56ee0d47c3320ad1b57e88eb8f1519d0aa346bd6
6 * This contains the functions necessary to detect and decode MIME
9 * @copyright 1999-2020 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 $errormessage .= '<pre>' . sm_encode_html_special_chars($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 .*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.
148 $wholemessage = implode('', $data);
149 if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) {
150 $ret = substr($wholemessage, 0, $regs[1]);
151 /* There is some information in the content info header that could be important
152 * in order to parse html messages. Let's get them here.
154 // if ($ret{0} == '<') {
155 // $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE);
157 } else if (preg_match('/"([^"]*)"/', $topline, $regs)) {
159 } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) {
160 $ret = $wholemessage;
162 global $where, $what, $mailbox, $passed_id, $startMessage;
163 $par = 'mailbox=' . urlencode($mailbox) . '&passed_id=' . $passed_id;
164 if (isset($where) && isset($what)) {
165 $par .= '&where=' . urlencode($where) . '&what=' . urlencode($what);
167 $par .= '&startMessage=' . $startMessage . '&show_more=0';
169 $par .= '&response=' . urlencode($response) .
170 '&message=' . urlencode($message) .
171 '&topline=' . urlencode($topline);
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.") .
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 />";
184 $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE);
186 $wholemessage = implode('', $data);
188 $ret = $wholemessage;
193 function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') {
195 /* Don't kill the connection if the browser is over a dialup
196 * and it would take over 30 seconds to download it.
197 * Don't call set_time_limit in safe mode.
200 if (!ini_get('safe_mode')) {
203 /* in case of base64 encoded attachments, do not buffer them.
204 Instead, echo the decoded attachment directly to screen */
205 if (strtolower($encoding) == 'base64') {
207 $query = "FETCH $id BODY[]";
209 $query = "FETCH $id BODY[$ent_id]";
211 sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
213 $body = mime_fetch_body ($imap_stream, $id, $ent_id);
214 if (is_resource($rStream)) {
215 fputs($rStream,decodeBody($body, $encoding, $force_crlf));
217 echo decodeBody($body, $encoding, $force_crlf);
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.
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.
239 /* -[ END MIME DECODING ]----------------------------------------------------------- */
241 /* This is here for debugging purposes. It will print out a list
242 * of all the entity IDs that are in the $message object.
244 function listEntities ($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++
) {
249 $msg = listEntities($message->entities
[$i]);
259 function getPriorityStr($priority) {
260 $priority_level = substr($priority,0,1);
262 switch($priority_level) {
263 /* Check for a higher then normal priority. */
266 $priority_string = _("High");
269 /* Check for a lower then normal priority. */
272 $priority_string = _("Low");
275 /* Check for a normal priority. */
278 $priority_level = '3';
279 $priority_string = _("Normal");
283 return $priority_string;
286 /* returns a $message object for a particular entity id */
287 function getEntity ($message, $ent_id) {
288 return $message->getEntity($ent_id);
292 * Extracted from strings.php 23/03/2002
295 function translateText(&$body, $wrap_at, $charset) {
296 global $where, $what; /* from searching */
297 global $color; /* color theme */
299 // require_once(SM_PATH . 'functions/url_parser.php');
301 $body_ary = explode("\n", $body);
302 for ($i=0; $i < count($body_ary); $i++
) {
303 $line = rtrim($body_ary[$i],"\r");
305 if (strlen($line) - 2 >= $wrap_at) {
306 sqWordWrap($line, $wrap_at, $charset);
308 $line = charset_decode($charset, $line);
309 $line = str_replace("\t", ' ', $line);
318 if ($line[$pos] == ' ') {
320 } else if (strpos($line, '>', $pos) === $pos) {
329 $line = '<span class="quote1">' . $line . '</span>';
331 $line = '<span class="quote2">' . $line . '</span>';
334 $body_ary[$i] = $line;
336 $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
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
345 * Since 1.2.0 function uses message_body hook.
346 * Till 1.3.0 function included output of formatAttachments().
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 $mailbox (since 1.3.0) imap folder name
355 * @return string html formated message text
357 function formatBody($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.
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, $unsafe_image_toggle_text,
368 // workaround for not updated config.php
369 if (! isset($use_iframe)) $use_iframe = false;
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);
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, $body_message->header
->encoding
);
383 if (isset($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);
390 /* As of 1.5.2, $body is passed (and modified) by reference */
391 do_hook('message_body', $body);
393 /* If there are other types that shouldn't be formatted, add
397 if ($body_message->header
->type1
== 'html') {
398 if ($show_html_default <> 1) {
399 $entity_conv = array(' ' => ' ',
408 $body = strtr($body, $entity_conv);
409 $body = strip_tags($body);
411 translateText($body, $wrap_at,
412 $body_message->header
->getParameter('charset'));
413 } elseif ($use_iframe) {
415 * If we don't add html message between iframe tags,
416 * we must detect unsafe images and modify $has_unsafe_images.
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);
422 // creating iframe url
423 $iframeurl=sqm_baseuri().'src/view_html.php?'
424 . 'mailbox=' . $urlmailbox
425 . '&passed_id=' . $id
426 . '&ent_id=' . $ent_num
427 . '&view_unsafe_images=' . (int) $view_unsafe_images;
430 $oTemplate->assign('iframe_url', $iframeurl);
431 $oTemplate->assign('iframe_height', $iframe_height);
432 $oTemplate->assign('html_body', $html_body);
434 $body = $oTemplate->fetch('read_html_iframe.tpl');
436 // old way of html rendering
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.
442 $charset = $body_message->header
->getParameter('charset');
443 if (!empty($charset)) {
444 $body = charset_decode($charset,$body,false,true);
446 $body = magicHTML($body, $id, $message, $mailbox);
449 translateText($body, $wrap_at,
450 $body_message->header
->getParameter('charset'));
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.
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 = '';
466 // Prepare and build a link for downloading the mail.
467 $link = 'passed_id=' . $id . '&ent_id='.$ent_num.
468 '&mailbox=' . $urlmailbox .'&sort=' . $sort .
469 '&startMessage=' . $startMessage . '&show_more=0';
470 if (isset($passed_ent_id)) {
471 $link .= '&passed_ent_id='.$passed_ent_id;
473 $download_href = SM_PATH
. 'src/download.php?absolute_dl=true&' . $link;
475 // Always add the link for downloading the mail as a file to the global
477 $download_and_unsafe_link .= "$nbsp|$nbsp"
478 . create_hyperlink($download_href, _("Download this as a file"));
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 $text = _("Hide Unsafe Images");
487 if (isset($has_unsafe_images) && $has_unsafe_images) {
488 $link .= '&view_unsafe_images=1';
489 $text = _("View Unsafe Images");
495 // Only create a link for unsafe images if there's need for one. If so:
496 // add it to the global variable.
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);
508 * Generate attachments array for passing to templates.
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 $id message id
516 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
517 global $where, $what, $startMessage, $color, $passed_ent_id,
518 $base_uri, $block_svg_download;
520 $att_ar = $message->getAttachments($exclude_id);
521 $urlMailbox = urlencode($mailbox);
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)
534 $links['download link']['text'] = _("Download");
535 $links['download link']['href'] = $base_uri .
