includes/Sanitizer.php

   1 <?php
   2 /**
   3  * XHTML sanitizer for MediaWiki
   4  *
   5  * Copyright (C) 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Parser
  25  */
  26
  27 /**
  28  * Regular expression to match various types of character references in
  29  * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  30  */
  31 define( 'MW_CHAR_REFS_REGEX',
  32         '/&([A-Za-z0-9\x80-\xff]+);
  33          |&\#([0-9]+);
  34          |&\#x([0-9A-Za-z]+);
  35          |&\#X([0-9A-Za-z]+);
  36          |(&)/x' );
  37
  38 /**
  39  * Regular expression to match HTML/XML attribute pairs within a tag.
  40  * Allows some... latitude.
  41  * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
  42  */
  43 $attrib = '[A-Za-z0-9]';
  44 $space = '[\x09\x0a\x0d\x20]';
  45 define( 'MW_ATTRIBS_REGEX',
  46         "/(?:^|$space)($attrib+)
  47           ($space*=$space*
  48                 (?:
  49                  # The attribute value: quoted or alone
  50                   \"([^<\"]*)\"
  51                  | '([^<']*)'
  52                  |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
  53                  |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
  54                                                          # colors are specified like this.
  55                                                          # We'll be normalizing it.
  56                 )
  57            )?(?=$space|\$)/sx" );
  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  * @private
  63  */
  64 global $wgHtmlEntities;
  65 $wgHtmlEntities = array(
  66         'Aacute'   => 193,
  67         'aacute'   => 225,
  68         'Acirc'    => 194,
  69         'acirc'    => 226,
  70         'acute'    => 180,
  71         'AElig'    => 198,
  72         'aelig'    => 230,
  73         'Agrave'   => 192,
  74         'agrave'   => 224,
  75         'alefsym'  => 8501,
  76         'Alpha'    => 913,
  77         'alpha'    => 945,
  78         'amp'      => 38,
  79         'and'      => 8743,
  80         'ang'      => 8736,
  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  * Character entity aliases accepted by MediaWiki
 321  */
 322 global $wgHtmlEntityAliases;
 323 $wgHtmlEntityAliases = array(
 324         'רלמ' => 'rlm',
 325         'رلم' => 'rlm',
 326 );
 327
 328
 329 /**
 330  * XHTML sanitizer for MediaWiki
 331  * @ingroup Parser
 332  */
 333 class Sanitizer {
 334         /**
 335          * Cleans up HTML, removes dangerous tags and attributes, and
 336          * removes HTML comments
 337          * @private
 338          * @param string $text
 339          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 340          * @param array $args for the processing callback
 341          * @return string
 342          */
 343         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array() ) {
 344                 global $wgUseTidy;
 345
 346                 static $htmlpairs, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 347                         $htmllist, $listtags, $htmlsingleallowed, $htmlelements, $staticInitialised;
 348
 349                 wfProfileIn( __METHOD__ );
 350
 351                 if ( !$staticInitialised ) {
 352
 353                         $htmlpairs = array_merge( $extratags, array( # Tags that must be closed
 354                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 355                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 356                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 357                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 358                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
 359                         ) );
 360                         $htmlsingle = array(
 361                                 'br', 'hr', 'li', 'dt', 'dd'
 362                         );
 363                         $htmlsingleonly = array( # Elements that cannot have close tags
 364                                 'br', 'hr'
 365                         );
 366                         $htmlnest = array( # Tags that can be nested--??
 367                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 368                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 369                         );
 370                         $tabletags = array( # Can only appear inside table, we will close them
 371                                 'td', 'th', 'tr',
 372                         );
 373                         $htmllist = array( # Tags used by list
 374                                 'ul','ol',
 375                         );
 376                         $listtags = array( # Tags that can appear in a list
 377                                 'li',
 378                         );
 379
 380                         $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
 381                         $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
 382
 383                         # Convert them all to hashtables for faster lookup
 384                         $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 385                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
 386                         foreach ( $vars as $var ) {
 387                                 $$var = array_flip( $$var );
 388                         }
 389                         $staticInitialised = true;
 390                 }
 391
 392                 # Remove HTML comments
 393                 $text = Sanitizer::removeHTMLcomments( $text );
 394                 $bits = explode( '<', $text );
 395                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 396                 if(!$wgUseTidy) {
 397                         $tagstack = $tablestack = array();
 398                         foreach ( $bits as $x ) {
 399                                 $regs = array();
 400                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 401                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 402                                 } else {
 403                                         $slash = $t = $params = $brace = $rest = null;
 404                                 }
 405
 406                                 $badtag = 0 ;
 407                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 408                                         # Check our stack
 409                                         if ( $slash ) {
 410                                                 # Closing a tag...
