6 * This contains the functions necessary to detect and decode MIME
9 * @copyright © 1999-2009 The SquirrelMail Project Team
10 * @license http://opensource.org/licenses/gpl-license.php GNU Public License
12 * @package squirrelmail
16 * dependency information
19 class/mime/Message.class.php
20 Message::parseStructure
21 functions/page_header.php
23 functions/display_messages.php
26 functions/imap_general.php
33 functions/attachment_common.php
34 functions/display_messages.php
36 magicHtml => url_parser
37 translateText => url_parser
42 /* -------------------------------------------------------------------------- */
44 /* -------------------------------------------------------------------------- */
47 * Get the MIME structure
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.
53 function mime_structure ($bodystructure, $flags=array()) {
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));
59 $msg = Message
::parseStructure($read,$i);
61 if (!is_object($msg)) {
62 global $color, $mailbox;
63 displayPageHeader( $color, $mailbox );
64 $errormessage = _("SquirrelMail could not decode the bodystructure of the message");
65 $errormessage .= '<br />'._("The bodystructure provided by your IMAP server:").'<br /><br />';
66 $errormessage .= '<pre>' . htmlspecialchars($read) . '</pre>';
67 plain_error_message( $errormessage );
68 echo '</body></html>';
72 foreach ($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});
77 if (strtolower($flag) == '\\seen') {
82 if (strtolower($flag) == '\\answered') {
83 $msg->is_answered
= true;
87 if (strtolower($flag) == '\\deleted') {
88 $msg->is_deleted
= true;
92 if (strtolower($flag) == '\\flagged') {
93 $msg->is_flagged
= true;
95 else if (strtolower($flag) == '$forwarded') {
96 $msg->is_forwarded
= true;
100 if (strtolower($flag) == '$mdnsent') {
101 $msg->is_mdnsent
= true;
109 // listEntities($msg);
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
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()
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.
130 $cmd = "FETCH $id BODY[]";
132 $cmd = "FETCH $id BODY[$ent_id]";
135 if ($fetch_size!=0) $cmd .= "<0.$fetch_size>";
137 $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, TRUE);
139 $topline = trim(array_shift($data));
140 } while($topline && ($topline[0] == '*') && !preg_match('/\* [0-9]+ FETCH.*/i', $topline)) ;
142 $wholemessage = implode('', $data);
143 if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) {
144 $ret = substr($wholemessage, 0, $regs[1]);
145 /* There is some information in the content info header that could be important
146 * in order to parse html messages. Let's get them here.
148 // if ($ret{0} == '<') {
149 // $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE);
151 } else if (preg_match('/"([^"]*)"/', $topline, $regs)) {
153 } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) {
154 $ret = $wholemessage;
156 global $where, $what, $mailbox, $passed_id, $startMessage;
157 $par = 'mailbox=' . urlencode($mailbox) . '&passed_id=' . $passed_id;
158 if (isset($where) && isset($what)) {
159 $par .= '&where=' . urlencode($where) . '&what=' . urlencode($what);
161 $par .= '&startMessage=' . $startMessage . '&show_more=0';
163 $par .= '&response=' . urlencode($response) .
164 '&message=' . urlencode($message) .
165 '&topline=' . urlencode($topline);
168 '<table width="80%"><tr>' .
169 '<tr><td colspan="2">' .
170 _("Body retrieval error. The reason for this is most probably that the message is malformed.") .
172 '<tr><td><b>' . _("Command:") . "</td><td>$cmd</td></tr>" .
173 '<tr><td><b>' . _("Response:") . "</td><td>$response</td></tr>" .
174 '<tr><td><b>' . _("Message:") . "</td><td>$message</td></tr>" .
175 '<tr><td><b>' . _("FETCH line:") . "</td><td>$topline</td></tr>" .
176 "</table><br /></tt></font><hr />";
178 $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE);
180 $wholemessage = implode('', $data);
182 $ret = $wholemessage;
187 function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') {
189 /* Don't kill the connection if the browser is over a dialup
190 * and it would take over 30 seconds to download it.
191 * Don't call set_time_limit in safe mode.
194 if (!ini_get('safe_mode')) {
197 /* in case of base64 encoded attachments, do not buffer them.
198 Instead, echo the decoded attachment directly to screen */
199 if (strtolower($encoding) == 'base64') {
201 $query = "FETCH $id BODY[]";
203 $query = "FETCH $id BODY[$ent_id]";
205 sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
207 $body = mime_fetch_body ($imap_stream, $id, $ent_id);
208 if (is_resource($rStream)) {
209 fputs($rStream,decodeBody($body, $encoding, $force_crlf));
211 echo decodeBody($body, $encoding, $force_crlf);
216 TODO, use the same method for quoted printable.
217 However, I assume that quoted printable attachments aren't that large
218 so the performancegain / memory usage drop will be minimal.
219 If we decide to add that then we need to adapt sqimap_fread because
220 we need to split te result on \n and fread doesn't stop at \n. That
221 means we also should provide $results from sqimap_fread (by ref) to
222 te function and set $no_return to false. The $filter function for
223 quoted printable should handle unsetting of $results.
226 TODO 2: find out how we write to the output stream php://stdout. fwrite
227 doesn't work because 'php://stdout isn't a stream.
233 /* -[ END MIME DECODING ]----------------------------------------------------------- */
235 /* This is here for debugging purposes. It will print out a list
236 * of all the entity IDs that are in the $message object.
238 function listEntities ($message) {
240 echo "<tt>" . $message->entity_id
. ' : ' . $message->type0
. '/' . $message->type1
. ' parent = '. $message->parent
->entity_id
. '<br />';
241 for ($i = 0; isset($message->entities
[$i]); $i++
) {
243 $msg = listEntities($message->entities
[$i]);
253 function getPriorityStr($priority) {
254 $priority_level = substr($priority,0,1);
256 switch($priority_level) {
257 /* Check for a higher then normal priority. */
260 $priority_string = _("High");
263 /* Check for a lower then normal priority. */
266 $priority_string = _("Low");
269 /* Check for a normal priority. */
272 $priority_level = '3';
273 $priority_string = _("Normal");
277 return $priority_string;
280 /* returns a $message object for a particular entity id */
281 function getEntity ($message, $ent_id) {
282 return $message->getEntity($ent_id);
286 * Extracted from strings.php 23/03/2002
289 function translateText(&$body, $wrap_at, $charset) {
290 global $where, $what; /* from searching */
291 global $color; /* color theme */
293 // require_once(SM_PATH . 'functions/url_parser.php');
295 $body_ary = explode("\n", $body);
296 for ($i=0; $i < count($body_ary); $i++
) {
297 $line = rtrim($body_ary[$i],"\r");
299 if (strlen($line) - 2 >= $wrap_at) {
300 sqWordWrap($line, $wrap_at, $charset);
302 $line = charset_decode($charset, $line);
303 $line = str_replace("\t", ' ', $line);
312 if ($line[$pos] == ' ') {
314 } else if (strpos($line, '>', $pos) === $pos) {
323 $line = '<span class="quote1">' . $line . '</span>';
325 $line = '<span class="quote2">' . $line . '</span>';
328 $body_ary[$i] = $line;
330 $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
334 * This returns a parsed string called $body. That string can then
335 * be displayed as the actual message in the HTML. It contains
336 * everything needed, including HTML Tags, Attachments at the
339 * Since 1.2.0 function uses message_body hook.
340 * Till 1.3.0 function included output of formatAttachments().
342 * @param resource $imap_stream imap connection resource
343 * @param object $message squirrelmail message object
344 * @param array $color squirrelmail color theme array
345 * @param integer $wrap_at number of characters per line
346 * @param string $ent_num (since 1.3.0) message part id
347 * @param integer $id (since 1.3.0) message id
348 * @param string $mailbox (since 1.3.0) imap folder name
349 * @return string html formated message text
351 function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') {
352 /* This if statement checks for the entity to show as the
353 * primary message. To add more of them, just put them in the
354 * order that is their priority.
356 global $startMessage, $languages, $squirrelmail_language,
357 $show_html_default, $sort, $has_unsafe_images, $passed_ent_id,
358 $use_iframe, $iframe_height, $download_and_unsafe_link,
359 $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text,
362 // workaround for not updated config.php
363 if (! isset($use_iframe)) $use_iframe = false;
365 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
366 // images off by default.
