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)((?:xml:|xmlns:)?$attrib+)
  47           ($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            )?(?=$space|\$)/sx" );
  58
  59 /**
  60  * Regular expression to match URIs that could trigger script execution
  61  */
  62 define( 'MW_EVIL_URI_PATTERN', '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i' );
  63
  64 /**
  65  * Regular expression to match namespace attributes
  66  */
  67 define( 'MW_XMLNS_ATTRIBUTE_PATTRN', "/^xmlns:$attrib+$/" );
  68
  69 /**
  70  * List of all named character entities defined in HTML 4.01
  71  * http://www.w3.org/TR/html4/sgml/entities.html
  72  * @private
  73  */
  74 global $wgHtmlEntities;
  75 $wgHtmlEntities = array(
  76         'Aacute'   => 193,
  77         'aacute'   => 225,
  78         'Acirc'    => 194,
  79         'acirc'    => 226,
  80         'acute'    => 180,
  81         'AElig'    => 198,
  82         'aelig'    => 230,
  83         'Agrave'   => 192,
  84         'agrave'   => 224,
  85         'alefsym'  => 8501,
  86         'Alpha'    => 913,
  87         'alpha'    => 945,
  88         'amp'      => 38,
  89         'and'      => 8743,
  90         'ang'      => 8736,
  91         'Aring'    => 197,
  92         'aring'    => 229,
  93         'asymp'    => 8776,
  94         'Atilde'   => 195,
  95         'atilde'   => 227,
  96         'Auml'     => 196,
  97         'auml'     => 228,
  98         'bdquo'    => 8222,
  99         'Beta'     => 914,
 100         'beta'     => 946,
 101         'brvbar'   => 166,
 102         'bull'     => 8226,
 103         'cap'      => 8745,
 104         'Ccedil'   => 199,
 105         'ccedil'   => 231,
 106         'cedil'    => 184,
 107         'cent'     => 162,
 108         'Chi'      => 935,
 109         'chi'      => 967,
 110         'circ'     => 710,
 111         'clubs'    => 9827,
 112         'cong'     => 8773,
 113         'copy'     => 169,
 114         'crarr'    => 8629,
 115         'cup'      => 8746,
 116         'curren'   => 164,
 117         'dagger'   => 8224,
 118         'Dagger'   => 8225,
 119         'darr'     => 8595,
 120         'dArr'     => 8659,
 121         'deg'      => 176,
 122         'Delta'    => 916,
 123         'delta'    => 948,
 124         'diams'    => 9830,
 125         'divide'   => 247,
 126         'Eacute'   => 201,
 127         'eacute'   => 233,
 128         'Ecirc'    => 202,
 129         'ecirc'    => 234,
 130         'Egrave'   => 200,
 131         'egrave'   => 232,
 132         'empty'    => 8709,
 133         'emsp'     => 8195,
 134         'ensp'     => 8194,
 135         'Epsilon'  => 917,
 136         'epsilon'  => 949,
 137         'equiv'    => 8801,
 138         'Eta'      => 919,
 139         'eta'      => 951,
 140         'ETH'      => 208,
 141         'eth'      => 240,
 142         'Euml'     => 203,
 143         'euml'     => 235,
 144         'euro'     => 8364,
 145         'exist'    => 8707,
 146         'fnof'     => 402,
 147         'forall'   => 8704,
 148         'frac12'   => 189,
 149         'frac14'   => 188,
 150         'frac34'   => 190,
 151         'frasl'    => 8260,
 152         'Gamma'    => 915,
 153         'gamma'    => 947,
 154         'ge'       => 8805,
 155         'gt'       => 62,
 156         'harr'     => 8596,
 157         'hArr'     => 8660,
 158         'hearts'   => 9829,
 159         'hellip'   => 8230,
 160         'Iacute'   => 205,
 161         'iacute'   => 237,
 162         'Icirc'    => 206,
 163         'icirc'    => 238,
 164         'iexcl'    => 161,
 165         'Igrave'   => 204,
 166         'igrave'   => 236,
 167         'image'    => 8465,
 168         'infin'    => 8734,
 169         'int'      => 8747,
 170         'Iota'     => 921,
 171         'iota'     => 953,
 172         'iquest'   => 191,
 173         'isin'     => 8712,
 174         'Iuml'     => 207,
 175         'iuml'     => 239,
 176         'Kappa'    => 922,
 177         'kappa'    => 954,
 178         'Lambda'   => 923,
 179         'lambda'   => 955,
 180         'lang'     => 9001,
 181         'laquo'    => 171,
 182         'larr'     => 8592,
 183         'lArr'     => 8656,
 184         'lceil'    => 8968,
 185         'ldquo'    => 8220,
 186         'le'       => 8804,
 187         'lfloor'   => 8970,
 188         'lowast'   => 8727,
 189         'loz'      => 9674,
 190         'lrm'      => 8206,
 191         'lsaquo'   => 8249,
 192         'lsquo'    => 8216,
 193         'lt'       => 60,
 194         'macr'     => 175,
 195         'mdash'    => 8212,
 196         'micro'    => 181,
 197         'middot'   => 183,
 198         'minus'    => 8722,
 199         'Mu'       => 924,
 200         'mu'       => 956,
 201         'nabla'    => 8711,
 202         'nbsp'     => 160,
 203         'ndash'    => 8211,
 204         'ne'       => 8800,
 205         'ni'       => 8715,
 206         'not'      => 172,
 207         'notin'    => 8713,
 208         'nsub'     => 8836,
 209         'Ntilde'   => 209,
 210         'ntilde'   => 241,
 211         'Nu'       => 925,
 212         'nu'       => 957,
 213         'Oacute'   => 211,
 214         'oacute'   => 243,
 215         'Ocirc'    => 212,
 216         'ocirc'    => 244,
 217         'OElig'    => 338,
 218         'oelig'    => 339,
 219         'Ograve'   => 210,
 220         'ograve'   => 242,
 221         'oline'    => 8254,
 222         'Omega'    => 937,
 223         'omega'    => 969,
