includes/Sanitizer.php

   1 <?php
   2 /**
   3  * HTML sanitizer for %MediaWiki.
   4  *
   5  * Copyright © 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * http://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          * Blacklist for evil uris like javascript:
  44          * WARNING: DO NOT use this in any place that actually requires blacklisting
  45          * for security reasons. There are NUMEROUS[1] ways to bypass blacklisting, the
  46          * only way to be secure from javascript: uri based xss vectors is to whitelist
  47          * things that you know are safe and deny everything else.
  48          * [1]: http://ha.ckers.org/xss.html
  49          */
  50         const EVIL_URI_PATTERN = '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i';
  51         const XMLNS_ATTRIBUTE_PATTERN = "/^xmlns:[:A-Z_a-z-.0-9]+$/";
  52
  53         /**
  54          * List of all named character entities defined in HTML 4.01
  55          * http://www.w3.org/TR/html4/sgml/entities.html
  56          * As well as &apos; which is only defined starting in XHTML1.
  57          */
  58         private static $htmlEntities = array(
  59                 'Aacute'   => 193,
  60                 'aacute'   => 225,
  61                 'Acirc'    => 194,
  62                 'acirc'    => 226,
  63                 'acute'    => 180,
  64                 'AElig'    => 198,
  65                 'aelig'    => 230,
  66                 'Agrave'   => 192,
  67                 'agrave'   => 224,
  68                 'alefsym'  => 8501,
  69                 'Alpha'    => 913,
  70                 'alpha'    => 945,
  71                 'amp'      => 38,
  72                 'and'      => 8743,
  73                 'ang'      => 8736,
  74                 'apos'     => 39, // New in XHTML & HTML 5; avoid in output for compatibility with IE.
  75                 'Aring'    => 197,
  76                 'aring'    => 229,
  77                 'asymp'    => 8776,
  78                 'Atilde'   => 195,
  79                 'atilde'   => 227,
  80                 'Auml'     => 196,
  81                 'auml'     => 228,
  82                 'bdquo'    => 8222,
  83                 'Beta'     => 914,
  84                 'beta'     => 946,
  85                 'brvbar'   => 166,
  86                 'bull'     => 8226,
  87                 'cap'      => 8745,
  88                 'Ccedil'   => 199,
  89                 'ccedil'   => 231,
  90                 'cedil'    => 184,
  91                 'cent'     => 162,
  92                 'Chi'      => 935,
  93                 'chi'      => 967,
  94                 'circ'     => 710,
  95                 'clubs'    => 9827,
  96                 'cong'     => 8773,
  97                 'copy'     => 169,
  98                 'crarr'    => 8629,
  99                 'cup'      => 8746,
 100                 'curren'   => 164,
 101                 'dagger'   => 8224,
 102                 'Dagger'   => 8225,
 103                 'darr'     => 8595,
 104                 'dArr'     => 8659,
 105                 'deg'      => 176,
 106                 'Delta'    => 916,
 107                 'delta'    => 948,
 108                 'diams'    => 9830,
 109                 'divide'   => 247,
 110                 'Eacute'   => 201,
 111                 'eacute'   => 233,
 112                 'Ecirc'    => 202,
 113                 'ecirc'    => 234,
 114                 'Egrave'   => 200,
 115                 'egrave'   => 232,
 116                 'empty'    => 8709,
 117                 'emsp'     => 8195,
 118                 'ensp'     => 8194,
 119                 'Epsilon'  => 917,
 120                 'epsilon'  => 949,
 121                 'equiv'    => 8801,
 122                 'Eta'      => 919,
 123                 'eta'      => 951,
 124                 'ETH'      => 208,
 125                 'eth'      => 240,
 126                 'Euml'     => 203,
 127                 'euml'     => 235,
 128                 'euro'     => 8364,
 129                 'exist'    => 8707,
 130                 'fnof'     => 402,
 131                 'forall'   => 8704,
 132                 'frac12'   => 189,
 133                 'frac14'   => 188,
 134                 'frac34'   => 190,
 135                 'frasl'    => 8260,
 136                 'Gamma'    => 915,
 137                 'gamma'    => 947,
 138                 'ge'       => 8805,
 139                 'gt'       => 62,
 140                 'harr'     => 8596,
 141                 'hArr'     => 8660,
 142                 'hearts'   => 9829,
 143                 'hellip'   => 8230,
 144                 'Iacute'   => 205,
 145                 'iacute'   => 237,
 146                 'Icirc'    => 206,
 147                 'icirc'    => 238,
 148                 'iexcl'    => 161,
 149                 'Igrave'   => 204,
 150                 'igrave'   => 236,
 151                 'image'    => 8465,
 152                 'infin'    => 8734,
 153                 'int'      => 8747,
 154                 'Iota'     => 921,
 155                 'iota'     => 953,
 156                 'iquest'   => 191,
 157                 'isin'     => 8712,
 158                 'Iuml'     => 207,
 159                 'iuml'     => 239,
 160                 'Kappa'    => 922,
 161                 'kappa'    => 954,
 162                 'Lambda'   => 923,
 163                 'lambda'   => 955,
 164                 'lang'     => 9001,
 165                 'laquo'    => 171,
 166                 'larr'     => 8592,
 167                 'lArr'     => 8656,
 168                 'lceil'    => 8968,
 169                 'ldquo'    => 8220,
 170                 'le'       => 8804,
 171                 'lfloor'   => 8970,
 172                 'lowast'   => 8727,
 173                 'loz'      => 9674,
 174                 'lrm'      => 8206,
 175                 'lsaquo'   => 8249,
 176                 'lsquo'    => 8216,
 177                 'lt'       => 60,
 178                 'macr'     => 175,
 179                 'mdash'    => 8212,
 180                 'micro'    => 181,
 181                 'middot'   => 183,
 182                 'minus'    => 8722,
 183                 'Mu'       => 924,
 184                 'mu'       => 956,
 185                 'nabla'    => 8711,
 186                 'nbsp'     => 160,
 187                 'ndash'    => 8211,
 188                 'ne'       => 8800,
 189                 'ni'       => 8715,
 190                 'not'      => 172,
 191                 'notin'    => 8713,
 192                 'nsub'     => 8836,
 193                 'Ntilde'   => 209,
 194                 'ntilde'   => 241,
 195                 'Nu'       => 925,
 196                 'nu'       => 957,
 197                 'Oacute'   => 211,
 198                 'oacute'   => 243,
 199                 'Ocirc'    => 212,
 200                 'ocirc'    => 244,
 201                 'OElig'    => 338,
 202                 'oelig'    => 339,
 203                 'Ograve'   => 210,
 204                 'ograve'   => 242,
 205                 'oline'    => 8254,
 206                 'Omega'    => 937,
 207                 'omega'    => 969,
 208                 'Omicron'  => 927,
 209                 'omicron'  => 959,
 210                 'oplus'    => 8853,
 211                 'or'       => 8744,
 212                 'ordf'     => 170,
 213                 'ordm'     => 186,
 214                 'Oslash'   => 216,
 215                 'oslash'   => 248,
 216                 'Otilde'   => 213,
 217                 'otilde'   => 245,
 218                 'otimes'   => 8855,
 219                 'Ouml'     => 214,
 220                 'ouml'     => 246,
 221                 'para'     => 182,
 222                 'part'     => 8706,
 223                 'permil'   => 8240,
 224                 'perp'     => 8869,
 225                 'Phi'      => 934,
 226                 'phi'      => 966,
 227                 'Pi'       => 928,
 228                 'pi'       => 960,
 229                 'piv'      => 982,
 230                 'plusmn'   => 177,
 231                 'pound'    => 163,
 232                 'prime'    => 8242,
 233                 'Prime'    => 8243,
 234                 'prod'     => 8719,
 235                 'prop'     => 8733,
 236                 'Psi'      => 936,
 237                 'psi'      => 968,
 238                 'quot'     => 34,
 239                 'radic'    => 8730,
 240                 'rang'     => 9002,
 241                 'raquo'    => 187,
 242                 'rarr'     => 8594,
 243                 'rArr'     => 8658,
 244                 'rceil'    => 8969,
 245                 'rdquo'    => 8221,
 246                 'real'     => 8476,
 247                 'reg'      => 174,
 248                 'rfloor'   => 8971,
 249                 'Rho'      => 929,
 250                 'rho'      => 961,
 251                 'rlm'      => 8207,
 252                 'rsaquo'   => 8250,
 253                 'rsquo'    => 8217,
 254                 'sbquo'    => 8218,
 255                 'Scaron'   => 352,
 256                 'scaron'   => 353,
 257                 'sdot'     => 8901,
 258                 'sect'     => 167,
 259                 'shy'      => 173,
 260                 'Sigma'    => 931,
 261                 'sigma'    => 963,
 262                 'sigmaf'   => 962,
 263                 'sim'      => 8764,
 264                 'spades'   => 9824,
 265                 'sub'      => 8834,
 266                 'sube'     => 8838,
 267                 'sum'      => 8721,
 268                 'sup'      => 8835,
 269                 'sup1'     => 185,
 270                 'sup2'     => 178,
 271                 'sup3'     => 179,
 272                 'supe'     => 8839,
 273                 'szlig'    => 223,
 274                 'Tau'      => 932,
 275                 'tau'      => 964,
