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  * @package MediaWiki
  24  * @subpackage 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]+);
  33          |&\#([0-9]+);
  34          |&\#x([0-9A-Za-z]+);
  35          |&\#X([0-9A-Za-z]+);
  36          |(&)/x' );
  37
  38 /**
  39  * Regular expression to match HTML/XML attribute pairs within a tag.
  40  * Allows some... latitude.
  41  * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
  42  */
  43 $attrib = '[A-Za-z0-9]';
  44 $space = '[\x09\x0a\x0d\x20]';
  45 define( 'MW_ATTRIBS_REGEX',
  46         "/(?:^|$space)($attrib+)
  47           ($space*=$space*
  48                 (?:
  49                  # The attribute value: quoted or alone
  50                   \"([^<\"]*)\"
  51                  | '([^<']*)'
  52                  |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
  53                  |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
  54                                                          # colors are specified like this.
  55                                                          # We'll be normalizing it.
  56                 )
  57            )?(?=$space|\$)/sx" );
  58
  59 /**
  60  * List of all named character entities defined in HTML 4.01
  61  * http://www.w3.org/TR/html4/sgml/entities.html
  62  * @private
  63  */
  64 global $wgHtmlEntities;
  65 $wgHtmlEntities = array(
  66         'Aacute'   => 193,
  67         'aacute'   => 225,
  68         'Acirc'    => 194,
  69         'acirc'    => 226,
  70         'acute'    => 180,
  71         'AElig'    => 198,
  72         'aelig'    => 230,
  73         'Agrave'   => 192,
  74         'agrave'   => 224,
  75         'alefsym'  => 8501,
  76         'Alpha'    => 913,
  77         'alpha'    => 945,
  78         'amp'      => 38,
  79         'and'      => 8743,
  80         'ang'      => 8736,
  81         'Aring'    => 197,
  82         'aring'    => 229,
  83         'asymp'    => 8776,
  84         'Atilde'   => 195,
  85         'atilde'   => 227,
  86         'Auml'     => 196,
  87         'auml'     => 228,
  88         'bdquo'    => 8222,
  89         'Beta'     => 914,
  90         'beta'     => 946,
  91         'brvbar'   => 166,
  92         'bull'     => 8226,
  93         'cap'      => 8745,
  94         'Ccedil'   => 199,
  95         'ccedil'   => 231,
  96         'cedil'    => 184,
  97         'cent'     => 162,
  98         'Chi'      => 935,
  99         'chi'      => 967,
 100         'circ'     => 710,
 101         'clubs'    => 9827,
 102         'cong'     => 8773,
 103         'copy'     => 169,
 104         'crarr'    => 8629,
 105         'cup'      => 8746,
 106         'curren'   => 164,
 107         'dagger'   => 8224,
 108         'Dagger'   => 8225,
 109         'darr'     => 8595,
 110         'dArr'     => 8659,
 111         'deg'      => 176,
 112         'Delta'    => 916,
 113         'delta'    => 948,
 114         'diams'    => 9830,
 115         'divide'   => 247,
 116         'Eacute'   => 201,
 117         'eacute'   => 233,
 118         'Ecirc'    => 202,
 119         'ecirc'    => 234,
 120         'Egrave'   => 200,
 121         'egrave'   => 232,
 122         'empty'    => 8709,
 123         'emsp'     => 8195,
 124         'ensp'     => 8194,
 125         'Epsilon'  => 917,
 126         'epsilon'  => 949,
 127         'equiv'    => 8801,
 128         'Eta'      => 919,
 129         'eta'      => 951,
 130         'ETH'      => 208,
 131         'eth'      => 240,
 132         'Euml'     => 203,
 133         'euml'     => 235,
 134         'euro'     => 8364,
 135         'exist'    => 8707,
 136         'fnof'     => 402,
 137         'forall'   => 8704,
 138         'frac12'   => 189,
 139         'frac14'   => 188,
 140         'frac34'   => 190,
 141         'frasl'    => 8260,
 142         'Gamma'    => 915,
 143         'gamma'    => 947,
 144         'ge'       => 8805,
 145         'gt'       => 62,
 146         'harr'     => 8596,
 147         'hArr'     => 8660,
 148         'hearts'   => 9829,
 149         'hellip'   => 8230,
 150         'Iacute'   => 205,
 151         'iacute'   => 237,
 152         'Icirc'    => 206,
 153         'icirc'    => 238,
 154         'iexcl'    => 161,
 155         'Igrave'   => 204,
 156         'igrave'   => 236,
 157         'image'    => 8465,
 158         'infin'    => 8734,
 159         'int'      => 8747,
 160         'Iota'     => 921,
 161         'iota'     => 953,
 162         'iquest'   => 191,
 163         'isin'     => 8712,
 164         'Iuml'     => 207,
 165         'iuml'     => 239,
 166         'Kappa'    => 922,
 167         'kappa'    => 954,
 168         'Lambda'   => 923,
 169         'lambda'   => 955,
 170         'lang'     => 9001,
 171         'laquo'    => 171,
 172         'larr'     => 8592,
 173         'lArr'     => 8656,
 174         'lceil'    => 8968,
 175         'ldquo'    => 8220,
 176         'le'       => 8804,
 177         'lfloor'   => 8970,
 178         'lowast'   => 8727,
 179         'loz'      => 9674,
 180         'lrm'      => 8206,
 181         'lsaquo'   => 8249,
 182         'lsquo'    => 8216,
 183         'lt'       => 60,
 184         'macr'     => 175,
 185         'mdash'    => 8212,
 186         'micro'    => 181,
 187         'middot'   => 183,
 188         'minus'    => 8722,
 189         'Mu'       => 924,
 190         'mu'       => 956,
 191         'nabla'    => 8711,
 192         'nbsp'     => 160,
 193         'ndash'    => 8211,
 194         'ne'       => 8800,
 195         'ni'       => 8715,
 196         'not'      => 172,
 197         'notin'    => 8713,
 198         'nsub'     => 8836,
 199         'Ntilde'   => 209,
 200         'ntilde'   => 241,
 201         'Nu'       => 925,
 202         'nu'       => 957,
 203         'Oacute'   => 211,
 204         'oacute'   => 243,
 205         'Ocirc'    => 212,
 206         'ocirc'    => 244,
 207         'OElig'    => 338,
 208         'oelig'    => 339,
 209         'Ograve'   => 210,
 210         'ograve'   => 242,
 211         'oline'    => 8254,
 212         'Omega'    => 937,
 213         'omega'    => 969,
 214         'Omicron'  => 927,
 215         'omicron'  => 959,
 216         'oplus'    => 8853,
 217         'or'       => 8744,
 218         'ordf'     => 170,
 219         'ordm'     => 186,
 220         'Oslash'   => 216,
 221         'oslash'   => 248,
 222         'Otilde'   => 213,
