lib/kses.php

   1 <?php
   2
   3 # kses 0.2.2 - HTML/XHTML filter that only allows some elements and attributes
   4 # Copyright (C) 2002, 2003, 2005  Ulf Harnhammar
   5 #
   6 # This program is free software and open source software; you can redistribute
   7 # it and/or modify it under the terms of the GNU General Public License as
   8 # published by the Free Software Foundation; either version 2 of the License,
   9 # or (at your option) any later version.
  10 #
  11 # This program is distributed in the hope that it will be useful, but WITHOUT
  12 # ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  13 # FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
  14 # more details.
  15 #
  16 # You should have received a copy of the GNU General Public License along
  17 # with this program; if not, write to the Free Software Foundation, Inc.,
  18 # 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA  or visit
  19 # http://www.gnu.org/licenses/gpl.html
  20 #
  21 # *** CONTACT INFORMATION ***
  22 #
  23 # E-mail:      metaur at users dot sourceforge dot net
  24 # Web page:    http://sourceforge.net/projects/kses
  25 # Paper mail:  Ulf Harnhammar
  26 #              Ymergatan 17 C
  27 #              753 25  Uppsala
  28 #              SWEDEN
  29 #
  30 # [kses strips evil scripts!]
  31
  32
  33 function kses($string, $allowed_html, $allowed_protocols =
  34                array('http', 'https', 'ftp', 'news', 'nntp', 'telnet',
  35                      'gopher', 'mailto'))
  36 ###############################################################################
  37 # This function makes sure that only the allowed HTML element names, attribute
  38 # names and attribute values plus only sane HTML entities will occur in
  39 # $string. You have to remove any slashes from PHP's magic quotes before you
  40 # call this function.
  41 ###############################################################################
  42 {
  43   $string = kses_no_null($string);
  44   $string = kses_js_entities($string);
  45   $string = kses_normalize_entities($string);
  46   $string = kses_hook($string);
  47   $allowed_html_fixed = kses_array_lc($allowed_html);
  48   return kses_split($string, $allowed_html_fixed, $allowed_protocols);
  49 } # function kses
  50
  51
  52 function kses_hook($string)
  53 ###############################################################################
  54 # You add any kses hooks here.
  55 ###############################################################################
  56 {
  57   return $string;
  58 } # function kses_hook
  59
  60
  61 function kses_version()
  62 ###############################################################################
  63 # This function returns kses' version number.
  64 ###############################################################################
  65 {
  66   return '0.2.2';
  67 } # function kses_version
  68
  69
  70 function kses_split($string, $allowed_html, $allowed_protocols)
  71 ###############################################################################
  72 # This function searches for HTML tags, no matter how malformed. It also
  73 # matches stray ">" characters.
  74 ###############################################################################
  75 {
  76   return preg_replace('%(<'.   # EITHER: <
  77                       '[^>]*'. # things that aren't >
  78                       '(>|$)'. # > or end of string
  79                       '|>)%e', # OR: just a >
  80                       "kses_split2('\\1', \$allowed_html, ".
  81                       '$allowed_protocols)',
  82                       $string);
  83 } # function kses_split
  84
  85
  86 function kses_split2($string, $allowed_html, $allowed_protocols)
  87 ###############################################################################
  88 # This function does a lot of work. It rejects some very malformed things
  89 # like <:::>. It returns an empty string, if the element isn't allowed (look
  90 # ma, no strip_tags()!). Otherwise it splits the tag into an element and an
  91 # attribute list.
  92 ###############################################################################
  93 {
  94   $string = kses_stripslashes($string);
  95
  96   if (substr($string, 0, 1) != '<')
  97     return '&gt;';
  98     # It matched a ">" character
  99
 100   if (!preg_match('%^<\s*(/\s*)?([a-zA-Z0-9]+)([^>]*)>?$%', $string, $matches))
 101     return '';
 102     # It's seriously malformed
 103
 104   $slash = trim($matches[1]);
 105   $elem = $matches[2];
 106   $attrlist = $matches[3];
 107
 108   if (!@isset($allowed_html[strtolower($elem)]))
 109     return '';
 110     # They are using a not allowed HTML element
 111
 112   if ($slash != '')
 113     return "<$slash$elem>";
 114   # No attributes are allowed for closing elements
 115
 116   return kses_attr("$slash$elem", $attrlist, $allowed_html,
 117                    $allowed_protocols);
 118 } # function kses_split2
 119
 120
 121 function kses_attr($element, $attr, $allowed_html, $allowed_protocols)
 122 ###############################################################################
 123 # This function removes all attributes, if none are allowed for this element.
