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 $text String
 339          * @param $processCallback Callback to do any variable or parameter replacements in HTML attribute values
 340          * @param $args Array for the processing callback
 341          * @param $extratags Array for any extra tags to include
 342          * @param $removetags Array for any tags (default or extra) to exclude
 343          * @return string
 344          */
 345         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array(), $removetags = array() ) {
 346                 global $wgUseTidy;
 347
 348                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 349                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 350
 351                 wfProfileIn( __METHOD__ );
 352
 353                 if ( !$staticInitialised ) {
 354
 355                         $htmlpairsStatic = array( # Tags that must be closed
 356                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 357                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 358                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 359                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 360                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u', 'abbr'
 361                         );
 362                         $htmlsingle = array(
 363                                 'br', 'hr', 'li', 'dt', 'dd'
 364                         );
 365                         $htmlsingleonly = array( # Elements that cannot have close tags
 366                                 'br', 'hr'
 367                         );
 368                         $htmlnest = array( # Tags that can be nested--??
 369                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 370                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 371                         );
 372                         $tabletags = array( # Can only appear inside table, we will close them
 373                                 'td', 'th', 'tr',
 374                         );
 375                         $htmllist = array( # Tags used by list
 376                                 'ul','ol',
 377                         );
 378                         $listtags = array( # Tags that can appear in a list
 379                                 'li',
 380                         );
 381
 382                         $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
 383                         $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
 384
 385                         # Convert them all to hashtables for faster lookup
 386                         $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 387                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
 388                         foreach ( $vars as $var ) {
 389                                 $$var = array_flip( $$var );
 390                         }
 391                         $staticInitialised = true;
 392                 }
 393                 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
 394                 $extratags = array_flip( $extratags );
 395                 $removetags = array_flip( $removetags );
 396                 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
 397                 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ) , $removetags );
 398
 399                 # Remove HTML comments
 400                 $text = Sanitizer::removeHTMLcomments( $text );
 401                 $bits = explode( '<', $text );
 402                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 403                 if(!$wgUseTidy) {
 404                         $tagstack = $tablestack = array();
 405                         foreach ( $bits as $x ) {
 406                                 $regs = array();
 407                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 408                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 409                                 } else {
 410                                         $slash = $t = $params = $brace = $rest = null;
 411                                 }
 412
 413                                 $badtag = 0 ;
 414                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 415                                         # Check our stack
 416                                         if ( $slash ) {
 417                                                 # Closing a tag...
 418                                                 if( isset( $htmlsingleonly[$t] ) ) {
 419                                                         $badtag = 1;
 420                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 421                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 422                                                                 # Pop all elements with an optional close tag
 423                                                                 # and see if we find a match below them
 424                                                                 $optstack = array();
 425                                                                 array_push ($optstack, $ot);
 426                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 427                                                                                 isset( $htmlsingleallowed[$ot] ) )
 428                                                                 {
 429                                                                         array_push ($optstack, $ot);
 430                                                                 }
 431                                                                 if ( $t != $ot ) {
 432                                                                         # No match. Push the optinal elements back again
 433                                                                         $badtag = 1;
 434                                                                         while ( $ot = @array_pop( $optstack ) ) {
 435                                                                                 array_push( $tagstack, $ot );
 436                                                                         }
 437                                                                 }
 438                                                         } else {
 439                                                                 @array_push( $tagstack, $ot );
 440                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 441                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 442                                                                         $badtag = 1;
 443                                                                 }
 444                                                         }
 445                                                 } else {
 446                                                         if ( $t == 'table' ) {
 447                                                                 $tagstack = array_pop( $tablestack );
 448                                                         }
 449                                                 }
 450                                                 $newparams = '';
 451                                         } else {
 452                                                 # Keep track for later
 453                                                 if ( isset( $tabletags[$t] ) &&
 454                                                 ! in_array( 'table', $tagstack ) ) {
 455                                                         $badtag = 1;
 456                                                 } else if ( in_array( $t, $tagstack ) &&
 457                                                 ! isset( $htmlnest [$t ] ) ) {
 458                                                         $badtag = 1 ;
 459                                                 # Is it a self closed htmlpair ? (bug 5487)
 460                                                 } else if( $brace == '/>' &&
 461                                                 isset( $htmlpairs[$t] ) ) {
 462                                                         $badtag = 1;
 463                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 464                                                         # Hack to force empty tag for uncloseable elements
 465                                                         $brace = '/>';
 466                                                 } else if( isset( $htmlsingle[$t] ) ) {
 467                                                         # Hack to not close $htmlsingle tags
 468                                                         $brace = NULL;
 469                                                 } else if( isset( $tabletags[$t] )
 470                                                 &&  in_array($t ,$tagstack) ) {
 471                                                         // New table tag but forgot to close the previous one
 472                                                         $text .= "</$t>";
 473                                                 } else {
 474                                                         if ( $t == 'table' ) {
 475                                                                 array_push( $tablestack, $tagstack );
 476                                                                 $tagstack = array();
 477                                                         }
 478                                                         array_push( $tagstack, $t );
 479                                                 }
 480
 481                                                 # Replace any variables or template parameters with
 482                                                 # plaintext results.
