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 1999-2019 The SquirrelMail Project Team
  10  * @license http://opensource.org/licenses/gpl-license.php GNU Public License
  11  * @version $Id$
  12  * @package squirrelmail
  13  */
  14
  15 /**
  16  * dependency information
  17    functions       dependency
  18    mime_structure
  19         class/mime/Message.class.php
  20             Message::parseStructure
  21         functions/page_header.php
  22             displayPageHeader
  23         functions/display_messages.php
  24             plain_error_message
  25    mime_fetch_body
  26         functions/imap_general.php
  27             sqimap_run_command
  28    mime_print_body_lines
  29
  30
  31
  32 functions/imap.php
  33 functions/attachment_common.php
  34 functions/display_messages.php
  35
  36 magicHtml => url_parser
  37 translateText => url_parser
  38
  39 */
  40
  41
  42 /* -------------------------------------------------------------------------- */
  43 /* MIME DECODING                                                              */
  44 /* -------------------------------------------------------------------------- */
  45
  46 /**
  47  * Get the MIME structure
  48  *
  49  * This function gets the structure of a message and stores it in the "message" class.
  50  * It will return this object for use with all relevant header information and
  51  * fully parsed into the standard "message" object format.
  52  */
  53 function mime_structure ($bodystructure, $flags=array()) {
  54
  55     /* Isolate the body structure and remove beginning and end parenthesis. */
  56     $read = trim(substr ($bodystructure, strpos(strtolower($bodystructure), 'bodystructure') + 13));
  57     $read = trim(substr ($read, 0, -1));
  58     $i = 0;
  59     $msg = Message::parseStructure($read,$i);
  60
  61     if (!is_object($msg)) {
  62         global $color, $mailbox;
  63         displayPageHeader( $color, $mailbox );
  64         $errormessage  = _("SquirrelMail could not decode the bodystructure of the message");
  65         $errormessage .= '<br />'._("The bodystructure provided by your IMAP server:").'<br /><br />';
  66         $errormessage .= '<pre>' . sm_encode_html_special_chars($read) . '</pre>';
  67         plain_error_message( $errormessage );
  68         echo '</body></html>';
  69         exit;
  70     }
  71     if (count($flags)) {
  72         foreach ($flags as $flag) {
  73 //FIXME: please document why it is we have to check the first char of the flag but we then go ahead and do a full string comparison anyway.  Is this a speed enhancement?  If not, let's keep it simple and just compare the full string and forget the switch block.
  74             $char = strtoupper($flag{1});
  75             switch ($char) {
  76                 case 'S':
  77                     if (strtolower($flag) == '\\seen') {
  78                         $msg->is_seen = true;
  79                     }
  80                     break;
  81                 case 'A':
  82                     if (strtolower($flag) == '\\answered') {
  83                         $msg->is_answered = true;
  84                     }
  85                     break;
  86                 case 'D':
  87                     if (strtolower($flag) == '\\deleted') {
  88                         $msg->is_deleted = true;
  89                     }
  90                     break;
  91                 case 'F':
  92                     if (strtolower($flag) == '\\flagged') {
  93                         $msg->is_flagged = true;
  94                     }
  95                     else if (strtolower($flag) == '$forwarded') {
  96                         $msg->is_forwarded = true;
  97                     }
  98                     break;
  99                 case 'M':
 100                     if (strtolower($flag) == '$mdnsent') {
 101                         $msg->is_mdnsent = true;
 102                     }
 103                     break;
 104                 default:
 105                     break;
 106             }
 107         }
 108     }
 109     //    listEntities($msg);
 110     return $msg;
 111 }
 112
 113
 114
 115 /* This starts the parsing of a particular structure.  It is called recursively,
 116  * so it can be passed different structures.  It returns an object of type
 117  * $message.
 118  * First, it checks to see if it is a multipart message.  If it is, then it
 119  * handles that as it sees is necessary.  If it is just a regular entity,
 120  * then it parses it and adds the necessary header information (by calling out
 121  * to mime_get_elements()
 122  */
 123
 124 function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) {
 125     /* Do a bit of error correction.  If we couldn't find the entity id, just guess
 126      * that it is the first one.  That is usually the case anyway.
 127      */
 128
 129     if (!$ent_id) {
 130         $cmd = "FETCH $id BODY[]";
 131     } else {
 132         $cmd = "FETCH $id BODY[$ent_id]";
 133     }
 134
 135     if ($fetch_size!=0) $cmd .= "<0.$fetch_size>";
 136
 137     $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, TRUE);
 138     do {
 139         $topline = trim(array_shift($data));
 140     } while($topline && ($topline[0] == '*') && !preg_match('/\* [0-9]+ FETCH .*BODY.*/i', $topline)) ;
 141     // Matching with "BODY" above is difficult: in most cases "FETCH \(BODY" would work
 142     // but some servers may put other things in the same result, perhaps something such
 143     // as "* 23 FETCH (FLAGS (\Seen) BODY[1] {174}".  There is some small chance that
 144     // if the character sequence "BODY" appears in a response where it isn't actually
 145     // a FETCH response data item name, the current regex will break things.  The better
 146     // way to do this would be to parse the response correctly and not use a regex.
 147
 148     $wholemessage = implode('', $data);
 149     if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) {
 150         $ret = substr($wholemessage, 0, $regs[1]);
 151         /* There is some information in the content info header that could be important
 152          * in order to parse html messages. Let's get them here.
 153          */
 154 //        if ($ret{0} == '<') {
 155 //            $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE);
 156 //        }
 157     } else if (preg_match('/"([^"]*)"/', $topline, $regs)) {
 158         $ret = $regs[1];
 159     } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) {
 160         $ret = $wholemessage;
 161     } else {
 162         global $where, $what, $mailbox, $passed_id, $startMessage;
 163         $par = 'mailbox=' . urlencode($mailbox) . '&amp;passed_id=' . $passed_id;
 164         if (isset($where) && isset($what)) {
 165             $par .= '&amp;where=' . urlencode($where) . '&amp;what=' . urlencode($what);
 166         } else {
 167             $par .= '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
 168         }
 169         $par .= '&amp;response=' . urlencode($response) .
 170             '&amp;message='  . urlencode($message)  .
 171             '&amp;topline='  . urlencode($topline);
 172
 173         echo '<tt><br />' .
 174             '<table width="80%"><tr>' .
 175             '<tr><td colspan="2">' .
 176             _("Body retrieval error. The reason for this is most probably that the message is malformed.") .
 177             '</td></tr>' .
 178             '<tr><td><b>' . _("Command:") . "</td><td>$cmd</td></tr>" .
 179             '<tr><td><b>' . _("Response:") . "</td><td>$response</td></tr>" .
 180             '<tr><td><b>' . _("Message:") . "</td><td>$message</td></tr>" .
 181             '<tr><td><b>' . _("FETCH line:") . "</td><td>$topline</td></tr>" .
 182             "</table><br /></tt></font><hr />";
 183
 184         $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE);
 185         array_shift($data);
 186         $wholemessage = implode('', $data);
 187
 188         $ret = $wholemessage;
 189     }
 190     return $ret;
 191 }
 192
 193 function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') {
 194
 195     /* Don't kill the connection if the browser is over a dialup
 196      * and it would take over 30 seconds to download it.
 197      * Don't call set_time_limit in safe mode.
 198      */
 199
 200     if (!ini_get('safe_mode')) {
 201         set_time_limit(0);
 202     }
 203     /* in case of base64 encoded attachments, do not buffer them.
 204        Instead, echo the decoded attachment directly to screen */
 205     if (strtolower($encoding) == 'base64') {
 206         if (!$ent_id) {
 207             $query = "FETCH $id BODY[]";
 208         } else {
 209             $query = "FETCH $id BODY[$ent_id]";
 210         }
 211         sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
 212     } else {
 213         $body = mime_fetch_body ($imap_stream, $id, $ent_id);
 214         if (is_resource($rStream)) {
 215             fputs($rStream,decodeBody($body, $encoding, $force_crlf));
 216         } else {
 217             echo decodeBody($body, $encoding, $force_crlf);
 218         }
 219     }
 220
 221     /*
 222        TODO, use the same method for quoted printable.
 223        However, I assume that quoted printable attachments aren't that large
 224        so the performancegain / memory usage drop will be minimal.
 225        If we decide to add that then we need to adapt sqimap_fread because
 226        we need to split te result on \n and fread doesn't stop at \n. That
 227        means we also should provide $results from sqimap_fread (by ref) to
 228        te function and set $no_return to false. The $filter function for
 229        quoted printable should handle unsetting of $results.
 230      */
 231     /*
 232        TODO 2: find out how we write to the output stream php://stdout. fwrite
 233        doesn't work because 'php://stdout isn't a stream.
 234      */
 235
 236     return;
 237 }
 238
 239 /* -[ END MIME DECODING ]----------------------------------------------------------- */
 240
 241 /* This is here for debugging purposes.  It will print out a list
 242  * of all the entity IDs that are in the $message object.
 243  */
 244 function listEntities ($message) {
 245     if ($message) {
 246         echo "<tt>" . $message->entity_id . ' : ' . $message->type0 . '/' . $message->type1 . ' parent = '. $message->parent->entity_id. '<br />';
 247         for ($i = 0; isset($message->entities[$i]); $i++) {
 248             echo "$i : ";
 249             $msg = listEntities($message->entities[$i]);
 250
 251             if ($msg) {
 252                 echo "return: ";
 253                 return $msg;
 254             }
 255         }
 256     }
 257 }
 258
 259 function getPriorityStr($priority) {
 260     $priority_level = substr($priority,0,1);
 261
 262     switch($priority_level) {
 263         /* Check for a higher then normal priority. */
 264         case '1':
 265         case '2':
 266             $priority_string = _("High");
 267             break;
 268
 269         /* Check for a lower then normal priority. */
 270         case '4':
 271         case '5':
 272             $priority_string = _("Low");
 273             break;
 274
 275         /* Check for a normal priority. */
 276         case '3':
 277         default:
 278             $priority_level = '3';
 279             $priority_string = _("Normal");
 280             break;
 281
 282     }
 283     return $priority_string;
 284 }
 285
 286 /* returns a $message object for a particular entity id */
 287 function getEntity ($message, $ent_id) {
 288     return $message->getEntity($ent_id);
 289 }
 290
 291 /* translateText
 292  * Extracted from strings.php 23/03/2002
 293  */
 294
 295 function translateText(&$body, $wrap_at, $charset) {
 296     global $where, $what;   /* from searching */
 297     global $color;          /* color theme */
 298
 299     // require_once(SM_PATH . 'functions/url_parser.php');
 300
 301     $body_ary = explode("\n", $body);
 302     for ($i=0; $i < count($body_ary); $i++) {
 303         $line = rtrim($body_ary[$i],"\r");
 304
 305         if (strlen($line) - 2 >= $wrap_at) {
 306             sqWordWrap($line, $wrap_at, $charset);
 307         }
 308         $line = charset_decode($charset, $line);
 309         $line = str_replace("\t", '        ', $line);
 310
 311         parseUrl ($line);
 312
 313         $quotes = 0;
 314         $pos = 0;
 315         $j = strlen($line);
 316
 317         while ($pos < $j) {
 318             if ($line[$pos] == ' ') {
 319                 $pos++;
 320             } else if (strpos($line, '&gt;', $pos) === $pos) {
 321                 $pos += 4;
 322                 $quotes++;
 323             } else {
 324                 break;
 325             }
 326         }
 327
 328         if ($quotes % 2) {
 329             $line = '<span class="quote1">' . $line . '</span>';
 330         } elseif ($quotes) {
 331             $line = '<span class="quote2">' . $line . '</span>';
 332         }
 333
 334         $body_ary[$i] = $line;
 335     }
 336     $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
 337 }
 338
 339 /**
 340  * This returns a parsed string called $body. That string can then
 341  * be displayed as the actual message in the HTML. It contains
 342  * everything needed, including HTML Tags, Attachments at the
 343  * bottom, etc.
 344  *
 345  * Since 1.2.0 function uses message_body hook.
 346  * Till 1.3.0 function included output of formatAttachments().