536 "src/download.php?absolute_dl=true&passed_id=$id&mailbox=$urlMailbox&ent_id=$ent";
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 . ']' ;
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));
556 $from_name = _("Unknown sender");
558 $description = _("From").': '.$from_name;
560 $default_page = $base_uri . 'src/download.php';
561 $filename = $att->getFilename();
562 if ($header->description
) {
563 $description = decodeHeader($header->description
);
569 $display_filename = $filename;
570 if (isset($passed_ent_id)) {
571 $passed_ent_id_link = '&passed_ent_id='.$passed_ent_id;
573 $passed_ent_id_link = '';
575 $defaultlink = $default_page . "?startMessage=$startMessage"
576 . "&passed_id=$id&mailbox=$urlMailbox"
577 . '&ent_id='.$ent.$passed_ent_id_link;
578 if ($where && $what) {
579 $defaultlink .= '&where='. urlencode($where).'&what='.urlencode($what);
581 // IE does make use of mime content sniffing. Forcing a download
582 // prohibit execution of XSS inside an application/octet-stream attachment
583 if ($type0 == 'application' && $type1 == 'octet-stream') {
584 $defaultlink .= '&absolute_dl=true';
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
592 // First remember the default link.
593 $defaultlink_orig = $defaultlink;
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) {
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) {
630 do_hook("attachment */*", $temp);
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']) ?
$links['attachment_common']['href'] : '';
639 // base64 encoded file sizes are misleading, so approximate real size
640 if (!empty($header->encoding
) && strtolower($header->encoding
) == 'base64')
641 $this_attachment['Size'] = $header->size
/ 4 * 3;
643 $this_attachment['Size'] = $header->size
;
645 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
646 $this_attachment['OtherLinks'] = array();
647 foreach ($links as $val) {
648 if ($val['text']==_("Download") ||
$val['text'] == _("View"))
650 if (empty($val['text']) && empty($val['extra']))
654 $temp['HREF'] = $val['href'];
655 $temp['Text'] = (empty($val['text']) ?
'' : $val['text']) . (empty($val['extra']) ?
'' : $val['extra']);
656 $this_attachment['OtherLinks'][] = $temp;
658 $attachments[] = $this_attachment;
667 * Displays attachment links and information
669 * Since 1.3.0 function is not included in formatBody() call.
671 * Since 1.0.2 uses attachment $type0/$type1 hook.
672 * Since 1.2.5 uses attachment $type0/* hook.
673 * Since 1.5.0 uses attachments_bottom hook.
674 * Since 1.5.2 uses templates and does *not* return a value.
676 * @param object $message SquirrelMail message object
677 * @param array $exclude_id message parts that are not attachments.
678 * @param string $mailbox mailbox name
679 * @param integer $id message id
681 function formatAttachments($message, $exclude_id, $mailbox, $id) {
684 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
686 $oTemplate->assign('attachments', $attach);
687 $oTemplate->display('read_attachments.tpl');
690 function sqimap_base64_decode(&$string) {
692 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
693 // fly decoding (to reduce memory usage) you have to check if the
694 // data has incomplete pairs
696 // Remove the noise in order to check if the 4 bytes pairs are complete
697 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
700 $iMod = strlen($string) %
4;
702 $sStringRem = substr($string,-$iMod);
703 // Check if $sStringRem contains padding characters
704 if (substr($sStringRem,-1) != '=') {
705 $string = substr($string,0,-$iMod);
710 $string = base64_decode($string);
715 * Decodes encoded string (usually message body)
717 * This function decodes a string (usually the message body)
718 * depending on the encoding type. Currently quoted-printable
719 * and base64 encodings are supported.
721 * The decode_body hook was added to this function in 1.4.2/1.5.0.
722 * The $force_crlf parameter was added in 1.5.2.
724 * @param string $string The encoded string
725 * @param string $encoding used encoding
726 * @param string $force_crlf Whether or not to force CRLF or LF
727 * line endings (or to leave as is).
728 * If given as "LF", line endings will
729 * all be converted to LF; if "CRLF",
730 * line endings will all be converted
731 * to CRLF. If given as an empty value,
732 * the global $force_crlf_default will
733 * be consulted (it can be specified in
734 * config/config_local.php). Otherwise,
735 * any other value will cause the string
736 * to be left alone. Note that this will
737 * be overridden to "LF" if not using at
738 * least PHP version 4.3.0. (OPTIONAL;
739 * default is empty - consult global
742 * @return string The decoded string
747 function decodeBody($string, $encoding, $force_crlf='') {
749 global $force_crlf_default;
750 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
751 $force_crlf = strtoupper($force_crlf);
753 // must force line endings to LF due to broken
754 // quoted_printable_decode() in PHP versions
755 // before 4.3.0 (see below)
757 if (!check_php_version(4, 3, 0) ||
$force_crlf == 'LF')
758 $string = str_replace("\r\n", "\n", $string);
759 else if ($force_crlf == 'CRLF')
760 $string = str_replace("\n", "\r\n", $string);
762 $encoding = strtolower($encoding);
764 $encoding_handler = do_hook('decode_body', $encoding);
767 // plugins get first shot at decoding the string
769 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
770 $string = $encoding_handler('decode', $string);
772 } elseif ($encoding == 'quoted-printable' ||
773 $encoding == 'quoted_printable') {
775 // quoted_printable_decode() function is broken in older
776 // php versions. Text with \r\n decoding was fixed only
777 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
778 // and the above call to: str_replace("\r\n", "\n", $string);
780 $string = quoted_printable_decode($string);
782 } elseif ($encoding == 'base64') {
783 $string = base64_decode($string);
786 // All other encodings are returned raw.
793 * This function decodes strings that are encoded according to
794 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
795 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
797 * @param string $string header string that has to be made readable
798 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
799 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
800 * @param boolean $decide decide if string can be utfencoded. defaults to false
801 * @return string decoded header string
803 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
804 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
805 if (is_array($string)) {
806 $string = implode("\n", $string);
809 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
810 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
811 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
812 // Do we need to return at this point?
819 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
820 $aString = explode(' ',$string);
822 foreach ($aString as $chunk) {
823 if ($encoded && $chunk === '') {
825 } elseif ($chunk === '') {
830 /* if encoded words are not separated by a linear-space-white we still catch them */
833 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
834 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
835 if ($iLastMatch !== $j) {
845 $ret .= sm_encode_html_special_chars($res[1]);
849 $encoding = ucfirst($res[3]);
851 /* decide about valid decoding */
852 if ($decide && is_conversion_safe($res[2])) {
854 $can_be_encoded=true;
856 $can_be_encoded=false;
861 // fix broken base64-encoded strings (remove end = padding,
862 // change any = to + in middle of string, add padding back
864 if ($fix_broken_base64_encoded_messages) {
865 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
866 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
868 $replace = base64_decode($res[4]);
870 if ($can_be_encoded) {
871 /* convert string to different charset,
872 * if functions asks for it (usually in compose)
874 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
876 // convert string to html codes in order to display it
877 $ret .= charset_decode($res[2],$replace);
881 $replace = sm_encode_html_special_chars($replace);
887 $replace = str_replace('_', ' ', $res[4]);
888 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
889 create_function ('$matches', 'return chr(hexdec($matches[1]));'),
892 if ($can_be_encoded) {
893 /* convert string to different charset,
894 * if functions asks for it (usually in compose)
896 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
898 // convert string to html codes in order to display it
899 $replace = charset_decode($res[2], $replace);
903 $replace = sm_encode_html_special_chars($replace);
922 if (!$encoded && $htmlsafe) {
923 $ret .= sm_encode_html_special_chars($chunk);
929 /* remove the first added space */
932 $ret = substr($ret,5);
934 $ret = substr($ret,1);
944 * Function uses XTRA_CODE _encodeheader function, if such function exists.
946 * Function uses Q encoding by default and encodes a string according to RFC
947 * 1522 for use in headers if it contains 8-bit characters or anything that
948 * looks like it should be encoded.
950 * Function switches to B encoding and encodeHeaderBase64() function, if
951 * string is 8bit and multibyte character set supported by mbstring extension
952 * is used. It can cause E_USER_NOTICE errors, if interface is used with
953 * multibyte character set unsupported by mbstring extension.
