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 1999-2022 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 // TODO: Needs documentation.  $ent_id default is usually 1
 194 function mime_print_body_lines ($imap_stream, $id, $ent_id, $encoding, $rStream='php://stdout', $force_crlf='') {
 195
 196     /* Don't kill the connection if the browser is over a dialup
 197      * and it would take over 30 seconds to download it.
 198      * Don't call set_time_limit in safe mode.
 199      */
 200
 201     if (!ini_get('safe_mode')) {
 202         set_time_limit(0);
 203     }
 204     /* in case of base64 encoded attachments, do not buffer them.
 205        Instead, echo the decoded attachment directly to screen */
 206     if (strtolower($encoding) == 'base64') {
 207         if (!$ent_id) {
 208             $query = "FETCH $id BODY[]";
 209         } else {
 210             $query = "FETCH $id BODY[$ent_id]";
 211         }
 212         sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
 213     } else {
 214         $body = mime_fetch_body ($imap_stream, $id, $ent_id);
 215         if (is_resource($rStream)) {
 216             fputs($rStream,decodeBody($body, $encoding, $force_crlf));
 217         } else {
 218             echo decodeBody($body, $encoding, $force_crlf);
 219         }
 220     }
 221
 222     /*
 223        TODO, use the same method for quoted printable.
 224        However, I assume that quoted printable attachments aren't that large
 225        so the performancegain / memory usage drop will be minimal.
 226        If we decide to add that then we need to adapt sqimap_fread because
 227        we need to split te result on \n and fread doesn't stop at \n. That
 228        means we also should provide $results from sqimap_fread (by ref) to
 229        te function and set $no_return to false. The $filter function for
 230        quoted printable should handle unsetting of $results.
 231      */
 232     /*
 233        TODO 2: find out how we write to the output stream php://stdout. fwrite
 234        doesn't work because 'php://stdout isn't a stream.
 235      */
 236
 237     return;
 238 }
 239
 240 /* -[ END MIME DECODING ]----------------------------------------------------------- */
 241
 242 /* This is here for debugging purposes.  It will print out a list
 243  * of all the entity IDs that are in the $message object.
 244  */
 245 function listEntities ($message) {
 246     if ($message) {
 247         echo "<tt>" . $message->entity_id . ' : ' . $message->type0 . '/' . $message->type1 . ' parent = '. $message->parent->entity_id. '<br />';
 248         for ($i = 0; isset($message->entities[$i]); $i++) {
 249             echo "$i : ";
 250             $msg = listEntities($message->entities[$i]);
 251
 252             if ($msg) {
 253                 echo "return: ";
 254                 return $msg;
 255             }
 256         }
 257     }
 258 }
 259
 260 function getPriorityStr($priority) {
 261     $priority_level = substr($priority,0,1);
 262
 263     switch($priority_level) {
 264         /* Check for a higher then normal priority. */
 265         case '1':
 266         case '2':
 267             $priority_string = _("High");
 268             break;
 269
 270         /* Check for a lower then normal priority. */
 271         case '4':
 272         case '5':
 273             $priority_string = _("Low");
 274             break;
 275
 276         /* Check for a normal priority. */
 277         case '3':
 278         default:
 279             $priority_level = '3';
 280             $priority_string = _("Normal");
 281             break;
 282
 283     }
 284     return $priority_string;
 285 }
 286
 287 /* returns a $message object for a particular entity id */
 288 function getEntity ($message, $ent_id) {
 289     return $message->getEntity($ent_id);
 290 }
 291
 292 /* translateText
 293  * Extracted from strings.php 23/03/2002
 294  */
 295
 296 function translateText(&$body, $wrap_at, $charset) {
 297     global $where, $what;   /* from searching */
 298     global $color;          /* color theme */
 299
 300     // require_once(SM_PATH . 'functions/url_parser.php');
 301
 302     $body_ary = explode("\n", $body);
 303     for ($i=0; $i < count($body_ary); $i++) {
 304         $line = rtrim($body_ary[$i],"\r");
 305
 306         if (strlen($line) - 2 >= $wrap_at) {
 307             sqWordWrap($line, $wrap_at, $charset);
 308         }
 309         $line = charset_decode($charset, $line);
 310         $line = str_replace("\t", '        ', $line);
 311
 312         parseUrl ($line);
 313
 314         $quotes = 0;
 315         $pos = 0;
 316         $j = strlen($line);
 317
 318         while ($pos < $j) {
 319             if ($line[$pos] == ' ') {
 320                 $pos++;
 321             } else if (strpos($line, '&gt;', $pos) === $pos) {
 322                 $pos += 4;
 323                 $quotes++;
 324             } else {
 325                 break;
 326             }
 327         }
 328
 329         if ($quotes % 2) {
 330             $line = '<span class="quote1">' . $line . '</span>';
 331         } elseif ($quotes) {
 332             $line = '<span class="quote2">' . $line . '</span>';
 333         }
 334
 335         $body_ary[$i] = $line;
 336     }
 337     $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
 338 }
 339
 340 /**
 341  * This returns a parsed string called $body. That string can then
 342  * be displayed as the actual message in the HTML. It contains
 343  * everything needed, including HTML Tags, Attachments at the
 344  * bottom, etc.
 345  *
 346  * Since 1.2.0 function uses message_body hook.
 347  * Till 1.3.0 function included output of formatAttachments().
 348  *
 349  * @param resource $imap_stream imap connection resource
 350  * @param object $message squirrelmail message object
 351  * @param array $color squirrelmail color theme array
 352  * @param integer $wrap_at number of characters per line
 353  * @param string $ent_num (since 1.3.0) message part id
 354  * @param integer $id (since 1.3.0) message id
 355  * @param string $mailbox (since 1.3.0) imap folder name
 356  * @return string html formated message text
 357  */
 358 function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') {
 359     /* This if statement checks for the entity to show as the
 360      * primary message. To add more of them, just put them in the
 361      * order that is their priority.
 362      */
 363     global $startMessage, $languages, $squirrelmail_language,
 364            $show_html_default, $sort, $has_unsafe_images, $passed_ent_id,
 365            $use_iframe, $iframe_height, $download_and_unsafe_link,
 366            $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text,
 367            $oTemplate, $nbsp;
 368
 369     // workaround for not updated config.php
 370     if (! isset($use_iframe)) $use_iframe = false;
 371
 372     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
 373     // images off by default.
 374     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
 375
 376     $body = '';
 377     $urlmailbox = urlencode($mailbox);
 378     $body_message = getEntity($message, $ent_num);
 379     if (($body_message->header->type0 == 'text') ||
 380             ($body_message->header->type0 == 'rfc822')) {
 381         $body = mime_fetch_body ($imap_stream, $id, $ent_num);
 382         $body = decodeBody($body, $body_message->header->encoding);
 383
 384         if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 385                 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) {
 386             if (mb_detect_encoding($body) != 'ASCII') {
 387                 $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body);
 388             }
 389         }
 390
 391         /* As of 1.5.2, $body is passed (and modified) by reference */
 392         do_hook('message_body', $body);
 393
 394         /* If there are other types that shouldn't be formatted, add
 395          * them here.
 396          */
 397
 398         if ($body_message->header->type1 == 'html') {
 399             // Do we need to make an HTML part viewable as non-HTML plain text?
 400             if ($show_html_default != 1) {
 401                 $entity_conv = array('&nbsp;' => ' ',
 402                                      // These are better done by regex (below)
 403                                      // '<p>'    => "\n",
 404                                      // '<P>'    => "\n",
 405                                      // '<br>'   => "\n",
 406                                      // '<BR>'   => "\n",
 407                                      // '<br />' => "\n",
 408                                      // '<BR />' => "\n",
 409                                      // '<tr>'   => "\n",
 410                                      // '<div>'  => "\n",
 411                                      '&gt;'   => '>',
 412                                      '&lt;'   => '<',
 413                                      '&amp;'   => '&',
 414                                      '&copy;'   => '©');
 415                 // first, completely remove <style> tags as they aren't useful in this context
 416                 $body = preg_replace('/<style.*>.*<\/style.*>/isU', '', $body);
 417                 // emulate how newlines are treated as spaces in HTML
 418                 $body = preg_replace('/(\r|\n)+/', ' ', $body);
 419                 // now replace the tags listed just above
 420                 $body = strtr($body, $entity_conv);
 421                 // <p>, <br>, <tr> and <div> are best replaced by a newline
 422                 $body = preg_replace('/<(p|br|tr|div).*>/isU', "\n", $body);
 423                 // remove the rest of the HTML tags
 424                 $body = strip_tags($body);
 425                 // condense multiple spaces into one
 426                 $body = preg_replace('/[ \t]+/', ' ', $body);
 427                 // trim each line
 428                 $body = preg_replace('/ *\n */', "\n", $body);
 429                 // allow maximum two newlines
 430                 $body = preg_replace('/\n\n\n+/', "\n\n", $body);
 431                 $body = trim($body);
 432                 translateText($body, $wrap_at,
 433                         $body_message->header->getParameter('charset'));
 434             } elseif ($use_iframe) {
 435                 /**
 436                  * If we don't add html message between iframe tags,
 437                  * we must detect unsafe images and modify $has_unsafe_images.
 438                  */
 439                 $html_body = magicHTML($body, $id, $message, $mailbox);
 440                 // Convert character set in order to display html mails in different character set
 441                 $html_body = charset_decode($body_message->header->getParameter('charset'),$html_body,false,true);
 442
 443                 // creating iframe url
 444                 $iframeurl=sqm_baseuri().'src/view_html.php?'
 445                     . 'mailbox=' . $urlmailbox
 446                     . '&amp;passed_id=' . $id
 447                     . '&amp;ent_id=' . $ent_num
 448                     . '&amp;view_unsafe_images=' . (int) $view_unsafe_images;
 449
 450                 global $oTemplate;
 451                 $oTemplate->assign('iframe_url', $iframeurl);
 452                 $oTemplate->assign('iframe_height', $iframe_height);
 453                 $oTemplate->assign('html_body', $html_body);
 454
 455                 $body = $oTemplate->fetch('read_html_iframe.tpl');
 456             } else {
 457                 // old way of html rendering
 458                 /**
 459                  * convert character set. charset_decode does not remove html special chars
 460                  * applied by magicHTML functions and does not sanitize them second time if
 461                  * fourth argument is true.
 462                  */
 463                 $charset = $body_message->header->getParameter('charset');
 464                 if (!empty($charset)) {
 465                     $body = charset_decode($charset,$body,false,true);
 466                 }
 467                 $body = magicHTML($body, $id, $message, $mailbox);
 468             }
 469         } else {
 470             translateText($body, $wrap_at,
 471                     $body_message->header->getParameter('charset'));
 472         }
 473
 474         /*
 475          * Previously the links for downloading and unsafe images were printed
 476          * under the mail. By putting the links in a global variable we can
 477          * print it in the toolbar where it belongs. Since the original code was
 478          * in this place it's left here. It might be possible to move it to some
 479          * other place if that makes sense. The possibility to do so has not
 480          * been evaluated yet.
 481          */
 482
 483         // Initialize the global variable to an empty string.
 484         // 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.
 485         $download_and_unsafe_link = '';
 486
 487         // Prepare and build a link for downloading the mail.
 488         $link = 'passed_id=' . $id . '&amp;ent_id='.$ent_num.
 489             '&amp;mailbox=' . $urlmailbox .'&amp;sort=' . $sort .
 490             '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
 491         if (isset($passed_ent_id)) {
 492             $link .= '&amp;passed_ent_id='.$passed_ent_id;
 493         }
 494         $download_href = SM_PATH . 'src/download.php?absolute_dl=true&amp;' . $link;
 495
 496         // Always add the link for downloading the mail as a file to the global
 497         // variable.
