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          * @param array $extratags for any extra tags to include
 342          * @param array $removetags 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'
 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( array_unique( 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 string $text
 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 array $attribs
 580          * @param string $element
 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 array $attribs
 600          * @param array $whitelist 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                         // If this attribute was previously set, override it.
 630                         // Output should only have one attribute of each name.
 631                         $out[$attribute] = $value;
 632                 }
 633                 return $out;
 634         }
 635
 636         /**
 637          * Merge two sets of HTML attributes.  Conflicting items in the second set
 638          * will override those in the first, except for 'class' attributes which
 639          * will be combined (if they're both strings).
 640          *
 641          * @todo implement merging for other attributes such as style
 642          * @param array $a
 643          * @param array $b
 644          * @return array
 645          */
 646         static function mergeAttributes( $a, $b ) {
 647                 $out = array_merge( $a, $b );
 648                 if( isset( $a['class'] ) && isset( $b['class'] )
 649                 && is_string( $a['class'] ) && is_string( $b['class'] )
 650                 && $a['class'] !== $b['class'] ) {
 651                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 652                                 -1, PREG_SPLIT_NO_EMPTY );
 653                         $out['class'] = implode( ' ', array_unique( $classes ) );
 654                 }
 655                 return $out;
 656         }
 657
 658         /**
 659          * Pick apart some CSS and check it for forbidden or unsafe structures.
 660          * Returns a sanitized string, or false if it was just too evil.
 661          *
 662          * Currently URL references, 'expression', 'tps' are forbidden.
 663          *
 664          * @param string $value
 665          * @return mixed
 666          */
 667         static function checkCss( $value ) {
 668                 $stripped = Sanitizer::decodeCharReferences( $value );
 669
 670                 // Remove any comments; IE gets token splitting wrong
 671                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 672
 673                 $value = $stripped;
 674
 675                 // ... and continue checks
 676                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 677                         'codepointToUtf8(hexdec("$1"))', $stripped );
 678                 $stripped = str_replace( '\\', '', $stripped );
 679                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 680                                 $stripped ) ) {
 681                         # haxx0r
 682                         return false;
 683                 }
 684
 685                 return $value;
 686         }
 687
 688         /**
 689          * Take a tag soup fragment listing an HTML element's attributes
 690          * and normalize it to well-formed XML, discarding unwanted attributes.
 691          * Output is safe for further wikitext processing, with escaping of
 692          * values that could trigger problems.
 693          *
 694          * - Normalizes attribute names to lowercase
 695          * - Discards attributes not on a whitelist for the given element
 696          * - Turns broken or invalid entities into plaintext
 697          * - Double-quotes all attribute values
 698          * - Attributes without values are given the name as attribute
 699          * - Double attributes are discarded
 700          * - Unsafe style attributes are discarded
 701          * - Prepends space if there are attributes.
 702          *
 703          * @param string $text
 704          * @param string $element
 705          * @return string
 706          */
 707         static function fixTagAttributes( $text, $element ) {
 708                 if( trim( $text ) == '' ) {
 709                         return '';
 710                 }
 711
 712                 $stripped = Sanitizer::validateTagAttributes(
 713                         Sanitizer::decodeTagAttributes( $text ), $element );
 714
 715                 $attribs = array();
 716                 foreach( $stripped as $attribute => $value ) {
 717                         $encAttribute = htmlspecialchars( $attribute );
 718                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 719
 720                         $attribs[] = "$encAttribute=\"$encValue\"";
 721                 }
 722                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 723         }
 724
 725         /**
 726          * Encode an attribute value for HTML output.
 727          * @param $text
 728          * @return HTML-encoded text fragment
 729          */
 730         static function encodeAttribute( $text ) {
 731                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 732
 733                 // Whitespace is normalized during attribute decoding,
 734                 // so if we've been passed non-spaces we must encode them
 735                 // ahead of time or they won't be preserved.