367 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
370 $urlmailbox = urlencode($mailbox);
371 $body_message = getEntity($message, $ent_num);
372 if (($body_message->header
->type0
== 'text') ||
373 ($body_message->header
->type0
== 'rfc822')) {
374 $body = mime_fetch_body ($imap_stream, $id, $ent_num);
375 $body = decodeBody($body, $body_message->header
->encoding
);
377 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
378 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) {
379 if (mb_detect_encoding($body) != 'ASCII') {
380 $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body);
384 /* As of 1.5.2, $body is passed (and modified) by reference */
385 do_hook('message_body', $body);
387 /* If there are other types that shouldn't be formatted, add
391 if ($body_message->header
->type1
== 'html') {
392 if ($show_html_default <> 1) {
393 $entity_conv = array(' ' => ' ',
402 $body = strtr($body, $entity_conv);
403 $body = strip_tags($body);
405 translateText($body, $wrap_at,
406 $body_message->header
->getParameter('charset'));
407 } elseif ($use_iframe) {
409 * If we don't add html message between iframe tags,
410 * we must detect unsafe images and modify $has_unsafe_images.
412 $html_body = magicHTML($body, $id, $message, $mailbox);
413 // Convert character set in order to display html mails in different character set
414 $html_body = charset_decode($body_message->header
->getParameter('charset'),$html_body,false,true);
416 // creating iframe url
417 $iframeurl=sqm_baseuri().'src/view_html.php?'
418 . 'mailbox=' . $urlmailbox
419 . '&passed_id=' . $id
420 . '&ent_id=' . $ent_num
421 . '&view_unsafe_images=' . (int) $view_unsafe_images;
424 $oTemplate->assign('iframe_url', $iframeurl);
425 $oTemplate->assign('iframe_height', $iframe_height);
426 $oTemplate->assign('html_body', $html_body);
428 $body = $oTemplate->fetch('read_html_iframe.tpl');
430 // old way of html rendering
432 * convert character set. charset_decode does not remove html special chars
433 * applied by magicHTML functions and does not sanitize them second time if
434 * fourth argument is true.
436 $charset = $body_message->header
->getParameter('charset');
437 if (!empty($charset)) {
438 $body = charset_decode($charset,$body,false,true);
440 $body = magicHTML($body, $id, $message, $mailbox);
443 translateText($body, $wrap_at,
444 $body_message->header
->getParameter('charset'));
448 * Previously the links for downloading and unsafe images were printed
449 * under the mail. By putting the links in a global variable we can
450 * print it in the toolbar where it belongs. Since the original code was
451 * in this place it's left here. It might be possible to move it to some
452 * other place if that makes sense. The possibility to do so has not
453 * been evaluated yet.
456 // Initialize the global variable to an empty string.
457 // FIXME: To have $download_and_unsafe_link as a global variable might not be needed since the use of separate variables ($download_href, $unsafe_image_toggle_href, and $unsafe_image_toggle_text) for the templates was introduced.
458 $download_and_unsafe_link = '';
460 // Prepare and build a link for downloading the mail.
461 $link = 'passed_id=' . $id . '&ent_id='.$ent_num.
462 '&mailbox=' . $urlmailbox .'&sort=' . $sort .
463 '&startMessage=' . $startMessage . '&show_more=0';
464 if (isset($passed_ent_id)) {
465 $link .= '&passed_ent_id='.$passed_ent_id;
467 $download_href = SM_PATH
. 'src/download.php?absolute_dl=true&' . $link;
469 // Always add the link for downloading the mail as a file to the global
471 $download_and_unsafe_link .= "$nbsp|$nbsp"
472 . create_hyperlink($download_href, _("Download this as a file"));
474 // Find out the right text to use in the link depending on the
475 // circumstances. If the unsafe images are displayed the link should
476 // hide them, if they aren't displayed the link should only appear if
477 // the mail really contains unsafe images.
478 if ($view_unsafe_images) {
479 $text = _("Hide Unsafe Images");
481 if (isset($has_unsafe_images) && $has_unsafe_images) {
482 $link .= '&view_unsafe_images=1';
483 $text = _("View Unsafe Images");
489 // Only create a link for unsafe images if there's need for one. If so:
490 // add it to the global variable.
492 $unsafe_image_toggle_href = SM_PATH
. 'src/read_body.php?'.$link;
493 $unsafe_image_toggle_text = $text;
494 $download_and_unsafe_link .= "$nbsp|$nbsp"
495 . create_hyperlink($unsafe_image_toggle_href, $text);
502 * Generate attachments array for passing to templates.
505 * @param object $message SquirrelMail message object
506 * @param array $exclude_id message parts that are not attachments.
507 * @param string $mailbox mailbox name
508 * @param integer $id message id
510 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
511 global $where, $what, $startMessage, $color, $passed_ent_id, $base_uri;
513 $att_ar = $message->getAttachments($exclude_id);
514 $urlMailbox = urlencode($mailbox);
516 $attachments = array();
517 foreach ($att_ar as $att) {
518 $ent = $att->entity_id
;
519 $header = $att->header
;
520 $type0 = strtolower($header->type0
);
521 $type1 = strtolower($header->type1
);
524 $links['download link']['text'] = _("Download");
525 $links['download link']['href'] = $base_uri .
526 "src/download.php?absolute_dl=true&passed_id=$id&mailbox=$urlMailbox&ent_id=$ent";
528 if ($type0 =='message' && $type1 == 'rfc822') {
529 $default_page = $base_uri . 'src/read_body.php';
530 $rfc822_header = $att->rfc822_header
;
531 $filename = $rfc822_header->subject
;
532 if (trim( $filename ) == '') {
533 $filename = 'untitled-[' . $ent . ']' ;
535 $from_o = $rfc822_header->from
;
536 if (is_object($from_o)) {
537 $from_name = decodeHeader($from_o->getAddress(false));
538 } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) {
539 // something weird happens when a digest message is opened and you return to the digest
540 // now the from object is part of an array. Probably the parseHeader call overwrites the info
541 // retrieved from the bodystructure in a different way. We need to fix this later.
542 // possible starting point, do not fetch header we already have and inspect how
543 // the rfc822_header object behaves.
544 $from_name = decodeHeader($from_o[0]->getAddress(false));
546 $from_name = _("Unknown sender");
548 $description = _("From").': '.$from_name;
550 $default_page = $base_uri . 'src/download.php';
551 $filename = $att->getFilename();
552 if ($header->description
) {
553 $description = decodeHeader($header->description
);
559 $display_filename = $filename;
560 if (isset($passed_ent_id)) {
561 $passed_ent_id_link = '&passed_ent_id='.$passed_ent_id;
563 $passed_ent_id_link = '';
565 $defaultlink = $default_page . "?startMessage=$startMessage"
566 . "&passed_id=$id&mailbox=$urlMailbox"
567 . '&ent_id='.$ent.$passed_ent_id_link;
568 if ($where && $what) {
569 $defaultlink .= '&where='. urlencode($where).'&what='.urlencode($what);
571 // IE does make use of mime content sniffing. Forcing a download
572 // prohibit execution of XSS inside an application/octet-stream attachment
573 if ($type0 == 'application' && $type1 == 'octet-stream') {
574 $defaultlink .= '&absolute_dl=true';
577 /* This executes the attachment hook with a specific MIME-type.
578 * If that doesn't have results, it tries if there's a rule
579 * for a more generic type. Finally, a hook for ALL attachment
580 * types is run as well.
582 // First remember the default link.
583 $defaultlink_orig = $defaultlink;
585 /* The API for this hook has changed as of 1.5.2 so that all plugin
586 arguments are passed in an array instead of each their own plugin
587 argument, and arguments are passed by reference, so instead of
588 returning any changes, changes should simply be made to the original
589 arguments themselves. */
590 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
591 &$defaultlink, &$display_filename, &$where, &$what);
592 do_hook("attachment $type0/$type1", $temp);
593 if(count($links) <= 1 && $defaultlink == $defaultlink_orig) {
594 /* The API for this hook has changed as of 1.5.2 so that all plugin
595 arguments are passed in an array instead of each their own plugin
596 argument, and arguments are passed by reference, so instead of
597 returning any changes, changes should simply be made to the original
598 arguments themselves. */
599 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
600 &$defaultlink, &$display_filename, &$where, &$what);
601 do_hook("attachment $type0/*", $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) {
616 do_hook("attachment */*", $temp);
618 $this_attachment = array();
619 $this_attachment['Name'] = decodeHeader($display_filename);
620 $this_attachment['Description'] = $description;
621 $this_attachment['DefaultHREF'] = $defaultlink;
622 $this_attachment['DownloadHREF'] = $links['download link']['href'];
623 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ?
