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         const NONE = 0;
 334         const INITIAL_NONLETTER = 1;
 335
 336         /**
 337          * Cleans up HTML, removes dangerous tags and attributes, and
 338          * removes HTML comments
 339          * @private
 340          * @param string $text
 341          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 342          * @param array $args for the processing callback
 343          * @return string
 344          */
 345         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array() ) {
 346                 global $wgUseTidy;
 347
 348                 static $htmlpairs, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 349                         $htmllist, $listtags, $htmlsingleallowed, $htmlelements, $staticInitialised;
 350
 351                 wfProfileIn( __METHOD__ );
 352
 353                 if ( !$staticInitialised ) {
 354
 355                         $htmlpairs = array_merge( $extratags, array( # Tags that must be closed
 356                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 357                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 358                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 359                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 360                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
 361                         ) );
 362                         $htmlsingle = array(
 363                                 'br', 'hr', 'li', 'dt', 'dd'
 364                         );
 365                         $htmlsingleonly = array( # Elements that cannot have close tags
 366                                 'br', 'hr'
 367                         );
 368                         $htmlnest = array( # Tags that can be nested--??
 369                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 370                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 371                         );
 372                         $tabletags = array( # Can only appear inside table, we will close them
 373                                 'td', 'th', 'tr',
 374                         );
 375                         $htmllist = array( # Tags used by list
 376                                 'ul','ol',
 377                         );
 378                         $listtags = array( # Tags that can appear in a list
 379                                 'li',
 380                         );
 381
 382                         $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
 383                         $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
 384
 385                         # Convert them all to hashtables for faster lookup
 386                         $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 387                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
 388                         foreach ( $vars as $var ) {
 389                                 $$var = array_flip( $$var );
 390                         }
 391                         $staticInitialised = true;
 392                 }
 393
 394                 # Remove HTML comments
 395                 $text = Sanitizer::removeHTMLcomments( $text );
 396                 $bits = explode( '<', $text );
 397                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 398                 if(!$wgUseTidy) {
 399                         $tagstack = $tablestack = array();
 400                         foreach ( $bits as $x ) {
 401                                 $regs = array();
 402                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 403                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 404                                 } else {
 405                                         $slash = $t = $params = $brace = $rest = null;
 406                                 }
 407
 408                                 $badtag = 0 ;
 409                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 410                                         # Check our stack
 411                                         if ( $slash ) {
 412                                                 # Closing a tag...
 413                                                 if( isset( $htmlsingleonly[$t] ) ) {
 414                                                         $badtag = 1;
 415                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 416                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 417                                                                 # Pop all elements with an optional close tag
 418                                                                 # and see if we find a match below them
 419                                                                 $optstack = array();
 420                                                                 array_push ($optstack, $ot);
 421                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 422                                                                                 isset( $htmlsingleallowed[$ot] ) )
 423                                                                 {
 424                                                                         array_push ($optstack, $ot);
 425                                                                 }
 426                                                                 if ( $t != $ot ) {
 427                                                                         # No match. Push the optinal elements back again
 428                                                                         $badtag = 1;
 429                                                                         while ( $ot = @array_pop( $optstack ) ) {
 430                                                                                 array_push( $tagstack, $ot );
 431                                                                         }
 432                                                                 }
 433                                                         } else {
 434                                                                 @array_push( $tagstack, $ot );
 435                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 436                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 437                                                                         $badtag = 1;
 438                                                                 }
 439                                                         }
 440                                                 } else {
 441                                                         if ( $t == 'table' ) {
 442                                                                 $tagstack = array_pop( $tablestack );
 443                                                         }
 444                                                 }
 445                                                 $newparams = '';
 446                                         } else {
 447                                                 # Keep track for later
 448                                                 if ( isset( $tabletags[$t] ) &&
 449                                                 ! in_array( 'table', $tagstack ) ) {
 450                                                         $badtag = 1;
 451                                                 } else if ( in_array( $t, $tagstack ) &&
 452                                                 ! isset( $htmlnest [$t ] ) ) {
 453                                                         $badtag = 1 ;
 454                                                 # Is it a self closed htmlpair ? (bug 5487)
 455                                                 } else if( $brace == '/>' &&
 456                                                 isset( $htmlpairs[$t] ) ) {
 457                                                         $badtag = 1;
 458                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 459                                                         # Hack to force empty tag for uncloseable elements
 460                                                         $brace = '/>';
 461                                                 } else if( isset( $htmlsingle[$t] ) ) {
 462                                                         # Hack to not close $htmlsingle tags
 463                                                         $brace = NULL;
 464                                                 } else if( isset( $tabletags[$t] )
 465                                                 &&  in_array($t ,$tagstack) ) {
 466                                                         // New table tag but forgot to close the previous one
 467                                                         $text .= "</$t>";
 468                                                 } else {
 469                                                         if ( $t == 'table' ) {
 470                                                                 array_push( $tablestack, $tagstack );
 471                                                                 $tagstack = array();
 472                                                         }
 473                                                         array_push( $tagstack, $t );
 474                                                 }
 475
 476                                                 # Replace any variables or template parameters with
 477                                                 # plaintext results.
