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  * @file
  24  * @ingroup Parser
  25  */
  26
  27 /**
  28  * Regular expression to match various types of character references in
  29  * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  30  */
  31 define( 'MW_CHAR_REFS_REGEX',
  32         '/&([A-Za-z0-9\x80-\xff]+);
  33          |&\#([0-9]+);
  34          |&\#x([0-9A-Za-z]+);
  35          |&\#X([0-9A-Za-z]+);
  36          |(&)/x' );
  37
  38 /**
  39  * Regular expression to match HTML/XML attribute pairs within a tag.
  40  * Allows some... latitude.
  41  * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
  42  */
  43 $attrib = '[A-Za-z0-9]';
  44 $space = '[\x09\x0a\x0d\x20]';
  45 define( 'MW_ATTRIBS_REGEX',
  46         "/(?:^|$space)($attrib+)
  47           ($space*=$space*
  48                 (?:
  49                  # The attribute value: quoted or alone
  50                   \"([^<\"]*)\"
  51                  | '([^<']*)'
  52                  |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
  53                  |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
  54                                                          # colors are specified like this.
  55                                                          # We'll be normalizing it.
  56                 )
  57            )?(?=$space|\$)/sx" );
  58
  59 /**
  60  * List of all named character entities defined in HTML 4.01
  61  * http://www.w3.org/TR/html4/sgml/entities.html
  62  * @private
  63  */
  64 global $wgHtmlEntities;
  65 $wgHtmlEntities = array(
  66         'Aacute'   => 193,
  67         'aacute'   => 225,
  68         'Acirc'    => 194,
  69         'acirc'    => 226,
  70         'acute'    => 180,
  71         'AElig'    => 198,
  72         'aelig'    => 230,
  73         'Agrave'   => 192,
  74         'agrave'   => 224,
  75         'alefsym'  => 8501,
  76         'Alpha'    => 913,
  77         'alpha'    => 945,
  78         'amp'      => 38,
  79         'and'      => 8743,
  80         'ang'      => 8736,
  81         'Aring'    => 197,
  82         'aring'    => 229,
  83         'asymp'    => 8776,
  84         'Atilde'   => 195,
  85         'atilde'   => 227,
  86         'Auml'     => 196,
  87         'auml'     => 228,
  88         'bdquo'    => 8222,
  89         'Beta'     => 914,
  90         'beta'     => 946,
  91         'brvbar'   => 166,
  92         'bull'     => 8226,
  93         'cap'      => 8745,
  94         'Ccedil'   => 199,
  95         'ccedil'   => 231,
  96         'cedil'    => 184,
  97         'cent'     => 162,
  98         'Chi'      => 935,
  99         'chi'      => 967,
 100         'circ'     => 710,
 101         'clubs'    => 9827,
 102         'cong'     => 8773,
 103         'copy'     => 169,
 104         'crarr'    => 8629,
 105         'cup'      => 8746,
 106         'curren'   => 164,
 107         'dagger'   => 8224,
 108         'Dagger'   => 8225,
 109         'darr'     => 8595,
 110         'dArr'     => 8659,
 111         'deg'      => 176,
 112         'Delta'    => 916,
 113         'delta'    => 948,
 114         'diams'    => 9830,
 115         'divide'   => 247,
 116         'Eacute'   => 201,
 117         'eacute'   => 233,
 118         'Ecirc'    => 202,
 119         'ecirc'    => 234,
 120         'Egrave'   => 200,
 121         'egrave'   => 232,
 122         'empty'    => 8709,
 123         'emsp'     => 8195,
 124         'ensp'     => 8194,
 125         'Epsilon'  => 917,
 126         'epsilon'  => 949,
 127         'equiv'    => 8801,
 128         'Eta'      => 919,
 129         'eta'      => 951,
 130         'ETH'      => 208,
 131         'eth'      => 240,
 132         'Euml'     => 203,
 133         'euml'     => 235,
 134         'euro'     => 8364,
 135         'exist'    => 8707,
 136         'fnof'     => 402,
 137         'forall'   => 8704,
 138         'frac12'   => 189,
 139         'frac14'   => 188,
 140         'frac34'   => 190,
 141         'frasl'    => 8260,
 142         'Gamma'    => 915,
 143         'gamma'    => 947,
 144         'ge'       => 8805,
 145         'gt'       => 62,
 146         'harr'     => 8596,
 147         'hArr'     => 8660,
 148         'hearts'   => 9829,
 149         'hellip'   => 8230,
 150         'Iacute'   => 205,
 151         'iacute'   => 237,
 152         'Icirc'    => 206,
 153         'icirc'    => 238,
 154         'iexcl'    => 161,
 155         'Igrave'   => 204,
 156         'igrave'   => 236,
 157         'image'    => 8465,
 158         'infin'    => 8734,
 159         'int'      => 8747,
 160         'Iota'     => 921,
 161         'iota'     => 953,
 162         'iquest'   => 191,
 163         'isin'     => 8712,
 164         'Iuml'     => 207,
 165         'iuml'     => 239,
 166         'Kappa'    => 922,
 167         'kappa'    => 954,
 168         'Lambda'   => 923,
 169         'lambda'   => 955,
 170         'lang'     => 9001,
 171         'laquo'    => 171,
 172         'larr'     => 8592,
 173         'lArr'     => 8656,
 174         'lceil'    => 8968,
 175         'ldquo'    => 8220,
 176         'le'       => 8804,
 177         'lfloor'   => 8970,
 178         'lowast'   => 8727,
 179         'loz'      => 9674,
 180         'lrm'      => 8206,
 181         'lsaquo'   => 8249,
 182         'lsquo'    => 8216,
 183         'lt'       => 60,
 184         'macr'     => 175,
 185         'mdash'    => 8212,
 186         'micro'    => 181,
 187         'middot'   => 183,
 188         'minus'    => 8722,
 189         'Mu'       => 924,
 190         'mu'       => 956,
 191         'nabla'    => 8711,
 192         'nbsp'     => 160,
 193         'ndash'    => 8211,
 194         'ne'       => 8800,
 195         'ni'       => 8715,
 196         'not'      => 172,
 197         'notin'    => 8713,
 198         'nsub'     => 8836,
 199         'Ntilde'   => 209,
 200         'ntilde'   => 241,
 201         'Nu'       => 925,
 202         'nu'       => 957,