 498         $download_and_unsafe_link .= "$nbsp|$nbsp"
 499             . create_hyperlink($download_href, _("Download this as a file"));
 500
 501         // Find out the right text to use in the link depending on the
 502         // circumstances. If the unsafe images are displayed the link should
 503         // hide them, if they aren't displayed the link should only appear if
 504         // the mail really contains unsafe images.
 505         if ($view_unsafe_images) {
 506             $text = _("Hide Unsafe Images");
 507         } else {
 508             if (isset($has_unsafe_images) && $has_unsafe_images) {
 509                 $link .= '&amp;view_unsafe_images=1';
 510                 $text = _("View Unsafe Images");
 511             } else {
 512                 $text = '';
 513             }
 514         }
 515
 516         // Only create a link for unsafe images if there's need for one. If so:
 517         // add it to the global variable.
 518         if($text != '') {
 519             $unsafe_image_toggle_href = SM_PATH . 'src/read_body.php?'.$link;
 520             $unsafe_image_toggle_text = $text;
 521             $download_and_unsafe_link .= "$nbsp|$nbsp"
 522                 . create_hyperlink($unsafe_image_toggle_href, $text);
 523         }
 524     }
 525     return $body;
 526 }
 527
 528 /**
 529  * Generate attachments array for passing to templates.
 530  *
 531  * @since 1.5.2
 532  * @param object $message SquirrelMail message object
 533  * @param array $exclude_id message parts that are not attachments.
 534  * @param string $mailbox mailbox name
 535  * @param integer $id message id
 536  */
 537 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
 538     global $where, $what, $startMessage, $color, $passed_ent_id,
 539            $base_uri, $block_svg_download;
 540
 541     $att_ar = $message->getAttachments($exclude_id);
 542     $urlMailbox = urlencode($mailbox);
 543
 544     $attachments = array();
 545     foreach ($att_ar as $att) {
 546         $ent = $att->entity_id;
 547         $header = $att->header;
 548         $type0 = strtolower($header->type0);
 549         $type1 = strtolower($header->type1);
 550         if ($block_svg_download && strpos($type1, 'svg') === 0)
 551             continue;
 552
 553         $name = '';
 554         $links = array();
 555         $links['download link']['text'] = _("Download");
 556         $links['download link']['href'] = $base_uri .
 557             "src/download.php?absolute_dl=true&amp;passed_id=$id&amp;mailbox=$urlMailbox&amp;ent_id=$ent";
 558
 559         if ($type0 =='message' && $type1 == 'rfc822') {
 560             $default_page = $base_uri  . 'src/read_body.php';
 561             $rfc822_header = $att->rfc822_header;
 562             $filename = $rfc822_header->subject;
 563             if (trim( $filename ) == '') {
 564                 $filename = 'untitled-[' . $ent . ']' ;
 565             }
 566             $from_o = $rfc822_header->from;
 567             if (is_object($from_o)) {
 568                 $from_name = decodeHeader($from_o->getAddress(false));
 569             } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) {
 570                 // something weird happens when a digest message is opened and you return to the digest
 571                 // now the from object is part of an array. Probably the parseHeader call overwrites the info
 572                 // retrieved from the bodystructure in a different way. We need to fix this later.
 573                 // possible starting point, do not fetch header we already have and inspect how
 574                 // the rfc822_header object behaves.
 575                 $from_name = decodeHeader($from_o[0]->getAddress(false));
 576             } else {
 577                 $from_name = _("Unknown sender");
 578             }
 579             $description = _("From").': '.$from_name;
 580         } else {
 581             $default_page = $base_uri  . 'src/download.php';
 582             $filename = $att->getFilename();
 583             if ($header->description) {
 584                 $description = decodeHeader($header->description);
 585             } else {
 586                 $description = '';
 587             }
 588         }
 589
 590         $display_filename = $filename;
 591         if (isset($passed_ent_id)) {
 592             $passed_ent_id_link = '&amp;passed_ent_id='.$passed_ent_id;
 593         } else {
 594             $passed_ent_id_link = '';
 595         }
 596         $defaultlink = $default_page . "?startMessage=$startMessage"
 597             . "&amp;passed_id=$id&amp;mailbox=$urlMailbox"
 598             . '&amp;ent_id='.$ent.$passed_ent_id_link;
 599         if ($where && $what) {
 600             $defaultlink .= '&amp;where='. urlencode($where).'&amp;what='.urlencode($what);
 601         }
 602         // IE does make use of mime content sniffing. Forcing a download
 603         // prohibit execution of XSS inside an application/octet-stream attachment
 604         if ($type0 == 'application' && $type1 == 'octet-stream') {
 605             $defaultlink .= '&amp;absolute_dl=true';
 606         }
 607
 608         /* This executes the attachment hook with a specific MIME-type.
 609          * It also allows plugins to run if there's a rule for a more
 610          * generic type. Finally, a hook for ALL attachment types is
 611          * run as well.
 612          */
 613         // First remember the default link.
 614         $defaultlink_orig = $defaultlink;
 615
 616         /* The API for this hook has changed as of 1.5.2 so that all plugin
 617            arguments are passed in an array instead of each their own plugin
 618            argument, and arguments are passed by reference, so instead of
 619            returning any changes, changes should simply be made to the original
 620            arguments themselves. */
 621         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 622                     &$defaultlink, &$display_filename, &$where, &$what,
 623                     &$type0, &$type1);
 624         do_hook("attachment $type0/$type1", $temp);
 625         /* The API for this hook has changed as of 1.5.2 so that all plugin
 626            arguments are passed in an array instead of each their own plugin
 627            argument, and arguments are passed by reference, so instead of
 628            returning any changes, changes should simply be made to the original
 629            arguments themselves. */
 630         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 631                       &$defaultlink, &$display_filename, &$where, &$what,
 632                       &$type0, &$type1);
 633         // Do not let a generic plugin change the default link if a more
 634         // specialized one already did it...
 635         if ($defaultlink != $defaultlink_orig) {
 636             $dummy = '';
 637             $temp[5] = &$dummy;
 638         }
 639         do_hook("attachment $type0/*", $temp);
 640         /* The API for this hook has changed as of 1.5.2 so that all plugin
 641            arguments are passed in an array instead of each their own plugin
 642            argument, and arguments are passed by reference, so instead of
 643            returning any changes, changes should simply be made to the original
 644            arguments themselves. */
 645         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 646                       &$defaultlink, &$display_filename, &$where, &$what,
 647                       &$type0, &$type1);
 648         // Do not let a generic plugin change the default link if a more
 649         // specialized one already did it...
 650         if ($defaultlink != $defaultlink_orig) {
 651             $dummy = '';
 652             $temp[5] = &$dummy;
 653         }
 654         do_hook("attachment */*", $temp);
 655
 656         $this_attachment = array();
 657         $this_attachment['Name'] = decodeHeader($display_filename);
 658         $this_attachment['Description'] = $description;
 659         $this_attachment['DefaultHREF'] = $defaultlink;
 660         $this_attachment['DownloadHREF'] = $links['download link']['href'];
 661         $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
 662
 663         // base64 encoded file sizes are misleading, so approximate real size
 664         if (!empty($header->encoding) && strtolower($header->encoding) == 'base64')
 665             $this_attachment['Size'] = $header->size / 4 * 3;
 666         else
 667             $this_attachment['Size'] = $header->size;
 668
 669         $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
 670         $this_attachment['OtherLinks'] = array();
 671         foreach ($links as $val) {
 672             if ($val['text']==_("Download")) {
 673                 $this_attachment['DownloadHREF'] = $val['href'];
 674                 continue;
 675             }
 676             if ($val['text']==_("View")) {
 677                 $this_attachment['ViewHREF'] = $val['href'];
 678                 continue;
 679             }
 680
 681             // This makes no sense - If 'text' and 'extra' are just concatenated,
 682             // there is no point in having 'extra'.... I am going to assume this
 683             // was a mistake and am changing 'extra' to be what I think it was
 684             // meant to be: additional tag attributes.  However, I'm not checking
 685             // extensively for plugins that were using this the wrong way (but why would they?)
 686             if (empty($val['text']))
 687                 continue;
 688
 689             $temp = array();
 690             $temp['HREF'] = $val['href'];
 691             $temp['Text'] = $val['text'];
 692             $temp['Extra'] = (empty($val['extra']) ? '' : $val['extra']);
 693             $this_attachment['OtherLinks'][] = $temp;
 694         }
 695         $attachments[] = $this_attachment;
 696
 697         unset($links);
 698     }
 699
 700     return $attachments;
 701 }
 702
 703 /**
 704  * Displays attachment links and information
 705  *
 706  * Since 1.3.0 function is not included in formatBody() call.
 707  *
 708  * Since 1.0.2 uses attachment $type0/$type1 hook.
 709  * Since 1.2.5 uses attachment $type0/* hook.
 710  * Since 1.5.0 uses attachments_bottom hook.
 711  * Since 1.5.2 uses templates and does *not* return a value.
 712  *
 713  * @param object $message SquirrelMail message object
 714  * @param array $exclude_id message parts that are not attachments.
 715  * @param string $mailbox mailbox name
 716  * @param integer $id message id
 717  */
 718 function formatAttachments($message, $exclude_id, $mailbox, $id) {
 719     global $oTemplate;
 720
 721     $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
 722
 723     $oTemplate->assign('attachments', $attach);
 724     $oTemplate->display('read_attachments.tpl');
 725 }
 726
 727 function sqimap_base64_decode(&$string) {
 728
 729     // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
 730     // fly decoding (to reduce memory usage) you have to check if the
 731     // data has incomplete pairs
 732
 733     // Remove the noise in order to check if the 4 bytes pairs are complete
 734     $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
 735
 736     $sStringRem = '';
 737     $iMod = strlen($string) % 4;
 738     if ($iMod) {
 739         $sStringRem = substr($string,-$iMod);
 740         // Check if $sStringRem contains padding characters
 741         if (substr($sStringRem,-1) != '=') {
 742             $string = substr($string,0,-$iMod);
 743         } else {
 744             $sStringRem = '';
 745         }
 746     }
 747     $string = base64_decode($string);
 748     return $sStringRem;
 749 }
 750
 751 /**
 752  * Decodes encoded string (usually message body)
 753  *
 754  * This function decodes a string (usually the message body)
 755  * depending on the encoding type.  Currently quoted-printable
 756  * and base64 encodings are supported.
 757  *
 758  * The decode_body hook was added to this function in 1.4.2/1.5.0.
 759  * The $force_crlf parameter was added in 1.5.2.
 760  *
 761  * @param string $string     The encoded string
 762  * @param string $encoding   used encoding
 763  * @param string $force_crlf Whether or not to force CRLF or LF
 764  *                           line endings (or to leave as is).
 765  *                           If given as "LF", line endings will
 766  *                           all be converted to LF; if "CRLF",
 767  *                           line endings will all be converted
 768  *                           to CRLF.  If given as an empty value,
 769  *                           the global $force_crlf_default will
 770  *                           be consulted (it can be specified in
 771  *                           config/config_local.php).  Otherwise,
 772  *                           any other value will cause the string
 773  *                           to be left alone.  Note that this will
 774  *                           be overridden to "LF" if not using at
 775  *                           least PHP version 4.3.0. (OPTIONAL;
 776  *                           default is empty - consult global
 777  *                           default value)
 778  *
 779  * @return string The decoded string
 780  *
 781  * @since 1.0
 782  *
 783  */
 784 function decodeBody($string, $encoding, $force_crlf='') {
 785
 786     global $force_crlf_default;
 787     if (empty($force_crlf)) $force_crlf = $force_crlf_default;
 788     $force_crlf = strtoupper($force_crlf);
 789
 790     // must force line endings to LF due to broken
 791     // quoted_printable_decode() in PHP versions
 792     // before 4.3.0 (see below)
 793     //
 794     if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
 795         $string = str_replace("\r\n", "\n", $string);
 796     else if ($force_crlf == 'CRLF')
 797         $string = str_replace("\n", "\r\n", $string);
 798
 799     $encoding = strtolower($encoding);
 800
 801     $encoding_handler = do_hook('decode_body', $encoding);
 802
 803
 804     // plugins get first shot at decoding the string
 805     //
 806     if (!empty($encoding_handler) && function_exists($encoding_handler)) {
 807         $string = $encoding_handler('decode', $string);
 808
 809     } elseif ($encoding == 'quoted-printable' ||
 810             $encoding == 'quoted_printable') {
 811
 812         // quoted_printable_decode() function is broken in older
 813         // php versions.  Text with \r\n decoding was fixed only
 814         // in php 4.3.0.  Minimal code requirement is PHP 4.0.4+
 815         // and the above call to:  str_replace("\r\n", "\n", $string);
 816         //
 817         $string = quoted_printable_decode($string);
 818
 819     } elseif ($encoding == 'base64') {
 820         $string = base64_decode($string);
 821     }
 822
 823     // All other encodings are returned raw.
