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