functions/mime.php

   1 <?php
   2
   3 /**
   4  * mime.php
   5  *
   6  * This contains the functions necessary to detect and decode MIME
   7  * messages.
   8  *
   9  * @copyright 1999-2012 The SquirrelMail Project Team
  10  * @license http://opensource.org/licenses/gpl-license.php GNU Public License
  11  * @version $Id$
  12  * @package squirrelmail
  13  */
  14
  15 /**
  16  * dependency information
  17    functions       dependency
  18    mime_structure
  19         class/mime/Message.class.php
  20             Message::parseStructure
  21         functions/page_header.php
  22             displayPageHeader
  23         functions/display_messages.php
  24             plain_error_message
  25    mime_fetch_body
  26         functions/imap_general.php
  27             sqimap_run_command
  28    mime_print_body_lines
  29
  30
  31
  32 functions/imap.php
  33 functions/attachment_common.php
  34 functions/display_messages.php
  35
  36 magicHtml => url_parser
  37 translateText => url_parser
  38
  39 */
  40
  41
  42 /* -------------------------------------------------------------------------- */
  43 /* MIME DECODING                                                              */
  44 /* -------------------------------------------------------------------------- */
  45
  46 /**
  47  * Get the MIME structure
  48  *
  49  * This function gets the structure of a message and stores it in the "message" class.
  50  * It will return this object for use with all relevant header information and
  51  * fully parsed into the standard "message" object format.
  52  */
  53 function mime_structure ($bodystructure, $flags=array()) {
  54
  55     /* Isolate the body structure and remove beginning and end parenthesis. */
  56     $read = trim(substr ($bodystructure, strpos(strtolower($bodystructure), 'bodystructure') + 13));
  57     $read = trim(substr ($read, 0, -1));
  58     $i = 0;
  59     $msg = Message::parseStructure($read,$i);
  60
  61     if (!is_object($msg)) {
  62         global $color, $mailbox;
  63         displayPageHeader( $color, $mailbox );
  64         $errormessage  = _("SquirrelMail could not decode the bodystructure of the message");
  65         $errormessage .= '<br />'._("The bodystructure provided by your IMAP server:").'<br /><br />';
  66         $errormessage .= '<pre>' . sm_encode_html_special_chars($read) . '</pre>';
  67         plain_error_message( $errormessage );
  68         echo '</body></html>';
  69         exit;
  70     }
  71     if (count($flags)) {
  72         foreach ($flags as $flag) {
  73 //FIXME: please document why it is we have to check the first char of the flag but we then go ahead and do a full string comparison anyway.  Is this a speed enhancement?  If not, let's keep it simple and just compare the full string and forget the switch block.
  74             $char = strtoupper($flag{1});
  75             switch ($char) {
  76                 case 'S':
  77                     if (strtolower($flag) == '\\seen') {
  78                         $msg->is_seen = true;
  79                     }
  80                     break;
  81                 case 'A':
  82                     if (strtolower($flag) == '\\answered') {
  83                         $msg->is_answered = true;
  84                     }
  85                     break;
  86                 case 'D':
  87                     if (strtolower($flag) == '\\deleted') {
  88                         $msg->is_deleted = true;
  89                     }
  90                     break;
  91                 case 'F':
  92                     if (strtolower($flag) == '\\flagged') {
  93                         $msg->is_flagged = true;
  94                     }
  95                     else if (strtolower($flag) == '$forwarded') {
  96                         $msg->is_forwarded = true;
  97                     }
  98                     break;
  99                 case 'M':
 100                     if (strtolower($flag) == '$mdnsent') {
 101                         $msg->is_mdnsent = true;
 102                     }
 103                     break;
 104                 default:
 105                     break;
 106             }
 107         }
 108     }
 109     //    listEntities($msg);
 110     return $msg;
 111 }
 112
 113
 114
 115 /* This starts the parsing of a particular structure.  It is called recursively,
 116  * so it can be passed different structures.  It returns an object of type
 117  * $message.
 118  * First, it checks to see if it is a multipart message.  If it is, then it
 119  * handles that as it sees is necessary.  If it is just a regular entity,
 120  * then it parses it and adds the necessary header information (by calling out
 121  * to mime_get_elements()
 122  */
 123
 124 function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) {
 125     /* Do a bit of error correction.  If we couldn't find the entity id, just guess
 126      * that it is the first one.  That is usually the case anyway.
 127      */
 128
 129     if (!$ent_id) {
 130         $cmd = "FETCH $id BODY[]";
 131     } else {
 132         $cmd = "FETCH $id BODY[$ent_id]";
 133     }
 134
 135     if ($fetch_size!=0) $cmd .= "<0.$fetch_size>";
 136
 137     $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, TRUE);
 138     do {
 139         $topline = trim(array_shift($data));
 140     } while($topline && ($topline[0] == '*') && !preg_match('/\* [0-9]+ FETCH .*BODY.*/i', $topline)) ;
 141     // Matching with "BODY" above is difficult: in most cases "FETCH \(BODY" would work
 142     // but some servers may put other things in the same result, perhaps something such
 143     // as "* 23 FETCH (FLAGS (\Seen) BODY[1] {174}".  There is some small chance that
 144     // if the character sequence "BODY" appears in a response where it isn't actually
 145     // a FETCH response data item name, the current regex will break things.  The better
 146     // way to do this would be to parse the response correctly and not use a regex.
 147
 148     $wholemessage = implode('', $data);
 149     if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) {
 150         $ret = substr($wholemessage, 0, $regs[1]);
 151         /* There is some information in the content info header that could be important
 152          * in order to parse html messages. Let's get them here.
 153          */
 154 //        if ($ret{0} == '<') {
 155 //            $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE);
 156 //        }
 157     } else if (preg_match('/"([^"]*)"/', $topline, $regs)) {
 158         $ret = $regs[1];
 159     } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) {
 160         $ret = $wholemessage;
 161     } else {
 162         global $where, $what, $mailbox, $passed_id, $startMessage;
 163         $par = 'mailbox=' . urlencode($mailbox) . '&amp;passed_id=' . $passed_id;
 164         if (isset($where) && isset($what)) {
 165             $par .= '&amp;where=' . urlencode($where) . '&amp;what=' . urlencode($what);
 166         } else {
 167             $par .= '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
 168         }
 169         $par .= '&amp;response=' . urlencode($response) .
 170             '&amp;message='  . urlencode($message)  .
 171             '&amp;topline='  . urlencode($topline);
 172
 173         echo '<tt><br />' .
 174             '<table width="80%"><tr>' .
 175             '<tr><td colspan="2">' .
 176             _("Body retrieval error. The reason for this is most probably that the message is malformed.") .
 177             '</td></tr>' .
 178             '<tr><td><b>' . _("Command:") . "</td><td>$cmd</td></tr>" .
 179             '<tr><td><b>' . _("Response:") . "</td><td>$response</td></tr>" .
 180             '<tr><td><b>' . _("Message:") . "</td><td>$message</td></tr>" .
 181             '<tr><td><b>' . _("FETCH line:") . "</td><td>$topline</td></tr>" .
 182             "</table><br /></tt></font><hr />";
 183
 184         $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE);
 185         array_shift($data);
 186         $wholemessage = implode('', $data);
 187
 188         $ret = $wholemessage;
 189     }
 190     return $ret;
 191 }
 192
 193 function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') {
 194
 195     /* Don't kill the connection if the browser is over a dialup
 196      * and it would take over 30 seconds to download it.
 197      * Don't call set_time_limit in safe mode.
 198      */
 199
 200     if (!ini_get('safe_mode')) {
 201         set_time_limit(0);
 202     }
 203     /* in case of base64 encoded attachments, do not buffer them.
 204        Instead, echo the decoded attachment directly to screen */
 205     if (strtolower($encoding) == 'base64') {
 206         if (!$ent_id) {
 207             $query = "FETCH $id BODY[]";
 208         } else {
 209             $query = "FETCH $id BODY[$ent_id]";
 210         }
 211         sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
 212     } else {
 213         $body = mime_fetch_body ($imap_stream, $id, $ent_id);
 214         if (is_resource($rStream)) {
 215             fputs($rStream,decodeBody($body, $encoding, $force_crlf));
 216         } else {
 217             echo decodeBody($body, $encoding, $force_crlf);
 218         }
 219     }
 220
 221     /*
 222        TODO, use the same method for quoted printable.
 223        However, I assume that quoted printable attachments aren't that large
 224        so the performancegain / memory usage drop will be minimal.
 225        If we decide to add that then we need to adapt sqimap_fread because
 226        we need to split te result on \n and fread doesn't stop at \n. That
 227        means we also should provide $results from sqimap_fread (by ref) to
 228        te function and set $no_return to false. The $filter function for
 229        quoted printable should handle unsetting of $results.
 230      */
 231     /*
 232        TODO 2: find out how we write to the output stream php://stdout. fwrite
 233        doesn't work because 'php://stdout isn't a stream.
 234      */
 235
 236     return;
 237 }
 238
 239 /* -[ END MIME DECODING ]----------------------------------------------------------- */
 240
 241 /* This is here for debugging purposes.  It will print out a list
 242  * of all the entity IDs that are in the $message object.
 243  */
 244 function listEntities ($message) {
 245     if ($message) {
 246         echo "<tt>" . $message->entity_id . ' : ' . $message->type0 . '/' . $message->type1 . ' parent = '. $message->parent->entity_id. '<br />';
 247         for ($i = 0; isset($message->entities[$i]); $i++) {
 248             echo "$i : ";
 249             $msg = listEntities($message->entities[$i]);
 250
 251             if ($msg) {
 252                 echo "return: ";
 253                 return $msg;
 254             }
 255         }
 256     }
 257 }
 258
 259 function getPriorityStr($priority) {
 260     $priority_level = substr($priority,0,1);
 261
 262     switch($priority_level) {
 263         /* Check for a higher then normal priority. */
 264         case '1':
 265         case '2':
 266             $priority_string = _("High");
 267             break;
 268
 269         /* Check for a lower then normal priority. */
 270         case '4':
 271         case '5':
 272             $priority_string = _("Low");
 273             break;
 274
 275         /* Check for a normal priority. */
 276         case '3':
 277         default:
 278             $priority_level = '3';
 279             $priority_string = _("Normal");
 280             break;
 281
 282     }
 283     return $priority_string;
 284 }
 285
 286 /* returns a $message object for a particular entity id */
 287 function getEntity ($message, $ent_id) {
 288     return $message->getEntity($ent_id);
 289 }
 290
 291 /* translateText
 292  * Extracted from strings.php 23/03/2002
 293  */
 294
 295 function translateText(&$body, $wrap_at, $charset) {
 296     global $where, $what;   /* from searching */
 297     global $color;          /* color theme */
 298
 299     // require_once(SM_PATH . 'functions/url_parser.php');
 300
 301     $body_ary = explode("\n", $body);
 302     for ($i=0; $i < count($body_ary); $i++) {
 303         $line = rtrim($body_ary[$i],"\r");
 304
 305         if (strlen($line) - 2 >= $wrap_at) {
 306             sqWordWrap($line, $wrap_at, $charset);
 307         }
 308         $line = charset_decode($charset, $line);
 309         $line = str_replace("\t", '        ', $line);
 310
 311         parseUrl ($line);
 312
 313         $quotes = 0;
 314         $pos = 0;
 315         $j = strlen($line);
 316
 317         while ($pos < $j) {
 318             if ($line[$pos] == ' ') {
 319                 $pos++;
 320             } else if (strpos($line, '&gt;', $pos) === $pos) {
 321                 $pos += 4;
 322                 $quotes++;
 323             } else {
 324                 break;
 325             }
 326         }
 327
 328         if ($quotes % 2) {
 329             $line = '<span class="quote1">' . $line . '</span>';
 330         } elseif ($quotes) {
 331             $line = '<span class="quote2">' . $line . '</span>';
 332         }
 333
 334         $body_ary[$i] = $line;
 335     }
 336     $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
 337 }
 338
 339 /**
 340  * This returns a parsed string called $body. That string can then
 341  * be displayed as the actual message in the HTML. It contains
 342  * everything needed, including HTML Tags, Attachments at the
 343  * bottom, etc.
