includes/Sanitizer.php

   1 <?php
   2 /**
   3  * HTML sanitizer for %MediaWiki.
   4  *
   5  * Copyright © 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Parser
  25  */
  26
  27 /**
  28  * HTML sanitizer for MediaWiki
  29  * @ingroup Parser
  30  */
  31 class Sanitizer {
  32         /**
  33          * Regular expression to match various types of character references in
  34          * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  35          */
  36         const CHAR_REFS_REGEX =
  37                 '/&([A-Za-z0-9\x80-\xff]+);
  38                  |&\#([0-9]+);
  39                  |&\#[xX]([0-9A-Fa-f]+);
  40                  |(&)/x';
  41
  42         /**
  43          * Acceptable tag name charset from HTML5 parsing spec
  44          * http://www.w3.org/TR/html5/syntax.html#tag-open-state
  45          */
  46         const ELEMENT_BITS_REGEX = '!^(/?)([A-Za-z][^\t\n\v />\0]*+)([^>]*?)(/?>)([^<]*)$!';
  47
  48         /**
  49          * Blacklist for evil uris like javascript:
  50          * WARNING: DO NOT use this in any place that actually requires blacklisting
  51          * for security reasons. There are NUMEROUS[1] ways to bypass blacklisting, the
  52          * only way to be secure from javascript: uri based xss vectors is to whitelist
  53          * things that you know are safe and deny everything else.
  54          * [1]: http://ha.ckers.org/xss.html
  55          */
  56         const EVIL_URI_PATTERN = '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i';
  57         const XMLNS_ATTRIBUTE_PATTERN = "/^xmlns:[:A-Z_a-z-.0-9]+$/";
  58
  59         /**
  60          * List of all named character entities defined in HTML 4.01
  61          * http://www.w3.org/TR/html4/sgml/entities.html
  62          * As well as &apos; which is only defined starting in XHTML1.
  63          */
  64         private static $htmlEntities = array(
  65                 'Aacute'   => 193,
  66                 'aacute'   => 225,
  67                 'Acirc'    => 194,
  68                 'acirc'    => 226,
  69                 'acute'    => 180,
  70                 'AElig'    => 198,
  71                 'aelig'    => 230,
  72                 'Agrave'   => 192,
  73                 'agrave'   => 224,
  74                 'alefsym'  => 8501,
  75                 'Alpha'    => 913,
  76                 'alpha'    => 945,
  77                 'amp'      => 38,
  78                 'and'      => 8743,
  79                 'ang'      => 8736,
  80                 'apos'     => 39, // New in XHTML & HTML 5; avoid in output for compatibility with IE.
  81                 'Aring'    => 197,
  82                 'aring'    => 229,
  83                 'asymp'    => 8776,
  84                 'Atilde'   => 195,
  85                 'atilde'   => 227,
  86                 'Auml'     => 196,
  87                 'auml'     => 228,
  88                 'bdquo'    => 8222,
  89                 'Beta'     => 914,
  90                 'beta'     => 946,
  91                 'brvbar'   => 166,
  92                 'bull'     => 8226,
  93                 'cap'      => 8745,
  94                 'Ccedil'   => 199,
  95                 'ccedil'   => 231,
  96                 'cedil'    => 184,
  97                 'cent'     => 162,
  98                 'Chi'      => 935,
  99                 'chi'      => 967,
 100                 'circ'     => 710,
 101                 'clubs'    => 9827,
 102                 'cong'     => 8773,
 103                 'copy'     => 169,
 104                 'crarr'    => 8629,
 105                 'cup'      => 8746,
 106                 'curren'   => 164,
 107                 'dagger'   => 8224,
 108                 'Dagger'   => 8225,
 109                 'darr'     => 8595,
 110                 'dArr'     => 8659,
 111                 'deg'      => 176,
 112                 'Delta'    => 916,
 113                 'delta'    => 948,
 114                 'diams'    => 9830,
 115                 'divide'   => 247,
 116                 'Eacute'   => 201,
 117                 'eacute'   => 233,
 118                 'Ecirc'    => 202,
 119                 'ecirc'    => 234,
 120                 'Egrave'   => 200,
 121                 'egrave'   => 232,
 122                 'empty'    => 8709,
 123                 'emsp'     => 8195,
 124                 'ensp'     => 8194,
 125                 'Epsilon'  => 917,
 126                 'epsilon'  => 949,
 127                 'equiv'    => 8801,
 128                 'Eta'      => 919,
 129                 'eta'      => 951,
 130                 'ETH'      => 208,
 131                 'eth'      => 240,
 132                 'Euml'     => 203,
 133                 'euml'     => 235,
 134                 'euro'     => 8364,
 135                 'exist'    => 8707,
 136                 'fnof'     => 402,
 137                 'forall'   => 8704,
 138                 'frac12'   => 189,
 139                 'frac14'   => 188,
 140                 'frac34'   => 190,
 141                 'frasl'    => 8260,
 142                 'Gamma'    => 915,
 143                 'gamma'    => 947,
 144                 'ge'       => 8805,
 145                 'gt'       => 62,
 146                 'harr'     => 8596,
 147                 'hArr'     => 8660,
 148                 'hearts'   => 9829,
 149                 'hellip'   => 8230,
 150                 'Iacute'   => 205,
 151                 'iacute'   => 237,
 152                 'Icirc'    => 206,
 153                 'icirc'    => 238,
 154                 'iexcl'    => 161,
 155                 'Igrave'   => 204,
 156                 'igrave'   => 236,
 157                 'image'    => 8465,
 158                 'infin'    => 8734,
 159                 'int'      => 8747,
 160                 'Iota'     => 921,
 161                 'iota'     => 953,
 162                 'iquest'   => 191,
 163                 'isin'     => 8712,
 164                 'Iuml'     => 207,
 165                 'iuml'     => 239,
 166                 'Kappa'    => 922,
 167                 'kappa'    => 954,
 168                 'Lambda'   => 923,
 169                 'lambda'   => 955,
 170                 'lang'     => 9001,
 171                 'laquo'    => 171,
 172                 'larr'     => 8592,
 173                 'lArr'     => 8656,
 174                 'lceil'    => 8968,
 175                 'ldquo'    => 8220,
 176                 'le'       => 8804,
 177                 'lfloor'   => 8970,
 178                 'lowast'   => 8727,
 179                 'loz'      => 9674,
 180                 'lrm'      => 8206,
 181                 'lsaquo'   => 8249,
 182                 'lsquo'    => 8216,
 183                 'lt'       => 60,
 184                 'macr'     => 175,
 185                 'mdash'    => 8212,
 186                 'micro'    => 181,
 187                 'middot'   => 183,
 188                 'minus'    => 8722,
 189                 'Mu'       => 924,
 190                 'mu'       => 956,
 191                 'nabla'    => 8711,
 192                 'nbsp'     => 160,
 193                 'ndash'    => 8211,
 194                 'ne'       => 8800,
 195                 'ni'       => 8715,
 196                 'not'      => 172,
 197                 'notin'    => 8713,
 198                 'nsub'     => 8836,
 199                 'Ntilde'   => 209,
 200                 'ntilde'   => 241,
 201                 'Nu'       => 925,
 202                 'nu'       => 957,
 203                 'Oacute'   => 211,
 204                 'oacute'   => 243,
 205                 'Ocirc'    => 212,
 206                 'ocirc'    => 244,
 207                 'OElig'    => 338,
 208                 'oelig'    => 339,
 209                 'Ograve'   => 210,
 210                 'ograve'   => 242,
 211                 'oline'    => 8254,
 212                 'Omega'    => 937,
 213                 'omega'    => 969,
 214                 'Omicron'  => 927,
 215                 'omicron'  => 959,
 216                 'oplus'    => 8853,
 217                 'or'       => 8744,
 218                 'ordf'     => 170,
 219                 'ordm'     => 186,
 220                 'Oslash'   => 216,
 221                 'oslash'   => 248,
 222                 'Otilde'   => 213,
 223                 'otilde'   => 245,
 224                 'otimes'   => 8855,
 225                 'Ouml'     => 214,
 226                 'ouml'     => 246,
 227                 'para'     => 182,
 228                 'part'     => 8706,
 229                 'permil'   => 8240,
 230                 'perp'     => 8869,
 231                 'Phi'      => 934,
 232                 'phi'      => 966,
 233                 'Pi'       => 928,
 234                 'pi'       => 960,
 235                 'piv'      => 982,
 236                 'plusmn'   => 177,
 237                 'pound'    => 163,
 238                 'prime'    => 8242,
 239                 'Prime'    => 8243,
 240                 'prod'     => 8719,
 241                 'prop'     => 8733,
 242                 'Psi'      => 936,
 243                 'psi'      => 968,
 244                 'quot'     => 34,
 245                 'radic'    => 8730,
 246                 'rang'     => 9002,
 247                 'raquo'    => 187,
 248                 'rarr'     => 8594,
 249                 'rArr'     => 8658,
 250                 'rceil'    => 8969,
 251                 'rdquo'    => 8221,
 252                 'real'     => 8476,
 253                 'reg'      => 174,
 254                 'rfloor'   => 8971,
 255                 'Rho'      => 929,
 256                 'rho'      => 961,
 257                 'rlm'      => 8207,
 258                 'rsaquo'   => 8250,
 259                 'rsquo'    => 8217,
 260                 'sbquo'    => 8218,
 261                 'Scaron'   => 352,
 262                 'scaron'   => 353,
 263                 'sdot'     => 8901,
 264                 'sect'     => 167,
 265                 'shy'      => 173,
 266                 'Sigma'    => 931,
 267                 'sigma'    => 963,
 268                 'sigmaf'   => 962,
 269                 'sim'      => 8764,
 270                 'spades'   => 9824,
 271                 'sub'      => 8834,
