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                 if( trim( $text ) == '' ) {
 529                         return '';
 530                 }
 531
 532                 # Unquoted attribute
 533                 # Since we quote this later, this can be anything distinguishable
 534                 # from the end of the attribute
 535                 $pairs = array();
 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' ) {
 558                                 $stripped = Sanitizer::decodeCharReferences( $value );
 559
 560                                 // Remove any comments; IE gets token splitting wrong
 561                                 $stripped = preg_replace( '!/\\*.*?\\*/!S', ' ', $stripped );
 562                                 $value = htmlspecialchars( $stripped );
 563
 564                                 // ... and continue checks
 565                                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 566                                         'codepointToUtf8(hexdec("$1"))', $stripped );
 567                                 $stripped = str_replace( '\\', '', $stripped );
 568                                 if( preg_match( '/(expression|tps*:\/\/|url\\s*\().*/is',
 569                                                 $stripped ) ) {
 570                                         # haxx0r
 571                                         continue;
 572                                 }
 573                         }
 574
 575                         # Templates and links may be expanded in later parsing,
 576                         # creating invalid or dangerous output. Suppress this.
 577                         $value = strtr( $value, array(
 578                                 '{'    => '&#123;',
 579                                 '['    => '&#91;',
 580                                 "''"   => '&#39;&#39;',
 581                                 'ISBN' => '&#73;SBN',
 582                                 'RFC'  => '&#82;FC',
 583                                 'PMID' => '&#80;MID',
 584                         ) );
 585
 586                         # Stupid hack
 587                         $value = preg_replace_callback(
 588                                 '/(' . wfUrlProtocols() . ')/',
 589                                 array( 'Sanitizer', 'armorLinksCallback' ),
 590                                 $value );
 591
 592                         // If this attribute was previously set, override it.
 593                         // Output should only have one attribute of each name.
 594                         $attribs[$attribute] = "$attribute=\"$value\"";
 595                 }
 596                 if( empty( $attribs ) ) {
 597                         return '';
 598                 } else {
 599                         return ' ' . implode( ' ', $attribs );
 600                 }
 601         }
 602
 603         /**
 604          * Regex replace callback for armoring links against further processing.
 605          * @param array $matches
 606          * @return string
 607          * @access private
 608          */
 609         function armorLinksCallback( $matches ) {
 610                 return str_replace( ':', '&#58;', $matches[1] );
 611         }
 612
 613         /**
 614          * Return an associative array of attribute names and values from
 615          * a partial tag string. Attribute names are forces to lowercase,
 616          * character references are decoded to UTF-8 text.
 617          *
 618          * @param string
 619          * @return array
 620          */
 621         function decodeTagAttributes( $text ) {
 622                 $attribs = array();
 623
 624                 if( trim( $text ) == '' ) {
 625                         return $attribs;
 626                 }
 627
 628                 $pairs = array();
 629                 if( !preg_match_all(
 630                         MW_ATTRIBS_REGEX,
 631                         $text,
 632                         $pairs,
 633                         PREG_SET_ORDER ) ) {
 634                         return $attribs;
 635                 }
 636
 637                 foreach( $pairs as $set ) {
 638                         $attribute = strtolower( $set[1] );
 639                         $value = Sanitizer::getTagAttributeCallback( $set );
 640                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 641                 }
 642                 return $attribs;
 643         }
 644
 645         /**
 646          * Pick the appropriate attribute value from a match set from the
 647          * MW_ATTRIBS_REGEX matches.
 648          *
 649          * @param array $set
 650          * @return string
 651          * @access private
 652          */
 653         function getTagAttributeCallback( $set ) {
 654                 if( isset( $set[6] ) ) {
 655                         # Illegal #XXXXXX color with no quotes.
 656                         return $set[6];
 657                 } elseif( isset( $set[5] ) ) {
 658                         # No quotes.
