includes/Sanitizer.php

   1 <?php
   2 /**
   3  * HTML sanitizer for %MediaWiki.
   4  *
   5  * Copyright © 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Parser
  25  */
  26
  27 /**
  28  * HTML sanitizer for MediaWiki
  29  * @ingroup Parser
  30  */
  31 class Sanitizer {
  32         /**
  33          * Regular expression to match various types of character references in
  34          * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  35          */
  36         const CHAR_REFS_REGEX =
  37                 '/&([A-Za-z0-9\x80-\xff]+);
  38                  |&\#([0-9]+);
  39                  |&\#[xX]([0-9A-Fa-f]+);
  40                  |(&)/x';
  41
  42         /**
  43          * 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 $attribsRegex;
 326
 327         /**
 328          * Regular expression to match HTML/XML attribute pairs within a tag.
 329          * Allows some... latitude.
 330          * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
 331          * @return string
 332          */
 333         static function getAttribsRegex() {
 334                 if ( self::$attribsRegex === null ) {
 335                         $attribFirst = '[:A-Z_a-z0-9]';
 336                         $attrib = '[:A-Z_a-z-.0-9]';
 337                         $space = '[\x09\x0a\x0d\x20]';
 338                         self::$attribsRegex =
 339                                 "/(?:^|$space)({$attribFirst}{$attrib}*)
 340                                   ($space*=$space*
 341                                         (?:
 342                                          # The attribute value: quoted or alone
 343                                           \"([^<\"]*)\"
 344                                          | '([^<']*)'
 345                                          |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
 346                                          |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
 347                                                                                  # colors are specified like this.
 348                                                                                  # We'll be normalizing it.
 349                                         )
 350                                 )?(?=$space|\$)/sx";
 351                 }
 352                 return self::$attribsRegex;
 353         }
 354
 355         /**
 356          * Cleans up HTML, removes dangerous tags and attributes, and
 357          * removes HTML comments
 358          * @private
 359          * @param string $text
 360          * @param callable $processCallback Callback to do any variable or parameter
 361          *   replacements in HTML attribute values
 362          * @param array|bool $args Arguments for the processing callback
 363          * @param array $extratags For any extra tags to include
 364          * @param array $removetags For any tags (default or extra) to exclude
 365          * @return string
 366          */
 367         static function removeHTMLtags( $text, $processCallback = null,
 368                 $args = array(), $extratags = array(), $removetags = array()
 369         ) {
 370                 global $wgUseTidy, $wgAllowMicrodataAttributes, $wgAllowImageTag;
 371
 372                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 373                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 374
 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', 'rb', 'rp', 'rt', 'rtc', '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                                                 wfSuppressWarnings();
 463                                                 $ot = array_pop( $tagstack );
 464                                                 wfRestoreWarnings();
 465
 466                                                 if ( $ot != $t ) {
 467                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 468                                                                 # Pop all elements with an optional close tag
 469                                                                 # and see if we find a match below them
 470                                                                 $optstack = array();
 471                                                                 array_push( $optstack, $ot );
 472                                                                 wfSuppressWarnings();
 473                                                                 $ot = array_pop( $tagstack );
 474                                                                 wfRestoreWarnings();
 475                                                                 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
 476                                                                         array_push( $optstack, $ot );
 477                                                                         wfSuppressWarnings();
 478                                                                         $ot = array_pop( $tagstack );
 479                                                                         wfRestoreWarnings();
 480                                                                 }
 481                                                                 if ( $t != $ot ) {
 482                                                                         # No match. Push the optional elements back again
 483                                                                         $badtag = true;
 484                                                                         wfSuppressWarnings();
 485                                                                         $ot = array_pop( $optstack );
 486                                                                         wfRestoreWarnings();
 487                                                                         while ( $ot ) {
 488                                                                                 array_push( $tagstack, $ot );
 489                                                                                 wfSuppressWarnings();
 490                                                                                 $ot = array_pop( $optstack );
 491                                                                                 wfRestoreWarnings();
 492                                                                         }
 493                                                                 }
 494                                                         } else {
 495                                                                 wfSuppressWarnings();
 496                                                                 array_push( $tagstack, $ot );
 497                                                                 wfRestoreWarnings();
 498
 499                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 500                                                                 if ( !isset( $htmllist[$ot] ) || !isset( $listtags[$t] ) ) {
 501                                                                         $badtag = true;
 502                                                                 }
 503                                                         }
 504                                                 } else {
 505                                                         if ( $t == 'table' ) {
 506                                                                 $tagstack = array_pop( $tablestack );
 507                                                         }
 508                                                 }
 509                                                 $newparams = '';
 510                                         } else {
 511                                                 # Keep track for later
 512                                                 if ( isset( $tabletags[$t] ) && !in_array( 'table', $tagstack ) ) {
 513                                                         $badtag = true;
 514                                                 } elseif ( in_array( $t, $tagstack ) && !isset( $htmlnest[$t] ) ) {
 515                                                         $badtag = true;
 516                                                 # Is it a self closed htmlpair ? (bug 5487)
 517                                                 } elseif ( $brace == '/>' && isset( $htmlpairs[$t] ) ) {
 518                                                         $badtag = true;
 519                                                 } elseif ( isset( $htmlsingleonly[$t] ) ) {
 520                                                         # Hack to force empty tag for unclosable elements
 521                                                         $brace = '/>';
 522                                                 } elseif ( isset( $htmlsingle[$t] ) ) {
 523                                                         # Hack to not close $htmlsingle tags
 524                                                         $brace = null;
 525                                                         # Still need to push this optionally-closed tag to
 526                                                         # the tag stack so that we can match end tags
 527                                                         # instead of marking them as bad.
 528                                                         array_push( $tagstack, $t );
 529                                                 } elseif ( isset( $tabletags[$t] ) && in_array( $t, $tagstack ) ) {
 530                                                         // New table tag but forgot to close the previous one
 531                                                         $text .= "</$t>";
 532                                                 } else {
 533                                                         if ( $t == 'table' ) {
 534                                                                 array_push( $tablestack, $tagstack );
 535                                                                 $tagstack = array();
 536                                                         }
 537                                                         array_push( $tagstack, $t );
 538                                                 }
 539
 540                                                 # Replace any variables or template parameters with
 541                                                 # plaintext results.