 736                 $encValue = strtr( $encValue, array(
 737                         "\n" => '&#10;',
 738                         "\r" => '&#13;',
 739                         "\t" => '&#9;',
 740                 ) );
 741
 742                 return $encValue;
 743         }
 744
 745         /**
 746          * Encode an attribute value for HTML tags, with extra armoring
 747          * against further wiki processing.
 748          * @param $text
 749          * @return HTML-encoded text fragment
 750          */
 751         static function safeEncodeAttribute( $text ) {
 752                 $encValue = Sanitizer::encodeAttribute( $text );
 753
 754                 # Templates and links may be expanded in later parsing,
 755                 # creating invalid or dangerous output. Suppress this.
 756                 $encValue = strtr( $encValue, array(
 757                         '<'    => '&lt;',   // This should never happen,
 758                         '>'    => '&gt;',   // we've received invalid input
 759                         '"'    => '&quot;', // which should have been escaped.
 760                         '{'    => '&#123;',
 761                         '['    => '&#91;',
 762                         "''"   => '&#39;&#39;',
 763                         'ISBN' => '&#73;SBN',
 764                         'RFC'  => '&#82;FC',
 765                         'PMID' => '&#80;MID',
 766                         '|'    => '&#124;',
 767                         '__'   => '&#95;_',
 768                 ) );
 769
 770                 # Stupid hack
 771                 $encValue = preg_replace_callback(
 772                         '/(' . wfUrlProtocols() . ')/',
 773                         array( 'Sanitizer', 'armorLinksCallback' ),
 774                         $encValue );
 775                 return $encValue;
 776         }
 777
 778         /**
 779          * Given a value escape it so that it can be used in an id attribute and
 780          * return it, this does not validate the value however (see first link)
 781          *
 782          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 783          *                                                          in the id and
 784          *                                                          name attributes
 785          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 786          *
 787          * @param string $id      Id to validate
 788          * @param mixed  $options String or array of strings (default is array()):
 789          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 790          *       so don't pay attention if the first character isn't valid at the
 791          *       beginning of an id.
 792          *   'xml': Don't restrict the id to be HTML4-compatible.  This option
 793          *       allows any alphabetic character to be used, per the XML standard.
 794          *       Therefore, it also completely changes the type of escaping: instead
 795          *       of weird dot-encoding, runs of invalid characters (mostly
 796          *       whitespace) are just compressed into a single underscore.
 797          * @return string
 798          */
 799         static function escapeId( $id, $options = array() ) {
 800                 $options = (array)$options;
 801
 802                 if ( !in_array( 'xml', $options ) ) {
 803                         # HTML4-style escaping
 804                         static $replace = array(
 805                                 '%3A' => ':',
 806                                 '%' => '.'
 807                         );
 808
 809                         $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 810                         $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 811
 812                         if ( !preg_match( '/^[a-zA-Z]/', $id )
 813                         && !in_array( 'noninitial', $options ) )  {
 814                                 // Initial character must be a letter!
 815                                 $id = "x$id";
 816                         }
 817                         return $id;
 818                 }
 819
 820                 # XML-style escaping.  For the patterns used, see the XML 1.0 standard,
 821                 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
 822                 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
 823                         . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
 824                         . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
 825                 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
 826                         . '\x{203F}-\x{2040}';
 827                 # Replace _ as well so we don't get multiple consecutive underscores
 828                 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
 829                 $id = trim( $id, '_' );
 830
 831                 if ( !preg_match( "/^[$nameStartChar]/u", $id )
 832                 && !in_array( 'noninitial', $options ) ) {
 833                         $id = "_$id";
 834                 }
 835
 836                 return $id;
 837         }
 838
 839         /**
 840          * Given a value, escape it so that it can be used as a CSS class and
 841          * return it.
 842          *
 843          * @todo For extra validity, input should be validated UTF-8.
 844          *
 845          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 846          *
 847          * @param string $class
 848          * @return string
 849          */
 850         static function escapeClass( $class ) {
 851                 // Convert ugly stuff to underscores and kill underscores in ugly places
 852                 return rtrim(preg_replace(
 853                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 854                         '_',
 855                         $class ), '_');
 856         }
 857
 858         /**
 859          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 860          * This allows (generally harmless) entities like &nbsp; to survive.