 659                         return $set[5];
 660                 } elseif( isset( $set[4] ) ) {
 661                         # Single-quoted
 662                         return $set[4];
 663                 } elseif( isset( $set[3] ) ) {
 664                         # Double-quoted
 665                         return $set[3];
 666                 } elseif( !isset( $set[2] ) ) {
 667                         # In XHTML, attributes must have a value.
 668                         # For 'reduced' form, return explicitly the attribute name here.
 669                         return $set[1];
 670                 } else {
 671                         wfDebugDieBacktrace( "Tag conditions not met. This should never happen and is a bug." );
 672                 }
 673         }
 674
 675         /**
 676          * Normalize whitespace and character references in an XML source-
 677          * encoded text for an attribute value.
 678          *
 679          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 680          * but note that we're not returning the value, but are returning
 681          * XML source fragments that will be slapped into output.
 682          *
 683          * @param string $text
 684          * @return string
 685          * @access private
 686          */
 687         function normalizeAttributeValue( $text ) {
 688                 return str_replace( '"', '&quot;',
 689                         preg_replace(
 690                                 '/\r\n|[\x20\x0d\x0a\x09]/',
 691                                 ' ',
 692                                 Sanitizer::normalizeCharReferences( $text ) ) );
 693         }
 694
 695         /**
 696          * Ensure that any entities and character references are legal
 697          * for XML and XHTML specifically. Any stray bits will be
 698          * &amp;-escaped to result in a valid text fragment.
 699          *
 700          * a. any named char refs must be known in XHTML
 701          * b. any numeric char refs must be legal chars, not invalid or forbidden
 702          * c. use &#x, not &#X
 703          * d. fix or reject non-valid attributes
 704          *
 705          * @param string $text
 706          * @return string
 707          * @access private
 708          */
 709         function normalizeCharReferences( $text ) {
 710                 return preg_replace_callback(
 711                         MW_CHAR_REFS_REGEX,
 712                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 713                         $text );
 714         }
 715         /**
 716          * @param string $matches
 717          * @return string
 718          */
 719         function normalizeCharReferencesCallback( $matches ) {
 720                 $ret = null;
 721                 if( $matches[1] != '' ) {
 722                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 723                 } elseif( $matches[2] != '' ) {
 724                         $ret = Sanitizer::decCharReference( $matches[2] );
 725                 } elseif( $matches[3] != ''  ) {
 726                         $ret = Sanitizer::hexCharReference( $matches[3] );
 727                 } elseif( $matches[4] != '' ) {
 728                         $ret = Sanitizer::hexCharReference( $matches[4] );
 729                 }
 730                 if( is_null( $ret ) ) {
 731                         return htmlspecialchars( $matches[0] );
 732                 } else {
 733                         return $ret;
 734                 }
 735         }
 736
 737         /**
 738          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 739          * return the named entity reference as is. Otherwise, returns
 740          * HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 741          *
 742          * @param string $name
 743          * @return string
 744          */
 745         function normalizeEntity( $name ) {
 746                 global $wgHtmlEntities;
 747                 if( isset( $wgHtmlEntities[$name] ) ) {
 748                         return "&$name;";
 749                 } else {
 750                         return "&amp;$name;";
 751                 }
 752         }
 753
 754         function decCharReference( $codepoint ) {
 755                 $point = intval( $codepoint );
 756                 if( Sanitizer::validateCodepoint( $point ) ) {
 757                         return sprintf( '&#%d;', $point );
 758                 } else {
 759                         return null;
 760                 }
 761         }
 762
 763         function hexCharReference( $codepoint ) {
 764                 $point = hexdec( $codepoint );
 765                 if( Sanitizer::validateCodepoint( $point ) ) {
 766                         return sprintf( '&#x%x;', $point );
 767                 } else {
 768                         return null;
 769                 }
 770         }
 771
 772         /**
 773          * Returns true if a given Unicode codepoint is a valid character in XML.