 223         'otilde'   => 245,
 224         'otimes'   => 8855,
 225         'Ouml'     => 214,
 226         'ouml'     => 246,
 227         'para'     => 182,
 228         'part'     => 8706,
 229         'permil'   => 8240,
 230         'perp'     => 8869,
 231         'Phi'      => 934,
 232         'phi'      => 966,
 233         'Pi'       => 928,
 234         'pi'       => 960,
 235         'piv'      => 982,
 236         'plusmn'   => 177,
 237         'pound'    => 163,
 238         'prime'    => 8242,
 239         'Prime'    => 8243,
 240         'prod'     => 8719,
 241         'prop'     => 8733,
 242         'Psi'      => 936,
 243         'psi'      => 968,
 244         'quot'     => 34,
 245         'radic'    => 8730,
 246         'rang'     => 9002,
 247         'raquo'    => 187,
 248         'rarr'     => 8594,
 249         'rArr'     => 8658,
 250         'rceil'    => 8969,
 251         'rdquo'    => 8221,
 252         'real'     => 8476,
 253         'reg'      => 174,
 254         'rfloor'   => 8971,
 255         'Rho'      => 929,
 256         'rho'      => 961,
 257         'rlm'      => 8207,
 258         'rsaquo'   => 8250,
 259         'rsquo'    => 8217,
 260         'sbquo'    => 8218,
 261         'Scaron'   => 352,
 262         'scaron'   => 353,
 263         'sdot'     => 8901,
 264         'sect'     => 167,
 265         'shy'      => 173,
 266         'Sigma'    => 931,
 267         'sigma'    => 963,
 268         'sigmaf'   => 962,
 269         'sim'      => 8764,
 270         'spades'   => 9824,
 271         'sub'      => 8834,
 272         'sube'     => 8838,
 273         'sum'      => 8721,
 274         'sup'      => 8835,
 275         'sup1'     => 185,
 276         'sup2'     => 178,
 277         'sup3'     => 179,
 278         'supe'     => 8839,
 279         'szlig'    => 223,
 280         'Tau'      => 932,
 281         'tau'      => 964,
 282         'there4'   => 8756,
 283         'Theta'    => 920,
 284         'theta'    => 952,
 285         'thetasym' => 977,
 286         'thinsp'   => 8201,
 287         'THORN'    => 222,
 288         'thorn'    => 254,
 289         'tilde'    => 732,
 290         'times'    => 215,
 291         'trade'    => 8482,
 292         'Uacute'   => 218,
 293         'uacute'   => 250,
 294         'uarr'     => 8593,
 295         'uArr'     => 8657,
 296         'Ucirc'    => 219,
 297         'ucirc'    => 251,
 298         'Ugrave'   => 217,
 299         'ugrave'   => 249,
 300         'uml'      => 168,
 301         'upsih'    => 978,
 302         'Upsilon'  => 933,
 303         'upsilon'  => 965,
 304         'Uuml'     => 220,
 305         'uuml'     => 252,
 306         'weierp'   => 8472,
 307         'Xi'       => 926,
 308         'xi'       => 958,
 309         'Yacute'   => 221,
 310         'yacute'   => 253,
 311         'yen'      => 165,
 312         'Yuml'     => 376,
 313         'yuml'     => 255,
 314         'Zeta'     => 918,
 315         'zeta'     => 950,
 316         'zwj'      => 8205,
 317         'zwnj'     => 8204 );
 318
 319 /** @package MediaWiki */
 320 class Sanitizer {
 321         /**
 322          * Cleans up HTML, removes dangerous tags and attributes, and
 323          * removes HTML comments
 324          * @private
 325          * @param string $text
 326          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 327          * @param array $args for the processing callback
 328          * @return string
 329          */
 330         static function removeHTMLtags( $text, $processCallback = null, $args = array() ) {
 331                 global $wgUseTidy, $wgUserHtml;
 332
 333                 static $htmlpairs, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 334                         $htmllist, $listtags, $htmlsingleallowed, $htmlelements, $staticInitialised;
 335
 336                 wfProfileIn( __METHOD__ );
 337
 338                 if ( !$staticInitialised ) {
 339                         if( $wgUserHtml ) {
 340                                 $htmlpairs = array( # Tags that must be closed
 341                                         'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 342                                         'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 343                                         'strike', 'strong', 'tt', 'var', 'div', 'center',
 344                                         'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 345                                         'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
 346                                 );
 347                                 $htmlsingle = array(
 348                                         'br', 'hr', 'li', 'dt', 'dd'
 349                                 );
 350                                 $htmlsingleonly = array( # Elements that cannot have close tags
 351                                         'br', 'hr'
 352                                 );
 353                                 $htmlnest = array( # Tags that can be nested--??
 354                                         'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 355                                         'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 356                                 );
 357                                 $tabletags = array( # Can only appear inside table
 358                                         'td', 'th', 'tr',
 359                                 );
 360                                 $htmllist = array( # Tags used by list
 361                                         'ul','ol',
 362                                 );
 363                                 $listtags = array( # Tags that can appear in a list
 364                                         'li',
 365                                 );
 366
 367                         } else {
 368                                 $htmlpairs = array();
 369                                 $htmlsingle = array();
 370                                 $htmlnest = array();
 371                                 $tabletags = array();
 372                         }
 373
 374                         $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
 375                         $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
 376
 377                         # Convert them all to hashtables for faster lookup