 542                                                 if ( is_callable( $processCallback ) ) {
 543                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 544                                                 }
 545
 546                                                 if ( !Sanitizer::validateTag( $params, $t ) ) {
 547                                                         $badtag = true;
 548                                                 }
 549
 550                                                 # Strip non-approved attributes from the tag
 551                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 552                                         }
 553                                         if ( !$badtag ) {
 554                                                 $rest = str_replace( '>', '&gt;', $rest );
 555                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 556                                                 $text .= "<$slash$t$newparams$close>$rest";
 557                                                 continue;
 558                                         }
 559                                 }
 560                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 561                         }
 562                         # Close off any remaining tags
 563                         while ( is_array( $tagstack ) && ( $t = array_pop( $tagstack ) ) ) {
 564                                 $text .= "</$t>\n";
 565                                 if ( $t == 'table' ) {
 566                                         $tagstack = array_pop( $tablestack );
 567                                 }
 568                         }
 569                 } else {
 570                         # this might be possible using tidy itself
 571                         foreach ( $bits as $x ) {
 572                                 preg_match(
 573                                         '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 574                                         $x,
 575                                         $regs
 576                                 );
 577
 578                                 wfSuppressWarnings();
 579                                 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 580                                 wfRestoreWarnings();
 581
 582                                 $badtag = false;
 583                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 584                                         if ( is_callable( $processCallback ) ) {
 585                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 586                                         }
 587
 588                                         if ( !Sanitizer::validateTag( $params, $t ) ) {
 589                                                 $badtag = true;
 590                                         }
 591
 592                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 593                                         if ( !$badtag ) {
 594                                                 $rest = str_replace( '>', '&gt;', $rest );
 595                                                 $text .= "<$slash$t$newparams$brace$rest";
 596                                                 continue;
 597                                         }
 598                                 }
 599                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 600                         }
 601                 }
 602                 return $text;
 603         }
 604
 605         /**
 606          * Remove '<!--', '-->', and everything between.
 607          * To avoid leaving blank lines, when a comment is both preceded
 608          * and followed by a newline (ignoring spaces), trim leading and
 609          * trailing spaces and one of the newlines.
 610          *
 611          * @private
 612          * @param string $text
 613          * @return string
 614          */
 615         static function removeHTMLcomments( $text ) {
 616                 while ( ( $start = strpos( $text, '<!--' ) ) !== false ) {
 617                         $end = strpos( $text, '-->', $start + 4 );
 618                         if ( $end === false ) {
 619                                 # Unterminated comment; bail out
 620                                 break;
 621                         }
 622
 623                         $end += 3;
 624
 625                         # Trim space and newline if the comment is both
 626                         # preceded and followed by a newline
 627                         $spaceStart = max( $start - 1, 0 );
 628                         $spaceLen = $end - $spaceStart;
 629                         while ( substr( $text, $spaceStart, 1 ) === ' ' && $spaceStart > 0 ) {
 630                                 $spaceStart--;
 631                                 $spaceLen++;
 632                         }
 633                         while ( substr( $text, $spaceStart + $spaceLen, 1 ) === ' ' ) {
 634                                 $spaceLen++;
 635                         }
 636                         if ( substr( $text, $spaceStart, 1 ) === "\n"
 637                                 && substr( $text, $spaceStart + $spaceLen, 1 ) === "\n" ) {
 638                                 # Remove the comment, leading and trailing
 639                                 # spaces, and leave only one newline.
 640                                 $text = substr_replace( $text, "\n", $spaceStart, $spaceLen + 1 );
 641                         } else {
 642                                 # Remove just the comment.
 643                                 $text = substr_replace( $text, '', $start, $end - $start );
 644                         }
 645                 }
 646                 return $text;
 647         }
 648
 649         /**
 650          * Takes attribute names and values for a tag and the tag name and
 651          * validates that the tag is allowed to be present.
 652          * This DOES NOT validate the attributes, nor does it validate the
 653          * tags themselves. This method only handles the special circumstances
 654          * where we may want to allow a tag within content but ONLY when it has
 655          * specific attributes set.
 656          *
 657          * @param string $params
 658          * @param string $element
 659          * @return bool
 660          */
 661         static function validateTag( $params, $element ) {
 662                 $params = Sanitizer::decodeTagAttributes( $params );
 663
 664                 if ( $element == 'meta' || $element == 'link' ) {
 665                         if ( !isset( $params['itemprop'] ) ) {
 666                                 // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content
 667                                 return false;
 668                         }
 669                         if ( $element == 'meta' && !isset( $params['content'] ) ) {
 670                                 // <meta> must have a content="" for the itemprop
 671                                 return false;
 672                         }
 673                         if ( $element == 'link' && !isset( $params['href'] ) ) {
 674                                 // <link> must have an associated href=""
 675                                 return false;
 676                         }
 677                 }
 678
 679                 return true;
 680         }
 681
 682         /**
 683          * Take an array of attribute names and values and normalize or discard
 684          * illegal values for the given element type.
 685          *
 686          * - Discards attributes not on a whitelist for the given element
 687          * - Unsafe style attributes are discarded
 688          * - Invalid id attributes are re-encoded
 689          *
 690          * @param array $attribs
 691          * @param string $element
 692          * @return array
 693          *
 694          * @todo Check for legal values where the DTD limits things.
 695          * @todo Check for unique id attribute :P
 696          */
 697         static function validateTagAttributes( $attribs, $element ) {
 698                 return Sanitizer::validateAttributes( $attribs,
 699                         Sanitizer::attributeWhitelist( $element ) );
 700         }
 701
 702         /**
 703          * Take an array of attribute names and values and normalize or discard
 704          * illegal values for the given whitelist.
 705          *
 706          * - Discards attributes not the given whitelist
 707          * - Unsafe style attributes are discarded
 708          * - Invalid id attributes are re-encoded
 709          *
 710          * @param array $attribs
 711          * @param array $whitelist List of allowed attribute names
 712          * @return array
 713          *
 714          * @todo Check for legal values where the DTD limits things.
 715          * @todo Check for unique id attribute :P
 716          */
 717         static function validateAttributes( $attribs, $whitelist ) {
 718                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
 719
 720                 $whitelist = array_flip( $whitelist );
 721                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 722
 723                 $out = array();
 724                 foreach ( $attribs as $attribute => $value ) {
 725                         #allow XML namespace declaration if RDFa is enabled
 726                         if ( $wgAllowRdfaAttributes && preg_match( self::XMLNS_ATTRIBUTE_PATTERN, $attribute ) ) {
 727                                 if ( !preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 728                                         $out[$attribute] = $value;
 729                                 }
 730
 731                                 continue;
 732                         }
 733
 734                         # Allow any attribute beginning with "data-"
 735                         if ( !preg_match( '/^data-/i', $attribute ) && !isset( $whitelist[$attribute] ) ) {
 736                                 continue;
 737                         }
 738
 739                         # Strip javascript "expression" from stylesheets.