 344  *
 345  * Since 1.2.0 function uses message_body hook.
 346  * Till 1.3.0 function included output of formatAttachments().
 347  *
 348  * @param resource $imap_stream imap connection resource
 349  * @param object $message squirrelmail message object
 350  * @param array $color squirrelmail color theme array
 351  * @param integer $wrap_at number of characters per line
 352  * @param string $ent_num (since 1.3.0) message part id
 353  * @param integer $id (since 1.3.0) message id
 354  * @param string $mailbox (since 1.3.0) imap folder name
 355  * @return string html formated message text
 356  */
 357 function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') {
 358     /* This if statement checks for the entity to show as the
 359      * primary message. To add more of them, just put them in the
 360      * order that is their priority.
 361      */
 362     global $startMessage, $languages, $squirrelmail_language,
 363            $show_html_default, $sort, $has_unsafe_images, $passed_ent_id,
 364            $use_iframe, $iframe_height, $download_and_unsafe_link,
 365            $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text,
 366            $oTemplate, $nbsp;
 367
 368     // workaround for not updated config.php
 369     if (! isset($use_iframe)) $use_iframe = false;
 370
 371     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
 372     // images off by default.
 373     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
 374
 375     $body = '';
 376     $urlmailbox = urlencode($mailbox);
 377     $body_message = getEntity($message, $ent_num);
 378     if (($body_message->header->type0 == 'text') ||
 379             ($body_message->header->type0 == 'rfc822')) {
 380         $body = mime_fetch_body ($imap_stream, $id, $ent_num);
 381         $body = decodeBody($body, $body_message->header->encoding);
 382
 383         if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 384                 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) {
 385             if (mb_detect_encoding($body) != 'ASCII') {
 386                 $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body);
 387             }
 388         }
 389
 390         /* As of 1.5.2, $body is passed (and modified) by reference */
 391         do_hook('message_body', $body);
 392
 393         /* If there are other types that shouldn't be formatted, add
 394          * them here.
 395          */
 396
 397         if ($body_message->header->type1 == 'html') {
 398             if ($show_html_default <> 1) {
 399                 $entity_conv = array('&nbsp;' => ' ',
 400                                      '<p>'    => "\n",
 401                                      '<P>'    => "\n",
 402                                      '<br>'   => "\n",
 403                                      '<BR>'   => "\n",
 404                                      '<br />' => "\n",
 405                                      '<BR />' => "\n",
 406                                      '&gt;'   => '>',
 407                                      '&lt;'   => '<');
 408                 $body = strtr($body, $entity_conv);
 409                 $body = strip_tags($body);
 410                 $body = trim($body);
 411                 translateText($body, $wrap_at,
 412                         $body_message->header->getParameter('charset'));
 413             } elseif ($use_iframe) {
 414                 /**
 415                  * If we don't add html message between iframe tags,
 416                  * we must detect unsafe images and modify $has_unsafe_images.
 417                  */
 418                 $html_body = magicHTML($body, $id, $message, $mailbox);
 419                 // Convert character set in order to display html mails in different character set
 420                 $html_body = charset_decode($body_message->header->getParameter('charset'),$html_body,false,true);
 421
 422                 // creating iframe url
 423                 $iframeurl=sqm_baseuri().'src/view_html.php?'
 424                     . 'mailbox=' . $urlmailbox
 425                     . '&amp;passed_id=' . $id
 426                     . '&amp;ent_id=' . $ent_num
 427                     . '&amp;view_unsafe_images=' . (int) $view_unsafe_images;
 428
 429                 global $oTemplate;
 430                 $oTemplate->assign('iframe_url', $iframeurl);
 431                 $oTemplate->assign('iframe_height', $iframe_height);
 432                 $oTemplate->assign('html_body', $html_body);
 433
 434                 $body = $oTemplate->fetch('read_html_iframe.tpl');
 435             } else {
 436                 // old way of html rendering
 437                 /**
 438                  * convert character set. charset_decode does not remove html special chars
 439                  * applied by magicHTML functions and does not sanitize them second time if
 440                  * fourth argument is true.
 441                  */
 442                 $charset = $body_message->header->getParameter('charset');
 443                 if (!empty($charset)) {
 444                     $body = charset_decode($charset,$body,false,true);
 445                 }
 446                 $body = magicHTML($body, $id, $message, $mailbox);
 447             }
 448         } else {
 449             translateText($body, $wrap_at,
 450                     $body_message->header->getParameter('charset'));
 451         }
 452
 453         /*
 454          * Previously the links for downloading and unsafe images were printed
 455          * under the mail. By putting the links in a global variable we can
 456          * print it in the toolbar where it belongs. Since the original code was
 457          * in this place it's left here. It might be possible to move it to some
 458          * other place if that makes sense. The possibility to do so has not
 459          * been evaluated yet.
 460          */
 461
 462         // Initialize the global variable to an empty string.
 463         // FIXME: To have $download_and_unsafe_link as a global variable might not be needed since the use of separate variables ($download_href, $unsafe_image_toggle_href, and $unsafe_image_toggle_text) for the templates was introduced.
 464         $download_and_unsafe_link = '';
 465
 466         // Prepare and build a link for downloading the mail.
 467         $link = 'passed_id=' . $id . '&amp;ent_id='.$ent_num.
 468             '&amp;mailbox=' . $urlmailbox .'&amp;sort=' . $sort .
 469             '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
 470         if (isset($passed_ent_id)) {
 471             $link .= '&amp;passed_ent_id='.$passed_ent_id;
 472         }
 473         $download_href = SM_PATH . 'src/download.php?absolute_dl=true&amp;' . $link;
 474
 475         // Always add the link for downloading the mail as a file to the global
 476         // variable.
 477         $download_and_unsafe_link .= "$nbsp|$nbsp"
 478             . create_hyperlink($download_href, _("Download this as a file"));
 479
 480         // Find out the right text to use in the link depending on the
 481         // circumstances. If the unsafe images are displayed the link should
 482         // hide them, if they aren't displayed the link should only appear if
 483         // the mail really contains unsafe images.
 484         if ($view_unsafe_images) {
 485             $text = _("Hide Unsafe Images");
 486         } else {
 487             if (isset($has_unsafe_images) && $has_unsafe_images) {
 488                 $link .= '&amp;view_unsafe_images=1';
 489                 $text = _("View Unsafe Images");
 490             } else {
 491                 $text = '';
 492             }
 493         }
 494
 495         // Only create a link for unsafe images if there's need for one. If so:
 496         // add it to the global variable.
 497         if($text != '') {
 498             $unsafe_image_toggle_href = SM_PATH . 'src/read_body.php?'.$link;
 499             $unsafe_image_toggle_text = $text;
 500             $download_and_unsafe_link .= "$nbsp|$nbsp"
 501                 . create_hyperlink($unsafe_image_toggle_href, $text);
 502         }
 503     }
 504     return $body;
 505 }
 506
 507 /**
 508  * Generate attachments array for passing to templates.
 509  *
 510  * @since 1.5.2
 511  * @param object $message SquirrelMail message object
 512  * @param array $exclude_id message parts that are not attachments.
 513  * @param string $mailbox mailbox name
 514  * @param integer $id message id
 515  */
 516 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
 517     global $where, $what, $startMessage, $color, $passed_ent_id, $base_uri;
 518
 519     $att_ar = $message->getAttachments($exclude_id);
 520     $urlMailbox = urlencode($mailbox);
 521
 522     $attachments = array();
 523     foreach ($att_ar as $att) {
 524         $ent = $att->entity_id;
 525         $header = $att->header;
 526         $type0 = strtolower($header->type0);
 527         $type1 = strtolower($header->type1);
 528         $name = '';
 529         $links = array();
 530         $links['download link']['text'] = _("Download");
 531         $links['download link']['href'] = $base_uri .
 532             "src/download.php?absolute_dl=true&amp;passed_id=$id&amp;mailbox=$urlMailbox&amp;ent_id=$ent";
 533
 534         if ($type0 =='message' && $type1 == 'rfc822') {
 535             $default_page = $base_uri  . 'src/read_body.php';
 536             $rfc822_header = $att->rfc822_header;
 537             $filename = $rfc822_header->subject;
 538             if (trim( $filename ) == '') {
 539                 $filename = 'untitled-[' . $ent . ']' ;
 540             }
 541             $from_o = $rfc822_header->from;
 542             if (is_object($from_o)) {
 543                 $from_name = decodeHeader($from_o->getAddress(false));
 544             } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) {
 545                 // something weird happens when a digest message is opened and you return to the digest
 546                 // now the from object is part of an array. Probably the parseHeader call overwrites the info
 547                 // retrieved from the bodystructure in a different way. We need to fix this later.
 548                 // possible starting point, do not fetch header we already have and inspect how
 549                 // the rfc822_header object behaves.
 550                 $from_name = decodeHeader($from_o[0]->getAddress(false));
 551             } else {
 552                 $from_name = _("Unknown sender");
 553             }
 554             $description = _("From").': '.$from_name;
 555         } else {
 556             $default_page = $base_uri  . 'src/download.php';
 557             $filename = $att->getFilename();
 558             if ($header->description) {
 559                 $description = decodeHeader($header->description);
 560             } else {
 561                 $description = '';
 562             }
 563         }
 564
 565         $display_filename = $filename;
 566         if (isset($passed_ent_id)) {
 567             $passed_ent_id_link = '&amp;passed_ent_id='.$passed_ent_id;
 568         } else {
 569             $passed_ent_id_link = '';
 570         }
 571         $defaultlink = $default_page . "?startMessage=$startMessage"
 572             . "&amp;passed_id=$id&amp;mailbox=$urlMailbox"
 573             . '&amp;ent_id='.$ent.$passed_ent_id_link;
 574         if ($where && $what) {
 575             $defaultlink .= '&amp;where='. urlencode($where).'&amp;what='.urlencode($what);
 576         }
 577         // IE does make use of mime content sniffing. Forcing a download
 578         // prohibit execution of XSS inside an application/octet-stream attachment
 579         if ($type0 == 'application' && $type1 == 'octet-stream') {
 580             $defaultlink .= '&amp;absolute_dl=true';
 581         }
 582
 583         /* This executes the attachment hook with a specific MIME-type.
 584          * It also allows plugins to run if there's a rule for a more
 585          * generic type. Finally, a hook for ALL attachment types is
 586          * run as well.
 587          */
 588         // First remember the default link.
 589         $defaultlink_orig = $defaultlink;
 590
 591         /* The API for this hook has changed as of 1.5.2 so that all plugin
 592            arguments are passed in an array instead of each their own plugin
 593            argument, and arguments are passed by reference, so instead of
 594            returning any changes, changes should simply be made to the original
 595            arguments themselves. */
 596         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 597                     &$defaultlink, &$display_filename, &$where, &$what);
 598         do_hook("attachment $type0/$type1", $temp);
 599         /* The API for this hook has changed as of 1.5.2 so that all plugin
 600            arguments are passed in an array instead of each their own plugin
 601            argument, and arguments are passed by reference, so instead of
 602            returning any changes, changes should simply be made to the original
 603            arguments themselves. */
 604         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 605                       &$defaultlink, &$display_filename, &$where, &$what);
 606         // Do not let a generic plugin change the default link if a more
 607         // specialized one already did it...
 608         if ($defaultlink != $defaultlink_orig) {
 609             $dummy = '';
 610             $temp[5] = &$dummy;
 611         }
 612         do_hook("attachment $type0/*", $temp);
 613         /* The API for this hook has changed as of 1.5.2 so that all plugin
 614            arguments are passed in an array instead of each their own plugin
 615            argument, and arguments are passed by reference, so instead of
 616            returning any changes, changes should simply be made to the original
 617            arguments themselves. */
 618         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 619                       &$defaultlink, &$display_filename, &$where, &$what);
 620         // Do not let a generic plugin change the default link if a more
 621         // specialized one already did it...
