includes/Sanitizer.php

   1 <?php
   2 /**
   3  * XHTML sanitizer for MediaWiki
   4  *
   5  * Copyright (C) 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @addtogroup Parser
  24  */
  25
  26 /**
  27  * Regular expression to match various types of character references in
  28  * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  29  */
  30 define( 'MW_CHAR_REFS_REGEX',
  31         '/&([A-Za-z0-9\x80-\xff]+);
  32          |&\#([0-9]+);
  33          |&\#x([0-9A-Za-z]+);
  34          |&\#X([0-9A-Za-z]+);
  35          |(&)/x' );
  36
  37 /**
  38  * Regular expression to match HTML/XML attribute pairs within a tag.
  39  * Allows some... latitude.
  40  * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
  41  */
  42 $attrib = '[A-Za-z0-9]';
  43 $space = '[\x09\x0a\x0d\x20]';
  44 define( 'MW_ATTRIBS_REGEX',
  45         "/(?:^|$space)($attrib+)
  46           ($space*=$space*
  47                 (?:
  48                  # The attribute value: quoted or alone
  49                   \"([^<\"]*)\"
  50                  | '([^<']*)'
  51                  |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
  52                  |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
  53                                                          # colors are specified like this.
  54                                                          # We'll be normalizing it.
  55                 )
  56            )?(?=$space|\$)/sx" );
  57
  58 /**
  59  * List of all named character entities defined in HTML 4.01
  60  * http://www.w3.org/TR/html4/sgml/entities.html
  61  * @private
  62  */
  63 global $wgHtmlEntities;
  64 $wgHtmlEntities = array(
  65         'Aacute'   => 193,
  66         'aacute'   => 225,
  67         'Acirc'    => 194,
  68         'acirc'    => 226,
  69         'acute'    => 180,
  70         'AElig'    => 198,
  71         'aelig'    => 230,
  72         'Agrave'   => 192,
  73         'agrave'   => 224,
  74         'alefsym'  => 8501,
  75         'Alpha'    => 913,
  76         'alpha'    => 945,
  77         'amp'      => 38,
  78         'and'      => 8743,
  79         'ang'      => 8736,
  80         'Aring'    => 197,
  81         'aring'    => 229,
  82         'asymp'    => 8776,
  83         'Atilde'   => 195,
  84         'atilde'   => 227,
  85         'Auml'     => 196,
  86         'auml'     => 228,
  87         'bdquo'    => 8222,
  88         'Beta'     => 914,
  89         'beta'     => 946,
  90         'brvbar'   => 166,
  91         'bull'     => 8226,
  92         'cap'      => 8745,
  93         'Ccedil'   => 199,
  94         'ccedil'   => 231,
  95         'cedil'    => 184,
  96         'cent'     => 162,
  97         'Chi'      => 935,
  98         'chi'      => 967,
  99         'circ'     => 710,
 100         'clubs'    => 9827,
 101         'cong'     => 8773,
 102         'copy'     => 169,
 103         'crarr'    => 8629,
 104         'cup'      => 8746,
 105         'curren'   => 164,
 106         'dagger'   => 8224,
 107         'Dagger'   => 8225,
 108         'darr'     => 8595,
 109         'dArr'     => 8659,
 110         'deg'      => 176,
 111         'Delta'    => 916,
 112         'delta'    => 948,
 113         'diams'    => 9830,
 114         'divide'   => 247,
 115         'Eacute'   => 201,
 116         'eacute'   => 233,
 117         'Ecirc'    => 202,
 118         'ecirc'    => 234,
 119         'Egrave'   => 200,
 120         'egrave'   => 232,
 121         'empty'    => 8709,
 122         'emsp'     => 8195,
 123         'ensp'     => 8194,
 124         'Epsilon'  => 917,
 125         'epsilon'  => 949,
 126         'equiv'    => 8801,
 127         'Eta'      => 919,
 128         'eta'      => 951,
 129         'ETH'      => 208,
 130         'eth'      => 240,
 131         'Euml'     => 203,
 132         'euml'     => 235,
 133         'euro'     => 8364,
 134         'exist'    => 8707,
 135         'fnof'     => 402,
 136         'forall'   => 8704,
 137         'frac12'   => 189,
 138         'frac14'   => 188,
 139         'frac34'   => 190,
 140         'frasl'    => 8260,
 141         'Gamma'    => 915,
 142         'gamma'    => 947,
 143         'ge'       => 8805,
 144         'gt'       => 62,
 145         'harr'     => 8596,
 146         'hArr'     => 8660,
 147         'hearts'   => 9829,
 148         'hellip'   => 8230,
 149         'Iacute'   => 205,
 150         'iacute'   => 237,
 151         'Icirc'    => 206,
 152         'icirc'    => 238,
 153         'iexcl'    => 161,
 154         'Igrave'   => 204,
 155         'igrave'   => 236,
 156         'image'    => 8465,
 157         'infin'    => 8734,
 158         'int'      => 8747,
 159         'Iota'     => 921,
 160         'iota'     => 953,
 161         'iquest'   => 191,
 162         'isin'     => 8712,
 163         'Iuml'     => 207,
 164         'iuml'     => 239,
 165         'Kappa'    => 922,
 166         'kappa'    => 954,
 167         'Lambda'   => 923,
 168         'lambda'   => 955,
 169         'lang'     => 9001,
 170         'laquo'    => 171,
 171         'larr'     => 8592,
 172         'lArr'     => 8656,
 173         'lceil'    => 8968,
 174         'ldquo'    => 8220,
 175         'le'       => 8804,
 176         'lfloor'   => 8970,
 177         'lowast'   => 8727,
 178         'loz'      => 9674,
 179         'lrm'      => 8206,
 180         'lsaquo'   => 8249,
 181         'lsquo'    => 8216,
 182         'lt'       => 60,
 183         'macr'     => 175,
 184         'mdash'    => 8212,
 185         'micro'    => 181,
 186         'middot'   => 183,
 187         'minus'    => 8722,
 188         'Mu'       => 924,
 189         'mu'       => 956,
 190         'nabla'    => 8711,
 191         'nbsp'     => 160,
 192         'ndash'    => 8211,
 193         'ne'       => 8800,
 194         'ni'       => 8715,
 195         'not'      => 172,
 196         'notin'    => 8713,
 197         'nsub'     => 8836,
 198         'Ntilde'   => 209,
 199         'ntilde'   => 241,
 200         'Nu'       => 925,
 201         'nu'       => 957,
 202         'Oacute'   => 211,
 203         'oacute'   => 243,
 204         'Ocirc'    => 212,
 205         'ocirc'    => 244,
 206         'OElig'    => 338,
 207         'oelig'    => 339,
 208         'Ograve'   => 210,
 209         'ograve'   => 242,
 210         'oline'    => 8254,
 211         'Omega'    => 937,
 212         'omega'    => 969,
 213         'Omicron'  => 927,
 214         'omicron'  => 959,
 215         'oplus'    => 8853,
