includes/Sanitizer.php

   1 <?php
   2 /**
   3  * XHTML sanitizer for MediaWiki
   4  *
   5  * Copyright (C) 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @addtogroup Parser
  24  */
  25
  26 /**
  27  * Regular expression to match various types of character references in
  28  * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  29  */
  30 define( 'MW_CHAR_REFS_REGEX',
  31         '/&([A-Za-z0-9]+);
  32          |&\#([0-9]+);
  33          |&\#x([0-9A-Za-z]+);
  34          |&\#X([0-9A-Za-z]+);
  35          |(&)/x' );
  36
  37 /**
  38  * Regular expression to match HTML/XML attribute pairs within a tag.
  39  * Allows some... latitude.
  40  * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
  41  */
  42 $attrib = '[A-Za-z0-9]';
  43 $space = '[\x09\x0a\x0d\x20]';
  44 define( 'MW_ATTRIBS_REGEX',
  45         "/(?:^|$space)($attrib+)
  46           ($space*=$space*
  47                 (?:
  48                  # The attribute value: quoted or alone
  49                   \"([^<\"]*)\"
  50                  | '([^<']*)'
  51                  |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
  52                  |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
  53                                                          # colors are specified like this.
  54                                                          # We'll be normalizing it.
  55                 )
  56            )?(?=$space|\$)/sx" );
  57
  58 /**
  59  * List of all named character entities defined in HTML 4.01
  60  * http://www.w3.org/TR/html4/sgml/entities.html
  61  * @private
  62  */
  63 global $wgHtmlEntities;
  64 $wgHtmlEntities = array(
  65         'Aacute'   => 193,
  66         'aacute'   => 225,
  67         'Acirc'    => 194,
  68         'acirc'    => 226,
  69         'acute'    => 180,
  70         'AElig'    => 198,
  71         'aelig'    => 230,
  72         'Agrave'   => 192,
  73         'agrave'   => 224,
  74         'alefsym'  => 8501,
  75         'Alpha'    => 913,
  76         'alpha'    => 945,
  77         'amp'      => 38,
  78         'and'      => 8743,
  79         'ang'      => 8736,
  80         'Aring'    => 197,
  81         'aring'    => 229,
  82         'asymp'    => 8776,
  83         'Atilde'   => 195,
  84         'atilde'   => 227,
  85         'Auml'     => 196,
  86         'auml'     => 228,
  87         'bdquo'    => 8222,
  88         'Beta'     => 914,
  89         'beta'     => 946,
  90         'brvbar'   => 166,
  91         'bull'     => 8226,
  92         'cap'      => 8745,
  93         'Ccedil'   => 199,
  94         'ccedil'   => 231,
  95         'cedil'    => 184,
  96         'cent'     => 162,
  97         'Chi'      => 935,
  98         'chi'      => 967,
  99         'circ'     => 710,
 100         'clubs'    => 9827,
 101         'cong'     => 8773,
 102         'copy'     => 169,
 103         'crarr'    => 8629,
 104         'cup'      => 8746,
 105         'curren'   => 164,
 106         'dagger'   => 8224,
 107         'Dagger'   => 8225,
 108         'darr'     => 8595,
 109         'dArr'     => 8659,
 110         'deg'      => 176,
 111         'Delta'    => 916,
 112         'delta'    => 948,
 113         'diams'    => 9830,
 114         'divide'   => 247,
 115         'Eacute'   => 201,
 116         'eacute'   => 233,
 117         'Ecirc'    => 202,
 118         'ecirc'    => 234,
 119         'Egrave'   => 200,
 120         'egrave'   => 232,
 121         'empty'    => 8709,
 122         'emsp'     => 8195,
 123         'ensp'     => 8194,
 124         'Epsilon'  => 917,
 125         'epsilon'  => 949,
 126         'equiv'    => 8801,
 127         'Eta'      => 919,
 128         'eta'      => 951,
 129         'ETH'      => 208,
 130         'eth'      => 240,
 131         'Euml'     => 203,
 132         'euml'     => 235,