$links['attachment_common']['href'] : '';
624 $this_attachment['Size'] = $header->size
;
625 $this_attachment['ContentType'] = htmlspecialchars($type0 .'/'. $type1);
626 $this_attachment['OtherLinks'] = array();
627 foreach ($links as $val) {
628 if ($val['text']==_("Download") ||
$val['text'] == _("View"))
630 if (empty($val['text']) && empty($val['extra']))
634 $temp['HREF'] = $val['href'];
635 $temp['Text'] = (empty($val['text']) ?
'' : $val['text']) . (empty($val['extra']) ?
'' : $val['extra']);
636 $this_attachment['OtherLinks'][] = $temp;
638 $attachments[] = $this_attachment;
647 * Displays attachment links and information
649 * Since 1.3.0 function is not included in formatBody() call.
651 * Since 1.0.2 uses attachment $type0/$type1 hook.
652 * Since 1.2.5 uses attachment $type0/* hook.
653 * Since 1.5.0 uses attachments_bottom hook.
654 * Since 1.5.2 uses templates and does *not* return a value.
656 * @param object $message SquirrelMail message object
657 * @param array $exclude_id message parts that are not attachments.
658 * @param string $mailbox mailbox name
659 * @param integer $id message id
661 function formatAttachments($message, $exclude_id, $mailbox, $id) {
664 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
666 $oTemplate->assign('attachments', $attach);
667 $oTemplate->display('read_attachments.tpl');
670 function sqimap_base64_decode(&$string) {
672 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
673 // fly decoding (to reduce memory usage) you have to check if the
674 // data has incomplete pairs
676 // Remove the noise in order to check if the 4 bytes pairs are complete
677 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
680 $iMod = strlen($string) %
4;
682 $sStringRem = substr($string,-$iMod);
683 // Check if $sStringRem contains padding characters
684 if (substr($sStringRem,-1) != '=') {
685 $string = substr($string,0,-$iMod);
690 $string = base64_decode($string);
695 * Decodes encoded string (usually message body)
697 * This function decodes a string (usually the message body)
698 * depending on the encoding type. Currently quoted-printable
699 * and base64 encodings are supported.
701 * The decode_body hook was added to this function in 1.4.2/1.5.0.
702 * The $force_crlf parameter was added in 1.5.2.
704 * @param string $string The encoded string
705 * @param string $encoding used encoding
706 * @param string $force_crlf Whether or not to force CRLF or LF
707 * line endings (or to leave as is).
708 * If given as "LF", line endings will
709 * all be converted to LF; if "CRLF",
710 * line endings will all be converted
711 * to CRLF. If given as an empty value,
712 * the global $force_crlf_default will
713 * be consulted (it can be specified in
714 * config/config_local.php). Otherwise,
715 * any other value will cause the string
716 * to be left alone. Note that this will
717 * be overridden to "LF" if not using at
718 * least PHP version 4.3.0. (OPTIONAL;
719 * default is empty - consult global
722 * @return string The decoded string
727 function decodeBody($string, $encoding, $force_crlf='') {
729 global $force_crlf_default;
730 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
731 $force_crlf = strtoupper($force_crlf);
733 // must force line endings to LF due to broken
734 // quoted_printable_decode() in PHP versions
735 // before 4.3.0 (see below)
737 if (!check_php_version(4, 3, 0) ||
$force_crlf == 'LF')
738 $string = str_replace("\r\n", "\n", $string);
739 else if ($force_crlf == 'CRLF')
740 $string = str_replace("\n", "\r\n", $string);
742 $encoding = strtolower($encoding);
744 $encoding_handler = do_hook('decode_body', $encoding);
747 // plugins get first shot at decoding the string
749 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
750 $string = $encoding_handler('decode', $string);
752 } elseif ($encoding == 'quoted-printable' ||
753 $encoding == 'quoted_printable') {
755 // quoted_printable_decode() function is broken in older
756 // php versions. Text with \r\n decoding was fixed only
757 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
758 // and the above call to: str_replace("\r\n", "\n", $string);
760 $string = quoted_printable_decode($string);
762 } elseif ($encoding == 'base64') {
763 $string = base64_decode($string);
766 // All other encodings are returned raw.
773 * This function decodes strings that are encoded according to
774 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
775 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
777 * @param string $string header string that has to be made readable
778 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
779 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
780 * @param boolean $decide decide if string can be utfencoded. defaults to false
781 * @return string decoded header string
783 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
784 global $languages, $squirrelmail_language,$default_charset;
785 if (is_array($string)) {
786 $string = implode("\n", $string);
789 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
790 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
791 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
792 // Do we need to return at this point?
799 $aString = explode(' ',$string);
801 foreach ($aString as $chunk) {
802 if ($encoded && $chunk === '') {
804 } elseif ($chunk === '') {
809 /* if encoded words are not separated by a linear-space-white we still catch them */
812 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
813 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
814 if ($iLastMatch !== $j) {
824 $ret .= htmlspecialchars($res[1]);
828 $encoding = ucfirst($res[3]);
830 /* decide about valid decoding */
831 if ($decide && is_conversion_safe($res[2])) {
833 $can_be_encoded=true;
835 $can_be_encoded=false;
840 $replace = base64_decode($res[4]);
842 if ($can_be_encoded) {
843 /* convert string to different charset,
844 * if functions asks for it (usually in compose)
846 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
848 // convert string to html codes in order to display it
849 $ret .= charset_decode($res[2],$replace);
853 $replace = htmlspecialchars($replace);
859 $replace = str_replace('_', ' ', $res[4]);
860 $replace = preg_replace('/=([0-9a-f]{2})/ie', 'chr(hexdec("\1"))',
863 if ($can_be_encoded) {
864 /* convert string to different charset,
865 * if functions asks for it (usually in compose)
867 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
869 // convert string to html codes in order to display it
870 $replace = charset_decode($res[2], $replace);
874 $replace = htmlspecialchars($replace);
893 if (!$encoded && $htmlsafe) {
894 $ret .= htmlspecialchars($chunk);
900 /* remove the first added space */
903 $ret = substr($ret,5);
905 $ret = substr($ret,1);
915 * Function uses XTRA_CODE _encodeheader function, if such function exists.
917 * Function uses Q encoding by default and encodes a string according to RFC
918 * 1522 for use in headers if it contains 8-bit characters or anything that
919 * looks like it should be encoded.
921 * Function switches to B encoding and encodeHeaderBase64() function, if
922 * string is 8bit and multibyte character set supported by mbstring extension
923 * is used. It can cause E_USER_NOTICE errors, if interface is used with
924 * multibyte character set unsupported by mbstring extension.
926 * @param string $string header string, that has to be encoded
927 * @return string quoted-printable encoded string
928 * @todo make $mb_charsets system wide constant
930 function encodeHeader ($string) {
931 global $default_charset, $languages, $squirrelmail_language;
933 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
934 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
935 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
938 // Use B encoding for multibyte charsets
939 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
940 if (in_array($default_charset,$mb_charsets) &&
941 in_array($default_charset,sq_mb_list_encodings()) &&
942 sq_is8bit($string)) {
943 return encodeHeaderBase64($string,$default_charset);
944 } elseif (in_array($default_charset,$mb_charsets) &&
945 sq_is8bit($string) &&
946 ! in_array($default_charset,sq_mb_list_encodings())) {
947 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
948 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
951 // Encode only if the string contains 8-bit characters or =?
952 $j = strlen($string);
953 $max_l = 75 - strlen($default_charset) - 7;
956 $iEncStart = $enc_init = false;
957 $cur_l = $iOffset = 0;
958 for($i = 0; $i < $j; ++
$i) {
967 if ($iEncStart === false) {
971 if ($cur_l > ($max_l-2)) {
972 /* if there is an stringpart that doesn't need encoding, add it */
973 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
974 $aRet[] = "=?$default_charset?Q?$ret?=";
980 $ret .= sprintf("=%02X",ord($string{$i}));
985 if ($iEncStart !== false) {
986 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
987 $aRet[] = "=?$default_charset?Q?$ret?=";
995 if ($iEncStart !== false) {
997 if ($cur_l > $max_l) {
998 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
999 $aRet[] = "=?$default_charset?Q?$ret?=";
1010 $k = ord($string{$i});
1012 if ($iEncStart === false) {
1013 // do not start encoding in the middle of a string, also take the rest of the word.