 378                         $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 379                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
 380                         foreach ( $vars as $var ) {
 381                                 $$var = array_flip( $$var );
 382                         }
 383                         $staticInitialised = true;
 384                 }
 385
 386                 # Remove HTML comments
 387                 $text = Sanitizer::removeHTMLcomments( $text );
 388                 $bits = explode( '<', $text );
 389                 $text = array_shift( $bits );
 390                 if(!$wgUseTidy) {
 391                         $tagstack = $tablestack = array();
 392                         foreach ( $bits as $x ) {
 393                                 $prev = error_reporting( E_ALL & ~( E_NOTICE | E_WARNING ) );
 394                                 $regs = array();
 395                                 preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs );
 396                                 list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 397                                 error_reporting( $prev );
 398
 399                                 $badtag = 0 ;
 400                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 401                                         # Check our stack
 402                                         if ( $slash ) {
 403                                                 # Closing a tag...
 404                                                 if( isset( $htmlsingleonly[$t] ) ) {
 405                                                         $badtag = 1;
 406                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 407                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 408                                                                 # Pop all elements with an optional close tag
 409                                                                 # and see if we find a match below them
 410                                                                 $optstack = array();
 411                                                                 array_push ($optstack, $ot);
 412                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 413                                                                                 isset( $htmlsingleallowed[$ot] ) )
 414                                                                 {
 415                                                                         array_push ($optstack, $ot);
 416                                                                 }
 417                                                                 if ( $t != $ot ) {
 418                                                                         # No match. Push the optinal elements back again
 419                                                                         $badtag = 1;
 420                                                                         while ( $ot = @array_pop( $optstack ) ) {
 421                                                                                 array_push( $tagstack, $ot );
 422                                                                         }
 423                                                                 }
 424                                                         } else {
 425                                                                 @array_push( $tagstack, $ot );
 426                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 427                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 428                                                                         $badtag = 1;
 429                                                                 }
 430                                                         }
 431                                                 } else {
 432                                                         if ( $t == 'table' ) {
 433                                                                 $tagstack = array_pop( $tablestack );
 434                                                         }
 435                                                 }
 436                                                 $newparams = '';
 437                                         } else {
 438                                                 # Keep track for later
 439                                                 if ( isset( $tabletags[$t] ) &&
 440                                                 ! in_array( 'table', $tagstack ) ) {
 441                                                         $badtag = 1;
 442                                                 } else if ( in_array( $t, $tagstack ) &&
 443                                                 ! isset( $htmlnest [$t ] ) ) {
 444                                                         $badtag = 1 ;
 445                                                 # Is it a self closed htmlpair ? (bug 5487)
 446                                                 } else if( $brace == '/>' &&
 447                                                 isset( $htmlpairs[$t] ) ) {
 448                                                         $badtag = 1;
 449                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 450                                                         # Hack to force empty tag for uncloseable elements
 451                                                         $brace = '/>';
 452                                                 } else if( isset( $htmlsingle[$t] ) ) {
 453                                                         # Hack to not close $htmlsingle tags
 454                                                         $brace = NULL;
 455                                                 } else {
 456                                                         if ( $t == 'table' ) {
 457                                                                 array_push( $tablestack, $tagstack );
 458                                                                 $tagstack = array();
 459                                                         }
 460                                                         array_push( $tagstack, $t );
 461                                                 }
 462
 463                                                 # Replace any variables or template parameters with
 464                                                 # plaintext results.
