includes/Sanitizer.php

   1 <?php
   2 /**
   3  * XHTML sanitizer for MediaWiki
   4  *
   5  * Copyright © 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Parser
  25  */
  26
  27 /**
  28  * Regular expression to match various types of character references in
  29  * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  30  */
  31 define( 'MW_CHAR_REFS_REGEX',
  32         '/&([A-Za-z0-9\x80-\xff]+);
  33          |&\#([0-9]+);
  34          |&\#x([0-9A-Za-z]+);
  35          |&\#X([0-9A-Za-z]+);
  36          |(&)/x' );
  37
  38 /**
  39  * Regular expression to match HTML/XML attribute pairs within a tag.
  40  * Allows some... latitude.
  41  * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
  42  */
  43 $attribFirst = '[:A-Z_a-z]';
  44 $attrib = '[:A-Z_a-z-.0-9]';
  45 $space = '[\x09\x0a\x0d\x20]';
  46 define( 'MW_ATTRIBS_REGEX',
  47         "/(?:^|$space)({$attribFirst}{$attrib}*)
  48           ($space*=$space*
  49                 (?:
  50                  # The attribute value: quoted or alone
  51                   \"([^<\"]*)\"
  52                  | '([^<']*)'
  53                  |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
  54                  |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
  55                                                          # colors are specified like this.
  56                                                          # We'll be normalizing it.
  57                 )
  58            )?(?=$space|\$)/sx" );
  59
  60 /**
  61  * Regular expression to match URIs that could trigger script execution
  62  */
  63 define( 'MW_EVIL_URI_PATTERN', '!(^|\s|\*/\s*)(javascript|vbscript)([^\w]|$)!i' );
  64
  65 /**
  66  * Regular expression to match namespace attributes
  67  */
  68 define( 'MW_XMLNS_ATTRIBUTE_PATTRN', "/^xmlns:$attrib+$/" );
  69
  70 /**
  71  * List of all named character entities defined in HTML 4.01
  72  * http://www.w3.org/TR/html4/sgml/entities.html
  73  * @private
  74  */
  75 global $wgHtmlEntities;
  76 $wgHtmlEntities = array(
  77         'Aacute'   => 193,
  78         'aacute'   => 225,
  79         'Acirc'    => 194,
  80         'acirc'    => 226,
  81         'acute'    => 180,
  82         'AElig'    => 198,
  83         'aelig'    => 230,
  84         'Agrave'   => 192,
  85         'agrave'   => 224,
  86         'alefsym'  => 8501,
  87         'Alpha'    => 913,
  88         'alpha'    => 945,
  89         'amp'      => 38,
  90         'and'      => 8743,
  91         'ang'      => 8736,
  92         'Aring'    => 197,
  93         'aring'    => 229,
  94         'asymp'    => 8776,
  95         'Atilde'   => 195,
  96         'atilde'   => 227,
  97         'Auml'     => 196,
  98         'auml'     => 228,
  99         'bdquo'    => 8222,
 100         'Beta'     => 914,
 101         'beta'     => 946,
 102         'brvbar'   => 166,
 103         'bull'     => 8226,
 104         'cap'      => 8745,
 105         'Ccedil'   => 199,
 106         'ccedil'   => 231,
 107         'cedil'    => 184,
 108         'cent'     => 162,
 109         'Chi'      => 935,
 110         'chi'      => 967,
 111         'circ'     => 710,
 112         'clubs'    => 9827,
 113         'cong'     => 8773,
 114         'copy'     => 169,
 115         'crarr'    => 8629,
 116         'cup'      => 8746,
 117         'curren'   => 164,
 118         'dagger'   => 8224,
 119         'Dagger'   => 8225,
 120         'darr'     => 8595,
 121         'dArr'     => 8659,
 122         'deg'      => 176,
 123         'Delta'    => 916,
 124         'delta'    => 948,
 125         'diams'    => 9830,
 126         'divide'   => 247,
 127         'Eacute'   => 201,
 128         'eacute'   => 233,
 129         'Ecirc'    => 202,
 130         'ecirc'    => 234,
 131         'Egrave'   => 200,
 132         'egrave'   => 232,
 133         'empty'    => 8709,
 134         'emsp'     => 8195,
 135         'ensp'     => 8194,
 136         'Epsilon'  => 917,
 137         'epsilon'  => 949,
 138         'equiv'    => 8801,
 139         'Eta'      => 919,
 140         'eta'      => 951,
 141         'ETH'      => 208,
 142         'eth'      => 240,
 143         'Euml'     => 203,
 144         'euml'     => 235,
 145         'euro'     => 8364,
 146         'exist'    => 8707,
 147         'fnof'     => 402,
 148         'forall'   => 8704,
 149         'frac12'   => 189,
 150         'frac14'   => 188,
 151         'frac34'   => 190,
 152         'frasl'    => 8260,
 153         'Gamma'    => 915,
 154         'gamma'    => 947,
 155         'ge'       => 8805,
 156         'gt'       => 62,
 157         'harr'     => 8596,
 158         'hArr'     => 8660,
 159         'hearts'   => 9829,
 160         'hellip'   => 8230,
 161         'Iacute'   => 205,
 162         'iacute'   => 237,
 163         'Icirc'    => 206,
 164         'icirc'    => 238,
 165         'iexcl'    => 161,
 166         'Igrave'   => 204,
 167         'igrave'   => 236,
 168         'image'    => 8465,
 169         'infin'    => 8734,
 170         'int'      => 8747,
 171         'Iota'     => 921,
 172         'iota'     => 953,
 173         'iquest'   => 191,
 174         'isin'     => 8712,
 175         'Iuml'     => 207,
 176         'iuml'     => 239,
 177         'Kappa'    => 922,
 178         'kappa'    => 954,
 179         'Lambda'   => 923,
 180         'lambda'   => 955,
 181         'lang'     => 9001,
 182         'laquo'    => 171,
 183         'larr'     => 8592,
 184         'lArr'     => 8656,
 185         'lceil'    => 8968,
 186         'ldquo'    => 8220,
 187         'le'       => 8804,
 188         'lfloor'   => 8970,
 189         'lowast'   => 8727,
 190         'loz'      => 9674,
 191         'lrm'      => 8206,
 192         'lsaquo'   => 8249,
 193         'lsquo'    => 8216,
 194         'lt'       => 60,
 195         'macr'     => 175,
 196         'mdash'    => 8212,
 197         'micro'    => 181,
 198         'middot'   => 183,
 199         'minus'    => 8722,
 200         'Mu'       => 924,
 201         'mu'       => 956,
 202         'nabla'    => 8711,
 203         'nbsp'     => 160,
 204         'ndash'    => 8211,
 205         'ne'       => 8800,
 206         'ni'       => 8715,