 224         'Omicron'  => 927,
 225         'omicron'  => 959,
 226         'oplus'    => 8853,
 227         'or'       => 8744,
 228         'ordf'     => 170,
 229         'ordm'     => 186,
 230         'Oslash'   => 216,
 231         'oslash'   => 248,
 232         'Otilde'   => 213,
 233         'otilde'   => 245,
 234         'otimes'   => 8855,
 235         'Ouml'     => 214,
 236         'ouml'     => 246,
 237         'para'     => 182,
 238         'part'     => 8706,
 239         'permil'   => 8240,
 240         'perp'     => 8869,
 241         'Phi'      => 934,
 242         'phi'      => 966,
 243         'Pi'       => 928,
 244         'pi'       => 960,
 245         'piv'      => 982,
 246         'plusmn'   => 177,
 247         'pound'    => 163,
 248         'prime'    => 8242,
 249         'Prime'    => 8243,
 250         'prod'     => 8719,
 251         'prop'     => 8733,
 252         'Psi'      => 936,
 253         'psi'      => 968,
 254         'quot'     => 34,
 255         'radic'    => 8730,
 256         'rang'     => 9002,
 257         'raquo'    => 187,
 258         'rarr'     => 8594,
 259         'rArr'     => 8658,
 260         'rceil'    => 8969,
 261         'rdquo'    => 8221,
 262         'real'     => 8476,
 263         'reg'      => 174,
 264         'rfloor'   => 8971,
 265         'Rho'      => 929,
 266         'rho'      => 961,
 267         'rlm'      => 8207,
 268         'rsaquo'   => 8250,
 269         'rsquo'    => 8217,
 270         'sbquo'    => 8218,
 271         'Scaron'   => 352,
 272         'scaron'   => 353,
 273         'sdot'     => 8901,
 274         'sect'     => 167,
 275         'shy'      => 173,
 276         'Sigma'    => 931,
 277         'sigma'    => 963,
 278         'sigmaf'   => 962,
 279         'sim'      => 8764,
 280         'spades'   => 9824,
 281         'sub'      => 8834,
 282         'sube'     => 8838,
 283         'sum'      => 8721,
 284         'sup'      => 8835,
 285         'sup1'     => 185,
 286         'sup2'     => 178,
 287         'sup3'     => 179,
 288         'supe'     => 8839,
 289         'szlig'    => 223,
 290         'Tau'      => 932,
 291         'tau'      => 964,
 292         'there4'   => 8756,
 293         'Theta'    => 920,
 294         'theta'    => 952,
 295         'thetasym' => 977,
 296         'thinsp'   => 8201,
 297         'THORN'    => 222,
 298         'thorn'    => 254,
 299         'tilde'    => 732,
 300         'times'    => 215,
 301         'trade'    => 8482,
 302         'Uacute'   => 218,
 303         'uacute'   => 250,
 304         'uarr'     => 8593,
 305         'uArr'     => 8657,
 306         'Ucirc'    => 219,
 307         'ucirc'    => 251,
 308         'Ugrave'   => 217,
 309         'ugrave'   => 249,
 310         'uml'      => 168,
 311         'upsih'    => 978,
 312         'Upsilon'  => 933,
 313         'upsilon'  => 965,
 314         'Uuml'     => 220,
 315         'uuml'     => 252,
 316         'weierp'   => 8472,
 317         'Xi'       => 926,
 318         'xi'       => 958,
 319         'Yacute'   => 221,
 320         'yacute'   => 253,
 321         'yen'      => 165,
 322         'Yuml'     => 376,
 323         'yuml'     => 255,
 324         'Zeta'     => 918,
 325         'zeta'     => 950,
 326         'zwj'      => 8205,
 327         'zwnj'     => 8204 );
 328
 329 /**
 330  * Character entity aliases accepted by MediaWiki
 331  */
 332 global $wgHtmlEntityAliases;
 333 $wgHtmlEntityAliases = array(
 334         'רלמ' => 'rlm',
 335         'رلم' => 'rlm',
 336 );
 337
 338
 339 /**
 340  * XHTML sanitizer for MediaWiki
 341  * @ingroup Parser
 342  */
 343 class Sanitizer {
 344         /**
 345          * Cleans up HTML, removes dangerous tags and attributes, and
 346          * removes HTML comments
 347          * @private
 348          * @param $text String
 349          * @param $processCallback Callback to do any variable or parameter replacements in HTML attribute values
 350          * @param $args Array for the processing callback
 351          * @param $extratags Array for any extra tags to include
 352          * @param $removetags Array for any tags (default or extra) to exclude
 353          * @return string
 354          */
 355         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array(), $removetags = array() ) {
 356                 global $wgUseTidy;
 357
 358                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 359                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 360
 361                 wfProfileIn( __METHOD__ );
 362
 363                 if ( !$staticInitialised ) {
 364
 365                         $htmlpairsStatic = array( # Tags that must be closed
 366                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 367                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 368                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 369                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 370                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u', 'abbr'
 371                         );
 372                         $htmlsingle = array(
 373                                 'br', 'hr', 'li', 'dt', 'dd'
 374                         );
 375                         $htmlsingleonly = array( # Elements that cannot have close tags
 376                                 'br', 'hr'
 377                         );
 378                         $htmlnest = array( # Tags that can be nested--??