 203         'Oacute'   => 211,
 204         'oacute'   => 243,
 205         'Ocirc'    => 212,
 206         'ocirc'    => 244,
 207         'OElig'    => 338,
 208         'oelig'    => 339,
 209         'Ograve'   => 210,
 210         'ograve'   => 242,
 211         'oline'    => 8254,
 212         'Omega'    => 937,
 213         'omega'    => 969,
 214         'Omicron'  => 927,
 215         'omicron'  => 959,
 216         'oplus'    => 8853,
 217         'or'       => 8744,
 218         'ordf'     => 170,
 219         'ordm'     => 186,
 220         'Oslash'   => 216,
 221         'oslash'   => 248,
 222         'Otilde'   => 213,
 223         'otilde'   => 245,
 224         'otimes'   => 8855,
 225         'Ouml'     => 214,
 226         'ouml'     => 246,
 227         'para'     => 182,
 228         'part'     => 8706,
 229         'permil'   => 8240,
 230         'perp'     => 8869,
 231         'Phi'      => 934,
 232         'phi'      => 966,
 233         'Pi'       => 928,
 234         'pi'       => 960,
 235         'piv'      => 982,
 236         'plusmn'   => 177,
 237         'pound'    => 163,
 238         'prime'    => 8242,
 239         'Prime'    => 8243,
 240         'prod'     => 8719,
 241         'prop'     => 8733,
 242         'Psi'      => 936,
 243         'psi'      => 968,
 244         'quot'     => 34,
 245         'radic'    => 8730,
 246         'rang'     => 9002,
 247         'raquo'    => 187,
 248         'rarr'     => 8594,
 249         'rArr'     => 8658,
 250         'rceil'    => 8969,
 251         'rdquo'    => 8221,
 252         'real'     => 8476,
 253         'reg'      => 174,
 254         'rfloor'   => 8971,
 255         'Rho'      => 929,
 256         'rho'      => 961,
 257         'rlm'      => 8207,
 258         'rsaquo'   => 8250,
 259         'rsquo'    => 8217,
 260         'sbquo'    => 8218,
 261         'Scaron'   => 352,
 262         'scaron'   => 353,
 263         'sdot'     => 8901,
 264         'sect'     => 167,
 265         'shy'      => 173,
 266         'Sigma'    => 931,
 267         'sigma'    => 963,
 268         'sigmaf'   => 962,
 269         'sim'      => 8764,
 270         'spades'   => 9824,
 271         'sub'      => 8834,
 272         'sube'     => 8838,
 273         'sum'      => 8721,
 274         'sup'      => 8835,
 275         'sup1'     => 185,
 276         'sup2'     => 178,
 277         'sup3'     => 179,
 278         'supe'     => 8839,
 279         'szlig'    => 223,
 280         'Tau'      => 932,
 281         'tau'      => 964,
 282         'there4'   => 8756,
 283         'Theta'    => 920,
 284         'theta'    => 952,
 285         'thetasym' => 977,
 286         'thinsp'   => 8201,
 287         'THORN'    => 222,
 288         'thorn'    => 254,
 289         'tilde'    => 732,
 290         'times'    => 215,
 291         'trade'    => 8482,
 292         'Uacute'   => 218,
 293         'uacute'   => 250,
 294         'uarr'     => 8593,
 295         'uArr'     => 8657,
 296         'Ucirc'    => 219,
 297         'ucirc'    => 251,
 298         'Ugrave'   => 217,
 299         'ugrave'   => 249,
 300         'uml'      => 168,
 301         'upsih'    => 978,
 302         'Upsilon'  => 933,
 303         'upsilon'  => 965,
 304         'Uuml'     => 220,
 305         'uuml'     => 252,
 306         'weierp'   => 8472,
 307         'Xi'       => 926,
 308         'xi'       => 958,
 309         'Yacute'   => 221,
 310         'yacute'   => 253,
 311         'yen'      => 165,
 312         'Yuml'     => 376,
 313         'yuml'     => 255,
 314         'Zeta'     => 918,
 315         'zeta'     => 950,
 316         'zwj'      => 8205,
 317         'zwnj'     => 8204 );
 318
 319 /**
 320  * Character entity aliases accepted by MediaWiki
 321  */
 322 global $wgHtmlEntityAliases;
 323 $wgHtmlEntityAliases = array(
 324         'רלמ' => 'rlm',
 325         'رلم' => 'rlm',
 326 );
 327
 328
 329 /**
 330  * XHTML sanitizer for MediaWiki
 331  * @ingroup Parser
 332  */
 333 class Sanitizer {
 334         /**
 335          * Cleans up HTML, removes dangerous tags and attributes, and
 336          * removes HTML comments
 337          * @private
 338          * @param string $text
 339          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 340          * @param array $args for the processing callback
 341          * @param array $extratags for any extra tags to include
 342          * @param array $removetags for any tags (default or extra) to exclude
 343          * @return string
 344          */
 345         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array(), $removetags = 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_unique( array_merge( $htmlsingle, $tabletags ) );
 383                         # Only allow elements that aren't specified in $removetags
 384                         # Doing it here since this is the top-level check
 385                         $htmlelements = array_diff( array_unique( array_merge( $htmlsingle, $htmlpairs, $htmlnest ) ), $removetags );
 386
 387                         # Convert them all to hashtables for faster lookup
 388                         $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 389                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
 390                         foreach ( $vars as $var ) {
 391                                 $$var = array_flip( $$var );
 392                         }
 393                         $staticInitialised = true;
 394                 }
 395
 396                 # Remove HTML comments
 397                 $text = Sanitizer::removeHTMLcomments( $text );
 398                 $bits = explode( '<', $text );
 399                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 400                 if(!$wgUseTidy) {
 401                         $tagstack = $tablestack = array();
 402                         foreach ( $bits as $x ) {
 403                                 $regs = array();