 483                                                 if( is_callable( $processCallback ) ) {
 484                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 485                                                 }
 486
 487                                                 # Strip non-approved attributes from the tag
 488                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 489                                         }
 490                                         if ( ! $badtag ) {
 491                                                 $rest = str_replace( '>', '&gt;', $rest );
 492                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 493                                                 $text .= "<$slash$t$newparams$close>$rest";
 494                                                 continue;
 495                                         }
 496                                 }
 497                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 498                         }
 499                         # Close off any remaining tags
 500                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 501                                 $text .= "</$t>\n";
 502                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 503                         }
 504                 } else {
 505                         # this might be possible using tidy itself
 506                         foreach ( $bits as $x ) {
 507                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 508                                 $x, $regs );
 509                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 510                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 511                                         if( is_callable( $processCallback ) ) {
 512                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 513                                         }
 514                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 515                                         $rest = str_replace( '>', '&gt;', $rest );
 516                                         $text .= "<$slash$t$newparams$brace$rest";
 517                                 } else {
 518                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 519                                 }
 520                         }
 521                 }
 522                 wfProfileOut( __METHOD__ );
 523                 return $text;
 524         }
 525
 526         /**
 527          * Remove '<!--', '-->', and everything between.
 528          * To avoid leaving blank lines, when a comment is both preceded
 529          * and followed by a newline (ignoring spaces), trim leading and
 530          * trailing spaces and one of the newlines.
 531          *
 532          * @private
 533          * @param $text String
 534          * @return string
 535          */
 536         static function removeHTMLcomments( $text ) {
 537                 wfProfileIn( __METHOD__ );
 538                 while (($start = strpos($text, '<!--')) !== false) {
 539                         $end = strpos($text, '-->', $start + 4);
 540                         if ($end === false) {
 541                                 # Unterminated comment; bail out
 542                                 break;
 543                         }
 544
 545                         $end += 3;
 546
 547                         # Trim space and newline if the comment is both
 548                         # preceded and followed by a newline
 549                         $spaceStart = max($start - 1, 0);
 550                         $spaceLen = $end - $spaceStart;
 551                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 552                                 $spaceStart--;
 553                                 $spaceLen++;
 554                         }
 555                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 556                                 $spaceLen++;
 557                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 558                                 # Remove the comment, leading and trailing
 559                                 # spaces, and leave only one newline.
 560                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 561                         }
 562                         else {
 563                                 # Remove just the comment.
 564                                 $text = substr_replace($text, '', $start, $end - $start);
 565                         }
 566                 }
 567                 wfProfileOut( __METHOD__ );
 568                 return $text;
 569         }
 570
 571         /**
 572          * Take an array of attribute names and values and normalize or discard
 573          * illegal values for the given element type.
 574          *
 575          * - Discards attributes not on a whitelist for the given element
 576          * - Unsafe style attributes are discarded
 577          * - Invalid id attributes are reencoded
 578          *
 579          * @param $attribs Array
 580          * @param $element String
 581          * @return Array
 582          *
 583          * @todo Check for legal values where the DTD limits things.
 584          * @todo Check for unique id attribute :P
 585          */
 586         static function validateTagAttributes( $attribs, $element ) {
 587                 return Sanitizer::validateAttributes( $attribs,
 588                         Sanitizer::attributeWhitelist( $element ) );
 589         }
 590
 591         /**
 592          * Take an array of attribute names and values and normalize or discard
 593          * illegal values for the given whitelist.
