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,
603 do_hook("attachment $type0/$type1", $temp);
604 /* The API for this hook has changed as of 1.5.2 so that all plugin
605 arguments are passed in an array instead of each their own plugin
606 argument, and arguments are passed by reference, so instead of
607 returning any changes, changes should simply be made to the original
608 arguments themselves. */
609 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
610 &$defaultlink, &$display_filename, &$where, &$what,
612 // Do not let a generic plugin change the default link if a more
613 // specialized one already did it...
614 if ($defaultlink != $defaultlink_orig) {
618 do_hook("attachment $type0/*", $temp);
619 /* The API for this hook has changed as of 1.5.2 so that all plugin
620 arguments are passed in an array instead of each their own plugin
621 argument, and arguments are passed by reference, so instead of
622 returning any changes, changes should simply be made to the original
623 arguments themselves. */
624 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
625 &$defaultlink, &$display_filename, &$where, &$what,
627 // Do not let a generic plugin change the default link if a more
628 // specialized one already did it...
629 if ($defaultlink != $defaultlink_orig) {
633 do_hook("attachment */*", $temp);
635 $this_attachment = array();
636 $this_attachment['Name'] = decodeHeader($display_filename);
637 $this_attachment['Description'] = $description;
638 $this_attachment['DefaultHREF'] = $defaultlink;
639 $this_attachment['DownloadHREF'] = $links['download link']['href'];
640 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ?
$links['attachment_common']['href'] : '';
642 // base64 encoded file sizes are misleading, so approximate real size
643 if (!empty($header->encoding
) && strtolower($header->encoding
) == 'base64')
644 $this_attachment['Size'] = $header->size
/ 4 * 3;
646 $this_attachment['Size'] = $header->size
;
648 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
649 $this_attachment['OtherLinks'] = array();
650 foreach ($links as $val) {
651 if ($val['text']==_("Download") ||
$val['text'] == _("View"))
653 if (empty($val['text']) && empty($val['extra']))
657 $temp['HREF'] = $val['href'];
658 $temp['Text'] = (empty($val['text']) ?
'' : $val['text']) . (empty($val['extra']) ?
'' : $val['extra']);
659 $this_attachment['OtherLinks'][] = $temp;
661 $attachments[] = $this_attachment;
670 * Displays attachment links and information
672 * Since 1.3.0 function is not included in formatBody() call.
674 * Since 1.0.2 uses attachment $type0/$type1 hook.
675 * Since 1.2.5 uses attachment $type0/* hook.
676 * Since 1.5.0 uses attachments_bottom hook.
677 * Since 1.5.2 uses templates and does *not* return a value.
679 * @param object $message SquirrelMail message object
680 * @param array $exclude_id message parts that are not attachments.
681 * @param string $mailbox mailbox name
682 * @param integer $id message id
684 function formatAttachments($message, $exclude_id, $mailbox, $id) {
687 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
689 $oTemplate->assign('attachments', $attach);
690 $oTemplate->display('read_attachments.tpl');
693 function sqimap_base64_decode(&$string) {
695 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
696 // fly decoding (to reduce memory usage) you have to check if the
697 // data has incomplete pairs
699 // Remove the noise in order to check if the 4 bytes pairs are complete
700 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
703 $iMod = strlen($string) %
4;
705 $sStringRem = substr($string,-$iMod);
706 // Check if $sStringRem contains padding characters
707 if (substr($sStringRem,-1) != '=') {
708 $string = substr($string,0,-$iMod);
713 $string = base64_decode($string);
718 * Decodes encoded string (usually message body)
720 * This function decodes a string (usually the message body)
721 * depending on the encoding type. Currently quoted-printable
722 * and base64 encodings are supported.
724 * The decode_body hook was added to this function in 1.4.2/1.5.0.
725 * The $force_crlf parameter was added in 1.5.2.
727 * @param string $string The encoded string
728 * @param string $encoding used encoding
729 * @param string $force_crlf Whether or not to force CRLF or LF
730 * line endings (or to leave as is).
731 * If given as "LF", line endings will
732 * all be converted to LF; if "CRLF",
733 * line endings will all be converted
734 * to CRLF. If given as an empty value,
735 * the global $force_crlf_default will
736 * be consulted (it can be specified in
737 * config/config_local.php). Otherwise,
738 * any other value will cause the string
739 * to be left alone. Note that this will
740 * be overridden to "LF" if not using at
741 * least PHP version 4.3.0. (OPTIONAL;
742 * default is empty - consult global
745 * @return string The decoded string
750 function decodeBody($string, $encoding, $force_crlf='') {
752 global $force_crlf_default;
753 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
754 $force_crlf = strtoupper($force_crlf);
756 // must force line endings to LF due to broken
757 // quoted_printable_decode() in PHP versions
758 // before 4.3.0 (see below)
760 if (!check_php_version(4, 3, 0) ||
$force_crlf == 'LF')
761 $string = str_replace("\r\n", "\n", $string);
762 else if ($force_crlf == 'CRLF')
763 $string = str_replace("\n", "\r\n", $string);
765 $encoding = strtolower($encoding);
767 $encoding_handler = do_hook('decode_body', $encoding);
770 // plugins get first shot at decoding the string
772 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
773 $string = $encoding_handler('decode', $string);
775 } elseif ($encoding == 'quoted-printable' ||
776 $encoding == 'quoted_printable') {
778 // quoted_printable_decode() function is broken in older
779 // php versions. Text with \r\n decoding was fixed only
780 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
781 // and the above call to: str_replace("\r\n", "\n", $string);
783 $string = quoted_printable_decode($string);
785 } elseif ($encoding == 'base64') {
786 $string = base64_decode($string);
789 // All other encodings are returned raw.
796 * This function decodes strings that are encoded according to
797 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
798 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
800 * @param string $string header string that has to be made readable
801 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
802 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
803 * @param boolean $decide decide if string can be utfencoded. defaults to false
804 * @return string decoded header string
806 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
807 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
808 if (is_array($string)) {
809 $string = implode("\n", $string);
812 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
813 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
814 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
815 // Do we need to return at this point?