 207         'not'      => 172,
 208         'notin'    => 8713,
 209         'nsub'     => 8836,
 210         'Ntilde'   => 209,
 211         'ntilde'   => 241,
 212         'Nu'       => 925,
 213         'nu'       => 957,
 214         'Oacute'   => 211,
 215         'oacute'   => 243,
 216         'Ocirc'    => 212,
 217         'ocirc'    => 244,
 218         'OElig'    => 338,
 219         'oelig'    => 339,
 220         'Ograve'   => 210,
 221         'ograve'   => 242,
 222         'oline'    => 8254,
 223         'Omega'    => 937,
 224         'omega'    => 969,
 225         'Omicron'  => 927,
 226         'omicron'  => 959,
 227         'oplus'    => 8853,
 228         'or'       => 8744,
 229         'ordf'     => 170,
 230         'ordm'     => 186,
 231         'Oslash'   => 216,
 232         'oslash'   => 248,
 233         'Otilde'   => 213,
 234         'otilde'   => 245,
 235         'otimes'   => 8855,
 236         'Ouml'     => 214,
 237         'ouml'     => 246,
 238         'para'     => 182,
 239         'part'     => 8706,
 240         'permil'   => 8240,
 241         'perp'     => 8869,
 242         'Phi'      => 934,
 243         'phi'      => 966,
 244         'Pi'       => 928,
 245         'pi'       => 960,
 246         'piv'      => 982,
 247         'plusmn'   => 177,
 248         'pound'    => 163,
 249         'prime'    => 8242,
 250         'Prime'    => 8243,
 251         'prod'     => 8719,
 252         'prop'     => 8733,
 253         'Psi'      => 936,
 254         'psi'      => 968,
 255         'quot'     => 34,
 256         'radic'    => 8730,
 257         'rang'     => 9002,
 258         'raquo'    => 187,
 259         'rarr'     => 8594,
 260         'rArr'     => 8658,
 261         'rceil'    => 8969,
 262         'rdquo'    => 8221,
 263         'real'     => 8476,
 264         'reg'      => 174,
 265         'rfloor'   => 8971,
 266         'Rho'      => 929,
 267         'rho'      => 961,
 268         'rlm'      => 8207,
 269         'rsaquo'   => 8250,
 270         'rsquo'    => 8217,
 271         'sbquo'    => 8218,
 272         'Scaron'   => 352,
 273         'scaron'   => 353,
 274         'sdot'     => 8901,
 275         'sect'     => 167,
 276         'shy'      => 173,
 277         'Sigma'    => 931,
 278         'sigma'    => 963,
 279         'sigmaf'   => 962,
 280         'sim'      => 8764,
 281         'spades'   => 9824,
 282         'sub'      => 8834,
 283         'sube'     => 8838,
 284         'sum'      => 8721,
 285         'sup'      => 8835,
 286         'sup1'     => 185,
 287         'sup2'     => 178,
 288         'sup3'     => 179,
 289         'supe'     => 8839,
 290         'szlig'    => 223,
 291         'Tau'      => 932,
 292         'tau'      => 964,
 293         'there4'   => 8756,
 294         'Theta'    => 920,
 295         'theta'    => 952,
 296         'thetasym' => 977,
 297         'thinsp'   => 8201,
 298         'THORN'    => 222,
 299         'thorn'    => 254,
 300         'tilde'    => 732,
 301         'times'    => 215,
 302         'trade'    => 8482,
 303         'Uacute'   => 218,
 304         'uacute'   => 250,
 305         'uarr'     => 8593,
 306         'uArr'     => 8657,
 307         'Ucirc'    => 219,
 308         'ucirc'    => 251,
 309         'Ugrave'   => 217,
 310         'ugrave'   => 249,
 311         'uml'      => 168,
 312         'upsih'    => 978,
 313         'Upsilon'  => 933,
 314         'upsilon'  => 965,
 315         'Uuml'     => 220,
 316         'uuml'     => 252,
 317         'weierp'   => 8472,
 318         'Xi'       => 926,
 319         'xi'       => 958,
 320         'Yacute'   => 221,
 321         'yacute'   => 253,
 322         'yen'      => 165,
 323         'Yuml'     => 376,
 324         'yuml'     => 255,
 325         'Zeta'     => 918,
 326         'zeta'     => 950,
 327         'zwj'      => 8205,
 328         'zwnj'     => 8204 );
 329
 330 /**
 331  * Character entity aliases accepted by MediaWiki
 332  */
 333 global $wgHtmlEntityAliases;
 334 $wgHtmlEntityAliases = array(
 335         'רלמ' => 'rlm',
 336         'رلم' => 'rlm',
 337 );
 338
 339
 340 /**
 341  * XHTML sanitizer for MediaWiki
 342  * @ingroup Parser
 343  */
 344 class Sanitizer {
 345         /**
 346          * Cleans up HTML, removes dangerous tags and attributes, and
 347          * removes HTML comments
 348          * @private
 349          * @param $text String
 350          * @param $processCallback Callback to do any variable or parameter replacements in HTML attribute values
 351          * @param $args Array for the processing callback
 352          * @param $extratags Array for any extra tags to include
 353          * @param $removetags Array for any tags (default or extra) to exclude
 354          * @return string
 355          */
 356         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array(), $removetags = array() ) {
 357                 global $wgUseTidy;
 358
 359                 static $htmlpairsStatic, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 360                         $htmllist, $listtags, $htmlsingleallowed, $htmlelementsStatic, $staticInitialised;
 361
 362                 wfProfileIn( __METHOD__ );
 363
 364                 if ( !$staticInitialised ) {
 365
 366                         $htmlpairsStatic = array( # Tags that must be closed
 367                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 368                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 369                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 370                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 371                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'abbr', 'dfn',
 372                                 'kbd', 'samp'
 373                         );
 374                         $htmlsingle = array(
 375                                 'br', 'hr', 'li', 'dt', 'dd'
 376                         );
 377                         $htmlsingleonly = array( # Elements that cannot have close tags
 378                                 'br', 'hr'
 379                         );
 380                         $htmlnest = array( # Tags that can be nested--??
