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         function removeHTMLtags( $text, $processCallback = null, $args = array() ) {
 331                 global $wgUseTidy, $wgUserHtml;
 332                 $fname = 'Parser::removeHTMLtags';
 333                 wfProfileIn( $fname );
 334
 335                 if( $wgUserHtml ) {
 336                         $htmlpairs = array( # Tags that must be closed
 337                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 338                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 339                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 340                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 341                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
 342                         );
 343                         $htmlsingle = array(
 344                                 'br', 'hr', 'li', 'dt', 'dd'
 345                         );
 346                         $htmlsingleonly = array( # Elements that cannot have close tags
 347                                 'br', 'hr'
 348                         );
 349                         $htmlnest = array( # Tags that can be nested--??
 350                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 351                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 352                         );
 353                         $tabletags = array( # Can only appear inside table
 354                                 'td', 'th', 'tr'
 355                         );
 356                 } else {
 357                         $htmlpairs = array();
 358                         $htmlsingle = array();
 359                         $htmlnest = array();
 360                         $tabletags = array();
 361                 }
 362
 363                 $htmlsingle = array_merge( $tabletags, $htmlsingle );
 364                 $htmlelements = array_merge( $htmlsingle, $htmlpairs );
 365
 366                 # Remove HTML comments
 367                 $text = Sanitizer::removeHTMLcomments( $text );
 368                 $bits = explode( '<', $text );
 369                 $text = array_shift( $bits );
 370                 if(!$wgUseTidy) {
 371                         $tagstack = array(); $tablestack = array();
 372                         foreach ( $bits as $x ) {
 373                                 $prev = error_reporting( E_ALL & ~( E_NOTICE | E_WARNING ) );
 374                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 375                                 $x, $regs );
 376                                 list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
 377                                 error_reporting( $prev );
 378
 379                                 $badtag = 0 ;
 380                                 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
 381                                         # Check our stack
 382                                         if ( $slash ) {
 383                                                 # Closing a tag...
 384                                                 if( in_array( $t, $htmlsingleonly ) ) {
 385                                                         $badtag = 1;
 386                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 387                                                         @array_push( $tagstack, $ot );
 388                                                         $badtag = 1;
 389                                                 } else {
 390                                                         if ( $t == 'table' ) {
 391                                                                 $tagstack = array_pop( $tablestack );
 392                                                         }
 393                                                         $newparams = '';
 394                                                 }
 395                                         } else {
 396                                                 # Keep track for later
 397                                                 if ( in_array( $t, $tabletags ) &&
 398                                                 ! in_array( 'table', $tagstack ) ) {
 399                                                         $badtag = 1;
 400                                                 } else if ( in_array( $t, $tagstack ) &&
 401                                                 ! in_array ( $t , $htmlnest ) ) {
 402                                                         $badtag = 1 ;
 403                                                 # Is it a self closed htmlpair ? (bug 5487)
 404                                                 } else if( $brace == '/>' &&
 405                                                 in_array($t, $htmlpairs) ) {
 406                                                         $badtag = 1;
 407                                                 } elseif( in_array( $t, $htmlsingleonly ) ) {
 408                                                         # Hack to force empty tag for uncloseable elements
 409                                                         $brace = '/>';
 410                                                 } else if( in_array( $t, $htmlsingle ) ) {
 411                                                         # Hack to not close $htmlsingle tags
 412                                                         $brace = NULL;
 413                                                 } else {
 414                                                         if ( $t == 'table' ) {
 415                                                                 array_push( $tablestack, $tagstack );
 416                                                                 $tagstack = array();
 417                                                         }
 418                                                         array_push( $tagstack, $t );
 419                                                 }
 420
 421                                                 # Replace any variables or template parameters with
 422                                                 # plaintext results.
