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