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 $attribsRegex;
 326
 327         /**
 328          * Regular expression to match HTML/XML attribute pairs within a tag.
 329          * Allows some... latitude.
 330          * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
 331          */
 332         static function getAttribsRegex() {
 333                 if ( self::$attribsRegex === null ) {
 334                         $attribFirst = '[:A-Z_a-z0-9]';
 335                         $attrib = '[:A-Z_a-z-.0-9]';
 336                         $space = '[\x09\x0a\x0d\x20]';
 337                         self::$attribsRegex =
 338                                 "/(?:^|$space)({$attribFirst}{$attrib}*)
 339                                   ($space*=$space*
 340                                         (?:
 341                                          # The attribute value: quoted or alone
 342                                           \"([^<\"]*)\"
 343                                          | '([^<']*)'
 344                                          |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
 345                                          |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
 346                                                                                  # colors are specified like this.
 347                                                                                  # We'll be normalizing it.
 348                                         )
 349                                 )?(?=$space|\$)/sx";
 350                 }
 351                 return self::$attribsRegex;
 352         }
 353
 354         /**
 355          * Cleans up HTML, removes dangerous tags and attributes, and
 356          * removes HTML comments
 357          * @private
 358          * @param string $text
 359          * @param callable $processCallback Callback to do any variable or parameter
 360          *   replacements in HTML attribute values
 361          * @param array|bool $args Arguments for the processing callback
 362          * @param array $extratags For any extra tags to include
 363          * @param array $removetags For any tags (default or extra) to exclude
 364          * @return string
 365          */
 366         static function removeHTMLtags( $text, $processCallback = null,
 367                 $args = array(), $extratags = array(), $removetags = array()
 368         ) {
 369                 global $wgUseTidy, $wgAllowMicrodataAttributes, $wgAllowImageTag;
 370
 371                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 372                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 373
 374                 wfProfileIn( __METHOD__ );
 375
 376                 // Base our staticInitialised variable off of the global config state so that if the globals
 377                 // are changed (like in the screwed up test system) we will re-initialise the settings.
 378                 $globalContext = implode( '-', compact( 'wgAllowMicrodataAttributes', 'wgAllowImageTag' ) );
 379                 if ( !$staticInitialised || $staticInitialised != $globalContext ) {
 380
 381                         $htmlpairsStatic = array( # Tags that must be closed
 382                                 'b', 'bdi', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 383                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 384                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 385                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 386                                 'ruby', '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] ) &&
 513                                                 !in_array( 'table', $tagstack ) ) {
 514                                                         $badtag = true;
 515                                                 } elseif ( in_array( $t, $tagstack ) &&
 516                                                 !isset( $htmlnest[$t] ) ) {
 517                                                         $badtag = true;
 518                                                 # Is it a self closed htmlpair ? (bug 5487)
 519                                                 } elseif ( $brace == '/>' &&
 520                                                 isset( $htmlpairs[$t] ) ) {
 521                                                         $badtag = true;
 522                                                 } elseif ( isset( $htmlsingleonly[$t] ) ) {
 523                                                         # Hack to force empty tag for unclosable elements
 524                                                         $brace = '/>';
 525                                                 } elseif ( isset( $htmlsingle[$t] ) ) {
 526                                                         # Hack to not close $htmlsingle tags
 527                                                         $brace = null;
 528                                                         # Still need to push this optionally-closed tag to
 529                                                         # the tag stack so that we can match end tags
 530                                                         # instead of marking them as bad.
 531                                                         array_push( $tagstack, $t );
 532                                                 } elseif ( isset( $tabletags[$t] )
 533                                                 && in_array( $t, $tagstack ) ) {
 534                                                         // New table tag but forgot to close the previous one
 535                                                         $text .= "</$t>";
 536                                                 } else {
 537                                                         if ( $t == 'table' ) {
 538                                                                 array_push( $tablestack, $tagstack );
 539                                                                 $tagstack = array();
 540                                                         }
 541                                                         array_push( $tagstack, $t );
 542                                                 }
 543
 544                                                 # Replace any variables or template parameters with
 545                                                 # plaintext results.
 546                                                 if ( is_callable( $processCallback ) ) {
 547                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 548                                                 }
 549
 550                                                 if ( !Sanitizer::validateTag( $params, $t ) ) {
 551                                                         $badtag = true;
 552                                                 }
 553
 554                                                 # Strip non-approved attributes from the tag
 555                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 556                                         }
 557                                         if ( !$badtag ) {
 558                                                 $rest = str_replace( '>', '&gt;', $rest );
 559                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 560                                                 $text .= "<$slash$t$newparams$close>$rest";
 561                                                 continue;
 562                                         }
 563                                 }
 564                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 565                         }
 566                         # Close off any remaining tags
 567                         while ( is_array( $tagstack ) && ( $t = array_pop( $tagstack ) ) ) {
 568                                 $text .= "</$t>\n";
 569                                 if ( $t == 'table' ) {
 570                                         $tagstack = array_pop( $tablestack );
 571                                 }
 572                         }
 573                 } else {
 574                         # this might be possible using tidy itself
 575                         foreach ( $bits as $x ) {
 576                                 preg_match(
 577                                         '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 578                                         $x,
 579                                         $regs
 580                                 );
 581
 582                                 wfSuppressWarnings();
 583                                 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 584                                 wfRestoreWarnings();
 585
 586                                 $badtag = false;
 587                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 588                                         if ( is_callable( $processCallback ) ) {
 589                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 590                                         }
 591
 592                                         if ( !Sanitizer::validateTag( $params, $t ) ) {
 593                                                 $badtag = true;
 594                                         }
 595
 596                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 597                                         if ( !$badtag ) {
 598                                                 $rest = str_replace( '>', '&gt;', $rest );
 599                                                 $text .= "<$slash$t$newparams$brace$rest";
 600                                                 continue;
 601                                         }
 602                                 }
 603                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x );
 604                         }
 605                 }
 606                 wfProfileOut( __METHOD__ );
 607                 return $text;
 608         }
 609
 610         /**
 611          * Remove '<!--', '-->', and everything between.
 612          * To avoid leaving blank lines, when a comment is both preceded
 613          * and followed by a newline (ignoring spaces), trim leading and
 614          * trailing spaces and one of the newlines.