 861          *
 862          * @param  string $html String to escape
 863          * @return string Escaped input
 864          */
 865         static function escapeHtmlAllowEntities( $html ) {
 866                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 867                 # hurt.
 868                 $html = htmlspecialchars( $html, ENT_QUOTES );
 869                 $html = str_replace( '&amp;', '&', $html );
 870                 $html = Sanitizer::normalizeCharReferences( $html );
 871                 return $html;
 872         }
 873
 874         /**
 875          * Regex replace callback for armoring links against further processing.
 876          * @param array $matches
 877          * @return string
 878          * @private
 879          */
 880         private static function armorLinksCallback( $matches ) {
 881                 return str_replace( ':', '&#58;', $matches[1] );
 882         }
 883
 884         /**
 885          * Return an associative array of attribute names and values from
 886          * a partial tag string. Attribute names are forces to lowercase,
 887          * character references are decoded to UTF-8 text.
 888          *
 889          * @param string
 890          * @return array
 891          */
 892         public static function decodeTagAttributes( $text ) {
 893                 $attribs = array();
 894
 895                 if( trim( $text ) == '' ) {
 896                         return $attribs;
 897                 }
 898
 899                 $pairs = array();
 900                 if( !preg_match_all(
 901                         MW_ATTRIBS_REGEX,
 902                         $text,
 903                         $pairs,
 904                         PREG_SET_ORDER ) ) {
 905                         return $attribs;
 906                 }
 907
 908                 foreach( $pairs as $set ) {
 909                         $attribute = strtolower( $set[1] );
 910                         $value = Sanitizer::getTagAttributeCallback( $set );
 911
 912                         // Normalize whitespace
 913                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 914                         $value = trim( $value );
 915
 916                         // Decode character references
 917                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 918                 }
 919                 return $attribs;
 920         }
 921
 922         /**
 923          * Pick the appropriate attribute value from a match set from the
 924          * MW_ATTRIBS_REGEX matches.
 925          *
 926          * @param array $set
 927          * @return string
 928          * @private
 929          */
 930         private static function getTagAttributeCallback( $set ) {
 931                 if( isset( $set[6] ) ) {
 932                         # Illegal #XXXXXX color with no quotes.
 933                         return $set[6];
 934                 } elseif( isset( $set[5] ) ) {
 935                         # No quotes.
 936                         return $set[5];
 937                 } elseif( isset( $set[4] ) ) {
 938                         # Single-quoted
 939                         return $set[4];
 940                 } elseif( isset( $set[3] ) ) {
 941                         # Double-quoted
 942                         return $set[3];
 943                 } elseif( !isset( $set[2] ) ) {
 944                         # In XHTML, attributes must have a value.
 945                         # For 'reduced' form, return explicitly the attribute name here.
 946                         return $set[1];
 947                 } else {
 948                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 949                 }
 950         }
 951
 952         /**
 953          * Normalize whitespace and character references in an XML source-
 954          * encoded text for an attribute value.
 955          *
 956          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 957          * but note that we're not returning the value, but are returning
 958          * XML source fragments that will be slapped into output.
 959          *
 960          * @param string $text
 961          * @return string
 962          * @private
 963          */
 964         private static function normalizeAttributeValue( $text ) {
 965                 return str_replace( '"', '&quot;',
 966                         self::normalizeWhitespace(
 967                                 Sanitizer::normalizeCharReferences( $text ) ) );
 968         }
 969
 970         private static function normalizeWhitespace( $text ) {
 971                 return preg_replace(
 972                         '/\r\n|[\x20\x0d\x0a\x09]/',
 973                         ' ',
 974                         $text );
 975         }
 976
 977         /**
 978          * Ensure that any entities and character references are legal
 979          * for XML and XHTML specifically. Any stray bits will be
 980          * &amp;-escaped to result in a valid text fragment.