 133         'euro'     => 8364,
 134         'exist'    => 8707,
 135         'fnof'     => 402,
 136         'forall'   => 8704,
 137         'frac12'   => 189,
 138         'frac14'   => 188,
 139         'frac34'   => 190,
 140         'frasl'    => 8260,
 141         'Gamma'    => 915,
 142         'gamma'    => 947,
 143         'ge'       => 8805,
 144         'gt'       => 62,
 145         'harr'     => 8596,
 146         'hArr'     => 8660,
 147         'hearts'   => 9829,
 148         'hellip'   => 8230,
 149         'Iacute'   => 205,
 150         'iacute'   => 237,
 151         'Icirc'    => 206,
 152         'icirc'    => 238,
 153         'iexcl'    => 161,
 154         'Igrave'   => 204,
 155         'igrave'   => 236,
 156         'image'    => 8465,
 157         'infin'    => 8734,
 158         'int'      => 8747,
 159         'Iota'     => 921,
 160         'iota'     => 953,
 161         'iquest'   => 191,
 162         'isin'     => 8712,
 163         'Iuml'     => 207,
 164         'iuml'     => 239,
 165         'Kappa'    => 922,
 166         'kappa'    => 954,
 167         'Lambda'   => 923,
 168         'lambda'   => 955,
 169         'lang'     => 9001,
 170         'laquo'    => 171,
 171         'larr'     => 8592,
 172         'lArr'     => 8656,
 173         'lceil'    => 8968,
 174         'ldquo'    => 8220,
 175         'le'       => 8804,
 176         'lfloor'   => 8970,
 177         'lowast'   => 8727,
 178         'loz'      => 9674,
 179         'lrm'      => 8206,
 180         'lsaquo'   => 8249,
 181         'lsquo'    => 8216,
 182         'lt'       => 60,
 183         'macr'     => 175,
 184         'mdash'    => 8212,
 185         'micro'    => 181,
 186         'middot'   => 183,
 187         'minus'    => 8722,
 188         'Mu'       => 924,
 189         'mu'       => 956,
 190         'nabla'    => 8711,
 191         'nbsp'     => 160,
 192         'ndash'    => 8211,
 193         'ne'       => 8800,
 194         'ni'       => 8715,
 195         'not'      => 172,
 196         'notin'    => 8713,
 197         'nsub'     => 8836,
 198         'Ntilde'   => 209,
 199         'ntilde'   => 241,
 200         'Nu'       => 925,
 201         'nu'       => 957,
 202         'Oacute'   => 211,
 203         'oacute'   => 243,
 204         'Ocirc'    => 212,
 205         'ocirc'    => 244,
 206         'OElig'    => 338,
 207         'oelig'    => 339,
 208         'Ograve'   => 210,
 209         'ograve'   => 242,
 210         'oline'    => 8254,
 211         'Omega'    => 937,
 212         'omega'    => 969,
 213         'Omicron'  => 927,
 214         'omicron'  => 959,
 215         'oplus'    => 8853,
 216         'or'       => 8744,
 217         'ordf'     => 170,
 218         'ordm'     => 186,
 219         'Oslash'   => 216,
 220         'oslash'   => 248,
 221         'Otilde'   => 213,
 222         'otilde'   => 245,
 223         'otimes'   => 8855,
 224         'Ouml'     => 214,
 225         'ouml'     => 246,
 226         'para'     => 182,
 227         'part'     => 8706,
 228         'permil'   => 8240,
 229         'perp'     => 8869,
 230         'Phi'      => 934,
 231         'phi'      => 966,
 232         'Pi'       => 928,
 233         'pi'       => 960,
 234         'piv'      => 982,
 235         'plusmn'   => 177,
 236         'pound'    => 163,
 237         'prime'    => 8242,
 238         'Prime'    => 8243,
 239         'prod'     => 8719,
 240         'prop'     => 8733,
 241         'Psi'      => 936,
 242         'psi'      => 968,
 243         'quot'     => 34,
 244         'radic'    => 8730,
 245         'rang'     => 9002,
 246         'raquo'    => 187,
 247         'rarr'     => 8594,
 248         'rArr'     => 8658,
 249         'rceil'    => 8969,
 250         'rdquo'    => 8221,
 251         'real'     => 8476,
 252         'reg'      => 174,