 465                                                 if( is_callable( $processCallback ) ) {
 466                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 467                                                 }
 468
 469                                                 # Strip non-approved attributes from the tag
 470                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 471                                         }
 472                                         if ( ! $badtag ) {
 473                                                 $rest = str_replace( '>', '&gt;', $rest );
 474                                                 $close = ( $brace == '/>' ) ? ' /' : '';
 475                                                 $text .= "<$slash$t$newparams$close>$rest";
 476                                                 continue;
 477                                         }
 478                                 }
 479                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 480                         }
 481                         # Close off any remaining tags
 482                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 483                                 $text .= "</$t>\n";
 484                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 485                         }
 486                 } else {
 487                         # this might be possible using tidy itself
 488                         foreach ( $bits as $x ) {
 489                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 490                                 $x, $regs );
 491                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 492                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 493                                         if( is_callable( $processCallback ) ) {
 494                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 495                                         }
 496                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 497                                         $rest = str_replace( '>', '&gt;', $rest );
 498                                         $text .= "<$slash$t$newparams$brace$rest";
 499                                 } else {
 500                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 501                                 }
 502                         }
 503                 }
 504                 wfProfileOut( __METHOD__ );
 505                 return $text;
 506         }
 507
 508         /**
 509          * Remove '<!--', '-->', and everything between.
 510          * To avoid leaving blank lines, when a comment is both preceded
 511          * and followed by a newline (ignoring spaces), trim leading and
 512          * trailing spaces and one of the newlines.
 513          *
 514          * @private
 515          * @param string $text
 516          * @return string
 517          */
 518         static function removeHTMLcomments( $text ) {
 519                 wfProfileIn( __METHOD__ );
 520                 while (($start = strpos($text, '<!--')) !== false) {
 521                         $end = strpos($text, '-->', $start + 4);
 522                         if ($end === false) {
 523                                 # Unterminated comment; bail out
 524                                 break;
 525                         }
 526
 527                         $end += 3;
 528
 529                         # Trim space and newline if the comment is both
 530                         # preceded and followed by a newline
 531                         $spaceStart = max($start - 1, 0);
 532                         $spaceLen = $end - $spaceStart;
 533                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 534                                 $spaceStart--;
 535                                 $spaceLen++;
 536                         }
 537                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 538                                 $spaceLen++;
 539                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 540                                 # Remove the comment, leading and trailing
 541                                 # spaces, and leave only one newline.
 542                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 543                         }
 544                         else {
 545                                 # Remove just the comment.
 546                                 $text = substr_replace($text, '', $start, $end - $start);
 547                         }
 548                 }
 549                 wfProfileOut( __METHOD__ );
 550                 return $text;
 551         }
 552
 553         /**
 554          * Take an array of attribute names and values and normalize or discard
 555          * illegal values for the given element type.
 556          *
 557          * - Discards attributes not on a whitelist for the given element
 558          * - Unsafe style attributes are discarded
 559          *
 560          * @param array $attribs
 561          * @param string $element
 562          * @return array
 563          *
 564          * @todo Check for legal values where the DTD limits things.
 565          * @todo Check for unique id attribute :P
 566          */
 567         static function validateTagAttributes( $attribs, $element ) {
 568                 $whitelist = array_flip( Sanitizer::attributeWhitelist( $element ) );
 569                 $out = array();
 570                 foreach( $attribs as $attribute => $value ) {
 571                         if( !isset( $whitelist[$attribute] ) ) {
 572                                 continue;
 573                         }
 574                         # Strip javascript "expression" from stylesheets.
 575                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 576                         if( $attribute == 'style' ) {
 577                                 $value = Sanitizer::checkCss( $value );
 578                                 if( $value === false ) {
 579                                         # haxx0r
 580                                         continue;
 581                                 }
 582                         }
 583
 584                         if ( $attribute === 'id' )
 585                                 $value = Sanitizer::escapeId( $value );
 586
 587                         // If this attribute was previously set, override it.
 588                         // Output should only have one attribute of each name.
 589                         $out[$attribute] = $value;
 590                 }
 591                 return $out;
 592         }
 593
 594         /**
 595          * Pick apart some CSS and check it for forbidden or unsafe structures.
 596          * Returns a sanitized string, or false if it was just too evil.
 597          *
 598          * Currently URL references, 'expression', 'tps' are forbidden.
