includes/Sanitizer.php

   1 <?php
   2 /**
   3  * (X)HTML 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  * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @package MediaWiki
  24  * @subpackage 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]+);
  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  * @access 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 /** @package MediaWiki */
 320 class Sanitizer {
 321         /**
 322          * Cleans up HTML, removes dangerous tags and attributes, and
 323          * removes HTML comments
 324          * @access private
 325          * @param string $text
 326          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 327          * @param array $args for the processing callback
 328          * @return string
 329          */
 330         function removeHTMLtags( $text, $processCallback = null, $args = array() ) {
 331                 global $wgUseTidy, $wgUserHtml;
 332                 $fname = 'Parser::removeHTMLtags';
 333                 wfProfileIn( $fname );
 334
 335                 if( $wgUserHtml ) {
 336                         $htmlpairs = array( # Tags that must be closed
 337                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 338                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 339                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 340                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 341                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span'
 342                         );
 343                         $htmlsingle = array(
 344                                 'br', 'hr', 'li', 'dt', 'dd'
 345                         );
 346                         $htmlsingleonly = array( # Elements that cannot have close tags
 347                                 'br', 'hr'
 348                         );
 349                         $htmlnest = array( # Tags that can be nested--??
 350                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 351                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 352                         );
 353                         $tabletags = array( # Can only appear inside table
 354                                 'td', 'th', 'tr'
 355                         );
 356                 } else {
 357                         $htmlpairs = array();
 358                         $htmlsingle = array();
 359                         $htmlnest = array();
 360                         $tabletags = array();
 361                 }
 362
 363                 $htmlsingle = array_merge( $tabletags, $htmlsingle );
 364                 $htmlelements = array_merge( $htmlsingle, $htmlpairs );
 365
 366                 # Remove HTML comments
 367                 $text = Sanitizer::removeHTMLcomments( $text );
 368
 369                 $bits = explode( '<', $text );
 370                 $text = array_shift( $bits );
 371                 if(!$wgUseTidy) {
 372                         $tagstack = array(); $tablestack = array();
 373                         foreach ( $bits as $x ) {
 374                                 $prev = error_reporting( E_ALL & ~( E_NOTICE | E_WARNING ) );
 375                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 376                                 $x, $regs );
 377                                 list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
 378                                 error_reporting( $prev );
 379
 380                                 $badtag = 0 ;
 381                                 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
 382                                         # Check our stack
 383                                         if ( $slash ) {
 384                                                 # Closing a tag...
 385                                                 if( in_array( $t, $htmlsingleonly ) ) {
 386                                                         $badtag = 1;
 387                                                 } elseif( !in_array( $t, $htmlsingle ) &&
 388                                                 ( $ot = @array_pop( $tagstack ) ) != $t ) {
 389                                                         @array_push( $tagstack, $ot );
 390                                                         $badtag = 1;
 391                                                 } else {
 392                                                         if ( $t == 'table' ) {
 393                                                                 $tagstack = array_pop( $tablestack );
 394                                                         }
 395                                                         $newparams = '';
 396                                                 }
 397                                         } else {
 398                                                 # Keep track for later
 399                                                 if ( in_array( $t, $tabletags ) &&
 400                                                 ! in_array( 'table', $tagstack ) ) {
 401                                                         $badtag = 1;
 402                                                 } else if ( in_array( $t, $tagstack ) &&
 403                                                 ! in_array ( $t , $htmlnest ) ) {
 404                                                         $badtag = 1 ;
 405                                                 } elseif( in_array( $t, $htmlsingleonly ) ) {
 406                                                         # Hack to force empty tag for uncloseable elements
 407                                                         $brace = '/>';
 408                                                 } else if ( ! in_array( $t, $htmlsingle ) ) {
 409                                                         if ( $t == 'table' ) {
 410                                                                 array_push( $tablestack, $tagstack );
 411                                                                 $tagstack = array();
 412                                                         }
 413                                                         array_push( $tagstack, $t );
 414                                                 }
 415
 416                                                 # Replace any variables or template parameters with
 417                                                 # plaintext results.