 774          * @param int $codepoint
 775          * @return bool
 776          */
 777         function validateCodepoint( $codepoint ) {
 778                 return ($codepoint ==    0x09)
 779                         || ($codepoint ==    0x0a)
 780                         || ($codepoint ==    0x0d)
 781                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 782                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 783                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 784         }
 785
 786         /**
 787          * Decode any character references, numeric or named entities,
 788          * in the text and return a UTF-8 string.
 789          *
 790          * @param string $text
 791          * @return string
 792          * @access public
 793          */
 794         function decodeCharReferences( $text ) {
 795                 return preg_replace_callback(
 796                         MW_CHAR_REFS_REGEX,
 797                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 798                         $text );
 799         }
 800
 801         /**
 802          * @param string $matches
 803          * @return string
 804          */
 805         function decodeCharReferencesCallback( $matches ) {
 806                 if( $matches[1] != '' ) {
 807                         return Sanitizer::decodeEntity( $matches[1] );
 808                 } elseif( $matches[2] != '' ) {
 809                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 810                 } elseif( $matches[3] != ''  ) {
 811                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 812                 } elseif( $matches[4] != '' ) {
 813                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 814                 }
 815                 # Last case should be an ampersand by itself
 816                 return $matches[0];
 817         }
 818
 819         /**
 820          * Return UTF-8 string for a codepoint if that is a valid
 821          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
 822          * @param int $codepoint
 823          * @return string
 824          * @access private
 825          */
 826         function decodeChar( $codepoint ) {
 827                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
 828                         return codepointToUtf8( $codepoint );
 829                 } else {
 830                         return UTF8_REPLACEMENT;
 831                 }
 832         }
 833
 834         /**
 835          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 836          * return the UTF-8 encoding of that character. Otherwise, returns
 837          * pseudo-entity source (eg &foo;)
 838          *
 839          * @param string $name
 840          * @return string
 841          */
 842         function decodeEntity( $name ) {
 843                 global $wgHtmlEntities;
 844                 if( isset( $wgHtmlEntities[$name] ) ) {
 845                         return codepointToUtf8( $wgHtmlEntities[$name] );
 846                 } else {
 847                         return "&$name;";
 848                 }
 849         }
 850
 851         /**
 852          * Fetch the whitelist of acceptable attributes for a given
 853          * element name.
 854          *
 855          * @param string $element
 856          * @return array
 857          */
 858         function attributeWhitelist( $element ) {
 859                 static $list;
 860                 if( !isset( $list ) ) {
 861                         $list = Sanitizer::setupAttributeWhitelist();
 862                 }
 863                 return isset( $list[$element] )
 864                         ? $list[$element]
 865                         : array();
 866         }
 867
 868         /**
 869          * @return array
 870          */
 871         function setupAttributeWhitelist() {
 872                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
 873                 $block = array_merge( $common, array( 'align' ) );
 874                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
 875                 $tablecell = array( 'abbr',
 876                                     'axis',
 877                                     'headers',
 878                                     'scope',
 879                                     'rowspan',
 880                                     'colspan',
 881                                     'nowrap', # deprecated
 882                                     'width',  # deprecated
 883                                     'height', # deprecated
 884                                     'bgcolor' # deprecated
 885                                     );
 886
 887                 # Numbers refer to sections in HTML 4.01 standard describing the element.
 888                 # See: http://www.w3.org/TR/html4/
 889                 $whitelist = array (
 890                         # 7.5.4
 891                         'div'        => $block,
 892                         'center'     => $common, # deprecated
 893                         'span'       => $block, # ??
