functions/mime.php

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