 253         'rfloor'   => 8971,
 254         'Rho'      => 929,
 255         'rho'      => 961,
 256         'rlm'      => 8207,
 257         'rsaquo'   => 8250,
 258         'rsquo'    => 8217,
 259         'sbquo'    => 8218,
 260         'Scaron'   => 352,
 261         'scaron'   => 353,
 262         'sdot'     => 8901,
 263         'sect'     => 167,
 264         'shy'      => 173,
 265         'Sigma'    => 931,
 266         'sigma'    => 963,
 267         'sigmaf'   => 962,
 268         'sim'      => 8764,
 269         'spades'   => 9824,
 270         'sub'      => 8834,
 271         'sube'     => 8838,
 272         'sum'      => 8721,
 273         'sup'      => 8835,
 274         'sup1'     => 185,
 275         'sup2'     => 178,
 276         'sup3'     => 179,
 277         'supe'     => 8839,
 278         'szlig'    => 223,
 279         'Tau'      => 932,
 280         'tau'      => 964,
 281         'there4'   => 8756,
 282         'Theta'    => 920,
 283         'theta'    => 952,
 284         'thetasym' => 977,
 285         'thinsp'   => 8201,
 286         'THORN'    => 222,
 287         'thorn'    => 254,
 288         'tilde'    => 732,
 289         'times'    => 215,
 290         'trade'    => 8482,
 291         'Uacute'   => 218,
 292         'uacute'   => 250,
 293         'uarr'     => 8593,
 294         'uArr'     => 8657,
 295         'Ucirc'    => 219,
 296         'ucirc'    => 251,
 297         'Ugrave'   => 217,
 298         'ugrave'   => 249,
 299         'uml'      => 168,
 300         'upsih'    => 978,
 301         'Upsilon'  => 933,
 302         'upsilon'  => 965,
 303         'Uuml'     => 220,
 304         'uuml'     => 252,
 305         'weierp'   => 8472,
 306         'Xi'       => 926,
 307         'xi'       => 958,
 308         'Yacute'   => 221,
 309         'yacute'   => 253,
 310         'yen'      => 165,
 311         'Yuml'     => 376,
 312         'yuml'     => 255,
 313         'Zeta'     => 918,
 314         'zeta'     => 950,
 315         'zwj'      => 8205,
 316         'zwnj'     => 8204 );
 317
 318 /**
 319  * XHTML sanitizer for MediaWiki
 320  * @addtogroup Parser
 321  */
 322 class Sanitizer {
 323         /**
 324          * Cleans up HTML, removes dangerous tags and attributes, and
 325          * removes HTML comments
 326          * @private
 327          * @param string $text
 328          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 329          * @param array $args for the processing callback
 330          * @return string
 331          */
 332         static function removeHTMLtags( $text, $processCallback = null, $args = array() ) {
 333                 global $wgUseTidy;
 334
 335                 static $htmlpairs, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 336                         $htmllist, $listtags, $htmlsingleallowed, $htmlelements, $staticInitialised;
 337
 338                 wfProfileIn( __METHOD__ );
 339
 340                 if ( !$staticInitialised ) {
 341
 342                         $htmlpairs = array( # Tags that must be closed
 343                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 344                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 345                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 346                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 347                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
 348                         );
 349                         $htmlsingle = array(
 350                                 'br', 'hr', 'li', 'dt', 'dd'
 351                         );
 352                         $htmlsingleonly = array( # Elements that cannot have close tags
 353                                 'br', 'hr'
 354                         );
 355                         $htmlnest = array( # Tags that can be nested--??