 272                 'sube'     => 8838,
 273                 'sum'      => 8721,
 274                 'sup'      => 8835,
 275                 'sup1'     => 185,
 276                 'sup2'     => 178,
 277                 'sup3'     => 179,
 278                 'supe'     => 8839,
 279                 'szlig'    => 223,
 280                 'Tau'      => 932,
 281                 'tau'      => 964,
 282                 'there4'   => 8756,
 283                 'Theta'    => 920,
 284                 'theta'    => 952,
 285                 'thetasym' => 977,
 286                 'thinsp'   => 8201,
 287                 'THORN'    => 222,
 288                 'thorn'    => 254,
 289                 'tilde'    => 732,
 290                 'times'    => 215,
 291                 'trade'    => 8482,
 292                 'Uacute'   => 218,
 293                 'uacute'   => 250,
 294                 'uarr'     => 8593,
 295                 'uArr'     => 8657,
 296                 'Ucirc'    => 219,
 297                 'ucirc'    => 251,
 298                 'Ugrave'   => 217,
 299                 'ugrave'   => 249,
 300                 'uml'      => 168,
 301                 'upsih'    => 978,
 302                 'Upsilon'  => 933,
 303                 'upsilon'  => 965,
 304                 'Uuml'     => 220,
 305                 'uuml'     => 252,
 306                 'weierp'   => 8472,
 307                 'Xi'       => 926,
 308                 'xi'       => 958,
 309                 'Yacute'   => 221,
 310                 'yacute'   => 253,
 311                 'yen'      => 165,
 312                 'Yuml'     => 376,
 313                 'yuml'     => 255,
 314                 'Zeta'     => 918,
 315                 'zeta'     => 950,
 316                 'zwj'      => 8205,
 317                 'zwnj'     => 8204
 318         );
 319
 320         /**
 321          * Character entity aliases accepted by MediaWiki
 322          */
 323         private static $htmlEntityAliases = array(
 324                 'רלמ' => 'rlm',
 325                 'رلم' => 'rlm',
 326         );
 327
 328         /**
 329          * Lazy-initialised attributes regex, see getAttribsRegex()
 330          */
 331         private static $attribsRegex;
 332
 333         /**
 334          * Regular expression to match HTML/XML attribute pairs within a tag.
 335          * Allows some... latitude.
 336          * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
 337          * @return string
 338          */
 339         static function getAttribsRegex() {
 340                 if ( self::$attribsRegex === null ) {
 341                         $attribFirst = '[:A-Z_a-z0-9]';
 342                         $attrib = '[:A-Z_a-z-.0-9]';
 343                         $space = '[\x09\x0a\x0d\x20]';
 344                         self::$attribsRegex =
 345                                 "/(?:^|$space)({$attribFirst}{$attrib}*)
 346                                   ($space*=$space*
 347                                         (?:
 348                                          # The attribute value: quoted or alone
 349                                           \"([^<\"]*)\"
 350                                          | '([^<']*)'
 351                                          |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
 352                                          |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
 353                                                                                  # colors are specified like this.
 354                                                                                  # We'll be normalizing it.
 355                                         )
 356                                 )?(?=$space|\$)/sx";
 357                 }
 358                 return self::$attribsRegex;
 359         }
 360
 361         /**
 362          * Return the various lists of recognized tags
 363          * @param array $extratags For any extra tags to include
 364          * @param array $removetags For any tags (default or extra) to exclude
 365          * @return array
 366          */
 367         public static function getRecognizedTagData( $extratags = array(), $removetags = array() ) {
 368                 global $wgAllowMicrodataAttributes, $wgAllowImageTag;
 369
 370                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 371                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 372
 373                 // Base our staticInitialised variable off of the global config state so that if the globals
 374                 // are changed (like in the screwed up test system) we will re-initialise the settings.
 375                 $globalContext = implode( '-', compact( 'wgAllowMicrodataAttributes', 'wgAllowImageTag' ) );
 376                 if ( !$staticInitialised || $staticInitialised != $globalContext ) {
 377
 378                         $htmlpairsStatic = array( # Tags that must be closed
 379                                 'b', 'bdi', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 380                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 381                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 382                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 383                                 'ruby', 'rb', 'rp', 'rt', 'rtc', 'p', 'span', 'abbr', 'dfn',
 384                                 'kbd', 'samp', 'data', 'time', 'mark'
 385                         );
 386                         $htmlsingle = array(
 387                                 'br', 'wbr', 'hr', 'li', 'dt', 'dd'
 388                         );
 389                         $htmlsingleonly = array( # Elements that cannot have close tags
 390                                 'br', 'wbr', 'hr'
 391                         );
 392                         if ( $wgAllowMicrodataAttributes ) {
 393                                 $htmlsingle[] = $htmlsingleonly[] = 'meta';
 394                                 $htmlsingle[] = $htmlsingleonly[] = 'link';
 395                         }
 396                         $htmlnest = array( # Tags that can be nested--??
 397                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 398                                 'li', 'dl', 'dt', 'dd', 'font', 'big', 'small', 'sub', 'sup', 'span',
 399                                 'var', 'kbd', 'samp', 'em', 'strong', 'q', 'ruby', 'bdo'
 400                         );
 401                         $tabletags = array( # Can only appear inside table, we will close them
 402                                 'td', 'th', 'tr',
 403                         );
 404                         $htmllist = array( # Tags used by list
 405                                 'ul', 'ol',
 406                         );
 407                         $listtags = array( # Tags that can appear in a list
 408                                 'li',
 409                         );
 410
 411                         if ( $wgAllowImageTag ) {
 412                                 $htmlsingle[] = 'img';
 413                                 $htmlsingleonly[] = 'img';
 414                         }
 415
 416                         $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
 417                         $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
 418
 419                         # Convert them all to hashtables for faster lookup
 420                         $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 421                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
 422                         foreach ( $vars as $var ) {
 423                                 $$var = array_flip( $$var );
 424                         }
 425                         $staticInitialised = $globalContext;
 426                 }
 427
 428                 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
 429                 $extratags = array_flip( $extratags );
 430                 $removetags = array_flip( $removetags );
 431                 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
 432                 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ), $removetags );
 433
 434                 return array(
 435                         'htmlpairs' => $htmlpairs,
 436                         'htmlsingle' => $htmlsingle,
 437                         'htmlsingleonly' => $htmlsingleonly,
 438                         'htmlnest' => $htmlnest,
 439                         'tabletags' => $tabletags,
 440                         'htmllist' => $htmllist,
 441                         'listtags' => $listtags,
 442                         'htmlsingleallowed' => $htmlsingleallowed,
 443                         'htmlelements' => $htmlelements,
 444                 );
 445         }
 446
 447         /**
 448          * Cleans up HTML, removes dangerous tags and attributes, and
 449          * removes HTML comments
 450          * @param string $text
 451          * @param callable $processCallback Callback to do any variable or parameter
 452          *   replacements in HTML attribute values
 453          * @param array|bool $args Arguments for the processing callback
 454          * @param array $extratags For any extra tags to include
 455          * @param array $removetags For any tags (default or extra) to exclude
 456          * @return string
 457          */
 458         public static function removeHTMLtags( $text, $processCallback = null,
 459                 $args = array(), $extratags = array(), $removetags = array()
 460         ) {
 461                 global $wgUseTidy;
 462
 463                 extract( self::getRecognizedTagData( $extratags, $removetags ) );
 464
 465                 # Remove HTML comments
 466                 $text = Sanitizer::removeHTMLcomments( $text );
 467                 $bits = explode( '<', $text );
 468                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 469                 if ( !$wgUseTidy ) {
 470                         $tagstack = $tablestack = array();
 471                         foreach ( $bits as $x ) {
 472                                 $regs = array();
 473                                 # $slash: Does the current element start with a '/'?