 740                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 741                         if ( $attribute == 'style' ) {
 742                                 $value = Sanitizer::checkCss( $value );
 743                         }
 744
 745                         if ( $attribute === 'id' ) {
 746                                 $value = Sanitizer::escapeId( $value, 'noninitial' );
 747                         }
 748
 749                         # WAI-ARIA
 750                         # http://www.w3.org/TR/wai-aria/
 751                         # http://www.whatwg.org/html/elements.html#wai-aria
 752                         # For now we only support role="presentation" until we work out what roles should be
 753                         # usable by content and we ensure that our code explicitly rejects patterns that
 754                         # violate HTML5's ARIA restrictions.
 755                         if ( $attribute === 'role' && $value !== 'presentation' ) {
 756                                 continue;
 757                         }
 758
 759                         // RDFa and microdata properties allow URLs, URIs and/or CURIs.
 760                         // Check them for sanity.
 761                         if ( $attribute === 'rel' || $attribute === 'rev'
 762                                 # RDFa
 763                                 || $attribute === 'about' || $attribute === 'property'
 764                                 || $attribute === 'resource' || $attribute === 'datatype'
 765                                 || $attribute === 'typeof'
 766                                 # HTML5 microdata
 767                                 || $attribute === 'itemid' || $attribute === 'itemprop'
 768                                 || $attribute === 'itemref' || $attribute === 'itemscope'
 769                                 || $attribute === 'itemtype'
 770                         ) {
 771                                 //Paranoia. Allow "simple" values but suppress javascript
 772                                 if ( preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 773                                         continue;
 774                                 }
 775                         }
 776
 777                         # NOTE: even though elements using href/src are not allowed directly, supply
 778                         #       validation code that can be used by tag hook handlers, etc
 779                         if ( $attribute === 'href' || $attribute === 'src' ) {
 780                                 if ( !preg_match( $hrefExp, $value ) ) {
 781                                         continue; //drop any href or src attributes not using an allowed protocol.
 782                                         // NOTE: this also drops all relative URLs
 783                                 }
 784                         }
 785
 786                         // If this attribute was previously set, override it.
 787                         // Output should only have one attribute of each name.
 788                         $out[$attribute] = $value;
 789                 }
 790
 791                 if ( $wgAllowMicrodataAttributes ) {
 792                         # itemtype, itemid, itemref don't make sense without itemscope
 793                         if ( !array_key_exists( 'itemscope', $out ) ) {
 794                                 unset( $out['itemtype'] );
 795                                 unset( $out['itemid'] );
 796                                 unset( $out['itemref'] );
 797                         }
 798                         # TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
 799                 }
 800                 return $out;
 801         }
 802
 803         /**
 804          * Merge two sets of HTML attributes.  Conflicting items in the second set
 805          * will override those in the first, except for 'class' attributes which
 806          * will be combined (if they're both strings).
 807          *
 808          * @todo implement merging for other attributes such as style
 809          * @param array $a
 810          * @param array $b
 811          * @return array
 812          */
 813         static function mergeAttributes( $a, $b ) {
 814                 $out = array_merge( $a, $b );
 815                 if ( isset( $a['class'] ) && isset( $b['class'] )
 816                         && is_string( $a['class'] ) && is_string( $b['class'] )
 817                         && $a['class'] !== $b['class']
 818                 ) {
 819                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 820                                 -1, PREG_SPLIT_NO_EMPTY );
 821                         $out['class'] = implode( ' ', array_unique( $classes ) );
 822                 }
 823                 return $out;
 824         }
 825
 826         /**
 827          * Normalize CSS into a format we can easily search for hostile input
 828          *  - decode character references
 829          *  - decode escape sequences
 830          *  - convert characters that IE6 interprets into ascii
 831          *  - remove comments, unless the entire value is one single comment
 832          * @param string $value the css string
 833          * @return string normalized css
 834          */
 835         public static function normalizeCss( $value ) {
 836
 837                 // Decode character references like &#123;
 838                 $value = Sanitizer::decodeCharReferences( $value );
 839
 840                 // Decode escape sequences and line continuation
 841                 // See the grammar in the CSS 2 spec, appendix D.
 842                 // This has to be done AFTER decoding character references.
 843                 // This means it isn't possible for this function to return
 844                 // unsanitized escape sequences. It is possible to manufacture
 845                 // input that contains character references that decode to
 846                 // escape sequences that decode to character references, but
 847                 // it's OK for the return value to contain character references
 848                 // because the caller is supposed to escape those anyway.
 849                 static $decodeRegex;
 850                 if ( !$decodeRegex ) {
 851                         $space = '[\\x20\\t\\r\\n\\f]';
 852                         $nl = '(?:\\n|\\r\\n|\\r|\\f)';
 853                         $backslash = '\\\\';
 854                         $decodeRegex = "/ $backslash
 855                                 (?:
 856                                         ($nl) |  # 1. Line continuation
 857                                         ([0-9A-Fa-f]{1,6})$space? |  # 2. character number
 858                                         (.) | # 3. backslash cancelling special meaning
 859                                         () | # 4. backslash at end of string
 860                                 )/xu";
 861                 }
 862                 $value = preg_replace_callback( $decodeRegex,
 863                         array( __CLASS__, 'cssDecodeCallback' ), $value );
 864
 865                 // Normalize Halfwidth and Fullwidth Unicode block that IE6 might treat as ascii
 866                 $value = preg_replace_callback(
 867                         '/[！-［］-ｚ]/u', // U+FF01 to U+FF5A, excluding U+FF3C (bug 58088)
 868                         function ( $matches ) {
 869                                 $cp = utf8ToCodepoint( $matches[0] );
 870                                 if ( $cp === false ) {
 871                                         return '';
 872                                 }
 873                                 return chr( $cp - 65248 ); // ASCII range \x21-\x7A
 874                         },
 875                         $value
 876                 );
 877
 878                 // Convert more characters IE6 might treat as ascii
 879                 // U+0280, U+0274, U+207F, U+029F, U+026A, U+207D, U+208D
 880                 $value = str_replace(
 881                         array( 'ʀ', 'ɴ', 'ⁿ', 'ʟ', 'ɪ', '⁽', '₍' ),
 882                         array( 'r', 'n', 'n', 'l', 'i', '(', '(' ),
 883                         $value
 884                 );
 885
 886                 // Let the value through if it's nothing but a single comment, to
 887                 // allow other functions which may reject it to pass some error
 888                 // message through.
