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  * 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 ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 388                                                         @array_push( $tagstack, $ot );
 389                                                         $badtag = 1;
 390                                                 } else {
 391                                                         if ( $t == 'table' ) {
 392                                                                 $tagstack = array_pop( $tablestack );
 393                                                         }
 394                                                         $newparams = '';
 395                                                 }
 396                                         } else {
 397                                                 # Keep track for later
 398                                                 if ( in_array( $t, $tabletags ) &&
 399                                                 ! in_array( 'table', $tagstack ) ) {
 400                                                         $badtag = 1;
 401                                                 } else if ( in_array( $t, $tagstack ) &&
 402                                                 ! in_array ( $t , $htmlnest ) ) {
 403                                                         $badtag = 1 ;
 404                                                 } elseif( in_array( $t, $htmlsingleonly ) ) {
 405                                                         # Hack to force empty tag for uncloseable elements
 406                                                         $brace = '/>';
 407                                                 } else {
 408                                                         if ( $t == 'table' ) {
 409                                                                 array_push( $tablestack, $tagstack );
 410                                                                 $tagstack = array();
 411                                                         }
 412                                                         array_push( $tagstack, $t );
 413                                                 }
 414
 415                                                 # Replace any variables or template parameters with
 416                                                 # plaintext results.
 417                                                 if( is_callable( $processCallback ) ) {
 418                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 419                                                 }
 420
 421                                                 # Strip non-approved attributes from the tag
 422                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 423                                         }
 424                                         if ( ! $badtag ) {
 425                                                 $rest = str_replace( '>', '&gt;', $rest );
 426                                                 $close = ( $brace == '/>' ) ? ' /' : '';
 427                                                 $text .= "<$slash$t$newparams$close>$rest";
 428                                                 continue;
 429                                         }
 430                                 }
 431                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 432                         }
 433                         # Close off any remaining tags
 434                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 435                                 $text .= "</$t>\n";
 436                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 437                         }
 438                 } else {
 439                         # this might be possible using tidy itself
 440                         foreach ( $bits as $x ) {
 441                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 442                                 $x, $regs );
 443                                 @list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
 444                                 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
 445                                         if( is_callable( $processCallback ) ) {
 446                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 447                                         }
 448                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 449                                         $rest = str_replace( '>', '&gt;', $rest );
 450                                         $text .= "<$slash$t$newparams$brace$rest";
 451                                 } else {
 452                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 453                                 }
 454                         }
 455                 }
 456                 wfProfileOut( $fname );
 457                 return $text;
 458         }
 459
 460         /**
 461          * Remove '<!--', '-->', and everything between.
 462          * To avoid leaving blank lines, when a comment is both preceded
 463          * and followed by a newline (ignoring spaces), trim leading and
 464          * trailing spaces and one of the newlines.
 465          *
 466          * @access private
 467          * @param string $text
 468          * @return string
 469          */
 470         function removeHTMLcomments( $text ) {
 471                 $fname='Parser::removeHTMLcomments';
 472                 wfProfileIn( $fname );
 473                 while (($start = strpos($text, '<!--')) !== false) {
 474                         $end = strpos($text, '-->', $start + 4);
 475                         if ($end === false) {
 476                                 # Unterminated comment; bail out
 477                                 break;
 478                         }
 479
 480                         $end += 3;
 481
 482                         # Trim space and newline if the comment is both
 483                         # preceded and followed by a newline
 484                         $spaceStart = max($start - 1, 0);
 485                         $spaceLen = $end - $spaceStart;
 486                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 487                                 $spaceStart--;
 488                                 $spaceLen++;
 489                         }
 490                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 491                                 $spaceLen++;
 492                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 493                                 # Remove the comment, leading and trailing
 494                                 # spaces, and leave only one newline.
 495                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 496                         }
 497                         else {
 498                                 # Remove just the comment.
 499                                 $text = substr_replace($text, '', $start, $end - $start);
 500                         }
 501                 }
 502                 wfProfileOut( $fname );
 503                 return $text;
 504         }
 505
 506         /**
 507          * Take a tag soup fragment listing an HTML element's attributes
 508          * and normalize it to well-formed XML, discarding unwanted attributes.
 509          *
 510          * - Normalizes attribute names to lowercase
 511          * - Discards attributes not on a whitelist for the given element
 512          * - Turns broken or invalid entities into plaintext
 513          * - Double-quotes all attribute values
 514          * - Attributes without values are given the name as attribute
 515          * - Double attributes are discarded
 516          * - Unsafe style attributes are discarded
 517          * - Prepends space if there are attributes.
