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);
611 // Do not let a generic plugin change the default link if a more
612 // specialized one already did it...
613 if ($defaultlink != $defaultlink_orig) {
617 do_hook("attachment $type0/*", $temp);
618 /* The API for this hook has changed as of 1.5.2 so that all plugin
619 arguments are passed in an array instead of each their own plugin
620 argument, and arguments are passed by reference, so instead of
621 returning any changes, changes should simply be made to the original
622 arguments themselves. */
623 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
624 &$defaultlink, &$display_filename, &$where, &$what);
625 // Do not let a generic plugin change the default link if a more
626 // specialized one already did it...
627 if ($defaultlink != $defaultlink_orig) {
631 do_hook("attachment */*", $temp);
633 $this_attachment = array();
634 $this_attachment['Name'] = decodeHeader($display_filename);
635 $this_attachment['Description'] = $description;
636 $this_attachment['DefaultHREF'] = $defaultlink;
637 $this_attachment['DownloadHREF'] = $links['download link']['href'];
638 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ?
$links['attachment_common']['href'] : '';
640 // base64 encoded file sizes are misleading, so approximate real size
641 if (!empty($header->encoding
) && strtolower($header->encoding
) == 'base64')
642 $this_attachment['Size'] = $header->size
/ 4 * 3;
644 $this_attachment['Size'] = $header->size
;
646 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
647 $this_attachment['OtherLinks'] = array();
648 foreach ($links as $val) {
649 if ($val['text']==_("Download") ||
$val['text'] == _("View"))
651 if (empty($val['text']) && empty($val['extra']))
655 $temp['HREF'] = $val['href'];
656 $temp['Text'] = (empty($val['text']) ?
'' : $val['text']) . (empty($val['extra']) ?
'' : $val['extra']);
657 $this_attachment['OtherLinks'][] = $temp;
659 $attachments[] = $this_attachment;
668 * Displays attachment links and information
670 * Since 1.3.0 function is not included in formatBody() call.
672 * Since 1.0.2 uses attachment $type0/$type1 hook.
673 * Since 1.2.5 uses attachment $type0/* hook.
674 * Since 1.5.0 uses attachments_bottom hook.
675 * Since 1.5.2 uses templates and does *not* return a value.
677 * @param object $message SquirrelMail message object
678 * @param array $exclude_id message parts that are not attachments.
679 * @param string $mailbox mailbox name
680 * @param integer $id message id
682 function formatAttachments($message, $exclude_id, $mailbox, $id) {
685 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
687 $oTemplate->assign('attachments', $attach);
688 $oTemplate->display('read_attachments.tpl');
691 function sqimap_base64_decode(&$string) {
693 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
694 // fly decoding (to reduce memory usage) you have to check if the
695 // data has incomplete pairs
697 // Remove the noise in order to check if the 4 bytes pairs are complete
698 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
701 $iMod = strlen($string) %
4;
703 $sStringRem = substr($string,-$iMod);
704 // Check if $sStringRem contains padding characters
705 if (substr($sStringRem,-1) != '=') {
706 $string = substr($string,0,-$iMod);
711 $string = base64_decode($string);
716 * Decodes encoded string (usually message body)
718 * This function decodes a string (usually the message body)
719 * depending on the encoding type. Currently quoted-printable
720 * and base64 encodings are supported.
722 * The decode_body hook was added to this function in 1.4.2/1.5.0.
723 * The $force_crlf parameter was added in 1.5.2.
725 * @param string $string The encoded string
726 * @param string $encoding used encoding
727 * @param string $force_crlf Whether or not to force CRLF or LF
728 * line endings (or to leave as is).