822 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
823 $aString = explode(' ',$string);
825 foreach ($aString as $chunk) {
826 if ($encoded && $chunk === '') {
828 } elseif ($chunk === '') {
833 /* if encoded words are not separated by a linear-space-white we still catch them */
836 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
837 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
838 if ($iLastMatch !== $j) {
848 $ret .= sm_encode_html_special_chars($res[1]);
852 $encoding = ucfirst($res[3]);
854 /* decide about valid decoding */
855 if ($decide && is_conversion_safe($res[2])) {
857 $can_be_encoded=true;
859 $can_be_encoded=false;
864 // fix broken base64-encoded strings (remove end = padding,
865 // change any = to + in middle of string, add padding back
867 if ($fix_broken_base64_encoded_messages) {
868 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
869 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
871 $replace = base64_decode($res[4]);
873 if ($can_be_encoded) {
874 /* convert string to different charset,
875 * if functions asks for it (usually in compose)
877 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
879 // convert string to html codes in order to display it
880 $ret .= charset_decode($res[2],$replace);
884 $replace = sm_encode_html_special_chars($replace);
890 $replace = str_replace('_', ' ', $res[4]);
891 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
892 create_function ('$matches', 'return chr(hexdec($matches[1]));'),
895 if ($can_be_encoded) {
896 /* convert string to different charset,
897 * if functions asks for it (usually in compose)
899 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
901 // convert string to html codes in order to display it
902 $replace = charset_decode($res[2], $replace);
906 $replace = sm_encode_html_special_chars($replace);
925 if (!$encoded && $htmlsafe) {
926 $ret .= sm_encode_html_special_chars($chunk);
932 /* remove the first added space */
935 $ret = substr($ret,5);
937 $ret = substr($ret,1);
947 * Function uses XTRA_CODE _encodeheader function, if such function exists.
949 * Function uses Q encoding by default and encodes a string according to RFC
950 * 1522 for use in headers if it contains 8-bit characters or anything that
951 * looks like it should be encoded.
953 * Function switches to B encoding and encodeHeaderBase64() function, if
954 * string is 8bit and multibyte character set supported by mbstring extension
955 * is used. It can cause E_USER_NOTICE errors, if interface is used with
956 * multibyte character set unsupported by mbstring extension.
958 * @param string $string header string, that has to be encoded
959 * @return string quoted-printable encoded string
960 * @todo make $mb_charsets system wide constant
962 function encodeHeader ($string) {
963 global $default_charset, $languages, $squirrelmail_language;
965 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
966 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
967 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
970 // Use B encoding for multibyte charsets
971 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
972 if (in_array($default_charset,$mb_charsets) &&
973 in_array($default_charset,sq_mb_list_encodings()) &&
974 sq_is8bit($string)) {
975 return encodeHeaderBase64($string,$default_charset);
976 } elseif (in_array($default_charset,$mb_charsets) &&
977 sq_is8bit($string) &&
978 ! in_array($default_charset,sq_mb_list_encodings())) {
979 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
980 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
983 // Encode only if the string contains 8-bit characters or =?
984 $j = strlen($string);
985 $max_l = 75 - strlen($default_charset) - 7;
988 $iEncStart = $enc_init = false;
989 $cur_l = $iOffset = 0;
990 for($i = 0; $i < $j; ++
$i) {
1000 if ($iEncStart === false) {
1004 if ($cur_l > ($max_l-2)) {
1005 /* if there is an stringpart that doesn't need encoding, add it */
1006 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1007 $aRet[] = "=?$default_charset?Q?$ret?=";
1013 $ret .= sprintf("=%02X",ord($string{$i}));
1018 if ($iEncStart !== false) {
1019 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1020 $aRet[] = "=?$default_charset?Q?$ret?=";
1028 if ($iEncStart !== false) {
1030 if ($cur_l > $max_l) {
1031 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1032 $aRet[] = "=?$default_charset?Q?$ret?=";
1043 $k = ord($string{$i});
1045 if ($iEncStart === false) {
1046 // do not start encoding in the middle of a string, also take the rest of the word.
1047 $sLeadString = substr($string,0,$i);
1048 $aLeadString = explode(' ',$sLeadString);
1049 $sToBeEncoded = array_pop($aLeadString);
1050 $iEncStart = $i - strlen($sToBeEncoded);
1051 $ret .= $sToBeEncoded;
1052 $cur_l +
= strlen($sToBeEncoded);
1055 /* first we add the encoded string that reached it's max size */
1056 if ($cur_l > ($max_l-2)) {
1057 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1058 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1065 $ret .= sprintf("=%02X", $k);
1067 if ($iEncStart !== false) {
1069 if ($cur_l > $max_l) {
1070 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1071 $aRet[] = "=?$default_charset?Q?$ret?=";
1077 $ret .= $string{$i};
1086 if ($iEncStart !== false) {
1087 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1088 $aRet[] = "=?$default_charset?Q?$ret?=";
1090 $aRet[] = substr($string,$iOffset);
1092 $string = implode('',$aRet);
1098 * Encodes string according to rfc2047 B encoding header formating rules
1100 * It is recommended way to encode headers with character sets that store
1101 * symbols in more than one byte.
1103 * Function requires mbstring support. If required mbstring functions are missing,
1104 * function returns false and sets E_USER_WARNING level error message.
1106 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1107 * that mbstring functions will generate E_WARNING errors, if unsupported
1108 * character set is used. mb_encode_mimeheader function provided by php
1109 * mbstring extension is not used in order to get better control of header
1112 * Used php code functions - function_exists(), trigger_error(), strlen()
1113 * (is used with charset names and base64 strings). Used php mbstring
1114 * functions - mb_strlen and mb_substr.
1116 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1117 * encoding), rfc 2822 (header folding)
1119 * @param string $string header string that must be encoded
1120 * @param string $charset character set. Must be supported by mbstring extension.
1121 * Use sq_mb_list_encodings() to detect supported charsets.
1122 * @return string string encoded according to rfc2047 B encoding formating rules
1124 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1125 * @todo Do we want to control max length of header?
1126 * @todo Do we want to control EOL (end-of-line) marker?
1127 * @todo Do we want to translate error message?
1129 function encodeHeaderBase64($string,$charset) {
1131 * Check mbstring function requirements.
1133 if (! function_exists('mb_strlen') ||
1134 ! function_exists('mb_substr')) {
1135 // set E_USER_WARNING
1136 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING
);
1141 // initial return array
1145 * header length = 75 symbols max (same as in encodeHeader)
1146 * remove $charset length
1147 * remove =? ? ?= (5 chars)
1148 * remove 2 more chars (\r\n ?)
1150 $iMaxLength = 75 - strlen($charset) - 7;
1152 // set first character position
1155 // loop through all characters. count characters and not bytes.
1156 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++
) {
1157 // encode string from starting character to current character.