955 * @param string $string header string, that has to be encoded
956 * @return string quoted-printable encoded string
957 * @todo make $mb_charsets system wide constant
959 function encodeHeader ($string) {
960 global $default_charset, $languages, $squirrelmail_language;
962 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
963 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
964 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
967 // Use B encoding for multibyte charsets
968 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
969 if (in_array($default_charset,$mb_charsets) &&
970 in_array($default_charset,sq_mb_list_encodings()) &&
971 sq_is8bit($string)) {
972 return encodeHeaderBase64($string,$default_charset);
973 } elseif (in_array($default_charset,$mb_charsets) &&
974 sq_is8bit($string) &&
975 ! in_array($default_charset,sq_mb_list_encodings())) {
976 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
977 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
980 // Encode only if the string contains 8-bit characters or =?
981 $j = strlen($string);
982 $max_l = 75 - strlen($default_charset) - 7;
985 $iEncStart = $enc_init = false;
986 $cur_l = $iOffset = 0;
987 for($i = 0; $i < $j; ++
$i) {
997 if ($iEncStart === false) {
1001 if ($cur_l > ($max_l-2)) {
1002 /* if there is an stringpart that doesn't need encoding, add it */
1003 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1004 $aRet[] = "=?$default_charset?Q?$ret?=";
1010 $ret .= sprintf("=%02X",ord($string{$i}));
1015 if ($iEncStart !== false) {
1016 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1017 $aRet[] = "=?$default_charset?Q?$ret?=";
1025 if ($iEncStart !== false) {
1027 if ($cur_l > $max_l) {
1028 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1029 $aRet[] = "=?$default_charset?Q?$ret?=";
1040 $k = ord($string{$i});
1042 if ($iEncStart === false) {
1043 // do not start encoding in the middle of a string, also take the rest of the word.
1044 $sLeadString = substr($string,0,$i);
1045 $aLeadString = explode(' ',$sLeadString);
1046 $sToBeEncoded = array_pop($aLeadString);
1047 $iEncStart = $i - strlen($sToBeEncoded);
1048 $ret .= $sToBeEncoded;
1049 $cur_l +
= strlen($sToBeEncoded);
1052 /* first we add the encoded string that reached it's max size */
1053 if ($cur_l > ($max_l-2)) {
1054 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1055 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1062 $ret .= sprintf("=%02X", $k);
1064 if ($iEncStart !== false) {
1066 if ($cur_l > $max_l) {
1067 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1068 $aRet[] = "=?$default_charset?Q?$ret?=";
1074 $ret .= $string{$i};
1083 if ($iEncStart !== false) {
1084 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1085 $aRet[] = "=?$default_charset?Q?$ret?=";
1087 $aRet[] = substr($string,$iOffset);
1089 $string = implode('',$aRet);
1095 * Encodes string according to rfc2047 B encoding header formating rules
1097 * It is recommended way to encode headers with character sets that store
1098 * symbols in more than one byte.
1100 * Function requires mbstring support. If required mbstring functions are missing,
1101 * function returns false and sets E_USER_WARNING level error message.
1103 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1104 * that mbstring functions will generate E_WARNING errors, if unsupported
1105 * character set is used. mb_encode_mimeheader function provided by php
1106 * mbstring extension is not used in order to get better control of header
1109 * Used php code functions - function_exists(), trigger_error(), strlen()
1110 * (is used with charset names and base64 strings). Used php mbstring
1111 * functions - mb_strlen and mb_substr.
1113 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1114 * encoding), rfc 2822 (header folding)
1116 * @param string $string header string that must be encoded
1117 * @param string $charset character set. Must be supported by mbstring extension.
1118 * Use sq_mb_list_encodings() to detect supported charsets.
1119 * @return string string encoded according to rfc2047 B encoding formating rules
1121 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1122 * @todo Do we want to control max length of header?
1123 * @todo Do we want to control EOL (end-of-line) marker?
1124 * @todo Do we want to translate error message?
1126 function encodeHeaderBase64($string,$charset) {
1128 * Check mbstring function requirements.
1130 if (! function_exists('mb_strlen') ||
1131 ! function_exists('mb_substr')) {
1132 // set E_USER_WARNING
1133 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING
);
1138 // initial return array
1142 * header length = 75 symbols max (same as in encodeHeader)
1143 * remove $charset length
1144 * remove =? ? ?= (5 chars)
1145 * remove 2 more chars (\r\n ?)
1147 $iMaxLength = 75 - strlen($charset) - 7;
1149 // set first character position
1152 // loop through all characters. count characters and not bytes.
1153 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++
) {
1154 // encode string from starting character to current character.
1155 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1157 // Check encoded string length
1158 if(strlen($encoded_string)>$iMaxLength) {
1159 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1160 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1162 // set new starting character
1163 $iStartCharNum = $iCharNum-1;
1165 // encode last char (in case it is last character in string)
1166 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1167 } // if string is shorter than max length - add next character
1170 // add last encoded string to array
1171 $aRet[] = $encoded_string;
1173 // set initial return string
1176 // loop through encoded strings
1177 foreach($aRet as $string) {
1178 // TODO: Do we want to control EOL (end-of-line) marker
1179 if ($sRet!='') $sRet.= " ";
1181 // add header tags and encoded string to return string
1182 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1188 /* This function trys to locate the entity_id of a specific mime element */
1189 function find_ent_id($id, $message) {
1190 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities
); $i++
) {
1191 if ($message->entities
[$i]->header
->type0
== 'multipart') {
1192 $ret = find_ent_id($id, $message->entities
[$i]);
1194 if (strcasecmp($message->entities
[$i]->header
->id
, $id) == 0) {
1195 // if (sq_check_save_extension($message->entities[$i])) {
1196 return $message->entities
[$i]->entity_id
;
1198 } elseif (!empty($message->entities
[$i]->header
->parameters
['name'])) {
1200 * This is part of a fix for Outlook Express 6.x generating
1201 * cid URLs without creating content-id headers
1204 if (strcasecmp($message->entities
[$i]->header
->parameters
['name'], $id) == 0) {
1205 return $message->entities
[$i]->entity_id
;
1213 function sq_check_save_extension($message) {
1214 $filename = $message->getFilename();
1215 $ext = substr($filename, strrpos($filename,'.')+
1);
1216 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1217 return in_array($ext, $save_extensions);
1222 ** HTMLFILTER ROUTINES
1226 * This function checks attribute values for entity-encoded values
1227 * and returns them translated into 8-bit strings so we can run
1230 * @param $attvalue A string to run entity check against.
1231 * @return Nothing, modifies a reference value.
1233 function sq_defang(&$attvalue){
1236 * Skip this if there aren't ampersands or backslashes.
1238 if (strpos($attvalue, '&') === false
1239 && strpos($attvalue, '\\') === false){
1243 // before deent, translate the dangerous unicode characters and ... to safe values
1244 // otherwise the regular expressions do not match.
1250 $m = $m ||
sq_deent($attvalue, '/\�*(\d+);*/s');
1251 $m = $m ||
sq_deent($attvalue, '/\�*((\d|[a-f])+);*/si', true);
1252 $m = $m ||
sq_deent($attvalue, '/\\\\(\d+)/s', true);
1253 } while ($m == true);
1254 $attvalue = stripslashes($attvalue);
1258 * Kill any tabs, newlines, or carriage returns. Our friends the
1259 * makers of the browser with 95% market value decided that it'd
1260 * be funny to make "java[tab]script" be just as good as "javascript".
1262 * @param attvalue The attribute value before extraneous spaces removed.