 824     return $string;
 825 }
 826
 827 /**
 828  * Decodes headers
 829  *
 830  * This function decodes strings that are encoded according to
 831  * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
 832  * Patched by Christian Schmidt <christian@ostenfeld.dk>  23/03/2002
 833  *
 834  * @param string $string header string that has to be made readable
 835  * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
 836  * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
 837  * @param boolean $decide decide if string can be utfencoded. defaults to false
 838  * @return string decoded header string
 839  */
 840 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
 841     global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
 842     if (is_array($string)) {
 843         $string = implode("\n", $string);
 844     }
 845
 846     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 847             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
 848         $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
 849         // Do we need to return at this point?
 850         // return $string;
 851     }
 852     $i = 0;
 853     $iLastMatch = -2;
 854     $encoded = true;
 855
 856 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
 857     $aString = explode(' ',$string);
 858     $ret = '';
 859     foreach ($aString as $chunk) {
 860         if ($encoded && $chunk === '') {
 861             continue;
 862         } elseif ($chunk === '') {
 863             $ret .= ' ';
 864             continue;
 865         }
 866         $encoded = false;
 867         /* if encoded words are not separated by a linear-space-white we still catch them */
 868         $j = $i-1;
 869
 870         while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
 871             /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
 872             if ($iLastMatch !== $j) {
 873                 if ($htmlsafe) {
 874                     $ret .= '&#32;';
 875                 } else {
 876                     $ret .= ' ';
 877                 }
 878             }
 879             $iLastMatch = $i;
 880             $j = $i;
 881             if ($htmlsafe) {
 882                 $ret .= sm_encode_html_special_chars($res[1]);
 883             } else {
 884                 $ret .= $res[1];
 885             }
 886             $encoding = ucfirst($res[3]);
 887
 888             /* decide about valid decoding */
 889             if ($decide && is_conversion_safe($res[2])) {
 890                 $utfencode=true;
 891                 $can_be_encoded=true;
 892             } else {
 893                 $can_be_encoded=false;
 894             }
 895             switch ($encoding)
 896             {
 897                 case 'B':
 898                     // fix broken base64-encoded strings (remove end = padding,
 899                     // change any = to + in middle of string, add padding back
 900                     // to the end)
 901                     if ($fix_broken_base64_encoded_messages) {
 902                         $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
 903                         $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
 904                     }
 905                     $replace = base64_decode($res[4]);
 906                     if ($utfencode) {
 907                         if ($can_be_encoded) {
 908                             /* convert string to different charset,
 909                              * if functions asks for it (usually in compose)
 910                              */
 911                             $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
 912                         } else {
 913                             // convert string to html codes in order to display it
 914                             $ret .= charset_decode($res[2],$replace);
 915                         }
 916                     } else {
 917                         if ($htmlsafe) {
 918                             $replace = sm_encode_html_special_chars($replace);
 919                         }
 920                         $ret.= $replace;
 921                     }
 922                     break;
 923                 case 'Q':
 924                     $replace = str_replace('_', ' ', $res[4]);
 925                     $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
 926                                              (check_php_version(5, 3, 0)
 927                                               ? function($matches) { return chr(hexdec($matches[1])); }
 928                                               : create_function ('$matches', 'return chr(hexdec($matches[1]));')
 929                                              ),
 930                                              $replace);
 931                     if ($utfencode) {
 932                         if ($can_be_encoded) {
 933                             /* convert string to different charset,
 934                              * if functions asks for it (usually in compose)
 935                              */
 936                             $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
 937                         } else {
 938                             // convert string to html codes in order to display it
 939                             $replace = charset_decode($res[2], $replace);
 940                         }
 941                     } else {
 942                         if ($htmlsafe) {
 943                             $replace = sm_encode_html_special_chars($replace);
 944                         }
 945                     }
 946                     $ret .= $replace;
 947                     break;
 948                 default:
 949                     break;
 950             }
 951             $chunk = $res[5];
 952             $encoded = true;
 953         }
 954         if (!$encoded) {
 955             if ($htmlsafe) {
 956                 $ret .= '&#32;';
 957             } else {
 958                 $ret .= ' ';
 959             }
 960         }
 961
 962         if (!$encoded && $htmlsafe) {
 963             $ret .= sm_encode_html_special_chars($chunk);
 964         } else {
 965             $ret .= $chunk;
 966         }
 967         ++$i;
 968     }
 969     /* remove the first added space */
 970     if ($ret) {
 971         if ($htmlsafe) {
 972             $ret = substr($ret,5);
 973         } else {
 974             $ret = substr($ret,1);
 975         }
 976     }
 977
 978     return $ret;
 979 }
 980
 981 /**
 982  * Encodes header
 983  *
 984  * Function uses XTRA_CODE _encodeheader function, if such function exists.
 985  *
 986  * Function uses Q encoding by default and encodes a string according to RFC
 987  * 1522 for use in headers if it contains 8-bit characters or anything that
 988  * looks like it should be encoded.
 989  *
 990  * Function switches to B encoding and encodeHeaderBase64() function, if
 991  * string is 8bit and multibyte character set supported by mbstring extension
 992  * is used. It can cause E_USER_NOTICE errors, if interface is used with
 993  * multibyte character set unsupported by mbstring extension.
 994  *
 995  * @param string $string header string, that has to be encoded
 996  * @return string quoted-printable encoded string
 997  * @todo make $mb_charsets system wide constant
 998  */
 999 function encodeHeader ($string) {
1000     global $default_charset, $languages, $squirrelmail_language;
1001
1002     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
1003             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
1004         return  call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
1005     }
1006
1007     // Use B encoding for multibyte charsets
1008     $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
1009     if (in_array($default_charset,$mb_charsets) &&
1010         in_array($default_charset,sq_mb_list_encodings()) &&
1011         sq_is8bit($string)) {
1012         return encodeHeaderBase64($string,$default_charset);
1013     } elseif (in_array($default_charset,$mb_charsets) &&
1014               sq_is8bit($string) &&
1015               ! in_array($default_charset,sq_mb_list_encodings())) {
1016         // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
1017         // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
1018     }
1019
1020     // Encode only if the string contains 8-bit characters or =?
1021     $j = strlen($string);
1022     $max_l = 75 - strlen($default_charset) - 7;
1023     $aRet = array();
1024     $ret = '';
1025     $iEncStart = $enc_init = false;
1026     $cur_l = $iOffset = 0;
1027     for($i = 0; $i < $j; ++$i) {
1028         switch($string[$i])
1029         {
1030             case '"':
1031             case '=':
1032             case '<':
1033             case '>':
1034             case ',':
1035             case '?':
1036             case '_':
1037                 if ($iEncStart === false) {
1038                     $iEncStart = $i;
1039                 }
1040                 $cur_l+=3;
1041                 if ($cur_l > ($max_l-2)) {
1042                     /* if there is an stringpart that doesn't need encoding, add it */
1043                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1044                     $aRet[] = "=?$default_charset?Q?$ret?=";
1045                     $iOffset = $i;
1046                     $cur_l = 0;
1047                     $ret = '';
1048                     $iEncStart = false;
1049                 } else {
1050                     $ret .= sprintf("=%02X",ord($string[$i]));
1051                 }
1052                 break;
1053             case '(':
1054             case ')':
1055                 if ($iEncStart !== false) {
1056                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1057                     $aRet[] = "=?$default_charset?Q?$ret?=";
1058                     $iOffset = $i;
1059                     $cur_l = 0;
1060                     $ret = '';
1061                     $iEncStart = false;
1062                 }
1063                 break;
1064             case ' ':
1065                 if ($iEncStart !== false) {
1066                     $cur_l++;
1067                     if ($cur_l > $max_l) {
1068                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1069                         $aRet[] = "=?$default_charset?Q?$ret?=";
1070                         $iOffset = $i;
1071                         $cur_l = 0;
1072                         $ret = '';
1073                         $iEncStart = false;
1074                     } else {
1075                         $ret .= '_';
1076                     }
1077                 }
1078                 break;
1079             default:
1080                 $k = ord($string[$i]);
1081                 if ($k > 126) {
1082                     if ($iEncStart === false) {
1083                         // do not start encoding in the middle of a string, also take the rest of the word.
1084                         $sLeadString = substr($string,0,$i);
1085                         $aLeadString = explode(' ',$sLeadString);
1086                         $sToBeEncoded = array_pop($aLeadString);
1087                         $iEncStart = $i - strlen($sToBeEncoded);
1088                         $ret .= $sToBeEncoded;
1089                         $cur_l += strlen($sToBeEncoded);
1090                     }
1091                     $cur_l += 3;
1092                     /* first we add the encoded string that reached it's max size */
1093                     if ($cur_l > ($max_l-2)) {
1094                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1095                         $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1096                         $cur_l = 3;
1097                         $ret = '';
1098                         $iOffset = $i;
1099                         $iEncStart = $i;
1100                     }
1101                     $enc_init = true;
1102                     $ret .= sprintf("=%02X", $k);
1103                 } else {
1104                     if ($iEncStart !== false) {
1105                         $cur_l++;
1106                         if ($cur_l > $max_l) {
1107                             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1108                             $aRet[] = "=?$default_charset?Q?$ret?=";
1109                             $iEncStart = false;
1110                             $iOffset = $i;
1111                             $cur_l = 0;
1112                             $ret = '';
1113                         } else {
1114                             $ret .= $string[$i];
1115                         }
1116                     }
1117                 }
1118                 break;
1119         }
1120     }
1121
1122     if ($enc_init) {
1123         if ($iEncStart !== false) {
1124             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1125             $aRet[] = "=?$default_charset?Q?$ret?=";
1126         } else {
1127             $aRet[] = substr($string,$iOffset);
1128         }
1129         $string = implode('',$aRet);
1130     }
1131     return $string;
1132 }
1133
1134 /**
1135  * Encodes string according to rfc2047 B encoding header formating rules
1136  *
1137  * It is recommended way to encode headers with character sets that store
1138  * symbols in more than one byte.
1139  *
1140  * Function requires mbstring support. If required mbstring functions are missing,
1141  * function returns false and sets E_USER_WARNING level error message.
1142  *
1143  * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1144  * that mbstring functions will generate E_WARNING errors, if unsupported
1145  * character set is used. mb_encode_mimeheader function provided by php
1146  * mbstring extension is not used in order to get better control of header
1147  * encoding.
1148  *
1149  * Used php code functions - function_exists(), trigger_error(), strlen()
1150  * (is used with charset names and base64 strings). Used php mbstring
1151  * functions - mb_strlen and mb_substr.
1152  *
1153  * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1154  * encoding), rfc 2822 (header folding)
1155  *
1156  * @param string $string header string that must be encoded
1157  * @param string $charset character set. Must be supported by mbstring extension.
1158  * Use sq_mb_list_encodings() to detect supported charsets.