 276                 'there4'   => 8756,
 277                 'Theta'    => 920,
 278                 'theta'    => 952,
 279                 'thetasym' => 977,
 280                 'thinsp'   => 8201,
 281                 'THORN'    => 222,
 282                 'thorn'    => 254,
 283                 'tilde'    => 732,
 284                 'times'    => 215,
 285                 'trade'    => 8482,
 286                 'Uacute'   => 218,
 287                 'uacute'   => 250,
 288                 'uarr'     => 8593,
 289                 'uArr'     => 8657,
 290                 'Ucirc'    => 219,
 291                 'ucirc'    => 251,
 292                 'Ugrave'   => 217,
 293                 'ugrave'   => 249,
 294                 'uml'      => 168,
 295                 'upsih'    => 978,
 296                 'Upsilon'  => 933,
 297                 'upsilon'  => 965,
 298                 'Uuml'     => 220,
 299                 'uuml'     => 252,
 300                 'weierp'   => 8472,
 301                 'Xi'       => 926,
 302                 'xi'       => 958,
 303                 'Yacute'   => 221,
 304                 'yacute'   => 253,
 305                 'yen'      => 165,
 306                 'Yuml'     => 376,
 307                 'yuml'     => 255,
 308                 'Zeta'     => 918,
 309                 'zeta'     => 950,
 310                 'zwj'      => 8205,
 311                 'zwnj'     => 8204
 312         );
 313
 314         /**
 315          * Character entity aliases accepted by MediaWiki
 316          */
 317         private static $htmlEntityAliases = array(
 318                 'רלמ' => 'rlm',
 319                 'رلم' => 'rlm',
 320         );
 321
 322         /**
 323          * Lazy-initialised attributes regex, see getAttribsRegex()
 324          */
 325         private static $attribsRegex;
 326
 327         /**
 328          * Regular expression to match HTML/XML attribute pairs within a tag.
 329          * Allows some... latitude.
 330          * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
 331          */
 332         static function getAttribsRegex() {
 333                 if ( self::$attribsRegex === null ) {
 334                         $attribFirst = '[:A-Z_a-z0-9]';
 335                         $attrib = '[:A-Z_a-z-.0-9]';
 336                         $space = '[\x09\x0a\x0d\x20]';
 337                         self::$attribsRegex =
 338                                 "/(?:^|$space)({$attribFirst}{$attrib}*)
 339                                   ($space*=$space*
 340                                         (?:
 341                                          # The attribute value: quoted or alone
 342                                           \"([^<\"]*)\"
 343                                          | '([^<']*)'
 344                                          |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
 345                                          |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
 346                                                                                  # colors are specified like this.
 347                                                                                  # We'll be normalizing it.
 348                                         )
 349                                 )?(?=$space|\$)/sx";
 350                 }
 351                 return self::$attribsRegex;
 352         }
 353
 354         /**
 355          * Cleans up HTML, removes dangerous tags and attributes, and
 356          * removes HTML comments
 357          * @private
 358          * @param $text String
 359          * @param $processCallback Callback to do any variable or parameter
 360          *        replacements in HTML attribute values
 361          * @param array $args for the processing callback
 362          * @param array $extratags for any extra tags to include
 363          * @param array $removetags for any tags (default or extra) to exclude
 364          * @return string
 365          */
 366         static function removeHTMLtags( $text, $processCallback = null,
 367                 $args = array(), $extratags = array(), $removetags = array()
 368         ) {
 369                 global $wgUseTidy, $wgAllowMicrodataAttributes, $wgAllowImageTag;
 370
 371                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 372                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 373
 374                 wfProfileIn( __METHOD__ );
 375
 376                 // Base our staticInitialised variable off of the global config state so that if the globals
 377                 // are changed (like in the screwed up test system) we will re-initialise the settings.
 378                 $globalContext = implode( '-', compact( 'wgAllowMicrodataAttributes', 'wgAllowImageTag' ) );
 379                 if ( !$staticInitialised || $staticInitialised != $globalContext ) {
 380
 381                         $htmlpairsStatic = array( # Tags that must be closed
 382                                 'b', 'bdi', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 383                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 384                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 385                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 386                                 'ruby', 'rt', 'rb', 'rp', 'p', 'span', 'abbr', 'dfn',
 387                                 'kbd', 'samp', 'data', 'time', 'mark'
 388                         );
 389                         $htmlsingle = array(
 390                                 'br', 'wbr', 'hr', 'li', 'dt', 'dd'
 391                         );
 392                         $htmlsingleonly = array( # Elements that cannot have close tags
 393                                 'br', 'wbr', 'hr'
 394                         );
 395                         if ( $wgAllowMicrodataAttributes ) {
 396                                 $htmlsingle[] = $htmlsingleonly[] = 'meta';
 397                                 $htmlsingle[] = $htmlsingleonly[] = 'link';
 398                         }
 399                         $htmlnest = array( # Tags that can be nested--??
 400                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 401                                 'li', 'dl', 'dt', 'dd', 'font', 'big', 'small', 'sub', 'sup', 'span',
 402                                 'var', 'kbd', 'samp', 'em', 'strong', 'q', 'ruby', 'bdo'
 403                         );
 404                         $tabletags = array( # Can only appear inside table, we will close them
 405                                 'td', 'th', 'tr',
 406                         );
 407                         $htmllist = array( # Tags used by list
 408                                 'ul', 'ol',
 409                         );
 410                         $listtags = array( # Tags that can appear in a list
 411                                 'li',
 412                         );
 413
 414                         if ( $wgAllowImageTag ) {
 415                                 $htmlsingle[] = 'img';
 416                                 $htmlsingleonly[] = 'img';
 417                         }
 418
 419                         $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
 420                         $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
 421
 422                         # Convert them all to hashtables for faster lookup
 423                         $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 424                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
 425                         foreach ( $vars as $var ) {
 426                                 $$var = array_flip( $$var );
 427                         }
 428                         $staticInitialised = $globalContext;
 429                 }
 430                 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
 431                 $extratags = array_flip( $extratags );
 432                 $removetags = array_flip( $removetags );
 433                 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
 434                 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ), $removetags );
 435
 436                 # Remove HTML comments
 437                 $text = Sanitizer::removeHTMLcomments( $text );
 438                 $bits = explode( '<', $text );
 439                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 440                 if ( !$wgUseTidy ) {
 441                         $tagstack = $tablestack = array();
 442                         foreach ( $bits as $x ) {
 443                                 $regs = array();
 444                                 # $slash: Does the current element start with a '/'?
 445                                 # $t: Current element name
 446                                 # $params: String between element name and >
 447                                 # $brace: Ending '>' or '/>'
 448                                 # $rest: Everything until the next element of $bits
 449                                 if ( preg_match( '!^(/?)([^\\s/>]+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 450                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 451                                 } else {
 452                                         $slash = $t = $params = $brace = $rest = null;
 453                                 }
 454
 455                                 $badtag = false;
 456                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 457                                         # Check our stack
 458                                         if ( $slash && isset( $htmlsingleonly[$t] ) ) {
 459                                                 $badtag = true;
 460                                         } elseif ( $slash ) {
 461                                                 # Closing a tag... is it the one we just opened?