 423                                                 if( is_callable( $processCallback ) ) {
 424                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 425                                                 }
 426
 427                                                 # Strip non-approved attributes from the tag
 428                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 429                                         }
 430                                         if ( ! $badtag ) {
 431                                                 $rest = str_replace( '>', '&gt;', $rest );
 432                                                 $close = ( $brace == '/>' ) ? ' /' : '';
 433                                                 $text .= "<$slash$t$newparams$close>$rest";
 434                                                 continue;
 435                                         }
 436                                 }
 437                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 438                         }
 439                         # Close off any remaining tags
 440                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 441                                 $text .= "</$t>\n";
 442                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 443                         }
 444                 } else {
 445                         # this might be possible using tidy itself
 446                         foreach ( $bits as $x ) {
 447                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 448                                 $x, $regs );
 449                                 @list( $qbar, $slash, $t, $params, $brace, $rest ) = $regs;
 450                                 if ( in_array( $t = strtolower( $t ), $htmlelements ) ) {
 451                                         if( is_callable( $processCallback ) ) {
 452                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 453                                         }
 454                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 455                                         $rest = str_replace( '>', '&gt;', $rest );
 456                                         $text .= "<$slash$t$newparams$brace$rest";
 457                                 } else {
 458                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 459                                 }
 460                         }
 461                 }
 462                 wfProfileOut( $fname );
 463                 return $text;
 464         }
 465
 466         /**
 467          * Remove '<!--', '-->', and everything between.
 468          * To avoid leaving blank lines, when a comment is both preceded
 469          * and followed by a newline (ignoring spaces), trim leading and
 470          * trailing spaces and one of the newlines.
 471          *
 472          * @private
 473          * @param string $text
 474          * @return string
 475          */
 476         function removeHTMLcomments( $text ) {
 477                 $fname='Parser::removeHTMLcomments';
 478                 wfProfileIn( $fname );
 479                 while (($start = strpos($text, '<!--')) !== false) {
 480                         $end = strpos($text, '-->', $start + 4);
 481                         if ($end === false) {
 482                                 # Unterminated comment; bail out
 483                                 break;
 484                         }
 485
 486                         $end += 3;
 487
 488                         # Trim space and newline if the comment is both
 489                         # preceded and followed by a newline
 490                         $spaceStart = max($start - 1, 0);
 491                         $spaceLen = $end - $spaceStart;
 492                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 493                                 $spaceStart--;
 494                                 $spaceLen++;
 495                         }
 496                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 497                                 $spaceLen++;
 498                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 499                                 # Remove the comment, leading and trailing
 500                                 # spaces, and leave only one newline.
 501                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 502                         }
 503                         else {
 504                                 # Remove just the comment.
 505                                 $text = substr_replace($text, '', $start, $end - $start);
 506                         }
 507                 }
 508                 wfProfileOut( $fname );
 509                 return $text;
 510         }
 511
 512         /**
 513          * Take a tag soup fragment listing an HTML element's attributes
 514          * and normalize it to well-formed XML, discarding unwanted attributes.
 515          *
 516          * - Normalizes attribute names to lowercase
 517          * - Discards attributes not on a whitelist for the given element
 518          * - Turns broken or invalid entities into plaintext
 519          * - Double-quotes all attribute values
 520          * - Attributes without values are given the name as attribute
 521          * - Double attributes are discarded
 522          * - Unsafe style attributes are discarded
 523          * - Prepends space if there are attributes.
 524          *
 525          * @param string $text
 526          * @param string $element
 527          * @return string
 528          *
 529          * @todo Check for legal values where the DTD limits things.
 530          * @todo Check for unique id attribute :P
 531          */
 532         function fixTagAttributes( $text, $element ) {
 533                 if( trim( $text ) == '' ) {
 534                         return '';
 535                 }
 536
 537                 # Unquoted attribute
 538                 # Since we quote this later, this can be anything distinguishable
 539                 # from the end of the attribute
 540                 $pairs = array();
 541                 if( !preg_match_all(
 542                         MW_ATTRIBS_REGEX,
 543                         $text,
 544                         $pairs,
 545                         PREG_SET_ORDER ) ) {
 546                         return '';
 547                 }
 548
 549                 $whitelist = array_flip( Sanitizer::attributeWhitelist( $element ) );
 550                 $attribs = array();
 551                 foreach( $pairs as $set ) {
 552                         $attribute = strtolower( $set[1] );
 553                         if( !isset( $whitelist[$attribute] ) ) {
 554                                 continue;
 555                         }
 556
 557                         $raw   = Sanitizer::getTagAttributeCallback( $set );
 558                         $value = Sanitizer::normalizeAttributeValue( $raw );
 559
 560                         # Strip javascript "expression" from stylesheets.