 216         'or'       => 8744,
 217         'ordf'     => 170,
 218         'ordm'     => 186,
 219         'Oslash'   => 216,
 220         'oslash'   => 248,
 221         'Otilde'   => 213,
 222         'otilde'   => 245,
 223         'otimes'   => 8855,
 224         'Ouml'     => 214,
 225         'ouml'     => 246,
 226         'para'     => 182,
 227         'part'     => 8706,
 228         'permil'   => 8240,
 229         'perp'     => 8869,
 230         'Phi'      => 934,
 231         'phi'      => 966,
 232         'Pi'       => 928,
 233         'pi'       => 960,
 234         'piv'      => 982,
 235         'plusmn'   => 177,
 236         'pound'    => 163,
 237         'prime'    => 8242,
 238         'Prime'    => 8243,
 239         'prod'     => 8719,
 240         'prop'     => 8733,
 241         'Psi'      => 936,
 242         'psi'      => 968,
 243         'quot'     => 34,
 244         'radic'    => 8730,
 245         'rang'     => 9002,
 246         'raquo'    => 187,
 247         'rarr'     => 8594,
 248         'rArr'     => 8658,
 249         'rceil'    => 8969,
 250         'rdquo'    => 8221,
 251         'real'     => 8476,
 252         'reg'      => 174,
 253         'rfloor'   => 8971,
 254         'Rho'      => 929,
 255         'rho'      => 961,
 256         'rlm'      => 8207,
 257         'rsaquo'   => 8250,
 258         'rsquo'    => 8217,
 259         'sbquo'    => 8218,
 260         'Scaron'   => 352,
 261         'scaron'   => 353,
 262         'sdot'     => 8901,
 263         'sect'     => 167,
 264         'shy'      => 173,
 265         'Sigma'    => 931,
 266         'sigma'    => 963,
 267         'sigmaf'   => 962,
 268         'sim'      => 8764,
 269         'spades'   => 9824,
 270         'sub'      => 8834,
 271         'sube'     => 8838,
 272         'sum'      => 8721,
 273         'sup'      => 8835,
 274         'sup1'     => 185,
 275         'sup2'     => 178,
 276         'sup3'     => 179,
 277         'supe'     => 8839,
 278         'szlig'    => 223,
 279         'Tau'      => 932,
 280         'tau'      => 964,
 281         'there4'   => 8756,
 282         'Theta'    => 920,
 283         'theta'    => 952,
 284         'thetasym' => 977,
 285         'thinsp'   => 8201,
 286         'THORN'    => 222,
 287         'thorn'    => 254,
 288         'tilde'    => 732,
 289         'times'    => 215,
 290         'trade'    => 8482,
 291         'Uacute'   => 218,
 292         'uacute'   => 250,
 293         'uarr'     => 8593,
 294         'uArr'     => 8657,
 295         'Ucirc'    => 219,
 296         'ucirc'    => 251,
 297         'Ugrave'   => 217,
 298         'ugrave'   => 249,
 299         'uml'      => 168,
 300         'upsih'    => 978,
 301         'Upsilon'  => 933,
 302         'upsilon'  => 965,
 303         'Uuml'     => 220,
 304         'uuml'     => 252,
 305         'weierp'   => 8472,
 306         'Xi'       => 926,
 307         'xi'       => 958,
 308         'Yacute'   => 221,
 309         'yacute'   => 253,
 310         'yen'      => 165,
 311         'Yuml'     => 376,
 312         'yuml'     => 255,
 313         'Zeta'     => 918,
 314         'zeta'     => 950,
 315         'zwj'      => 8205,
 316         'zwnj'     => 8204 );
 317
 318 /**
 319  * Character entity aliases accepted by MediaWiki
 320  */
 321 global $wgHtmlEntityAliases;
 322 $wgHtmlEntityAliases = array(
 323         'רלמ' => 'rlm',
 324         'رلم' => 'rlm',
 325 );
 326
 327
 328 /**
 329  * XHTML sanitizer for MediaWiki
 330  * @addtogroup Parser
 331  */
 332 class Sanitizer {
 333         /**
 334          * Cleans up HTML, removes dangerous tags and attributes, and
 335          * removes HTML comments
 336          * @private
 337          * @param string $text
 338          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 339          * @param array $args for the processing callback
 340          * @return string
 341          */
 342         static function removeHTMLtags( $text, $processCallback = null, $args = array() ) {
 343                 global $wgUseTidy;
 344
 345                 static $htmlpairs, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 346                         $htmllist, $listtags, $htmlsingleallowed, $htmlelements, $staticInitialised;
 347
 348                 wfProfileIn( __METHOD__ );
 349
 350                 if ( !$staticInitialised ) {
 351
 352                         $htmlpairs = array( # Tags that must be closed
 353                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 354                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 355                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 356                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 357                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
 358                         );
 359                         $htmlsingle = array(
 360                                 'br', 'hr', 'li', 'dt', 'dd'
 361                         );
 362                         $htmlsingleonly = array( # Elements that cannot have close tags
 363                                 'br', 'hr'
 364                         );
 365                         $htmlnest = array( # Tags that can be nested--??
 366                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 367                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 368                         );
 369                         $tabletags = array( # Can only appear inside table, we will close them
 370                                 'td', 'th', 'tr',
 371                         );
 372                         $htmllist = array( # Tags used by list
 373                                 'ul','ol',
 374                         );
 375                         $listtags = array( # Tags that can appear in a list
 376                                 'li',
 377                         );
 378
 379                         $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
 380                         $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
 381
 382                         # Convert them all to hashtables for faster lookup
 383                         $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 384                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
 385                         foreach ( $vars as $var ) {
 386                                 $$var = array_flip( $$var );
 387                         }
 388                         $staticInitialised = true;