 418                                                 if( is_callable( $processCallback ) ) {
 419                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 420                                                 }
 421
 422                                                 # Strip non-approved attributes from the tag
 423                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 424                                         }
 425                                         if ( ! $badtag ) {
 426                                                 $rest = str_replace( '>', '&gt;', $rest );
 427                                                 $close = ( $brace == '/>' ) ? ' /' : '';
 428                                                 $text .= "<$slash$t$newparams$close>$rest";
 429                                                 continue;
 430                                         }
 431                                 }
 432                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 433                         }
 434                         # Close off any remaining tags
 435                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 436                                 $text .= "</$t>\n";
 437                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 438                         }
 439                 } else {
 440                         # this might be possible using tidy itself
 441                         foreach ( $bits as $x ) {
 442                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 443                                 $x, $regs );
 444                                 @list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
 445                                 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
 446                                         if( is_callable( $processCallback ) ) {
 447                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 448                                         }
 449                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 450                                         $rest = str_replace( '>', '&gt;', $rest );
 451                                         $text .= "<$slash$t$newparams$brace$rest";
 452                                 } else {
 453                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 454                                 }
 455                         }
 456                 }
 457                 wfProfileOut( $fname );
 458                 return $text;
 459         }
 460
 461         /**
 462          * Remove '<!--', '-->', and everything between.
 463          * To avoid leaving blank lines, when a comment is both preceded
 464          * and followed by a newline (ignoring spaces), trim leading and
 465          * trailing spaces and one of the newlines.
 466          *
 467          * @access private
 468          * @param string $text
 469          * @return string
 470          */
 471         function removeHTMLcomments( $text ) {
 472                 $fname='Parser::removeHTMLcomments';
 473                 wfProfileIn( $fname );
 474                 while (($start = strpos($text, '<!--')) !== false) {
 475                         $end = strpos($text, '-->', $start + 4);
 476                         if ($end === false) {
 477                                 # Unterminated comment; bail out
 478                                 break;
 479                         }
 480
 481                         $end += 3;
 482
 483                         # Trim space and newline if the comment is both
 484                         # preceded and followed by a newline
 485                         $spaceStart = max($start - 1, 0);
 486                         $spaceLen = $end - $spaceStart;
 487                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 488                                 $spaceStart--;
 489                                 $spaceLen++;
 490                         }
 491                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 492                                 $spaceLen++;
 493                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 494                                 # Remove the comment, leading and trailing
 495                                 # spaces, and leave only one newline.
 496                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 497                         }
 498                         else {
 499                                 # Remove just the comment.
 500                                 $text = substr_replace($text, '', $start, $end - $start);
 501                         }
 502                 }
 503                 wfProfileOut( $fname );
 504                 return $text;
 505         }
 506
 507         /**
 508          * Take a tag soup fragment listing an HTML element's attributes
 509          * and normalize it to well-formed XML, discarding unwanted attributes.
 510          *
 511          * - Normalizes attribute names to lowercase
 512          * - Discards attributes not on a whitelist for the given element
 513          * - Turns broken or invalid entities into plaintext
 514          * - Double-quotes all attribute values
 515          * - Attributes without values are given the name as attribute
 516          * - Double attributes are discarded
 517          * - Unsafe style attributes are discarded
 518          * - Prepends space if there are attributes.
 519          *
 520          * @param string $text
 521          * @param string $element
 522          * @return string
 523          *
 524          * @todo Check for legal values where the DTD limits things.
 525          * @todo Check for unique id attribute :P
 526          */
 527         function fixTagAttributes( $text, $element ) {
 528                 global $wgUrlProtcols;
 529                 if( trim( $text ) == '' ) {
 530                         return '';
 531                 }
 532
 533                 # Unquoted attribute
 534                 # Since we quote this later, this can be anything distinguishable
 535                 # from the end of the attribute
 536                 if( !preg_match_all(
 537                         MW_ATTRIBS_REGEX,
 538                         $text,
 539                         $pairs,
 540                         PREG_SET_ORDER ) ) {
 541                         return '';
 542                 }
 543
 544                 $whitelist = array_flip( Sanitizer::attributeWhitelist( $element ) );
 545                 $attribs = array();
 546                 foreach( $pairs as $set ) {
 547                         $attribute = strtolower( $set[1] );
 548                         if( !isset( $whitelist[$attribute] ) ) {
 549                                 continue;
 550                         }
 551
 552                         $raw   = Sanitizer::getTagAttributeCallback( $set );
 553                         $value = Sanitizer::normalizeAttributeValue( $raw );
 554
 555                         # Strip javascript "expression" from stylesheets.
 556                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 557                         if( $attribute == 'style' && preg_match(
 558                                 '/(expression|tps*:\/\/|url\\s*\().*/is',
 559                                         Sanitizer::decodeCharReferences( $value ) ) ) {
 560                                 # haxx0r
 561                                 continue;
 562                         }
 563
 564                         # Templates and links may be expanded in later parsing,
 565                         # creating invalid or dangerous output. Suppress this.