729 * If given as "LF", line endings will
730 * all be converted to LF; if "CRLF",
731 * line endings will all be converted
732 * to CRLF. If given as an empty value,
733 * the global $force_crlf_default will
734 * be consulted (it can be specified in
735 * config/config_local.php). Otherwise,
736 * any other value will cause the string
737 * to be left alone. Note that this will
738 * be overridden to "LF" if not using at
739 * least PHP version 4.3.0. (OPTIONAL;
740 * default is empty - consult global
743 * @return string The decoded string
748 function decodeBody($string, $encoding, $force_crlf='') {
750 global $force_crlf_default;
751 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
752 $force_crlf = strtoupper($force_crlf);
754 // must force line endings to LF due to broken
755 // quoted_printable_decode() in PHP versions
756 // before 4.3.0 (see below)
758 if (!check_php_version(4, 3, 0) ||
$force_crlf == 'LF')
759 $string = str_replace("\r\n", "\n", $string);
760 else if ($force_crlf == 'CRLF')
761 $string = str_replace("\n", "\r\n", $string);
763 $encoding = strtolower($encoding);
765 $encoding_handler = do_hook('decode_body', $encoding);
768 // plugins get first shot at decoding the string
770 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
771 $string = $encoding_handler('decode', $string);
773 } elseif ($encoding == 'quoted-printable' ||
774 $encoding == 'quoted_printable') {
776 // quoted_printable_decode() function is broken in older
777 // php versions. Text with \r\n decoding was fixed only
778 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
779 // and the above call to: str_replace("\r\n", "\n", $string);
781 $string = quoted_printable_decode($string);
783 } elseif ($encoding == 'base64') {
784 $string = base64_decode($string);
787 // All other encodings are returned raw.
794 * This function decodes strings that are encoded according to
795 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
796 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
798 * @param string $string header string that has to be made readable
799 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
800 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
801 * @param boolean $decide decide if string can be utfencoded. defaults to false
802 * @return string decoded header string
804 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
805 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
806 if (is_array($string)) {
807 $string = implode("\n", $string);
810 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
811 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
812 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
813 // Do we need to return at this point?
820 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
821 $aString = explode(' ',$string);
823 foreach ($aString as $chunk) {
824 if ($encoded && $chunk === '') {
826 } elseif ($chunk === '') {
831 /* if encoded words are not separated by a linear-space-white we still catch them */
834 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
835 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
836 if ($iLastMatch !== $j) {
846 $ret .= sm_encode_html_special_chars($res[1]);
850 $encoding = ucfirst($res[3]);
852 /* decide about valid decoding */
853 if ($decide && is_conversion_safe($res[2])) {
855 $can_be_encoded=true;
857 $can_be_encoded=false;
862 // fix broken base64-encoded strings (remove end = padding,
863 // change any = to + in middle of string, add padding back
865 if ($fix_broken_base64_encoded_messages) {
866 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
867 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
869 $replace = base64_decode($res[4]);
871 if ($can_be_encoded) {
872 /* convert string to different charset,
873 * if functions asks for it (usually in compose)
875 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
877 // convert string to html codes in order to display it
878 $ret .= charset_decode($res[2],$replace);
882 $replace = sm_encode_html_special_chars($replace);
888 $replace = str_replace('_', ' ', $res[4]);
889 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
890 create_function ('$matches', 'return chr(hexdec($matches[1]));'),
893 if ($can_be_encoded) {
894 /* convert string to different charset,
895 * if functions asks for it (usually in compose)
897 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
899 // convert string to html codes in order to display it
900 $replace = charset_decode($res[2], $replace);
904 $replace = sm_encode_html_special_chars($replace);
923 if (!$encoded && $htmlsafe) {
924 $ret .= sm_encode_html_special_chars($chunk);
930 /* remove the first added space */
933 $ret = substr($ret,5);
935 $ret = substr($ret,1);
945 * Function uses XTRA_CODE _encodeheader function, if such function exists.
947 * Function uses Q encoding by default and encodes a string according to RFC
948 * 1522 for use in headers if it contains 8-bit characters or anything that
949 * looks like it should be encoded.
951 * Function switches to B encoding and encodeHeaderBase64() function, if
952 * string is 8bit and multibyte character set supported by mbstring extension
953 * is used. It can cause E_USER_NOTICE errors, if interface is used with
954 * multibyte character set unsupported by mbstring extension.
956 * @param string $string header string, that has to be encoded
957 * @return string quoted-printable encoded string
958 * @todo make $mb_charsets system wide constant
960 function encodeHeader ($string) {
961 global $default_charset, $languages, $squirrelmail_language;
963 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
964 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
965 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
968 // Use B encoding for multibyte charsets
969 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
970 if (in_array($default_charset,$mb_charsets) &&
971 in_array($default_charset,sq_mb_list_encodings()) &&
972 sq_is8bit($string)) {
973 return encodeHeaderBase64($string,$default_charset);
974 } elseif (in_array($default_charset,$mb_charsets) &&
975 sq_is8bit($string) &&
976 ! in_array($default_charset,sq_mb_list_encodings())) {
977 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
978 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
981 // Encode only if the string contains 8-bit characters or =?
