6 * This contains the functions necessary to detect and decode MIME
9 * @copyright 1999-2021 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 // TODO: Needs documentation. $ent_id default is usually 1
194 function mime_print_body_lines ($imap_stream, $id, $ent_id, $encoding, $rStream='php://stdout', $force_crlf='') {
196 /* Don't kill the connection if the browser is over a dialup
197 * and it would take over 30 seconds to download it.
198 * Don't call set_time_limit in safe mode.
201 if (!ini_get('safe_mode')) {
204 /* in case of base64 encoded attachments, do not buffer them.
205 Instead, echo the decoded attachment directly to screen */
206 if (strtolower($encoding) == 'base64') {
208 $query = "FETCH $id BODY[]";
210 $query = "FETCH $id BODY[$ent_id]";
212 sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
214 $body = mime_fetch_body ($imap_stream, $id, $ent_id);
215 if (is_resource($rStream)) {
216 fputs($rStream,decodeBody($body, $encoding, $force_crlf));
218 echo decodeBody($body, $encoding, $force_crlf);
223 TODO, use the same method for quoted printable.
224 However, I assume that quoted printable attachments aren't that large
225 so the performancegain / memory usage drop will be minimal.
226 If we decide to add that then we need to adapt sqimap_fread because
227 we need to split te result on \n and fread doesn't stop at \n. That
228 means we also should provide $results from sqimap_fread (by ref) to
229 te function and set $no_return to false. The $filter function for
230 quoted printable should handle unsetting of $results.
233 TODO 2: find out how we write to the output stream php://stdout. fwrite
234 doesn't work because 'php://stdout isn't a stream.
240 /* -[ END MIME DECODING ]----------------------------------------------------------- */
242 /* This is here for debugging purposes. It will print out a list
243 * of all the entity IDs that are in the $message object.
245 function listEntities ($message) {
247 echo "<tt>" . $message->entity_id
. ' : ' . $message->type0
. '/' . $message->type1
. ' parent = '. $message->parent
->entity_id
. '<br />';
248 for ($i = 0; isset($message->entities
[$i]); $i++
) {
250 $msg = listEntities($message->entities
[$i]);
260 function getPriorityStr($priority) {
261 $priority_level = substr($priority,0,1);
263 switch($priority_level) {
264 /* Check for a higher then normal priority. */
267 $priority_string = _("High");
270 /* Check for a lower then normal priority. */
273 $priority_string = _("Low");
276 /* Check for a normal priority. */
279 $priority_level = '3';
280 $priority_string = _("Normal");
284 return $priority_string;
287 /* returns a $message object for a particular entity id */
288 function getEntity ($message, $ent_id) {
289 return $message->getEntity($ent_id);
293 * Extracted from strings.php 23/03/2002
296 function translateText(&$body, $wrap_at, $charset) {
297 global $where, $what; /* from searching */
298 global $color; /* color theme */
300 // require_once(SM_PATH . 'functions/url_parser.php');
302 $body_ary = explode("\n", $body);
303 for ($i=0; $i < count($body_ary); $i++
) {
304 $line = rtrim($body_ary[$i],"\r");
306 if (strlen($line) - 2 >= $wrap_at) {
307 sqWordWrap($line, $wrap_at, $charset);
309 $line = charset_decode($charset, $line);
310 $line = str_replace("\t", ' ', $line);
319 if ($line[$pos] == ' ') {
321 } else if (strpos($line, '>', $pos) === $pos) {
330 $line = '<span class="quote1">' . $line . '</span>';
332 $line = '<span class="quote2">' . $line . '</span>';
335 $body_ary[$i] = $line;
337 $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
341 * This returns a parsed string called $body. That string can then
342 * be displayed as the actual message in the HTML. It contains
343 * everything needed, including HTML Tags, Attachments at the
346 * Since 1.2.0 function uses message_body hook.
347 * Till 1.3.0 function included output of formatAttachments().
349 * @param resource $imap_stream imap connection resource
350 * @param object $message squirrelmail message object
351 * @param array $color squirrelmail color theme array
352 * @param integer $wrap_at number of characters per line
353 * @param string $ent_num (since 1.3.0) message part id
354 * @param integer $id (since 1.3.0) message id
355 * @param string $mailbox (since 1.3.0) imap folder name
356 * @return string html formated message text
358 function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') {
359 /* This if statement checks for the entity to show as the
360 * primary message. To add more of them, just put them in the
361 * order that is their priority.
363 global $startMessage, $languages, $squirrelmail_language,
364 $show_html_default, $sort, $has_unsafe_images, $passed_ent_id,
365 $use_iframe, $iframe_height, $download_and_unsafe_link,
366 $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text,
369 // workaround for not updated config.php
370 if (! isset($use_iframe)) $use_iframe = false;
372 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
373 // images off by default.
374 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
377 $urlmailbox = urlencode($mailbox);
378 $body_message = getEntity($message, $ent_num);
379 if (($body_message->header
->type0
== 'text') ||
380 ($body_message->header
->type0
== 'rfc822')) {
381 $body = mime_fetch_body ($imap_stream, $id, $ent_num);
382 $body = decodeBody($body, $body_message->header
->encoding
);
384 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
385 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) {
386 if (mb_detect_encoding($body) != 'ASCII') {
387 $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body);
391 /* As of 1.5.2, $body is passed (and modified) by reference */
392 do_hook('message_body', $body);
394 /* If there are other types that shouldn't be formatted, add
398 if ($body_message->header
->type1
== 'html') {
399 if ($show_html_default <> 1) {
400 $entity_conv = array(' ' => ' ',
409 $body = strtr($body, $entity_conv);
410 $body = strip_tags($body);
412 translateText($body, $wrap_at,
413 $body_message->header
->getParameter('charset'));
414 } elseif ($use_iframe) {
416 * If we don't add html message between iframe tags,
417 * we must detect unsafe images and modify $has_unsafe_images.
419 $html_body = magicHTML($body, $id, $message, $mailbox);
420 // Convert character set in order to display html mails in different character set
421 $html_body = charset_decode($body_message->header
->getParameter('charset'),$html_body,false,true);
423 // creating iframe url
424 $iframeurl=sqm_baseuri().'src/view_html.php?'
425 . 'mailbox=' . $urlmailbox
426 . '&passed_id=' . $id
427 . '&ent_id=' . $ent_num
428 . '&view_unsafe_images=' . (int) $view_unsafe_images;
431 $oTemplate->assign('iframe_url', $iframeurl);
432 $oTemplate->assign('iframe_height', $iframe_height);
433 $oTemplate->assign('html_body', $html_body);
435 $body = $oTemplate->fetch('read_html_iframe.tpl');
437 // old way of html rendering
439 * convert character set. charset_decode does not remove html special chars
440 * applied by magicHTML functions and does not sanitize them second time if
441 * fourth argument is true.
443 $charset = $body_message->header
->getParameter('charset');
444 if (!empty($charset)) {
445 $body = charset_decode($charset,$body,false,true);
447 $body = magicHTML($body, $id, $message, $mailbox);
450 translateText($body, $wrap_at,
451 $body_message->header
->getParameter('charset'));
455 * Previously the links for downloading and unsafe images were printed
456 * under the mail. By putting the links in a global variable we can
457 * print it in the toolbar where it belongs. Since the original code was
458 * in this place it's left here. It might be possible to move it to some
459 * other place if that makes sense. The possibility to do so has not
460 * been evaluated yet.
463 // Initialize the global variable to an empty string.
464 // 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.
465 $download_and_unsafe_link = '';
467 // Prepare and build a link for downloading the mail.
468 $link = 'passed_id=' . $id . '&ent_id='.$ent_num.
469 '&mailbox=' . $urlmailbox .'&sort=' . $sort .
470 '&startMessage=' . $startMessage . '&show_more=0';
471 if (isset($passed_ent_id)) {
472 $link .= '&passed_ent_id='.$passed_ent_id;
474 $download_href = SM_PATH
. 'src/download.php?absolute_dl=true&' . $link;
476 // Always add the link for downloading the mail as a file to the global
478 $download_and_unsafe_link .= "$nbsp|$nbsp"
479 . create_hyperlink($download_href, _("Download this as a file"));
481 // Find out the right text to use in the link depending on the
482 // circumstances. If the unsafe images are displayed the link should
483 // hide them, if they aren't displayed the link should only appear if
484 // the mail really contains unsafe images.