 615          *
 616          * @private
 617          * @param string $text
 618          * @return string
 619          */
 620         static function removeHTMLcomments( $text ) {
 621                 wfProfileIn( __METHOD__ );
 622                 while ( ( $start = strpos( $text, '<!--' ) ) !== false ) {
 623                         $end = strpos( $text, '-->', $start + 4 );
 624                         if ( $end === false ) {
 625                                 # Unterminated comment; bail out
 626                                 break;
 627                         }
 628
 629                         $end += 3;
 630
 631                         # Trim space and newline if the comment is both
 632                         # preceded and followed by a newline
 633                         $spaceStart = max( $start - 1, 0 );
 634                         $spaceLen = $end - $spaceStart;
 635                         while ( substr( $text, $spaceStart, 1 ) === ' ' && $spaceStart > 0 ) {
 636                                 $spaceStart--;
 637                                 $spaceLen++;
 638                         }
 639                         while ( substr( $text, $spaceStart + $spaceLen, 1 ) === ' ' ) {
 640                                 $spaceLen++;
 641                         }
 642                         if ( substr( $text, $spaceStart, 1 ) === "\n"
 643                                 && substr( $text, $spaceStart + $spaceLen, 1 ) === "\n" ) {
 644                                 # Remove the comment, leading and trailing
 645                                 # spaces, and leave only one newline.
 646                                 $text = substr_replace( $text, "\n", $spaceStart, $spaceLen + 1 );
 647                         } else {
 648                                 # Remove just the comment.
 649                                 $text = substr_replace( $text, '', $start, $end - $start );
 650                         }
 651                 }
 652                 wfProfileOut( __METHOD__ );
 653                 return $text;
 654         }
 655
 656         /**
 657          * Takes attribute names and values for a tag and the tag name and
 658          * validates that the tag is allowed to be present.
 659          * This DOES NOT validate the attributes, nor does it validate the
 660          * tags themselves. This method only handles the special circumstances
 661          * where we may want to allow a tag within content but ONLY when it has
 662          * specific attributes set.
 663          *
 664          * @param string $params
 665          * @param string $element
 666          * @return bool
 667          */
 668         static function validateTag( $params, $element ) {
 669                 $params = Sanitizer::decodeTagAttributes( $params );
 670
 671                 if ( $element == 'meta' || $element == 'link' ) {
 672                         if ( !isset( $params['itemprop'] ) ) {
 673                                 // <meta> and <link> must have an itemprop="" otherwise they are not valid or safe in content
 674                                 return false;
 675                         }
 676                         if ( $element == 'meta' && !isset( $params['content'] ) ) {
 677                                 // <meta> must have a content="" for the itemprop
 678                                 return false;
 679                         }
 680                         if ( $element == 'link' && !isset( $params['href'] ) ) {
 681                                 // <link> must have an associated href=""
 682                                 return false;
 683                         }
 684                 }
 685
 686                 return true;
 687         }
 688
 689         /**
 690          * Take an array of attribute names and values and normalize or discard
 691          * illegal values for the given element type.
 692          *
 693          * - Discards attributes not on a whitelist for the given element
 694          * - Unsafe style attributes are discarded
 695          * - Invalid id attributes are re-encoded
 696          *
 697          * @param array $attribs
 698          * @param string $element
 699          * @return array
 700          *
 701          * @todo Check for legal values where the DTD limits things.
 702          * @todo Check for unique id attribute :P
 703          */
 704         static function validateTagAttributes( $attribs, $element ) {
 705                 return Sanitizer::validateAttributes( $attribs,
 706                         Sanitizer::attributeWhitelist( $element ) );
 707         }
 708
 709         /**
 710          * Take an array of attribute names and values and normalize or discard
 711          * illegal values for the given whitelist.
 712          *
 713          * - Discards attributes not the given whitelist
 714          * - Unsafe style attributes are discarded
 715          * - Invalid id attributes are re-encoded
 716          *
 717          * @param array $attribs
 718          * @param array $whitelist list of allowed attribute names
 719          * @return array
 720          *
 721          * @todo Check for legal values where the DTD limits things.
 722          * @todo Check for unique id attribute :P
 723          */
 724         static function validateAttributes( $attribs, $whitelist ) {
 725                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
 726
 727                 $whitelist = array_flip( $whitelist );
 728                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 729
 730                 $out = array();
 731                 foreach ( $attribs as $attribute => $value ) {
 732                         #allow XML namespace declaration if RDFa is enabled
 733                         if ( $wgAllowRdfaAttributes && preg_match( self::XMLNS_ATTRIBUTE_PATTERN, $attribute ) ) {
 734                                 if ( !preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 735                                         $out[$attribute] = $value;
 736                                 }
 737
 738                                 continue;
 739                         }
 740
 741                         # Allow any attribute beginning with "data-"
 742                         if ( !preg_match( '/^data-/i', $attribute ) && !isset( $whitelist[$attribute] ) ) {
 743                                 continue;
 744                         }
 745
 746                         # Strip javascript "expression" from stylesheets.
 747                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 748                         if ( $attribute == 'style' ) {
 749                                 $value = Sanitizer::checkCss( $value );
 750                         }
 751
 752                         if ( $attribute === 'id' ) {
 753                                 $value = Sanitizer::escapeId( $value, 'noninitial' );
 754                         }
 755
 756                         # WAI-ARIA
 757                         # http://www.w3.org/TR/wai-aria/
 758                         # http://www.whatwg.org/html/elements.html#wai-aria
 759                         # For now we only support role="presentation" until we work out what roles should be
 760                         # usable by content and we ensure that our code explicitly rejects patterns that
 761                         # violate HTML5's ARIA restrictions.
 762                         if ( $attribute === 'role' && $value !== 'presentation' ) {
 763                                 continue;
 764                         }
 765
 766                         // RDFa and microdata properties allow URLs, URIs and/or CURIs.
 767                         // Check them for sanity.
 768                         if ( $attribute === 'rel' || $attribute === 'rev'
 769                                 # RDFa
 770                                 || $attribute === 'about' || $attribute === 'property'
 771                                 || $attribute === 'resource' || $attribute === 'datatype'
 772                                 || $attribute === 'typeof'
 773                                 # HTML5 microdata
 774                                 || $attribute === 'itemid' || $attribute === 'itemprop'
 775                                 || $attribute === 'itemref' || $attribute === 'itemscope'
 776                                 || $attribute === 'itemtype'
 777                         ) {
 778                                 //Paranoia. Allow "simple" values but suppress javascript
 779                                 if ( preg_match( self::EVIL_URI_PATTERN, $value ) ) {
 780                                         continue;
 781                                 }
 782                         }
 783
 784                         # NOTE: even though elements using href/src are not allowed directly, supply
 785                         #       validation code that can be used by tag hook handlers, etc
 786                         if ( $attribute === 'href' || $attribute === 'src' ) {
 787                                 if ( !preg_match( $hrefExp, $value ) ) {
 788                                         continue; //drop any href or src attributes not using an allowed protocol.