 594          *
 595          * - Discards attributes not the given whitelist
 596          * - Unsafe style attributes are discarded
 597          * - Invalid id attributes are reencoded
 598          *
 599          * @param $attribs Array
 600          * @param $whitelist Array: list of allowed attribute names
 601          * @return Array
 602          *
 603          * @todo Check for legal values where the DTD limits things.
 604          * @todo Check for unique id attribute :P
 605          */
 606         static function validateAttributes( $attribs, $whitelist ) {
 607                 $whitelist = array_flip( $whitelist );
 608                 $out = array();
 609                 foreach( $attribs as $attribute => $value ) {
 610                         if( !isset( $whitelist[$attribute] ) ) {
 611                                 continue;
 612                         }
 613                         # Strip javascript "expression" from stylesheets.
 614                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 615                         if( $attribute == 'style' ) {
 616                                 $value = Sanitizer::checkCss( $value );
 617                                 if( $value === false ) {
 618                                         # haxx0r
 619                                         continue;
 620                                 }
 621                         }
 622
 623                         if ( $attribute === 'id' ) {
 624                                 global $wgEnforceHtmlIds;
 625                                 $value = Sanitizer::escapeId( $value,
 626                                         $wgEnforceHtmlIds ? 'noninitial' : 'xml' );
 627                         }
 628
 629                         //RDFa properties allow URIs. check them
 630                         if ( $attribute === 'rel' || $attribute === 'rev' ||
 631                                 $attribute === 'about' || $attribute === 'property' || $attribute === 'resource' ||
 632                                 $attribute === 'datatype' || $attribute === 'typeof' ) {
 633                                 //Paranoia. Allow "simple" values but suppress javascript
 634                                 if ( preg_match( '/(^|\s)javascript\s*:/i', $value ) ) {
 635                                         continue;
 636                                 }
 637                         }
 638
 639                         // If this attribute was previously set, override it.
 640                         // Output should only have one attribute of each name.
 641                         $out[$attribute] = $value;
 642                 }
 643                 return $out;
 644         }
 645
 646         /**
 647          * Merge two sets of HTML attributes.  Conflicting items in the second set
 648          * will override those in the first, except for 'class' attributes which
 649          * will be combined (if they're both strings).
 650          *
 651          * @todo implement merging for other attributes such as style
 652          * @param $a Array
 653          * @param $b Array
 654          * @return array
 655          */
 656         static function mergeAttributes( $a, $b ) {
 657                 $out = array_merge( $a, $b );
 658                 if( isset( $a['class'] ) && isset( $b['class'] )
 659                 && is_string( $a['class'] ) && is_string( $b['class'] )
 660                 && $a['class'] !== $b['class'] ) {
 661                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 662                                 -1, PREG_SPLIT_NO_EMPTY );
 663                         $out['class'] = implode( ' ', array_unique( $classes ) );
 664                 }
 665                 return $out;
 666         }
 667
 668         /**
 669          * Pick apart some CSS and check it for forbidden or unsafe structures.
 670          * Returns a sanitized string, or false if it was just too evil.
 671          *
 672          * Currently URL references, 'expression', 'tps' are forbidden.
 673          *
 674          * @param $value String
 675          * @return Mixed
 676          */
 677         static function checkCss( $value ) {
 678                 $stripped = Sanitizer::decodeCharReferences( $value );
 679
 680                 // Remove any comments; IE gets token splitting wrong
 681                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 682
 683                 $value = $stripped;
 684
 685                 // ... and continue checks
 686                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 687                         'codepointToUtf8(hexdec("$1"))', $stripped );
 688                 $stripped = str_replace( '\\', '', $stripped );
 689                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 690                                 $stripped ) ) {
 691                         # haxx0r
 692                         return false;
 693                 }
 694
 695                 return $value;
 696         }
 697
 698         /**
 699          * Take a tag soup fragment listing an HTML element's attributes
 700          * and normalize it to well-formed XML, discarding unwanted attributes.
 701          * Output is safe for further wikitext processing, with escaping of
 702          * values that could trigger problems.