1158 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1160 // Check encoded string length
1161 if(strlen($encoded_string)>$iMaxLength) {
1162 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1163 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1165 // set new starting character
1166 $iStartCharNum = $iCharNum-1;
1168 // encode last char (in case it is last character in string)
1169 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1170 } // if string is shorter than max length - add next character
1173 // add last encoded string to array
1174 $aRet[] = $encoded_string;
1176 // set initial return string
1179 // loop through encoded strings
1180 foreach($aRet as $string) {
1181 // TODO: Do we want to control EOL (end-of-line) marker
1182 if ($sRet!='') $sRet.= " ";
1184 // add header tags and encoded string to return string
1185 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1191 /* This function trys to locate the entity_id of a specific mime element */
1192 function find_ent_id($id, $message) {
1193 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities
); $i++
) {
1194 if ($message->entities
[$i]->header
->type0
== 'multipart') {
1195 $ret = find_ent_id($id, $message->entities
[$i]);
1197 if (strcasecmp($message->entities
[$i]->header
->id
, $id) == 0) {
1198 // if (sq_check_save_extension($message->entities[$i])) {
1199 return $message->entities
[$i]->entity_id
;
1201 } elseif (!empty($message->entities
[$i]->header
->parameters
['name'])) {
1203 * This is part of a fix for Outlook Express 6.x generating
1204 * cid URLs without creating content-id headers
1207 if (strcasecmp($message->entities
[$i]->header
->parameters
['name'], $id) == 0) {
1208 return $message->entities
[$i]->entity_id
;
1216 function sq_check_save_extension($message) {
1217 $filename = $message->getFilename();
1218 $ext = substr($filename, strrpos($filename,'.')+
1);
1219 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1220 return in_array($ext, $save_extensions);
1225 ** HTMLFILTER ROUTINES
1229 * This function checks attribute values for entity-encoded values
1230 * and returns them translated into 8-bit strings so we can run
1233 * @param $attvalue A string to run entity check against.
1234 * @return Nothing, modifies a reference value.
1236 function sq_defang(&$attvalue){
1239 * Skip this if there aren't ampersands or backslashes.
1241 if (strpos($attvalue, '&') === false
1242 && strpos($attvalue, '\\') === false){
1246 // before deent, translate the dangerous unicode characters and ... to safe values
1247 // otherwise the regular expressions do not match.
1253 $m = $m ||
sq_deent($attvalue, '/\�*(\d+);*/s');
1254 $m = $m ||
sq_deent($attvalue, '/\�*((\d|[a-f])+);*/si', true);
1255 $m = $m ||
sq_deent($attvalue, '/\\\\(\d+)/s', true);
1256 } while ($m == true);
1257 $attvalue = stripslashes($attvalue);
1261 * Kill any tabs, newlines, or carriage returns. Our friends the
1262 * makers of the browser with 95% market value decided that it'd
1263 * be funny to make "java[tab]script" be just as good as "javascript".
1265 * @param attvalue The attribute value before extraneous spaces removed.
1266 * @return attvalue Nothing, modifies a reference value.
1268 function sq_unspace(&$attvalue){
1270 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1271 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1272 Array('', '', '', '', ''), $attvalue);
1277 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1278 * IE as regular characters.
1280 * @param attvalue The attribute value before dangerous characters are translated.
1281 * @return attvalue Nothing, modifies a reference value.
1282 * @author Marc Groot Koerkamp.
1284 function sq_fixIE_idiocy(&$attvalue) {
1286 $attvalue = str_replace("\0", "", $attvalue);
1288 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1290 // IE has the evil habit of accepting every possible value for the attribute expression.
1291 // The table below contains characters which are parsed by IE if they are used in the "expression"
1293 $aDangerousCharsReplacementTable = array(
1294 array('ʟ', 'ʟ' ,/* L UNICODE IPA Extension */
1295 'ʀ', 'ʀ' ,/* R UNICODE IPA Extension */
1296 'ɴ', 'ɴ' ,/* N UNICODE IPA Extension */
1297 'E', 'E' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1298 'e', 'e' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1299 'X', 'X',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1300 'x', 'x',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1301 'P', 'P',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1302 'p', 'p',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1303 'R', 'R',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1304 'r', 'r',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1305 'S', 'S',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1306 's', 's',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1307 'I', 'I',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1308 'i', 'i',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1309 'O', 'O',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1310 'o', 'o',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1311 'N', 'N',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1312 'n', 'n',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1313 'L', 'L',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1314 'l', 'l',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1315 'U', 'U',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1316 'u', 'u',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1317 'ⁿ', 'ⁿ' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1318 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1319 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1320 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1321 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1322 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1323 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1324 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1325 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1326 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1327 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1328 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1329 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1330 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1331 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1332 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1333 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1334 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1335 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1336 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1337 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1338 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1339 "\xCA\x9F", /* L UNICODE IPA Extension */
1340 "\xCA\x80", /* R UNICODE IPA Extension */
1341 "\xC9\xB4"), /* N UNICODE IPA Extension */
1342 array('l', 'l', 'r','r','n','n',
1343 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1344 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1345 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1346 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1348 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1350 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1354 * This function returns the final tag out of the tag name, an array
1355 * of attributes, and the type of the tag. This function is called by
1356 * sq_sanitize internally.
1358 * @param $tagname the name of the tag.
1359 * @param $attary the array of attributes and their values
1360 * @param $tagtype The type of the tag (see in comments).
1361 * @return a string with the final tag representation.
1363 function sq_tagprint($tagname, $attary, $tagtype){
1364 $me = 'sq_tagprint';
1367 $fulltag = '</' . $tagname . '>';
1369 $fulltag = '<' . $tagname;
1370 if (is_array($attary) && sizeof($attary)){
1372 while (list($attname, $attvalue) = each($attary)){
1373 array_push($atts, "$attname=$attvalue");
1375 $fulltag .= ' ' . join(" ", $atts);
1386 * A small helper function to use with array_walk. Modifies a by-ref
1387 * value and makes it lowercase.
1389 * @param $val a value passed by-ref.
1390 * @return void since it modifies a by-ref value.
1392 function sq_casenormalize(&$val){
1393 $val = strtolower($val);
1397 * This function skips any whitespace from the current position within
1398 * a string and to the next non-whitespace value.
1400 * @param $body the string
1401 * @param $offset the offset within the string where we should start
1402 * looking for the next non-whitespace character.
1403 * @return the location within the $body where the next
1404 * non-whitespace char is located.
1406 function sq_skipspace($body, $offset){
1407 $me = 'sq_skipspace';
1408 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1409 if (!empty($matches[1])){
1410 $offset +
= strlen($matches[1]);
1416 * This function looks for the next character within a string. It's
1417 * really just a glorified "strpos", except it catches if failures
1420 * @param $body The string to look for needle in.
1421 * @param $offset Start looking from this position.
1422 * @param $needle The character/string to look for.
1423 * @return location of the next occurance of the needle, or
1424 * strlen($body) if needle wasn't found.
1426 function sq_findnxstr($body, $offset, $needle){
1427 $me = 'sq_findnxstr';
1428 $pos = strpos($body, $needle, $offset);
1429 if ($pos === FALSE){
1430 $pos = strlen($body);
1436 * This function takes a PCRE-style regexp and tries to match it
1437 * within the string.
1439 * @param $body The string to look for needle in.
1440 * @param $offset Start looking from here.