 381                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 382                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 383                         );
 384                         $tabletags = array( # Can only appear inside table, we will close them
 385                                 'td', 'th', 'tr',
 386                         );
 387                         $htmllist = array( # Tags used by list
 388                                 'ul','ol',
 389                         );
 390                         $listtags = array( # Tags that can appear in a list
 391                                 'li',
 392                         );
 393
 394                         global $wgAllowImageTag;
 395                         if ( $wgAllowImageTag ) {
 396                                 $htmlsingle[] = 'img';
 397                                 $htmlsingleonly[] = 'img';
 398                         }
 399
 400                         $htmlsingleallowed = array_unique( array_merge( $htmlsingle, $tabletags ) );
 401                         $htmlelementsStatic = array_unique( array_merge( $htmlsingle, $htmlpairsStatic, $htmlnest ) );
 402
 403                         # Convert them all to hashtables for faster lookup
 404                         $vars = array( 'htmlpairsStatic', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 405                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelementsStatic' );
 406                         foreach ( $vars as $var ) {
 407                                 $$var = array_flip( $$var );
 408                         }
 409                         $staticInitialised = true;
 410                 }
 411                 # Populate $htmlpairs and $htmlelements with the $extratags and $removetags arrays
 412                 $extratags = array_flip( $extratags );
 413                 $removetags = array_flip( $removetags );
 414                 $htmlpairs = array_merge( $extratags, $htmlpairsStatic );
 415                 $htmlelements = array_diff_key( array_merge( $extratags, $htmlelementsStatic ) , $removetags );
 416
 417                 # Remove HTML comments
 418                 $text = Sanitizer::removeHTMLcomments( $text );
 419                 $bits = explode( '<', $text );
 420                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 421                 if ( !$wgUseTidy ) {
 422                         $tagstack = $tablestack = array();
 423                         foreach ( $bits as $x ) {
 424                                 $regs = array();
 425                                 # $slash: Does the current element start with a '/'?
 426                                 # $t: Current element name
 427                                 # $params: String between element name and >
 428                                 # $brace: Ending '>' or '/>'
 429                                 # $rest: Everything until the next element of $bits
 430                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 431                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 432                                 } else {
 433                                         $slash = $t = $params = $brace = $rest = null;
 434                                 }
 435
 436                                 $badtag = false;
 437                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 438                                         # Check our stack
 439                                         if ( $slash && isset( $htmlsingleonly[$t] ) ) {
 440                                                 $badtag = true;
 441                                         } elseif ( $slash ) {
 442                                                 # Closing a tag... is it the one we just opened?
 443                                                 $ot = @array_pop( $tagstack );
 444                                                 if ( $ot != $t ) {
 445                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 446                                                                 # Pop all elements with an optional close tag
 447                                                                 # and see if we find a match below them
 448                                                                 $optstack = array();
 449                                                                 array_push( $optstack, $ot );
 450                                                                 $ot = @array_pop( $tagstack );
 451                                                                 while ( $ot != $t && isset( $htmlsingleallowed[$ot] ) ) {
 452                                                                         array_push( $optstack, $ot );
 453                                                                         $ot = @array_pop( $tagstack );
 454                                                                 }
 455                                                                 if ( $t != $ot ) {
 456                                                                         # No match. Push the optional elements back again
 457                                                                         $badtag = true;
 458                                                                         while ( $ot = @array_pop( $optstack ) ) {
 459                                                                                 array_push( $tagstack, $ot );
 460                                                                         }
 461                                                                 }
 462                                                         } else {
 463                                                                 @array_push( $tagstack, $ot );
 464                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 465                                                                 if ( !isset( $htmllist[$ot] ) || !isset( $listtags[$t] ) ) {
 466                                                                         $badtag = true;
 467                                                                 }
 468                                                         }
 469                                                 } else {
 470                                                         if ( $t == 'table' ) {
 471                                                                 $tagstack = array_pop( $tablestack );
 472                                                         }
 473                                                 }
 474                                                 $newparams = '';
 475                                         } else {
 476                                                 # Keep track for later
 477                                                 if ( isset( $tabletags[$t] ) &&
 478                                                 !in_array( 'table', $tagstack ) ) {
 479                                                         $badtag = true;
 480                                                 } elseif ( in_array( $t, $tagstack ) &&
 481                                                 !isset( $htmlnest [$t ] ) ) {
 482                                                         $badtag = true;
 483                                                 # Is it a self closed htmlpair ? (bug 5487)
 484                                                 } elseif ( $brace == '/>' &&
 485                                                 isset( $htmlpairs[$t] ) ) {
 486                                                         $badtag = true;
 487                                                 } elseif ( isset( $htmlsingleonly[$t] ) ) {
 488                                                         # Hack to force empty tag for uncloseable elements
 489                                                         $brace = '/>';
 490                                                 } elseif ( isset( $htmlsingle[$t] ) ) {
 491                                                         # Hack to not close $htmlsingle tags
 492                                                         $brace = null;
 493                                                 } elseif ( isset( $tabletags[$t] )
 494                                                 && in_array( $t, $tagstack ) ) {
 495                                                         // New table tag but forgot to close the previous one
 496                                                         $text .= "</$t>";
 497                                                 } else {
 498                                                         if ( $t == 'table' ) {
 499                                                                 array_push( $tablestack, $tagstack );
 500                                                                 $tagstack = array();
 501                                                         }
 502                                                         array_push( $tagstack, $t );
 503                                                 }
 504
 505                                                 # Replace any variables or template parameters with
 506                                                 # plaintext results.
 507                                                 if( is_callable( $processCallback ) ) {
 508                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 509                                                 }
 510
 511                                                 # Strip non-approved attributes from the tag
 512                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 513                                         }
 514                                         if ( !$badtag ) {
 515                                                 $rest = str_replace( '>', '&gt;', $rest );
 516                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 517                                                 $text .= "<$slash$t$newparams$close>$rest";
 518                                                 continue;
 519                                         }
 520                                 }
 521                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 522                         }
 523                         # Close off any remaining tags
 524                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 525                                 $text .= "</$t>\n";
 526                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 527                         }
 528                 } else {
 529                         # this might be possible using tidy itself
 530                         foreach ( $bits as $x ) {
 531                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 532                                 $x, $regs );
 533                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 534                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 535                                         if( is_callable( $processCallback ) ) {
 536                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 537                                         }
 538                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 539                                         $rest = str_replace( '>', '&gt;', $rest );
 540                                         $text .= "<$slash$t$newparams$brace$rest";
 541                                 } else {
 542                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 543                                 }
 544                         }
 545                 }
 546                 wfProfileOut( __METHOD__ );
 547                 return $text;
 548         }
 549
 550         /**
 551          * Remove '<!--', '-->', and everything between.
 552          * To avoid leaving blank lines, when a comment is both preceded
 553          * and followed by a newline (ignoring spaces), trim leading and
 554          * trailing spaces and one of the newlines.
 555          *
 556          * @private
 557          * @param $text String
 558          * @return string
 559          */
 560         static function removeHTMLcomments( $text ) {
 561                 wfProfileIn( __METHOD__ );
 562                 while (($start = strpos($text, '<!--')) !== false) {
 563                         $end = strpos($text, '-->', $start + 4);
 564                         if ($end === false) {
 565                                 # Unterminated comment; bail out
 566                                 break;
 567                         }
 568
 569                         $end += 3;
 570
 571                         # Trim space and newline if the comment is both
 572                         # preceded and followed by a newline
 573                         $spaceStart = max($start - 1, 0);
 574                         $spaceLen = $end - $spaceStart;
 575                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 576                                 $spaceStart--;
 577                                 $spaceLen++;
 578                         }
 579                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 580                                 $spaceLen++;
 581                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 582                                 # Remove the comment, leading and trailing
 583                                 # spaces, and leave only one newline.