 389                 }
 390
 391                 # Remove HTML comments
 392                 $text = Sanitizer::removeHTMLcomments( $text );
 393                 $bits = explode( '<', $text );
 394                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 395                 if(!$wgUseTidy) {
 396                         $tagstack = $tablestack = array();
 397                         foreach ( $bits as $x ) {
 398                                 $regs = array();
 399                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 400                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 401                                 } else {
 402                                         $slash = $t = $params = $brace = $rest = null;
 403                                 }
 404
 405                                 $badtag = 0 ;
 406                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 407                                         # Check our stack
 408                                         if ( $slash ) {
 409                                                 # Closing a tag...
 410                                                 if( isset( $htmlsingleonly[$t] ) ) {
 411                                                         $badtag = 1;
 412                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 413                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 414                                                                 # Pop all elements with an optional close tag
 415                                                                 # and see if we find a match below them
 416                                                                 $optstack = array();
 417                                                                 array_push ($optstack, $ot);
 418                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 419                                                                                 isset( $htmlsingleallowed[$ot] ) )
 420                                                                 {
 421                                                                         array_push ($optstack, $ot);
 422                                                                 }
 423                                                                 if ( $t != $ot ) {
 424                                                                         # No match. Push the optinal elements back again
 425                                                                         $badtag = 1;
 426                                                                         while ( $ot = @array_pop( $optstack ) ) {
 427                                                                                 array_push( $tagstack, $ot );
 428                                                                         }
 429                                                                 }
 430                                                         } else {
 431                                                                 @array_push( $tagstack, $ot );
 432                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 433                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 434                                                                         $badtag = 1;
 435                                                                 }
 436                                                         }
 437                                                 } else {
 438                                                         if ( $t == 'table' ) {
 439                                                                 $tagstack = array_pop( $tablestack );
 440                                                         }
 441                                                 }
 442                                                 $newparams = '';
 443                                         } else {
 444                                                 # Keep track for later
 445                                                 if ( isset( $tabletags[$t] ) &&
 446                                                 ! in_array( 'table', $tagstack ) ) {
 447                                                         $badtag = 1;
 448                                                 } else if ( in_array( $t, $tagstack ) &&
 449                                                 ! isset( $htmlnest [$t ] ) ) {
 450                                                         $badtag = 1 ;
 451                                                 # Is it a self closed htmlpair ? (bug 5487)
 452                                                 } else if( $brace == '/>' &&
 453                                                 isset( $htmlpairs[$t] ) ) {
 454                                                         $badtag = 1;
 455                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 456                                                         # Hack to force empty tag for uncloseable elements
 457                                                         $brace = '/>';
 458                                                 } else if( isset( $htmlsingle[$t] ) ) {
 459                                                         # Hack to not close $htmlsingle tags
 460                                                         $brace = NULL;
 461                                                 } else if( isset( $tabletags[$t] )
 462                                                 &&  in_array($t ,$tagstack) ) {
 463                                                         // New table tag but forgot to close the previous one
 464                                                         $text .= "</$t>";
 465                                                 } else {
 466                                                         if ( $t == 'table' ) {
 467                                                                 array_push( $tablestack, $tagstack );
 468                                                                 $tagstack = array();
 469                                                         }
 470                                                         array_push( $tagstack, $t );
 471                                                 }
 472
 473                                                 # Replace any variables or template parameters with
 474                                                 # plaintext results.