1441 * @param $reg A PCRE-style regex to match.
1442 * @return Returns a false if no matches found, or an array
1443 * with the following members:
1444 * - integer with the location of the match within $body
1445 * - string with whatever content between offset and the match
1446 * - string with whatever it is we matched
1448 function sq_findnxreg($body, $offset, $reg){
1449 $me = 'sq_findnxreg';
1452 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1453 if (!isset($matches{0}) ||
!$matches{0}){
1456 $retarr{0} = $offset +
strlen($matches{1});
1457 $retarr{1} = $matches{1};
1458 $retarr{2} = $matches{2};
1464 * This function looks for the next tag.
1466 * @param $body String where to look for the next tag.
1467 * @param $offset Start looking from here.
1468 * @return false if no more tags exist in the body, or
1469 * an array with the following members:
1470 * - string with the name of the tag
1471 * - array with attributes and their values
1472 * - integer with tag type (1, 2, or 3)
1473 * - integer where the tag starts (starting "<")
1474 * - integer where the tag ends (ending ">")
1475 * first three members will be false, if the tag is invalid.
1477 function sq_getnxtag($body, $offset){
1478 $me = 'sq_getnxtag';
1479 if ($offset > strlen($body)){
1482 $lt = sq_findnxstr($body, $offset, "<");
1483 if ($lt == strlen($body)){
1488 * blah blah <tag attribute="value">
1491 $pos = sq_skipspace($body, $lt+
1);
1492 if ($pos >= strlen($body)){
1493 return Array(false, false, false, $lt, strlen($body));
1496 * There are 3 kinds of tags:
1497 * 1. Opening tag, e.g.:
1499 * 2. Closing tag, e.g.:
1501 * 3. XHTML-style content-less tag, e.g.:
1502 * <img src="blah" />
1505 switch (substr($body, $pos, 1)){
1512 * A comment or an SGML declaration.
1514 if (substr($body, $pos+
1, 2) == "--"){
1515 $gt = strpos($body, "-->", $pos);
1517 $gt = strlen($body);
1521 return Array(false, false, false, $lt, $gt);
1523 $gt = sq_findnxstr($body, $pos, ">");
1524 return Array(false, false, false, $lt, $gt);
1529 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1539 * Look for next [\W-_], which will indicate the end of the tag name.
1541 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1542 if ($regary == false){
1543 return Array(false, false, false, $lt, strlen($body));
1545 list($pos, $tagname, $match) = $regary;
1546 $tagname = strtolower($tagname);
1549 * $match can be either of these:
1550 * '>' indicating the end of the tag entirely.
1551 * '\s' indicating the end of the tag name.
1552 * '/' indicating that this is type-3 xhtml tag.
1554 * Whatever else we find there indicates an invalid tag.
1559 * This is an xhtml-style tag with a closing / at the
1560 * end, like so: <img src="blah" />. Check if it's followed
1561 * by the closing bracket. If not, then this tag is invalid
1563 if (substr($body, $pos, 2) == "/>"){
1567 $gt = sq_findnxstr($body, $pos, ">");
1568 $retary = Array(false, false, false, $lt, $gt);
1572 return Array($tagname, false, $tagtype, $lt, $pos);
1576 * Check if it's whitespace
1578 if (!preg_match('/\s/', $match)){
1580 * This is an invalid tag! Look for the next closing ">".
1582 $gt = sq_findnxstr($body, $lt, ">");
1583 return Array(false, false, false, $lt, $gt);
1589 * At this point we're here:
1590 * <tagname attribute='blah'>
1593 * At this point we loop in order to find all attributes.
1599 while ($pos <= strlen($body)){
1600 $pos = sq_skipspace($body, $pos);
1601 if ($pos == strlen($body)){
1605 return Array(false, false, false, $lt, $pos);
1608 * See if we arrived at a ">" or "/>", which means that we reached
1609 * the end of the tag.
1612 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1616 $pos +
= strlen($matches{1});
1617 if ($matches{2} == "/>"){
1621 return Array($tagname, $attary, $tagtype, $lt, $pos);
1625 * There are several types of attributes, with optional
1626 * [:space:] between members.
1628 * attrname[:space:]=[:space:]'CDATA'
1630 * attrname[:space:]=[:space:]"CDATA"
1632 * attr[:space:]=[:space:]CDATA
1636 * We leave types 1 and 2 the same, type 3 we check for
1637 * '"' and convert to """ if needed, then wrap in
1638 * double quotes. Type 4 we convert into:
1641 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1642 if ($regary == false){
1644 * Looks like body ended before the end of tag.
1646 return Array(false, false, false, $lt, strlen($body));
1648 list($pos, $attname, $match) = $regary;
1649 $attname = strtolower($attname);
1651 * We arrived at the end of attribute name. Several things possible
1653 * '>' means the end of the tag and this is attribute type 4
1654 * '/' if followed by '>' means the same thing as above
1655 * '\s' means a lot of things -- look what it's followed by.
1656 * anything else means the attribute is invalid.
1661 * This is an xhtml-style tag with a closing / at the
1662 * end, like so: <img src="blah" />. Check if it's followed
1663 * by the closing bracket. If not, then this tag is invalid
1665 if (substr($body, $pos, 2) == "/>"){
1669 $gt = sq_findnxstr($body, $pos, ">");
1670 $retary = Array(false, false, false, $lt, $gt);
1674 $attary{$attname} = '"yes"';
1675 return Array($tagname, $attary, $tagtype, $lt, $pos);
1679 * Skip whitespace and see what we arrive at.
1681 $pos = sq_skipspace($body, $pos);
1682 $char = substr($body, $pos, 1);
1684 * Two things are valid here:
1685 * '=' means this is attribute type 1 2 or 3.
1686 * \w means this was attribute type 4.
1687 * anything else we ignore and re-loop. End of tag and
1688 * invalid stuff will be caught by our checks at the beginning
1693 $pos = sq_skipspace($body, $pos);
1695 * Here are 3 possibilities:
1696 * "'" attribute type 1
1697 * '"' attribute type 2
1698 * everything else is the content of tag type 3
1700 $quot = substr($body, $pos, 1);
1702 $regary = sq_findnxreg($body, $pos+
1, "\'");
1703 if ($regary == false){
1704 return Array(false, false, false, $lt, strlen($body));
1706 list($pos, $attval, $match) = $regary;
1708 $attary{$attname} = "'" . $attval . "'";
1709 } else if ($quot == '"'){
1710 $regary = sq_findnxreg($body, $pos+
1, '\"');
1711 if ($regary == false){
1712 return Array(false, false, false, $lt, strlen($body));
1714 list($pos, $attval, $match) = $regary;
1716 $attary{$attname} = '"' . $attval . '"';
1719 * These are hateful. Look for \s, or >.