 599          *
 600          * @param string $value
 601          * @return mixed
 602          */
 603         static function checkCss( $value ) {
 604                 $stripped = Sanitizer::decodeCharReferences( $value );
 605
 606                 // Remove any comments; IE gets token splitting wrong
 607                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 608
 609                 $value = $stripped;
 610
 611                 // ... and continue checks
 612                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 613                         'codepointToUtf8(hexdec("$1"))', $stripped );
 614                 $stripped = str_replace( '\\', '', $stripped );
 615                 if( preg_match( '/(expression|tps*:\/\/|url\\s*\().*/is',
 616                                 $stripped ) ) {
 617                         # haxx0r
 618                         return false;
 619                 }
 620
 621                 return $value;
 622         }
 623
 624         /**
 625          * Take a tag soup fragment listing an HTML element's attributes
 626          * and normalize it to well-formed XML, discarding unwanted attributes.
 627          * Output is safe for further wikitext processing, with escaping of
 628          * values that could trigger problems.
 629          *
 630          * - Normalizes attribute names to lowercase
 631          * - Discards attributes not on a whitelist for the given element
 632          * - Turns broken or invalid entities into plaintext
 633          * - Double-quotes all attribute values
 634          * - Attributes without values are given the name as attribute
 635          * - Double attributes are discarded
 636          * - Unsafe style attributes are discarded
 637          * - Prepends space if there are attributes.
 638          *
 639          * @param string $text
 640          * @param string $element
 641          * @return string
 642          */
 643         static function fixTagAttributes( $text, $element ) {
 644                 if( trim( $text ) == '' ) {
 645                         return '';
 646                 }
 647
 648                 $stripped = Sanitizer::validateTagAttributes(
 649                         Sanitizer::decodeTagAttributes( $text ), $element );
 650
 651                 $attribs = array();
 652                 foreach( $stripped as $attribute => $value ) {
 653                         $encAttribute = htmlspecialchars( $attribute );
 654                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 655
 656                         $attribs[] = "$encAttribute=\"$encValue\"";
 657                 }
 658                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 659         }
 660
 661         /**
 662          * Encode an attribute value for HTML output.
 663          * @param $text
 664          * @return HTML-encoded text fragment
 665          */
 666         static function encodeAttribute( $text ) {
 667                 $encValue = htmlspecialchars( $text );
 668
 669                 // Whitespace is normalized during attribute decoding,
 670                 // so if we've been passed non-spaces we must encode them
 671                 // ahead of time or they won't be preserved.
 672                 $encValue = strtr( $encValue, array(
 673                         "\n" => '&#10;',
 674                         "\r" => '&#13;',
 675                         "\t" => '&#9;',
 676                 ) );
 677
 678                 return $encValue;
 679         }
 680
 681         /**
 682          * Encode an attribute value for HTML tags, with extra armoring
 683          * against further wiki processing.
 684          * @param $text
 685          * @return HTML-encoded text fragment
 686          */
 687         static function safeEncodeAttribute( $text ) {
 688                 $encValue = Sanitizer::encodeAttribute( $text );
 689
 690                 # Templates and links may be expanded in later parsing,
 691                 # creating invalid or dangerous output. Suppress this.
 692                 $encValue = strtr( $encValue, array(
 693                         '<'    => '&lt;',   // This should never happen,
 694                         '>'    => '&gt;',   // we've received invalid input
 695                         '"'    => '&quot;', // which should have been escaped.
 696                         '{'    => '&#123;',
 697                         '['    => '&#91;',
 698                         "''"   => '&#39;&#39;',
 699                         'ISBN' => '&#73;SBN',
 700                         'RFC'  => '&#82;FC',
 701                         'PMID' => '&#80;MID',
 702                         '|'    => '&#124;',
 703                         '__'   => '&#95;_',
 704                 ) );
 705
 706                 # Stupid hack
 707                 $encValue = preg_replace_callback(
 708                         '/(' . wfUrlProtocols() . ')/',
 709                         array( 'Sanitizer', 'armorLinksCallback' ),
 710                         $encValue );
 711                 return $encValue;
 712         }
 713
 714         /**
 715          * Given a value escape it so that it can be used in an id attribute and
 716          * return it, this does not validate the value however (see first link)
 717          *
 718          * @link http://www.w3.org/TR/html401/types.html#type-name Valid characters
 719          *                                                          in the id and
 720          *                                                          name attributes
 721          * @link http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 722          *
 723          * @bug 4461
 724          *
 725          * @static
 726          *
 727          * @param string $id
 728          * @return string
 729          */
 730         static function escapeId( $id ) {
 731                 static $replace = array(
 732                         '%3A' => ':',
 733                         '%' => '.'
 734                 );
 735
 736                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 737
 738                 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
 739         }
 740
 741         /**
 742          * Given a value, escape it so that it can be used as a CSS class and
 743          * return it.
 744          *
 745          * TODO: For extra validity, input should be validated UTF-8.