 411                                                 if( isset( $htmlsingleonly[$t] ) ) {
 412                                                         $badtag = 1;
 413                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 414                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 415                                                                 # Pop all elements with an optional close tag
 416                                                                 # and see if we find a match below them
 417                                                                 $optstack = array();
 418                                                                 array_push ($optstack, $ot);
 419                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 420                                                                                 isset( $htmlsingleallowed[$ot] ) )
 421                                                                 {
 422                                                                         array_push ($optstack, $ot);
 423                                                                 }
 424                                                                 if ( $t != $ot ) {
 425                                                                         # No match. Push the optinal elements back again
 426                                                                         $badtag = 1;
 427                                                                         while ( $ot = @array_pop( $optstack ) ) {
 428                                                                                 array_push( $tagstack, $ot );
 429                                                                         }
 430                                                                 }
 431                                                         } else {
 432                                                                 @array_push( $tagstack, $ot );
 433                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 434                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 435                                                                         $badtag = 1;
 436                                                                 }
 437                                                         }
 438                                                 } else {
 439                                                         if ( $t == 'table' ) {
 440                                                                 $tagstack = array_pop( $tablestack );
 441                                                         }
 442                                                 }
 443                                                 $newparams = '';
 444                                         } else {
 445                                                 # Keep track for later
 446                                                 if ( isset( $tabletags[$t] ) &&
 447                                                 ! in_array( 'table', $tagstack ) ) {
 448                                                         $badtag = 1;
 449                                                 } else if ( in_array( $t, $tagstack ) &&
 450                                                 ! isset( $htmlnest [$t ] ) ) {
 451                                                         $badtag = 1 ;
 452                                                 # Is it a self closed htmlpair ? (bug 5487)
 453                                                 } else if( $brace == '/>' &&
 454                                                 isset( $htmlpairs[$t] ) ) {
 455                                                         $badtag = 1;
 456                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 457                                                         # Hack to force empty tag for uncloseable elements
 458                                                         $brace = '/>';
 459                                                 } else if( isset( $htmlsingle[$t] ) ) {
 460                                                         # Hack to not close $htmlsingle tags
 461                                                         $brace = NULL;
 462                                                 } else if( isset( $tabletags[$t] )
 463                                                 &&  in_array($t ,$tagstack) ) {
 464                                                         // New table tag but forgot to close the previous one
 465                                                         $text .= "</$t>";
 466                                                 } else {
 467                                                         if ( $t == 'table' ) {
 468                                                                 array_push( $tablestack, $tagstack );
 469                                                                 $tagstack = array();
 470                                                         }
 471                                                         array_push( $tagstack, $t );
 472                                                 }
 473
 474                                                 # Replace any variables or template parameters with
 475                                                 # plaintext results.
 476                                                 if( is_callable( $processCallback ) ) {
 477                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 478                                                 }
 479
 480                                                 # Strip non-approved attributes from the tag
 481                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 482                                         }
 483                                         if ( ! $badtag ) {
 484                                                 $rest = str_replace( '>', '&gt;', $rest );
 485                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 486                                                 $text .= "<$slash$t$newparams$close>$rest";
 487                                                 continue;
 488                                         }
 489                                 }
 490                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 491                         }
 492                         # Close off any remaining tags
 493                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 494                                 $text .= "</$t>\n";
 495                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 496                         }
 497                 } else {
 498                         # this might be possible using tidy itself
 499                         foreach ( $bits as $x ) {
 500                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 501                                 $x, $regs );
 502                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 503                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 504                                         if( is_callable( $processCallback ) ) {
 505                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 506                                         }
 507                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 508                                         $rest = str_replace( '>', '&gt;', $rest );
 509                                         $text .= "<$slash$t$newparams$brace$rest";
 510                                 } else {
 511                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 512                                 }
 513                         }
 514                 }
 515                 wfProfileOut( __METHOD__ );
 516                 return $text;
 517         }
 518
 519         /**
 520          * Remove '<!--', '-->', and everything between.
 521          * To avoid leaving blank lines, when a comment is both preceded
 522          * and followed by a newline (ignoring spaces), trim leading and
 523          * trailing spaces and one of the newlines.
