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)((?:xml:|xmlns:)?$attrib+)
  47           ($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            )?(?=$space|\$)/sx" );
  58
  59 /**
  60  * Regular expression to match URIs that could trigger script execution
  61  */
  62 define( 'MW_EVIL_URI_PATTERN', '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i' );
  63
  64 /**
  65  * Regular expression to match namespace attributes
  66  */
  67 define( 'MW_XMLNS_ATTRIBUTE_PATTRN', "/^xmlns:$attrib+$/" );
  68
  69 /**
  70  * List of all named character entities defined in HTML 4.01
  71  * http://www.w3.org/TR/html4/sgml/entities.html
  72  * @private
  73  */
  74 global $wgHtmlEntities;
  75 $wgHtmlEntities = array(
  76         'Aacute'   => 193,
  77         'aacute'   => 225,
  78         'Acirc'    => 194,
  79         'acirc'    => 226,
  80         'acute'    => 180,
  81         'AElig'    => 198,
  82         'aelig'    => 230,
  83         'Agrave'   => 192,
  84         'agrave'   => 224,
  85         'alefsym'  => 8501,
  86         'Alpha'    => 913,
  87         'alpha'    => 945,
  88         'amp'      => 38,
  89         'and'      => 8743,
  90         'ang'      => 8736,
  91         'Aring'    => 197,
  92         'aring'    => 229,
  93         'asymp'    => 8776,
  94         'Atilde'   => 195,
  95         'atilde'   => 227,
  96         'Auml'     => 196,
  97         'auml'     => 228,
  98         'bdquo'    => 8222,
  99         'Beta'     => 914,
 100         'beta'     => 946,
 101         'brvbar'   => 166,
 102         'bull'     => 8226,
 103         'cap'      => 8745,
 104         'Ccedil'   => 199,
 105         'ccedil'   => 231,
 106         'cedil'    => 184,
 107         'cent'     => 162,
 108         'Chi'      => 935,
 109         'chi'      => 967,
 110         'circ'     => 710,
 111         'clubs'    => 9827,
 112         'cong'     => 8773,
 113         'copy'     => 169,
 114         'crarr'    => 8629,
 115         'cup'      => 8746,
 116         'curren'   => 164,
 117         'dagger'   => 8224,
 118         'Dagger'   => 8225,
 119         'darr'     => 8595,
 120         'dArr'     => 8659,
 121         'deg'      => 176,
 122         'Delta'    => 916,
 123         'delta'    => 948,
 124         'diams'    => 9830,
 125         'divide'   => 247,
 126         'Eacute'   => 201,
 127         'eacute'   => 233,
 128         'Ecirc'    => 202,
 129         'ecirc'    => 234,
 130         'Egrave'   => 200,
 131         'egrave'   => 232,
 132         'empty'    => 8709,
 133         'emsp'     => 8195,
 134         'ensp'     => 8194,
 135         'Epsilon'  => 917,
 136         'epsilon'  => 949,
 137         'equiv'    => 8801,
 138         'Eta'      => 919,
 139         'eta'      => 951,
 140         'ETH'      => 208,
 141         'eth'      => 240,
 142         'Euml'     => 203,
 143         'euml'     => 235,
 144         'euro'     => 8364,
 145         'exist'    => 8707,
 146         'fnof'     => 402,
 147         'forall'   => 8704,
 148         'frac12'   => 189,
 149         'frac14'   => 188,
 150         'frac34'   => 190,
 151         'frasl'    => 8260,
 152         'Gamma'    => 915,
 153         'gamma'    => 947,
 154         'ge'       => 8805,
 155         'gt'       => 62,
 156         'harr'     => 8596,
 157         'hArr'     => 8660,
 158         'hearts'   => 9829,
 159         'hellip'   => 8230,
 160         'Iacute'   => 205,
 161         'iacute'   => 237,
 162         'Icirc'    => 206,
 163         'icirc'    => 238,
 164         'iexcl'    => 161,
 165         'Igrave'   => 204,
 166         'igrave'   => 236,
 167         'image'    => 8465,
 168         'infin'    => 8734,
 169         'int'      => 8747,
 170         'Iota'     => 921,
 171         'iota'     => 953,
 172         'iquest'   => 191,
 173         'isin'     => 8712,
 174         'Iuml'     => 207,
 175         'iuml'     => 239,
 176         'Kappa'    => 922,
 177         'kappa'    => 954,
 178         'Lambda'   => 923,
 179         'lambda'   => 955,
 180         'lang'     => 9001,
 181         'laquo'    => 171,
 182         'larr'     => 8592,
 183         'lArr'     => 8656,
 184         'lceil'    => 8968,
 185         'ldquo'    => 8220,
 186         'le'       => 8804,
 187         'lfloor'   => 8970,
 188         'lowast'   => 8727,
 189         'loz'      => 9674,
 190         'lrm'      => 8206,
 191         'lsaquo'   => 8249,
 192         'lsquo'    => 8216,
 193         'lt'       => 60,
 194         'macr'     => 175,
 195         'mdash'    => 8212,
 196         'micro'    => 181,
 197         'middot'   => 183,
 198         'minus'    => 8722,
 199         'Mu'       => 924,
 200         'mu'       => 956,
 201         'nabla'    => 8711,
 202         'nbsp'     => 160,
 203         'ndash'    => 8211,
 204         'ne'       => 8800,
 205         'ni'       => 8715,
 206         'not'      => 172,
 207         'notin'    => 8713,
 208         'nsub'     => 8836,
 209         'Ntilde'   => 209,
 210         'ntilde'   => 241,
 211         'Nu'       => 925,
 212         'nu'       => 957,
 213         'Oacute'   => 211,
 214         'oacute'   => 243,
 215         'Ocirc'    => 212,
 216         'ocirc'    => 244,
 217         'OElig'    => 338,
 218         'oelig'    => 339,
 219         'Ograve'   => 210,
 220         'ograve'   => 242,
 221         'oline'    => 8254,
 222         'Omega'    => 937,
 223         'omega'    => 969,
 224         'Omicron'  => 927,
 225         'omicron'  => 959,
 226         'oplus'    => 8853,
 227         'or'       => 8744,
 228         'ordf'     => 170,
 229         'ordm'     => 186,
 230         'Oslash'   => 216,
 231         'oslash'   => 248,
 232         'Otilde'   => 213,
 233         'otilde'   => 245,
 234         'otimes'   => 8855,
 235         'Ouml'     => 214,
 236         'ouml'     => 246,
 237         'para'     => 182,
 238         'part'     => 8706,
 239         'permil'   => 8240,
 240         'perp'     => 8869,
 241         'Phi'      => 934,
 242         'phi'      => 966,
 243         'Pi'       => 928,
 244         'pi'       => 960,
 245         'piv'      => 982,
 246         'plusmn'   => 177,
 247         'pound'    => 163,
 248         'prime'    => 8242,
 249         'Prime'    => 8243,
 250         'prod'     => 8719,
 251         'prop'     => 8733,
 252         'Psi'      => 936,
 253         'psi'      => 968,
 254         'quot'     => 34,
 255         'radic'    => 8730,
 256         'rang'     => 9002,
 257         'raquo'    => 187,
 258         'rarr'     => 8594,