1014 $sLeadString = substr($string,0,$i);
1015 $aLeadString = explode(' ',$sLeadString);
1016 $sToBeEncoded = array_pop($aLeadString);
1017 $iEncStart = $i - strlen($sToBeEncoded);
1018 $ret .= $sToBeEncoded;
1019 $cur_l +
= strlen($sToBeEncoded);
1022 /* first we add the encoded string that reached it's max size */
1023 if ($cur_l > ($max_l-2)) {
1024 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1025 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1032 $ret .= sprintf("=%02X", $k);
1034 if ($iEncStart !== false) {
1036 if ($cur_l > $max_l) {
1037 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1038 $aRet[] = "=?$default_charset?Q?$ret?=";
1044 $ret .= $string{$i};
1053 if ($iEncStart !== false) {
1054 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1055 $aRet[] = "=?$default_charset?Q?$ret?=";
1057 $aRet[] = substr($string,$iOffset);
1059 $string = implode('',$aRet);
1065 * Encodes string according to rfc2047 B encoding header formating rules
1067 * It is recommended way to encode headers with character sets that store
1068 * symbols in more than one byte.
1070 * Function requires mbstring support. If required mbstring functions are missing,
1071 * function returns false and sets E_USER_WARNING level error message.
1073 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1074 * that mbstring functions will generate E_WARNING errors, if unsupported
1075 * character set is used. mb_encode_mimeheader function provided by php
1076 * mbstring extension is not used in order to get better control of header
1079 * Used php code functions - function_exists(), trigger_error(), strlen()
1080 * (is used with charset names and base64 strings). Used php mbstring
1081 * functions - mb_strlen and mb_substr.
1083 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1084 * encoding), rfc 2822 (header folding)
1086 * @param string $string header string that must be encoded
1087 * @param string $charset character set. Must be supported by mbstring extension.
1088 * Use sq_mb_list_encodings() to detect supported charsets.
1089 * @return string string encoded according to rfc2047 B encoding formating rules
1091 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1092 * @todo Do we want to control max length of header?
1093 * @todo Do we want to control EOL (end-of-line) marker?
1094 * @todo Do we want to translate error message?
1096 function encodeHeaderBase64($string,$charset) {
1098 * Check mbstring function requirements.
1100 if (! function_exists('mb_strlen') ||
1101 ! function_exists('mb_substr')) {
1102 // set E_USER_WARNING
1103 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING
);
1108 // initial return array
1112 * header length = 75 symbols max (same as in encodeHeader)
1113 * remove $charset length
1114 * remove =? ? ?= (5 chars)
1115 * remove 2 more chars (\r\n ?)
1117 $iMaxLength = 75 - strlen($charset) - 7;
1119 // set first character position
1122 // loop through all characters. count characters and not bytes.
1123 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++
) {
1124 // encode string from starting character to current character.
1125 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1127 // Check encoded string length
1128 if(strlen($encoded_string)>$iMaxLength) {
1129 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1130 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1132 // set new starting character
1133 $iStartCharNum = $iCharNum-1;
1135 // encode last char (in case it is last character in string)
1136 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1137 } // if string is shorter than max length - add next character
1140 // add last encoded string to array
1141 $aRet[] = $encoded_string;
1143 // set initial return string
1146 // loop through encoded strings
1147 foreach($aRet as $string) {
1148 // TODO: Do we want to control EOL (end-of-line) marker
1149 if ($sRet!='') $sRet.= " ";
1151 // add header tags and encoded string to return string
1152 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1158 /* This function trys to locate the entity_id of a specific mime element */
1159 function find_ent_id($id, $message) {
1160 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities
); $i++
) {
1161 if ($message->entities
[$i]->header
->type0
== 'multipart') {
1162 $ret = find_ent_id($id, $message->entities
[$i]);
1164 if (strcasecmp($message->entities
[$i]->header
->id
, $id) == 0) {
1165 // if (sq_check_save_extension($message->entities[$i])) {
1166 return $message->entities
[$i]->entity_id
;
1168 } elseif (!empty($message->entities
[$i]->header
->parameters
['name'])) {
1170 * This is part of a fix for Outlook Express 6.x generating
1171 * cid URLs without creating content-id headers
1174 if (strcasecmp($message->entities
[$i]->header
->parameters
['name'], $id) == 0) {
1175 return $message->entities
[$i]->entity_id
;
1183 function sq_check_save_extension($message) {
1184 $filename = $message->getFilename();
1185 $ext = substr($filename, strrpos($filename,'.')+
1);
1186 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1187 return in_array($ext, $save_extensions);
1192 ** HTMLFILTER ROUTINES
1196 * This function checks attribute values for entity-encoded values
1197 * and returns them translated into 8-bit strings so we can run
1200 * @param $attvalue A string to run entity check against.
1201 * @return Nothing, modifies a reference value.
1203 function sq_defang(&$attvalue){
1206 * Skip this if there aren't ampersands or backslashes.
1208 if (strpos($attvalue, '&') === false
1209 && strpos($attvalue, '\\') === false){
1213 // before deent, translate the dangerous unicode characters and ... to safe values
1214 // otherwise the regular expressions do not match.
1220 $m = $m ||
sq_deent($attvalue, '/\�*(\d+);*/s');
1221 $m = $m ||
sq_deent($attvalue, '/\�*((\d|[a-f])+);*/si', true);
1222 $m = $m ||
sq_deent($attvalue, '/\\\\(\d+)/s', true);
1223 } while ($m == true);
1224 $attvalue = stripslashes($attvalue);
1228 * Kill any tabs, newlines, or carriage returns. Our friends the
1229 * makers of the browser with 95% market value decided that it'd
1230 * be funny to make "java[tab]script" be just as good as "javascript".
1232 * @param attvalue The attribute value before extraneous spaces removed.
1233 * @return attvalue Nothing, modifies a reference value.
1235 function sq_unspace(&$attvalue){
1237 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1238 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1239 Array('', '', '', '', ''), $attvalue);
1244 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1245 * IE as regular characters.
1247 * @param attvalue The attribute value before dangerous characters are translated.
1248 * @return attvalue Nothing, modifies a reference value.
1249 * @author Marc Groot Koerkamp.
1251 function sq_fixIE_idiocy(&$attvalue) {
1253 $attvalue = str_replace("\0", "", $attvalue);
1255 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1257 // IE has the evil habit of accepting every possible value for the attribute expression.
1258 // The table below contains characters which are parsed by IE if they are used in the "expression"
1260 $aDangerousCharsReplacementTable = array(
1261 array('ʟ', 'ʟ' ,/* L UNICODE IPA Extension */
1262 'ʀ', 'ʀ' ,/* R UNICODE IPA Extension */
1263 'ɴ', 'ɴ' ,/* N UNICODE IPA Extension */
1264 'E', 'E' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1265 'e', 'e' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1266 'X', 'X',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1267 'x', 'x',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1268 'P', 'P',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1269 'p', 'p',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1270 'R', 'R',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1271 'r', 'r',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1272 'S', 'S',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1273 's', 's',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1274 'I', 'I',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1275 'i', 'i',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1276 'O', 'O',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1277 'o', 'o',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1278 'N', 'N',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1279 'n', 'n',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1280 'L', 'L',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1281 'l', 'l',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1282 'U', 'U',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1283 'u', 'u',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1284 'ⁿ', 'ⁿ' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1285 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1286 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1287 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1288 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1289 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1290 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1291 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1292 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1293 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1294 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1295 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1296 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1297 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1298 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1299 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1300 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1301 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1302 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1303 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1304 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1305 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1306 "\xCA\x9F", /* L UNICODE IPA Extension */
1307 "\xCA\x80", /* R UNICODE IPA Extension */
1308 "\xC9\xB4"), /* N UNICODE IPA Extension */
1309 array('l', 'l', 'r','r','n','n',
1310 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1311 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1312 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1313 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1315 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1317 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1321 * This function returns the final tag out of the tag name, an array
1322 * of attributes, and the type of the tag. This function is called by
1323 * sq_sanitize internally.
1325 * @param $tagname the name of the tag.