1159  * @return string string encoded according to rfc2047 B encoding formating rules
1160  * @since 1.5.1
1161  * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1162  * @todo Do we want to control max length of header?
1163  * @todo Do we want to control EOL (end-of-line) marker?
1164  * @todo Do we want to translate error message?
1165  */
1166 function encodeHeaderBase64($string,$charset) {
1167     /**
1168      * Check mbstring function requirements.
1169      */
1170     if (! function_exists('mb_strlen') ||
1171         ! function_exists('mb_substr')) {
1172         // set E_USER_WARNING
1173         trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1174         // return false
1175         return false;
1176     }
1177
1178     // initial return array
1179     $aRet = array();
1180
1181     /**
1182      * header length = 75 symbols max (same as in encodeHeader)
1183      * remove $charset length
1184      * remove =? ? ?= (5 chars)
1185      * remove 2 more chars (\r\n ?)
1186      */
1187     $iMaxLength = 75 - strlen($charset) - 7;
1188
1189     // set first character position
1190     $iStartCharNum = 0;
1191
1192     // loop through all characters. count characters and not bytes.
1193     for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1194         // encode string from starting character to current character.
1195         $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1196
1197         // Check encoded string length
1198         if(strlen($encoded_string)>$iMaxLength) {
1199             // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1200             $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1201
1202             // set new starting character
1203             $iStartCharNum = $iCharNum-1;
1204
1205             // encode last char (in case it is last character in string)
1206             $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1207         } // if string is shorter than max length - add next character
1208     }
1209
1210     // add last encoded string to array
1211     $aRet[] = $encoded_string;
1212
1213     // set initial return string
1214     $sRet = '';
1215
1216     // loop through encoded strings
1217     foreach($aRet as $string) {
1218         // TODO: Do we want to control EOL (end-of-line) marker
1219         if ($sRet!='') $sRet.= " ";
1220
1221         // add header tags and encoded string to return string
1222         $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1223     }
1224
1225     return $sRet;
1226 }
1227
1228 /* This function trys to locate the entity_id of a specific mime element */
1229 function find_ent_id($id, $message) {
1230     for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1231         if ($message->entities[$i]->header->type0 == 'multipart')  {
1232             $ret = find_ent_id($id, $message->entities[$i]);
1233         } else {
1234             if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1235 //                if (sq_check_save_extension($message->entities[$i])) {
1236                 return $message->entities[$i]->entity_id;
1237 //                }
1238             } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1239                 /**
1240                  * This is part of a fix for Outlook Express 6.x generating
1241                  * cid URLs without creating content-id headers
1242                  * @@JA - 20050207
1243                  */
1244                 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1245                     return $message->entities[$i]->entity_id;
1246                 }
1247             }
1248         }
1249     }
1250     return $ret;
1251 }
1252
1253 function sq_check_save_extension($message) {
1254     $filename = $message->getFilename();
1255     $ext = substr($filename, strrpos($filename,'.')+1);
1256     $save_extensions = array('jpg','jpeg','gif','png','bmp');
1257     return in_array($ext, $save_extensions);
1258 }
1259
1260
1261 /**
1262  ** HTMLFILTER ROUTINES
1263  */
1264
1265 /**
1266  * This function checks attribute values for entity-encoded values
1267  * and returns them translated into 8-bit strings so we can run
1268  * checks on them.
1269  *
1270  * @param  $attvalue A string to run entity check against.
1271  * @return           Nothing, modifies a reference value.
1272  */
1273 function sq_defang(&$attvalue){
1274     $me = 'sq_defang';
1275     /**
1276      * Skip this if there aren't ampersands or backslashes.
1277      */
1278     if (strpos($attvalue, '&') === false
1279         && strpos($attvalue, '\\') === false){
1280         return;
1281     }
1282     $m = false;
1283     // before deent, translate the dangerous unicode characters and ... to safe values
1284     // otherwise the regular expressions do not match.
1285
1286
1287
1288     do {
1289         $m = false;
1290         $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1291         $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1292         $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1293     } while ($m == true);
1294     $attvalue = stripslashes($attvalue);
1295 }
1296
1297 /**
1298  * Kill any tabs, newlines, or carriage returns. Our friends the
1299  * makers of the browser with 95% market value decided that it'd
1300  * be funny to make "java[tab]script" be just as good as "javascript".
1301  *
1302  * @param  attvalue  The attribute value before extraneous spaces removed.
1303  * @return attvalue  Nothing, modifies a reference value.
1304  */
1305 function sq_unspace(&$attvalue){
1306     $me = 'sq_unspace';
1307     if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1308         $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1309                                 Array('',   '',   '',   '',   ''), $attvalue);
1310     }
1311 }
1312
1313 /**
1314  * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1315  * IE as regular characters.
1316  *
1317  * @param  attvalue  The attribute value before dangerous characters are translated.
1318  * @return attvalue  Nothing, modifies a reference value.
1319  * @author Marc Groot Koerkamp.
1320  */
1321 function sq_fixIE_idiocy(&$attvalue) {
1322     // remove NUL
1323     $attvalue = str_replace("\0", "", $attvalue);
1324     // remove comments
1325     $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1326
1327     // IE has the evil habit of accepting every possible value for the attribute expression.
1328     // The table below contains characters which are parsed by IE if they are used in the "expression"
1329     // attribute value.
1330     $aDangerousCharsReplacementTable = array(
1331                         array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1332                               '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1333                               '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1334                               '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1335                               '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1336                               '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1337                               '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1338                               '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1339                               '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1340                               '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1341                               '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1342                               '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1343                               '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1344                               '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1345                               '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1346                               '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1347                               '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1348                               '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1349                               '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1350                               '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1351                               '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1352                               '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1353                               '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1354                               '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1355                               "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */   // in unicode this is some Chinese char range
1356                               "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1357                               "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1358                               "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1359                               "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1360                               "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1361                               "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1362                               "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1363                               "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1364                               "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1365                               "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1366                               "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1367                               "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1368                               "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1369                               "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1370                               "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1371                               "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1372                               "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1373                               "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1374                               "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1375                               "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1376                               "\xCA\x9F", /* L UNICODE IPA Extension */
1377                               "\xCA\x80", /* R UNICODE IPA Extension */
1378                               "\xC9\xB4"),  /* N UNICODE IPA Extension */
1379                        array('l', 'l', 'r','r','n','n',
1380                              'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1381                              'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1382                              'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1383     $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1384
1385     // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1386     // used for XSS.
1387     $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1388 }
1389
1390 /**
1391  * This function returns the final tag out of the tag name, an array
1392  * of attributes, and the type of the tag. This function is called by
1393  * sq_sanitize internally.
1394  *
1395  * @param  $tagname  the name of the tag.
1396  * @param  $attary   the array of attributes and their values
1397  * @param  $tagtype  The type of the tag (see in comments).
1398  * @return           a string with the final tag representation.
1399  */
1400 function sq_tagprint($tagname, $attary, $tagtype){
1401     $me = 'sq_tagprint';
1402
1403     if ($tagtype == 2){
1404         $fulltag = '</' . $tagname . '>';
1405     } else {
1406         $fulltag = '<' . $tagname;
1407         if (is_array($attary) && sizeof($attary)){
1408             $atts = Array();
1409             foreach ($attary as $attname => $attvalue){
1410                 array_push($atts, "$attname=$attvalue");
1411             }
1412             $fulltag .= ' ' . join(" ", $atts);
1413         }
1414         if ($tagtype == 3){
1415             $fulltag .= ' /';
1416         }
1417         $fulltag .= '>';
1418     }
1419     return $fulltag;
1420 }
1421
1422 /**
1423  * A small helper function to use with array_walk. Modifies a by-ref
1424  * value and makes it lowercase.
1425  *
1426  * @param  $val a value passed by-ref.
1427  * @return      void since it modifies a by-ref value.
1428  */
1429 function sq_casenormalize(&$val){
1430     $val = strtolower($val);
1431 }
1432
1433 /**
1434  * This function skips any whitespace from the current position within
1435  * a string and to the next non-whitespace value.
1436  *
1437  * @param  $body   the string
1438  * @param  $offset the offset within the string where we should start
1439  *                 looking for the next non-whitespace character.
1440  * @return         the location within the $body where the next
1441  *                 non-whitespace char is located.
1442  */
1443 function sq_skipspace($body, $offset){
1444     $me = 'sq_skipspace';
1445     preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1446     if (!empty($matches[1])){
1447         $offset += strlen($matches[1]);
1448     }
1449     return $offset;
1450 }
1451
1452 /**
1453  * This function looks for the next character within a string.  It's
1454  * really just a glorified "strpos", except it catches if failures
1455  * nicely.
1456  *
1457  * @param  $body   The string to look for needle in.
1458  * @param  $offset Start looking from this position.
1459  * @param  $needle The character/string to look for.
1460  * @return         location of the next occurance of the needle, or
1461  *                 strlen($body) if needle wasn't found.
1462  */
1463 function sq_findnxstr($body, $offset, $needle){
1464     $me  = 'sq_findnxstr';
1465     $pos = strpos($body, $needle, $offset);
1466     if ($pos === FALSE){
1467         $pos = strlen($body);
1468     }
1469     return $pos;
1470 }
1471
1472 /**
1473  * This function takes a PCRE-style regexp and tries to match it
1474  * within the string.
1475  *
1476  * @param  $body   The string to look for needle in.
1477  * @param  $offset Start looking from here.
1478  * @param  $reg    A PCRE-style regex to match.
1479  * @return         Returns a false if no matches found, or an array
1480  *                 with the following members:
1481  *                 - integer with the location of the match within $body
1482  *                 - string with whatever content between offset and the match
1483  *                 - string with whatever it is we matched
1484  */
1485 function sq_findnxreg($body, $offset, $reg){
1486     $me = 'sq_findnxreg';
1487     $matches = Array();
1488     $retarr = Array();
1489     preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1490     if (!isset($matches[0]) || !$matches[0]){
1491         $retarr = false;
1492     } else {
1493         $retarr[0] = $offset + strlen($matches[1]);
1494         $retarr[1] = $matches[1];
1495         $retarr[2] = $matches[2];
1496     }
1497     return $retarr;
1498 }
1499
1500 /**
1501  * This function looks for the next tag.
1502  *
1503  * @param  $body   String where to look for the next tag.
1504  * @param  $offset Start looking from here.
1505  * @return         false if no more tags exist in the body, or
1506  *                 an array with the following members:
1507  *                 - string with the name of the tag
1508  *                 - array with attributes and their values
1509  *                 - integer with tag type (1, 2, or 3)
1510  *                 - integer where the tag starts (starting "<")
1511  *                 - integer where the tag ends (ending ">")
1512  *                 first three members will be false, if the tag is invalid.
1513  */
1514 function sq_getnxtag($body, $offset){
1515     $me = 'sq_getnxtag';
1516     if ($offset > strlen($body)){
1517         return false;
1518     }
1519     $lt = sq_findnxstr($body, $offset, "<");
1520     if ($lt == strlen($body)){
1521         return false;
1522     }
1523     /**
1524      * We are here:
1525      * blah blah <tag attribute="value">
1526      * \---------^
1527      */
1528     $pos = sq_skipspace($body, $lt+1);
1529     if ($pos >= strlen($body)){
1530         return Array(false, false, false, $lt, strlen($body));
1531     }
1532     /**
1533      * There are 3 kinds of tags:
1534      * 1. Opening tag, e.g.:
1535      *    <a href="blah">
1536      * 2. Closing tag, e.g.:
1537      *    </a>
1538      * 3. XHTML-style content-less tag, e.g.:
1539      *    <img src="blah" />
1540      */
1541     $tagtype = false;
1542     switch (substr($body, $pos, 1)){
1543         case '/':
1544             $tagtype = 2;
1545             $pos++;
1546             break;
1547         case '!':
1548             /**
1549              * A comment or an SGML declaration.
