functions/mime.php

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