 746          *
 747          * @link http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 748          *
 749          * @param string $class
 750          * @return string
 751          */
 752         static function escapeClass( $class ) {
 753                 // Convert ugly stuff to underscores and kill underscores in ugly places
 754                 return rtrim(preg_replace(
 755                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 756                         '_',
 757                         $class ), '_');
 758         }
 759
 760         /**
 761          * Regex replace callback for armoring links against further processing.
 762          * @param array $matches
 763          * @return string
 764          * @private
 765          */
 766         private static function armorLinksCallback( $matches ) {
 767                 return str_replace( ':', '&#58;', $matches[1] );
 768         }
 769
 770         /**
 771          * Return an associative array of attribute names and values from
 772          * a partial tag string. Attribute names are forces to lowercase,
 773          * character references are decoded to UTF-8 text.
 774          *
 775          * @param string
 776          * @return array
 777          */
 778         static function decodeTagAttributes( $text ) {
 779                 $attribs = array();
 780
 781                 if( trim( $text ) == '' ) {
 782                         return $attribs;
 783                 }
 784
 785                 $pairs = array();
 786                 if( !preg_match_all(
 787                         MW_ATTRIBS_REGEX,
 788                         $text,
 789                         $pairs,
 790                         PREG_SET_ORDER ) ) {
 791                         return $attribs;
 792                 }
 793
 794                 foreach( $pairs as $set ) {
 795                         $attribute = strtolower( $set[1] );
 796                         $value = Sanitizer::getTagAttributeCallback( $set );
 797
 798                         // Normalize whitespace
 799                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 800                         $value = trim( $value );
 801
 802                         // Decode character references
 803                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 804                 }
 805                 return $attribs;
 806         }
 807
 808         /**
 809          * Pick the appropriate attribute value from a match set from the
 810          * MW_ATTRIBS_REGEX matches.
 811          *
 812          * @param array $set
 813          * @return string
 814          * @private
 815          */
 816         private static function getTagAttributeCallback( $set ) {
 817                 if( isset( $set[6] ) ) {
 818                         # Illegal #XXXXXX color with no quotes.
 819                         return $set[6];
 820                 } elseif( isset( $set[5] ) ) {
 821                         # No quotes.
 822                         return $set[5];
 823                 } elseif( isset( $set[4] ) ) {
 824                         # Single-quoted
 825                         return $set[4];
 826                 } elseif( isset( $set[3] ) ) {
 827                         # Double-quoted
 828                         return $set[3];
 829                 } elseif( !isset( $set[2] ) ) {
 830                         # In XHTML, attributes must have a value.
 831                         # For 'reduced' form, return explicitly the attribute name here.
 832                         return $set[1];
 833                 } else {
 834                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 835                 }
 836         }
 837
 838         /**
 839          * Normalize whitespace and character references in an XML source-
 840          * encoded text for an attribute value.
 841          *
 842          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 843          * but note that we're not returning the value, but are returning
 844          * XML source fragments that will be slapped into output.
 845          *
 846          * @param string $text
 847          * @return string
 848          * @private
 849          */
 850         private static function normalizeAttributeValue( $text ) {
 851                 return str_replace( '"', '&quot;',
 852                         preg_replace(
 853                                 '/\r\n|[\x20\x0d\x0a\x09]/',
 854                                 ' ',
 855                                 Sanitizer::normalizeCharReferences( $text ) ) );
 856         }
 857
 858         /**
 859          * Ensure that any entities and character references are legal
 860          * for XML and XHTML specifically. Any stray bits will be
 861          * &amp;-escaped to result in a valid text fragment.
 862          *
 863          * a. any named char refs must be known in XHTML
 864          * b. any numeric char refs must be legal chars, not invalid or forbidden
 865          * c. use &#x, not &#X
 866          * d. fix or reject non-valid attributes
 867          *
 868          * @param string $text
 869          * @return string
 870          * @private
 871          */
 872         static function normalizeCharReferences( $text ) {
 873                 return preg_replace_callback(
 874                         MW_CHAR_REFS_REGEX,
 875                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 876                         $text );
 877         }
 878         /**
 879          * @param string $matches
 880          * @return string
 881          */
 882         static function normalizeCharReferencesCallback( $matches ) {
 883                 $ret = null;
 884                 if( $matches[1] != '' ) {
 885                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 886                 } elseif( $matches[2] != '' ) {
 887                         $ret = Sanitizer::decCharReference( $matches[2] );
 888                 } elseif( $matches[3] != ''  ) {
 889                         $ret = Sanitizer::hexCharReference( $matches[3] );
 890                 } elseif( $matches[4] != '' ) {
 891                         $ret = Sanitizer::hexCharReference( $matches[4] );
 892                 }
 893                 if( is_null( $ret ) ) {
 894                         return htmlspecialchars( $matches[0] );
 895                 } else {
 896                         return $ret;
 897                 }
 898         }
 899
 900         /**
 901          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 902          * return the named entity reference as is. Otherwise, returns
 903          * HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 904          *
 905          * @param string $name
 906          * @return string
 907          * @static
 908          */
 909         static function normalizeEntity( $name ) {
 910                 global $wgHtmlEntities;
 911                 if( isset( $wgHtmlEntities[$name] ) ) {
 912                         return "&$name;";
 913                 } else {
 914                         return "&amp;$name;";
 915                 }
 916         }
 917
 918         static function decCharReference( $codepoint ) {
 919                 $point = intval( $codepoint );
 920                 if( Sanitizer::validateCodepoint( $point ) ) {
 921                         return sprintf( '&#%d;', $point );
 922                 } else {
 923                         return null;
 924                 }
 925         }
 926
 927         static function hexCharReference( $codepoint ) {
 928                 $point = hexdec( $codepoint );
 929                 if( Sanitizer::validateCodepoint( $point ) ) {
 930                         return sprintf( '&#x%x;', $point );
 931                 } else {
 932                         return null;
 933                 }
 934         }
 935
 936         /**
 937          * Returns true if a given Unicode codepoint is a valid character in XML.