 475                                                 if( is_callable( $processCallback ) ) {
 476                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 477                                                 }
 478
 479                                                 # Strip non-approved attributes from the tag
 480                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 481                                         }
 482                                         if ( ! $badtag ) {
 483                                                 $rest = str_replace( '>', '&gt;', $rest );
 484                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 485                                                 $text .= "<$slash$t$newparams$close>$rest";
 486                                                 continue;
 487                                         }
 488                                 }
 489                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 490                         }
 491                         # Close off any remaining tags
 492                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 493                                 $text .= "</$t>\n";
 494                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 495                         }
 496                 } else {
 497                         # this might be possible using tidy itself
 498                         foreach ( $bits as $x ) {
 499                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 500                                 $x, $regs );
 501                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 502                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 503                                         if( is_callable( $processCallback ) ) {
 504                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 505                                         }
 506                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 507                                         $rest = str_replace( '>', '&gt;', $rest );
 508                                         $text .= "<$slash$t$newparams$brace$rest";
 509                                 } else {
 510                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 511                                 }
 512                         }
 513                 }
 514                 wfProfileOut( __METHOD__ );
 515                 return $text;
 516         }
 517
 518         /**
 519          * Remove '<!--', '-->', and everything between.
 520          * To avoid leaving blank lines, when a comment is both preceded
 521          * and followed by a newline (ignoring spaces), trim leading and
 522          * trailing spaces and one of the newlines.
 523          *
 524          * @private
 525          * @param string $text
 526          * @return string
 527          */
 528         static function removeHTMLcomments( $text ) {
 529                 wfProfileIn( __METHOD__ );
 530                 while (($start = strpos($text, '<!--')) !== false) {
 531                         $end = strpos($text, '-->', $start + 4);
 532                         if ($end === false) {
 533                                 # Unterminated comment; bail out
 534                                 break;
 535                         }
 536
 537                         $end += 3;
 538
 539                         # Trim space and newline if the comment is both
 540                         # preceded and followed by a newline
 541                         $spaceStart = max($start - 1, 0);
 542                         $spaceLen = $end - $spaceStart;
 543                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 544                                 $spaceStart--;
 545                                 $spaceLen++;
 546                         }
 547                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 548                                 $spaceLen++;
 549                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 550                                 # Remove the comment, leading and trailing
 551                                 # spaces, and leave only one newline.
 552                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 553                         }
 554                         else {
 555                                 # Remove just the comment.
 556                                 $text = substr_replace($text, '', $start, $end - $start);
 557                         }
 558                 }
 559                 wfProfileOut( __METHOD__ );
 560                 return $text;
 561         }
 562
 563         /**
 564          * Take an array of attribute names and values and normalize or discard
 565          * illegal values for the given element type.
 566          *
 567          * - Discards attributes not on a whitelist for the given element
 568          * - Unsafe style attributes are discarded
 569          *
 570          * @param array $attribs
 571          * @param string $element
 572          * @return array
 573          *
 574          * @todo Check for legal values where the DTD limits things.
 575          * @todo Check for unique id attribute :P
 576          */
 577         static function validateTagAttributes( $attribs, $element ) {
 578                 $whitelist = array_flip( Sanitizer::attributeWhitelist( $element ) );
 579                 $out = array();
 580                 foreach( $attribs as $attribute => $value ) {
 581                         if( !isset( $whitelist[$attribute] ) ) {
 582                                 continue;
 583                         }
 584                         # Strip javascript "expression" from stylesheets.
 585                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 586                         if( $attribute == 'style' ) {
 587                                 $value = Sanitizer::checkCss( $value );
 588                                 if( $value === false ) {
 589                                         # haxx0r
 590                                         continue;
 591                                 }
 592                         }
 593
 594                         if ( $attribute === 'id' )
 595                                 $value = Sanitizer::escapeId( $value );
 596
 597                         // If this attribute was previously set, override it.
 598                         // Output should only have one attribute of each name.