 789                                         // NOTE: this also drops all relative URLs
 790                                 }
 791                         }
 792
 793                         // If this attribute was previously set, override it.
 794                         // Output should only have one attribute of each name.
 795                         $out[$attribute] = $value;
 796                 }
 797
 798                 if ( $wgAllowMicrodataAttributes ) {
 799                         # itemtype, itemid, itemref don't make sense without itemscope
 800                         if ( !array_key_exists( 'itemscope', $out ) ) {
 801                                 unset( $out['itemtype'] );
 802                                 unset( $out['itemid'] );
 803                                 unset( $out['itemref'] );
 804                         }
 805                         # TODO: Strip itemprop if we aren't descendants of an itemscope or pointed to by an itemref.
 806                 }
 807                 return $out;
 808         }
 809
 810         /**
 811          * Merge two sets of HTML attributes.  Conflicting items in the second set
 812          * will override those in the first, except for 'class' attributes which
 813          * will be combined (if they're both strings).
 814          *
 815          * @todo implement merging for other attributes such as style
 816          * @param array $a
 817          * @param array $b
 818          * @return array
 819          */
 820         static function mergeAttributes( $a, $b ) {
 821                 $out = array_merge( $a, $b );
 822                 if ( isset( $a['class'] ) && isset( $b['class'] )
 823                         && is_string( $a['class'] ) && is_string( $b['class'] )
 824                         && $a['class'] !== $b['class']
 825                 ) {
 826                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 827                                 -1, PREG_SPLIT_NO_EMPTY );
 828                         $out['class'] = implode( ' ', array_unique( $classes ) );
 829                 }
 830                 return $out;
 831         }
 832
 833         /**
 834          * Pick apart some CSS and check it for forbidden or unsafe structures.
 835          * Returns a sanitized string. This sanitized string will have
 836          * character references and escape sequences decoded and comments
 837          * stripped (unless it is itself one valid comment, in which case the value
 838          * will be passed through). If the input is just too evil, only a comment
 839          * complaining about evilness will be returned.
 840          *
 841          * Currently URL references, 'expression', 'tps' are forbidden.
 842          *
 843          * NOTE: Despite the fact that character references are decoded, the
 844          * returned string may contain character references given certain
 845          * clever input strings. These character references must
 846          * be escaped before the return value is embedded in HTML.
 847          *
 848          * @param string $value
 849          * @return string
 850          */
 851         static function checkCss( $value ) {
 852                 global $wgDisallowedCss;
 853
 854                 // Decode character references like &#123;
 855                 $value = Sanitizer::decodeCharReferences( $value );
 856
 857                 // Decode escape sequences and line continuation
 858                 // See the grammar in the CSS 2 spec, appendix D.
 859                 // This has to be done AFTER decoding character references.
 860                 // This means it isn't possible for this function to return
 861                 // unsanitized escape sequences. It is possible to manufacture
 862                 // input that contains character references that decode to
 863                 // escape sequences that decode to character references, but
 864                 // it's OK for the return value to contain character references
 865                 // because the caller is supposed to escape those anyway.
 866                 static $decodeRegex;
 867                 if ( !$decodeRegex ) {
 868                         $space = '[\\x20\\t\\r\\n\\f]';
 869                         $nl = '(?:\\n|\\r\\n|\\r|\\f)';
 870                         $backslash = '\\\\';
 871                         $decodeRegex = "/ $backslash
 872                                 (?:
 873                                         ($nl) |  # 1. Line continuation
 874                                         ([0-9A-Fa-f]{1,6})$space? |  # 2. character number
 875                                         (.) | # 3. backslash cancelling special meaning
 876                                         () | # 4. backslash at end of string
 877                                 )/xu";
 878                 }
 879                 $value = preg_replace_callback( $decodeRegex,
 880                         array( __CLASS__, 'cssDecodeCallback' ), $value );
 881
 882                 // Normalize Halfwidth and Fullwidth Unicode block that IE6 might treat as ascii
 883                 $value = preg_replace_callback(
 884                         '/[！-［］-ｚ]/u', // U+FF01 to U+FF5A, excluding U+FF3C (bug 58088)
 885                         function ( $matches ) {
 886                                 $cp = utf8ToCodepoint( $matches[0] );
 887                                 if ( $cp === false ) {
 888                                         return '';
 889                                 }
 890                                 return chr( $cp - 65248 ); // ASCII range \x21-\x7A
 891                         },
 892                         $value
 893                 );
 894
 895                 // Convert more characters IE6 might treat as ascii
 896                 // U+0280, U+0274, U+207F, U+029F, U+026A, U+207D, U+208D
 897                 $value = str_replace(
 898                         array( 'ʀ', 'ɴ', 'ⁿ', 'ʟ', 'ɪ', '⁽', '₍' ),
 899                         array( 'r', 'n', 'n', 'l', 'i', '(', '(' ),
 900                         $value
 901                 );
 902
 903                 // Let the value through if it's nothing but a single comment, to
 904                 // allow other functions which may reject it to pass some error
 905                 // message through.
 906                 if ( !preg_match( '! ^ \s* /\* [^*\\/]* \*/ \s* $ !x', $value ) ) {
 907                         // Remove any comments; IE gets token splitting wrong
 908                         // This must be done AFTER decoding character references and
 909                         // escape sequences, because those steps can introduce comments
 910                         // This step cannot introduce character references or escape
 911                         // sequences, because it replaces comments with spaces rather
 912                         // than removing them completely.
 913                         $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
 914
 915                         // Remove anything after a comment-start token, to guard against
 916                         // incorrect client implementations.