 938          * @param int $codepoint
 939          * @return bool
 940          */
 941         private static function validateCodepoint( $codepoint ) {
 942                 return ($codepoint ==    0x09)
 943                         || ($codepoint ==    0x0a)
 944                         || ($codepoint ==    0x0d)
 945                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 946                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 947                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 948         }
 949
 950         /**
 951          * Decode any character references, numeric or named entities,
 952          * in the text and return a UTF-8 string.
 953          *
 954          * @param string $text
 955          * @return string
 956          * @public
 957          * @static
 958          */
 959         public static function decodeCharReferences( $text ) {
 960                 return preg_replace_callback(
 961                         MW_CHAR_REFS_REGEX,
 962                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 963                         $text );
 964         }
 965
 966         /**
 967          * @param string $matches
 968          * @return string
 969          */
 970         static function decodeCharReferencesCallback( $matches ) {
 971                 if( $matches[1] != '' ) {
 972                         return Sanitizer::decodeEntity( $matches[1] );
 973                 } elseif( $matches[2] != '' ) {
 974                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 975                 } elseif( $matches[3] != ''  ) {
 976                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 977                 } elseif( $matches[4] != '' ) {
 978                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 979                 }
 980                 # Last case should be an ampersand by itself
 981                 return $matches[0];
 982         }
 983
 984         /**
 985          * Return UTF-8 string for a codepoint if that is a valid
 986          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
 987          * @param int $codepoint
 988          * @return string
 989          * @private
 990          */
 991         static function decodeChar( $codepoint ) {
 992                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
 993                         return codepointToUtf8( $codepoint );
 994                 } else {
 995                         return UTF8_REPLACEMENT;
 996                 }
 997         }
 998
 999         /**
1000          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1001          * return the UTF-8 encoding of that character. Otherwise, returns
1002          * pseudo-entity source (eg &foo;)
1003          *
1004          * @param string $name
1005          * @return string
1006          */
1007         static function decodeEntity( $name ) {
1008                 global $wgHtmlEntities;
1009                 if( isset( $wgHtmlEntities[$name] ) ) {
1010                         return codepointToUtf8( $wgHtmlEntities[$name] );
1011                 } else {
1012                         return "&$name;";
1013                 }
1014         }
1015
1016         /**
1017          * Fetch the whitelist of acceptable attributes for a given
1018          * element name.
1019          *
1020          * @param string $element
1021          * @return array
1022          */
1023         static function attributeWhitelist( $element ) {
1024                 static $list;
1025                 if( !isset( $list ) ) {
1026                         $list = Sanitizer::setupAttributeWhitelist();
1027                 }
1028                 return isset( $list[$element] )
1029                         ? $list[$element]
1030                         : array();
1031         }
1032
1033         /**
1034          * @todo Document it a bit
1035          * @return array
1036          */
1037         static function setupAttributeWhitelist() {
1038                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1039                 $block = array_merge( $common, array( 'align' ) );
1040                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1041                 $tablecell = array( 'abbr',
1042                                     'axis',
1043                                     'headers',
1044                                     'scope',
1045                                     'rowspan',
1046                                     'colspan',
1047                                     'nowrap', # deprecated
1048                                     'width',  # deprecated
1049                                     'height', # deprecated
1050                                     'bgcolor' # deprecated
1051                                     );
1052
1053                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1054                 # See: http://www.w3.org/TR/html4/
1055                 $whitelist = array (
1056                         # 7.5.4
1057                         'div'        => $block,
1058                         'center'     => $common, # deprecated
1059                         'span'       => $block, # ??
