functions/mime.php

   1 <?php
   2
   3 /**
   4  * mime.php
   5  *
   6  * This contains the functions necessary to detect and decode MIME
   7  * messages.
   8  *
   9  * @copyright 1999-2012 The SquirrelMail Project Team
  10  * @license http://opensource.org/licenses/gpl-license.php GNU Public License
  11  * @version $Id$
  12  * @package squirrelmail
  13  */
  14
  15 /**
  16  * dependency information
  17    functions       dependency
  18    mime_structure
  19         class/mime/Message.class.php
  20             Message::parseStructure
  21         functions/page_header.php
  22             displayPageHeader
  23         functions/display_messages.php
  24             plain_error_message
  25    mime_fetch_body
  26         functions/imap_general.php
  27             sqimap_run_command
  28    mime_print_body_lines
  29
  30
  31
  32 functions/imap.php
  33 functions/attachment_common.php
  34 functions/display_messages.php
  35
  36 magicHtml => url_parser
  37 translateText => url_parser
  38
  39 */
  40
  41
  42 /* -------------------------------------------------------------------------- */
  43 /* MIME DECODING                                                              */
  44 /* -------------------------------------------------------------------------- */
  45
  46 /**
  47  * Get the MIME structure
  48  *
  49  * This function gets the structure of a message and stores it in the "message" class.
  50  * It will return this object for use with all relevant header information and
  51  * fully parsed into the standard "message" object format.
  52  */
  53 function mime_structure ($bodystructure, $flags=array()) {
  54
  55     /* Isolate the body structure and remove beginning and end parenthesis. */
  56     $read = trim(substr ($bodystructure, strpos(strtolower($bodystructure), 'bodystructure') + 13));
  57     $read = trim(substr ($read, 0, -1));
  58     $i = 0;
  59     $msg = Message::parseStructure($read,$i);
  60
  61     if (!is_object($msg)) {
  62         global $color, $mailbox;
  63         displayPageHeader( $color, $mailbox );
  64         $errormessage  = _("SquirrelMail could not decode the bodystructure of the message");
  65         $errormessage .= '<br />'._("The bodystructure provided by your IMAP server:").'<br /><br />';
  66         $errormessage .= '<pre>' . sm_encode_html_special_chars($read) . '</pre>';
  67         plain_error_message( $errormessage );
  68         echo '</body></html>';
  69         exit;
  70     }
  71     if (count($flags)) {
  72         foreach ($flags as $flag) {
  73 //FIXME: please document why it is we have to check the first char of the flag but we then go ahead and do a full string comparison anyway.  Is this a speed enhancement?  If not, let's keep it simple and just compare the full string and forget the switch block.
  74             $char = strtoupper($flag{1});
  75             switch ($char) {
  76                 case 'S':
  77                     if (strtolower($flag) == '\\seen') {
  78                         $msg->is_seen = true;
  79                     }
  80                     break;
  81                 case 'A':
  82                     if (strtolower($flag) == '\\answered') {
  83                         $msg->is_answered = true;
  84                     }
  85                     break;
  86                 case 'D':
  87                     if (strtolower($flag) == '\\deleted') {
  88                         $msg->is_deleted = true;
  89                     }
  90                     break;
  91                 case 'F':
  92                     if (strtolower($flag) == '\\flagged') {
  93                         $msg->is_flagged = true;
  94                     }
  95                     else if (strtolower($flag) == '$forwarded') {
  96                         $msg->is_forwarded = true;
  97                     }
  98                     break;
  99                 case 'M':
 100                     if (strtolower($flag) == '$mdnsent') {
 101                         $msg->is_mdnsent = true;
 102                     }
 103                     break;
 104                 default:
 105                     break;
 106             }
 107         }
 108     }
 109     //    listEntities($msg);
 110     return $msg;
 111 }
 112
 113
 114
 115 /* This starts the parsing of a particular structure.  It is called recursively,
 116  * so it can be passed different structures.  It returns an object of type
 117  * $message.
 118  * First, it checks to see if it is a multipart message.  If it is, then it
 119  * handles that as it sees is necessary.  If it is just a regular entity,
 120  * then it parses it and adds the necessary header information (by calling out
 121  * to mime_get_elements()
 122  */
 123
 124 function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) {
 125     /* Do a bit of error correction.  If we couldn't find the entity id, just guess
 126      * that it is the first one.  That is usually the case anyway.
 127      */
 128
 129     if (!$ent_id) {
 130         $cmd = "FETCH $id BODY[]";
 131     } else {
 132         $cmd = "FETCH $id BODY[$ent_id]";
 133     }
 134
 135     if ($fetch_size!=0) $cmd .= "<0.$fetch_size>";
 136
 137     $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, TRUE);
 138     do {
 139         $topline = trim(array_shift($data));
 140     } while($topline && ($topline[0] == '*') && !preg_match('/\* [0-9]+ FETCH .*BODY.*/i', $topline)) ;
 141     // Matching with "BODY" above is difficult: in most cases "FETCH \(BODY" would work
 142     // but some servers may put other things in the same result, perhaps something such
 143     // as "* 23 FETCH (FLAGS (\Seen) BODY[1] {174}".  There is some small chance that
 144     // if the character sequence "BODY" appears in a response where it isn't actually
 145     // a FETCH response data item name, the current regex will break things.  The better
 146     // way to do this would be to parse the response correctly and not use a regex.
 147
 148     $wholemessage = implode('', $data);
 149     if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) {
 150         $ret = substr($wholemessage, 0, $regs[1]);
 151         /* There is some information in the content info header that could be important
 152          * in order to parse html messages. Let's get them here.
 153          */
 154 //        if ($ret{0} == '<') {
 155 //            $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE);
 156 //        }
 157     } else if (preg_match('/"([^"]*)"/', $topline, $regs)) {
 158         $ret = $regs[1];
 159     } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) {
 160         $ret = $wholemessage;
 161     } else {
 162         global $where, $what, $mailbox, $passed_id, $startMessage;
 163         $par = 'mailbox=' . urlencode($mailbox) . '&amp;passed_id=' . $passed_id;
 164         if (isset($where) && isset($what)) {
 165             $par .= '&amp;where=' . urlencode($where) . '&amp;what=' . urlencode($what);
 166         } else {
 167             $par .= '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
 168         }
 169         $par .= '&amp;response=' . urlencode($response) .
 170             '&amp;message='  . urlencode($message)  .
 171             '&amp;topline='  . urlencode($topline);
 172
 173         echo '<tt><br />' .
 174             '<table width="80%"><tr>' .
 175             '<tr><td colspan="2">' .
 176             _("Body retrieval error. The reason for this is most probably that the message is malformed.") .
 177             '</td></tr>' .
 178             '<tr><td><b>' . _("Command:") . "</td><td>$cmd</td></tr>" .
 179             '<tr><td><b>' . _("Response:") . "</td><td>$response</td></tr>" .
 180             '<tr><td><b>' . _("Message:") . "</td><td>$message</td></tr>" .
 181             '<tr><td><b>' . _("FETCH line:") . "</td><td>$topline</td></tr>" .
 182             "</table><br /></tt></font><hr />";
 183
 184         $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE);
 185         array_shift($data);
 186         $wholemessage = implode('', $data);
 187
 188         $ret = $wholemessage;
 189     }
 190     return $ret;
 191 }
 192
 193 function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') {
 194
 195     /* Don't kill the connection if the browser is over a dialup
 196      * and it would take over 30 seconds to download it.
 197      * Don't call set_time_limit in safe mode.
 198      */
 199
 200     if (!ini_get('safe_mode')) {
 201         set_time_limit(0);
 202     }
 203     /* in case of base64 encoded attachments, do not buffer them.
 204        Instead, echo the decoded attachment directly to screen */
 205     if (strtolower($encoding) == 'base64') {
 206         if (!$ent_id) {
 207             $query = "FETCH $id BODY[]";
 208         } else {
 209             $query = "FETCH $id BODY[$ent_id]";
 210         }
 211         sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
 212     } else {
 213         $body = mime_fetch_body ($imap_stream, $id, $ent_id);
 214         if (is_resource($rStream)) {
 215             fputs($rStream,decodeBody($body, $encoding, $force_crlf));
 216         } else {
 217             echo decodeBody($body, $encoding, $force_crlf);
 218         }
 219     }
 220
 221     /*
 222        TODO, use the same method for quoted printable.
 223        However, I assume that quoted printable attachments aren't that large
 224        so the performancegain / memory usage drop will be minimal.
 225        If we decide to add that then we need to adapt sqimap_fread because
 226        we need to split te result on \n and fread doesn't stop at \n. That
 227        means we also should provide $results from sqimap_fread (by ref) to
 228        te function and set $no_return to false. The $filter function for
 229        quoted printable should handle unsetting of $results.
 230      */
 231     /*
 232        TODO 2: find out how we write to the output stream php://stdout. fwrite
 233        doesn't work because 'php://stdout isn't a stream.
 234      */
 235
 236     return;
 237 }
 238
 239 /* -[ END MIME DECODING ]----------------------------------------------------------- */
 240
 241 /* This is here for debugging purposes.  It will print out a list
 242  * of all the entity IDs that are in the $message object.
 243  */
 244 function listEntities ($message) {
 245     if ($message) {
 246         echo "<tt>" . $message->entity_id . ' : ' . $message->type0 . '/' . $message->type1 . ' parent = '. $message->parent->entity_id. '<br />';
 247         for ($i = 0; isset($message->entities[$i]); $i++) {
 248             echo "$i : ";
 249             $msg = listEntities($message->entities[$i]);
 250
 251             if ($msg) {
 252                 echo "return: ";
 253                 return $msg;
 254             }
 255         }
 256     }
 257 }
 258
 259 function getPriorityStr($priority) {
 260     $priority_level = substr($priority,0,1);
 261
 262     switch($priority_level) {
 263         /* Check for a higher then normal priority. */
 264         case '1':
 265         case '2':
 266             $priority_string = _("High");
 267             break;
 268
 269         /* Check for a lower then normal priority. */
 270         case '4':
 271         case '5':
 272             $priority_string = _("Low");
 273             break;
 274
 275         /* Check for a normal priority. */
 276         case '3':
 277         default:
 278             $priority_level = '3';
 279             $priority_string = _("Normal");
 280             break;
 281
 282     }
 283     return $priority_string;
 284 }
 285
 286 /* returns a $message object for a particular entity id */
 287 function getEntity ($message, $ent_id) {
 288     return $message->getEntity($ent_id);
 289 }
 290
 291 /* translateText
 292  * Extracted from strings.php 23/03/2002
 293  */
 294
 295 function translateText(&$body, $wrap_at, $charset) {
 296     global $where, $what;   /* from searching */
 297     global $color;          /* color theme */
 298
 299     // require_once(SM_PATH . 'functions/url_parser.php');
 300
 301     $body_ary = explode("\n", $body);
 302     for ($i=0; $i < count($body_ary); $i++) {
 303         $line = rtrim($body_ary[$i],"\r");
 304
 305         if (strlen($line) - 2 >= $wrap_at) {
 306             sqWordWrap($line, $wrap_at, $charset);
 307         }
 308         $line = charset_decode($charset, $line);
 309         $line = str_replace("\t", '        ', $line);
 310
 311         parseUrl ($line);
 312
 313         $quotes = 0;
 314         $pos = 0;
 315         $j = strlen($line);
 316
 317         while ($pos < $j) {
 318             if ($line[$pos] == ' ') {
 319                 $pos++;
 320             } else if (strpos($line, '&gt;', $pos) === $pos) {
 321                 $pos += 4;
 322                 $quotes++;
 323             } else {
 324                 break;
 325             }
 326         }
 327
 328         if ($quotes % 2) {
 329             $line = '<span class="quote1">' . $line . '</span>';
 330         } elseif ($quotes) {
 331             $line = '<span class="quote2">' . $line . '</span>';
 332         }
 333
 334         $body_ary[$i] = $line;
 335     }
 336     $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
 337 }
 338
 339 /**
 340  * This returns a parsed string called $body. That string can then
 341  * be displayed as the actual message in the HTML. It contains
 342  * everything needed, including HTML Tags, Attachments at the
 343  * bottom, etc.