 379                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 380                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 381                         );
 382                         $tabletags = array( # Can only appear inside table, we will close them
 383                                 'td', 'th', 'tr',
 384                         );
 385                         $htmllist = array( # Tags used by list
 386                                 'ul','ol',
 387                         );
 388                         $listtags = array( # Tags that can appear in a list
 389                                 'li',
 390                         );
 391
 392                         $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
 393                         $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
 394
 395                         # Convert them all to hashtables for faster lookup
 396                         $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 397                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
 398                         foreach ( $vars as $var ) {
 399                                 $$var = array_flip( $$var );
 400                         }
 401                         $staticInitialised = true;
 402                 }
 403                 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
 404                 $extratags = array_flip( $extratags );
 405                 $removetags = array_flip( $removetags );
 406                 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
 407                 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ) , $removetags );
 408
 409                 # Remove HTML comments
 410                 $text = Sanitizer::removeHTMLcomments( $text );
 411                 $bits = explode( '<', $text );
 412                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 413                 if(!$wgUseTidy) {
 414                         $tagstack = $tablestack = array();
 415                         foreach ( $bits as $x ) {
 416                                 $regs = array();
 417                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 418                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 419                                 } else {
 420                                         $slash = $t = $params = $brace = $rest = null;
 421                                 }
 422
 423                                 $badtag = 0 ;
 424                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 425                                         # Check our stack
 426                                         if ( $slash ) {
 427                                                 # Closing a tag...
 428                                                 if( isset( $htmlsingleonly[$t] ) ) {
 429                                                         $badtag = 1;
 430                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 431                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 432                                                                 # Pop all elements with an optional close tag
 433                                                                 # and see if we find a match below them
 434                                                                 $optstack = array();
 435                                                                 array_push ($optstack, $ot);
 436                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 437                                                                                 isset( $htmlsingleallowed[$ot] ) )
 438                                                                 {
 439                                                                         array_push ($optstack, $ot);
 440                                                                 }
 441                                                                 if ( $t != $ot ) {
 442                                                                         # No match. Push the optinal elements back again
 443                                                                         $badtag = 1;
 444                                                                         while ( $ot = @array_pop( $optstack ) ) {
 445                                                                                 array_push( $tagstack, $ot );
 446                                                                         }
 447                                                                 }
 448                                                         } else {
 449                                                                 @array_push( $tagstack, $ot );
 450                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 451                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 452                                                                         $badtag = 1;
 453                                                                 }
 454                                                         }
 455                                                 } else {
 456                                                         if ( $t == 'table' ) {
 457                                                                 $tagstack = array_pop( $tablestack );
 458                                                         }
 459                                                 }
 460                                                 $newparams = '';
 461                                         } else {
 462                                                 # Keep track for later
 463                                                 if ( isset( $tabletags[$t] ) &&
 464                                                 ! in_array( 'table', $tagstack ) ) {
 465                                                         $badtag = 1;
 466                                                 } else if ( in_array( $t, $tagstack ) &&
 467                                                 ! isset( $htmlnest [$t ] ) ) {
 468                                                         $badtag = 1 ;
 469                                                 # Is it a self closed htmlpair ? (bug 5487)
 470                                                 } else if( $brace == '/>' &&
 471                                                 isset( $htmlpairs[$t] ) ) {
 472                                                         $badtag = 1;
 473                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 474                                                         # Hack to force empty tag for uncloseable elements
 475                                                         $brace = '/>';
 476                                                 } else if( isset( $htmlsingle[$t] ) ) {
 477                                                         # Hack to not close $htmlsingle tags
 478                                                         $brace = NULL;
 479                                                 } else if( isset( $tabletags[$t] )
 480                                                 &&  in_array($t ,$tagstack) ) {
 481                                                         // New table tag but forgot to close the previous one
 482                                                         $text .= "</$t>";
 483                                                 } else {
 484                                                         if ( $t == 'table' ) {
 485                                                                 array_push( $tablestack, $tagstack );
 486                                                                 $tagstack = array();
 487                                                         }
 488                                                         array_push( $tagstack, $t );
 489                                                 }
 490
 491                                                 # Replace any variables or template parameters with
 492                                                 # plaintext results.
 493                                                 if( is_callable( $processCallback ) ) {
 494                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 495                                                 }
 496
 497                                                 # Strip non-approved attributes from the tag
 498                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 499                                         }
 500                                         if ( ! $badtag ) {
 501                                                 $rest = str_replace( '>', '&gt;', $rest );
 502                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 503                                                 $text .= "<$slash$t$newparams$close>$rest";
 504                                                 continue;
 505                                         }
 506                                 }
 507                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 508                         }
 509                         # Close off any remaining tags
 510                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 511                                 $text .= "</$t>\n";
 512                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 513                         }
 514                 } else {
 515                         # this might be possible using tidy itself
 516                         foreach ( $bits as $x ) {
 517                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 518                                 $x, $regs );
 519                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 520                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 521                                         if( is_callable( $processCallback ) ) {
 522                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 523                                         }
 524                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 525                                         $rest = str_replace( '>', '&gt;', $rest );
 526                                         $text .= "<$slash$t$newparams$brace$rest";
 527                                 } else {
 528                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 529                                 }
 530                         }
 531                 }
 532                 wfProfileOut( __METHOD__ );
 533                 return $text;
 534         }
 535
 536         /**
 537          * Remove '<!--', '-->', and everything between.
 538          * To avoid leaving blank lines, when a comment is both preceded
 539          * and followed by a newline (ignoring spaces), trim leading and
 540          * trailing spaces and one of the newlines.