 889                 if ( !preg_match( '! ^ \s* /\* [^*\\/]* \*/ \s* $ !x', $value ) ) {
 890                         // Remove any comments; IE gets token splitting wrong
 891                         // This must be done AFTER decoding character references and
 892                         // escape sequences, because those steps can introduce comments
 893                         // This step cannot introduce character references or escape
 894                         // sequences, because it replaces comments with spaces rather
 895                         // than removing them completely.
 896                         $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
 897
 898                         // Remove anything after a comment-start token, to guard against
 899                         // incorrect client implementations.
 900                         $commentPos = strpos( $value, '/*' );
 901                         if ( $commentPos !== false ) {
 902                                 $value = substr( $value, 0, $commentPos );
 903                         }
 904                 }
 905
 906                 // S followed by repeat, iteration, or prolonged sound marks,
 907                 // which IE will treat as "ss"
 908                 $value = preg_replace(
 909                         '/s(?:
 910                                 \xE3\x80\xB1 | # U+3031
 911                                 \xE3\x82\x9D | # U+309D
 912                                 \xE3\x83\xBC | # U+30FC
 913                                 \xE3\x83\xBD | # U+30FD
 914                                 \xEF\xB9\xBC | # U+FE7C
 915                                 \xEF\xB9\xBD | # U+FE7D
 916                                 \xEF\xBD\xB0   # U+FF70
 917                         )/ix',
 918                         'ss',
 919                         $value
 920                 );
 921
 922                 return $value;
 923         }
 924
 925
 926         /**
 927          * Pick apart some CSS and check it for forbidden or unsafe structures.
 928          * Returns a sanitized string. This sanitized string will have
 929          * character references and escape sequences decoded and comments
 930          * stripped (unless it is itself one valid comment, in which case the value
 931          * will be passed through). If the input is just too evil, only a comment
 932          * complaining about evilness will be returned.
 933          *
 934          * Currently URL references, 'expression', 'tps' are forbidden.
 935          *
 936          * NOTE: Despite the fact that character references are decoded, the
 937          * returned string may contain character references given certain
 938          * clever input strings. These character references must
 939          * be escaped before the return value is embedded in HTML.
 940          *
 941          * @param string $value
 942          * @return string
 943          */
 944         static function checkCss( $value ) {
 945                 $value = self::normalizeCss( $value );
 946
 947                 // Reject problematic keywords and control characters
 948                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
 949                         return '/* invalid control char */';
 950                 } elseif ( preg_match(
 951                         '! expression
 952                                 | filter\s*:
 953                                 | accelerator\s*:
 954                                 | -o-link\s*:
 955                                 | -o-link-source\s*:
 956                                 | -o-replace\s*:
 957                                 | url\s*\(
 958                                 | image\s*\(
 959                                 | image-set\s*\(
 960                         !ix', $value ) ) {
 961                         return '/* insecure input */';
 962                 }
 963                 return $value;
 964         }
 965
 966         /**
 967          * @param array $matches
 968          * @return string
 969          */
 970         static function cssDecodeCallback( $matches ) {
 971                 if ( $matches[1] !== '' ) {
 972                         // Line continuation
 973                         return '';
 974                 } elseif ( $matches[2] !== '' ) {
 975                         $char = codepointToUtf8( hexdec( $matches[2] ) );
 976                 } elseif ( $matches[3] !== '' ) {
 977                         $char = $matches[3];
 978                 } else {
 979                         $char = '\\';
 980                 }
 981                 if ( $char == "\n" || $char == '"' || $char == "'" || $char == '\\' ) {
 982                         // These characters need to be escaped in strings
 983                         // Clean up the escape sequence to avoid parsing errors by clients
 984                         return '\\' . dechex( ord( $char ) ) . ' ';
 985                 } else {
 986                         // Decode unnecessary escape
 987                         return $char;
 988                 }
 989         }
 990
 991         /**
 992          * Take a tag soup fragment listing an HTML element's attributes
 993          * and normalize it to well-formed XML, discarding unwanted attributes.
 994          * Output is safe for further wikitext processing, with escaping of
 995          * values that could trigger problems.
 996          *
 997          * - Normalizes attribute names to lowercase
 998          * - Discards attributes not on a whitelist for the given element
 999          * - Turns broken or invalid entities into plaintext
1000          * - Double-quotes all attribute values
1001          * - Attributes without values are given the name as attribute
1002          * - Double attributes are discarded
1003          * - Unsafe style attributes are discarded
1004          * - Prepends space if there are attributes.
1005          *
1006          * @param string $text
1007          * @param string $element
1008          * @return string
1009          */
1010         static function fixTagAttributes( $text, $element ) {
1011                 if ( trim( $text ) == '' ) {
1012                         return '';
1013                 }
1014
1015                 $decoded = Sanitizer::decodeTagAttributes( $text );
1016                 $stripped = Sanitizer::validateTagAttributes( $decoded, $element );
1017
1018                 return Sanitizer::safeEncodeTagAttributes( $stripped );
1019         }
1020
1021         /**
1022          * Encode an attribute value for HTML output.
1023          * @param string $text
1024          * @return string HTML-encoded text fragment
1025          */
1026         static function encodeAttribute( $text ) {
1027                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
1028
1029                 // Whitespace is normalized during attribute decoding,
1030                 // so if we've been passed non-spaces we must encode them
1031                 // ahead of time or they won't be preserved.
1032                 $encValue = strtr( $encValue, array(
1033                         "\n" => '&#10;',
1034                         "\r" => '&#13;',
1035                         "\t" => '&#9;',
1036                 ) );
1037
1038                 return $encValue;
1039         }
1040
1041         /**
1042          * Encode an attribute value for HTML tags, with extra armoring
1043          * against further wiki processing.
1044          * @param string $text
1045          * @return string HTML-encoded text fragment
1046          */
1047         static function safeEncodeAttribute( $text ) {
1048                 $encValue = Sanitizer::encodeAttribute( $text );
1049
1050                 # Templates and links may be expanded in later parsing,
1051                 # creating invalid or dangerous output. Suppress this.
1052                 $encValue = strtr( $encValue, array(
1053                         '<'    => '&lt;',   // This should never happen,
1054                         '>'    => '&gt;',   // we've received invalid input
1055                         '"'    => '&quot;', // which should have been escaped.
1056                         '{'    => '&#123;',
1057                         '['    => '&#91;',
1058                         "''"   => '&#39;&#39;',
1059                         'ISBN' => '&#73;SBN',
1060                         'RFC'  => '&#82;FC',
1061                         'PMID' => '&#80;MID',
1062                         '|'    => '&#124;',
1063                         '__'   => '&#95;_',
1064                 ) );
1065
1066                 # Stupid hack
1067                 $encValue = preg_replace_callback(
1068                         '/((?i)' . wfUrlProtocols() . ')/',
1069                         array( 'Sanitizer', 'armorLinksCallback' ),
1070                         $encValue );
1071                 return $encValue;
1072         }
1073
1074         /**
1075          * Given a value, escape it so that it can be used in an id attribute and
1076          * return it.  This will use HTML5 validation if $wgExperimentalHtmlIds is
1077          * true, allowing anything but ASCII whitespace.  Otherwise it will use
1078          * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
1079          * escaped with lots of dots.