 259         'rArr'     => 8658,
 260         'rceil'    => 8969,
 261         'rdquo'    => 8221,
 262         'real'     => 8476,
 263         'reg'      => 174,
 264         'rfloor'   => 8971,
 265         'Rho'      => 929,
 266         'rho'      => 961,
 267         'rlm'      => 8207,
 268         'rsaquo'   => 8250,
 269         'rsquo'    => 8217,
 270         'sbquo'    => 8218,
 271         'Scaron'   => 352,
 272         'scaron'   => 353,
 273         'sdot'     => 8901,
 274         'sect'     => 167,
 275         'shy'      => 173,
 276         'Sigma'    => 931,
 277         'sigma'    => 963,
 278         'sigmaf'   => 962,
 279         'sim'      => 8764,
 280         'spades'   => 9824,
 281         'sub'      => 8834,
 282         'sube'     => 8838,
 283         'sum'      => 8721,
 284         'sup'      => 8835,
 285         'sup1'     => 185,
 286         'sup2'     => 178,
 287         'sup3'     => 179,
 288         'supe'     => 8839,
 289         'szlig'    => 223,
 290         'Tau'      => 932,
 291         'tau'      => 964,
 292         'there4'   => 8756,
 293         'Theta'    => 920,
 294         'theta'    => 952,
 295         'thetasym' => 977,
 296         'thinsp'   => 8201,
 297         'THORN'    => 222,
 298         'thorn'    => 254,
 299         'tilde'    => 732,
 300         'times'    => 215,
 301         'trade'    => 8482,
 302         'Uacute'   => 218,
 303         'uacute'   => 250,
 304         'uarr'     => 8593,
 305         'uArr'     => 8657,
 306         'Ucirc'    => 219,
 307         'ucirc'    => 251,
 308         'Ugrave'   => 217,
 309         'ugrave'   => 249,
 310         'uml'      => 168,
 311         'upsih'    => 978,
 312         'Upsilon'  => 933,
 313         'upsilon'  => 965,
 314         'Uuml'     => 220,
 315         'uuml'     => 252,
 316         'weierp'   => 8472,
 317         'Xi'       => 926,
 318         'xi'       => 958,
 319         'Yacute'   => 221,
 320         'yacute'   => 253,
 321         'yen'      => 165,
 322         'Yuml'     => 376,
 323         'yuml'     => 255,
 324         'Zeta'     => 918,
 325         'zeta'     => 950,
 326         'zwj'      => 8205,
 327         'zwnj'     => 8204 );
 328
 329 /**
 330  * Character entity aliases accepted by MediaWiki
 331  */
 332 global $wgHtmlEntityAliases;
 333 $wgHtmlEntityAliases = array(
 334         'רלמ' => 'rlm',
 335         'رلم' => 'rlm',
 336 );
 337
 338
 339 /**
 340  * XHTML sanitizer for MediaWiki
 341  * @ingroup Parser
 342  */
 343 class Sanitizer {
 344         /**
 345          * Cleans up HTML, removes dangerous tags and attributes, and
 346          * removes HTML comments
 347          * @private
 348          * @param $text String
 349          * @param $processCallback Callback to do any variable or parameter replacements in HTML attribute values
 350          * @param $args Array for the processing callback
 351          * @param $extratags Array for any extra tags to include
 352          * @param $removetags Array for any tags (default or extra) to exclude
 353          * @return string
 354          */
 355         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array(), $removetags = array() ) {
 356                 global $wgUseTidy;
 357
 358                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 359                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 360
 361                 wfProfileIn( __METHOD__ );
 362
 363                 if ( !$staticInitialised ) {
 364
 365                         $htmlpairsStatic = array( # Tags that must be closed
 366                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 367                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 368                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 369                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 370                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u', 'abbr'
 371                         );
 372                         $htmlsingle = array(
 373                                 'br', 'hr', 'li', 'dt', 'dd'
 374                         );
 375                         $htmlsingleonly = array( # Elements that cannot have close tags
 376                                 'br', 'hr'
 377                         );
 378                         $htmlnest = array( # Tags that can be nested--??
 379                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 380                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 381                         );
 382                         $tabletags = array( # Can only appear inside table, we will close them
 383                                 'td', 'th', 'tr',
 384                         );
 385                         $htmllist = array( # Tags used by list
 386                                 'ul','ol',
 387                         );
 388                         $listtags = array( # Tags that can appear in a list
 389                                 'li',
 390                         );
 391
 392                         $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
 393                         $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
 394
 395                         # Convert them all to hashtables for faster lookup
 396                         $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 397                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
 398                         foreach ( $vars as $var ) {
 399                                 $$var = array_flip( $$var );
 400                         }
 401                         $staticInitialised = true;
 402                 }
 403                 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
 404                 $extratags = array_flip( $extratags );
 405                 $removetags = array_flip( $removetags );
 406                 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
 407                 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ) , $removetags );
 408
 409                 # Remove HTML comments
 410                 $text = Sanitizer::removeHTMLcomments( $text );
 411                 $bits = explode( '<', $text );
 412                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 413                 if ( !$wgUseTidy ) {
 414                         $tagstack = $tablestack = array();
 415                         foreach ( $bits as $x ) {
 416                                 $regs = array();
 417                                 # $slash: Does the current element start with a '/'?