 524          *
 525          * @private
 526          * @param string $text
 527          * @return string
 528          */
 529         static function removeHTMLcomments( $text ) {
 530                 wfProfileIn( __METHOD__ );
 531                 while (($start = strpos($text, '<!--')) !== false) {
 532                         $end = strpos($text, '-->', $start + 4);
 533                         if ($end === false) {
 534                                 # Unterminated comment; bail out
 535                                 break;
 536                         }
 537
 538                         $end += 3;
 539
 540                         # Trim space and newline if the comment is both
 541                         # preceded and followed by a newline
 542                         $spaceStart = max($start - 1, 0);
 543                         $spaceLen = $end - $spaceStart;
 544                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 545                                 $spaceStart--;
 546                                 $spaceLen++;
 547                         }
 548                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 549                                 $spaceLen++;
 550                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 551                                 # Remove the comment, leading and trailing
 552                                 # spaces, and leave only one newline.
 553                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 554                         }
 555                         else {
 556                                 # Remove just the comment.
 557                                 $text = substr_replace($text, '', $start, $end - $start);
 558                         }
 559                 }
 560                 wfProfileOut( __METHOD__ );
 561                 return $text;
 562         }
 563
 564         /**
 565          * Take an array of attribute names and values and normalize or discard
 566          * illegal values for the given element type.
 567          *
 568          * - Discards attributes not on a whitelist for the given element
 569          * - Unsafe style attributes are discarded
 570          * - Invalid id attributes are reencoded
 571          *
 572          * @param array $attribs
 573          * @param string $element
 574          * @return array
 575          *
 576          * @todo Check for legal values where the DTD limits things.
 577          * @todo Check for unique id attribute :P
 578          */
 579         static function validateTagAttributes( $attribs, $element ) {
 580                 return Sanitizer::validateAttributes( $attribs,
 581                         Sanitizer::attributeWhitelist( $element ) );
 582         }
 583
 584         /**
 585          * Take an array of attribute names and values and normalize or discard
 586          * illegal values for the given whitelist.
 587          *
 588          * - Discards attributes not the given whitelist
 589          * - Unsafe style attributes are discarded
 590          * - Invalid id attributes are reencoded
 591          *
 592          * @param array $attribs
 593          * @param array $whitelist list of allowed attribute names
 594          * @return array
 595          *
 596          * @todo Check for legal values where the DTD limits things.
 597          * @todo Check for unique id attribute :P
 598          */
 599         static function validateAttributes( $attribs, $whitelist ) {
 600                 $whitelist = array_flip( $whitelist );
 601                 $out = array();
 602                 foreach( $attribs as $attribute => $value ) {
 603                         if( !isset( $whitelist[$attribute] ) ) {
 604                                 continue;
 605                         }
 606                         # Strip javascript "expression" from stylesheets.
 607                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 608                         if( $attribute == 'style' ) {
 609                                 $value = Sanitizer::checkCss( $value );
 610                                 if( $value === false ) {
 611                                         # haxx0r
 612                                         continue;
 613                                 }
 614                         }
 615
 616                         if ( $attribute === 'id' ) {
 617                                 global $wgEnforceHtmlIds;
 618                                 $value = Sanitizer::escapeId( $value,
 619                                         $wgEnforceHtmlIds ? 'noninitial' : 'xml' );
 620                         }
 621
 622                         // If this attribute was previously set, override it.
 623                         // Output should only have one attribute of each name.
 624                         $out[$attribute] = $value;
 625                 }
 626                 return $out;
 627         }
 628
 629         /**
 630          * Merge two sets of HTML attributes.  Conflicting items in the second set
 631          * will override those in the first, except for 'class' attributes which
 632          * will be combined (if they're both strings).
 633          *
 634          * @todo implement merging for other attributes such as style
 635          * @param array $a
 636          * @param array $b
 637          * @return array
 638          */
 639         static function mergeAttributes( $a, $b ) {
 640                 $out = array_merge( $a, $b );
 641                 if( isset( $a['class'] ) && isset( $b['class'] )
 642                 && is_string( $a['class'] ) && is_string( $b['class'] )
 643                 && $a['class'] !== $b['class'] ) {
 644                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 645                                 -1, PREG_SPLIT_NO_EMPTY );
 646                         $out['class'] = implode( ' ', array_unique( $classes ) );
 647                 }
 648                 return $out;
 649         }
 650
 651         /**
 652          * Pick apart some CSS and check it for forbidden or unsafe structures.
 653          * Returns a sanitized string, or false if it was just too evil.