 541          *
 542          * @private
 543          * @param $text String
 544          * @return string
 545          */
 546         static function removeHTMLcomments( $text ) {
 547                 wfProfileIn( __METHOD__ );
 548                 while (($start = strpos($text, '<!--')) !== false) {
 549                         $end = strpos($text, '-->', $start + 4);
 550                         if ($end === false) {
 551                                 # Unterminated comment; bail out
 552                                 break;
 553                         }
 554
 555                         $end += 3;
 556
 557                         # Trim space and newline if the comment is both
 558                         # preceded and followed by a newline
 559                         $spaceStart = max($start - 1, 0);
 560                         $spaceLen = $end - $spaceStart;
 561                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 562                                 $spaceStart--;
 563                                 $spaceLen++;
 564                         }
 565                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 566                                 $spaceLen++;
 567                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 568                                 # Remove the comment, leading and trailing
 569                                 # spaces, and leave only one newline.
 570                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 571                         }
 572                         else {
 573                                 # Remove just the comment.
 574                                 $text = substr_replace($text, '', $start, $end - $start);
 575                         }
 576                 }
 577                 wfProfileOut( __METHOD__ );
 578                 return $text;
 579         }
 580
 581         /**
 582          * Take an array of attribute names and values and normalize or discard
 583          * illegal values for the given element type.
 584          *
 585          * - Discards attributes not on a whitelist for the given element
 586          * - Unsafe style attributes are discarded
 587          * - Invalid id attributes are reencoded
 588          *
 589          * @param $attribs Array
 590          * @param $element String
 591          * @return Array
 592          *
 593          * @todo Check for legal values where the DTD limits things.
 594          * @todo Check for unique id attribute :P
 595          */
 596         static function validateTagAttributes( $attribs, $element ) {
 597                 return Sanitizer::validateAttributes( $attribs,
 598                         Sanitizer::attributeWhitelist( $element ) );
 599         }
 600
 601         /**
 602          * Take an array of attribute names and values and normalize or discard
 603          * illegal values for the given whitelist.
 604          *
 605          * - Discards attributes not the given whitelist
 606          * - Unsafe style attributes are discarded
 607          * - Invalid id attributes are reencoded
 608          *
 609          * @param $attribs Array
 610          * @param $whitelist Array: list of allowed attribute names
 611          * @return Array
 612          *
 613          * @todo Check for legal values where the DTD limits things.
 614          * @todo Check for unique id attribute :P
 615          */
 616         static function validateAttributes( $attribs, $whitelist ) {
 617                 $whitelist = array_flip( $whitelist );
 618                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 619
 620                 $out = array();
 621                 foreach( $attribs as $attribute => $value ) {
 622                         #allow XML namespace declaration. Useful especially with RDFa
 623                         print "($attribute=$value)";
 624
 625                         if ( preg_match( MW_XMLNS_ATTRIBUTE_PATTRN, $attribute ) ) {
 626                                 if ( !preg_match( MW_EVIL_URI_PATTERN, $value ) ) {
 627                                         $out[$attribute] = $value;
 628                                 }
 629
 630                                 continue;
 631                         }
 632
 633                         if( !isset( $whitelist[$attribute] ) ) {
 634                                 continue;
 635                         }
 636
 637                         # Strip javascript "expression" from stylesheets.
 638                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 639                         if( $attribute == 'style' ) {
 640                                 $value = Sanitizer::checkCss( $value );
 641                                 if( $value === false ) {
 642                                         # haxx0r
 643                                         continue;
 644                                 }
 645                         }
 646
 647                         if ( $attribute === 'id' ) {
 648                                 global $wgEnforceHtmlIds;
 649                                 $value = Sanitizer::escapeId( $value,
 650                                         $wgEnforceHtmlIds ? 'noninitial' : 'xml' );
 651                         }
 652
 653                         //RDFa and microdata properties allow URIs. check them
 654                         if ( $attribute === 'rel' || $attribute === 'rev' ||
 655                                 $attribute === 'about' || $attribute === 'property' || $attribute === 'resource' ||
 656                                 $attribute === 'datatype' || $attribute === 'typeof' ||
 657                                 $attribute === 'item' || $attribute === 'itemprop' || $attribute === 'subject' ) {
 658
 659                                 //Paranoia. Allow "simple" values but suppress javascript
 660                                 if ( preg_match( MW_EVIL_URI_PATTERN, $value ) ) {
 661                                         continue;
 662                                 }
 663                         }
 664
 665                         # NOTE: even though elements using href/src are not allowed directly, supply
 666                         #       validation code that can be used by tag hook handlers, etc
 667                         if ( $attribute === 'href' || $attribute === 'src' ) {
 668                                 if ( !preg_match( $hrefExp, $value ) ) {
 669                                         continue; //drop any href or src attributes not using an allowed protocol.
 670                                                   //NOTE: this also drops all relative URLs
 671                                 }
 672                         }
 673
 674                         // If this attribute was previously set, override it.
 675                         // Output should only have one attribute of each name.
 676                         $out[$attribute] = $value;
 677                 }
 678                 return $out;
 679         }
 680
 681         /**
 682          * Merge two sets of HTML attributes.  Conflicting items in the second set
 683          * will override those in the first, except for 'class' attributes which
 684          * will be combined (if they're both strings).
 685          *
 686          * @todo implement merging for other attributes such as style
 687          * @param $a Array
 688          * @param $b Array
 689          * @return array
 690          */
 691         static function mergeAttributes( $a, $b ) {
 692                 $out = array_merge( $a, $b );
 693                 if( isset( $a['class'] ) && isset( $b['class'] )
 694                 && is_string( $a['class'] ) && is_string( $b['class'] )
 695                 && $a['class'] !== $b['class'] ) {
 696                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 697                                 -1, PREG_SPLIT_NO_EMPTY );
 698                         $out['class'] = implode( ' ', array_unique( $classes ) );
 699                 }
 700                 return $out;
 701         }
 702
 703         /**
 704          * Pick apart some CSS and check it for forbidden or unsafe structures.
 705          * Returns a sanitized string, or false if it was just too evil.
 706          *
 707          * Currently URL references, 'expression', 'tps' are forbidden.