 622         if ($defaultlink != $defaultlink_orig) {
 623             $dummy = '';
 624             $temp[5] = &$dummy;
 625         }
 626         do_hook("attachment */*", $temp);
 627
 628         $this_attachment = array();
 629         $this_attachment['Name'] = decodeHeader($display_filename);
 630         $this_attachment['Description'] = $description;
 631         $this_attachment['DefaultHREF'] = $defaultlink;
 632         $this_attachment['DownloadHREF'] = $links['download link']['href'];
 633         $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
 634         $this_attachment['Size'] = $header->size;
 635         $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
 636         $this_attachment['OtherLinks'] = array();
 637         foreach ($links as $val) {
 638             if ($val['text']==_("Download") || $val['text'] == _("View"))
 639                 continue;
 640             if (empty($val['text']) && empty($val['extra']))
 641                 continue;
 642
 643             $temp = array();
 644             $temp['HREF'] = $val['href'];
 645             $temp['Text'] = (empty($val['text']) ? '' : $val['text']) . (empty($val['extra']) ? '' : $val['extra']);
 646             $this_attachment['OtherLinks'][] = $temp;
 647         }
 648         $attachments[] = $this_attachment;
 649
 650         unset($links);
 651     }
 652
 653     return $attachments;
 654 }
 655
 656 /**
 657  * Displays attachment links and information
 658  *
 659  * Since 1.3.0 function is not included in formatBody() call.
 660  *
 661  * Since 1.0.2 uses attachment $type0/$type1 hook.
 662  * Since 1.2.5 uses attachment $type0/* hook.
 663  * Since 1.5.0 uses attachments_bottom hook.
 664  * Since 1.5.2 uses templates and does *not* return a value.
 665  *
 666  * @param object $message SquirrelMail message object
 667  * @param array $exclude_id message parts that are not attachments.
 668  * @param string $mailbox mailbox name
 669  * @param integer $id message id
 670  */
 671 function formatAttachments($message, $exclude_id, $mailbox, $id) {
 672     global $oTemplate;
 673
 674     $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
 675
 676     $oTemplate->assign('attachments', $attach);
 677     $oTemplate->display('read_attachments.tpl');
 678 }
 679
 680 function sqimap_base64_decode(&$string) {
 681
 682     // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
 683     // fly decoding (to reduce memory usage) you have to check if the
 684     // data has incomplete pairs
 685
 686     // Remove the noise in order to check if the 4 bytes pairs are complete
 687     $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
 688
 689     $sStringRem = '';
 690     $iMod = strlen($string) % 4;
 691     if ($iMod) {
 692         $sStringRem = substr($string,-$iMod);
 693         // Check if $sStringRem contains padding characters
 694         if (substr($sStringRem,-1) != '=') {
 695             $string = substr($string,0,-$iMod);
 696         } else {
 697             $sStringRem = '';
 698         }
 699     }
 700     $string = base64_decode($string);
 701     return $sStringRem;
 702 }
 703
 704 /**
 705  * Decodes encoded string (usually message body)
 706  *
 707  * This function decodes a string (usually the message body)
 708  * depending on the encoding type.  Currently quoted-printable
 709  * and base64 encodings are supported.
 710  *
 711  * The decode_body hook was added to this function in 1.4.2/1.5.0.
 712  * The $force_crlf parameter was added in 1.5.2.
 713  *
 714  * @param string $string     The encoded string
 715  * @param string $encoding   used encoding
 716  * @param string $force_crlf Whether or not to force CRLF or LF
 717  *                           line endings (or to leave as is).
 718  *                           If given as "LF", line endings will
 719  *                           all be converted to LF; if "CRLF",
 720  *                           line endings will all be converted
 721  *                           to CRLF.  If given as an empty value,
 722  *                           the global $force_crlf_default will
 723  *                           be consulted (it can be specified in
 724  *                           config/config_local.php).  Otherwise,
 725  *                           any other value will cause the string
 726  *                           to be left alone.  Note that this will
 727  *                           be overridden to "LF" if not using at
 728  *                           least PHP version 4.3.0. (OPTIONAL;
 729  *                           default is empty - consult global
 730  *                           default value)
 731  *
 732  * @return string The decoded string
 733  *
 734  * @since 1.0
 735  *
 736  */
 737 function decodeBody($string, $encoding, $force_crlf='') {
 738
 739     global $force_crlf_default;
 740     if (empty($force_crlf)) $force_crlf = $force_crlf_default;
 741     $force_crlf = strtoupper($force_crlf);
 742
 743     // must force line endings to LF due to broken
 744     // quoted_printable_decode() in PHP versions
 745     // before 4.3.0 (see below)
 746     //
 747     if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
 748         $string = str_replace("\r\n", "\n", $string);
 749     else if ($force_crlf == 'CRLF')
 750         $string = str_replace("\n", "\r\n", $string);
 751
 752     $encoding = strtolower($encoding);
 753
 754     $encoding_handler = do_hook('decode_body', $encoding);
 755
 756
 757     // plugins get first shot at decoding the string
 758     //
 759     if (!empty($encoding_handler) && function_exists($encoding_handler)) {
 760         $string = $encoding_handler('decode', $string);
 761
 762     } elseif ($encoding == 'quoted-printable' ||
 763             $encoding == 'quoted_printable') {
 764
 765         // quoted_printable_decode() function is broken in older
 766         // php versions.  Text with \r\n decoding was fixed only
 767         // in php 4.3.0.  Minimal code requirement is PHP 4.0.4+
 768         // and the above call to:  str_replace("\r\n", "\n", $string);
 769         //
 770         $string = quoted_printable_decode($string);
 771
 772     } elseif ($encoding == 'base64') {
 773         $string = base64_decode($string);
 774     }
 775
 776     // All other encodings are returned raw.
 777     return $string;
 778 }
 779
 780 /**
 781  * Decodes headers
 782  *
 783  * This function decodes strings that are encoded according to
 784  * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
 785  * Patched by Christian Schmidt <christian@ostenfeld.dk>  23/03/2002
 786  *
 787  * @param string $string header string that has to be made readable
 788  * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
 789  * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
 790  * @param boolean $decide decide if string can be utfencoded. defaults to false
 791  * @return string decoded header string
 792  */
 793 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
 794     global $languages, $squirrelmail_language,$default_charset;
 795     if (is_array($string)) {
 796         $string = implode("\n", $string);
 797     }
 798
 799     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 800             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
 801         $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
 802         // Do we need to return at this point?
 803         // return $string;
 804     }
 805     $i = 0;
 806     $iLastMatch = -2;
 807     $encoded = true;
 808
 809     $aString = explode(' ',$string);
 810     $ret = '';
 811     foreach ($aString as $chunk) {
 812         if ($encoded && $chunk === '') {
 813             continue;
 814         } elseif ($chunk === '') {
 815             $ret .= ' ';
 816             continue;
 817         }
 818         $encoded = false;
 819         /* if encoded words are not separated by a linear-space-white we still catch them */
 820         $j = $i-1;
 821
 822         while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
 823             /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
 824             if ($iLastMatch !== $j) {
 825                 if ($htmlsafe) {
 826                     $ret .= '&#32;';
 827                 } else {
 828                     $ret .= ' ';
 829                 }
 830             }
 831             $iLastMatch = $i;
 832             $j = $i;
 833             if ($htmlsafe) {
 834                 $ret .= sm_encode_html_special_chars($res[1]);
 835             } else {
 836                 $ret .= $res[1];
 837             }
 838             $encoding = ucfirst($res[3]);
 839
 840             /* decide about valid decoding */
 841             if ($decide && is_conversion_safe($res[2])) {
 842                 $utfencode=true;
 843                 $can_be_encoded=true;
 844             } else {
 845                 $can_be_encoded=false;
 846             }
 847             switch ($encoding)
 848             {
 849                 case 'B':
 850                     $replace = base64_decode($res[4]);
 851                     if ($utfencode) {
 852                         if ($can_be_encoded) {
 853                             /* convert string to different charset,
 854                              * if functions asks for it (usually in compose)
 855                              */
 856                             $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
 857                         } else {
 858                             // convert string to html codes in order to display it
 859                             $ret .= charset_decode($res[2],$replace);
 860                         }
 861                     } else {
 862                         if ($htmlsafe) {
 863                             $replace = sm_encode_html_special_chars($replace);
 864                         }
 865                         $ret.= $replace;
 866                     }
 867                     break;
 868                 case 'Q':
 869                     $replace = str_replace('_', ' ', $res[4]);
 870                     $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
 871                             create_function ('$matches', 'return chr(hexdec($matches[1]));'),
 872                             $replace);
 873                     if ($utfencode) {
 874                         if ($can_be_encoded) {
 875                             /* convert string to different charset,
 876                              * if functions asks for it (usually in compose)
 877                              */
 878                             $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
 879                         } else {
 880                             // convert string to html codes in order to display it
 881                             $replace = charset_decode($res[2], $replace);
 882                         }
 883                     } else {
 884                         if ($htmlsafe) {
 885                             $replace = sm_encode_html_special_chars($replace);
 886                         }
 887                     }
 888                     $ret .= $replace;
 889                     break;
 890                 default:
 891                     break;
 892             }
 893             $chunk = $res[5];
 894             $encoded = true;
 895         }
 896         if (!$encoded) {
 897             if ($htmlsafe) {
 898                 $ret .= '&#32;';
 899             } else {
 900                 $ret .= ' ';
 901             }
 902         }
 903
 904         if (!$encoded && $htmlsafe) {
 905             $ret .= sm_encode_html_special_chars($chunk);
 906         } else {
 907             $ret .= $chunk;
 908         }
 909         ++$i;
 910     }
 911     /* remove the first added space */
 912     if ($ret) {
 913         if ($htmlsafe) {
 914             $ret = substr($ret,5);
 915         } else {
 916             $ret = substr($ret,1);
 917         }
 918     }
 919
 920     return $ret;
 921 }
 922
 923 /**
 924  * Encodes header
 925  *
 926  * Function uses XTRA_CODE _encodeheader function, if such function exists.
 927  *
 928  * Function uses Q encoding by default and encodes a string according to RFC
 929  * 1522 for use in headers if it contains 8-bit characters or anything that
 930  * looks like it should be encoded.
 931  *
 932  * Function switches to B encoding and encodeHeaderBase64() function, if
 933  * string is 8bit and multibyte character set supported by mbstring extension
 934  * is used. It can cause E_USER_NOTICE errors, if interface is used with
 935  * multibyte character set unsupported by mbstring extension.
 936  *
 937  * @param string $string header string, that has to be encoded
 938  * @return string quoted-printable encoded string
 939  * @todo make $mb_charsets system wide constant
 940  */
 941 function encodeHeader ($string) {
 942     global $default_charset, $languages, $squirrelmail_language;
 943
 944     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 945             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
 946         return  call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
 947     }
 948
 949     // Use B encoding for multibyte charsets
 950     $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
 951     if (in_array($default_charset,$mb_charsets) &&
 952         in_array($default_charset,sq_mb_list_encodings()) &&
 953         sq_is8bit($string)) {
 954         return encodeHeaderBase64($string,$default_charset);
 955     } elseif (in_array($default_charset,$mb_charsets) &&
 956               sq_is8bit($string) &&
 957               ! in_array($default_charset,sq_mb_list_encodings())) {
 958         // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
 959         // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
 960     }
 961
 962     // Encode only if the string contains 8-bit characters or =?