 462                                                 $ot = @array_pop( $tagstack );
 463                                                 if ( $ot != $t ) {
 464                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 465                                                                 # Pop all elements with an optional close tag
 466                                                                 # and see if we find a match below them
 467                                                                 $optstack = array();
 468                                                                 array_push( $optstack, $ot );
 469                                                                 wfSuppressWarnings();
 470                                                                 $ot = array_pop( $tagstack );
 471                                                                 wfRestoreWarnings();
 472                                                                 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
 473                                                                         array_push( $optstack, $ot );
 474                                                                         wfSuppressWarnings();
 475                                                                         $ot = array_pop( $tagstack );
 476                                                                         wfRestoreWarnings();
 477                                                                 }
 478                                                                 if ( $t != $ot ) {
 479                                                                         # No match. Push the optional elements back again
 480                                                                         $badtag = true;
 481                                                                         wfSuppressWarnings();
 482                                                                         $ot = array_pop( $optstack );
 483                                                                         wfRestoreWarnings();
 484                                                                         while ( $ot ) {
 485                                                                                 array_push( $tagstack, $ot );
 486                                                                                 wfSuppressWarnings();
 487                                                                                 $ot = array_pop( $optstack );
 488                                                                                 wfRestoreWarnings();
 489                                                                         }
 490                                                                 }
 491                                                         } else {
 492                                                                 @array_push( $tagstack, $ot );
 493                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 494                                                                 if ( !isset( $htmllist[$ot] ) || !isset( $listtags[$t] ) ) {
 495                                                                         $badtag = true;
 496                                                                 }
 497                                                         }
 498                                                 } else {
 499                                                         if ( $t == 'table' ) {
 500                                                                 $tagstack = array_pop( $tablestack );
 501                                                         }
 502                                                 }
 503                                                 $newparams = '';
 504                                         } else {
 505                                                 # Keep track for later
 506                                                 if ( isset( $tabletags[$t] ) &&
 507                                                 !in_array( 'table', $tagstack ) ) {
 508                                                         $badtag = true;
 509                                                 } elseif ( in_array( $t, $tagstack ) &&
 510                                                 !isset( $htmlnest[$t] ) ) {
 511                                                         $badtag = true;
 512                                                 # Is it a self closed htmlpair ? (bug 5487)
 513                                                 } elseif ( $brace == '/>' &&
 514                                                 isset( $htmlpairs[$t] ) ) {
 515                                                         $badtag = true;
 516                                                 } elseif ( isset( $htmlsingleonly[$t] ) ) {
 517                                                         # Hack to force empty tag for unclosable elements
 518                                                         $brace = '/>';
 519                                                 } elseif ( isset( $htmlsingle[$t] ) ) {
 520                                                         # Hack to not close $htmlsingle tags
 521                                                         $brace = null;
 522                                                         # Still need to push this optionally-closed tag to
 523                                                         # the tag stack so that we can match end tags
 524                                                         # instead of marking them as bad.
 525                                                         array_push( $tagstack, $t );
 526                                                 } elseif ( isset( $tabletags[$t] )
 527                                                 && in_array( $t, $tagstack ) ) {
 528                                                         // New table tag but forgot to close the previous one
 529                                                         $text .= "</$t>";
 530                                                 } else {
 531                                                         if ( $t == 'table' ) {
 532                                                                 array_push( $tablestack, $tagstack );
 533                                                                 $tagstack = array();
 534                                                         }
 535                                                         array_push( $tagstack, $t );
 536                                                 }
 537
 538                                                 # Replace any variables or template parameters with
 539                                                 # plaintext results.
 540                                                 if ( is_callable( $processCallback ) ) {
 541                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 542                                                 }
 543
 544                                                 if ( !Sanitizer::validateTag( $params, $t ) ) {
 545                                                         $badtag = true;
 546                                                 }
 547
 548                                                 # Strip non-approved attributes from the tag
 549                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 550                                         }
 551                                         if ( !$badtag ) {
 552                                                 $rest = str_replace( '>', '&gt;', $rest );
 553                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 554                                                 $text .= "<$slash$t$newparams$close>$rest";
 555                                                 continue;
 556                                         }
 557                                 }
 558                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 559                         }
 560                         # Close off any remaining tags
 561                         while ( is_array( $tagstack ) && ( $t = array_pop( $tagstack ) ) ) {
 562                                 $text .= "</$t>\n";
 563                                 if ( $t == 'table' ) {
 564                                         $tagstack = array_pop( $tablestack );
 565                                 }
 566                         }
 567                 } else {
 568                         # this might be possible using tidy itself
 569                         foreach ( $bits as $x ) {
 570                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 571                                 $x, $regs );
 572                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 573                                 $badtag = false;
 574                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 575                                         if ( is_callable( $processCallback ) ) {
 576                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 577                                         }
 578
 579                                         if ( !Sanitizer::validateTag( $params, $t ) ) {
 580                                                 $badtag = true;
 581                                         }
 582
 583                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 584                                         if ( !$badtag ) {
 585                                                 $rest = str_replace( '>', '&gt;', $rest );
 586                                                 $text .= "<$slash$t$newparams$brace$rest";
 587                                                 continue;
 588                                         }
 589                                 }
 590                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 591                         }
 592                 }
 593                 wfProfileOut( __METHOD__ );
 594                 return $text;
 595         }
 596
 597         /**
 598          * Remove '<!--', '-->', and everything between.
 599          * To avoid leaving blank lines, when a comment is both preceded
 600          * and followed by a newline (ignoring spaces), trim leading and
 601          * trailing spaces and one of the newlines.
 602          *
 603          * @private
 604          * @param $text String
 605          * @return string
 606          */
 607         static function removeHTMLcomments( $text ) {
 608                 wfProfileIn( __METHOD__ );
 609                 while ( ( $start = strpos( $text, '<!--' ) ) !== false ) {
 610                         $end = strpos( $text, '-->', $start + 4 );
 611                         if ( $end === false ) {
 612                                 # Unterminated comment; bail out
 613                                 break;
 614                         }
 615
 616                         $end += 3;
 617
 618                         # Trim space and newline if the comment is both
 619                         # preceded and followed by a newline
 620                         $spaceStart = max( $start - 1, 0 );
 621                         $spaceLen = $end - $spaceStart;
 622                         while ( substr( $text, $spaceStart, 1 ) === ' ' && $spaceStart > 0 ) {
 623                                 $spaceStart--;
 624                                 $spaceLen++;
 625                         }
 626                         while ( substr( $text, $spaceStart + $spaceLen, 1 ) === ' ' ) {
 627                                 $spaceLen++;
 628                         }
 629                         if ( substr( $text, $spaceStart, 1 ) === "\n"
 630                                 && substr( $text, $spaceStart + $spaceLen, 1 ) === "\n" ) {
 631                                 # Remove the comment, leading and trailing
 632                                 # spaces, and leave only one newline.
 633                                 $text = substr_replace( $text, "\n", $spaceStart, $spaceLen + 1 );
 634                         }
 635                         else {
 636                                 # Remove just the comment.
 637                                 $text = substr_replace( $text, '', $start, $end - $start );
 638                         }
 639                 }
 640                 wfProfileOut( __METHOD__ );
 641                 return $text;
 642         }
 643
 644         /**
 645          * Takes attribute names and values for a tag and the tag name and
 646          * validates that the tag is allowed to be present.
 647          * This DOES NOT validate the attributes, nor does it validate the
 648          * tags themselves. This method only handles the special circumstances
 649          * where we may want to allow a tag within content but ONLY when it has
 650          * specific attributes set.
 651          *
 652          * @param $params
 653          * @param $element
 654          * @return bool
 655          */
 656         static function validateTag( $params, $element ) {
 657                 $params = Sanitizer::decodeTagAttributes( $params );
 658
 659                 if ( $element == 'meta' || $element == 'link' ) {
 660                         if ( !isset( $params['itemprop'] ) ) {
 661                                 // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content
 662                                 return false;
 663                         }
 664                         if ( $element == 'meta' && !isset( $params['content'] ) ) {
 665                                 // <meta> must have a content="" for the itemprop
 666                                 return false;
 667                         }
 668                         if ( $element == 'link' && !isset( $params['href'] ) ) {
 669                                 // <link> must have an associated href=""
 670                                 return false;
 671                         }
 672                 }
 673
 674                 return true;
 675         }
 676
 677         /**
 678          * Take an array of attribute names and values and normalize or discard
 679          * illegal values for the given element type.