 418                                 # $t: Current element name
 419                                 # $params: String between element name and >
 420                                 # $brace: Ending '>' or '/>'
 421                                 # $rest: Everything until the next element of $bits
 422                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 423                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 424                                 } else {
 425                                         $slash = $t = $params = $brace = $rest = null;
 426                                 }
 427
 428                                 $badtag = false;
 429                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 430                                         # Check our stack
 431                                         if ( $slash && isset( $htmlsingleonly[$t] ) ) {
 432                                                 $badtag = true;
 433                                         } elseif ( $slash ) {
 434                                                 # Closing a tag... is it the one we just opened?
 435                                                 $ot = @array_pop( $tagstack );
 436                                                 if ( $ot != $t ) {
 437                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 438                                                                 # Pop all elements with an optional close tag
 439                                                                 # and see if we find a match below them
 440                                                                 $optstack = array();
 441                                                                 array_push( $optstack, $ot );
 442                                                                 $ot = @array_pop( $tagstack );
 443                                                                 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
 444                                                                         array_push( $optstack, $ot );
 445                                                                         $ot = @array_pop( $tagstack );
 446                                                                 }
 447                                                                 if ( $t != $ot ) {
 448                                                                         # No match. Push the optional elements back again
 449                                                                         $badtag = true;
 450                                                                         while ( $ot = @array_pop( $optstack ) ) {
 451                                                                                 array_push( $tagstack, $ot );
 452                                                                         }
 453                                                                 }
 454                                                         } else {
 455                                                                 @array_push( $tagstack, $ot );
 456                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 457                                                                 if ( !isset( $htmllist[$ot] ) || !isset( $listtags[$t] ) ) {
 458                                                                         $badtag = true;
 459                                                                 }
 460                                                         }
 461                                                 } else {
 462                                                         if ( $t == 'table' ) {
 463                                                                 $tagstack = array_pop( $tablestack );
 464                                                         }
 465                                                 }
 466                                                 $newparams = '';
 467                                         } else {
 468                                                 # Keep track for later
 469                                                 if ( isset( $tabletags[$t] ) &&
 470                                                 !in_array( 'table', $tagstack ) ) {
 471                                                         $badtag = true;
 472                                                 } elseif ( in_array( $t, $tagstack ) &&
 473                                                 !isset( $htmlnest [$t ] ) ) {
 474                                                         $badtag = true;
 475                                                 # Is it a self closed htmlpair ? (bug 5487)
 476                                                 } elseif ( $brace == '/>' &&
 477                                                 isset( $htmlpairs[$t] ) ) {
 478                                                         $badtag = true;
 479                                                 } elseif ( isset( $htmlsingleonly[$t] ) ) {
 480                                                         # Hack to force empty tag for uncloseable elements
 481                                                         $brace = '/>';
 482                                                 } elseif ( isset( $htmlsingle[$t] ) ) {
 483                                                         # Hack to not close $htmlsingle tags
 484                                                         $brace = null;
 485                                                 } elseif ( isset( $tabletags[$t] )
 486                                                 && in_array( $t, $tagstack ) ) {
 487                                                         // New table tag but forgot to close the previous one
 488                                                         $text .= "</$t>";
 489                                                 } else {
 490                                                         if ( $t == 'table' ) {
 491                                                                 array_push( $tablestack, $tagstack );
 492                                                                 $tagstack = array();
 493                                                         }
 494                                                         array_push( $tagstack, $t );
 495                                                 }
 496
 497                                                 # Replace any variables or template parameters with
 498                                                 # plaintext results.
 499                                                 if( is_callable( $processCallback ) ) {
 500                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 501                                                 }
 502
 503                                                 # Strip non-approved attributes from the tag
 504                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 505                                         }
 506                                         if ( !$badtag ) {
 507                                                 $rest = str_replace( '>', '&gt;', $rest );
 508                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 509                                                 $text .= "<$slash$t$newparams$close>$rest";
 510                                                 continue;
 511                                         }
 512                                 }
 513                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 514                         }
 515                         # Close off any remaining tags
 516                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 517                                 $text .= "</$t>\n";
 518                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 519                         }
 520                 } else {
 521                         # this might be possible using tidy itself
 522                         foreach ( $bits as $x ) {
 523                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 524                                 $x, $regs );
 525                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 526                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 527                                         if( is_callable( $processCallback ) ) {
 528                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 529                                         }
 530                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 531                                         $rest = str_replace( '>', '&gt;', $rest );
 532                                         $text .= "<$slash$t$newparams$brace$rest";
 533                                 } else {
 534                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 535                                 }
 536                         }
 537                 }
 538                 wfProfileOut( __METHOD__ );
 539                 return $text;
 540         }
 541
 542         /**
 543          * Remove '<!--', '-->', and everything between.
 544          * To avoid leaving blank lines, when a comment is both preceded
 545          * and followed by a newline (ignoring spaces), trim leading and
 546          * trailing spaces and one of the newlines.
 547          *
 548          * @private
 549          * @param $text String
 550          * @return string
 551          */
 552         static function removeHTMLcomments( $text ) {
 553                 wfProfileIn( __METHOD__ );
 554                 while (($start = strpos($text, '<!--')) !== false) {
 555                         $end = strpos($text, '-->', $start + 4);
 556                         if ($end === false) {
 557                                 # Unterminated comment; bail out
 558                                 break;
 559                         }
 560
 561                         $end += 3;
 562
 563                         # Trim space and newline if the comment is both
 564                         # preceded and followed by a newline
 565                         $spaceStart = max($start - 1, 0);
 566                         $spaceLen = $end - $spaceStart;
 567                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 568                                 $spaceStart--;
 569                                 $spaceLen++;
 570                         }
 571                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 572                                 $spaceLen++;
 573                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 574                                 # Remove the comment, leading and trailing
 575                                 # spaces, and leave only one newline.