 474                                 # $t: Current element name
 475                                 # $params: String between element name and >
 476                                 # $brace: Ending '>' or '/>'
 477                                 # $rest: Everything until the next element of $bits
 478                                 if ( preg_match( self::ELEMENT_BITS_REGEX, $x, $regs ) ) {
 479                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 480                                 } else {
 481                                         $slash = $t = $params = $brace = $rest = null;
 482                                 }
 483
 484                                 $badtag = false;
 485                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 486                                         # Check our stack
 487                                         if ( $slash && isset( $htmlsingleonly[$t] ) ) {
 488                                                 $badtag = true;
 489                                         } elseif ( $slash ) {
 490                                                 # Closing a tag... is it the one we just opened?
 491                                                 wfSuppressWarnings();
 492                                                 $ot = array_pop( $tagstack );
 493                                                 wfRestoreWarnings();
 494
 495                                                 if ( $ot != $t ) {
 496                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 497                                                                 # Pop all elements with an optional close tag
 498                                                                 # and see if we find a match below them
 499                                                                 $optstack = array();
 500                                                                 array_push( $optstack, $ot );
 501                                                                 wfSuppressWarnings();
 502                                                                 $ot = array_pop( $tagstack );
 503                                                                 wfRestoreWarnings();
 504                                                                 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
 505                                                                         array_push( $optstack, $ot );
 506                                                                         wfSuppressWarnings();
 507                                                                         $ot = array_pop( $tagstack );
 508                                                                         wfRestoreWarnings();
 509                                                                 }
 510                                                                 if ( $t != $ot ) {
 511                                                                         # No match. Push the optional elements back again
 512                                                                         $badtag = true;
 513                                                                         wfSuppressWarnings();
 514                                                                         $ot = array_pop( $optstack );
 515                                                                         wfRestoreWarnings();
 516                                                                         while ( $ot ) {
 517                                                                                 array_push( $tagstack, $ot );
 518                                                                                 wfSuppressWarnings();
 519                                                                                 $ot = array_pop( $optstack );
 520                                                                                 wfRestoreWarnings();
 521                                                                         }
 522                                                                 }
 523                                                         } else {
 524                                                                 wfSuppressWarnings();
 525                                                                 array_push( $tagstack, $ot );
 526                                                                 wfRestoreWarnings();
 527
 528                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 529                                                                 if ( !isset( $htmllist[$ot] ) || !isset( $listtags[$t] ) ) {
 530                                                                         $badtag = true;
 531                                                                 }
 532                                                         }
 533                                                 } else {
 534                                                         if ( $t == 'table' ) {
 535                                                                 $tagstack = array_pop( $tablestack );
 536                                                         }
 537                                                 }
 538                                                 $newparams = '';
 539                                         } else {
 540                                                 # Keep track for later
 541                                                 if ( isset( $tabletags[$t] ) && !in_array( 'table', $tagstack ) ) {
 542                                                         $badtag = true;
 543                                                 } elseif ( in_array( $t, $tagstack ) && !isset( $htmlnest[$t] ) ) {
 544                                                         $badtag = true;
 545                                                 # Is it a self closed htmlpair ? (bug 5487)
 546                                                 } elseif ( $brace == '/>' && isset( $htmlpairs[$t] ) ) {
 547                                                         $badtag = true;
 548                                                 } elseif ( isset( $htmlsingleonly[$t] ) ) {
 549                                                         # Hack to force empty tag for unclosable elements
 550                                                         $brace = '/>';
 551                                                 } elseif ( isset( $htmlsingle[$t] ) ) {
 552                                                         # Hack to not close $htmlsingle tags
 553                                                         $brace = null;
 554                                                         # Still need to push this optionally-closed tag to
 555                                                         # the tag stack so that we can match end tags
 556                                                         # instead of marking them as bad.
 557                                                         array_push( $tagstack, $t );
 558                                                 } elseif ( isset( $tabletags[$t] ) && in_array( $t, $tagstack ) ) {
 559                                                         // New table tag but forgot to close the previous one
 560                                                         $text .= "</$t>";
 561                                                 } else {
 562                                                         if ( $t == 'table' ) {
 563                                                                 array_push( $tablestack, $tagstack );
 564                                                                 $tagstack = array();
 565                                                         }
 566                                                         array_push( $tagstack, $t );
 567                                                 }
 568
 569                                                 # Replace any variables or template parameters with
 570                                                 # plaintext results.
 571                                                 if ( is_callable( $processCallback ) ) {
 572                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 573                                                 }
 574
 575                                                 if ( !Sanitizer::validateTag( $params, $t ) ) {
 576                                                         $badtag = true;
 577                                                 }
 578
 579                                                 # Strip non-approved attributes from the tag
 580                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 581                                         }
 582                                         if ( !$badtag ) {
 583                                                 $rest = str_replace( '>', '&gt;', $rest );
 584                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 585                                                 $text .= "<$slash$t$newparams$close>$rest";
 586                                                 continue;
 587                                         }
 588                                 }
 589                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 590                         }
 591                         # Close off any remaining tags
 592                         while ( is_array( $tagstack ) && ( $t = array_pop( $tagstack ) ) ) {
 593                                 $text .= "</$t>\n";
 594                                 if ( $t == 'table' ) {
 595                                         $tagstack = array_pop( $tablestack );
 596                                 }
 597                         }
 598                 } else {
 599                         # this might be possible using tidy itself
 600                         foreach ( $bits as $x ) {
 601                                 if ( preg_match( self::ELEMENT_BITS_REGEX, $x, $regs ) ) {
 602                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 603
 604                                         $badtag = false;
 605                                         if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 606                                                 if ( is_callable( $processCallback ) ) {
 607                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 608                                                 }
 609
 610                                                 if ( !Sanitizer::validateTag( $params, $t ) ) {
 611                                                         $badtag = true;
 612                                                 }
 613
 614                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 615                                                 if ( !$badtag ) {
 616                                                         $rest = str_replace( '>', '&gt;', $rest );
 617                                                         $text .= "<$slash$t$newparams$brace$rest";
 618                                                         continue;
 619                                                 }
 620                                         }
 621                                 }
 622                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 623                         }
 624                 }
 625                 return $text;
 626         }
 627
 628         /**
 629          * Remove '<!--', '-->', and everything between.
 630          * To avoid leaving blank lines, when a comment is both preceded
 631          * and followed by a newline (ignoring spaces), trim leading and
 632          * trailing spaces and one of the newlines.
 633          *
 634          * @param string $text
 635          * @return string
 636          */
 637         public static function removeHTMLcomments( $text ) {
 638                 while ( ( $start = strpos( $text, '<!--' ) ) !== false ) {
 639                         $end = strpos( $text, '-->', $start + 4 );
 640                         if ( $end === false ) {
 641                                 # Unterminated comment; bail out
 642                                 break;
 643                         }
 644
 645                         $end += 3;
 646
 647                         # Trim space and newline if the comment is both
 648                         # preceded and followed by a newline
 649                         $spaceStart = max( $start - 1, 0 );
 650                         $spaceLen = $end - $spaceStart;
 651                         while ( substr( $text, $spaceStart, 1 ) === ' ' && $spaceStart > 0 ) {
 652                                 $spaceStart--;
 653                                 $spaceLen++;
 654                         }
 655                         while ( substr( $text, $spaceStart + $spaceLen, 1 ) === ' ' ) {
 656                                 $spaceLen++;
 657                         }
 658                         if ( substr( $text, $spaceStart, 1 ) === "\n"
 659                                 && substr( $text, $spaceStart + $spaceLen, 1 ) === "\n" ) {
 660                                 # Remove the comment, leading and trailing
 661                                 # spaces, and leave only one newline.
 662                                 $text = substr_replace( $text, "\n", $spaceStart, $spaceLen + 1 );
 663                         } else {
 664                                 # Remove just the comment.
 665                                 $text = substr_replace( $text, '', $start, $end - $start );
 666                         }
 667                 }
 668                 return $text;
 669         }
 670
 671         /**
 672          * Takes attribute names and values for a tag and the tag name and
 673          * validates that the tag is allowed to be present.