1550              */
1551             if (substr($body, $pos+1, 2) == "--"){
1552                 $gt = strpos($body, "-->", $pos);
1553                 if ($gt === false){
1554                     $gt = strlen($body);
1555                 } else {
1556                     $gt += 2;
1557                 }
1558                 return Array(false, false, false, $lt, $gt);
1559             } else {
1560                 $gt = sq_findnxstr($body, $pos, ">");
1561                 return Array(false, false, false, $lt, $gt);
1562             }
1563             break;
1564         default:
1565             /**
1566              * Assume tagtype 1 for now. If it's type 3, we'll switch values
1567              * later.
1568              */
1569             $tagtype = 1;
1570             break;
1571     }
1572
1573     $tag_start = $pos;
1574     $tagname = '';
1575     /**
1576      * Look for next [\W-_], which will indicate the end of the tag name.
1577      */
1578     $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1579     if ($regary == false){
1580         return Array(false, false, false, $lt, strlen($body));
1581     }
1582     list($pos, $tagname, $match) = $regary;
1583     $tagname = strtolower($tagname);
1584
1585     /**
1586      * $match can be either of these:
1587      * '>'  indicating the end of the tag entirely.
1588      * '\s' indicating the end of the tag name.
1589      * '/'  indicating that this is type-3 xhtml tag.
1590      *
1591      * Whatever else we find there indicates an invalid tag.
1592      */
1593     switch ($match){
1594         case '/':
1595             /**
1596              * This is an xhtml-style tag with a closing / at the
1597              * end, like so: <img src="blah" />. Check if it's followed
1598              * by the closing bracket. If not, then this tag is invalid
1599              */
1600             if (substr($body, $pos, 2) == "/>"){
1601                 $pos++;
1602                 $tagtype = 3;
1603             } else {
1604                 $gt = sq_findnxstr($body, $pos, ">");
1605                 $retary = Array(false, false, false, $lt, $gt);
1606                 return $retary;
1607             }
1608         case '>':
1609             return Array($tagname, false, $tagtype, $lt, $pos);
1610             break;
1611         default:
1612             /**
1613              * Check if it's whitespace
1614              */
1615             if (!preg_match('/\s/', $match)){
1616                 /**
1617                  * This is an invalid tag! Look for the next closing ">".
1618                  */
1619                 $gt = sq_findnxstr($body, $lt, ">");
1620                 return Array(false, false, false, $lt, $gt);
1621             }
1622             break;
1623     }
1624
1625     /**
1626      * At this point we're here:
1627      * <tagname  attribute='blah'>
1628      * \-------^
1629      *
1630      * At this point we loop in order to find all attributes.
1631      */
1632     $attname = '';
1633     $atttype = false;
1634     $attary = Array();
1635
1636     while ($pos <= strlen($body)){
1637         $pos = sq_skipspace($body, $pos);
1638         if ($pos == strlen($body)){
1639             /**
1640              * Non-closed tag.
1641              */
1642             return Array(false, false, false, $lt, $pos);
1643         }
1644         /**
1645          * See if we arrived at a ">" or "/>", which means that we reached
1646          * the end of the tag.
1647          */
1648         $matches = Array();
1649         if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1650             /**
1651              * Yep. So we did.
1652              */
1653             $pos += strlen($matches[1]);
1654             if ($matches[2] == "/>"){
1655                 $tagtype = 3;
1656                 $pos++;
1657             }
1658             return Array($tagname, $attary, $tagtype, $lt, $pos);
1659         }
1660
1661         /**
1662          * There are several types of attributes, with optional
1663          * [:space:] between members.
1664          * Type 1:
1665          *   attrname[:space:]=[:space:]'CDATA'
1666          * Type 2:
1667          *   attrname[:space:]=[:space:]"CDATA"
1668          * Type 3:
1669          *   attr[:space:]=[:space:]CDATA
1670          * Type 4:
1671          *   attrname
1672          *
1673          * We leave types 1 and 2 the same, type 3 we check for
1674          * '"' and convert to "&quot" if needed, then wrap in
1675          * double quotes. Type 4 we convert into:
1676          * attrname="yes".
1677          */
1678         $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1679         if ($regary == false){
1680             /**
1681              * Looks like body ended before the end of tag.
1682              */
1683             return Array(false, false, false, $lt, strlen($body));
1684         }
1685         list($pos, $attname, $match) = $regary;
1686         $attname = strtolower($attname);
1687         /**
1688          * We arrived at the end of attribute name. Several things possible
1689          * here:
1690          * '>'  means the end of the tag and this is attribute type 4
1691          * '/'  if followed by '>' means the same thing as above
1692          * '\s' means a lot of things -- look what it's followed by.
1693          *      anything else means the attribute is invalid.
1694          */
1695         switch($match){
1696             case '/':
1697                 /**
1698                  * This is an xhtml-style tag with a closing / at the
1699                  * end, like so: <img src="blah" />. Check if it's followed
1700                  * by the closing bracket. If not, then this tag is invalid
1701                  */
1702                 if (substr($body, $pos, 2) == "/>"){
1703                     $pos++;
1704                     $tagtype = 3;
1705                 } else {
1706                     $gt = sq_findnxstr($body, $pos, ">");
1707                     $retary = Array(false, false, false, $lt, $gt);
1708                     return $retary;
1709                 }
1710             case '>':
1711                 $attary[$attname] = '"yes"';
1712                 return Array($tagname, $attary, $tagtype, $lt, $pos);
1713                 break;
1714             default:
1715                 /**
1716                  * Skip whitespace and see what we arrive at.
1717                  */
1718                 $pos = sq_skipspace($body, $pos);
1719                 $char = substr($body, $pos, 1);
1720                 /**
1721                  * Two things are valid here:
1722                  * '=' means this is attribute type 1 2 or 3.
1723                  * \w means this was attribute type 4.
1724                  * anything else we ignore and re-loop. End of tag and
1725                  * invalid stuff will be caught by our checks at the beginning
1726                  * of the loop.
1727                  */
1728                 if ($char == "="){
1729                     $pos++;
1730                     $pos = sq_skipspace($body, $pos);
1731                     /**
1732                      * Here are 3 possibilities:
1733                      * "'"  attribute type 1
1734                      * '"'  attribute type 2
1735                      * everything else is the content of tag type 3
1736                      */
1737                     $quot = substr($body, $pos, 1);
1738                     if ($quot == "'"){
1739                         $regary = sq_findnxreg($body, $pos+1, "\'");
1740                         if ($regary == false){
1741                             return Array(false, false, false, $lt, strlen($body));
1742                         }
1743                         list($pos, $attval, $match) = $regary;
1744                         $pos++;
1745                         $attary[$attname] = "'" . $attval . "'";
1746                     } else if ($quot == '"'){
1747                         $regary = sq_findnxreg($body, $pos+1, '\"');
1748                         if ($regary == false){
1749                             return Array(false, false, false, $lt, strlen($body));
1750                         }
1751                         list($pos, $attval, $match) = $regary;
1752                         $pos++;
1753                         $attary[$attname] = '"' . $attval . '"';
1754                     } else {
1755                         /**
1756                          * These are hateful. Look for \s, or >.
1757                          */
1758                         $regary = sq_findnxreg($body, $pos, "[\s>]");
1759                         if ($regary == false){
1760                             return Array(false, false, false, $lt, strlen($body));
1761                         }
1762                         list($pos, $attval, $match) = $regary;
1763                         /**
1764                          * If it's ">" it will be caught at the top.
1765                          */
1766                         $attval = preg_replace("/\"/s", "&quot;", $attval);
1767                         $attary[$attname] = '"' . $attval . '"';
1768                     }
1769                 } else if (preg_match("|[\w/>]|", $char)) {
1770                     /**
1771                      * That was attribute type 4.
1772                      */
1773                     $attary[$attname] = '"yes"';
1774                 } else {
1775                     /**
1776                      * An illegal character. Find next '>' and return.
1777                      */
1778                     $gt = sq_findnxstr($body, $pos, ">");
1779                     return Array(false, false, false, $lt, $gt);
1780                 }
1781                 break;
1782         }
1783     }
1784     /**
1785      * The fact that we got here indicates that the tag end was never
1786      * found. Return invalid tag indication so it gets stripped.
1787      */
1788     return Array(false, false, false, $lt, strlen($body));
1789 }
1790
1791 /**
1792  * Translates entities into literal values so they can be checked.
1793  *
1794  * @param $attvalue the by-ref value to check.
1795  * @param $regex    the regular expression to check against.
1796  * @param $hex      whether the entites are hexadecimal.
1797  * @return          True or False depending on whether there were matches.
1798  */
1799 function sq_deent(&$attvalue, $regex, $hex=false){
1800     $me = 'sq_deent';
1801     $ret_match = false;
1802     // remove comments
1803     //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1804     preg_match_all($regex, $attvalue, $matches);
1805     if (is_array($matches) && sizeof($matches[0]) > 0){
1806         $repl = Array();
1807         for ($i = 0; $i < sizeof($matches[0]); $i++){
1808             $numval = $matches[1][$i];
1809             if ($hex){
1810                 $numval = hexdec($numval);
1811             }
1812             $repl[$matches[0][$i]] = chr($numval);
1813         }
1814         $attvalue = strtr($attvalue, $repl);
1815         return true;
1816     } else {
1817         return false;
1818     }
1819 }
1820
1821 /**
1822  * This function runs various checks against the attributes.
1823  *
1824  * @param  $tagname         String with the name of the tag.
1825  * @param  $attary          Array with all tag attributes.
1826  * @param  $rm_attnames     See description for sq_sanitize
1827  * @param  $bad_attvals     See description for sq_sanitize
1828  * @param  $add_attr_to_tag See description for sq_sanitize
1829  * @param  $message         message object
1830  * @param  $id              message id
1831  * @return                  Array with modified attributes.
1832  */
1833 function sq_fixatts($tagname,
1834                     $attary,
1835                     $rm_attnames,
1836                     $bad_attvals,
1837                     $add_attr_to_tag,
1838                     $message,
1839                     $id,
1840                     $mailbox
1841                     ){
1842     $me = 'sq_fixatts';
1843     foreach ($attary as $attname => $attvalue){
1844         /**
1845          * See if this attribute should be removed.
1846          */
1847         foreach ($rm_attnames as $matchtag=>$matchattrs){
1848             if (preg_match($matchtag, $tagname)){
1849                 foreach ($matchattrs as $matchattr){
1850                     if (preg_match($matchattr, $attname)){
1851                         unset($attary[$attname]);
1852                         continue;
1853                     }
1854                 }
1855             }
1856         }
1857         /**
1858          * Workaround for IE quirks
1859          */
1860         sq_fixIE_idiocy($attvalue);
1861
1862         /**
1863          * Remove any backslashes, entities, and extraneous whitespace.
1864          */
1865
1866         $oldattvalue = $attvalue;
1867         sq_defang($attvalue);
1868         if ($attname == 'style' && $attvalue !== $oldattvalue) {
1869             // entities are used in the attribute value. In 99% of the cases it's there as XSS
1870             // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1871             $attvalue = "idiocy";
1872             $attary[$attname] = $attvalue;
1873         }
1874         sq_unspace($attvalue);
1875
1876         /**
1877          * Now let's run checks on the attvalues.
1878          * I don't expect anyone to comprehend this. If you do,
1879          * get in touch with me so I can drive to where you live and
1880          * shake your hand personally. :)
1881          */
1882         foreach ($bad_attvals as $matchtag=>$matchattrs){
1883             if (preg_match($matchtag, $tagname)){
1884                 foreach ($matchattrs as $matchattr=>$valary){
1885                     if (preg_match($matchattr, $attname)){
1886                         /**
1887                          * There are two arrays in valary.
1888                          * First is matches.