 708          *
 709          * @param $value String
 710          * @return Mixed
 711          */
 712         static function checkCss( $value ) {
 713                 $stripped = Sanitizer::decodeCharReferences( $value );
 714
 715                 // Remove any comments; IE gets token splitting wrong
 716                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 717
 718                 $value = $stripped;
 719
 720                 // ... and continue checks
 721                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 722                         'codepointToUtf8(hexdec("$1"))', $stripped );
 723                 $stripped = str_replace( '\\', '', $stripped );
 724                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 725                                 $stripped ) ) {
 726                         # haxx0r
 727                         return false;
 728                 }
 729
 730                 return $value;
 731         }
 732
 733         /**
 734          * Take a tag soup fragment listing an HTML element's attributes
 735          * and normalize it to well-formed XML, discarding unwanted attributes.
 736          * Output is safe for further wikitext processing, with escaping of
 737          * values that could trigger problems.
 738          *
 739          * - Normalizes attribute names to lowercase
 740          * - Discards attributes not on a whitelist for the given element
 741          * - Turns broken or invalid entities into plaintext
 742          * - Double-quotes all attribute values
 743          * - Attributes without values are given the name as attribute
 744          * - Double attributes are discarded
 745          * - Unsafe style attributes are discarded
 746          * - Prepends space if there are attributes.
 747          *
 748          * @param $text String
 749          * @param $element String
 750          * @return String
 751          */
 752         static function fixTagAttributes( $text, $element ) {
 753                 if( trim( $text ) == '' ) {
 754                         return '';
 755                 }
 756
 757                 $stripped = Sanitizer::validateTagAttributes(
 758                         Sanitizer::decodeTagAttributes( $text ), $element );
 759
 760                 $attribs = array();
 761                 foreach( $stripped as $attribute => $value ) {
 762                         $encAttribute = htmlspecialchars( $attribute );
 763                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 764
 765                         $attribs[] = "$encAttribute=\"$encValue\"";
 766                 }
 767                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 768         }
 769
 770         /**
 771          * Encode an attribute value for HTML output.
 772          * @param $text String
 773          * @return HTML-encoded text fragment
 774          */
 775         static function encodeAttribute( $text ) {
 776                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 777
 778                 // Whitespace is normalized during attribute decoding,
 779                 // so if we've been passed non-spaces we must encode them
 780                 // ahead of time or they won't be preserved.
 781                 $encValue = strtr( $encValue, array(
 782                         "\n" => '&#10;',
 783                         "\r" => '&#13;',
 784                         "\t" => '&#9;',
 785                 ) );
 786
 787                 return $encValue;
 788         }
 789
 790         /**
 791          * Encode an attribute value for HTML tags, with extra armoring
 792          * against further wiki processing.
 793          * @param $text String
 794          * @return HTML-encoded text fragment
 795          */
 796         static function safeEncodeAttribute( $text ) {
 797                 $encValue = Sanitizer::encodeAttribute( $text );
 798
 799                 # Templates and links may be expanded in later parsing,
 800                 # creating invalid or dangerous output. Suppress this.
 801                 $encValue = strtr( $encValue, array(
 802                         '<'    => '&lt;',   // This should never happen,
 803                         '>'    => '&gt;',   // we've received invalid input
 804                         '"'    => '&quot;', // which should have been escaped.
 805                         '{'    => '&#123;',
 806                         '['    => '&#91;',
 807                         "''"   => '&#39;&#39;',
 808                         'ISBN' => '&#73;SBN',
 809                         'RFC'  => '&#82;FC',
 810                         'PMID' => '&#80;MID',
 811                         '|'    => '&#124;',
 812                         '__'   => '&#95;_',
 813                 ) );
 814
 815                 # Stupid hack
 816                 $encValue = preg_replace_callback(
 817                         '/(' . wfUrlProtocols() . ')/',
 818                         array( 'Sanitizer', 'armorLinksCallback' ),
 819                         $encValue );
 820                 return $encValue;
 821         }
 822
 823         /**
 824          * Given a value escape it so that it can be used in an id attribute and
 825          * return it, this does not validate the value however (see first link)
 826          *
 827          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 828          *                                                          in the id and
 829          *                                                          name attributes
 830          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 831          *
 832          * @param $id String: id to validate
 833          * @param $options Mixed: string or array of strings (default is array()):
 834          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 835          *       so don't pay attention if the first character isn't valid at the
 836          *       beginning of an id.
 837          *   'xml': Don't restrict the id to be HTML4-compatible.  This option
 838          *       allows any alphabetic character to be used, per the XML standard.
 839          *       Therefore, it also completely changes the type of escaping: instead
 840          *       of weird dot-encoding, runs of invalid characters (mostly
 841          *       whitespace) are just compressed into a single underscore.
 842          * @return String
 843          */
 844         static function escapeId( $id, $options = array() ) {
 845                 $options = (array)$options;
 846
 847                 if ( !in_array( 'xml', $options ) ) {
 848                         # HTML4-style escaping
 849                         static $replace = array(
 850                                 '%3A' => ':',
 851                                 '%' => '.'
 852                         );
 853
 854                         $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 855                         $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 856
 857                         if ( !preg_match( '/^[a-zA-Z]/', $id )
 858                         && !in_array( 'noninitial', $options ) )  {
 859                                 // Initial character must be a letter!
 860                                 $id = "x$id";
 861                         }
 862                         return $id;
 863                 }
 864
 865                 # XML-style escaping.  For the patterns used, see the XML 1.0 standard,
 866                 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
 867                 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
 868                         . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
 869                         . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
 870                 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
 871                         . '\x{203F}-\x{2040}';
 872                 # Replace _ as well so we don't get multiple consecutive underscores
 873                 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
 874                 $id = trim( $id, '_' );
 875
 876                 if ( !preg_match( "/^[$nameStartChar]/u", $id )
 877                 && !in_array( 'noninitial', $options ) ) {
 878                         $id = "_$id";
 879                 }
 880
 881                 return $id;
 882         }
 883
 884         /**
 885          * Given a value, escape it so that it can be used as a CSS class and
 886          * return it.