 478                                                 if( is_callable( $processCallback ) ) {
 479                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 480                                                 }
 481
 482                                                 # Strip non-approved attributes from the tag
 483                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 484                                         }
 485                                         if ( ! $badtag ) {
 486                                                 $rest = str_replace( '>', '&gt;', $rest );
 487                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 488                                                 $text .= "<$slash$t$newparams$close>$rest";
 489                                                 continue;
 490                                         }
 491                                 }
 492                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 493                         }
 494                         # Close off any remaining tags
 495                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 496                                 $text .= "</$t>\n";
 497                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 498                         }
 499                 } else {
 500                         # this might be possible using tidy itself
 501                         foreach ( $bits as $x ) {
 502                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 503                                 $x, $regs );
 504                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 505                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 506                                         if( is_callable( $processCallback ) ) {
 507                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 508                                         }
 509                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 510                                         $rest = str_replace( '>', '&gt;', $rest );
 511                                         $text .= "<$slash$t$newparams$brace$rest";
 512                                 } else {
 513                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 514                                 }
 515                         }
 516                 }
 517                 wfProfileOut( __METHOD__ );
 518                 return $text;
 519         }
 520
 521         /**
 522          * Remove '<!--', '-->', and everything between.
 523          * To avoid leaving blank lines, when a comment is both preceded
 524          * and followed by a newline (ignoring spaces), trim leading and
 525          * trailing spaces and one of the newlines.
 526          *
 527          * @private
 528          * @param string $text
 529          * @return string
 530          */
 531         static function removeHTMLcomments( $text ) {
 532                 wfProfileIn( __METHOD__ );
 533                 while (($start = strpos($text, '<!--')) !== false) {
 534                         $end = strpos($text, '-->', $start + 4);
 535                         if ($end === false) {
 536                                 # Unterminated comment; bail out
 537                                 break;
 538                         }
 539
 540                         $end += 3;
 541
 542                         # Trim space and newline if the comment is both
 543                         # preceded and followed by a newline
 544                         $spaceStart = max($start - 1, 0);
 545                         $spaceLen = $end - $spaceStart;
 546                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 547                                 $spaceStart--;
 548                                 $spaceLen++;
 549                         }
 550                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 551                                 $spaceLen++;
 552                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 553                                 # Remove the comment, leading and trailing
 554                                 # spaces, and leave only one newline.
 555                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 556                         }
 557                         else {
 558                                 # Remove just the comment.
 559                                 $text = substr_replace($text, '', $start, $end - $start);
 560                         }
 561                 }
 562                 wfProfileOut( __METHOD__ );
 563                 return $text;
 564         }
 565
 566         /**
 567          * Take an array of attribute names and values and normalize or discard
 568          * illegal values for the given element type.
 569          *
 570          * - Discards attributes not on a whitelist for the given element
 571          * - Unsafe style attributes are discarded
 572          * - Invalid id attributes are reencoded
 573          *
 574          * @param array $attribs
 575          * @param string $element
 576          * @return array
 577          *
 578          * @todo Check for legal values where the DTD limits things.