1889                          * Second one is replacements
1890                          */
1891                         list($valmatch, $valrepl) = $valary;
1892                         $newvalue =
1893                             preg_replace($valmatch, $valrepl, $attvalue);
1894                         if ($newvalue != $attvalue){
1895                             $attary[$attname] = $newvalue;
1896                             $attvalue = $newvalue;
1897                         }
1898                     }
1899                 }
1900             }
1901         }
1902         if ($attname == 'style') {
1903             if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1904                 // 8bit and control characters in style attribute values can be used for XSS, remove them
1905                 $attary[$attname] = '"disallowed character"';
1906             }
1907             preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1908             if (count($aMatch)) {
1909                 foreach($aMatch[1] as $sMatch) {
1910                     // url value
1911                     $urlvalue = $sMatch;
1912                     sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1913                     $attary[$attname] = str_replace($sMatch,$urlvalue,$attvalue);
1914                 }
1915             }
1916         }
1917         /**
1918          * Use white list based filtering on attributes which can contain url's
1919          */
1920         else if ($attname == 'href' || $attname == 'xlink:href' || $attname == 'src'
1921               || $attname == 'poster' || $attname == 'formaction'
1922               || $attname == 'background' || $attname == 'action') {
1923             sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1924             $attary[$attname] = $attvalue;
1925         }
1926     }
1927     /**
1928      * See if we need to append any attributes to this tag.
1929      */
1930     foreach ($add_attr_to_tag as $matchtag=>$addattary){
1931         if (preg_match($matchtag, $tagname)){
1932             $attary = array_merge($attary, $addattary);
1933         }
1934     }
1935     return $attary;
1936 }
1937
1938 /**
1939  * This function filters url's
1940  *
1941  * @param  $attvalue        String with attribute value to filter
1942  * @param  $message         message object
1943  * @param  $id               message id
1944  * @param  $mailbox         mailbox
1945  * @param  $sQuote          quoting characters around url's
1946  */
1947 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1948     $attvalue = trim($attvalue);
1949     if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1950         // remove the double quotes
1951         $sQuote = $attvalue[0];
1952         $attvalue = trim(substr($attvalue,1,-1));
1953     }
1954
1955     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1956     // images off by default.
1957     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1958
1959     global $use_transparent_security_image;
1960     if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1961     else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1962
1963     /**
1964      * Replace empty src tags with the blank image.  src is only used
1965      * for frames, images, and image inputs.  Doing a replace should
1966      * not affect them working as should be, however it will stop
1967      * IE from being kicked off when src for img tags are not set
1968      */
1969     if ($attvalue == '') {
1970         $attvalue = '"' . SM_PATH . 'images/blank.png"';
1971     } else {
1972         // first, disallow 8 bit characters and control characters
1973         if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1974             switch ($attname) {
1975                 case 'href':
1976                     $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1977                     break;
1978                 default:
1979                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1980                     break;
1981             }
1982         } else {
1983             $aUrl = parse_url($attvalue);
1984             if (isset($aUrl['scheme'])) {
1985                 switch(strtolower($aUrl['scheme'])) {
1986                     case 'mailto':
1987                     case 'http':
1988                     case 'https':
1989                     case 'ftp':
1990                         if ($attname != 'href') {
1991                             if ($view_unsafe_images == false) {
1992                                 $attvalue = $sQuote . $secremoveimg . $sQuote;
1993                             } else {
1994                                 if (isset($aUrl['path'])) {
1995
1996                                     // No one has been able to show that image URIs
1997                                     // can be exploited, so for now, no restrictions
1998                                     // are made at all.  If this proves to be a problem,
1999                                     // the commented-out code below can be of help.
2000                                     // (One consideration is that I see nothing in this
2001                                     // function that specifically says that we will
2002                                     // only ever arrive here when inspecting an image
2003                                     // tag, although that does seem to be the end
2004                                     // result - e.g., <script src="..."> where malicious
2005                                     // image URIs are in fact a problem are already
2006                                     // filtered out elsewhere.
2007                                     /* ---------------------------------
2008                                     // validate image extension.
2009                                     $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
2010                                     if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
2011                                         // If URI is to something other than
2012                                         // a regular image file, get the contents
2013                                         // and try to see if it is an image.
2014                                         // Don't use Fileinfo (finfo_file()) because
2015                                         // we'd need to make the admin configure the
2016                                         // location of the magic.mime file (FIXME: add finfo_file() support later?)
2017                                         //
2018                                         $mime_type = '';
2019                                         if (function_exists('mime_content_type')
2020                                          && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
2021
2022                                             // fetch file
2023                                             //
2024                                             $file_contents = '';
2025                                             while (!feof($FILE)) {
2026                                                 $file_contents .= fread($FILE, 8192);
2027                                             }
2028                                             fclose($FILE);
2029
2030                                             // store file locally
2031                                             //
2032                                             global $attachment_dir, $username;
2033                                             $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
2034                                             $localfilename = GenerateRandomString(32, '', 7);
2035                                             $full_localfilename = "$hashed_attachment_dir/$localfilename";
2036                                             while (file_exists($full_localfilename)) {
2037                                                 $localfilename = GenerateRandomString(32, '', 7);
2038                                                 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2039                                             }
2040                                             $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2041                                             fwrite($FILE, $file_contents);
2042                                             fclose($FILE);
2043
2044                                             // get mime type and remove file
2045                                             //
2046                                             $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2047                                             unlink("$hashed_attachment_dir/$localfilename");
2048                                         }
2049                                         // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2050                                         if (substr(strtolower($mime_type), 0, 5) != 'image') {
2051                                             $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2052                                         }
2053                                     }
2054                                     --------------------------------- */
2055                                 } else {
2056                                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2057                                 }
2058                             }
2059                         } else {
2060                             $attvalue = $sQuote . $attvalue . $sQuote;
2061                         }
2062                         break;
2063                     case 'outbind':
2064                         /**
2065                          * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2066                          * One day MS might actually make it match something useful, for now, falling
2067                          * back to using cid2http, so we can grab the blank.png.
2068                          */
2069                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2070                         break;
2071                     case 'cid':
2072                         /**
2073                             * Turn cid: urls into http-friendly ones.
2074                             */
2075                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2076                         break;
2077                     default:
2078                         $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2079                         break;
2080                 }
2081             } else {
2082                 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2083                     // parse_url did not lead to satisfying result
2084                     $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2085                 }
2086             }
2087         }
2088     }
2089 }
2090
2091 /**
2092  * This function edits the style definition to make them friendly and
2093  * usable in SquirrelMail.
2094  *
2095  * @param  $message  the message object
2096  * @param  $id       the message id
2097  * @param  $content  a string with whatever is between <style> and </style>
2098  * @param  $mailbox  the message mailbox
2099  * @return           a string with edited content.
2100  */
2101 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2102     $me = 'sq_fixstyle';
2103     // workaround for </style> in between comments
2104     $iCurrentPos = $pos;
2105     $content = '';
2106     $sToken = '';
2107     $bSucces = false;
2108     $bEndTag = false;
2109     for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2110         $char = $body[$i];
2111         switch ($char) {
2112             case '<':
2113                 $sToken = $char;
2114                 break;
2115             case '/':
2116                  if ($sToken == '<') {
2117                     $sToken .= $char;
2118                     $bEndTag = true;
2119                  } else {
2120                     $content .= $char;
2121                  }
2122                  break;
2123             case '>':
2124                  if ($bEndTag) {
2125                     $sToken .= $char;
2126                     if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2127                         $newpos = $i + 1;
2128                         $bSucces = true;
2129                         break 2;
2130                     } else {
2131                         $content .= $sToken;
2132                     }
2133                     $bEndTag = false;
2134                  } else {
2135                     $content .= $char;
2136                  }
2137                  break;
2138             case '!':
2139                 if ($sToken == '<') {
2140                     // possible comment
2141                     if (isset($body[$i+2]) && substr($body,$i,3) == '!--') {
2142                         $i = strpos($body,'-->',$i+3);
2143                         if ($i === false) { // no end comment
2144                             $i = strlen($body);
2145                         }
2146                         $sToken = '';
2147                     }
2148                 } else {
2149                     $content .= $char;
2150                 }
2151                 break;
2152             default:
2153                 if ($bEndTag) {
2154                     $sToken .= $char;
2155                 } else {
2156                     $content .= $char;
2157                 }
2158                 break;
2159         }
2160     }
2161     if ($bSucces == FALSE){
2162         return array(FALSE, strlen($body));
2163     }
2164
2165
2166
2167     /**
2168      * First look for general BODY style declaration, which would be
2169      * like so:
2170      * body {background: blah-blah}
2171      * and change it to .bodyclass so we can just assign it to a <div>
2172      */
2173     // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2174     // Nah, this is even better - try to preface all CSS selectors with
2175     // our <div> class ID "bodyclass" then correct generic "body" selectors
2176     // TODO: this works pretty good but breaks stuff like this:
2177     //       @media print { body { font-size: 10pt; } }
2178     //       but there isn't an easy way to make this regex skip @media
2179     //       definitions... though lots of the ones in the wild will be
2180     //       correctly handled because they tend to end with a parenthesis, like:
2181     //       @media screen and (max-width:480px) { ...
2182     $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2183     $content = str_replace('.bodyclass body', '.bodyclass', $content);
2184
2185     global $use_transparent_security_image;
2186     if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2187     else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2188
2189     /**
2190     * Fix url('blah') declarations.
2191     */
2192     //   $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2193     //                           "url(\\1$secremoveimg\\2)", $content);
2194
2195     // first check for 8bit sequences and disallowed control characters
2196     if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2197         $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2198         return array($content, $newpos);
2199     }
2200
2201     // IE Sucks hard. We have a special function for it.
2202     sq_fixIE_idiocy($content);
2203
2204     // remove @import line
2205     $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2206
2207     // translate ur\l and variations (IE parses that)
2208     // TODO check if the sq_fixIE_idiocy function already handles this.
2209     $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2210     preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2211     if (count($aMatch)) {
2212         $aValue = $aReplace = array();
2213         foreach($aMatch[1] as $sMatch) {
2214             // url value
2215             $urlvalue = $sMatch;
2216             sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2217             $aValue[] = $sMatch;
2218             $aReplace[] = $urlvalue;
2219         }
2220         $content = str_replace($aValue,$aReplace,$content);
2221     }
2222
2223     /**
2224      * Remove any backslashes, entities, and extraneous whitespace.
2225      */
2226     $contentTemp = $content;
2227     sq_defang($contentTemp);
2228     sq_unspace($contentTemp);
2229
2230     /**
2231      * Fix stupid css declarations which lead to vulnerabilities
2232      * in IE.
2233      *
2234      * Also remove "position" attribute, as it can easily be set
2235      * to "fixed" or "absolute" with "left" and "top" attributes
2236      * of zero, taking over the whole content frame.  It can also
2237      * be set to relative and move itself anywhere it wants to,
2238      * displaying content in areas it shouldn't be allowed to touch.
2239      */
2240     $match   = Array('/\/\*.*\*\//', // removes /* blah blah */
2241                     '/expression/i',
2242                     '/behaviou*r/i',
2243                     '/binding/i',
2244                     '/include-source/i',
2245                     '/javascript/i',
2246                     '/script/i',
2247                     '/position/i');
2248     $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2249     $contentNew = preg_replace($match, $replace, $contentTemp);
2250     if ($contentNew !== $contentTemp) {
2251         // insecure css declarations are used. From now on we don't care
2252         // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2253         $content = $contentNew;
2254     }
2255     return array($content, $newpos);
2256 }
2257
2258
2259 /**
2260  * This function converts cid: url's into the ones that can be viewed in
2261  * the browser.
2262  *
2263  * @param  $message  the message object
2264  * @param  $id       the message id
2265  * @param  $cidurl   the cid: url.