 347  *
 348  * @param resource $imap_stream imap connection resource
 349  * @param object $message squirrelmail message object
 350  * @param array $color squirrelmail color theme array
 351  * @param integer $wrap_at number of characters per line
 352  * @param string $ent_num (since 1.3.0) message part id
 353  * @param integer $id (since 1.3.0) message id
 354  * @param string $mailbox (since 1.3.0) imap folder name
 355  * @return string html formated message text
 356  */
 357 function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') {
 358     /* This if statement checks for the entity to show as the
 359      * primary message. To add more of them, just put them in the
 360      * order that is their priority.
 361      */
 362     global $startMessage, $languages, $squirrelmail_language,
 363            $show_html_default, $sort, $has_unsafe_images, $passed_ent_id,
 364            $use_iframe, $iframe_height, $download_and_unsafe_link,
 365            $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text,
 366            $oTemplate, $nbsp;
 367
 368     // workaround for not updated config.php
 369     if (! isset($use_iframe)) $use_iframe = false;
 370
 371     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
 372     // images off by default.
 373     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
 374
 375     $body = '';
 376     $urlmailbox = urlencode($mailbox);
 377     $body_message = getEntity($message, $ent_num);
 378     if (($body_message->header->type0 == 'text') ||
 379             ($body_message->header->type0 == 'rfc822')) {
 380         $body = mime_fetch_body ($imap_stream, $id, $ent_num);
 381         $body = decodeBody($body, $body_message->header->encoding);
 382
 383         if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 384                 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) {
 385             if (mb_detect_encoding($body) != 'ASCII') {
 386                 $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body);
 387             }
 388         }
 389
 390         /* As of 1.5.2, $body is passed (and modified) by reference */
 391         do_hook('message_body', $body);
 392
 393         /* If there are other types that shouldn't be formatted, add
 394          * them here.
 395          */
 396
 397         if ($body_message->header->type1 == 'html') {
 398             if ($show_html_default <> 1) {
 399                 $entity_conv = array('&nbsp;' => ' ',
 400                                      '<p>'    => "\n",
 401                                      '<P>'    => "\n",
 402                                      '<br>'   => "\n",
 403                                      '<BR>'   => "\n",
 404                                      '<br />' => "\n",
 405                                      '<BR />' => "\n",
 406                                      '&gt;'   => '>',
 407                                      '&lt;'   => '<');
 408                 $body = strtr($body, $entity_conv);
 409                 $body = strip_tags($body);
 410                 $body = trim($body);
 411                 translateText($body, $wrap_at,
 412                         $body_message->header->getParameter('charset'));
 413             } elseif ($use_iframe) {
 414                 /**
 415                  * If we don't add html message between iframe tags,
 416                  * we must detect unsafe images and modify $has_unsafe_images.
 417                  */
 418                 $html_body = magicHTML($body, $id, $message, $mailbox);
 419                 // Convert character set in order to display html mails in different character set
 420                 $html_body = charset_decode($body_message->header->getParameter('charset'),$html_body,false,true);
 421
 422                 // creating iframe url
 423                 $iframeurl=sqm_baseuri().'src/view_html.php?'
 424                     . 'mailbox=' . $urlmailbox
 425                     . '&amp;passed_id=' . $id
 426                     . '&amp;ent_id=' . $ent_num
 427                     . '&amp;view_unsafe_images=' . (int) $view_unsafe_images;
 428
 429                 global $oTemplate;
 430                 $oTemplate->assign('iframe_url', $iframeurl);
 431                 $oTemplate->assign('iframe_height', $iframe_height);
 432                 $oTemplate->assign('html_body', $html_body);
 433
 434                 $body = $oTemplate->fetch('read_html_iframe.tpl');
 435             } else {
 436                 // old way of html rendering
 437                 /**
 438                  * convert character set. charset_decode does not remove html special chars
 439                  * applied by magicHTML functions and does not sanitize them second time if
 440                  * fourth argument is true.
 441                  */
 442                 $charset = $body_message->header->getParameter('charset');
 443                 if (!empty($charset)) {
 444                     $body = charset_decode($charset,$body,false,true);
 445                 }
 446                 $body = magicHTML($body, $id, $message, $mailbox);
 447             }
 448         } else {
 449             translateText($body, $wrap_at,
 450                     $body_message->header->getParameter('charset'));
 451         }
 452
 453         /*
 454          * Previously the links for downloading and unsafe images were printed
 455          * under the mail. By putting the links in a global variable we can
 456          * print it in the toolbar where it belongs. Since the original code was
 457          * in this place it's left here. It might be possible to move it to some
 458          * other place if that makes sense. The possibility to do so has not
 459          * been evaluated yet.
 460          */
 461
 462         // Initialize the global variable to an empty string.
 463         // FIXME: To have $download_and_unsafe_link as a global variable might not be needed since the use of separate variables ($download_href, $unsafe_image_toggle_href, and $unsafe_image_toggle_text) for the templates was introduced.
 464         $download_and_unsafe_link = '';
 465
 466         // Prepare and build a link for downloading the mail.
 467         $link = 'passed_id=' . $id . '&amp;ent_id='.$ent_num.
 468             '&amp;mailbox=' . $urlmailbox .'&amp;sort=' . $sort .
 469             '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
 470         if (isset($passed_ent_id)) {
 471             $link .= '&amp;passed_ent_id='.$passed_ent_id;
 472         }
 473         $download_href = SM_PATH . 'src/download.php?absolute_dl=true&amp;' . $link;
 474
 475         // Always add the link for downloading the mail as a file to the global
 476         // variable.
 477         $download_and_unsafe_link .= "$nbsp|$nbsp"
 478             . create_hyperlink($download_href, _("Download this as a file"));
 479
 480         // Find out the right text to use in the link depending on the
 481         // circumstances. If the unsafe images are displayed the link should
 482         // hide them, if they aren't displayed the link should only appear if
 483         // the mail really contains unsafe images.
 484         if ($view_unsafe_images) {
 485             $text = _("Hide Unsafe Images");
 486         } else {
 487             if (isset($has_unsafe_images) && $has_unsafe_images) {
 488                 $link .= '&amp;view_unsafe_images=1';
 489                 $text = _("View Unsafe Images");
 490             } else {
 491                 $text = '';
 492             }
 493         }
 494
 495         // Only create a link for unsafe images if there's need for one. If so:
 496         // add it to the global variable.
 497         if($text != '') {
 498             $unsafe_image_toggle_href = SM_PATH . 'src/read_body.php?'.$link;
 499             $unsafe_image_toggle_text = $text;
 500             $download_and_unsafe_link .= "$nbsp|$nbsp"
 501                 . create_hyperlink($unsafe_image_toggle_href, $text);
 502         }
 503     }
 504     return $body;
 505 }
 506
 507 /**
 508  * Generate attachments array for passing to templates.
 509  *
 510  * @since 1.5.2
 511  * @param object $message SquirrelMail message object
 512  * @param array $exclude_id message parts that are not attachments.
 513  * @param string $mailbox mailbox name
 514  * @param integer $id message id
 515  */
 516 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
 517     global $where, $what, $startMessage, $color, $passed_ent_id,
 518            $base_uri, $block_svg_download;
 519
 520     $att_ar = $message->getAttachments($exclude_id);
 521     $urlMailbox = urlencode($mailbox);
 522
 523     $attachments = array();
 524     foreach ($att_ar as $att) {
 525         $ent = $att->entity_id;
 526         $header = $att->header;
 527         $type0 = strtolower($header->type0);
 528         $type1 = strtolower($header->type1);
 529         if ($block_svg_download && strpos($type1, 'svg') === 0)
 530             continue;
 531
 532         $name = '';
 533         $links = array();
 534         $links['download link']['text'] = _("Download");
 535         $links['download link']['href'] = $base_uri .
 536             "src/download.php?absolute_dl=true&amp;passed_id=$id&amp;mailbox=$urlMailbox&amp;ent_id=$ent";
 537
 538         if ($type0 =='message' && $type1 == 'rfc822') {
 539             $default_page = $base_uri  . 'src/read_body.php';
 540             $rfc822_header = $att->rfc822_header;
 541             $filename = $rfc822_header->subject;
 542             if (trim( $filename ) == '') {
 543                 $filename = 'untitled-[' . $ent . ']' ;
 544             }
 545             $from_o = $rfc822_header->from;
 546             if (is_object($from_o)) {
 547                 $from_name = decodeHeader($from_o->getAddress(false));
 548             } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) {
 549                 // something weird happens when a digest message is opened and you return to the digest
 550                 // now the from object is part of an array. Probably the parseHeader call overwrites the info
 551                 // retrieved from the bodystructure in a different way. We need to fix this later.
 552                 // possible starting point, do not fetch header we already have and inspect how
 553                 // the rfc822_header object behaves.
 554                 $from_name = decodeHeader($from_o[0]->getAddress(false));
 555             } else {
 556                 $from_name = _("Unknown sender");
 557             }
 558             $description = _("From").': '.$from_name;
 559         } else {
 560             $default_page = $base_uri  . 'src/download.php';
 561             $filename = $att->getFilename();
 562             if ($header->description) {
 563                 $description = decodeHeader($header->description);
 564             } else {
 565                 $description = '';
 566             }
 567         }
 568
 569         $display_filename = $filename;
 570         if (isset($passed_ent_id)) {
 571             $passed_ent_id_link = '&amp;passed_ent_id='.$passed_ent_id;
 572         } else {
 573             $passed_ent_id_link = '';
 574         }
 575         $defaultlink = $default_page . "?startMessage=$startMessage"
 576             . "&amp;passed_id=$id&amp;mailbox=$urlMailbox"
 577             . '&amp;ent_id='.$ent.$passed_ent_id_link;
 578         if ($where && $what) {
 579             $defaultlink .= '&amp;where='. urlencode($where).'&amp;what='.urlencode($what);
 580         }
 581         // IE does make use of mime content sniffing. Forcing a download
 582         // prohibit execution of XSS inside an application/octet-stream attachment
 583         if ($type0 == 'application' && $type1 == 'octet-stream') {
 584             $defaultlink .= '&amp;absolute_dl=true';
 585         }
 586
 587         /* This executes the attachment hook with a specific MIME-type.
 588          * It also allows plugins to run if there's a rule for a more
 589          * generic type. Finally, a hook for ALL attachment types is
 590          * run as well.
 591          */
 592         // First remember the default link.
 593         $defaultlink_orig = $defaultlink;
 594
 595         /* The API for this hook has changed as of 1.5.2 so that all plugin
 596            arguments are passed in an array instead of each their own plugin
 597            argument, and arguments are passed by reference, so instead of
 598            returning any changes, changes should simply be made to the original
 599            arguments themselves. */
 600         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 601                     &$defaultlink, &$display_filename, &$where, &$what);
 602         do_hook("attachment $type0/$type1", $temp);
 603         /* The API for this hook has changed as of 1.5.2 so that all plugin
 604            arguments are passed in an array instead of each their own plugin
 605            argument, and arguments are passed by reference, so instead of
 606            returning any changes, changes should simply be made to the original
 607            arguments themselves. */
 608         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 609                       &$defaultlink, &$display_filename, &$where, &$what);
 610         // Do not let a generic plugin change the default link if a more
 611         // specialized one already did it...
 612         if ($defaultlink != $defaultlink_orig) {
 613             $dummy = '';
 614             $temp[5] = &$dummy;
 615         }
 616         do_hook("attachment $type0/*", $temp);
 617         /* The API for this hook has changed as of 1.5.2 so that all plugin
 618            arguments are passed in an array instead of each their own plugin
 619            argument, and arguments are passed by reference, so instead of
 620            returning any changes, changes should simply be made to the original
 621            arguments themselves. */
 622         $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
 623                       &$defaultlink, &$display_filename, &$where, &$what);
 624         // Do not let a generic plugin change the default link if a more
 625         // specialized one already did it...
 626         if ($defaultlink != $defaultlink_orig) {
 627             $dummy = '';
 628             $temp[5] = &$dummy;
 629         }
 630         do_hook("attachment */*", $temp);
 631
 632         $this_attachment = array();
 633         $this_attachment['Name'] = decodeHeader($display_filename);
 634         $this_attachment['Description'] = $description;
 635         $this_attachment['DefaultHREF'] = $defaultlink;
 636         $this_attachment['DownloadHREF'] = $links['download link']['href'];
 637         $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
 638         $this_attachment['Size'] = $header->size;
 639         $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
 640         $this_attachment['OtherLinks'] = array();
 641         foreach ($links as $val) {
 642             if ($val['text']==_("Download") || $val['text'] == _("View"))
 643                 continue;
 644             if (empty($val['text']) && empty($val['extra']))
 645                 continue;
 646
 647             $temp = array();
 648             $temp['HREF'] = $val['href'];
 649             $temp['Text'] = (empty($val['text']) ? '' : $val['text']) . (empty($val['extra']) ? '' : $val['extra']);
 650             $this_attachment['OtherLinks'][] = $temp;
 651         }
 652         $attachments[] = $this_attachment;
 653
 654         unset($links);
 655     }
 656
 657     return $attachments;
 658 }
 659
 660 /**
 661  * Displays attachment links and information
 662  *
 663  * Since 1.3.0 function is not included in formatBody() call.