 981          *
 982          * a. any named char refs must be known in XHTML
 983          * b. any numeric char refs must be legal chars, not invalid or forbidden
 984          * c. use &#x, not &#X
 985          * d. fix or reject non-valid attributes
 986          *
 987          * @param string $text
 988          * @return string
 989          * @private
 990          */
 991         static function normalizeCharReferences( $text ) {
 992                 return preg_replace_callback(
 993                         MW_CHAR_REFS_REGEX,
 994                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 995                         $text );
 996         }
 997         /**
 998          * @param string $matches
 999          * @return string
1000          */
1001         static function normalizeCharReferencesCallback( $matches ) {
1002                 $ret = null;
1003                 if( $matches[1] != '' ) {
1004                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1005                 } elseif( $matches[2] != '' ) {
1006                         $ret = Sanitizer::decCharReference( $matches[2] );
1007                 } elseif( $matches[3] != ''  ) {
1008                         $ret = Sanitizer::hexCharReference( $matches[3] );
1009                 } elseif( $matches[4] != '' ) {
1010                         $ret = Sanitizer::hexCharReference( $matches[4] );
1011                 }
1012                 if( is_null( $ret ) ) {
1013                         return htmlspecialchars( $matches[0] );
1014                 } else {
1015                         return $ret;
1016                 }
1017         }
1018
1019         /**
1020          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1021          * return the named entity reference as is. If the entity is a
1022          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1023          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1024          *
1025          * @param string $name
1026          * @return string
1027          * @static
1028          */
1029         static function normalizeEntity( $name ) {
1030                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1031                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1032                         return "&{$wgHtmlEntityAliases[$name]};";
1033                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1034                         return "&$name;";
1035                 } else {
1036                         return "&amp;$name;";
1037                 }
1038         }
1039
1040         static function decCharReference( $codepoint ) {
1041                 $point = intval( $codepoint );
1042                 if( Sanitizer::validateCodepoint( $point ) ) {
1043                         return sprintf( '&#%d;', $point );
1044                 } else {
1045                         return null;
1046                 }
1047         }
1048
1049         static function hexCharReference( $codepoint ) {
1050                 $point = hexdec( $codepoint );
1051                 if( Sanitizer::validateCodepoint( $point ) ) {
1052                         return sprintf( '&#x%x;', $point );
1053                 } else {
1054                         return null;
1055                 }
1056         }
1057
1058         /**
1059          * Returns true if a given Unicode codepoint is a valid character in XML.
1060          * @param int $codepoint
1061          * @return bool
1062          */
1063         private static function validateCodepoint( $codepoint ) {
1064                 return ($codepoint ==    0x09)
1065                         || ($codepoint ==    0x0a)
1066                         || ($codepoint ==    0x0d)
1067                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1068                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1069                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1070         }
1071
1072         /**
1073          * Decode any character references, numeric or named entities,
1074          * in the text and return a UTF-8 string.
1075          *
1076          * @param string $text
1077          * @return string
1078          * @public
1079          * @static
1080          */
1081         public static function decodeCharReferences( $text ) {
1082                 return preg_replace_callback(
1083                         MW_CHAR_REFS_REGEX,
1084                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1085                         $text );
1086         }
1087
1088         /**
1089          * @param string $matches
1090          * @return string
1091          */
1092         static function decodeCharReferencesCallback( $matches ) {
1093                 if( $matches[1] != '' ) {
1094                         return Sanitizer::decodeEntity( $matches[1] );
1095                 } elseif( $matches[2] != '' ) {
1096                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1097                 } elseif( $matches[3] != ''  ) {
1098                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1099                 } elseif( $matches[4] != '' ) {
1100                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1101                 }
1102                 # Last case should be an ampersand by itself
1103                 return $matches[0];
1104         }
1105
1106         /**
1107          * Return UTF-8 string for a codepoint if that is a valid
1108          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1109          * @param int $codepoint
1110          * @return string
1111          * @private
1112          */
1113         static function decodeChar( $codepoint ) {
1114                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1115                         return codepointToUtf8( $codepoint );
1116                 } else {
1117                         return UTF8_REPLACEMENT;
1118                 }
1119         }
1120
1121         /**
1122          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1123          * return the UTF-8 encoding of that character. Otherwise, returns
1124          * pseudo-entity source (eg &foo;)
1125          *
1126          * @param string $name
1127          * @return string
1128          */
1129         static function decodeEntity( $name ) {
1130                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1131                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1132                         $name = $wgHtmlEntityAliases[$name];
1133                 }
1134                 if( isset( $wgHtmlEntities[$name] ) ) {
1135                         return codepointToUtf8( $wgHtmlEntities[$name] );
1136                 } else {
1137                         return "&$name;";
1138                 }
1139         }
1140
1141         /**
1142          * Fetch the whitelist of acceptable attributes for a given
1143          * element name.