 680          *
 681          * - Discards attributes not on a whitelist for the given element
 682          * - Unsafe style attributes are discarded
 683          * - Invalid id attributes are re-encoded
 684          *
 685          * @param $attribs Array
 686          * @param $element String
 687          * @return Array
 688          *
 689          * @todo Check for legal values where the DTD limits things.
 690          * @todo Check for unique id attribute :P
 691          */
 692         static function validateTagAttributes( $attribs, $element ) {
 693                 return Sanitizer::validateAttributes( $attribs,
 694                         Sanitizer::attributeWhitelist( $element ) );
 695         }
 696
 697         /**
 698          * Take an array of attribute names and values and normalize or discard
 699          * illegal values for the given whitelist.
 700          *
 701          * - Discards attributes not the given whitelist
 702          * - Unsafe style attributes are discarded
 703          * - Invalid id attributes are re-encoded
 704          *
 705          * @param $attribs Array
 706          * @param array $whitelist list of allowed attribute names
 707          * @return Array
 708          *
 709          * @todo Check for legal values where the DTD limits things.
 710          * @todo Check for unique id attribute :P
 711          */
 712         static function validateAttributes( $attribs, $whitelist ) {
 713                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
 714
 715                 $whitelist = array_flip( $whitelist );
 716                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 717
 718                 $out = array();
 719                 foreach ( $attribs as $attribute => $value ) {
 720                         #allow XML namespace declaration if RDFa is enabled
 721                         if ( $wgAllowRdfaAttributes && preg_match( self::XMLNS_ATTRIBUTE_PATTERN, $attribute ) ) {
 722                                 if ( !preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 723                                         $out[$attribute] = $value;
 724                                 }
 725
 726                                 continue;
 727                         }
 728
 729                         # Allow any attribute beginning with "data-"
 730                         if ( !preg_match( '/^data-/i', $attribute ) && !isset( $whitelist[$attribute] ) ) {
 731                                 continue;
 732                         }
 733
 734                         # Strip javascript "expression" from stylesheets.
 735                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 736                         if ( $attribute == 'style' ) {
 737                                 $value = Sanitizer::checkCss( $value );
 738                         }
 739
 740                         if ( $attribute === 'id' ) {
 741                                 $value = Sanitizer::escapeId( $value, 'noninitial' );
 742                         }
 743
 744                         # WAI-ARIA
 745                         # http://www.w3.org/TR/wai-aria/
 746                         # http://www.whatwg.org/html/elements.html#wai-aria
 747                         # For now we only support role="presentation" until we work out what roles should be
 748                         # usable by content and we ensure that our code explicitly rejects patterns that
 749                         # violate HTML5's ARIA restrictions.
 750                         if ( $attribute === 'role' && $value !== 'presentation' ) {
 751                                 continue;
 752                         }
 753
 754                         // RDFa and microdata properties allow URLs, URIs and/or CURIs.
 755                         // Check them for sanity.
 756                         if ( $attribute === 'rel' || $attribute === 'rev'
 757                                 # RDFa
 758                                 || $attribute === 'about' || $attribute === 'property'
 759                                 || $attribute === 'resource' || $attribute === 'datatype'
 760                                 || $attribute === 'typeof'
 761                                 # HTML5 microdata
 762                                 || $attribute === 'itemid' || $attribute === 'itemprop'
 763                                 || $attribute === 'itemref' || $attribute === 'itemscope'
 764                                 || $attribute === 'itemtype'
 765                         ) {
 766                                 //Paranoia. Allow "simple" values but suppress javascript
 767                                 if ( preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 768                                         continue;
 769                                 }
 770                         }
 771
 772                         # NOTE: even though elements using href/src are not allowed directly, supply
 773                         #       validation code that can be used by tag hook handlers, etc
 774                         if ( $attribute === 'href' || $attribute === 'src' ) {
 775                                 if ( !preg_match( $hrefExp, $value ) ) {
 776                                         continue; //drop any href or src attributes not using an allowed protocol.
 777                                         // NOTE: this also drops all relative URLs
 778                                 }
 779                         }
 780
 781                         // If this attribute was previously set, override it.
 782                         // Output should only have one attribute of each name.
 783                         $out[$attribute] = $value;
 784                 }
 785
 786                 if ( $wgAllowMicrodataAttributes ) {
 787                         # itemtype, itemid, itemref don't make sense without itemscope
 788                         if ( !array_key_exists( 'itemscope', $out ) ) {
 789                                 unset( $out['itemtype'] );
 790                                 unset( $out['itemid'] );
 791                                 unset( $out['itemref'] );
 792                         }
 793                         # TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
 794                 }
 795                 return $out;
 796         }
 797
 798         /**
 799          * Merge two sets of HTML attributes.  Conflicting items in the second set
 800          * will override those in the first, except for 'class' attributes which
 801          * will be combined (if they're both strings).
 802          *
 803          * @todo implement merging for other attributes such as style
 804          * @param $a Array
 805          * @param $b Array
 806          * @return array
 807          */
 808         static function mergeAttributes( $a, $b ) {
 809                 $out = array_merge( $a, $b );
 810                 if ( isset( $a['class'] ) && isset( $b['class'] )
 811                         && is_string( $a['class'] ) && is_string( $b['class'] )
 812                         && $a['class'] !== $b['class']
 813                 ) {
 814                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 815                                 -1, PREG_SPLIT_NO_EMPTY );
 816                         $out['class'] = implode( ' ', array_unique( $classes ) );
 817                 }
 818                 return $out;
 819         }
 820
 821         /**
 822          * Pick apart some CSS and check it for forbidden or unsafe structures.
 823          * Returns a sanitized string. This sanitized string will have
 824          * character references and escape sequences decoded and comments
 825          * stripped (unless it is itself one valid comment, in which case the value
 826          * will be passed through). If the input is just too evil, only a comment
 827          * complaining about evilness will be returned.
 828          *
 829          * Currently URL references, 'expression', 'tps' are forbidden.
 830          *
 831          * NOTE: Despite the fact that character references are decoded, the
 832          * returned string may contain character references given certain
 833          * clever input strings. These character references must
 834          * be escaped before the return value is embedded in HTML.
 835          *
 836          * @param $value String
 837          * @return String
 838          */
 839         static function checkCss( $value ) {
 840                 // Decode character references like &#123;
 841                 $value = Sanitizer::decodeCharReferences( $value );
 842
 843                 // Decode escape sequences and line continuation
 844                 // See the grammar in the CSS 2 spec, appendix D.
 845                 // This has to be done AFTER decoding character references.
 846                 // This means it isn't possible for this function to return
 847                 // unsanitized escape sequences. It is possible to manufacture
 848                 // input that contains character references that decode to
 849                 // escape sequences that decode to character references, but
 850                 // it's OK for the return value to contain character references
 851                 // because the caller is supposed to escape those anyway.
 852                 static $decodeRegex;
 853                 if ( !$decodeRegex ) {
 854                         $space = '[\\x20\\t\\r\\n\\f]';
 855                         $nl = '(?:\\n|\\r\\n|\\r|\\f)';
 856                         $backslash = '\\\\';
 857                         $decodeRegex = "/ $backslash
 858                                 (?:
 859                                         ($nl) |  # 1. Line continuation
 860                                         ([0-9A-Fa-f]{1,6})$space? |  # 2. character number
 861                                         (.) | # 3. backslash cancelling special meaning
 862                                         () | # 4. backslash at end of string
 863                                 )/xu";
 864                 }
 865                 $value = preg_replace_callback( $decodeRegex,
 866                         array( __CLASS__, 'cssDecodeCallback' ), $value );
 867
 868                 // Normalize Halfwidth and Fullwidth Unicode block that IE6 might treat as ascii
 869                 $value = preg_replace_callback(
 870                         '/[！-ｚ]/u', // U+FF01 to U+FF5A
 871                         function ( $matches ) {
 872                                 $cp = utf8ToCodepoint( $matches[0] );
 873                                 if ( $cp === false ) {
 874                                         return '';
 875                                 }
 876                                 return chr( $cp - 65248 ); // ASCII range \x21-\x7A
 877                         },
 878                         $value
 879                 );
 880
 881                 // Convert more characters IE6 might treat as ascii
 882                 // U+0280, U+0274, U+207F, U+029F, U+026A, U+207D, U+208D
 883                 $value = str_replace(
 884                         array( 'ʀ', 'ɴ', 'ⁿ', 'ʟ', 'ɪ', '⁽', '₍' ),
 885                         array( 'r', 'n', 'n', 'l', 'i', '(', '(' ),
 886                         $value
 887                 );
 888
 889                 // Let the value through if it's nothing but a single comment, to
 890                 // allow other functions which may reject it to pass some error
 891                 // message through.