485 if ($view_unsafe_images) {
486 $text = _("Hide Unsafe Images");
488 if (isset($has_unsafe_images) && $has_unsafe_images) {
489 $link .= '&view_unsafe_images=1';
490 $text = _("View Unsafe Images");
496 // Only create a link for unsafe images if there's need for one. If so:
497 // add it to the global variable.
499 $unsafe_image_toggle_href = SM_PATH
. 'src/read_body.php?'.$link;
500 $unsafe_image_toggle_text = $text;
501 $download_and_unsafe_link .= "$nbsp|$nbsp"
502 . create_hyperlink($unsafe_image_toggle_href, $text);
509 * Generate attachments array for passing to templates.
512 * @param object $message SquirrelMail message object
513 * @param array $exclude_id message parts that are not attachments.
514 * @param string $mailbox mailbox name
515 * @param integer $id message id
517 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
518 global $where, $what, $startMessage, $color, $passed_ent_id,
519 $base_uri, $block_svg_download;
521 $att_ar = $message->getAttachments($exclude_id);
522 $urlMailbox = urlencode($mailbox);
524 $attachments = array();
525 foreach ($att_ar as $att) {
526 $ent = $att->entity_id
;
527 $header = $att->header
;
528 $type0 = strtolower($header->type0
);
529 $type1 = strtolower($header->type1
);
530 if ($block_svg_download && strpos($type1, 'svg') === 0)
535 $links['download link']['text'] = _("Download");
536 $links['download link']['href'] = $base_uri .
537 "src/download.php?absolute_dl=true&passed_id=$id&mailbox=$urlMailbox&ent_id=$ent";
539 if ($type0 =='message' && $type1 == 'rfc822') {
540 $default_page = $base_uri . 'src/read_body.php';
541 $rfc822_header = $att->rfc822_header
;
542 $filename = $rfc822_header->subject
;
543 if (trim( $filename ) == '') {
544 $filename = 'untitled-[' . $ent . ']' ;
546 $from_o = $rfc822_header->from
;
547 if (is_object($from_o)) {
548 $from_name = decodeHeader($from_o->getAddress(false));
549 } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) {
550 // something weird happens when a digest message is opened and you return to the digest
551 // now the from object is part of an array. Probably the parseHeader call overwrites the info
552 // retrieved from the bodystructure in a different way. We need to fix this later.
553 // possible starting point, do not fetch header we already have and inspect how
554 // the rfc822_header object behaves.
555 $from_name = decodeHeader($from_o[0]->getAddress(false));
557 $from_name = _("Unknown sender");
559 $description = _("From").': '.$from_name;
561 $default_page = $base_uri . 'src/download.php';
562 $filename = $att->getFilename();
563 if ($header->description
) {
564 $description = decodeHeader($header->description
);
570 $display_filename = $filename;
571 if (isset($passed_ent_id)) {
572 $passed_ent_id_link = '&passed_ent_id='.$passed_ent_id;
574 $passed_ent_id_link = '';
576 $defaultlink = $default_page . "?startMessage=$startMessage"
577 . "&passed_id=$id&mailbox=$urlMailbox"
578 . '&ent_id='.$ent.$passed_ent_id_link;
579 if ($where && $what) {
580 $defaultlink .= '&where='. urlencode($where).'&what='.urlencode($what);
582 // IE does make use of mime content sniffing. Forcing a download
583 // prohibit execution of XSS inside an application/octet-stream attachment
584 if ($type0 == 'application' && $type1 == 'octet-stream') {
585 $defaultlink .= '&absolute_dl=true';
588 /* This executes the attachment hook with a specific MIME-type.
589 * It also allows plugins to run if there's a rule for a more
590 * generic type. Finally, a hook for ALL attachment types is
593 // First remember the default link.
594 $defaultlink_orig = $defaultlink;
596 /* The API for this hook has changed as of 1.5.2 so that all plugin
597 arguments are passed in an array instead of each their own plugin
598 argument, and arguments are passed by reference, so instead of
599 returning any changes, changes should simply be made to the original
600 arguments themselves. */
601 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
602 &$defaultlink, &$display_filename, &$where, &$what,
604 do_hook("attachment $type0/$type1", $temp);
605 /* The API for this hook has changed as of 1.5.2 so that all plugin
606 arguments are passed in an array instead of each their own plugin
607 argument, and arguments are passed by reference, so instead of
608 returning any changes, changes should simply be made to the original
609 arguments themselves. */
610 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
611 &$defaultlink, &$display_filename, &$where, &$what,
613 // Do not let a generic plugin change the default link if a more
614 // specialized one already did it...
615 if ($defaultlink != $defaultlink_orig) {
619 do_hook("attachment $type0/*", $temp);
620 /* The API for this hook has changed as of 1.5.2 so that all plugin
621 arguments are passed in an array instead of each their own plugin
622 argument, and arguments are passed by reference, so instead of
623 returning any changes, changes should simply be made to the original
624 arguments themselves. */
625 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
626 &$defaultlink, &$display_filename, &$where, &$what,
628 // Do not let a generic plugin change the default link if a more
629 // specialized one already did it...
630 if ($defaultlink != $defaultlink_orig) {
634 do_hook("attachment */*", $temp);
636 $this_attachment = array();
637 $this_attachment['Name'] = decodeHeader($display_filename);
638 $this_attachment['Description'] = $description;
639 $this_attachment['DefaultHREF'] = $defaultlink;
640 $this_attachment['DownloadHREF'] = $links['download link']['href'];
641 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ?
$links['attachment_common']['href'] : '';
643 // base64 encoded file sizes are misleading, so approximate real size
644 if (!empty($header->encoding
) && strtolower($header->encoding
) == 'base64')
645 $this_attachment['Size'] = $header->size
/ 4 * 3;
647 $this_attachment['Size'] = $header->size
;
649 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
650 $this_attachment['OtherLinks'] = array();
651 foreach ($links as $val) {
652 if ($val['text']==_("Download")) {
653 $this_attachment['DownloadHREF'] = $val['href'];
656 if ($val['text']==_("View")) {
657 $this_attachment['ViewHREF'] = $val['href'];
661 // This makes no sense - If 'text' and 'extra' are just concatenated,
662 // there is no point in having 'extra'.... I am going to assume this
663 // was a mistake and am changing 'extra' to be what I think it was
664 // meant to be: additional tag attributes. However, I'm not checking
665 // extensively for plugins that were using this the wrong way (but why would they?)
666 if (empty($val['text']))
670 $temp['HREF'] = $val['href'];
671 $temp['Text'] = $val['text'];
672 $temp['Extra'] = (empty($val['extra']) ?
'' : $val['extra']);
673 $this_attachment['OtherLinks'][] = $temp;
675 $attachments[] = $this_attachment;
684 * Displays attachment links and information
686 * Since 1.3.0 function is not included in formatBody() call.
688 * Since 1.0.2 uses attachment $type0/$type1 hook.
689 * Since 1.2.5 uses attachment $type0/* hook.
690 * Since 1.5.0 uses attachments_bottom hook.
691 * Since 1.5.2 uses templates and does *not* return a value.
693 * @param object $message SquirrelMail message object
694 * @param array $exclude_id message parts that are not attachments.
695 * @param string $mailbox mailbox name
696 * @param integer $id message id
698 function formatAttachments($message, $exclude_id, $mailbox, $id) {
701 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
703 $oTemplate->assign('attachments', $attach);
704 $oTemplate->display('read_attachments.tpl');
707 function sqimap_base64_decode(&$string) {
709 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
710 // fly decoding (to reduce memory usage) you have to check if the
711 // data has incomplete pairs
713 // Remove the noise in order to check if the 4 bytes pairs are complete
714 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
717 $iMod = strlen($string) %
4;
719 $sStringRem = substr($string,-$iMod);
720 // Check if $sStringRem contains padding characters
721 if (substr($sStringRem,-1) != '=') {
722 $string = substr($string,0,-$iMod);
727 $string = base64_decode($string);
732 * Decodes encoded string (usually message body)
734 * This function decodes a string (usually the message body)
735 * depending on the encoding type. Currently quoted-printable
736 * and base64 encodings are supported.
738 * The decode_body hook was added to this function in 1.4.2/1.5.0.
739 * The $force_crlf parameter was added in 1.5.2.
741 * @param string $string The encoded string
742 * @param string $encoding used encoding
743 * @param string $force_crlf Whether or not to force CRLF or LF
744 * line endings (or to leave as is).