 599                         $out[$attribute] = $value;
 600                 }
 601                 return $out;
 602         }
 603
 604         /**
 605          * Pick apart some CSS and check it for forbidden or unsafe structures.
 606          * Returns a sanitized string, or false if it was just too evil.
 607          *
 608          * Currently URL references, 'expression', 'tps' are forbidden.
 609          *
 610          * @param string $value
 611          * @return mixed
 612          */
 613         static function checkCss( $value ) {
 614                 $stripped = Sanitizer::decodeCharReferences( $value );
 615
 616                 // Remove any comments; IE gets token splitting wrong
 617                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 618
 619                 $value = $stripped;
 620
 621                 // ... and continue checks
 622                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 623                         'codepointToUtf8(hexdec("$1"))', $stripped );
 624                 $stripped = str_replace( '\\', '', $stripped );
 625                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 626                                 $stripped ) ) {
 627                         # haxx0r
 628                         return false;
 629                 }
 630
 631                 return $value;
 632         }
 633
 634         /**
 635          * Take a tag soup fragment listing an HTML element's attributes
 636          * and normalize it to well-formed XML, discarding unwanted attributes.
 637          * Output is safe for further wikitext processing, with escaping of
 638          * values that could trigger problems.
 639          *
 640          * - Normalizes attribute names to lowercase
 641          * - Discards attributes not on a whitelist for the given element
 642          * - Turns broken or invalid entities into plaintext
 643          * - Double-quotes all attribute values
 644          * - Attributes without values are given the name as attribute
 645          * - Double attributes are discarded
 646          * - Unsafe style attributes are discarded
 647          * - Prepends space if there are attributes.
 648          *
 649          * @param string $text
 650          * @param string $element
 651          * @return string
 652          */
 653         static function fixTagAttributes( $text, $element ) {
 654                 if( trim( $text ) == '' ) {
 655                         return '';
 656                 }
 657
 658                 $stripped = Sanitizer::validateTagAttributes(
 659                         Sanitizer::decodeTagAttributes( $text ), $element );
 660
 661                 $attribs = array();
 662                 foreach( $stripped as $attribute => $value ) {
 663                         $encAttribute = htmlspecialchars( $attribute );
 664                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 665
 666                         $attribs[] = "$encAttribute=\"$encValue\"";
 667                 }
 668                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 669         }
 670
 671         /**
 672          * Encode an attribute value for HTML output.
 673          * @param $text
 674          * @return HTML-encoded text fragment
 675          */
 676         static function encodeAttribute( $text ) {
 677                 $encValue = htmlspecialchars( $text );
 678
 679                 // Whitespace is normalized during attribute decoding,
 680                 // so if we've been passed non-spaces we must encode them
 681                 // ahead of time or they won't be preserved.
 682                 $encValue = strtr( $encValue, array(
 683                         "\n" => '&#10;',
 684                         "\r" => '&#13;',
 685                         "\t" => '&#9;',
 686                 ) );
 687
 688                 return $encValue;
 689         }
 690
 691         /**
 692          * Encode an attribute value for HTML tags, with extra armoring
 693          * against further wiki processing.
 694          * @param $text
 695          * @return HTML-encoded text fragment
 696          */
 697         static function safeEncodeAttribute( $text ) {
 698                 $encValue = Sanitizer::encodeAttribute( $text );
 699
 700                 # Templates and links may be expanded in later parsing,
 701                 # creating invalid or dangerous output. Suppress this.
 702                 $encValue = strtr( $encValue, array(
 703                         '<'    => '&lt;',   // This should never happen,
 704                         '>'    => '&gt;',   // we've received invalid input
 705                         '"'    => '&quot;', // which should have been escaped.
 706                         '{'    => '&#123;',
 707                         '['    => '&#91;',
 708                         "''"   => '&#39;&#39;',
 709                         'ISBN' => '&#73;SBN',
 710                         'RFC'  => '&#82;FC',
 711                         'PMID' => '&#80;MID',
 712                         '|'    => '&#124;',
 713                         '__'   => '&#95;_',
 714                 ) );
 715
 716                 # Stupid hack
 717                 $encValue = preg_replace_callback(
 718                         '/(' . wfUrlProtocols() . ')/',
 719                         array( 'Sanitizer', 'armorLinksCallback' ),
 720                         $encValue );
 721                 return $encValue;
 722         }
 723
 724         /**
 725          * Given a value escape it so that it can be used in an id attribute and
 726          * return it, this does not validate the value however (see first link)
 727          *
 728          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 729          *                                                          in the id and
 730          *                                                          name attributes
 731          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 732          *
 733          * @static
 734          *
 735          * @param string $id
 736          * @return string
 737          */
 738         static function escapeId( $id ) {
 739                 static $replace = array(
 740                         '%3A' => ':',
 741                         '%' => '.'
 742                 );
 743
 744                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 745
 746                 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
 747         }
 748
 749         /**
 750          * Given a value, escape it so that it can be used as a CSS class and
 751          * return it.
 752          *
 753          * @todo For extra validity, input should be validated UTF-8.