 887          *
 888          * @todo For extra validity, input should be validated UTF-8.
 889          *
 890          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 891          *
 892          * @param $class String
 893          * @return String
 894          */
 895         static function escapeClass( $class ) {
 896                 // Convert ugly stuff to underscores and kill underscores in ugly places
 897                 return rtrim(preg_replace(
 898                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 899                         '_',
 900                         $class ), '_');
 901         }
 902
 903         /**
 904          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 905          * This allows (generally harmless) entities like &nbsp; to survive.
 906          *
 907          * @param $html String to escape
 908          * @return String: escaped input
 909          */
 910         static function escapeHtmlAllowEntities( $html ) {
 911                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 912                 # hurt.
 913                 $html = htmlspecialchars( $html, ENT_QUOTES );
 914                 $html = str_replace( '&amp;', '&', $html );
 915                 $html = Sanitizer::normalizeCharReferences( $html );
 916                 return $html;
 917         }
 918
 919         /**
 920          * Regex replace callback for armoring links against further processing.
 921          * @param $matches Array
 922          * @return string
 923          */
 924         private static function armorLinksCallback( $matches ) {
 925                 return str_replace( ':', '&#58;', $matches[1] );
 926         }
 927
 928         /**
 929          * Return an associative array of attribute names and values from
 930          * a partial tag string. Attribute names are forces to lowercase,
 931          * character references are decoded to UTF-8 text.
 932          *
 933          * @param $text String
 934          * @return Array
 935          */
 936         public static function decodeTagAttributes( $text ) {
 937                 $attribs = array();
 938
 939                 if( trim( $text ) == '' ) {
 940                         return $attribs;
 941                 }
 942
 943                 $pairs = array();
 944                 if( !preg_match_all(
 945                         MW_ATTRIBS_REGEX,
 946                         $text,
 947                         $pairs,
 948                         PREG_SET_ORDER ) ) {
 949                         return $attribs;
 950                 }
 951
 952                 foreach( $pairs as $set ) {
 953                         $attribute = strtolower( $set[1] );
 954                         $value = Sanitizer::getTagAttributeCallback( $set );
 955
 956                         // Normalize whitespace
 957                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 958                         $value = trim( $value );
 959
 960                         // Decode character references
 961                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 962                 }
 963                 return $attribs;
 964         }
 965
 966         /**
 967          * Pick the appropriate attribute value from a match set from the
 968          * MW_ATTRIBS_REGEX matches.
 969          *
 970          * @param $set Array
 971          * @return String
 972          */
 973         private static function getTagAttributeCallback( $set ) {
 974                 if( isset( $set[6] ) ) {
 975                         # Illegal #XXXXXX color with no quotes.
 976                         return $set[6];
 977                 } elseif( isset( $set[5] ) ) {
 978                         # No quotes.
 979                         return $set[5];
 980                 } elseif( isset( $set[4] ) ) {
 981                         # Single-quoted
 982                         return $set[4];
 983                 } elseif( isset( $set[3] ) ) {
 984                         # Double-quoted
 985                         return $set[3];
 986                 } elseif( !isset( $set[2] ) ) {
 987                         # In XHTML, attributes must have a value.
 988                         # For 'reduced' form, return explicitly the attribute name here.
 989                         return $set[1];
 990                 } else {
 991                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 992                 }
 993         }
 994
 995         /**
 996          * Normalize whitespace and character references in an XML source-
 997          * encoded text for an attribute value.
 998          *
 999          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
1000          * but note that we're not returning the value, but are returning
1001          * XML source fragments that will be slapped into output.
1002          *
1003          * @param $text String
1004          * @return String
1005          */
1006         private static function normalizeAttributeValue( $text ) {
1007                 return str_replace( '"', '&quot;',
1008                         self::normalizeWhitespace(
1009                                 Sanitizer::normalizeCharReferences( $text ) ) );
1010         }
1011
1012         private static function normalizeWhitespace( $text ) {
1013                 return preg_replace(
1014                         '/\r\n|[\x20\x0d\x0a\x09]/',
1015                         ' ',
1016                         $text );
1017         }
1018
1019         /**
1020          * Ensure that any entities and character references are legal
1021          * for XML and XHTML specifically. Any stray bits will be
1022          * &amp;-escaped to result in a valid text fragment.