745 * If given as "LF", line endings will
746 * all be converted to LF; if "CRLF",
747 * line endings will all be converted
748 * to CRLF. If given as an empty value,
749 * the global $force_crlf_default will
750 * be consulted (it can be specified in
751 * config/config_local.php). Otherwise,
752 * any other value will cause the string
753 * to be left alone. Note that this will
754 * be overridden to "LF" if not using at
755 * least PHP version 4.3.0. (OPTIONAL;
756 * default is empty - consult global
759 * @return string The decoded string
764 function decodeBody($string, $encoding, $force_crlf='') {
766 global $force_crlf_default;
767 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
768 $force_crlf = strtoupper($force_crlf);
770 // must force line endings to LF due to broken
771 // quoted_printable_decode() in PHP versions
772 // before 4.3.0 (see below)
774 if (!check_php_version(4, 3, 0) ||
$force_crlf == 'LF')
775 $string = str_replace("\r\n", "\n", $string);
776 else if ($force_crlf == 'CRLF')
777 $string = str_replace("\n", "\r\n", $string);
779 $encoding = strtolower($encoding);
781 $encoding_handler = do_hook('decode_body', $encoding);
784 // plugins get first shot at decoding the string
786 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
787 $string = $encoding_handler('decode', $string);
789 } elseif ($encoding == 'quoted-printable' ||
790 $encoding == 'quoted_printable') {
792 // quoted_printable_decode() function is broken in older
793 // php versions. Text with \r\n decoding was fixed only
794 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
795 // and the above call to: str_replace("\r\n", "\n", $string);
797 $string = quoted_printable_decode($string);
799 } elseif ($encoding == 'base64') {
800 $string = base64_decode($string);
803 // All other encodings are returned raw.
810 * This function decodes strings that are encoded according to
811 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
812 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
814 * @param string $string header string that has to be made readable
815 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
816 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
817 * @param boolean $decide decide if string can be utfencoded. defaults to false
818 * @return string decoded header string
820 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
821 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
822 if (is_array($string)) {
823 $string = implode("\n", $string);
826 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
827 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
828 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
829 // Do we need to return at this point?
836 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
837 $aString = explode(' ',$string);
839 foreach ($aString as $chunk) {
840 if ($encoded && $chunk === '') {
842 } elseif ($chunk === '') {
847 /* if encoded words are not separated by a linear-space-white we still catch them */
850 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
851 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
852 if ($iLastMatch !== $j) {
862 $ret .= sm_encode_html_special_chars($res[1]);
866 $encoding = ucfirst($res[3]);
868 /* decide about valid decoding */
869 if ($decide && is_conversion_safe($res[2])) {
871 $can_be_encoded=true;
873 $can_be_encoded=false;
878 // fix broken base64-encoded strings (remove end = padding,
879 // change any = to + in middle of string, add padding back
881 if ($fix_broken_base64_encoded_messages) {
882 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
883 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
885 $replace = base64_decode($res[4]);
887 if ($can_be_encoded) {
888 /* convert string to different charset,
889 * if functions asks for it (usually in compose)
891 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
893 // convert string to html codes in order to display it
894 $ret .= charset_decode($res[2],$replace);
898 $replace = sm_encode_html_special_chars($replace);
904 $replace = str_replace('_', ' ', $res[4]);
905 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
906 (check_php_version(5, 3, 0)
907 ?
function($matches) { return chr(hexdec($matches[1])); }
908 : create_function ('$matches', 'return chr(hexdec($matches[1]));')
912 if ($can_be_encoded) {
913 /* convert string to different charset,
914 * if functions asks for it (usually in compose)
916 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
918 // convert string to html codes in order to display it
919 $replace = charset_decode($res[2], $replace);
923 $replace = sm_encode_html_special_chars($replace);
942 if (!$encoded && $htmlsafe) {
943 $ret .= sm_encode_html_special_chars($chunk);
949 /* remove the first added space */
952 $ret = substr($ret,5);
954 $ret = substr($ret,1);
964 * Function uses XTRA_CODE _encodeheader function, if such function exists.
966 * Function uses Q encoding by default and encodes a string according to RFC
967 * 1522 for use in headers if it contains 8-bit characters or anything that
968 * looks like it should be encoded.
970 * Function switches to B encoding and encodeHeaderBase64() function, if
971 * string is 8bit and multibyte character set supported by mbstring extension
972 * is used. It can cause E_USER_NOTICE errors, if interface is used with
973 * multibyte character set unsupported by mbstring extension.
975 * @param string $string header string, that has to be encoded
976 * @return string quoted-printable encoded string
977 * @todo make $mb_charsets system wide constant
979 function encodeHeader ($string) {
980 global $default_charset, $languages, $squirrelmail_language;
982 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
983 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
984 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
987 // Use B encoding for multibyte charsets
988 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
989 if (in_array($default_charset,$mb_charsets) &&
990 in_array($default_charset,sq_mb_list_encodings()) &&
991 sq_is8bit($string)) {
992 return encodeHeaderBase64($string,$default_charset);
993 } elseif (in_array($default_charset,$mb_charsets) &&
994 sq_is8bit($string) &&
995 ! in_array($default_charset,sq_mb_list_encodings())) {
996 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
997 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
1000 // Encode only if the string contains 8-bit characters or =?
1001 $j = strlen($string);
1002 $max_l = 75 - strlen($default_charset) - 7;
1005 $iEncStart = $enc_init = false;
1006 $cur_l = $iOffset = 0;
1007 for($i = 0; $i < $j; ++
$i) {
1017 if ($iEncStart === false) {
1021 if ($cur_l > ($max_l-2)) {
1022 /* if there is an stringpart that doesn't need encoding, add it */
1023 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1024 $aRet[] = "=?$default_charset?Q?$ret?=";
1030 $ret .= sprintf("=%02X",ord($string[$i]));
1035 if ($iEncStart !== false) {
1036 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1037 $aRet[] = "=?$default_charset?Q?$ret?=";
1045 if ($iEncStart !== false) {
1047 if ($cur_l > $max_l) {
1048 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1049 $aRet[] = "=?$default_charset?Q?$ret?=";
1060 $k = ord($string[$i]);
1062 if ($iEncStart === false) {
1063 // do not start encoding in the middle of a string, also take the rest of the word.
1064 $sLeadString = substr($string,0,$i);
1065 $aLeadString = explode(' ',$sLeadString);
1066 $sToBeEncoded = array_pop($aLeadString);
1067 $iEncStart = $i - strlen($sToBeEncoded);
1068 $ret .= $sToBeEncoded;
1069 $cur_l +
= strlen($sToBeEncoded);
1072 /* first we add the encoded string that reached it's max size */
1073 if ($cur_l > ($max_l-2)) {
1074 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1075 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1082 $ret .= sprintf("=%02X", $k);
1084 if ($iEncStart !== false) {
1086 if ($cur_l > $max_l) {
1087 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1088 $aRet[] = "=?$default_charset?Q?$ret?=";
1094 $ret .= $string[$i];
1103 if ($iEncStart !== false) {
1104 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1105 $aRet[] = "=?$default_charset?Q?$ret?=";
1107 $aRet[] = substr($string,$iOffset);
1109 $string = implode('',$aRet);
1115 * Encodes string according to rfc2047 B encoding header formating rules
1117 * It is recommended way to encode headers with character sets that store
1118 * symbols in more than one byte.
1120 * Function requires mbstring support. If required mbstring functions are missing,
1121 * function returns false and sets E_USER_WARNING level error message.
1123 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1124 * that mbstring functions will generate E_WARNING errors, if unsupported
1125 * character set is used. mb_encode_mimeheader function provided by php
1126 * mbstring extension is not used in order to get better control of header
1129 * Used php code functions - function_exists(), trigger_error(), strlen()
1130 * (is used with charset names and base64 strings). Used php mbstring
1131 * functions - mb_strlen and mb_substr.
1133 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1134 * encoding), rfc 2822 (header folding)
1136 * @param string $string header string that must be encoded
1137 * @param string $charset character set. Must be supported by mbstring extension.
1138 * Use sq_mb_list_encodings() to detect supported charsets.
1139 * @return string string encoded according to rfc2047 B encoding formating rules
1141 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1142 * @todo Do we want to control max length of header?
1143 * @todo Do we want to control EOL (end-of-line) marker?
1144 * @todo Do we want to translate error message?
1146 function encodeHeaderBase64($string,$charset) {
1148 * Check mbstring function requirements.