 584                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 585                         }
 586                         else {
 587                                 # Remove just the comment.
 588                                 $text = substr_replace($text, '', $start, $end - $start);
 589                         }
 590                 }
 591                 wfProfileOut( __METHOD__ );
 592                 return $text;
 593         }
 594
 595         /**
 596          * Take an array of attribute names and values and normalize or discard
 597          * illegal values for the given element type.
 598          *
 599          * - Discards attributes not on a whitelist for the given element
 600          * - Unsafe style attributes are discarded
 601          * - Invalid id attributes are reencoded
 602          *
 603          * @param $attribs Array
 604          * @param $element String
 605          * @return Array
 606          *
 607          * @todo Check for legal values where the DTD limits things.
 608          * @todo Check for unique id attribute :P
 609          */
 610         static function validateTagAttributes( $attribs, $element ) {
 611                 return Sanitizer::validateAttributes( $attribs,
 612                         Sanitizer::attributeWhitelist( $element ) );
 613         }
 614
 615         /**
 616          * Take an array of attribute names and values and normalize or discard
 617          * illegal values for the given whitelist.
 618          *
 619          * - Discards attributes not the given whitelist
 620          * - Unsafe style attributes are discarded
 621          * - Invalid id attributes are reencoded
 622          *
 623          * @param $attribs Array
 624          * @param $whitelist Array: list of allowed attribute names
 625          * @return Array
 626          *
 627          * @todo Check for legal values where the DTD limits things.
 628          * @todo Check for unique id attribute :P
 629          */
 630         static function validateAttributes( $attribs, $whitelist ) {
 631                 global $wgAllowRdfaAttributes, $wgAllowMicrodataAttributes, $wgHtml5;
 632
 633                 $whitelist = array_flip( $whitelist );
 634                 $hrefExp = '/^(' . wfUrlProtocols() . ')[^\s]+$/';
 635
 636                 $out = array();
 637                 foreach( $attribs as $attribute => $value ) {
 638                         #allow XML namespace declaration if RDFa is enabled
 639                         if ( $wgAllowRdfaAttributes && preg_match( MW_XMLNS_ATTRIBUTE_PATTRN, $attribute ) ) {
 640                                 if ( !preg_match( MW_EVIL_URI_PATTERN, $value ) ) {
 641                                         $out[$attribute] = $value;
 642                                 }
 643
 644                                 continue;
 645                         }
 646
 647                         # Allow any attribute beginning with "data-", if in HTML5 mode
 648                         if ( !($wgHtml5 && preg_match( '/^data-/i', $attribute )) && !isset( $whitelist[$attribute] ) ) {
 649                                 continue;
 650                         }
 651
 652                         # Strip javascript "expression" from stylesheets.
 653                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 654                         if( $attribute == 'style' ) {
 655                                 $value = Sanitizer::checkCss( $value );
 656                         }
 657
 658                         if ( $attribute === 'id' ) {
 659                                 $value = Sanitizer::escapeId( $value, 'noninitial' );
 660                         }
 661
 662                         //RDFa and microdata properties allow URLs, URIs and/or CURIs. check them for sanity
 663                         if ( $attribute === 'rel' || $attribute === 'rev' ||
 664                                 $attribute === 'about' || $attribute === 'property' || $attribute === 'resource' || #RDFa
 665                                 $attribute === 'datatype' || $attribute === 'typeof' ||                             #RDFa
 666                                 $attribute === 'itemid' || $attribute === 'itemprop' || $attribute === 'itemref' || #HTML5 microdata
 667                                 $attribute === 'itemscope' || $attribute === 'itemtype' ) {                         #HTML5 microdata
 668
 669                                 //Paranoia. Allow "simple" values but suppress javascript
 670                                 if ( preg_match( MW_EVIL_URI_PATTERN, $value ) ) {
 671                                         continue;
 672                                 }
 673                         }
 674
 675                         # NOTE: even though elements using href/src are not allowed directly, supply
 676                         #       validation code that can be used by tag hook handlers, etc
 677                         if ( $attribute === 'href' || $attribute === 'src' ) {
 678                                 if ( !preg_match( $hrefExp, $value ) ) {
 679                                         continue; //drop any href or src attributes not using an allowed protocol.
 680                                                   //NOTE: this also drops all relative URLs
 681                                 }
 682                         }
 683
 684                         // If this attribute was previously set, override it.
 685                         // Output should only have one attribute of each name.
 686                         $out[$attribute] = $value;
 687                 }
 688
 689                 if ( $wgAllowMicrodataAttributes ) {
 690                         # There are some complicated validity constraints we need to
 691                         # enforce here.  First of all, we don't want to allow non-standard
 692                         # itemtypes.
 693                         $allowedTypes = array(
 694                                 'http://microformats.org/profile/hcard',
 695                                 'http://microformats.org/profile/hcalendar#vevent',
 696                                 'http://n.whatwg.org/work',
 697                         );
 698                         if ( isset( $out['itemtype'] ) && !in_array( $out['itemtype'],
 699                         $allowedTypes ) ) {
 700                                 # Kill everything
 701                                 unset( $out['itemscope'] );
 702                         }
 703                         # itemtype, itemid, itemref don't make sense without itemscope
 704                         if ( !array_key_exists( 'itemscope', $out ) ) {
 705                                 unset( $out['itemtype'] );
 706                                 unset( $out['itemid'] );
 707                                 unset( $out['itemref'] );
 708                         }
 709                         # TODO: Strip itemprop if we aren't descendants of an itemscope.
 710                 }
 711                 return $out;
 712         }
 713
 714         /**
 715          * Merge two sets of HTML attributes.  Conflicting items in the second set
 716          * will override those in the first, except for 'class' attributes which
 717          * will be combined (if they're both strings).
 718          *
 719          * @todo implement merging for other attributes such as style
 720          * @param $a Array
 721          * @param $b Array
 722          * @return array
 723          */
 724         static function mergeAttributes( $a, $b ) {
 725                 $out = array_merge( $a, $b );
 726                 if( isset( $a['class'] ) && isset( $b['class'] )
 727                 && is_string( $a['class'] ) && is_string( $b['class'] )
 728                 && $a['class'] !== $b['class'] ) {
 729                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 730                                 -1, PREG_SPLIT_NO_EMPTY );
 731                         $out['class'] = implode( ' ', array_unique( $classes ) );
 732                 }
 733                 return $out;
 734         }
 735
 736         /**
 737          * Pick apart some CSS and check it for forbidden or unsafe structures.