 356                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 357                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 358                         );
 359                         $tabletags = array( # Can only appear inside table, we will close them
 360                                 'td', 'th', 'tr',
 361                         );
 362                         $htmllist = array( # Tags used by list
 363                                 'ul','ol',
 364                         );
 365                         $listtags = array( # Tags that can appear in a list
 366                                 'li',
 367                         );
 368
 369                         $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
 370                         $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
 371
 372                         # Convert them all to hashtables for faster lookup
 373                         $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 374                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
 375                         foreach ( $vars as $var ) {
 376                                 $$var = array_flip( $$var );
 377                         }
 378                         $staticInitialised = true;
 379                 }
 380
 381                 # Remove HTML comments
 382                 $text = Sanitizer::removeHTMLcomments( $text );
 383                 $bits = explode( '<', $text );
 384                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 385                 if(!$wgUseTidy) {
 386                         $tagstack = $tablestack = array();
 387                         foreach ( $bits as $x ) {
 388                                 $regs = array();
 389                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 390                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 391                                 } else {
 392                                         $slash = $t = $params = $brace = $rest = null;
 393                                 }
 394
 395                                 $badtag = 0 ;
 396                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 397                                         # Check our stack
 398                                         if ( $slash ) {
 399                                                 # Closing a tag...
 400                                                 if( isset( $htmlsingleonly[$t] ) ) {
 401                                                         $badtag = 1;
 402                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 403                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 404                                                                 # Pop all elements with an optional close tag
 405                                                                 # and see if we find a match below them
 406                                                                 $optstack = array();
 407                                                                 array_push ($optstack, $ot);
 408                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 409                                                                                 isset( $htmlsingleallowed[$ot] ) )
 410                                                                 {
 411                                                                         array_push ($optstack, $ot);
 412                                                                 }
 413                                                                 if ( $t != $ot ) {
 414                                                                         # No match. Push the optinal elements back again
 415                                                                         $badtag = 1;
 416                                                                         while ( $ot = @array_pop( $optstack ) ) {
 417                                                                                 array_push( $tagstack, $ot );
 418                                                                         }
 419                                                                 }
 420                                                         } else {
 421                                                                 @array_push( $tagstack, $ot );
 422                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 423                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 424                                                                         $badtag = 1;
 425                                                                 }
 426                                                         }
 427                                                 } else {
 428                                                         if ( $t == 'table' ) {
 429                                                                 $tagstack = array_pop( $tablestack );
 430                                                         }
 431                                                 }
 432                                                 $newparams = '';
 433                                         } else {
 434                                                 # Keep track for later
 435                                                 if ( isset( $tabletags[$t] ) &&
 436                                                 ! in_array( 'table', $tagstack ) ) {
 437                                                         $badtag = 1;
 438                                                 } else if ( in_array( $t, $tagstack ) &&
 439                                                 ! isset( $htmlnest [$t ] ) ) {
 440                                                         $badtag = 1 ;
 441                                                 # Is it a self closed htmlpair ? (bug 5487)
 442                                                 } else if( $brace == '/>' &&
 443                                                 isset( $htmlpairs[$t] ) ) {
 444                                                         $badtag = 1;
 445                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 446                                                         # Hack to force empty tag for uncloseable elements
 447                                                         $brace = '/>';
 448                                                 } else if( isset( $htmlsingle[$t] ) ) {
 449                                                         # Hack to not close $htmlsingle tags
 450                                                         $brace = NULL;
 451                                                 } else if( isset( $tabletags[$t] )
 452                                                 &&  in_array($t ,$tagstack) ) {
 453                                                         // New table tag but forgot to close the previous one
 454                                                         $text .= "</$t>";
 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 = ( $brace == '/>' && !$slash ) ? ' /' : '';
 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          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 719          *                                                          in the id and
 720          *                                                          name attributes
 721          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 722          *
 723          * @static
 724          *
 725          * @param string $id
 726          * @return string
 727          */
 728         static function escapeId( $id ) {
 729                 static $replace = array(
 730                         '%3A' => ':',
 731                         '%' => '.'