1721 $regary = sq_findnxreg($body, $pos, "[\s>]");
1722 if ($regary == false){
1723 return Array(false, false, false, $lt, strlen($body));
1725 list($pos, $attval, $match) = $regary;
1727 * If it's ">" it will be caught at the top.
1729 $attval = preg_replace("/\"/s", """, $attval);
1730 $attary{$attname} = '"' . $attval . '"';
1732 } else if (preg_match("|[\w/>]|", $char)) {
1734 * That was attribute type 4.
1736 $attary{$attname} = '"yes"';
1739 * An illegal character. Find next '>' and return.
1741 $gt = sq_findnxstr($body, $pos, ">");
1742 return Array(false, false, false, $lt, $gt);
1748 * The fact that we got here indicates that the tag end was never
1749 * found. Return invalid tag indication so it gets stripped.
1751 return Array(false, false, false, $lt, strlen($body));
1755 * Translates entities into literal values so they can be checked.
1757 * @param $attvalue the by-ref value to check.
1758 * @param $regex the regular expression to check against.
1759 * @param $hex whether the entites are hexadecimal.
1760 * @return True or False depending on whether there were matches.
1762 function sq_deent(&$attvalue, $regex, $hex=false){
1766 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1767 preg_match_all($regex, $attvalue, $matches);
1768 if (is_array($matches) && sizeof($matches[0]) > 0){
1770 for ($i = 0; $i < sizeof($matches[0]); $i++
){
1771 $numval = $matches[1][$i];
1773 $numval = hexdec($numval);
1775 $repl{$matches[0][$i]} = chr($numval);
1777 $attvalue = strtr($attvalue, $repl);
1785 * This function runs various checks against the attributes.
1787 * @param $tagname String with the name of the tag.
1788 * @param $attary Array with all tag attributes.
1789 * @param $rm_attnames See description for sq_sanitize
1790 * @param $bad_attvals See description for sq_sanitize
1791 * @param $add_attr_to_tag See description for sq_sanitize
1792 * @param $message message object
1793 * @param $id message id
1794 * @return Array with modified attributes.
1796 function sq_fixatts($tagname,
1806 while (list($attname, $attvalue) = each($attary)){
1808 * See if this attribute should be removed.
1810 foreach ($rm_attnames as $matchtag=>$matchattrs){
1811 if (preg_match($matchtag, $tagname)){
1812 foreach ($matchattrs as $matchattr){
1813 if (preg_match($matchattr, $attname)){
1814 unset($attary{$attname});
1821 * Workaround for IE quirks
1823 sq_fixIE_idiocy($attvalue);
1826 * Remove any backslashes, entities, and extraneous whitespace.
1829 $oldattvalue = $attvalue;
1830 sq_defang($attvalue);
1831 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1832 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1833 // i.e.<div style="{ left:expʀessioɴ( alert('XSS') ) }">
1834 $attvalue = "idiocy";
1835 $attary{$attname} = $attvalue;
1837 sq_unspace($attvalue);
1840 * Now let's run checks on the attvalues.
1841 * I don't expect anyone to comprehend this. If you do,
1842 * get in touch with me so I can drive to where you live and
1843 * shake your hand personally. :)
1845 foreach ($bad_attvals as $matchtag=>$matchattrs){
1846 if (preg_match($matchtag, $tagname)){
1847 foreach ($matchattrs as $matchattr=>$valary){
1848 if (preg_match($matchattr, $attname)){
1850 * There are two arrays in valary.
1852 * Second one is replacements
1854 list($valmatch, $valrepl) = $valary;
1856 preg_replace($valmatch, $valrepl, $attvalue);
1857 if ($newvalue != $attvalue){
1858 $attary{$attname} = $newvalue;
1859 $attvalue = $newvalue;
1865 if ($attname == 'style') {
1866 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1867 // 8bit and control characters in style attribute values can be used for XSS, remove them
1868 $attary{$attname} = '"disallowed character"';
1870 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1871 if (count($aMatch)) {
1872 foreach($aMatch[1] as $sMatch) {
1874 $urlvalue = $sMatch;
1875 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1876 $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1881 * Use white list based filtering on attributes which can contain url's
1883 else if ($attname == 'href' ||
$attname == 'xlink:href' ||
$attname == 'src'
1884 ||
$attname == 'poster' ||
$attname == 'formaction'
1885 ||
$attname == 'background' ||
$attname == 'action') {
1886 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1887 $attary{$attname} = $attvalue;
1891 * See if we need to append any attributes to this tag.
1893 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1894 if (preg_match($matchtag, $tagname)){
1895 $attary = array_merge($attary, $addattary);
1902 * This function filters url's
1904 * @param $attvalue String with attribute value to filter
1905 * @param $message message object
1906 * @param $id message id
1907 * @param $mailbox mailbox
1908 * @param $sQuote quoting characters around url's
1910 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1911 $attvalue = trim($attvalue);
1912 if ($attvalue && ($attvalue[0] =='"'||
$attvalue[0] == "'")) {
1913 // remove the double quotes
1914 $sQuote = $attvalue[0];
1915 $attvalue = trim(substr($attvalue,1,-1));
1918 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1919 // images off by default.
1920 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
1922 global $use_transparent_security_image;
1923 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1924 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1927 * Replace empty src tags with the blank image. src is only used
1928 * for frames, images, and image inputs. Doing a replace should
1929 * not affect them working as should be, however it will stop
1930 * IE from being kicked off when src for img tags are not set
1932 if ($attvalue == '') {
1933 $attvalue = '"' . SM_PATH
. 'images/blank.png"';
1935 // first, disallow 8 bit characters and control characters
1936 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1939 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1942 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
1946 $aUrl = parse_url($attvalue);
1947 if (isset($aUrl['scheme'])) {
1948 switch(strtolower($aUrl['scheme'])) {
1953 if ($attname != 'href') {
1954 if ($view_unsafe_images == false) {
1955 $attvalue = $sQuote . $secremoveimg . $sQuote;
1957 if (isset($aUrl['path'])) {
1959 // No one has been able to show that image URIs
1960 // can be exploited, so for now, no restrictions
1961 // are made at all. If this proves to be a problem,
1962 // the commented-out code below can be of help.
1963 // (One consideration is that I see nothing in this
1964 // function that specifically says that we will
1965 // only ever arrive here when inspecting an image
1966 // tag, although that does seem to be the end
1967 // result - e.g., <script src="..."> where malicious
1968 // image URIs are in fact a problem are already
1969 // filtered out elsewhere.
1970 /* ---------------------------------
1971 // validate image extension.
1972 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1973 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1974 // If URI is to something other than
1975 // a regular image file, get the contents
1976 // and try to see if it is an image.