1023          *
1024          * a. any named char refs must be known in XHTML
1025          * b. any numeric char refs must be legal chars, not invalid or forbidden
1026          * c. use &#x, not &#X
1027          * d. fix or reject non-valid attributes
1028          *
1029          * @param $text String
1030          * @return String
1031          * @private
1032          */
1033         static function normalizeCharReferences( $text ) {
1034                 return preg_replace_callback(
1035                         MW_CHAR_REFS_REGEX,
1036                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1037                         $text );
1038         }
1039         /**
1040          * @param $matches String
1041          * @return String
1042          */
1043         static function normalizeCharReferencesCallback( $matches ) {
1044                 $ret = null;
1045                 if( $matches[1] != '' ) {
1046                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1047                 } elseif( $matches[2] != '' ) {
1048                         $ret = Sanitizer::decCharReference( $matches[2] );
1049                 } elseif( $matches[3] != ''  ) {
1050                         $ret = Sanitizer::hexCharReference( $matches[3] );
1051                 } elseif( $matches[4] != '' ) {
1052                         $ret = Sanitizer::hexCharReference( $matches[4] );
1053                 }
1054                 if( is_null( $ret ) ) {
1055                         return htmlspecialchars( $matches[0] );
1056                 } else {
1057                         return $ret;
1058                 }
1059         }
1060
1061         /**
1062          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1063          * return the named entity reference as is. If the entity is a
1064          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1065          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1066          *
1067          * @param $name String
1068          * @return String
1069          */
1070         static function normalizeEntity( $name ) {
1071                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1072                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1073                         return "&{$wgHtmlEntityAliases[$name]};";
1074                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1075                         return "&$name;";
1076                 } else {
1077                         return "&amp;$name;";
1078                 }
1079         }
1080
1081         static function decCharReference( $codepoint ) {
1082                 $point = intval( $codepoint );
1083                 if( Sanitizer::validateCodepoint( $point ) ) {
1084                         return sprintf( '&#%d;', $point );
1085                 } else {
1086                         return null;
1087                 }
1088         }
1089
1090         static function hexCharReference( $codepoint ) {
1091                 $point = hexdec( $codepoint );
1092                 if( Sanitizer::validateCodepoint( $point ) ) {
1093                         return sprintf( '&#x%x;', $point );
1094                 } else {
1095                         return null;
1096                 }
1097         }
1098
1099         /**
1100          * Returns true if a given Unicode codepoint is a valid character in XML.
1101          * @param $codepoint Integer
1102          * @return Boolean
1103          */
1104         private static function validateCodepoint( $codepoint ) {
1105                 return ($codepoint ==    0x09)
1106                         || ($codepoint ==    0x0a)
1107                         || ($codepoint ==    0x0d)
1108                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1109                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1110                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1111         }
1112
1113         /**
1114          * Decode any character references, numeric or named entities,
1115          * in the text and return a UTF-8 string.
1116          *
1117          * @param $text String
1118          * @return String
1119          */
1120         public static function decodeCharReferences( $text ) {
1121                 return preg_replace_callback(
1122                         MW_CHAR_REFS_REGEX,
1123                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1124                         $text );
1125         }
1126
1127         /**
1128          * @param $matches String
1129          * @return String
1130          */
1131         static function decodeCharReferencesCallback( $matches ) {
1132                 if( $matches[1] != '' ) {
1133                         return Sanitizer::decodeEntity( $matches[1] );
1134                 } elseif( $matches[2] != '' ) {
1135                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1136                 } elseif( $matches[3] != ''  ) {
1137                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1138                 } elseif( $matches[4] != '' ) {
1139                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1140                 }
1141                 # Last case should be an ampersand by itself
1142                 return $matches[0];
1143         }
1144
1145         /**
1146          * Return UTF-8 string for a codepoint if that is a valid
1147          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1148          * @param $codepoint Integer
1149          * @return String
1150          * @private
1151          */
1152         static function decodeChar( $codepoint ) {
1153                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1154                         return codepointToUtf8( $codepoint );
1155                 } else {
1156                         return UTF8_REPLACEMENT;
1157                 }
1158         }
1159
1160         /**
1161          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1162          * return the UTF-8 encoding of that character. Otherwise, returns
1163          * pseudo-entity source (eg &foo;)
1164          *
1165          * @param $name Strings
1166          * @return String
1167          */
1168         static function decodeEntity( $name ) {
1169                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1170                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1171                         $name = $wgHtmlEntityAliases[$name];
1172                 }
1173                 if( isset( $wgHtmlEntities[$name] ) ) {
1174                         return codepointToUtf8( $wgHtmlEntities[$name] );
1175                 } else {
1176                         return "&$name;";
1177                 }
1178         }
1179
1180         /**
1181          * Fetch the whitelist of acceptable attributes for a given element name.
1182          *
1183          * @param $element String
1184          * @return Array
1185          */
1186         static function attributeWhitelist( $element ) {
1187                 static $list;
1188                 if( !isset( $list ) ) {
1189                         $list = Sanitizer::setupAttributeWhitelist();
1190                 }
1191                 return isset( $list[$element] )
1192                         ? $list[$element]
1193                         : array();
1194         }
1195
1196         /**
1197          * Foreach array key (an allowed HTML element), return an array
1198          * of allowed attributes
1199          * @return Array
1200          */
1201         static function setupAttributeWhitelist() {
1202                 global $wgAllowRdfaAttributes, $wgHtml5, $wgAllowItemAttributes;
1203
1204                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style', 'xml:lang' );
1205
1206                 if ( $wgAllowRdfaAttributes ) {
1207                         #RDFa attributes as specified in section 9 of http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1208                         $common = array_merge( $common, array(
1209                             'about', 'property', 'resource', 'datatype', 'typeof',
1210                         ) );
1211                 }
1212
1213                 if ( $wgHtml5 && $wgAllowItemAttributes ) {
1214                         # add HTML5 microdata tages as pecified by http://www.w3.org/TR/html5/microdata.html
1215                         $common = array_merge( $common, array(
1216                             'item', 'itemprop', 'subject'
1217                         ) );
1218                 }
1219
1220                 $block = array_merge( $common, array( 'align' ) );
1221                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1222                 $tablecell = array( 'abbr',
1223                                     'axis',
1224                                     'headers',
1225                                     'scope',
1226                                     'rowspan',
1227                                     'colspan',
1228                                     'nowrap', # deprecated
1229                                     'width',  # deprecated
1230                                     'height', # deprecated
1231                                     'bgcolor' # deprecated
1232                                     );
1233
1234                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1235                 # See: http://www.w3.org/TR/html4/
1236                 $whitelist = array (
1237                         # 7.5.4
1238                         'div'        => $block,
1239                         'center'     => $common, # deprecated
1240                         'span'       => $block, # ??