 963     $j = strlen($string);
 964     $max_l = 75 - strlen($default_charset) - 7;
 965     $aRet = array();
 966     $ret = '';
 967     $iEncStart = $enc_init = false;
 968     $cur_l = $iOffset = 0;
 969     for($i = 0; $i < $j; ++$i) {
 970         switch($string{$i})
 971         {
 972             case '"':
 973             case '=':
 974             case '<':
 975             case '>':
 976             case ',':
 977             case '?':
 978             case '_':
 979                 if ($iEncStart === false) {
 980                     $iEncStart = $i;
 981                 }
 982                 $cur_l+=3;
 983                 if ($cur_l > ($max_l-2)) {
 984                     /* if there is an stringpart that doesn't need encoding, add it */
 985                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
 986                     $aRet[] = "=?$default_charset?Q?$ret?=";
 987                     $iOffset = $i;
 988                     $cur_l = 0;
 989                     $ret = '';
 990                     $iEncStart = false;
 991                 } else {
 992                     $ret .= sprintf("=%02X",ord($string{$i}));
 993                 }
 994                 break;
 995             case '(':
 996             case ')':
 997                 if ($iEncStart !== false) {
 998                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
 999                     $aRet[] = "=?$default_charset?Q?$ret?=";
1000                     $iOffset = $i;
1001                     $cur_l = 0;
1002                     $ret = '';
1003                     $iEncStart = false;
1004                 }
1005                 break;
1006             case ' ':
1007                 if ($iEncStart !== false) {
1008                     $cur_l++;
1009                     if ($cur_l > $max_l) {
1010                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1011                         $aRet[] = "=?$default_charset?Q?$ret?=";
1012                         $iOffset = $i;
1013                         $cur_l = 0;
1014                         $ret = '';
1015                         $iEncStart = false;
1016                     } else {
1017                         $ret .= '_';
1018                     }
1019                 }
1020                 break;
1021             default:
1022                 $k = ord($string{$i});
1023                 if ($k > 126) {
1024                     if ($iEncStart === false) {
1025                         // do not start encoding in the middle of a string, also take the rest of the word.
1026                         $sLeadString = substr($string,0,$i);
1027                         $aLeadString = explode(' ',$sLeadString);
1028                         $sToBeEncoded = array_pop($aLeadString);
1029                         $iEncStart = $i - strlen($sToBeEncoded);
1030                         $ret .= $sToBeEncoded;
1031                         $cur_l += strlen($sToBeEncoded);
1032                     }
1033                     $cur_l += 3;
1034                     /* first we add the encoded string that reached it's max size */
1035                     if ($cur_l > ($max_l-2)) {
1036                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1037                         $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1038                         $cur_l = 3;
1039                         $ret = '';
1040                         $iOffset = $i;
1041                         $iEncStart = $i;
1042                     }
1043                     $enc_init = true;
1044                     $ret .= sprintf("=%02X", $k);
1045                 } else {
1046                     if ($iEncStart !== false) {
1047                         $cur_l++;
1048                         if ($cur_l > $max_l) {
1049                             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1050                             $aRet[] = "=?$default_charset?Q?$ret?=";
1051                             $iEncStart = false;
1052                             $iOffset = $i;
1053                             $cur_l = 0;
1054                             $ret = '';
1055                         } else {
1056                             $ret .= $string{$i};
1057                         }
1058                     }
1059                 }
1060                 break;
1061         }
1062     }
1063
1064     if ($enc_init) {
1065         if ($iEncStart !== false) {
1066             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1067             $aRet[] = "=?$default_charset?Q?$ret?=";
1068         } else {
1069             $aRet[] = substr($string,$iOffset);
1070         }
1071         $string = implode('',$aRet);
1072     }
1073     return $string;
1074 }
1075
1076 /**
1077  * Encodes string according to rfc2047 B encoding header formating rules
1078  *
1079  * It is recommended way to encode headers with character sets that store
1080  * symbols in more than one byte.
1081  *
1082  * Function requires mbstring support. If required mbstring functions are missing,
1083  * function returns false and sets E_USER_WARNING level error message.
1084  *
1085  * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1086  * that mbstring functions will generate E_WARNING errors, if unsupported
1087  * character set is used. mb_encode_mimeheader function provided by php
1088  * mbstring extension is not used in order to get better control of header
1089  * encoding.
1090  *
1091  * Used php code functions - function_exists(), trigger_error(), strlen()
1092  * (is used with charset names and base64 strings). Used php mbstring
1093  * functions - mb_strlen and mb_substr.
1094  *
1095  * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1096  * encoding), rfc 2822 (header folding)
1097  *
1098  * @param string $string header string that must be encoded
1099  * @param string $charset character set. Must be supported by mbstring extension.
1100  * Use sq_mb_list_encodings() to detect supported charsets.
1101  * @return string string encoded according to rfc2047 B encoding formating rules
1102  * @since 1.5.1
1103  * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1104  * @todo Do we want to control max length of header?
1105  * @todo Do we want to control EOL (end-of-line) marker?
1106  * @todo Do we want to translate error message?
1107  */
1108 function encodeHeaderBase64($string,$charset) {
1109     /**
1110      * Check mbstring function requirements.
1111      */
1112     if (! function_exists('mb_strlen') ||
1113         ! function_exists('mb_substr')) {
1114         // set E_USER_WARNING
1115         trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1116         // return false
1117         return false;
1118     }
1119
1120     // initial return array
1121     $aRet = array();
1122
1123     /**
1124      * header length = 75 symbols max (same as in encodeHeader)
1125      * remove $charset length
1126      * remove =? ? ?= (5 chars)
1127      * remove 2 more chars (\r\n ?)
1128      */
1129     $iMaxLength = 75 - strlen($charset) - 7;
1130
1131     // set first character position
1132     $iStartCharNum = 0;
1133
1134     // loop through all characters. count characters and not bytes.
1135     for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1136         // encode string from starting character to current character.
1137         $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1138
1139         // Check encoded string length
1140         if(strlen($encoded_string)>$iMaxLength) {
1141             // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1142             $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1143
1144             // set new starting character
1145             $iStartCharNum = $iCharNum-1;
1146
1147             // encode last char (in case it is last character in string)
1148             $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1149         } // if string is shorter than max length - add next character
1150     }
1151
1152     // add last encoded string to array
1153     $aRet[] = $encoded_string;
1154
1155     // set initial return string
1156     $sRet = '';
1157
1158     // loop through encoded strings
1159     foreach($aRet as $string) {
1160         // TODO: Do we want to control EOL (end-of-line) marker
1161         if ($sRet!='') $sRet.= " ";
1162
1163         // add header tags and encoded string to return string
1164         $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1165     }
1166
1167     return $sRet;
1168 }
1169
1170 /* This function trys to locate the entity_id of a specific mime element */
1171 function find_ent_id($id, $message) {
1172     for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1173         if ($message->entities[$i]->header->type0 == 'multipart')  {
1174             $ret = find_ent_id($id, $message->entities[$i]);
1175         } else {
1176             if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1177 //                if (sq_check_save_extension($message->entities[$i])) {
1178                 return $message->entities[$i]->entity_id;
1179 //                }
1180             } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1181                 /**
1182                  * This is part of a fix for Outlook Express 6.x generating
1183                  * cid URLs without creating content-id headers
1184                  * @@JA - 20050207
1185                  */
1186                 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1187                     return $message->entities[$i]->entity_id;
1188                 }
1189             }
1190         }
1191     }
1192     return $ret;
1193 }
1194
1195 function sq_check_save_extension($message) {
1196     $filename = $message->getFilename();
1197     $ext = substr($filename, strrpos($filename,'.')+1);
1198     $save_extensions = array('jpg','jpeg','gif','png','bmp');
1199     return in_array($ext, $save_extensions);
1200 }
1201
1202
1203 /**
1204  ** HTMLFILTER ROUTINES
1205  */
1206
1207 /**
1208  * This function checks attribute values for entity-encoded values
1209  * and returns them translated into 8-bit strings so we can run
1210  * checks on them.
1211  *
1212  * @param  $attvalue A string to run entity check against.
1213  * @return           Nothing, modifies a reference value.
1214  */
1215 function sq_defang(&$attvalue){
1216     $me = 'sq_defang';
1217     /**
1218      * Skip this if there aren't ampersands or backslashes.
1219      */
1220     if (strpos($attvalue, '&') === false
1221         && strpos($attvalue, '\\') === false){
1222         return;
1223     }
1224     $m = false;
1225     // before deent, translate the dangerous unicode characters and ... to safe values
1226     // otherwise the regular expressions do not match.
1227
1228
1229
1230     do {
1231         $m = false;
1232         $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1233         $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1234         $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1235     } while ($m == true);
1236     $attvalue = stripslashes($attvalue);
1237 }
1238
1239 /**
1240  * Kill any tabs, newlines, or carriage returns. Our friends the
1241  * makers of the browser with 95% market value decided that it'd
1242  * be funny to make "java[tab]script" be just as good as "javascript".
1243  *
1244  * @param  attvalue  The attribute value before extraneous spaces removed.
1245  * @return attvalue  Nothing, modifies a reference value.
1246  */
1247 function sq_unspace(&$attvalue){
1248     $me = 'sq_unspace';
1249     if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1250         $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1251                                 Array('',   '',   '',   '',   ''), $attvalue);
1252     }
1253 }
1254
1255 /**
1256  * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1257  * IE as regular characters.
1258  *
1259  * @param  attvalue  The attribute value before dangerous characters are translated.
1260  * @return attvalue  Nothing, modifies a reference value.
1261  * @author Marc Groot Koerkamp.
1262  */
1263 function sq_fixIE_idiocy(&$attvalue) {
1264     // remove NUL
1265     $attvalue = str_replace("\0", "", $attvalue);
1266     // remove comments
1267     $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1268
1269     // IE has the evil habit of accepting every possible value for the attribute expression.
1270     // The table below contains characters which are parsed by IE if they are used in the "expression"
1271     // attribute value.
1272     $aDangerousCharsReplacementTable = array(
1273                         array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1274                               '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1275                               '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1276                               '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1277                               '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1278                               '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1279                               '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1280                               '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1281                               '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1282                               '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1283                               '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1284                               '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1285                               '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1286                               '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1287                               '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1288                               '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1289                               '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1290                               '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1291                               '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1292                               '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1293                               '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1294                               '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1295                               '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1296                               '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1297                               "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */   // in unicode this is some Chinese char range
1298                               "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1299                               "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1300                               "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1301                               "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1302                               "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1303                               "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1304                               "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1305                               "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1306                               "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1307                               "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1308                               "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1309                               "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1310                               "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1311                               "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1312                               "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1313                               "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1314                               "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1315                               "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1316                               "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1317                               "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1318                               "\xCA\x9F", /* L UNICODE IPA Extension */
1319                               "\xCA\x80", /* R UNICODE IPA Extension */
1320                               "\xC9\xB4"),  /* N UNICODE IPA Extension */
1321                        array('l', 'l', 'r','r','n','n',
1322                              'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1323                              'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1324                              'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1325     $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1326
1327     // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1328     // used for XSS.
1329     $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1330 }
1331
1332 /**
1333  * This function returns the final tag out of the tag name, an array
1334  * of attributes, and the type of the tag. This function is called by
1335  * sq_sanitize internally.
1336  *
1337  * @param  $tagname  the name of the tag.
1338  * @param  $attary   the array of attributes and their values
1339  * @param  $tagtype  The type of the tag (see in comments).
1340  * @return           a string with the final tag representation.
1341  */
1342 function sq_tagprint($tagname, $attary, $tagtype){
1343     $me = 'sq_tagprint';
1344
1345     if ($tagtype == 2){
1346         $fulltag = '</' . $tagname . '>';
1347     } else {
1348         $fulltag = '<' . $tagname;
1349         if (is_array($attary) && sizeof($attary)){
1350             $atts = Array();
1351             while (list($attname, $attvalue) = each($attary)){
1352                 array_push($atts, "$attname=$attvalue");
1353             }
1354             $fulltag .= ' ' . join(" ", $atts);
1355         }
1356         if ($tagtype == 3){
1357             $fulltag .= ' /';
1358         }
1359         $fulltag .= '>';
1360     }
1361     return $fulltag;
1362 }
1363
1364 /**
1365  * A small helper function to use with array_walk. Modifies a by-ref
1366  * value and makes it lowercase.
1367  *
1368  * @param  $val a value passed by-ref.