982 $j = strlen($string);
983 $max_l = 75 - strlen($default_charset) - 7;
986 $iEncStart = $enc_init = false;
987 $cur_l = $iOffset = 0;
988 for($i = 0; $i < $j; ++
$i) {
998 if ($iEncStart === false) {
1002 if ($cur_l > ($max_l-2)) {
1003 /* if there is an stringpart that doesn't need encoding, add it */
1004 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1005 $aRet[] = "=?$default_charset?Q?$ret?=";
1011 $ret .= sprintf("=%02X",ord($string{$i}));
1016 if ($iEncStart !== false) {
1017 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1018 $aRet[] = "=?$default_charset?Q?$ret?=";
1026 if ($iEncStart !== false) {
1028 if ($cur_l > $max_l) {
1029 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1030 $aRet[] = "=?$default_charset?Q?$ret?=";
1041 $k = ord($string{$i});
1043 if ($iEncStart === false) {
1044 // do not start encoding in the middle of a string, also take the rest of the word.
1045 $sLeadString = substr($string,0,$i);
1046 $aLeadString = explode(' ',$sLeadString);
1047 $sToBeEncoded = array_pop($aLeadString);
1048 $iEncStart = $i - strlen($sToBeEncoded);
1049 $ret .= $sToBeEncoded;
1050 $cur_l +
= strlen($sToBeEncoded);
1053 /* first we add the encoded string that reached it's max size */
1054 if ($cur_l > ($max_l-2)) {
1055 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1056 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1063 $ret .= sprintf("=%02X", $k);
1065 if ($iEncStart !== false) {
1067 if ($cur_l > $max_l) {
1068 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1069 $aRet[] = "=?$default_charset?Q?$ret?=";
1075 $ret .= $string{$i};
1084 if ($iEncStart !== false) {
1085 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1086 $aRet[] = "=?$default_charset?Q?$ret?=";
1088 $aRet[] = substr($string,$iOffset);
1090 $string = implode('',$aRet);
1096 * Encodes string according to rfc2047 B encoding header formating rules
1098 * It is recommended way to encode headers with character sets that store
1099 * symbols in more than one byte.
1101 * Function requires mbstring support. If required mbstring functions are missing,
1102 * function returns false and sets E_USER_WARNING level error message.
1104 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1105 * that mbstring functions will generate E_WARNING errors, if unsupported
1106 * character set is used. mb_encode_mimeheader function provided by php
1107 * mbstring extension is not used in order to get better control of header
1110 * Used php code functions - function_exists(), trigger_error(), strlen()
1111 * (is used with charset names and base64 strings). Used php mbstring
1112 * functions - mb_strlen and mb_substr.
1114 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1115 * encoding), rfc 2822 (header folding)
1117 * @param string $string header string that must be encoded
1118 * @param string $charset character set. Must be supported by mbstring extension.
1119 * Use sq_mb_list_encodings() to detect supported charsets.
1120 * @return string string encoded according to rfc2047 B encoding formating rules
1122 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1123 * @todo Do we want to control max length of header?
1124 * @todo Do we want to control EOL (end-of-line) marker?
1125 * @todo Do we want to translate error message?
1127 function encodeHeaderBase64($string,$charset) {
1129 * Check mbstring function requirements.
1131 if (! function_exists('mb_strlen') ||
1132 ! function_exists('mb_substr')) {
1133 // set E_USER_WARNING
1134 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING
);
1139 // initial return array
1143 * header length = 75 symbols max (same as in encodeHeader)
1144 * remove $charset length
1145 * remove =? ? ?= (5 chars)
1146 * remove 2 more chars (\r\n ?)
1148 $iMaxLength = 75 - strlen($charset) - 7;
1150 // set first character position
1153 // loop through all characters. count characters and not bytes.
1154 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++
) {
1155 // encode string from starting character to current character.