 664  *
 665  * Since 1.0.2 uses attachment $type0/$type1 hook.
 666  * Since 1.2.5 uses attachment $type0/* hook.
 667  * Since 1.5.0 uses attachments_bottom hook.
 668  * Since 1.5.2 uses templates and does *not* return a value.
 669  *
 670  * @param object $message SquirrelMail message object
 671  * @param array $exclude_id message parts that are not attachments.
 672  * @param string $mailbox mailbox name
 673  * @param integer $id message id
 674  */
 675 function formatAttachments($message, $exclude_id, $mailbox, $id) {
 676     global $oTemplate;
 677
 678     $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
 679
 680     $oTemplate->assign('attachments', $attach);
 681     $oTemplate->display('read_attachments.tpl');
 682 }
 683
 684 function sqimap_base64_decode(&$string) {
 685
 686     // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
 687     // fly decoding (to reduce memory usage) you have to check if the
 688     // data has incomplete pairs
 689
 690     // Remove the noise in order to check if the 4 bytes pairs are complete
 691     $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
 692
 693     $sStringRem = '';
 694     $iMod = strlen($string) % 4;
 695     if ($iMod) {
 696         $sStringRem = substr($string,-$iMod);
 697         // Check if $sStringRem contains padding characters
 698         if (substr($sStringRem,-1) != '=') {
 699             $string = substr($string,0,-$iMod);
 700         } else {
 701             $sStringRem = '';
 702         }
 703     }
 704     $string = base64_decode($string);
 705     return $sStringRem;
 706 }
 707
 708 /**
 709  * Decodes encoded string (usually message body)
 710  *
 711  * This function decodes a string (usually the message body)
 712  * depending on the encoding type.  Currently quoted-printable
 713  * and base64 encodings are supported.
 714  *
 715  * The decode_body hook was added to this function in 1.4.2/1.5.0.
 716  * The $force_crlf parameter was added in 1.5.2.
 717  *
 718  * @param string $string     The encoded string
 719  * @param string $encoding   used encoding
 720  * @param string $force_crlf Whether or not to force CRLF or LF
 721  *                           line endings (or to leave as is).
 722  *                           If given as "LF", line endings will
 723  *                           all be converted to LF; if "CRLF",
 724  *                           line endings will all be converted
 725  *                           to CRLF.  If given as an empty value,
 726  *                           the global $force_crlf_default will
 727  *                           be consulted (it can be specified in
 728  *                           config/config_local.php).  Otherwise,
 729  *                           any other value will cause the string
 730  *                           to be left alone.  Note that this will
 731  *                           be overridden to "LF" if not using at
 732  *                           least PHP version 4.3.0. (OPTIONAL;
 733  *                           default is empty - consult global
 734  *                           default value)
 735  *
 736  * @return string The decoded string
 737  *
 738  * @since 1.0
 739  *
 740  */
 741 function decodeBody($string, $encoding, $force_crlf='') {
 742
 743     global $force_crlf_default;
 744     if (empty($force_crlf)) $force_crlf = $force_crlf_default;
 745     $force_crlf = strtoupper($force_crlf);
 746
 747     // must force line endings to LF due to broken
 748     // quoted_printable_decode() in PHP versions
 749     // before 4.3.0 (see below)
 750     //
 751     if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
 752         $string = str_replace("\r\n", "\n", $string);
 753     else if ($force_crlf == 'CRLF')
 754         $string = str_replace("\n", "\r\n", $string);
 755
 756     $encoding = strtolower($encoding);
 757
 758     $encoding_handler = do_hook('decode_body', $encoding);
 759
 760
 761     // plugins get first shot at decoding the string
 762     //
 763     if (!empty($encoding_handler) && function_exists($encoding_handler)) {
 764         $string = $encoding_handler('decode', $string);
 765
 766     } elseif ($encoding == 'quoted-printable' ||
 767             $encoding == 'quoted_printable') {
 768
 769         // quoted_printable_decode() function is broken in older
 770         // php versions.  Text with \r\n decoding was fixed only
 771         // in php 4.3.0.  Minimal code requirement is PHP 4.0.4+
 772         // and the above call to:  str_replace("\r\n", "\n", $string);
 773         //
 774         $string = quoted_printable_decode($string);
 775
 776     } elseif ($encoding == 'base64') {
 777         $string = base64_decode($string);
 778     }
 779
 780     // All other encodings are returned raw.
 781     return $string;
 782 }
 783
 784 /**
 785  * Decodes headers
 786  *
 787  * This function decodes strings that are encoded according to
 788  * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
 789  * Patched by Christian Schmidt <christian@ostenfeld.dk>  23/03/2002
 790  *
 791  * @param string $string header string that has to be made readable
 792  * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
 793  * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
 794  * @param boolean $decide decide if string can be utfencoded. defaults to false
 795  * @return string decoded header string
 796  */
 797 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
 798     global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
 799     if (is_array($string)) {
 800         $string = implode("\n", $string);
 801     }
 802
 803     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 804             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
 805         $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
 806         // Do we need to return at this point?
 807         // return $string;
 808     }
 809     $i = 0;
 810     $iLastMatch = -2;
 811     $encoded = true;
 812
 813 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
 814     $aString = explode(' ',$string);
 815     $ret = '';
 816     foreach ($aString as $chunk) {
 817         if ($encoded && $chunk === '') {
 818             continue;
 819         } elseif ($chunk === '') {
 820             $ret .= ' ';
 821             continue;
 822         }
 823         $encoded = false;
 824         /* if encoded words are not separated by a linear-space-white we still catch them */
 825         $j = $i-1;
 826
 827         while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
 828             /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
 829             if ($iLastMatch !== $j) {
 830                 if ($htmlsafe) {
 831                     $ret .= '&#32;';
 832                 } else {
 833                     $ret .= ' ';
 834                 }
 835             }
 836             $iLastMatch = $i;
 837             $j = $i;
 838             if ($htmlsafe) {
 839                 $ret .= sm_encode_html_special_chars($res[1]);
 840             } else {
 841                 $ret .= $res[1];
 842             }
 843             $encoding = ucfirst($res[3]);
 844
 845             /* decide about valid decoding */
 846             if ($decide && is_conversion_safe($res[2])) {
 847                 $utfencode=true;
 848                 $can_be_encoded=true;
 849             } else {
 850                 $can_be_encoded=false;
 851             }
 852             switch ($encoding)
 853             {
 854                 case 'B':
 855                     // fix broken base64-encoded strings (remove end = padding,
 856                     // change any = to + in middle of string, add padding back
 857                     // to the end)
 858                     if ($fix_broken_base64_encoded_messages) {
 859                         $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
 860                         $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
 861                     }
 862                     $replace = base64_decode($res[4]);
 863                     if ($utfencode) {
 864                         if ($can_be_encoded) {
 865                             /* convert string to different charset,
 866                              * if functions asks for it (usually in compose)
 867                              */
 868                             $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
 869                         } else {
 870                             // convert string to html codes in order to display it
 871                             $ret .= charset_decode($res[2],$replace);
 872                         }
 873                     } else {
 874                         if ($htmlsafe) {
 875                             $replace = sm_encode_html_special_chars($replace);
 876                         }
 877                         $ret.= $replace;
 878                     }
 879                     break;
 880                 case 'Q':
 881                     $replace = str_replace('_', ' ', $res[4]);
 882                     $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
 883                             create_function ('$matches', 'return chr(hexdec($matches[1]));'),
 884                             $replace);
 885                     if ($utfencode) {
 886                         if ($can_be_encoded) {
 887                             /* convert string to different charset,
 888                              * if functions asks for it (usually in compose)
 889                              */
 890                             $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
 891                         } else {
 892                             // convert string to html codes in order to display it
 893                             $replace = charset_decode($res[2], $replace);
 894                         }
 895                     } else {
 896                         if ($htmlsafe) {
 897                             $replace = sm_encode_html_special_chars($replace);
 898                         }
 899                     }
 900                     $ret .= $replace;
 901                     break;
 902                 default:
 903                     break;
 904             }
 905             $chunk = $res[5];
 906             $encoded = true;
 907         }
 908         if (!$encoded) {
 909             if ($htmlsafe) {
 910                 $ret .= '&#32;';
 911             } else {
 912                 $ret .= ' ';
 913             }
 914         }
 915
 916         if (!$encoded && $htmlsafe) {
 917             $ret .= sm_encode_html_special_chars($chunk);
 918         } else {
 919             $ret .= $chunk;
 920         }
 921         ++$i;
 922     }
 923     /* remove the first added space */
 924     if ($ret) {
 925         if ($htmlsafe) {
 926             $ret = substr($ret,5);
 927         } else {
 928             $ret = substr($ret,1);
 929         }
 930     }
 931
 932     return $ret;
 933 }
 934
 935 /**
 936  * Encodes header
 937  *
 938  * Function uses XTRA_CODE _encodeheader function, if such function exists.
 939  *
 940  * Function uses Q encoding by default and encodes a string according to RFC
 941  * 1522 for use in headers if it contains 8-bit characters or anything that
 942  * looks like it should be encoded.
 943  *
 944  * Function switches to B encoding and encodeHeaderBase64() function, if
 945  * string is 8bit and multibyte character set supported by mbstring extension
 946  * is used. It can cause E_USER_NOTICE errors, if interface is used with
 947  * multibyte character set unsupported by mbstring extension.
 948  *
 949  * @param string $string header string, that has to be encoded
 950  * @return string quoted-printable encoded string
 951  * @todo make $mb_charsets system wide constant
 952  */
 953 function encodeHeader ($string) {
 954     global $default_charset, $languages, $squirrelmail_language;
 955
 956     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
 957             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
 958         return  call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
 959     }
 960
 961     // Use B encoding for multibyte charsets
 962     $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
 963     if (in_array($default_charset,$mb_charsets) &&
 964         in_array($default_charset,sq_mb_list_encodings()) &&
 965         sq_is8bit($string)) {
 966         return encodeHeaderBase64($string,$default_charset);
 967     } elseif (in_array($default_charset,$mb_charsets) &&
 968               sq_is8bit($string) &&
 969               ! in_array($default_charset,sq_mb_list_encodings())) {
 970         // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
 971         // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
 972     }
 973
 974     // Encode only if the string contains 8-bit characters or =?
 975     $j = strlen($string);
 976     $max_l = 75 - strlen($default_charset) - 7;
 977     $aRet = array();
 978     $ret = '';
 979     $iEncStart = $enc_init = false;
 980     $cur_l = $iOffset = 0;
 981     for($i = 0; $i < $j; ++$i) {
 982         switch($string{$i})
 983         {
 984             case '"':
 985             case '=':
 986             case '<':
 987             case '>':
 988             case ',':
 989             case '?':
 990             case '_':
 991                 if ($iEncStart === false) {
 992                     $iEncStart = $i;
 993                 }
 994                 $cur_l+=3;
 995                 if ($cur_l > ($max_l-2)) {
 996                     /* if there is an stringpart that doesn't need encoding, add it */
 997                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
 998                     $aRet[] = "=?$default_charset?Q?$ret?=";
 999                     $iOffset = $i;
1000                     $cur_l = 0;
1001                     $ret = '';
1002                     $iEncStart = false;
1003                 } else {
1004                     $ret .= sprintf("=%02X",ord($string{$i}));
1005                 }
1006                 break;
1007             case '(':
1008             case ')':
1009                 if ($iEncStart !== false) {
1010                     $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1011                     $aRet[] = "=?$default_charset?Q?$ret?=";
1012                     $iOffset = $i;
1013                     $cur_l = 0;
1014                     $ret = '';
1015                     $iEncStart = false;
1016                 }
1017                 break;
1018             case ' ':
1019                 if ($iEncStart !== false) {
1020                     $cur_l++;
1021                     if ($cur_l > $max_l) {
1022                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1023                         $aRet[] = "=?$default_charset?Q?$ret?=";
1024                         $iOffset = $i;
1025                         $cur_l = 0;
1026                         $ret = '';
1027                         $iEncStart = false;
1028                     } else {
1029                         $ret .= '_';
1030                     }
1031                 }
1032                 break;
1033             default:
1034                 $k = ord($string{$i});
1035                 if ($k > 126) {
1036                     if ($iEncStart === false) {
1037                         // do not start encoding in the middle of a string, also take the rest of the word.