 566                         $value = strtr( $value, array(
 567                                 '{'    => '&#123;',
 568                                 '['    => '&#91;',
 569                                 "''"   => '&#39;&#39;',
 570                                 'ISBN' => '&#73;SBN',
 571                                 'RFC'  => '&#82;FC',
 572                                 'PMID' => '&#80;MID',
 573                         ) );
 574                         $value = preg_replace(
 575                                 '/(' . $wgUrlProtcols . '):/',
 576                                 '\\1&#58;', $value );
 577
 578                         // If this attribute was previously set, override it.
 579                         // Output should only have one attribute of each name.
 580                         $attribs[$attribute] = "$attribute=\"$value\"";
 581                 }
 582                 if( empty( $attribs ) ) {
 583                         return '';
 584                 } else {
 585                         return ' ' . implode( ' ', $attribs );
 586                 }
 587         }
 588
 589         /**
 590          * Return an associative array of attribute names and values from
 591          * a partial tag string. Attribute names are forces to lowercase,
 592          * character references are decoded to UTF-8 text.
 593          *
 594          * @param string
 595          * @return array
 596          */
 597         function decodeTagAttributes( $text ) {
 598                 $attribs = array();
 599
 600                 if( trim( $text ) == '' ) {
 601                         return $attribs;
 602                 }
 603
 604                 if( !preg_match_all(
 605                         MW_ATTRIBS_REGEX,
 606                         $text,
 607                         $pairs,
 608                         PREG_SET_ORDER ) ) {
 609                         return $attribs;
 610                 }
 611
 612                 foreach( $pairs as $set ) {
 613                         $attribute = strtolower( $set[1] );
 614                         $value = Sanitizer::getTagAttributeCallback( $set );
 615                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 616                 }
 617                 return $attribs;
 618         }
 619
 620         /**
 621          * Pick the appropriate attribute value from a match set from the
 622          * MW_ATTRIBS_REGEX matches.
 623          *
 624          * @param array $set
 625          * @return string
 626          * @access private
 627          */
 628         function getTagAttributeCallback( $set ) {
 629                 if( isset( $set[6] ) ) {
 630                         # Illegal #XXXXXX color with no quotes.
 631                         return $set[6];
 632                 } elseif( isset( $set[5] ) ) {
 633                         # No quotes.
 634                         return $set[5];
 635                 } elseif( isset( $set[4] ) ) {
 636                         # Single-quoted
 637                         return $set[4];
 638                 } elseif( isset( $set[3] ) ) {
 639                         # Double-quoted
 640                         return $set[3];
 641                 } elseif( !isset( $set[2] ) ) {
 642                         # In XHTML, attributes must have a value.
 643                         # For 'reduced' form, return explicitly the attribute name here.
 644                         return $set[1];
 645                 } else {
 646                         wfDebugDieBacktrace( "Tag conditions not met. This should never happen and is a bug." );
 647                 }
 648         }
 649
 650         /**
 651          * Normalize whitespace and character references in an XML source-
 652          * encoded text for an attribute value.
 653          *
 654          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 655          * but note that we're not returning the value, but are returning
 656          * XML source fragments that will be slapped into output.
 657          *
 658          * @param string $text
 659          * @return string
 660          * @access private
 661          */
 662         function normalizeAttributeValue( $text ) {
 663                 return str_replace( '"', '&quot;',
 664                         preg_replace(
 665                                 '/\r\n|[\x20\x0d\x0a\x09]/',
 666                                 ' ',
 667                                 Sanitizer::normalizeCharReferences( $text ) ) );
 668         }
 669
 670         /**
 671          * Ensure that any entities and character references are legal
 672          * for XML and XHTML specifically. Any stray bits will be
 673          * &amp;-escaped to result in a valid text fragment.