 124 # If some are allowed it calls kses_hair() to split them further, and then it
 125 # builds up new HTML code from the data that kses_hair() returns. It also
 126 # removes "<" and ">" characters, if there are any left. One more thing it
 127 # does is to check if the tag has a closing XHTML slash, and if it does,
 128 # it puts one in the returned code as well.
 129 ###############################################################################
 130 {
 131 # Is there a closing XHTML slash at the end of the attributes?
 132
 133   $xhtml_slash = '';
 134   if (preg_match('%\s/\s*$%', $attr))
 135     $xhtml_slash = ' /';
 136
 137 # Are any attributes allowed at all for this element?
 138
 139   if (@count($allowed_html[strtolower($element)]) == 0)
 140     return "<$element$xhtml_slash>";
 141
 142 # Split it
 143
 144   $attrarr = kses_hair($attr, $allowed_protocols);
 145
 146 # Go through $attrarr, and save the allowed attributes for this element
 147 # in $attr2
 148
 149   $attr2 = '';
 150
 151   foreach ($attrarr as $arreach)
 152   {
 153     if (!@isset($allowed_html[strtolower($element)]
 154                             [strtolower($arreach['name'])]))
 155       continue; # the attribute is not allowed
 156
 157     $current = $allowed_html[strtolower($element)]
 158                             [strtolower($arreach['name'])];
 159
 160     if (!is_array($current))
 161       $attr2 .= ' '.$arreach['whole'];
 162     # there are no checks
 163
 164     else
 165     {
 166     # there are some checks
 167       $ok = true;
 168       foreach ($current as $currkey => $currval)
 169         if (!kses_check_attr_val($arreach['value'], $arreach['vless'],
 170                                  $currkey, $currval))
 171         { $ok = false; break; }
 172
 173       if ($ok)
 174         $attr2 .= ' '.$arreach['whole']; # it passed them
 175     } # if !is_array($current)
 176   } # foreach
 177
 178 # Remove any "<" or ">" characters
 179
 180   $attr2 = preg_replace('/[<>]/', '', $attr2);
 181
 182   return "<$element$attr2$xhtml_slash>";
 183 } # function kses_attr
 184
 185
 186 function kses_hair($attr, $allowed_protocols)
 187 ###############################################################################
 188 # This function does a lot of work. It parses an attribute list into an array
 189 # with attribute data, and tries to do the right thing even if it gets weird
 190 # input. It will add quotes around attribute values that don't have any quotes
 191 # or apostrophes around them, to make it easier to produce HTML code that will
 192 # conform to W3C's HTML specification. It will also remove bad URL protocols
 193 # from attribute values.
 194 ###############################################################################
 195 {
 196   $attrarr = array();
 197   $mode = 0;
 198   $attrname = '';
 199
 200 # Loop through the whole attribute list
 201
 202   while (strlen($attr) != 0)
 203   {
 204     $working = 0; # Was the last operation successful?