1150 if (! function_exists('mb_strlen') ||
1151 ! function_exists('mb_substr')) {
1152 // set E_USER_WARNING
1153 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING
);
1158 // initial return array
1162 * header length = 75 symbols max (same as in encodeHeader)
1163 * remove $charset length
1164 * remove =? ? ?= (5 chars)
1165 * remove 2 more chars (\r\n ?)
1167 $iMaxLength = 75 - strlen($charset) - 7;
1169 // set first character position
1172 // loop through all characters. count characters and not bytes.
1173 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++
) {
1174 // encode string from starting character to current character.
1175 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1177 // Check encoded string length
1178 if(strlen($encoded_string)>$iMaxLength) {
1179 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1180 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1182 // set new starting character
1183 $iStartCharNum = $iCharNum-1;
1185 // encode last char (in case it is last character in string)
1186 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1187 } // if string is shorter than max length - add next character
1190 // add last encoded string to array
1191 $aRet[] = $encoded_string;
1193 // set initial return string
1196 // loop through encoded strings
1197 foreach($aRet as $string) {
1198 // TODO: Do we want to control EOL (end-of-line) marker
1199 if ($sRet!='') $sRet.= " ";
1201 // add header tags and encoded string to return string
1202 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1208 /* This function trys to locate the entity_id of a specific mime element */
1209 function find_ent_id($id, $message) {
1210 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities
); $i++
) {
1211 if ($message->entities
[$i]->header
->type0
== 'multipart') {
1212 $ret = find_ent_id($id, $message->entities
[$i]);
1214 if (strcasecmp($message->entities
[$i]->header
->id
, $id) == 0) {
1215 // if (sq_check_save_extension($message->entities[$i])) {
1216 return $message->entities
[$i]->entity_id
;
1218 } elseif (!empty($message->entities
[$i]->header
->parameters
['name'])) {
1220 * This is part of a fix for Outlook Express 6.x generating
1221 * cid URLs without creating content-id headers
1224 if (strcasecmp($message->entities
[$i]->header
->parameters
['name'], $id) == 0) {
1225 return $message->entities
[$i]->entity_id
;
1233 function sq_check_save_extension($message) {
1234 $filename = $message->getFilename();
1235 $ext = substr($filename, strrpos($filename,'.')+
1);
1236 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1237 return in_array($ext, $save_extensions);
1242 ** HTMLFILTER ROUTINES
1246 * This function checks attribute values for entity-encoded values
1247 * and returns them translated into 8-bit strings so we can run
1250 * @param $attvalue A string to run entity check against.
1251 * @return Nothing, modifies a reference value.
1253 function sq_defang(&$attvalue){
1256 * Skip this if there aren't ampersands or backslashes.
1258 if (strpos($attvalue, '&') === false
1259 && strpos($attvalue, '\\') === false){
1263 // before deent, translate the dangerous unicode characters and ... to safe values
1264 // otherwise the regular expressions do not match.
1270 $m = $m ||
sq_deent($attvalue, '/\�*(\d+);*/s');
1271 $m = $m ||
sq_deent($attvalue, '/\�*((\d|[a-f])+);*/si', true);
1272 $m = $m ||
sq_deent($attvalue, '/\\\\(\d+)/s', true);
1273 } while ($m == true);
1274 $attvalue = stripslashes($attvalue);
1278 * Kill any tabs, newlines, or carriage returns. Our friends the
1279 * makers of the browser with 95% market value decided that it'd
1280 * be funny to make "java[tab]script" be just as good as "javascript".
1282 * @param attvalue The attribute value before extraneous spaces removed.
1283 * @return attvalue Nothing, modifies a reference value.
1285 function sq_unspace(&$attvalue){
1287 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1288 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1289 Array('', '', '', '', ''), $attvalue);
1294 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1295 * IE as regular characters.
1297 * @param attvalue The attribute value before dangerous characters are translated.
1298 * @return attvalue Nothing, modifies a reference value.
1299 * @author Marc Groot Koerkamp.
1301 function sq_fixIE_idiocy(&$attvalue) {
1303 $attvalue = str_replace("\0", "", $attvalue);
1305 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1307 // IE has the evil habit of accepting every possible value for the attribute expression.
1308 // The table below contains characters which are parsed by IE if they are used in the "expression"
1310 $aDangerousCharsReplacementTable = array(
1311 array('ʟ', 'ʟ' ,/* L UNICODE IPA Extension */
1312 'ʀ', 'ʀ' ,/* R UNICODE IPA Extension */
1313 'ɴ', 'ɴ' ,/* N UNICODE IPA Extension */
1314 'E', 'E' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1315 'e', 'e' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1316 'X', 'X',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1317 'x', 'x',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1318 'P', 'P',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1319 'p', 'p',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1320 'R', 'R',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1321 'r', 'r',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1322 'S', 'S',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1323 's', 's',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1324 'I', 'I',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1325 'i', 'i',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1326 'O', 'O',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1327 'o', 'o',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1328 'N', 'N',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1329 'n', 'n',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1330 'L', 'L',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1331 'l', 'l',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1332 'U', 'U',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1333 'u', 'u',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1334 'ⁿ', 'ⁿ' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1335 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1336 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1337 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1338 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1339 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1340 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1341 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1342 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1343 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1344 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1345 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1346 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1347 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1348 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1349 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1350 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1351 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1352 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1353 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1354 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1355 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1356 "\xCA\x9F", /* L UNICODE IPA Extension */
1357 "\xCA\x80", /* R UNICODE IPA Extension */
1358 "\xC9\xB4"), /* N UNICODE IPA Extension */
1359 array('l', 'l', 'r','r','n','n',
1360 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1361 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1362 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1363 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1365 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1367 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1371 * This function returns the final tag out of the tag name, an array
1372 * of attributes, and the type of the tag. This function is called by
1373 * sq_sanitize internally.
1375 * @param $tagname the name of the tag.
1376 * @param $attary the array of attributes and their values
1377 * @param $tagtype The type of the tag (see in comments).
1378 * @return a string with the final tag representation.
1380 function sq_tagprint($tagname, $attary, $tagtype){
1381 $me = 'sq_tagprint';
1384 $fulltag = '</' . $tagname . '>';
1386 $fulltag = '<' . $tagname;
1387 if (is_array($attary) && sizeof($attary)){
1389 foreach ($attary as $attname => $attvalue){
1390 array_push($atts, "$attname=$attvalue");
1392 $fulltag .= ' ' . join(" ", $atts);
1403 * A small helper function to use with array_walk. Modifies a by-ref
1404 * value and makes it lowercase.
1406 * @param $val a value passed by-ref.
1407 * @return void since it modifies a by-ref value.
1409 function sq_casenormalize(&$val){
1410 $val = strtolower($val);
1414 * This function skips any whitespace from the current position within
1415 * a string and to the next non-whitespace value.
1417 * @param $body the string
1418 * @param $offset the offset within the string where we should start
1419 * looking for the next non-whitespace character.
1420 * @return the location within the $body where the next
1421 * non-whitespace char is located.
1423 function sq_skipspace($body, $offset){
1424 $me = 'sq_skipspace';
1425 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1426 if (!empty($matches[1])){
1427 $offset +
= strlen($matches[1]);
1433 * This function looks for the next character within a string. It's
1434 * really just a glorified "strpos", except it catches if failures
1437 * @param $body The string to look for needle in.
1438 * @param $offset Start looking from this position.
1439 * @param $needle The character/string to look for.
1440 * @return location of the next occurance of the needle, or
1441 * strlen($body) if needle wasn't found.
1443 function sq_findnxstr($body, $offset, $needle){
1444 $me = 'sq_findnxstr';
1445 $pos = strpos($body, $needle, $offset);
1446 if ($pos === FALSE){
1447 $pos = strlen($body);
1453 * This function takes a PCRE-style regexp and tries to match it
1454 * within the string.
1456 * @param $body The string to look for needle in.
1457 * @param $offset Start looking from here.
1458 * @param $reg A PCRE-style regex to match.
1459 * @return Returns a false if no matches found, or an array
1460 * with the following members:
1461 * - integer with the location of the match within $body
1462 * - string with whatever content between offset and the match
1463 * - string with whatever it is we matched
1465 function sq_findnxreg($body, $offset, $reg){
1466 $me = 'sq_findnxreg';
1469 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1470 if (!isset($matches[0]) ||
!$matches[0]){
1473 $retarr[0] = $offset +
strlen($matches[1]);
1474 $retarr[1] = $matches[1];
1475 $retarr[2] = $matches[2];
1481 * This function looks for the next tag.