 518          *
 519          * @param string $text
 520          * @param string $element
 521          * @return string
 522          *
 523          * @todo Check for legal values where the DTD limits things.
 524          * @todo Check for unique id attribute :P
 525          */
 526         function fixTagAttributes( $text, $element ) {
 527                 if( trim( $text ) == '' ) {
 528                         return '';
 529                 }
 530
 531                 # Unquoted attribute
 532                 # Since we quote this later, this can be anything distinguishable
 533                 # from the end of the attribute
 534                 $pairs = array();
 535                 if( !preg_match_all(
 536                         MW_ATTRIBS_REGEX,
 537                         $text,
 538                         $pairs,
 539                         PREG_SET_ORDER ) ) {
 540                         return '';
 541                 }
 542
 543                 $whitelist = array_flip( Sanitizer::attributeWhitelist( $element ) );
 544                 $attribs = array();
 545                 foreach( $pairs as $set ) {
 546                         $attribute = strtolower( $set[1] );
 547                         if( !isset( $whitelist[$attribute] ) ) {
 548                                 continue;
 549                         }
 550
 551                         $raw   = Sanitizer::getTagAttributeCallback( $set );
 552                         $value = Sanitizer::normalizeAttributeValue( $raw );
 553
 554                         # Strip javascript "expression" from stylesheets.
 555                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 556                         if( $attribute == 'style' ) {
 557                                 $stripped = Sanitizer::decodeCharReferences( $value );
 558
 559                                 // Remove any comments; IE gets token splitting wrong
 560                                 $stripped = preg_replace( '!/\\*.*?\\*/!S', ' ', $stripped );
 561                                 $value = htmlspecialchars( $stripped );
 562
 563                                 // ... and continue checks
 564                                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 565                                         'codepointToUtf8(hexdec("$1"))', $stripped );
 566                                 $stripped = str_replace( '\\', '', $stripped );
 567                                 if( preg_match( '/(expression|tps*:\/\/|url\\s*\().*/is',
 568                                                 $stripped ) ) {
 569                                         # haxx0r
 570                                         continue;
 571                                 }
 572                         }
 573
 574                         if ( $attribute === 'id' )
 575                                 $value = Sanitizer::escapeId( $value );
 576
 577                         # Templates and links may be expanded in later parsing,
 578                         # creating invalid or dangerous output. Suppress this.
 579                         $value = strtr( $value, array(
 580                                 '<'    => '&lt;',   // This should never happen,
 581                                 '>'    => '&gt;',   // we've received invalid input
 582                                 '"'    => '&quot;', // which should have been escaped.
 583                                 '{'    => '&#123;',
 584                                 '['    => '&#91;',
 585                                 "''"   => '&#39;&#39;',
 586                                 'ISBN' => '&#73;SBN',
 587                                 'RFC'  => '&#82;FC',
 588                                 'PMID' => '&#80;MID',
 589                         ) );
 590
 591                         # Stupid hack
 592                         $value = preg_replace_callback(
 593                                 '/(' . wfUrlProtocols() . ')/',
 594                                 array( 'Sanitizer', 'armorLinksCallback' ),
 595                                 $value );
 596
 597                         // If this attribute was previously set, override it.
 598                         // Output should only have one attribute of each name.
 599                         $attribs[$attribute] = "$attribute=\"$value\"";
 600                 }
 601
 602                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 603         }
 604
 605         /**
 606          * Given a value escape it so that it can be used in an id attribute and
 607          * return it, this does not validate the value however (see first link)
 608          *
 609          * @link http://www.w3.org/TR/html401/types.html#type-name Valid characters
 610          *                                                          in the id and
 611          *                                                          name attributes
 612          * @link http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 613          *
 614          * @bug 4461
 615          *
 616          * @static
 617          *
 618          * @param string $id
 619          * @return string
 620          */
 621         function escapeId( $id ) {
 622                 static $replace = array(
 623                         '%3A' => ':',
 624                         '%' => '.'
 625                 );
 626
 627                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 628
 629                 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
 630         }
 631
 632         /**
 633          * Regex replace callback for armoring links against further processing.
 634          * @param array $matches
 635          * @return string
 636          * @access private
 637          */
 638         function armorLinksCallback( $matches ) {
 639                 return str_replace( ':', '&#58;', $matches[1] );
 640         }
 641
 642         /**
 643          * Return an associative array of attribute names and values from
 644          * a partial tag string. Attribute names are forces to lowercase,
 645          * character references are decoded to UTF-8 text.