1369  * @return      void since it modifies a by-ref value.
1370  */
1371 function sq_casenormalize(&$val){
1372     $val = strtolower($val);
1373 }
1374
1375 /**
1376  * This function skips any whitespace from the current position within
1377  * a string and to the next non-whitespace value.
1378  *
1379  * @param  $body   the string
1380  * @param  $offset the offset within the string where we should start
1381  *                 looking for the next non-whitespace character.
1382  * @return         the location within the $body where the next
1383  *                 non-whitespace char is located.
1384  */
1385 function sq_skipspace($body, $offset){
1386     $me = 'sq_skipspace';
1387     preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1388     if (sizeof($matches{1})){
1389         $count = strlen($matches{1});
1390         $offset += $count;
1391     }
1392     return $offset;
1393 }
1394
1395 /**
1396  * This function looks for the next character within a string.  It's
1397  * really just a glorified "strpos", except it catches if failures
1398  * nicely.
1399  *
1400  * @param  $body   The string to look for needle in.
1401  * @param  $offset Start looking from this position.
1402  * @param  $needle The character/string to look for.
1403  * @return         location of the next occurance of the needle, or
1404  *                 strlen($body) if needle wasn't found.
1405  */
1406 function sq_findnxstr($body, $offset, $needle){
1407     $me  = 'sq_findnxstr';
1408     $pos = strpos($body, $needle, $offset);
1409     if ($pos === FALSE){
1410         $pos = strlen($body);
1411     }
1412     return $pos;
1413 }
1414
1415 /**
1416  * This function takes a PCRE-style regexp and tries to match it
1417  * within the string.
1418  *
1419  * @param  $body   The string to look for needle in.
1420  * @param  $offset Start looking from here.
1421  * @param  $reg    A PCRE-style regex to match.
1422  * @return         Returns a false if no matches found, or an array
1423  *                 with the following members:
1424  *                 - integer with the location of the match within $body
1425  *                 - string with whatever content between offset and the match
1426  *                 - string with whatever it is we matched
1427  */
1428 function sq_findnxreg($body, $offset, $reg){
1429     $me = 'sq_findnxreg';
1430     $matches = Array();
1431     $retarr = Array();
1432     preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1433     if (!isset($matches{0}) || !$matches{0}){
1434         $retarr = false;
1435     } else {
1436         $retarr{0} = $offset + strlen($matches{1});
1437         $retarr{1} = $matches{1};
1438         $retarr{2} = $matches{2};
1439     }
1440     return $retarr;
1441 }
1442
1443 /**
1444  * This function looks for the next tag.
1445  *
1446  * @param  $body   String where to look for the next tag.
1447  * @param  $offset Start looking from here.
1448  * @return         false if no more tags exist in the body, or
1449  *                 an array with the following members:
1450  *                 - string with the name of the tag
1451  *                 - array with attributes and their values
1452  *                 - integer with tag type (1, 2, or 3)
1453  *                 - integer where the tag starts (starting "<")
1454  *                 - integer where the tag ends (ending ">")
1455  *                 first three members will be false, if the tag is invalid.
1456  */
1457 function sq_getnxtag($body, $offset){
1458     $me = 'sq_getnxtag';
1459     if ($offset > strlen($body)){
1460         return false;
1461     }
1462     $lt = sq_findnxstr($body, $offset, "<");
1463     if ($lt == strlen($body)){
1464         return false;
1465     }
1466     /**
1467      * We are here:
1468      * blah blah <tag attribute="value">
1469      * \---------^
1470      */
1471     $pos = sq_skipspace($body, $lt+1);
1472     if ($pos >= strlen($body)){
1473         return Array(false, false, false, $lt, strlen($body));
1474     }
1475     /**
1476      * There are 3 kinds of tags:
1477      * 1. Opening tag, e.g.:
1478      *    <a href="blah">
1479      * 2. Closing tag, e.g.:
1480      *    </a>
1481      * 3. XHTML-style content-less tag, e.g.:
1482      *    <img src="blah" />
1483      */
1484     $tagtype = false;
1485     switch (substr($body, $pos, 1)){
1486         case '/':
1487             $tagtype = 2;
1488             $pos++;
1489             break;
1490         case '!':
1491             /**
1492              * A comment or an SGML declaration.
1493              */
1494             if (substr($body, $pos+1, 2) == "--"){
1495                 $gt = strpos($body, "-->", $pos);
1496                 if ($gt === false){
1497                     $gt = strlen($body);
1498                 } else {
1499                     $gt += 2;
1500                 }
1501                 return Array(false, false, false, $lt, $gt);
1502             } else {
1503                 $gt = sq_findnxstr($body, $pos, ">");
1504                 return Array(false, false, false, $lt, $gt);
1505             }
1506             break;
1507         default:
1508             /**
1509              * Assume tagtype 1 for now. If it's type 3, we'll switch values
1510              * later.
1511              */
1512             $tagtype = 1;
1513             break;
1514     }
1515
1516     $tag_start = $pos;
1517     $tagname = '';
1518     /**
1519      * Look for next [\W-_], which will indicate the end of the tag name.
1520      */
1521     $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1522     if ($regary == false){
1523         return Array(false, false, false, $lt, strlen($body));
1524     }
1525     list($pos, $tagname, $match) = $regary;
1526     $tagname = strtolower($tagname);
1527
1528     /**
1529      * $match can be either of these:
1530      * '>'  indicating the end of the tag entirely.
1531      * '\s' indicating the end of the tag name.
1532      * '/'  indicating that this is type-3 xhtml tag.
1533      *
1534      * Whatever else we find there indicates an invalid tag.
1535      */
1536     switch ($match){
1537         case '/':
1538             /**
1539              * This is an xhtml-style tag with a closing / at the
1540              * end, like so: <img src="blah" />. Check if it's followed
1541              * by the closing bracket. If not, then this tag is invalid
1542              */
1543             if (substr($body, $pos, 2) == "/>"){
1544                 $pos++;
1545                 $tagtype = 3;
1546             } else {
1547                 $gt = sq_findnxstr($body, $pos, ">");
1548                 $retary = Array(false, false, false, $lt, $gt);
1549                 return $retary;
1550             }
1551         case '>':
1552             return Array($tagname, false, $tagtype, $lt, $pos);
1553             break;
1554         default:
1555             /**
1556              * Check if it's whitespace
1557              */
1558             if (!preg_match('/\s/', $match)){
1559                 /**
1560                  * This is an invalid tag! Look for the next closing ">".
1561                  */
1562                 $gt = sq_findnxstr($body, $lt, ">");
1563                 return Array(false, false, false, $lt, $gt);
1564             }
1565             break;
1566     }
1567
1568     /**
1569      * At this point we're here:
1570      * <tagname  attribute='blah'>
1571      * \-------^
1572      *
1573      * At this point we loop in order to find all attributes.
1574      */
1575     $attname = '';
1576     $atttype = false;
1577     $attary = Array();
1578
1579     while ($pos <= strlen($body)){
1580         $pos = sq_skipspace($body, $pos);
1581         if ($pos == strlen($body)){
1582             /**
1583              * Non-closed tag.
1584              */
1585             return Array(false, false, false, $lt, $pos);
1586         }
1587         /**
1588          * See if we arrived at a ">" or "/>", which means that we reached
1589          * the end of the tag.
1590          */
1591         $matches = Array();
1592         if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1593             /**
1594              * Yep. So we did.
1595              */
1596             $pos += strlen($matches{1});
1597             if ($matches{2} == "/>"){
1598                 $tagtype = 3;
1599                 $pos++;
1600             }
1601             return Array($tagname, $attary, $tagtype, $lt, $pos);
1602         }
1603
1604         /**
1605          * There are several types of attributes, with optional
1606          * [:space:] between members.
1607          * Type 1:
1608          *   attrname[:space:]=[:space:]'CDATA'
1609          * Type 2:
1610          *   attrname[:space:]=[:space:]"CDATA"
1611          * Type 3:
1612          *   attr[:space:]=[:space:]CDATA
1613          * Type 4:
1614          *   attrname
1615          *
1616          * We leave types 1 and 2 the same, type 3 we check for
1617          * '"' and convert to "&quot" if needed, then wrap in
1618          * double quotes. Type 4 we convert into:
1619          * attrname="yes".
1620          */
1621         $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1622         if ($regary == false){
1623             /**
1624              * Looks like body ended before the end of tag.
1625              */
1626             return Array(false, false, false, $lt, strlen($body));
1627         }
1628         list($pos, $attname, $match) = $regary;
1629         $attname = strtolower($attname);
1630         /**
1631          * We arrived at the end of attribute name. Several things possible
1632          * here:
1633          * '>'  means the end of the tag and this is attribute type 4
1634          * '/'  if followed by '>' means the same thing as above
1635          * '\s' means a lot of things -- look what it's followed by.
1636          *      anything else means the attribute is invalid.
1637          */
1638         switch($match){
1639             case '/':
1640                 /**
1641                  * This is an xhtml-style tag with a closing / at the
1642                  * end, like so: <img src="blah" />. Check if it's followed
1643                  * by the closing bracket. If not, then this tag is invalid
1644                  */
1645                 if (substr($body, $pos, 2) == "/>"){
1646                     $pos++;
1647                     $tagtype = 3;
1648                 } else {
1649                     $gt = sq_findnxstr($body, $pos, ">");
1650                     $retary = Array(false, false, false, $lt, $gt);
1651                     return $retary;
1652                 }
1653             case '>':
1654                 $attary{$attname} = '"yes"';
1655                 return Array($tagname, $attary, $tagtype, $lt, $pos);
1656                 break;
1657             default:
1658                 /**
1659                  * Skip whitespace and see what we arrive at.
1660                  */
1661                 $pos = sq_skipspace($body, $pos);
1662                 $char = substr($body, $pos, 1);
1663                 /**
1664                  * Two things are valid here:
1665                  * '=' means this is attribute type 1 2 or 3.
1666                  * \w means this was attribute type 4.
1667                  * anything else we ignore and re-loop. End of tag and
1668                  * invalid stuff will be caught by our checks at the beginning
1669                  * of the loop.
1670                  */
1671                 if ($char == "="){
1672                     $pos++;
1673                     $pos = sq_skipspace($body, $pos);
1674                     /**
1675                      * Here are 3 possibilities:
1676                      * "'"  attribute type 1
1677                      * '"'  attribute type 2
1678                      * everything else is the content of tag type 3
1679                      */
1680                     $quot = substr($body, $pos, 1);
1681                     if ($quot == "'"){
1682                         $regary = sq_findnxreg($body, $pos+1, "\'");
1683                         if ($regary == false){
1684                             return Array(false, false, false, $lt, strlen($body));
1685                         }
1686                         list($pos, $attval, $match) = $regary;
1687                         $pos++;
1688                         $attary{$attname} = "'" . $attval . "'";
1689                     } else if ($quot == '"'){
1690                         $regary = sq_findnxreg($body, $pos+1, '\"');
1691                         if ($regary == false){
1692                             return Array(false, false, false, $lt, strlen($body));
1693                         }
1694                         list($pos, $attval, $match) = $regary;
1695                         $pos++;
1696                         $attary{$attname} = '"' . $attval . '"';
1697                     } else {
1698                         /**
1699                          * These are hateful. Look for \s, or >.
1700                          */
1701                         $regary = sq_findnxreg($body, $pos, "[\s>]");
1702                         if ($regary == false){
1703                             return Array(false, false, false, $lt, strlen($body));
1704                         }
1705                         list($pos, $attval, $match) = $regary;
1706                         /**
1707                          * If it's ">" it will be caught at the top.
1708                          */
1709                         $attval = preg_replace("/\"/s", "&quot;", $attval);
1710                         $attary{$attname} = '"' . $attval . '"';
1711                     }
1712                 } else if (preg_match("|[\w/>]|", $char)) {
1713                     /**
1714                      * That was attribute type 4.