 703          *
 704          * - Normalizes attribute names to lowercase
 705          * - Discards attributes not on a whitelist for the given element
 706          * - Turns broken or invalid entities into plaintext
 707          * - Double-quotes all attribute values
 708          * - Attributes without values are given the name as attribute
 709          * - Double attributes are discarded
 710          * - Unsafe style attributes are discarded
 711          * - Prepends space if there are attributes.
 712          *
 713          * @param $text String
 714          * @param $element String
 715          * @return String
 716          */
 717         static function fixTagAttributes( $text, $element ) {
 718                 if( trim( $text ) == '' ) {
 719                         return '';
 720                 }
 721
 722                 $stripped = Sanitizer::validateTagAttributes(
 723                         Sanitizer::decodeTagAttributes( $text ), $element );
 724
 725                 $attribs = array();
 726                 foreach( $stripped as $attribute => $value ) {
 727                         $encAttribute = htmlspecialchars( $attribute );
 728                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 729
 730                         $attribs[] = "$encAttribute=\"$encValue\"";
 731                 }
 732                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 733         }
 734
 735         /**
 736          * Encode an attribute value for HTML output.
 737          * @param $text String
 738          * @return HTML-encoded text fragment
 739          */
 740         static function encodeAttribute( $text ) {
 741                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 742
 743                 // Whitespace is normalized during attribute decoding,
 744                 // so if we've been passed non-spaces we must encode them
 745                 // ahead of time or they won't be preserved.
 746                 $encValue = strtr( $encValue, array(
 747                         "\n" => '&#10;',
 748                         "\r" => '&#13;',
 749                         "\t" => '&#9;',
 750                 ) );
 751
 752                 return $encValue;
 753         }
 754
 755         /**
 756          * Encode an attribute value for HTML tags, with extra armoring
 757          * against further wiki processing.
 758          * @param $text String
 759          * @return HTML-encoded text fragment
 760          */
 761         static function safeEncodeAttribute( $text ) {
 762                 $encValue = Sanitizer::encodeAttribute( $text );
 763
 764                 # Templates and links may be expanded in later parsing,
 765                 # creating invalid or dangerous output. Suppress this.
 766                 $encValue = strtr( $encValue, array(
 767                         '<'    => '&lt;',   // This should never happen,
 768                         '>'    => '&gt;',   // we've received invalid input
 769                         '"'    => '&quot;', // which should have been escaped.
 770                         '{'    => '&#123;',
 771                         '['    => '&#91;',
 772                         "''"   => '&#39;&#39;',
 773                         'ISBN' => '&#73;SBN',
 774                         'RFC'  => '&#82;FC',
 775                         'PMID' => '&#80;MID',
 776                         '|'    => '&#124;',
 777                         '__'   => '&#95;_',
 778                 ) );
 779
 780                 # Stupid hack
 781                 $encValue = preg_replace_callback(
 782                         '/(' . wfUrlProtocols() . ')/',
 783                         array( 'Sanitizer', 'armorLinksCallback' ),
 784                         $encValue );
 785                 return $encValue;
 786         }
 787
 788         /**
 789          * Given a value escape it so that it can be used in an id attribute and
 790          * return it, this does not validate the value however (see first link)
 791          *
 792          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 793          *                                                          in the id and
 794          *                                                          name attributes
 795          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 796          *
 797          * @param $id String: id to validate
 798          * @param $options Mixed: string or array of strings (default is array()):
 799          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 800          *       so don't pay attention if the first character isn't valid at the
 801          *       beginning of an id.
 802          *   'xml': Don't restrict the id to be HTML4-compatible.  This option
 803          *       allows any alphabetic character to be used, per the XML standard.
 804          *       Therefore, it also completely changes the type of escaping: instead
 805          *       of weird dot-encoding, runs of invalid characters (mostly
 806          *       whitespace) are just compressed into a single underscore.
 807          * @return String
 808          */
 809         static function escapeId( $id, $options = array() ) {
 810                 $options = (array)$options;
 811
 812                 if ( !in_array( 'xml', $options ) ) {
 813                         # HTML4-style escaping
 814                         static $replace = array(
 815                                 '%3A' => ':',
 816                                 '%' => '.'
 817                         );
 818
 819                         $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 820                         $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 821
 822                         if ( !preg_match( '/^[a-zA-Z]/', $id )
 823                         && !in_array( 'noninitial', $options ) )  {
 824                                 // Initial character must be a letter!