 917                         $commentPos = strpos( $value, '/*' );
 918                         if ( $commentPos !== false ) {
 919                                 $value = substr( $value, 0, $commentPos );
 920                         }
 921                 }
 922
 923                 // S followed by repeat, iteration, or prolonged sound marks,
 924                 // which IE will treat as "ss"
 925                 $value = preg_replace(
 926                         '/s(?:
 927                                 \xE3\x80\xB1 | # U+3031
 928                                 \xE3\x82\x9D | # U+309D
 929                                 \xE3\x83\xBC | # U+30FC
 930                                 \xE3\x83\xBD | # U+30FD
 931                                 \xEF\xB9\xBC | # U+FE7C
 932                                 \xEF\xB9\xBD | # U+FE7D
 933                                 \xEF\xBD\xB0   # U+FF70
 934                         )/ix',
 935                         'ss',
 936                         $value
 937                 );
 938
 939                 // Reject problematic keywords and control characters
 940                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
 941                         return '/* invalid control char */';
 942                 } else {
 943                         if ( $wgDisallowedCss ) {
 944                                 if ( preg_match( $wgDisallowedCss, $value ) ) {
 945                                         return '/* insecure input */';
 946                                 }
 947                         }
 948                 }
 949                 return $value;
 950         }
 951
 952         /**
 953          * @param array $matches
 954          * @return string
 955          */
 956         static function cssDecodeCallback( $matches ) {
 957                 if ( $matches[1] !== '' ) {
 958                         // Line continuation
 959                         return '';
 960                 } elseif ( $matches[2] !== '' ) {
 961                         $char = codepointToUtf8( hexdec( $matches[2] ) );
 962                 } elseif ( $matches[3] !== '' ) {
 963                         $char = $matches[3];
 964                 } else {
 965                         $char = '\\';
 966                 }
 967                 if ( $char == "\n" || $char == '"' || $char == "'" || $char == '\\' ) {
 968                         // These characters need to be escaped in strings
 969                         // Clean up the escape sequence to avoid parsing errors by clients
 970                         return '\\' . dechex( ord( $char ) ) . ' ';
 971                 } else {
 972                         // Decode unnecessary escape
 973                         return $char;
 974                 }
 975         }
 976
 977         /**
 978          * Take a tag soup fragment listing an HTML element's attributes
 979          * and normalize it to well-formed XML, discarding unwanted attributes.
 980          * Output is safe for further wikitext processing, with escaping of
 981          * values that could trigger problems.
 982          *
 983          * - Normalizes attribute names to lowercase
 984          * - Discards attributes not on a whitelist for the given element
 985          * - Turns broken or invalid entities into plaintext
 986          * - Double-quotes all attribute values
 987          * - Attributes without values are given the name as attribute
 988          * - Double attributes are discarded
 989          * - Unsafe style attributes are discarded
 990          * - Prepends space if there are attributes.
 991          *
 992          * @param string $text
 993          * @param string $element
 994          * @return string
 995          */
 996         static function fixTagAttributes( $text, $element ) {
 997                 if ( trim( $text ) == '' ) {
 998                         return '';
 999                 }
1000
1001                 $decoded = Sanitizer::decodeTagAttributes( $text );
1002                 $stripped = Sanitizer::validateTagAttributes( $decoded, $element );
1003
1004                 return Sanitizer::safeEncodeTagAttributes( $stripped );
1005         }
1006
1007         /**
1008          * Encode an attribute value for HTML output.
1009          * @param string $text
1010          * @return string HTML-encoded text fragment
1011          */
1012         static function encodeAttribute( $text ) {
1013                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
1014
1015                 // Whitespace is normalized during attribute decoding,
1016                 // so if we've been passed non-spaces we must encode them
1017                 // ahead of time or they won't be preserved.
1018                 $encValue = strtr( $encValue, array(
1019                         "\n" => '&#10;',
1020                         "\r" => '&#13;',
1021                         "\t" => '&#9;',
1022                 ) );
1023
1024                 return $encValue;
1025         }
1026
1027         /**
1028          * Encode an attribute value for HTML tags, with extra armoring
1029          * against further wiki processing.
1030          * @param string $text
1031          * @return string HTML-encoded text fragment
1032          */
1033         static function safeEncodeAttribute( $text ) {
1034                 $encValue = Sanitizer::encodeAttribute( $text );
1035
1036                 # Templates and links may be expanded in later parsing,
1037                 # creating invalid or dangerous output. Suppress this.
1038                 $encValue = strtr( $encValue, array(
1039                         '<'    => '&lt;',   // This should never happen,
1040                         '>'    => '&gt;',   // we've received invalid input
1041                         '"'    => '&quot;', // which should have been escaped.
1042                         '{'    => '&#123;',
1043                         '['    => '&#91;',
1044                         "''"   => '&#39;&#39;',
1045                         'ISBN' => '&#73;SBN',
1046                         'RFC'  => '&#82;FC',
1047                         'PMID' => '&#80;MID',
1048                         '|'    => '&#124;',
1049                         '__'   => '&#95;_',
1050                 ) );
1051
1052                 # Stupid hack
1053                 $encValue = preg_replace_callback(
1054                         '/((?i)' . wfUrlProtocols() . ')/',
1055                         array( 'Sanitizer', 'armorLinksCallback' ),
1056                         $encValue );
1057                 return $encValue;
1058         }
1059
1060         /**
1061          * Given a value, escape it so that it can be used in an id attribute and
1062          * return it.  This will use HTML5 validation if $wgExperimentalHtmlIds is
1063          * true, allowing anything but ASCII whitespace.  Otherwise it will use
1064          * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
1065          * escaped with lots of dots.
1066          *
1067          * To ensure we don't have to bother escaping anything, we also strip ', ",
1068          * & even if $wgExperimentalIds is true.  TODO: Is this the best tactic?
1069          * We also strip # because it upsets IE, and % because it could be
1070          * ambiguous if it's part of something that looks like a percent escape
1071          * (which don't work reliably in fragments cross-browser).
1072          *
1073          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
1074          *   in the id and name attributes
1075          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with
1076          *   the id attribute
1077          * @see http://www.whatwg.org/html/elements.html#the-id-attribute
1078          *   HTML5 definition of id attribute
1079          *
1080          * @param string $id id to escape
1081          * @param string|array $options String or array of strings (default is array()):
1082          *   'noninitial': This is a non-initial fragment of an id, not a full id,
1083          *       so don't pay attention if the first character isn't valid at the
1084          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
1085          *       false.
1086          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
1087          *       if $wgExperimentalHtmlIds is used, so we can generate extra
1088          *       anchors and links won't break.