 754          *
 755          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 756          *
 757          * @param string $class
 758          * @return string
 759          */
 760         static function escapeClass( $class ) {
 761                 // Convert ugly stuff to underscores and kill underscores in ugly places
 762                 return rtrim(preg_replace(
 763                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 764                         '_',
 765                         $class ), '_');
 766         }
 767
 768         /**
 769          * Regex replace callback for armoring links against further processing.
 770          * @param array $matches
 771          * @return string
 772          * @private
 773          */
 774         private static function armorLinksCallback( $matches ) {
 775                 return str_replace( ':', '&#58;', $matches[1] );
 776         }
 777
 778         /**
 779          * Return an associative array of attribute names and values from
 780          * a partial tag string. Attribute names are forces to lowercase,
 781          * character references are decoded to UTF-8 text.
 782          *
 783          * @param string
 784          * @return array
 785          */
 786         static function decodeTagAttributes( $text ) {
 787                 $attribs = array();
 788
 789                 if( trim( $text ) == '' ) {
 790                         return $attribs;
 791                 }
 792
 793                 $pairs = array();
 794                 if( !preg_match_all(
 795                         MW_ATTRIBS_REGEX,
 796                         $text,
 797                         $pairs,
 798                         PREG_SET_ORDER ) ) {
 799                         return $attribs;
 800                 }
 801
 802                 foreach( $pairs as $set ) {
 803                         $attribute = strtolower( $set[1] );
 804                         $value = Sanitizer::getTagAttributeCallback( $set );
 805
 806                         // Normalize whitespace
 807                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 808                         $value = trim( $value );
 809
 810                         // Decode character references
 811                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 812                 }
 813                 return $attribs;
 814         }
 815
 816         /**
 817          * Pick the appropriate attribute value from a match set from the
 818          * MW_ATTRIBS_REGEX matches.
 819          *
 820          * @param array $set
 821          * @return string
 822          * @private
 823          */
 824         private static function getTagAttributeCallback( $set ) {
 825                 if( isset( $set[6] ) ) {
 826                         # Illegal #XXXXXX color with no quotes.
 827                         return $set[6];
 828                 } elseif( isset( $set[5] ) ) {
 829                         # No quotes.
 830                         return $set[5];
 831                 } elseif( isset( $set[4] ) ) {
 832                         # Single-quoted
 833                         return $set[4];
 834                 } elseif( isset( $set[3] ) ) {
 835                         # Double-quoted
 836                         return $set[3];
 837                 } elseif( !isset( $set[2] ) ) {
 838                         # In XHTML, attributes must have a value.
 839                         # For 'reduced' form, return explicitly the attribute name here.
 840                         return $set[1];
 841                 } else {
 842                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 843                 }
 844         }
 845
 846         /**
 847          * Normalize whitespace and character references in an XML source-
 848          * encoded text for an attribute value.
 849          *
 850          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 851          * but note that we're not returning the value, but are returning
 852          * XML source fragments that will be slapped into output.
 853          *
 854          * @param string $text
 855          * @return string
 856          * @private
 857          */
 858         private static function normalizeAttributeValue( $text ) {
 859                 return str_replace( '"', '&quot;',
 860                         self::normalizeWhitespace(
 861                                 Sanitizer::normalizeCharReferences( $text ) ) );
 862         }
 863
 864         private static function normalizeWhitespace( $text ) {
 865                 return preg_replace(
 866                         '/\r\n|[\x20\x0d\x0a\x09]/',
 867                         ' ',
 868                         $text );
 869         }
 870
 871         /**
 872          * Ensure that any entities and character references are legal
 873          * for XML and XHTML specifically. Any stray bits will be
 874          * &amp;-escaped to result in a valid text fragment.
 875          *
 876          * a. any named char refs must be known in XHTML
 877          * b. any numeric char refs must be legal chars, not invalid or forbidden
 878          * c. use &#x, not &#X
 879          * d. fix or reject non-valid attributes
 880          *
 881          * @param string $text
 882          * @return string
 883          * @private
 884          */
 885         static function normalizeCharReferences( $text ) {
 886                 return preg_replace_callback(
 887                         MW_CHAR_REFS_REGEX,
 888                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 889                         $text );
 890         }
 891         /**
 892          * @param string $matches
 893          * @return string
 894          */
 895         static function normalizeCharReferencesCallback( $matches ) {
 896                 $ret = null;
 897                 if( $matches[1] != '' ) {
 898                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 899                 } elseif( $matches[2] != '' ) {
 900                         $ret = Sanitizer::decCharReference( $matches[2] );
 901                 } elseif( $matches[3] != ''  ) {
 902                         $ret = Sanitizer::hexCharReference( $matches[3] );
 903                 } elseif( $matches[4] != '' ) {
 904                         $ret = Sanitizer::hexCharReference( $matches[4] );
 905                 }
 906                 if( is_null( $ret ) ) {
 907                         return htmlspecialchars( $matches[0] );
 908                 } else {
 909                         return $ret;
 910                 }
 911         }
 912
 913         /**
 914          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 915          * return the named entity reference as is. If the entity is a
 916          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
 917          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 918          *
 919          * @param string $name
 920          * @return string
 921          * @static
 922          */
 923         static function normalizeEntity( $name ) {
 924                 global $wgHtmlEntities, $wgHtmlEntityAliases;
 925                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
 926                         return "&{$wgHtmlEntityAliases[$name]};";
 927                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
 928                         return "&$name;";
 929                 } else {
 930                         return "&amp;$name;";
 931                 }
 932         }
 933
 934         static function decCharReference( $codepoint ) {
 935                 $point = intval( $codepoint );
 936                 if( Sanitizer::validateCodepoint( $point ) ) {
 937                         return sprintf( '&#%d;', $point );
 938                 } else {
 939                         return null;
 940                 }
 941         }
 942
 943         static function hexCharReference( $codepoint ) {
 944                 $point = hexdec( $codepoint );
 945                 if( Sanitizer::validateCodepoint( $point ) ) {
 946                         return sprintf( '&#x%x;', $point );
 947                 } else {
 948                         return null;
 949                 }
 950         }
 951
 952         /**
 953          * Returns true if a given Unicode codepoint is a valid character in XML.