1038                         $sLeadString = substr($string,0,$i);
1039                         $aLeadString = explode(' ',$sLeadString);
1040                         $sToBeEncoded = array_pop($aLeadString);
1041                         $iEncStart = $i - strlen($sToBeEncoded);
1042                         $ret .= $sToBeEncoded;
1043                         $cur_l += strlen($sToBeEncoded);
1044                     }
1045                     $cur_l += 3;
1046                     /* first we add the encoded string that reached it's max size */
1047                     if ($cur_l > ($max_l-2)) {
1048                         $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1049                         $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1050                         $cur_l = 3;
1051                         $ret = '';
1052                         $iOffset = $i;
1053                         $iEncStart = $i;
1054                     }
1055                     $enc_init = true;
1056                     $ret .= sprintf("=%02X", $k);
1057                 } else {
1058                     if ($iEncStart !== false) {
1059                         $cur_l++;
1060                         if ($cur_l > $max_l) {
1061                             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1062                             $aRet[] = "=?$default_charset?Q?$ret?=";
1063                             $iEncStart = false;
1064                             $iOffset = $i;
1065                             $cur_l = 0;
1066                             $ret = '';
1067                         } else {
1068                             $ret .= $string{$i};
1069                         }
1070                     }
1071                 }
1072                 break;
1073         }
1074     }
1075
1076     if ($enc_init) {
1077         if ($iEncStart !== false) {
1078             $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1079             $aRet[] = "=?$default_charset?Q?$ret?=";
1080         } else {
1081             $aRet[] = substr($string,$iOffset);
1082         }
1083         $string = implode('',$aRet);
1084     }
1085     return $string;
1086 }
1087
1088 /**
1089  * Encodes string according to rfc2047 B encoding header formating rules
1090  *
1091  * It is recommended way to encode headers with character sets that store
1092  * symbols in more than one byte.
1093  *
1094  * Function requires mbstring support. If required mbstring functions are missing,
1095  * function returns false and sets E_USER_WARNING level error message.
1096  *
1097  * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1098  * that mbstring functions will generate E_WARNING errors, if unsupported
1099  * character set is used. mb_encode_mimeheader function provided by php
1100  * mbstring extension is not used in order to get better control of header
1101  * encoding.
1102  *
1103  * Used php code functions - function_exists(), trigger_error(), strlen()
1104  * (is used with charset names and base64 strings). Used php mbstring
1105  * functions - mb_strlen and mb_substr.
1106  *
1107  * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1108  * encoding), rfc 2822 (header folding)
1109  *
1110  * @param string $string header string that must be encoded
1111  * @param string $charset character set. Must be supported by mbstring extension.
1112  * Use sq_mb_list_encodings() to detect supported charsets.
1113  * @return string string encoded according to rfc2047 B encoding formating rules
1114  * @since 1.5.1
1115  * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1116  * @todo Do we want to control max length of header?
1117  * @todo Do we want to control EOL (end-of-line) marker?
1118  * @todo Do we want to translate error message?
1119  */
1120 function encodeHeaderBase64($string,$charset) {
1121     /**
1122      * Check mbstring function requirements.
1123      */
1124     if (! function_exists('mb_strlen') ||
1125         ! function_exists('mb_substr')) {
1126         // set E_USER_WARNING
1127         trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1128         // return false
1129         return false;
1130     }
1131
1132     // initial return array
1133     $aRet = array();
1134
1135     /**
1136      * header length = 75 symbols max (same as in encodeHeader)
1137      * remove $charset length
1138      * remove =? ? ?= (5 chars)
1139      * remove 2 more chars (\r\n ?)
1140      */
1141     $iMaxLength = 75 - strlen($charset) - 7;
1142
1143     // set first character position
1144     $iStartCharNum = 0;
1145
1146     // loop through all characters. count characters and not bytes.
1147     for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1148         // encode string from starting character to current character.
1149         $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1150
1151         // Check encoded string length
1152         if(strlen($encoded_string)>$iMaxLength) {
1153             // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1154             $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1155
1156             // set new starting character
1157             $iStartCharNum = $iCharNum-1;
1158
1159             // encode last char (in case it is last character in string)
1160             $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1161         } // if string is shorter than max length - add next character
1162     }
1163
1164     // add last encoded string to array
1165     $aRet[] = $encoded_string;
1166
1167     // set initial return string
1168     $sRet = '';
1169
1170     // loop through encoded strings
1171     foreach($aRet as $string) {
1172         // TODO: Do we want to control EOL (end-of-line) marker
1173         if ($sRet!='') $sRet.= " ";
1174
1175         // add header tags and encoded string to return string
1176         $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1177     }
1178
1179     return $sRet;
1180 }
1181
1182 /* This function trys to locate the entity_id of a specific mime element */
1183 function find_ent_id($id, $message) {
1184     for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1185         if ($message->entities[$i]->header->type0 == 'multipart')  {
1186             $ret = find_ent_id($id, $message->entities[$i]);
1187         } else {
1188             if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1189 //                if (sq_check_save_extension($message->entities[$i])) {
1190                 return $message->entities[$i]->entity_id;
1191 //                }
1192             } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1193                 /**
1194                  * This is part of a fix for Outlook Express 6.x generating
1195                  * cid URLs without creating content-id headers
1196                  * @@JA - 20050207
1197                  */
1198                 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1199                     return $message->entities[$i]->entity_id;
1200                 }
1201             }
1202         }
1203     }
1204     return $ret;
1205 }
1206
1207 function sq_check_save_extension($message) {
1208     $filename = $message->getFilename();
1209     $ext = substr($filename, strrpos($filename,'.')+1);
1210     $save_extensions = array('jpg','jpeg','gif','png','bmp');
1211     return in_array($ext, $save_extensions);
1212 }
1213
1214
1215 /**
1216  ** HTMLFILTER ROUTINES
1217  */
1218
1219 /**
1220  * This function checks attribute values for entity-encoded values
1221  * and returns them translated into 8-bit strings so we can run
1222  * checks on them.
1223  *
1224  * @param  $attvalue A string to run entity check against.
1225  * @return           Nothing, modifies a reference value.
1226  */
1227 function sq_defang(&$attvalue){
1228     $me = 'sq_defang';
1229     /**
1230      * Skip this if there aren't ampersands or backslashes.
1231      */
1232     if (strpos($attvalue, '&') === false
1233         && strpos($attvalue, '\\') === false){
1234         return;
1235     }
1236     $m = false;
1237     // before deent, translate the dangerous unicode characters and ... to safe values
1238     // otherwise the regular expressions do not match.
1239
1240
1241
1242     do {
1243         $m = false;
1244         $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1245         $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1246         $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1247     } while ($m == true);
1248     $attvalue = stripslashes($attvalue);
1249 }
1250
1251 /**
1252  * Kill any tabs, newlines, or carriage returns. Our friends the
1253  * makers of the browser with 95% market value decided that it'd
1254  * be funny to make "java[tab]script" be just as good as "javascript".
1255  *
1256  * @param  attvalue  The attribute value before extraneous spaces removed.
1257  * @return attvalue  Nothing, modifies a reference value.
1258  */
1259 function sq_unspace(&$attvalue){
1260     $me = 'sq_unspace';
1261     if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1262         $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1263                                 Array('',   '',   '',   '',   ''), $attvalue);
1264     }
1265 }
1266
1267 /**
1268  * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1269  * IE as regular characters.
1270  *
1271  * @param  attvalue  The attribute value before dangerous characters are translated.
1272  * @return attvalue  Nothing, modifies a reference value.
1273  * @author Marc Groot Koerkamp.
1274  */
1275 function sq_fixIE_idiocy(&$attvalue) {
1276     // remove NUL
1277     $attvalue = str_replace("\0", "", $attvalue);
1278     // remove comments
1279     $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1280
1281     // IE has the evil habit of accepting every possible value for the attribute expression.
1282     // The table below contains characters which are parsed by IE if they are used in the "expression"
1283     // attribute value.
1284     $aDangerousCharsReplacementTable = array(
1285                         array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1286                               '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1287                               '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1288                               '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1289                               '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1290                               '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1291                               '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1292                               '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1293                               '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1294                               '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1295                               '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1296                               '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1297                               '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1298                               '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1299                               '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1300                               '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1301                               '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1302                               '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1303                               '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1304                               '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1305                               '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1306                               '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1307                               '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1308                               '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1309                               "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */   // in unicode this is some Chinese char range
1310                               "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1311                               "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1312                               "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1313                               "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1314                               "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1315                               "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1316                               "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1317                               "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1318                               "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1319                               "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1320                               "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1321                               "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1322                               "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1323                               "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1324                               "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1325                               "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1326                               "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1327                               "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1328                               "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1329                               "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1330                               "\xCA\x9F", /* L UNICODE IPA Extension */
1331                               "\xCA\x80", /* R UNICODE IPA Extension */
1332                               "\xC9\xB4"),  /* N UNICODE IPA Extension */
1333                        array('l', 'l', 'r','r','n','n',
1334                              'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1335                              'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1336                              'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1337     $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1338
1339     // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1340     // used for XSS.
1341     $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1342 }
1343
1344 /**
1345  * This function returns the final tag out of the tag name, an array
1346  * of attributes, and the type of the tag. This function is called by
1347  * sq_sanitize internally.
1348  *
1349  * @param  $tagname  the name of the tag.
1350  * @param  $attary   the array of attributes and their values
1351  * @param  $tagtype  The type of the tag (see in comments).
1352  * @return           a string with the final tag representation.
1353  */
1354 function sq_tagprint($tagname, $attary, $tagtype){
1355     $me = 'sq_tagprint';
1356
1357     if ($tagtype == 2){
1358         $fulltag = '</' . $tagname . '>';
1359     } else {
1360         $fulltag = '<' . $tagname;
1361         if (is_array($attary) && sizeof($attary)){
1362             $atts = Array();
1363             while (list($attname, $attvalue) = each($attary)){
1364                 array_push($atts, "$attname=$attvalue");
1365             }
1366             $fulltag .= ' ' . join(" ", $atts);
1367         }
1368         if ($tagtype == 3){
1369             $fulltag .= ' /';
1370         }
1371         $fulltag .= '>';
1372     }
1373     return $fulltag;
1374 }
1375
1376 /**
1377  * A small helper function to use with array_walk. Modifies a by-ref
1378  * value and makes it lowercase.
1379  *
1380  * @param  $val a value passed by-ref.
1381  * @return      void since it modifies a by-ref value.
1382  */
1383 function sq_casenormalize(&$val){
1384     $val = strtolower($val);
1385 }
1386
1387 /**
1388  * This function skips any whitespace from the current position within
1389  * a string and to the next non-whitespace value.
1390  *
1391  * @param  $body   the string
1392  * @param  $offset the offset within the string where we should start
1393  *                 looking for the next non-whitespace character.
1394  * @return         the location within the $body where the next
1395  *                 non-whitespace char is located.
1396  */
1397 function sq_skipspace($body, $offset){
1398     $me = 'sq_skipspace';
1399     preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1400     if (!empty($matches[1])){
1401         $offset += strlen($matches[1]);
1402     }
1403     return $offset;
1404 }
1405
1406 /**
1407  * This function looks for the next character within a string.  It's
1408  * really just a glorified "strpos", except it catches if failures
1409  * nicely.
1410  *
1411  * @param  $body   The string to look for needle in.
1412  * @param  $offset Start looking from this position.
1413  * @param  $needle The character/string to look for.
1414  * @return         location of the next occurance of the needle, or
1415  *                 strlen($body) if needle wasn't found.
1416  */
1417 function sq_findnxstr($body, $offset, $needle){
1418     $me  = 'sq_findnxstr';
1419     $pos = strpos($body, $needle, $offset);
1420     if ($pos === FALSE){
1421         $pos = strlen($body);
1422     }
1423     return $pos;
1424 }
1425
1426 /**
1427  * This function takes a PCRE-style regexp and tries to match it
1428  * within the string.