 579          * @todo Check for unique id attribute :P
 580          */
 581         static function validateTagAttributes( $attribs, $element ) {
 582                 return Sanitizer::validateAttributes( $attribs,
 583                         Sanitizer::attributeWhitelist( $element ) );
 584         }
 585
 586         /**
 587          * Take an array of attribute names and values and normalize or discard
 588          * illegal values for the given whitelist.
 589          *
 590          * - Discards attributes not the given whitelist
 591          * - Unsafe style attributes are discarded
 592          * - Invalid id attributes are reencoded
 593          *
 594          * @param array $attribs
 595          * @param array $whitelist list of allowed attribute names
 596          * @return array
 597          *
 598          * @todo Check for legal values where the DTD limits things.
 599          * @todo Check for unique id attribute :P
 600          */
 601         static function validateAttributes( $attribs, $whitelist ) {
 602                 $whitelist = array_flip( $whitelist );
 603                 $out = array();
 604                 foreach( $attribs as $attribute => $value ) {
 605                         if( !isset( $whitelist[$attribute] ) ) {
 606                                 continue;
 607                         }
 608                         # Strip javascript "expression" from stylesheets.
 609                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 610                         if( $attribute == 'style' ) {
 611                                 $value = Sanitizer::checkCss( $value );
 612                                 if( $value === false ) {
 613                                         # haxx0r
 614                                         continue;
 615                                 }
 616                         }
 617
 618                         if ( $attribute === 'id' )
 619                                 $value = Sanitizer::escapeId( $value );
 620
 621                         // If this attribute was previously set, override it.
 622                         // Output should only have one attribute of each name.
 623                         $out[$attribute] = $value;
 624                 }
 625                 return $out;
 626         }
 627
 628         /**
 629          * Merge two sets of HTML attributes.
 630          * Conflicting items in the second set will override those
 631          * in the first, except for 'class' attributes which will be
 632          * combined.
 633          *
 634          * @todo implement merging for other attributes such as style
 635          * @param array $a
 636          * @param array $b
 637          * @return array
 638          */
 639         static function mergeAttributes( $a, $b ) {
 640                 $out = array_merge( $a, $b );
 641                 if( isset( $a['class'] )
 642                         && isset( $b['class'] )
 643                         && $a['class'] !== $b['class'] ) {
 644
 645                         $out['class'] = implode( ' ',
 646                                 array_unique(
 647                                         preg_split( '/\s+/',
 648                                                 $a['class'] . ' ' . $b['class'],
 649                                                 -1,
 650                                                 PREG_SPLIT_NO_EMPTY ) ) );
 651                 }
 652                 return $out;
 653         }
 654
 655         /**
 656          * Pick apart some CSS and check it for forbidden or unsafe structures.
 657          * Returns a sanitized string, or false if it was just too evil.
 658          *
 659          * Currently URL references, 'expression', 'tps' are forbidden.
 660          *
 661          * @param string $value
 662          * @return mixed
 663          */
 664         static function checkCss( $value ) {
 665                 $stripped = Sanitizer::decodeCharReferences( $value );
 666
 667                 // Remove any comments; IE gets token splitting wrong
 668                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 669
 670                 $value = $stripped;
 671
 672                 // ... and continue checks
 673                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 674                         'codepointToUtf8(hexdec("$1"))', $stripped );
 675                 $stripped = str_replace( '\\', '', $stripped );
 676                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 677                                 $stripped ) ) {
 678                         # haxx0r
 679                         return false;
 680                 }
 681
 682                 return $value;
 683         }
 684
 685         /**
 686          * Take a tag soup fragment listing an HTML element's attributes
 687          * and normalize it to well-formed XML, discarding unwanted attributes.
 688          * Output is safe for further wikitext processing, with escaping of
 689          * values that could trigger problems.