 825                                 $id = "x$id";
 826                         }
 827                         return $id;
 828                 }
 829
 830                 # XML-style escaping.  For the patterns used, see the XML 1.0 standard,
 831                 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
 832                 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
 833                         . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
 834                         . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
 835                 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
 836                         . '\x{203F}-\x{2040}';
 837                 # Replace _ as well so we don't get multiple consecutive underscores
 838                 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
 839                 $id = trim( $id, '_' );
 840
 841                 if ( !preg_match( "/^[$nameStartChar]/u", $id )
 842                 && !in_array( 'noninitial', $options ) ) {
 843                         $id = "_$id";
 844                 }
 845
 846                 return $id;
 847         }
 848
 849         /**
 850          * Given a value, escape it so that it can be used as a CSS class and
 851          * return it.
 852          *
 853          * @todo For extra validity, input should be validated UTF-8.
 854          *
 855          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 856          *
 857          * @param $class String
 858          * @return String
 859          */
 860         static function escapeClass( $class ) {
 861                 // Convert ugly stuff to underscores and kill underscores in ugly places
 862                 return rtrim(preg_replace(
 863                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 864                         '_',
 865                         $class ), '_');
 866         }
 867
 868         /**
 869          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 870          * This allows (generally harmless) entities like &nbsp; to survive.
 871          *
 872          * @param $html String to escape
 873          * @return String: escaped input
 874          */
 875         static function escapeHtmlAllowEntities( $html ) {
 876                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 877                 # hurt.
 878                 $html = htmlspecialchars( $html, ENT_QUOTES );
 879                 $html = str_replace( '&amp;', '&', $html );
 880                 $html = Sanitizer::normalizeCharReferences( $html );
 881                 return $html;
 882         }
 883
 884         /**
 885          * Regex replace callback for armoring links against further processing.
 886          * @param $matches Array
 887          * @return string
 888          */
 889         private static function armorLinksCallback( $matches ) {
 890                 return str_replace( ':', '&#58;', $matches[1] );
 891         }
 892
 893         /**
 894          * Return an associative array of attribute names and values from
 895          * a partial tag string. Attribute names are forces to lowercase,
 896          * character references are decoded to UTF-8 text.
 897          *
 898          * @param $text String
 899          * @return Array
 900          */
 901         public static function decodeTagAttributes( $text ) {
 902                 $attribs = array();
 903
 904                 if( trim( $text ) == '' ) {
 905                         return $attribs;
 906                 }
 907
 908                 $pairs = array();
 909                 if( !preg_match_all(
 910                         MW_ATTRIBS_REGEX,
 911                         $text,
 912                         $pairs,
 913                         PREG_SET_ORDER ) ) {
 914                         return $attribs;
 915                 }
 916
 917                 foreach( $pairs as $set ) {
 918                         $attribute = strtolower( $set[1] );
 919                         $value = Sanitizer::getTagAttributeCallback( $set );
 920
 921                         // Normalize whitespace
 922                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 923                         $value = trim( $value );
 924
 925                         // Decode character references
 926                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 927                 }
 928                 return $attribs;
 929         }
 930
 931         /**
 932          * Pick the appropriate attribute value from a match set from the
 933          * MW_ATTRIBS_REGEX matches.
 934          *
 935          * @param $set Array
 936          * @return String
 937          */
 938         private static function getTagAttributeCallback( $set ) {
 939                 if( isset( $set[6] ) ) {
 940                         # Illegal #XXXXXX color with no quotes.
 941                         return $set[6];
 942                 } elseif( isset( $set[5] ) ) {
 943                         # No quotes.
 944                         return $set[5];
 945                 } elseif( isset( $set[4] ) ) {
 946                         # Single-quoted
 947                         return $set[4];
 948                 } elseif( isset( $set[3] ) ) {
 949                         # Double-quoted
 950                         return $set[3];
 951                 } elseif( !isset( $set[2] ) ) {
 952                         # In XHTML, attributes must have a value.
 953                         # For 'reduced' form, return explicitly the attribute name here.
 954                         return $set[1];
 955                 } else {
 956                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 957                 }
 958         }
 959
 960         /**
 961          * Normalize whitespace and character references in an XML source-
 962          * encoded text for an attribute value.