 404                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 405                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 406                                 } else {
 407                                         $slash = $t = $params = $brace = $rest = null;
 408                                 }
 409
 410                                 $badtag = 0 ;
 411                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 412                                         # Check our stack
 413                                         if ( $slash ) {
 414                                                 # Closing a tag...
 415                                                 if( isset( $htmlsingleonly[$t] ) ) {
 416                                                         $badtag = 1;
 417                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 418                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 419                                                                 # Pop all elements with an optional close tag
 420                                                                 # and see if we find a match below them
 421                                                                 $optstack = array();
 422                                                                 array_push ($optstack, $ot);
 423                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 424                                                                                 isset( $htmlsingleallowed[$ot] ) )
 425                                                                 {
 426                                                                         array_push ($optstack, $ot);
 427                                                                 }
 428                                                                 if ( $t != $ot ) {
 429                                                                         # No match. Push the optinal elements back again
 430                                                                         $badtag = 1;
 431                                                                         while ( $ot = @array_pop( $optstack ) ) {
 432                                                                                 array_push( $tagstack, $ot );
 433                                                                         }
 434                                                                 }
 435                                                         } else {
 436                                                                 @array_push( $tagstack, $ot );
 437                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 438                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 439                                                                         $badtag = 1;
 440                                                                 }
 441                                                         }
 442                                                 } else {
 443                                                         if ( $t == 'table' ) {
 444                                                                 $tagstack = array_pop( $tablestack );
 445                                                         }
 446                                                 }
 447                                                 $newparams = '';
 448                                         } else {
 449                                                 # Keep track for later
 450                                                 if ( isset( $tabletags[$t] ) &&
 451                                                 ! in_array( 'table', $tagstack ) ) {
 452                                                         $badtag = 1;
 453                                                 } else if ( in_array( $t, $tagstack ) &&
 454                                                 ! isset( $htmlnest [$t ] ) ) {
 455                                                         $badtag = 1 ;
 456                                                 # Is it a self closed htmlpair ? (bug 5487)
 457                                                 } else if( $brace == '/>' &&
 458                                                 isset( $htmlpairs[$t] ) ) {
 459                                                         $badtag = 1;
 460                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 461                                                         # Hack to force empty tag for uncloseable elements
 462                                                         $brace = '/>';
 463                                                 } else if( isset( $htmlsingle[$t] ) ) {
 464                                                         # Hack to not close $htmlsingle tags
 465                                                         $brace = NULL;
 466                                                 } else if( isset( $tabletags[$t] )
 467                                                 &&  in_array($t ,$tagstack) ) {
 468                                                         // New table tag but forgot to close the previous one
 469                                                         $text .= "</$t>";
 470                                                 } else {
 471                                                         if ( $t == 'table' ) {
 472                                                                 array_push( $tablestack, $tagstack );
 473                                                                 $tagstack = array();
 474                                                         }
 475                                                         array_push( $tagstack, $t );
 476                                                 }
 477
 478                                                 # Replace any variables or template parameters with
 479                                                 # plaintext results.