 344  *
 345  * Since 1.2.0 function uses message_body hook.
 346  * Till 1.3.0 function included output of formatAttachments().
 347  *
 348  * @param resource $imap_stream imap connection resource
 349  * @param object $message squirrelmail message object
 350  * @param array $color squirrelmail color theme array
 351  * @param integer $wrap_at number of characters per line
 352  * @param string $ent_num (since 1.3.0) message part id
 353  * @param integer $id (since 1.3.0) message id
 354  * @param string $mailbox (since 1.3.0) imap folder name
 355  * @return string html formated message text
 356  */
 357 function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') {
 358     /* This if statement checks for the entity to show as the
 359      * primary message. To add more of them, just put them in the
 360      * order that is their priority.
 361      */
 362     global $startMessage, $languages, $squirrelmail_language,
 363            $show_html_default, $sort, $has_unsafe_images, $passed_ent_id,
 364            $use_iframe, $iframe_height, $download_and_unsafe_link,
 365            $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text,
 366            $oTemplate, $nbsp;
 367
 368     // workaround for not updated config.php
 369     if (! isset($use_iframe)) $use_iframe = false;
 370
 371     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
 372     // images off by default.
 373     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
 374
 375     $body = '';
 376     $urlmailbox = urlencode($mailbox);
 377     $body_message = getEntity($message, $ent_num);
 378     if (($body_message->header->type0 == 'text') ||
 379             ($body_message->header->type0 == 'rfc822')) {
 380         $body = mime_fetch_body ($imap_stream, $id, $ent_num);
 381         $body = decodeBody($body, $body_message->header->encoding);
 382
 383         if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 384                 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) {
 385             if (mb_detect_encoding($body) != 'ASCII') {
 386                 $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body);
 387             }
 388         }
 389
 390         /* As of 1.5.2, $body is passed (and modified) by reference */
 391         do_hook('message_body', $body);
 392
 393         /* If there are other types that shouldn't be formatted, add
 394          * them here.
 395          */
 396
 397         if ($body_message->header->type1 == 'html') {
 398             if ($show_html_default <> 1) {
 399                 $entity_conv = array('&nbsp;' => ' ',
 400                                      '<p>'    => "\n",
 401                                      '<P>'    => "\n",
 402                                      '<br>'   => "\n",
 403                                      '<BR>'   => "\n",
 404                                      '<br />' => "\n",
 405                                      '<BR />' => "\n",
 406                                      '&gt;'   => '>',
 407                                      '&lt;'   => '<');
 408                 $body = strtr($body, $entity_conv);
 409                 $body = strip_tags($body);
 410                 $body = trim($body);
 411                 translateText($body, $wrap_at,
 412                         $body_message->header->getParameter('charset'));
 413             } elseif ($use_iframe) {
 414                 /**
 415                  * If we don't add html message between iframe tags,
 416                  * we must detect unsafe images and modify $has_unsafe_images.
 417                  */
 418                 $html_body = magicHTML($body, $id, $message, $mailbox);
 419                 // Convert character set in order to display html mails in different character set
 420                 $html_body = charset_decode($body_message->header->getParameter('charset'),$html_body,false,true);
 421
 422                 // creating iframe url
 423                 $iframeurl=sqm_baseuri().'src/view_html.php?'
 424                     . 'mailbox=' . $urlmailbox
 425                     . '&amp;passed_id=' . $id
 426                     . '&amp;ent_id=' . $ent_num
 427                     . '&amp;view_unsafe_images=' . (int) $view_unsafe_images;
 428
 429                 global $oTemplate;
 430                 $oTemplate->assign('iframe_url', $iframeurl);
 431                 $oTemplate->assign('iframe_height', $iframe_height);
 432                 $oTemplate->assign('html_body', $html_body);
 433
 434                 $body = $oTemplate->fetch('read_html_iframe.tpl');
 435             } else {
 436                 // old way of html rendering
 437                 /**
 438                  * convert character set. charset_decode does not remove html special chars
 439                  * applied by magicHTML functions and does not sanitize them second time if
 440                  * fourth argument is true.
 441                  */
 442                 $charset = $body_message->header->getParameter('charset');
 443                 if (!empty($charset)) {
 444                     $body = charset_decode($charset,$body,false,true);
 445                 }
 446                 $body = magicHTML($body, $id, $message, $mailbox);
 447             }
 448         } else {
 449             translateText($body, $wrap_at,
 450                     $body_message->header->getParameter('charset'));
 451         }
 452
 453         /*
 454          * Previously the links for downloading and unsafe images were printed
 455          * under the mail. By putting the links in a global variable we can
 456          * print it in the toolbar where it belongs. Since the original code was
 457          * in this place it's left here. It might be possible to move it to some
 458          * other place if that makes sense. The possibility to do so has not
 459          * been evaluated yet.
 460          */
 461
 462         // Initialize the global variable to an empty string.
 463         // FIXME: To have $download_and_unsafe_link as a global variable might not be needed since the use of separate variables ($download_href, $unsafe_image_toggle_href, and $unsafe_image_toggle_text) for the templates was introduced.
 464         $download_and_unsafe_link = '';
 465
 466         // Prepare and build a link for downloading the mail.
 467         $link = 'passed_id=' . $id . '&amp;ent_id='.$ent_num.
 468             '&amp;mailbox=' . $urlmailbox .'&amp;sort=' . $sort .
 469             '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
 470         if (isset($passed_ent_id)) {
 471             $link .= '&amp;passed_ent_id='.$passed_ent_id;
 472         }
 473         $download_href = SM_PATH . 'src/download.php?absolute_dl=true&amp;' . $link;
 474
 475         // Always add the link for downloading the mail as a file to the global
 476         // variable.
 477         $download_and_unsafe_link .= "$nbsp|$nbsp"
 478             . create_hyperlink($download_href, _("Download this as a file"));
 479
 480         // Find out the right text to use in the link depending on the
 481         // circumstances. If the unsafe images are displayed the link should
 482         // hide them, if they aren't displayed the link should only appear if
 483         // the mail really contains unsafe images.
 484         if ($view_unsafe_images) {
 485             $text = _("Hide Unsafe Images");
 486         } else {
 487             if (isset($has_unsafe_images) && $has_unsafe_images) {
 488                 $link .= '&amp;view_unsafe_images=1';
 489                 $text = _("View Unsafe Images");
 490             } else {
 491                 $text = '';
 492             }
 493         }
 494
 495         // Only create a link for unsafe images if there's need for one. If so:
 496         // add it to the global variable.
 497         if($text != '') {
 498             $unsafe_image_toggle_href = SM_PATH . 'src/read_body.php?'.$link;
 499             $unsafe_image_toggle_text = $text;
 500             $download_and_unsafe_link .= "$nbsp|$nbsp"
 501                 . create_hyperlink($unsafe_image_toggle_href, $text);
 502         }
 503     }
 504     return $body;
 505 }
 506
 507 /**
 508  * Generate attachments array for passing to templates.
 509  *
 510  * @since 1.5.2
 511  * @param object $message SquirrelMail message object
 512  * @param array $exclude_id message parts that are not attachments.
 513  * @param string $mailbox mailbox name
 514  * @param integer $id message id
 515  */
 516 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
 517     global $where, $what, $startMessage, $color, $passed_ent_id, $base_uri;
 518
 519     $att_ar = $message->getAttachments($exclude_id);
 520     $urlMailbox = urlencode($mailbox);
 521
 522     $attachments = array();
 523     foreach ($att_ar as $att) {
 524         $ent = $att->entity_id;
 525         $header = $att->header;
 526         $type0 = strtolower($header->type0);
 527         $type1 = strtolower($header->type1);
 528         $name = '';
 529         $links = array();
 530         $links['download link']['text'] = _("Download");
 531         $links['download link']['href'] = $base_uri .
 532             "src/download.php?absolute_dl=true&amp;passed_id=$id&amp;mailbox=$urlMailbox&amp;ent_id=$ent";
 533
 534         if ($type0 =='message' && $type1 == 'rfc822') {
 535             $default_page = $base_uri  . 'src/read_body.php';
 536             $rfc822_header = $att->rfc822_header;
 537             $filename = $rfc822_header->subject;
 538             if (trim( $filename ) == '') {
 539                 $filename = 'untitled-[' . $ent . ']' ;
 540             }
 541             $from_o = $rfc822_header->from;
 542             if (is_object($from_o)) {
 543                 $from_name = decodeHeader($from_o->getAddress(false));
 544             } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) {
 545                 // something weird happens when a digest message is opened and you return to the digest
 546                 // now the from object is part of an array. Probably the parseHeader call overwrites the info
 547                 // retrieved from the bodystructure in a different way. We need to fix this later.
 548                 // possible starting point, do not fetch header we already have and inspect how
 549                 // the rfc822_header object behaves.
 550                 $from_name = decodeHeader($from_o[0]->getAddress(false));
 551             } else {
 552                 $from_name = _("Unknown sender");
 553             }
 554             $description = _("From").': '.$from_name;
 555         } else {
 556             $default_page = $base_uri  . 'src/download.php';
 557             $filename = $att->getFilename();
 558             if ($header->description) {
 559                 $description = decodeHeader($header->description);
 560             } else {
 561                 $description = '';
 562             }
 563         }
 564
 565         $display_filename = $filename;
 566         if (isset($passed_ent_id)) {
 567             $passed_ent_id_link = '&amp;passed_ent_id='.$passed_ent_id;
 568         } else {
 569             $passed_ent_id_link = '';
 570         }
 571         $defaultlink = $default_page . "?startMessage=$startMessage"
 572             . "&amp;passed_id=$id&amp;mailbox=$urlMailbox"
 573             . '&amp;ent_id='.$ent.$passed_ent_id_link;
 574         if ($where && $what) {
 575             $defaultlink .= '&amp;where='. urlencode($where).'&amp;what='.urlencode($what);
 576         }
 577         // IE does make use of mime content sniffing. Forcing a download
 578         // prohibit execution of XSS inside an application/octet-stream attachment
 579         if ($type0 == 'application' && $type1 == 'octet-stream') {
 580             $defaultlink .= '&amp;absolute_dl=true';
 581         }
 582
 583         /* This executes the attachment hook with a specific MIME-type.
 584          * It also allows plugins to run if there's a rule for a more
 585          * generic type. Finally, a hook for ALL attachment types is
 586          * run as well.
 587          */
 588         // First remember the default link.
 589         $defaultlink_orig = $defaultlink;
 590
 591         /* The API for this hook has changed as of 1.5.2 so that all plugin
 592            arguments are passed in an array instead of each their own plugin
 593            argument, and arguments are passed by reference, so instead of
 594            returning any changes, changes should simply be made to the original
 595            arguments themselves. */
 596         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 597                     &$defaultlink, &$display_filename, &$where, &$what);
 598         do_hook("attachment $type0/$type1", $temp);
 599         /* The API for this hook has changed as of 1.5.2 so that all plugin
 600            arguments are passed in an array instead of each their own plugin
 601            argument, and arguments are passed by reference, so instead of
 602            returning any changes, changes should simply be made to the original
 603            arguments themselves. */
 604         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 605                       &$defaultlink, &$display_filename, &$where, &$what);
 606         // Do not let a generic plugin change the default link if a more
 607         // specialized one already did it...
 608         if ($defaultlink != $defaultlink_orig) {
 609             $dummy = '';
 610             $temp[5] = &$dummy;
 611         }
 612         do_hook("attachment $type0/*", $temp);
 613         /* The API for this hook has changed as of 1.5.2 so that all plugin
 614            arguments are passed in an array instead of each their own plugin
 615            argument, and arguments are passed by reference, so instead of
 616            returning any changes, changes should simply be made to the original
 617            arguments themselves. */
 618         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 619                       &$defaultlink, &$display_filename, &$where, &$what);
 620         // Do not let a generic plugin change the default link if a more
 621         // specialized one already did it...