1483 * @param $body String where to look for the next tag.
1484 * @param $offset Start looking from here.
1485 * @return false if no more tags exist in the body, or
1486 * an array with the following members:
1487 * - string with the name of the tag
1488 * - array with attributes and their values
1489 * - integer with tag type (1, 2, or 3)
1490 * - integer where the tag starts (starting "<")
1491 * - integer where the tag ends (ending ">")
1492 * first three members will be false, if the tag is invalid.
1494 function sq_getnxtag($body, $offset){
1495 $me = 'sq_getnxtag';
1496 if ($offset > strlen($body)){
1499 $lt = sq_findnxstr($body, $offset, "<");
1500 if ($lt == strlen($body)){
1505 * blah blah <tag attribute="value">
1508 $pos = sq_skipspace($body, $lt+
1);
1509 if ($pos >= strlen($body)){
1510 return Array(false, false, false, $lt, strlen($body));
1513 * There are 3 kinds of tags:
1514 * 1. Opening tag, e.g.:
1516 * 2. Closing tag, e.g.:
1518 * 3. XHTML-style content-less tag, e.g.:
1519 * <img src="blah" />
1522 switch (substr($body, $pos, 1)){
1529 * A comment or an SGML declaration.
1531 if (substr($body, $pos+
1, 2) == "--"){
1532 $gt = strpos($body, "-->", $pos);
1534 $gt = strlen($body);
1538 return Array(false, false, false, $lt, $gt);
1540 $gt = sq_findnxstr($body, $pos, ">");
1541 return Array(false, false, false, $lt, $gt);
1546 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1556 * Look for next [\W-_], which will indicate the end of the tag name.
1558 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1559 if ($regary == false){
1560 return Array(false, false, false, $lt, strlen($body));
1562 list($pos, $tagname, $match) = $regary;
1563 $tagname = strtolower($tagname);
1566 * $match can be either of these:
1567 * '>' indicating the end of the tag entirely.
1568 * '\s' indicating the end of the tag name.
1569 * '/' indicating that this is type-3 xhtml tag.
1571 * Whatever else we find there indicates an invalid tag.
1576 * This is an xhtml-style tag with a closing / at the
1577 * end, like so: <img src="blah" />. Check if it's followed
1578 * by the closing bracket. If not, then this tag is invalid
1580 if (substr($body, $pos, 2) == "/>"){
1584 $gt = sq_findnxstr($body, $pos, ">");
1585 $retary = Array(false, false, false, $lt, $gt);
1589 return Array($tagname, false, $tagtype, $lt, $pos);
1593 * Check if it's whitespace
1595 if (!preg_match('/\s/', $match)){
1597 * This is an invalid tag! Look for the next closing ">".
1599 $gt = sq_findnxstr($body, $lt, ">");
1600 return Array(false, false, false, $lt, $gt);
1606 * At this point we're here:
1607 * <tagname attribute='blah'>
1610 * At this point we loop in order to find all attributes.
1616 while ($pos <= strlen($body)){
1617 $pos = sq_skipspace($body, $pos);
1618 if ($pos == strlen($body)){
1622 return Array(false, false, false, $lt, $pos);
1625 * See if we arrived at a ">" or "/>", which means that we reached
1626 * the end of the tag.
1629 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1633 $pos +
= strlen($matches[1]);
1634 if ($matches[2] == "/>"){
1638 return Array($tagname, $attary, $tagtype, $lt, $pos);
1642 * There are several types of attributes, with optional
1643 * [:space:] between members.
1645 * attrname[:space:]=[:space:]'CDATA'
1647 * attrname[:space:]=[:space:]"CDATA"
1649 * attr[:space:]=[:space:]CDATA
1653 * We leave types 1 and 2 the same, type 3 we check for
1654 * '"' and convert to """ if needed, then wrap in
1655 * double quotes. Type 4 we convert into:
1658 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1659 if ($regary == false){
1661 * Looks like body ended before the end of tag.
1663 return Array(false, false, false, $lt, strlen($body));
1665 list($pos, $attname, $match) = $regary;
1666 $attname = strtolower($attname);
1668 * We arrived at the end of attribute name. Several things possible
1670 * '>' means the end of the tag and this is attribute type 4
1671 * '/' if followed by '>' means the same thing as above
1672 * '\s' means a lot of things -- look what it's followed by.
1673 * anything else means the attribute is invalid.
1678 * This is an xhtml-style tag with a closing / at the
1679 * end, like so: <img src="blah" />. Check if it's followed
1680 * by the closing bracket. If not, then this tag is invalid
1682 if (substr($body, $pos, 2) == "/>"){
1686 $gt = sq_findnxstr($body, $pos, ">");
1687 $retary = Array(false, false, false, $lt, $gt);
1691 $attary[$attname] = '"yes"';
1692 return Array($tagname, $attary, $tagtype, $lt, $pos);
1696 * Skip whitespace and see what we arrive at.
1698 $pos = sq_skipspace($body, $pos);
1699 $char = substr($body, $pos, 1);
1701 * Two things are valid here:
1702 * '=' means this is attribute type 1 2 or 3.
1703 * \w means this was attribute type 4.
1704 * anything else we ignore and re-loop. End of tag and
1705 * invalid stuff will be caught by our checks at the beginning
1710 $pos = sq_skipspace($body, $pos);
1712 * Here are 3 possibilities:
1713 * "'" attribute type 1
1714 * '"' attribute type 2
1715 * everything else is the content of tag type 3
1717 $quot = substr($body, $pos, 1);
1719 $regary = sq_findnxreg($body, $pos+
1, "\'");
1720 if ($regary == false){
1721 return Array(false, false, false, $lt, strlen($body));
1723 list($pos, $attval, $match) = $regary;
1725 $attary[$attname] = "'" . $attval . "'";
1726 } else if ($quot == '"'){
1727 $regary = sq_findnxreg($body, $pos+
1, '\"');
1728 if ($regary == false){
1729 return Array(false, false, false, $lt, strlen($body));
1731 list($pos, $attval, $match) = $regary;
1733 $attary[$attname] = '"' . $attval . '"';
1736 * These are hateful. Look for \s, or >.
1738 $regary = sq_findnxreg($body, $pos, "[\s>]");
1739 if ($regary == false){
1740 return Array(false, false, false, $lt, strlen($body));
1742 list($pos, $attval, $match) = $regary;
1744 * If it's ">" it will be caught at the top.
1746 $attval = preg_replace("/\"/s", """, $attval);
1747 $attary[$attname] = '"' . $attval . '"';
1749 } else if (preg_match("|[\w/>]|", $char)) {
1751 * That was attribute type 4.
1753 $attary[$attname] = '"yes"';
1756 * An illegal character. Find next '>' and return.
1758 $gt = sq_findnxstr($body, $pos, ">");
1759 return Array(false, false, false, $lt, $gt);
1765 * The fact that we got here indicates that the tag end was never
1766 * found. Return invalid tag indication so it gets stripped.
1768 return Array(false, false, false, $lt, strlen($body));
1772 * Translates entities into literal values so they can be checked.
1774 * @param $attvalue the by-ref value to check.
1775 * @param $regex the regular expression to check against.
1776 * @param $hex whether the entites are hexadecimal.
1777 * @return True or False depending on whether there were matches.
1779 function sq_deent(&$attvalue, $regex, $hex=false){
1783 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1784 preg_match_all($regex, $attvalue, $matches);
1785 if (is_array($matches) && sizeof($matches[0]) > 0){
1787 for ($i = 0; $i < sizeof($matches[0]); $i++
){
1788 $numval = $matches[1][$i];
1790 $numval = hexdec($numval);
1792 $repl[$matches[0][$i]] = chr($numval);
1794 $attvalue = strtr($attvalue, $repl);
1802 * This function runs various checks against the attributes.
1804 * @param $tagname String with the name of the tag.
1805 * @param $attary Array with all tag attributes.
1806 * @param $rm_attnames See description for sq_sanitize
1807 * @param $bad_attvals See description for sq_sanitize
1808 * @param $add_attr_to_tag See description for sq_sanitize
1809 * @param $message message object
1810 * @param $id message id
1811 * @return Array with modified attributes.