1060
1061                         # 7.5.5
1062                         'h1'         => $block,
1063                         'h2'         => $block,
1064                         'h3'         => $block,
1065                         'h4'         => $block,
1066                         'h5'         => $block,
1067                         'h6'         => $block,
1068
1069                         # 7.5.6
1070                         # address
1071
1072                         # 8.2.4
1073                         # bdo
1074
1075                         # 9.2.1
1076                         'em'         => $common,
1077                         'strong'     => $common,
1078                         'cite'       => $common,
1079                         # dfn
1080                         'code'       => $common,
1081                         # samp
1082                         # kbd
1083                         'var'        => $common,
1084                         # abbr
1085                         # acronym
1086
1087                         # 9.2.2
1088                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1089                         # q
1090
1091                         # 9.2.3
1092                         'sub'        => $common,
1093                         'sup'        => $common,
1094
1095                         # 9.3.1
1096                         'p'          => $block,
1097
1098                         # 9.3.2
1099                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1100
1101                         # 9.3.4
1102                         'pre'        => array_merge( $common, array( 'width' ) ),
1103
1104                         # 9.4
1105                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1106                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1107
1108                         # 10.2
1109                         'ul'         => array_merge( $common, array( 'type' ) ),
1110                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1111                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1112
1113                         # 10.3
1114                         'dl'         => $common,
1115                         'dd'         => $common,
1116                         'dt'         => $common,
1117
1118                         # 11.2.1
1119                         'table'      => array_merge( $common,
1120                                                                 array( 'summary', 'width', 'border', 'frame',
1121                                                                                 'rules', 'cellspacing', 'cellpadding',
1122                                                                                 'align', 'bgcolor',
1123                                                                 ) ),
1124
1125                         # 11.2.2
1126                         'caption'    => array_merge( $common, array( 'align' ) ),
1127
1128                         # 11.2.3
1129                         'thead'      => array_merge( $common, $tablealign ),
1130                         'tfoot'      => array_merge( $common, $tablealign ),
1131                         'tbody'      => array_merge( $common, $tablealign ),
1132
1133                         # 11.2.4
1134                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1135                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1136
1137                         # 11.2.5
1138                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1139
1140                         # 11.2.6
1141                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1142                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1143
1144                         # 15.2.1
1145                         'tt'         => $common,
1146                         'b'          => $common,
1147                         'i'          => $common,
1148                         'big'        => $common,
1149                         'small'      => $common,
1150                         'strike'     => $common,
1151                         's'          => $common,
1152                         'u'          => $common,
1153
1154                         # 15.2.2
1155                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1156                         # basefont
1157
1158                         # 15.3
1159                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1160
1161                         # XHTML Ruby annotation text module, simple ruby only.
1162                         # http://www.w3c.org/TR/ruby/
1163                         'ruby'       => $common,
1164                         # rbc
1165                         # rtc
1166                         'rb'         => $common,
1167                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1168                         'rp'         => $common,
1169                         );
1170                 return $whitelist;
1171         }
1172
1173         /**
1174          * Take a fragment of (potentially invalid) HTML and return
1175          * a version with any tags removed, encoded suitably for literal
1176          * inclusion in an attribute value.
1177          *
1178          * @param string $text HTML fragment
1179          * @return string
1180          */
1181         static function stripAllTags( $text ) {
1182                 # Actual <tags>
1183                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1184
1185                 # Normalize &entities and whitespace
1186                 $text = Sanitizer::normalizeAttributeValue( $text );
1187
1188                 # Will be placed into "double-quoted" attributes,
1189                 # make sure remaining bits are safe.
1190                 $text = str_replace(
1191                         array('<', '>', '"'),
1192                         array('&lt;', '&gt;', '&quot;'),
1193                         $text );
1194
1195                 return $text;
1196         }
1197
1198         /**
1199          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1200          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1201          * PHP 5.1 doesn't.
1202          *
1203          * Use for passing XHTML fragments to PHP's XML parsing functions
1204          *
1205          * @return string
1206          * @static
1207          */
1208         static function hackDocType() {
1209                 global $wgHtmlEntities;
1210                 $out = "<!DOCTYPE html [\n";
1211                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1212                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1213                 }
1214                 $out .= "]>\n";
1215                 return $out;
1216         }
1217
1218         static function cleanUrl( $url, $hostname=true ) {
1219                 # Normalize any HTML entities in input. They will be
1220                 # re-escaped by makeExternalLink().
1221                 $url = Sanitizer::decodeCharReferences( $url );
1222
1223                 # Escape any control characters introduced by the above step
1224                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1225
1226                 # Validate hostname portion
1227                 $matches = array();
1228                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1229                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1230
1231                         // Characters that will be ignored in IDNs.
1232                         // http://tools.ietf.org/html/3454#section-3.1
1233                         // Strip them before further processing so blacklists and such work.
1234                         $strip = "/
1235                                 \\s|          # general whitespace
1236                                 \xc2\xad|     # 00ad SOFT HYPHEN
1237                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1238                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1239                                 \xe2\x81\xa0| # 2060 WORD JOINER
1240                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1241                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1242                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1243                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1244                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1245                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1246                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1247                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1248                                 /xuD";
1249
1250                         $host = preg_replace( $strip, '', $host );
1251
1252                         // @fixme: validate hostnames here
1253
1254                         return $protocol . $host . $rest;
1255                 } else {
1256                         return $url;
1257                 }
1258         }
1259
1260 }
1261
1262 ?>