 654          *
 655          * Currently URL references, 'expression', 'tps' are forbidden.
 656          *
 657          * @param string $value
 658          * @return mixed
 659          */
 660         static function checkCss( $value ) {
 661                 $stripped = Sanitizer::decodeCharReferences( $value );
 662
 663                 // Remove any comments; IE gets token splitting wrong
 664                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 665
 666                 $value = $stripped;
 667
 668                 // ... and continue checks
 669                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 670                         'codepointToUtf8(hexdec("$1"))', $stripped );
 671                 $stripped = str_replace( '\\', '', $stripped );
 672                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 673                                 $stripped ) ) {
 674                         # haxx0r
 675                         return false;
 676                 }
 677
 678                 return $value;
 679         }
 680
 681         /**
 682          * Take a tag soup fragment listing an HTML element's attributes
 683          * and normalize it to well-formed XML, discarding unwanted attributes.
 684          * Output is safe for further wikitext processing, with escaping of
 685          * values that could trigger problems.
 686          *
 687          * - Normalizes attribute names to lowercase
 688          * - Discards attributes not on a whitelist for the given element
 689          * - Turns broken or invalid entities into plaintext
 690          * - Double-quotes all attribute values
 691          * - Attributes without values are given the name as attribute
 692          * - Double attributes are discarded
 693          * - Unsafe style attributes are discarded
 694          * - Prepends space if there are attributes.
 695          *
 696          * @param string $text
 697          * @param string $element
 698          * @return string
 699          */
 700         static function fixTagAttributes( $text, $element ) {
 701                 if( trim( $text ) == '' ) {
 702                         return '';
 703                 }
 704
 705                 $stripped = Sanitizer::validateTagAttributes(
 706                         Sanitizer::decodeTagAttributes( $text ), $element );
 707
 708                 $attribs = array();
 709                 foreach( $stripped as $attribute => $value ) {
 710                         $encAttribute = htmlspecialchars( $attribute );
 711                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 712
 713                         $attribs[] = "$encAttribute=\"$encValue\"";
 714                 }
 715                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 716         }
 717
 718         /**
 719          * Encode an attribute value for HTML output.
 720          * @param $text
 721          * @return HTML-encoded text fragment
 722          */
 723         static function encodeAttribute( $text ) {
 724                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 725
 726                 // Whitespace is normalized during attribute decoding,
 727                 // so if we've been passed non-spaces we must encode them
 728                 // ahead of time or they won't be preserved.
 729                 $encValue = strtr( $encValue, array(
 730                         "\n" => '&#10;',
 731                         "\r" => '&#13;',
 732                         "\t" => '&#9;',
 733                 ) );
 734
 735                 return $encValue;
 736         }
 737
 738         /**
 739          * Encode an attribute value for HTML tags, with extra armoring
 740          * against further wiki processing.
 741          * @param $text
 742          * @return HTML-encoded text fragment
 743          */
 744         static function safeEncodeAttribute( $text ) {
 745                 $encValue = Sanitizer::encodeAttribute( $text );
 746
 747                 # Templates and links may be expanded in later parsing,
 748                 # creating invalid or dangerous output. Suppress this.
 749                 $encValue = strtr( $encValue, array(
 750                         '<'    => '&lt;',   // This should never happen,
 751                         '>'    => '&gt;',   // we've received invalid input
 752                         '"'    => '&quot;', // which should have been escaped.
 753                         '{'    => '&#123;',
 754                         '['    => '&#91;',
 755                         "''"   => '&#39;&#39;',
 756                         'ISBN' => '&#73;SBN',
 757                         'RFC'  => '&#82;FC',
 758                         'PMID' => '&#80;MID',
 759                         '|'    => '&#124;',
 760                         '__'   => '&#95;_',
 761                 ) );
 762
 763                 # Stupid hack
 764                 $encValue = preg_replace_callback(
 765                         '/(' . wfUrlProtocols() . ')/',
 766                         array( 'Sanitizer', 'armorLinksCallback' ),
 767                         $encValue );
 768                 return $encValue;
 769         }
 770
 771         /**
 772          * Given a value escape it so that it can be used in an id attribute and
 773          * return it, this does not validate the value however (see first link)
 774          *
 775          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 776          *                                                          in the id and
 777          *                                                          name attributes
 778          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 779          *
 780          * @param string $id      Id to validate
 781          * @param mixed  $options String or array of strings (default is array()):
 782          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 783          *       so don't pay attention if the first character isn't valid at the
 784          *       beginning of an id.