1089          * @return string
1090          */
1091         static function escapeId( $id, $options = array() ) {
1092                 global $wgExperimentalHtmlIds;
1093                 $options = (array)$options;
1094
1095                 if ( $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
1096                         $id = Sanitizer::decodeCharReferences( $id );
1097                         $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
1098                         $id = trim( $id, '_' );
1099                         if ( $id === '' ) {
1100                                 # Must have been all whitespace to start with.
1101                                 return '_';
1102                         } else {
1103                                 return $id;
1104                         }
1105                 }
1106
1107                 # HTML4-style escaping
1108                 static $replace = array(
1109                         '%3A' => ':',
1110                         '%' => '.'
1111                 );
1112
1113                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
1114                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
1115
1116                 if ( !preg_match( '/^[a-zA-Z]/', $id )
1117                 && !in_array( 'noninitial', $options ) ) {
1118                         // Initial character must be a letter!
1119                         $id = "x$id";
1120                 }
1121                 return $id;
1122         }
1123
1124         /**
1125          * Given a value, escape it so that it can be used as a CSS class and
1126          * return it.
1127          *
1128          * @todo For extra validity, input should be validated UTF-8.
1129          *
1130          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
1131          *
1132          * @param string $class
1133          * @return string
1134          */
1135         static function escapeClass( $class ) {
1136                 // Convert ugly stuff to underscores and kill underscores in ugly places
1137                 return rtrim( preg_replace(
1138                         array( '/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/', '/_+/' ),
1139                         '_',
1140                         $class ), '_' );
1141         }
1142
1143         /**
1144          * Given HTML input, escape with htmlspecialchars but un-escape entities.
1145          * This allows (generally harmless) entities like &#160; to survive.
1146          *
1147          * @param string $html HTML to escape
1148          * @return string Escaped input
1149          */
1150         static function escapeHtmlAllowEntities( $html ) {
1151                 $html = Sanitizer::decodeCharReferences( $html );
1152                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
1153                 # hurt.
1154                 $html = htmlspecialchars( $html, ENT_QUOTES );
1155                 return $html;
1156         }
1157
1158         /**
1159          * Regex replace callback for armoring links against further processing.
1160          * @param array $matches
1161          * @return string
1162          */
1163         private static function armorLinksCallback( $matches ) {
1164                 return str_replace( ':', '&#58;', $matches[1] );
1165         }
1166
1167         /**
1168          * Return an associative array of attribute names and values from
1169          * a partial tag string. Attribute names are forces to lowercase,
1170          * character references are decoded to UTF-8 text.
1171          *
1172          * @param string $text
1173          * @return array
1174          */
1175         public static function decodeTagAttributes( $text ) {
1176                 if ( trim( $text ) == '' ) {
1177                         return array();
1178                 }
1179
1180                 $attribs = array();
1181                 $pairs = array();
1182                 if ( !preg_match_all(
1183                         self::getAttribsRegex(),
1184                         $text,
1185                         $pairs,
1186                         PREG_SET_ORDER ) ) {
1187                         return $attribs;
1188                 }
1189
1190                 foreach ( $pairs as $set ) {
1191                         $attribute = strtolower( $set[1] );
1192                         $value = Sanitizer::getTagAttributeCallback( $set );
1193
1194                         // Normalize whitespace
1195                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1196                         $value = trim( $value );
1197
1198                         // Decode character references
1199                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
1200                 }
1201                 return $attribs;
1202         }
1203
1204         /**
1205          * Build a partial tag string from an associative array of attribute
1206          * names and values as returned by decodeTagAttributes.
1207          *
1208          * @param array $assoc_array
1209          * @return string
1210          */
1211         public static function safeEncodeTagAttributes( $assoc_array ) {
1212                 $attribs = array();
1213                 foreach ( $assoc_array as $attribute => $value ) {
1214                         $encAttribute = htmlspecialchars( $attribute );
1215                         $encValue = Sanitizer::safeEncodeAttribute( $value );
1216
1217                         $attribs[] = "$encAttribute=\"$encValue\"";
1218                 }
1219                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
1220         }
1221
1222         /**
1223          * Pick the appropriate attribute value from a match set from the
1224          * attribs regex matches.
1225          *
1226          * @param array $set
1227          * @throws MWException when tag conditions are not met.
1228          * @return string
1229          */
1230         private static function getTagAttributeCallback( $set ) {
1231                 if ( isset( $set[6] ) ) {
1232                         # Illegal #XXXXXX color with no quotes.
1233                         return $set[6];
1234                 } elseif ( isset( $set[5] ) ) {
1235                         # No quotes.
1236                         return $set[5];
1237                 } elseif ( isset( $set[4] ) ) {
1238                         # Single-quoted
1239                         return $set[4];
1240                 } elseif ( isset( $set[3] ) ) {
1241                         # Double-quoted
1242                         return $set[3];
1243                 } elseif ( !isset( $set[2] ) ) {
1244                         # In XHTML, attributes must have a value.
1245                         # For 'reduced' form, return explicitly the attribute name here.
1246                         return $set[1];
1247                 } else {
1248                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1249                 }
1250         }
1251
1252         /**
1253          * Normalize whitespace and character references in an XML source-
1254          * encoded text for an attribute value.
1255          *
1256          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
1257          * but note that we're not returning the value, but are returning
1258          * XML source fragments that will be slapped into output.
1259          *
1260          * @param string $text
1261          * @return string
1262          * @todo Remove, unused?
1263          */
1264         private static function normalizeAttributeValue( $text ) {
1265                 return str_replace( '"', '&quot;',
1266                         self::normalizeWhitespace(
1267                                 Sanitizer::normalizeCharReferences( $text ) ) );
1268         }
1269
1270         /**
1271          * @param string $text
1272          * @return string
1273          */
1274         private static function normalizeWhitespace( $text ) {
1275                 return preg_replace(
1276                         '/\r\n|[\x20\x0d\x0a\x09]/',
1277                         ' ',
1278                         $text );
1279         }
1280
1281         /**
1282          * Normalizes whitespace in a section name, such as might be returned
1283          * by Parser::stripSectionName(), for use in the id's that are used for
1284          * section links.
1285          *
1286          * @param string $section
1287          * @return string
1288          */
1289         static function normalizeSectionNameWhitespace( $section ) {
1290                 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1291         }
1292
1293         /**
1294          * Ensure that any entities and character references are legal
1295          * for XML and XHTML specifically. Any stray bits will be
1296          * &amp;-escaped to result in a valid text fragment.