 205
 206     switch ($mode)
 207     {
 208       case 0: # attribute name, href for instance
 209
 210         if (preg_match('/^([-a-zA-Z]+)/', $attr, $match))
 211         {
 212           $attrname = $match[1];
 213           $working = $mode = 1;
 214           $attr = preg_replace('/^[-a-zA-Z]+/', '', $attr);
 215         }
 216
 217         break;
 218
 219       case 1: # equals sign or valueless ("selected")
 220
 221         if (preg_match('/^\s*=\s*/', $attr)) # equals sign
 222         {
 223           $working = 1; $mode = 2;
 224           $attr = preg_replace('/^\s*=\s*/', '', $attr);
 225           break;
 226         }
 227
 228         if (preg_match('/^\s+/', $attr)) # valueless
 229         {
 230           $working = 1; $mode = 0;
 231           $attrarr[] = array
 232                         ('name'  => $attrname,
 233                          'value' => '',
 234                          'whole' => $attrname,
 235                          'vless' => 'y');
 236           $attr = preg_replace('/^\s+/', '', $attr);
 237         }
 238
 239         break;
 240
 241       case 2: # attribute value, a URL after href= for instance
 242
 243         if (preg_match('/^"([^"]*)"(\s+|$)/', $attr, $match))
 244          # "value"
 245         {
 246           $thisval = kses_bad_protocol($match[1], $allowed_protocols);
 247
 248           $attrarr[] = array
 249                         ('name'  => $attrname,
 250                          'value' => $thisval,
 251                          'whole' => "$attrname=\"$thisval\"",
 252                          'vless' => 'n');
 253           $working = 1; $mode = 0;
 254           $attr = preg_replace('/^"[^"]*"(\s+|$)/', '', $attr);
 255           break;
 256         }
 257
 258         if (preg_match("/^'([^']*)'(\s+|$)/", $attr, $match))
 259          # 'value'
 260         {
 261           $thisval = kses_bad_protocol($match[1], $allowed_protocols);
 262
 263           $attrarr[] = array
 264                         ('name'  => $attrname,
 265                          'value' => $thisval,
 266                          'whole' => "$attrname='$thisval'",
 267                          'vless' => 'n');
 268           $working = 1; $mode = 0;
 269           $attr = preg_replace("/^'[^']*'(\s+|$)/", '', $attr);
 270           break;
 271         }
 272
 273         if (preg_match("%^([^\s\"']+)(\s+|$)%", $attr, $match))
 274          # value
 275         {
 276           $thisval = kses_bad_protocol($match[1], $allowed_protocols);
 277
 278           $attrarr[] = array
 279                         ('name'  => $attrname,
 280                          'value' => $thisval,
 281                          'whole' => "$attrname=\"$thisval\"",
 282                          'vless' => 'n');
 283                          # We add quotes to conform to W3C's HTML spec.
 284           $working = 1; $mode = 0;
 285           $attr = preg_replace("%^[^\s\"']+(\s+|$)%", '', $attr);
 286         }
 287
 288         break;
 289     } # switch
 290
 291     if ($working == 0) # not well formed, remove and try again
 292     {
 293       $attr = kses_html_error($attr);
 294       $mode = 0;
 295     }
 296   } # while
 297
 298   if ($mode == 1)
 299   # special case, for when the attribute list ends with a valueless
 300   # attribute like "selected"
 301     $attrarr[] = array
 302                   ('name'  => $attrname,
 303                    'value' => '',
 304                    'whole' => $attrname,
 305                    'vless' => 'y');
 306
 307   return $attrarr;
 308 } # function kses_hair
 309
 310
 311 function kses_check_attr_val($value, $vless, $checkname, $checkvalue)
 312 ###############################################################################
 313 # This function performs different checks for attribute values. The currently
 314 # implemented checks are "maxlen", "minlen", "maxval", "minval" and "valueless"
 315 # with even more checks to come soon.
 316 ###############################################################################
 317 {
 318   $ok = true;
 319
 320   switch (strtolower($checkname))
 321   {
 322     case 'maxlen':
 323     # The maxlen check makes sure that the attribute value has a length not
 324     # greater than the given value. This can be used to avoid Buffer Overflows
 325     # in WWW clients and various Internet servers.
 326
 327       if (strlen($value) > $checkvalue)
 328         $ok = false;
 329       break;
 330
 331     case 'minlen':
 332     # The minlen check makes sure that the attribute value has a length not
 333     # smaller than the given value.