 622         if ($defaultlink != $defaultlink_orig) {
 623             $dummy = '';
 624             $temp[5] = &$dummy;
 625         }
 626         do_hook("attachment */*", $temp);
 627
 628         $this_attachment = array();
 629         $this_attachment['Name'] = decodeHeader($display_filename);
 630         $this_attachment['Description'] = $description;
 631         $this_attachment['DefaultHREF'] = $defaultlink;
 632         $this_attachment['DownloadHREF'] = $links['download link']['href'];
 633         $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
 634         $this_attachment['Size'] = $header->size;
 635         $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
 636         $this_attachment['OtherLinks'] = array();
 637         foreach ($links as $val) {
 638             if ($val['text']==_("Download") || $val['text'] == _("View"))
 639                 continue;
 640             if (empty($val['text']) && empty($val['extra']))
 641                 continue;
 642
 643             $temp = array();
 644             $temp['HREF'] = $val['href'];
 645             $temp['Text'] = (empty($val['text']) ? '' : $val['text']) . (empty($val['extra']) ? '' : $val['extra']);
 646             $this_attachment['OtherLinks'][] = $temp;
 647         }
 648         $attachments[] = $this_attachment;
 649
 650         unset($links);
 651     }
 652
 653     return $attachments;
 654 }
 655
 656 /**
 657  * Displays attachment links and information
 658  *
 659  * Since 1.3.0 function is not included in formatBody() call.
 660  *
 661  * Since 1.0.2 uses attachment $type0/$type1 hook.
 662  * Since 1.2.5 uses attachment $type0/* hook.
 663  * Since 1.5.0 uses attachments_bottom hook.
 664  * Since 1.5.2 uses templates and does *not* return a value.
 665  *
 666  * @param object $message SquirrelMail message object
 667  * @param array $exclude_id message parts that are not attachments.
 668  * @param string $mailbox mailbox name
 669  * @param integer $id message id
 670  */
 671 function formatAttachments($message, $exclude_id, $mailbox, $id) {
 672     global $oTemplate;
 673
 674     $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
 675
 676     $oTemplate->assign('attachments', $attach);
 677     $oTemplate->display('read_attachments.tpl');
 678 }
 679
 680 function sqimap_base64_decode(&$string) {
 681
 682     // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
 683     // fly decoding (to reduce memory usage) you have to check if the
 684     // data has incomplete pairs
 685
 686     // Remove the noise in order to check if the 4 bytes pairs are complete
 687     $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
 688
 689     $sStringRem = '';
 690     $iMod = strlen($string) % 4;
 691     if ($iMod) {
 692         $sStringRem = substr($string,-$iMod);
 693         // Check if $sStringRem contains padding characters
 694         if (substr($sStringRem,-1) != '=') {
 695             $string = substr($string,0,-$iMod);
 696         } else {
 697             $sStringRem = '';
 698         }
 699     }
 700     $string = base64_decode($string);
 701     return $sStringRem;
 702 }
 703
 704 /**
 705  * Decodes encoded string (usually message body)
 706  *
 707  * This function decodes a string (usually the message body)
 708  * depending on the encoding type.  Currently quoted-printable
 709  * and base64 encodings are supported.
 710  *
 711  * The decode_body hook was added to this function in 1.4.2/1.5.0.
 712  * The $force_crlf parameter was added in 1.5.2.
 713  *
 714  * @param string $string     The encoded string
 715  * @param string $encoding   used encoding
 716  * @param string $force_crlf Whether or not to force CRLF or LF
 717  *                           line endings (or to leave as is).
 718  *                           If given as "LF", line endings will
 719  *                           all be converted to LF; if "CRLF",
 720  *                           line endings will all be converted
 721  *                           to CRLF.  If given as an empty value,
 722  *                           the global $force_crlf_default will
 723  *                           be consulted (it can be specified in
 724  *                           config/config_local.php).  Otherwise,
 725  *                           any other value will cause the string
 726  *                           to be left alone.  Note that this will
 727  *                           be overridden to "LF" if not using at
 728  *                           least PHP version 4.3.0. (OPTIONAL;
 729  *                           default is empty - consult global
 730  *                           default value)
 731  *
 732  * @return string The decoded string
 733  *
 734  * @since 1.0
 735  *
 736  */
 737 function decodeBody($string, $encoding, $force_crlf='') {
 738
 739     global $force_crlf_default;
 740     if (empty($force_crlf)) $force_crlf = $force_crlf_default;
 741     $force_crlf = strtoupper($force_crlf);
 742
 743     // must force line endings to LF due to broken
 744     // quoted_printable_decode() in PHP versions
 745     // before 4.3.0 (see below)
 746     //
 747     if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
 748         $string = str_replace("\r\n", "\n", $string);
 749     else if ($force_crlf == 'CRLF')
 750         $string = str_replace("\n", "\r\n", $string);
 751
 752     $encoding = strtolower($encoding);
 753
 754     $encoding_handler = do_hook('decode_body', $encoding);
 755
 756
 757     // plugins get first shot at decoding the string
 758     //
 759     if (!empty($encoding_handler) && function_exists($encoding_handler)) {
 760         $string = $encoding_handler('decode', $string);
 761
 762     } elseif ($encoding == 'quoted-printable' ||
 763             $encoding == 'quoted_printable') {
 764
 765         // quoted_printable_decode() function is broken in older
 766         // php versions.  Text with \r\n decoding was fixed only
 767         // in php 4.3.0.  Minimal code requirement is PHP 4.0.4+
 768         // and the above call to:  str_replace("\r\n", "\n", $string);
 769         //
 770         $string = quoted_printable_decode($string);
 771
 772     } elseif ($encoding == 'base64') {
 773         $string = base64_decode($string);
 774     }
 775
 776     // All other encodings are returned raw.
 777     return $string;
 778 }
 779
 780 /**
 781  * Decodes headers
 782  *
 783  * This function decodes strings that are encoded according to
 784  * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
 785  * Patched by Christian Schmidt <christian@ostenfeld.dk>  23/03/2002
 786  *
 787  * @param string $string header string that has to be made readable
 788  * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
 789  * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
 790  * @param boolean $decide decide if string can be utfencoded. defaults to false
 791  * @return string decoded header string
 792  */
 793 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
 794     global $languages, $squirrelmail_language,$default_charset;
 795     if (is_array($string)) {
 796         $string = implode("\n", $string);
 797     }
 798
 799     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 800             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
 801         $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
 802         // Do we need to return at this point?
 803         // return $string;
 804     }
 805     $i = 0;
 806     $iLastMatch = -2;
 807     $encoded = true;
 808
 809 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
 810     $aString = explode(' ',$string);
 811     $ret = '';
 812     foreach ($aString as $chunk) {
 813         if ($encoded && $chunk === '') {
 814             continue;
 815         } elseif ($chunk === '') {
 816             $ret .= ' ';
 817             continue;
 818         }
 819         $encoded = false;
 820         /* if encoded words are not separated by a linear-space-white we still catch them */
 821         $j = $i-1;
 822
 823         while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
 824             /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
 825             if ($iLastMatch !== $j) {
 826                 if ($htmlsafe) {
 827                     $ret .= '&#32;';
 828                 } else {
 829                     $ret .= ' ';
 830                 }
 831             }
 832             $iLastMatch = $i;
 833             $j = $i;
 834             if ($htmlsafe) {
 835                 $ret .= sm_encode_html_special_chars($res[1]);
 836             } else {
 837                 $ret .= $res[1];
 838             }
 839             $encoding = ucfirst($res[3]);
 840
 841             /* decide about valid decoding */
 842             if ($decide && is_conversion_safe($res[2])) {
 843                 $utfencode=true;
 844                 $can_be_encoded=true;
 845             } else {
 846                 $can_be_encoded=false;
 847             }
 848             switch ($encoding)
 849             {
 850                 case 'B':
 851                     $replace = base64_decode($res[4]);
 852                     if ($utfencode) {
 853                         if ($can_be_encoded) {
 854                             /* convert string to different charset,
 855                              * if functions asks for it (usually in compose)
 856                              */
 857                             $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
 858                         } else {
 859                             // convert string to html codes in order to display it
 860                             $ret .= charset_decode($res[2],$replace);
 861                         }
 862                     } else {
 863                         if ($htmlsafe) {
 864                             $replace = sm_encode_html_special_chars($replace);
 865                         }
 866                         $ret.= $replace;
 867                     }
 868                     break;
 869                 case 'Q':
 870                     $replace = str_replace('_', ' ', $res[4]);
 871                     $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
 872                             create_function ('$matches', 'return chr(hexdec($matches[1]));'),
 873                             $replace);
 874                     if ($utfencode) {
 875                         if ($can_be_encoded) {
 876                             /* convert string to different charset,
 877                              * if functions asks for it (usually in compose)
 878                              */
 879                             $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
 880                         } else {
 881                             // convert string to html codes in order to display it
 882                             $replace = charset_decode($res[2], $replace);
 883                         }
 884                     } else {
 885                         if ($htmlsafe) {
 886                             $replace = sm_encode_html_special_chars($replace);
 887                         }
 888                     }
 889                     $ret .= $replace;
 890                     break;
 891                 default:
 892                     break;
 893             }
 894             $chunk = $res[5];
 895             $encoded = true;
 896         }
 897         if (!$encoded) {
 898             if ($htmlsafe) {
 899                 $ret .= '&#32;';
 900             } else {
 901                 $ret .= ' ';
 902             }
 903         }
 904
 905         if (!$encoded && $htmlsafe) {
 906             $ret .= sm_encode_html_special_chars($chunk);
 907         } else {
 908             $ret .= $chunk;
 909         }
 910         ++$i;
 911     }
 912     /* remove the first added space */
 913     if ($ret) {
 914         if ($htmlsafe) {
 915             $ret = substr($ret,5);
 916         } else {
 917             $ret = substr($ret,1);
 918         }
 919     }
 920
 921     return $ret;
 922 }
 923
 924 /**
 925  * Encodes header
 926  *
 927  * Function uses XTRA_CODE _encodeheader function, if such function exists.
 928  *
 929  * Function uses Q encoding by default and encodes a string according to RFC
 930  * 1522 for use in headers if it contains 8-bit characters or anything that
 931  * looks like it should be encoded.
 932  *
 933  * Function switches to B encoding and encodeHeaderBase64() function, if
 934  * string is 8bit and multibyte character set supported by mbstring extension
 935  * is used. It can cause E_USER_NOTICE errors, if interface is used with
 936  * multibyte character set unsupported by mbstring extension.
 937  *
 938  * @param string $string header string, that has to be encoded
 939  * @return string quoted-printable encoded string
 940  * @todo make $mb_charsets system wide constant
 941  */
 942 function encodeHeader ($string) {
 943     global $default_charset, $languages, $squirrelmail_language;
 944
 945     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 946             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
 947         return  call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
 948     }
 949
 950     // Use B encoding for multibyte charsets
 951     $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
 952     if (in_array($default_charset,$mb_charsets) &&
 953         in_array($default_charset,sq_mb_list_encodings()) &&
 954         sq_is8bit($string)) {
 955         return encodeHeaderBase64($string,$default_charset);
 956     } elseif (in_array($default_charset,$mb_charsets) &&
 957               sq_is8bit($string) &&
 958               ! in_array($default_charset,sq_mb_list_encodings())) {
 959         // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
 960         // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
 961     }
 962
 963     // Encode only if the string contains 8-bit characters or =?