1429  *
1430  * @param  $body   The string to look for needle in.
1431  * @param  $offset Start looking from here.
1432  * @param  $reg    A PCRE-style regex to match.
1433  * @return         Returns a false if no matches found, or an array
1434  *                 with the following members:
1435  *                 - integer with the location of the match within $body
1436  *                 - string with whatever content between offset and the match
1437  *                 - string with whatever it is we matched
1438  */
1439 function sq_findnxreg($body, $offset, $reg){
1440     $me = 'sq_findnxreg';
1441     $matches = Array();
1442     $retarr = Array();
1443     preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1444     if (!isset($matches{0}) || !$matches{0}){
1445         $retarr = false;
1446     } else {
1447         $retarr{0} = $offset + strlen($matches{1});
1448         $retarr{1} = $matches{1};
1449         $retarr{2} = $matches{2};
1450     }
1451     return $retarr;
1452 }
1453
1454 /**
1455  * This function looks for the next tag.
1456  *
1457  * @param  $body   String where to look for the next tag.
1458  * @param  $offset Start looking from here.
1459  * @return         false if no more tags exist in the body, or
1460  *                 an array with the following members:
1461  *                 - string with the name of the tag
1462  *                 - array with attributes and their values
1463  *                 - integer with tag type (1, 2, or 3)
1464  *                 - integer where the tag starts (starting "<")
1465  *                 - integer where the tag ends (ending ">")
1466  *                 first three members will be false, if the tag is invalid.
1467  */
1468 function sq_getnxtag($body, $offset){
1469     $me = 'sq_getnxtag';
1470     if ($offset > strlen($body)){
1471         return false;
1472     }
1473     $lt = sq_findnxstr($body, $offset, "<");
1474     if ($lt == strlen($body)){
1475         return false;
1476     }
1477     /**
1478      * We are here:
1479      * blah blah <tag attribute="value">
1480      * \---------^
1481      */
1482     $pos = sq_skipspace($body, $lt+1);
1483     if ($pos >= strlen($body)){
1484         return Array(false, false, false, $lt, strlen($body));
1485     }
1486     /**
1487      * There are 3 kinds of tags:
1488      * 1. Opening tag, e.g.:
1489      *    <a href="blah">
1490      * 2. Closing tag, e.g.:
1491      *    </a>
1492      * 3. XHTML-style content-less tag, e.g.:
1493      *    <img src="blah" />
1494      */
1495     $tagtype = false;
1496     switch (substr($body, $pos, 1)){
1497         case '/':
1498             $tagtype = 2;
1499             $pos++;
1500             break;
1501         case '!':
1502             /**
1503              * A comment or an SGML declaration.
1504              */
1505             if (substr($body, $pos+1, 2) == "--"){
1506                 $gt = strpos($body, "-->", $pos);
1507                 if ($gt === false){
1508                     $gt = strlen($body);
1509                 } else {
1510                     $gt += 2;
1511                 }
1512                 return Array(false, false, false, $lt, $gt);
1513             } else {
1514                 $gt = sq_findnxstr($body, $pos, ">");
1515                 return Array(false, false, false, $lt, $gt);
1516             }
1517             break;
1518         default:
1519             /**
1520              * Assume tagtype 1 for now. If it's type 3, we'll switch values
1521              * later.
1522              */
1523             $tagtype = 1;
1524             break;
1525     }
1526
1527     $tag_start = $pos;
1528     $tagname = '';
1529     /**
1530      * Look for next [\W-_], which will indicate the end of the tag name.
1531      */
1532     $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1533     if ($regary == false){
1534         return Array(false, false, false, $lt, strlen($body));
1535     }
1536     list($pos, $tagname, $match) = $regary;
1537     $tagname = strtolower($tagname);
1538
1539     /**
1540      * $match can be either of these:
1541      * '>'  indicating the end of the tag entirely.
1542      * '\s' indicating the end of the tag name.
1543      * '/'  indicating that this is type-3 xhtml tag.
1544      *
1545      * Whatever else we find there indicates an invalid tag.
1546      */
1547     switch ($match){
1548         case '/':
1549             /**
1550              * This is an xhtml-style tag with a closing / at the
1551              * end, like so: <img src="blah" />. Check if it's followed
1552              * by the closing bracket. If not, then this tag is invalid
1553              */
1554             if (substr($body, $pos, 2) == "/>"){
1555                 $pos++;
1556                 $tagtype = 3;
1557             } else {
1558                 $gt = sq_findnxstr($body, $pos, ">");
1559                 $retary = Array(false, false, false, $lt, $gt);
1560                 return $retary;
1561             }
1562         case '>':
1563             return Array($tagname, false, $tagtype, $lt, $pos);
1564             break;
1565         default:
1566             /**
1567              * Check if it's whitespace
1568              */
1569             if (!preg_match('/\s/', $match)){
1570                 /**
1571                  * This is an invalid tag! Look for the next closing ">".
1572                  */
1573                 $gt = sq_findnxstr($body, $lt, ">");
1574                 return Array(false, false, false, $lt, $gt);
1575             }
1576             break;
1577     }
1578
1579     /**
1580      * At this point we're here:
1581      * <tagname  attribute='blah'>
1582      * \-------^
1583      *
1584      * At this point we loop in order to find all attributes.
1585      */
1586     $attname = '';
1587     $atttype = false;
1588     $attary = Array();
1589
1590     while ($pos <= strlen($body)){
1591         $pos = sq_skipspace($body, $pos);
1592         if ($pos == strlen($body)){
1593             /**
1594              * Non-closed tag.
1595              */
1596             return Array(false, false, false, $lt, $pos);
1597         }
1598         /**
1599          * See if we arrived at a ">" or "/>", which means that we reached
1600          * the end of the tag.
1601          */
1602         $matches = Array();
1603         if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1604             /**
1605              * Yep. So we did.
1606              */
1607             $pos += strlen($matches{1});
1608             if ($matches{2} == "/>"){
1609                 $tagtype = 3;
1610                 $pos++;
1611             }
1612             return Array($tagname, $attary, $tagtype, $lt, $pos);
1613         }
1614
1615         /**
1616          * There are several types of attributes, with optional
1617          * [:space:] between members.
1618          * Type 1:
1619          *   attrname[:space:]=[:space:]'CDATA'
1620          * Type 2:
1621          *   attrname[:space:]=[:space:]"CDATA"
1622          * Type 3:
1623          *   attr[:space:]=[:space:]CDATA
1624          * Type 4:
1625          *   attrname
1626          *
1627          * We leave types 1 and 2 the same, type 3 we check for
1628          * '"' and convert to "&quot" if needed, then wrap in
1629          * double quotes. Type 4 we convert into:
1630          * attrname="yes".
1631          */
1632         $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1633         if ($regary == false){
1634             /**
1635              * Looks like body ended before the end of tag.
1636              */
1637             return Array(false, false, false, $lt, strlen($body));
1638         }
1639         list($pos, $attname, $match) = $regary;
1640         $attname = strtolower($attname);
1641         /**
1642          * We arrived at the end of attribute name. Several things possible
1643          * here:
1644          * '>'  means the end of the tag and this is attribute type 4
1645          * '/'  if followed by '>' means the same thing as above
1646          * '\s' means a lot of things -- look what it's followed by.
1647          *      anything else means the attribute is invalid.
1648          */
1649         switch($match){
1650             case '/':
1651                 /**
1652                  * This is an xhtml-style tag with a closing / at the
1653                  * end, like so: <img src="blah" />. Check if it's followed
1654                  * by the closing bracket. If not, then this tag is invalid
1655                  */
1656                 if (substr($body, $pos, 2) == "/>"){
1657                     $pos++;
1658                     $tagtype = 3;
1659                 } else {
1660                     $gt = sq_findnxstr($body, $pos, ">");
1661                     $retary = Array(false, false, false, $lt, $gt);
1662                     return $retary;
1663                 }
1664             case '>':
1665                 $attary{$attname} = '"yes"';
1666                 return Array($tagname, $attary, $tagtype, $lt, $pos);
1667                 break;
1668             default:
1669                 /**
1670                  * Skip whitespace and see what we arrive at.
1671                  */
1672                 $pos = sq_skipspace($body, $pos);
1673                 $char = substr($body, $pos, 1);
1674                 /**
1675                  * Two things are valid here:
1676                  * '=' means this is attribute type 1 2 or 3.
1677                  * \w means this was attribute type 4.
1678                  * anything else we ignore and re-loop. End of tag and
1679                  * invalid stuff will be caught by our checks at the beginning
1680                  * of the loop.
1681                  */
1682                 if ($char == "="){
1683                     $pos++;
1684                     $pos = sq_skipspace($body, $pos);
1685                     /**
1686                      * Here are 3 possibilities:
1687                      * "'"  attribute type 1
1688                      * '"'  attribute type 2
1689                      * everything else is the content of tag type 3
1690                      */
1691                     $quot = substr($body, $pos, 1);
1692                     if ($quot == "'"){
1693                         $regary = sq_findnxreg($body, $pos+1, "\'");
1694                         if ($regary == false){
1695                             return Array(false, false, false, $lt, strlen($body));
1696                         }
1697                         list($pos, $attval, $match) = $regary;
1698                         $pos++;
1699                         $attary{$attname} = "'" . $attval . "'";
1700                     } else if ($quot == '"'){
1701                         $regary = sq_findnxreg($body, $pos+1, '\"');
1702                         if ($regary == false){
1703                             return Array(false, false, false, $lt, strlen($body));
1704                         }
1705                         list($pos, $attval, $match) = $regary;
1706                         $pos++;
1707                         $attary{$attname} = '"' . $attval . '"';
1708                     } else {
1709                         /**
1710                          * These are hateful. Look for \s, or >.
1711                          */
1712                         $regary = sq_findnxreg($body, $pos, "[\s>]");
1713                         if ($regary == false){
1714                             return Array(false, false, false, $lt, strlen($body));
1715                         }
1716                         list($pos, $attval, $match) = $regary;
1717                         /**
1718                          * If it's ">" it will be caught at the top.
1719                          */
1720                         $attval = preg_replace("/\"/s", "&quot;", $attval);
1721                         $attary{$attname} = '"' . $attval . '"';
1722                     }
1723                 } else if (preg_match("|[\w/>]|", $char)) {
1724                     /**
1725                      * That was attribute type 4.
1726                      */
1727                     $attary{$attname} = '"yes"';
1728                 } else {
1729                     /**
1730                      * An illegal character. Find next '>' and return.
1731                      */
1732                     $gt = sq_findnxstr($body, $pos, ">");
1733                     return Array(false, false, false, $lt, $gt);
1734                 }
1735                 break;
1736         }
1737     }
1738     /**
1739      * The fact that we got here indicates that the tag end was never
1740      * found. Return invalid tag indication so it gets stripped.
1741      */
1742     return Array(false, false, false, $lt, strlen($body));
1743 }
1744
1745 /**
1746  * Translates entities into literal values so they can be checked.
1747  *
1748  * @param $attvalue the by-ref value to check.
1749  * @param $regex    the regular expression to check against.
1750  * @param $hex      whether the entites are hexadecimal.
1751  * @return          True or False depending on whether there were matches.
1752  */
1753 function sq_deent(&$attvalue, $regex, $hex=false){
1754     $me = 'sq_deent';
1755     $ret_match = false;
1756     // remove comments
1757     //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1758     preg_match_all($regex, $attvalue, $matches);
1759     if (is_array($matches) && sizeof($matches[0]) > 0){
1760         $repl = Array();
1761         for ($i = 0; $i < sizeof($matches[0]); $i++){
1762             $numval = $matches[1][$i];
1763             if ($hex){
1764                 $numval = hexdec($numval);
1765             }
1766             $repl{$matches[0][$i]} = chr($numval);
1767         }
1768         $attvalue = strtr($attvalue, $repl);
1769         return true;
1770     } else {
1771         return false;
1772     }
1773 }
1774
1775 /**
1776  * This function runs various checks against the attributes.
1777  *
1778  * @param  $tagname         String with the name of the tag.
1779  * @param  $attary          Array with all tag attributes.
1780  * @param  $rm_attnames     See description for sq_sanitize
1781  * @param  $bad_attvals     See description for sq_sanitize
1782  * @param  $add_attr_to_tag See description for sq_sanitize
1783  * @param  $message         message object
1784  * @param  $id              message id
1785  * @return                  Array with modified attributes.