 892                 if ( !preg_match( '! ^ \s* /\* [^*\\/]* \*/ \s* $ !x', $value ) ) {
 893                         // Remove any comments; IE gets token splitting wrong
 894                         // This must be done AFTER decoding character references and
 895                         // escape sequences, because those steps can introduce comments
 896                         // This step cannot introduce character references or escape
 897                         // sequences, because it replaces comments with spaces rather
 898                         // than removing them completely.
 899                         $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
 900
 901                         // Remove anything after a comment-start token, to guard against
 902                         // incorrect client implementations.
 903                         $commentPos = strpos( $value, '/*' );
 904                         if ( $commentPos !== false ) {
 905                                 $value = substr( $value, 0, $commentPos );
 906                         }
 907                 }
 908
 909                 // S followed by repeat, iteration, or prolonged sound marks,
 910                 // which IE will treat as "ss"
 911                 $value = preg_replace(
 912                         '/s(?:
 913                                 \xE3\x80\xB1 | # U+3031
 914                                 \xE3\x82\x9D | # U+309D
 915                                 \xE3\x83\xBC | # U+30FC
 916                                 \xE3\x83\xBD | # U+30FD
 917                                 \xEF\xB9\xBC | # U+FE7C
 918                                 \xEF\xB9\xBD | # U+FE7D
 919                                 \xEF\xBD\xB0   # U+FF70
 920                         )/ix',
 921                         'ss',
 922                         $value
 923                 );
 924
 925                 // Reject problematic keywords and control characters
 926                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
 927                         return '/* invalid control char */';
 928                 } elseif ( preg_match( '! expression | filter\s*: | accelerator\s*: | url\s*\( | image\s*\( | image-set\s*\( !ix', $value ) ) {
 929                         return '/* insecure input */';
 930                 }
 931                 return $value;
 932         }
 933
 934         /**
 935          * @param $matches array
 936          * @return String
 937          */
 938         static function cssDecodeCallback( $matches ) {
 939                 if ( $matches[1] !== '' ) {
 940                         // Line continuation
 941                         return '';
 942                 } elseif ( $matches[2] !== '' ) {
 943                         $char = codepointToUtf8( hexdec( $matches[2] ) );
 944                 } elseif ( $matches[3] !== '' ) {
 945                         $char = $matches[3];
 946                 } else {
 947                         $char = '\\';
 948                 }
 949                 if ( $char == "\n" || $char == '"' || $char == "'" || $char == '\\' ) {
 950                         // These characters need to be escaped in strings
 951                         // Clean up the escape sequence to avoid parsing errors by clients
 952                         return '\\' . dechex( ord( $char ) ) . ' ';
 953                 } else {
 954                         // Decode unnecessary escape
 955                         return $char;
 956                 }
 957         }
 958
 959         /**
 960          * Take a tag soup fragment listing an HTML element's attributes
 961          * and normalize it to well-formed XML, discarding unwanted attributes.
 962          * Output is safe for further wikitext processing, with escaping of
 963          * values that could trigger problems.
 964          *
 965          * - Normalizes attribute names to lowercase
 966          * - Discards attributes not on a whitelist for the given element
 967          * - Turns broken or invalid entities into plaintext
 968          * - Double-quotes all attribute values
 969          * - Attributes without values are given the name as attribute
 970          * - Double attributes are discarded
 971          * - Unsafe style attributes are discarded
 972          * - Prepends space if there are attributes.
 973          *
 974          * @param $text String
 975          * @param $element String
 976          * @return String
 977          */
 978         static function fixTagAttributes( $text, $element ) {
 979                 if ( trim( $text ) == '' ) {
 980                         return '';
 981                 }
 982
 983                 $decoded = Sanitizer::decodeTagAttributes( $text );
 984                 $stripped = Sanitizer::validateTagAttributes( $decoded, $element );
 985
 986                 return Sanitizer::safeEncodeTagAttributes( $stripped );
 987         }
 988
 989         /**
 990          * Encode an attribute value for HTML output.
 991          * @param $text String
 992          * @return HTML-encoded text fragment
 993          */
 994         static function encodeAttribute( $text ) {
 995                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 996
 997                 // Whitespace is normalized during attribute decoding,
 998                 // so if we've been passed non-spaces we must encode them
 999                 // ahead of time or they won't be preserved.
1000                 $encValue = strtr( $encValue, array(
1001                         "\n" => '&#10;',
1002                         "\r" => '&#13;',
1003                         "\t" => '&#9;',
1004                 ) );
1005
1006                 return $encValue;
1007         }
1008
1009         /**
1010          * Encode an attribute value for HTML tags, with extra armoring
1011          * against further wiki processing.
1012          * @param $text String
1013          * @return HTML-encoded text fragment
1014          */
1015         static function safeEncodeAttribute( $text ) {
1016                 $encValue = Sanitizer::encodeAttribute( $text );
1017
1018                 # Templates and links may be expanded in later parsing,
1019                 # creating invalid or dangerous output. Suppress this.
1020                 $encValue = strtr( $encValue, array(
1021                         '<'    => '&lt;',   // This should never happen,
1022                         '>'    => '&gt;',   // we've received invalid input
1023                         '"'    => '&quot;', // which should have been escaped.
1024                         '{'    => '&#123;',
1025                         '['    => '&#91;',
1026                         "''"   => '&#39;&#39;',
1027                         'ISBN' => '&#73;SBN',
1028                         'RFC'  => '&#82;FC',
1029                         'PMID' => '&#80;MID',
1030                         '|'    => '&#124;',
1031                         '__'   => '&#95;_',
1032                 ) );
1033
1034                 # Stupid hack
1035                 $encValue = preg_replace_callback(
1036                         '/((?i)' . wfUrlProtocols() . ')/',
1037                         array( 'Sanitizer', 'armorLinksCallback' ),
1038                         $encValue );
1039                 return $encValue;
1040         }
1041
1042         /**
1043          * Given a value, escape it so that it can be used in an id attribute and
1044          * return it.  This will use HTML5 validation if $wgExperimentalHtmlIds is
1045          * true, allowing anything but ASCII whitespace.  Otherwise it will use
1046          * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
1047          * escaped with lots of dots.
1048          *
1049          * To ensure we don't have to bother escaping anything, we also strip ', ",
1050          * & even if $wgExperimentalIds is true.  TODO: Is this the best tactic?
1051          * We also strip # because it upsets IE, and % because it could be
1052          * ambiguous if it's part of something that looks like a percent escape
1053          * (which don't work reliably in fragments cross-browser).
1054          *
1055          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
1056          *                                                          in the id and
1057          *                                                          name attributes
1058          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
1059          * @see http://www.whatwg.org/html/elements.html#the-id-attribute
1060          *   HTML5 definition of id attribute
1061          *
1062          * @param string $id id to escape
1063          * @param $options Mixed: string or array of strings (default is array()):
1064          *   'noninitial': This is a non-initial fragment of an id, not a full id,
1065          *       so don't pay attention if the first character isn't valid at the
1066          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
1067          *       false.
1068          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
1069          *       if $wgExperimentalHtmlIds is used, so we can generate extra
1070          *       anchors and links won't break.
1071          * @return String
1072          */
1073         static function escapeId( $id, $options = array() ) {
1074                 global $wgExperimentalHtmlIds;
1075                 $options = (array)$options;
1076
1077                 if ( $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
1078                         $id = Sanitizer::decodeCharReferences( $id );
1079                         $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
1080                         $id = trim( $id, '_' );
1081                         if ( $id === '' ) {
1082                                 # Must have been all whitespace to start with.
1083                                 return '_';
1084                         } else {
1085                                 return $id;
1086                         }
1087                 }
1088
1089                 # HTML4-style escaping
1090                 static $replace = array(
1091                         '%3A' => ':',
1092                         '%' => '.'
