *       (cf. FormatJson::XMLMETA_OK). Use Xml::encodeJsVar() instead in such cases.
 119          *
 120          * @param mixed $value The value to encode. Can be any type except a resource.
 121          * @param string|bool $pretty If a string, add non-significant whitespace to improve
 122          *   readability, using that string for indentation. If true, use the default indent
 123          *   string (four spaces).
 124          * @param int $escaping Bitfield consisting of _OK class constants
 125          * @return string|false String if successful; false upon failure
 126          */
 127         public static function encode( $value, $pretty = false, $escaping = 0 ) {
 128                 if ( !is_string( $pretty ) ) {
 129                         $pretty = $pretty ? '    ' : false;
 130                 }
 131
 132                 static $bug66021;
 133                 if ( $pretty !== false && $bug66021 === null ) {
 134                         $bug66021 = json_encode( [], JSON_PRETTY_PRINT ) !== '[]';
 135                 }
 136
 137                 // PHP escapes '/' to prevent breaking out of inline script blocks using '</script>',
 138                 // which is hardly useful when '<' and '>' are escaped (and inadequate), and such
 139                 // escaping negatively impacts the human readability of URLs and similar strings.
 140                 $options = JSON_UNESCAPED_SLASHES;
 141                 $options |= $pretty !== false ? JSON_PRETTY_PRINT : 0;
 142                 $options |= ( $escaping & self::UTF8_OK ) ? JSON_UNESCAPED_UNICODE : 0;
 143                 $options |= ( $escaping & self::XMLMETA_OK ) ? 0 : ( JSON_HEX_TAG | JSON_HEX_AMP );
 144                 $json = json_encode( $value, $options );
 145                 if ( $json === false ) {
 146                         return false;
 147                 }
 148
 149                 if ( $pretty !== false ) {
 150                         // Workaround for <https://bugs.php.net/bug.php?id=66021>
 151                         if ( $bug66021 ) {
 152                                 $json = preg_replace( self::WS_CLEANUP_REGEX, '', $json );
 153                         }
 154                         if ( $pretty !== '    ' ) {
 155                                 // Change the four-space indent to a tab indent
 156                                 $json = str_replace( "\n    ", "\n\t", $json );
 157                                 while ( strpos( $json, "\t    " ) !== false ) {
 158                                         $json = str_replace( "\t    ", "\t\t", $json );
 159                                 }
 160
 161                                 if ( $pretty !== "\t" ) {
 162                                         // Change the tab indent to the provided indent
 163                                         $json = str_replace( "\t", $pretty, $json );
 164                                 }
 165                         }
 166                 }
 167                 if ( $escaping & self::UTF8_OK ) {
 168                         $json = str_replace( self::$badChars, self::$badCharsEscaped, $json );
 169                 }
 170
 171                 return $json;
 172         }
 173
 174         /**
 175          * Decodes a JSON string. It is recommended to use FormatJson::parse(),
 176          * which returns more comprehensive result in case of an error, and has
 177          * more parsing options.
 178          *
 179          * @param string $value The JSON string being decoded
 180          * @param bool $assoc When true, returned objects will be converted into associative arrays.
 181          *
 182          * @return mixed The value encoded in JSON in appropriate PHP type.
 183          * `null` is returned if $value represented `null`, if $value could not be decoded,
 184          * or if the encoded data was deeper than the recursion limit.
 185          * Use FormatJson::parse() to distinguish between types of `null` and to get proper error code.
 186          */
 187         public static function decode( $value, $assoc = false ) {
 188                 return json_decode( $value, $assoc );
 189         }
 190
 191         /**
 192          * Decodes a JSON string.
 193          * Unlike FormatJson::decode(), if $value represents null value, it will be
 194          * properly decoded as valid.
 195          *
 196          * @param string $value The JSON string being decoded
 197          * @param int $options A bit field that allows FORCE_ASSOC, TRY_FIXING,
 198          * STRIP_COMMENTS
 199          * @return Status If valid JSON, the value is available in $result->getValue()
 200          */
 201         public static function parse( $value, $options = 0 ) {
 202                 if ( $options & self::STRIP_COMMENTS ) {
 203                         $value = self::stripComments( $value );
 204                 }
 205                 $assoc = ( $options & self::FORCE_ASSOC ) !== 0;
 206                 $result = json_decode( $value, $assoc );
 207                 $code = json_last_error();
 208
 209                 if ( $code === JSON_ERROR_SYNTAX && ( $options & self::TRY_FIXING ) !== 0 ) {
 210                         // The most common error is the trailing comma in a list or an object.
 211                         // We cannot simply replace /,\s*[}\]]/ because it could be inside a string value.
 212                         // But we could use the fact that JSON does not allow multi-line string values,
 213                         // And remove trailing commas if they are et the end of a line.
 214                         // JSON only allows 4 control characters: [ \t\r\n].  So we must not use '\s' for matching.
 215                         // Regex match   ,]<any non-quote chars>\n   or   ,\n]   with optional spaces/tabs.
