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                                 '{'    => '&#123;',
 581                                 '['    => '&#91;',
 582                                 "''"   => '&#39;&#39;',
 583                                 'ISBN' => '&#73;SBN',
 584                                 'RFC'  => '&#82;FC',
 585                                 'PMID' => '&#80;MID',
 586                         ) );
 587
 588                         # Stupid hack
 589                         $value = preg_replace_callback(
 590                                 '/(' . wfUrlProtocols() . ')/',
 591                                 array( 'Sanitizer', 'armorLinksCallback' ),
 592                                 $value );
 593
 594                         // If this attribute was previously set, override it.
 595                         // Output should only have one attribute of each name.
 596                         $attribs[$attribute] = "$attribute=\"$value\"";
 597                 }
 598
 599                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 600         }
 601
 602         /**
 603          * Given a value escape it so that it can be used in an id attribute and
 604          * return it, this does not validate the value however (see first link)
 605          *
 606          * @link http://www.w3.org/TR/html401/types.html#type-name Valid characters
 607          *                                                          in the id and
 608          *                                                          name attributes
 609          * @link http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 610          *
 611          * @bug 4461
 612          *
 613          * @static
 614          *
 615          * @param string $id
 616          * @return string
 617          */
 618         function escapeId( $id ) {
 619                 static $replace = array(
 620                         '%3A' => ':',
 621                         '%' => '.'
 622                 );
 623
 624                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 625
 626                 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
 627         }
 628
 629         /**
 630          * Regex replace callback for armoring links against further processing.
 631          * @param array $matches
 632          * @return string
 633          * @access private
 634          */
 635         function armorLinksCallback( $matches ) {
 636                 return str_replace( ':', '&#58;', $matches[1] );
 637         }
 638
 639         /**
 640          * Return an associative array of attribute names and values from
 641          * a partial tag string. Attribute names are forces to lowercase,
 642          * character references are decoded to UTF-8 text.
 643          *
 644          * @param string
 645          * @return array
 646          */
 647         function decodeTagAttributes( $text ) {
 648                 $attribs = array();
 649
 650                 if( trim( $text ) == '' ) {
 651                         return $attribs;
 652                 }
 653
 654                 $pairs = array();
 655                 if( !preg_match_all(
 656                         MW_ATTRIBS_REGEX,
 657                         $text,
 658                         $pairs,
 659                         PREG_SET_ORDER ) ) {
 660                         return $attribs;
 661                 }
 662
 663                 foreach( $pairs as $set ) {
 664                         $attribute = strtolower( $set[1] );
 665                         $value = Sanitizer::getTagAttributeCallback( $set );
 666                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 667                 }
 668                 return $attribs;
 669         }
 670
 671         /**
 672          * Pick the appropriate attribute value from a match set from the
 673          * MW_ATTRIBS_REGEX matches.
 674          *
 675          * @param array $set
 676          * @return string
 677          * @access private
 678          */
 679         function getTagAttributeCallback( $set ) {
 680                 if( isset( $set[6] ) ) {
 681                         # Illegal #XXXXXX color with no quotes.
 682                         return $set[6];
 683                 } elseif( isset( $set[5] ) ) {
 684                         # No quotes.
 685                         return $set[5];
 686                 } elseif( isset( $set[4] ) ) {
 687                         # Single-quoted
 688                         return $set[4];
 689                 } elseif( isset( $set[3] ) ) {
 690                         # Double-quoted
 691                         return $set[3];
 692                 } elseif( !isset( $set[2] ) ) {
 693                         # In XHTML, attributes must have a value.
 694                         # For 'reduced' form, return explicitly the attribute name here.
 695                         return $set[1];
 696                 } else {
 697                         wfDebugDieBacktrace( "Tag conditions not met. This should never happen and is a bug." );
 698                 }
 699         }
 700
 701         /**
 702          * Normalize whitespace and character references in an XML source-
 703          * encoded text for an attribute value.
 704          *
 705          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 706          * but note that we're not returning the value, but are returning
 707          * XML source fragments that will be slapped into output.
 708          *
 709          * @param string $text
 710          * @return string
 711          * @access private
 712          */
 713         function normalizeAttributeValue( $text ) {
 714                 return str_replace( '"', '&quot;',
 715                         preg_replace(
 716                                 '/\r\n|[\x20\x0d\x0a\x09]/',
 717                                 ' ',
 718                                 Sanitizer::normalizeCharReferences( $text ) ) );
 719         }
 720
 721         /**
 722          * Ensure that any entities and character references are legal
 723          * for XML and XHTML specifically. Any stray bits will be
 724          * &amp;-escaped to result in a valid text fragment.