 785          *   'xml': Don't restrict the id to be HTML4-compatible.  This option
 786          *       allows any alphabetic character to be used, per the XML standard.
 787          *       Therefore, it also completely changes the type of escaping: instead
 788          *       of weird dot-encoding, runs of invalid characters (mostly
 789          *       whitespace) are just compressed into a single underscore.
 790          * @return string
 791          */
 792         static function escapeId( $id, $options = array() ) {
 793                 $options = (array)$options;
 794
 795                 if ( !in_array( 'xml', $options ) ) {
 796                         # HTML4-style escaping
 797                         static $replace = array(
 798                                 '%3A' => ':',
 799                                 '%' => '.'
 800                         );
 801
 802                         $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 803                         $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 804
 805                         if ( !preg_match( '/^[a-zA-Z]/', $id )
 806                         && !in_array( 'noninitial', $options ) )  {
 807                                 // Initial character must be a letter!
 808                                 $id = "x$id";
 809                         }
 810                         return $id;
 811                 }
 812
 813                 # XML-style escaping.  For the patterns used, see the XML 1.0 standard,
 814                 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
 815                 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
 816                         . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
 817                         . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
 818                 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
 819                         . '\x{203F}-\x{2040}';
 820                 # Replace _ as well so we don't get multiple consecutive underscores
 821                 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
 822                 $id = trim( $id, '_' );
 823
 824                 if ( !preg_match( "/^[$nameStartChar]/u", $id )
 825                 && !in_array( 'noninitial', $options ) ) {
 826                         $id = "_$id";
 827                 }
 828
 829                 return $id;
 830         }
 831
 832         /**
 833          * Given a value, escape it so that it can be used as a CSS class and
 834          * return it.
 835          *
 836          * @todo For extra validity, input should be validated UTF-8.
 837          *
 838          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 839          *
 840          * @param string $class
 841          * @return string
 842          */
 843         static function escapeClass( $class ) {
 844                 // Convert ugly stuff to underscores and kill underscores in ugly places
 845                 return rtrim(preg_replace(
 846                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 847                         '_',
 848                         $class ), '_');
 849         }
 850
 851         /**
 852          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 853          * This allows (generally harmless) entities like &nbsp; to survive.
 854          *
 855          * @param  string $html String to escape
 856          * @return string Escaped input
 857          */
 858         static function escapeHtmlAllowEntities( $html ) {
 859                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 860                 # hurt.
 861                 $html = htmlspecialchars( $html, ENT_QUOTES );
 862                 $html = str_replace( '&amp;', '&', $html );
 863                 $html = Sanitizer::normalizeCharReferences( $html );
 864                 return $html;
 865         }
 866
 867         /**
 868          * Regex replace callback for armoring links against further processing.
 869          * @param array $matches
 870          * @return string
 871          * @private
 872          */
 873         private static function armorLinksCallback( $matches ) {
 874                 return str_replace( ':', '&#58;', $matches[1] );
 875         }
 876
 877         /**
 878          * Return an associative array of attribute names and values from
 879          * a partial tag string. Attribute names are forces to lowercase,
 880          * character references are decoded to UTF-8 text.
 881          *
 882          * @param string
 883          * @return array
 884          */
 885         public static function decodeTagAttributes( $text ) {
 886                 $attribs = array();
 887
 888                 if( trim( $text ) == '' ) {
 889                         return $attribs;
 890                 }
 891
 892                 $pairs = array();
 893                 if( !preg_match_all(
 894                         MW_ATTRIBS_REGEX,
 895                         $text,
 896                         $pairs,
 897                         PREG_SET_ORDER ) ) {
 898                         return $attribs;
 899                 }
 900
 901                 foreach( $pairs as $set ) {
 902                         $attribute = strtolower( $set[1] );
 903                         $value = Sanitizer::getTagAttributeCallback( $set );
 904
 905                         // Normalize whitespace
 906                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 907                         $value = trim( $value );
 908
 909                         // Decode character references
 910                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 911                 }
 912                 return $attribs;
 913         }
 914
 915         /**
 916          * Pick the appropriate attribute value from a match set from the
 917          * MW_ATTRIBS_REGEX matches.
 918          *
 919          * @param array $set
 920          * @return string
 921          * @private
 922          */
 923         private static function getTagAttributeCallback( $set ) {
 924                 if( isset( $set[6] ) ) {
 925                         # Illegal #XXXXXX color with no quotes.
 926                         return $set[6];
 927                 } elseif( isset( $set[5] ) ) {
 928                         # No quotes.