1297          *
1298          * a. named char refs can only be &lt; &gt; &amp; &quot;, others are
1299          *   numericized (this way we're well-formed even without a DTD)
1300          * b. any numeric char refs must be legal chars, not invalid or forbidden
1301          * c. use lower cased "&#x", not "&#X"
1302          * d. fix or reject non-valid attributes
1303          *
1304          * @param string $text
1305          * @return string
1306          * @private
1307          */
1308         static function normalizeCharReferences( $text ) {
1309                 return preg_replace_callback(
1310                         self::CHAR_REFS_REGEX,
1311                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1312                         $text );
1313         }
1314
1315         /**
1316          * @param string $matches
1317          * @return string
1318          */
1319         static function normalizeCharReferencesCallback( $matches ) {
1320                 $ret = null;
1321                 if ( $matches[1] != '' ) {
1322                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1323                 } elseif ( $matches[2] != '' ) {
1324                         $ret = Sanitizer::decCharReference( $matches[2] );
1325                 } elseif ( $matches[3] != '' ) {
1326                         $ret = Sanitizer::hexCharReference( $matches[3] );
1327                 }
1328                 if ( is_null( $ret ) ) {
1329                         return htmlspecialchars( $matches[0] );
1330                 } else {
1331                         return $ret;
1332                 }
1333         }
1334
1335         /**
1336          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1337          * return the equivalent numeric entity reference (except for the core &lt;
1338          * &gt; &amp; &quot;). If the entity is a MediaWiki-specific alias, returns
1339          * the HTML equivalent. Otherwise, returns HTML-escaped text of
1340          * pseudo-entity source (eg &amp;foo;)
1341          *
1342          * @param string $name
1343          * @return string
1344          */
1345         static function normalizeEntity( $name ) {
1346                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1347                         return '&' . self::$htmlEntityAliases[$name] . ';';
1348                 } elseif ( in_array( $name,
1349                 array( 'lt', 'gt', 'amp', 'quot' ) ) ) {
1350                         return "&$name;";
1351                 } elseif ( isset( self::$htmlEntities[$name] ) ) {
1352                         return '&#' . self::$htmlEntities[$name] . ';';
1353                 } else {
1354                         return "&amp;$name;";
1355                 }
1356         }
1357
1358         /**
1359          * @param int $codepoint
1360          * @return null|string
1361          */
1362         static function decCharReference( $codepoint ) {
1363                 $point = intval( $codepoint );
1364                 if ( Sanitizer::validateCodepoint( $point ) ) {
1365                         return sprintf( '&#%d;', $point );
1366                 } else {
1367                         return null;
1368                 }
1369         }
1370
1371         /**
1372          * @param int $codepoint
1373          * @return null|string
1374          */
1375         static function hexCharReference( $codepoint ) {
1376                 $point = hexdec( $codepoint );
1377                 if ( Sanitizer::validateCodepoint( $point ) ) {
1378                         return sprintf( '&#x%x;', $point );
1379                 } else {
1380                         return null;
1381                 }
1382         }
1383
1384         /**
1385          * Returns true if a given Unicode codepoint is a valid character in XML.
1386          * @param int $codepoint
1387          * @return bool
1388          */
1389         private static function validateCodepoint( $codepoint ) {
1390                 return $codepoint == 0x09
1391                         || $codepoint == 0x0a
1392                         || $codepoint == 0x0d
1393                         || ( $codepoint >= 0x20 && $codepoint <= 0xd7ff )
1394                         || ( $codepoint >= 0xe000 && $codepoint <= 0xfffd )
1395                         || ( $codepoint >= 0x10000 && $codepoint <= 0x10ffff );
1396         }
1397
1398         /**
1399          * Decode any character references, numeric or named entities,
1400          * in the text and return a UTF-8 string.
1401          *
1402          * @param string $text
1403          * @return string
1404          */
1405         public static function decodeCharReferences( $text ) {
1406                 return preg_replace_callback(
1407                         self::CHAR_REFS_REGEX,
1408                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1409                         $text );
1410         }
1411
1412         /**
1413          * Decode any character references, numeric or named entities,
1414          * in the next and normalize the resulting string. (bug 14952)
1415          *
1416          * This is useful for page titles, not for text to be displayed,
1417          * MediaWiki allows HTML entities to escape normalization as a feature.
1418          *
1419          * @param string $text Already normalized, containing entities
1420          * @return string Still normalized, without entities
1421          */
1422         public static function decodeCharReferencesAndNormalize( $text ) {
1423                 global $wgContLang;
1424                 $text = preg_replace_callback(
1425                         self::CHAR_REFS_REGEX,
1426                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1427                         $text, /* limit */ -1, $count );
1428
1429                 if ( $count ) {
1430                         return $wgContLang->normalize( $text );
1431                 } else {
1432                         return $text;
1433                 }
1434         }
1435
1436         /**
1437          * @param string $matches
1438          * @return string
1439          */
1440         static function decodeCharReferencesCallback( $matches ) {
1441                 if ( $matches[1] != '' ) {
1442                         return Sanitizer::decodeEntity( $matches[1] );
1443                 } elseif ( $matches[2] != '' ) {
1444                         return Sanitizer::decodeChar( intval( $matches[2] ) );
1445                 } elseif ( $matches[3] != '' ) {
1446                         return Sanitizer::decodeChar( hexdec( $matches[3] ) );
1447                 }
1448                 # Last case should be an ampersand by itself
1449                 return $matches[0];
1450         }
1451
1452         /**
1453          * Return UTF-8 string for a codepoint if that is a valid
1454          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1455          * @param int $codepoint
1456          * @return string
1457          * @private
1458          */
1459         static function decodeChar( $codepoint ) {
1460                 if ( Sanitizer::validateCodepoint( $codepoint ) ) {
1461                         return codepointToUtf8( $codepoint );
1462                 } else {
1463                         return UTF8_REPLACEMENT;
1464                 }
1465         }
1466
1467         /**
1468          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1469          * return the UTF-8 encoding of that character. Otherwise, returns
1470          * pseudo-entity source (eg "&foo;")
1471          *
1472          * @param string $name
1473          * @return string
1474          */
1475         static function decodeEntity( $name ) {
1476                 if ( isset( self::$htmlEntityAliases[$name] ) ) {
1477                         $name = self::$htmlEntityAliases[$name];
1478                 }
1479                 if ( isset( self::$htmlEntities[$name] ) ) {
1480                         return codepointToUtf8( self::$htmlEntities[$name] );
1481                 } else {
1482                         return "&$name;";
1483                 }
1484         }
1485
1486         /**
1487          * Fetch the whitelist of acceptable attributes for a given element name.