 963          *
 964          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 965          * but note that we're not returning the value, but are returning
 966          * XML source fragments that will be slapped into output.
 967          *
 968          * @param $text String
 969          * @return String
 970          */
 971         private static function normalizeAttributeValue( $text ) {
 972                 return str_replace( '"', '&quot;',
 973                         self::normalizeWhitespace(
 974                                 Sanitizer::normalizeCharReferences( $text ) ) );
 975         }
 976
 977         private static function normalizeWhitespace( $text ) {
 978                 return preg_replace(
 979                         '/\r\n|[\x20\x0d\x0a\x09]/',
 980                         ' ',
 981                         $text );
 982         }
 983
 984         /**
 985          * Ensure that any entities and character references are legal
 986          * for XML and XHTML specifically. Any stray bits will be
 987          * &amp;-escaped to result in a valid text fragment.
 988          *
 989          * a. any named char refs must be known in XHTML
 990          * b. any numeric char refs must be legal chars, not invalid or forbidden
 991          * c. use &#x, not &#X
 992          * d. fix or reject non-valid attributes
 993          *
 994          * @param $text String
 995          * @return String
 996          * @private
 997          */
 998         static function normalizeCharReferences( $text ) {
 999                 return preg_replace_callback(
1000                         MW_CHAR_REFS_REGEX,
1001                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1002                         $text );
1003         }
1004         /**
1005          * @param $matches String
1006          * @return String
1007          */
1008         static function normalizeCharReferencesCallback( $matches ) {
1009                 $ret = null;
1010                 if( $matches[1] != '' ) {
1011                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1012                 } elseif( $matches[2] != '' ) {
1013                         $ret = Sanitizer::decCharReference( $matches[2] );
1014                 } elseif( $matches[3] != ''  ) {
1015                         $ret = Sanitizer::hexCharReference( $matches[3] );
1016                 } elseif( $matches[4] != '' ) {
1017                         $ret = Sanitizer::hexCharReference( $matches[4] );
1018                 }
1019                 if( is_null( $ret ) ) {
1020                         return htmlspecialchars( $matches[0] );
1021                 } else {
1022                         return $ret;
1023                 }
1024         }
1025
1026         /**
1027          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1028          * return the named entity reference as is. If the entity is a
1029          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1030          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1031          *
1032          * @param $name String
1033          * @return String
1034          */
1035         static function normalizeEntity( $name ) {
1036                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1037                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1038                         return "&{$wgHtmlEntityAliases[$name]};";
1039                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1040                         return "&$name;";
1041                 } else {
1042                         return "&amp;$name;";
1043                 }
1044         }
1045
1046         static function decCharReference( $codepoint ) {
1047                 $point = intval( $codepoint );
1048                 if( Sanitizer::validateCodepoint( $point ) ) {
1049                         return sprintf( '&#%d;', $point );
1050                 } else {
1051                         return null;
1052                 }
1053         }
1054
1055         static function hexCharReference( $codepoint ) {
1056                 $point = hexdec( $codepoint );
1057                 if( Sanitizer::validateCodepoint( $point ) ) {
1058                         return sprintf( '&#x%x;', $point );
1059                 } else {
1060                         return null;
1061                 }
1062         }
1063
1064         /**
1065          * Returns true if a given Unicode codepoint is a valid character in XML.
1066          * @param $codepoint Integer
1067          * @return Boolean
1068          */
1069         private static function validateCodepoint( $codepoint ) {
1070                 return ($codepoint ==    0x09)
1071                         || ($codepoint ==    0x0a)
1072                         || ($codepoint ==    0x0d)
1073                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1074                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1075                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1076         }
1077
1078         /**
1079          * Decode any character references, numeric or named entities,
1080          * in the text and return a UTF-8 string.