 964     $j = strlen($string);
 965     $max_l = 75 - strlen($default_charset) - 7;
 966     $aRet = array();
 967     $ret = '';
 968     $iEncStart = $enc_init = false;
 969     $cur_l = $iOffset = 0;
 970     for($i = 0; $i < $j; ++$i) {
 971         switch($string{$i})
 972         {
 973             case '"':
 974             case '=':
 975             case '<':
 976             case '>':
 977             case ',':
 978             case '?':
 979             case '_':
 980                 if ($iEncStart === false) {
 981                     $iEncStart = $i;
 982                 }
 983                 $cur_l+=3;
 984                 if ($cur_l > ($max_l-2)) {
 985                     /* if there is an stringpart that doesn't need encoding, add it */
 986                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
 987                     $aRet[] = "=?$default_charset?Q?$ret?=";
 988                     $iOffset = $i;
 989                     $cur_l = 0;
 990                     $ret = '';
 991                     $iEncStart = false;
 992                 } else {
 993                     $ret .= sprintf("=%02X",ord($string{$i}));
 994                 }
 995                 break;
 996             case '(':
 997             case ')':
 998                 if ($iEncStart !== false) {
 999                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1000                     $aRet[] = "=?$default_charset?Q?$ret?=";
1001                     $iOffset = $i;
1002                     $cur_l = 0;
1003                     $ret = '';
1004                     $iEncStart = false;
1005                 }
1006                 break;
1007             case ' ':
1008                 if ($iEncStart !== false) {
1009                     $cur_l++;
1010                     if ($cur_l > $max_l) {
1011                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1012                         $aRet[] = "=?$default_charset?Q?$ret?=";
1013                         $iOffset = $i;
1014                         $cur_l = 0;
1015                         $ret = '';
1016                         $iEncStart = false;
1017                     } else {
1018                         $ret .= '_';
1019                     }
1020                 }
1021                 break;
1022             default:
1023                 $k = ord($string{$i});
1024                 if ($k > 126) {
1025                     if ($iEncStart === false) {
1026                         // do not start encoding in the middle of a string, also take the rest of the word.
1027                         $sLeadString = substr($string,0,$i);
1028                         $aLeadString = explode(' ',$sLeadString);
1029                         $sToBeEncoded = array_pop($aLeadString);
1030                         $iEncStart = $i - strlen($sToBeEncoded);
1031                         $ret .= $sToBeEncoded;
1032                         $cur_l += strlen($sToBeEncoded);
1033                     }
1034                     $cur_l += 3;
1035                     /* first we add the encoded string that reached it's max size */
1036                     if ($cur_l > ($max_l-2)) {
1037                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1038                         $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1039                         $cur_l = 3;
1040                         $ret = '';
1041                         $iOffset = $i;
1042                         $iEncStart = $i;
1043                     }
1044                     $enc_init = true;
1045                     $ret .= sprintf("=%02X", $k);
1046                 } else {
1047                     if ($iEncStart !== false) {
1048                         $cur_l++;
1049                         if ($cur_l > $max_l) {
1050                             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1051                             $aRet[] = "=?$default_charset?Q?$ret?=";
1052                             $iEncStart = false;
1053                             $iOffset = $i;
1054                             $cur_l = 0;
1055                             $ret = '';
1056                         } else {
1057                             $ret .= $string{$i};
1058                         }
1059                     }
1060                 }
1061                 break;
1062         }
1063     }
1064
1065     if ($enc_init) {
1066         if ($iEncStart !== false) {
1067             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1068             $aRet[] = "=?$default_charset?Q?$ret?=";
1069         } else {
1070             $aRet[] = substr($string,$iOffset);
1071         }
1072         $string = implode('',$aRet);
1073     }
1074     return $string;
1075 }
1076
1077 /**
1078  * Encodes string according to rfc2047 B encoding header formating rules
1079  *
1080  * It is recommended way to encode headers with character sets that store
1081  * symbols in more than one byte.
1082  *
1083  * Function requires mbstring support. If required mbstring functions are missing,
1084  * function returns false and sets E_USER_WARNING level error message.
1085  *
1086  * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1087  * that mbstring functions will generate E_WARNING errors, if unsupported
1088  * character set is used. mb_encode_mimeheader function provided by php
1089  * mbstring extension is not used in order to get better control of header
1090  * encoding.
1091  *
1092  * Used php code functions - function_exists(), trigger_error(), strlen()
1093  * (is used with charset names and base64 strings). Used php mbstring
1094  * functions - mb_strlen and mb_substr.
1095  *
1096  * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1097  * encoding), rfc 2822 (header folding)
1098  *
1099  * @param string $string header string that must be encoded
1100  * @param string $charset character set. Must be supported by mbstring extension.
1101  * Use sq_mb_list_encodings() to detect supported charsets.
1102  * @return string string encoded according to rfc2047 B encoding formating rules
1103  * @since 1.5.1
1104  * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1105  * @todo Do we want to control max length of header?
1106  * @todo Do we want to control EOL (end-of-line) marker?
1107  * @todo Do we want to translate error message?
1108  */
1109 function encodeHeaderBase64($string,$charset) {
1110     /**
1111      * Check mbstring function requirements.
1112      */
1113     if (! function_exists('mb_strlen') ||
1114         ! function_exists('mb_substr')) {
1115         // set E_USER_WARNING
1116         trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1117         // return false
1118         return false;
1119     }
1120
1121     // initial return array
1122     $aRet = array();
1123
1124     /**
1125      * header length = 75 symbols max (same as in encodeHeader)
1126      * remove $charset length
1127      * remove =? ? ?= (5 chars)
1128      * remove 2 more chars (\r\n ?)
1129      */
1130     $iMaxLength = 75 - strlen($charset) - 7;
1131
1132     // set first character position
1133     $iStartCharNum = 0;
1134
1135     // loop through all characters. count characters and not bytes.
1136     for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1137         // encode string from starting character to current character.
1138         $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1139
1140         // Check encoded string length
1141         if(strlen($encoded_string)>$iMaxLength) {
1142             // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1143             $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1144
1145             // set new starting character
1146             $iStartCharNum = $iCharNum-1;
1147
1148             // encode last char (in case it is last character in string)
1149             $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1150         } // if string is shorter than max length - add next character
1151     }
1152
1153     // add last encoded string to array
1154     $aRet[] = $encoded_string;
1155
1156     // set initial return string
1157     $sRet = '';
1158
1159     // loop through encoded strings
1160     foreach($aRet as $string) {
1161         // TODO: Do we want to control EOL (end-of-line) marker
1162         if ($sRet!='') $sRet.= " ";
1163
1164         // add header tags and encoded string to return string
1165         $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1166     }
1167
1168     return $sRet;
1169 }
1170
1171 /* This function trys to locate the entity_id of a specific mime element */
1172 function find_ent_id($id, $message) {
1173     for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1174         if ($message->entities[$i]->header->type0 == 'multipart')  {
1175             $ret = find_ent_id($id, $message->entities[$i]);
1176         } else {
1177             if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1178 //                if (sq_check_save_extension($message->entities[$i])) {
1179                 return $message->entities[$i]->entity_id;
1180 //                }
1181             } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1182                 /**
1183                  * This is part of a fix for Outlook Express 6.x generating
1184                  * cid URLs without creating content-id headers
1185                  * @@JA - 20050207
1186                  */
1187                 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1188                     return $message->entities[$i]->entity_id;
1189                 }
1190             }
1191         }
1192     }
1193     return $ret;
1194 }
1195
1196 function sq_check_save_extension($message) {
1197     $filename = $message->getFilename();
1198     $ext = substr($filename, strrpos($filename,'.')+1);
1199     $save_extensions = array('jpg','jpeg','gif','png','bmp');
1200     return in_array($ext, $save_extensions);
1201 }
1202
1203
1204 /**
1205  ** HTMLFILTER ROUTINES
1206  */
1207
1208 /**
1209  * This function checks attribute values for entity-encoded values
1210  * and returns them translated into 8-bit strings so we can run
1211  * checks on them.
1212  *
1213  * @param  $attvalue A string to run entity check against.
1214  * @return           Nothing, modifies a reference value.
1215  */
1216 function sq_defang(&$attvalue){
1217     $me = 'sq_defang';
1218     /**
1219      * Skip this if there aren't ampersands or backslashes.
1220      */
1221     if (strpos($attvalue, '&') === false
1222         && strpos($attvalue, '\\') === false){
1223         return;
1224     }
1225     $m = false;
1226     // before deent, translate the dangerous unicode characters and ... to safe values
1227     // otherwise the regular expressions do not match.
1228
1229
1230
1231     do {
1232         $m = false;
1233         $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1234         $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1235         $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1236     } while ($m == true);
1237     $attvalue = stripslashes($attvalue);
1238 }
1239
1240 /**
1241  * Kill any tabs, newlines, or carriage returns. Our friends the
1242  * makers of the browser with 95% market value decided that it'd
1243  * be funny to make "java[tab]script" be just as good as "javascript".
1244  *
1245  * @param  attvalue  The attribute value before extraneous spaces removed.
1246  * @return attvalue  Nothing, modifies a reference value.
1247  */
1248 function sq_unspace(&$attvalue){
1249     $me = 'sq_unspace';
1250     if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1251         $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1252                                 Array('',   '',   '',   '',   ''), $attvalue);
1253     }
1254 }
1255
1256 /**
1257  * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1258  * IE as regular characters.
1259  *
1260  * @param  attvalue  The attribute value before dangerous characters are translated.
1261  * @return attvalue  Nothing, modifies a reference value.
1262  * @author Marc Groot Koerkamp.
1263  */
1264 function sq_fixIE_idiocy(&$attvalue) {
1265     // remove NUL
1266     $attvalue = str_replace("\0", "", $attvalue);
1267     // remove comments
1268     $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1269
1270     // IE has the evil habit of accepting every possible value for the attribute expression.
1271     // The table below contains characters which are parsed by IE if they are used in the "expression"
1272     // attribute value.
1273     $aDangerousCharsReplacementTable = array(
1274                         array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1275                               '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1276                               '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1277                               '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1278                               '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1279                               '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1280                               '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1281                               '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1282                               '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1283                               '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1284                               '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1285                               '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1286                               '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1287                               '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1288                               '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1289                               '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1290                               '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1291                               '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1292                               '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1293                               '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1294                               '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1295                               '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1296                               '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1297                               '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1298                               "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */   // in unicode this is some Chinese char range
1299                               "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1300                               "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1301                               "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1302                               "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1303                               "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1304                               "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1305                               "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1306                               "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1307                               "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1308                               "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1309                               "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1310                               "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1311                               "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1312                               "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1313                               "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1314                               "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1315                               "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1316                               "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1317                               "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1318                               "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1319                               "\xCA\x9F", /* L UNICODE IPA Extension */
1320                               "\xCA\x80", /* R UNICODE IPA Extension */
1321                               "\xC9\xB4"),  /* N UNICODE IPA Extension */
1322                        array('l', 'l', 'r','r','n','n',
1323                              'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1324                              'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1325                              'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1326     $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1327
1328     // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1329     // used for XSS.
1330     $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1331 }
1332
1333 /**
1334  * This function returns the final tag out of the tag name, an array
1335  * of attributes, and the type of the tag. This function is called by
1336  * sq_sanitize internally.
1337  *
1338  * @param  $tagname  the name of the tag.
1339  * @param  $attary   the array of attributes and their values
1340  * @param  $tagtype  The type of the tag (see in comments).
1341  * @return           a string with the final tag representation.
1342  */
1343 function sq_tagprint($tagname, $attary, $tagtype){
1344     $me = 'sq_tagprint';
1345
1346     if ($tagtype == 2){
1347         $fulltag = '</' . $tagname . '>';
1348     } else {
1349         $fulltag = '<' . $tagname;
1350         if (is_array($attary) && sizeof($attary)){
1351             $atts = Array();
1352             while (list($attname, $attvalue) = each($attary)){
1353                 array_push($atts, "$attname=$attvalue");
1354             }
1355             $fulltag .= ' ' . join(" ", $atts);
1356         }
1357         if ($tagtype == 3){
1358             $fulltag .= ' /';
1359         }
1360         $fulltag .= '>';
1361     }
1362     return $fulltag;
1363 }
1364
1365 /**
1366  * A small helper function to use with array_walk. Modifies a by-ref
1367  * value and makes it lowercase.