1326 * @param $attary the array of attributes and their values
1327 * @param $tagtype The type of the tag (see in comments).
1328 * @return a string with the final tag representation.
1330 function sq_tagprint($tagname, $attary, $tagtype){
1331 $me = 'sq_tagprint';
1334 $fulltag = '</' . $tagname . '>';
1336 $fulltag = '<' . $tagname;
1337 if (is_array($attary) && sizeof($attary)){
1339 while (list($attname, $attvalue) = each($attary)){
1340 array_push($atts, "$attname=$attvalue");
1342 $fulltag .= ' ' . join(" ", $atts);
1353 * A small helper function to use with array_walk. Modifies a by-ref
1354 * value and makes it lowercase.
1356 * @param $val a value passed by-ref.
1357 * @return void since it modifies a by-ref value.
1359 function sq_casenormalize(&$val){
1360 $val = strtolower($val);
1364 * This function skips any whitespace from the current position within
1365 * a string and to the next non-whitespace value.
1367 * @param $body the string
1368 * @param $offset the offset within the string where we should start
1369 * looking for the next non-whitespace character.
1370 * @return the location within the $body where the next
1371 * non-whitespace char is located.
1373 function sq_skipspace($body, $offset){
1374 $me = 'sq_skipspace';
1375 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1376 if (sizeof($matches{1})){
1377 $count = strlen($matches{1});
1384 * This function looks for the next character within a string. It's
1385 * really just a glorified "strpos", except it catches if failures
1388 * @param $body The string to look for needle in.
1389 * @param $offset Start looking from this position.
1390 * @param $needle The character/string to look for.
1391 * @return location of the next occurance of the needle, or
1392 * strlen($body) if needle wasn't found.
1394 function sq_findnxstr($body, $offset, $needle){
1395 $me = 'sq_findnxstr';
1396 $pos = strpos($body, $needle, $offset);
1397 if ($pos === FALSE){
1398 $pos = strlen($body);
1404 * This function takes a PCRE-style regexp and tries to match it
1405 * within the string.
1407 * @param $body The string to look for needle in.
1408 * @param $offset Start looking from here.
1409 * @param $reg A PCRE-style regex to match.
1410 * @return Returns a false if no matches found, or an array
1411 * with the following members:
1412 * - integer with the location of the match within $body
1413 * - string with whatever content between offset and the match
1414 * - string with whatever it is we matched
1416 function sq_findnxreg($body, $offset, $reg){
1417 $me = 'sq_findnxreg';
1420 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1421 if (!isset($matches{0}) ||
!$matches{0}){
1424 $retarr{0} = $offset +
strlen($matches{1});
1425 $retarr{1} = $matches{1};
1426 $retarr{2} = $matches{2};
1432 * This function looks for the next tag.
1434 * @param $body String where to look for the next tag.
1435 * @param $offset Start looking from here.
1436 * @return false if no more tags exist in the body, or
1437 * an array with the following members:
1438 * - string with the name of the tag
1439 * - array with attributes and their values
1440 * - integer with tag type (1, 2, or 3)
1441 * - integer where the tag starts (starting "<")
1442 * - integer where the tag ends (ending ">")
1443 * first three members will be false, if the tag is invalid.
1445 function sq_getnxtag($body, $offset){
1446 $me = 'sq_getnxtag';
1447 if ($offset > strlen($body)){
1450 $lt = sq_findnxstr($body, $offset, "<");
1451 if ($lt == strlen($body)){
1456 * blah blah <tag attribute="value">
1459 $pos = sq_skipspace($body, $lt+
1);
1460 if ($pos >= strlen($body)){
1461 return Array(false, false, false, $lt, strlen($body));
1464 * There are 3 kinds of tags:
1465 * 1. Opening tag, e.g.:
1467 * 2. Closing tag, e.g.:
1469 * 3. XHTML-style content-less tag, e.g.:
1470 * <img src="blah" />
1473 switch (substr($body, $pos, 1)){
1480 * A comment or an SGML declaration.
1482 if (substr($body, $pos+
1, 2) == "--"){
1483 $gt = strpos($body, "-->", $pos);
1485 $gt = strlen($body);
1489 return Array(false, false, false, $lt, $gt);
1491 $gt = sq_findnxstr($body, $pos, ">");
1492 return Array(false, false, false, $lt, $gt);
1497 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1507 * Look for next [\W-_], which will indicate the end of the tag name.
1509 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1510 if ($regary == false){
1511 return Array(false, false, false, $lt, strlen($body));
1513 list($pos, $tagname, $match) = $regary;
1514 $tagname = strtolower($tagname);
1517 * $match can be either of these:
1518 * '>' indicating the end of the tag entirely.
1519 * '\s' indicating the end of the tag name.
1520 * '/' indicating that this is type-3 xhtml tag.
1522 * Whatever else we find there indicates an invalid tag.
1527 * This is an xhtml-style tag with a closing / at the
1528 * end, like so: <img src="blah" />. Check if it's followed
1529 * by the closing bracket. If not, then this tag is invalid
1531 if (substr($body, $pos, 2) == "/>"){
1535 $gt = sq_findnxstr($body, $pos, ">");
1536 $retary = Array(false, false, false, $lt, $gt);
1540 return Array($tagname, false, $tagtype, $lt, $pos);
1544 * Check if it's whitespace
1546 if (!preg_match('/\s/', $match)){
1548 * This is an invalid tag! Look for the next closing ">".
1550 $gt = sq_findnxstr($body, $lt, ">");
1551 return Array(false, false, false, $lt, $gt);
1557 * At this point we're here:
1558 * <tagname attribute='blah'>
1561 * At this point we loop in order to find all attributes.
1567 while ($pos <= strlen($body)){
1568 $pos = sq_skipspace($body, $pos);
1569 if ($pos == strlen($body)){
1573 return Array(false, false, false, $lt, $pos);
1576 * See if we arrived at a ">" or "/>", which means that we reached
1577 * the end of the tag.
1580 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1584 $pos +
= strlen($matches{1});
1585 if ($matches{2} == "/>"){
1589 return Array($tagname, $attary, $tagtype, $lt, $pos);
1593 * There are several types of attributes, with optional
1594 * [:space:] between members.
1596 * attrname[:space:]=[:space:]'CDATA'
1598 * attrname[:space:]=[:space:]"CDATA"
1600 * attr[:space:]=[:space:]CDATA
1604 * We leave types 1 and 2 the same, type 3 we check for
1605 * '"' and convert to """ if needed, then wrap in
1606 * double quotes. Type 4 we convert into:
1609 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1610 if ($regary == false){
1612 * Looks like body ended before the end of tag.
1614 return Array(false, false, false, $lt, strlen($body));
1616 list($pos, $attname, $match) = $regary;
1617 $attname = strtolower($attname);
1619 * We arrived at the end of attribute name. Several things possible
1621 * '>' means the end of the tag and this is attribute type 4
1622 * '/' if followed by '>' means the same thing as above
1623 * '\s' means a lot of things -- look what it's followed by.
1624 * anything else means the attribute is invalid.
1629 * This is an xhtml-style tag with a closing / at the
1630 * end, like so: <img src="blah" />. Check if it's followed
1631 * by the closing bracket. If not, then this tag is invalid
1633 if (substr($body, $pos, 2) == "/>"){
1637 $gt = sq_findnxstr($body, $pos, ">");
1638 $retary = Array(false, false, false, $lt, $gt);
1642 $attary{$attname} = '"yes"';
1643 return Array($tagname, $attary, $tagtype, $lt, $pos);
1647 * Skip whitespace and see what we arrive at.
1649 $pos = sq_skipspace($body, $pos);
1650 $char = substr($body, $pos, 1);
1652 * Two things are valid here:
1653 * '=' means this is attribute type 1 2 or 3.
1654 * \w means this was attribute type 4.
1655 * anything else we ignore and re-loop. End of tag and
1656 * invalid stuff will be caught by our checks at the beginning
1661 $pos = sq_skipspace($body, $pos);
1663 * Here are 3 possibilities:
1664 * "'" attribute type 1
1665 * '"' attribute type 2
1666 * everything else is the content of tag type 3
1668 $quot = substr($body, $pos, 1);
1670 $regary = sq_findnxreg($body, $pos+
1, "\'");
1671 if ($regary == false){
1672 return Array(false, false, false, $lt, strlen($body));
1674 list($pos, $attval, $match) = $regary;
1676 $attary{$attname} = "'" . $attval . "'";
1677 } else if ($quot == '"'){
1678 $regary = sq_findnxreg($body, $pos+
1, '\"');
1679 if ($regary == false){
1680 return Array(false, false, false, $lt, strlen($body));
1682 list($pos, $attval, $match) = $regary;
1684 $attary{$attname} = '"' . $attval . '"';
1687 * These are hateful. Look for \s, or >.