 929                         return $set[5];
 930                 } elseif( isset( $set[4] ) ) {
 931                         # Single-quoted
 932                         return $set[4];
 933                 } elseif( isset( $set[3] ) ) {
 934                         # Double-quoted
 935                         return $set[3];
 936                 } elseif( !isset( $set[2] ) ) {
 937                         # In XHTML, attributes must have a value.
 938                         # For 'reduced' form, return explicitly the attribute name here.
 939                         return $set[1];
 940                 } else {
 941                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 942                 }
 943         }
 944
 945         /**
 946          * Normalize whitespace and character references in an XML source-
 947          * encoded text for an attribute value.
 948          *
 949          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 950          * but note that we're not returning the value, but are returning
 951          * XML source fragments that will be slapped into output.
 952          *
 953          * @param string $text
 954          * @return string
 955          * @private
 956          */
 957         private static function normalizeAttributeValue( $text ) {
 958                 return str_replace( '"', '&quot;',
 959                         self::normalizeWhitespace(
 960                                 Sanitizer::normalizeCharReferences( $text ) ) );
 961         }
 962
 963         private static function normalizeWhitespace( $text ) {
 964                 return preg_replace(
 965                         '/\r\n|[\x20\x0d\x0a\x09]/',
 966                         ' ',
 967                         $text );
 968         }
 969
 970         /**
 971          * Ensure that any entities and character references are legal
 972          * for XML and XHTML specifically. Any stray bits will be
 973          * &amp;-escaped to result in a valid text fragment.
 974          *
 975          * a. any named char refs must be known in XHTML
 976          * b. any numeric char refs must be legal chars, not invalid or forbidden
 977          * c. use &#x, not &#X
 978          * d. fix or reject non-valid attributes
 979          *
 980          * @param string $text
 981          * @return string
 982          * @private
 983          */
 984         static function normalizeCharReferences( $text ) {
 985                 return preg_replace_callback(
 986                         MW_CHAR_REFS_REGEX,
 987                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 988                         $text );
 989         }
 990         /**
 991          * @param string $matches
 992          * @return string
 993          */
 994         static function normalizeCharReferencesCallback( $matches ) {
 995                 $ret = null;
 996                 if( $matches[1] != '' ) {
 997                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 998                 } elseif( $matches[2] != '' ) {
 999                         $ret = Sanitizer::decCharReference( $matches[2] );
1000                 } elseif( $matches[3] != ''  ) {
1001                         $ret = Sanitizer::hexCharReference( $matches[3] );
1002                 } elseif( $matches[4] != '' ) {
1003                         $ret = Sanitizer::hexCharReference( $matches[4] );
1004                 }
1005                 if( is_null( $ret ) ) {
1006                         return htmlspecialchars( $matches[0] );
1007                 } else {
1008                         return $ret;
1009                 }
1010         }
1011
1012         /**
1013          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1014          * return the named entity reference as is. If the entity is a
1015          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1016          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1017          *
1018          * @param string $name
1019          * @return string
1020          * @static
1021          */
1022         static function normalizeEntity( $name ) {
1023                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1024                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1025                         return "&{$wgHtmlEntityAliases[$name]};";
1026                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1027                         return "&$name;";
1028                 } else {
1029                         return "&amp;$name;";
1030                 }
1031         }
1032
1033         static function decCharReference( $codepoint ) {
1034                 $point = intval( $codepoint );
1035                 if( Sanitizer::validateCodepoint( $point ) ) {
1036                         return sprintf( '&#%d;', $point );
1037                 } else {
1038                         return null;
1039                 }
1040         }
1041
1042         static function hexCharReference( $codepoint ) {
1043                 $point = hexdec( $codepoint );
1044                 if( Sanitizer::validateCodepoint( $point ) ) {
1045                         return sprintf( '&#x%x;', $point );
1046                 } else {
1047                         return null;
1048                 }
1049         }
1050
1051         /**
1052          * Returns true if a given Unicode codepoint is a valid character in XML.
1053          * @param int $codepoint
1054          * @return bool
1055          */
1056         private static function validateCodepoint( $codepoint ) {
1057                 return ($codepoint ==    0x09)
1058                         || ($codepoint ==    0x0a)
1059                         || ($codepoint ==    0x0d)
1060                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1061                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1062                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1063         }
1064
1065         /**
1066          * Decode any character references, numeric or named entities,
1067          * in the text and return a UTF-8 string.