1263 * @return attvalue Nothing, modifies a reference value.
1265 function sq_unspace(&$attvalue){
1267 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1268 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1269 Array('', '', '', '', ''), $attvalue);
1274 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1275 * IE as regular characters.
1277 * @param attvalue The attribute value before dangerous characters are translated.
1278 * @return attvalue Nothing, modifies a reference value.
1279 * @author Marc Groot Koerkamp.
1281 function sq_fixIE_idiocy(&$attvalue) {
1283 $attvalue = str_replace("\0", "", $attvalue);
1285 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1287 // IE has the evil habit of accepting every possible value for the attribute expression.
1288 // The table below contains characters which are parsed by IE if they are used in the "expression"
1290 $aDangerousCharsReplacementTable = array(
1291 array('ʟ', 'ʟ' ,/* L UNICODE IPA Extension */
1292 'ʀ', 'ʀ' ,/* R UNICODE IPA Extension */
1293 'ɴ', 'ɴ' ,/* N UNICODE IPA Extension */
1294 'E', 'E' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1295 'e', 'e' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1296 'X', 'X',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1297 'x', 'x',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1298 'P', 'P',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1299 'p', 'p',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1300 'R', 'R',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1301 'r', 'r',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1302 'S', 'S',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1303 's', 's',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1304 'I', 'I',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1305 'i', 'i',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1306 'O', 'O',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1307 'o', 'o',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1308 'N', 'N',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1309 'n', 'n',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1310 'L', 'L',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1311 'l', 'l',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1312 'U', 'U',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1313 'u', 'u',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1314 'ⁿ', 'ⁿ' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1315 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1316 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1317 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1318 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1319 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1320 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1321 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1322 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1323 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1324 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1325 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1326 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1327 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1328 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1329 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1330 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1331 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1332 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1333 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1334 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1335 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1336 "\xCA\x9F", /* L UNICODE IPA Extension */
1337 "\xCA\x80", /* R UNICODE IPA Extension */
1338 "\xC9\xB4"), /* N UNICODE IPA Extension */
1339 array('l', 'l', 'r','r','n','n',
1340 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1341 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1342 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1343 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1345 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1347 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1351 * This function returns the final tag out of the tag name, an array
1352 * of attributes, and the type of the tag. This function is called by
1353 * sq_sanitize internally.
1355 * @param $tagname the name of the tag.
1356 * @param $attary the array of attributes and their values
1357 * @param $tagtype The type of the tag (see in comments).
1358 * @return a string with the final tag representation.
1360 function sq_tagprint($tagname, $attary, $tagtype){
1361 $me = 'sq_tagprint';
1364 $fulltag = '</' . $tagname . '>';
1366 $fulltag = '<' . $tagname;
1367 if (is_array($attary) && sizeof($attary)){
1369 while (list($attname, $attvalue) = each($attary)){
1370 array_push($atts, "$attname=$attvalue");
1372 $fulltag .= ' ' . join(" ", $atts);
1383 * A small helper function to use with array_walk. Modifies a by-ref
1384 * value and makes it lowercase.
1386 * @param $val a value passed by-ref.
1387 * @return void since it modifies a by-ref value.
1389 function sq_casenormalize(&$val){
1390 $val = strtolower($val);
1394 * This function skips any whitespace from the current position within
1395 * a string and to the next non-whitespace value.
1397 * @param $body the string
1398 * @param $offset the offset within the string where we should start
1399 * looking for the next non-whitespace character.
1400 * @return the location within the $body where the next
1401 * non-whitespace char is located.
1403 function sq_skipspace($body, $offset){
1404 $me = 'sq_skipspace';
1405 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1406 if (!empty($matches[1])){
1407 $offset +
= strlen($matches[1]);
1413 * This function looks for the next character within a string. It's
1414 * really just a glorified "strpos", except it catches if failures
1417 * @param $body The string to look for needle in.
1418 * @param $offset Start looking from this position.
1419 * @param $needle The character/string to look for.
1420 * @return location of the next occurance of the needle, or
1421 * strlen($body) if needle wasn't found.
1423 function sq_findnxstr($body, $offset, $needle){
1424 $me = 'sq_findnxstr';
1425 $pos = strpos($body, $needle, $offset);
1426 if ($pos === FALSE){
1427 $pos = strlen($body);
1433 * This function takes a PCRE-style regexp and tries to match it
1434 * within the string.
1436 * @param $body The string to look for needle in.
1437 * @param $offset Start looking from here.
1438 * @param $reg A PCRE-style regex to match.
1439 * @return Returns a false if no matches found, or an array
1440 * with the following members:
1441 * - integer with the location of the match within $body
1442 * - string with whatever content between offset and the match
1443 * - string with whatever it is we matched
1445 function sq_findnxreg($body, $offset, $reg){
1446 $me = 'sq_findnxreg';
1449 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1450 if (!isset($matches{0}) ||
!$matches{0}){
1453 $retarr{0} = $offset +
strlen($matches{1});
1454 $retarr{1} = $matches{1};
1455 $retarr{2} = $matches{2};
1461 * This function looks for the next tag.
1463 * @param $body String where to look for the next tag.
1464 * @param $offset Start looking from here.
1465 * @return false if no more tags exist in the body, or
1466 * an array with the following members:
1467 * - string with the name of the tag
1468 * - array with attributes and their values
1469 * - integer with tag type (1, 2, or 3)
1470 * - integer where the tag starts (starting "<")
1471 * - integer where the tag ends (ending ">")
1472 * first three members will be false, if the tag is invalid.
1474 function sq_getnxtag($body, $offset){
1475 $me = 'sq_getnxtag';
1476 if ($offset > strlen($body)){
1479 $lt = sq_findnxstr($body, $offset, "<");
1480 if ($lt == strlen($body)){
1485 * blah blah <tag attribute="value">
1488 $pos = sq_skipspace($body, $lt+
1);
1489 if ($pos >= strlen($body)){
1490 return Array(false, false, false, $lt, strlen($body));
1493 * There are 3 kinds of tags:
1494 * 1. Opening tag, e.g.:
1496 * 2. Closing tag, e.g.:
1498 * 3. XHTML-style content-less tag, e.g.:
1499 * <img src="blah" />
1502 switch (substr($body, $pos, 1)){
1509 * A comment or an SGML declaration.
1511 if (substr($body, $pos+
1, 2) == "--"){
1512 $gt = strpos($body, "-->", $pos);
1514 $gt = strlen($body);
1518 return Array(false, false, false, $lt, $gt);
1520 $gt = sq_findnxstr($body, $pos, ">");
1521 return Array(false, false, false, $lt, $gt);
1526 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1536 * Look for next [\W-_], which will indicate the end of the tag name.
1538 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1539 if ($regary == false){
1540 return Array(false, false, false, $lt, strlen($body));
1542 list($pos, $tagname, $match) = $regary;
1543 $tagname = strtolower($tagname);
1546 * $match can be either of these:
1547 * '>' indicating the end of the tag entirely.
1548 * '\s' indicating the end of the tag name.
1549 * '/' indicating that this is type-3 xhtml tag.
1551 * Whatever else we find there indicates an invalid tag.
1556 * This is an xhtml-style tag with a closing / at the
1557 * end, like so: <img src="blah" />. Check if it's followed
1558 * by the closing bracket. If not, then this tag is invalid
1560 if (substr($body, $pos, 2) == "/>"){
1564 $gt = sq_findnxstr($body, $pos, ">");
1565 $retary = Array(false, false, false, $lt, $gt);
1569 return Array($tagname, false, $tagtype, $lt, $pos);
1573 * Check if it's whitespace
1575 if (!preg_match('/\s/', $match)){
1577 * This is an invalid tag! Look for the next closing ">".