1093                 );
1094
1095                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
1096                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
1097
1098                 if ( !preg_match( '/^[a-zA-Z]/', $id )
1099                 && !in_array( 'noninitial', $options ) ) {
1100                         // Initial character must be a letter!
1101                         $id = "x$id";
1102                 }
1103                 return $id;
1104         }
1105
1106         /**
1107          * Given a value, escape it so that it can be used as a CSS class and
1108          * return it.
1109          *
1110          * @todo For extra validity, input should be validated UTF-8.
1111          *
1112          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
1113          *
1114          * @param $class String
1115          * @return String
1116          */
1117         static function escapeClass( $class ) {
1118                 // Convert ugly stuff to underscores and kill underscores in ugly places
1119                 return rtrim( preg_replace(
1120                         array( '/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/', '/_+/' ),
1121                         '_',
1122                         $class ), '_' );
1123         }
1124
1125         /**
1126          * Given HTML input, escape with htmlspecialchars but un-escape entities.
1127          * This allows (generally harmless) entities like &#160; to survive.
1128          *
1129          * @param string $html to escape
1130          * @return String: escaped input
1131          */
1132         static function escapeHtmlAllowEntities( $html ) {
1133                 $html = Sanitizer::decodeCharReferences( $html );
1134                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
1135                 # hurt.
1136                 $html = htmlspecialchars( $html, ENT_QUOTES );
1137                 return $html;
1138         }
1139
1140         /**
1141          * Regex replace callback for armoring links against further processing.
1142          * @param $matches Array
1143          * @return string
1144          */
1145         private static function armorLinksCallback( $matches ) {
1146                 return str_replace( ':', '&#58;', $matches[1] );
1147         }
1148
1149         /**
1150          * Return an associative array of attribute names and values from
1151          * a partial tag string. Attribute names are forces to lowercase,
1152          * character references are decoded to UTF-8 text.
1153          *
1154          * @param $text String
1155          * @return Array
1156          */
1157         public static function decodeTagAttributes( $text ) {
1158                 if ( trim( $text ) == '' ) {
1159                         return array();
1160                 }
1161
1162                 $attribs = array();
1163                 $pairs = array();
1164                 if ( !preg_match_all(
1165                         self::getAttribsRegex(),
1166                         $text,
1167                         $pairs,
1168                         PREG_SET_ORDER ) ) {
1169                         return $attribs;
1170                 }
1171
1172                 foreach ( $pairs as $set ) {
1173                         $attribute = strtolower( $set[1] );
1174                         $value = Sanitizer::getTagAttributeCallback( $set );
1175
1176                         // Normalize whitespace
1177                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1178                         $value = trim( $value );
1179
1180                         // Decode character references
1181                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
1182                 }
1183                 return $attribs;
1184         }
1185
1186         /**
1187          * Build a partial tag string from an associative array of attribute
1188          * names and values as returned by decodeTagAttributes.
1189          *
1190          * @param $assoc_array Array
1191          * @return String
1192          */
1193         public static function safeEncodeTagAttributes( $assoc_array ) {
1194                 $attribs = array();
1195                 foreach ( $assoc_array as $attribute => $value ) {
1196                         $encAttribute = htmlspecialchars( $attribute );
1197                         $encValue = Sanitizer::safeEncodeAttribute( $value );
1198
1199                         $attribs[] = "$encAttribute=\"$encValue\"";
1200                 }
1201                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
1202         }
1203
1204         /**
1205          * Pick the appropriate attribute value from a match set from the
1206          * attribs regex matches.
1207          *
1208          * @param $set Array
1209          * @throws MWException
1210          * @return String
1211          */
1212         private static function getTagAttributeCallback( $set ) {
1213                 if ( isset( $set[6] ) ) {
1214                         # Illegal #XXXXXX color with no quotes.
1215                         return $set[6];
1216                 } elseif ( isset( $set[5] ) ) {
1217                         # No quotes.
1218                         return $set[5];
1219                 } elseif ( isset( $set[4] ) ) {
1220                         # Single-quoted
1221                         return $set[4];
1222                 } elseif ( isset( $set[3] ) ) {
1223                         # Double-quoted
1224                         return $set[3];
1225                 } elseif ( !isset( $set[2] ) ) {
1226                         # In XHTML, attributes must have a value.
1227                         # For 'reduced' form, return explicitly the attribute name here.
1228                         return $set[1];
1229                 } else {
1230                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1231                 }
1232         }
1233
1234         /**
1235          * Normalize whitespace and character references in an XML source-
1236          * encoded text for an attribute value.
1237          *
1238          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
1239          * but note that we're not returning the value, but are returning
1240          * XML source fragments that will be slapped into output.
1241          *
1242          * @param $text String
1243          * @return String
1244          */
1245         private static function normalizeAttributeValue( $text ) {
1246                 return str_replace( '"', '&quot;',
1247                         self::normalizeWhitespace(
1248                                 Sanitizer::normalizeCharReferences( $text ) ) );
1249         }
1250
1251         /**
1252          * @param $text string
1253          * @return mixed
1254          */
1255         private static function normalizeWhitespace( $text ) {
1256                 return preg_replace(
1257                         '/\r\n|[\x20\x0d\x0a\x09]/',
1258                         ' ',
1259                         $text );
1260         }
1261
1262         /**
1263          * Normalizes whitespace in a section name, such as might be returned
1264          * by Parser::stripSectionName(), for use in the id's that are used for
1265          * section links.
1266          *
1267          * @param $section String
1268          * @return String
1269          */
1270         static function normalizeSectionNameWhitespace( $section ) {
1271                 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1272         }
1273
1274         /**
1275          * Ensure that any entities and character references are legal
1276          * for XML and XHTML specifically. Any stray bits will be
1277          * &amp;-escaped to result in a valid text fragment.
1278          *
1279          * a. named char refs can only be &lt; &gt; &amp; &quot;, others are
1280          *   numericized (this way we're well-formed even without a DTD)
1281          * b. any numeric char refs must be legal chars, not invalid or forbidden
1282          * c. use lower cased "&#x", not "&#X"
1283          * d. fix or reject non-valid attributes
1284          *
1285          * @param $text String
1286          * @return String
1287          * @private
1288          */
1289         static function normalizeCharReferences( $text ) {
1290                 return preg_replace_callback(
1291                         self::CHAR_REFS_REGEX,
1292                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1293                         $text );
1294         }
1295         /**
1296          * @param $matches String
1297          * @return String
1298          */
1299         static function normalizeCharReferencesCallback( $matches ) {
1300                 $ret = null;
1301                 if ( $matches[1] != '' ) {
1302                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1303                 } elseif ( $matches[2] != '' ) {
1304                         $ret = Sanitizer::decCharReference( $matches[2] );
1305                 } elseif ( $matches[3] != '' ) {
1306                         $ret = Sanitizer::hexCharReference( $matches[3] );
1307                 }
1308                 if ( is_null( $ret ) ) {
1309                         return htmlspecialchars( $matches[0] );
1310                 } else {
1311                         return $ret;
1312                 }
1313         }
1314
1315         /**
1316          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1317          * return the equivalent numeric entity reference (except for the core &lt;
1318          * &gt; &amp; &quot;). If the entity is a MediaWiki-specific alias, returns
1319          * the HTML equivalent. Otherwise, returns HTML-escaped text of
1320          * pseudo-entity source (eg &amp;foo;)
1321          *
1322          * @param $name String
1323          * @return String
1324          */
1325         static function normalizeEntity( $name ) {
1326                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1327                         return '&' . self::$htmlEntityAliases[$name] . ';';
1328                 } elseif ( in_array( $name,
1329                 array( 'lt', 'gt', 'amp', 'quot' ) ) ) {
1330                         return "&$name;";
1331                 } elseif ( isset( self::$htmlEntities[$name] ) ) {
1332                         return '&#' . self::$htmlEntities[$name] . ';';
1333                 } else {
1334                         return "&amp;$name;";
1335                 }
1336         }
1337
1338         /**
1339          * @param $codepoint
1340          * @return null|string
1341          */
1342         static function decCharReference( $codepoint ) {
1343                 $point = intval( $codepoint );
1344                 if ( Sanitizer::validateCodepoint( $point ) ) {
1345                         return sprintf( '&#%d;', $point );
1346                 } else {
1347                         return null;
1348                 }
1349         }
1350
1351         /**
1352          * @param $codepoint
1353          * @return null|string
1354          */
1355         static function hexCharReference( $codepoint ) {
1356                 $point = hexdec( $codepoint );
1357                 if ( Sanitizer::validateCodepoint( $point ) ) {
1358                         return sprintf( '&#x%x;', $point );
1359                 } else {
1360                         return null;
1361                 }
1362         }
1363
1364         /**
1365          * Returns true if a given Unicode codepoint is a valid character in XML.