1080          *
1081          * To ensure we don't have to bother escaping anything, we also strip ', ",
1082          * & even if $wgExperimentalIds is true.  TODO: Is this the best tactic?
1083          * We also strip # because it upsets IE, and % because it could be
1084          * ambiguous if it's part of something that looks like a percent escape
1085          * (which don't work reliably in fragments cross-browser).
1086          *
1087          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
1088          *   in the id and name attributes
1089          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with
1090          *   the id attribute
1091          * @see http://www.whatwg.org/html/elements.html#the-id-attribute
1092          *   HTML5 definition of id attribute
1093          *
1094          * @param string $id Id to escape
1095          * @param string|array $options String or array of strings (default is array()):
1096          *   'noninitial': This is a non-initial fragment of an id, not a full id,
1097          *       so don't pay attention if the first character isn't valid at the
1098          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
1099          *       false.
1100          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
1101          *       if $wgExperimentalHtmlIds is used, so we can generate extra
1102          *       anchors and links won't break.
1103          * @return string
1104          */
1105         static function escapeId( $id, $options = array() ) {
1106                 global $wgExperimentalHtmlIds;
1107                 $options = (array)$options;
1108
1109                 $id = Sanitizer::decodeCharReferences( $id );
1110
1111                 if ( $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
1112                         $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
1113                         $id = trim( $id, '_' );
1114                         if ( $id === '' ) {
1115                                 // Must have been all whitespace to start with.
1116                                 return '_';
1117                         } else {
1118                                 return $id;
1119                         }
1120                 }
1121
1122                 // HTML4-style escaping
1123                 static $replace = array(
1124                         '%3A' => ':',
1125                         '%' => '.'
1126                 );
1127
1128                 $id = urlencode( strtr( $id, ' ', '_' ) );
1129                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
1130
1131                 if ( !preg_match( '/^[a-zA-Z]/', $id ) && !in_array( 'noninitial', $options ) ) {
1132                         // Initial character must be a letter!
1133                         $id = "x$id";
1134                 }
1135                 return $id;
1136         }
1137
1138         /**
1139          * Given a value, escape it so that it can be used as a CSS class and
1140          * return it.
1141          *
1142          * @todo For extra validity, input should be validated UTF-8.
1143          *
1144          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
1145          *
1146          * @param string $class
1147          * @return string
1148          */
1149         static function escapeClass( $class ) {
1150                 // Convert ugly stuff to underscores and kill underscores in ugly places
1151                 return rtrim( preg_replace(
1152                         array( '/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/', '/_+/' ),
1153                         '_',
1154                         $class ), '_' );
1155         }
1156
1157         /**
1158          * Given HTML input, escape with htmlspecialchars but un-escape entities.
1159          * This allows (generally harmless) entities like &#160; to survive.
1160          *
1161          * @param string $html HTML to escape
1162          * @return string Escaped input
1163          */
1164         static function escapeHtmlAllowEntities( $html ) {
1165                 $html = Sanitizer::decodeCharReferences( $html );
1166                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
1167                 # hurt.
1168                 $html = htmlspecialchars( $html, ENT_QUOTES );
1169                 return $html;
1170         }
1171
1172         /**
1173          * Regex replace callback for armoring links against further processing.
1174          * @param array $matches
1175          * @return string
1176          */
1177         private static function armorLinksCallback( $matches ) {
1178                 return str_replace( ':', '&#58;', $matches[1] );
1179         }
1180
1181         /**
1182          * Return an associative array of attribute names and values from
1183          * a partial tag string. Attribute names are forces to lowercase,
1184          * character references are decoded to UTF-8 text.
1185          *
1186          * @param string $text
1187          * @return array
1188          */
1189         public static function decodeTagAttributes( $text ) {
1190                 if ( trim( $text ) == '' ) {
1191                         return array();
1192                 }
1193
1194                 $attribs = array();
1195                 $pairs = array();
1196                 if ( !preg_match_all(
1197                         self::getAttribsRegex(),
1198                         $text,
1199                         $pairs,
1200                         PREG_SET_ORDER ) ) {
1201                         return $attribs;
1202                 }
1203
1204                 foreach ( $pairs as $set ) {
1205                         $attribute = strtolower( $set[1] );
1206                         $value = Sanitizer::getTagAttributeCallback( $set );
1207
1208                         // Normalize whitespace
1209                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1210                         $value = trim( $value );
1211
1212                         // Decode character references
1213                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
1214                 }
1215                 return $attribs;
1216         }
1217
1218         /**
1219          * Build a partial tag string from an associative array of attribute
1220          * names and values as returned by decodeTagAttributes.
1221          *
1222          * @param array $assoc_array
1223          * @return string
1224          */
1225         public static function safeEncodeTagAttributes( $assoc_array ) {
1226                 $attribs = array();
1227                 foreach ( $assoc_array as $attribute => $value ) {
1228                         $encAttribute = htmlspecialchars( $attribute );
1229                         $encValue = Sanitizer::safeEncodeAttribute( $value );
1230
1231                         $attribs[] = "$encAttribute=\"$encValue\"";
1232                 }
1233                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
1234         }
1235
1236         /**
1237          * Pick the appropriate attribute value from a match set from the
1238          * attribs regex matches.
1239          *
1240          * @param array $set
1241          * @throws MWException When tag conditions are not met.
1242          * @return string
1243          */
1244         private static function getTagAttributeCallback( $set ) {
1245                 if ( isset( $set[6] ) ) {
1246                         # Illegal #XXXXXX color with no quotes.
1247                         return $set[6];
1248                 } elseif ( isset( $set[5] ) ) {
1249                         # No quotes.
1250                         return $set[5];
1251                 } elseif ( isset( $set[4] ) ) {
1252                         # Single-quoted
1253                         return $set[4];
1254                 } elseif ( isset( $set[3] ) ) {
1255                         # Double-quoted
1256                         return $set[3];
1257                 } elseif ( !isset( $set[2] ) ) {
1258                         # In XHTML, attributes must have a value.
1259                         # For 'reduced' form, return explicitly the attribute name here.