1241
1242                         # 7.5.5
1243                         'h1'         => $block,
1244                         'h2'         => $block,
1245                         'h3'         => $block,
1246                         'h4'         => $block,
1247                         'h5'         => $block,
1248                         'h6'         => $block,
1249
1250                         # 7.5.6
1251                         # address
1252
1253                         # 8.2.4
1254                         # bdo
1255
1256                         # 9.2.1
1257                         'em'         => $common,
1258                         'strong'     => $common,
1259                         'cite'       => $common,
1260                         # dfn
1261                         'code'       => $common,
1262                         # samp
1263                         # kbd
1264                         'var'        => $common,
1265                         'abbr'       => $common,
1266                         # acronym
1267
1268                         # 9.2.2
1269                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1270                         # q
1271
1272                         # 9.2.3
1273                         'sub'        => $common,
1274                         'sup'        => $common,
1275
1276                         # 9.3.1
1277                         'p'          => $block,
1278
1279                         # 9.3.2
1280                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1281
1282                         # 9.3.4
1283                         'pre'        => array_merge( $common, array( 'width' ) ),
1284
1285                         # 9.4
1286                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1287                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1288
1289                         # 10.2
1290                         'ul'         => array_merge( $common, array( 'type' ) ),
1291                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1292                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1293
1294                         # 10.3
1295                         'dl'         => $common,
1296                         'dd'         => $common,
1297                         'dt'         => $common,
1298
1299                         # 11.2.1
1300                         'table'      => array_merge( $common,
1301                                                                 array( 'summary', 'width', 'border', 'frame',
1302                                                                                 'rules', 'cellspacing', 'cellpadding',
1303                                                                                 'align', 'bgcolor',
1304                                                                 ) ),
1305
1306                         # 11.2.2
1307                         'caption'    => array_merge( $common, array( 'align' ) ),
1308
1309                         # 11.2.3
1310                         'thead'      => array_merge( $common, $tablealign ),
1311                         'tfoot'      => array_merge( $common, $tablealign ),
1312                         'tbody'      => array_merge( $common, $tablealign ),
1313
1314                         # 11.2.4
1315                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1316                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1317
1318                         # 11.2.5
1319                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1320
1321                         # 11.2.6
1322                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1323                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1324
1325                         # 12.2 # NOTE: <a> is not allowed directly, but the attrib whitelist is used from the Parser object
1326                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1327
1328                         # 13.2
1329                         # Not usually allowed, but may be used for extension-style hooks
1330                         # such as <math> when it is rasterized
1331                         'img'        => array_merge( $common, array( 'alt' ) ),
1332
1333                         # 15.2.1
1334                         'tt'         => $common,
1335                         'b'          => $common,
1336                         'i'          => $common,
1337                         'big'        => $common,
1338                         'small'      => $common,
1339                         'strike'     => $common,
1340                         's'          => $common,
1341                         'u'          => $common,
1342
1343                         # 15.2.2
1344                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1345                         # basefont
1346
1347                         # 15.3
1348                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1349
1350                         # XHTML Ruby annotation text module, simple ruby only.
1351                         # http://www.w3c.org/TR/ruby/
1352                         'ruby'       => $common,
1353                         # rbc
1354                         # rtc
1355                         'rb'         => $common,
1356                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1357                         'rp'         => $common,
1358
1359                         # MathML root element, where used for extensions
1360                         # 'title' may not be 100% valid here; it's XHTML
1361                         # http://www.w3.org/TR/REC-MathML/
1362                         'math'       => array( 'class', 'style', 'id', 'title' ),
1363                         );
1364                 return $whitelist;
1365         }
1366
1367         /**
1368          * Take a fragment of (potentially invalid) HTML and return
1369          * a version with any tags removed, encoded as plain text.
1370          *
1371          * Warning: this return value must be further escaped for literal
1372          * inclusion in HTML output as of 1.10!
1373          *
1374          * @param $text String: HTML fragment
1375          * @return String
1376          */
1377         static function stripAllTags( $text ) {
1378                 # Actual <tags>
1379                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1380
1381                 # Normalize &entities and whitespace
1382                 $text = self::decodeCharReferences( $text );
1383                 $text = self::normalizeWhitespace( $text );
1384
1385                 return $text;
1386         }
1387
1388         /**
1389          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1390          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1391          * PHP 5.1 doesn't.
1392          *
1393          * Use for passing XHTML fragments to PHP's XML parsing functions
1394          *
1395          * @return String
1396          */
1397         static function hackDocType() {
1398                 global $wgHtmlEntities;
1399                 $out = "<!DOCTYPE html [\n";
1400                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1401                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1402                 }
1403                 $out .= "]>\n";
1404                 return $out;
1405         }
1406
1407         static function cleanUrl( $url ) {
1408                 # Normalize any HTML entities in input. They will be
1409                 # re-escaped by makeExternalLink().
1410                 $url = Sanitizer::decodeCharReferences( $url );
1411
1412                 # Escape any control characters introduced by the above step
1413                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1414
1415                 # Validate hostname portion
1416                 $matches = array();
1417                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1418                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1419
1420                         // Characters that will be ignored in IDNs.
1421                         // http://tools.ietf.org/html/3454#section-3.1
1422                         // Strip them before further processing so blacklists and such work.
1423                         $strip = "/
1424                                 \\s|          # general whitespace
1425                                 \xc2\xad|     # 00ad SOFT HYPHEN
1426                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1427                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1428                                 \xe2\x81\xa0| # 2060 WORD JOINER
1429                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1430                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1431                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1432                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1433                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1434                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1435                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1436                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1437                                 /xuD";
1438
1439                         $host = preg_replace( $strip, '', $host );
1440
1441                         // @fixme: validate hostnames here
1442
1443                         return $protocol . $host . $rest;
1444                 } else {
1445                         return $url;
1446                 }
1447         }
1448
1449 }