 480                                                 if( is_callable( $processCallback ) ) {
 481                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 482                                                 }
 483
 484                                                 # Strip non-approved attributes from the tag
 485                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 486                                         }
 487                                         if ( ! $badtag ) {
 488                                                 $rest = str_replace( '>', '&gt;', $rest );
 489                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 490                                                 $text .= "<$slash$t$newparams$close>$rest";
 491                                                 continue;
 492                                         }
 493                                 }
 494                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 495                         }
 496                         # Close off any remaining tags
 497                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 498                                 $text .= "</$t>\n";
 499                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 500                         }
 501                 } else {
 502                         # this might be possible using tidy itself
 503                         foreach ( $bits as $x ) {
 504                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 505                                 $x, $regs );
 506                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 507                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 508                                         if( is_callable( $processCallback ) ) {
 509                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 510                                         }
 511                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 512                                         $rest = str_replace( '>', '&gt;', $rest );
 513                                         $text .= "<$slash$t$newparams$brace$rest";
 514                                 } else {
 515                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 516                                 }
 517                         }
 518                 }
 519                 wfProfileOut( __METHOD__ );
 520                 return $text;
 521         }
 522
 523         /**
 524          * Remove '<!--', '-->', and everything between.
 525          * To avoid leaving blank lines, when a comment is both preceded
 526          * and followed by a newline (ignoring spaces), trim leading and
 527          * trailing spaces and one of the newlines.
 528          *
 529          * @private
 530          * @param string $text
 531          * @return string
 532          */
 533         static function removeHTMLcomments( $text ) {
 534                 wfProfileIn( __METHOD__ );
 535                 while (($start = strpos($text, '<!--')) !== false) {
 536                         $end = strpos($text, '-->', $start + 4);
 537                         if ($end === false) {
 538                                 # Unterminated comment; bail out
 539                                 break;
 540                         }
 541
 542                         $end += 3;
 543
 544                         # Trim space and newline if the comment is both
 545                         # preceded and followed by a newline
 546                         $spaceStart = max($start - 1, 0);
 547                         $spaceLen = $end - $spaceStart;
 548                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 549                                 $spaceStart--;
 550                                 $spaceLen++;
 551                         }
 552                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 553                                 $spaceLen++;
 554                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 555                                 # Remove the comment, leading and trailing
 556                                 # spaces, and leave only one newline.
 557                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 558                         }
 559                         else {
 560                                 # Remove just the comment.
 561                                 $text = substr_replace($text, '', $start, $end - $start);
 562                         }
 563                 }
 564                 wfProfileOut( __METHOD__ );
 565                 return $text;
 566         }
 567
 568         /**
 569          * Take an array of attribute names and values and normalize or discard
 570          * illegal values for the given element type.
 571          *
 572          * - Discards attributes not on a whitelist for the given element
 573          * - Unsafe style attributes are discarded
 574          * - Invalid id attributes are reencoded
 575          *
 576          * @param array $attribs
 577          * @param string $element
 578          * @return array
 579          *
 580          * @todo Check for legal values where the DTD limits things.
 581          * @todo Check for unique id attribute :P
 582          */
 583         static function validateTagAttributes( $attribs, $element ) {
 584                 return Sanitizer::validateAttributes( $attribs,
 585                         Sanitizer::attributeWhitelist( $element ) );
 586         }
 587
 588         /**
 589          * Take an array of attribute names and values and normalize or discard
 590          * illegal values for the given whitelist.
 591          *
 592          * - Discards attributes not the given whitelist
 593          * - Unsafe style attributes are discarded
 594          * - Invalid id attributes are reencoded
 595          *
 596          * @param array $attribs
 597          * @param array $whitelist list of allowed attribute names
 598          * @return array
 599          *
 600          * @todo Check for legal values where the DTD limits things.
 601          * @todo Check for unique id attribute :P
 602          */
 603         static function validateAttributes( $attribs, $whitelist ) {
 604                 $whitelist = array_flip( $whitelist );
 605                 $out = array();
 606                 foreach( $attribs as $attribute => $value ) {
 607                         if( !isset( $whitelist[$attribute] ) ) {
 608                                 continue;
 609                         }
 610                         # Strip javascript "expression" from stylesheets.