 690          *
 691          * - Normalizes attribute names to lowercase
 692          * - Discards attributes not on a whitelist for the given element
 693          * - Turns broken or invalid entities into plaintext
 694          * - Double-quotes all attribute values
 695          * - Attributes without values are given the name as attribute
 696          * - Double attributes are discarded
 697          * - Unsafe style attributes are discarded
 698          * - Prepends space if there are attributes.
 699          *
 700          * @param string $text
 701          * @param string $element
 702          * @return string
 703          */
 704         static function fixTagAttributes( $text, $element ) {
 705                 if( trim( $text ) == '' ) {
 706                         return '';
 707                 }
 708
 709                 $stripped = Sanitizer::validateTagAttributes(
 710                         Sanitizer::decodeTagAttributes( $text ), $element );
 711
 712                 $attribs = array();
 713                 foreach( $stripped as $attribute => $value ) {
 714                         $encAttribute = htmlspecialchars( $attribute );
 715                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 716
 717                         $attribs[] = "$encAttribute=\"$encValue\"";
 718                 }
 719                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 720         }
 721
 722         /**
 723          * Encode an attribute value for HTML output.
 724          * @param $text
 725          * @return HTML-encoded text fragment
 726          */
 727         static function encodeAttribute( $text ) {
 728                 $encValue = htmlspecialchars( $text );
 729
 730                 // Whitespace is normalized during attribute decoding,
 731                 // so if we've been passed non-spaces we must encode them
 732                 // ahead of time or they won't be preserved.
 733                 $encValue = strtr( $encValue, array(
 734                         "\n" => '&#10;',
 735                         "\r" => '&#13;',
 736                         "\t" => '&#9;',
 737                 ) );
 738
 739                 return $encValue;
 740         }
 741
 742         /**
 743          * Encode an attribute value for HTML tags, with extra armoring
 744          * against further wiki processing.
 745          * @param $text
 746          * @return HTML-encoded text fragment
 747          */
 748         static function safeEncodeAttribute( $text ) {
 749                 $encValue = Sanitizer::encodeAttribute( $text );
 750
 751                 # Templates and links may be expanded in later parsing,
 752                 # creating invalid or dangerous output. Suppress this.
 753                 $encValue = strtr( $encValue, array(
 754                         '<'    => '&lt;',   // This should never happen,
 755                         '>'    => '&gt;',   // we've received invalid input
 756                         '"'    => '&quot;', // which should have been escaped.
 757                         '{'    => '&#123;',
 758                         '['    => '&#91;',
 759                         "''"   => '&#39;&#39;',
 760                         'ISBN' => '&#73;SBN',
 761                         'RFC'  => '&#82;FC',
 762                         'PMID' => '&#80;MID',
 763                         '|'    => '&#124;',
 764                         '__'   => '&#95;_',
 765                 ) );
 766
 767                 # Stupid hack
 768                 $encValue = preg_replace_callback(
 769                         '/(' . wfUrlProtocols() . ')/',
 770                         array( 'Sanitizer', 'armorLinksCallback' ),
 771                         $encValue );
 772                 return $encValue;
 773         }
 774
 775         /**
 776          * Given a value escape it so that it can be used in an id attribute and
 777          * return it, this does not validate the value however (see first link)
 778          *
 779          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 780          *                                                          in the id and
 781          *                                                          name attributes
 782          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 783          *
 784          * @param string $id    Id to validate
 785          * @param int    $flags Currently only two values: Sanitizer::INITIAL_NONLETTER
 786          *                      (default) permits initial non-letter characters,
 787          *                      such as if you're adding a prefix to them.
 788          *                      Sanitizer::NONE will prepend an 'x' if the id
 789          *                      would otherwise start with a nonletter.
 790          * @return string
 791          */
 792         static function escapeId( $id, $flags = Sanitizer::INITIAL_NONLETTER ) {
 793                 static $replace = array(
 794                         '%3A' => ':',
 795                         '%' => '.'