1081          *
1082          * @param $text String
1083          * @return String
1084          */
1085         public static function decodeCharReferences( $text ) {
1086                 return preg_replace_callback(
1087                         MW_CHAR_REFS_REGEX,
1088                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1089                         $text );
1090         }
1091
1092         /**
1093          * @param $matches String
1094          * @return String
1095          */
1096         static function decodeCharReferencesCallback( $matches ) {
1097                 if( $matches[1] != '' ) {
1098                         return Sanitizer::decodeEntity( $matches[1] );
1099                 } elseif( $matches[2] != '' ) {
1100                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1101                 } elseif( $matches[3] != ''  ) {
1102                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1103                 } elseif( $matches[4] != '' ) {
1104                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1105                 }
1106                 # Last case should be an ampersand by itself
1107                 return $matches[0];
1108         }
1109
1110         /**
1111          * Return UTF-8 string for a codepoint if that is a valid
1112          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1113          * @param $codepoint Integer
1114          * @return String
1115          * @private
1116          */
1117         static function decodeChar( $codepoint ) {
1118                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1119                         return codepointToUtf8( $codepoint );
1120                 } else {
1121                         return UTF8_REPLACEMENT;
1122                 }
1123         }
1124
1125         /**
1126          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1127          * return the UTF-8 encoding of that character. Otherwise, returns
1128          * pseudo-entity source (eg &foo;)
1129          *
1130          * @param $name Strings
1131          * @return String
1132          */
1133         static function decodeEntity( $name ) {
1134                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1135                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1136                         $name = $wgHtmlEntityAliases[$name];
1137                 }
1138                 if( isset( $wgHtmlEntities[$name] ) ) {
1139                         return codepointToUtf8( $wgHtmlEntities[$name] );
1140                 } else {
1141                         return "&$name;";
1142                 }
1143         }
1144
1145         /**
1146          * Fetch the whitelist of acceptable attributes for a given element name.
1147          *
1148          * @param $element String
1149          * @return Array
1150          */
1151         static function attributeWhitelist( $element ) {
1152                 static $list;
1153                 if( !isset( $list ) ) {
1154                         $list = Sanitizer::setupAttributeWhitelist();
1155                 }
1156                 return isset( $list[$element] )
1157                         ? $list[$element]
1158                         : array();
1159         }
1160
1161         /**
1162          * Foreach array key (an allowed HTML element), return an array
1163          * of allowed attributes
1164          * @return Array
1165          */
1166         static function setupAttributeWhitelist() {
1167                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style',
1168                                  #RDFa attributes as specified in section 9 of http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1169                                  'about', 'property', 'resource', 'datatype', 'typeof',
1170                                 );
1171
1172                 $block = array_merge( $common, array( 'align' ) );
1173                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1174                 $tablecell = array( 'abbr',
1175                                     'axis',
1176                                     'headers',
1177                                     'scope',
1178                                     'rowspan',
1179                                     'colspan',
1180                                     'nowrap', # deprecated
1181                                     'width',  # deprecated
1182                                     'height', # deprecated
1183                                     'bgcolor' # deprecated
1184                                     );
1185
1186                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1187                 # See: http://www.w3.org/TR/html4/
1188                 $whitelist = array (
1189                         # 7.5.4
1190                         'div'        => $block,
1191                         'center'     => $common, # deprecated
1192                         'span'       => $block, # ??
1193
1194                         # 7.5.5
1195                         'h1'         => $block,
1196                         'h2'         => $block,
1197                         'h3'         => $block,
1198                         'h4'         => $block,
1199                         'h5'         => $block,
1200                         'h6'         => $block,
1201
1202                         # 7.5.6
1203                         # address
1204
1205                         # 8.2.4
1206                         # bdo
1207
1208                         # 9.2.1
1209                         'em'         => $common,
1210                         'strong'     => $common,
1211                         'cite'       => $common,
1212                         # dfn
1213                         'code'       => $common,
1214                         # samp
1215                         # kbd
1216                         'var'        => $common,
1217                         'abbr'       => $common,
1218                         # acronym
1219
1220                         # 9.2.2
1221                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1222                         # q
1223
1224                         # 9.2.3
1225                         'sub'        => $common,
1226                         'sup'        => $common,
1227
1228                         # 9.3.1
1229                         'p'          => $block,
1230
1231                         # 9.3.2
1232                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1233
1234                         # 9.3.4
1235                         'pre'        => array_merge( $common, array( 'width' ) ),
1236
1237                         # 9.4
1238                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1239                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1240
1241                         # 10.2
1242                         'ul'         => array_merge( $common, array( 'type' ) ),
1243                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1244                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1245
1246                         # 10.3
1247                         'dl'         => $common,