 611                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 612                         if( $attribute == 'style' ) {
 613                                 $value = Sanitizer::checkCss( $value );
 614                                 if( $value === false ) {
 615                                         # haxx0r
 616                                         continue;
 617                                 }
 618                         }
 619
 620                         if ( $attribute === 'id' ) {
 621                                 global $wgEnforceHtmlIds;
 622                                 $value = Sanitizer::escapeId( $value,
 623                                         $wgEnforceHtmlIds ? 'noninitial' : 'xml' );
 624                         }
 625
 626                         // If this attribute was previously set, override it.
 627                         // Output should only have one attribute of each name.
 628                         $out[$attribute] = $value;
 629                 }
 630                 return $out;
 631         }
 632
 633         /**
 634          * Merge two sets of HTML attributes.  Conflicting items in the second set
 635          * will override those in the first, except for 'class' attributes which
 636          * will be combined (if they're both strings).
 637          *
 638          * @todo implement merging for other attributes such as style
 639          * @param array $a
 640          * @param array $b
 641          * @return array
 642          */
 643         static function mergeAttributes( $a, $b ) {
 644                 $out = array_merge( $a, $b );
 645                 if( isset( $a['class'] ) && isset( $b['class'] )
 646                 && is_string( $a['class'] ) && is_string( $b['class'] )
 647                 && $a['class'] !== $b['class'] ) {
 648                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 649                                 -1, PREG_SPLIT_NO_EMPTY );
 650                         $out['class'] = implode( ' ', array_unique( $classes ) );
 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, ENT_QUOTES );
 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 mixed  $options String or array of strings (default is array()):
 786          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 787          *       so don't pay attention if the first character isn't valid at the
 788          *       beginning of an id.
 789          *   'xml': Don't restrict the id to be HTML4-compatible.  This option
 790          *       allows any alphabetic character to be used, per the XML standard.
 791          *       Therefore, it also completely changes the type of escaping: instead
 792          *       of weird dot-encoding, runs of invalid characters (mostly
 793          *       whitespace) are just compressed into a single underscore.
 794          * @return string
 795          */
 796         static function escapeId( $id, $options = array() ) {
 797                 $options = (array)$options;
 798
 799                 if ( !in_array( 'xml', $options ) ) {
 800                         # HTML4-style escaping
 801                         static $replace = array(
 802                                 '%3A' => ':',
 803                                 '%' => '.'
 804                         );
 805
 806                         $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 807                         $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 808
 809                         if ( !preg_match( '/^[a-zA-Z]/', $id )
 810                         && !in_array( 'noninitial', $options ) )  {
 811                                 // Initial character must be a letter!
 812                                 $id = "x$id";
 813                         }
 814                         return $id;
 815                 }
 816
 817                 # XML-style escaping.  For the patterns used, see the XML 1.0 standard,
 818                 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
 819                 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
 820                         . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
 821                         . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
 822                 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
 823                         . '\x{203F}-\x{2040}';
 824                 # Replace _ as well so we don't get multiple consecutive underscores
 825                 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
 826                 $id = trim( $id, '_' );
 827
 828                 if ( !preg_match( "/^[$nameStartChar]/u", $id )
 829                 && !in_array( 'noninitial', $options ) ) {
 830                         $id = "_$id";
 831                 }
 832
 833                 return $id;
 834         }
 835
 836         /**
 837          * Given a value, escape it so that it can be used as a CSS class and
 838          * return it.
 839          *
 840          * @todo For extra validity, input should be validated UTF-8.
 841          *
 842          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 843          *
 844          * @param string $class
 845          * @return string
 846          */
 847         static function escapeClass( $class ) {
 848                 // Convert ugly stuff to underscores and kill underscores in ugly places
 849                 return rtrim(preg_replace(
 850                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 851                         '_',
 852                         $class ), '_');
 853         }
 854
 855         /**
 856          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 857          * This allows (generally harmless) entities like &nbsp; to survive.
 858          *
 859          * @param  string $html String to escape
 860          * @return string Escaped input
 861          */
 862         static function escapeHtmlAllowEntities( $html ) {
 863                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 864                 # hurt.
 865                 $html = htmlspecialchars( $html, ENT_QUOTES );
 866                 $html = str_replace( '&amp;', '&', $html );
 867                 $html = Sanitizer::normalizeCharReferences( $html );
 868                 return $html;
 869         }
 870
 871         /**
 872          * Regex replace callback for armoring links against further processing.
 873          * @param array $matches
 874          * @return string
 875          * @private
 876          */
 877         private static function armorLinksCallback( $matches ) {
 878                 return str_replace( ':', '&#58;', $matches[1] );
 879         }
 880
 881         /**
 882          * Return an associative array of attribute names and values from
 883          * a partial tag string. Attribute names are forces to lowercase,
 884          * character references are decoded to UTF-8 text.