1715                      */
1716                     $attary{$attname} = '"yes"';
1717                 } else {
1718                     /**
1719                      * An illegal character. Find next '>' and return.
1720                      */
1721                     $gt = sq_findnxstr($body, $pos, ">");
1722                     return Array(false, false, false, $lt, $gt);
1723                 }
1724                 break;
1725         }
1726     }
1727     /**
1728      * The fact that we got here indicates that the tag end was never
1729      * found. Return invalid tag indication so it gets stripped.
1730      */
1731     return Array(false, false, false, $lt, strlen($body));
1732 }
1733
1734 /**
1735  * Translates entities into literal values so they can be checked.
1736  *
1737  * @param $attvalue the by-ref value to check.
1738  * @param $regex    the regular expression to check against.
1739  * @param $hex      whether the entites are hexadecimal.
1740  * @return          True or False depending on whether there were matches.
1741  */
1742 function sq_deent(&$attvalue, $regex, $hex=false){
1743     $me = 'sq_deent';
1744     $ret_match = false;
1745     // remove comments
1746     //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1747     preg_match_all($regex, $attvalue, $matches);
1748     if (is_array($matches) && sizeof($matches[0]) > 0){
1749         $repl = Array();
1750         for ($i = 0; $i < sizeof($matches[0]); $i++){
1751             $numval = $matches[1][$i];
1752             if ($hex){
1753                 $numval = hexdec($numval);
1754             }
1755             $repl{$matches[0][$i]} = chr($numval);
1756         }
1757         $attvalue = strtr($attvalue, $repl);
1758         return true;
1759     } else {
1760         return false;
1761     }
1762 }
1763
1764 /**
1765  * This function runs various checks against the attributes.
1766  *
1767  * @param  $tagname         String with the name of the tag.
1768  * @param  $attary          Array with all tag attributes.
1769  * @param  $rm_attnames     See description for sq_sanitize
1770  * @param  $bad_attvals     See description for sq_sanitize
1771  * @param  $add_attr_to_tag See description for sq_sanitize
1772  * @param  $message         message object
1773  * @param  $id              message id
1774  * @return                  Array with modified attributes.
1775  */
1776 function sq_fixatts($tagname,
1777                     $attary,
1778                     $rm_attnames,
1779                     $bad_attvals,
1780                     $add_attr_to_tag,
1781                     $message,
1782                     $id,
1783                     $mailbox
1784                     ){
1785     $me = 'sq_fixatts';
1786     while (list($attname, $attvalue) = each($attary)){
1787         /**
1788          * See if this attribute should be removed.
1789          */
1790         foreach ($rm_attnames as $matchtag=>$matchattrs){
1791             if (preg_match($matchtag, $tagname)){
1792                 foreach ($matchattrs as $matchattr){
1793                     if (preg_match($matchattr, $attname)){
1794                         unset($attary{$attname});
1795                         continue;
1796                     }
1797                 }
1798             }
1799         }
1800         /**
1801          * Workaround for IE quirks
1802          */
1803         sq_fixIE_idiocy($attvalue);
1804
1805         /**
1806          * Remove any backslashes, entities, and extraneous whitespace.
1807          */
1808
1809         $oldattvalue = $attvalue;
1810         sq_defang($attvalue);
1811         if ($attname == 'style' && $attvalue !== $oldattvalue) {
1812             // entities are used in the attribute value. In 99% of the cases it's there as XSS
1813             // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1814             $attvalue = "idiocy";
1815             $attary{$attname} = $attvalue;
1816         }
1817         sq_unspace($attvalue);
1818
1819         /**
1820          * Now let's run checks on the attvalues.
1821          * I don't expect anyone to comprehend this. If you do,
1822          * get in touch with me so I can drive to where you live and
1823          * shake your hand personally. :)
1824          */
1825         foreach ($bad_attvals as $matchtag=>$matchattrs){
1826             if (preg_match($matchtag, $tagname)){
1827                 foreach ($matchattrs as $matchattr=>$valary){
1828                     if (preg_match($matchattr, $attname)){
1829                         /**
1830                          * There are two arrays in valary.
1831                          * First is matches.
1832                          * Second one is replacements
1833                          */
1834                         list($valmatch, $valrepl) = $valary;
1835                         $newvalue =
1836                             preg_replace($valmatch, $valrepl, $attvalue);
1837                         if ($newvalue != $attvalue){
1838                             $attary{$attname} = $newvalue;
1839                             $attvalue = $newvalue;
1840                         }
1841                     }
1842                 }
1843             }
1844         }
1845         if ($attname == 'style') {
1846             if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1847                 // 8bit and control characters in style attribute values can be used for XSS, remove them
1848                 $attary{$attname} = '"disallowed character"';
1849             }
1850             preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1851             if (count($aMatch)) {
1852                 foreach($aMatch[1] as $sMatch) {
1853                     // url value
1854                     $urlvalue = $sMatch;
1855                     sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1856                     $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1857                 }
1858             }
1859         }
1860         /**
1861          * Use white list based filtering on attributes which can contain url's
1862          */
1863         else if ($attname == 'href' || $attname == 'src' || $attname == 'background') {
1864             sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1865             $attary{$attname} = $attvalue;
1866         }
1867     }
1868     /**
1869      * See if we need to append any attributes to this tag.
1870      */
1871     foreach ($add_attr_to_tag as $matchtag=>$addattary){
1872         if (preg_match($matchtag, $tagname)){
1873             $attary = array_merge($attary, $addattary);
1874         }
1875     }
1876     return $attary;
1877 }
1878
1879 /**
1880  * This function filters url's
1881  *
1882  * @param  $attvalue        String with attribute value to filter
1883  * @param  $message         message object
1884  * @param  $id               message id
1885  * @param  $mailbox         mailbox
1886  * @param  $sQuote          quoting characters around url's
1887  */
1888 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1889     $attvalue = trim($attvalue);
1890     if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1891         // remove the double quotes
1892         $sQuote = $attvalue[0];
1893         $attvalue = trim(substr($attvalue,1,-1));
1894     }
1895
1896     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1897     // images off by default.
1898     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1899
1900     $secremoveimg = '../images/' . _("sec_remove_eng.png");
1901
1902     /**
1903      * Replace empty src tags with the blank image.  src is only used
1904      * for frames, images, and image inputs.  Doing a replace should
1905      * not affect them working as should be, however it will stop
1906      * IE from being kicked off when src for img tags are not set
1907      */
1908     if ($attvalue == '') {
1909         $attvalue = '"' . SM_PATH . 'images/blank.png"';
1910     } else {
1911         // first, disallow 8 bit characters and control characters
1912         if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1913             switch ($attname) {
1914                 case 'href':
1915                     $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1916                     break;
1917                 default:
1918                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1919                     break;
1920             }
1921         } else {
1922             $aUrl = parse_url($attvalue);
1923             if (isset($aUrl['scheme'])) {
1924                 switch(strtolower($aUrl['scheme'])) {
1925                     case 'mailto':
1926                     case 'http':
1927                     case 'https':
1928                     case 'ftp':
1929                         if ($attname != 'href') {
1930                             if ($view_unsafe_images == false) {
1931                                 $attvalue = $sQuote . $secremoveimg . $sQuote;
1932                             } else {
1933                                 if (isset($aUrl['path'])) {
1934
1935                                     // No one has been able to show that image URIs
1936                                     // can be exploited, so for now, no restrictions
1937                                     // are made at all.  If this proves to be a problem,
1938                                     // the commented-out code below can be of help.
1939                                     // (One consideration is that I see nothing in this
1940                                     // function that specifically says that we will
1941                                     // only ever arrive here when inspecting an image
1942                                     // tag, although that does seem to be the end
1943                                     // result - e.g., <script src="..."> where malicious
1944                                     // image URIs are in fact a problem are already
1945                                     // filtered out elsewhere.
1946                                     /* ---------------------------------
1947                                     // validate image extension.
1948                                     $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1949                                     if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1950                                         // If URI is to something other than
1951                                         // a regular image file, get the contents
1952                                         // and try to see if it is an image.
1953                                         // Don't use Fileinfo (finfo_file()) because
1954                                         // we'd need to make the admin configure the
1955                                         // location of the magic.mime file (FIXME: add finfo_file() support later?)
1956                                         //
1957                                         $mime_type = '';
1958                                         if (function_exists('mime_content_type')
1959                                          && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1960
1961                                             // fetch file
1962                                             //
1963                                             $file_contents = '';
1964                                             while (!feof($FILE)) {
1965                                                 $file_contents .= fread($FILE, 8192);
1966                                             }
1967                                             fclose($FILE);
1968
1969                                             // store file locally
1970                                             //
1971                                             global $attachment_dir, $username;
1972                                             $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1973                                             $localfilename = GenerateRandomString(32, '', 7);
1974                                             $full_localfilename = "$hashed_attachment_dir/$localfilename";
1975                                             while (file_exists($full_localfilename)) {
1976                                                 $localfilename = GenerateRandomString(32, '', 7);
1977                                                 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1978                                             }
1979                                             $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
1980                                             fwrite($FILE, $file_contents);
1981                                             fclose($FILE);
1982
1983                                             // get mime type and remove file
1984                                             //
1985                                             $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
1986                                             unlink("$hashed_attachment_dir/$localfilename");
1987                                         }
1988                                         // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
1989                                         if (substr(strtolower($mime_type), 0, 5) != 'image') {
1990                                             $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1991                                         }
1992                                     }
1993                                     --------------------------------- */
1994                                 } else {
1995                                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1996                                 }
1997                             }
1998                         } else {
1999                             $attvalue = $sQuote . $attvalue . $sQuote;
2000                         }
2001                         break;
2002                     case 'outbind':
2003                         /**
2004                          * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2005                          * One day MS might actually make it match something useful, for now, falling
2006                          * back to using cid2http, so we can grab the blank.png.
2007                          */
2008                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2009                         break;
2010                     case 'cid':
2011                         /**
2012                             * Turn cid: urls into http-friendly ones.
2013                             */
2014                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2015                         break;
2016                     default:
2017                         $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2018                         break;
2019                 }
2020             } else {
2021                 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2022                     // parse_url did not lead to satisfying result
2023                     $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2024                 }
2025             }
2026         }
2027     }
2028 }
2029
2030 /**
2031  * This function edits the style definition to make them friendly and
2032  * usable in SquirrelMail.
2033  *
2034  * @param  $message  the message object
2035  * @param  $id       the message id
2036  * @param  $content  a string with whatever is between <style> and </style>
2037  * @param  $mailbox  the message mailbox
2038  * @return           a string with edited content.
2039  */
2040 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2041     $me = 'sq_fixstyle';
2042     // workaround for </style> in between comments
2043     $iCurrentPos = $pos;
2044     $content = '';
2045     $sToken = '';
2046     $bSucces = false;
2047     $bEndTag = false;
2048     for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2049         $char = $body{$i};
2050         switch ($char) {
2051             case '<':
2052                 $sToken = $char;
2053                 break;
2054             case '/':
2055                  if ($sToken == '<') {
2056                     $sToken .= $char;
2057                     $bEndTag = true;
2058                  } else {
2059                     $content .= $char;
2060                  }
2061                  break;
2062             case '>':
2063                  if ($bEndTag) {
2064                     $sToken .= $char;
2065                     if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2066                         $newpos = $i + 1;
2067                         $bSucces = true;
2068                         break 2;
2069                     } else {
2070                         $content .= $sToken;
2071                     }
2072                     $bEndTag = false;
2073                  } else {
2074                     $content .= $char;
2075                  }
2076                  break;
2077             case '!':
2078                 if ($sToken == '<') {
2079                     // possible comment
2080                     if (isset($body{$i+2}) && substr($body,$i,3) == '!--') {
2081                         $i = strpos($body,'-->',$i+3);
2082                         if ($i === false) { // no end comment
2083                             $i = strlen($body);
2084                         }
2085                         $sToken = '';
2086                     }
2087                 } else {
2088                     $content .= $char;
2089                 }
2090                 break;
2091             default:
2092                 if ($bEndTag) {
2093                     $sToken .= $char;
2094                 } else {
2095                     $content .= $char;
2096                 }
2097                 break;
2098         }
2099     }
2100     if ($bSucces == FALSE){
2101         return array(FALSE, strlen($body));
2102     }
2103
2104
2105
2106     /**
2107      * First look for general BODY style declaration, which would be
2108      * like so:
2109      * body {background: blah-blah}
2110      * and change it to .bodyclass so we can just assign it to a <div>
2111      */
2112     $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2113     $secremoveimg = '../images/' . _("sec_remove_eng.png");
2114     /**
2115     * Fix url('blah') declarations.
