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          *
 861          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 862          *                                                          in the id and
 863          *                                                          name attributes
 864          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 865          * @see http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#the-id-attribute
 866          *   HTML5 definition of id attribute
 867          *
 868          * @param $id String: id to escape
 869          * @param $options Mixed: string or array of strings (default is array()):
 870          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 871          *       so don't pay attention if the first character isn't valid at the
 872          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
 873          *       false.
 874          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
 875          *       if $wgExperimentalHtmlIds is used, so we can generate extra
 876          *       anchors and links won't break.
 877          * @return String
 878          */
 879         static function escapeId( $id, $options = array() ) {
 880                 global $wgHtml5, $wgExperimentalHtmlIds;
 881                 $options = (array)$options;
 882
 883                 if ( $wgHtml5 && $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
 884                         $id = Sanitizer::decodeCharReferences( $id );
 885                         $id = preg_replace( '/[ \t\n\r\f_\'"&]+/', '_', $id );
 886                         $id = trim( $id, '_' );
 887                         if ( $id === '' ) {
 888                                 # Must have been all whitespace to start with.
 889                                 return '_';
 890                         } else {
 891                                 return $id;
 892                         }
 893                 }
 894
 895                 # HTML4-style escaping
 896                 static $replace = array(
 897                         '%3A' => ':',
 898                         '%' => '.'
 899                 );
 900
 901                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 902                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 903
 904                 if ( !preg_match( '/^[a-zA-Z]/', $id )
 905                 && !in_array( 'noninitial', $options ) )  {
 906                         // Initial character must be a letter!
 907                         $id = "x$id";
 908                 }
 909                 return $id;
 910         }
 911
 912         /**
 913          * Given a value, escape it so that it can be used as a CSS class and
 914          * return it.
 915          *
 916          * @todo For extra validity, input should be validated UTF-8.
 917          *
 918          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 919          *
 920          * @param $class String
 921          * @return String
 922          */
 923         static function escapeClass( $class ) {
 924                 // Convert ugly stuff to underscores and kill underscores in ugly places
 925                 return rtrim(preg_replace(
 926                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 927                         '_',
 928                         $class ), '_');
 929         }
 930
 931         /**
 932          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 933          * This allows (generally harmless) entities like &nbsp; to survive.
 934          *
 935          * @param $html String to escape
 936          * @return String: escaped input
 937          */
 938         static function escapeHtmlAllowEntities( $html ) {
 939                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 940                 # hurt.
 941                 $html = htmlspecialchars( $html, ENT_QUOTES );
 942                 $html = str_replace( '&amp;', '&', $html );
 943                 $html = Sanitizer::normalizeCharReferences( $html );
 944                 return $html;
 945         }
 946
 947         /**
 948          * Regex replace callback for armoring links against further processing.
 949          * @param $matches Array
 950          * @return string
 951          */
 952         private static function armorLinksCallback( $matches ) {
 953                 return str_replace( ':', '&#58;', $matches[1] );
 954         }
 955
 956         /**
 957          * Return an associative array of attribute names and values from
 958          * a partial tag string. Attribute names are forces to lowercase,
 959          * character references are decoded to UTF-8 text.
 960          *
 961          * @param $text String
 962          * @return Array
 963          */
 964         public static function decodeTagAttributes( $text ) {
 965                 if( trim( $text ) == '' ) {
 966                         return array();
 967                 }
 968
 969                 $attribs = array();
 970                 $pairs = array();
 971                 if( !preg_match_all(
 972                         MW_ATTRIBS_REGEX,
 973                         $text,
 974                         $pairs,
 975                         PREG_SET_ORDER ) ) {
 976                         return $attribs;
 977                 }
 978
 979                 foreach( $pairs as $set ) {
 980                         $attribute = strtolower( $set[1] );
 981                         $value = Sanitizer::getTagAttributeCallback( $set );
 982
 983                         // Normalize whitespace
 984                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 985                         $value = trim( $value );
 986
 987                         // Decode character references
 988                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 989                 }
 990                 return $attribs;
 991         }
 992
 993         /**
 994          * Pick the appropriate attribute value from a match set from the
 995          * MW_ATTRIBS_REGEX matches.
 996          *
 997          * @param $set Array
 998          * @return String
 999          */
1000         private static function getTagAttributeCallback( $set ) {
1001                 if( isset( $set[6] ) ) {
1002                         # Illegal #XXXXXX color with no quotes.
1003                         return $set[6];
1004                 } elseif( isset( $set[5] ) ) {
1005                         # No quotes.