1786  */
1787 function sq_fixatts($tagname,
1788                     $attary,
1789                     $rm_attnames,
1790                     $bad_attvals,
1791                     $add_attr_to_tag,
1792                     $message,
1793                     $id,
1794                     $mailbox
1795                     ){
1796     $me = 'sq_fixatts';
1797     while (list($attname, $attvalue) = each($attary)){
1798         /**
1799          * See if this attribute should be removed.
1800          */
1801         foreach ($rm_attnames as $matchtag=>$matchattrs){
1802             if (preg_match($matchtag, $tagname)){
1803                 foreach ($matchattrs as $matchattr){
1804                     if (preg_match($matchattr, $attname)){
1805                         unset($attary{$attname});
1806                         continue;
1807                     }
1808                 }
1809             }
1810         }
1811         /**
1812          * Workaround for IE quirks
1813          */
1814         sq_fixIE_idiocy($attvalue);
1815
1816         /**
1817          * Remove any backslashes, entities, and extraneous whitespace.
1818          */
1819
1820         $oldattvalue = $attvalue;
1821         sq_defang($attvalue);
1822         if ($attname == 'style' && $attvalue !== $oldattvalue) {
1823             // entities are used in the attribute value. In 99% of the cases it's there as XSS
1824             // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1825             $attvalue = "idiocy";
1826             $attary{$attname} = $attvalue;
1827         }
1828         sq_unspace($attvalue);
1829
1830         /**
1831          * Now let's run checks on the attvalues.
1832          * I don't expect anyone to comprehend this. If you do,
1833          * get in touch with me so I can drive to where you live and
1834          * shake your hand personally. :)
1835          */
1836         foreach ($bad_attvals as $matchtag=>$matchattrs){
1837             if (preg_match($matchtag, $tagname)){
1838                 foreach ($matchattrs as $matchattr=>$valary){
1839                     if (preg_match($matchattr, $attname)){
1840                         /**
1841                          * There are two arrays in valary.
1842                          * First is matches.
1843                          * Second one is replacements
1844                          */
1845                         list($valmatch, $valrepl) = $valary;
1846                         $newvalue =
1847                             preg_replace($valmatch, $valrepl, $attvalue);
1848                         if ($newvalue != $attvalue){
1849                             $attary{$attname} = $newvalue;
1850                             $attvalue = $newvalue;
1851                         }
1852                     }
1853                 }
1854             }
1855         }
1856         if ($attname == 'style') {
1857             if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1858                 // 8bit and control characters in style attribute values can be used for XSS, remove them
1859                 $attary{$attname} = '"disallowed character"';
1860             }
1861             preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1862             if (count($aMatch)) {
1863                 foreach($aMatch[1] as $sMatch) {
1864                     // url value
1865                     $urlvalue = $sMatch;
1866                     sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1867                     $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1868                 }
1869             }
1870         }
1871         /**
1872          * Use white list based filtering on attributes which can contain url's
1873          */
1874         else if ($attname == 'href' || $attname == 'xlink:href' || $attname == 'src'
1875               || $attname == 'poster' || $attname == 'formaction'
1876               || $attname == 'background' || $attname == 'action') {
1877             sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1878             $attary{$attname} = $attvalue;
1879         }
1880     }
1881     /**
1882      * See if we need to append any attributes to this tag.
1883      */
1884     foreach ($add_attr_to_tag as $matchtag=>$addattary){
1885         if (preg_match($matchtag, $tagname)){
1886             $attary = array_merge($attary, $addattary);
1887         }
1888     }
1889     return $attary;
1890 }
1891
1892 /**
1893  * This function filters url's
1894  *
1895  * @param  $attvalue        String with attribute value to filter
1896  * @param  $message         message object
1897  * @param  $id               message id
1898  * @param  $mailbox         mailbox
1899  * @param  $sQuote          quoting characters around url's
1900  */
1901 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1902     $attvalue = trim($attvalue);
1903     if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1904         // remove the double quotes
1905         $sQuote = $attvalue[0];
1906         $attvalue = trim(substr($attvalue,1,-1));
1907     }
1908
1909     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1910     // images off by default.
1911     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1912
1913     global $use_transparent_security_image;
1914     if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1915     else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1916
1917     /**
1918      * Replace empty src tags with the blank image.  src is only used
1919      * for frames, images, and image inputs.  Doing a replace should
1920      * not affect them working as should be, however it will stop
1921      * IE from being kicked off when src for img tags are not set
1922      */
1923     if ($attvalue == '') {
1924         $attvalue = '"' . SM_PATH . 'images/blank.png"';
1925     } else {
1926         // first, disallow 8 bit characters and control characters
1927         if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1928             switch ($attname) {
1929                 case 'href':
1930                     $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1931                     break;
1932                 default:
1933                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1934                     break;
1935             }
1936         } else {
1937             $aUrl = parse_url($attvalue);
1938             if (isset($aUrl['scheme'])) {
1939                 switch(strtolower($aUrl['scheme'])) {
1940                     case 'mailto':
1941                     case 'http':
1942                     case 'https':
1943                     case 'ftp':
1944                         if ($attname != 'href') {
1945                             if ($view_unsafe_images == false) {
1946                                 $attvalue = $sQuote . $secremoveimg . $sQuote;
1947                             } else {
1948                                 if (isset($aUrl['path'])) {
1949
1950                                     // No one has been able to show that image URIs
1951                                     // can be exploited, so for now, no restrictions
1952                                     // are made at all.  If this proves to be a problem,
1953                                     // the commented-out code below can be of help.
1954                                     // (One consideration is that I see nothing in this
1955                                     // function that specifically says that we will
1956                                     // only ever arrive here when inspecting an image
1957                                     // tag, although that does seem to be the end
1958                                     // result - e.g., <script src="..."> where malicious
1959                                     // image URIs are in fact a problem are already
1960                                     // filtered out elsewhere.
1961                                     /* ---------------------------------
1962                                     // validate image extension.
1963                                     $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1964                                     if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1965                                         // If URI is to something other than
1966                                         // a regular image file, get the contents
1967                                         // and try to see if it is an image.
1968                                         // Don't use Fileinfo (finfo_file()) because
1969                                         // we'd need to make the admin configure the
1970                                         // location of the magic.mime file (FIXME: add finfo_file() support later?)
1971                                         //
1972                                         $mime_type = '';
1973                                         if (function_exists('mime_content_type')
1974                                          && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1975
1976                                             // fetch file
1977                                             //
1978                                             $file_contents = '';
1979                                             while (!feof($FILE)) {
1980                                                 $file_contents .= fread($FILE, 8192);
1981                                             }
1982                                             fclose($FILE);
1983
1984                                             // store file locally
1985                                             //
1986                                             global $attachment_dir, $username;
1987                                             $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1988                                             $localfilename = GenerateRandomString(32, '', 7);
1989                                             $full_localfilename = "$hashed_attachment_dir/$localfilename";
1990                                             while (file_exists($full_localfilename)) {
1991                                                 $localfilename = GenerateRandomString(32, '', 7);
1992                                                 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1993                                             }
1994                                             $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
1995                                             fwrite($FILE, $file_contents);
1996                                             fclose($FILE);
1997
1998                                             // get mime type and remove file
1999                                             //
2000                                             $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2001                                             unlink("$hashed_attachment_dir/$localfilename");
2002                                         }
2003                                         // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2004                                         if (substr(strtolower($mime_type), 0, 5) != 'image') {
2005                                             $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2006                                         }
2007                                     }
2008                                     --------------------------------- */
2009                                 } else {
2010                                     $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2011                                 }
2012                             }
2013                         } else {
2014                             $attvalue = $sQuote . $attvalue . $sQuote;
2015                         }
2016                         break;
2017                     case 'outbind':
2018                         /**
2019                          * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2020                          * One day MS might actually make it match something useful, for now, falling
2021                          * back to using cid2http, so we can grab the blank.png.
2022                          */
2023                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2024                         break;
2025                     case 'cid':
2026                         /**
2027                             * Turn cid: urls into http-friendly ones.
2028                             */
2029                         $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2030                         break;
2031                     default:
2032                         $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2033                         break;
2034                 }
2035             } else {
2036                 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2037                     // parse_url did not lead to satisfying result
2038                     $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2039                 }
2040             }
2041         }
2042     }
2043 }
2044
2045 /**
2046  * This function edits the style definition to make them friendly and
2047  * usable in SquirrelMail.
2048  *
2049  * @param  $message  the message object
2050  * @param  $id       the message id
2051  * @param  $content  a string with whatever is between <style> and </style>
2052  * @param  $mailbox  the message mailbox
2053  * @return           a string with edited content.
2054  */
2055 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2056     $me = 'sq_fixstyle';
2057     // workaround for </style> in between comments
2058     $iCurrentPos = $pos;
2059     $content = '';
2060     $sToken = '';
2061     $bSucces = false;
2062     $bEndTag = false;
2063     for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2064         $char = $body{$i};
2065         switch ($char) {
2066             case '<':
2067                 $sToken = $char;
2068                 break;
2069             case '/':
2070                  if ($sToken == '<') {
2071                     $sToken .= $char;
2072                     $bEndTag = true;
2073                  } else {
2074                     $content .= $char;
2075                  }
2076                  break;
2077             case '>':
2078                  if ($bEndTag) {
2079                     $sToken .= $char;
2080                     if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2081                         $newpos = $i + 1;
2082                         $bSucces = true;
2083                         break 2;
2084                     } else {
2085                         $content .= $sToken;
2086                     }
2087                     $bEndTag = false;
2088                  } else {
2089                     $content .= $char;
2090                  }
2091                  break;
2092             case '!':
2093                 if ($sToken == '<') {
2094                     // possible comment
2095                     if (isset($body{$i+2}) && substr($body,$i,3) == '!--') {
2096                         $i = strpos($body,'-->',$i+3);
2097                         if ($i === false) { // no end comment
2098                             $i = strlen($body);
2099                         }
2100                         $sToken = '';
2101                     }
2102                 } else {
2103                     $content .= $char;
2104                 }
2105                 break;
2106             default:
2107                 if ($bEndTag) {
2108                     $sToken .= $char;
2109                 } else {
2110                     $content .= $char;
2111                 }
2112                 break;
2113         }
2114     }
2115     if ($bSucces == FALSE){
2116         return array(FALSE, strlen($body));
2117     }
2118
2119
2120
2121     /**
2122      * First look for general BODY style declaration, which would be
2123      * like so:
2124      * body {background: blah-blah}
2125      * and change it to .bodyclass so we can just assign it to a <div>
2126      */
2127     // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2128     // Nah, this is even better - try to preface all CSS selectors with
2129     // our <div> class ID "bodyclass" then correct generic "body" selectors
2130     // TODO: this works pretty good but breaks stuff like this:
2131     //       @media print { body { font-size: 10pt; } }
2132     //       but there isn't an easy way to make this regex skip @media
2133     //       definitions... though lots of the ones in the wild will be
2134     //       correctly handled because they tend to end with a parenthesis, like:
2135     //       @media screen and (max-width:480px) { ...
2136     $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2137     $content = str_replace('.bodyclass body', '.bodyclass', $content);
2138
2139     global $use_transparent_security_image;
2140     if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2141     else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2142
2143     /**
2144     * Fix url('blah') declarations.
2145     */
2146     //   $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2147     //                           "url(\\1$secremoveimg\\2)", $content);
2148
2149     // first check for 8bit sequences and disallowed control characters
2150     if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2151         $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2152         return array($content, $newpos);
2153     }
2154
2155     // IE Sucks hard. We have a special function for it.
2156     sq_fixIE_idiocy($content);
2157
2158     // remove @import line
2159     $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2160
2161     // translate ur\l and variations (IE parses that)
2162     // TODO check if the sq_fixIE_idiocy function already handles this.
2163     $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2164     preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2165     if (count($aMatch)) {
2166         $aValue = $aReplace = array();
2167         foreach($aMatch[1] as $sMatch) {
2168             // url value
2169             $urlvalue = $sMatch;
2170             sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2171             $aValue[] = $sMatch;
2172             $aReplace[] = $urlvalue;
2173         }
2174         $content = str_replace($aValue,$aReplace,$content);
2175     }
2176
2177     /**
2178      * Remove any backslashes, entities, and extraneous whitespace.