 954          * @param int $codepoint
 955          * @return bool
 956          */
 957         private static function validateCodepoint( $codepoint ) {
 958                 return ($codepoint ==    0x09)
 959                         || ($codepoint ==    0x0a)
 960                         || ($codepoint ==    0x0d)
 961                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 962                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 963                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 964         }
 965
 966         /**
 967          * Decode any character references, numeric or named entities,
 968          * in the text and return a UTF-8 string.
 969          *
 970          * @param string $text
 971          * @return string
 972          * @public
 973          * @static
 974          */
 975         public static function decodeCharReferences( $text ) {
 976                 return preg_replace_callback(
 977                         MW_CHAR_REFS_REGEX,
 978                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 979                         $text );
 980         }
 981
 982         /**
 983          * @param string $matches
 984          * @return string
 985          */
 986         static function decodeCharReferencesCallback( $matches ) {
 987                 if( $matches[1] != '' ) {
 988                         return Sanitizer::decodeEntity( $matches[1] );
 989                 } elseif( $matches[2] != '' ) {
 990                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 991                 } elseif( $matches[3] != ''  ) {
 992                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 993                 } elseif( $matches[4] != '' ) {
 994                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 995                 }
 996                 # Last case should be an ampersand by itself
 997                 return $matches[0];
 998         }
 999
1000         /**
1001          * Return UTF-8 string for a codepoint if that is a valid
1002          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1003          * @param int $codepoint
1004          * @return string
1005          * @private
1006          */
1007         static function decodeChar( $codepoint ) {
1008                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1009                         return codepointToUtf8( $codepoint );
1010                 } else {
1011                         return UTF8_REPLACEMENT;
1012                 }
1013         }
1014
1015         /**
1016          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1017          * return the UTF-8 encoding of that character. Otherwise, returns
1018          * pseudo-entity source (eg &foo;)
1019          *
1020          * @param string $name
1021          * @return string
1022          */
1023         static function decodeEntity( $name ) {
1024                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1025                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1026                         $name = $wgHtmlEntityAliases[$name];
1027                 }
1028                 if( isset( $wgHtmlEntities[$name] ) ) {
1029                         return codepointToUtf8( $wgHtmlEntities[$name] );
1030                 } else {
1031                         return "&$name;";
1032                 }
1033         }
1034
1035         /**
1036          * Fetch the whitelist of acceptable attributes for a given
1037          * element name.
1038          *
1039          * @param string $element
1040          * @return array
1041          */
1042         static function attributeWhitelist( $element ) {
1043                 static $list;
1044                 if( !isset( $list ) ) {
1045                         $list = Sanitizer::setupAttributeWhitelist();
1046                 }
1047                 return isset( $list[$element] )
1048                         ? $list[$element]
1049                         : array();
1050         }
1051
1052         /**
1053          * @todo Document it a bit
1054          * @return array
1055          */
1056         static function setupAttributeWhitelist() {
1057                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1058                 $block = array_merge( $common, array( 'align' ) );
1059                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1060                 $tablecell = array( 'abbr',
1061                                     'axis',
1062                                     'headers',
1063                                     'scope',
1064                                     'rowspan',
1065                                     'colspan',
1066                                     'nowrap', # deprecated
1067                                     'width',  # deprecated
1068                                     'height', # deprecated
1069                                     'bgcolor' # deprecated
1070                                     );
1071
1072                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1073                 # See: http://www.w3.org/TR/html4/
1074                 $whitelist = array (
1075                         # 7.5.4
1076                         'div'        => $block,
1077                         'center'     => $common, # deprecated
1078                         'span'       => $block, # ??