 576                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 577                         }
 578                         else {
 579                                 # Remove just the comment.
 580                                 $text = substr_replace($text, '', $start, $end - $start);
 581                         }
 582                 }
 583                 wfProfileOut( __METHOD__ );
 584                 return $text;
 585         }
 586
 587         /**
 588          * Take an array of attribute names and values and normalize or discard
 589          * illegal values for the given element type.
 590          *
 591          * - Discards attributes not on a whitelist for the given element
 592          * - Unsafe style attributes are discarded
 593          * - Invalid id attributes are reencoded
 594          *
 595          * @param $attribs Array
 596          * @param $element String
 597          * @return Array
 598          *
 599          * @todo Check for legal values where the DTD limits things.
 600          * @todo Check for unique id attribute :P
 601          */
 602         static function validateTagAttributes( $attribs, $element ) {
 603                 return Sanitizer::validateAttributes( $attribs,
 604                         Sanitizer::attributeWhitelist( $element ) );
 605         }
 606
 607         /**
 608          * Take an array of attribute names and values and normalize or discard
 609          * illegal values for the given whitelist.
 610          *
 611          * - Discards attributes not the given whitelist
 612          * - Unsafe style attributes are discarded
 613          * - Invalid id attributes are reencoded
 614          *
 615          * @param $attribs Array
 616          * @param $whitelist Array: list of allowed attribute names
 617          * @return Array
 618          *
 619          * @todo Check for legal values where the DTD limits things.
 620          * @todo Check for unique id attribute :P
 621          */
 622         static function validateAttributes( $attribs, $whitelist ) {
 623                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes;
 624
 625                 $whitelist = array_flip( $whitelist );
 626                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 627
 628                 $out = array();
 629                 foreach( $attribs as $attribute => $value ) {
 630                         #allow XML namespace declaration if RDFa is enabled
 631                         if ( $wgAllowRdfaAttributes && preg_match( MW_XMLNS_ATTRIBUTE_PATTRN, $attribute ) ) {
 632                                 if ( !preg_match( MW_EVIL_URI_PATTERN, $value ) ) {
 633                                         $out[$attribute] = $value;
 634                                 }
 635
 636                                 continue;
 637                         }
 638
 639                         if( !isset( $whitelist[$attribute] ) ) {
 640                                 continue;
 641                         }
 642
 643                         # Strip javascript "expression" from stylesheets.
 644                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 645                         if( $attribute == 'style' ) {
 646                                 $value = Sanitizer::checkCss( $value );
 647                                 if( $value === false ) {
 648                                         # haxx0r
 649                                         continue;
 650                                 }
 651                         }
 652
 653                         if ( $attribute === 'id' ) {
 654                                 $value = Sanitizer::escapeId( $value, 'noninitial' );
 655                         }
 656
 657                         //RDFa and microdata properties allow URLs, URIs and/or CURIs. check them for sanity
 658                         if ( $attribute === 'rel' || $attribute === 'rev' ||
 659                                 $attribute === 'about' || $attribute === 'property' || $attribute === 'resource' || #RDFa
 660                                 $attribute === 'datatype' || $attribute === 'typeof' ||                             #RDFa
 661                                 $attribute === 'itemid' || $attribute === 'itemprop' || $attribute === 'itemref' || #HTML5 microdata
 662                                 $attribute === 'itemscope' || $attribute === 'itemtype' ) {                         #HTML5 microdata
 663
 664                                 //Paranoia. Allow "simple" values but suppress javascript
 665                                 if ( preg_match( MW_EVIL_URI_PATTERN, $value ) ) {
 666                                         continue;
 667                                 }
 668                         }
 669
 670                         # NOTE: even though elements using href/src are not allowed directly, supply
 671                         #       validation code that can be used by tag hook handlers, etc
 672                         if ( $attribute === 'href' || $attribute === 'src' ) {
 673                                 if ( !preg_match( $hrefExp, $value ) ) {
 674                                         continue; //drop any href or src attributes not using an allowed protocol.
 675                                                   //NOTE: this also drops all relative URLs
 676                                 }
 677                         }
 678
 679                         // If this attribute was previously set, override it.
 680                         // Output should only have one attribute of each name.
 681                         $out[$attribute] = $value;
 682                 }
 683
 684                 if ( $wgAllowMicrodataAttributes ) {
 685                         # There are some complicated validity constraints we need to
 686                         # enforce here.  First of all, we don't want to allow non-standard
 687                         # itemtypes.
 688                         $allowedTypes = array(
 689                                 'http://microformats.org/profile/hcard',
 690                                 'http://microformats.org/profile/hcalendar#vevent',
 691                                 'http://n.whatwg.org/work',
 692                         );
 693                         if ( isset( $out['itemtype'] ) && !in_array( $out['itemtype'],
 694                         $allowedTypes ) ) {
 695                                 # Kill everything
 696                                 unset( $out['itemscope'] );
 697                         }
 698                         # itemtype, itemid, itemref don't make sense without itemscope
 699                         if ( !array_key_exists( 'itemscope', $out ) ) {
 700                                 unset( $out['itemtype'] );
 701                                 unset( $out['itemid'] );
 702                                 unset( $out['itemref'] );
 703                         }
 704                         # TODO: Strip itemprop if we aren't descendants of an itemscope.
 705                 }
 706                 return $out;
 707         }
 708
 709         /**
 710          * Merge two sets of HTML attributes.  Conflicting items in the second set
 711          * will override those in the first, except for 'class' attributes which
 712          * will be combined (if they're both strings).