 796                 );
 797
 798                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 799                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 800
 801                 echo "flags = $flags, ~flags & Sanitizer::INITIAL_NONLETTER = ".(~$flags&Sanitizer::INITIAL_NONLETTER).", id=$id\n";
 802
 803                 if( ~$flags & Sanitizer::INITIAL_NONLETTER
 804                 && !preg_match( '/[a-zA-Z]/', $id[0] ) ) {
 805                         // Initial character must be a letter!
 806                         $id = "x$id";
 807                 }
 808                 return $id;
 809         }
 810
 811         /**
 812          * Given a value, escape it so that it can be used as a CSS class and
 813          * return it.
 814          *
 815          * @todo For extra validity, input should be validated UTF-8.
 816          *
 817          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 818          *
 819          * @param string $class
 820          * @return string
 821          */
 822         static function escapeClass( $class ) {
 823                 // Convert ugly stuff to underscores and kill underscores in ugly places
 824                 return rtrim(preg_replace(
 825                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 826                         '_',
 827                         $class ), '_');
 828         }
 829
 830         /**
 831          * Regex replace callback for armoring links against further processing.
 832          * @param array $matches
 833          * @return string
 834          * @private
 835          */
 836         private static function armorLinksCallback( $matches ) {
 837                 return str_replace( ':', '&#58;', $matches[1] );
 838         }
 839
 840         /**
 841          * Return an associative array of attribute names and values from
 842          * a partial tag string. Attribute names are forces to lowercase,
 843          * character references are decoded to UTF-8 text.
 844          *
 845          * @param string
 846          * @return array
 847          */
 848         static function decodeTagAttributes( $text ) {
 849                 $attribs = array();
 850
 851                 if( trim( $text ) == '' ) {
 852                         return $attribs;
 853                 }
 854
 855                 $pairs = array();
 856                 if( !preg_match_all(
 857                         MW_ATTRIBS_REGEX,
 858                         $text,
 859                         $pairs,
 860                         PREG_SET_ORDER ) ) {
 861                         return $attribs;
 862                 }
 863
 864                 foreach( $pairs as $set ) {
 865                         $attribute = strtolower( $set[1] );
 866                         $value = Sanitizer::getTagAttributeCallback( $set );
 867
 868                         // Normalize whitespace
 869                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 870                         $value = trim( $value );
 871
 872                         // Decode character references
 873                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 874                 }
 875                 return $attribs;
 876         }
 877
 878         /**
 879          * Pick the appropriate attribute value from a match set from the
 880          * MW_ATTRIBS_REGEX matches.
 881          *
 882          * @param array $set
 883          * @return string
 884          * @private
 885          */
 886         private static function getTagAttributeCallback( $set ) {
 887                 if( isset( $set[6] ) ) {
 888                         # Illegal #XXXXXX color with no quotes.
 889                         return $set[6];
 890                 } elseif( isset( $set[5] ) ) {
 891                         # No quotes.
 892                         return $set[5];
 893                 } elseif( isset( $set[4] ) ) {
 894                         # Single-quoted
 895                         return $set[4];
 896                 } elseif( isset( $set[3] ) ) {
 897                         # Double-quoted
 898                         return $set[3];
 899                 } elseif( !isset( $set[2] ) ) {
 900                         # In XHTML, attributes must have a value.
 901                         # For 'reduced' form, return explicitly the attribute name here.
 902                         return $set[1];
 903                 } else {
 904                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 905                 }
 906         }
 907
 908         /**
 909          * Normalize whitespace and character references in an XML source-
 910          * encoded text for an attribute value.
 911          *
 912          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 913          * but note that we're not returning the value, but are returning
 914          * XML source fragments that will be slapped into output.
 915          *
 916          * @param string $text
 917          * @return string
 918          * @private
 919          */
 920         private static function normalizeAttributeValue( $text ) {
 921                 return str_replace( '"', '&quot;',
 922                         self::normalizeWhitespace(
 923                                 Sanitizer::normalizeCharReferences( $text ) ) );
 924         }
 925
 926         private static function normalizeWhitespace( $text ) {
 927                 return preg_replace(
 928                         '/\r\n|[\x20\x0d\x0a\x09]/',
 929                         ' ',
 930                         $text );
 931         }
 932
 933         /**
 934          * Ensure that any entities and character references are legal
 935          * for XML and XHTML specifically. Any stray bits will be
 936          * &amp;-escaped to result in a valid text fragment.