 885          *
 886          * @param string
 887          * @return array
 888          */
 889         public static function decodeTagAttributes( $text ) {
 890                 $attribs = array();
 891
 892                 if( trim( $text ) == '' ) {
 893                         return $attribs;
 894                 }
 895
 896                 $pairs = array();
 897                 if( !preg_match_all(
 898                         MW_ATTRIBS_REGEX,
 899                         $text,
 900                         $pairs,
 901                         PREG_SET_ORDER ) ) {
 902                         return $attribs;
 903                 }
 904
 905                 foreach( $pairs as $set ) {
 906                         $attribute = strtolower( $set[1] );
 907                         $value = Sanitizer::getTagAttributeCallback( $set );
 908
 909                         // Normalize whitespace
 910                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 911                         $value = trim( $value );
 912
 913                         // Decode character references
 914                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 915                 }
 916                 return $attribs;
 917         }
 918
 919         /**
 920          * Pick the appropriate attribute value from a match set from the
 921          * MW_ATTRIBS_REGEX matches.
 922          *
 923          * @param array $set
 924          * @return string
 925          * @private
 926          */
 927         private static function getTagAttributeCallback( $set ) {
 928                 if( isset( $set[6] ) ) {
 929                         # Illegal #XXXXXX color with no quotes.
 930                         return $set[6];
 931                 } elseif( isset( $set[5] ) ) {
 932                         # No quotes.
 933                         return $set[5];
 934                 } elseif( isset( $set[4] ) ) {
 935                         # Single-quoted
 936                         return $set[4];
 937                 } elseif( isset( $set[3] ) ) {
 938                         # Double-quoted
 939                         return $set[3];
 940                 } elseif( !isset( $set[2] ) ) {
 941                         # In XHTML, attributes must have a value.
 942                         # For 'reduced' form, return explicitly the attribute name here.
 943                         return $set[1];
 944                 } else {
 945                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 946                 }
 947         }
 948
 949         /**
 950          * Normalize whitespace and character references in an XML source-
 951          * encoded text for an attribute value.
 952          *
 953          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 954          * but note that we're not returning the value, but are returning
 955          * XML source fragments that will be slapped into output.
 956          *
 957          * @param string $text
 958          * @return string
 959          * @private
 960          */
 961         private static function normalizeAttributeValue( $text ) {
 962                 return str_replace( '"', '&quot;',
 963                         self::normalizeWhitespace(
 964                                 Sanitizer::normalizeCharReferences( $text ) ) );
 965         }
 966
 967         private static function normalizeWhitespace( $text ) {
 968                 return preg_replace(
 969                         '/\r\n|[\x20\x0d\x0a\x09]/',
 970                         ' ',
 971                         $text );
 972         }
 973
 974         /**
 975          * Ensure that any entities and character references are legal
 976          * for XML and XHTML specifically. Any stray bits will be
 977          * &amp;-escaped to result in a valid text fragment.
 978          *
 979          * a. any named char refs must be known in XHTML
 980          * b. any numeric char refs must be legal chars, not invalid or forbidden
 981          * c. use &#x, not &#X
 982          * d. fix or reject non-valid attributes
 983          *
 984          * @param string $text
 985          * @return string
 986          * @private
 987          */
 988         static function normalizeCharReferences( $text ) {
 989                 return preg_replace_callback(
 990                         MW_CHAR_REFS_REGEX,
 991                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 992                         $text );
 993         }
 994         /**
 995          * @param string $matches
 996          * @return string
 997          */
 998         static function normalizeCharReferencesCallback( $matches ) {
 999                 $ret = null;
1000                 if( $matches[1] != '' ) {
1001                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1002                 } elseif( $matches[2] != '' ) {
1003                         $ret = Sanitizer::decCharReference( $matches[2] );
1004                 } elseif( $matches[3] != ''  ) {
1005                         $ret = Sanitizer::hexCharReference( $matches[3] );
1006                 } elseif( $matches[4] != '' ) {
1007                         $ret = Sanitizer::hexCharReference( $matches[4] );
1008                 }
1009                 if( is_null( $ret ) ) {
1010                         return htmlspecialchars( $matches[0] );
1011                 } else {
1012                         return $ret;
1013                 }
1014         }
1015
1016         /**
1017          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1018          * return the named entity reference as is. If the entity is a
1019          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1020          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1021          *
1022          * @param string $name
1023          * @return string
1024          * @static
1025          */
1026         static function normalizeEntity( $name ) {
1027                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1028                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1029                         return "&{$wgHtmlEntityAliases[$name]};";
1030                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1031                         return "&$name;";
1032                 } else {
1033                         return "&amp;$name;";
1034                 }
1035         }
1036
1037         static function decCharReference( $codepoint ) {
1038                 $point = intval( $codepoint );
1039                 if( Sanitizer::validateCodepoint( $point ) ) {
1040                         return sprintf( '&#%d;', $point );
1041                 } else {
1042                         return null;
1043                 }
1044         }
1045
1046         static function hexCharReference( $codepoint ) {
1047                 $point = hexdec( $codepoint );
1048                 if( Sanitizer::validateCodepoint( $point ) ) {
1049                         return sprintf( '&#x%x;', $point );
1050                 } else {
1051                         return null;
1052                 }
1053         }
1054
1055         /**
1056          * Returns true if a given Unicode codepoint is a valid character in XML.