 713          *
 714          * @todo implement merging for other attributes such as style
 715          * @param $a Array
 716          * @param $b Array
 717          * @return array
 718          */
 719         static function mergeAttributes( $a, $b ) {
 720                 $out = array_merge( $a, $b );
 721                 if( isset( $a['class'] ) && isset( $b['class'] )
 722                 && is_string( $a['class'] ) && is_string( $b['class'] )
 723                 && $a['class'] !== $b['class'] ) {
 724                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 725                                 -1, PREG_SPLIT_NO_EMPTY );
 726                         $out['class'] = implode( ' ', array_unique( $classes ) );
 727                 }
 728                 return $out;
 729         }
 730
 731         /**
 732          * Pick apart some CSS and check it for forbidden or unsafe structures.
 733          * Returns a sanitized string, or false if it was just too evil.
 734          *
 735          * Currently URL references, 'expression', 'tps' are forbidden.
 736          *
 737          * @param $value String
 738          * @return Mixed
 739          */
 740         static function checkCss( $value ) {
 741                 $stripped = Sanitizer::decodeCharReferences( $value );
 742
 743                 // Remove any comments; IE gets token splitting wrong
 744                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 745
 746                 $value = $stripped;
 747
 748                 // ... and continue checks
 749                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 750                         'codepointToUtf8(hexdec("$1"))', $stripped );
 751                 $stripped = str_replace( '\\', '', $stripped );
 752                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 753                                 $stripped ) ) {
 754                         # haxx0r
 755                         return false;
 756                 }
 757
 758                 return $value;
 759         }
 760
 761         /**
 762          * Take a tag soup fragment listing an HTML element's attributes
 763          * and normalize it to well-formed XML, discarding unwanted attributes.
 764          * Output is safe for further wikitext processing, with escaping of
 765          * values that could trigger problems.
 766          *
 767          * - Normalizes attribute names to lowercase
 768          * - Discards attributes not on a whitelist for the given element
 769          * - Turns broken or invalid entities into plaintext
 770          * - Double-quotes all attribute values
 771          * - Attributes without values are given the name as attribute
 772          * - Double attributes are discarded
 773          * - Unsafe style attributes are discarded
 774          * - Prepends space if there are attributes.
 775          *
 776          * @param $text String
 777          * @param $element String
 778          * @return String
 779          */
 780         static function fixTagAttributes( $text, $element ) {
 781                 if( trim( $text ) == '' ) {
 782                         return '';
 783                 }
 784
 785                 $stripped = Sanitizer::validateTagAttributes(
 786                         Sanitizer::decodeTagAttributes( $text ), $element );
 787
 788                 $attribs = array();
 789                 foreach( $stripped as $attribute => $value ) {
 790                         $encAttribute = htmlspecialchars( $attribute );
 791                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 792
 793                         $attribs[] = "$encAttribute=\"$encValue\"";
 794                 }
 795                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 796         }
 797
 798         /**
 799          * Encode an attribute value for HTML output.
 800          * @param $text String
 801          * @return HTML-encoded text fragment
 802          */
 803         static function encodeAttribute( $text ) {
 804                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 805
 806                 // Whitespace is normalized during attribute decoding,
 807                 // so if we've been passed non-spaces we must encode them
 808                 // ahead of time or they won't be preserved.
 809                 $encValue = strtr( $encValue, array(
 810                         "\n" => '&#10;',
 811                         "\r" => '&#13;',
 812                         "\t" => '&#9;',
 813                 ) );
 814
 815                 return $encValue;
 816         }
 817
 818         /**
 819          * Encode an attribute value for HTML tags, with extra armoring
 820          * against further wiki processing.
 821          * @param $text String
 822          * @return HTML-encoded text fragment
 823          */
 824         static function safeEncodeAttribute( $text ) {
 825                 $encValue = Sanitizer::encodeAttribute( $text );
 826
 827                 # Templates and links may be expanded in later parsing,
 828                 # creating invalid or dangerous output. Suppress this.
 829                 $encValue = strtr( $encValue, array(
 830                         '<'    => '&lt;',   // This should never happen,
 831                         '>'    => '&gt;',   // we've received invalid input
 832                         '"'    => '&quot;', // which should have been escaped.
 833                         '{'    => '&#123;',
 834                         '['    => '&#91;',
 835                         "''"   => '&#39;&#39;',
 836                         'ISBN' => '&#73;SBN',
 837                         'RFC'  => '&#82;FC',
 838                         'PMID' => '&#80;MID',
 839                         '|'    => '&#124;',
 840                         '__'   => '&#95;_',
 841                 ) );
 842
 843                 # Stupid hack
 844                 $encValue = preg_replace_callback(
 845                         '/(' . wfUrlProtocols() . ')/',
 846                         array( 'Sanitizer', 'armorLinksCallback' ),
 847                         $encValue );
 848                 return $encValue;
 849         }
 850
 851         /**
 852          * Given a value, escape it so that it can be used in an id attribute and
 853          * return it.  This will use HTML5 validation if $wgExperimentalHtmlIds is
 854          * true, allowing anything but ASCII whitespace.  Otherwise it will use
 855          * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
 856          * escaped with lots of dots.
 857          *
 858          * To ensure we don't have to bother escaping anything, we also strip ', ",
 859          * & even if $wgExperimentalIds is true.  TODO: Is this the best tactic?
 860          * We also strip # because it upsets IE6.
 861          *
 862          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 863          *                                                          in the id and
 864          *                                                          name attributes
 865          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 866          * @see http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#the-id-attribute
 867          *   HTML5 definition of id attribute
 868          *
 869          * @param $id String: id to escape
 870          * @param $options Mixed: string or array of strings (default is array()):
 871          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 872          *       so don't pay attention if the first character isn't valid at the
 873          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
 874          *       false.
 875          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
 876          *       if $wgExperimentalHtmlIds is used, so we can generate extra
 877          *       anchors and links won't break.