 937          *
 938          * a. any named char refs must be known in XHTML
 939          * b. any numeric char refs must be legal chars, not invalid or forbidden
 940          * c. use &#x, not &#X
 941          * d. fix or reject non-valid attributes
 942          *
 943          * @param string $text
 944          * @return string
 945          * @private
 946          */
 947         static function normalizeCharReferences( $text ) {
 948                 return preg_replace_callback(
 949                         MW_CHAR_REFS_REGEX,
 950                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 951                         $text );
 952         }
 953         /**
 954          * @param string $matches
 955          * @return string
 956          */
 957         static function normalizeCharReferencesCallback( $matches ) {
 958                 $ret = null;
 959                 if( $matches[1] != '' ) {
 960                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 961                 } elseif( $matches[2] != '' ) {
 962                         $ret = Sanitizer::decCharReference( $matches[2] );
 963                 } elseif( $matches[3] != ''  ) {
 964                         $ret = Sanitizer::hexCharReference( $matches[3] );
 965                 } elseif( $matches[4] != '' ) {
 966                         $ret = Sanitizer::hexCharReference( $matches[4] );
 967                 }
 968                 if( is_null( $ret ) ) {
 969                         return htmlspecialchars( $matches[0] );
 970                 } else {
 971                         return $ret;
 972                 }
 973         }
 974
 975         /**
 976          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 977          * return the named entity reference as is. If the entity is a
 978          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
 979          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 980          *
 981          * @param string $name
 982          * @return string
 983          * @static
 984          */
 985         static function normalizeEntity( $name ) {
 986                 global $wgHtmlEntities, $wgHtmlEntityAliases;
 987                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
 988                         return "&{$wgHtmlEntityAliases[$name]};";
 989                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
 990                         return "&$name;";
 991                 } else {
 992                         return "&amp;$name;";
 993                 }
 994         }
 995
 996         static function decCharReference( $codepoint ) {
 997                 $point = intval( $codepoint );
 998                 if( Sanitizer::validateCodepoint( $point ) ) {
 999                         return sprintf( '&#%d;', $point );
1000                 } else {
1001                         return null;
1002                 }
1003         }
1004
1005         static function hexCharReference( $codepoint ) {
1006                 $point = hexdec( $codepoint );
1007                 if( Sanitizer::validateCodepoint( $point ) ) {
1008                         return sprintf( '&#x%x;', $point );
1009                 } else {
1010                         return null;
1011                 }
1012         }
1013
1014         /**
1015          * Returns true if a given Unicode codepoint is a valid character in XML.
1016          * @param int $codepoint
1017          * @return bool
1018          */
1019         private static function validateCodepoint( $codepoint ) {
1020                 return ($codepoint ==    0x09)
1021                         || ($codepoint ==    0x0a)
1022                         || ($codepoint ==    0x0d)
1023                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1024                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1025                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1026         }
1027
1028         /**
1029          * Decode any character references, numeric or named entities,
1030          * in the text and return a UTF-8 string.