1079
1080                         # 7.5.5
1081                         'h1'         => $block,
1082                         'h2'         => $block,
1083                         'h3'         => $block,
1084                         'h4'         => $block,
1085                         'h5'         => $block,
1086                         'h6'         => $block,
1087
1088                         # 7.5.6
1089                         # address
1090
1091                         # 8.2.4
1092                         # bdo
1093
1094                         # 9.2.1
1095                         'em'         => $common,
1096                         'strong'     => $common,
1097                         'cite'       => $common,
1098                         # dfn
1099                         'code'       => $common,
1100                         # samp
1101                         # kbd
1102                         'var'        => $common,
1103                         # abbr
1104                         # acronym
1105
1106                         # 9.2.2
1107                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1108                         # q
1109
1110                         # 9.2.3
1111                         'sub'        => $common,
1112                         'sup'        => $common,
1113
1114                         # 9.3.1
1115                         'p'          => $block,
1116
1117                         # 9.3.2
1118                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1119
1120                         # 9.3.4
1121                         'pre'        => array_merge( $common, array( 'width' ) ),
1122
1123                         # 9.4
1124                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1125                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1126
1127                         # 10.2
1128                         'ul'         => array_merge( $common, array( 'type' ) ),
1129                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1130                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1131
1132                         # 10.3
1133                         'dl'         => $common,
1134                         'dd'         => $common,
1135                         'dt'         => $common,
1136
1137                         # 11.2.1
1138                         'table'      => array_merge( $common,
1139                                                                 array( 'summary', 'width', 'border', 'frame',
1140                                                                                 'rules', 'cellspacing', 'cellpadding',
1141                                                                                 'align', 'bgcolor',
1142                                                                 ) ),
1143
1144                         # 11.2.2
1145                         'caption'    => array_merge( $common, array( 'align' ) ),
1146
1147                         # 11.2.3
1148                         'thead'      => array_merge( $common, $tablealign ),
1149                         'tfoot'      => array_merge( $common, $tablealign ),
1150                         'tbody'      => array_merge( $common, $tablealign ),
1151
1152                         # 11.2.4
1153                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1154                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1155
1156                         # 11.2.5
1157                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1158
1159                         # 11.2.6
1160                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1161                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1162
1163                         # 15.2.1
1164                         'tt'         => $common,
1165                         'b'          => $common,
1166                         'i'          => $common,
1167                         'big'        => $common,
1168                         'small'      => $common,
1169                         'strike'     => $common,
1170                         's'          => $common,
1171                         'u'          => $common,
1172
1173                         # 15.2.2
1174                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1175                         # basefont
1176
1177                         # 15.3
1178                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1179
1180                         # XHTML Ruby annotation text module, simple ruby only.
1181                         # http://www.w3c.org/TR/ruby/
1182                         'ruby'       => $common,
1183                         # rbc
1184                         # rtc
1185                         'rb'         => $common,
1186                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1187                         'rp'         => $common,
1188                         );
1189                 return $whitelist;
1190         }
1191
1192         /**
1193          * Take a fragment of (potentially invalid) HTML and return
1194          * a version with any tags removed, encoded as plain text.
1195          *
1196          * Warning: this return value must be further escaped for literal
1197          * inclusion in HTML output as of 1.10!
1198          *
1199          * @param string $text HTML fragment
1200          * @return string
1201          */
1202         static function stripAllTags( $text ) {
1203                 # Actual <tags>
1204                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1205
1206                 # Normalize &entities and whitespace
1207                 $text = self::decodeCharReferences( $text );
1208                 $text = self::normalizeWhitespace( $text );
1209
1210                 return $text;
1211         }
1212
1213         /**
1214          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1215          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1216          * PHP 5.1 doesn't.
1217          *
1218          * Use for passing XHTML fragments to PHP's XML parsing functions
1219          *
1220          * @return string
1221          * @static
1222          */
1223         static function hackDocType() {
1224                 global $wgHtmlEntities;
1225                 $out = "<!DOCTYPE html [\n";
1226                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1227                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1228                 }
1229                 $out .= "]>\n";
1230                 return $out;
1231         }
1232
1233         static function cleanUrl( $url, $hostname=true ) {
1234                 # Normalize any HTML entities in input. They will be
1235                 # re-escaped by makeExternalLink().
1236                 $url = Sanitizer::decodeCharReferences( $url );
1237
1238                 # Escape any control characters introduced by the above step
1239                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1240
1241                 # Validate hostname portion
1242                 $matches = array();
1243                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1244                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1245
1246                         // Characters that will be ignored in IDNs.
1247                         // http://tools.ietf.org/html/3454#section-3.1
1248                         // Strip them before further processing so blacklists and such work.
1249                         $strip = "/
1250                                 \\s|          # general whitespace
1251                                 \xc2\xad|     # 00ad SOFT HYPHEN
1252                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1253                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1254                                 \xe2\x81\xa0| # 2060 WORD JOINER
1255                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1256                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1257                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1258                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1259                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1260                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1261                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1262                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1263                                 /xuD";
1264
1265                         $host = preg_replace( $strip, '', $host );
1266
1267                         // @fixme: validate hostnames here
1268
1269                         return $protocol . $host . $rest;
1270                 } else {
1271                         return $url;
1272                 }
1273         }
1274
1275 }
1276
1277