 732                 );
 733
 734                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 735
 736                 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
 737         }
 738
 739         /**
 740          * Given a value, escape it so that it can be used as a CSS class and
 741          * return it.
 742          *
 743          * @todo For extra validity, input should be validated UTF-8.
 744          *
 745          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 746          *
 747          * @param string $class
 748          * @return string
 749          */
 750         static function escapeClass( $class ) {
 751                 // Convert ugly stuff to underscores and kill underscores in ugly places
 752                 return rtrim(preg_replace(
 753                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 754                         '_',
 755                         $class ), '_');
 756         }
 757
 758         /**
 759          * Regex replace callback for armoring links against further processing.
 760          * @param array $matches
 761          * @return string
 762          * @private
 763          */
 764         private static function armorLinksCallback( $matches ) {
 765                 return str_replace( ':', '&#58;', $matches[1] );
 766         }
 767
 768         /**
 769          * Return an associative array of attribute names and values from
 770          * a partial tag string. Attribute names are forces to lowercase,
 771          * character references are decoded to UTF-8 text.
 772          *
 773          * @param string
 774          * @return array
 775          */
 776         static function decodeTagAttributes( $text ) {
 777                 $attribs = array();
 778
 779                 if( trim( $text ) == '' ) {
 780                         return $attribs;
 781                 }
 782
 783                 $pairs = array();
 784                 if( !preg_match_all(
 785                         MW_ATTRIBS_REGEX,
 786                         $text,
 787                         $pairs,
 788                         PREG_SET_ORDER ) ) {
 789                         return $attribs;
 790                 }
 791
 792                 foreach( $pairs as $set ) {
 793                         $attribute = strtolower( $set[1] );
 794                         $value = Sanitizer::getTagAttributeCallback( $set );
 795
 796                         // Normalize whitespace
 797                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 798                         $value = trim( $value );
 799
 800                         // Decode character references
 801                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 802                 }
 803                 return $attribs;
 804         }
 805
 806         /**
 807          * Pick the appropriate attribute value from a match set from the
 808          * MW_ATTRIBS_REGEX matches.
 809          *
 810          * @param array $set
 811          * @return string
 812          * @private
 813          */
 814         private static function getTagAttributeCallback( $set ) {
 815                 if( isset( $set[6] ) ) {
 816                         # Illegal #XXXXXX color with no quotes.
 817                         return $set[6];
 818                 } elseif( isset( $set[5] ) ) {
 819                         # No quotes.
 820                         return $set[5];
 821                 } elseif( isset( $set[4] ) ) {
 822                         # Single-quoted
 823                         return $set[4];
 824                 } elseif( isset( $set[3] ) ) {
 825                         # Double-quoted
 826                         return $set[3];
 827                 } elseif( !isset( $set[2] ) ) {
 828                         # In XHTML, attributes must have a value.
 829                         # For 'reduced' form, return explicitly the attribute name here.
 830                         return $set[1];
 831                 } else {
 832                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 833                 }
 834         }
 835
 836         /**
 837          * Normalize whitespace and character references in an XML source-
 838          * encoded text for an attribute value.
 839          *
 840          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 841          * but note that we're not returning the value, but are returning
 842          * XML source fragments that will be slapped into output.
 843          *
 844          * @param string $text
 845          * @return string
 846          * @private
 847          */
 848         private static function normalizeAttributeValue( $text ) {
 849                 return str_replace( '"', '&quot;',
 850                         preg_replace(
 851                                 '/\r\n|[\x20\x0d\x0a\x09]/',
 852                                 ' ',
 853                                 Sanitizer::normalizeCharReferences( $text ) ) );
 854         }
 855
 856         /**
 857          * Ensure that any entities and character references are legal
 858          * for XML and XHTML specifically. Any stray bits will be
 859          * &amp;-escaped to result in a valid text fragment.