2179      */
2180     $contentTemp = $content;
2181     sq_defang($contentTemp);
2182     sq_unspace($contentTemp);
2183
2184     /**
2185      * Fix stupid css declarations which lead to vulnerabilities
2186      * in IE.
2187      *
2188      * Also remove "position" attribute, as it can easily be set
2189      * to "fixed" or "absolute" with "left" and "top" attributes
2190      * of zero, taking over the whole content frame.  It can also
2191      * be set to relative and move itself anywhere it wants to,
2192      * displaying content in areas it shouldn't be allowed to touch.
2193      */
2194     $match   = Array('/\/\*.*\*\//', // removes /* blah blah */
2195                     '/expression/i',
2196                     '/behaviou*r/i',
2197                     '/binding/i',
2198                     '/include-source/i',
2199                     '/javascript/i',
2200                     '/script/i',
2201                     '/position/i');
2202     $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2203     $contentNew = preg_replace($match, $replace, $contentTemp);
2204     if ($contentNew !== $contentTemp) {
2205         // insecure css declarations are used. From now on we don't care
2206         // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2207         $content = $contentNew;
2208     }
2209     return array($content, $newpos);
2210 }
2211
2212
2213 /**
2214  * This function converts cid: url's into the ones that can be viewed in
2215  * the browser.
2216  *
2217  * @param  $message  the message object
2218  * @param  $id       the message id
2219  * @param  $cidurl   the cid: url.
2220  * @param  $mailbox  the message mailbox
2221  * @return           a string with a http-friendly url
2222  */
2223 function sq_cid2http($message, $id, $cidurl, $mailbox){
2224     /**
2225      * Get rid of quotes.
2226      */
2227     $quotchar = substr($cidurl, 0, 1);
2228     if ($quotchar == '"' || $quotchar == "'"){
2229         $cidurl = str_replace($quotchar, "", $cidurl);
2230     } else {
2231         $quotchar = '';
2232     }
2233     $cidurl = substr(trim($cidurl), 4);
2234
2235     $match_str = '/\{.*?\}\//';
2236     $str_rep = '';
2237     $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2238
2239     $linkurl = find_ent_id($cidurl, $message);
2240     /* in case of non-safe cid links $httpurl should be replaced by a sort of
2241        unsafe link image */
2242     $httpurl = '';
2243
2244     /**
2245      * This is part of a fix for Outlook Express 6.x generating
2246      * cid URLs without creating content-id headers. These images are
2247      * not part of the multipart/related html mail. The html contains
2248      * <img src="cid:{some_id}/image_filename.ext"> references to
2249      * attached images with as goal to render them inline although
2250      * the attachment disposition property is not inline.
2251      */
2252
2253     if (empty($linkurl)) {
2254         if (preg_match('/{.*}\//', $cidurl)) {
2255             $cidurl = preg_replace('/{.*}\//','', $cidurl);
2256             if (!empty($cidurl)) {
2257                 $linkurl = find_ent_id($cidurl, $message);
2258             }
2259         }
2260     }
2261
2262     if (!empty($linkurl)) {
2263         $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2264             "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2265             '&amp;ent_id=' . $linkurl . $quotchar;
2266     } else {
2267         /**
2268          * If we couldn't generate a proper img url, drop in a blank image
2269          * instead of sending back empty, otherwise it causes unusual behaviour
2270          */
2271         $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2272     }
2273
2274     return $httpurl;
2275 }
2276
2277 /**
2278  * This function changes the <body> tag into a <div> tag since we
2279  * can't really have a body-within-body.
2280  *
2281  * @param  $attary   an array of attributes and values of <body>
2282  * @param  $mailbox  mailbox we're currently reading (for cid2http)
2283  * @param  $message  current message (for cid2http)
2284  * @param  $id       current message id (for cid2http)
2285  * @return           a modified array of attributes to be set for <div>
2286  */
2287 function sq_body2div($attary, $mailbox, $message, $id){
2288     $me = 'sq_body2div';
2289     $divattary = Array('class' => "'bodyclass'");
2290     $text = '#000000';
2291     $has_bgc_stl = $has_txt_stl = false;
2292     $styledef = '';
2293     if (is_array($attary) && sizeof($attary) > 0){
2294         foreach ($attary as $attname=>$attvalue){
2295             $quotchar = substr($attvalue, 0, 1);
2296             $attvalue = str_replace($quotchar, "", $attvalue);
2297             switch ($attname){
2298                 case 'background':
2299                     $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2300                     $styledef .= "background-image: url('$attvalue'); ";
2301                     break;
2302                 case 'bgcolor':
2303                     $has_bgc_stl = true;
2304                     $styledef .= "background-color: $attvalue; ";
2305                     break;
2306                 case 'text':
2307                     $has_txt_stl = true;
2308                     $styledef .= "color: $attvalue; ";
2309                     break;
2310             }
2311         }
2312         // Outlook defines a white bgcolor and no text color. This can lead to
2313         // white text on a white bg with certain themes.
2314         if ($has_bgc_stl && !$has_txt_stl) {
2315             $styledef .= "color: $text; ";
2316         }
2317         if (strlen($styledef) > 0){
2318             $divattary{"style"} = "\"$styledef\"";
2319         }
2320     }
2321     return $divattary;
2322 }
2323
2324 /**
2325  * This is the main function and the one you should actually be calling.
2326  * There are several variables you should be aware of an which need
2327  * special description.
2328  *
2329  * Since the description is quite lengthy, see it here:
2330  * http://linux.duke.edu/projects/mini/htmlfilter/
2331  *
2332  * @param $body                 the string with HTML you wish to filter
2333  * @param $tag_list             see description above
2334  * @param $rm_tags_with_content see description above
2335  * @param $self_closing_tags    see description above
2336  * @param $force_tag_closing    see description above
2337  * @param $rm_attnames          see description above
2338  * @param $bad_attvals          see description above
2339  * @param $add_attr_to_tag      see description above
2340  * @param $message              message object
2341  * @param $id                   message id
2342  * @param $recursively_called   boolean flag for recursive calls into this function (optional; default FALSE)
2343  * @return                      sanitized html safe to show on your pages.
2344  */
2345 function sq_sanitize($body,
2346                      $tag_list,
2347                      $rm_tags_with_content,
2348                      $self_closing_tags,
2349                      $force_tag_closing,
2350                      $rm_attnames,
2351                      $bad_attvals,
2352                      $add_attr_to_tag,
2353                      $message,
2354                      $id,
2355                      $mailbox,
2356                      $recursively_called=FALSE
2357                      ){
2358     $me = 'sq_sanitize';
2359
2360     /**
2361      * See if tag_list is of tags to remove or tags to allow.
2362      * false  means remove these tags
2363      * true   means allow these tags
2364      */
2365     $orig_tag_list = $tag_list;
2366     $rm_tags = array_shift($tag_list);
2367
2368     /**
2369      * Normalize rm_tags and rm_tags_with_content.
2370      */
2371     @array_walk($tag_list, 'sq_casenormalize');
2372     @array_walk($rm_tags_with_content, 'sq_casenormalize');
2373     @array_walk($self_closing_tags, 'sq_casenormalize');
2374
2375     $curpos = 0;
2376     $open_tags = Array();
2377     $trusted = "\n<!-- begin sanitized html -->\n";
2378     $skip_content = false;
2379     /**
2380      * Take care of netscape's stupid javascript entities like
2381      * &{alert('boo')};
2382      */
2383     $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2384
2385     while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2386         list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2387
2388         /**
2389          * RCDATA and RAWTEXT tags are handled differently:
2390          * next instance of closing tag is used, whether or not
2391          * the HTML is well formed before that
2392          */
2393         global $rcdata_rawtext_tags;
2394         if (!$recursively_called
2395          && in_array($tagname, $rcdata_rawtext_tags)
2396          && $tagtype === 1){
2397             $closing_tag = false;
2398             $closing_tag_offset = $curpos;
2399             // seek out the closing tag for the current RCDATA/RAWTEXT tag
2400             while (1) {
2401                 // first we need to move forward to next available closing tag
2402                 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2403                 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2404                 if ($next_tag === false) {
2405                     $closing_tag = false;
2406                     break;
2407                 }
2408                 // but then we have to make sure it's a well-formed tag
2409                 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2410                 if ($closing_tag === false)
2411                     break;
2412                 else if ($closing_tag[0] !== false
2413                  // these should be redundant
2414                  && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2415                     $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos + 1),
2416                                             $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2417                                             $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2418                                             $message, $id, $mailbox, true);
2419                     $curpos = $closing_tag[4] + 1;
2420                     continue 2;
2421                 }
2422                 $closing_tag_offset = $next_tag[0] + 1;
2423             }
2424             if ($closing_tag === false)
2425             { /* 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 */ }
2426         }
2427
2428         $free_content = substr($body, $curpos, $lt-$curpos);
2429         /**
2430          * Take care of <style>
2431          */
2432         if ($tagname == "style" && $tagtype == 1){
2433             list($free_content, $curpos) =
2434                 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2435             if ($free_content != FALSE){
2436                 if ( !empty($attary) ) {
2437                     $attary = sq_fixatts($tagname,
2438                                          $attary,
2439                                          $rm_attnames,
2440                                          $bad_attvals,
2441                                          $add_attr_to_tag,
2442                                          $message,
2443                                          $id,
2444                                          $mailbox
2445                                          );
2446                 }
2447                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2448                 $trusted .= $free_content;
2449                 $trusted .= sq_tagprint($tagname, false, 2);
2450             }
2451             continue;
2452         }
2453         if ($skip_content == false){
2454             $trusted .= $free_content;
2455         }
2456         if ($tagname != FALSE){
2457             if ($tagtype == 2){
2458                 if ($skip_content == $tagname){
2459                     /**
2460                      * Got to the end of tag we needed to remove.
2461                      */
2462                     $tagname = false;
2463                     $skip_content = false;
2464                 } else {
2465                     if ($skip_content == false){
2466                         if ($tagname == "body"){
2467                             $tagname = "div";
2468                         }
2469                         if (isset($open_tags{$tagname}) &&
2470                                 $open_tags{$tagname} > 0){
2471                             $open_tags{$tagname}--;
2472                         } else {
2473                             $tagname = false;
2474                         }
2475                     }
2476                 }
2477             } else {
2478                 /**
2479                  * $rm_tags_with_content
2480                  */
2481                 if ($skip_content == false){
2482                     /**
2483                      * See if this is a self-closing type and change
2484                      * tagtype appropriately.
2485                      */
2486                     if ($tagtype == 1
2487                             && in_array($tagname, $self_closing_tags)){
2488                         $tagtype = 3;
2489                     }
2490                     /**
2491                      * See if we should skip this tag and any content
2492                      * inside it.
2493                      */
2494                     if ($tagtype == 1 &&
2495                             in_array($tagname, $rm_tags_with_content)){
2496                         $skip_content = $tagname;
2497                     } else {
2498                         if (($rm_tags == false
2499                                     && in_array($tagname, $tag_list)) ||
2500                                 ($rm_tags == true &&
2501                                  !in_array($tagname, $tag_list))){
2502                             $tagname = false;
2503                         } else {
2504                             /**
2505                              * Convert body into div.
2506                              */
2507                             if ($tagname == "body"){
2508                                 $tagname = "div";
2509                                 $attary = sq_body2div($attary, $mailbox,
2510                                         $message, $id);
2511                             }
2512                             if ($tagtype == 1){
2513                                 if (isset($open_tags{$tagname})){
2514                                     $open_tags{$tagname}++;
2515                                 } else {
2516                                     $open_tags{$tagname}=1;
2517                                 }
2518                             }
2519                             /**
2520                              * This is where we run other checks.
2521                              */
2522                             if (is_array($attary) && sizeof($attary) > 0){
2523                                 $attary = sq_fixatts($tagname,
2524                                                      $attary,
2525                                                      $rm_attnames,
2526                                                      $bad_attvals,
2527                                                      $add_attr_to_tag,
2528                                                      $message,
2529                                                      $id,
2530                                                      $mailbox
2531                                                      );
2532                             }
2533                         }
2534                     }
2535                 }
2536             }
2537             if ($tagname != false && $skip_content == false){
2538                 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2539             }
2540         }
2541         $curpos = $gt+1;
2542     }
2543     $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2544     if ($force_tag_closing == true){
2545         foreach ($open_tags as $tagname=>$opentimes){
2546             while ($opentimes > 0){
2547                 $trusted .= '</' . $tagname . '>';
2548                 $opentimes--;
2549             }
2550         }
2551         $trusted .= "\n";
2552     }
2553     $trusted .= "<!-- end sanitized html -->\n";
2554     return $trusted;
2555 }
2556
2557 /**
2558  * This is a wrapper function to call html sanitizing routines.