 334
 335       if (strlen($value) < $checkvalue)
 336         $ok = false;
 337       break;
 338
 339     case 'maxval':
 340     # The maxval check does two things: it checks that the attribute value is
 341     # an integer from 0 and up, without an excessive amount of zeroes or
 342     # whitespace (to avoid Buffer Overflows). It also checks that the attribute
 343     # value is not greater than the given value.
 344     # This check can be used to avoid Denial of Service attacks.
 345
 346       if (!preg_match('/^\s{0,6}[0-9]{1,6}\s{0,6}$/', $value))
 347         $ok = false;
 348       if ($value > $checkvalue)
 349         $ok = false;
 350       break;
 351
 352     case 'minval':
 353     # The minval check checks that the attribute value is a positive integer,
 354     # and that it is not smaller than the given value.
 355
 356       if (!preg_match('/^\s{0,6}[0-9]{1,6}\s{0,6}$/', $value))
 357         $ok = false;
 358       if ($value < $checkvalue)
 359         $ok = false;
 360       break;
 361
 362     case 'valueless':
 363     # The valueless check checks if the attribute has a value
 364     # (like <a href="blah">) or not (<option selected>). If the given value
 365     # is a "y" or a "Y", the attribute must not have a value.
 366     # If the given value is an "n" or an "N", the attribute must have one.
 367
 368       if (strtolower($checkvalue) != $vless)
 369         $ok = false;
 370       break;
 371   } # switch
 372
 373   return $ok;
 374 } # function kses_check_attr_val
 375
 376
 377 function kses_bad_protocol($string, $allowed_protocols)
 378 ###############################################################################
 379 # This function removes all non-allowed protocols from the beginning of
 380 # $string. It ignores whitespace and the case of the letters, and it does
 381 # understand HTML entities. It does its work in a while loop, so it won't be
 382 # fooled by a string like "javascript:javascript:alert(57)".
 383 ###############################################################################
 384 {
 385   $string = kses_no_null($string);
 386   $string = preg_replace('/\xad+/', '', $string); # deals with Opera "feature"
 387   $string2 = $string.'a';
 388
 389   while ($string != $string2)
 390   {
 391     $string2 = $string;
 392     $string = kses_bad_protocol_once($string, $allowed_protocols);
 393   } # while
 394
 395   return $string;
 396 } # function kses_bad_protocol
 397
 398
 399 function kses_no_null($string)
 400 ###############################################################################
 401 # This function removes any NULL characters in $string.
 402 ###############################################################################
 403 {
 404   $string = preg_replace('/\0+/', '', $string);
 405   $string = preg_replace('/(\\\\0)+/', '', $string);
 406
 407   return $string;
 408 } # function kses_no_null
 409
 410
 411 function kses_stripslashes($string)
 412 ###############################################################################
 413 # This function changes the character sequence  \"  to just  "
 414 # It leaves all other slashes alone. It's really weird, but the quoting from
 415 # preg_replace(//e) seems to require this.
 416 ###############################################################################
 417 {
 418   return preg_replace('%\\\\"%', '"', $string);
 419 } # function kses_stripslashes
 420
 421
 422 function kses_array_lc($inarray)
 423 ###############################################################################
 424 # This function goes through an array, and changes the keys to all lower case.
 425 ###############################################################################
 426 {
 427   $outarray = array();
 428
 429   foreach ($inarray as $inkey => $inval)
 430   {
 431     $outkey = strtolower($inkey);
 432     $outarray[$outkey] = array();
 433
 434     foreach ($inval as $inkey2 => $inval2)
 435     {
 436       $outkey2 = strtolower($inkey2);
 437       $outarray[$outkey][$outkey2] = $inval2;
 438     } # foreach $inval
 439   } # foreach $inarray
 440
 441   return $outarray;
 442 } # function kses_array_lc
 443
 444
 445 function kses_js_entities($string)
 446 ###############################################################################
 447 # This function removes the HTML JavaScript entities found in early versions of
 448 # Netscape 4.
 449 ###############################################################################
 450 {
 451   return preg_replace('%&\s*\{[^}]*(\}\s*;?|$)%', '', $string);
 452 } # function kses_js_entities
 453
 454
 455 function kses_html_error($string)
 456 ###############################################################################
 457 # This function deals with parsing errors in kses_hair(). The general plan is
 458 # to remove everything to and including some whitespace, but it deals with
 459 # quotes and apostrophes as well.