 561                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 562                         if( $attribute == 'style' ) {
 563                                 $stripped = Sanitizer::decodeCharReferences( $value );
 564
 565                                 // Remove any comments; IE gets token splitting wrong
 566                                 $stripped = preg_replace( '!/\\*.*?\\*/!S', ' ', $stripped );
 567                                 $value = htmlspecialchars( $stripped );
 568
 569                                 // ... and continue checks
 570                                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 571                                         'codepointToUtf8(hexdec("$1"))', $stripped );
 572                                 $stripped = str_replace( '\\', '', $stripped );
 573                                 if( preg_match( '/(expression|tps*:\/\/|url\\s*\().*/is',
 574                                                 $stripped ) ) {
 575                                         # haxx0r
 576                                         continue;
 577                                 }
 578                         }
 579
 580                         if ( $attribute === 'id' )
 581                                 $value = Sanitizer::escapeId( $value );
 582
 583                         # Templates and links may be expanded in later parsing,
 584                         # creating invalid or dangerous output. Suppress this.
 585                         $value = strtr( $value, array(
 586                                 '<'    => '&lt;',   // This should never happen,
 587                                 '>'    => '&gt;',   // we've received invalid input
 588                                 '"'    => '&quot;', // which should have been escaped.
 589                                 '{'    => '&#123;',
 590                                 '['    => '&#91;',
 591                                 "''"   => '&#39;&#39;',
 592                                 'ISBN' => '&#73;SBN',
 593                                 'RFC'  => '&#82;FC',
 594                                 'PMID' => '&#80;MID',
 595                         ) );
 596
 597                         # Stupid hack
 598                         $value = preg_replace_callback(
 599                                 '/(' . wfUrlProtocols() . ')/',
 600                                 array( 'Sanitizer', 'armorLinksCallback' ),
 601                                 $value );
 602
 603                         // If this attribute was previously set, override it.
 604                         // Output should only have one attribute of each name.
 605                         $attribs[$attribute] = "$attribute=\"$value\"";
 606                 }
 607
 608                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 609         }
 610
 611         /**
 612          * Given a value escape it so that it can be used in an id attribute and
 613          * return it, this does not validate the value however (see first link)
 614          *
 615          * @link http://www.w3.org/TR/html401/types.html#type-name Valid characters
 616          *                                                          in the id and
 617          *                                                          name attributes
 618          * @link http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 619          *
 620          * @bug 4461
 621          *
 622          * @static
 623          *
 624          * @param string $id
 625          * @return string
 626          */
 627         function escapeId( $id ) {
 628                 static $replace = array(
 629                         '%3A' => ':',
 630                         '%' => '.'
 631                 );
 632
 633                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 634
 635                 return str_replace( array_keys( $replace ), array_values( $replace ), $id );
 636         }
 637
 638         /**
 639          * Regex replace callback for armoring links against further processing.
 640          * @param array $matches
 641          * @return string
 642          * @private
 643          */
 644         function armorLinksCallback( $matches ) {
 645                 return str_replace( ':', '&#58;', $matches[1] );
 646         }
 647
 648         /**
 649          * Return an associative array of attribute names and values from
 650          * a partial tag string. Attribute names are forces to lowercase,
 651          * character references are decoded to UTF-8 text.
 652          *
 653          * @param string
 654          * @return array
 655          */
 656         function decodeTagAttributes( $text ) {
 657                 $attribs = array();
 658
 659                 if( trim( $text ) == '' ) {
 660                         return $attribs;
 661                 }
 662
 663                 $pairs = array();
 664                 if( !preg_match_all(
 665                         MW_ATTRIBS_REGEX,
 666                         $text,
 667                         $pairs,
 668                         PREG_SET_ORDER ) ) {
 669                         return $attribs;
 670                 }
 671
 672                 foreach( $pairs as $set ) {
 673                         $attribute = strtolower( $set[1] );
 674                         $value = Sanitizer::getTagAttributeCallback( $set );
 675                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 676                 }
 677                 return $attribs;
 678         }
 679
 680         /**
 681          * Pick the appropriate attribute value from a match set from the
 682          * MW_ATTRIBS_REGEX matches.