1156 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1158 // Check encoded string length
1159 if(strlen($encoded_string)>$iMaxLength) {
1160 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1161 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1163 // set new starting character
1164 $iStartCharNum = $iCharNum-1;
1166 // encode last char (in case it is last character in string)
1167 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1168 } // if string is shorter than max length - add next character
1171 // add last encoded string to array
1172 $aRet[] = $encoded_string;
1174 // set initial return string
1177 // loop through encoded strings
1178 foreach($aRet as $string) {
1179 // TODO: Do we want to control EOL (end-of-line) marker
1180 if ($sRet!='') $sRet.= " ";
1182 // add header tags and encoded string to return string
1183 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1189 /* This function trys to locate the entity_id of a specific mime element */
1190 function find_ent_id($id, $message) {
1191 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities
); $i++
) {
1192 if ($message->entities
[$i]->header
->type0
== 'multipart') {
1193 $ret = find_ent_id($id, $message->entities
[$i]);
1195 if (strcasecmp($message->entities
[$i]->header
->id
, $id) == 0) {
1196 // if (sq_check_save_extension($message->entities[$i])) {
1197 return $message->entities
[$i]->entity_id
;
1199 } elseif (!empty($message->entities
[$i]->header
->parameters
['name'])) {
1201 * This is part of a fix for Outlook Express 6.x generating
1202 * cid URLs without creating content-id headers
1205 if (strcasecmp($message->entities
[$i]->header
->parameters
['name'], $id) == 0) {
1206 return $message->entities
[$i]->entity_id
;
1214 function sq_check_save_extension($message) {
1215 $filename = $message->getFilename();
1216 $ext = substr($filename, strrpos($filename,'.')+
1);
1217 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1218 return in_array($ext, $save_extensions);
1223 ** HTMLFILTER ROUTINES
1227 * This function checks attribute values for entity-encoded values
1228 * and returns them translated into 8-bit strings so we can run
1231 * @param $attvalue A string to run entity check against.
1232 * @return Nothing, modifies a reference value.
1234 function sq_defang(&$attvalue){
1237 * Skip this if there aren't ampersands or backslashes.
1239 if (strpos($attvalue, '&') === false
1240 && strpos($attvalue, '\\') === false){
1244 // before deent, translate the dangerous unicode characters and ... to safe values
1245 // otherwise the regular expressions do not match.
1251 $m = $m ||
sq_deent($attvalue, '/\�*(\d+);*/s');
1252 $m = $m ||
sq_deent($attvalue, '/\�*((\d|[a-f])+);*/si', true);
1253 $m = $m ||
sq_deent($attvalue, '/\\\\(\d+)/s', true);
1254 } while ($m == true);
1255 $attvalue = stripslashes($attvalue);
1259 * Kill any tabs, newlines, or carriage returns. Our friends the
1260 * makers of the browser with 95% market value decided that it'd
1261 * be funny to make "java[tab]script" be just as good as "javascript".
1263 * @param attvalue The attribute value before extraneous spaces removed.
1264 * @return attvalue Nothing, modifies a reference value.
1266 function sq_unspace(&$attvalue){
1268 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1269 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1270 Array('', '', '', '', ''), $attvalue);
1275 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1276 * IE as regular characters.
1278 * @param attvalue The attribute value before dangerous characters are translated.
1279 * @return attvalue Nothing, modifies a reference value.
1280 * @author Marc Groot Koerkamp.
1282 function sq_fixIE_idiocy(&$attvalue) {
1284 $attvalue = str_replace("\0", "", $attvalue);
1286 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1288 // IE has the evil habit of accepting every possible value for the attribute expression.
1289 // The table below contains characters which are parsed by IE if they are used in the "expression"
1291 $aDangerousCharsReplacementTable = array(
1292 array('ʟ', 'ʟ' ,/* L UNICODE IPA Extension */
1293 'ʀ', 'ʀ' ,/* R UNICODE IPA Extension */
1294 'ɴ', 'ɴ' ,/* N UNICODE IPA Extension */
1295 'E', 'E' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1296 'e', 'e' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1297 'X', 'X',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1298 'x', 'x',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1299 'P', 'P',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1300 'p', 'p',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1301 'R', 'R',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1302 'r', 'r',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1303 'S', 'S',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1304 's', 's',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1305 'I', 'I',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1306 'i', 'i',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1307 'O', 'O',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1308 'o', 'o',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1309 'N', 'N',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1310 'n', 'n',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1311 'L', 'L',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1312 'l', 'l',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1313 'U', 'U',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1314 'u', 'u',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1315 'ⁿ', 'ⁿ' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1316 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1317 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1318 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1319 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1320 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1321 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1322 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1323 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1324 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1325 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1326 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1327 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1328 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1329 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1330 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1331 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1332 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1333 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1334 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1335 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1336 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1337 "\xCA\x9F", /* L UNICODE IPA Extension */
1338 "\xCA\x80", /* R UNICODE IPA Extension */
1339 "\xC9\xB4"), /* N UNICODE IPA Extension */
1340 array('l', 'l', 'r','r','n','n',
1341 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1342 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1343 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1344 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1346 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1348 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1352 * This function returns the final tag out of the tag name, an array
1353 * of attributes, and the type of the tag. This function is called by
1354 * sq_sanitize internally.