1006                         return $set[5];
1007                 } elseif( isset( $set[4] ) ) {
1008                         # Single-quoted
1009                         return $set[4];
1010                 } elseif( isset( $set[3] ) ) {
1011                         # Double-quoted
1012                         return $set[3];
1013                 } elseif( !isset( $set[2] ) ) {
1014                         # In XHTML, attributes must have a value.
1015                         # For 'reduced' form, return explicitly the attribute name here.
1016                         return $set[1];
1017                 } else {
1018                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1019                 }
1020         }
1021
1022         /**
1023          * Normalize whitespace and character references in an XML source-
1024          * encoded text for an attribute value.
1025          *
1026          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
1027          * but note that we're not returning the value, but are returning
1028          * XML source fragments that will be slapped into output.
1029          *
1030          * @param $text String
1031          * @return String
1032          */
1033         private static function normalizeAttributeValue( $text ) {
1034                 return str_replace( '"', '&quot;',
1035                         self::normalizeWhitespace(
1036                                 Sanitizer::normalizeCharReferences( $text ) ) );
1037         }
1038
1039         private static function normalizeWhitespace( $text ) {
1040                 return preg_replace(
1041                         '/\r\n|[\x20\x0d\x0a\x09]/',
1042                         ' ',
1043                         $text );
1044         }
1045
1046         /**
1047          * Ensure that any entities and character references are legal
1048          * for XML and XHTML specifically. Any stray bits will be
1049          * &amp;-escaped to result in a valid text fragment.
1050          *
1051          * a. any named char refs must be known in XHTML
1052          * b. any numeric char refs must be legal chars, not invalid or forbidden
1053          * c. use &#x, not &#X
1054          * d. fix or reject non-valid attributes
1055          *
1056          * @param $text String
1057          * @return String
1058          * @private
1059          */
1060         static function normalizeCharReferences( $text ) {
1061                 return preg_replace_callback(
1062                         MW_CHAR_REFS_REGEX,
1063                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1064                         $text );
1065         }
1066         /**
1067          * @param $matches String
1068          * @return String
1069          */
1070         static function normalizeCharReferencesCallback( $matches ) {
1071                 $ret = null;
1072                 if( $matches[1] != '' ) {
1073                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1074                 } elseif( $matches[2] != '' ) {
1075                         $ret = Sanitizer::decCharReference( $matches[2] );
1076                 } elseif( $matches[3] != ''  ) {
1077                         $ret = Sanitizer::hexCharReference( $matches[3] );
1078                 } elseif( $matches[4] != '' ) {
1079                         $ret = Sanitizer::hexCharReference( $matches[4] );
1080                 }
1081                 if( is_null( $ret ) ) {
1082                         return htmlspecialchars( $matches[0] );
1083                 } else {
1084                         return $ret;
1085                 }
1086         }
1087
1088         /**
1089          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1090          * return the named entity reference as is. If the entity is a
1091          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1092          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1093          *
1094          * @param $name String
1095          * @return String
1096          */
1097         static function normalizeEntity( $name ) {
1098                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1099                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1100                         return "&{$wgHtmlEntityAliases[$name]};";
1101                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1102                         return "&$name;";
1103                 } else {
1104                         return "&amp;$name;";
1105                 }
1106         }
1107
1108         static function decCharReference( $codepoint ) {
1109                 $point = intval( $codepoint );
1110                 if( Sanitizer::validateCodepoint( $point ) ) {
1111                         return sprintf( '&#%d;', $point );
1112                 } else {
1113                         return null;
1114                 }
1115         }
1116
1117         static function hexCharReference( $codepoint ) {
1118                 $point = hexdec( $codepoint );
1119                 if( Sanitizer::validateCodepoint( $point ) ) {
1120                         return sprintf( '&#x%x;', $point );
1121                 } else {
1122                         return null;
1123                 }
1124         }
1125
1126         /**
1127          * Returns true if a given Unicode codepoint is a valid character in XML.
1128          * @param $codepoint Integer
1129          * @return Boolean
1130          */
1131         private static function validateCodepoint( $codepoint ) {
1132                 return ($codepoint ==    0x09)
1133                         || ($codepoint ==    0x0a)
1134                         || ($codepoint ==    0x0d)
1135                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1136                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1137                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1138         }
1139
1140         /**
1141          * Decode any character references, numeric or named entities,
1142          * in the text and return a UTF-8 string.