2116     */
2117     //   $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2118     //                           "url(\\1$secremoveimg\\2)", $content);
2119
2120     // first check for 8bit sequences and disallowed control characters
2121     if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2122         $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2123         return array($content, $newpos);
2124     }
2125
2126     // IE Sucks hard. We have a special function for it.
2127     sq_fixIE_idiocy($content);
2128
2129     // remove @import line
2130     $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2131
2132     // translate ur\l and variations (IE parses that)
2133     // TODO check if the sq_fixIE_idiocy function already handles this.
2134     $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2135     preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2136     if (count($aMatch)) {
2137         $aValue = $aReplace = array();
2138         foreach($aMatch[1] as $sMatch) {
2139             // url value
2140             $urlvalue = $sMatch;
2141             sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2142             $aValue[] = $sMatch;
2143             $aReplace[] = $urlvalue;
2144         }
2145         $content = str_replace($aValue,$aReplace,$content);
2146     }
2147
2148     /**
2149      * Remove any backslashes, entities, and extraneous whitespace.
2150      */
2151     $contentTemp = $content;
2152     sq_defang($contentTemp);
2153     sq_unspace($contentTemp);
2154
2155     /**
2156      * Fix stupid css declarations which lead to vulnerabilities
2157      * in IE.
2158      *
2159      * Also remove "position" attribute, as it can easily be set
2160      * to "fixed" or "absolute" with "left" and "top" attributes
2161      * of zero, taking over the whole content frame.  It can also
2162      * be set to relative and move itself anywhere it wants to,
2163      * displaying content in areas it shouldn't be allowed to touch.
2164      */
2165     $match   = Array('/\/\*.*\*\//',
2166                     '/expression/i',
2167                     '/behaviou*r/i',
2168                     '/binding/i',
2169                     '/include-source/i',
2170                     '/javascript/i',
2171                     '/script/i',
2172                     '/position/i');
2173     $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2174     $contentNew = preg_replace($match, $replace, $contentTemp);
2175     if ($contentNew !== $contentTemp) {
2176         // insecure css declarations are used. From now on we don't care
2177         // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2178         $content = $contentNew;
2179     }
2180     return array($content, $newpos);
2181 }
2182
2183
2184 /**
2185  * This function converts cid: url's into the ones that can be viewed in
2186  * the browser.
2187  *
2188  * @param  $message  the message object
2189  * @param  $id       the message id
2190  * @param  $cidurl   the cid: url.
2191  * @param  $mailbox  the message mailbox
2192  * @return           a string with a http-friendly url
2193  */
2194 function sq_cid2http($message, $id, $cidurl, $mailbox){
2195     /**
2196      * Get rid of quotes.
2197      */
2198     $quotchar = substr($cidurl, 0, 1);
2199     if ($quotchar == '"' || $quotchar == "'"){
2200         $cidurl = str_replace($quotchar, "", $cidurl);
2201     } else {
2202         $quotchar = '';
2203     }
2204     $cidurl = substr(trim($cidurl), 4);
2205
2206     $match_str = '/\{.*?\}\//';
2207     $str_rep = '';
2208     $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2209
2210     $linkurl = find_ent_id($cidurl, $message);
2211     /* in case of non-safe cid links $httpurl should be replaced by a sort of
2212        unsafe link image */
2213     $httpurl = '';
2214
2215     /**
2216      * This is part of a fix for Outlook Express 6.x generating
2217      * cid URLs without creating content-id headers. These images are
2218      * not part of the multipart/related html mail. The html contains
2219      * <img src="cid:{some_id}/image_filename.ext"> references to
2220      * attached images with as goal to render them inline although
2221      * the attachment disposition property is not inline.
2222      */
2223
2224     if (empty($linkurl)) {
2225         if (preg_match('/{.*}\//', $cidurl)) {
2226             $cidurl = preg_replace('/{.*}\//','', $cidurl);
2227             if (!empty($cidurl)) {
2228                 $linkurl = find_ent_id($cidurl, $message);
2229             }
2230         }
2231     }
2232
2233     if (!empty($linkurl)) {
2234         $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2235             "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2236             '&amp;ent_id=' . $linkurl . $quotchar;
2237     } else {
2238         /**
2239          * If we couldn't generate a proper img url, drop in a blank image
2240          * instead of sending back empty, otherwise it causes unusual behaviour
2241          */
2242         $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2243     }
2244
2245     return $httpurl;
2246 }
2247
2248 /**
2249  * This function changes the <body> tag into a <div> tag since we
2250  * can't really have a body-within-body.
2251  *
2252  * @param  $attary   an array of attributes and values of <body>
2253  * @param  $mailbox  mailbox we're currently reading (for cid2http)
2254  * @param  $message  current message (for cid2http)
2255  * @param  $id       current message id (for cid2http)
2256  * @return           a modified array of attributes to be set for <div>
2257  */
2258 function sq_body2div($attary, $mailbox, $message, $id){
2259     $me = 'sq_body2div';
2260     $divattary = Array('class' => "'bodyclass'");
2261     $text = '#000000';
2262     $has_bgc_stl = $has_txt_stl = false;
2263     $styledef = '';
2264     if (is_array($attary) && sizeof($attary) > 0){
2265         foreach ($attary as $attname=>$attvalue){
2266             $quotchar = substr($attvalue, 0, 1);
2267             $attvalue = str_replace($quotchar, "", $attvalue);
2268             switch ($attname){
2269                 case 'background':
2270                     $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2271                     $styledef .= "background-image: url('$attvalue'); ";
2272                     break;
2273                 case 'bgcolor':
2274                     $has_bgc_stl = true;
2275                     $styledef .= "background-color: $attvalue; ";
2276                     break;
2277                 case 'text':
2278                     $has_txt_stl = true;
2279                     $styledef .= "color: $attvalue; ";
2280                     break;
2281             }
2282         }
2283         // Outlook defines a white bgcolor and no text color. This can lead to
2284         // white text on a white bg with certain themes.
2285         if ($has_bgc_stl && !$has_txt_stl) {
2286             $styledef .= "color: $text; ";
2287         }
2288         if (strlen($styledef) > 0){
2289             $divattary{"style"} = "\"$styledef\"";
2290         }
2291     }
2292     return $divattary;
2293 }
2294
2295 /**
2296  * This is the main function and the one you should actually be calling.
2297  * There are several variables you should be aware of an which need
2298  * special description.
2299  *
2300  * Since the description is quite lengthy, see it here:
2301  * http://linux.duke.edu/projects/mini/htmlfilter/
2302  *
2303  * @param $body                 the string with HTML you wish to filter
2304  * @param $tag_list             see description above
2305  * @param $rm_tags_with_content see description above
2306  * @param $self_closing_tags    see description above
2307  * @param $force_tag_closing    see description above
2308  * @param $rm_attnames          see description above
2309  * @param $bad_attvals          see description above
2310  * @param $add_attr_to_tag      see description above
2311  * @param $message              message object
2312  * @param $id                   message id
2313  * @return                      sanitized html safe to show on your pages.
2314  */
2315 function sq_sanitize($body,
2316                      $tag_list,
2317                      $rm_tags_with_content,
2318                      $self_closing_tags,
2319                      $force_tag_closing,
2320                      $rm_attnames,
2321                      $bad_attvals,
2322                      $add_attr_to_tag,
2323                      $message,
2324                      $id,
2325                      $mailbox
2326                      ){
2327     $me = 'sq_sanitize';
2328     $rm_tags = array_shift($tag_list);
2329     /**
2330      * Normalize rm_tags and rm_tags_with_content.
2331      */
2332     @array_walk($tag_list, 'sq_casenormalize');
2333     @array_walk($rm_tags_with_content, 'sq_casenormalize');
2334     @array_walk($self_closing_tags, 'sq_casenormalize');
2335     /**
2336      * See if tag_list is of tags to remove or tags to allow.
2337      * false  means remove these tags
2338      * true   means allow these tags
2339      */
2340     $curpos = 0;
2341     $open_tags = Array();
2342     $trusted = "\n<!-- begin sanitized html -->\n";
2343     $skip_content = false;
2344     /**
2345      * Take care of netscape's stupid javascript entities like
2346      * &{alert('boo')};
2347      */
2348     $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2349
2350     while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2351         list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2352         $free_content = substr($body, $curpos, $lt-$curpos);
2353         /**
2354          * Take care of <style>
2355          */
2356         if ($tagname == "style" && $tagtype == 1){
2357             list($free_content, $curpos) =
2358                 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2359             if ($free_content != FALSE){
2360                 if ( !empty($attary) ) {
2361                     $attary = sq_fixatts($tagname,
2362                                          $attary,
2363                                          $rm_attnames,
2364                                          $bad_attvals,
2365                                          $add_attr_to_tag,
2366                                          $message,
2367                                          $id,
2368                                          $mailbox
2369                                          );
2370                 }
2371                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2372                 $trusted .= $free_content;
2373                 $trusted .= sq_tagprint($tagname, false, 2);
2374             }
2375             continue;
2376         }
2377         if ($skip_content == false){
2378             $trusted .= $free_content;
2379         }
2380         if ($tagname != FALSE){
2381             if ($tagtype == 2){
2382                 if ($skip_content == $tagname){
2383                     /**
2384                      * Got to the end of tag we needed to remove.
2385                      */
2386                     $tagname = false;
2387                     $skip_content = false;
2388                 } else {
2389                     if ($skip_content == false){
2390                         if ($tagname == "body"){
2391                             $tagname = "div";
2392                         }
2393                         if (isset($open_tags{$tagname}) &&
2394                                 $open_tags{$tagname} > 0){
2395                             $open_tags{$tagname}--;
2396                         } else {
2397                             $tagname = false;
2398                         }
2399                     }
2400                 }
2401             } else {
2402                 /**
2403                  * $rm_tags_with_content
2404                  */
2405                 if ($skip_content == false){
2406                     /**
2407                      * See if this is a self-closing type and change
2408                      * tagtype appropriately.
2409                      */
2410                     if ($tagtype == 1
2411                             && in_array($tagname, $self_closing_tags)){
2412                         $tagtype = 3;
2413                     }
2414                     /**
2415                      * See if we should skip this tag and any content
2416                      * inside it.
2417                      */
2418                     if ($tagtype == 1 &&
2419                             in_array($tagname, $rm_tags_with_content)){
2420                         $skip_content = $tagname;
2421                     } else {
2422                         if (($rm_tags == false
2423                                     && in_array($tagname, $tag_list)) ||
2424                                 ($rm_tags == true &&
2425                                  !in_array($tagname, $tag_list))){
2426                             $tagname = false;
2427                         } else {
2428                             /**
2429                              * Convert body into div.
2430                              */
2431                             if ($tagname == "body"){
2432                                 $tagname = "div";
2433                                 $attary = sq_body2div($attary, $mailbox,
2434                                         $message, $id);
2435                             }
2436                             if ($tagtype == 1){
2437                                 if (isset($open_tags{$tagname})){
2438                                     $open_tags{$tagname}++;
2439                                 } else {
2440                                     $open_tags{$tagname}=1;
2441                                 }
2442                             }
2443                             /**
2444                              * This is where we run other checks.