1977 // Don't use Fileinfo (finfo_file()) because
1978 // we'd need to make the admin configure the
1979 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1982 if (function_exists('mime_content_type')
1983 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1987 $file_contents = '';
1988 while (!feof($FILE)) {
1989 $file_contents .= fread($FILE, 8192);
1993 // store file locally
1995 global $attachment_dir, $username;
1996 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1997 $localfilename = GenerateRandomString(32, '', 7);
1998 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1999 while (file_exists($full_localfilename)) {
2000 $localfilename = GenerateRandomString(32, '', 7);
2001 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2003 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2004 fwrite($FILE, $file_contents);
2007 // get mime type and remove file
2009 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2010 unlink("$hashed_attachment_dir/$localfilename");
2012 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2013 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2014 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2017 --------------------------------- */
2019 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
2023 $attvalue = $sQuote . $attvalue . $sQuote;
2028 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2029 * One day MS might actually make it match something useful, for now, falling
2030 * back to using cid2http, so we can grab the blank.png.
2032 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2036 * Turn cid: urls into http-friendly ones.
2038 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2041 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2045 if (!isset($aUrl['path']) ||
$aUrl['path'] != $secremoveimg) {
2046 // parse_url did not lead to satisfying result
2047 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2055 * This function edits the style definition to make them friendly and
2056 * usable in SquirrelMail.
2058 * @param $message the message object
2059 * @param $id the message id
2060 * @param $content a string with whatever is between <style> and </style>
2061 * @param $mailbox the message mailbox
2062 * @return a string with edited content.
2064 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2065 $me = 'sq_fixstyle';
2066 // workaround for </style> in between comments
2067 $iCurrentPos = $pos;
2072 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++
$i) {
2079 if ($sToken == '<') {
2089 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2094 $content .= $sToken;
2102 if ($sToken == '<') {
2104 if (isset($body{$i+
2}) && substr($body,$i,3) == '!--') {
2105 $i = strpos($body,'-->',$i+
3);
2106 if ($i === false) { // no end comment
2124 if ($bSucces == FALSE){
2125 return array(FALSE, strlen($body));
2131 * First look for general BODY style declaration, which would be
2133 * body {background: blah-blah}
2134 * and change it to .bodyclass so we can just assign it to a <div>
2136 // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2137 // Nah, this is even better - try to preface all CSS selectors with
2138 // our <div> class ID "bodyclass" then correct generic "body" selectors
2139 // TODO: this works pretty good but breaks stuff like this:
2140 // @media print { body { font-size: 10pt; } }
2141 // but there isn't an easy way to make this regex skip @media
2142 // definitions... though lots of the ones in the wild will be
2143 // correctly handled because they tend to end with a parenthesis, like:
2144 // @media screen and (max-width:480px) { ...
2145 $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2146 $content = str_replace('.bodyclass body', '.bodyclass', $content);
2148 global $use_transparent_security_image;
2149 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2150 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2153 * Fix url('blah') declarations.
2155 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2156 // "url(\\1$secremoveimg\\2)", $content);
2158 // first check for 8bit sequences and disallowed control characters
2159 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2160 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2161 return array($content, $newpos);
2164 // IE Sucks hard. We have a special function for it.
2165 sq_fixIE_idiocy($content);
2167 // remove @import line
2168 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2170 // translate ur\l and variations (IE parses that)
2171 // TODO check if the sq_fixIE_idiocy function already handles this.
2172 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2173 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2174 if (count($aMatch)) {
2175 $aValue = $aReplace = array();
2176 foreach($aMatch[1] as $sMatch) {
2178 $urlvalue = $sMatch;
2179 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2180 $aValue[] = $sMatch;
2181 $aReplace[] = $urlvalue;
2183 $content = str_replace($aValue,$aReplace,$content);
2187 * Remove any backslashes, entities, and extraneous whitespace.
2189 $contentTemp = $content;
2190 sq_defang($contentTemp);
2191 sq_unspace($contentTemp);
2194 * Fix stupid css declarations which lead to vulnerabilities
2197 * Also remove "position" attribute, as it can easily be set
2198 * to "fixed" or "absolute" with "left" and "top" attributes
2199 * of zero, taking over the whole content frame. It can also
2200 * be set to relative and move itself anywhere it wants to,
2201 * displaying content in areas it shouldn't be allowed to touch.
2203 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2207 '/include-source/i',
2211 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2212 $contentNew = preg_replace($match, $replace, $contentTemp);
2213 if ($contentNew !== $contentTemp) {
2214 // insecure css declarations are used. From now on we don't care
2215 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2216 $content = $contentNew;
2218 return array($content, $newpos);
2223 * This function converts cid: url's into the ones that can be viewed in
2226 * @param $message the message object
2227 * @param $id the message id
2228 * @param $cidurl the cid: url.
2229 * @param $mailbox the message mailbox
2230 * @return a string with a http-friendly url
2232 function sq_cid2http($message, $id, $cidurl, $mailbox){
2234 * Get rid of quotes.
2236 $quotchar = substr($cidurl, 0, 1);
2237 if ($quotchar == '"' ||
$quotchar == "'"){
2238 $cidurl = str_replace($quotchar, "", $cidurl);
2242 $cidurl = substr(trim($cidurl), 4);
2244 $match_str = '/\{.*?\}\//';
2246 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2248 $linkurl = find_ent_id($cidurl, $message);
2249 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2250 unsafe link image */
2254 * This is part of a fix for Outlook Express 6.x generating
2255 * cid URLs without creating content-id headers. These images are
2256 * not part of the multipart/related html mail. The html contains
2257 * <img src="cid:{some_id}/image_filename.ext"> references to
2258 * attached images with as goal to render them inline although
2259 * the attachment disposition property is not inline.
2262 if (empty($linkurl)) {
2263 if (preg_match('/{.*}\//', $cidurl)) {
2264 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2265 if (!empty($cidurl)) {
2266 $linkurl = find_ent_id($cidurl, $message);
2271 if (!empty($linkurl)) {
2272 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&' .
2273 "passed_id=$id&mailbox=" . urlencode($mailbox) .
2274 '&ent_id=' . $linkurl . $quotchar;
2277 * If we couldn't generate a proper img url, drop in a blank image
2278 * instead of sending back empty, otherwise it causes unusual behaviour
2280 $httpurl = $quotchar . SM_PATH
. 'images/blank.png' . $quotchar;
2287 * This function changes the <body> tag into a <div> tag since we
2288 * can't really have a body-within-body.