 725          *
 726          * a. any named char refs must be known in XHTML
 727          * b. any numeric char refs must be legal chars, not invalid or forbidden
 728          * c. use &#x, not &#X
 729          * d. fix or reject non-valid attributes
 730          *
 731          * @param string $text
 732          * @return string
 733          * @access private
 734          */
 735         function normalizeCharReferences( $text ) {
 736                 return preg_replace_callback(
 737                         MW_CHAR_REFS_REGEX,
 738                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 739                         $text );
 740         }
 741         /**
 742          * @param string $matches
 743          * @return string
 744          */
 745         function normalizeCharReferencesCallback( $matches ) {
 746                 $ret = null;
 747                 if( $matches[1] != '' ) {
 748                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 749                 } elseif( $matches[2] != '' ) {
 750                         $ret = Sanitizer::decCharReference( $matches[2] );
 751                 } elseif( $matches[3] != ''  ) {
 752                         $ret = Sanitizer::hexCharReference( $matches[3] );
 753                 } elseif( $matches[4] != '' ) {
 754                         $ret = Sanitizer::hexCharReference( $matches[4] );
 755                 }
 756                 if( is_null( $ret ) ) {
 757                         return htmlspecialchars( $matches[0] );
 758                 } else {
 759                         return $ret;
 760                 }
 761         }
 762
 763         /**
 764          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 765          * return the named entity reference as is. Otherwise, returns
 766          * HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 767          *
 768          * @param string $name
 769          * @return string
 770          */
 771         function normalizeEntity( $name ) {
 772                 global $wgHtmlEntities;
 773                 if( isset( $wgHtmlEntities[$name] ) ) {
 774                         return "&$name;";
 775                 } else {
 776                         return "&amp;$name;";
 777                 }
 778         }
 779
 780         function decCharReference( $codepoint ) {
 781                 $point = intval( $codepoint );
 782                 if( Sanitizer::validateCodepoint( $point ) ) {
 783                         return sprintf( '&#%d;', $point );
 784                 } else {
 785                         return null;
 786                 }
 787         }
 788
 789         function hexCharReference( $codepoint ) {
 790                 $point = hexdec( $codepoint );
 791                 if( Sanitizer::validateCodepoint( $point ) ) {
 792                         return sprintf( '&#x%x;', $point );
 793                 } else {
 794                         return null;
 795                 }
 796         }
 797
 798         /**
 799          * Returns true if a given Unicode codepoint is a valid character in XML.
 800          * @param int $codepoint
 801          * @return bool
 802          */
 803         function validateCodepoint( $codepoint ) {
 804                 return ($codepoint ==    0x09)
 805                         || ($codepoint ==    0x0a)
 806                         || ($codepoint ==    0x0d)
 807                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 808                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 809                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 810         }
 811
 812         /**
 813          * Decode any character references, numeric or named entities,
 814          * in the text and return a UTF-8 string.
 815          *
 816          * @param string $text
 817          * @return string
 818          * @access public
 819          */
 820         function decodeCharReferences( $text ) {
 821                 return preg_replace_callback(
 822                         MW_CHAR_REFS_REGEX,
 823                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 824                         $text );
 825         }
 826
 827         /**
 828          * @param string $matches
 829          * @return string
 830          */
 831         function decodeCharReferencesCallback( $matches ) {
 832                 if( $matches[1] != '' ) {
 833                         return Sanitizer::decodeEntity( $matches[1] );
 834                 } elseif( $matches[2] != '' ) {
 835                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 836                 } elseif( $matches[3] != ''  ) {
 837                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 838                 } elseif( $matches[4] != '' ) {
 839                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 840                 }
 841                 # Last case should be an ampersand by itself
 842                 return $matches[0];
 843         }
 844
 845         /**
 846          * Return UTF-8 string for a codepoint if that is a valid
 847          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
 848          * @param int $codepoint
 849          * @return string
 850          * @access private
 851          */
 852         function decodeChar( $codepoint ) {
 853                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
 854                         return codepointToUtf8( $codepoint );
 855                 } else {
 856                         return UTF8_REPLACEMENT;
 857                 }
 858         }
 859
 860         /**
 861          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 862          * return the UTF-8 encoding of that character. Otherwise, returns
 863          * pseudo-entity source (eg &foo;)
 864          *
 865          * @param string $name
 866          * @return string
 867          */
 868         function decodeEntity( $name ) {
 869                 global $wgHtmlEntities;
 870                 if( isset( $wgHtmlEntities[$name] ) ) {
 871                         return codepointToUtf8( $wgHtmlEntities[$name] );
 872                 } else {
 873                         return "&$name;";
 874                 }
 875         }
 876
 877         /**
 878          * Fetch the whitelist of acceptable attributes for a given
 879          * element name.
 880          *
 881          * @param string $element
 882          * @return array
 883          */
 884         function attributeWhitelist( $element ) {
 885                 static $list;
 886                 if( !isset( $list ) ) {
 887                         $list = Sanitizer::setupAttributeWhitelist();
 888                 }
 889                 return isset( $list[$element] )
 890                         ? $list[$element]
 891                         : array();
 892         }
 893
 894         /**
 895          * @return array
 896          */
 897         function setupAttributeWhitelist() {
 898                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
 899                 $block = array_merge( $common, array( 'align' ) );
 900                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
 901                 $tablecell = array( 'abbr',
 902                                     'axis',
 903                                     'headers',
 904                                     'scope',
 905                                     'rowspan',
 906                                     'colspan',
 907                                     'nowrap', # deprecated
 908                                     'width',  # deprecated
 909                                     'height', # deprecated
 910                                     'bgcolor' # deprecated
 911                                     );
 912
 913                 # Numbers refer to sections in HTML 4.01 standard describing the element.