1260                         return $set[1];
1261                 } else {
1262                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1263                 }
1264         }
1265
1266         /**
1267          * @param string $text
1268          * @return string
1269          */
1270         private static function normalizeWhitespace( $text ) {
1271                 return preg_replace(
1272                         '/\r\n|[\x20\x0d\x0a\x09]/',
1273                         ' ',
1274                         $text );
1275         }
1276
1277         /**
1278          * Normalizes whitespace in a section name, such as might be returned
1279          * by Parser::stripSectionName(), for use in the id's that are used for
1280          * section links.
1281          *
1282          * @param string $section
1283          * @return string
1284          */
1285         static function normalizeSectionNameWhitespace( $section ) {
1286                 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1287         }
1288
1289         /**
1290          * Ensure that any entities and character references are legal
1291          * for XML and XHTML specifically. Any stray bits will be
1292          * &amp;-escaped to result in a valid text fragment.
1293          *
1294          * a. named char refs can only be &lt; &gt; &amp; &quot;, others are
1295          *   numericized (this way we're well-formed even without a DTD)
1296          * b. any numeric char refs must be legal chars, not invalid or forbidden
1297          * c. use lower cased "&#x", not "&#X"
1298          * d. fix or reject non-valid attributes
1299          *
1300          * @param string $text
1301          * @return string
1302          * @private
1303          */
1304         static function normalizeCharReferences( $text ) {
1305                 return preg_replace_callback(
1306                         self::CHAR_REFS_REGEX,
1307                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1308                         $text );
1309         }
1310
1311         /**
1312          * @param string $matches
1313          * @return string
1314          */
1315         static function normalizeCharReferencesCallback( $matches ) {
1316                 $ret = null;
1317                 if ( $matches[1] != '' ) {
1318                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1319                 } elseif ( $matches[2] != '' ) {
1320                         $ret = Sanitizer::decCharReference( $matches[2] );
1321                 } elseif ( $matches[3] != '' ) {
1322                         $ret = Sanitizer::hexCharReference( $matches[3] );
1323                 }
1324                 if ( is_null( $ret ) ) {
1325                         return htmlspecialchars( $matches[0] );
1326                 } else {
1327                         return $ret;
1328                 }
1329         }
1330
1331         /**
1332          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1333          * return the equivalent numeric entity reference (except for the core &lt;
1334          * &gt; &amp; &quot;). If the entity is a MediaWiki-specific alias, returns
1335          * the HTML equivalent. Otherwise, returns HTML-escaped text of
1336          * pseudo-entity source (eg &amp;foo;)
1337          *
1338          * @param string $name
1339          * @return string
1340          */
1341         static function normalizeEntity( $name ) {
1342                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1343                         return '&' . self::$htmlEntityAliases[$name] . ';';
1344                 } elseif ( in_array( $name, array( 'lt', 'gt', 'amp', 'quot' ) ) ) {
1345                         return "&$name;";
1346                 } elseif ( isset( self::$htmlEntities[$name] ) ) {
1347                         return '&#' . self::$htmlEntities[$name] . ';';
1348                 } else {
1349                         return "&amp;$name;";
1350                 }
1351         }
1352
1353         /**
1354          * @param int $codepoint
1355          * @return null|string
1356          */
1357         static function decCharReference( $codepoint ) {
1358                 $point = intval( $codepoint );
1359                 if ( Sanitizer::validateCodepoint( $point ) ) {
1360                         return sprintf( '&#%d;', $point );
1361                 } else {
1362                         return null;
1363                 }
1364         }
1365
1366         /**
1367          * @param int $codepoint
1368          * @return null|string
1369          */
1370         static function hexCharReference( $codepoint ) {
1371                 $point = hexdec( $codepoint );
1372                 if ( Sanitizer::validateCodepoint( $point ) ) {
1373                         return sprintf( '&#x%x;', $point );
1374                 } else {
1375                         return null;
1376                 }
1377         }
1378
1379         /**
1380          * Returns true if a given Unicode codepoint is a valid character in XML.
1381          * @param int $codepoint
1382          * @return bool
1383          */
1384         private static function validateCodepoint( $codepoint ) {
1385                 return $codepoint == 0x09
1386                         || $codepoint == 0x0a
1387                         || $codepoint == 0x0d
1388                         || ( $codepoint >= 0x20 && $codepoint <= 0xd7ff )
1389                         || ( $codepoint >= 0xe000 && $codepoint <= 0xfffd )
1390                         || ( $codepoint >= 0x10000 && $codepoint <= 0x10ffff );
1391         }
1392
1393         /**
1394          * Decode any character references, numeric or named entities,
1395          * in the text and return a UTF-8 string.
1396          *
1397          * @param string $text
1398          * @return string
1399          */
1400         public static function decodeCharReferences( $text ) {
1401                 return preg_replace_callback(
1402                         self::CHAR_REFS_REGEX,
1403                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1404                         $text );
1405         }
1406
1407         /**
1408          * Decode any character references, numeric or named entities,
1409          * in the next and normalize the resulting string. (bug 14952)
1410          *
1411          * This is useful for page titles, not for text to be displayed,
1412          * MediaWiki allows HTML entities to escape normalization as a feature.
1413          *
1414          * @param string $text Already normalized, containing entities
1415          * @return string Still normalized, without entities
1416          */
1417         public static function decodeCharReferencesAndNormalize( $text ) {
1418                 global $wgContLang;
1419                 $text = preg_replace_callback(
1420                         self::CHAR_REFS_REGEX,
1421                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1422                         $text, /* limit */ -1, $count );
1423
1424                 if ( $count ) {
1425                         return $wgContLang->normalize( $text );
1426                 } else {
1427                         return $text;
1428                 }
1429         }
1430
1431         /**
1432          * @param string $matches
1433          * @return string
1434          */
1435         static function decodeCharReferencesCallback( $matches ) {
1436                 if ( $matches[1] != '' ) {
1437                         return Sanitizer::decodeEntity( $matches[1] );
1438                 } elseif ( $matches[2] != '' ) {
1439                         return Sanitizer::decodeChar( intval( $matches[2] ) );
1440                 } elseif ( $matches[3] != '' ) {
1441                         return Sanitizer::decodeChar( hexdec( $matches[3] ) );
1442                 }
1443                 # Last case should be an ampersand by itself
1444                 return $matches[0];
1445         }
1446
1447         /**
1448          * Return UTF-8 string for a codepoint if that is a valid
1449          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1450          * @param int $codepoint
1451          * @return string
1452          * @private
1453          */
1454         static function decodeChar( $codepoint ) {
1455                 if ( Sanitizer::validateCodepoint( $codepoint ) ) {
1456                         return codepointToUtf8( $codepoint );
1457                 } else {
1458                         return UTF8_REPLACEMENT;
1459                 }
1460         }
1461
1462         /**
1463          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1464          * return the UTF-8 encoding of that character. Otherwise, returns
1465          * pseudo-entity source (eg "&foo;")
1466          *
1467          * @param string $name
1468          * @return string
1469          */
1470         static function decodeEntity( $name ) {
1471                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1472                         $name = self::$htmlEntityAliases[$name];
1473                 }
1474                 if ( isset( self::$htmlEntities[$name] ) ) {
1475                         return codepointToUtf8( self::$htmlEntities[$name] );
1476                 } else {
1477                         return "&$name;";
1478                 }
1479         }
1480
1481         /**
1482          * Fetch the whitelist of acceptable attributes for a given element name.