 460 ###############################################################################
 461 {
 462   return preg_replace('/^("[^"]*("|$)|\'[^\']*(\'|$)|\S)*\s*/', '', $string);
 463 } # function kses_html_error
 464
 465
 466 function kses_bad_protocol_once($string, $allowed_protocols)
 467 ###############################################################################
 468 # This function searches for URL protocols at the beginning of $string, while
 469 # handling whitespace and HTML entities.
 470 ###############################################################################
 471 {
 472   return preg_replace('/^((&[^;]*;|[\sA-Za-z0-9])*)'.
 473                       '(:|&#0*58;|&#[Xx]3[Aa];)\s*/e',
 474                       'kses_bad_protocol_once2("\\1", $allowed_protocols)',
 475                       $string);
 476 } # function kses_bad_protocol_once
 477
 478
 479 function kses_bad_protocol_once2($string, $allowed_protocols)
 480 ###############################################################################
 481 # This function processes URL protocols, checks to see if they're in the white-
 482 # list or not, and returns different data depending on the answer.
 483 ###############################################################################
 484 {
 485   $string2 = kses_decode_entities($string);
 486   $string2 = preg_replace('/\s/', '', $string2);
 487   $string2 = kses_no_null($string2);
 488   $string2 = preg_replace('/\xad+/', '', $string2);
 489    # deals with Opera "feature"
 490   $string2 = strtolower($string2);
 491
 492   $allowed = false;
 493   foreach ($allowed_protocols as $one_protocol)
 494     if (strtolower($one_protocol) == $string2)
 495     {
 496       $allowed = true;
 497       break;
 498     }
 499
 500   if ($allowed)
 501     return "$string2:";
 502   else
 503     return '';
 504 } # function kses_bad_protocol_once2
 505
 506
 507 function kses_normalize_entities($string)
 508 ###############################################################################
 509 # This function normalizes HTML entities. It will convert "AT&T" to the correct
 510 # "AT&amp;T", "&#00058;" to "&#58;", "&#XYZZY;" to "&amp;#XYZZY;" and so on.
 511 ###############################################################################
 512 {
 513 # Disarm all entities by converting & to &amp;
 514
 515   $string = str_replace('&', '&amp;', $string);
 516
 517 # Change back the allowed entities in our entity whitelist
 518
 519   $string = preg_replace('/&amp;([A-Za-z][A-Za-z0-9]{0,19});/',
 520                          '&\\1;', $string);
 521   $string = preg_replace('/&amp;#0*([0-9]{1,5});/e',
 522                          'kses_normalize_entities2("\\1")', $string);
 523   $string = preg_replace('/&amp;#([Xx])0*(([0-9A-Fa-f]{2}){1,2});/',
 524                          '&#\\1\\2;', $string);
 525
 526   return $string;
 527 } # function kses_normalize_entities
 528
 529
 530 function kses_normalize_entities2($i)
 531 ###############################################################################
 532 # This function helps kses_normalize_entities() to only accept 16 bit values
 533 # and nothing more for &#number; entities.
 534 ###############################################################################
 535 {
 536   return (($i > 65535) ? "&amp;#$i;" : "&#$i;");
 537 } # function kses_normalize_entities2
 538
 539
 540 function kses_decode_entities($string)
 541 ###############################################################################
 542 # This function decodes numeric HTML entities (&#65; and &#x41;). It doesn't
 543 # do anything with other entities like &auml;, but we don't need them in the
 544 # URL protocol whitelisting system anyway.
 545 ###############################################################################
 546 {
 547   $string = preg_replace('/&#([0-9]+);/e', 'chr("\\1")', $string);
 548   $string = preg_replace('/&#[Xx]([0-9A-Fa-f]+);/e', 'chr(hexdec("\\1"))',
 549                          $string);
 550
 551   return $string;
 552 } # function kses_decode_entities
 553
 554 ?>