1057          * @param int $codepoint
1058          * @return bool
1059          */
1060         private static function validateCodepoint( $codepoint ) {
1061                 return ($codepoint ==    0x09)
1062                         || ($codepoint ==    0x0a)
1063                         || ($codepoint ==    0x0d)
1064                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1065                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1066                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1067         }
1068
1069         /**
1070          * Decode any character references, numeric or named entities,
1071          * in the text and return a UTF-8 string.
1072          *
1073          * @param string $text
1074          * @return string
1075          * @public
1076          * @static
1077          */
1078         public static function decodeCharReferences( $text ) {
1079                 return preg_replace_callback(
1080                         MW_CHAR_REFS_REGEX,
1081                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1082                         $text );
1083         }
1084
1085         /**
1086          * @param string $matches
1087          * @return string
1088          */
1089         static function decodeCharReferencesCallback( $matches ) {
1090                 if( $matches[1] != '' ) {
1091                         return Sanitizer::decodeEntity( $matches[1] );
1092                 } elseif( $matches[2] != '' ) {
1093                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1094                 } elseif( $matches[3] != ''  ) {
1095                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1096                 } elseif( $matches[4] != '' ) {
1097                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1098                 }
1099                 # Last case should be an ampersand by itself
1100                 return $matches[0];
1101         }
1102
1103         /**
1104          * Return UTF-8 string for a codepoint if that is a valid
1105          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1106          * @param int $codepoint
1107          * @return string
1108          * @private
1109          */
1110         static function decodeChar( $codepoint ) {
1111                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1112                         return codepointToUtf8( $codepoint );
1113                 } else {
1114                         return UTF8_REPLACEMENT;
1115                 }
1116         }
1117
1118         /**
1119          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1120          * return the UTF-8 encoding of that character. Otherwise, returns
1121          * pseudo-entity source (eg &foo;)
1122          *
1123          * @param string $name
1124          * @return string
1125          */
1126         static function decodeEntity( $name ) {
1127                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1128                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1129                         $name = $wgHtmlEntityAliases[$name];
1130                 }
1131                 if( isset( $wgHtmlEntities[$name] ) ) {
1132                         return codepointToUtf8( $wgHtmlEntities[$name] );
1133                 } else {
1134                         return "&$name;";
1135                 }
1136         }
1137
1138         /**
1139          * Fetch the whitelist of acceptable attributes for a given
1140          * element name.
1141          *
1142          * @param string $element
1143          * @return array
1144          */
1145         static function attributeWhitelist( $element ) {
1146                 static $list;
1147                 if( !isset( $list ) ) {
1148                         $list = Sanitizer::setupAttributeWhitelist();
1149                 }
1150                 return isset( $list[$element] )
1151                         ? $list[$element]
1152                         : array();
1153         }
1154
1155         /**
1156          * Foreach array key (an allowed HTML element), return an array
1157          * of allowed attributes
1158          * @return array
1159          */
1160         static function setupAttributeWhitelist() {
1161                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1162                 $block = array_merge( $common, array( 'align' ) );
1163                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1164                 $tablecell = array( 'abbr',
1165                                     'axis',
1166                                     'headers',
1167                                     'scope',
1168                                     'rowspan',
1169                                     'colspan',
1170                                     'nowrap', # deprecated
1171                                     'width',  # deprecated
1172                                     'height', # deprecated
1173                                     'bgcolor' # deprecated
1174                                     );
1175
1176                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1177                 # See: http://www.w3.org/TR/html4/
1178                 $whitelist = array (
1179                         # 7.5.4
1180                         'div'        => $block,
1181                         'center'     => $common, # deprecated
1182                         'span'       => $block, # ??