1813 function sq_fixatts($tagname,
1823 foreach ($attary as $attname => $attvalue){
1825 * See if this attribute should be removed.
1827 foreach ($rm_attnames as $matchtag=>$matchattrs){
1828 if (preg_match($matchtag, $tagname)){
1829 foreach ($matchattrs as $matchattr){
1830 if (preg_match($matchattr, $attname)){
1831 unset($attary[$attname]);
1838 * Workaround for IE quirks
1840 sq_fixIE_idiocy($attvalue);
1843 * Remove any backslashes, entities, and extraneous whitespace.
1846 $oldattvalue = $attvalue;
1847 sq_defang($attvalue);
1848 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1849 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1850 // i.e.<div style="{ left:expʀessioɴ( alert('XSS') ) }">
1851 $attvalue = "idiocy";
1852 $attary[$attname] = $attvalue;
1854 sq_unspace($attvalue);
1857 * Now let's run checks on the attvalues.
1858 * I don't expect anyone to comprehend this. If you do,
1859 * get in touch with me so I can drive to where you live and
1860 * shake your hand personally. :)
1862 foreach ($bad_attvals as $matchtag=>$matchattrs){
1863 if (preg_match($matchtag, $tagname)){
1864 foreach ($matchattrs as $matchattr=>$valary){
1865 if (preg_match($matchattr, $attname)){
1867 * There are two arrays in valary.
1869 * Second one is replacements
1871 list($valmatch, $valrepl) = $valary;
1873 preg_replace($valmatch, $valrepl, $attvalue);
1874 if ($newvalue != $attvalue){
1875 $attary[$attname] = $newvalue;
1876 $attvalue = $newvalue;
1882 if ($attname == 'style') {
1883 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1884 // 8bit and control characters in style attribute values can be used for XSS, remove them
1885 $attary[$attname] = '"disallowed character"';
1887 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1888 if (count($aMatch)) {
1889 foreach($aMatch[1] as $sMatch) {
1891 $urlvalue = $sMatch;
1892 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1893 $attary[$attname] = str_replace($sMatch,$urlvalue,$attvalue);
1898 * Use white list based filtering on attributes which can contain url's
1900 else if ($attname == 'href' ||
$attname == 'xlink:href' ||
$attname == 'src'
1901 ||
$attname == 'poster' ||
$attname == 'formaction'
1902 ||
$attname == 'background' ||
$attname == 'action') {
1903 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1904 $attary[$attname] = $attvalue;
1908 * See if we need to append any attributes to this tag.
1910 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1911 if (preg_match($matchtag, $tagname)){
1912 $attary = array_merge($attary, $addattary);
1919 * This function filters url's
1921 * @param $attvalue String with attribute value to filter
1922 * @param $message message object
1923 * @param $id message id
1924 * @param $mailbox mailbox
1925 * @param $sQuote quoting characters around url's
1927 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1928 $attvalue = trim($attvalue);
1929 if ($attvalue && ($attvalue[0] =='"'||
$attvalue[0] == "'")) {
1930 // remove the double quotes
1931 $sQuote = $attvalue[0];
1932 $attvalue = trim(substr($attvalue,1,-1));
1935 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1936 // images off by default.
1937 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
1939 global $use_transparent_security_image;
1940 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1941 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1944 * Replace empty src tags with the blank image. src is only used
1945 * for frames, images, and image inputs. Doing a replace should
1946 * not affect them working as should be, however it will stop
1947 * IE from being kicked off when src for img tags are not set
1949 if ($attvalue == '') {
1950 $attvalue = '"' . SM_PATH
. 'images/blank.png"';
1952 // first, disallow 8 bit characters and control characters
1953 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1956 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1959 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
1963 $aUrl = parse_url($attvalue);
1964 if (isset($aUrl['scheme'])) {
1965 switch(strtolower($aUrl['scheme'])) {
1970 if ($attname != 'href') {
1971 if ($view_unsafe_images == false) {
1972 $attvalue = $sQuote . $secremoveimg . $sQuote;
1974 if (isset($aUrl['path'])) {
1976 // No one has been able to show that image URIs
1977 // can be exploited, so for now, no restrictions
1978 // are made at all. If this proves to be a problem,
1979 // the commented-out code below can be of help.
1980 // (One consideration is that I see nothing in this
1981 // function that specifically says that we will
1982 // only ever arrive here when inspecting an image
1983 // tag, although that does seem to be the end
1984 // result - e.g., <script src="..."> where malicious
1985 // image URIs are in fact a problem are already
1986 // filtered out elsewhere.
1987 /* ---------------------------------
1988 // validate image extension.
1989 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1990 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1991 // If URI is to something other than
1992 // a regular image file, get the contents
1993 // and try to see if it is an image.
1994 // Don't use Fileinfo (finfo_file()) because
1995 // we'd need to make the admin configure the
1996 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1999 if (function_exists('mime_content_type')
2000 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
2004 $file_contents = '';
2005 while (!feof($FILE)) {
2006 $file_contents .= fread($FILE, 8192);
2010 // store file locally
2012 global $attachment_dir, $username;
2013 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
2014 $localfilename = GenerateRandomString(32, '', 7);
2015 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2016 while (file_exists($full_localfilename)) {
2017 $localfilename = GenerateRandomString(32, '', 7);
2018 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2020 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2021 fwrite($FILE, $file_contents);
2024 // get mime type and remove file
2026 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2027 unlink("$hashed_attachment_dir/$localfilename");
2029 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2030 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2031 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2034 --------------------------------- */
2036 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
2040 $attvalue = $sQuote . $attvalue . $sQuote;
2045 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2046 * One day MS might actually make it match something useful, for now, falling
2047 * back to using cid2http, so we can grab the blank.png.
2049 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2053 * Turn cid: urls into http-friendly ones.
2055 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2058 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2062 if (!isset($aUrl['path']) ||
$aUrl['path'] != $secremoveimg) {
2063 // parse_url did not lead to satisfying result
2064 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2072 * This function edits the style definition to make them friendly and
2073 * usable in SquirrelMail.
2075 * @param $message the message object
2076 * @param $id the message id
2077 * @param $content a string with whatever is between <style> and </style>
2078 * @param $mailbox the message mailbox
2079 * @return a string with edited content.
2081 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2082 $me = 'sq_fixstyle';
2083 // workaround for </style> in between comments
2084 $iCurrentPos = $pos;
2089 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++
$i) {
2096 if ($sToken == '<') {
2106 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2111 $content .= $sToken;
2119 if ($sToken == '<') {
2121 if (isset($body[$i+
2]) && substr($body,$i,3) == '!--') {
2122 $i = strpos($body,'-->',$i+
3);
2123 if ($i === false) { // no end comment
2141 if ($bSucces == FALSE){
2142 return array(FALSE, strlen($body));
2148 * First look for general BODY style declaration, which would be
2150 * body {background: blah-blah}
2151 * and change it to .bodyclass so we can just assign it to a <div>
2153 // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2154 // Nah, this is even better - try to preface all CSS selectors with
2155 // our <div> class ID "bodyclass" then correct generic "body" selectors
2156 // TODO: this works pretty good but breaks stuff like this:
2157 // @media print { body { font-size: 10pt; } }
2158 // but there isn't an easy way to make this regex skip @media
2159 // definitions... though lots of the ones in the wild will be
2160 // correctly handled because they tend to end with a parenthesis, like:
2161 // @media screen and (max-width:480px) { ...
2162 $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2163 $content = str_replace('.bodyclass body', '.bodyclass', $content);
2165 global $use_transparent_security_image;
2166 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2167 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2170 * Fix url('blah') declarations.
2172 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2173 // "url(\\1$secremoveimg\\2)", $content);
2175 // first check for 8bit sequences and disallowed control characters
2176 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2177 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2178 return array($content, $newpos);
2181 // IE Sucks hard. We have a special function for it.
2182 sq_fixIE_idiocy($content);
2184 // remove @import line
2185 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2187 // translate ur\l and variations (IE parses that)
2188 // TODO check if the sq_fixIE_idiocy function already handles this.
2189 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2190 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2191 if (count($aMatch)) {
2192 $aValue = $aReplace = array();
2193 foreach($aMatch[1] as $sMatch) {
2195 $urlvalue = $sMatch;
2196 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2197 $aValue[] = $sMatch;
2198 $aReplace[] = $urlvalue;
2200 $content = str_replace($aValue,$aReplace,$content);
2204 * Remove any backslashes, entities, and extraneous whitespace.