1356 * @param $tagname the name of the tag.
1357 * @param $attary the array of attributes and their values
1358 * @param $tagtype The type of the tag (see in comments).
1359 * @return a string with the final tag representation.
1361 function sq_tagprint($tagname, $attary, $tagtype){
1362 $me = 'sq_tagprint';
1365 $fulltag = '</' . $tagname . '>';
1367 $fulltag = '<' . $tagname;
1368 if (is_array($attary) && sizeof($attary)){
1370 while (list($attname, $attvalue) = each($attary)){
1371 array_push($atts, "$attname=$attvalue");
1373 $fulltag .= ' ' . join(" ", $atts);
1384 * A small helper function to use with array_walk. Modifies a by-ref
1385 * value and makes it lowercase.
1387 * @param $val a value passed by-ref.
1388 * @return void since it modifies a by-ref value.
1390 function sq_casenormalize(&$val){
1391 $val = strtolower($val);
1395 * This function skips any whitespace from the current position within
1396 * a string and to the next non-whitespace value.
1398 * @param $body the string
1399 * @param $offset the offset within the string where we should start
1400 * looking for the next non-whitespace character.
1401 * @return the location within the $body where the next
1402 * non-whitespace char is located.
1404 function sq_skipspace($body, $offset){
1405 $me = 'sq_skipspace';
1406 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1407 if (!empty($matches[1])){
1408 $offset +
= strlen($matches[1]);
1414 * This function looks for the next character within a string. It's
1415 * really just a glorified "strpos", except it catches if failures
1418 * @param $body The string to look for needle in.
1419 * @param $offset Start looking from this position.
1420 * @param $needle The character/string to look for.
1421 * @return location of the next occurance of the needle, or
1422 * strlen($body) if needle wasn't found.
1424 function sq_findnxstr($body, $offset, $needle){
1425 $me = 'sq_findnxstr';
1426 $pos = strpos($body, $needle, $offset);
1427 if ($pos === FALSE){
1428 $pos = strlen($body);
1434 * This function takes a PCRE-style regexp and tries to match it
1435 * within the string.
1437 * @param $body The string to look for needle in.
1438 * @param $offset Start looking from here.
1439 * @param $reg A PCRE-style regex to match.
1440 * @return Returns a false if no matches found, or an array
1441 * with the following members:
1442 * - integer with the location of the match within $body
1443 * - string with whatever content between offset and the match
1444 * - string with whatever it is we matched
1446 function sq_findnxreg($body, $offset, $reg){
1447 $me = 'sq_findnxreg';
1450 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1451 if (!isset($matches{0}) ||
!$matches{0}){
1454 $retarr{0} = $offset +
strlen($matches{1});
1455 $retarr{1} = $matches{1};
1456 $retarr{2} = $matches{2};
1462 * This function looks for the next tag.
1464 * @param $body String where to look for the next tag.
1465 * @param $offset Start looking from here.
1466 * @return false if no more tags exist in the body, or
1467 * an array with the following members:
1468 * - string with the name of the tag
1469 * - array with attributes and their values
1470 * - integer with tag type (1, 2, or 3)
1471 * - integer where the tag starts (starting "<")
1472 * - integer where the tag ends (ending ">")
1473 * first three members will be false, if the tag is invalid.
1475 function sq_getnxtag($body, $offset){
1476 $me = 'sq_getnxtag';
1477 if ($offset > strlen($body)){
1480 $lt = sq_findnxstr($body, $offset, "<");
1481 if ($lt == strlen($body)){
1486 * blah blah <tag attribute="value">
1489 $pos = sq_skipspace($body, $lt+
1);
1490 if ($pos >= strlen($body)){
1491 return Array(false, false, false, $lt, strlen($body));
1494 * There are 3 kinds of tags:
1495 * 1. Opening tag, e.g.:
1497 * 2. Closing tag, e.g.:
1499 * 3. XHTML-style content-less tag, e.g.:
1500 * <img src="blah" />
1503 switch (substr($body, $pos, 1)){
1510 * A comment or an SGML declaration.