 894
 895                         # 7.5.5
 896                         'h1'         => $block,
 897                         'h2'         => $block,
 898                         'h3'         => $block,
 899                         'h4'         => $block,
 900                         'h5'         => $block,
 901                         'h6'         => $block,
 902
 903                         # 7.5.6
 904                         # address
 905
 906                         # 8.2.4
 907                         # bdo
 908
 909                         # 9.2.1
 910                         'em'         => $common,
 911                         'strong'     => $common,
 912                         'cite'       => $common,
 913                         # dfn
 914                         'code'       => $common,
 915                         # samp
 916                         # kbd
 917                         'var'        => $common,
 918                         # abbr
 919                         # acronym
 920
 921                         # 9.2.2
 922                         'blockquote' => array_merge( $common, array( 'cite' ) ),
 923                         # q
 924
 925                         # 9.2.3
 926                         'sub'        => $common,
 927                         'sup'        => $common,
 928
 929                         # 9.3.1
 930                         'p'          => $block,
 931
 932                         # 9.3.2
 933                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
 934
 935                         # 9.3.4
 936                         'pre'        => array_merge( $common, array( 'width' ) ),
 937
 938                         # 9.4
 939                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 940                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 941
 942                         # 10.2
 943                         'ul'         => array_merge( $common, array( 'type' ) ),
 944                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
 945                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
 946
 947                         # 10.3
 948                         'dl'         => $common,
 949                         'dd'         => $common,
 950                         'dt'         => $common,
 951
 952                         # 11.2.1
 953                         'table'      => array_merge( $common,
 954                                                                 array( 'summary', 'width', 'border', 'frame',
 955                                                                                          'rules', 'cellspacing', 'cellpadding',
 956                                                                                          'align', 'bgcolor', 'frame', 'rules',
 957                                                                                          'border' ) ),
 958
 959                         # 11.2.2
 960                         'caption'    => array_merge( $common, array( 'align' ) ),
 961
 962                         # 11.2.3
 963                         'thead'      => array_merge( $common, $tablealign ),
 964                         'tfoot'      => array_merge( $common, $tablealign ),
 965                         'tbody'      => array_merge( $common, $tablealign ),
 966
 967                         # 11.2.4
 968                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 969                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 970
 971                         # 11.2.5
 972                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
 973
 974                         # 11.2.6
 975                         'td'         => array_merge( $common, $tablecell, $tablealign ),
 976                         'th'         => array_merge( $common, $tablecell, $tablealign ),
 977
 978                         # 15.2.1
 979                         'tt'         => $common,
 980                         'b'          => $common,
 981                         'i'          => $common,
 982                         'big'        => $common,
 983                         'small'      => $common,
 984                         'strike'     => $common,
 985                         's'          => $common,
 986                         'u'          => $common,
 987
 988                         # 15.2.2
 989                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
 990                         # basefont
 991
 992                         # 15.3
 993                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
 994
 995                         # XHTML Ruby annotation text module, simple ruby only.
 996                         # http://www.w3c.org/TR/ruby/
 997                         'ruby'       => $common,
 998                         # rbc
 999                         # rtc
1000                         'rb'         => $common,
1001                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1002                         'rp'         => $common,
1003                         );
1004                 return $whitelist;
1005         }
1006
1007         /**
1008          * Take a fragment of (potentially invalid) HTML and return
1009          * a version with any tags removed, encoded suitably for literal
1010          * inclusion in an attribute value.
1011          *
1012          * @param string $text HTML fragment
1013          * @return string
1014          */
1015         function stripAllTags( $text ) {
1016                 # Actual <tags>
1017                 $text = preg_replace( '/<[^>]*>/', '', $text );
1018
1019                 # Normalize &entities and whitespace
1020                 $text = Sanitizer::normalizeAttributeValue( $text );
1021
1022                 # Will be placed into "double-quoted" attributes,
1023                 # make sure remaining bits are safe.
1024                 $text = str_replace(
1025                         array('<', '>', '"'),
1026                         array('&lt;', '&gt;', '&quot;'),
1027                         $text );
1028
1029                 return $text;
1030         }
1031
1032 }
1033
1034 ?>