1248                         'dd'         => $common,
1249                         'dt'         => $common,
1250
1251                         # 11.2.1
1252                         'table'      => array_merge( $common,
1253                                                                 array( 'summary', 'width', 'border', 'frame',
1254                                                                                 'rules', 'cellspacing', 'cellpadding',
1255                                                                                 'align', 'bgcolor',
1256                                                                 ) ),
1257
1258                         # 11.2.2
1259                         'caption'    => array_merge( $common, array( 'align' ) ),
1260
1261                         # 11.2.3
1262                         'thead'      => array_merge( $common, $tablealign ),
1263                         'tfoot'      => array_merge( $common, $tablealign ),
1264                         'tbody'      => array_merge( $common, $tablealign ),
1265
1266                         # 11.2.4
1267                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1268                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1269
1270                         # 11.2.5
1271                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1272
1273                         # 11.2.6
1274                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1275                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1276
1277                         # 13.2
1278                         # Not usually allowed, but may be used for extension-style hooks
1279                         # such as <math> when it is rasterized
1280                         'img'        => array_merge( $common, array( 'alt' ) ),
1281
1282                         # 15.2.1
1283                         'tt'         => $common,
1284                         'b'          => $common,
1285                         'i'          => $common,
1286                         'big'        => $common,
1287                         'small'      => $common,
1288                         'strike'     => $common,
1289                         's'          => $common,
1290                         'u'          => $common,
1291
1292                         # 15.2.2
1293                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1294                         # basefont
1295
1296                         # 15.3
1297                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1298
1299                         # XHTML Ruby annotation text module, simple ruby only.
1300                         # http://www.w3c.org/TR/ruby/
1301                         'ruby'       => $common,
1302                         # rbc
1303                         # rtc
1304                         'rb'         => $common,
1305                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1306                         'rp'         => $common,
1307
1308                         # MathML root element, where used for extensions
1309                         # 'title' may not be 100% valid here; it's XHTML
1310                         # http://www.w3.org/TR/REC-MathML/
1311                         'math'       => array( 'class', 'style', 'id', 'title' ),
1312                         );
1313                 return $whitelist;
1314         }
1315
1316         /**
1317          * Take a fragment of (potentially invalid) HTML and return
1318          * a version with any tags removed, encoded as plain text.
1319          *
1320          * Warning: this return value must be further escaped for literal
1321          * inclusion in HTML output as of 1.10!
1322          *
1323          * @param $text String: HTML fragment
1324          * @return String
1325          */
1326         static function stripAllTags( $text ) {
1327                 # Actual <tags>
1328                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1329
1330                 # Normalize &entities and whitespace
1331                 $text = self::decodeCharReferences( $text );
1332                 $text = self::normalizeWhitespace( $text );
1333
1334                 return $text;
1335         }
1336
1337         /**
1338          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1339          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1340          * PHP 5.1 doesn't.
1341          *
1342          * Use for passing XHTML fragments to PHP's XML parsing functions
1343          *
1344          * @return String
1345          */
1346         static function hackDocType() {
1347                 global $wgHtmlEntities;
1348                 $out = "<!DOCTYPE html [\n";
1349                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1350                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1351                 }
1352                 $out .= "]>\n";
1353                 return $out;
1354         }
1355
1356         static function cleanUrl( $url ) {
1357                 # Normalize any HTML entities in input. They will be
1358                 # re-escaped by makeExternalLink().
1359                 $url = Sanitizer::decodeCharReferences( $url );
1360
1361                 # Escape any control characters introduced by the above step
1362                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1363
1364                 # Validate hostname portion
1365                 $matches = array();
1366                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1367                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1368
1369                         // Characters that will be ignored in IDNs.
1370                         // http://tools.ietf.org/html/3454#section-3.1
1371                         // Strip them before further processing so blacklists and such work.
1372                         $strip = "/
1373                                 \\s|          # general whitespace
1374                                 \xc2\xad|     # 00ad SOFT HYPHEN
1375                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1376                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1377                                 \xe2\x81\xa0| # 2060 WORD JOINER
1378                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1379                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1380                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1381                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1382                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1383                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1384                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1385                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1386                                 /xuD";
1387
1388                         $host = preg_replace( $strip, '', $host );
1389
1390                         // @fixme: validate hostnames here
1391
1392                         return $protocol . $host . $rest;
1393                 } else {
1394                         return $url;
1395                 }
1396         }
1397
1398 }