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