1031          *
1032          * @param string $text
1033          * @return string
1034          * @public
1035          * @static
1036          */
1037         public static function decodeCharReferences( $text ) {
1038                 return preg_replace_callback(
1039                         MW_CHAR_REFS_REGEX,
1040                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1041                         $text );
1042         }
1043
1044         /**
1045          * @param string $matches
1046          * @return string
1047          */
1048         static function decodeCharReferencesCallback( $matches ) {
1049                 if( $matches[1] != '' ) {
1050                         return Sanitizer::decodeEntity( $matches[1] );
1051                 } elseif( $matches[2] != '' ) {
1052                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1053                 } elseif( $matches[3] != ''  ) {
1054                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1055                 } elseif( $matches[4] != '' ) {
1056                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1057                 }
1058                 # Last case should be an ampersand by itself
1059                 return $matches[0];
1060         }
1061
1062         /**
1063          * Return UTF-8 string for a codepoint if that is a valid
1064          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1065          * @param int $codepoint
1066          * @return string
1067          * @private
1068          */
1069         static function decodeChar( $codepoint ) {
1070                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1071                         return codepointToUtf8( $codepoint );
1072                 } else {
1073                         return UTF8_REPLACEMENT;
1074                 }
1075         }
1076
1077         /**
1078          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1079          * return the UTF-8 encoding of that character. Otherwise, returns
1080          * pseudo-entity source (eg &foo;)
1081          *
1082          * @param string $name
1083          * @return string
1084          */
1085         static function decodeEntity( $name ) {
1086                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1087                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1088                         $name = $wgHtmlEntityAliases[$name];
1089                 }
1090                 if( isset( $wgHtmlEntities[$name] ) ) {
1091                         return codepointToUtf8( $wgHtmlEntities[$name] );
1092                 } else {
1093                         return "&$name;";
1094                 }
1095         }
1096
1097         /**
1098          * Fetch the whitelist of acceptable attributes for a given
1099          * element name.
1100          *
1101          * @param string $element
1102          * @return array
1103          */
1104         static function attributeWhitelist( $element ) {
1105                 static $list;
1106                 if( !isset( $list ) ) {
1107                         $list = Sanitizer::setupAttributeWhitelist();
1108                 }
1109                 return isset( $list[$element] )
1110                         ? $list[$element]
1111                         : array();
1112         }
1113
1114         /**
1115          * @todo Document it a bit
1116          * @return array
1117          */
1118         static function setupAttributeWhitelist() {
1119                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1120                 $block = array_merge( $common, array( 'align' ) );
1121                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1122                 $tablecell = array( 'abbr',
1123                                     'axis',
1124                                     'headers',
1125                                     'scope',
1126                                     'rowspan',
1127                                     'colspan',
1128                                     'nowrap', # deprecated
1129                                     'width',  # deprecated
1130                                     'height', # deprecated
1131                                     'bgcolor' # deprecated
1132                                     );
1133
1134                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1135                 # See: http://www.w3.org/TR/html4/
1136                 $whitelist = array (
1137                         # 7.5.4
1138                         'div'        => $block,
1139                         'center'     => $common, # deprecated
1140                         'span'       => $block, # ??
1141
1142                         # 7.5.5
1143                         'h1'         => $block,
1144                         'h2'         => $block,
1145                         'h3'         => $block,
1146                         'h4'         => $block,
1147                         'h5'         => $block,
1148                         'h6'         => $block,
1149
1150                         # 7.5.6
1151                         # address
1152
1153                         # 8.2.4
1154                         # bdo
1155
1156                         # 9.2.1
1157                         'em'         => $common,
1158                         'strong'     => $common,
1159                         'cite'       => $common,
1160                         # dfn
1161                         'code'       => $common,
1162                         # samp
1163                         # kbd
1164                         'var'        => $common,
1165                         # abbr
1166                         # acronym
1167
1168                         # 9.2.2
1169                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1170                         # q
1171
1172                         # 9.2.3
1173                         'sub'        => $common,
1174                         'sup'        => $common,
1175
1176                         # 9.3.1
1177                         'p'          => $block,
1178
1179                         # 9.3.2
1180                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1181
1182                         # 9.3.4
1183                         'pre'        => array_merge( $common, array( 'width' ) ),
1184
1185                         # 9.4
1186                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1187                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1188
1189                         # 10.2
1190                         'ul'         => array_merge( $common, array( 'type' ) ),
1191                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1192                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1193
1194                         # 10.3
1195                         'dl'         => $common,
1196                         'dd'         => $common,
1197                         'dt'         => $common,
1198
1199                         # 11.2.1
1200                         'table'      => array_merge( $common,
1201                                                                 array( 'summary', 'width', 'border', 'frame',
1202                                                                                 'rules', 'cellspacing', 'cellpadding',
1203                                                                                 'align', 'bgcolor',
1204                                                                 ) ),
1205
1206                         # 11.2.2
1207                         'caption'    => array_merge( $common, array( 'align' ) ),
1208
1209                         # 11.2.3
1210                         'thead'      => array_merge( $common, $tablealign ),
1211                         'tfoot'      => array_merge( $common, $tablealign ),
1212                         'tbody'      => array_merge( $common, $tablealign ),
1213
1214                         # 11.2.4
1215                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1216                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1217
1218                         # 11.2.5
1219                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1220
1221                         # 11.2.6
1222                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1223                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1224
1225                         # 13.2
1226                         # Not usually allowed, but may be used for extension-style hooks
1227                         # such as <math> when it is rasterized
1228                         'img'        => array_merge( $common, array( 'alt' ) ),
1229
1230                         # 15.2.1
1231                         'tt'         => $common,
1232                         'b'          => $common,
1233                         'i'          => $common,
1234                         'big'        => $common,
1235                         'small'      => $common,
1236                         'strike'     => $common,
1237                         's'          => $common,
1238                         'u'          => $common,
1239
1240                         # 15.2.2
1241                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1242                         # basefont
1243
1244                         # 15.3
1245                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1246
1247                         # XHTML Ruby annotation text module, simple ruby only.
