includes/Sanitizer.php

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