 914                 # See: http://www.w3.org/TR/html4/
 915                 $whitelist = array (
 916                         # 7.5.4
 917                         'div'        => $block,
 918                         'center'     => $common, # deprecated
 919                         'span'       => $block, # ??
 920
 921                         # 7.5.5
 922                         'h1'         => $block,
 923                         'h2'         => $block,
 924                         'h3'         => $block,
 925                         'h4'         => $block,
 926                         'h5'         => $block,
 927                         'h6'         => $block,
 928
 929                         # 7.5.6
 930                         # address
 931
 932                         # 8.2.4
 933                         # bdo
 934
 935                         # 9.2.1
 936                         'em'         => $common,
 937                         'strong'     => $common,
 938                         'cite'       => $common,
 939                         # dfn
 940                         'code'       => $common,
 941                         # samp
 942                         # kbd
 943                         'var'        => $common,
 944                         # abbr
 945                         # acronym
 946
 947                         # 9.2.2
 948                         'blockquote' => array_merge( $common, array( 'cite' ) ),
 949                         # q
 950
 951                         # 9.2.3
 952                         'sub'        => $common,
 953                         'sup'        => $common,
 954
 955                         # 9.3.1
 956                         'p'          => $block,
 957
 958                         # 9.3.2
 959                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
 960
 961                         # 9.3.4
 962                         'pre'        => array_merge( $common, array( 'width' ) ),
 963
 964                         # 9.4
 965                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 966                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 967
 968                         # 10.2
 969                         'ul'         => array_merge( $common, array( 'type' ) ),
 970                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
 971                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
 972
 973                         # 10.3
 974                         'dl'         => $common,
 975                         'dd'         => $common,
 976                         'dt'         => $common,
 977
 978                         # 11.2.1
 979                         'table'      => array_merge( $common,
 980                                                                 array( 'summary', 'width', 'border', 'frame',
 981                                                                                          'rules', 'cellspacing', 'cellpadding',
 982                                                                                          'align', 'bgcolor', 'frame', 'rules',
 983                                                                                          'border' ) ),
 984
 985                         # 11.2.2
 986                         'caption'    => array_merge( $common, array( 'align' ) ),
 987
 988                         # 11.2.3
 989                         'thead'      => array_merge( $common, $tablealign ),
 990                         'tfoot'      => array_merge( $common, $tablealign ),
 991                         'tbody'      => array_merge( $common, $tablealign ),
 992
 993                         # 11.2.4
 994                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 995                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
 996
 997                         # 11.2.5
 998                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
 999
1000                         # 11.2.6
1001                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1002                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1003
1004                         # 15.2.1
1005                         'tt'         => $common,
1006                         'b'          => $common,
1007                         'i'          => $common,
1008                         'big'        => $common,
1009                         'small'      => $common,
1010                         'strike'     => $common,
1011                         's'          => $common,
1012                         'u'          => $common,
1013
1014                         # 15.2.2
1015                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1016                         # basefont
1017
1018                         # 15.3
1019                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1020
1021                         # XHTML Ruby annotation text module, simple ruby only.
1022                         # http://www.w3c.org/TR/ruby/
1023                         'ruby'       => $common,
1024                         # rbc
1025                         # rtc
1026                         'rb'         => $common,
1027                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1028                         'rp'         => $common,
1029                         );
1030                 return $whitelist;
1031         }
1032
1033         /**
1034          * Take a fragment of (potentially invalid) HTML and return
1035          * a version with any tags removed, encoded suitably for literal
1036          * inclusion in an attribute value.
1037          *
1038          * @param string $text HTML fragment
1039          * @return string
1040          */
1041         function stripAllTags( $text ) {
1042                 # Actual <tags>
1043                 $text = preg_replace( '/ < .*? > /x', '', $text );
1044
1045                 # Normalize &entities and whitespace
1046                 $text = Sanitizer::normalizeAttributeValue( $text );
1047
1048                 # Will be placed into "double-quoted" attributes,
1049                 # make sure remaining bits are safe.
1050                 $text = str_replace(
1051                         array('<', '>', '"'),
1052                         array('&lt;', '&gt;', '&quot;'),
1053                         $text );
1054
1055                 return $text;
1056         }
1057
1058         /**
1059          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1060          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1061          * PHP 5.1 doesn't.
1062          *
1063          * Use for passing XHTML fragments to PHP's XML parsing functions
1064          *
1065          * @return string
1066          * @static
1067          */
1068         function hackDocType() {
1069                 global $wgHtmlEntities;
1070                 $out = "<!DOCTYPE html [\n";
1071                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1072                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1073                 }
1074                 $out .= "]>\n";
1075                 return $out;
1076         }
1077
1078 }
1079
1080 ?>