2206 $contentTemp = $content;
2207 sq_defang($contentTemp);
2208 sq_unspace($contentTemp);
2211 * Fix stupid css declarations which lead to vulnerabilities
2214 * Also remove "position" attribute, as it can easily be set
2215 * to "fixed" or "absolute" with "left" and "top" attributes
2216 * of zero, taking over the whole content frame. It can also
2217 * be set to relative and move itself anywhere it wants to,
2218 * displaying content in areas it shouldn't be allowed to touch.
2220 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2224 '/include-source/i',
2228 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2229 $contentNew = preg_replace($match, $replace, $contentTemp);
2230 if ($contentNew !== $contentTemp) {
2231 // insecure css declarations are used. From now on we don't care
2232 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2233 $content = $contentNew;
2235 return array($content, $newpos);
2240 * This function converts cid: url's into the ones that can be viewed in
2243 * @param $message the message object
2244 * @param $id the message id
2245 * @param $cidurl the cid: url.
2246 * @param $mailbox the message mailbox
2247 * @return a string with a http-friendly url
2249 function sq_cid2http($message, $id, $cidurl, $mailbox){
2251 * Get rid of quotes.
2253 $quotchar = substr($cidurl, 0, 1);
2254 if ($quotchar == '"' ||
$quotchar == "'"){
2255 $cidurl = str_replace($quotchar, "", $cidurl);
2259 $cidurl = substr(trim($cidurl), 4);
2261 $match_str = '/\{.*?\}\//';
2263 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2265 $linkurl = find_ent_id($cidurl, $message);
2266 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2267 unsafe link image */
2271 * This is part of a fix for Outlook Express 6.x generating
2272 * cid URLs without creating content-id headers. These images are
2273 * not part of the multipart/related html mail. The html contains
2274 * <img src="cid:{some_id}/image_filename.ext"> references to
2275 * attached images with as goal to render them inline although
2276 * the attachment disposition property is not inline.
2279 if (empty($linkurl)) {
2280 if (preg_match('/{.*}\//', $cidurl)) {
2281 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2282 if (!empty($cidurl)) {
2283 $linkurl = find_ent_id($cidurl, $message);
2288 if (!empty($linkurl)) {
2289 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&' .
2290 "passed_id=$id&mailbox=" . urlencode($mailbox) .
2291 '&ent_id=' . $linkurl . $quotchar;
2294 * If we couldn't generate a proper img url, drop in a blank image
2295 * instead of sending back empty, otherwise it causes unusual behaviour
2297 $httpurl = $quotchar . SM_PATH
. 'images/blank.png' . $quotchar;
2304 * This function changes the <body> tag into a <div> tag since we
2305 * can't really have a body-within-body.
2307 * @param $attary an array of attributes and values of <body>
2308 * @param $mailbox mailbox we're currently reading (for cid2http)
2309 * @param $message current message (for cid2http)
2310 * @param $id current message id (for cid2http)
2311 * @return a modified array of attributes to be set for <div>
2313 function sq_body2div($attary, $mailbox, $message, $id){
2314 $me = 'sq_body2div';
2315 $divattary = Array('class' => "'bodyclass'");
2317 $has_bgc_stl = $has_txt_stl = false;
2319 if (is_array($attary) && sizeof($attary) > 0){
2320 foreach ($attary as $attname=>$attvalue){
2321 $quotchar = substr($attvalue, 0, 1);
2322 $attvalue = str_replace($quotchar, "", $attvalue);
2325 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2326 $styledef .= "background-image: url('$attvalue'); ";
2329 $has_bgc_stl = true;
2330 $styledef .= "background-color: $attvalue; ";
2333 $has_txt_stl = true;
2334 $styledef .= "color: $attvalue; ";
2338 // Outlook defines a white bgcolor and no text color. This can lead to
2339 // white text on a white bg with certain themes.
2340 if ($has_bgc_stl && !$has_txt_stl) {
2341 $styledef .= "color: $text; ";
2343 if (strlen($styledef) > 0){
2344 $divattary["style"] = "\"$styledef\"";
2351 * This is the main function and the one you should actually be calling.
2352 * There are several variables you should be aware of an which need
2353 * special description.
2355 * Since the description is quite lengthy, see it here:
2356 * http://linux.duke.edu/projects/mini/htmlfilter/
2358 * @param $body the string with HTML you wish to filter
2359 * @param $tag_list see description above
2360 * @param $rm_tags_with_content see description above
2361 * @param $self_closing_tags see description above
2362 * @param $force_tag_closing see description above
2363 * @param $rm_attnames see description above
2364 * @param $bad_attvals see description above
2365 * @param $add_attr_to_tag see description above
2366 * @param $message message object
2367 * @param $id message id
2368 * @param $recursively_called boolean flag for recursive calls into this function (optional; default FALSE)
2369 * @return sanitized html safe to show on your pages.
2371 function sq_sanitize($body,
2373 $rm_tags_with_content,
2382 $recursively_called=FALSE
2384 $me = 'sq_sanitize';
2387 * See if tag_list is of tags to remove or tags to allow.
2388 * false means remove these tags
2389 * true means allow these tags
2391 $orig_tag_list = $tag_list;
2392 $rm_tags = array_shift($tag_list);
2395 * Normalize rm_tags and rm_tags_with_content.
2397 @array_walk
($tag_list, 'sq_casenormalize');
2398 @array_walk
($rm_tags_with_content, 'sq_casenormalize');
2399 @array_walk
($self_closing_tags, 'sq_casenormalize');
2402 $open_tags = Array();
2403 $trusted = "\n<!-- begin sanitized html -->\n";
2404 $skip_content = false;
2406 * Take care of netscape's stupid javascript entities like
2409 $body = preg_replace("/&(\{.*?\};)/si", "&\\1", $body);
2411 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2412 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2415 * RCDATA and RAWTEXT tags are handled differently:
2416 * next instance of closing tag is used, whether or not
2417 * the HTML is well formed before that
2419 global $rcdata_rawtext_tags;
2420 if (!$recursively_called
2421 && in_array($tagname, $rcdata_rawtext_tags)
2423 $closing_tag = false;
2424 $closing_tag_offset = $curpos;
2425 // seek out the closing tag for the current RCDATA/RAWTEXT tag
2427 // first we need to move forward to next available closing tag
2428 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2429 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2430 if ($next_tag === false) {
2431 $closing_tag = false;
2434 // but then we have to make sure it's a well-formed tag
2435 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2436 if ($closing_tag === false)
2438 else if ($closing_tag[0] !== false
2439 // these should be redundant
2440 && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2441 $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos +
1),
2442 $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2443 $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2444 $message, $id, $mailbox, true);
2445 $curpos = $closing_tag[4] +
1;
2448 $closing_tag_offset = $next_tag[0] +
1;
2450 if ($closing_tag === false)
2451 { /* 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 */ }
2454 $free_content = substr($body, $curpos, $lt-$curpos);
2456 * Take care of <style>
2458 if ($tagname == "style" && $tagtype == 1){
2459 list($free_content, $curpos) =
2460 sq_fixstyle($body, $gt+
1, $message, $id, $mailbox);
2461 if ($free_content != FALSE){
2462 if ( !empty($attary) ) {
2463 $attary = sq_fixatts($tagname,
2473 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2474 $trusted .= $free_content;
2475 $trusted .= sq_tagprint($tagname, false, 2);
2479 if ($skip_content == false){
2480 $trusted .= $free_content;
2482 if ($tagname != FALSE){
2484 if ($skip_content == $tagname){
2486 * Got to the end of tag we needed to remove.
2489 $skip_content = false;
2491 if ($skip_content == false){
2492 if ($tagname == "body"){
2495 if (isset($open_tags[$tagname]) &&
2496 $open_tags[$tagname] > 0){
2497 $open_tags[$tagname]--;
2505 * $rm_tags_with_content
2507 if ($skip_content == false){
2509 * See if this is a self-closing type and change
2510 * tagtype appropriately.
2513 && in_array($tagname, $self_closing_tags)){
2517 * See if we should skip this tag and any content
2520 if ($tagtype == 1 &&
2521 in_array($tagname, $rm_tags_with_content)){
2522 $skip_content = $tagname;
2524 if (($rm_tags == false
2525 && in_array($tagname, $tag_list)) ||
2526 ($rm_tags == true &&
2527 !in_array($tagname, $tag_list))){
2531 * Convert body into div.