1368  *
1369  * @param  $val a value passed by-ref.
1370  * @return      void since it modifies a by-ref value.
1371  */
1372 function sq_casenormalize(&$val){
1373     $val = strtolower($val);
1374 }
1375
1376 /**
1377  * This function skips any whitespace from the current position within
1378  * a string and to the next non-whitespace value.
1379  *
1380  * @param  $body   the string
1381  * @param  $offset the offset within the string where we should start
1382  *                 looking for the next non-whitespace character.
1383  * @return         the location within the $body where the next
1384  *                 non-whitespace char is located.
1385  */
1386 function sq_skipspace($body, $offset){
1387     $me = 'sq_skipspace';
1388     preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1389     if (sizeof($matches{1})){
1390         $count = strlen($matches{1});
1391         $offset += $count;
1392     }
1393     return $offset;
1394 }
1395
1396 /**
1397  * This function looks for the next character within a string.  It's
1398  * really just a glorified "strpos", except it catches if failures
1399  * nicely.
1400  *
1401  * @param  $body   The string to look for needle in.
1402  * @param  $offset Start looking from this position.
1403  * @param  $needle The character/string to look for.
1404  * @return         location of the next occurance of the needle, or
1405  *                 strlen($body) if needle wasn't found.
1406  */
1407 function sq_findnxstr($body, $offset, $needle){
1408     $me  = 'sq_findnxstr';
1409     $pos = strpos($body, $needle, $offset);
1410     if ($pos === FALSE){
1411         $pos = strlen($body);
1412     }
1413     return $pos;
1414 }
1415
1416 /**
1417  * This function takes a PCRE-style regexp and tries to match it
1418  * within the string.
1419  *
1420  * @param  $body   The string to look for needle in.
1421  * @param  $offset Start looking from here.
1422  * @param  $reg    A PCRE-style regex to match.
1423  * @return         Returns a false if no matches found, or an array
1424  *                 with the following members:
1425  *                 - integer with the location of the match within $body
1426  *                 - string with whatever content between offset and the match
1427  *                 - string with whatever it is we matched
1428  */
1429 function sq_findnxreg($body, $offset, $reg){
1430     $me = 'sq_findnxreg';
1431     $matches = Array();
1432     $retarr = Array();
1433     preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1434     if (!isset($matches{0}) || !$matches{0}){
1435         $retarr = false;
1436     } else {
1437         $retarr{0} = $offset + strlen($matches{1});
1438         $retarr{1} = $matches{1};
1439         $retarr{2} = $matches{2};
1440     }
1441     return $retarr;
1442 }
1443
1444 /**
1445  * This function looks for the next tag.
1446  *
1447  * @param  $body   String where to look for the next tag.
1448  * @param  $offset Start looking from here.
1449  * @return         false if no more tags exist in the body, or
1450  *                 an array with the following members:
1451  *                 - string with the name of the tag
1452  *                 - array with attributes and their values
1453  *                 - integer with tag type (1, 2, or 3)
1454  *                 - integer where the tag starts (starting "<")
1455  *                 - integer where the tag ends (ending ">")
1456  *                 first three members will be false, if the tag is invalid.
1457  */
1458 function sq_getnxtag($body, $offset){
1459     $me = 'sq_getnxtag';
1460     if ($offset > strlen($body)){
1461         return false;
1462     }
1463     $lt = sq_findnxstr($body, $offset, "<");
1464     if ($lt == strlen($body)){
1465         return false;
1466     }
1467     /**
1468      * We are here:
1469      * blah blah <tag attribute="value">
1470      * \---------^
1471      */
1472     $pos = sq_skipspace($body, $lt+1);
1473     if ($pos >= strlen($body)){
1474         return Array(false, false, false, $lt, strlen($body));
1475     }
1476     /**
1477      * There are 3 kinds of tags:
1478      * 1. Opening tag, e.g.:
1479      *    <a href="blah">
1480      * 2. Closing tag, e.g.:
1481      *    </a>
1482      * 3. XHTML-style content-less tag, e.g.:
1483      *    <img src="blah" />
1484      */
1485     $tagtype = false;
1486     switch (substr($body, $pos, 1)){
1487         case '/':
1488             $tagtype = 2;
1489             $pos++;
1490             break;
1491         case '!':
1492             /**
1493              * A comment or an SGML declaration.
1494              */
1495             if (substr($body, $pos+1, 2) == "--"){
1496                 $gt = strpos($body, "-->", $pos);
1497                 if ($gt === false){
1498                     $gt = strlen($body);
1499                 } else {
1500                     $gt += 2;
1501                 }
1502                 return Array(false, false, false, $lt, $gt);
1503             } else {
1504                 $gt = sq_findnxstr($body, $pos, ">");
1505                 return Array(false, false, false, $lt, $gt);
1506             }
1507             break;
1508         default:
1509             /**
1510              * Assume tagtype 1 for now. If it's type 3, we'll switch values
1511              * later.
1512              */
1513             $tagtype = 1;
1514             break;
1515     }
1516
1517     $tag_start = $pos;
1518     $tagname = '';
1519     /**
1520      * Look for next [\W-_], which will indicate the end of the tag name.
1521      */
1522     $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1523     if ($regary == false){
1524         return Array(false, false, false, $lt, strlen($body));
1525     }
1526     list($pos, $tagname, $match) = $regary;
1527     $tagname = strtolower($tagname);
1528
1529     /**
1530      * $match can be either of these:
1531      * '>'  indicating the end of the tag entirely.
1532      * '\s' indicating the end of the tag name.
1533      * '/'  indicating that this is type-3 xhtml tag.
1534      *
1535      * Whatever else we find there indicates an invalid tag.
1536      */
1537     switch ($match){
1538         case '/':
1539             /**
1540              * This is an xhtml-style tag with a closing / at the
1541              * end, like so: <img src="blah" />. Check if it's followed
1542              * by the closing bracket. If not, then this tag is invalid
1543              */
1544             if (substr($body, $pos, 2) == "/>"){
1545                 $pos++;
1546                 $tagtype = 3;
1547             } else {
1548                 $gt = sq_findnxstr($body, $pos, ">");
1549                 $retary = Array(false, false, false, $lt, $gt);
1550                 return $retary;
1551             }
1552         case '>':
1553             return Array($tagname, false, $tagtype, $lt, $pos);
1554             break;
1555         default:
1556             /**
1557              * Check if it's whitespace
1558              */
1559             if (!preg_match('/\s/', $match)){
1560                 /**
1561                  * This is an invalid tag! Look for the next closing ">".
1562                  */
1563                 $gt = sq_findnxstr($body, $lt, ">");
1564                 return Array(false, false, false, $lt, $gt);
1565             }
1566             break;
1567     }
1568
1569     /**
1570      * At this point we're here:
1571      * <tagname  attribute='blah'>
1572      * \-------^
1573      *
1574      * At this point we loop in order to find all attributes.
1575      */
1576     $attname = '';
1577     $atttype = false;
1578     $attary = Array();
1579
1580     while ($pos <= strlen($body)){
1581         $pos = sq_skipspace($body, $pos);
1582         if ($pos == strlen($body)){
1583             /**
1584              * Non-closed tag.
1585              */
1586             return Array(false, false, false, $lt, $pos);
1587         }
1588         /**
1589          * See if we arrived at a ">" or "/>", which means that we reached
1590          * the end of the tag.
1591          */
1592         $matches = Array();
1593         if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1594             /**
1595              * Yep. So we did.
1596              */
1597             $pos += strlen($matches{1});
1598             if ($matches{2} == "/>"){
1599                 $tagtype = 3;
1600                 $pos++;
1601             }
1602             return Array($tagname, $attary, $tagtype, $lt, $pos);
1603         }
1604
1605         /**
1606          * There are several types of attributes, with optional
1607          * [:space:] between members.
1608          * Type 1:
1609          *   attrname[:space:]=[:space:]'CDATA'
1610          * Type 2:
1611          *   attrname[:space:]=[:space:]"CDATA"
1612          * Type 3:
1613          *   attr[:space:]=[:space:]CDATA
1614          * Type 4:
1615          *   attrname
1616          *
1617          * We leave types 1 and 2 the same, type 3 we check for
1618          * '"' and convert to "&quot" if needed, then wrap in
1619          * double quotes. Type 4 we convert into:
1620          * attrname="yes".
1621          */
1622         $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1623         if ($regary == false){
1624             /**
1625              * Looks like body ended before the end of tag.
1626              */
1627             return Array(false, false, false, $lt, strlen($body));
1628         }
1629         list($pos, $attname, $match) = $regary;
1630         $attname = strtolower($attname);
1631         /**
1632          * We arrived at the end of attribute name. Several things possible
1633          * here:
1634          * '>'  means the end of the tag and this is attribute type 4
1635          * '/'  if followed by '>' means the same thing as above
1636          * '\s' means a lot of things -- look what it's followed by.
1637          *      anything else means the attribute is invalid.
1638          */
1639         switch($match){
1640             case '/':
1641                 /**
1642                  * This is an xhtml-style tag with a closing / at the
1643                  * end, like so: <img src="blah" />. Check if it's followed
1644                  * by the closing bracket. If not, then this tag is invalid
1645                  */
1646                 if (substr($body, $pos, 2) == "/>"){
1647                     $pos++;
1648                     $tagtype = 3;
1649                 } else {
1650                     $gt = sq_findnxstr($body, $pos, ">");
1651                     $retary = Array(false, false, false, $lt, $gt);
1652                     return $retary;
1653                 }
1654             case '>':
1655                 $attary{$attname} = '"yes"';
1656                 return Array($tagname, $attary, $tagtype, $lt, $pos);
1657                 break;
1658             default:
1659                 /**
1660                  * Skip whitespace and see what we arrive at.
1661                  */
1662                 $pos = sq_skipspace($body, $pos);
1663                 $char = substr($body, $pos, 1);
1664                 /**
1665                  * Two things are valid here:
1666                  * '=' means this is attribute type 1 2 or 3.
1667                  * \w means this was attribute type 4.
1668                  * anything else we ignore and re-loop. End of tag and
1669                  * invalid stuff will be caught by our checks at the beginning
1670                  * of the loop.
1671                  */
1672                 if ($char == "="){
1673                     $pos++;
1674                     $pos = sq_skipspace($body, $pos);
1675                     /**
1676                      * Here are 3 possibilities:
1677                      * "'"  attribute type 1
1678                      * '"'  attribute type 2
1679                      * everything else is the content of tag type 3
1680                      */
1681                     $quot = substr($body, $pos, 1);
1682                     if ($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} = "'" . $attval . "'";
1690                     } else if ($quot == '"'){
1691                         $regary = sq_findnxreg($body, $pos+1, '\"');
1692                         if ($regary == false){
1693                             return Array(false, false, false, $lt, strlen($body));
1694                         }
1695                         list($pos, $attval, $match) = $regary;
1696                         $pos++;
1697                         $attary{$attname} = '"' . $attval . '"';
1698                     } else {
1699                         /**
1700                          * These are hateful. Look for \s, or >.
1701                          */
1702                         $regary = sq_findnxreg($body, $pos, "[\s>]");
1703                         if ($regary == false){
1704                             return Array(false, false, false, $lt, strlen($body));
1705                         }
1706                         list($pos, $attval, $match) = $regary;
1707                         /**
1708                          * If it's ">" it will be caught at the top.
1709                          */
1710                         $attval = preg_replace("/\"/s", "&quot;", $attval);
1711                         $attary{$attname} = '"' . $attval . '"';
1712                     }
1713                 } else if (preg_match("|[\w/>]|", $char)) {
1714                     /**
1715                      * That was attribute type 4.
1716                      */
1717                     $attary{$attname} = '"yes"';
1718                 } else {
1719                     /**
1720                      * An illegal character. Find next '>' and return.