 674          *
 675          * a. any named char refs must be known in XHTML
 676          * b. any numeric char refs must be legal chars, not invalid or forbidden
 677          * c. use &#x, not &#X
 678          * d. fix or reject non-valid attributes
 679          *
 680          * @param string $text
 681          * @return string
 682          * @access private
 683          */
 684         function normalizeCharReferences( $text ) {
 685                 return preg_replace_callback(
 686                         MW_CHAR_REFS_REGEX,
 687                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 688                         $text );
 689         }
 690         /**
 691          * @param string $matches
 692          * @return string
 693          */
 694         function normalizeCharReferencesCallback( $matches ) {
 695                 $ret = null;
 696                 if( $matches[1] != '' ) {
 697                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 698                 } elseif( $matches[2] != '' ) {
 699                         $ret = Sanitizer::decCharReference( $matches[2] );
 700                 } elseif( $matches[3] != ''  ) {
 701                         $ret = Sanitizer::hexCharReference( $matches[3] );
 702                 } elseif( $matches[4] != '' ) {
 703                         $ret = Sanitizer::hexCharReference( $matches[4] );
 704                 }
 705                 if( is_null( $ret ) ) {
 706                         return htmlspecialchars( $matches[0] );
 707                 } else {
 708                         return $ret;
 709                 }
 710         }
 711
 712         /**
 713          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 714          * return the named entity reference as is. Otherwise, returns
 715          * HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 716          *
 717          * @param string $name
 718          * @return string
 719          */
 720         function normalizeEntity( $name ) {
 721                 global $wgHtmlEntities;
 722                 if( isset( $wgHtmlEntities[$name] ) ) {
 723                         return "&$name;";
 724                 } else {
 725                         return "&amp;$name;";
 726                 }
 727         }
 728
 729         function decCharReference( $codepoint ) {
 730                 $point = IntVal( $codepoint );
 731                 if( Sanitizer::validateCodepoint( $point ) ) {
 732                         return sprintf( '&#%d;', $point );
 733                 } else {
 734                         return null;
 735                 }
 736         }
 737
 738         function hexCharReference( $codepoint ) {
 739                 $point = hexdec( $codepoint );
 740                 if( Sanitizer::validateCodepoint( $point ) ) {
 741                         return sprintf( '&#x%x;', $point );
 742                 } else {
 743                         return null;
 744                 }
 745         }
 746
 747         /**
 748          * Returns true if a given Unicode codepoint is a valid character in XML.
 749          * @param int $codepoint
 750          * @return bool
 751          */
 752         function validateCodepoint( $codepoint ) {
 753                 return ($codepoint ==    0x09)
 754                         || ($codepoint ==    0x0a)
 755                         || ($codepoint ==    0x0d)
 756                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 757                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 758                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 759         }
 760
 761         /**
 762          * Decode any character references, numeric or named entities,
 763          * in the text and return a UTF-8 string.
 764          *
 765          * @param string $text
 766          * @return string
 767          * @access public
 768          */
 769         function decodeCharReferences( $text ) {
 770                 return preg_replace_callback(
 771                         MW_CHAR_REFS_REGEX,
 772                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 773                         $text );
 774         }
 775
 776         /**
 777          * @param string $matches
 778          * @return string
 779          */
 780         function decodeCharReferencesCallback( $matches ) {
 781                 if( $matches[1] != '' ) {
 782                         return Sanitizer::decodeEntity( $matches[1] );
 783                 } elseif( $matches[2] != '' ) {
 784                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 785                 } elseif( $matches[3] != ''  ) {
 786                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 787                 } elseif( $matches[4] != '' ) {
 788                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 789                 }
 790                 # Last case should be an ampersand by itself
 791                 return $matches[0];
 792         }
 793
 794         /**
 795          * Return UTF-8 string for a codepoint if that is a valid
 796          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
 797          * @param int $codepoint
 798          * @return string
 799          * @access private
 800          */
 801         function decodeChar( $codepoint ) {
 802                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
 803                         return codepointToUtf8( $codepoint );
 804                 } else {
 805                         return UTF8_REPLACEMENT;
 806                 }
 807         }
 808
 809         /**
 810          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 811          * return the UTF-8 encoding of that character. Otherwise, returns
 812          * pseudo-entity source (eg &foo;)
 813          *
 814          * @param string $name
 815          * @return string
 816          */
 817         function decodeEntity( $name ) {
 818                 global $wgHtmlEntities;
 819                 if( isset( $wgHtmlEntities[$name] ) ) {
 820                         return codepointToUtf8( $wgHtmlEntities[$name] );
 821                 } else {
 822                         return "&$name;";
 823                 }
 824         }
 825
 826         /**
 827          * Fetch the whitelist of acceptable attributes for a given
 828          * element name.
 829          *
 830          * @param string $element
 831          * @return array
 832          */
 833         function attributeWhitelist( $element ) {
 834                 static $list;
 835                 if( !isset( $list ) ) {
 836                         $list = Sanitizer::setupAttributeWhitelist();
 837                 }
 838                 return isset( $list[$element] )
 839                         ? $list[$element]
 840                         : array();
 841         }
 842
 843         /**
 844          * @return array
 845          */
 846         function setupAttributeWhitelist() {
 847                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
 848                 $block = array_merge( $common, array( 'align' ) );
 849                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
 850                 $tablecell = array( 'abbr',
 851                                     'axis',
 852                                     'headers',
 853                                     'scope',
 854                                     'rowspan',
 855                                     'colspan',
 856                                     'nowrap', # deprecated
 857                                     'width',  # deprecated
 858                                     'height', # deprecated
 859                                     'bgcolor' # deprecated
 860                                     );
 861
 862                 # Numbers refer to sections in HTML 4.01 standard describing the element.