 738          * Returns a sanitized string, or false if it was just too evil.
 739          *
 740          * Currently URL references, 'expression', 'tps' are forbidden.
 741          *
 742          * @param $value String
 743          * @return Mixed
 744          */
 745         static function checkCss( $value ) {
 746                 $value = Sanitizer::decodeCharReferences( $value );
 747
 748                 // Remove any comments; IE gets token splitting wrong
 749                 $value = StringUtils::delimiterReplace( '/*', '*/', ' ', $value );
 750
 751                 // Decode escape sequences and line continuation
 752                 // See the grammar in the CSS 2 spec, appendix D.
 753                 static $decodeRegex;
 754                 if ( !$decodeRegex ) {
 755                         $space = '[\\x20\\t\\r\\n\\f]';
 756                         $nl = '(?:\\n|\\r\\n|\\r|\\f)';
 757                         $backslash = '\\\\';
 758                         $decodeRegex = "/ $backslash
 759                                 (?:
 760                                         ($nl) |  # 1. Line continuation
 761                                         ([0-9A-Fa-f]{1,6})$space? |  # 2. character number
 762                                         (.) | # 3. backslash cancelling special meaning
 763                                         () | # 4. backslash at end of string
 764                                 )/xu";
 765                 }
 766                 $value = preg_replace_callback( $decodeRegex,
 767                         array( __CLASS__, 'cssDecodeCallback' ), $value );
 768
 769                 // Reject problematic keywords and control characters
 770                 if ( preg_match( '/[\000-\010\016-\037\177]/', $value ) ) {
 771                         return '/* invalid control char */';
 772                 } elseif ( preg_match( '! expression | filter\s*: | accelerator\s*: | url\s*\( !ix', $value ) ) {
 773                         return '/* insecure input */';
 774                 }
 775                 return $value;
 776         }
 777
 778         static function cssDecodeCallback( $matches ) {
 779                 if ( $matches[1] !== '' ) {
 780                         // Line continuation
 781                         return '';
 782                 } elseif ( $matches[2] !== '' ) {
 783                         $char = codepointToUtf8( hexdec( $matches[2] ) );
 784                 } elseif ( $matches[3] !== '' ) {
 785                         $char = $matches[3];
 786                 } else {
 787                         $char = '\\';
 788                 }
 789                 if ( $char == "\n" || $char == '"' || $char == "'" || $char == '\\' ) {
 790                         // These characters need to be escaped in strings
 791                         // Clean up the escape sequence to avoid parsing errors by clients
 792                         return '\\' . dechex( ord( $char ) ) . ' ';
 793                 } else {
 794                         // Decode unnecessary escape
 795                         return $char;
 796                 }
 797         }
 798
 799         /**
 800          * Take a tag soup fragment listing an HTML element's attributes
 801          * and normalize it to well-formed XML, discarding unwanted attributes.
 802          * Output is safe for further wikitext processing, with escaping of
 803          * values that could trigger problems.
 804          *
 805          * - Normalizes attribute names to lowercase
 806          * - Discards attributes not on a whitelist for the given element
 807          * - Turns broken or invalid entities into plaintext
 808          * - Double-quotes all attribute values
 809          * - Attributes without values are given the name as attribute
 810          * - Double attributes are discarded
 811          * - Unsafe style attributes are discarded
 812          * - Prepends space if there are attributes.
 813          *
 814          * @param $text String
 815          * @param $element String
 816          * @return String
 817          */
 818         static function fixTagAttributes( $text, $element ) {
 819                 if( trim( $text ) == '' ) {
 820                         return '';
 821                 }
 822
 823                 $stripped = Sanitizer::validateTagAttributes(
 824                         Sanitizer::decodeTagAttributes( $text ), $element );
 825
 826                 $attribs = array();
 827                 foreach( $stripped as $attribute => $value ) {
 828                         $encAttribute = htmlspecialchars( $attribute );
 829                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 830
 831                         $attribs[] = "$encAttribute=\"$encValue\"";
 832                 }
 833                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 834         }
 835
 836         /**
 837          * Encode an attribute value for HTML output.
 838          * @param $text String
 839          * @return HTML-encoded text fragment
 840          */
 841         static function encodeAttribute( $text ) {
 842                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 843
 844                 // Whitespace is normalized during attribute decoding,
 845                 // so if we've been passed non-spaces we must encode them
 846                 // ahead of time or they won't be preserved.
 847                 $encValue = strtr( $encValue, array(
 848                         "\n" => '&#10;',
 849                         "\r" => '&#13;',
 850                         "\t" => '&#9;',
 851                 ) );
 852
 853                 return $encValue;
 854         }
 855
 856         /**
 857          * Encode an attribute value for HTML tags, with extra armoring
 858          * against further wiki processing.
 859          * @param $text String
 860          * @return HTML-encoded text fragment
 861          */
 862         static function safeEncodeAttribute( $text ) {
 863                 $encValue = Sanitizer::encodeAttribute( $text );
 864
 865                 # Templates and links may be expanded in later parsing,
 866                 # creating invalid or dangerous output. Suppress this.
 867                 $encValue = strtr( $encValue, array(
 868                         '<'    => '&lt;',   // This should never happen,
 869                         '>'    => '&gt;',   // we've received invalid input
 870                         '"'    => '&quot;', // which should have been escaped.
 871                         '{'    => '&#123;',
 872                         '['    => '&#91;',
 873                         "''"   => '&#39;&#39;',
 874                         'ISBN' => '&#73;SBN',
 875                         'RFC'  => '&#82;FC',
 876                         'PMID' => '&#80;MID',
 877                         '|'    => '&#124;',
 878                         '__'   => '&#95;_',
 879                 ) );
 880
 881                 # Stupid hack
 882                 $encValue = preg_replace_callback(
 883                         '/(' . wfUrlProtocols() . ')/',
 884                         array( 'Sanitizer', 'armorLinksCallback' ),
 885                         $encValue );
 886                 return $encValue;
 887         }
 888
 889         /**
 890          * Given a value, escape it so that it can be used in an id attribute and
 891          * return it.  This will use HTML5 validation if $wgExperimentalHtmlIds is
 892          * true, allowing anything but ASCII whitespace.  Otherwise it will use
 893          * HTML 4 rules, which means a narrow subset of ASCII, with bad characters
 894          * escaped with lots of dots.
 895          *
 896          * To ensure we don't have to bother escaping anything, we also strip ', ",
 897          * & even if $wgExperimentalIds is true.  TODO: Is this the best tactic?
 898          * We also strip # because it upsets IE, and % because it could be
 899          * ambiguous if it's part of something that looks like a percent escape
 900          * (which don't work reliably in fragments cross-browser).