2559  *
2560  * @param  $body  the body of the message
2561  * @param  $id    the id of the message
2562
2563  * @param  $message
2564  * @param  $mailbox
2565  * @param  boolean $take_mailto_links When TRUE, converts mailto: links
2566  *                                    into internal SM compose links
2567  *                                    (optional; default = TRUE)
2568  * @return        a string with html safe to display in the browser.
2569  */
2570 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2571
2572     // require_once(SM_PATH . 'functions/url_parser.php');  // for $MailTo_PReg_Match
2573
2574     global $attachment_common_show_images, $view_unsafe_images,
2575            $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2576            $remove_rcdata_rawtext_tags_and_content;
2577
2578     $rcdata_rawtext_tags = array(
2579         "noscript",
2580         "noframes",
2581         "noembed",
2582         "textarea",
2583         // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2584     );
2585
2586     /**
2587      * Don't display attached images in HTML mode.
2588      *
2589      * SB: why?
2590      */
2591     $attachment_common_show_images = false;
2592     $tag_list = Array(
2593             false, // remove these tags
2594             "meta",
2595             "html",
2596             "head",
2597             "base",
2598             "link",
2599             "frame",
2600             "iframe",
2601             "plaintext",
2602             "marquee",
2603             );
2604
2605     $rm_tags_with_content = Array(
2606             "script",
2607             "object",
2608             "applet",
2609             "embed",
2610             "title",
2611             "frameset",
2612             "xmp",
2613             "xml",
2614             );
2615     if (!$allow_svg_display)
2616         $rm_tags_with_content[] = 'svg';
2617     /**
2618      * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2619      * case that they are, but if you prefer to remove them and their contents entirely
2620      * (in most cases, should be a safe thing with minimal impact), you can add the
2621      * following to config/config_local.php
2622      *    $remove_rcdata_rawtext_tags_and_content = TRUE;
2623      */
2624     if ($remove_rcdata_rawtext_tags_and_content)
2625         $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2626
2627     $self_closing_tags =  Array(
2628             "img",
2629             "br",
2630             "hr",
2631             "input",
2632             "outbind",
2633             );
2634
2635     $force_tag_closing = true;
2636
2637     $rm_attnames = Array(
2638             "/.*/" =>
2639             Array(
2640                 "/target/i",
2641                 "/^on.*/i",
2642                 "/^dynsrc/i",
2643                 "/^data.*/i",
2644                 "/^lowsrc.*/i",
2645                 )
2646             );
2647
2648     global $use_transparent_security_image;
2649     if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2650     else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2651
2652     $bad_attvals = Array(
2653             "/.*/" =>
2654             Array(
2655                 "/^src|background/i" =>
2656                 Array(
2657                     Array(
2658                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2659                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2660                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2661                         ),
2662                     Array(
2663                         "\\1$secremoveimg\\2",
2664                         "\\1$secremoveimg\\2",
2665                         "\\1$secremoveimg\\2",
2666                         )
2667                     ),
2668                 "/^href|action/i" =>
2669                 Array(
2670                     Array(
2671                         "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2672                         "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2673                         "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2674                         ),
2675                     Array(
2676                         "\\1#\\1",
2677                         "\\1#\\1",
2678                         "\\1#\\1"
2679                         )
2680                     ),
2681         "/^style/i" =>
2682             Array(
2683                 Array(
2684                     "/\/\*.*\*\//",
2685                     "/expression/i",
2686                     "/binding/i",
2687                     "/behaviou*r/i",
2688                     "/include-source/i",
2689
2690                     // position:relative can also be exploited
2691                     // to put content outside of email body area
2692                     // and position:fixed is similarly exploitable
2693                     // as position:absolute, so we'll remove it
2694                     // altogether....
2695                     //
2696                     // Does this screw up legitimate HTML messages?
2697                     // If so, the only fix I see is to allow position
2698                     // attributes (any values?  I think we still have
2699                     // to block static and fixed) only if $use_iframe
2700                     // is enabled (1.5.0+)
2701                     //
2702                     // was:   "/position\s*:\s*absolute/i",
2703                     //
2704                     "/position\s*:/i",
2705
2706                     "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2707                     "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2708                     "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2709                     "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2710                     "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2711                     ),
2712                 Array(
2713                     "",
2714                     "idiocy",
2715                     "idiocy",
2716                     "idiocy",
2717                     "idiocy",
2718                     "idiocy",
2719                     "url",
2720                     "url(\\1#\\1)",
2721                     "url(\\1#\\1)",
2722                     "url(\\1#\\1)",
2723                     "\\1:url(\\2#\\3)"
2724                     )
2725                 )
2726             )
2727         );
2728
2729     // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2730     // images off by default.
2731     sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2732
2733     if (!$view_unsafe_images){
2734         /**
2735          * Remove any references to http/https if view_unsafe_images set
2736          * to false.
2737          */
2738         array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2739                 '/^([\'\"])\s*https*:.*([\'\"])/si');
2740         array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2741                 "\\1$secremoveimg\\1");
2742         array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2743                 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2744         array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2745                 "url(\\1$secremoveimg\\1)");
2746     }
2747
2748     $add_attr_to_tag = Array(
2749             "/^a$/i" =>
2750             Array('target'=>'"_blank"',
2751                 'title'=>'"'._("This external link will open in a new window").'"'
2752                 )
2753             );
2754     $trusted = sq_sanitize($body,
2755                            $tag_list,
2756                            $rm_tags_with_content,
2757                            $self_closing_tags,
2758                            $force_tag_closing,
2759                            $rm_attnames,
2760                            $bad_attvals,
2761                            $add_attr_to_tag,
2762                            $message,
2763                            $id,
2764                            $mailbox
2765                            );
2766     if (strpos($trusted,$secremoveimg)){
2767         $has_unsafe_images = true;
2768     }
2769
2770     // we want to parse mailto's in HTML output, change to SM compose links
2771     // this is a modified version of code from url_parser.php... but Marc is
2772     // right: we need a better filtering implementation; adding this randomly
2773     // here is not a great solution
2774     //
2775     if ($take_mailto_links) {
2776         // parseUrl($trusted);   // this even parses URLs inside of tags... too aggressive
2777         global $MailTo_PReg_Match;
2778         // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2779         // inside an anchor (link) tag, so we have to make sure the regex looks for the
2780         // quote before mailto, and we'll also try to convert the non-links back into links
2781         $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2782         if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2783             foreach ($regs[0] as $i => $mailto_before) {
2784                 $mailto_params = $regs[11][$i];
2785
2786                 // get rid of any leading quote we may have captured but don't care about
2787                 //
2788                 $mailto_before = ltrim($mailto_before, '"\'');
2789
2790                 // get rid of any tailing quote since we have to add send_to to the end
2791                 //
2792                 $mailto_before = rtrim($mailto_before, '"\'');
2793                 $mailto_params = rtrim($mailto_params, '"\'');
2794
2795                 if ($regs[2][$i]) {    //if there is an email addr before '?', we need to merge it with the params
2796                     $to = 'to=' . $regs[2][$i];
2797                     if (strpos($mailto_params, 'to=') > -1)    //already a 'to='
2798                         $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2799                     else {
2800                         if ($mailto_params)    //already some params, append to them
2801                             $mailto_params .= '&amp;' . $to;
2802                         else
2803                             $mailto_params .= '?' . $to;
2804                     }
2805                 }
2806
2807                 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2808
2809                 // we'll already have target=_blank, no need to allow comp_in_new
2810                 // here (which would be a lot more work anyway)
2811                 //
2812                 global $compose_new_win;
2813                 $temp_comp_in_new = $compose_new_win;
2814                 $compose_new_win = 0;
2815                 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2816                 $compose_new_win = $temp_comp_in_new;
2817
2818                 // remove <a href=" and anything after the next quote (we only
2819                 // need the uri, not the link HTML) in compose uri
2820                 //
2821                 // but only do this if the original mailto was in a real anchor tag
2822                 //
2823                 if (!empty($regs[1][$i])) {
2824                     $comp_uri = substr($comp_uri, 9);
2825                     $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2826                 }
2827                 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2828             }
2829         }
2830     }
2831
2832     return $trusted;
2833 }
2834
2835 /**
2836  * function SendDownloadHeaders - send file to the browser
2837  *
2838  * Original Source: SM core src/download.php
2839  * moved here to make it available to other code, and separate
2840  * front end from back end functionality.
2841  *
2842  * @param string $type0 first half of mime type
2843  * @param string $type1 second half of mime type
2844  * @param string $filename filename to tell the browser for downloaded file
2845  * @param boolean $force whether to force the download dialog to pop
2846  * @param optional integer $filesize send the Content-Header and length to the browser
2847  * @return void
2848  */
2849 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2850     global $languages, $squirrelmail_language;
2851     $isIE = $isIE6plus = false;
2852
2853     sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2854
2855     if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2856             strstr($HTTP_USER_AGENT, 'Opera') === false) {
2857         $isIE = true;
2858     }
2859
2860     if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2861         ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2862         $isIE6plus = true;
2863     }
2864
2865     if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2866             function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2867         $filename =
2868             call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2869     } else {
2870         $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2871     }
2872
2873     // A Pox on Microsoft and it's Internet Explorer!
2874     //
2875     // IE has lots of bugs with file downloads.
2876     // It also has problems with SSL.  Both of these cause problems
2877     // for us in this function.
2878     //
2879     // See this article on Cache Control headers and SSL
2880     // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2881     //
2882     // The best thing you can do for IE is to upgrade to the latest
2883     // version
2884     //set all the Cache Control Headers for IE
2885     if ($isIE) {
2886         $filename=rawurlencode($filename);
2887         header ("Pragma: public");
2888         header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2889         // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2890         // header ("Cache-Control: post-check=0, pre-check=0", false);
2891         header ("Cache-Control: private");
2892
2893         //set the inline header for IE, we'll add the attachment header later if we need it
2894         header ("Content-Disposition: inline; filename=$filename");
2895     }
2896
2897     if (!$force) {
2898         // Try to show in browser window
2899         header ("Content-Disposition: inline; filename=\"$filename\"");
2900         header ("Content-Type: $type0/$type1; name=\"$filename\"");
2901     } else {
2902         // Try to pop up the "save as" box
2903
2904         // IE makes this hard.  It pops up 2 save boxes, or none.
2905         // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2906         // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2907         // But, according to Microsoft, it is "RFC compliant but doesn't
2908         // take into account some deviations that allowed within the
2909         // specification."  Doesn't that mean RFC non-compliant?
2910         // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2911
2912         // all browsers need the application/octet-stream header for this
2913         header ("Content-Type: application/octet-stream; name=\"$filename\"");
2914
2915         // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2916         // Do not have quotes around filename, but that applied to
2917         // "attachment"... does it apply to inline too?
2918         header ("Content-Disposition: attachment; filename=\"$filename\"");
2919
2920         if ($isIE && !$isIE6plus) {
2921             // This combination seems to work mostly.  IE 5.5 SP 1 has
2922             // known issues (see the Microsoft Knowledge Base)
2923
2924             // This works for most types, but doesn't work with Word files
2925             header ("Content-Type: application/download; name=\"$filename\"");
2926             header ("Content-Type: application/force-download; name=\"$filename\"");
2927             // These are spares, just in case.  :-)
2928             //header("Content-Type: $type0/$type1; name=\"$filename\"");
2929             //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2930             //header("Content-Type: application/octet-stream; name=\"$filename\"");
2931         } else if ($isIE) {
2932              // This is to prevent IE for MIME sniffing and auto open a file in IE
2933              header ("Content-Type: application/force-download; name=\"$filename\"");
2934         } else {
2935             // another application/octet-stream forces download for Netscape
2936             header ("Content-Type: application/octet-stream; name=\"$filename\"");
2937         }
2938     }
2939
2940     //send the content-length header if the calling function provides it
2941     if ($filesize > 0) {
2942         header("Content-Length: $filesize");
2943     }
2944
2945 }  // end fn SendDownloadHeaders