 860          *
 861          * a. any named char refs must be known in XHTML
 862          * b. any numeric char refs must be legal chars, not invalid or forbidden
 863          * c. use &#x, not &#X
 864          * d. fix or reject non-valid attributes
 865          *
 866          * @param string $text
 867          * @return string
 868          * @private
 869          */
 870         static function normalizeCharReferences( $text ) {
 871                 return preg_replace_callback(
 872                         MW_CHAR_REFS_REGEX,
 873                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 874                         $text );
 875         }
 876         /**
 877          * @param string $matches
 878          * @return string
 879          */
 880         static function normalizeCharReferencesCallback( $matches ) {
 881                 $ret = null;
 882                 if( $matches[1] != '' ) {
 883                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 884                 } elseif( $matches[2] != '' ) {
 885                         $ret = Sanitizer::decCharReference( $matches[2] );
 886                 } elseif( $matches[3] != ''  ) {
 887                         $ret = Sanitizer::hexCharReference( $matches[3] );
 888                 } elseif( $matches[4] != '' ) {
 889                         $ret = Sanitizer::hexCharReference( $matches[4] );
 890                 }
 891                 if( is_null( $ret ) ) {
 892                         return htmlspecialchars( $matches[0] );
 893                 } else {
 894                         return $ret;
 895                 }
 896         }
 897
 898         /**
 899          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 900          * return the named entity reference as is. Otherwise, returns
 901          * HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 902          *
 903          * @param string $name
 904          * @return string
 905          * @static
 906          */
 907         static function normalizeEntity( $name ) {
 908                 global $wgHtmlEntities;
 909                 if( isset( $wgHtmlEntities[$name] ) ) {
 910                         return "&$name;";
 911                 } else {
 912                         return "&amp;$name;";
 913                 }
 914         }
 915
 916         static function decCharReference( $codepoint ) {
 917                 $point = intval( $codepoint );
 918                 if( Sanitizer::validateCodepoint( $point ) ) {
 919                         return sprintf( '&#%d;', $point );
 920                 } else {
 921                         return null;
 922                 }
 923         }
 924
 925         static function hexCharReference( $codepoint ) {
 926                 $point = hexdec( $codepoint );
 927                 if( Sanitizer::validateCodepoint( $point ) ) {
 928                         return sprintf( '&#x%x;', $point );
 929                 } else {
 930                         return null;
 931                 }
 932         }
 933
 934         /**
 935          * Returns true if a given Unicode codepoint is a valid character in XML.
 936          * @param int $codepoint
 937          * @return bool
 938          */
 939         private static function validateCodepoint( $codepoint ) {
 940                 return ($codepoint ==    0x09)
 941                         || ($codepoint ==    0x0a)
 942                         || ($codepoint ==    0x0d)
 943                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 944                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 945                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 946         }
 947
 948         /**
 949          * Decode any character references, numeric or named entities,
 950          * in the text and return a UTF-8 string.