1143          *
1144          * @param $text String
1145          * @return String
1146          */
1147         public static function decodeCharReferences( $text ) {
1148                 return preg_replace_callback(
1149                         MW_CHAR_REFS_REGEX,
1150                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1151                         $text );
1152         }
1153
1154         /**
1155          * @param $matches String
1156          * @return String
1157          */
1158         static function decodeCharReferencesCallback( $matches ) {
1159                 if( $matches[1] != '' ) {
1160                         return Sanitizer::decodeEntity( $matches[1] );
1161                 } elseif( $matches[2] != '' ) {
1162                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1163                 } elseif( $matches[3] != ''  ) {
1164                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1165                 } elseif( $matches[4] != '' ) {
1166                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1167                 }
1168                 # Last case should be an ampersand by itself
1169                 return $matches[0];
1170         }
1171
1172         /**
1173          * Return UTF-8 string for a codepoint if that is a valid
1174          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1175          * @param $codepoint Integer
1176          * @return String
1177          * @private
1178          */
1179         static function decodeChar( $codepoint ) {
1180                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1181                         return codepointToUtf8( $codepoint );
1182                 } else {
1183                         return UTF8_REPLACEMENT;
1184                 }
1185         }
1186
1187         /**
1188          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1189          * return the UTF-8 encoding of that character. Otherwise, returns
1190          * pseudo-entity source (eg &foo;)
1191          *
1192          * @param $name Strings
1193          * @return String
1194          */
1195         static function decodeEntity( $name ) {
1196                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1197                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1198                         $name = $wgHtmlEntityAliases[$name];
1199                 }
1200                 if( isset( $wgHtmlEntities[$name] ) ) {
1201                         return codepointToUtf8( $wgHtmlEntities[$name] );
1202                 } else {
1203                         return "&$name;";
1204                 }
1205         }
1206
1207         /**
1208          * Fetch the whitelist of acceptable attributes for a given element name.
1209          *
1210          * @param $element String
1211          * @return Array
1212          */
1213         static function attributeWhitelist( $element ) {
1214                 static $list;
1215                 if( !isset( $list ) ) {
1216                         $list = Sanitizer::setupAttributeWhitelist();
1217                 }
1218                 return isset( $list[$element] )
1219                         ? $list[$element]
1220                         : array();
1221         }
1222
1223         /**
1224          * Foreach array key (an allowed HTML element), return an array
1225          * of allowed attributes
1226          * @return Array
1227          */
1228         static function setupAttributeWhitelist() {
1229                 global $wgAllowRdfaAttributes, $wgHtml5, $wgAllowMicrodataAttributes;
1230
1231                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style', 'xml:lang' );
1232
1233                 if ( $wgAllowRdfaAttributes ) {
1234                         #RDFa attributes as specified in section 9 of http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1235                         $common = array_merge( $common, array(
1236                             'about', 'property', 'resource', 'datatype', 'typeof',
1237                         ) );
1238                 }
1239
1240                 if ( $wgHtml5 && $wgAllowMicrodataAttributes ) {
1241                         # add HTML5 microdata tages as pecified by http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#the-microdata-model
1242                         $common = array_merge( $common, array(
1243                             'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1244                         ) );
1245                 }
1246
1247                 $block = array_merge( $common, array( 'align' ) );
1248                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1249                 $tablecell = array( 'abbr',
1250                                     'axis',
1251                                     'headers',
1252                                     'scope',
1253                                     'rowspan',
1254                                     'colspan',
1255                                     'nowrap', # deprecated
1256                                     'width',  # deprecated
1257                                     'height', # deprecated
1258                                     'bgcolor' # deprecated
1259                                     );
1260
1261                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1262                 # See: http://www.w3.org/TR/html4/
1263                 $whitelist = array (
1264                         # 7.5.4
1265                         'div'        => $block,
1266                         'center'     => $common, # deprecated
1267                         'span'       => $block, # ??