1183
1184                         # 7.5.5
1185                         'h1'         => $block,
1186                         'h2'         => $block,
1187                         'h3'         => $block,
1188                         'h4'         => $block,
1189                         'h5'         => $block,
1190                         'h6'         => $block,
1191
1192                         # 7.5.6
1193                         # address
1194
1195                         # 8.2.4
1196                         # bdo
1197
1198                         # 9.2.1
1199                         'em'         => $common,
1200                         'strong'     => $common,
1201                         'cite'       => $common,
1202                         # dfn
1203                         'code'       => $common,
1204                         # samp
1205                         # kbd
1206                         'var'        => $common,
1207                         # abbr
1208                         # acronym
1209
1210                         # 9.2.2
1211                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1212                         # q
1213
1214                         # 9.2.3
1215                         'sub'        => $common,
1216                         'sup'        => $common,
1217
1218                         # 9.3.1
1219                         'p'          => $block,
1220
1221                         # 9.3.2
1222                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1223
1224                         # 9.3.4
1225                         'pre'        => array_merge( $common, array( 'width' ) ),
1226
1227                         # 9.4
1228                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1229                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1230
1231                         # 10.2
1232                         'ul'         => array_merge( $common, array( 'type' ) ),
1233                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1234                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1235
1236                         # 10.3
1237                         'dl'         => $common,
1238                         'dd'         => $common,
1239                         'dt'         => $common,
1240
1241                         # 11.2.1
1242                         'table'      => array_merge( $common,
1243                                                                 array( 'summary', 'width', 'border', 'frame',
1244                                                                                 'rules', 'cellspacing', 'cellpadding',
1245                                                                                 'align', 'bgcolor',
1246                                                                 ) ),
1247
1248                         # 11.2.2
1249                         'caption'    => array_merge( $common, array( 'align' ) ),
1250
1251                         # 11.2.3
1252                         'thead'      => array_merge( $common, $tablealign ),
1253                         'tfoot'      => array_merge( $common, $tablealign ),
1254                         'tbody'      => array_merge( $common, $tablealign ),
1255
1256                         # 11.2.4
1257                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1258                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1259
1260                         # 11.2.5
1261                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1262
1263                         # 11.2.6
1264                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1265                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1266
1267                         # 13.2
1268                         # Not usually allowed, but may be used for extension-style hooks
1269                         # such as <math> when it is rasterized
1270                         'img'        => array_merge( $common, array( 'alt' ) ),
1271
1272                         # 15.2.1
1273                         'tt'         => $common,
1274                         'b'          => $common,
1275                         'i'          => $common,
1276                         'big'        => $common,
1277                         'small'      => $common,
1278                         'strike'     => $common,
1279                         's'          => $common,
1280                         'u'          => $common,
1281
1282                         # 15.2.2
1283                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1284                         # basefont
1285
1286                         # 15.3
1287                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1288
1289                         # XHTML Ruby annotation text module, simple ruby only.
1290                         # http://www.w3c.org/TR/ruby/
1291                         'ruby'       => $common,
1292                         # rbc
1293                         # rtc
1294                         'rb'         => $common,
1295                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1296                         'rp'         => $common,
1297
1298                         # MathML root element, where used for extensions
1299                         # 'title' may not be 100% valid here; it's XHTML
1300                         # http://www.w3.org/TR/REC-MathML/
1301                         'math'       => array( 'class', 'style', 'id', 'title' ),
1302                         );
1303                 return $whitelist;
1304         }
1305
1306         /**
1307          * Take a fragment of (potentially invalid) HTML and return
1308          * a version with any tags removed, encoded as plain text.
1309          *
1310          * Warning: this return value must be further escaped for literal
1311          * inclusion in HTML output as of 1.10!
1312          *
1313          * @param string $text HTML fragment
1314          * @return string
1315          */
1316         static function stripAllTags( $text ) {
1317                 # Actual <tags>
1318                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1319
1320                 # Normalize &entities and whitespace
1321                 $text = self::decodeCharReferences( $text );
1322                 $text = self::normalizeWhitespace( $text );
1323
1324                 return $text;
1325         }
1326
1327         /**
1328          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1329          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1330          * PHP 5.1 doesn't.
1331          *
1332          * Use for passing XHTML fragments to PHP's XML parsing functions
1333          *
1334          * @return string
1335          * @static
1336          */
1337         static function hackDocType() {
1338                 global $wgHtmlEntities;
1339                 $out = "<!DOCTYPE html [\n";
1340                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1341                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1342                 }
1343                 $out .= "]>\n";
1344                 return $out;
1345         }
1346
1347         static function cleanUrl( $url ) {
1348                 # Normalize any HTML entities in input. They will be
1349                 # re-escaped by makeExternalLink().
1350                 $url = Sanitizer::decodeCharReferences( $url );
1351
1352                 # Escape any control characters introduced by the above step
1353                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1354
1355                 # Validate hostname portion
1356                 $matches = array();
1357                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1358                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1359
1360                         // Characters that will be ignored in IDNs.
1361                         // http://tools.ietf.org/html/3454#section-3.1
1362                         // Strip them before further processing so blacklists and such work.
1363                         $strip = "/
1364                                 \\s|          # general whitespace
1365                                 \xc2\xad|     # 00ad SOFT HYPHEN
1366                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1367                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1368                                 \xe2\x81\xa0| # 2060 WORD JOINER
1369                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1370                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1371                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1372                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1373                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1374                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1375                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1376                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1377                                 /xuD";
1378
1379                         $host = preg_replace( $strip, '', $host );
1380
1381                         // @fixme: validate hostnames here
1382
1383                         return $protocol . $host . $rest;
1384                 } else {
1385                         return $url;
1386                 }
1387         }
1388
1389 }