2290 * @param $attary an array of attributes and values of <body>
2291 * @param $mailbox mailbox we're currently reading (for cid2http)
2292 * @param $message current message (for cid2http)
2293 * @param $id current message id (for cid2http)
2294 * @return a modified array of attributes to be set for <div>
2296 function sq_body2div($attary, $mailbox, $message, $id){
2297 $me = 'sq_body2div';
2298 $divattary = Array('class' => "'bodyclass'");
2300 $has_bgc_stl = $has_txt_stl = false;
2302 if (is_array($attary) && sizeof($attary) > 0){
2303 foreach ($attary as $attname=>$attvalue){
2304 $quotchar = substr($attvalue, 0, 1);
2305 $attvalue = str_replace($quotchar, "", $attvalue);
2308 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2309 $styledef .= "background-image: url('$attvalue'); ";
2312 $has_bgc_stl = true;
2313 $styledef .= "background-color: $attvalue; ";
2316 $has_txt_stl = true;
2317 $styledef .= "color: $attvalue; ";
2321 // Outlook defines a white bgcolor and no text color. This can lead to
2322 // white text on a white bg with certain themes.
2323 if ($has_bgc_stl && !$has_txt_stl) {
2324 $styledef .= "color: $text; ";
2326 if (strlen($styledef) > 0){
2327 $divattary{"style"} = "\"$styledef\"";
2334 * This is the main function and the one you should actually be calling.
2335 * There are several variables you should be aware of an which need
2336 * special description.
2338 * Since the description is quite lengthy, see it here:
2339 * http://linux.duke.edu/projects/mini/htmlfilter/
2341 * @param $body the string with HTML you wish to filter
2342 * @param $tag_list see description above
2343 * @param $rm_tags_with_content see description above
2344 * @param $self_closing_tags see description above
2345 * @param $force_tag_closing see description above
2346 * @param $rm_attnames see description above
2347 * @param $bad_attvals see description above
2348 * @param $add_attr_to_tag see description above
2349 * @param $message message object
2350 * @param $id message id
2351 * @param $recursively_called boolean flag for recursive calls into this function (optional; default FALSE)
2352 * @return sanitized html safe to show on your pages.
2354 function sq_sanitize($body,
2356 $rm_tags_with_content,
2365 $recursively_called=FALSE
2367 $me = 'sq_sanitize';
2370 * See if tag_list is of tags to remove or tags to allow.
2371 * false means remove these tags
2372 * true means allow these tags
2374 $orig_tag_list = $tag_list;
2375 $rm_tags = array_shift($tag_list);
2378 * Normalize rm_tags and rm_tags_with_content.
2380 @array_walk
($tag_list, 'sq_casenormalize');
2381 @array_walk
($rm_tags_with_content, 'sq_casenormalize');
2382 @array_walk
($self_closing_tags, 'sq_casenormalize');
2385 $open_tags = Array();
2386 $trusted = "\n<!-- begin sanitized html -->\n";
2387 $skip_content = false;
2389 * Take care of netscape's stupid javascript entities like
2392 $body = preg_replace("/&(\{.*?\};)/si", "&\\1", $body);
2394 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2395 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2398 * RCDATA and RAWTEXT tags are handled differently:
2399 * next instance of closing tag is used, whether or not
2400 * the HTML is well formed before that
2402 global $rcdata_rawtext_tags;
2403 if (!$recursively_called
2404 && in_array($tagname, $rcdata_rawtext_tags)
2406 $closing_tag = false;
2407 $closing_tag_offset = $curpos;
2408 // seek out the closing tag for the current RCDATA/RAWTEXT tag
2410 // first we need to move forward to next available closing tag
2411 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2412 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2413 if ($next_tag === false) {
2414 $closing_tag = false;
2417 // but then we have to make sure it's a well-formed tag
2418 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2419 if ($closing_tag === false)
2421 else if ($closing_tag[0] !== false
2422 // these should be redundant
2423 && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2424 $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos +
1),
2425 $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2426 $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2427 $message, $id, $mailbox, true);
2428 $curpos = $closing_tag[4] +
1;
2431 $closing_tag_offset = $next_tag[0] +
1;
2433 if ($closing_tag === false)
2434 { /* 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 */ }
2437 $free_content = substr($body, $curpos, $lt-$curpos);
2439 * Take care of <style>
2441 if ($tagname == "style" && $tagtype == 1){
2442 list($free_content, $curpos) =
2443 sq_fixstyle($body, $gt+
1, $message, $id, $mailbox);
2444 if ($free_content != FALSE){
2445 if ( !empty($attary) ) {
2446 $attary = sq_fixatts($tagname,
2456 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2457 $trusted .= $free_content;
2458 $trusted .= sq_tagprint($tagname, false, 2);
2462 if ($skip_content == false){
2463 $trusted .= $free_content;
2465 if ($tagname != FALSE){
2467 if ($skip_content == $tagname){
2469 * Got to the end of tag we needed to remove.
2472 $skip_content = false;
2474 if ($skip_content == false){
2475 if ($tagname == "body"){
2478 if (isset($open_tags{$tagname}) &&
2479 $open_tags{$tagname} > 0){
2480 $open_tags{$tagname}--;
2488 * $rm_tags_with_content
2490 if ($skip_content == false){
2492 * See if this is a self-closing type and change
2493 * tagtype appropriately.
2496 && in_array($tagname, $self_closing_tags)){
2500 * See if we should skip this tag and any content
2503 if ($tagtype == 1 &&
2504 in_array($tagname, $rm_tags_with_content)){
2505 $skip_content = $tagname;
2507 if (($rm_tags == false
2508 && in_array($tagname, $tag_list)) ||
2509 ($rm_tags == true &&
2510 !in_array($tagname, $tag_list))){
2514 * Convert body into div.
2516 if ($tagname == "body"){
2518 $attary = sq_body2div($attary, $mailbox,
2522 if (isset($open_tags{$tagname})){
2523 $open_tags{$tagname}++
;
2525 $open_tags{$tagname}=1;
2529 * This is where we run other checks.
2531 if (is_array($attary) && sizeof($attary) > 0){
2532 $attary = sq_fixatts($tagname,
2546 if ($tagname != false && $skip_content == false){
2547 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2552 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2553 if ($force_tag_closing == true){
2554 foreach ($open_tags as $tagname=>$opentimes){
2555 while ($opentimes > 0){
2556 $trusted .= '</' . $tagname . '>';
2562 $trusted .= "<!-- end sanitized html -->\n";
2567 * This is a wrapper function to call html sanitizing routines.
2569 * @param $body the body of the message
2570 * @param $id the id of the message
2574 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2575 * into internal SM compose links
2576 * (optional; default = TRUE)
2577 * @return a string with html safe to display in the browser.
2579 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2581 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2583 global $attachment_common_show_images, $view_unsafe_images,
2584 $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2585 $remove_rcdata_rawtext_tags_and_content;
2587 $rcdata_rawtext_tags = array(
2592 // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2596 * Don't display attached images in HTML mode.