1689 $regary = sq_findnxreg($body, $pos, "[\s>]");
1690 if ($regary == false){
1691 return Array(false, false, false, $lt, strlen($body));
1693 list($pos, $attval, $match) = $regary;
1695 * If it's ">" it will be caught at the top.
1697 $attval = preg_replace("/\"/s", """, $attval);
1698 $attary{$attname} = '"' . $attval . '"';
1700 } else if (preg_match("|[\w/>]|", $char)) {
1702 * That was attribute type 4.
1704 $attary{$attname} = '"yes"';
1707 * An illegal character. Find next '>' and return.
1709 $gt = sq_findnxstr($body, $pos, ">");
1710 return Array(false, false, false, $lt, $gt);
1716 * The fact that we got here indicates that the tag end was never
1717 * found. Return invalid tag indication so it gets stripped.
1719 return Array(false, false, false, $lt, strlen($body));
1723 * Translates entities into literal values so they can be checked.
1725 * @param $attvalue the by-ref value to check.
1726 * @param $regex the regular expression to check against.
1727 * @param $hex whether the entites are hexadecimal.
1728 * @return True or False depending on whether there were matches.
1730 function sq_deent(&$attvalue, $regex, $hex=false){
1734 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1735 preg_match_all($regex, $attvalue, $matches);
1736 if (is_array($matches) && sizeof($matches[0]) > 0){
1738 for ($i = 0; $i < sizeof($matches[0]); $i++
){
1739 $numval = $matches[1][$i];
1741 $numval = hexdec($numval);
1743 $repl{$matches[0][$i]} = chr($numval);
1745 $attvalue = strtr($attvalue, $repl);
1753 * This function runs various checks against the attributes.
1755 * @param $tagname String with the name of the tag.
1756 * @param $attary Array with all tag attributes.
1757 * @param $rm_attnames See description for sq_sanitize
1758 * @param $bad_attvals See description for sq_sanitize
1759 * @param $add_attr_to_tag See description for sq_sanitize
1760 * @param $message message object
1761 * @param $id message id
1762 * @return Array with modified attributes.
1764 function sq_fixatts($tagname,
1774 while (list($attname, $attvalue) = each($attary)){
1776 * See if this attribute should be removed.
1778 foreach ($rm_attnames as $matchtag=>$matchattrs){
1779 if (preg_match($matchtag, $tagname)){
1780 foreach ($matchattrs as $matchattr){
1781 if (preg_match($matchattr, $attname)){
1782 unset($attary{$attname});
1789 * Workaround for IE quirks
1791 sq_fixIE_idiocy($attvalue);
1794 * Remove any backslashes, entities, and extraneous whitespace.
1797 $oldattvalue = $attvalue;
1798 sq_defang($attvalue);
1799 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1800 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1801 // i.e.<div style="{ left:expʀessioɴ( alert('XSS') ) }">
1802 $attvalue = "idiocy";
1803 $attary{$attname} = $attvalue;
1805 sq_unspace($attvalue);
1808 * Now let's run checks on the attvalues.
1809 * I don't expect anyone to comprehend this. If you do,
1810 * get in touch with me so I can drive to where you live and
1811 * shake your hand personally. :)
1813 foreach ($bad_attvals as $matchtag=>$matchattrs){
1814 if (preg_match($matchtag, $tagname)){
1815 foreach ($matchattrs as $matchattr=>$valary){
1816 if (preg_match($matchattr, $attname)){
1818 * There are two arrays in valary.
1820 * Second one is replacements
1822 list($valmatch, $valrepl) = $valary;
1824 preg_replace($valmatch, $valrepl, $attvalue);
1825 if ($newvalue != $attvalue){
1826 $attary{$attname} = $newvalue;
1827 $attvalue = $newvalue;
1833 if ($attname == 'style') {
1834 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1835 // 8bit and control characters in style attribute values can be used for XSS, remove them
1836 $attary{$attname} = '"disallowed character"';
1838 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1839 if (count($aMatch)) {
1840 foreach($aMatch[1] as $sMatch) {
1842 $urlvalue = $sMatch;
1843 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1844 $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1849 * Use white list based filtering on attributes which can contain url's
1851 else if ($attname == 'href' ||
$attname == 'src' ||
$attname == 'background') {
1852 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1853 $attary{$attname} = $attvalue;
1857 * See if we need to append any attributes to this tag.
1859 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1860 if (preg_match($matchtag, $tagname)){
1861 $attary = array_merge($attary, $addattary);
1868 * This function filters url's
1870 * @param $attvalue String with attribute value to filter
1871 * @param $message message object
1872 * @param $id message id
1873 * @param $mailbox mailbox
1874 * @param $sQuote quoting characters around url's
1876 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1877 $attvalue = trim($attvalue);
1878 if ($attvalue && ($attvalue[0] =='"'||
$attvalue[0] == "'")) {
1879 // remove the double quotes
1880 $sQuote = $attvalue[0];
1881 $attvalue = trim(substr($attvalue,1,-1));
1884 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1885 // images off by default.
1886 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
1888 $secremoveimg = '../images/' . _("sec_remove_eng.png");
1891 * Replace empty src tags with the blank image. src is only used
1892 * for frames, images, and image inputs. Doing a replace should
1893 * not affect them working as should be, however it will stop
1894 * IE from being kicked off when src for img tags are not set
1896 if ($attvalue == '') {
1897 $attvalue = '"' . SM_PATH
. 'images/blank.png"';
1899 // first, disallow 8 bit characters and control characters
1900 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1903 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1906 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
1910 $aUrl = parse_url($attvalue);
1911 if (isset($aUrl['scheme'])) {
1912 switch(strtolower($aUrl['scheme'])) {
1917 if ($attname != 'href') {
1918 if ($view_unsafe_images == false) {
1919 $attvalue = $sQuote . $secremoveimg . $sQuote;
1921 if (isset($aUrl['path'])) {
1923 // No one has been able to show that image URIs
1924 // can be exploited, so for now, no restrictions
1925 // are made at all. If this proves to be a problem,
1926 // the commented-out code below can be of help.
1927 // (One consideration is that I see nothing in this
1928 // function that specifically says that we will
1929 // only ever arrive here when inspecting an image
1930 // tag, although that does seem to be the end
1931 // result - e.g., <script src="..."> where malicious
1932 // image URIs are in fact a problem are already
1933 // filtered out elsewhere.
1934 /* ---------------------------------
1935 // validate image extension.
1936 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1937 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1938 // If URI is to something other than
1939 // a regular image file, get the contents
1940 // and try to see if it is an image.
1941 // Don't use Fileinfo (finfo_file()) because
1942 // we'd need to make the admin configure the
1943 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1946 if (function_exists('mime_content_type')
1947 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1951 $file_contents = '';
1952 while (!feof($FILE)) {
1953 $file_contents .= fread($FILE, 8192);
1957 // store file locally
1959 global $attachment_dir, $username;
1960 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1961 $localfilename = GenerateRandomString(32, '', 7);
1962 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1963 while (file_exists($full_localfilename)) {
1964 $localfilename = GenerateRandomString(32, '', 7);
1965 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1967 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
1968 fwrite($FILE, $file_contents);
1971 // get mime type and remove file
1973 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
1974 unlink("$hashed_attachment_dir/$localfilename");
1976 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
1977 if (substr(strtolower($mime_type), 0, 5) != 'image') {
1978 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1981 --------------------------------- */
1983 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
1987 $attvalue = $sQuote . $attvalue . $sQuote;
1992 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
1993 * One day MS might actually make it match something useful, for now, falling
1994 * back to using cid2http, so we can grab the blank.png.
1996 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2000 * Turn cid: urls into http-friendly ones.
2002 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2005 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2009 if (!(isset($aUrl['path']) && $aUrl['path'] == $secremoveimg)) {
2010 // parse_url did not lead to satisfying result
2011 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2019 * This function edits the style definition to make them friendly and
2020 * usable in SquirrelMail.