 878          * @return String
 879          */
 880         static function escapeId( $id, $options = array() ) {
 881                 global $wgHtml5, $wgExperimentalHtmlIds;
 882                 $options = (array)$options;
 883
 884                 if ( $wgHtml5 && $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
 885                         $id = Sanitizer::decodeCharReferences( $id );
 886                         $id = preg_replace( '/[ \t\n\r\f_\'"&#]+/', '_', $id );
 887                         $id = trim( $id, '_' );
 888                         if ( $id === '' ) {
 889                                 # Must have been all whitespace to start with.
 890                                 return '_';
 891                         } else {
 892                                 return $id;
 893                         }
 894                 }
 895
 896                 # HTML4-style escaping
 897                 static $replace = array(
 898                         '%3A' => ':',
 899                         '%' => '.'
 900                 );
 901
 902                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 903                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 904
 905                 if ( !preg_match( '/^[a-zA-Z]/', $id )
 906                 && !in_array( 'noninitial', $options ) )  {
 907                         // Initial character must be a letter!
 908                         $id = "x$id";
 909                 }
 910                 return $id;
 911         }
 912
 913         /**
 914          * Given a value, escape it so that it can be used as a CSS class and
 915          * return it.
 916          *
 917          * @todo For extra validity, input should be validated UTF-8.
 918          *
 919          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 920          *
 921          * @param $class String
 922          * @return String
 923          */
 924         static function escapeClass( $class ) {
 925                 // Convert ugly stuff to underscores and kill underscores in ugly places
 926                 return rtrim(preg_replace(
 927                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 928                         '_',
 929                         $class ), '_');
 930         }
 931
 932         /**
 933          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 934          * This allows (generally harmless) entities like &nbsp; to survive.
 935          *
 936          * @param $html String to escape
 937          * @return String: escaped input
 938          */
 939         static function escapeHtmlAllowEntities( $html ) {
 940                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 941                 # hurt.
 942                 $html = htmlspecialchars( $html, ENT_QUOTES );
 943                 $html = str_replace( '&amp;', '&', $html );
 944                 $html = Sanitizer::normalizeCharReferences( $html );
 945                 return $html;
 946         }
 947
 948         /**
 949          * Regex replace callback for armoring links against further processing.
 950          * @param $matches Array
 951          * @return string
 952          */
 953         private static function armorLinksCallback( $matches ) {
 954                 return str_replace( ':', '&#58;', $matches[1] );
 955         }
 956
 957         /**
 958          * Return an associative array of attribute names and values from
 959          * a partial tag string. Attribute names are forces to lowercase,
 960          * character references are decoded to UTF-8 text.
 961          *
 962          * @param $text String
 963          * @return Array
 964          */
 965         public static function decodeTagAttributes( $text ) {
 966                 if( trim( $text ) == '' ) {
 967                         return array();
 968                 }
 969
 970                 $attribs = array();
 971                 $pairs = array();
 972                 if( !preg_match_all(
 973                         MW_ATTRIBS_REGEX,
 974                         $text,
 975                         $pairs,
 976                         PREG_SET_ORDER ) ) {
 977                         return $attribs;
 978                 }
 979
 980                 foreach( $pairs as $set ) {
 981                         $attribute = strtolower( $set[1] );
 982                         $value = Sanitizer::getTagAttributeCallback( $set );
 983
 984                         // Normalize whitespace
 985                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 986                         $value = trim( $value );
 987
 988                         // Decode character references
 989                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 990                 }
 991                 return $attribs;
 992         }
 993
 994         /**
 995          * Pick the appropriate attribute value from a match set from the
 996          * MW_ATTRIBS_REGEX matches.
 997          *
 998          * @param $set Array
 999          * @return String
1000          */
1001         private static function getTagAttributeCallback( $set ) {
1002                 if( isset( $set[6] ) ) {
1003                         # Illegal #XXXXXX color with no quotes.
1004                         return $set[6];
1005                 } elseif( isset( $set[5] ) ) {
1006                         # No quotes.
1007                         return $set[5];
1008                 } elseif( isset( $set[4] ) ) {
1009                         # Single-quoted
1010                         return $set[4];
1011                 } elseif( isset( $set[3] ) ) {
1012                         # Double-quoted
1013                         return $set[3];
1014                 } elseif( !isset( $set[2] ) ) {
1015                         # In XHTML, attributes must have a value.
1016                         # For 'reduced' form, return explicitly the attribute name here.
1017                         return $set[1];
1018                 } else {
1019                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1020                 }
1021         }
1022
1023         /**
1024          * Normalize whitespace and character references in an XML source-
1025          * encoded text for an attribute value.
1026          *
1027          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
1028          * but note that we're not returning the value, but are returning
1029          * XML source fragments that will be slapped into output.
1030          *
1031          * @param $text String
1032          * @return String
1033          */
1034         private static function normalizeAttributeValue( $text ) {
1035                 return str_replace( '"', '&quot;',
1036                         self::normalizeWhitespace(
1037                                 Sanitizer::normalizeCharReferences( $text ) ) );
1038         }
1039
1040         private static function normalizeWhitespace( $text ) {
1041                 return preg_replace(
1042                         '/\r\n|[\x20\x0d\x0a\x09]/',
1043                         ' ',
1044                         $text );
1045         }
1046
1047         /**
1048          * Ensure that any entities and character references are legal
1049          * for XML and XHTML specifically. Any stray bits will be
1050          * &amp;-escaped to result in a valid text fragment.