1483          *
1484          * @param string $element
1485          * @return array
1486          */
1487         static function attributeWhitelist( $element ) {
1488                 $list = Sanitizer::setupAttributeWhitelist();
1489                 return isset( $list[$element] )
1490                         ? $list[$element]
1491                         : array();
1492         }
1493
1494         /**
1495          * Foreach array key (an allowed HTML element), return an array
1496          * of allowed attributes
1497          * @return array
1498          */
1499         static function setupAttributeWhitelist() {
1500                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
1501                 static $whitelist, $staticInitialised;
1502
1503                 $globalContext = implode( '-', compact( 'wgAllowRdfaAttributes', 'wgAllowMicrodataAttributes' ) );
1504
1505                 if ( $whitelist !== null && $staticInitialised == $globalContext ) {
1506                         return $whitelist;
1507                 }
1508
1509                 $common = array(
1510                         # HTML
1511                         'id',
1512                         'class',
1513                         'style',
1514                         'lang',
1515                         'dir',
1516                         'title',
1517
1518                         # WAI-ARIA
1519                         'role',
1520                 );
1521
1522                 if ( $wgAllowRdfaAttributes ) {
1523                         # RDFa attributes as specified in section 9 of
1524                         # http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1525                         $common = array_merge( $common, array(
1526                                 'about', 'property', 'resource', 'datatype', 'typeof',
1527                         ) );
1528                 }
1529
1530                 if ( $wgAllowMicrodataAttributes ) {
1531                         # add HTML5 microdata tags as specified by
1532                         # http://www.whatwg.org/html/microdata.html#the-microdata-model
1533                         $common = array_merge( $common, array(
1534                                 'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1535                         ) );
1536                 }
1537
1538                 $block = array_merge( $common, array( 'align' ) );
1539                 $tablealign = array( 'align', 'valign' );
1540                 $tablecell = array(
1541                         'abbr',
1542                         'axis',
1543                         'headers',
1544                         'scope',
1545                         'rowspan',
1546                         'colspan',
1547                         'nowrap', # deprecated
1548                         'width', # deprecated
1549                         'height', # deprecated
1550                         'bgcolor', # deprecated
1551                 );
1552
1553                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1554                 # See: http://www.w3.org/TR/html4/
1555                 $whitelist = array(
1556                         # 7.5.4
1557                         'div'        => $block,
1558                         'center'     => $common, # deprecated
1559                         'span'       => $common,
1560
1561                         # 7.5.5
1562                         'h1'         => $block,
1563                         'h2'         => $block,
1564                         'h3'         => $block,
1565                         'h4'         => $block,
1566                         'h5'         => $block,
1567                         'h6'         => $block,
1568
1569                         # 7.5.6
1570                         # address
1571
1572                         # 8.2.4
1573                         'bdo'        => $common,
1574
1575                         # 9.2.1
1576                         'em'         => $common,
1577                         'strong'     => $common,
1578                         'cite'       => $common,
1579                         'dfn'        => $common,
1580                         'code'       => $common,
1581                         'samp'       => $common,
1582                         'kbd'        => $common,
1583                         'var'        => $common,
1584                         'abbr'       => $common,
1585                         # acronym
1586
1587                         # 9.2.2
1588                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1589                         'q'          => array_merge( $common, array( 'cite' ) ),
1590
1591                         # 9.2.3
1592                         'sub'        => $common,
1593                         'sup'        => $common,
1594
1595                         # 9.3.1
1596                         'p'          => $block,
1597
1598                         # 9.3.2
1599                         'br'         => array_merge( $common, array( 'clear' ) ),
1600
1601                         # http://www.whatwg.org/html/text-level-semantics.html#the-wbr-element
1602                         'wbr'        => $common,
1603
1604                         # 9.3.4
1605                         'pre'        => array_merge( $common, array( 'width' ) ),
1606
1607                         # 9.4
1608                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1609                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1610
1611                         # 10.2
1612                         'ul'         => array_merge( $common, array( 'type' ) ),
1613                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1614                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1615
1616                         # 10.3
1617                         'dl'         => $common,
1618                         'dd'         => $common,
1619                         'dt'         => $common,
1620
1621                         # 11.2.1
1622                         'table'      => array_merge( $common,
1623                                                                 array( 'summary', 'width', 'border', 'frame',
1624                                                                                 'rules', 'cellspacing', 'cellpadding',
1625                                                                                 'align', 'bgcolor',
1626                                                                 ) ),
1627
1628                         # 11.2.2
1629                         'caption'    => $block,
1630
1631                         # 11.2.3
1632                         'thead'      => $common,
1633                         'tfoot'      => $common,
1634                         'tbody'      => $common,
1635
1636                         # 11.2.4
1637                         'colgroup'   => array_merge( $common, array( 'span' ) ),
1638                         'col'        => array_merge( $common, array( 'span' ) ),
1639
1640                         # 11.2.5
1641                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1642
1643                         # 11.2.6
1644                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1645                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1646
1647                         # 12.2
1648                         # NOTE: <a> is not allowed directly, but the attrib
1649                         # whitelist is used from the Parser object
1650                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1651
1652                         # 13.2
1653                         # Not usually allowed, but may be used for extension-style hooks
1654                         # such as <math> when it is rasterized, or if $wgAllowImageTag is
1655                         # true
1656                         'img'        => array_merge( $common, array( 'alt', 'src', 'width', 'height' ) ),
1657
1658                         # 15.2.1
1659                         'tt'         => $common,
1660                         'b'          => $common,
1661                         'i'          => $common,
1662                         'big'        => $common,
1663                         'small'      => $common,
1664                         'strike'     => $common,
1665                         's'          => $common,
1666                         'u'          => $common,
1667
1668                         # 15.2.2
1669                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1670                         # basefont
1671
1672                         # 15.3
1673                         'hr'         => array_merge( $common, array( 'width' ) ),
1674
1675                         # HTML Ruby annotation text module, simple ruby only.