2022 * @param $message the message object
2023 * @param $id the message id
2024 * @param $content a string with whatever is between <style> and </style>
2025 * @param $mailbox the message mailbox
2026 * @return a string with edited content.
2028 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2029 $me = 'sq_fixstyle';
2030 // workaround for </style> in between comments
2031 $iCurrentPos = $pos;
2036 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++
$i) {
2043 if ($sToken == '<') {
2053 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2058 $content .= $sToken;
2066 if ($sToken == '<') {
2068 if (isset($body{$i+
2}) && substr($body,$i,3) == '!--') {
2069 $i = strpos($body,'-->',$i+
3);
2070 if ($i === false) { // no end comment
2088 if ($bSucces == FALSE){
2089 return array(FALSE, strlen($body));
2095 * First look for general BODY style declaration, which would be
2097 * body {background: blah-blah}
2098 * and change it to .bodyclass so we can just assign it to a <div>
2100 $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2101 $secremoveimg = '../images/' . _("sec_remove_eng.png");
2103 * Fix url('blah') declarations.
2105 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2106 // "url(\\1$secremoveimg\\2)", $content);
2108 // first check for 8bit sequences and disallowed control characters
2109 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2110 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2111 return array($content, $newpos);
2114 // IE Sucks hard. We have a special function for it.
2115 sq_fixIE_idiocy($content);
2117 // remove @import line
2118 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2120 // translate ur\l and variations (IE parses that)
2121 // TODO check if the sq_fixIE_idiocy function already handles this.
2122 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2123 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2124 if (count($aMatch)) {
2125 $aValue = $aReplace = array();
2126 foreach($aMatch[1] as $sMatch) {
2128 $urlvalue = $sMatch;
2129 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2130 $aValue[] = $sMatch;
2131 $aReplace[] = $urlvalue;
2133 $content = str_replace($aValue,$aReplace,$content);
2137 * Remove any backslashes, entities, and extraneous whitespace.
2139 $contentTemp = $content;
2140 sq_defang($contentTemp);
2141 sq_unspace($contentTemp);
2144 * Fix stupid css declarations which lead to vulnerabilities
2147 * Also remove "position" attribute, as it can easily be set
2148 * to "fixed" or "absolute" with "left" and "top" attributes
2149 * of zero, taking over the whole content frame. It can also
2150 * be set to relative and move itself anywhere it wants to,
2151 * displaying content in areas it shouldn't be allowed to touch.
2153 $match = Array('/\/\*.*\*\//',
2157 '/include-source/i',
2161 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2162 $contentNew = preg_replace($match, $replace, $contentTemp);
2163 if ($contentNew !== $contentTemp) {
2164 // insecure css declarations are used. From now on we don't care
2165 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2166 $content = $contentNew;
2168 return array($content, $newpos);
2173 * This function converts cid: url's into the ones that can be viewed in
2176 * @param $message the message object
2177 * @param $id the message id
2178 * @param $cidurl the cid: url.
2179 * @param $mailbox the message mailbox
2180 * @return a string with a http-friendly url
2182 function sq_cid2http($message, $id, $cidurl, $mailbox){
2184 * Get rid of quotes.
2186 $quotchar = substr($cidurl, 0, 1);
2187 if ($quotchar == '"' ||
$quotchar == "'"){
2188 $cidurl = str_replace($quotchar, "", $cidurl);
2192 $cidurl = substr(trim($cidurl), 4);
2194 $match_str = '/\{.*?\}\//';
2196 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2198 $linkurl = find_ent_id($cidurl, $message);
2199 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2200 unsafe link image */
2204 * This is part of a fix for Outlook Express 6.x generating
2205 * cid URLs without creating content-id headers. These images are
2206 * not part of the multipart/related html mail. The html contains
2207 * <img src="cid:{some_id}/image_filename.ext"> references to
2208 * attached images with as goal to render them inline although
2209 * the attachment disposition property is not inline.
2212 if (empty($linkurl)) {
2213 if (preg_match('/{.*}\//', $cidurl)) {
2214 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2215 if (!empty($cidurl)) {
2216 $linkurl = find_ent_id($cidurl, $message);
2221 if (!empty($linkurl)) {
2222 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&' .
2223 "passed_id=$id&mailbox=" . urlencode($mailbox) .
2224 '&ent_id=' . $linkurl . $quotchar;
2227 * If we couldn't generate a proper img url, drop in a blank image
2228 * instead of sending back empty, otherwise it causes unusual behaviour
2230 $httpurl = $quotchar . SM_PATH
. 'images/blank.png' . $quotchar;
2237 * This function changes the <body> tag into a <div> tag since we
2238 * can't really have a body-within-body.
2240 * @param $attary an array of attributes and values of <body>
2241 * @param $mailbox mailbox we're currently reading (for cid2http)
2242 * @param $message current message (for cid2http)
2243 * @param $id current message id (for cid2http)
2244 * @return a modified array of attributes to be set for <div>
2246 function sq_body2div($attary, $mailbox, $message, $id){
2247 $me = 'sq_body2div';
2248 $divattary = Array('class' => "'bodyclass'");
2250 $has_bgc_stl = $has_txt_stl = false;
2252 if (is_array($attary) && sizeof($attary) > 0){
2253 foreach ($attary as $attname=>$attvalue){
2254 $quotchar = substr($attvalue, 0, 1);
2255 $attvalue = str_replace($quotchar, "", $attvalue);
2258 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2259 $styledef .= "background-image: url('$attvalue'); ";
2262 $has_bgc_stl = true;
2263 $styledef .= "background-color: $attvalue; ";
2266 $has_txt_stl = true;
2267 $styledef .= "color: $attvalue; ";
2271 // Outlook defines a white bgcolor and no text color. This can lead to
2272 // white text on a white bg with certain themes.
2273 if ($has_bgc_stl && !$has_txt_stl) {
2274 $styledef .= "color: $text; ";
2276 if (strlen($styledef) > 0){
2277 $divattary{"style"} = "\"$styledef\"";
2284 * This is the main function and the one you should actually be calling.
2285 * There are several variables you should be aware of an which need
2286 * special description.
2288 * Since the description is quite lengthy, see it here:
2289 * http://linux.duke.edu/projects/mini/htmlfilter/
2291 * @param $body the string with HTML you wish to filter
2292 * @param $tag_list see description above
2293 * @param $rm_tags_with_content see description above
2294 * @param $self_closing_tags see description above
2295 * @param $force_tag_closing see description above
2296 * @param $rm_attnames see description above
2297 * @param $bad_attvals see description above
2298 * @param $add_attr_to_tag see description above
2299 * @param $message message object
2300 * @param $id message id
2301 * @return sanitized html safe to show on your pages.
2303 function sq_sanitize($body,
2305 $rm_tags_with_content,
2315 $me = 'sq_sanitize';
2316 $rm_tags = array_shift($tag_list);
2318 * Normalize rm_tags and rm_tags_with_content.
2320 @array_walk
($tag_list, 'sq_casenormalize');
2321 @array_walk
($rm_tags_with_content, 'sq_casenormalize');
2322 @array_walk
($self_closing_tags, 'sq_casenormalize');
2324 * See if tag_list is of tags to remove or tags to allow.
2325 * false means remove these tags
2326 * true means allow these tags
2329 $open_tags = Array();
2330 $trusted = "\n<!-- begin sanitized html -->\n";
2331 $skip_content = false;
2333 * Take care of netscape's stupid javascript entities like
2336 $body = preg_replace("/&(\{.*?\};)/si", "&\\1", $body);
2338 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2339 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2340 $free_content = substr($body, $curpos, $lt-$curpos);
2342 * Take care of <style>
2344 if ($tagname == "style" && $tagtype == 1){
2345 list($free_content, $curpos) =
2346 sq_fixstyle($body, $gt+
1, $message, $id, $mailbox);
2347 if ($free_content != FALSE){
2348 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2349 $trusted .= $free_content;
2350 $trusted .= sq_tagprint($tagname, false, 2);
2354 if ($skip_content == false){
2355 $trusted .= $free_content;
2357 if ($tagname != FALSE){
2359 if ($skip_content == $tagname){
2361 * Got to the end of tag we needed to remove.