1051          *
1052          * a. any named char refs must be known in XHTML
1053          * b. any numeric char refs must be legal chars, not invalid or forbidden
1054          * c. use &#x, not &#X
1055          * d. fix or reject non-valid attributes
1056          *
1057          * @param $text String
1058          * @return String
1059          * @private
1060          */
1061         static function normalizeCharReferences( $text ) {
1062                 return preg_replace_callback(
1063                         MW_CHAR_REFS_REGEX,
1064                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1065                         $text );
1066         }
1067         /**
1068          * @param $matches String
1069          * @return String
1070          */
1071         static function normalizeCharReferencesCallback( $matches ) {
1072                 $ret = null;
1073                 if( $matches[1] != '' ) {
1074                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1075                 } elseif( $matches[2] != '' ) {
1076                         $ret = Sanitizer::decCharReference( $matches[2] );
1077                 } elseif( $matches[3] != ''  ) {
1078                         $ret = Sanitizer::hexCharReference( $matches[3] );
1079                 } elseif( $matches[4] != '' ) {
1080                         $ret = Sanitizer::hexCharReference( $matches[4] );
1081                 }
1082                 if( is_null( $ret ) ) {
1083                         return htmlspecialchars( $matches[0] );
1084                 } else {
1085                         return $ret;
1086                 }
1087         }
1088
1089         /**
1090          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1091          * return the named entity reference as is. If the entity is a
1092          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1093          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1094          *
1095          * @param $name String
1096          * @return String
1097          */
1098         static function normalizeEntity( $name ) {
1099                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1100                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1101                         return "&{$wgHtmlEntityAliases[$name]};";
1102                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1103                         return "&$name;";
1104                 } else {
1105                         return "&amp;$name;";
1106                 }
1107         }
1108
1109         static function decCharReference( $codepoint ) {
1110                 $point = intval( $codepoint );
1111                 if( Sanitizer::validateCodepoint( $point ) ) {
1112                         return sprintf( '&#%d;', $point );
1113                 } else {
1114                         return null;
1115                 }
1116         }
1117
1118         static function hexCharReference( $codepoint ) {
1119                 $point = hexdec( $codepoint );
1120                 if( Sanitizer::validateCodepoint( $point ) ) {
1121                         return sprintf( '&#x%x;', $point );
1122                 } else {
1123                         return null;
1124                 }
1125         }
1126
1127         /**
1128          * Returns true if a given Unicode codepoint is a valid character in XML.
1129          * @param $codepoint Integer
1130          * @return Boolean
1131          */
1132         private static function validateCodepoint( $codepoint ) {
1133                 return ($codepoint ==    0x09)
1134                         || ($codepoint ==    0x0a)
1135                         || ($codepoint ==    0x0d)
1136                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1137                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1138                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1139         }
1140
1141         /**
1142          * Decode any character references, numeric or named entities,
1143          * in the text and return a UTF-8 string.
1144          *
1145          * @param $text String
1146          * @return String
1147          */
1148         public static function decodeCharReferences( $text ) {
1149                 return preg_replace_callback(
1150                         MW_CHAR_REFS_REGEX,
1151                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1152                         $text );
1153         }
1154
1155         /**
1156          * @param $matches String
1157          * @return String
1158          */
1159         static function decodeCharReferencesCallback( $matches ) {
1160                 if( $matches[1] != '' ) {
1161                         return Sanitizer::decodeEntity( $matches[1] );
1162                 } elseif( $matches[2] != '' ) {
1163                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1164                 } elseif( $matches[3] != ''  ) {
1165                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1166                 } elseif( $matches[4] != '' ) {
1167                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1168                 }
1169                 # Last case should be an ampersand by itself
1170                 return $matches[0];
1171         }
1172
1173         /**
1174          * Return UTF-8 string for a codepoint if that is a valid
1175          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1176          * @param $codepoint Integer
1177          * @return String
1178          * @private
1179          */
1180         static function decodeChar( $codepoint ) {
1181                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1182                         return codepointToUtf8( $codepoint );
1183                 } else {
1184                         return UTF8_REPLACEMENT;
1185                 }
1186         }
1187
1188         /**
1189          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1190          * return the UTF-8 encoding of that character. Otherwise, returns
1191          * pseudo-entity source (eg &foo;)
1192          *
1193          * @param $name Strings
1194          * @return String
1195          */
1196         static function decodeEntity( $name ) {
1197                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1198                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1199                         $name = $wgHtmlEntityAliases[$name];
1200                 }
1201                 if( isset( $wgHtmlEntities[$name] ) ) {
1202                         return codepointToUtf8( $wgHtmlEntities[$name] );
1203                 } else {
1204                         return "&$name;";
1205                 }
1206         }
1207
1208         /**
1209          * Fetch the whitelist of acceptable attributes for a given element name.
1210          *
1211          * @param $element String
1212          * @return Array
1213          */
1214         static function attributeWhitelist( $element ) {
1215                 static $list;
1216                 if( !isset( $list ) ) {
1217                         $list = Sanitizer::setupAttributeWhitelist();
1218                 }
1219                 return isset( $list[$element] )
1220                         ? $list[$element]
1221                         : array();
1222         }
1223
1224         /**
1225          * Foreach array key (an allowed HTML element), return an array
1226          * of allowed attributes
1227          * @return Array
1228          */
1229         static function setupAttributeWhitelist() {
1230                 global $wgAllowRdfaAttributes, $wgHtml5, $wgAllowMicrodataAttributes;
1231
1232                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style', 'xml:lang' );
1233
1234                 if ( $wgAllowRdfaAttributes ) {
1235                         #RDFa attributes as specified in section 9 of http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1236                         $common = array_merge( $common, array(
1237                             'about', 'property', 'resource', 'datatype', 'typeof',
1238                         ) );
1239                 }
1240
1241                 if ( $wgHtml5 && $wgAllowMicrodataAttributes ) {
1242                         # add HTML5 microdata tages as pecified by http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#the-microdata-model
1243                         $common = array_merge( $common, array(
1244                             'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1245                         ) );
1246                 }
1247
1248                 $block = array_merge( $common, array( 'align' ) );
1249                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1250                 $tablecell = array( 'abbr',
1251                                     'axis',
1252                                     'headers',
1253                                     'scope',
1254                                     'rowspan',
1255                                     'colspan',
1256                                     'nowrap', # deprecated
1257                                     'width',  # deprecated
1258                                     'height', # deprecated
1259                                     'bgcolor' # deprecated
1260                                     );
1261
1262                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1263                 # See: http://www.w3.org/TR/html4/
1264                 $whitelist = array (
1265                         # 7.5.4
1266                         'div'        => $block,
1267                         'center'     => $common, # deprecated
1268                         'span'       => $block, # ??