 683          *
 684          * @param array $set
 685          * @return string
 686          * @private
 687          */
 688         function getTagAttributeCallback( $set ) {
 689                 if( isset( $set[6] ) ) {
 690                         # Illegal #XXXXXX color with no quotes.
 691                         return $set[6];
 692                 } elseif( isset( $set[5] ) ) {
 693                         # No quotes.
 694                         return $set[5];
 695                 } elseif( isset( $set[4] ) ) {
 696                         # Single-quoted
 697                         return $set[4];
 698                 } elseif( isset( $set[3] ) ) {
 699                         # Double-quoted
 700                         return $set[3];
 701                 } elseif( !isset( $set[2] ) ) {
 702                         # In XHTML, attributes must have a value.
 703                         # For 'reduced' form, return explicitly the attribute name here.
 704                         return $set[1];
 705                 } else {
 706                         wfDebugDieBacktrace( "Tag conditions not met. This should never happen and is a bug." );
 707                 }
 708         }
 709
 710         /**
 711          * Normalize whitespace and character references in an XML source-
 712          * encoded text for an attribute value.
 713          *
 714          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 715          * but note that we're not returning the value, but are returning
 716          * XML source fragments that will be slapped into output.
 717          *
 718          * @param string $text
 719          * @return string
 720          * @private
 721          */
 722         function normalizeAttributeValue( $text ) {
 723                 return str_replace( '"', '&quot;',
 724                         preg_replace(
 725                                 '/\r\n|[\x20\x0d\x0a\x09]/',
 726                                 ' ',
 727                                 Sanitizer::normalizeCharReferences( $text ) ) );
 728         }
 729
 730         /**
 731          * Ensure that any entities and character references are legal
 732          * for XML and XHTML specifically. Any stray bits will be
 733          * &amp;-escaped to result in a valid text fragment.
 734          *
 735          * a. any named char refs must be known in XHTML
 736          * b. any numeric char refs must be legal chars, not invalid or forbidden
 737          * c. use &#x, not &#X
 738          * d. fix or reject non-valid attributes
 739          *
 740          * @param string $text
 741          * @return string
 742          * @private
 743          */
 744         function normalizeCharReferences( $text ) {
 745                 return preg_replace_callback(
 746                         MW_CHAR_REFS_REGEX,
 747                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 748                         $text );
 749         }
 750         /**
 751          * @param string $matches
 752          * @return string
 753          */
 754         function normalizeCharReferencesCallback( $matches ) {
 755                 $ret = null;
 756                 if( $matches[1] != '' ) {
 757                         $ret = Sanitizer::normalizeEntity( $matches[1] );
 758                 } elseif( $matches[2] != '' ) {
 759                         $ret = Sanitizer::decCharReference( $matches[2] );
 760                 } elseif( $matches[3] != ''  ) {
 761                         $ret = Sanitizer::hexCharReference( $matches[3] );
 762                 } elseif( $matches[4] != '' ) {
 763                         $ret = Sanitizer::hexCharReference( $matches[4] );
 764                 }
 765                 if( is_null( $ret ) ) {
 766                         return htmlspecialchars( $matches[0] );
 767                 } else {
 768                         return $ret;
 769                 }
 770         }
 771
 772         /**
 773          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 774          * return the named entity reference as is. Otherwise, returns
 775          * HTML-escaped text of pseudo-entity source (eg &amp;foo;)
 776          *
 777          * @param string $name
 778          * @return string
 779          */
 780         function normalizeEntity( $name ) {
 781                 global $wgHtmlEntities;
 782                 if( isset( $wgHtmlEntities[$name] ) ) {
 783                         return "&$name;";
 784                 } else {
 785                         return "&amp;$name;";
 786                 }
 787         }
 788
 789         function decCharReference( $codepoint ) {
 790                 $point = intval( $codepoint );
 791                 if( Sanitizer::validateCodepoint( $point ) ) {
 792                         return sprintf( '&#%d;', $point );
 793                 } else {
 794                         return null;
 795                 }
 796         }
 797
 798         function hexCharReference( $codepoint ) {
 799                 $point = hexdec( $codepoint );
 800                 if( Sanitizer::validateCodepoint( $point ) ) {
 801                         return sprintf( '&#x%x;', $point );
 802                 } else {
 803                         return null;
 804                 }
 805         }
 806
 807         /**
 808          * Returns true if a given Unicode codepoint is a valid character in XML.