2445                              */
2446                             if (is_array($attary) && sizeof($attary) > 0){
2447                                 $attary = sq_fixatts($tagname,
2448                                                      $attary,
2449                                                      $rm_attnames,
2450                                                      $bad_attvals,
2451                                                      $add_attr_to_tag,
2452                                                      $message,
2453                                                      $id,
2454                                                      $mailbox
2455                                                      );
2456                             }
2457                         }
2458                     }
2459                 }
2460             }
2461             if ($tagname != false && $skip_content == false){
2462                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2463             }
2464         }
2465         $curpos = $gt+1;
2466     }
2467     $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2468     if ($force_tag_closing == true){
2469         foreach ($open_tags as $tagname=>$opentimes){
2470             while ($opentimes > 0){
2471                 $trusted .= '</' . $tagname . '>';
2472                 $opentimes--;
2473             }
2474         }
2475         $trusted .= "\n";
2476     }
2477     $trusted .= "<!-- end sanitized html -->\n";
2478     return $trusted;
2479 }
2480
2481 /**
2482  * This is a wrapper function to call html sanitizing routines.
2483  *
2484  * @param  $body  the body of the message
2485  * @param  $id    the id of the message
2486
2487  * @param  $message
2488  * @param  $mailbox
2489  * @param  boolean $take_mailto_links When TRUE, converts mailto: links
2490  *                                    into internal SM compose links
2491  *                                    (optional; default = TRUE)
2492  * @return        a string with html safe to display in the browser.
2493  */
2494 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2495
2496     // require_once(SM_PATH . 'functions/url_parser.php');  // for $MailTo_PReg_Match
2497
2498     global $attachment_common_show_images, $view_unsafe_images,
2499            $has_unsafe_images;
2500     /**
2501      * Don't display attached images in HTML mode.
2502      *
2503      * SB: why?
2504      */
2505     $attachment_common_show_images = false;
2506     $tag_list = Array(
2507             false,
2508             "object",
2509             "meta",
2510             "html",
2511             "head",
2512             "base",
2513             "link",
2514             "frame",
2515             "iframe",
2516             "plaintext",
2517             "marquee"
2518             );
2519
2520     $rm_tags_with_content = Array(
2521             "script",
2522             "applet",
2523             "embed",
2524             "title",
2525             "frameset",
2526             "xmp",
2527             "xml"
2528             );
2529
2530     $self_closing_tags =  Array(
2531             "img",
2532             "br",
2533             "hr",
2534             "input",
2535             "outbind"
2536             );
2537
2538     $force_tag_closing = true;
2539
2540     $rm_attnames = Array(
2541             "/.*/" =>
2542             Array(
2543                 "/target/i",
2544                 "/^on.*/i",
2545                 "/^dynsrc/i",
2546                 "/^data.*/i",
2547                 "/^lowsrc.*/i"
2548                 )
2549             );
2550
2551     $secremoveimg = "../images/" . _("sec_remove_eng.png");
2552     $bad_attvals = Array(
2553             "/.*/" =>
2554             Array(
2555                 "/^src|background/i" =>
2556                 Array(
2557                     Array(
2558                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2559                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2560                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2561                         ),
2562                     Array(
2563                         "\\1$secremoveimg\\2",
2564                         "\\1$secremoveimg\\2",
2565                         "\\1$secremoveimg\\2",
2566                         )
2567                     ),
2568                 "/^href|action/i" =>
2569                 Array(
2570                     Array(
2571                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2572                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2573                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2574                         ),
2575                     Array(
2576                         "\\1#\\1",
2577                         "\\1#\\1",
2578                         "\\1#\\1"
2579                         )
2580                     ),
2581         "/^style/i" =>
2582             Array(
2583                 Array(
2584                     "/\/\*.*\*\//",
2585                     "/expression/i",
2586                     "/binding/i",
2587                     "/behaviou*r/i",
2588                     "/include-source/i",
2589
2590                     // position:relative can also be exploited
2591                     // to put content outside of email body area
2592                     // and position:fixed is similarly exploitable
2593                     // as position:absolute, so we'll remove it
2594                     // altogether....
2595                     //
2596                     // Does this screw up legitimate HTML messages?
2597                     // If so, the only fix I see is to allow position
2598                     // attributes (any values?  I think we still have
2599                     // to block static and fixed) only if $use_iframe
2600                     // is enabled (1.5.0+)
2601                     //
2602                     // was:   "/position\s*:\s*absolute/i",
2603                     //
2604                     "/position\s*:/i",
2605
2606                     "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2607                     "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2608                     "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2609                     "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2610                     "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2611                     ),
2612                 Array(
2613                     "",
2614                     "idiocy",
2615                     "idiocy",
2616                     "idiocy",
2617                     "idiocy",
2618                     "idiocy",
2619                     "url",
2620                     "url(\\1#\\1)",
2621                     "url(\\1#\\1)",
2622                     "url(\\1#\\1)",
2623                     "\\1:url(\\2#\\3)"
2624                     )
2625                 )
2626             )
2627         );
2628
2629     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2630     // images off by default.
2631     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2632
2633     if (!$view_unsafe_images){
2634         /**
2635          * Remove any references to http/https if view_unsafe_images set
2636          * to false.
2637          */
2638         array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2639                 '/^([\'\"])\s*https*:.*([\'\"])/si');
2640         array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2641                 "\\1$secremoveimg\\1");
2642         array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2643                 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2644         array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2645                 "url(\\1$secremoveimg\\1)");
2646     }
2647
2648     $add_attr_to_tag = Array(
2649             "/^a$/i" =>
2650             Array('target'=>'"_blank"',
2651                 'title'=>'"'._("This external link will open in a new window").'"'
2652                 )
2653             );
2654     $trusted = sq_sanitize($body,
2655                            $tag_list,
2656                            $rm_tags_with_content,
2657                            $self_closing_tags,
2658                            $force_tag_closing,
2659                            $rm_attnames,
2660                            $bad_attvals,
2661                            $add_attr_to_tag,
2662                            $message,
2663                            $id,
2664                            $mailbox
2665                            );
2666     if (strpos($trusted,$secremoveimg)){
2667         $has_unsafe_images = true;
2668     }
2669
2670     // we want to parse mailto's in HTML output, change to SM compose links
2671     // this is a modified version of code from url_parser.php... but Marc is
2672     // right: we need a better filtering implementation; adding this randomly
2673     // here is not a great solution
2674     //
2675     if ($take_mailto_links) {
2676         // parseUrl($trusted);   // this even parses URLs inside of tags... too aggressive
2677         global $MailTo_PReg_Match;
2678         $MailTo_PReg_Match = '/mailto:' . substr($MailTo_PReg_Match, 1) ;
2679         if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2680             foreach ($regs[0] as $i => $mailto_before) {
2681                 $mailto_params = $regs[10][$i];
2682                 // get rid of any tailing quote since we have to add send_to to the end
2683                 //
2684                 if (substr($mailto_before, strlen($mailto_before) - 1) == '"')
2685                     $mailto_before = substr($mailto_before, 0, strlen($mailto_before) - 1);
2686                 if (substr($mailto_params, strlen($mailto_params) - 1) == '"')
2687                     $mailto_params = substr($mailto_params, 0, strlen($mailto_params) - 1);
2688
2689                 if ($regs[1][$i]) {    //if there is an email addr before '?', we need to merge it with the params
2690                     $to = 'to=' . $regs[1][$i];
2691                     if (strpos($mailto_params, 'to=') > -1)    //already a 'to='
2692                         $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2693                     else {
2694                         if ($mailto_params)    //already some params, append to them
2695                             $mailto_params .= '&amp;' . $to;
2696                         else
2697                             $mailto_params .= '?' . $to;
2698                     }
2699                 }
2700
2701                 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2702
2703                 // we'll already have target=_blank, no need to allow comp_in_new
2704                 // here (which would be a lot more work anyway)
2705                 //
2706                 global $compose_new_win;
2707                 $temp_comp_in_new = $compose_new_win;
2708                 $compose_new_win = 0;
2709                 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2710                 $compose_new_win = $temp_comp_in_new;
2711
2712                 // remove <a href=" and anything after the next quote (we only
2713                 // need the uri, not the link HTML) in compose uri
2714                 //
2715                 $comp_uri = substr($comp_uri, 9);
2716                 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2717                 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2718             }
2719         }
2720     }
2721
2722     return $trusted;
2723 }
2724
2725 /**
2726  * function SendDownloadHeaders - send file to the browser
2727  *
2728  * Original Source: SM core src/download.php
2729  * moved here to make it available to other code, and separate
2730  * front end from back end functionality.
2731  *
2732  * @param string $type0 first half of mime type
2733  * @param string $type1 second half of mime type
2734  * @param string $filename filename to tell the browser for downloaded file
2735  * @param boolean $force whether to force the download dialog to pop
2736  * @param optional integer $filesize send the Content-Header and length to the browser
2737  * @return void
2738  */
2739 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2740     global $languages, $squirrelmail_language;
2741     $isIE = $isIE6plus = false;
2742
2743     sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2744
2745     if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2746             strstr($HTTP_USER_AGENT, 'Opera') === false) {
2747         $isIE = true;
2748     }
2749
2750     if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2751         ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2752         $isIE6plus = true;
2753     }
2754
2755     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2756             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2757         $filename =
2758             call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2759     } else {
2760         $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2761     }
2762
2763     // A Pox on Microsoft and it's Internet Explorer!
2764     //
2765     // IE has lots of bugs with file downloads.
2766     // It also has problems with SSL.  Both of these cause problems
2767     // for us in this function.
2768     //
2769     // See this article on Cache Control headers and SSL
2770     // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2771     //
2772     // The best thing you can do for IE is to upgrade to the latest
2773     // version
2774     //set all the Cache Control Headers for IE
2775     if ($isIE) {
2776         $filename=rawurlencode($filename);
2777         header ("Pragma: public");
2778         header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2779         header ("Cache-Control: post-check=0, pre-check=0", false);
2780         header ("Cache-Control: private");
2781
2782         //set the inline header for IE, we'll add the attachment header later if we need it
2783         header ("Content-Disposition: inline; filename=$filename");
2784     }
2785
2786     if (!$force) {
2787         // Try to show in browser window
2788         header ("Content-Disposition: inline; filename=\"$filename\"");
2789         header ("Content-Type: $type0/$type1; name=\"$filename\"");
2790     } else {
2791         // Try to pop up the "save as" box
2792
2793         // IE makes this hard.  It pops up 2 save boxes, or none.
2794         // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2795         // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2796         // But, according to Microsoft, it is "RFC compliant but doesn't
2797         // take into account some deviations that allowed within the
2798         // specification."  Doesn't that mean RFC non-compliant?
2799         // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2800
2801         // all browsers need the application/octet-stream header for this
2802         header ("Content-Type: application/octet-stream; name=\"$filename\"");
2803
2804         // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2805         // Do not have quotes around filename, but that applied to
2806         // "attachment"... does it apply to inline too?
2807         header ("Content-Disposition: attachment; filename=\"$filename\"");
2808
2809         if ($isIE && !$isIE6plus) {
2810             // This combination seems to work mostly.  IE 5.5 SP 1 has
2811             // known issues (see the Microsoft Knowledge Base)
2812
2813             // This works for most types, but doesn't work with Word files
2814             header ("Content-Type: application/download; name=\"$filename\"");
2815             header ("Content-Type: application/force-download; name=\"$filename\"");
2816             // These are spares, just in case.  :-)
2817             //header("Content-Type: $type0/$type1; name=\"$filename\"");
2818             //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2819             //header("Content-Type: application/octet-stream; name=\"$filename\"");
2820         } else if ($isIE) {
2821              // This is to prevent IE for MIME sniffing and auto open a file in IE
2822              header ("Content-Type: application/force-download; name=\"$filename\"");
2823         } else {
2824             // another application/octet-stream forces download for Netscape
2825             header ("Content-Type: application/octet-stream; name=\"$filename\"");
2826         }
2827     }
2828
2829     //send the content-length header if the calling function provides it
2830     if ($filesize > 0) {
2831         header("Content-Length: $filesize");
2832     }
2833
2834 }  // end fn SendDownloadHeaders