1579 $gt = sq_findnxstr($body, $lt, ">");
1580 return Array(false, false, false, $lt, $gt);
1586 * At this point we're here:
1587 * <tagname attribute='blah'>
1590 * At this point we loop in order to find all attributes.
1596 while ($pos <= strlen($body)){
1597 $pos = sq_skipspace($body, $pos);
1598 if ($pos == strlen($body)){
1602 return Array(false, false, false, $lt, $pos);
1605 * See if we arrived at a ">" or "/>", which means that we reached
1606 * the end of the tag.
1609 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1613 $pos +
= strlen($matches{1});
1614 if ($matches{2} == "/>"){
1618 return Array($tagname, $attary, $tagtype, $lt, $pos);
1622 * There are several types of attributes, with optional
1623 * [:space:] between members.
1625 * attrname[:space:]=[:space:]'CDATA'
1627 * attrname[:space:]=[:space:]"CDATA"
1629 * attr[:space:]=[:space:]CDATA
1633 * We leave types 1 and 2 the same, type 3 we check for
1634 * '"' and convert to """ if needed, then wrap in
1635 * double quotes. Type 4 we convert into:
1638 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1639 if ($regary == false){
1641 * Looks like body ended before the end of tag.
1643 return Array(false, false, false, $lt, strlen($body));
1645 list($pos, $attname, $match) = $regary;
1646 $attname = strtolower($attname);
1648 * We arrived at the end of attribute name. Several things possible
1650 * '>' means the end of the tag and this is attribute type 4
1651 * '/' if followed by '>' means the same thing as above
1652 * '\s' means a lot of things -- look what it's followed by.
1653 * anything else means the attribute is invalid.
1658 * This is an xhtml-style tag with a closing / at the
1659 * end, like so: <img src="blah" />. Check if it's followed
1660 * by the closing bracket. If not, then this tag is invalid
1662 if (substr($body, $pos, 2) == "/>"){
1666 $gt = sq_findnxstr($body, $pos, ">");
1667 $retary = Array(false, false, false, $lt, $gt);
1671 $attary{$attname} = '"yes"';
1672 return Array($tagname, $attary, $tagtype, $lt, $pos);
1676 * Skip whitespace and see what we arrive at.
1678 $pos = sq_skipspace($body, $pos);
1679 $char = substr($body, $pos, 1);
1681 * Two things are valid here:
1682 * '=' means this is attribute type 1 2 or 3.
1683 * \w means this was attribute type 4.
1684 * anything else we ignore and re-loop. End of tag and
1685 * invalid stuff will be caught by our checks at the beginning
1690 $pos = sq_skipspace($body, $pos);
1692 * Here are 3 possibilities:
1693 * "'" attribute type 1
1694 * '"' attribute type 2
1695 * everything else is the content of tag type 3
1697 $quot = substr($body, $pos, 1);
1699 $regary = sq_findnxreg($body, $pos+
1, "\'");
1700 if ($regary == false){
1701 return Array(false, false, false, $lt, strlen($body));
1703 list($pos, $attval, $match) = $regary;
1705 $attary{$attname} = "'" . $attval . "'";
1706 } else if ($quot == '"'){
1707 $regary = sq_findnxreg($body, $pos+
1, '\"');
1708 if ($regary == false){
1709 return Array(false, false, false, $lt, strlen($body));
1711 list($pos, $attval, $match) = $regary;
1713 $attary{$attname} = '"' . $attval . '"';
1716 * These are hateful. Look for \s, or >.
1718 $regary = sq_findnxreg($body, $pos, "[\s>]");
1719 if ($regary == false){
1720 return Array(false, false, false, $lt, strlen($body));
1722 list($pos, $attval, $match) = $regary;
1724 * If it's ">" it will be caught at the top.
1726 $attval = preg_replace("/\"/s", """, $attval);
1727 $attary{$attname} = '"' . $attval . '"';
1729 } else if (preg_match("|[\w/>]|", $char)) {
1731 * That was attribute type 4.
1733 $attary{$attname} = '"yes"';
1736 * An illegal character. Find next '>' and return.
1738 $gt = sq_findnxstr($body, $pos, ">");
1739 return Array(false, false, false, $lt, $gt);
1745 * The fact that we got here indicates that the tag end was never
1746 * found. Return invalid tag indication so it gets stripped.
1748 return Array(false, false, false, $lt, strlen($body));
1752 * Translates entities into literal values so they can be checked.
1754 * @param $attvalue the by-ref value to check.
1755 * @param $regex the regular expression to check against.
1756 * @param $hex whether the entites are hexadecimal.
1757 * @return True or False depending on whether there were matches.
1759 function sq_deent(&$attvalue, $regex, $hex=false){
1763 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1764 preg_match_all($regex, $attvalue, $matches);
1765 if (is_array($matches) && sizeof($matches[0]) > 0){
1767 for ($i = 0; $i < sizeof($matches[0]); $i++
){
1768 $numval = $matches[1][$i];
1770 $numval = hexdec($numval);
1772 $repl{$matches[0][$i]} = chr($numval);
1774 $attvalue = strtr($attvalue, $repl);
1782 * This function runs various checks against the attributes.
1784 * @param $tagname String with the name of the tag.
1785 * @param $attary Array with all tag attributes.
1786 * @param $rm_attnames See description for sq_sanitize
1787 * @param $bad_attvals See description for sq_sanitize
1788 * @param $add_attr_to_tag See description for sq_sanitize
1789 * @param $message message object
1790 * @param $id message id
1791 * @return Array with modified attributes.
1793 function sq_fixatts($tagname,
1803 while (list($attname, $attvalue) = each($attary)){
1805 * See if this attribute should be removed.
1807 foreach ($rm_attnames as $matchtag=>$matchattrs){
1808 if (preg_match($matchtag, $tagname)){
1809 foreach ($matchattrs as $matchattr){
1810 if (preg_match($matchattr, $attname)){
1811 unset($attary{$attname});
1818 * Workaround for IE quirks
1820 sq_fixIE_idiocy($attvalue);
1823 * Remove any backslashes, entities, and extraneous whitespace.
1826 $oldattvalue = $attvalue;
1827 sq_defang($attvalue);
1828 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1829 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1830 // i.e.<div style="{ left:expʀessioɴ( alert('XSS') ) }">
1831 $attvalue = "idiocy";
1832 $attary{$attname} = $attvalue;
1834 sq_unspace($attvalue);
1837 * Now let's run checks on the attvalues.
1838 * I don't expect anyone to comprehend this. If you do,
1839 * get in touch with me so I can drive to where you live and
1840 * shake your hand personally. :)
1842 foreach ($bad_attvals as $matchtag=>$matchattrs){
1843 if (preg_match($matchtag, $tagname)){
1844 foreach ($matchattrs as $matchattr=>$valary){
1845 if (preg_match($matchattr, $attname)){
1847 * There are two arrays in valary.
1849 * Second one is replacements
1851 list($valmatch, $valrepl) = $valary;
1853 preg_replace($valmatch, $valrepl, $attvalue);
1854 if ($newvalue != $attvalue){
1855 $attary{$attname} = $newvalue;
1856 $attvalue = $newvalue;
1862 if ($attname == 'style') {
1863 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1864 // 8bit and control characters in style attribute values can be used for XSS, remove them
1865 $attary{$attname} = '"disallowed character"';
1867 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1868 if (count($aMatch)) {
1869 foreach($aMatch[1] as $sMatch) {
1871 $urlvalue = $sMatch;
1872 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1873 $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1878 * Use white list based filtering on attributes which can contain url's
1880 else if ($attname == 'href' ||
$attname == 'xlink:href' ||
$attname == 'src'
1881 ||
$attname == 'poster' ||
$attname == 'formaction'
1882 ||
$attname == 'background' ||
$attname == 'action') {
1883 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1884 $attary{$attname} = $attvalue;
1888 * See if we need to append any attributes to this tag.