 901          *
 902          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 903          *                                                          in the id and
 904          *                                                          name attributes
 905          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 906          * @see http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#the-id-attribute
 907          *   HTML5 definition of id attribute
 908          *
 909          * @param $id String: id to escape
 910          * @param $options Mixed: string or array of strings (default is array()):
 911          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 912          *       so don't pay attention if the first character isn't valid at the
 913          *       beginning of an id.  Only matters if $wgExperimentalHtmlIds is
 914          *       false.
 915          *   'legacy': Behave the way the old HTML 4-based ID escaping worked even
 916          *       if $wgExperimentalHtmlIds is used, so we can generate extra
 917          *       anchors and links won't break.
 918          * @return String
 919          */
 920         static function escapeId( $id, $options = array() ) {
 921                 global $wgHtml5, $wgExperimentalHtmlIds;
 922                 $options = (array)$options;
 923
 924                 if ( $wgHtml5 && $wgExperimentalHtmlIds && !in_array( 'legacy', $options ) ) {
 925                         $id = Sanitizer::decodeCharReferences( $id );
 926                         $id = preg_replace( '/[ \t\n\r\f_\'"&#%]+/', '_', $id );
 927                         $id = trim( $id, '_' );
 928                         if ( $id === '' ) {
 929                                 # Must have been all whitespace to start with.
 930                                 return '_';
 931                         } else {
 932                                 return $id;
 933                         }
 934                 }
 935
 936                 # HTML4-style escaping
 937                 static $replace = array(
 938                         '%3A' => ':',
 939                         '%' => '.'
 940                 );
 941
 942                 $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 943                 $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 944
 945                 if ( !preg_match( '/^[a-zA-Z]/', $id )
 946                 && !in_array( 'noninitial', $options ) )  {
 947                         // Initial character must be a letter!
 948                         $id = "x$id";
 949                 }
 950                 return $id;
 951         }
 952
 953         /**
 954          * Given a value, escape it so that it can be used as a CSS class and
 955          * return it.
 956          *
 957          * @todo For extra validity, input should be validated UTF-8.
 958          *
 959          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 960          *
 961          * @param $class String
 962          * @return String
 963          */
 964         static function escapeClass( $class ) {
 965                 // Convert ugly stuff to underscores and kill underscores in ugly places
 966                 return rtrim(preg_replace(
 967                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 968                         '_',
 969                         $class ), '_');
 970         }
 971
 972         /**
 973          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 974          * This allows (generally harmless) entities like &#160; to survive.
 975          *
 976          * @param $html String to escape
 977          * @return String: escaped input
 978          */
 979         static function escapeHtmlAllowEntities( $html ) {
 980                 $html = Sanitizer::decodeCharReferences( $html );
 981                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 982                 # hurt.
 983                 $html = htmlspecialchars( $html, ENT_QUOTES );
 984                 return $html;
 985         }
 986
 987         /**
 988          * Regex replace callback for armoring links against further processing.
 989          * @param $matches Array
 990          * @return string
 991          */
 992         private static function armorLinksCallback( $matches ) {
 993                 return str_replace( ':', '&#58;', $matches[1] );
 994         }
 995
 996         /**
 997          * Return an associative array of attribute names and values from
 998          * a partial tag string. Attribute names are forces to lowercase,
 999          * character references are decoded to UTF-8 text.
1000          *
1001          * @param $text String
1002          * @return Array
1003          */
1004         public static function decodeTagAttributes( $text ) {
1005                 if( trim( $text ) == '' ) {
1006                         return array();
1007                 }
1008
1009                 $attribs = array();
1010                 $pairs = array();
1011                 if( !preg_match_all(
1012                         MW_ATTRIBS_REGEX,
1013                         $text,
1014                         $pairs,
1015                         PREG_SET_ORDER ) ) {
1016                         return $attribs;
1017                 }
1018
1019                 foreach( $pairs as $set ) {
1020                         $attribute = strtolower( $set[1] );
1021                         $value = Sanitizer::getTagAttributeCallback( $set );
1022
1023                         // Normalize whitespace
1024                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
1025                         $value = trim( $value );
1026
1027                         // Decode character references
1028                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
1029                 }
1030                 return $attribs;
1031         }
1032
1033         /**
1034          * Pick the appropriate attribute value from a match set from the
1035          * MW_ATTRIBS_REGEX matches.
1036          *
1037          * @param $set Array
1038          * @return String
1039          */
1040         private static function getTagAttributeCallback( $set ) {
1041                 if( isset( $set[6] ) ) {
1042                         # Illegal #XXXXXX color with no quotes.
1043                         return $set[6];
1044                 } elseif( isset( $set[5] ) ) {
1045                         # No quotes.
1046                         return $set[5];
1047                 } elseif( isset( $set[4] ) ) {
1048                         # Single-quoted
1049                         return $set[4];
1050                 } elseif( isset( $set[3] ) ) {
1051                         # Double-quoted
1052                         return $set[3];
1053                 } elseif( !isset( $set[2] ) ) {
1054                         # In XHTML, attributes must have a value.
1055                         # For 'reduced' form, return explicitly the attribute name here.
1056                         return $set[1];
1057                 } else {
1058                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
1059                 }
1060         }
1061
1062         /**
1063          * Normalize whitespace and character references in an XML source-
1064          * encoded text for an attribute value.
1065          *
1066          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
1067          * but note that we're not returning the value, but are returning
1068          * XML source fragments that will be slapped into output.
1069          *
1070          * @param $text String
1071          * @return String
1072          */
1073         private static function normalizeAttributeValue( $text ) {
1074                 return str_replace( '"', '&quot;',
1075                         self::normalizeWhitespace(
1076                                 Sanitizer::normalizeCharReferences( $text ) ) );
1077         }
1078
1079         private static function normalizeWhitespace( $text ) {
1080                 return preg_replace(
1081                         '/\r\n|[\x20\x0d\x0a\x09]/',
1082                         ' ',
1083                         $text );
1084         }
1085
1086         /**
1087          * Normalizes whitespace in a section name, such as might be returned
1088          * by Parser::stripSectionName(), for use in the id's that are used for
1089          * section links.
1090          *
1091          * @param $section String
1092          * @return String
1093          */
1094         static function normalizeSectionNameWhitespace( $section ) {
1095                 return trim( preg_replace( '/[ _]+/', ' ', $section ) );
1096         }
1097
1098         /**
1099          * Ensure that any entities and character references are legal
1100          * for XML and XHTML specifically. Any stray bits will be
1101          * &amp;-escaped to result in a valid text fragment.