1366          * @param $codepoint Integer
1367          * @return Boolean
1368          */
1369         private static function validateCodepoint( $codepoint ) {
1370                 return $codepoint == 0x09
1371                         || $codepoint == 0x0a
1372                         || $codepoint == 0x0d
1373                         || ( $codepoint >= 0x20 && $codepoint <= 0xd7ff )
1374                         || ( $codepoint >= 0xe000 && $codepoint <= 0xfffd )
1375                         || ( $codepoint >= 0x10000 && $codepoint <= 0x10ffff );
1376         }
1377
1378         /**
1379          * Decode any character references, numeric or named entities,
1380          * in the text and return a UTF-8 string.
1381          *
1382          * @param $text String
1383          * @return String
1384          */
1385         public static function decodeCharReferences( $text ) {
1386                 return preg_replace_callback(
1387                         self::CHAR_REFS_REGEX,
1388                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1389                         $text );
1390         }
1391
1392         /**
1393          * Decode any character references, numeric or named entities,
1394          * in the next and normalize the resulting string. (bug 14952)
1395          *
1396          * This is useful for page titles, not for text to be displayed,
1397          * MediaWiki allows HTML entities to escape normalization as a feature.
1398          *
1399          * @param string $text (already normalized, containing entities)
1400          * @return String (still normalized, without entities)
1401          */
1402         public static function decodeCharReferencesAndNormalize( $text ) {
1403                 global $wgContLang;
1404                 $text = preg_replace_callback(
1405                         self::CHAR_REFS_REGEX,
1406                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1407                         $text, /* limit */ -1, $count );
1408
1409                 if ( $count ) {
1410                         return $wgContLang->normalize( $text );
1411                 } else {
1412                         return $text;
1413                 }
1414         }
1415
1416         /**
1417          * @param $matches String
1418          * @return String
1419          */
1420         static function decodeCharReferencesCallback( $matches ) {
1421                 if ( $matches[1] != '' ) {
1422                         return Sanitizer::decodeEntity( $matches[1] );
1423                 } elseif ( $matches[2] != '' ) {
1424                         return Sanitizer::decodeChar( intval( $matches[2] ) );
1425                 } elseif ( $matches[3] != '' ) {
1426                         return Sanitizer::decodeChar( hexdec( $matches[3] ) );
1427                 }
1428                 # Last case should be an ampersand by itself
1429                 return $matches[0];
1430         }
1431
1432         /**
1433          * Return UTF-8 string for a codepoint if that is a valid
1434          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1435          * @param $codepoint Integer
1436          * @return String
1437          * @private
1438          */
1439         static function decodeChar( $codepoint ) {
1440                 if ( Sanitizer::validateCodepoint( $codepoint ) ) {
1441                         return codepointToUtf8( $codepoint );
1442                 } else {
1443                         return UTF8_REPLACEMENT;
1444                 }
1445         }
1446
1447         /**
1448          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1449          * return the UTF-8 encoding of that character. Otherwise, returns
1450          * pseudo-entity source (eg "&foo;")
1451          *
1452          * @param $name String
1453          * @return String
1454          */
1455         static function decodeEntity( $name ) {
1456                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1457                         $name = self::$htmlEntityAliases[$name];
1458                 }
1459                 if ( isset( self::$htmlEntities[$name] ) ) {
1460                         return codepointToUtf8( self::$htmlEntities[$name] );
1461                 } else {
1462                         return "&$name;";
1463                 }
1464         }
1465
1466         /**
1467          * Fetch the whitelist of acceptable attributes for a given element name.
1468          *
1469          * @param $element String
1470          * @return Array
1471          */
1472         static function attributeWhitelist( $element ) {
1473                 $list = Sanitizer::setupAttributeWhitelist();
1474                 return isset( $list[$element] )
1475                         ? $list[$element]
1476                         : array();
1477         }
1478
1479         /**
1480          * Foreach array key (an allowed HTML element), return an array
1481          * of allowed attributes
1482          * @return Array
1483          */
1484         static function setupAttributeWhitelist() {
1485                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
1486
1487                 static $whitelist, $staticInitialised;
1488                 $globalContext = implode( '-', compact( 'wgAllowRdfaAttributes', 'wgAllowMicrodataAttributes' ) );
1489
1490                 if ( isset( $whitelist ) && $staticInitialised == $globalContext ) {
1491                         return $whitelist;
1492                 }
1493
1494                 $common = array(
1495                         # HTML
1496                         'id',
1497                         'class',
1498                         'style',
1499                         'lang',
1500                         'dir',
1501                         'title',
1502
1503                         # WAI-ARIA
1504                         'role',
1505                 );
1506
1507                 if ( $wgAllowRdfaAttributes ) {
1508                         # RDFa attributes as specified in section 9 of
1509                         # http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1510                         $common = array_merge( $common, array(
1511                                 'about', 'property', 'resource', 'datatype', 'typeof',
1512                         ) );
1513                 }
1514
1515                 if ( $wgAllowMicrodataAttributes ) {
1516                         # add HTML5 microdata tags as specified by
1517                         # http://www.whatwg.org/html/microdata.html#the-microdata-model
1518                         $common = array_merge( $common, array(
1519                                 'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1520                         ) );
1521                 }
1522
1523                 $block = array_merge( $common, array( 'align' ) );
1524                 $tablealign = array( 'align', 'valign' );
1525                 $tablecell = array(
1526                         'abbr',
1527                         'axis',
1528                         'headers',
1529                         'scope',
1530                         'rowspan',
1531                         'colspan',
1532                         'nowrap', # deprecated
1533                         'width', # deprecated
1534                         'height', # deprecated
1535                         'bgcolor', # deprecated
1536                 );
1537
1538                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1539                 # See: http://www.w3.org/TR/html4/
1540                 $whitelist = array(
1541                         # 7.5.4
1542                         'div'        => $block,
1543                         'center'     => $common, # deprecated
1544                         'span'       => $common,
1545
1546                         # 7.5.5
1547                         'h1'         => $block,
1548                         'h2'         => $block,
1549                         'h3'         => $block,
1550                         'h4'         => $block,
1551                         'h5'         => $block,
1552                         'h6'         => $block,
1553
1554                         # 7.5.6
1555                         # address
1556
1557                         # 8.2.4
1558                         'bdo'        => $common,
1559
1560                         # 9.2.1
1561                         'em'         => $common,
1562                         'strong'     => $common,
1563                         'cite'       => $common,
1564                         'dfn'        => $common,
1565                         'code'       => $common,
1566                         'samp'       => $common,
1567                         'kbd'        => $common,
1568                         'var'        => $common,
1569                         'abbr'       => $common,
1570                         # acronym
1571
1572                         # 9.2.2
1573                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1574                         'q'          => array_merge( $common, array( 'cite' ) ),
1575
1576                         # 9.2.3
1577                         'sub'        => $common,
1578                         'sup'        => $common,
1579
1580                         # 9.3.1
1581                         'p'          => $block,
1582
1583                         # 9.3.2
1584                         'br'         => array_merge( $common, array( 'clear' ) ),
1585
1586                         # http://www.whatwg.org/html/text-level-semantics.html#the-wbr-element
1587                         'wbr'        => $common,
1588
1589                         # 9.3.4
1590                         'pre'        => array_merge( $common, array( 'width' ) ),
1591
1592                         # 9.4
1593                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1594                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1595
1596                         # 10.2
1597                         'ul'         => array_merge( $common, array( 'type' ) ),
1598                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1599                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1600
1601                         # 10.3
1602                         'dl'         => $common,
1603                         'dd'         => $common,
1604                         'dt'         => $common,
1605
1606                         # 11.2.1
1607                         'table'      => array_merge( $common,
1608                                                                 array( 'summary', 'width', 'border', 'frame',
1609                                                                                 'rules', 'cellspacing', 'cellpadding',
1610                                                                                 'align', 'bgcolor',
1611                                                                 ) ),
1612
1613                         # 11.2.2
1614                         'caption'    => $block,
1615
1616                         # 11.2.3
1617                         'thead'      => $common,
1618                         'tfoot'      => $common,
1619                         'tbody'      => $common,
1620
1621                         # 11.2.4
1622                         'colgroup'   => array_merge( $common, array( 'span' ) ),
1623                         'col'        => array_merge( $common, array( 'span' ) ),
1624
1625                         # 11.2.5
1626                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1627
1628                         # 11.2.6
1629                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1630                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1631
1632                         # 12.2
1633                         # NOTE: <a> is not allowed directly, but the attrib
1634                         # whitelist is used from the Parser object
1635                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1636
1637                         # 13.2
1638                         # Not usually allowed, but may be used for extension-style hooks
1639                         # such as <math> when it is rasterized, or if $wgAllowImageTag is
1640                         # true
1641                         'img'        => array_merge( $common, array( 'alt', 'src', 'width', 'height' ) ),
1642
1643                         # 15.2.1
1644                         'tt'         => $common,
1645                         'b'          => $common,
1646                         'i'          => $common,
1647                         'big'        => $common,
1648                         'small'      => $common,
1649                         'strike'     => $common,
1650                         's'          => $common,
1651                         'u'          => $common,
1652
1653                         # 15.2.2
1654                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1655                         # basefont
1656
1657                         # 15.3
1658                         'hr'         => array_merge( $common, array( 'width' ) ),
1659
1660                         # HTML Ruby annotation text module, simple ruby only.