1890 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1891 if (preg_match($matchtag, $tagname)){
1892 $attary = array_merge($attary, $addattary);
1899 * This function filters url's
1901 * @param $attvalue String with attribute value to filter
1902 * @param $message message object
1903 * @param $id message id
1904 * @param $mailbox mailbox
1905 * @param $sQuote quoting characters around url's
1907 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1908 $attvalue = trim($attvalue);
1909 if ($attvalue && ($attvalue[0] =='"'||
$attvalue[0] == "'")) {
1910 // remove the double quotes
1911 $sQuote = $attvalue[0];
1912 $attvalue = trim(substr($attvalue,1,-1));
1915 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1916 // images off by default.
1917 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
1919 global $use_transparent_security_image;
1920 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1921 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1924 * Replace empty src tags with the blank image. src is only used
1925 * for frames, images, and image inputs. Doing a replace should
1926 * not affect them working as should be, however it will stop
1927 * IE from being kicked off when src for img tags are not set
1929 if ($attvalue == '') {
1930 $attvalue = '"' . SM_PATH
. 'images/blank.png"';
1932 // first, disallow 8 bit characters and control characters
1933 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1936 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1939 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
1943 $aUrl = parse_url($attvalue);
1944 if (isset($aUrl['scheme'])) {
1945 switch(strtolower($aUrl['scheme'])) {
1950 if ($attname != 'href') {
1951 if ($view_unsafe_images == false) {
1952 $attvalue = $sQuote . $secremoveimg . $sQuote;
1954 if (isset($aUrl['path'])) {
1956 // No one has been able to show that image URIs
1957 // can be exploited, so for now, no restrictions
1958 // are made at all. If this proves to be a problem,
1959 // the commented-out code below can be of help.
1960 // (One consideration is that I see nothing in this
1961 // function that specifically says that we will
1962 // only ever arrive here when inspecting an image
1963 // tag, although that does seem to be the end
1964 // result - e.g., <script src="..."> where malicious
1965 // image URIs are in fact a problem are already
1966 // filtered out elsewhere.
1967 /* ---------------------------------
1968 // validate image extension.
1969 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1970 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1971 // If URI is to something other than
1972 // a regular image file, get the contents
1973 // and try to see if it is an image.
1974 // Don't use Fileinfo (finfo_file()) because
1975 // we'd need to make the admin configure the
1976 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1979 if (function_exists('mime_content_type')
1980 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1984 $file_contents = '';
1985 while (!feof($FILE)) {
1986 $file_contents .= fread($FILE, 8192);
1990 // store file locally
1992 global $attachment_dir, $username;
1993 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1994 $localfilename = GenerateRandomString(32, '', 7);
1995 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1996 while (file_exists($full_localfilename)) {
1997 $localfilename = GenerateRandomString(32, '', 7);
1998 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2000 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2001 fwrite($FILE, $file_contents);
2004 // get mime type and remove file
2006 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2007 unlink("$hashed_attachment_dir/$localfilename");
2009 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2010 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2011 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2014 --------------------------------- */
2016 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
2020 $attvalue = $sQuote . $attvalue . $sQuote;
2025 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2026 * One day MS might actually make it match something useful, for now, falling
2027 * back to using cid2http, so we can grab the blank.png.
2029 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2033 * Turn cid: urls into http-friendly ones.
2035 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2038 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2042 if (!isset($aUrl['path']) ||
$aUrl['path'] != $secremoveimg) {
2043 // parse_url did not lead to satisfying result
2044 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2052 * This function edits the style definition to make them friendly and
2053 * usable in SquirrelMail.
2055 * @param $message the message object
2056 * @param $id the message id
2057 * @param $content a string with whatever is between <style> and </style>
2058 * @param $mailbox the message mailbox
2059 * @return a string with edited content.
2061 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2062 $me = 'sq_fixstyle';
2063 // workaround for </style> in between comments
2064 $iCurrentPos = $pos;
2069 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++
$i) {
2076 if ($sToken == '<') {
2086 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2091 $content .= $sToken;
2099 if ($sToken == '<') {
2101 if (isset($body{$i+
2}) && substr($body,$i,3) == '!--') {
2102 $i = strpos($body,'-->',$i+
3);
2103 if ($i === false) { // no end comment
2121 if ($bSucces == FALSE){
2122 return array(FALSE, strlen($body));
2128 * First look for general BODY style declaration, which would be
2130 * body {background: blah-blah}
2131 * and change it to .bodyclass so we can just assign it to a <div>
2133 // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2134 // Nah, this is even better - try to preface all CSS selectors with
2135 // our <div> class ID "bodyclass" then correct generic "body" selectors
2136 // TODO: this works pretty good but breaks stuff like this:
2137 // @media print { body { font-size: 10pt; } }
2138 // but there isn't an easy way to make this regex skip @media
2139 // definitions... though lots of the ones in the wild will be
2140 // correctly handled because they tend to end with a parenthesis, like:
2141 // @media screen and (max-width:480px) { ...
2142 $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2143 $content = str_replace('.bodyclass body', '.bodyclass', $content);
2145 global $use_transparent_security_image;
2146 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2147 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2150 * Fix url('blah') declarations.
2152 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2153 // "url(\\1$secremoveimg\\2)", $content);
2155 // first check for 8bit sequences and disallowed control characters
2156 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2157 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2158 return array($content, $newpos);
2161 // IE Sucks hard. We have a special function for it.
2162 sq_fixIE_idiocy($content);
2164 // remove @import line
2165 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2167 // translate ur\l and variations (IE parses that)
2168 // TODO check if the sq_fixIE_idiocy function already handles this.
2169 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2170 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2171 if (count($aMatch)) {
2172 $aValue = $aReplace = array();
2173 foreach($aMatch[1] as $sMatch) {
2175 $urlvalue = $sMatch;
2176 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2177 $aValue[] = $sMatch;
2178 $aReplace[] = $urlvalue;
2180 $content = str_replace($aValue,$aReplace,$content);
2184 * Remove any backslashes, entities, and extraneous whitespace.
2186 $contentTemp = $content;
2187 sq_defang($contentTemp);
2188 sq_unspace($contentTemp);
2191 * Fix stupid css declarations which lead to vulnerabilities
2194 * Also remove "position" attribute, as it can easily be set
2195 * to "fixed" or "absolute" with "left" and "top" attributes
2196 * of zero, taking over the whole content frame. It can also
2197 * be set to relative and move itself anywhere it wants to,
2198 * displaying content in areas it shouldn't be allowed to touch.
2200 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2204 '/include-source/i',
2208 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2209 $contentNew = preg_replace($match, $replace, $contentTemp);
2210 if ($contentNew !== $contentTemp) {
2211 // insecure css declarations are used. From now on we don't care
2212 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2213 $content = $contentNew;
2215 return array($content, $newpos);
2220 * This function converts cid: url's into the ones that can be viewed in
2223 * @param $message the message object
2224 * @param $id the message id
2225 * @param $cidurl the cid: url.
2226 * @param $mailbox the message mailbox
2227 * @return a string with a http-friendly url
2229 function sq_cid2http($message, $id, $cidurl, $mailbox){
2231 * Get rid of quotes.