1102          *
1103          * a. any named char refs must be known in XHTML
1104          * b. any numeric char refs must be legal chars, not invalid or forbidden
1105          * c. use &#x, not &#X
1106          * d. fix or reject non-valid attributes
1107          *
1108          * @param $text String
1109          * @return String
1110          * @private
1111          */
1112         static function normalizeCharReferences( $text ) {
1113                 return preg_replace_callback(
1114                         MW_CHAR_REFS_REGEX,
1115                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
1116                         $text );
1117         }
1118         /**
1119          * @param $matches String
1120          * @return String
1121          */
1122         static function normalizeCharReferencesCallback( $matches ) {
1123                 $ret = null;
1124                 if( $matches[1] != '' ) {
1125                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1126                 } elseif( $matches[2] != '' ) {
1127                         $ret = Sanitizer::decCharReference( $matches[2] );
1128                 } elseif( $matches[3] != ''  ) {
1129                         $ret = Sanitizer::hexCharReference( $matches[3] );
1130                 } elseif( $matches[4] != '' ) {
1131                         $ret = Sanitizer::hexCharReference( $matches[4] );
1132                 }
1133                 if( is_null( $ret ) ) {
1134                         return htmlspecialchars( $matches[0] );
1135                 } else {
1136                         return $ret;
1137                 }
1138         }
1139
1140         /**
1141          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1142          * return the named entity reference as is. If the entity is a
1143          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1144          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1145          *
1146          * @param $name String
1147          * @return String
1148          */
1149         static function normalizeEntity( $name ) {
1150                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1151                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1152                         return "&{$wgHtmlEntityAliases[$name]};";
1153                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1154                         return "&$name;";
1155                 } else {
1156                         return "&amp;$name;";
1157                 }
1158         }
1159
1160         static function decCharReference( $codepoint ) {
1161                 $point = intval( $codepoint );
1162                 if( Sanitizer::validateCodepoint( $point ) ) {
1163                         return sprintf( '&#%d;', $point );
1164                 } else {
1165                         return null;
1166                 }
1167         }
1168
1169         static function hexCharReference( $codepoint ) {
1170                 $point = hexdec( $codepoint );
1171                 if( Sanitizer::validateCodepoint( $point ) ) {
1172                         return sprintf( '&#x%x;', $point );
1173                 } else {
1174                         return null;
1175                 }
1176         }
1177
1178         /**
1179          * Returns true if a given Unicode codepoint is a valid character in XML.
1180          * @param $codepoint Integer
1181          * @return Boolean
1182          */
1183         private static function validateCodepoint( $codepoint ) {
1184                 return ($codepoint ==    0x09)
1185                         || ($codepoint ==    0x0a)
1186                         || ($codepoint ==    0x0d)
1187                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1188                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1189                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1190         }
1191
1192         /**
1193          * Decode any character references, numeric or named entities,
1194          * in the text and return a UTF-8 string.
1195          *
1196          * @param $text String
1197          * @return String
1198          */
1199         public static function decodeCharReferences( $text ) {
1200                 return preg_replace_callback(
1201                         MW_CHAR_REFS_REGEX,
1202                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1203                         $text );
1204         }
1205
1206         /**
1207          * Decode any character references, numeric or named entities,
1208          * in the next and normalize the resulting string. (bug 14952)
1209          *
1210          * This is useful for page titles, not for text to be displayed,
1211          * MediaWiki allows HTML entities to escape normalization as a feature.
1212          *
1213          * @param $text String (already normalized, containing entities)
1214          * @return String (still normalized, without entities)
1215          */
1216         public static function decodeCharReferencesAndNormalize( $text ) {
1217                 global $wgContLang;
1218                 $text = preg_replace_callback(
1219                         MW_CHAR_REFS_REGEX,
1220                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1221                         $text, /* limit */ -1, $count );
1222
1223                 if ( $count ) {
1224                         return $wgContLang->normalize( $text );
1225                 } else {
1226                         return $text;
1227                 }
1228         }
1229
1230         /**
1231          * @param $matches String
1232          * @return String
1233          */
1234         static function decodeCharReferencesCallback( $matches ) {
1235                 if( $matches[1] != '' ) {
1236                         return Sanitizer::decodeEntity( $matches[1] );
1237                 } elseif( $matches[2] != '' ) {
1238                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1239                 } elseif( $matches[3] != ''  ) {
1240                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1241                 } elseif( $matches[4] != '' ) {
1242                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1243                 }
1244                 # Last case should be an ampersand by itself
1245                 return $matches[0];
1246         }
1247
1248         /**
1249          * Return UTF-8 string for a codepoint if that is a valid
1250          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1251          * @param $codepoint Integer
1252          * @return String
1253          * @private
1254          */
1255         static function decodeChar( $codepoint ) {
1256                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1257                         return codepointToUtf8( $codepoint );
1258                 } else {
1259                         return UTF8_REPLACEMENT;
1260                 }
1261         }
1262
1263         /**
1264          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1265          * return the UTF-8 encoding of that character. Otherwise, returns
1266          * pseudo-entity source (eg &foo;)
1267          *
1268          * @param $name Strings
1269          * @return String
1270          */
1271         static function decodeEntity( $name ) {
1272                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1273                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1274                         $name = $wgHtmlEntityAliases[$name];
1275                 }
1276                 if( isset( $wgHtmlEntities[$name] ) ) {
1277                         return codepointToUtf8( $wgHtmlEntities[$name] );
1278                 } else {
1279                         return "&$name;";
1280                 }
1281         }
1282
1283         /**
1284          * Fetch the whitelist of acceptable attributes for a given element name.
1285          *
1286          * @param $element String
1287          * @return Array
1288          */
1289         static function attributeWhitelist( $element ) {
1290                 static $list;
1291                 if( !isset( $list ) ) {
1292                         $list = Sanitizer::setupAttributeWhitelist();
1293                 }
1294                 return isset( $list[$element] )
1295                         ? $list[$element]
1296                         : array();
1297         }
1298
1299         /**
1300          * Foreach array key (an allowed HTML element), return an array
1301          * of allowed attributes
1302          * @return Array
1303          */
1304         static function setupAttributeWhitelist() {
1305                 global $wgAllowRdfaAttributes, $wgHtml5, $wgAllowMicrodataAttributes;
1306
1307                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1308
1309                 if ( $wgAllowRdfaAttributes ) {
1310                         #RDFa attributes as specified in section 9 of http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014
1311                         $common = array_merge( $common, array(
1312                             'about', 'property', 'resource', 'datatype', 'typeof',
1313                         ) );
1314                 }
1315
1316                 if ( $wgHtml5 && $wgAllowMicrodataAttributes ) {
1317                         # add HTML5 microdata tages as pecified by http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html#the-microdata-model
1318                         $common = array_merge( $common, array(
1319                             'itemid', 'itemprop', 'itemref', 'itemscope', 'itemtype'
1320                         ) );
1321                 }
1322
1323                 $block = array_merge( $common, array( 'align' ) );
1324                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1325                 $tablecell = array( 'abbr',
1326                                     'axis',
1327                                     'headers',
1328                                     'scope',
1329                                     'rowspan',
1330                                     'colspan',
1331                                     'nowrap', # deprecated
1332                                     'width',  # deprecated
1333                                     'height', # deprecated
1334                                     'bgcolor' # deprecated
1335                                     );
1336
1337                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1338                 # See: http://www.w3.org/TR/html4/
1339                 $whitelist = array (
1340                         # 7.5.4
1341                         'div'        => $block,
1342                         'center'     => $common, # deprecated
1343                         'span'       => $block, # ??