2266  * @param  $mailbox  the message mailbox
2267  * @return           a string with a http-friendly url
2268  */
2269 function sq_cid2http($message, $id, $cidurl, $mailbox){
2270     /**
2271      * Get rid of quotes.
2272      */
2273     $quotchar = substr($cidurl, 0, 1);
2274     if ($quotchar == '"' || $quotchar == "'"){
2275         $cidurl = str_replace($quotchar, "", $cidurl);
2276     } else {
2277         $quotchar = '';
2278     }
2279     $cidurl = substr(trim($cidurl), 4);
2280
2281     $match_str = '/\{.*?\}\//';
2282     $str_rep = '';
2283     $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2284
2285     $linkurl = find_ent_id($cidurl, $message);
2286     /* in case of non-safe cid links $httpurl should be replaced by a sort of
2287        unsafe link image */
2288     $httpurl = '';
2289
2290     /**
2291      * This is part of a fix for Outlook Express 6.x generating
2292      * cid URLs without creating content-id headers. These images are
2293      * not part of the multipart/related html mail. The html contains
2294      * <img src="cid:{some_id}/image_filename.ext"> references to
2295      * attached images with as goal to render them inline although
2296      * the attachment disposition property is not inline.
2297      */
2298
2299     if (empty($linkurl)) {
2300         if (preg_match('/{.*}\//', $cidurl)) {
2301             $cidurl = preg_replace('/{.*}\//','', $cidurl);
2302             if (!empty($cidurl)) {
2303                 $linkurl = find_ent_id($cidurl, $message);
2304             }
2305         }
2306     }
2307
2308     if (!empty($linkurl)) {
2309         $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2310             "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2311             '&amp;ent_id=' . $linkurl . $quotchar;
2312     } else {
2313         /**
2314          * If we couldn't generate a proper img url, drop in a blank image
2315          * instead of sending back empty, otherwise it causes unusual behaviour
2316          */
2317         $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2318     }
2319
2320     return $httpurl;
2321 }
2322
2323 /**
2324  * This function changes the <body> tag into a <div> tag since we
2325  * can't really have a body-within-body.
2326  *
2327  * @param  $attary   an array of attributes and values of <body>
2328  * @param  $mailbox  mailbox we're currently reading (for cid2http)
2329  * @param  $message  current message (for cid2http)
2330  * @param  $id       current message id (for cid2http)
2331  * @return           a modified array of attributes to be set for <div>
2332  */
2333 function sq_body2div($attary, $mailbox, $message, $id){
2334     $me = 'sq_body2div';
2335     $divattary = Array('class' => "'bodyclass'");
2336     $text = '#000000';
2337     $has_bgc_stl = $has_txt_stl = false;
2338     $styledef = '';
2339     if (is_array($attary) && sizeof($attary) > 0){
2340         foreach ($attary as $attname=>$attvalue){
2341             $quotchar = substr($attvalue, 0, 1);
2342             $attvalue = str_replace($quotchar, "", $attvalue);
2343             switch ($attname){
2344                 case 'background':
2345                     $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2346                     $styledef .= "background-image: url('$attvalue'); ";
2347                     break;
2348                 case 'bgcolor':
2349                     $has_bgc_stl = true;
2350                     $styledef .= "background-color: $attvalue; ";
2351                     break;
2352                 case 'text':
2353                     $has_txt_stl = true;
2354                     $styledef .= "color: $attvalue; ";
2355                     break;
2356             }
2357         }
2358         // Outlook defines a white bgcolor and no text color. This can lead to
2359         // white text on a white bg with certain themes.
2360         if ($has_bgc_stl && !$has_txt_stl) {
2361             $styledef .= "color: $text; ";
2362         }
2363         if (strlen($styledef) > 0){
2364             $divattary["style"] = "\"$styledef\"";
2365         }
2366     }
2367     return $divattary;
2368 }
2369
2370 /**
2371  * This is the main function and the one you should actually be calling.
2372  * There are several variables you should be aware of an which need
2373  * special description.
2374  *
2375  * Since the description is quite lengthy, see it here:
2376  * http://linux.duke.edu/projects/mini/htmlfilter/
2377  *
2378  * @param $body                 the string with HTML you wish to filter
2379  * @param $tag_list             see description above
2380  * @param $rm_tags_with_content see description above
2381  * @param $self_closing_tags    see description above
2382  * @param $force_tag_closing    see description above
2383  * @param $rm_attnames          see description above
2384  * @param $bad_attvals          see description above
2385  * @param $add_attr_to_tag      see description above
2386  * @param $message              message object
2387  * @param $id                   message id
2388  * @param $recursively_called   boolean flag for recursive calls into this function (optional; default FALSE)
2389  * @return                      sanitized html safe to show on your pages.
2390  */
2391 function sq_sanitize($body,
2392                      $tag_list,
2393                      $rm_tags_with_content,
2394                      $self_closing_tags,
2395                      $force_tag_closing,
2396                      $rm_attnames,
2397                      $bad_attvals,
2398                      $add_attr_to_tag,
2399                      $message,
2400                      $id,
2401                      $mailbox,
2402                      $recursively_called=FALSE
2403                      ){
2404     $me = 'sq_sanitize';
2405
2406     /**
2407      * See if tag_list is of tags to remove or tags to allow.
2408      * false  means remove these tags
2409      * true   means allow these tags
2410      */
2411     $orig_tag_list = $tag_list;
2412     $rm_tags = array_shift($tag_list);
2413
2414     /**
2415      * Normalize rm_tags and rm_tags_with_content.
2416      */
2417     @array_walk($tag_list, 'sq_casenormalize');
2418     @array_walk($rm_tags_with_content, 'sq_casenormalize');
2419     @array_walk($self_closing_tags, 'sq_casenormalize');
2420
2421     $curpos = 0;
2422     $open_tags = Array();
2423     $trusted = "\n<!-- begin sanitized html -->\n";
2424     $skip_content = false;
2425     /**
2426      * Take care of netscape's stupid javascript entities like
2427      * &{alert('boo')};
2428      */
2429     $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2430
2431     while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2432         list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2433
2434         /**
2435          * RCDATA and RAWTEXT tags are handled differently:
2436          * next instance of closing tag is used, whether or not
2437          * the HTML is well formed before that
2438          */
2439         global $rcdata_rawtext_tags;
2440         if (!$recursively_called
2441          && in_array($tagname, $rcdata_rawtext_tags)
2442          && $tagtype === 1){
2443             $closing_tag = false;
2444             $closing_tag_offset = $curpos;
2445             // seek out the closing tag for the current RCDATA/RAWTEXT tag
2446             while (1) {
2447                 // first we need to move forward to next available closing tag
2448                 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2449                 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2450                 if ($next_tag === false) {
2451                     $closing_tag = false;
2452                     break;
2453                 }
2454                 // but then we have to make sure it's a well-formed tag
2455                 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2456                 if ($closing_tag === false)
2457                     break;
2458                 else if ($closing_tag[0] !== false
2459                  // these should be redundant
2460                  && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2461                     $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos + 1),
2462                                             $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2463                                             $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2464                                             $message, $id, $mailbox, true);
2465                     $curpos = $closing_tag[4] + 1;
2466                     continue 2;
2467                 }
2468                 $closing_tag_offset = $next_tag[0] + 1;
2469             }
2470             if ($closing_tag === false)
2471             { /* no-op... there was no closing tag for this RCDATA/RAWTEXT tag - we could probably set $curpos to the end of $body, but this HTML is malformed anyway and should just fall apart on its own */ }
2472         }
2473
2474         $free_content = substr($body, $curpos, $lt-$curpos);
2475         /**
2476          * Take care of <style>
2477          */
2478         if ($tagname == "style" && $tagtype == 1){
2479             list($free_content, $curpos) =
2480                 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2481             if ($free_content != FALSE){
2482                 if ( !empty($attary) ) {
2483                     $attary = sq_fixatts($tagname,
2484                                          $attary,
2485                                          $rm_attnames,
2486                                          $bad_attvals,
2487                                          $add_attr_to_tag,
2488                                          $message,
2489                                          $id,
2490                                          $mailbox
2491                                          );
2492                 }
2493                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2494                 $trusted .= $free_content;
2495                 $trusted .= sq_tagprint($tagname, false, 2);
2496             }
2497             continue;
2498         }
2499         if ($skip_content == false){
2500             $trusted .= $free_content;
2501         }
2502         if ($tagname != FALSE){
2503             if ($tagtype == 2){
2504                 if ($skip_content == $tagname){
2505                     /**
2506                      * Got to the end of tag we needed to remove.
2507                      */
2508                     $tagname = false;
2509                     $skip_content = false;
2510                 } else {
2511                     if ($skip_content == false){
2512                         if ($tagname == "body"){
2513                             $tagname = "div";
2514                         }
2515                         if (isset($open_tags[$tagname]) &&
2516                                 $open_tags[$tagname] > 0){
2517                             $open_tags[$tagname]--;
2518                         } else {
2519                             $tagname = false;
2520                         }
2521                     }
2522                 }
2523             } else {
2524                 /**
2525                  * $rm_tags_with_content
2526                  */
2527                 if ($skip_content == false){
2528                     /**
2529                      * See if this is a self-closing type and change
2530                      * tagtype appropriately.
2531                      */
2532                     if ($tagtype == 1
2533                             && in_array($tagname, $self_closing_tags)){
2534                         $tagtype = 3;
2535                     }
2536                     /**
2537                      * See if we should skip this tag and any content
2538                      * inside it.
2539                      */
2540                     if ($tagtype == 1 &&
2541                             in_array($tagname, $rm_tags_with_content)){
2542                         $skip_content = $tagname;
2543                     } else {
2544                         if (($rm_tags == false
2545                                     && in_array($tagname, $tag_list)) ||
2546                                 ($rm_tags == true &&
2547                                  !in_array($tagname, $tag_list))){
2548                             $tagname = false;
2549                         } else {
2550                             /**
2551                              * Convert body into div.
2552                              */
2553                             if ($tagname == "body"){
2554                                 $tagname = "div";
2555                                 $attary = sq_body2div($attary, $mailbox,
2556                                         $message, $id);
2557                             }
2558                             if ($tagtype == 1){
2559                                 if (isset($open_tags[$tagname])){
2560                                     $open_tags[$tagname]++;
2561                                 } else {
2562                                     $open_tags[$tagname]=1;
2563                                 }
2564                             }
2565                             /**
2566                              * This is where we run other checks.
2567                              */
2568                             if (is_array($attary) && sizeof($attary) > 0){
2569                                 $attary = sq_fixatts($tagname,
2570                                                      $attary,
2571                                                      $rm_attnames,
2572                                                      $bad_attvals,
2573                                                      $add_attr_to_tag,
2574                                                      $message,
2575                                                      $id,
2576                                                      $mailbox
2577                                                      );
2578                             }
2579                         }
2580                     }
2581                 }
2582             }
2583             if ($tagname != false && $skip_content == false){
2584                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2585             }
2586         }
2587         $curpos = $gt+1;
2588     }
2589     $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2590     if ($force_tag_closing == true){
2591         foreach ($open_tags as $tagname=>$opentimes){
2592             while ($opentimes > 0){
2593                 $trusted .= '</' . $tagname . '>';
2594                 $opentimes--;
2595             }
2596         }
2597         $trusted .= "\n";
2598     }
2599     $trusted .= "<!-- end sanitized html -->\n";
2600     return $trusted;
2601 }
2602
2603 /**
2604  * This is a wrapper function to call html sanitizing routines.
2605  *
2606  * @param  $body  the body of the message
2607  * @param  $id    the id of the message
2608
2609  * @param  $message
2610  * @param  $mailbox
2611  * @param  boolean $take_mailto_links When TRUE, converts mailto: links
2612  *                                    into internal SM compose links
2613  *                                    (optional; default = TRUE)
2614  * @return        a string with html safe to display in the browser.