 646          *
 647          * @param string
 648          * @return array
 649          */
 650         function decodeTagAttributes( $text ) {
 651                 $attribs = array();
 652
 653                 if( trim( $text ) == '' ) {
 654                         return $attribs;
 655                 }
 656
 657                 $pairs = array();
 658                 if( !preg_match_all(
 659                         MW_ATTRIBS_REGEX,
 660                         $text,
 661                         $pairs,
 662                         PREG_SET_ORDER ) ) {
 663                         return $attribs;
 664                 }
 665
 666                 foreach( $pairs as $set ) {
 667                         $attribute = strtolower( $set[1] );
 668                         $value = Sanitizer::getTagAttributeCallback( $set );
 669                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 670                 }
 671                 return $attribs;
 672         }
 673
 674         /**
 675          * Pick the appropriate attribute value from a match set from the
 676          * MW_ATTRIBS_REGEX matches.
 677          *
 678          * @param array $set
 679          * @return string
 680          * @access private
 681          */
 682         function getTagAttributeCallback( $set ) {
 683                 if( isset( $set[6] ) ) {
 684                         # Illegal #XXXXXX color with no quotes.
 685                         return $set[6];
 686                 } elseif( isset( $set[5] ) ) {
 687                         # No quotes.
 688                         return $set[5];
 689                 } elseif( isset( $set[4] ) ) {
 690                         # Single-quoted
 691                         return $set[4];
 692                 } elseif( isset( $set[3] ) ) {
 693                         # Double-quoted
 694                         return $set[3];
 695                 } elseif( !isset( $set[2] ) ) {
 696                         # In XHTML, attributes must have a value.
 697                         # For 'reduced' form, return explicitly the attribute name here.
 698                         return $set[1];
 699                 } else {
 700                         wfDebugDieBacktrace( "Tag conditions not met. This should never happen and is a bug." );
 701                 }
 702         }
 703
 704         /**
 705          * Normalize whitespace and character references in an XML source-
 706          * encoded text for an attribute value.
 707          *
 708          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 709          * but note that we're not returning the value, but are returning
 710          * XML source fragments that will be slapped into output.
 711          *
 712          * @param string $text
 713          * @return string
 714          * @access private
 715          */
 716         function normalizeAttributeValue( $text ) {
 717                 return str_replace( '"', '&quot;',
 718                         preg_replace(
 719                                 '/\r\n|[\x20\x0d\x0a\x09]/',
 720                                 ' ',
 721                                 Sanitizer::normalizeCharReferences( $text ) ) );
 722         }
 723
 724         /**
 725          * Ensure that any entities and character references are legal
 726          * for XML and XHTML specifically. Any stray bits will be
 727          * &amp;-escaped to result in a valid text fragment.
 728          *
 729          * a. any named char refs must be known in XHTML
 730          * b. any numeric char refs must be legal chars, not invalid or forbidden
 731          * c. use &#x, not &#X
 732          * d. fix or reject non-valid attributes
 733          *
 734          * @param string $text
 735          * @return string
 736          * @access private
 737          */
 738         function normalizeCharReferences( $text ) {
 739                 return preg_replace_callback(
 740                         MW_CHAR_REFS_REGEX,
 741                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 742                         $text );
 743         }
 744         /**
 745          * @param string $matches
 746          * @return string
 747          */
 748         function normalizeCharReferencesCallback( $matches ) {
 749                 $ret = null;
 750                 if( $matches[1] != '' ) {
 751                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 752                 } elseif( $matches[2] != '' ) {
 753                         $ret = Sanitizer::decCharReference( $matches[2] );
 754                 } elseif( $matches[3] != ''  ) {
 755                         $ret = Sanitizer::hexCharReference( $matches[3] );
 756                 } elseif( $matches[4] != '' ) {
 757                         $ret = Sanitizer::hexCharReference( $matches[4] );
 758                 }
 759                 if( is_null( $ret ) ) {
 760                         return htmlspecialchars( $matches[0] );
 761                 } else {
 762                         return $ret;
 763                 }
 764         }
 765
 766         /**
 767          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 768          * return the named entity reference as is. Otherwise, returns
 769          * HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 770          *
 771          * @param string $name
 772          * @return string
 773          */
 774         function normalizeEntity( $name ) {
 775                 global $wgHtmlEntities;
 776                 if( isset( $wgHtmlEntities[$name] ) ) {
 777                         return "&$name;";
 778                 } else {
 779                         return "&amp;$name;";
 780                 }
 781         }
 782
 783         function decCharReference( $codepoint ) {
 784                 $point = intval( $codepoint );
 785                 if( Sanitizer::validateCodepoint( $point ) ) {
 786                         return sprintf( '&#%d;', $point );
 787                 } else {
 788                         return null;
 789                 }
 790         }
 791
 792         function hexCharReference( $codepoint ) {
 793                 $point = hexdec( $codepoint );
 794                 if( Sanitizer::validateCodepoint( $point ) ) {
 795                         return sprintf( '&#x%x;', $point );
 796                 } else {
 797                         return null;
 798                 }
 799         }
 800
 801         /**
 802          * Returns true if a given Unicode codepoint is a valid character in XML.