1144          *
1145          * @param string $element
1146          * @return array
1147          */
1148         static function attributeWhitelist( $element ) {
1149                 static $list;
1150                 if( !isset( $list ) ) {
1151                         $list = Sanitizer::setupAttributeWhitelist();
1152                 }
1153                 return isset( $list[$element] )
1154                         ? $list[$element]
1155                         : array();
1156         }
1157
1158         /**
1159          * Foreach array key (an allowed HTML element), return an array
1160          * of allowed attributes
1161          * @return array
1162          */
1163         static function setupAttributeWhitelist() {
1164                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1165                 $block = array_merge( $common, array( 'align' ) );
1166                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1167                 $tablecell = array( 'abbr',
1168                                     'axis',
1169                                     'headers',
1170                                     'scope',
1171                                     'rowspan',
1172                                     'colspan',
1173                                     'nowrap', # deprecated
1174                                     'width',  # deprecated
1175                                     'height', # deprecated
1176                                     'bgcolor' # deprecated
1177                                     );
1178
1179                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1180                 # See: http://www.w3.org/TR/html4/
1181                 $whitelist = array (
1182                         # 7.5.4
1183                         'div'        => $block,
1184                         'center'     => $common, # deprecated
1185                         'span'       => $block, # ??
1186
1187                         # 7.5.5
1188                         'h1'         => $block,
1189                         'h2'         => $block,
1190                         'h3'         => $block,
1191                         'h4'         => $block,
1192                         'h5'         => $block,
1193                         'h6'         => $block,
1194
1195                         # 7.5.6
1196                         # address
1197
1198                         # 8.2.4
1199                         # bdo
1200
1201                         # 9.2.1
1202                         'em'         => $common,
1203                         'strong'     => $common,
1204                         'cite'       => $common,
1205                         # dfn
1206                         'code'       => $common,
1207                         # samp
1208                         # kbd
1209                         'var'        => $common,
1210                         # abbr
1211                         # acronym
1212
1213                         # 9.2.2
1214                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1215                         # q
1216
1217                         # 9.2.3
1218                         'sub'        => $common,
1219                         'sup'        => $common,
1220
1221                         # 9.3.1
1222                         'p'          => $block,
1223
1224                         # 9.3.2
1225                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1226
1227                         # 9.3.4
1228                         'pre'        => array_merge( $common, array( 'width' ) ),
1229
1230                         # 9.4
1231                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1232                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1233
1234                         # 10.2
1235                         'ul'         => array_merge( $common, array( 'type' ) ),
1236                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1237                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1238
1239                         # 10.3
1240                         'dl'         => $common,
1241                         'dd'         => $common,
1242                         'dt'         => $common,
1243
1244                         # 11.2.1
1245                         'table'      => array_merge( $common,
1246                                                                 array( 'summary', 'width', 'border', 'frame',
1247                                                                                 'rules', 'cellspacing', 'cellpadding',
1248                                                                                 'align', 'bgcolor',
1249                                                                 ) ),
1250
1251                         # 11.2.2
1252                         'caption'    => array_merge( $common, array( 'align' ) ),
1253
1254                         # 11.2.3
1255                         'thead'      => array_merge( $common, $tablealign ),
1256                         'tfoot'      => array_merge( $common, $tablealign ),
1257                         'tbody'      => array_merge( $common, $tablealign ),
1258
1259                         # 11.2.4
1260                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1261                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1262
1263                         # 11.2.5
1264                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1265
1266                         # 11.2.6
1267                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1268                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1269
1270                         # 13.2
1271                         # Not usually allowed, but may be used for extension-style hooks
1272                         # such as <math> when it is rasterized
1273                         'img'        => array_merge( $common, array( 'alt' ) ),
1274
1275                         # 15.2.1
1276                         'tt'         => $common,
1277                         'b'          => $common,
1278                         'i'          => $common,
1279                         'big'        => $common,
1280                         'small'      => $common,
1281                         'strike'     => $common,
1282                         's'          => $common,
1283                         'u'          => $common,
1284
1285                         # 15.2.2
1286                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1287                         # basefont
1288
1289                         # 15.3
1290                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1291
1292                         # XHTML Ruby annotation text module, simple ruby only.