 216                         $count = 0;
 217                         $value =
 218                                 preg_replace( '/,([ \t]*[}\]][^"\r\n]*([\r\n]|$)|[ \t]*[\r\n][ \t\r\n]*[}\]])/', '$1',
 219                                         $value, -1, $count );
 220                         if ( $count > 0 ) {
 221                                 $result = json_decode( $value, $assoc );
 222                                 if ( JSON_ERROR_NONE === json_last_error() ) {
 223                                         // Report warning
 224                                         $st = Status::newGood( $result );
 225                                         $st->warning( wfMessage( 'json-warn-trailing-comma' )->numParams( $count ) );
 226                                         return $st;
 227                                 }
 228                         }
 229                 }
 230
 231                 switch ( $code ) {
 232                         case JSON_ERROR_NONE:
 233                                 return Status::newGood( $result );
 234                         default:
 235                                 return Status::newFatal( wfMessage( 'json-error-unknown' )->numParams( $code ) );
 236                         case JSON_ERROR_DEPTH:
 237                                 $msg = 'json-error-depth';
 238                                 break;
 239                         case JSON_ERROR_STATE_MISMATCH:
 240                                 $msg = 'json-error-state-mismatch';
 241                                 break;
 242                         case JSON_ERROR_CTRL_CHAR:
 243                                 $msg = 'json-error-ctrl-char';
 244                                 break;
 245                         case JSON_ERROR_SYNTAX:
 246                                 $msg = 'json-error-syntax';
 247                                 break;
 248                         case JSON_ERROR_UTF8:
 249                                 $msg = 'json-error-utf8';
 250                                 break;
 251                         case JSON_ERROR_RECURSION:
 252                                 $msg = 'json-error-recursion';
 253                                 break;
 254                         case JSON_ERROR_INF_OR_NAN:
 255                                 $msg = 'json-error-inf-or-nan';
 256                                 break;
 257                         case JSON_ERROR_UNSUPPORTED_TYPE:
 258                                 $msg = 'json-error-unsupported-type';
 259                                 break;
 260                 }
 261                 return Status::newFatal( $msg );
 262         }
 263
 264         /**
 265          * Remove multiline and single line comments from an otherwise valid JSON
 266          * input string. This can be used as a preprocessor for to allow JSON
 267          * formatted configuration files to contain comments.
 268          *
 269          * @param string $json
 270          * @return string JSON with comments removed
 271          */
 272         public static function stripComments( $json ) {
 273                 // Ensure we have a string
 274                 $str = (string)$json;
 275                 $buffer = '';
 276                 $maxLen = strlen( $str );
 277                 $mark = 0;
 278
 279                 $inString = false;
 280                 $inComment = false;
 281                 $multiline = false;
 282
 283                 for ( $idx = 0; $idx < $maxLen; $idx++ ) {
 284                         switch ( $str[$idx] ) {
 285                                 case '"':
 286                                         $lookBehind = ( $idx - 1 >= 0 ) ? $str[$idx - 1] : '';
 287                                         if ( !$inComment && $lookBehind !== '\\' ) {
 288                                                 // Either started or ended a string
 289                                                 $inString = !$inString;
 290                                         }
 291                                         break;
 292
 293                                 case '/':
 294                                         $lookAhead = ( $idx + 1 < $maxLen ) ? $str[$idx + 1] : '';
 295                                         $lookBehind = ( $idx - 1 >= 0 ) ? $str[$idx - 1] : '';
 296                                         if ( $inString ) {
 297                                                 continue;
 298
 299                                         } elseif ( !$inComment &&
 300                                                 ( $lookAhead === '/' || $lookAhead === '*' )
 301                                         ) {
 302                                                 // Transition into a comment
 303                                                 // Add characters seen to buffer
 304                                                 $buffer .= substr( $str, $mark, $idx - $mark );
 305                                                 // Consume the look ahead character
 306                                                 $idx++;
 307                                                 // Track state
 308                                                 $inComment = true;
 309                                                 $multiline = $lookAhead === '*';
 310
 311                                         } elseif ( $multiline && $lookBehind === '*' ) {
 312                                                 // Found the end of the current comment
 313                                                 $mark = $idx + 1;
 314                                                 $inComment = false;
 315                                                 $multiline = false;
 316                                         }
 317                                         break;
 318
 319                                 case "\n":
 320                                         if ( $inComment && !$multiline ) {
 321                                                 // Found the end of the current comment
 322                                                 $mark = $idx + 1;
 323                                                 $inComment = false;
 324                                         }
 325                                         break;
 326                         }
 327                 }
 328                 if ( $inComment ) {
 329                         // Comment ends with input
 330                         // Technically we should check to ensure that we aren't in
 331                         // a multiline comment that hasn't been properly ended, but this
 332                         // is a strip filter, not a validating parser.
 333                         $mark = $maxLen;
 334                 }
 335                 // Add final chunk to buffer before returning
 336                 return $buffer . substr( $str, $mark, $maxLen - $mark );
 337         }
 338 }