1248                         # http://www.w3c.org/TR/ruby/
1249                         'ruby'       => $common,
1250                         # rbc
1251                         # rtc
1252                         'rb'         => $common,
1253                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1254                         'rp'         => $common,
1255
1256                         # MathML root element, where used for extensions
1257                         # 'title' may not be 100% valid here; it's XHTML
1258                         # http://www.w3.org/TR/REC-MathML/
1259                         'math'       => array( 'class', 'style', 'id', 'title' ),
1260                         );
1261                 return $whitelist;
1262         }
1263
1264         /**
1265          * Take a fragment of (potentially invalid) HTML and return
1266          * a version with any tags removed, encoded as plain text.
1267          *
1268          * Warning: this return value must be further escaped for literal
1269          * inclusion in HTML output as of 1.10!
1270          *
1271          * @param string $text HTML fragment
1272          * @return string
1273          */
1274         static function stripAllTags( $text ) {
1275                 # Actual <tags>
1276                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1277
1278                 # Normalize &entities and whitespace
1279                 $text = self::decodeCharReferences( $text );
1280                 $text = self::normalizeWhitespace( $text );
1281
1282                 return $text;
1283         }
1284
1285         /**
1286          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1287          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1288          * PHP 5.1 doesn't.
1289          *
1290          * Use for passing XHTML fragments to PHP's XML parsing functions
1291          *
1292          * @return string
1293          * @static
1294          */
1295         static function hackDocType() {
1296                 global $wgHtmlEntities;
1297                 $out = "<!DOCTYPE html [\n";
1298                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1299                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1300                 }
1301                 $out .= "]>\n";
1302                 return $out;
1303         }
1304
1305         static function cleanUrl( $url, $hostname=true ) {
1306                 # Normalize any HTML entities in input. They will be
1307                 # re-escaped by makeExternalLink().
1308                 $url = Sanitizer::decodeCharReferences( $url );
1309
1310                 # Escape any control characters introduced by the above step
1311                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1312
1313                 # Validate hostname portion
1314                 $matches = array();
1315                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1316                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1317
1318                         // Characters that will be ignored in IDNs.
1319                         // http://tools.ietf.org/html/3454#section-3.1
1320                         // Strip them before further processing so blacklists and such work.
1321                         $strip = "/
1322                                 \\s|          # general whitespace
1323                                 \xc2\xad|     # 00ad SOFT HYPHEN
1324                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1325                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1326                                 \xe2\x81\xa0| # 2060 WORD JOINER
1327                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1328                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1329                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1330                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1331                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1332                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1333                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1334                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1335                                 /xuD";
1336
1337                         $host = preg_replace( $strip, '', $host );
1338
1339                         // @fixme: validate hostnames here
1340
1341                         return $protocol . $host . $rest;
1342                 } else {
1343                         return $url;
1344                 }
1345         }
1346
1347 }
1348
1349