 809          * @param int $codepoint
 810          * @return bool
 811          */
 812         function validateCodepoint( $codepoint ) {
 813                 return ($codepoint ==    0x09)
 814                         || ($codepoint ==    0x0a)
 815                         || ($codepoint ==    0x0d)
 816                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
 817                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
 818                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
 819         }
 820
 821         /**
 822          * Decode any character references, numeric or named entities,
 823          * in the text and return a UTF-8 string.
 824          *
 825          * @param string $text
 826          * @return string
 827          * @public
 828          */
 829         function decodeCharReferences( $text ) {
 830                 return preg_replace_callback(
 831                         MW_CHAR_REFS_REGEX,
 832                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
 833                         $text );
 834         }
 835
 836         /**
 837          * @param string $matches
 838          * @return string
 839          */
 840         function decodeCharReferencesCallback( $matches ) {
 841                 if( $matches[1] != '' ) {
 842                         return Sanitizer::decodeEntity( $matches[1] );
 843                 } elseif( $matches[2] != '' ) {
 844                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
 845                 } elseif( $matches[3] != ''  ) {
 846                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
 847                 } elseif( $matches[4] != '' ) {
 848                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
 849                 }
 850                 # Last case should be an ampersand by itself
 851                 return $matches[0];
 852         }
 853
 854         /**
 855          * Return UTF-8 string for a codepoint if that is a valid
 856          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
 857          * @param int $codepoint
 858          * @return string
 859          * @private
 860          */
 861         function decodeChar( $codepoint ) {
 862                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
 863                         return codepointToUtf8( $codepoint );
 864                 } else {
 865                         return UTF8_REPLACEMENT;
 866                 }
 867         }
 868
 869         /**
 870          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
 871          * return the UTF-8 encoding of that character. Otherwise, returns
 872          * pseudo-entity source (eg &foo;)
 873          *
 874          * @param string $name
 875          * @return string
 876          */
 877         function decodeEntity( $name ) {
 878                 global $wgHtmlEntities;
 879                 if( isset( $wgHtmlEntities[$name] ) ) {
 880                         return codepointToUtf8( $wgHtmlEntities[$name] );
 881                 } else {
 882                         return "&$name;";
 883                 }
 884         }
 885
 886         /**
 887          * Fetch the whitelist of acceptable attributes for a given
 888          * element name.
 889          *
 890          * @param string $element
 891          * @return array
 892          */
 893         function attributeWhitelist( $element ) {
 894                 static $list;
 895                 if( !isset( $list ) ) {
 896                         $list = Sanitizer::setupAttributeWhitelist();
 897                 }
 898                 return isset( $list[$element] )
 899                         ? $list[$element]
 900                         : array();
 901         }
 902
 903         /**
 904          * @return array
 905          */
 906         function setupAttributeWhitelist() {
 907                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
 908                 $block = array_merge( $common, array( 'align' ) );
 909                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
 910                 $tablecell = array( 'abbr',
 911                                     'axis',
 912                                     'headers',
 913                                     'scope',
 914                                     'rowspan',
 915                                     'colspan',
 916                                     'nowrap', # deprecated
 917                                     'width',  # deprecated
 918                                     'height', # deprecated
 919                                     'bgcolor' # deprecated
 920                                     );
 921
 922                 # Numbers refer to sections in HTML 4.01 standard describing the element.
 923                 # See: http://www.w3.org/TR/html4/
 924                 $whitelist = array (
 925                         # 7.5.4
 926                         'div'        => $block,
 927                         'center'     => $common, # deprecated
 928                         'span'       => $block, # ??