1721                      */
1722                     $gt = sq_findnxstr($body, $pos, ">");
1723                     return Array(false, false, false, $lt, $gt);
1724                 }
1725                 break;
1726         }
1727     }
1728     /**
1729      * The fact that we got here indicates that the tag end was never
1730      * found. Return invalid tag indication so it gets stripped.
1731      */
1732     return Array(false, false, false, $lt, strlen($body));
1733 }
1734
1735 /**
1736  * Translates entities into literal values so they can be checked.
1737  *
1738  * @param $attvalue the by-ref value to check.
1739  * @param $regex    the regular expression to check against.
1740  * @param $hex      whether the entites are hexadecimal.
1741  * @return          True or False depending on whether there were matches.
1742  */
1743 function sq_deent(&$attvalue, $regex, $hex=false){
1744     $me = 'sq_deent';
1745     $ret_match = false;
1746     // remove comments
1747     //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1748     preg_match_all($regex, $attvalue, $matches);
1749     if (is_array($matches) && sizeof($matches[0]) > 0){
1750         $repl = Array();
1751         for ($i = 0; $i < sizeof($matches[0]); $i++){
1752             $numval = $matches[1][$i];
1753             if ($hex){
1754                 $numval = hexdec($numval);
1755             }
1756             $repl{$matches[0][$i]} = chr($numval);
1757         }
1758         $attvalue = strtr($attvalue, $repl);
1759         return true;
1760     } else {
1761         return false;
1762     }
1763 }
1764
1765 /**
1766  * This function runs various checks against the attributes.
1767  *
1768  * @param  $tagname         String with the name of the tag.
1769  * @param  $attary          Array with all tag attributes.
1770  * @param  $rm_attnames     See description for sq_sanitize
1771  * @param  $bad_attvals     See description for sq_sanitize
1772  * @param  $add_attr_to_tag See description for sq_sanitize
1773  * @param  $message         message object
1774  * @param  $id              message id
1775  * @return                  Array with modified attributes.
1776  */
1777 function sq_fixatts($tagname,
1778                     $attary,
1779                     $rm_attnames,
1780                     $bad_attvals,
1781                     $add_attr_to_tag,
1782                     $message,
1783                     $id,
1784                     $mailbox
1785                     ){
1786     $me = 'sq_fixatts';
1787     while (list($attname, $attvalue) = each($attary)){
1788         /**
1789          * See if this attribute should be removed.
1790          */
1791         foreach ($rm_attnames as $matchtag=>$matchattrs){
1792             if (preg_match($matchtag, $tagname)){
1793                 foreach ($matchattrs as $matchattr){
1794                     if (preg_match($matchattr, $attname)){
1795                         unset($attary{$attname});
1796                         continue;
1797                     }
1798                 }
1799             }
1800         }
1801         /**
1802          * Workaround for IE quirks
1803          */
1804         sq_fixIE_idiocy($attvalue);
1805
1806         /**
1807          * Remove any backslashes, entities, and extraneous whitespace.
1808          */
1809
1810         $oldattvalue = $attvalue;
1811         sq_defang($attvalue);
1812         if ($attname == 'style' && $attvalue !== $oldattvalue) {
1813             // entities are used in the attribute value. In 99% of the cases it's there as XSS
1814             // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1815             $attvalue = "idiocy";
1816             $attary{$attname} = $attvalue;
1817         }
1818         sq_unspace($attvalue);
1819
1820         /**
1821          * Now let's run checks on the attvalues.
1822          * I don't expect anyone to comprehend this. If you do,
1823          * get in touch with me so I can drive to where you live and
1824          * shake your hand personally. :)
1825          */
1826         foreach ($bad_attvals as $matchtag=>$matchattrs){
1827             if (preg_match($matchtag, $tagname)){
1828                 foreach ($matchattrs as $matchattr=>$valary){
1829                     if (preg_match($matchattr, $attname)){
1830                         /**
1831                          * There are two arrays in valary.
1832                          * First is matches.
1833                          * Second one is replacements
1834                          */
1835                         list($valmatch, $valrepl) = $valary;
1836                         $newvalue =
1837                             preg_replace($valmatch, $valrepl, $attvalue);
1838                         if ($newvalue != $attvalue){
1839                             $attary{$attname} = $newvalue;
1840                             $attvalue = $newvalue;
1841                         }
1842                     }
1843                 }
1844             }
1845         }
1846         if ($attname == 'style') {
1847             if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1848                 // 8bit and control characters in style attribute values can be used for XSS, remove them
1849                 $attary{$attname} = '"disallowed character"';
1850             }
1851             preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1852             if (count($aMatch)) {
1853                 foreach($aMatch[1] as $sMatch) {
1854                     // url value
1855                     $urlvalue = $sMatch;
1856                     sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1857                     $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1858                 }
1859             }
1860         }
1861         /**
1862          * Use white list based filtering on attributes which can contain url's
1863          */
1864         else if ($attname == 'href' || $attname == 'src' || $attname == 'background') {
1865             sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1866             $attary{$attname} = $attvalue;
1867         }
1868     }
1869     /**
1870      * See if we need to append any attributes to this tag.
1871      */
1872     foreach ($add_attr_to_tag as $matchtag=>$addattary){
1873         if (preg_match($matchtag, $tagname)){
1874             $attary = array_merge($attary, $addattary);
1875         }
1876     }
1877     return $attary;
1878 }
1879
1880 /**
1881  * This function filters url's
1882  *
1883  * @param  $attvalue        String with attribute value to filter
1884  * @param  $message         message object
1885  * @param  $id               message id
1886  * @param  $mailbox         mailbox
1887  * @param  $sQuote          quoting characters around url's
1888  */
1889 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1890     $attvalue = trim($attvalue);
1891     if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1892         // remove the double quotes
1893         $sQuote = $attvalue[0];
1894         $attvalue = trim(substr($attvalue,1,-1));
1895     }
1896
1897     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1898     // images off by default.
1899     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1900
1901     $secremoveimg = '../images/' . _("sec_remove_eng.png");
1902
1903     /**
1904      * Replace empty src tags with the blank image.  src is only used
1905      * for frames, images, and image inputs.  Doing a replace should
1906      * not affect them working as should be, however it will stop
1907      * IE from being kicked off when src for img tags are not set
1908      */
1909     if ($attvalue == '') {
1910         $attvalue = '"' . SM_PATH . 'images/blank.png"';
1911     } else {
1912         // first, disallow 8 bit characters and control characters
1913         if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1914             switch ($attname) {
1915                 case 'href':
1916                     $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1917                     break;
1918                 default:
1919                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1920                     break;
1921             }
1922         } else {
1923             $aUrl = parse_url($attvalue);
1924             if (isset($aUrl['scheme'])) {
1925                 switch(strtolower($aUrl['scheme'])) {
1926                     case 'mailto':
1927                     case 'http':
1928                     case 'https':
1929                     case 'ftp':
1930                         if ($attname != 'href') {
1931                             if ($view_unsafe_images == false) {
1932                                 $attvalue = $sQuote . $secremoveimg . $sQuote;
1933                             } else {
1934                                 if (isset($aUrl['path'])) {
1935
1936                                     // No one has been able to show that image URIs
1937                                     // can be exploited, so for now, no restrictions
1938                                     // are made at all.  If this proves to be a problem,
1939                                     // the commented-out code below can be of help.
1940                                     // (One consideration is that I see nothing in this
1941                                     // function that specifically says that we will
1942                                     // only ever arrive here when inspecting an image
1943                                     // tag, although that does seem to be the end
1944                                     // result - e.g., <script src="..."> where malicious
1945                                     // image URIs are in fact a problem are already
1946                                     // filtered out elsewhere.
1947                                     /* ---------------------------------
1948                                     // validate image extension.
1949                                     $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1950                                     if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1951                                         // If URI is to something other than
1952                                         // a regular image file, get the contents
1953                                         // and try to see if it is an image.
1954                                         // Don't use Fileinfo (finfo_file()) because
1955                                         // we'd need to make the admin configure the
1956                                         // location of the magic.mime file (FIXME: add finfo_file() support later?)
1957                                         //
1958                                         $mime_type = '';
1959                                         if (function_exists('mime_content_type')
1960                                          && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1961
1962                                             // fetch file
1963                                             //
1964                                             $file_contents = '';
1965                                             while (!feof($FILE)) {
1966                                                 $file_contents .= fread($FILE, 8192);
1967                                             }
1968                                             fclose($FILE);
1969
1970                                             // store file locally
1971                                             //
1972                                             global $attachment_dir, $username;
1973                                             $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1974                                             $localfilename = GenerateRandomString(32, '', 7);
1975                                             $full_localfilename = "$hashed_attachment_dir/$localfilename";
1976                                             while (file_exists($full_localfilename)) {
1977                                                 $localfilename = GenerateRandomString(32, '', 7);
1978                                                 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1979                                             }
1980                                             $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
1981                                             fwrite($FILE, $file_contents);
1982                                             fclose($FILE);
1983
1984                                             // get mime type and remove file
1985                                             //
1986                                             $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
1987                                             unlink("$hashed_attachment_dir/$localfilename");
1988                                         }
1989                                         // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
1990                                         if (substr(strtolower($mime_type), 0, 5) != 'image') {
1991                                             $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1992                                         }
1993                                     }
1994                                     --------------------------------- */
1995                                 } else {
1996                                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1997                                 }
1998                             }
1999                         } else {
2000                             $attvalue = $sQuote . $attvalue . $sQuote;
2001                         }
2002                         break;
2003                     case 'outbind':
2004                         /**
2005                          * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2006                          * One day MS might actually make it match something useful, for now, falling
2007                          * back to using cid2http, so we can grab the blank.png.
2008                          */
2009                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2010                         break;
2011                     case 'cid':
2012                         /**
2013                             * Turn cid: urls into http-friendly ones.
2014                             */
2015                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2016                         break;
2017                     default:
2018                         $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2019                         break;
2020                 }
2021             } else {
2022                 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2023                     // parse_url did not lead to satisfying result
2024                     $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2025                 }
2026             }
2027         }
2028     }
2029 }
2030
2031 /**
2032  * This function edits the style definition to make them friendly and
2033  * usable in SquirrelMail.
2034  *
2035  * @param  $message  the message object
2036  * @param  $id       the message id
2037  * @param  $content  a string with whatever is between <style> and </style>
2038  * @param  $mailbox  the message mailbox
2039  * @return           a string with edited content.
2040  */
2041 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2042     $me = 'sq_fixstyle';
2043     // workaround for </style> in between comments
2044     $iCurrentPos = $pos;
2045     $content = '';
2046     $sToken = '';
2047     $bSucces = false;
2048     $bEndTag = false;
2049     for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2050         $char = $body{$i};
2051         switch ($char) {
2052             case '<':
2053                 $sToken = $char;
2054                 break;
2055             case '/':
2056                  if ($sToken == '<') {
2057                     $sToken .= $char;
2058                     $bEndTag = true;
2059                  } else {
2060                     $content .= $char;
2061                  }
2062                  break;
2063             case '>':
2064                  if ($bEndTag) {
2065                     $sToken .= $char;
2066                     if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2067                         $newpos = $i + 1;
2068                         $bSucces = true;
2069                         break 2;
2070                     } else {
2071                         $content .= $sToken;
2072                     }
2073                     $bEndTag = false;
2074                  } else {
2075                     $content .= $char;
2076                  }
2077                  break;
2078             case '!':
2079                 if ($sToken == '<') {
2080                     // possible comment
2081                     if (isset($body{$i+2}) && substr($body,$i,3) == '!--') {
2082                         $i = strpos($body,'-->',$i+3);
2083                         if ($i === false) { // no end comment
2084                             $i = strlen($body);
2085                         }
2086                         $sToken = '';
2087                     }
2088                 } else {
2089                     $content .= $char;
2090                 }
2091                 break;
2092             default:
2093                 if ($bEndTag) {
2094                     $sToken .= $char;
2095                 } else {
2096                     $content .= $char;
2097                 }
2098                 break;
2099         }
2100     }
2101     if ($bSucces == FALSE){
2102         return array(FALSE, strlen($body));
2103     }
2104
2105
2106
2107     /**
2108      * First look for general BODY style declaration, which would be
2109      * like so:
2110      * body {background: blah-blah}
2111      * and change it to .bodyclass so we can just assign it to a <div>
2112      */
2113     $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2114     $secremoveimg = '../images/' . _("sec_remove_eng.png");
2115     /**
2116     * Fix url('blah') declarations.