 951          *
 952          * @param string $text
 953          * @return string
 954          * @public
 955          * @static
 956          */
 957         public static function decodeCharReferences( $text ) {
 958                 return preg_replace_callback(
 959                         MW_CHAR_REFS_REGEX,
 960                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 961                         $text );
 962         }
 963
 964         /**
 965          * @param string $matches
 966          * @return string
 967          */
 968         static function decodeCharReferencesCallback( $matches ) {
 969                 if( $matches[1] != '' ) {
 970                         return Sanitizer::decodeEntity( $matches[1] );
 971                 } elseif( $matches[2] != '' ) {
 972                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 973                 } elseif( $matches[3] != ''  ) {
 974                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 975                 } elseif( $matches[4] != '' ) {
 976                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 977                 }
 978                 # Last case should be an ampersand by itself
 979                 return $matches[0];
 980         }
 981
 982         /**
 983          * Return UTF-8 string for a codepoint if that is a valid
 984          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
 985          * @param int $codepoint
 986          * @return string
 987          * @private
 988          */
 989         static function decodeChar( $codepoint ) {
 990                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
 991                         return codepointToUtf8( $codepoint );
 992                 } else {
 993                         return UTF8_REPLACEMENT;
 994                 }
 995         }
 996
 997         /**
 998          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 999          * return the UTF-8 encoding of that character. Otherwise, returns
1000          * pseudo-entity source (eg &foo;)
1001          *
1002          * @param string $name
1003          * @return string
1004          */
1005         static function decodeEntity( $name ) {
1006                 global $wgHtmlEntities;
1007                 if( isset( $wgHtmlEntities[$name] ) ) {
1008                         return codepointToUtf8( $wgHtmlEntities[$name] );
1009                 } else {
1010                         return "&$name;";
1011                 }
1012         }
1013
1014         /**
1015          * Fetch the whitelist of acceptable attributes for a given
1016          * element name.
1017          *
1018          * @param string $element
1019          * @return array
1020          */
1021         static function attributeWhitelist( $element ) {
1022                 static $list;
1023                 if( !isset( $list ) ) {
1024                         $list = Sanitizer::setupAttributeWhitelist();
1025                 }
1026                 return isset( $list[$element] )
1027                         ? $list[$element]
1028                         : array();
1029         }
1030
1031         /**
1032          * @todo Document it a bit
1033          * @return array
1034          */
1035         static function setupAttributeWhitelist() {
1036                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1037                 $block = array_merge( $common, array( 'align' ) );
1038                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1039                 $tablecell = array( 'abbr',
1040                                     'axis',
1041                                     'headers',
1042                                     'scope',
1043                                     'rowspan',
1044                                     'colspan',
1045                                     'nowrap', # deprecated
1046                                     'width',  # deprecated
1047                                     'height', # deprecated
1048                                     'bgcolor' # deprecated
1049                                     );
1050
1051                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1052                 # See: http://www.w3.org/TR/html4/
1053                 $whitelist = array (
1054                         # 7.5.4
1055                         'div'        => $block,
1056                         'center'     => $common, # deprecated
1057                         'span'       => $block, # ??
1058
1059                         # 7.5.5
1060                         'h1'         => $block,
1061                         'h2'         => $block,
1062                         'h3'         => $block,
1063                         'h4'         => $block,
1064                         'h5'         => $block,
1065                         'h6'         => $block,
1066
1067                         # 7.5.6
1068                         # address
1069
1070                         # 8.2.4
1071                         # bdo
1072
1073                         # 9.2.1
1074                         'em'         => $common,
1075                         'strong'     => $common,
1076                         'cite'       => $common,
1077                         # dfn
1078                         'code'       => $common,
1079                         # samp
1080                         # kbd
1081                         'var'        => $common,
1082                         # abbr
1083                         # acronym
1084
1085                         # 9.2.2
1086                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1087                         # q
1088
1089                         # 9.2.3
1090                         'sub'        => $common,
1091                         'sup'        => $common,
1092
1093                         # 9.3.1
1094                         'p'          => $block,
1095
1096                         # 9.3.2
1097                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1098
1099                         # 9.3.4
1100                         'pre'        => array_merge( $common, array( 'width' ) ),
1101
1102                         # 9.4
1103                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1104                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1105
1106                         # 10.2
1107                         'ul'         => array_merge( $common, array( 'type' ) ),
1108                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1109                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1110
1111                         # 10.3
1112                         'dl'         => $common,
1113                         'dd'         => $common,
1114                         'dt'         => $common,
1115
1116                         # 11.2.1
1117                         'table'      => array_merge( $common,
1118                                                                 array( 'summary', 'width', 'border', 'frame',
1119                                                                                 'rules', 'cellspacing', 'cellpadding',
1120                                                                                 'align', 'bgcolor',
1121                                                                 ) ),
1122
1123                         # 11.2.2
1124                         'caption'    => array_merge( $common, array( 'align' ) ),
1125
1126                         # 11.2.3
1127                         'thead'      => array_merge( $common, $tablealign ),
1128                         'tfoot'      => array_merge( $common, $tablealign ),
1129                         'tbody'      => array_merge( $common, $tablealign ),
1130
1131                         # 11.2.4
1132                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1133                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1134
1135                         # 11.2.5
1136                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1137
1138                         # 11.2.6
1139                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1140                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1141
1142                         # 15.2.1
1143                         'tt'         => $common,
1144                         'b'          => $common,
1145                         'i'          => $common,
1146                         'big'        => $common,
1147                         'small'      => $common,
1148                         'strike'     => $common,
1149                         's'          => $common,
1150                         'u'          => $common,
1151
1152                         # 15.2.2
1153                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1154                         # basefont
1155
1156                         # 15.3
1157                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1158
1159                         # XHTML Ruby annotation text module, simple ruby only.