 674          * This DOES NOT validate the attributes, nor does it validate the
 675          * tags themselves. This method only handles the special circumstances
 676          * where we may want to allow a tag within content but ONLY when it has
 677          * specific attributes set.
 678          *
 679          * @param string $params
 680          * @param string $element
 681          * @return bool
 682          */
 683         static function validateTag( $params, $element ) {
 684                 $params = Sanitizer::decodeTagAttributes( $params );
 685
 686                 if ( $element == 'meta' || $element == 'link' ) {
 687                         if ( !isset( $params['itemprop'] ) ) {
 688                                 // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content
 689                                 return false;
 690                         }
 691                         if ( $element == 'meta' && !isset( $params['content'] ) ) {
 692                                 // <meta> must have a content="" for the itemprop
 693                                 return false;
 694                         }
 695                         if ( $element == 'link' && !isset( $params['href'] ) ) {
 696                                 // <link> must have an associated href=""
 697                                 return false;
 698                         }
 699                 }
 700
 701                 return true;
 702         }
 703
 704         /**
 705          * Take an array of attribute names and values and normalize or discard
 706          * illegal values for the given element type.
 707          *
 708          * - Discards attributes not on a whitelist for the given element
 709          * - Unsafe style attributes are discarded
 710          * - Invalid id attributes are re-encoded
 711          *
 712          * @param array $attribs
 713          * @param string $element
 714          * @return array
 715          *
 716          * @todo Check for legal values where the DTD limits things.
 717          * @todo Check for unique id attribute :P
 718          */
 719         static function validateTagAttributes( $attribs, $element ) {
 720                 return Sanitizer::validateAttributes( $attribs,
 721                         Sanitizer::attributeWhitelist( $element ) );
 722         }
 723
 724         /**
 725          * Take an array of attribute names and values and normalize or discard
 726          * illegal values for the given whitelist.
 727          *
 728          * - Discards attributes not the given whitelist
 729          * - Unsafe style attributes are discarded
 730          * - Invalid id attributes are re-encoded
 731          *
 732          * @param array $attribs
 733          * @param array $whitelist List of allowed attribute names
 734          * @return array
 735          *
 736          * @todo Check for legal values where the DTD limits things.
 737          * @todo Check for unique id attribute :P
 738          */
 739         static function validateAttributes( $attribs, $whitelist ) {
 740                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
 741
 742                 $whitelist = array_flip( $whitelist );
 743                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 744
 745                 $out = array();
 746                 foreach ( $attribs as $attribute => $value ) {
 747                         #allow XML namespace declaration if RDFa is enabled
 748                         if ( $wgAllowRdfaAttributes && preg_match( self::XMLNS_ATTRIBUTE_PATTERN, $attribute ) ) {
 749                                 if ( !preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 750                                         $out[$attribute] = $value;
 751                                 }
 752
 753                                 continue;
 754                         }
 755
 756                         # Allow any attribute beginning with "data-"
 757                         if ( !preg_match( '/^data-/i', $attribute ) && !isset( $whitelist[$attribute] ) ) {
 758                                 continue;
 759                         }
 760
 761                         # Strip javascript "expression" from stylesheets.
 762                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 763                         if ( $attribute == 'style' ) {
 764                                 $value = Sanitizer::checkCss( $value );
 765                         }
 766
 767                         if ( $attribute === 'id' ) {
 768                                 $value = Sanitizer::escapeId( $value, 'noninitial' );
 769                         }
 770
 771                         # WAI-ARIA
 772                         # http://www.w3.org/TR/wai-aria/
 773                         # http://www.whatwg.org/html/elements.html#wai-aria
 774                         # For now we only support role="presentation" until we work out what roles should be
 775                         # usable by content and we ensure that our code explicitly rejects patterns that
 776                         # violate HTML5's ARIA restrictions.
 777                         if ( $attribute === 'role' && $value !== 'presentation' ) {
 778                                 continue;
 779                         }
 780
 781                         // RDFa and microdata properties allow URLs, URIs and/or CURIs.
 782                         // Check them for sanity.
 783                         if ( $attribute === 'rel' || $attribute === 'rev'
 784                                 # RDFa
 785                                 || $attribute === 'about' || $attribute === 'property'
 786                                 || $attribute === 'resource' || $attribute === 'datatype'
 787                                 || $attribute === 'typeof'
 788                                 # HTML5 microdata
 789                                 || $attribute === 'itemid' || $attribute === 'itemprop'
 790                                 || $attribute === 'itemref' || $attribute === 'itemscope'
 791                                 || $attribute === 'itemtype'
 792                         ) {
 793                                 //Paranoia. Allow "simple" values but suppress javascript
 794                                 if ( preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 795                                         continue;
 796                                 }
 797                         }
 798
 799                         # NOTE: even though elements using href/src are not allowed directly, supply
 800                         #       validation code that can be used by tag hook handlers, etc
 801                         if ( $attribute === 'href' || $attribute === 'src' ) {
 802                                 if ( !preg_match( $hrefExp, $value ) ) {
 803                                         continue; //drop any href or src attributes not using an allowed protocol.
 804                                         // NOTE: this also drops all relative URLs
 805                                 }
 806                         }
 807
 808                         // If this attribute was previously set, override it.
 809                         // Output should only have one attribute of each name.
 810                         $out[$attribute] = $value;
 811                 }
 812
 813                 if ( $wgAllowMicrodataAttributes ) {
 814                         # itemtype, itemid, itemref don't make sense without itemscope
 815                         if ( !array_key_exists( 'itemscope', $out ) ) {
 816                                 unset( $out['itemtype'] );
 817                                 unset( $out['itemid'] );
 818                                 unset( $out['itemref'] );
 819                         }
 820                         # TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
 821                 }
 822                 return $out;
 823         }
 824
 825         /**
 826          * Merge two sets of HTML attributes.  Conflicting items in the second set
 827          * will override those in the first, except for 'class' attributes which
 828          * will be combined (if they're both strings).
 829          *
 830          * @todo implement merging for other attributes such as style
 831          * @param array $a
 832          * @param array $b
 833          * @return array
 834          */
 835         static function mergeAttributes( $a, $b ) {
 836                 $out = array_merge( $a, $b );
 837                 if ( isset( $a['class'] ) && isset( $b['class'] )
 838                         && is_string( $a['class'] ) && is_string( $b['class'] )
 839                         && $a['class'] !== $b['class']
 840                 ) {
 841                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 842                                 -1, PREG_SPLIT_NO_EMPTY );
 843                         $out['class'] = implode( ' ', array_unique( $classes ) );
 844                 }
 845                 return $out;
 846         }
 847
 848         /**
 849          * Normalize CSS into a format we can easily search for hostile input
 850          *  - decode character references
 851          *  - decode escape sequences
 852          *  - convert characters that IE6 interprets into ascii
 853          *  - remove comments, unless the entire value is one single comment
 854          * @param string $value the css string
 855          * @return string normalized css
 856          */
 857         public static function normalizeCss( $value ) {
 858
 859                 // Decode character references like &#123;
 860                 $value = Sanitizer::decodeCharReferences( $value );
 861
 862                 // Decode escape sequences and line continuation
 863                 // See the grammar in the CSS 2 spec, appendix D.
 864                 // This has to be done AFTER decoding character references.
 865                 // This means it isn't possible for this function to return
 866                 // unsanitized escape sequences. It is possible to manufacture
 867                 // input that contains character references that decode to
 868                 // escape sequences that decode to character references, but
 869                 // it's OK for the return value to contain character references
 870                 // because the caller is supposed to escape those anyway.
 871                 static $decodeRegex;
 872                 if ( !$decodeRegex ) {
 873                         $space = '[\\x20\\t\\r\\n\\f]';
 874                         $nl = '(?:\\n|\\r\\n|\\r|\\f)';
 875                         $backslash = '\\\\';
 876                         $decodeRegex = "/ $backslash
 877                                 (?:
 878                                         ($nl) |  # 1. Line continuation
 879                                         ([0-9A-Fa-f]{1,6})$space? |  # 2. character number
 880                                         (.) | # 3. backslash cancelling special meaning
 881                                         () | # 4. backslash at end of string
 882                                 )/xu";
 883                 }
 884                 $value = preg_replace_callback( $decodeRegex,
 885                         array( __CLASS__, 'cssDecodeCallback' ), $value );
 886
 887                 // Normalize Halfwidth and Fullwidth Unicode block that IE6 might treat as ascii
 888                 $value = preg_replace_callback(
 889                         '/[！-［］-ｚ]/u', // U+FF01 to U+FF5A, excluding U+FF3C (bug 58088)
 890                         function ( $matches ) {
 891                                 $cp = UtfNormal\Utils::utf8ToCodepoint( $matches[0] );
 892                                 if ( $cp === false ) {
 893                                         return '';
 894                                 }
 895                                 return chr( $cp - 65248 ); // ASCII range \x21-\x7A
 896                         },
 897                         $value
 898                 );
 899
 900                 // Convert more characters IE6 might treat as ascii
 901                 // U+0280, U+0274, U+207F, U+029F, U+026A, U+207D, U+208D
 902                 $value = str_replace(
 903                         array( 'ʀ', 'ɴ', 'ⁿ', 'ʟ', 'ɪ', '⁽', '₍' ),
 904                         array( 'r', 'n', 'n', 'l', 'i', '(', '(' ),
 905                         $value
 906                 );
 907
 908                 // Let the value through if it's nothing but a single comment, to
 909                 // allow other functions which may reject it to pass some error
 910                 // message through.