2615  */
2616 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2617
2618     // require_once(SM_PATH . 'functions/url_parser.php');  // for $MailTo_PReg_Match
2619
2620     global $attachment_common_show_images, $view_unsafe_images,
2621            $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2622            $remove_rcdata_rawtext_tags_and_content;
2623
2624     $rcdata_rawtext_tags = array(
2625         "noscript",
2626         "noframes",
2627         "noembed",
2628         "textarea",
2629         // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2630     );
2631
2632     /**
2633      * Don't display attached images in HTML mode.
2634      *
2635      * SB: why?
2636      */
2637     $attachment_common_show_images = false;
2638     $tag_list = Array(
2639             false, // remove these tags
2640             "meta",
2641             "html",
2642             "head",
2643             "base",
2644             "link",
2645             "frame",
2646             "iframe",
2647             "plaintext",
2648             "marquee",
2649             );
2650
2651     $rm_tags_with_content = Array(
2652             "script",
2653             "object",
2654             "applet",
2655             "embed",
2656             "title",
2657             "frameset",
2658             "xmp",
2659             "xml",
2660             );
2661     if (!$allow_svg_display)
2662         $rm_tags_with_content[] = 'svg';
2663     /**
2664      * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2665      * case that they are, but if you prefer to remove them and their contents entirely
2666      * (in most cases, should be a safe thing with minimal impact), you can add the
2667      * following to config/config_local.php
2668      *    $remove_rcdata_rawtext_tags_and_content = TRUE;
2669      */
2670     if ($remove_rcdata_rawtext_tags_and_content)
2671         $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2672
2673     $self_closing_tags =  Array(
2674             "img",
2675             "br",
2676             "hr",
2677             "input",
2678             "outbind",
2679             );
2680
2681     $force_tag_closing = true;
2682
2683     $rm_attnames = Array(
2684             "/.*/" =>
2685             Array(
2686                 "/target/i",
2687                 "/^on.*/i",
2688                 "/^dynsrc/i",
2689                 "/^data.*/i",
2690                 "/^lowsrc.*/i",
2691                 )
2692             );
2693
2694     global $use_transparent_security_image;
2695     if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2696     else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2697
2698     $bad_attvals = Array(
2699             "/.*/" =>
2700             Array(
2701                 "/^src|background/i" =>
2702                 Array(
2703                     Array(
2704                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2705                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2706                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2707                         ),
2708                     Array(
2709                         "\\1$secremoveimg\\2",
2710                         "\\1$secremoveimg\\2",
2711                         "\\1$secremoveimg\\2",
2712                         )
2713                     ),
2714                 "/^href|action/i" =>
2715                 Array(
2716                     Array(
2717                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2718                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2719                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2720                         ),
2721                     Array(
2722                         "\\1#\\1",
2723                         "\\1#\\1",
2724                         "\\1#\\1"
2725                         )
2726                     ),
2727         "/^style/i" =>
2728             Array(
2729                 Array(
2730                     "/\/\*.*\*\//",
2731                     "/expression/i",
2732                     "/binding/i",
2733                     "/behaviou*r/i",
2734                     "/include-source/i",
2735
2736                     // position:relative can also be exploited
2737                     // to put content outside of email body area
2738                     // and position:fixed is similarly exploitable
2739                     // as position:absolute, so we'll remove it
2740                     // altogether....
2741                     //
2742                     // Does this screw up legitimate HTML messages?
2743                     // If so, the only fix I see is to allow position
2744                     // attributes (any values?  I think we still have
2745                     // to block static and fixed) only if $use_iframe
2746                     // is enabled (1.5.0+)
2747                     //
2748                     // was:   "/position\s*:\s*absolute/i",
2749                     //
2750                     "/position\s*:/i",
2751
2752                     "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2753                     "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2754                     "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2755                     "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2756                     "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2757                     ),
2758                 Array(
2759                     "",
2760                     "idiocy",
2761                     "idiocy",
2762                     "idiocy",
2763                     "idiocy",
2764                     "idiocy",
2765                     "url",
2766                     "url(\\1#\\1)",
2767                     "url(\\1#\\1)",
2768                     "url(\\1#\\1)",
2769                     "\\1:url(\\2#\\3)"
2770                     )
2771                 )
2772             )
2773         );
2774
2775     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2776     // images off by default.
2777     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2778
2779     if (!$view_unsafe_images){
2780         /**
2781          * Remove any references to http/https if view_unsafe_images set
2782          * to false.
2783          */
2784         array_push($bad_attvals['/.*/']['/^src|background/i'][0],
2785                 '/^([\'\"])\s*https*:.*([\'\"])/si');
2786         array_push($bad_attvals['/.*/']['/^src|background/i'][1],
2787                 "\\1$secremoveimg\\1");
2788         array_push($bad_attvals['/.*/']['/^style/i'][0],
2789                 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2790         array_push($bad_attvals['/.*/']['/^style/i'][1],
2791                 "url(\\1$secremoveimg\\1)");
2792     }
2793
2794     $add_attr_to_tag = Array(
2795             "/^a$/i" =>
2796             Array('target'=>'"_blank"',
2797                 'title'=>'"'._("This external link will open in a new window").'"'
2798                 )
2799             );
2800     $trusted = sq_sanitize($body,
2801                            $tag_list,
2802                            $rm_tags_with_content,
2803                            $self_closing_tags,
2804                            $force_tag_closing,
2805                            $rm_attnames,
2806                            $bad_attvals,
2807                            $add_attr_to_tag,
2808                            $message,
2809                            $id,
2810                            $mailbox
2811                            );
2812     if (strpos($trusted,$secremoveimg)){
2813         $has_unsafe_images = true;
2814     }
2815
2816     // we want to parse mailto's in HTML output, change to SM compose links
2817     // this is a modified version of code from url_parser.php... but Marc is
2818     // right: we need a better filtering implementation; adding this randomly
2819     // here is not a great solution
2820     //
2821     if ($take_mailto_links) {
2822         // parseUrl($trusted);   // this even parses URLs inside of tags... too aggressive
2823         global $MailTo_PReg_Match;
2824         // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2825         // inside an anchor (link) tag, so we have to make sure the regex looks for the
2826         // quote before mailto, and we'll also try to convert the non-links back into links
2827         $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2828         if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2829             foreach ($regs[0] as $i => $mailto_before) {
2830                 $mailto_params = $regs[11][$i];
2831
2832                 // get rid of any leading quote we may have captured but don't care about
2833                 //
2834                 $mailto_before = ltrim($mailto_before, '"\'');
2835
2836                 // get rid of any tailing quote since we have to add send_to to the end
2837                 //
2838                 $mailto_before = rtrim($mailto_before, '"\'');
2839                 $mailto_params = rtrim($mailto_params, '"\'');
2840
2841                 if ($regs[2][$i]) {    //if there is an email addr before '?', we need to merge it with the params
2842                     $to = 'to=' . $regs[2][$i];
2843                     if (strpos($mailto_params, 'to=') > -1)    //already a 'to='
2844                         $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2845                     else {
2846                         if ($mailto_params)    //already some params, append to them
2847                             $mailto_params .= '&amp;' . $to;
2848                         else
2849                             $mailto_params .= '?' . $to;
2850                     }
2851                 }
2852
2853                 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2854
2855                 // we'll already have target=_blank, no need to allow comp_in_new
2856                 // here (which would be a lot more work anyway)
2857                 //
2858                 global $compose_new_win;
2859                 $temp_comp_in_new = $compose_new_win;
2860                 $compose_new_win = 0;
2861                 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2862                 $compose_new_win = $temp_comp_in_new;
2863
2864                 // remove <a href=" and anything after the next quote (we only
2865                 // need the uri, not the link HTML) in compose uri
2866                 //
2867                 // but only do this if the original mailto was in a real anchor tag
2868                 //
2869                 if (!empty($regs[1][$i])) {
2870                     $comp_uri = substr($comp_uri, 9);
2871                     $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2872                 }
2873                 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2874             }
2875         }
2876     }
2877
2878     return $trusted;
2879 }
2880
2881 /**
2882  * function SendDownloadHeaders - send file to the browser
2883  *
2884  * Original Source: SM core src/download.php
2885  * moved here to make it available to other code, and separate
2886  * front end from back end functionality.
2887  *
2888  * @param string $type0 first half of mime type
2889  * @param string $type1 second half of mime type
2890  * @param string $filename filename to tell the browser for downloaded file
2891  * @param boolean $force whether to force the download dialog to pop
2892  * @param optional integer $filesize send the Content-Header and length to the browser
2893  * @return void
2894  */
2895 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2896     global $languages, $squirrelmail_language;
2897     $isIE = $isIE6plus = false;
2898
2899     sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2900
2901     if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2902             strstr($HTTP_USER_AGENT, 'Opera') === false) {
2903         $isIE = true;
2904     }
2905
2906     if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2907         ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2908         $isIE6plus = true;
2909     }
2910
2911     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2912             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2913         $filename =
2914             call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2915     } else {
2916         $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2917     }
2918
2919     // A Pox on Microsoft and it's Internet Explorer!
2920     //
2921     // IE has lots of bugs with file downloads.
2922     // It also has problems with SSL.  Both of these cause problems
2923     // for us in this function.
2924     //
2925     // See this article on Cache Control headers and SSL
2926     // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2927     //
2928     // The best thing you can do for IE is to upgrade to the latest
2929     // version
2930     //set all the Cache Control Headers for IE
2931     if ($isIE) {
2932         $filename=rawurlencode($filename);
2933         header ("Pragma: public");
2934         header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2935         // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2936         // header ("Cache-Control: post-check=0, pre-check=0", false);
2937         header ("Cache-Control: private");
2938
2939         //set the inline header for IE, we'll add the attachment header later if we need it
2940         header ("Content-Disposition: inline; filename=$filename");
2941     }
2942
2943     if (!$force) {
2944         // Try to show in browser window
2945         header ("Content-Disposition: inline; filename=\"$filename\"");
2946         header ("Content-Type: $type0/$type1; name=\"$filename\"");
2947     } else {
2948         // Try to pop up the "save as" box
2949
2950         // IE makes this hard.  It pops up 2 save boxes, or none.
2951         // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2952         // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2953         // But, according to Microsoft, it is "RFC compliant but doesn't
2954         // take into account some deviations that allowed within the
2955         // specification."  Doesn't that mean RFC non-compliant?
2956         // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2957
2958         // all browsers need the application/octet-stream header for this
2959         header ("Content-Type: application/octet-stream; name=\"$filename\"");
2960
2961         // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2962         // Do not have quotes around filename, but that applied to
2963         // "attachment"... does it apply to inline too?
2964         header ("Content-Disposition: attachment; filename=\"$filename\"");
2965
2966         if ($isIE && !$isIE6plus) {
2967             // This combination seems to work mostly.  IE 5.5 SP 1 has
2968             // known issues (see the Microsoft Knowledge Base)
2969
2970             // This works for most types, but doesn't work with Word files
2971             header ("Content-Type: application/download; name=\"$filename\"");
2972             header ("Content-Type: application/force-download; name=\"$filename\"");
2973             // These are spares, just in case.  :-)
2974             //header("Content-Type: $type0/$type1; name=\"$filename\"");
2975             //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2976             //header("Content-Type: application/octet-stream; name=\"$filename\"");
2977         } else if ($isIE) {
2978              // This is to prevent IE for MIME sniffing and auto open a file in IE
2979              header ("Content-Type: application/force-download; name=\"$filename\"");
2980         } else {
2981             // another application/octet-stream forces download for Netscape
2982             header ("Content-Type: application/octet-stream; name=\"$filename\"");
2983         }
2984     }
2985
2986     //send the content-length header if the calling function provides it
2987     if ($filesize > 0) {
2988         header("Content-Length: $filesize");
2989     }
2990
2991 }  // end fn SendDownloadHeaders