1661                         # http://www.whatwg.org/html/text-level-semantics.html#the-ruby-element
1662                         'ruby'       => $common,
1663                         # rbc
1664                         # rtc
1665                         'rb'         => $common,
1666                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1667                         'rp'         => $common,
1668
1669                         # MathML root element, where used for extensions
1670                         # 'title' may not be 100% valid here; it's XHTML
1671                         # http://www.w3.org/TR/REC-MathML/
1672                         'math'       => array( 'class', 'style', 'id', 'title' ),
1673
1674                         # HTML 5 section 4.6
1675                         'bdi' => $common,
1676
1677                         # HTML5 elements, defined by:
1678                         # http://www.whatwg.org/html/
1679                         'data' => array_merge( $common, array( 'value' ) ),
1680                         'time' => array_merge( $common, array( 'datetime' ) ),
1681                         'mark' => $common,
1682
1683                         // meta and link are only permitted by removeHTMLtags when Microdata
1684                         // is enabled so we don't bother adding a conditional to hide these
1685                         // Also meta and link are only valid in WikiText as Microdata elements
1686                         // (ie: validateTag rejects tags missing the attributes needed for Microdata)
1687                         // So we don't bother including $common attributes that have no purpose.
1688                         'meta' => array( 'itemprop', 'content' ),
1689                         'link' => array( 'itemprop', 'href' ),
1690                 );
1691
1692                 $staticInitialised = $globalContext;
1693
1694                 return $whitelist;
1695         }
1696
1697         /**
1698          * Take a fragment of (potentially invalid) HTML and return
1699          * a version with any tags removed, encoded as plain text.
1700          *
1701          * Warning: this return value must be further escaped for literal
1702          * inclusion in HTML output as of 1.10!
1703          *
1704          * @param string $text HTML fragment
1705          * @return String
1706          */
1707         static function stripAllTags( $text ) {
1708                 # Actual <tags>
1709                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1710
1711                 # Normalize &entities and whitespace
1712                 $text = self::decodeCharReferences( $text );
1713                 $text = self::normalizeWhitespace( $text );
1714
1715                 return $text;
1716         }
1717
1718         /**
1719          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1720          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1721          * PHP 5.1 doesn't.
1722          *
1723          * Use for passing XHTML fragments to PHP's XML parsing functions
1724          *
1725          * @return String
1726          */
1727         static function hackDocType() {
1728                 $out = "<!DOCTYPE html [\n";
1729                 foreach ( self::$htmlEntities as $entity => $codepoint ) {
1730                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1731                 }
1732                 $out .= "]>\n";
1733                 return $out;
1734         }
1735
1736         /**
1737          * @param $url string
1738          * @return mixed|string
1739          */
1740         static function cleanUrl( $url ) {
1741                 # Normalize any HTML entities in input. They will be
1742                 # re-escaped by makeExternalLink().
1743                 $url = Sanitizer::decodeCharReferences( $url );
1744
1745                 # Escape any control characters introduced by the above step
1746                 $url = preg_replace_callback( '/[\][<>"\\x00-\\x20\\x7F\|]/',
1747                         array( __CLASS__, 'cleanUrlCallback' ), $url );
1748
1749                 # Validate hostname portion
1750                 $matches = array();
1751                 if ( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1752                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1753
1754                         // Characters that will be ignored in IDNs.
1755                         // http://tools.ietf.org/html/3454#section-3.1
1756                         // Strip them before further processing so blacklists and such work.
1757                         $strip = "/
1758                                 \\s|          # general whitespace
1759                                 \xc2\xad|     # 00ad SOFT HYPHEN
1760                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1761                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1762                                 \xe2\x81\xa0| # 2060 WORD JOINER
1763                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1764                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1765                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1766                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1767                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1768                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1769                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1770                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe0f VARIATION SELECTOR-1-16
1771                                 /xuD";
1772
1773                         $host = preg_replace( $strip, '', $host );
1774
1775                         // @todo FIXME: Validate hostnames here
1776
1777                         return $protocol . $host . $rest;
1778                 } else {
1779                         return $url;
1780                 }
1781         }
1782
1783         /**
1784          * @param $matches array
1785          * @return string
1786          */
1787         static function cleanUrlCallback( $matches ) {
1788                 return urlencode( $matches[0] );
1789         }
1790
1791         /**
1792          * Does a string look like an e-mail address?
1793          *
1794          * This validates an email address using an HTML5 specification found at:
1795          * http://www.whatwg.org/html/states-of-the-type-attribute.html#valid-e-mail-address
1796          * Which as of 2011-01-24 says:
1797          *
1798          *   A valid e-mail address is a string that matches the ABNF production
1799          *   1*( atext / "." ) "@" ldh-str *( "." ldh-str ) where atext is defined
1800          *   in RFC 5322 section 3.2.3, and ldh-str is defined in RFC 1034 section
1801          *   3.5.
1802          *
1803          * This function is an implementation of the specification as requested in
1804          * bug 22449.
1805          *
1806          * Client-side forms will use the same standard validation rules via JS or
1807          * HTML 5 validation; additional restrictions can be enforced server-side
1808          * by extensions via the 'isValidEmailAddr' hook.
1809          *
1810          * Note that this validation doesn't 100% match RFC 2822, but is believed
1811          * to be liberal enough for wide use. Some invalid addresses will still
1812          * pass validation here.
1813          *
1814          * @since 1.18
1815          *
1816          * @param string $addr E-mail address
1817          * @return Bool
1818          */
1819         public static function validateEmail( $addr ) {
1820                 $result = null;
1821                 if ( !wfRunHooks( 'isValidEmailAddr', array( $addr, &$result ) ) ) {
1822                         return $result;
1823                 }
1824
1825                 // Please note strings below are enclosed in brackets [], this make the
1826                 // hyphen "-" a range indicator. Hence it is double backslashed below.
1827                 // See bug 26948
1828                 $rfc5322_atext = "a-z0-9!#$%&'*+\\-\/=?^_`{|}~";
1829                 $rfc1034_ldh_str = "a-z0-9\\-";
1830
1831                 $HTML5_email_regexp = "/
1832                 ^                      # start of string
1833                 [$rfc5322_atext\\.]+    # user part which is liberal :p
1834                 @                      # 'apostrophe'
1835                 [$rfc1034_ldh_str]+       # First domain part
1836                 (\\.[$rfc1034_ldh_str]+)*  # Following part prefixed with a dot
1837                 $                      # End of string
1838                 /ix"; // case Insensitive, eXtended
1839
1840                 return (bool)preg_match( $HTML5_email_regexp, $addr );
1841         }
1842 }