 803          * @param int $codepoint
 804          * @return bool
 805          */
 806         function validateCodepoint( $codepoint ) {
 807                 return ($codepoint ==    0x09)
 808                         || ($codepoint ==    0x0a)
 809                         || ($codepoint ==    0x0d)
 810                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 811                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 812                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 813         }
 814
 815         /**
 816          * Decode any character references, numeric or named entities,
 817          * in the text and return a UTF-8 string.
 818          *
 819          * @param string $text
 820          * @return string
 821          * @access public
 822          */
 823         function decodeCharReferences( $text ) {
 824                 return preg_replace_callback(
 825                         MW_CHAR_REFS_REGEX,
 826                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 827                         $text );
 828         }
 829
 830         /**
 831          * @param string $matches
 832          * @return string
 833          */
 834         function decodeCharReferencesCallback( $matches ) {
 835                 if( $matches[1] != '' ) {
 836                         return Sanitizer::decodeEntity( $matches[1] );
 837                 } elseif( $matches[2] != '' ) {
 838                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 839                 } elseif( $matches[3] != ''  ) {
 840                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 841                 } elseif( $matches[4] != '' ) {
 842                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 843                 }
 844                 # Last case should be an ampersand by itself
 845                 return $matches[0];
 846         }
 847
 848         /**
 849          * Return UTF-8 string for a codepoint if that is a valid
 850          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
 851          * @param int $codepoint
 852          * @return string
 853          * @access private
 854          */
 855         function decodeChar( $codepoint ) {
 856                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
 857                         return codepointToUtf8( $codepoint );
 858                 } else {
 859                         return UTF8_REPLACEMENT;
 860                 }
 861         }
 862
 863         /**
 864          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 865          * return the UTF-8 encoding of that character. Otherwise, returns
 866          * pseudo-entity source (eg &foo;)
 867          *
 868          * @param string $name
 869          * @return string
 870          */
 871         function decodeEntity( $name ) {
 872                 global $wgHtmlEntities;
 873                 if( isset( $wgHtmlEntities[$name] ) ) {
 874                         return codepointToUtf8( $wgHtmlEntities[$name] );
 875                 } else {
 876                         return "&$name;";
 877                 }
 878         }
 879
 880         /**
 881          * Fetch the whitelist of acceptable attributes for a given
 882          * element name.
 883          *
 884          * @param string $element
 885          * @return array
 886          */
 887         function attributeWhitelist( $element ) {
 888                 static $list;
 889                 if( !isset( $list ) ) {
 890                         $list = Sanitizer::setupAttributeWhitelist();
 891                 }
 892                 return isset( $list[$element] )
 893                         ? $list[$element]
 894                         : array();
 895         }
 896
 897         /**
 898          * @return array
 899          */
 900         function setupAttributeWhitelist() {
 901                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
 902                 $block = array_merge( $common, array( 'align' ) );
 903                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
 904                 $tablecell = array( 'abbr',
 905                                     'axis',
 906                                     'headers',
 907                                     'scope',
 908                                     'rowspan',
 909                                     'colspan',
 910                                     'nowrap', # deprecated
 911                                     'width',  # deprecated
 912                                     'height', # deprecated
 913                                     'bgcolor' # deprecated
 914                                     );
 915
 916                 # Numbers refer to sections in HTML 4.01 standard describing the element.
 917                 # See: http://www.w3.org/TR/html4/
 918                 $whitelist = array (
 919                         # 7.5.4
 920                         'div'        => $block,
 921                         'center'     => $common, # deprecated
 922                         'span'       => $block, # ??