2533 if ($tagname == "body"){
2535 $attary = sq_body2div($attary, $mailbox,
2539 if (isset($open_tags[$tagname])){
2540 $open_tags[$tagname]++
;
2542 $open_tags[$tagname]=1;
2546 * This is where we run other checks.
2548 if (is_array($attary) && sizeof($attary) > 0){
2549 $attary = sq_fixatts($tagname,
2563 if ($tagname != false && $skip_content == false){
2564 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2569 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2570 if ($force_tag_closing == true){
2571 foreach ($open_tags as $tagname=>$opentimes){
2572 while ($opentimes > 0){
2573 $trusted .= '</' . $tagname . '>';
2579 $trusted .= "<!-- end sanitized html -->\n";
2584 * This is a wrapper function to call html sanitizing routines.
2586 * @param $body the body of the message
2587 * @param $id the id of the message
2591 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2592 * into internal SM compose links
2593 * (optional; default = TRUE)
2594 * @return a string with html safe to display in the browser.
2596 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2598 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2600 global $attachment_common_show_images, $view_unsafe_images,
2601 $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2602 $remove_rcdata_rawtext_tags_and_content;
2604 $rcdata_rawtext_tags = array(
2609 // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2613 * Don't display attached images in HTML mode.
2617 $attachment_common_show_images = false;
2619 false, // remove these tags
2631 $rm_tags_with_content = Array(
2641 if (!$allow_svg_display)
2642 $rm_tags_with_content[] = 'svg';
2644 * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2645 * case that they are, but if you prefer to remove them and their contents entirely
2646 * (in most cases, should be a safe thing with minimal impact), you can add the
2647 * following to config/config_local.php
2648 * $remove_rcdata_rawtext_tags_and_content = TRUE;
2650 if ($remove_rcdata_rawtext_tags_and_content)
2651 $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2653 $self_closing_tags = Array(
2661 $force_tag_closing = true;
2663 $rm_attnames = Array(
2674 global $use_transparent_security_image;
2675 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2676 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2678 $bad_attvals = Array(
2681 "/^src|background/i" =>
2684 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2685 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2686 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2689 "\\1$secremoveimg\\2",
2690 "\\1$secremoveimg\\2",
2691 "\\1$secremoveimg\\2",
2694 "/^href|action/i" =>
2697 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2698 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2699 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2714 "/include-source/i",
2716 // position:relative can also be exploited
2717 // to put content outside of email body area
2718 // and position:fixed is similarly exploitable
2719 // as position:absolute, so we'll remove it
2722 // Does this screw up legitimate HTML messages?
2723 // If so, the only fix I see is to allow position
2724 // attributes (any values? I think we still have
2725 // to block static and fixed) only if $use_iframe
2726 // is enabled (1.5.0+)
2728 // was: "/position\s*:\s*absolute/i",
2732 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2733 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2734 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2735 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2736 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2755 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2756 // images off by default.
2757 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
2759 if (!$view_unsafe_images){
2761 * Remove any references to http/https if view_unsafe_images set
2764 array_push($bad_attvals['/.*/']['/^src|background/i'][0],
2765 '/^([\'\"])\s*https*:.*([\'\"])/si');
2766 array_push($bad_attvals['/.*/']['/^src|background/i'][1],
2767 "\\1$secremoveimg\\1");
2768 array_push($bad_attvals['/.*/']['/^style/i'][0],
2769 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2770 array_push($bad_attvals['/.*/']['/^style/i'][1],
2771 "url(\\1$secremoveimg\\1)");
2774 $add_attr_to_tag = Array(
2776 Array('target'=>'"_blank"',
2777 'title'=>'"'._("This external link will open in a new window").'"'
2780 $trusted = sq_sanitize($body,
2782 $rm_tags_with_content,
2792 if (strpos($trusted,$secremoveimg)){
2793 $has_unsafe_images = true;
2796 // we want to parse mailto's in HTML output, change to SM compose links
2797 // this is a modified version of code from url_parser.php... but Marc is
2798 // right: we need a better filtering implementation; adding this randomly
2799 // here is not a great solution
2801 if ($take_mailto_links) {
2802 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2803 global $MailTo_PReg_Match;
2804 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2805 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2806 // quote before mailto, and we'll also try to convert the non-links back into links
2807 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2808 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2809 foreach ($regs[0] as $i => $mailto_before) {
2810 $mailto_params = $regs[11][$i];
2812 // get rid of any leading quote we may have captured but don't care about
2814 $mailto_before = ltrim($mailto_before, '"\'');
2816 // get rid of any tailing quote since we have to add send_to to the end
2818 $mailto_before = rtrim($mailto_before, '"\'');
2819 $mailto_params = rtrim($mailto_params, '"\'');
2821 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2822 $to = 'to=' . $regs[2][$i];
2823 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2824 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2826 if ($mailto_params) //already some params, append to them
2827 $mailto_params .= '&' . $to;
2829 $mailto_params .= '?' . $to;
2833 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2835 // we'll already have target=_blank, no need to allow comp_in_new
2836 // here (which would be a lot more work anyway)
2838 global $compose_new_win;
2839 $temp_comp_in_new = $compose_new_win;
2840 $compose_new_win = 0;
2841 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2842 $compose_new_win = $temp_comp_in_new;
2844 // remove <a href=" and anything after the next quote (we only
2845 // need the uri, not the link HTML) in compose uri
2847 // but only do this if the original mailto was in a real anchor tag
2849 if (!empty($regs[1][$i])) {
2850 $comp_uri = substr($comp_uri, 9);
2851 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2853 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2862 * function SendDownloadHeaders - send file to the browser
2864 * Original Source: SM core src/download.php
2865 * moved here to make it available to other code, and separate
2866 * front end from back end functionality.
2868 * @param string $type0 first half of mime type
2869 * @param string $type1 second half of mime type
2870 * @param string $filename filename to tell the browser for downloaded file
2871 * @param boolean $force whether to force the download dialog to pop
2872 * @param optional integer $filesize send the Content-Header and length to the browser
2875 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2876 global $languages, $squirrelmail_language;
2877 $isIE = $isIE6plus = false;
2879 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER
);
2881 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2882 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2886 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2887 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2891 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2892 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2894 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2896 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace(' ', ' ', $filename));
2899 // A Pox on Microsoft and it's Internet Explorer!
2901 // IE has lots of bugs with file downloads.
2902 // It also has problems with SSL. Both of these cause problems
2903 // for us in this function.
2905 // See this article on Cache Control headers and SSL
2906 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2908 // The best thing you can do for IE is to upgrade to the latest
2910 //set all the Cache Control Headers for IE
2912 $filename=rawurlencode($filename);
2913 header ("Pragma: public");
2914 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2915 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2916 // header ("Cache-Control: post-check=0, pre-check=0", false);
2917 header ("Cache-Control: private");
2919 //set the inline header for IE, we'll add the attachment header later if we need it
2920 header ("Content-Disposition: inline; filename=$filename");
2924 // Try to show in browser window
2925 header ("Content-Disposition: inline; filename=\"$filename\"");
2926 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2928 // Try to pop up the "save as" box
2930 // IE makes this hard. It pops up 2 save boxes, or none.
2931 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2932 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2933 // But, according to Microsoft, it is "RFC compliant but doesn't
2934 // take into account some deviations that allowed within the
2935 // specification." Doesn't that mean RFC non-compliant?
2936 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2938 // all browsers need the application/octet-stream header for this
2939 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2941 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2942 // Do not have quotes around filename, but that applied to
2943 // "attachment"... does it apply to inline too?
2944 header ("Content-Disposition: attachment; filename=\"$filename\"");
2946 if ($isIE && !$isIE6plus) {
2947 // This combination seems to work mostly. IE 5.5 SP 1 has
2948 // known issues (see the Microsoft Knowledge Base)
2950 // This works for most types, but doesn't work with Word files
2951 header ("Content-Type: application/download; name=\"$filename\"");
2952 header ("Content-Type: application/force-download; name=\"$filename\"");
2953 // These are spares, just in case. :-)
2954 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2955 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2956 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2958 // This is to prevent IE for MIME sniffing and auto open a file in IE
2959 header ("Content-Type: application/force-download; name=\"$filename\"");
2961 // another application/octet-stream forces download for Netscape
2962 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2966 //send the content-length header if the calling function provides it
2967 if ($filesize > 0) {
2968 header("Content-Length: $filesize");
2971 } // end fn SendDownloadHeaders