 911                 if ( !preg_match( '! ^ \s* /\* [^*\\/]* \*/ \s* $ !x', $value ) ) {
 912                         // Remove any comments; IE gets token splitting wrong
 913                         // This must be done AFTER decoding character references and
 914                         // escape sequences, because those steps can introduce comments
 915                         // This step cannot introduce character references or escape
 916                         // sequences, because it replaces comments with spaces rather
 917                         // than removing them completely.
 918                         $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
 919
 920                         // Remove anything after a comment-start token, to guard against
 921                         // incorrect client implementations.
 922                         $commentPos = strpos( $value, '/*' );
 923                         if ( $commentPos !== false ) {
 924                                 $value = substr( $value, 0, $commentPos );
 925                         }
 926                 }
 927
 928                 // S followed by repeat, iteration, or prolonged sound marks,
 929                 // which IE will treat as "ss"
 930                 $value = preg_replace(
 931                         '/s(?:
 932                                 \xE3\x80\xB1 | # U+3031
 933                                 \xE3\x82\x9D | # U+309D
 934                                 \xE3\x83\xBC | # U+30FC
 935                                 \xE3\x83\xBD | # U+30FD
 936                                 \xEF\xB9\xBC | # U+FE7C
 937                                 \xEF\xB9\xBD | # U+FE7D
 938                                 \xEF\xBD\xB0   # U+FF70
 939                         )/ix',
 940                         'ss',
 941                         $value
 942                 );
 943
 944                 return $value;
 945         }
 946
 947
 948         /**
 949          * Pick apart some CSS and check it for forbidden or unsafe structures.
 950          * Returns a sanitized string. This sanitized string will have
 951          * character references and escape sequences decoded and comments
 952          * stripped (unless it is itself one valid comment, in which case the value
 953          * will be passed through). If the input is just too evil, only a comment
 954          * complaining about evilness will be returned.
 955          *
 956          * Currently URL references, 'expression', 'tps' are forbidden.
 957          *
 958          * NOTE: Despite the fact that character references are decoded, the
 959          * returned string may contain character references given certain
 960          * clever input strings. These character references must
 961          * be escaped before the return value is embedded in HTML.
 962          *
 963          * @param string $value
 964          * @return string
 965          */
 966         static function checkCss( $value ) {
 967                 $value = self::normalizeCss( $value );
 968
 969                 // Reject problematic keywords and control characters
 970                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
 971                         return '/* invalid control char */';
 972                 } elseif ( preg_match(
 973                         '! expression
 974                                 | filter\s*:
 975                                 | accelerator\s*:
 976                                 | -o-link\s*:
 977                                 | -o-link-source\s*:
 978                                 | -o-replace\s*:
 979                                 | url\s*\(
 980                                 | image\s*\(
 981                                 | image-set\s*\(
 982                         !ix', $value ) ) {
 983                         return '/* insecure input */';
 984                 }
 985                 return $value;
 986         }
 987
 988         /**
 989          * @param array $matches
 990          * @return string
 991          */
 992         static function cssDecodeCallback( $matches ) {
 993                 if ( $matches[1] !== '' ) {
 994                         // Line continuation
 995                         return '';
 996                 } elseif ( $matches[2] !== '' ) {
 997                         $char = UtfNormal\Utils::codepointToUtf8( hexdec( $matches[2] ) );
 998                 } elseif ( $matches[3] !== '' ) {
 999                         $char = $matches[3];
1000                 } else {
1001                         $char = '\\';
1002                 }
1003                 if ( $char == "\n" || $char == '"' || $char == "'" || $char == '\\' ) {
1004                         // These characters need to be escaped in strings
1005                         // Clean up the escape sequence to avoid parsing errors by clients
1006                         return '\\' . dechex( ord( $char ) ) . ' ';
1007                 } else {
1008                         // Decode unnecessary escape
1009                         return $char;
1010                 }
1011         }
1012
1013         /**
1014          * Take a tag soup fragment listing an HTML element's attributes
1015          * and normalize it to well-formed XML, discarding unwanted attributes.
1016          * Output is safe for further wikitext processing, with escaping of
1017          * values that could trigger problems.
1018          *
1019          * - Normalizes attribute names to lowercase
1020          * - Discards attributes not on a whitelist for the given element
1021          * - Turns broken or invalid entities into plaintext
1022          * - Double-quotes all attribute values
1023          * - Attributes without values are given the name as attribute
1024          * - Double attributes are discarded
1025          * - Unsafe style attributes are discarded
1026          * - Prepends space if there are attributes.
1027          *
1028          * @param string $text
1029          * @param string $element
1030          * @return string
1031          */
1032         static function fixTagAttributes( $text, $element ) {
1033                 if ( trim( $text ) == '' ) {
1034                         return '';
1035                 }
1036
1037                 $decoded = Sanitizer::decodeTagAttributes( $text );
1038                 $stripped = Sanitizer::validateTagAttributes( $decoded, $element );
1039
1040                 return Sanitizer::safeEncodeTagAttributes( $stripped );
1041         }
1042
1043         /**
1044          * Encode an attribute value for HTML output.
1045          * @param string $text
1046          * @return string HTML-encoded text fragment
1047          */
1048         static function encodeAttribute( $text ) {
1049                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
1050
1051                 // Whitespace is normalized during attribute decoding,
1052                 // so if we've been passed non-spaces we must encode them
1053                 // ahead of time or they won't be preserved.
1054                 $encValue = strtr( $encValue, array(
1055                         "\n" => '&#10;',
1056                         "\r" => '&#13;',
1057                         "\t" => '&#9;',
1058                 ) );
1059
1060                 return $encValue;
1061         }
1062
1063         /**
1064          * Encode an attribute value for HTML tags, with extra armoring
1065          * against further wiki processing.
1066          * @param string $text
1067          * @return string HTML-encoded text fragment
1068          */
1069         static function safeEncodeAttribute( $text ) {
1070                 $encValue = Sanitizer::encodeAttribute( $text );
1071
1072                 # Templates and links may be expanded in later parsing,
1073                 # creating invalid or dangerous output. Suppress this.
1074                 $encValue = strtr( $encValue, array(
1075                         '<'    => '&lt;',   // This should never happen,
1076                         '>'    => '&gt;',   // we've received invalid input
1077                         '"'    => '&quot;', // which should have been escaped.
1078                         '{'    => '&#123;',
1079                         '['    => '&#91;',
1080                         "''"   => '&#39;&#39;',
1081                         'ISBN' => '&#73;SBN',
1082                         'RFC'  => '&#82;FC',
1083                         'PMID' => '&#80;MID',
1084                         '|'    => '&#124;',
1085                         '__'   => '&#95;_',
1086                 ) );
1087
1088                 # Stupid hack
1089                 $encValue = preg_replace_callback(
1090                         '/((?i)' . wfUrlProtocols() . ')/',
1091                         array( 'Sanitizer', 'armorLinksCallback' ),
1092                         $encValue );
1093                 return $encValue;
1094         }
1095
1096         /**
1097          * Given a value, escape it so that it can be used in an id attribute and
1098          * return it.  This will use HTML5 validation if $wgExperimentalHtmlIds is
1099          * true, allowing anything but ASCII whitespace.  Otherwise it will use
1100          * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
1101          * escaped with lots of dots.
1102          *
1103          * To ensure we don't have to bother escaping anything, we also strip ', ",
1104          * & even if $wgExperimentalIds is true.  TODO: Is this the best tactic?
1105          * We also strip # because it upsets IE, and % because it could be
1106          * ambiguous if it's part of something that looks like a percent escape
1107          * (which don't work reliably in fragments cross-browser).