1160                         # http://www.w3c.org/TR/ruby/
1161                         'ruby'       => $common,
1162                         # rbc
1163                         # rtc
1164                         'rb'         => $common,
1165                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1166                         'rp'         => $common,
1167                         );
1168                 return $whitelist;
1169         }
1170
1171         /**
1172          * Take a fragment of (potentially invalid) HTML and return
1173          * a version with any tags removed, encoded suitably for literal
1174          * inclusion in an attribute value.
1175          *
1176          * @param string $text HTML fragment
1177          * @return string
1178          */
1179         static function stripAllTags( $text ) {
1180                 # Actual <tags>
1181                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1182
1183                 # Normalize &entities and whitespace
1184                 $text = Sanitizer::normalizeAttributeValue( $text );
1185
1186                 # Will be placed into "double-quoted" attributes,
1187                 # make sure remaining bits are safe.
1188                 $text = str_replace(
1189                         array('<', '>', '"'),
1190                         array('&lt;', '&gt;', '&quot;'),
1191                         $text );
1192
1193                 return $text;
1194         }
1195
1196         /**
1197          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1198          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1199          * PHP 5.1 doesn't.
1200          *
1201          * Use for passing XHTML fragments to PHP's XML parsing functions
1202          *
1203          * @return string
1204          * @static
1205          */
1206         static function hackDocType() {
1207                 global $wgHtmlEntities;
1208                 $out = "<!DOCTYPE html [\n";
1209                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1210                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1211                 }
1212                 $out .= "]>\n";
1213                 return $out;
1214         }
1215
1216         static function cleanUrl( $url, $hostname=true ) {
1217                 # Normalize any HTML entities in input. They will be
1218                 # re-escaped by makeExternalLink().
1219                 $url = Sanitizer::decodeCharReferences( $url );
1220
1221                 # Escape any control characters introduced by the above step
1222                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1223
1224                 # Validate hostname portion
1225                 $matches = array();
1226                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1227                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1228
1229                         // Characters that will be ignored in IDNs.
1230                         // http://tools.ietf.org/html/3454#section-3.1
1231                         // Strip them before further processing so blacklists and such work.
1232                         $strip = "/
1233                                 \\s|          # general whitespace
1234                                 \xc2\xad|     # 00ad SOFT HYPHEN
1235                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1236                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1237                                 \xe2\x81\xa0| # 2060 WORD JOINER
1238                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1239                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1240                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1241                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1242                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1243                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1244                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1245                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1246                                 /xuD";
1247
1248                         $host = preg_replace( $strip, '', $host );
1249
1250                         // @fixme: validate hostnames here
1251
1252                         return $protocol . $host . $rest;
1253                 } else {
1254                         return $url;
1255                 }
1256         }
1257
1258 }
1259
1260 ?>