 863                 # See: http://www.w3.org/TR/html4/
 864                 $whitelist = array (
 865                         # 7.5.4
 866                         'div'        => $block,
 867                         'center'     => $common, # deprecated
 868                         'span'       => $block, # ??
 869
 870                         # 7.5.5
 871                         'h1'         => $block,
 872                         'h2'         => $block,
 873                         'h3'         => $block,
 874                         'h4'         => $block,
 875                         'h5'         => $block,
 876                         'h6'         => $block,
 877
 878                         # 7.5.6
 879                         # address
 880
 881                         # 8.2.4
 882                         # bdo
 883
 884                         # 9.2.1
 885                         'em'         => $common,
 886                         'strong'     => $common,
 887                         'cite'       => $common,
 888                         # dfn
 889                         'code'       => $common,
 890                         # samp
 891                         # kbd
 892                         'var'        => $common,
 893                         # abbr
 894                         # acronym
 895
 896                         # 9.2.2
 897                         'blockquote' => array_merge( $common, array( 'cite' ) ),
 898                         # q
 899
 900                         # 9.2.3
 901                         'sub'        => $common,
 902                         'sup'        => $common,
 903
 904                         # 9.3.1
 905                         'p'          => $block,
 906
 907                         # 9.3.2
 908                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
 909
 910                         # 9.3.4
 911                         'pre'        => array_merge( $common, array( 'width' ) ),
 912
 913                         # 9.4
 914                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 915                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 916
 917                         # 10.2
 918                         'ul'         => array_merge( $common, array( 'type' ) ),
 919                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
 920                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
 921
 922                         # 10.3
 923                         'dl'         => $common,
 924                         'dd'         => $common,
 925                         'dt'         => $common,
 926
 927                         # 11.2.1
 928                         'table'      => array_merge( $common,
 929                                                                 array( 'summary', 'width', 'border', 'frame',
 930                                                                                          'rules', 'cellspacing', 'cellpadding',
 931                                                                                          'align', 'bgcolor', 'frame', 'rules',
 932                                                                                          'border' ) ),
 933
 934                         # 11.2.2
 935                         'caption'    => array_merge( $common, array( 'align' ) ),
 936
 937                         # 11.2.3
 938                         'thead'      => array_merge( $common, $tablealign ),
 939                         'tfoot'      => array_merge( $common, $tablealign ),
 940                         'tbody'      => array_merge( $common, $tablealign ),
 941
 942                         # 11.2.4
 943                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 944                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 945
 946                         # 11.2.5
 947                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
 948
 949                         # 11.2.6
 950                         'td'         => array_merge( $common, $tablecell, $tablealign ),
 951                         'th'         => array_merge( $common, $tablecell, $tablealign ),
 952
 953                         # 15.2.1
 954                         'tt'         => $common,
 955                         'b'          => $common,
 956                         'i'          => $common,
 957                         'big'        => $common,
 958                         'small'      => $common,
 959                         'strike'     => $common,
 960                         's'          => $common,
 961                         'u'          => $common,
 962
 963                         # 15.2.2
 964                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
 965                         # basefont
 966
 967                         # 15.3
 968                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
 969
 970                         # XHTML Ruby annotation text module, simple ruby only.
 971                         # http://www.w3c.org/TR/ruby/
 972                         'ruby'       => $common,
 973                         # rbc
 974                         # rtc
 975                         'rb'         => $common,
 976                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
 977                         'rp'         => $common,
 978                         );
 979                 return $whitelist;
 980         }
 981
 982         /**
 983          * Take a fragment of (potentially invalid) HTML and return
 984          * a version with any tags removed, encoded suitably for literal
 985          * inclusion in an attribute value.
 986          *
 987          * @param string $text HTML fragment
 988          * @return string
 989          */
 990         function stripAllTags( $text ) {
 991                 # Actual <tags>
 992                 $text = preg_replace( '/<[^>]*>/', '', $text );
 993
 994                 # Normalize &entities and whitespace
 995                 $text = Sanitizer::normalizeAttributeValue( $text );
 996
 997                 # Will be placed into "double-quoted" attributes,
 998                 # make sure remaining bits are safe.
 999                 $text = str_replace(
1000                         array('<', '>', '"'),
1001                         array('&lt;', '&gt;', '&quot;'),
1002                         $text );
1003
1004                 return $text;
1005         }
1006
1007 }
1008
1009 ?>