2117     */
2118     //   $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2119     //                           "url(\\1$secremoveimg\\2)", $content);
2120
2121     // first check for 8bit sequences and disallowed control characters
2122     if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2123         $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2124         return array($content, $newpos);
2125     }
2126
2127     // IE Sucks hard. We have a special function for it.
2128     sq_fixIE_idiocy($content);
2129
2130     // remove @import line
2131     $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2132
2133     // translate ur\l and variations (IE parses that)
2134     // TODO check if the sq_fixIE_idiocy function already handles this.
2135     $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2136     preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2137     if (count($aMatch)) {
2138         $aValue = $aReplace = array();
2139         foreach($aMatch[1] as $sMatch) {
2140             // url value
2141             $urlvalue = $sMatch;
2142             sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2143             $aValue[] = $sMatch;
2144             $aReplace[] = $urlvalue;
2145         }
2146         $content = str_replace($aValue,$aReplace,$content);
2147     }
2148
2149     /**
2150      * Remove any backslashes, entities, and extraneous whitespace.
2151      */
2152     $contentTemp = $content;
2153     sq_defang($contentTemp);
2154     sq_unspace($contentTemp);
2155
2156     /**
2157      * Fix stupid css declarations which lead to vulnerabilities
2158      * in IE.
2159      *
2160      * Also remove "position" attribute, as it can easily be set
2161      * to "fixed" or "absolute" with "left" and "top" attributes
2162      * of zero, taking over the whole content frame.  It can also
2163      * be set to relative and move itself anywhere it wants to,
2164      * displaying content in areas it shouldn't be allowed to touch.
2165      */
2166     $match   = Array('/\/\*.*\*\//',
2167                     '/expression/i',
2168                     '/behaviou*r/i',
2169                     '/binding/i',
2170                     '/include-source/i',
2171                     '/javascript/i',
2172                     '/script/i',
2173                     '/position/i');
2174     $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2175     $contentNew = preg_replace($match, $replace, $contentTemp);
2176     if ($contentNew !== $contentTemp) {
2177         // insecure css declarations are used. From now on we don't care
2178         // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2179         $content = $contentNew;
2180     }
2181     return array($content, $newpos);
2182 }
2183
2184
2185 /**
2186  * This function converts cid: url's into the ones that can be viewed in
2187  * the browser.
2188  *
2189  * @param  $message  the message object
2190  * @param  $id       the message id
2191  * @param  $cidurl   the cid: url.
2192  * @param  $mailbox  the message mailbox
2193  * @return           a string with a http-friendly url
2194  */
2195 function sq_cid2http($message, $id, $cidurl, $mailbox){
2196     /**
2197      * Get rid of quotes.
2198      */
2199     $quotchar = substr($cidurl, 0, 1);
2200     if ($quotchar == '"' || $quotchar == "'"){
2201         $cidurl = str_replace($quotchar, "", $cidurl);
2202     } else {
2203         $quotchar = '';
2204     }
2205     $cidurl = substr(trim($cidurl), 4);
2206
2207     $match_str = '/\{.*?\}\//';
2208     $str_rep = '';
2209     $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2210
2211     $linkurl = find_ent_id($cidurl, $message);
2212     /* in case of non-safe cid links $httpurl should be replaced by a sort of
2213        unsafe link image */
2214     $httpurl = '';
2215
2216     /**
2217      * This is part of a fix for Outlook Express 6.x generating
2218      * cid URLs without creating content-id headers. These images are
2219      * not part of the multipart/related html mail. The html contains
2220      * <img src="cid:{some_id}/image_filename.ext"> references to
2221      * attached images with as goal to render them inline although
2222      * the attachment disposition property is not inline.
2223      */
2224
2225     if (empty($linkurl)) {
2226         if (preg_match('/{.*}\//', $cidurl)) {
2227             $cidurl = preg_replace('/{.*}\//','', $cidurl);
2228             if (!empty($cidurl)) {
2229                 $linkurl = find_ent_id($cidurl, $message);
2230             }
2231         }
2232     }
2233
2234     if (!empty($linkurl)) {
2235         $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2236             "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2237             '&amp;ent_id=' . $linkurl . $quotchar;
2238     } else {
2239         /**
2240          * If we couldn't generate a proper img url, drop in a blank image
2241          * instead of sending back empty, otherwise it causes unusual behaviour
2242          */
2243         $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2244     }
2245
2246     return $httpurl;
2247 }
2248
2249 /**
2250  * This function changes the <body> tag into a <div> tag since we
2251  * can't really have a body-within-body.
2252  *
2253  * @param  $attary   an array of attributes and values of <body>
2254  * @param  $mailbox  mailbox we're currently reading (for cid2http)
2255  * @param  $message  current message (for cid2http)
2256  * @param  $id       current message id (for cid2http)
2257  * @return           a modified array of attributes to be set for <div>
2258  */
2259 function sq_body2div($attary, $mailbox, $message, $id){
2260     $me = 'sq_body2div';
2261     $divattary = Array('class' => "'bodyclass'");
2262     $text = '#000000';
2263     $has_bgc_stl = $has_txt_stl = false;
2264     $styledef = '';
2265     if (is_array($attary) && sizeof($attary) > 0){
2266         foreach ($attary as $attname=>$attvalue){
2267             $quotchar = substr($attvalue, 0, 1);
2268             $attvalue = str_replace($quotchar, "", $attvalue);
2269             switch ($attname){
2270                 case 'background':
2271                     $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2272                     $styledef .= "background-image: url('$attvalue'); ";
2273                     break;
2274                 case 'bgcolor':
2275                     $has_bgc_stl = true;
2276                     $styledef .= "background-color: $attvalue; ";
2277                     break;
2278                 case 'text':
2279                     $has_txt_stl = true;
2280                     $styledef .= "color: $attvalue; ";
2281                     break;
2282             }
2283         }
2284         // Outlook defines a white bgcolor and no text color. This can lead to
2285         // white text on a white bg with certain themes.
2286         if ($has_bgc_stl && !$has_txt_stl) {
2287             $styledef .= "color: $text; ";
2288         }
2289         if (strlen($styledef) > 0){
2290             $divattary{"style"} = "\"$styledef\"";
2291         }
2292     }
2293     return $divattary;
2294 }
2295
2296 /**
2297  * This is the main function and the one you should actually be calling.
2298  * There are several variables you should be aware of an which need
2299  * special description.
2300  *
2301  * Since the description is quite lengthy, see it here:
2302  * http://linux.duke.edu/projects/mini/htmlfilter/
2303  *
2304  * @param $body                 the string with HTML you wish to filter
2305  * @param $tag_list             see description above
2306  * @param $rm_tags_with_content see description above
2307  * @param $self_closing_tags    see description above
2308  * @param $force_tag_closing    see description above
2309  * @param $rm_attnames          see description above
2310  * @param $bad_attvals          see description above
2311  * @param $add_attr_to_tag      see description above
2312  * @param $message              message object
2313  * @param $id                   message id
2314  * @return                      sanitized html safe to show on your pages.
2315  */
2316 function sq_sanitize($body,
2317                      $tag_list,
2318                      $rm_tags_with_content,
2319                      $self_closing_tags,
2320                      $force_tag_closing,
2321                      $rm_attnames,
2322                      $bad_attvals,
2323                      $add_attr_to_tag,
2324                      $message,
2325                      $id,
2326                      $mailbox
2327                      ){
2328     $me = 'sq_sanitize';
2329     $rm_tags = array_shift($tag_list);
2330     /**
2331      * Normalize rm_tags and rm_tags_with_content.
2332      */
2333     @array_walk($tag_list, 'sq_casenormalize');
2334     @array_walk($rm_tags_with_content, 'sq_casenormalize');
2335     @array_walk($self_closing_tags, 'sq_casenormalize');
2336     /**
2337      * See if tag_list is of tags to remove or tags to allow.
2338      * false  means remove these tags
2339      * true   means allow these tags
2340      */
2341     $curpos = 0;
2342     $open_tags = Array();
2343     $trusted = "\n<!-- begin sanitized html -->\n";
2344     $skip_content = false;
2345     /**
2346      * Take care of netscape's stupid javascript entities like
2347      * &{alert('boo')};
2348      */
2349     $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2350
2351     while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2352         list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2353         $free_content = substr($body, $curpos, $lt-$curpos);
2354         /**
2355          * Take care of <style>
2356          */
2357         if ($tagname == "style" && $tagtype == 1){
2358             list($free_content, $curpos) =
2359                 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2360             if ($free_content != FALSE){
2361                 if ( !empty($attary) ) {
2362                     $attary = sq_fixatts($tagname,
2363                                          $attary,
2364                                          $rm_attnames,
2365                                          $bad_attvals,
2366                                          $add_attr_to_tag,
2367                                          $message,
2368                                          $id,
2369                                          $mailbox
2370                                          );
2371                 }
2372                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2373                 $trusted .= $free_content;
2374                 $trusted .= sq_tagprint($tagname, false, 2);
2375             }
2376             continue;
2377         }
2378         if ($skip_content == false){
2379             $trusted .= $free_content;
2380         }
2381         if ($tagname != FALSE){
2382             if ($tagtype == 2){
2383                 if ($skip_content == $tagname){
2384                     /**
2385                      * Got to the end of tag we needed to remove.
2386                      */
2387                     $tagname = false;
2388                     $skip_content = false;
2389                 } else {
2390                     if ($skip_content == false){
2391                         if ($tagname == "body"){
2392                             $tagname = "div";
2393                         }
2394                         if (isset($open_tags{$tagname}) &&
2395                                 $open_tags{$tagname} > 0){
2396                             $open_tags{$tagname}--;
2397                         } else {
2398                             $tagname = false;
2399                         }
2400                     }
2401                 }
2402             } else {
2403                 /**
2404                  * $rm_tags_with_content
2405                  */
2406                 if ($skip_content == false){
2407                     /**
2408                      * See if this is a self-closing type and change
2409                      * tagtype appropriately.
2410                      */
2411                     if ($tagtype == 1
2412                             && in_array($tagname, $self_closing_tags)){
2413                         $tagtype = 3;
2414                     }
2415                     /**
2416                      * See if we should skip this tag and any content
2417                      * inside it.
2418                      */
2419                     if ($tagtype == 1 &&
2420                             in_array($tagname, $rm_tags_with_content)){
2421                         $skip_content = $tagname;
2422                     } else {
2423                         if (($rm_tags == false
2424                                     && in_array($tagname, $tag_list)) ||
2425                                 ($rm_tags == true &&
2426                                  !in_array($tagname, $tag_list))){
2427                             $tagname = false;
2428                         } else {
2429                             /**
2430                              * Convert body into div.
2431                              */
2432                             if ($tagname == "body"){
2433                                 $tagname = "div";
2434                                 $attary = sq_body2div($attary, $mailbox,
2435                                         $message, $id);
2436                             }
2437                             if ($tagtype == 1){
2438                                 if (isset($open_tags{$tagname})){
2439                                     $open_tags{$tagname}++;
2440                                 } else {
2441                                     $open_tags{$tagname}=1;
2442                                 }
2443                             }
2444                             /**
2445                              * This is where we run other checks.