1676                         # http://www.whatwg.org/html/text-level-semantics.html#the-ruby-element
1677                         'ruby'       => $common,
1678                         # rbc
1679                         'rb'         => $common,
1680                         'rp'         => $common,
1681                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1682                         'rtc'         => $common,
1683
1684                         # MathML root element, where used for extensions
1685                         # 'title' may not be 100% valid here; it's XHTML
1686                         # http://www.w3.org/TR/REC-MathML/
1687                         'math'       => array( 'class', 'style', 'id', 'title' ),
1688
1689                         # HTML 5 section 4.6
1690                         'bdi' => $common,
1691
1692                         # HTML5 elements, defined by:
1693                         # http://www.whatwg.org/html/
1694                         'data' => array_merge( $common, array( 'value' ) ),
1695                         'time' => array_merge( $common, array( 'datetime' ) ),
1696                         'mark' => $common,
1697
1698                         // meta and link are only permitted by removeHTMLtags when Microdata
1699                         // is enabled so we don't bother adding a conditional to hide these
1700                         // Also meta and link are only valid in WikiText as Microdata elements
1701                         // (ie: validateTag rejects tags missing the attributes needed for Microdata)
1702                         // So we don't bother including $common attributes that have no purpose.
1703                         'meta' => array( 'itemprop', 'content' ),
1704                         'link' => array( 'itemprop', 'href' ),
1705                 );
1706
1707                 $staticInitialised = $globalContext;
1708
1709                 return $whitelist;
1710         }
1711
1712         /**
1713          * Take a fragment of (potentially invalid) HTML and return
1714          * a version with any tags removed, encoded as plain text.
1715          *
1716          * Warning: this return value must be further escaped for literal
1717          * inclusion in HTML output as of 1.10!
1718          *
1719          * @param string $text HTML fragment
1720          * @return string
1721          */
1722         static function stripAllTags( $text ) {
1723                 # Actual <tags>
1724                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1725
1726                 # Normalize &entities and whitespace
1727                 $text = self::decodeCharReferences( $text );
1728                 $text = self::normalizeWhitespace( $text );
1729
1730                 return $text;
1731         }
1732
1733         /**
1734          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1735          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1736          * PHP 5.1 doesn't.
1737          *
1738          * Use for passing XHTML fragments to PHP's XML parsing functions
1739          *
1740          * @return string
1741          */
1742         static function hackDocType() {
1743                 $out = "<!DOCTYPE html [\n";
1744                 foreach ( self::$htmlEntities as $entity => $codepoint ) {
1745                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1746                 }
1747                 $out .= "]>\n";
1748                 return $out;
1749         }
1750
1751         /**
1752          * @param string $url
1753          * @return mixed|string
1754          */
1755         static function cleanUrl( $url ) {
1756                 # Normalize any HTML entities in input. They will be
1757                 # re-escaped by makeExternalLink().
1758                 $url = Sanitizer::decodeCharReferences( $url );
1759
1760                 # Escape any control characters introduced by the above step
1761                 $url = preg_replace_callback( '/[\][<>"\\x00-\\x20\\x7F\|]/',
1762                         array( __CLASS__, 'cleanUrlCallback' ), $url );
1763
1764                 # Validate hostname portion
1765                 $matches = array();
1766                 if ( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1767                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1768
1769                         // Characters that will be ignored in IDNs.
1770                         // http://tools.ietf.org/html/3454#section-3.1
1771                         // Strip them before further processing so blacklists and such work.
1772                         $strip = "/
1773                                 \\s|          # general whitespace
1774                                 \xc2\xad|     # 00ad SOFT HYPHEN
1775                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1776                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1777                                 \xe2\x81\xa0| # 2060 WORD JOINER
1778                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1779                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1780                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1781                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1782                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1783                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1784                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1785                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe0f VARIATION SELECTOR-1-16
1786                                 /xuD";
1787
1788                         $host = preg_replace( $strip, '', $host );
1789
1790                         // @todo FIXME: Validate hostnames here
1791
1792                         return $protocol . $host . $rest;
1793                 } else {
1794                         return $url;
1795                 }
1796         }
1797
1798         /**
1799          * @param array $matches
1800          * @return string
1801          */
1802         static function cleanUrlCallback( $matches ) {
1803                 return urlencode( $matches[0] );
1804         }
1805
1806         /**
1807          * Does a string look like an e-mail address?
1808          *
1809          * This validates an email address using an HTML5 specification found at:
1810          * http://www.whatwg.org/html/states-of-the-type-attribute.html#valid-e-mail-address
1811          * Which as of 2011-01-24 says:
1812          *
1813          *   A valid e-mail address is a string that matches the ABNF production
1814          *   1*( atext / "." ) "@" ldh-str *( "." ldh-str ) where atext is defined
1815          *   in RFC 5322 section 3.2.3, and ldh-str is defined in RFC 1034 section
1816          *   3.5.
1817          *
1818          * This function is an implementation of the specification as requested in
1819          * bug 22449.
1820          *
1821          * Client-side forms will use the same standard validation rules via JS or
1822          * HTML 5 validation; additional restrictions can be enforced server-side
1823          * by extensions via the 'isValidEmailAddr' hook.
1824          *
1825          * Note that this validation doesn't 100% match RFC 2822, but is believed
1826          * to be liberal enough for wide use. Some invalid addresses will still
1827          * pass validation here.
1828          *
1829          * @since 1.18
1830          *
1831          * @param string $addr E-mail address
1832          * @return bool
1833          */
1834         public static function validateEmail( $addr ) {
1835                 $result = null;
1836                 if ( !Hooks::run( 'isValidEmailAddr', array( $addr, &$result ) ) ) {
1837                         return $result;
1838                 }
1839
1840                 // Please note strings below are enclosed in brackets [], this make the
1841                 // hyphen "-" a range indicator. Hence it is double backslashed below.
1842                 // See bug 26948
1843                 $rfc5322_atext = "a-z0-9!#$%&'*+\\-\/=?^_`{|}~";
1844                 $rfc1034_ldh_str = "a-z0-9\\-";
1845
1846                 $html5_email_regexp = "/
1847                 ^                      # start of string
1848                 [$rfc5322_atext\\.]+    # user part which is liberal :p
1849                 @                      # 'apostrophe'
1850                 [$rfc1034_ldh_str]+       # First domain part
1851                 (\\.[$rfc1034_ldh_str]+)*  # Following part prefixed with a dot
1852                 $                      # End of string
1853                 /ix"; // case Insensitive, eXtended
1854
1855                 return (bool)preg_match( $html5_email_regexp, $addr );
1856         }
1857 }