1268
1269                         # 7.5.5
1270                         'h1'         => $block,
1271                         'h2'         => $block,
1272                         'h3'         => $block,
1273                         'h4'         => $block,
1274                         'h5'         => $block,
1275                         'h6'         => $block,
1276
1277                         # 7.5.6
1278                         # address
1279
1280                         # 8.2.4
1281                         # bdo
1282
1283                         # 9.2.1
1284                         'em'         => $common,
1285                         'strong'     => $common,
1286                         'cite'       => $common,
1287                         # dfn
1288                         'code'       => $common,
1289                         # samp
1290                         # kbd
1291                         'var'        => $common,
1292                         'abbr'       => $common,
1293                         # acronym
1294
1295                         # 9.2.2
1296                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1297                         # q
1298
1299                         # 9.2.3
1300                         'sub'        => $common,
1301                         'sup'        => $common,
1302
1303                         # 9.3.1
1304                         'p'          => $block,
1305
1306                         # 9.3.2
1307                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1308
1309                         # 9.3.4
1310                         'pre'        => array_merge( $common, array( 'width' ) ),
1311
1312                         # 9.4
1313                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1314                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1315
1316                         # 10.2
1317                         'ul'         => array_merge( $common, array( 'type' ) ),
1318                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1319                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1320
1321                         # 10.3
1322                         'dl'         => $common,
1323                         'dd'         => $common,
1324                         'dt'         => $common,
1325
1326                         # 11.2.1
1327                         'table'      => array_merge( $common,
1328                                                                 array( 'summary', 'width', 'border', 'frame',
1329                                                                                 'rules', 'cellspacing', 'cellpadding',
1330                                                                                 'align', 'bgcolor',
1331                                                                 ) ),
1332
1333                         # 11.2.2
1334                         'caption'    => array_merge( $common, array( 'align' ) ),
1335
1336                         # 11.2.3
1337                         'thead'      => array_merge( $common, $tablealign ),
1338                         'tfoot'      => array_merge( $common, $tablealign ),
1339                         'tbody'      => array_merge( $common, $tablealign ),
1340
1341                         # 11.2.4
1342                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1343                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1344
1345                         # 11.2.5
1346                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1347
1348                         # 11.2.6
1349                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1350                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1351
1352                         # 12.2 # NOTE: <a> is not allowed directly, but the attrib whitelist is used from the Parser object
1353                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1354
1355                         # 13.2
1356                         # Not usually allowed, but may be used for extension-style hooks
1357                         # such as <math> when it is rasterized
1358                         'img'        => array_merge( $common, array( 'alt' ) ),
1359
1360                         # 15.2.1
1361                         'tt'         => $common,
1362                         'b'          => $common,
1363                         'i'          => $common,
1364                         'big'        => $common,
1365                         'small'      => $common,
1366                         'strike'     => $common,
1367                         's'          => $common,
1368                         'u'          => $common,
1369
1370                         # 15.2.2
1371                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1372                         # basefont
1373
1374                         # 15.3
1375                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1376
1377                         # XHTML Ruby annotation text module, simple ruby only.
1378                         # http://www.w3c.org/TR/ruby/
1379                         'ruby'       => $common,
1380                         # rbc
1381                         # rtc
1382                         'rb'         => $common,
1383                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1384                         'rp'         => $common,
1385
1386                         # MathML root element, where used for extensions
1387                         # 'title' may not be 100% valid here; it's XHTML
1388                         # http://www.w3.org/TR/REC-MathML/
1389                         'math'       => array( 'class', 'style', 'id', 'title' ),
1390                         );
1391                 return $whitelist;
1392         }
1393
1394         /**
1395          * Take a fragment of (potentially invalid) HTML and return
1396          * a version with any tags removed, encoded as plain text.
1397          *
1398          * Warning: this return value must be further escaped for literal
1399          * inclusion in HTML output as of 1.10!
1400          *
1401          * @param $text String: HTML fragment
1402          * @return String
1403          */
1404         static function stripAllTags( $text ) {
1405                 # Actual <tags>
1406                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1407
1408                 # Normalize &entities and whitespace
1409                 $text = self::decodeCharReferences( $text );
1410                 $text = self::normalizeWhitespace( $text );
1411
1412                 return $text;
1413         }
1414
1415         /**
1416          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1417          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1418          * PHP 5.1 doesn't.
1419          *
1420          * Use for passing XHTML fragments to PHP's XML parsing functions
1421          *
1422          * @return String
1423          */
1424         static function hackDocType() {
1425                 global $wgHtmlEntities;
1426                 $out = "<!DOCTYPE html [\n";
1427                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1428                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1429                 }
1430                 $out .= "]>\n";
1431                 return $out;
1432         }
1433
1434         static function cleanUrl( $url ) {
1435                 # Normalize any HTML entities in input. They will be
1436                 # re-escaped by makeExternalLink().
1437                 $url = Sanitizer::decodeCharReferences( $url );
1438
1439                 # Escape any control characters introduced by the above step
1440                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1441
1442                 # Validate hostname portion
1443                 $matches = array();
1444                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1445                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1446
1447                         // Characters that will be ignored in IDNs.
1448                         // http://tools.ietf.org/html/3454#section-3.1
1449                         // Strip them before further processing so blacklists and such work.
1450                         $strip = "/
1451                                 \\s|          # general whitespace
1452                                 \xc2\xad|     # 00ad SOFT HYPHEN
1453                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1454                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1455                                 \xe2\x81\xa0| # 2060 WORD JOINER
1456                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1457                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1458                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1459                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1460                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1461                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1462                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1463                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1464                                 /xuD";
1465
1466                         $host = preg_replace( $strip, '', $host );
1467
1468                         // @todo Fixme: validate hostnames here
1469
1470                         return $protocol . $host . $rest;
1471                 } else {
1472                         return $url;
1473                 }
1474         }
1475
1476 }