2600 $attachment_common_show_images = false;
2602 false, // remove these tags
2614 $rm_tags_with_content = Array(
2624 if (!$allow_svg_display)
2625 $rm_tags_with_content[] = 'svg';
2627 * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2628 * case that they are, but if you prefer to remove them and their contents entirely
2629 * (in most cases, should be a safe thing with minimal impact), you can add the
2630 * following to config/config_local.php
2631 * $remove_rcdata_rawtext_tags_and_content = TRUE;
2633 if ($remove_rcdata_rawtext_tags_and_content)
2634 $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2636 $self_closing_tags = Array(
2644 $force_tag_closing = true;
2646 $rm_attnames = Array(
2657 global $use_transparent_security_image;
2658 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2659 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2661 $bad_attvals = Array(
2664 "/^src|background/i" =>
2667 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2668 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2669 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2672 "\\1$secremoveimg\\2",
2673 "\\1$secremoveimg\\2",
2674 "\\1$secremoveimg\\2",
2677 "/^href|action/i" =>
2680 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2681 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2682 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2697 "/include-source/i",
2699 // position:relative can also be exploited
2700 // to put content outside of email body area
2701 // and position:fixed is similarly exploitable
2702 // as position:absolute, so we'll remove it
2705 // Does this screw up legitimate HTML messages?
2706 // If so, the only fix I see is to allow position
2707 // attributes (any values? I think we still have
2708 // to block static and fixed) only if $use_iframe
2709 // is enabled (1.5.0+)
2711 // was: "/position\s*:\s*absolute/i",
2715 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2716 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2717 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2718 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2719 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2738 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2739 // images off by default.
2740 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
2742 if (!$view_unsafe_images){
2744 * Remove any references to http/https if view_unsafe_images set
2747 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2748 '/^([\'\"])\s*https*:.*([\'\"])/si');
2749 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2750 "\\1$secremoveimg\\1");
2751 array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2752 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2753 array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2754 "url(\\1$secremoveimg\\1)");
2757 $add_attr_to_tag = Array(
2759 Array('target'=>'"_blank"',
2760 'title'=>'"'._("This external link will open in a new window").'"'
2763 $trusted = sq_sanitize($body,
2765 $rm_tags_with_content,
2775 if (strpos($trusted,$secremoveimg)){
2776 $has_unsafe_images = true;
2779 // we want to parse mailto's in HTML output, change to SM compose links
2780 // this is a modified version of code from url_parser.php... but Marc is
2781 // right: we need a better filtering implementation; adding this randomly
2782 // here is not a great solution
2784 if ($take_mailto_links) {
2785 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2786 global $MailTo_PReg_Match;
2787 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2788 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2789 // quote before mailto, and we'll also try to convert the non-links back into links
2790 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2791 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2792 foreach ($regs[0] as $i => $mailto_before) {
2793 $mailto_params = $regs[11][$i];
2795 // get rid of any leading quote we may have captured but don't care about
2797 $mailto_before = ltrim($mailto_before, '"\'');
2799 // get rid of any tailing quote since we have to add send_to to the end
2801 $mailto_before = rtrim($mailto_before, '"\'');
2802 $mailto_params = rtrim($mailto_params, '"\'');
2804 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2805 $to = 'to=' . $regs[2][$i];
2806 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2807 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2809 if ($mailto_params) //already some params, append to them
2810 $mailto_params .= '&' . $to;
2812 $mailto_params .= '?' . $to;
2816 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2818 // we'll already have target=_blank, no need to allow comp_in_new
2819 // here (which would be a lot more work anyway)
2821 global $compose_new_win;
2822 $temp_comp_in_new = $compose_new_win;
2823 $compose_new_win = 0;
2824 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2825 $compose_new_win = $temp_comp_in_new;
2827 // remove <a href=" and anything after the next quote (we only
2828 // need the uri, not the link HTML) in compose uri
2830 // but only do this if the original mailto was in a real anchor tag
2832 if (!empty($regs[1][$i])) {
2833 $comp_uri = substr($comp_uri, 9);
2834 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2836 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2845 * function SendDownloadHeaders - send file to the browser
2847 * Original Source: SM core src/download.php
2848 * moved here to make it available to other code, and separate
2849 * front end from back end functionality.
2851 * @param string $type0 first half of mime type
2852 * @param string $type1 second half of mime type
2853 * @param string $filename filename to tell the browser for downloaded file
2854 * @param boolean $force whether to force the download dialog to pop
2855 * @param optional integer $filesize send the Content-Header and length to the browser
2858 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2859 global $languages, $squirrelmail_language;
2860 $isIE = $isIE6plus = false;
2862 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER
);
2864 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2865 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2869 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2870 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2874 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2875 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2877 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2879 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace(' ', ' ', $filename));
2882 // A Pox on Microsoft and it's Internet Explorer!
2884 // IE has lots of bugs with file downloads.
2885 // It also has problems with SSL. Both of these cause problems
2886 // for us in this function.
2888 // See this article on Cache Control headers and SSL
2889 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2891 // The best thing you can do for IE is to upgrade to the latest
2893 //set all the Cache Control Headers for IE
2895 $filename=rawurlencode($filename);
2896 header ("Pragma: public");
2897 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2898 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2899 // header ("Cache-Control: post-check=0, pre-check=0", false);
2900 header ("Cache-Control: private");
2902 //set the inline header for IE, we'll add the attachment header later if we need it
2903 header ("Content-Disposition: inline; filename=$filename");
2907 // Try to show in browser window
2908 header ("Content-Disposition: inline; filename=\"$filename\"");
2909 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2911 // Try to pop up the "save as" box
2913 // IE makes this hard. It pops up 2 save boxes, or none.
2914 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2915 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2916 // But, according to Microsoft, it is "RFC compliant but doesn't
2917 // take into account some deviations that allowed within the
2918 // specification." Doesn't that mean RFC non-compliant?
2919 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2921 // all browsers need the application/octet-stream header for this
2922 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2924 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2925 // Do not have quotes around filename, but that applied to
2926 // "attachment"... does it apply to inline too?
2927 header ("Content-Disposition: attachment; filename=\"$filename\"");
2929 if ($isIE && !$isIE6plus) {
2930 // This combination seems to work mostly. IE 5.5 SP 1 has
2931 // known issues (see the Microsoft Knowledge Base)
2933 // This works for most types, but doesn't work with Word files
2934 header ("Content-Type: application/download; name=\"$filename\"");
2935 header ("Content-Type: application/force-download; name=\"$filename\"");
2936 // These are spares, just in case. :-)
2937 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2938 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2939 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2941 // This is to prevent IE for MIME sniffing and auto open a file in IE
2942 header ("Content-Type: application/force-download; name=\"$filename\"");
2944 // another application/octet-stream forces download for Netscape
2945 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2949 //send the content-length header if the calling function provides it
2950 if ($filesize > 0) {
2951 header("Content-Length: $filesize");
2954 } // end fn SendDownloadHeaders