1293                         # http://www.w3c.org/TR/ruby/
1294                         'ruby'       => $common,
1295                         # rbc
1296                         # rtc
1297                         'rb'         => $common,
1298                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1299                         'rp'         => $common,
1300
1301                         # MathML root element, where used for extensions
1302                         # 'title' may not be 100% valid here; it's XHTML
1303                         # http://www.w3.org/TR/REC-MathML/
1304                         'math'       => array( 'class', 'style', 'id', 'title' ),
1305                         );
1306                 return $whitelist;
1307         }
1308
1309         /**
1310          * Take a fragment of (potentially invalid) HTML and return
1311          * a version with any tags removed, encoded as plain text.
1312          *
1313          * Warning: this return value must be further escaped for literal
1314          * inclusion in HTML output as of 1.10!
1315          *
1316          * @param string $text HTML fragment
1317          * @return string
1318          */
1319         static function stripAllTags( $text ) {
1320                 # Actual <tags>
1321                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1322
1323                 # Normalize &entities and whitespace
1324                 $text = self::decodeCharReferences( $text );
1325                 $text = self::normalizeWhitespace( $text );
1326
1327                 return $text;
1328         }
1329
1330         /**
1331          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1332          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1333          * PHP 5.1 doesn't.
1334          *
1335          * Use for passing XHTML fragments to PHP's XML parsing functions
1336          *
1337          * @return string
1338          * @static
1339          */
1340         static function hackDocType() {
1341                 global $wgHtmlEntities;
1342                 $out = "<!DOCTYPE html [\n";
1343                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1344                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1345                 }
1346                 $out .= "]>\n";
1347                 return $out;
1348         }
1349
1350         static function cleanUrl( $url ) {
1351                 # Normalize any HTML entities in input. They will be
1352                 # re-escaped by makeExternalLink().
1353                 $url = Sanitizer::decodeCharReferences( $url );
1354
1355                 # Escape any control characters introduced by the above step
1356                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1357
1358                 # Validate hostname portion
1359                 $matches = array();
1360                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1361                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1362
1363                         // Characters that will be ignored in IDNs.
1364                         // http://tools.ietf.org/html/3454#section-3.1
1365                         // Strip them before further processing so blacklists and such work.
1366                         $strip = "/
1367                                 \\s|          # general whitespace
1368                                 \xc2\xad|     # 00ad SOFT HYPHEN
1369                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1370                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1371                                 \xe2\x81\xa0| # 2060 WORD JOINER
1372                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1373                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1374                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1375                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1376                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1377                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1378                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1379                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1380                                 /xuD";
1381
1382                         $host = preg_replace( $strip, '', $host );
1383
1384                         // @fixme: validate hostnames here
1385
1386                         return $protocol . $host . $rest;
1387                 } else {
1388                         return $url;
1389                 }
1390         }
1391
1392 }