1068          *
1069          * @param string $text
1070          * @return string
1071          * @public
1072          * @static
1073          */
1074         public static function decodeCharReferences( $text ) {
1075                 return preg_replace_callback(
1076                         MW_CHAR_REFS_REGEX,
1077                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1078                         $text );
1079         }
1080
1081         /**
1082          * @param string $matches
1083          * @return string
1084          */
1085         static function decodeCharReferencesCallback( $matches ) {
1086                 if( $matches[1] != '' ) {
1087                         return Sanitizer::decodeEntity( $matches[1] );
1088                 } elseif( $matches[2] != '' ) {
1089                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1090                 } elseif( $matches[3] != ''  ) {
1091                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1092                 } elseif( $matches[4] != '' ) {
1093                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1094                 }
1095                 # Last case should be an ampersand by itself
1096                 return $matches[0];
1097         }
1098
1099         /**
1100          * Return UTF-8 string for a codepoint if that is a valid
1101          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1102          * @param int $codepoint
1103          * @return string
1104          * @private
1105          */
1106         static function decodeChar( $codepoint ) {
1107                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1108                         return codepointToUtf8( $codepoint );
1109                 } else {
1110                         return UTF8_REPLACEMENT;
1111                 }
1112         }
1113
1114         /**
1115          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1116          * return the UTF-8 encoding of that character. Otherwise, returns
1117          * pseudo-entity source (eg &foo;)
1118          *
1119          * @param string $name
1120          * @return string
1121          */
1122         static function decodeEntity( $name ) {
1123                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1124                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1125                         $name = $wgHtmlEntityAliases[$name];
1126                 }
1127                 if( isset( $wgHtmlEntities[$name] ) ) {
1128                         return codepointToUtf8( $wgHtmlEntities[$name] );
1129                 } else {
1130                         return "&$name;";
1131                 }
1132         }
1133
1134         /**
1135          * Fetch the whitelist of acceptable attributes for a given
1136          * element name.
1137          *
1138          * @param string $element
1139          * @return array
1140          */
1141         static function attributeWhitelist( $element ) {
1142                 static $list;
1143                 if( !isset( $list ) ) {
1144                         $list = Sanitizer::setupAttributeWhitelist();
1145                 }
1146                 return isset( $list[$element] )
1147                         ? $list[$element]
1148                         : array();
1149         }
1150
1151         /**
1152          * Foreach array key (an allowed HTML element), return an array
1153          * of allowed attributes
1154          * @return array
1155          */
1156         static function setupAttributeWhitelist() {
1157                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1158                 $block = array_merge( $common, array( 'align' ) );
1159                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1160                 $tablecell = array( 'abbr',
1161                                     'axis',
1162                                     'headers',
1163                                     'scope',
1164                                     'rowspan',
1165                                     'colspan',
1166                                     'nowrap', # deprecated
1167                                     'width',  # deprecated
1168                                     'height', # deprecated
1169                                     'bgcolor' # deprecated
1170                                     );
1171
1172                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1173                 # See: http://www.w3.org/TR/html4/
1174                 $whitelist = array (
1175                         # 7.5.4
1176                         'div'        => $block,
1177                         'center'     => $common, # deprecated
1178                         'span'       => $block, # ??