2233 $quotchar = substr($cidurl, 0, 1);
2234 if ($quotchar == '"' ||
$quotchar == "'"){
2235 $cidurl = str_replace($quotchar, "", $cidurl);
2239 $cidurl = substr(trim($cidurl), 4);
2241 $match_str = '/\{.*?\}\//';
2243 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2245 $linkurl = find_ent_id($cidurl, $message);
2246 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2247 unsafe link image */
2251 * This is part of a fix for Outlook Express 6.x generating
2252 * cid URLs without creating content-id headers. These images are
2253 * not part of the multipart/related html mail. The html contains
2254 * <img src="cid:{some_id}/image_filename.ext"> references to
2255 * attached images with as goal to render them inline although
2256 * the attachment disposition property is not inline.
2259 if (empty($linkurl)) {
2260 if (preg_match('/{.*}\//', $cidurl)) {
2261 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2262 if (!empty($cidurl)) {
2263 $linkurl = find_ent_id($cidurl, $message);
2268 if (!empty($linkurl)) {
2269 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&' .
2270 "passed_id=$id&mailbox=" . urlencode($mailbox) .
2271 '&ent_id=' . $linkurl . $quotchar;
2274 * If we couldn't generate a proper img url, drop in a blank image
2275 * instead of sending back empty, otherwise it causes unusual behaviour
2277 $httpurl = $quotchar . SM_PATH
. 'images/blank.png' . $quotchar;
2284 * This function changes the <body> tag into a <div> tag since we
2285 * can't really have a body-within-body.
2287 * @param $attary an array of attributes and values of <body>
2288 * @param $mailbox mailbox we're currently reading (for cid2http)
2289 * @param $message current message (for cid2http)
2290 * @param $id current message id (for cid2http)
2291 * @return a modified array of attributes to be set for <div>
2293 function sq_body2div($attary, $mailbox, $message, $id){
2294 $me = 'sq_body2div';
2295 $divattary = Array('class' => "'bodyclass'");
2297 $has_bgc_stl = $has_txt_stl = false;
2299 if (is_array($attary) && sizeof($attary) > 0){
2300 foreach ($attary as $attname=>$attvalue){
2301 $quotchar = substr($attvalue, 0, 1);
2302 $attvalue = str_replace($quotchar, "", $attvalue);
2305 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2306 $styledef .= "background-image: url('$attvalue'); ";
2309 $has_bgc_stl = true;
2310 $styledef .= "background-color: $attvalue; ";
2313 $has_txt_stl = true;
2314 $styledef .= "color: $attvalue; ";
2318 // Outlook defines a white bgcolor and no text color. This can lead to
2319 // white text on a white bg with certain themes.
2320 if ($has_bgc_stl && !$has_txt_stl) {
2321 $styledef .= "color: $text; ";
2323 if (strlen($styledef) > 0){
2324 $divattary{"style"} = "\"$styledef\"";
2331 * This is the main function and the one you should actually be calling.
2332 * There are several variables you should be aware of an which need
2333 * special description.
2335 * Since the description is quite lengthy, see it here:
2336 * http://linux.duke.edu/projects/mini/htmlfilter/
2338 * @param $body the string with HTML you wish to filter
2339 * @param $tag_list see description above
2340 * @param $rm_tags_with_content see description above
2341 * @param $self_closing_tags see description above
2342 * @param $force_tag_closing see description above
2343 * @param $rm_attnames see description above
2344 * @param $bad_attvals see description above
2345 * @param $add_attr_to_tag see description above
2346 * @param $message message object
2347 * @param $id message id
2348 * @param $recursively_called boolean flag for recursive calls into this function (optional; default FALSE)
2349 * @return sanitized html safe to show on your pages.
2351 function sq_sanitize($body,
2353 $rm_tags_with_content,
2362 $recursively_called=FALSE
2364 $me = 'sq_sanitize';
2367 * See if tag_list is of tags to remove or tags to allow.
2368 * false means remove these tags
2369 * true means allow these tags
2371 $orig_tag_list = $tag_list;
2372 $rm_tags = array_shift($tag_list);
2375 * Normalize rm_tags and rm_tags_with_content.
2377 @array_walk
($tag_list, 'sq_casenormalize');
2378 @array_walk
($rm_tags_with_content, 'sq_casenormalize');
2379 @array_walk
($self_closing_tags, 'sq_casenormalize');
2382 $open_tags = Array();
2383 $trusted = "\n<!-- begin sanitized html -->\n";
2384 $skip_content = false;
2386 * Take care of netscape's stupid javascript entities like
2389 $body = preg_replace("/&(\{.*?\};)/si", "&\\1", $body);
2391 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2392 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2395 * RCDATA and RAWTEXT tags are handled differently:
2396 * next instance of closing tag is used, whether or not
2397 * the HTML is well formed before that
2399 global $rcdata_rawtext_tags;
2400 if (!$recursively_called
2401 && in_array($tagname, $rcdata_rawtext_tags)
2403 $closing_tag = false;
2404 $closing_tag_offset = $curpos;
2405 // seek out the closing tag for the current RCDATA/RAWTEXT tag
2407 // first we need to move forward to next available closing tag
2408 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2409 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2410 if ($next_tag === false) {
2411 $closing_tag = false;
2414 // but then we have to make sure it's a well-formed tag
2415 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2416 if ($closing_tag === false)
2418 else if ($closing_tag[0] !== false
2419 // these should be redundant
2420 && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2421 $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos +
1),
2422 $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2423 $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2424 $message, $id, $mailbox, true);
2425 $curpos = $closing_tag[4] +
1;
2428 $closing_tag_offset = $next_tag[0] +
1;
2430 if ($closing_tag === false)
2431 { /* 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 */ }
2434 $free_content = substr($body, $curpos, $lt-$curpos);
2436 * Take care of <style>
2438 if ($tagname == "style" && $tagtype == 1){
2439 list($free_content, $curpos) =
2440 sq_fixstyle($body, $gt+
1, $message, $id, $mailbox);
2441 if ($free_content != FALSE){
2442 if ( !empty($attary) ) {
2443 $attary = sq_fixatts($tagname,
2453 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2454 $trusted .= $free_content;
2455 $trusted .= sq_tagprint($tagname, false, 2);
2459 if ($skip_content == false){
2460 $trusted .= $free_content;
2462 if ($tagname != FALSE){
2464 if ($skip_content == $tagname){
2466 * Got to the end of tag we needed to remove.
2469 $skip_content = false;
2471 if ($skip_content == false){
2472 if ($tagname == "body"){
2475 if (isset($open_tags{$tagname}) &&
2476 $open_tags{$tagname} > 0){
2477 $open_tags{$tagname}--;
2485 * $rm_tags_with_content
2487 if ($skip_content == false){
2489 * See if this is a self-closing type and change
2490 * tagtype appropriately.
2493 && in_array($tagname, $self_closing_tags)){
2497 * See if we should skip this tag and any content
2500 if ($tagtype == 1 &&
2501 in_array($tagname, $rm_tags_with_content)){
2502 $skip_content = $tagname;
2504 if (($rm_tags == false
2505 && in_array($tagname, $tag_list)) ||
2506 ($rm_tags == true &&
2507 !in_array($tagname, $tag_list))){
2511 * Convert body into div.
2513 if ($tagname == "body"){
2515 $attary = sq_body2div($attary, $mailbox,
2519 if (isset($open_tags{$tagname})){
2520 $open_tags{$tagname}++
;
2522 $open_tags{$tagname}=1;
2526 * This is where we run other checks.
2528 if (is_array($attary) && sizeof($attary) > 0){
2529 $attary = sq_fixatts($tagname,
2543 if ($tagname != false && $skip_content == false){
2544 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2549 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2550 if ($force_tag_closing == true){
2551 foreach ($open_tags as $tagname=>$opentimes){
2552 while ($opentimes > 0){
2553 $trusted .= '</' . $tagname . '>';
2559 $trusted .= "<!-- end sanitized html -->\n";
2564 * This is a wrapper function to call html sanitizing routines.