1488          *
1489          * @param string $element
1490          * @return array
1491          */
1492         static function attributeWhitelist( $element ) {
1493                 $list = Sanitizer::setupAttributeWhitelist();
1494                 return isset( $list[$element] )
1495                         ? $list[$element]
1496                         : array();
1497         }
1498
1499         /**
1500          * Foreach array key (an allowed HTML element), return an array
1501          * of allowed attributes
1502          * @return array
1503          */
1504         static function setupAttributeWhitelist() {
1505                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
1506                 static $whitelist, $staticInitialised;
1507
1508                 $globalContext = implode( '-', compact( 'wgAllowRdfaAttributes', 'wgAllowMicrodataAttributes' ) );
1509
1510                 if ( $whitelist !== null && $staticInitialised == $globalContext ) {
1511                         return $whitelist;
1512                 }
1513
1514                 $common = array(
1515                         # HTML
1516                         'id',
1517                         'class',
1518                         'style',
1519                         'lang',
1520                         'dir',
1521                         'title',
1522
1523                         # WAI-ARIA
1524                         'role',
1525                 );
1526
1527                 if ( $wgAllowRdfaAttributes ) {
1528                         # RDFa attributes as specified in section 9 of
1529                         # http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1530                         $common = array_merge( $common, array(
1531                                 'about', 'property', 'resource', 'datatype', 'typeof',
1532                         ) );
1533                 }
1534
1535                 if ( $wgAllowMicrodataAttributes ) {
1536                         # add HTML5 microdata tags as specified by
1537                         # http://www.whatwg.org/html/microdata.html#the-microdata-model
1538                         $common = array_merge( $common, array(
1539                                 'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1540                         ) );
1541                 }
1542
1543                 $block = array_merge( $common, array( 'align' ) );
1544                 $tablealign = array( 'align', 'valign' );
1545                 $tablecell = array(
1546                         'abbr',
1547                         'axis',
1548                         'headers',
1549                         'scope',
1550                         'rowspan',
1551                         'colspan',
1552                         'nowrap', # deprecated
1553                         'width', # deprecated
1554                         'height', # deprecated
1555                         'bgcolor', # deprecated
1556                 );
1557
1558                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1559                 # See: http://www.w3.org/TR/html4/
1560                 $whitelist = array(
1561                         # 7.5.4
1562                         'div'        => $block,
1563                         'center'     => $common, # deprecated
1564                         'span'       => $common,
1565
1566                         # 7.5.5
1567                         'h1'         => $block,
1568                         'h2'         => $block,
1569                         'h3'         => $block,
1570                         'h4'         => $block,
1571                         'h5'         => $block,
1572                         'h6'         => $block,
1573
1574                         # 7.5.6
1575                         # address
1576
1577                         # 8.2.4
1578                         'bdo'        => $common,
1579
1580                         # 9.2.1
1581                         'em'         => $common,
1582                         'strong'     => $common,
1583                         'cite'       => $common,
1584                         'dfn'        => $common,
1585                         'code'       => $common,
1586                         'samp'       => $common,
1587                         'kbd'        => $common,
1588                         'var'        => $common,
1589                         'abbr'       => $common,
1590                         # acronym
1591
1592                         # 9.2.2
1593                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1594                         'q'          => array_merge( $common, array( 'cite' ) ),
1595
1596                         # 9.2.3
1597                         'sub'        => $common,
1598                         'sup'        => $common,
1599
1600                         # 9.3.1
1601                         'p'          => $block,
1602
1603                         # 9.3.2
1604                         'br'         => array_merge( $common, array( 'clear' ) ),
1605
1606                         # http://www.whatwg.org/html/text-level-semantics.html#the-wbr-element
1607                         'wbr'        => $common,
1608
1609                         # 9.3.4
1610                         'pre'        => array_merge( $common, array( 'width' ) ),
1611
1612                         # 9.4
1613                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1614                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1615
1616                         # 10.2
1617                         'ul'         => array_merge( $common, array( 'type' ) ),
1618                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1619                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1620
1621                         # 10.3
1622                         'dl'         => $common,
1623                         'dd'         => $common,
1624                         'dt'         => $common,
1625
1626                         # 11.2.1
1627                         'table'      => array_merge( $common,
1628                                                                 array( 'summary', 'width', 'border', 'frame',
1629                                                                                 'rules', 'cellspacing', 'cellpadding',
1630                                                                                 'align', 'bgcolor',
1631                                                                 ) ),
1632
1633                         # 11.2.2
1634                         'caption'    => $block,
1635
1636                         # 11.2.3
1637                         'thead'      => $common,
1638                         'tfoot'      => $common,
1639                         'tbody'      => $common,
1640
1641                         # 11.2.4
1642                         'colgroup'   => array_merge( $common, array( 'span' ) ),
1643                         'col'        => array_merge( $common, array( 'span' ) ),
1644
1645                         # 11.2.5
1646                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1647
1648                         # 11.2.6
1649                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1650                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1651
1652                         # 12.2
1653                         # NOTE: <a> is not allowed directly, but the attrib
1654                         # whitelist is used from the Parser object
1655                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1656
1657                         # 13.2
1658                         # Not usually allowed, but may be used for extension-style hooks
1659                         # such as <math> when it is rasterized, or if $wgAllowImageTag is
1660                         # true
1661                         'img'        => array_merge( $common, array( 'alt', 'src', 'width', 'height' ) ),
1662
1663                         # 15.2.1
1664                         'tt'         => $common,
1665                         'b'          => $common,
1666                         'i'          => $common,
1667                         'big'        => $common,
1668                         'small'      => $common,
1669                         'strike'     => $common,
1670                         's'          => $common,
1671                         'u'          => $common,
1672
1673                         # 15.2.2
1674                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1675                         # basefont
1676
1677                         # 15.3
1678                         'hr'         => array_merge( $common, array( 'width' ) ),
1679
1680                         # HTML Ruby annotation text module, simple ruby only.