1108          *
1109          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
1110          *   in the id and name attributes
1111          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with
1112          *   the id attribute
1113          * @see http://www.whatwg.org/html/elements.html#the-id-attribute
1114          *   HTML5 definition of id attribute
1115          *
1116          * @param string $id Id to escape
1117          * @param string|array $options String or array of strings (default is array()):
1118          *   'noninitial': This is a non-initial fragment of an id, not a full id,
1119          *       so don't pay attention if the first character isn't valid at the
1120          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
1121          *       false.
1122          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
1123          *       if $wgExperimentalHtmlIds is used, so we can generate extra
1124          *       anchors and links won't break.
1125          * @return string
1126          */
1127         static function escapeId( $id, $options = array() ) {
1128                 global $wgExperimentalHtmlIds;
1129                 $options = (array)$options;
1130
1131                 $id = Sanitizer::decodeCharReferences( $id );
1132
1133                 if ( $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
1134                         $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
1135                         $id = trim( $id, '_' );
1136                         if ( $id === '' ) {
1137                                 // Must have been all whitespace to start with.
1138                                 return '_';
1139                         } else {
1140                                 return $id;
1141                         }
1142                 }
1143
1144                 // HTML4-style escaping
1145                 static $replace = array(
1146                         '%3A' => ':',
1147                         '%' => '.'
1148                 );
1149
1150                 $id = urlencode( strtr( $id, ' ', '_' ) );
1151                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
1152
1153                 if ( !preg_match( '/^[a-zA-Z]/', $id ) && !in_array( 'noninitial', $options ) ) {
1154                         // Initial character must be a letter!
1155                         $id = "x$id";
1156                 }
1157                 return $id;
1158         }
1159
1160         /**
1161          * Given a value, escape it so that it can be used as a CSS class and
1162          * return it.
1163          *
1164          * @todo For extra validity, input should be validated UTF-8.
1165          *
1166          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
1167          *
1168          * @param string $class
1169          * @return string
1170          */
1171         static function escapeClass( $class ) {
1172                 // Convert ugly stuff to underscores and kill underscores in ugly places
1173                 return rtrim( preg_replace(
1174                         array( '/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/', '/_+/' ),
1175                         '_',
1176                         $class ), '_' );
1177         }
1178
1179         /**
1180          * Given HTML input, escape with htmlspecialchars but un-escape entities.
1181          * This allows (generally harmless) entities like &#160; to survive.
1182          *
1183          * @param string $html HTML to escape
1184          * @return string Escaped input
1185          */
1186         static function escapeHtmlAllowEntities( $html ) {
1187                 $html = Sanitizer::decodeCharReferences( $html );
1188                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
1189                 # hurt.
1190                 $html = htmlspecialchars( $html, ENT_QUOTES );
1191                 return $html;
1192         }
1193
1194         /**
1195          * Regex replace callback for armoring links against further processing.
1196          * @param array $matches
1197          * @return string
1198          */
1199         private static function armorLinksCallback( $matches ) {
1200                 return str_replace( ':', '&#58;', $matches[1] );
1201         }
1202
1203         /**
1204          * Return an associative array of attribute names and values from
1205          * a partial tag string. Attribute names are forces to lowercase,
1206          * character references are decoded to UTF-8 text.
1207          *
1208          * @param string $text
1209          * @return array
1210          */
1211         public static function decodeTagAttributes( $text ) {
1212                 if ( trim( $text ) == '' ) {
1213                         return array();
1214                 }
1215
1216                 $attribs = array();
1217                 $pairs = array();
1218                 if ( !preg_match_all(
1219                         self::getAttribsRegex(),
1220                         $text,
1221                         $pairs,
1222                         PREG_SET_ORDER ) ) {
1223                         return $attribs;
1224                 }
1225
1226                 foreach ( $pairs as $set ) {
1227                         $attribute = strtolower( $set[1] );
1228                         $value = Sanitizer::getTagAttributeCallback( $set );
1229
1230                         // Normalize whitespace
1231                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1232                         $value = trim( $value );
1233
1234                         // Decode character references
1235                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
1236                 }
1237                 return $attribs;
1238         }
1239
1240         /**
1241          * Build a partial tag string from an associative array of attribute
1242          * names and values as returned by decodeTagAttributes.
1243          *
1244          * @param array $assoc_array
1245          * @return string
1246          */
1247         public static function safeEncodeTagAttributes( $assoc_array ) {
1248                 $attribs = array();
1249                 foreach ( $assoc_array as $attribute => $value ) {
1250                         $encAttribute = htmlspecialchars( $attribute );
1251                         $encValue = Sanitizer::safeEncodeAttribute( $value );
1252
1253                         $attribs[] = "$encAttribute=\"$encValue\"";
1254                 }
1255                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
1256         }
1257
1258         /**
1259          * Pick the appropriate attribute value from a match set from the
1260          * attribs regex matches.
1261          *
1262          * @param array $set
1263          * @throws MWException When tag conditions are not met.
1264          * @return string
1265          */
1266         private static function getTagAttributeCallback( $set ) {
1267                 if ( isset( $set[6] ) ) {
1268                         # Illegal #XXXXXX color with no quotes.
1269                         return $set[6];
1270                 } elseif ( isset( $set[5] ) ) {
1271                         # No quotes.
1272                         return $set[5];
1273                 } elseif ( isset( $set[4] ) ) {
1274                         # Single-quoted
1275                         return $set[4];
1276                 } elseif ( isset( $set[3] ) ) {
1277                         # Double-quoted
1278                         return $set[3];
1279                 } elseif ( !isset( $set[2] ) ) {
1280                         # In XHTML, attributes must have a value.
1281                         # For 'reduced' form, return explicitly the attribute name here.
1282                         return $set[1];
1283                 } else {
1284                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1285                 }
1286         }
1287
1288         /**
1289          * @param string $text
1290          * @return string
1291          */
1292         private static function normalizeWhitespace( $text ) {
1293                 return preg_replace(
1294                         '/\r\n|[\x20\x0d\x0a\x09]/',
1295                         ' ',
1296                         $text );
1297         }
1298
1299         /**
1300          * Normalizes whitespace in a section name, such as might be returned
1301          * by Parser::stripSectionName(), for use in the id's that are used for
1302          * section links.
1303          *
1304          * @param string $section
1305          * @return string
1306          */
1307         static function normalizeSectionNameWhitespace( $section ) {
1308                 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1309         }
1310
1311         /**
1312          * Ensure that any entities and character references are legal
1313          * for XML and XHTML specifically. Any stray bits will be
1314          * &amp;-escaped to result in a valid text fragment.
1315          *
1316          * a. named char refs can only be &lt; &gt; &amp; &quot;, others are
1317          *   numericized (this way we're well-formed even without a DTD)
1318          * b. any numeric char refs must be legal chars, not invalid or forbidden
1319          * c. use lower cased "&#x", not "&#X"
1320          * d. fix or reject non-valid attributes
1321          *
1322          * @param string $text
1323          * @return string
1324          * @private
1325          */
1326         static function normalizeCharReferences( $text ) {
1327                 return preg_replace_callback(
1328                         self::CHAR_REFS_REGEX,
1329                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1330                         $text );
1331         }
1332
1333         /**
1334          * @param string $matches
1335          * @return string
1336          */
1337         static function normalizeCharReferencesCallback( $matches ) {
1338                 $ret = null;
1339                 if ( $matches[1] != '' ) {
1340                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1341                 } elseif ( $matches[2] != '' ) {
1342                         $ret = Sanitizer::decCharReference( $matches[2] );
1343                 } elseif ( $matches[3] != '' ) {
1344                         $ret = Sanitizer::hexCharReference( $matches[3] );
1345                 }
1346                 if ( is_null( $ret ) ) {
1347                         return htmlspecialchars( $matches[0] );
1348                 } else {
1349                         return $ret;
1350                 }
1351         }
1352
1353         /**
1354          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1355          * return the equivalent numeric entity reference (except for the core &lt;
1356          * &gt; &amp; &quot;). If the entity is a MediaWiki-specific alias, returns
1357          * the HTML equivalent. Otherwise, returns HTML-escaped text of
1358          * pseudo-entity source (eg &amp;foo;)
1359          *
1360          * @param string $name
1361          * @return string
1362          */
1363         static function normalizeEntity( $name ) {
1364                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1365                         return '&' . self::$htmlEntityAliases[$name] . ';';
1366                 } elseif ( in_array( $name, array( 'lt', 'gt', 'amp', 'quot' ) ) ) {
1367                         return "&$name;";
1368                 } elseif ( isset( self::$htmlEntities[$name] ) ) {
1369                         return '&#' . self::$htmlEntities[$name] . ';';
1370                 } else {
1371                         return "&amp;$name;";
1372                 }
1373         }
1374
1375         /**
1376          * @param int $codepoint
1377          * @return null|string
1378          */
1379         static function decCharReference( $codepoint ) {
1380                 $point = intval( $codepoint );
1381                 if ( Sanitizer::validateCodepoint( $point ) ) {
1382                         return sprintf( '&#%d;', $point );
1383                 } else {
1384                         return null;
1385                 }
1386         }
1387
1388         /**
1389          * @param int $codepoint
1390          * @return null|string
1391          */
1392         static function hexCharReference( $codepoint ) {
1393                 $point = hexdec( $codepoint );
1394                 if ( Sanitizer::validateCodepoint( $point ) ) {
1395                         return sprintf( '&#x%x;', $point );
1396                 } else {
1397                         return null;
1398                 }
1399         }
1400
1401         /**
1402          * Returns true if a given Unicode codepoint is a valid character in XML.