1344
1345                         # 7.5.5
1346                         'h1'         => $block,
1347                         'h2'         => $block,
1348                         'h3'         => $block,
1349                         'h4'         => $block,
1350                         'h5'         => $block,
1351                         'h6'         => $block,
1352
1353                         # 7.5.6
1354                         # address
1355
1356                         # 8.2.4
1357                         # bdo
1358
1359                         # 9.2.1
1360                         'em'         => $common,
1361                         'strong'     => $common,
1362                         'cite'       => $common,
1363                         'dfn'        => $common,
1364                         'code'       => $common,
1365                         'samp'       => $common,
1366                         'kbd'        => $common,
1367                         'var'        => $common,
1368                         'abbr'       => $common,
1369                         # acronym
1370
1371                         # 9.2.2
1372                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1373                         # q
1374
1375                         # 9.2.3
1376                         'sub'        => $common,
1377                         'sup'        => $common,
1378
1379                         # 9.3.1
1380                         'p'          => $block,
1381
1382                         # 9.3.2
1383                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1384
1385                         # 9.3.4
1386                         'pre'        => array_merge( $common, array( 'width' ) ),
1387
1388                         # 9.4
1389                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1390                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1391
1392                         # 10.2
1393                         'ul'         => array_merge( $common, array( 'type' ) ),
1394                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1395                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1396
1397                         # 10.3
1398                         'dl'         => $common,
1399                         'dd'         => $common,
1400                         'dt'         => $common,
1401
1402                         # 11.2.1
1403                         'table'      => array_merge( $common,
1404                                                                 array( 'summary', 'width', 'border', 'frame',
1405                                                                                 'rules', 'cellspacing', 'cellpadding',
1406                                                                                 'align', 'bgcolor',
1407                                                                 ) ),
1408
1409                         # 11.2.2
1410                         'caption'    => array_merge( $common, array( 'align' ) ),
1411
1412                         # 11.2.3
1413                         'thead'      => array_merge( $common, $tablealign ),
1414                         'tfoot'      => array_merge( $common, $tablealign ),
1415                         'tbody'      => array_merge( $common, $tablealign ),
1416
1417                         # 11.2.4
1418                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1419                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1420
1421                         # 11.2.5
1422                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1423
1424                         # 11.2.6
1425                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1426                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1427
1428                         # 12.2 # NOTE: <a> is not allowed directly, but the attrib whitelist is used from the Parser object
1429                         'a'          => array_merge( $common, array( 'href', 'rel', 'rev' ) ), # rel/rev esp. for RDFa
1430
1431                         # 13.2
1432                         # Not usually allowed, but may be used for extension-style hooks
1433                         # such as <math> when it is rasterized, or if $wgAllowImageTag is
1434                         # true
1435                         'img'        => array_merge( $common, array( 'alt', 'src', 'width', 'height' ) ),
1436
1437                         # 15.2.1
1438                         'tt'         => $common,
1439                         'b'          => $common,
1440                         'i'          => $common,
1441                         'big'        => $common,
1442                         'small'      => $common,
1443                         'strike'     => $common,
1444                         's'          => $common,
1445                         'u'          => $common,
1446
1447                         # 15.2.2
1448                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1449                         # basefont
1450
1451                         # 15.3
1452                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1453
1454                         # XHTML Ruby annotation text module, simple ruby only.
1455                         # http://www.w3c.org/TR/ruby/
1456                         'ruby'       => $common,
1457                         # rbc
1458                         # rtc
1459                         'rb'         => $common,
1460                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1461                         'rp'         => $common,
1462
1463                         # MathML root element, where used for extensions
1464                         # 'title' may not be 100% valid here; it's XHTML
1465                         # http://www.w3.org/TR/REC-MathML/
1466                         'math'       => array( 'class', 'style', 'id', 'title' ),
1467                         );
1468                 return $whitelist;
1469         }
1470
1471         /**
1472          * Take a fragment of (potentially invalid) HTML and return
1473          * a version with any tags removed, encoded as plain text.
1474          *
1475          * Warning: this return value must be further escaped for literal
1476          * inclusion in HTML output as of 1.10!
1477          *
1478          * @param $text String: HTML fragment
1479          * @return String
1480          */
1481         static function stripAllTags( $text ) {
1482                 # Actual <tags>
1483                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1484
1485                 # Normalize &entities and whitespace
1486                 $text = self::decodeCharReferences( $text );
1487                 $text = self::normalizeWhitespace( $text );
1488
1489                 return $text;
1490         }
1491
1492         /**
1493          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1494          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1495          * PHP 5.1 doesn't.
1496          *
1497          * Use for passing XHTML fragments to PHP's XML parsing functions
1498          *
1499          * @return String
1500          */
1501         static function hackDocType() {
1502                 global $wgHtmlEntities;
1503                 $out = "<!DOCTYPE html [\n";
1504                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1505                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1506                 }
1507                 $out .= "]>\n";
1508                 return $out;
1509         }
1510
1511         static function cleanUrl( $url ) {
1512                 # Normalize any HTML entities in input. They will be
1513                 # re-escaped by makeExternalLink().
1514                 $url = Sanitizer::decodeCharReferences( $url );
1515
1516                 # Escape any control characters introduced by the above step
1517                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F\|]/e', "urlencode('\\0')", $url );
1518
1519                 # Validate hostname portion
1520                 $matches = array();
1521                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1522                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1523
1524                         // Characters that will be ignored in IDNs.
1525                         // http://tools.ietf.org/html/3454#section-3.1
1526                         // Strip them before further processing so blacklists and such work.
1527                         $strip = "/
1528                                 \\s|          # general whitespace
1529                                 \xc2\xad|     # 00ad SOFT HYPHEN
1530                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1531                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1532                                 \xe2\x81\xa0| # 2060 WORD JOINER
1533                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1534                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1535                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1536                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1537                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1538                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1539                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1540                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1541                                 /xuD";
1542
1543                         $host = preg_replace( $strip, '', $host );
1544
1545                         // @todo Fixme: validate hostnames here
1546
1547                         return $protocol . $host . $rest;
1548                 } else {
1549                         return $url;
1550                 }
1551         }
1552
1553 }