2364 $skip_content = false;
2366 if ($skip_content == false){
2367 if ($tagname == "body"){
2370 if (isset($open_tags{$tagname}) &&
2371 $open_tags{$tagname} > 0){
2372 $open_tags{$tagname}--;
2380 * $rm_tags_with_content
2382 if ($skip_content == false){
2384 * See if this is a self-closing type and change
2385 * tagtype appropriately.
2388 && in_array($tagname, $self_closing_tags)){
2392 * See if we should skip this tag and any content
2395 if ($tagtype == 1 &&
2396 in_array($tagname, $rm_tags_with_content)){
2397 $skip_content = $tagname;
2399 if (($rm_tags == false
2400 && in_array($tagname, $tag_list)) ||
2401 ($rm_tags == true &&
2402 !in_array($tagname, $tag_list))){
2406 * Convert body into div.
2408 if ($tagname == "body"){
2410 $attary = sq_body2div($attary, $mailbox,
2414 if (isset($open_tags{$tagname})){
2415 $open_tags{$tagname}++
;
2417 $open_tags{$tagname}=1;
2421 * This is where we run other checks.
2423 if (is_array($attary) && sizeof($attary) > 0){
2424 $attary = sq_fixatts($tagname,
2438 if ($tagname != false && $skip_content == false){
2439 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2444 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2445 if ($force_tag_closing == true){
2446 foreach ($open_tags as $tagname=>$opentimes){
2447 while ($opentimes > 0){
2448 $trusted .= '</' . $tagname . '>';
2454 $trusted .= "<!-- end sanitized html -->\n";
2459 * This is a wrapper function to call html sanitizing routines.
2461 * @param $body the body of the message
2462 * @param $id the id of the message
2466 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2467 * into internal SM compose links
2468 * (optional; default = TRUE)
2469 * @return a string with html safe to display in the browser.
2471 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2473 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2475 global $attachment_common_show_images, $view_unsafe_images,
2478 * Don't display attached images in HTML mode.
2482 $attachment_common_show_images = false;
2497 $rm_tags_with_content = Array(
2507 $self_closing_tags = Array(
2515 $force_tag_closing = true;
2517 $rm_attnames = Array(
2528 $secremoveimg = "../images/" . _("sec_remove_eng.png");
2529 $bad_attvals = Array(
2532 "/^src|background/i" =>
2535 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2536 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2537 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2540 "\\1$secremoveimg\\2",
2541 "\\1$secremoveimg\\2",
2542 "\\1$secremoveimg\\2",
2545 "/^href|action/i" =>
2548 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2549 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2550 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2565 "/include-source/i",
2567 // position:relative can also be exploited
2568 // to put content outside of email body area
2569 // and position:fixed is similarly exploitable
2570 // as position:absolute, so we'll remove it
2573 // Does this screw up legitimate HTML messages?
2574 // If so, the only fix I see is to allow position
2575 // attributes (any values? I think we still have
2576 // to block static and fixed) only if $use_iframe
2577 // is enabled (1.5.0+)
2579 // was: "/position\s*:\s*absolute/i",
2583 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2584 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2585 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2586 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2587 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2606 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2607 // images off by default.
2608 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
2610 if (!$view_unsafe_images){
2612 * Remove any references to http/https if view_unsafe_images set
2615 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2616 '/^([\'\"])\s*https*:.*([\'\"])/si');
2617 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2618 "\\1$secremoveimg\\1");
2619 array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2620 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2621 array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2622 "url(\\1$secremoveimg\\1)");
2625 $add_attr_to_tag = Array(
2627 Array('target'=>'"_blank"',
2628 'title'=>'"'._("This external link will open in a new window").'"'
2631 $trusted = sq_sanitize($body,
2633 $rm_tags_with_content,
2643 if (strpos($trusted,$secremoveimg)){
2644 $has_unsafe_images = true;
2647 // we want to parse mailto's in HTML output, change to SM compose links
2648 // this is a modified version of code from url_parser.php... but Marc is
2649 // right: we need a better filtering implementation; adding this randomly
2650 // here is not a great solution
2652 if ($take_mailto_links) {
2653 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2654 global $MailTo_PReg_Match;
2655 $MailTo_PReg_Match = '/mailto:' . substr($MailTo_PReg_Match, 1) ;
2656 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2657 foreach ($regs[0] as $i => $mailto_before) {
2658 $mailto_params = $regs[10][$i];
2659 // get rid of any tailing quote since we have to add send_to to the end
2661 if (substr($mailto_before, strlen($mailto_before) - 1) == '"')
2662 $mailto_before = substr($mailto_before, 0, strlen($mailto_before) - 1);
2663 if (substr($mailto_params, strlen($mailto_params) - 1) == '"')
2664 $mailto_params = substr($mailto_params, 0, strlen($mailto_params) - 1);
2666 if ($regs[1][$i]) { //if there is an email addr before '?', we need to merge it with the params
2667 $to = 'to=' . $regs[1][$i];
2668 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2669 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2671 if ($mailto_params) //already some params, append to them
2672 $mailto_params .= '&' . $to;
2674 $mailto_params .= '?' . $to;
2678 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2680 // we'll already have target=_blank, no need to allow comp_in_new
2681 // here (which would be a lot more work anyway)
2683 global $compose_new_win;
2684 $temp_comp_in_new = $compose_new_win;
2685 $compose_new_win = 0;
2686 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2687 $compose_new_win = $temp_comp_in_new;
2689 // remove <a href=" and anything after the next quote (we only
2690 // need the uri, not the link HTML) in compose uri
2692 $comp_uri = substr($comp_uri, 9);
2693 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2694 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2703 * function SendDownloadHeaders - send file to the browser
2705 * Original Source: SM core src/download.php
2706 * moved here to make it available to other code, and separate
2707 * front end from back end functionality.
2709 * @param string $type0 first half of mime type
2710 * @param string $type1 second half of mime type
2711 * @param string $filename filename to tell the browser for downloaded file
2712 * @param boolean $force whether to force the download dialog to pop
2713 * @param optional integer $filesize send the Content-Header and length to the browser
2716 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2717 global $languages, $squirrelmail_language;
2718 $isIE = $isIE6plus = false;
2720 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER
);
2722 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2723 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2727 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2728 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2732 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2733 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2735 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2737 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace(' ', ' ', $filename));
2740 // A Pox on Microsoft and it's Internet Explorer!
2742 // IE has lots of bugs with file downloads.
2743 // It also has problems with SSL. Both of these cause problems
2744 // for us in this function.
2746 // See this article on Cache Control headers and SSL
2747 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2749 // The best thing you can do for IE is to upgrade to the latest
2751 //set all the Cache Control Headers for IE
2753 $filename=rawurlencode($filename);
2754 header ("Pragma: public");
2755 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2756 header ("Cache-Control: post-check=0, pre-check=0", false);
2757 header ("Cache-Control: private");
2759 //set the inline header for IE, we'll add the attachment header later if we need it
2760 header ("Content-Disposition: inline; filename=$filename");
2764 // Try to show in browser window
2765 header ("Content-Disposition: inline; filename=\"$filename\"");
2766 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2768 // Try to pop up the "save as" box
2770 // IE makes this hard. It pops up 2 save boxes, or none.
2771 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2772 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2773 // But, according to Microsoft, it is "RFC compliant but doesn't
2774 // take into account some deviations that allowed within the
2775 // specification." Doesn't that mean RFC non-compliant?
2776 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2778 // all browsers need the application/octet-stream header for this
2779 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2781 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2782 // Do not have quotes around filename, but that applied to
2783 // "attachment"... does it apply to inline too?
2784 header ("Content-Disposition: attachment; filename=\"$filename\"");
2786 if ($isIE && !$isIE6plus) {
2787 // This combination seems to work mostly. IE 5.5 SP 1 has
2788 // known issues (see the Microsoft Knowledge Base)
2790 // This works for most types, but doesn't work with Word files
2791 header ("Content-Type: application/download; name=\"$filename\"");
2792 header ("Content-Type: application/force-download; name=\"$filename\"");
2793 // These are spares, just in case. :-)
2794 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2795 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2796 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2798 // This is to prevent IE for MIME sniffing and auto open a file in IE
2799 header ("Content-Type: application/force-download; name=\"$filename\"");
2801 // another application/octet-stream forces download for Netscape
2802 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2806 //send the content-length header if the calling function provides it
2807 if ($filesize > 0) {
2808 header("Content-Length: $filesize");
2811 } // end fn SendDownloadHeaders