2566 * @param $body the body of the message
2567 * @param $id the id of the message
2571 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2572 * into internal SM compose links
2573 * (optional; default = TRUE)
2574 * @return a string with html safe to display in the browser.
2576 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2578 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2580 global $attachment_common_show_images, $view_unsafe_images,
2581 $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2582 $remove_rcdata_rawtext_tags_and_content;
2584 $rcdata_rawtext_tags = array(
2589 // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2593 * Don't display attached images in HTML mode.
2597 $attachment_common_show_images = false;
2599 false, // remove these tags
2611 $rm_tags_with_content = Array(
2621 if (!$allow_svg_display)
2622 $rm_tags_with_content[] = 'svg';
2624 * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2625 * case that they are, but if you prefer to remove them and their contents entirely
2626 * (in most cases, should be a safe thing with minimal impact), you can add the
2627 * following to config/config_local.php
2628 * $remove_rcdata_rawtext_tags_and_content = TRUE;
2630 if ($remove_rcdata_rawtext_tags_and_content)
2631 $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2633 $self_closing_tags = Array(
2641 $force_tag_closing = true;
2643 $rm_attnames = Array(
2654 global $use_transparent_security_image;
2655 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2656 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2658 $bad_attvals = Array(
2661 "/^src|background/i" =>
2664 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2665 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2666 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2669 "\\1$secremoveimg\\2",
2670 "\\1$secremoveimg\\2",
2671 "\\1$secremoveimg\\2",
2674 "/^href|action/i" =>
2677 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2678 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2679 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2694 "/include-source/i",
2696 // position:relative can also be exploited
2697 // to put content outside of email body area
2698 // and position:fixed is similarly exploitable
2699 // as position:absolute, so we'll remove it
2702 // Does this screw up legitimate HTML messages?
2703 // If so, the only fix I see is to allow position
2704 // attributes (any values? I think we still have
2705 // to block static and fixed) only if $use_iframe
2706 // is enabled (1.5.0+)
2708 // was: "/position\s*:\s*absolute/i",
2712 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2713 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2714 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2715 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2716 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2735 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2736 // images off by default.
2737 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
2739 if (!$view_unsafe_images){
2741 * Remove any references to http/https if view_unsafe_images set
2744 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2745 '/^([\'\"])\s*https*:.*([\'\"])/si');
2746 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2747 "\\1$secremoveimg\\1");
2748 array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2749 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2750 array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2751 "url(\\1$secremoveimg\\1)");
2754 $add_attr_to_tag = Array(
2756 Array('target'=>'"_blank"',
2757 'title'=>'"'._("This external link will open in a new window").'"'
2760 $trusted = sq_sanitize($body,
2762 $rm_tags_with_content,
2772 if (strpos($trusted,$secremoveimg)){
2773 $has_unsafe_images = true;
2776 // we want to parse mailto's in HTML output, change to SM compose links
2777 // this is a modified version of code from url_parser.php... but Marc is
2778 // right: we need a better filtering implementation; adding this randomly
2779 // here is not a great solution
2781 if ($take_mailto_links) {
2782 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2783 global $MailTo_PReg_Match;
2784 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2785 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2786 // quote before mailto, and we'll also try to convert the non-links back into links
2787 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2788 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2789 foreach ($regs[0] as $i => $mailto_before) {
2790 $mailto_params = $regs[11][$i];
2792 // get rid of any leading quote we may have captured but don't care about
2794 $mailto_before = ltrim($mailto_before, '"\'');
2796 // get rid of any tailing quote since we have to add send_to to the end
2798 $mailto_before = rtrim($mailto_before, '"\'');
2799 $mailto_params = rtrim($mailto_params, '"\'');
2801 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2802 $to = 'to=' . $regs[2][$i];
2803 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2804 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2806 if ($mailto_params) //already some params, append to them
2807 $mailto_params .= '&' . $to;
2809 $mailto_params .= '?' . $to;
2813 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2815 // we'll already have target=_blank, no need to allow comp_in_new
2816 // here (which would be a lot more work anyway)
2818 global $compose_new_win;
2819 $temp_comp_in_new = $compose_new_win;
2820 $compose_new_win = 0;
2821 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2822 $compose_new_win = $temp_comp_in_new;
2824 // remove <a href=" and anything after the next quote (we only
2825 // need the uri, not the link HTML) in compose uri
2827 // but only do this if the original mailto was in a real anchor tag
2829 if (!empty($regs[1][$i])) {
2830 $comp_uri = substr($comp_uri, 9);
2831 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2833 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2842 * function SendDownloadHeaders - send file to the browser
2844 * Original Source: SM core src/download.php
2845 * moved here to make it available to other code, and separate
2846 * front end from back end functionality.
2848 * @param string $type0 first half of mime type
2849 * @param string $type1 second half of mime type
2850 * @param string $filename filename to tell the browser for downloaded file
2851 * @param boolean $force whether to force the download dialog to pop
2852 * @param optional integer $filesize send the Content-Header and length to the browser
2855 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2856 global $languages, $squirrelmail_language;
2857 $isIE = $isIE6plus = false;
2859 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER
);
2861 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2862 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2866 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2867 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2871 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2872 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2874 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2876 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace(' ', ' ', $filename));
2879 // A Pox on Microsoft and it's Internet Explorer!
2881 // IE has lots of bugs with file downloads.
2882 // It also has problems with SSL. Both of these cause problems
2883 // for us in this function.
2885 // See this article on Cache Control headers and SSL
2886 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2888 // The best thing you can do for IE is to upgrade to the latest
2890 //set all the Cache Control Headers for IE
2892 $filename=rawurlencode($filename);
2893 header ("Pragma: public");
2894 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2895 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2896 // header ("Cache-Control: post-check=0, pre-check=0", false);
2897 header ("Cache-Control: private");
2899 //set the inline header for IE, we'll add the attachment header later if we need it
2900 header ("Content-Disposition: inline; filename=$filename");
2904 // Try to show in browser window
2905 header ("Content-Disposition: inline; filename=\"$filename\"");
2906 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2908 // Try to pop up the "save as" box
2910 // IE makes this hard. It pops up 2 save boxes, or none.
2911 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2912 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2913 // But, according to Microsoft, it is "RFC compliant but doesn't
2914 // take into account some deviations that allowed within the
2915 // specification." Doesn't that mean RFC non-compliant?
2916 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2918 // all browsers need the application/octet-stream header for this
2919 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2921 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2922 // Do not have quotes around filename, but that applied to
2923 // "attachment"... does it apply to inline too?
2924 header ("Content-Disposition: attachment; filename=\"$filename\"");
2926 if ($isIE && !$isIE6plus) {
2927 // This combination seems to work mostly. IE 5.5 SP 1 has
2928 // known issues (see the Microsoft Knowledge Base)
2930 // This works for most types, but doesn't work with Word files
2931 header ("Content-Type: application/download; name=\"$filename\"");
2932 header ("Content-Type: application/force-download; name=\"$filename\"");
2933 // These are spares, just in case. :-)
2934 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2935 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2936 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2938 // This is to prevent IE for MIME sniffing and auto open a file in IE
2939 header ("Content-Type: application/force-download; name=\"$filename\"");
2941 // another application/octet-stream forces download for Netscape
2942 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2946 //send the content-length header if the calling function provides it
2947 if ($filesize > 0) {
2948 header("Content-Length: $filesize");
2951 } // end fn SendDownloadHeaders