1403          * @param int $codepoint
1404          * @return bool
1405          */
1406         private static function validateCodepoint( $codepoint ) {
1407                 return $codepoint == 0x09
1408                         || $codepoint == 0x0a
1409                         || $codepoint == 0x0d
1410                         || ( $codepoint >= 0x20 && $codepoint <= 0xd7ff )
1411                         || ( $codepoint >= 0xe000 && $codepoint <= 0xfffd )
1412                         || ( $codepoint >= 0x10000 && $codepoint <= 0x10ffff );
1413         }
1414
1415         /**
1416          * Decode any character references, numeric or named entities,
1417          * in the text and return a UTF-8 string.
1418          *
1419          * @param string $text
1420          * @return string
1421          */
1422         public static function decodeCharReferences( $text ) {
1423                 return preg_replace_callback(
1424                         self::CHAR_REFS_REGEX,
1425                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1426                         $text );
1427         }
1428
1429         /**
1430          * Decode any character references, numeric or named entities,
1431          * in the next and normalize the resulting string. (bug 14952)
1432          *
1433          * This is useful for page titles, not for text to be displayed,
1434          * MediaWiki allows HTML entities to escape normalization as a feature.
1435          *
1436          * @param string $text Already normalized, containing entities
1437          * @return string Still normalized, without entities
1438          */
1439         public static function decodeCharReferencesAndNormalize( $text ) {
1440                 global $wgContLang;
1441                 $text = preg_replace_callback(
1442                         self::CHAR_REFS_REGEX,
1443                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1444                         $text, /* limit */ -1, $count );
1445
1446                 if ( $count ) {
1447                         return $wgContLang->normalize( $text );
1448                 } else {
1449                         return $text;
1450                 }
1451         }
1452
1453         /**
1454          * @param string $matches
1455          * @return string
1456          */
1457         static function decodeCharReferencesCallback( $matches ) {
1458                 if ( $matches[1] != '' ) {
1459                         return Sanitizer::decodeEntity( $matches[1] );
1460                 } elseif ( $matches[2] != '' ) {
1461                         return Sanitizer::decodeChar( intval( $matches[2] ) );
1462                 } elseif ( $matches[3] != '' ) {
1463                         return Sanitizer::decodeChar( hexdec( $matches[3] ) );
1464                 }
1465                 # Last case should be an ampersand by itself
1466                 return $matches[0];
1467         }
1468
1469         /**
1470          * Return UTF-8 string for a codepoint if that is a valid
1471          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1472          * @param int $codepoint
1473          * @return string
1474          * @private
1475          */
1476         static function decodeChar( $codepoint ) {
1477                 if ( Sanitizer::validateCodepoint( $codepoint ) ) {
1478                         return UtfNormal\Utils::codepointToUtf8( $codepoint );
1479                 } else {
1480                         return UtfNormal\Constants::UTF8_REPLACEMENT;
1481                 }
1482         }
1483
1484         /**
1485          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1486          * return the UTF-8 encoding of that character. Otherwise, returns
1487          * pseudo-entity source (eg "&foo;")
1488          *
1489          * @param string $name
1490          * @return string
1491          */
1492         static function decodeEntity( $name ) {
1493                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1494                         $name = self::$htmlEntityAliases[$name];
1495                 }
1496                 if ( isset( self::$htmlEntities[$name] ) ) {
1497                         return UtfNormal\Utils::codepointToUtf8( self::$htmlEntities[$name] );
1498                 } else {
1499                         return "&$name;";
1500                 }
1501         }
1502
1503         /**
1504          * Fetch the whitelist of acceptable attributes for a given element name.
1505          *
1506          * @param string $element
1507          * @return array
1508          */
1509         static function attributeWhitelist( $element ) {
1510                 $list = Sanitizer::setupAttributeWhitelist();
1511                 return isset( $list[$element] )
1512                         ? $list[$element]
1513                         : array();
1514         }
1515
1516         /**
1517          * Foreach array key (an allowed HTML element), return an array
1518          * of allowed attributes
1519          * @return array
1520          */
1521         static function setupAttributeWhitelist() {
1522                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
1523                 static $whitelist, $staticInitialised;
1524
1525                 $globalContext = implode( '-', compact( 'wgAllowRdfaAttributes', 'wgAllowMicrodataAttributes' ) );
1526
1527                 if ( $whitelist !== null && $staticInitialised == $globalContext ) {
1528                         return $whitelist;
1529                 }
1530
1531                 $common = array(
1532                         # HTML
1533                         'id',
1534                         'class',
1535                         'style',
1536                         'lang',
1537                         'dir',
1538                         'title',
1539
1540                         # WAI-ARIA
1541                         'role',
1542                 );
1543
1544                 if ( $wgAllowRdfaAttributes ) {
1545                         # RDFa attributes as specified in section 9 of
1546                         # http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1547                         $common = array_merge( $common, array(
1548                                 'about', 'property', 'resource', 'datatype', 'typeof',
1549                         ) );
1550                 }
1551
1552                 if ( $wgAllowMicrodataAttributes ) {
1553                         # add HTML5 microdata tags as specified by
1554                         # http://www.whatwg.org/html/microdata.html#the-microdata-model
1555                         $common = array_merge( $common, array(
1556                                 'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1557                         ) );
1558                 }
1559
1560                 $block = array_merge( $common, array( 'align' ) );
1561                 $tablealign = array( 'align', 'valign' );
1562                 $tablecell = array(
1563                         'abbr',
1564                         'axis',
1565                         'headers',
1566                         'scope',
1567                         'rowspan',
1568                         'colspan',
1569                         'nowrap', # deprecated
1570                         'width', # deprecated
1571                         'height', # deprecated
1572                         'bgcolor', # deprecated
1573                 );
1574
1575                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1576                 # See: http://www.w3.org/TR/html4/
1577                 $whitelist = array(
1578                         # 7.5.4
1579                         'div'        => $block,
1580                         'center'     => $common, # deprecated
1581                         'span'       => $common,
1582
1583                         # 7.5.5
1584                         'h1'         => $block,
1585                         'h2'         => $block,
1586                         'h3'         => $block,
1587                         'h4'         => $block,
1588                         'h5'         => $block,
1589                         'h6'         => $block,
1590
1591                         # 7.5.6
1592                         # address
1593
1594                         # 8.2.4
1595                         'bdo'        => $common,
1596
1597                         # 9.2.1
1598                         'em'         => $common,
1599                         'strong'     => $common,
1600                         'cite'       => $common,
1601                         'dfn'        => $common,
1602                         'code'       => $common,
1603                         'samp'       => $common,
1604                         'kbd'        => $common,
1605                         'var'        => $common,
1606                         'abbr'       => $common,
1607                         # acronym
1608
1609                         # 9.2.2
1610                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1611                         'q'          => array_merge( $common, array( 'cite' ) ),
1612
1613                         # 9.2.3
1614                         'sub'        => $common,
1615                         'sup'        => $common,
1616
1617                         # 9.3.1
1618                         'p'          => $block,
1619
1620                         # 9.3.2
1621                         'br'         => array_merge( $common, array( 'clear' ) ),
1622
1623                         # http://www.whatwg.org/html/text-level-semantics.html#the-wbr-element
1624                         'wbr'        => $common,
1625
1626                         # 9.3.4
1627                         'pre'        => array_merge( $common, array( 'width' ) ),
1628
1629                         # 9.4
1630                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1631                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1632
1633                         # 10.2
1634                         'ul'         => array_merge( $common, array( 'type' ) ),
1635                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1636                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1637
1638                         # 10.3
1639                         'dl'         => $common,
1640                         'dd'         => $common,
1641                         'dt'         => $common,
1642
1643                         # 11.2.1
1644                         'table'      => array_merge( $common,
1645                                                                 array( 'summary', 'width', 'border', 'frame',
1646                                                                                 'rules', 'cellspacing', 'cellpadding',
1647                                                                                 'align', 'bgcolor',
1648                                                                 ) ),
1649
1650                         # 11.2.2
1651                         'caption'    => $block,
1652
1653                         # 11.2.3
1654                         'thead'      => $common,
1655                         'tfoot'      => $common,
1656                         'tbody'      => $common,
1657
1658                         # 11.2.4
1659                         'colgroup'   => array_merge( $common, array( 'span' ) ),
1660                         'col'        => array_merge( $common, array( 'span' ) ),
1661
1662                         # 11.2.5
1663                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1664
1665                         # 11.2.6
1666                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1667                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1668
1669                         # 12.2
1670                         # NOTE: <a> is not allowed directly, but the attrib
1671                         # whitelist is used from the Parser object
1672                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1673
1674                         # 13.2
1675                         # Not usually allowed, but may be used for extension-style hooks
1676                         # such as <math> when it is rasterized, or if $wgAllowImageTag is
1677                         # true
1678                         'img'        => array_merge( $common, array( 'alt', 'src', 'width', 'height' ) ),
1679
1680                         # 15.2.1
1681                         'tt'         => $common,
1682                         'b'          => $common,
1683                         'i'          => $common,
1684                         'big'        => $common,
1685                         'small'      => $common,
1686                         'strike'     => $common,
1687                         's'          => $common,
1688                         'u'          => $common,
1689
1690                         # 15.2.2
1691                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1692                         # basefont
1693
1694                         # 15.3
1695                         'hr'         => array_merge( $common, array( 'width' ) ),
1696
1697                         # HTML Ruby annotation text module, simple ruby only.