1681                         # http://www.whatwg.org/html/text-level-semantics.html#the-ruby-element
1682                         'ruby'       => $common,
1683                         # rbc
1684                         'rb'         => $common,
1685                         'rp'         => $common,
1686                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1687                         'rtc'         => $common,
1688
1689                         # MathML root element, where used for extensions
1690                         # 'title' may not be 100% valid here; it's XHTML
1691                         # http://www.w3.org/TR/REC-MathML/
1692                         'math'       => array( 'class', 'style', 'id', 'title' ),
1693
1694                         # HTML 5 section 4.6
1695                         'bdi' => $common,
1696
1697                         # HTML5 elements, defined by:
1698                         # http://www.whatwg.org/html/
1699                         'data' => array_merge( $common, array( 'value' ) ),
1700                         'time' => array_merge( $common, array( 'datetime' ) ),
1701                         'mark' => $common,
1702
1703                         // meta and link are only permitted by removeHTMLtags when Microdata
1704                         // is enabled so we don't bother adding a conditional to hide these
1705                         // Also meta and link are only valid in WikiText as Microdata elements
1706                         // (ie: validateTag rejects tags missing the attributes needed for Microdata)
1707                         // So we don't bother including $common attributes that have no purpose.
1708                         'meta' => array( 'itemprop', 'content' ),
1709                         'link' => array( 'itemprop', 'href' ),
1710                 );
1711
1712                 $staticInitialised = $globalContext;
1713
1714                 return $whitelist;
1715         }
1716
1717         /**
1718          * Take a fragment of (potentially invalid) HTML and return
1719          * a version with any tags removed, encoded as plain text.
1720          *
1721          * Warning: this return value must be further escaped for literal
1722          * inclusion in HTML output as of 1.10!
1723          *
1724          * @param string $text HTML fragment
1725          * @return string
1726          */
1727         static function stripAllTags( $text ) {
1728                 # Actual <tags>
1729                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1730
1731                 # Normalize &entities and whitespace
1732                 $text = self::decodeCharReferences( $text );
1733                 $text = self::normalizeWhitespace( $text );
1734
1735                 return $text;
1736         }
1737
1738         /**
1739          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1740          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1741          * PHP 5.1 doesn't.
1742          *
1743          * Use for passing XHTML fragments to PHP's XML parsing functions
1744          *
1745          * @return string
1746          */
1747         static function hackDocType() {
1748                 $out = "<!DOCTYPE html [\n";
1749                 foreach ( self::$htmlEntities as $entity => $codepoint ) {
1750                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1751                 }
1752                 $out .= "]>\n";
1753                 return $out;
1754         }
1755
1756         /**
1757          * @param string $url
1758          * @return mixed|string
1759          */
1760         static function cleanUrl( $url ) {
1761                 # Normalize any HTML entities in input. They will be
1762                 # re-escaped by makeExternalLink().
1763                 $url = Sanitizer::decodeCharReferences( $url );
1764
1765                 # Escape any control characters introduced by the above step
1766                 $url = preg_replace_callback( '/[\][<>"\\x00-\\x20\\x7F\|]/',
1767                         array( __CLASS__, 'cleanUrlCallback' ), $url );
1768
1769                 # Validate hostname portion
1770                 $matches = array();
1771                 if ( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1772                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1773
1774                         // Characters that will be ignored in IDNs.
1775                         // http://tools.ietf.org/html/3454#section-3.1
1776                         // Strip them before further processing so blacklists and such work.
1777                         $strip = "/
1778                                 \\s|          # general whitespace
1779                                 \xc2\xad|     # 00ad SOFT HYPHEN
1780                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1781                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1782                                 \xe2\x81\xa0| # 2060 WORD JOINER
1783                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1784                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1785                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1786                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1787                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1788                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1789                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1790                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe0f VARIATION SELECTOR-1-16
1791                                 /xuD";
1792
1793                         $host = preg_replace( $strip, '', $host );
1794
1795                         // @todo FIXME: Validate hostnames here
1796
1797                         return $protocol . $host . $rest;
1798                 } else {
1799                         return $url;
1800                 }
1801         }
1802
1803         /**
1804          * @param array $matches
1805          * @return string
1806          */
1807         static function cleanUrlCallback( $matches ) {
1808                 return urlencode( $matches[0] );
1809         }
1810
1811         /**
1812          * Does a string look like an e-mail address?
1813          *
1814          * This validates an email address using an HTML5 specification found at:
1815          * http://www.whatwg.org/html/states-of-the-type-attribute.html#valid-e-mail-address
1816          * Which as of 2011-01-24 says:
1817          *
1818          *   A valid e-mail address is a string that matches the ABNF production
1819          *   1*( atext / "." ) "@" ldh-str *( "." ldh-str ) where atext is defined
1820          *   in RFC 5322 section 3.2.3, and ldh-str is defined in RFC 1034 section
1821          *   3.5.
1822          *
1823          * This function is an implementation of the specification as requested in
1824          * bug 22449.
1825          *
1826          * Client-side forms will use the same standard validation rules via JS or
1827          * HTML 5 validation; additional restrictions can be enforced server-side
1828          * by extensions via the 'isValidEmailAddr' hook.
1829          *
1830          * Note that this validation doesn't 100% match RFC 2822, but is believed
1831          * to be liberal enough for wide use. Some invalid addresses will still
1832          * pass validation here.
1833          *
1834          * @since 1.18
1835          *
1836          * @param string $addr E-mail address
1837          * @return bool
1838          */
1839         public static function validateEmail( $addr ) {
1840                 $result = null;
1841                 if ( !wfRunHooks( 'isValidEmailAddr', array( $addr, &$result ) ) ) {
1842                         return $result;
1843                 }
1844
1845                 // Please note strings below are enclosed in brackets [], this make the
1846                 // hyphen "-" a range indicator. Hence it is double backslashed below.
1847                 // See bug 26948
1848                 $rfc5322_atext = "a-z0-9!#$%&'*+\\-\/=?^_`{|}~";
1849                 $rfc1034_ldh_str = "a-z0-9\\-";
1850
1851                 $html5_email_regexp = "/
1852                 ^                      # start of string
1853                 [$rfc5322_atext\\.]+    # user part which is liberal :p
1854                 @                      # 'apostrophe'
1855                 [$rfc1034_ldh_str]+       # First domain part
1856                 (\\.[$rfc1034_ldh_str]+)*  # Following part prefixed with a dot
1857                 $                      # End of string
1858                 /ix"; // case Insensitive, eXtended
1859
1860                 return (bool)preg_match( $html5_email_regexp, $addr );
1861         }
1862 }