1179
1180                         # 7.5.5
1181                         'h1'         => $block,
1182                         'h2'         => $block,
1183                         'h3'         => $block,
1184                         'h4'         => $block,
1185                         'h5'         => $block,
1186                         'h6'         => $block,
1187
1188                         # 7.5.6
1189                         # address
1190
1191                         # 8.2.4
1192                         # bdo
1193
1194                         # 9.2.1
1195                         'em'         => $common,
1196                         'strong'     => $common,
1197                         'cite'       => $common,
1198                         # dfn
1199                         'code'       => $common,
1200                         # samp
1201                         # kbd
1202                         'var'        => $common,
1203                         # abbr
1204                         # acronym
1205
1206                         # 9.2.2
1207                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1208                         # q
1209
1210                         # 9.2.3
1211                         'sub'        => $common,
1212                         'sup'        => $common,
1213
1214                         # 9.3.1
1215                         'p'          => $block,
1216
1217                         # 9.3.2
1218                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1219
1220                         # 9.3.4
1221                         'pre'        => array_merge( $common, array( 'width' ) ),
1222
1223                         # 9.4
1224                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1225                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1226
1227                         # 10.2
1228                         'ul'         => array_merge( $common, array( 'type' ) ),
1229                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1230                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1231
1232                         # 10.3
1233                         'dl'         => $common,
1234                         'dd'         => $common,
1235                         'dt'         => $common,
1236
1237                         # 11.2.1
1238                         'table'      => array_merge( $common,
1239                                                                 array( 'summary', 'width', 'border', 'frame',
1240                                                                                 'rules', 'cellspacing', 'cellpadding',
1241                                                                                 'align', 'bgcolor',
1242                                                                 ) ),
1243
1244                         # 11.2.2
1245                         'caption'    => array_merge( $common, array( 'align' ) ),
1246
1247                         # 11.2.3
1248                         'thead'      => array_merge( $common, $tablealign ),
1249                         'tfoot'      => array_merge( $common, $tablealign ),
1250                         'tbody'      => array_merge( $common, $tablealign ),
1251
1252                         # 11.2.4
1253                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1254                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1255
1256                         # 11.2.5
1257                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1258
1259                         # 11.2.6
1260                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1261                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1262
1263                         # 13.2
1264                         # Not usually allowed, but may be used for extension-style hooks
1265                         # such as <math> when it is rasterized
1266                         'img'        => array_merge( $common, array( 'alt' ) ),
1267
1268                         # 15.2.1
1269                         'tt'         => $common,
1270                         'b'          => $common,
1271                         'i'          => $common,
1272                         'big'        => $common,
1273                         'small'      => $common,
1274                         'strike'     => $common,
1275                         's'          => $common,
1276                         'u'          => $common,
1277
1278                         # 15.2.2
1279                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1280                         # basefont
1281
1282                         # 15.3
1283                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1284
1285                         # XHTML Ruby annotation text module, simple ruby only.
1286                         # http://www.w3c.org/TR/ruby/
1287                         'ruby'       => $common,
1288                         # rbc
1289                         # rtc
1290                         'rb'         => $common,
1291                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1292                         'rp'         => $common,
1293
1294                         # MathML root element, where used for extensions
1295                         # 'title' may not be 100% valid here; it's XHTML
1296                         # http://www.w3.org/TR/REC-MathML/
1297                         'math'       => array( 'class', 'style', 'id', 'title' ),
1298                         );
1299                 return $whitelist;
1300         }
1301
1302         /**
1303          * Take a fragment of (potentially invalid) HTML and return
1304          * a version with any tags removed, encoded as plain text.
1305          *
1306          * Warning: this return value must be further escaped for literal
1307          * inclusion in HTML output as of 1.10!
1308          *
1309          * @param string $text HTML fragment
1310          * @return string
1311          */
1312         static function stripAllTags( $text ) {
1313                 # Actual <tags>
1314                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1315
1316                 # Normalize &entities and whitespace
1317                 $text = self::decodeCharReferences( $text );
1318                 $text = self::normalizeWhitespace( $text );
1319
1320                 return $text;
1321         }
1322
1323         /**
1324          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1325          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1326          * PHP 5.1 doesn't.
1327          *
1328          * Use for passing XHTML fragments to PHP's XML parsing functions
1329          *
1330          * @return string
1331          * @static
1332          */
1333         static function hackDocType() {
1334                 global $wgHtmlEntities;
1335                 $out = "<!DOCTYPE html [\n";
1336                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1337                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1338                 }
1339                 $out .= "]>\n";
1340                 return $out;
1341         }
1342
1343         static function cleanUrl( $url ) {
1344                 # Normalize any HTML entities in input. They will be
1345                 # re-escaped by makeExternalLink().
1346                 $url = Sanitizer::decodeCharReferences( $url );
1347
1348                 # Escape any control characters introduced by the above step
1349                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1350
1351                 # Validate hostname portion
1352                 $matches = array();
1353                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1354                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1355
1356                         // Characters that will be ignored in IDNs.
1357                         // http://tools.ietf.org/html/3454#section-3.1
1358                         // Strip them before further processing so blacklists and such work.
1359                         $strip = "/
1360                                 \\s|          # general whitespace
1361                                 \xc2\xad|     # 00ad SOFT HYPHEN
1362                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1363                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1364                                 \xe2\x81\xa0| # 2060 WORD JOINER
1365                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1366                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1367                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1368                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1369                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1370                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1371                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1372                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1373                                 /xuD";
1374
1375                         $host = preg_replace( $strip, '', $host );
1376
1377                         // @fixme: validate hostnames here
1378
1379                         return $protocol . $host . $rest;
1380                 } else {
1381                         return $url;
1382                 }
1383         }
1384
1385 }