1698                         # http://www.whatwg.org/html/text-level-semantics.html#the-ruby-element
1699                         'ruby'       => $common,
1700                         # rbc
1701                         'rb'         => $common,
1702                         'rp'         => $common,
1703                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1704                         'rtc'         => $common,
1705
1706                         # MathML root element, where used for extensions
1707                         # 'title' may not be 100% valid here; it's XHTML
1708                         # http://www.w3.org/TR/REC-MathML/
1709                         'math'       => array( 'class', 'style', 'id', 'title' ),
1710
1711                         # HTML 5 section 4.6
1712                         'bdi' => $common,
1713
1714                         # HTML5 elements, defined by:
1715                         # http://www.whatwg.org/html/
1716                         'data' => array_merge( $common, array( 'value' ) ),
1717                         'time' => array_merge( $common, array( 'datetime' ) ),
1718                         'mark' => $common,
1719
1720                         // meta and link are only permitted by removeHTMLtags when Microdata
1721                         // is enabled so we don't bother adding a conditional to hide these
1722                         // Also meta and link are only valid in WikiText as Microdata elements
1723                         // (ie: validateTag rejects tags missing the attributes needed for Microdata)
1724                         // So we don't bother including $common attributes that have no purpose.
1725                         'meta' => array( 'itemprop', 'content' ),
1726                         'link' => array( 'itemprop', 'href' ),
1727                 );
1728
1729                 $staticInitialised = $globalContext;
1730
1731                 return $whitelist;
1732         }
1733
1734         /**
1735          * Take a fragment of (potentially invalid) HTML and return
1736          * a version with any tags removed, encoded as plain text.
1737          *
1738          * Warning: this return value must be further escaped for literal
1739          * inclusion in HTML output as of 1.10!
1740          *
1741          * @param string $text HTML fragment
1742          * @return string
1743          */
1744         static function stripAllTags( $text ) {
1745                 # Actual <tags>
1746                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1747
1748                 # Normalize &entities and whitespace
1749                 $text = self::decodeCharReferences( $text );
1750                 $text = self::normalizeWhitespace( $text );
1751
1752                 return $text;
1753         }
1754
1755         /**
1756          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1757          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1758          * PHP 5.1 doesn't.
1759          *
1760          * Use for passing XHTML fragments to PHP's XML parsing functions
1761          *
1762          * @return string
1763          */
1764         static function hackDocType() {
1765                 $out = "<!DOCTYPE html [\n";
1766                 foreach ( self::$htmlEntities as $entity => $codepoint ) {
1767                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1768                 }
1769                 $out .= "]>\n";
1770                 return $out;
1771         }
1772
1773         /**
1774          * @param string $url
1775          * @return mixed|string
1776          */
1777         static function cleanUrl( $url ) {
1778                 # Normalize any HTML entities in input. They will be
1779                 # re-escaped by makeExternalLink().
1780                 $url = Sanitizer::decodeCharReferences( $url );
1781
1782                 # Escape any control characters introduced by the above step
1783                 $url = preg_replace_callback( '/[\][<>"\\x00-\\x20\\x7F\|]/',
1784                         array( __CLASS__, 'cleanUrlCallback' ), $url );
1785
1786                 # Validate hostname portion
1787                 $matches = array();
1788                 if ( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1789                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1790
1791                         // Characters that will be ignored in IDNs.
1792                         // http://tools.ietf.org/html/3454#section-3.1
1793                         // Strip them before further processing so blacklists and such work.
1794                         $strip = "/
1795                                 \\s|          # general whitespace
1796                                 \xc2\xad|     # 00ad SOFT HYPHEN
1797                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1798                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1799                                 \xe2\x81\xa0| # 2060 WORD JOINER
1800                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1801                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1802                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1803                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1804                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1805                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1806                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1807                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe0f VARIATION SELECTOR-1-16
1808                                 /xuD";
1809
1810                         $host = preg_replace( $strip, '', $host );
1811
1812                         // @todo FIXME: Validate hostnames here
1813
1814                         return $protocol . $host . $rest;
1815                 } else {
1816                         return $url;
1817                 }
1818         }
1819
1820         /**
1821          * @param array $matches
1822          * @return string
1823          */
1824         static function cleanUrlCallback( $matches ) {
1825                 return urlencode( $matches[0] );
1826         }
1827
1828         /**
1829          * Does a string look like an e-mail address?
1830          *
1831          * This validates an email address using an HTML5 specification found at:
1832          * http://www.whatwg.org/html/states-of-the-type-attribute.html#valid-e-mail-address
1833          * Which as of 2011-01-24 says:
1834          *
1835          *   A valid e-mail address is a string that matches the ABNF production
1836          *   1*( atext / "." ) "@" ldh-str *( "." ldh-str ) where atext is defined
1837          *   in RFC 5322 section 3.2.3, and ldh-str is defined in RFC 1034 section
1838          *   3.5.
1839          *
1840          * This function is an implementation of the specification as requested in
1841          * bug 22449.
1842          *
1843          * Client-side forms will use the same standard validation rules via JS or
1844          * HTML 5 validation; additional restrictions can be enforced server-side
1845          * by extensions via the 'isValidEmailAddr' hook.
1846          *
1847          * Note that this validation doesn't 100% match RFC 2822, but is believed
1848          * to be liberal enough for wide use. Some invalid addresses will still
1849          * pass validation here.
1850          *
1851          * @since 1.18
1852          *
1853          * @param string $addr E-mail address
1854          * @return bool
1855          */
1856         public static function validateEmail( $addr ) {
1857                 $result = null;
1858                 if ( !Hooks::run( 'isValidEmailAddr', array( $addr, &$result ) ) ) {
1859                         return $result;
1860                 }
1861
1862                 // Please note strings below are enclosed in brackets [], this make the
1863                 // hyphen "-" a range indicator. Hence it is double backslashed below.
1864                 // See bug 26948
1865                 $rfc5322_atext = "a-z0-9!#$%&'*+\\-\/=?^_`{|}~";
1866                 $rfc1034_ldh_str = "a-z0-9\\-";
1867
1868                 $html5_email_regexp = "/
1869                 ^                      # start of string
1870                 [$rfc5322_atext\\.]+    # user part which is liberal :p
1871                 @                      # 'apostrophe'
1872                 [$rfc1034_ldh_str]+       # First domain part
1873                 (\\.[$rfc1034_ldh_str]+)*  # Following part prefixed with a dot
1874                 $                      # End of string
1875                 /ix"; // case Insensitive, eXtended
1876
1877                 return (bool)preg_match( $html5_email_regexp, $addr );
1878         }
1879 }