1269
1270                         # 7.5.5
1271                         'h1'         => $block,
1272                         'h2'         => $block,
1273                         'h3'         => $block,
1274                         'h4'         => $block,
1275                         'h5'         => $block,
1276                         'h6'         => $block,
1277
1278                         # 7.5.6
1279                         # address
1280
1281                         # 8.2.4
1282                         # bdo
1283
1284                         # 9.2.1
1285                         'em'         => $common,
1286                         'strong'     => $common,
1287                         'cite'       => $common,
1288                         # dfn
1289                         'code'       => $common,
1290                         # samp
1291                         # kbd
1292                         'var'        => $common,
1293                         'abbr'       => $common,
1294                         # acronym
1295
1296                         # 9.2.2
1297                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1298                         # q
1299
1300                         # 9.2.3
1301                         'sub'        => $common,
1302                         'sup'        => $common,
1303
1304                         # 9.3.1
1305                         'p'          => $block,
1306
1307                         # 9.3.2
1308                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1309
1310                         # 9.3.4
1311                         'pre'        => array_merge( $common, array( 'width' ) ),
1312
1313                         # 9.4
1314                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1315                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1316
1317                         # 10.2
1318                         'ul'         => array_merge( $common, array( 'type' ) ),
1319                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1320                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1321
1322                         # 10.3
1323                         'dl'         => $common,
1324                         'dd'         => $common,
1325                         'dt'         => $common,
1326
1327                         # 11.2.1
1328                         'table'      => array_merge( $common,
1329                                                                 array( 'summary', 'width', 'border', 'frame',
1330                                                                                 'rules', 'cellspacing', 'cellpadding',
1331                                                                                 'align', 'bgcolor',
1332                                                                 ) ),
1333
1334                         # 11.2.2
1335                         'caption'    => array_merge( $common, array( 'align' ) ),
1336
1337                         # 11.2.3
1338                         'thead'      => array_merge( $common, $tablealign ),
1339                         'tfoot'      => array_merge( $common, $tablealign ),
1340                         'tbody'      => array_merge( $common, $tablealign ),
1341
1342                         # 11.2.4
1343                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1344                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1345
1346                         # 11.2.5
1347                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1348
1349                         # 11.2.6
1350                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1351                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1352
1353                         # 12.2 # NOTE: <a> is not allowed directly, but the attrib whitelist is used from the Parser object
1354                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1355
1356                         # 13.2
1357                         # Not usually allowed, but may be used for extension-style hooks
1358                         # such as <math> when it is rasterized
1359                         'img'        => array_merge( $common, array( 'alt' ) ),
1360
1361                         # 15.2.1
1362                         'tt'         => $common,
1363                         'b'          => $common,
1364                         'i'          => $common,
1365                         'big'        => $common,
1366                         'small'      => $common,
1367                         'strike'     => $common,
1368                         's'          => $common,
1369                         'u'          => $common,
1370
1371                         # 15.2.2
1372                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1373                         # basefont
1374
1375                         # 15.3
1376                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1377
1378                         # XHTML Ruby annotation text module, simple ruby only.
1379                         # http://www.w3c.org/TR/ruby/
1380                         'ruby'       => $common,
1381                         # rbc
1382                         # rtc
1383                         'rb'         => $common,
1384                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1385                         'rp'         => $common,
1386
1387                         # MathML root element, where used for extensions
1388                         # 'title' may not be 100% valid here; it's XHTML
1389                         # http://www.w3.org/TR/REC-MathML/
1390                         'math'       => array( 'class', 'style', 'id', 'title' ),
1391                         );
1392                 return $whitelist;
1393         }
1394
1395         /**
1396          * Take a fragment of (potentially invalid) HTML and return
1397          * a version with any tags removed, encoded as plain text.
1398          *
1399          * Warning: this return value must be further escaped for literal
1400          * inclusion in HTML output as of 1.10!
1401          *
1402          * @param $text String: HTML fragment
1403          * @return String
1404          */
1405         static function stripAllTags( $text ) {
1406                 # Actual <tags>
1407                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1408
1409                 # Normalize &entities and whitespace
1410                 $text = self::decodeCharReferences( $text );
1411                 $text = self::normalizeWhitespace( $text );
1412
1413                 return $text;
1414         }
1415
1416         /**
1417          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1418          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1419          * PHP 5.1 doesn't.
1420          *
1421          * Use for passing XHTML fragments to PHP's XML parsing functions
1422          *
1423          * @return String
1424          */
1425         static function hackDocType() {
1426                 global $wgHtmlEntities;
1427                 $out = "<!DOCTYPE html [\n";
1428                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1429                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1430                 }
1431                 $out .= "]>\n";
1432                 return $out;
1433         }
1434
1435         static function cleanUrl( $url ) {
1436                 # Normalize any HTML entities in input. They will be
1437                 # re-escaped by makeExternalLink().
1438                 $url = Sanitizer::decodeCharReferences( $url );
1439
1440                 # Escape any control characters introduced by the above step
1441                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1442
1443                 # Validate hostname portion
1444                 $matches = array();
1445                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1446                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1447
1448                         // Characters that will be ignored in IDNs.
1449                         // http://tools.ietf.org/html/3454#section-3.1
1450                         // Strip them before further processing so blacklists and such work.
1451                         $strip = "/
1452                                 \\s|          # general whitespace
1453                                 \xc2\xad|     # 00ad SOFT HYPHEN
1454                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1455                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1456                                 \xe2\x81\xa0| # 2060 WORD JOINER
1457                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1458                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1459                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1460                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1461                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1462                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1463                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1464                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1465                                 /xuD";
1466
1467                         $host = preg_replace( $strip, '', $host );
1468
1469                         // @todo Fixme: validate hostnames here
1470
1471                         return $protocol . $host . $rest;
1472                 } else {
1473                         return $url;
1474                 }
1475         }
1476
1477 }