 923
 924                         # 7.5.5
 925                         'h1'         => $block,
 926                         'h2'         => $block,
 927                         'h3'         => $block,
 928                         'h4'         => $block,
 929                         'h5'         => $block,
 930                         'h6'         => $block,
 931
 932                         # 7.5.6
 933                         # address
 934
 935                         # 8.2.4
 936                         # bdo
 937
 938                         # 9.2.1
 939                         'em'         => $common,
 940                         'strong'     => $common,
 941                         'cite'       => $common,
 942                         # dfn
 943                         'code'       => $common,
 944                         # samp
 945                         # kbd
 946                         'var'        => $common,
 947                         # abbr
 948                         # acronym
 949
 950                         # 9.2.2
 951                         'blockquote' => array_merge( $common, array( 'cite' ) ),
 952                         # q
 953
 954                         # 9.2.3
 955                         'sub'        => $common,
 956                         'sup'        => $common,
 957
 958                         # 9.3.1
 959                         'p'          => $block,
 960
 961                         # 9.3.2
 962                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
 963
 964                         # 9.3.4
 965                         'pre'        => array_merge( $common, array( 'width' ) ),
 966
 967                         # 9.4
 968                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 969                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 970
 971                         # 10.2
 972                         'ul'         => array_merge( $common, array( 'type' ) ),
 973                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
 974                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
 975
 976                         # 10.3
 977                         'dl'         => $common,
 978                         'dd'         => $common,
 979                         'dt'         => $common,
 980
 981                         # 11.2.1
 982                         'table'      => array_merge( $common,
 983                                                                 array( 'summary', 'width', 'border', 'frame',
 984                                                                                          'rules', 'cellspacing', 'cellpadding',
 985                                                                                          'align', 'bgcolor', 'frame', 'rules',
 986                                                                                          'border' ) ),
 987
 988                         # 11.2.2
 989                         'caption'    => array_merge( $common, array( 'align' ) ),
 990
 991                         # 11.2.3
 992                         'thead'      => array_merge( $common, $tablealign ),
 993                         'tfoot'      => array_merge( $common, $tablealign ),
 994                         'tbody'      => array_merge( $common, $tablealign ),
 995
 996                         # 11.2.4
 997                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 998                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 999
1000                         # 11.2.5
1001                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1002
1003                         # 11.2.6
1004                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1005                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1006
1007                         # 15.2.1
1008                         'tt'         => $common,
1009                         'b'          => $common,
1010                         'i'          => $common,
1011                         'big'        => $common,
1012                         'small'      => $common,
1013                         'strike'     => $common,
1014                         's'          => $common,
1015                         'u'          => $common,
1016
1017                         # 15.2.2
1018                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1019                         # basefont
1020
1021                         # 15.3
1022                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1023
1024                         # XHTML Ruby annotation text module, simple ruby only.
1025                         # http://www.w3c.org/TR/ruby/
1026                         'ruby'       => $common,
1027                         # rbc
1028                         # rtc
1029                         'rb'         => $common,
1030                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1031                         'rp'         => $common,
1032                         );
1033                 return $whitelist;
1034         }
1035
1036         /**
1037          * Take a fragment of (potentially invalid) HTML and return
1038          * a version with any tags removed, encoded suitably for literal
1039          * inclusion in an attribute value.
1040          *
1041          * @param string $text HTML fragment
1042          * @return string
1043          */
1044         function stripAllTags( $text ) {
1045                 # Actual <tags>
1046                 $text = preg_replace( '/ < .*? > /x', '', $text );
1047
1048                 # Normalize &entities and whitespace
1049                 $text = Sanitizer::normalizeAttributeValue( $text );
1050
1051                 # Will be placed into "double-quoted" attributes,
1052                 # make sure remaining bits are safe.
1053                 $text = str_replace(
1054                         array('<', '>', '"'),
1055                         array('&lt;', '&gt;', '&quot;'),
1056                         $text );
1057
1058                 return $text;
1059         }
1060
1061         /**
1062          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1063          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1064          * PHP 5.1 doesn't.
1065          *
1066          * Use for passing XHTML fragments to PHP's XML parsing functions
1067          *
1068          * @return string
1069          * @static
1070          */
1071         function hackDocType() {
1072                 global $wgHtmlEntities;
1073                 $out = "<!DOCTYPE html [\n";
1074                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1075                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1076                 }
1077                 $out .= "]>\n";
1078                 return $out;
1079         }
1080
1081 }
1082
1083 ?>