 929
 930                         # 7.5.5
 931                         'h1'         => $block,
 932                         'h2'         => $block,
 933                         'h3'         => $block,
 934                         'h4'         => $block,
 935                         'h5'         => $block,
 936                         'h6'         => $block,
 937
 938                         # 7.5.6
 939                         # address
 940
 941                         # 8.2.4
 942                         # bdo
 943
 944                         # 9.2.1
 945                         'em'         => $common,
 946                         'strong'     => $common,
 947                         'cite'       => $common,
 948                         # dfn
 949                         'code'       => $common,
 950                         # samp
 951                         # kbd
 952                         'var'        => $common,
 953                         # abbr
 954                         # acronym
 955
 956                         # 9.2.2
 957                         'blockquote' => array_merge( $common, array( 'cite' ) ),
 958                         # q
 959
 960                         # 9.2.3
 961                         'sub'        => $common,
 962                         'sup'        => $common,
 963
 964                         # 9.3.1
 965                         'p'          => $block,
 966
 967                         # 9.3.2
 968                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
 969
 970                         # 9.3.4
 971                         'pre'        => array_merge( $common, array( 'width' ) ),
 972
 973                         # 9.4
 974                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 975                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
 976
 977                         # 10.2
 978                         'ul'         => array_merge( $common, array( 'type' ) ),
 979                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
 980                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
 981
 982                         # 10.3
 983                         'dl'         => $common,
 984                         'dd'         => $common,
 985                         'dt'         => $common,
 986
 987                         # 11.2.1
 988                         'table'      => array_merge( $common,
 989                                                                 array( 'summary', 'width', 'border', 'frame',
 990                                                                                          'rules', 'cellspacing', 'cellpadding',
 991                                                                                          'align', 'bgcolor', 'frame', 'rules',
 992                                                                                          'border' ) ),
 993
 994                         # 11.2.2
 995                         'caption'    => array_merge( $common, array( 'align' ) ),
 996
 997                         # 11.2.3
 998                         'thead'      => array_merge( $common, $tablealign ),
 999                         'tfoot'      => array_merge( $common, $tablealign ),
1000                         'tbody'      => array_merge( $common, $tablealign ),
1001
1002                         # 11.2.4
1003                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1004                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1005
1006                         # 11.2.5
1007                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1008
1009                         # 11.2.6
1010                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1011                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1012
1013                         # 15.2.1
1014                         'tt'         => $common,
1015                         'b'          => $common,
1016                         'i'          => $common,
1017                         'big'        => $common,
1018                         'small'      => $common,
1019                         'strike'     => $common,
1020                         's'          => $common,
1021                         'u'          => $common,
1022
1023                         # 15.2.2
1024                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1025                         # basefont
1026
1027                         # 15.3
1028                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1029
1030                         # XHTML Ruby annotation text module, simple ruby only.
1031                         # http://www.w3c.org/TR/ruby/
1032                         'ruby'       => $common,
1033                         # rbc
1034                         # rtc
1035                         'rb'         => $common,
1036                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1037                         'rp'         => $common,
1038                         );
1039                 return $whitelist;
1040         }
1041
1042         /**
1043          * Take a fragment of (potentially invalid) HTML and return
1044          * a version with any tags removed, encoded suitably for literal
1045          * inclusion in an attribute value.
1046          *
1047          * @param string $text HTML fragment
1048          * @return string
1049          */
1050         function stripAllTags( $text ) {
1051                 # Actual <tags>
1052                 $text = preg_replace( '/ < .*? > /x', '', $text );
1053
1054                 # Normalize &entities and whitespace
1055                 $text = Sanitizer::normalizeAttributeValue( $text );
1056
1057                 # Will be placed into "double-quoted" attributes,
1058                 # make sure remaining bits are safe.
1059                 $text = str_replace(
1060                         array('<', '>', '"'),
1061                         array('&lt;', '&gt;', '&quot;'),
1062                         $text );
1063
1064                 return $text;
1065         }
1066
1067         /**
1068          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1069          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1070          * PHP 5.1 doesn't.
1071          *
1072          * Use for passing XHTML fragments to PHP's XML parsing functions
1073          *
1074          * @return string
1075          * @static
1076          */
1077         function hackDocType() {
1078                 global $wgHtmlEntities;
1079                 $out = "<!DOCTYPE html [\n";
1080                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1081                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1082                 }
1083                 $out .= "]>\n";
1084                 return $out;
1085         }
1086
1087 }
1088
1089 ?>