2446                              */
2447                             if (is_array($attary) && sizeof($attary) > 0){
2448                                 $attary = sq_fixatts($tagname,
2449                                                      $attary,
2450                                                      $rm_attnames,
2451                                                      $bad_attvals,
2452                                                      $add_attr_to_tag,
2453                                                      $message,
2454                                                      $id,
2455                                                      $mailbox
2456                                                      );
2457                             }
2458                         }
2459                     }
2460                 }
2461             }
2462             if ($tagname != false && $skip_content == false){
2463                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2464             }
2465         }
2466         $curpos = $gt+1;
2467     }
2468     $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2469     if ($force_tag_closing == true){
2470         foreach ($open_tags as $tagname=>$opentimes){
2471             while ($opentimes > 0){
2472                 $trusted .= '</' . $tagname . '>';
2473                 $opentimes--;
2474             }
2475         }
2476         $trusted .= "\n";
2477     }
2478     $trusted .= "<!-- end sanitized html -->\n";
2479     return $trusted;
2480 }
2481
2482 /**
2483  * This is a wrapper function to call html sanitizing routines.
2484  *
2485  * @param  $body  the body of the message
2486  * @param  $id    the id of the message
2487
2488  * @param  $message
2489  * @param  $mailbox
2490  * @param  boolean $take_mailto_links When TRUE, converts mailto: links
2491  *                                    into internal SM compose links
2492  *                                    (optional; default = TRUE)
2493  * @return        a string with html safe to display in the browser.
2494  */
2495 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2496
2497     // require_once(SM_PATH . 'functions/url_parser.php');  // for $MailTo_PReg_Match
2498
2499     global $attachment_common_show_images, $view_unsafe_images,
2500            $has_unsafe_images;
2501     /**
2502      * Don't display attached images in HTML mode.
2503      *
2504      * SB: why?
2505      */
2506     $attachment_common_show_images = false;
2507     $tag_list = Array(
2508             false,
2509             "object",
2510             "meta",
2511             "html",
2512             "head",
2513             "base",
2514             "link",
2515             "frame",
2516             "iframe",
2517             "plaintext",
2518             "marquee"
2519             );
2520
2521     $rm_tags_with_content = Array(
2522             "script",
2523             "applet",
2524             "embed",
2525             "title",
2526             "frameset",
2527             "xmp",
2528             "xml"
2529             );
2530
2531     $self_closing_tags =  Array(
2532             "img",
2533             "br",
2534             "hr",
2535             "input",
2536             "outbind"
2537             );
2538
2539     $force_tag_closing = true;
2540
2541     $rm_attnames = Array(
2542             "/.*/" =>
2543             Array(
2544                 "/target/i",
2545                 "/^on.*/i",
2546                 "/^dynsrc/i",
2547                 "/^data.*/i",
2548                 "/^lowsrc.*/i"
2549                 )
2550             );
2551
2552     $secremoveimg = "../images/" . _("sec_remove_eng.png");
2553     $bad_attvals = Array(
2554             "/.*/" =>
2555             Array(
2556                 "/^src|background/i" =>
2557                 Array(
2558                     Array(
2559                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2560                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2561                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2562                         ),
2563                     Array(
2564                         "\\1$secremoveimg\\2",
2565                         "\\1$secremoveimg\\2",
2566                         "\\1$secremoveimg\\2",
2567                         )
2568                     ),
2569                 "/^href|action/i" =>
2570                 Array(
2571                     Array(
2572                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2573                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2574                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2575                         ),
2576                     Array(
2577                         "\\1#\\1",
2578                         "\\1#\\1",
2579                         "\\1#\\1"
2580                         )
2581                     ),
2582         "/^style/i" =>
2583             Array(
2584                 Array(
2585                     "/\/\*.*\*\//",
2586                     "/expression/i",
2587                     "/binding/i",
2588                     "/behaviou*r/i",
2589                     "/include-source/i",
2590
2591                     // position:relative can also be exploited
2592                     // to put content outside of email body area
2593                     // and position:fixed is similarly exploitable
2594                     // as position:absolute, so we'll remove it
2595                     // altogether....
2596                     //
2597                     // Does this screw up legitimate HTML messages?
2598                     // If so, the only fix I see is to allow position
2599                     // attributes (any values?  I think we still have
2600                     // to block static and fixed) only if $use_iframe
2601                     // is enabled (1.5.0+)
2602                     //
2603                     // was:   "/position\s*:\s*absolute/i",
2604                     //
2605                     "/position\s*:/i",
2606
2607                     "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2608                     "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2609                     "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2610                     "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2611                     "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2612                     ),
2613                 Array(
2614                     "",
2615                     "idiocy",
2616                     "idiocy",
2617                     "idiocy",
2618                     "idiocy",
2619                     "idiocy",
2620                     "url",
2621                     "url(\\1#\\1)",
2622                     "url(\\1#\\1)",
2623                     "url(\\1#\\1)",
2624                     "\\1:url(\\2#\\3)"
2625                     )
2626                 )
2627             )
2628         );
2629
2630     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2631     // images off by default.
2632     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2633
2634     if (!$view_unsafe_images){
2635         /**
2636          * Remove any references to http/https if view_unsafe_images set
2637          * to false.
2638          */
2639         array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2640                 '/^([\'\"])\s*https*:.*([\'\"])/si');
2641         array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2642                 "\\1$secremoveimg\\1");
2643         array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2644                 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2645         array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2646                 "url(\\1$secremoveimg\\1)");
2647     }
2648
2649     $add_attr_to_tag = Array(
2650             "/^a$/i" =>
2651             Array('target'=>'"_blank"',
2652                 'title'=>'"'._("This external link will open in a new window").'"'
2653                 )
2654             );
2655     $trusted = sq_sanitize($body,
2656                            $tag_list,
2657                            $rm_tags_with_content,
2658                            $self_closing_tags,
2659                            $force_tag_closing,
2660                            $rm_attnames,
2661                            $bad_attvals,
2662                            $add_attr_to_tag,
2663                            $message,
2664                            $id,
2665                            $mailbox
2666                            );
2667     if (strpos($trusted,$secremoveimg)){
2668         $has_unsafe_images = true;
2669     }
2670
2671     // we want to parse mailto's in HTML output, change to SM compose links
2672     // this is a modified version of code from url_parser.php... but Marc is
2673     // right: we need a better filtering implementation; adding this randomly
2674     // here is not a great solution
2675     //
2676     if ($take_mailto_links) {
2677         // parseUrl($trusted);   // this even parses URLs inside of tags... too aggressive
2678         global $MailTo_PReg_Match;
2679         $MailTo_PReg_Match = '/mailto:' . substr($MailTo_PReg_Match, 1) ;
2680         if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2681             foreach ($regs[0] as $i => $mailto_before) {
2682                 $mailto_params = $regs[10][$i];
2683                 // get rid of any tailing quote since we have to add send_to to the end
2684                 //
2685                 if (substr($mailto_before, strlen($mailto_before) - 1) == '"')
2686                     $mailto_before = substr($mailto_before, 0, strlen($mailto_before) - 1);
2687                 if (substr($mailto_params, strlen($mailto_params) - 1) == '"')
2688                     $mailto_params = substr($mailto_params, 0, strlen($mailto_params) - 1);
2689
2690                 if ($regs[1][$i]) {    //if there is an email addr before '?', we need to merge it with the params
2691                     $to = 'to=' . $regs[1][$i];
2692                     if (strpos($mailto_params, 'to=') > -1)    //already a 'to='
2693                         $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2694                     else {
2695                         if ($mailto_params)    //already some params, append to them
2696                             $mailto_params .= '&amp;' . $to;
2697                         else
2698                             $mailto_params .= '?' . $to;
2699                     }
2700                 }
2701
2702                 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2703
2704                 // we'll already have target=_blank, no need to allow comp_in_new
2705                 // here (which would be a lot more work anyway)
2706                 //
2707                 global $compose_new_win;
2708                 $temp_comp_in_new = $compose_new_win;
2709                 $compose_new_win = 0;
2710                 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2711                 $compose_new_win = $temp_comp_in_new;
2712
2713                 // remove <a href=" and anything after the next quote (we only
2714                 // need the uri, not the link HTML) in compose uri
2715                 //
2716                 $comp_uri = substr($comp_uri, 9);
2717                 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2718                 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2719             }
2720         }
2721     }
2722
2723     return $trusted;
2724 }
2725
2726 /**
2727  * function SendDownloadHeaders - send file to the browser
2728  *
2729  * Original Source: SM core src/download.php
2730  * moved here to make it available to other code, and separate
2731  * front end from back end functionality.
2732  *
2733  * @param string $type0 first half of mime type
2734  * @param string $type1 second half of mime type
2735  * @param string $filename filename to tell the browser for downloaded file
2736  * @param boolean $force whether to force the download dialog to pop
2737  * @param optional integer $filesize send the Content-Header and length to the browser
2738  * @return void
2739  */
2740 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2741     global $languages, $squirrelmail_language;
2742     $isIE = $isIE6plus = false;
2743
2744     sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2745
2746     if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2747             strstr($HTTP_USER_AGENT, 'Opera') === false) {
2748         $isIE = true;
2749     }
2750
2751     if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2752         ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2753         $isIE6plus = true;
2754     }
2755
2756     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2757             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2758         $filename =
2759             call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2760     } else {
2761         $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2762     }
2763
2764     // A Pox on Microsoft and it's Internet Explorer!
2765     //
2766     // IE has lots of bugs with file downloads.
2767     // It also has problems with SSL.  Both of these cause problems
2768     // for us in this function.
2769     //
2770     // See this article on Cache Control headers and SSL
2771     // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2772     //
2773     // The best thing you can do for IE is to upgrade to the latest
2774     // version
2775     //set all the Cache Control Headers for IE
2776     if ($isIE) {
2777         $filename=rawurlencode($filename);
2778         header ("Pragma: public");
2779         header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2780         header ("Cache-Control: post-check=0, pre-check=0", false);
2781         header ("Cache-Control: private");
2782
2783         //set the inline header for IE, we'll add the attachment header later if we need it
2784         header ("Content-Disposition: inline; filename=$filename");
2785     }
2786
2787     if (!$force) {
2788         // Try to show in browser window
2789         header ("Content-Disposition: inline; filename=\"$filename\"");
2790         header ("Content-Type: $type0/$type1; name=\"$filename\"");
2791     } else {
2792         // Try to pop up the "save as" box
2793
2794         // IE makes this hard.  It pops up 2 save boxes, or none.
2795         // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2796         // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2797         // But, according to Microsoft, it is "RFC compliant but doesn't
2798         // take into account some deviations that allowed within the
2799         // specification."  Doesn't that mean RFC non-compliant?
2800         // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2801
2802         // all browsers need the application/octet-stream header for this
2803         header ("Content-Type: application/octet-stream; name=\"$filename\"");
2804
2805         // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2806         // Do not have quotes around filename, but that applied to
2807         // "attachment"... does it apply to inline too?
2808         header ("Content-Disposition: attachment; filename=\"$filename\"");
2809
2810         if ($isIE && !$isIE6plus) {
2811             // This combination seems to work mostly.  IE 5.5 SP 1 has
2812             // known issues (see the Microsoft Knowledge Base)
2813
2814             // This works for most types, but doesn't work with Word files
2815             header ("Content-Type: application/download; name=\"$filename\"");
2816             header ("Content-Type: application/force-download; name=\"$filename\"");
2817             // These are spares, just in case.  :-)
2818             //header("Content-Type: $type0/$type1; name=\"$filename\"");
2819             //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2820             //header("Content-Type: application/octet-stream; name=\"$filename\"");
2821         } else if ($isIE) {
2822              // This is to prevent IE for MIME sniffing and auto open a file in IE
2823              header ("Content-Type: application/force-download; name=\"$filename\"");
2824         } else {
2825             // another application/octet-stream forces download for Netscape
2826             header ("Content-Type: application/octet-stream; name=\"$filename\"");
2827         }
2828     }
2829
2830     //send the content-length header if the calling function provides it
2831     if ($filesize > 0) {
2832         header("Content-Length: $filesize");
2833     }
2834
2835 }  // end fn SendDownloadHeaders