1512 if (substr($body, $pos+
1, 2) == "--"){
1513 $gt = strpos($body, "-->", $pos);
1515 $gt = strlen($body);
1519 return Array(false, false, false, $lt, $gt);
1521 $gt = sq_findnxstr($body, $pos, ">");
1522 return Array(false, false, false, $lt, $gt);
1527 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1537 * Look for next [\W-_], which will indicate the end of the tag name.
1539 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1540 if ($regary == false){
1541 return Array(false, false, false, $lt, strlen($body));
1543 list($pos, $tagname, $match) = $regary;
1544 $tagname = strtolower($tagname);
1547 * $match can be either of these:
1548 * '>' indicating the end of the tag entirely.
1549 * '\s' indicating the end of the tag name.
1550 * '/' indicating that this is type-3 xhtml tag.
1552 * Whatever else we find there indicates an invalid tag.
1557 * This is an xhtml-style tag with a closing / at the
1558 * end, like so: <img src="blah" />. Check if it's followed
1559 * by the closing bracket. If not, then this tag is invalid
1561 if (substr($body, $pos, 2) == "/>"){
1565 $gt = sq_findnxstr($body, $pos, ">");
1566 $retary = Array(false, false, false, $lt, $gt);
1570 return Array($tagname, false, $tagtype, $lt, $pos);
1574 * Check if it's whitespace
1576 if (!preg_match('/\s/', $match)){
1578 * This is an invalid tag! Look for the next closing ">".
1580 $gt = sq_findnxstr($body, $lt, ">");
1581 return Array(false, false, false, $lt, $gt);
1587 * At this point we're here:
1588 * <tagname attribute='blah'>
1591 * At this point we loop in order to find all attributes.
1597 while ($pos <= strlen($body)){
1598 $pos = sq_skipspace($body, $pos);
1599 if ($pos == strlen($body)){
1603 return Array(false, false, false, $lt, $pos);
1606 * See if we arrived at a ">" or "/>", which means that we reached
1607 * the end of the tag.
1610 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1614 $pos +
= strlen($matches{1});
1615 if ($matches{2} == "/>"){
1619 return Array($tagname, $attary, $tagtype, $lt, $pos);
1623 * There are several types of attributes, with optional
1624 * [:space:] between members.
1626 * attrname[:space:]=[:space:]'CDATA'
1628 * attrname[:space:]=[:space:]"CDATA"
1630 * attr[:space:]=[:space:]CDATA
1634 * We leave types 1 and 2 the same, type 3 we check for
1635 * '"' and convert to """ if needed, then wrap in
1636 * double quotes. Type 4 we convert into:
1639 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1640 if ($regary == false){
1642 * Looks like body ended before the end of tag.
1644 return Array(false, false, false, $lt, strlen($body));
1646 list($pos, $attname, $match) = $regary;
1647 $attname = strtolower($attname);
1649 * We arrived at the end of attribute name. Several things possible
1651 * '>' means the end of the tag and this is attribute type 4
1652 * '/' if followed by '>' means the same thing as above
1653 * '\s' means a lot of things -- look what it's followed by.
1654 * anything else means the attribute is invalid.
1659 * This is an xhtml-style tag with a closing / at the
1660 * end, like so: <img src="blah" />. Check if it's followed
1661 * by the closing bracket. If not, then this tag is invalid
1663 if (substr($body, $pos, 2) == "/>"){
1667 $gt = sq_findnxstr($body, $pos, ">");
1668 $retary = Array(false, false, false, $lt, $gt);
1672 $attary{$attname} = '"yes"';
1673 return Array($tagname, $attary, $tagtype, $lt, $pos);
1677 * Skip whitespace and see what we arrive at.
1679 $pos = sq_skipspace($body, $pos);
1680 $char = substr($body, $pos, 1);
1682 * Two things are valid here:
1683 * '=' means this is attribute type 1 2 or 3.
1684 * \w means this was attribute type 4.
1685 * anything else we ignore and re-loop. End of tag and
1686 * invalid stuff will be caught by our checks at the beginning
1691 $pos = sq_skipspace($body, $pos);
1693 * Here are 3 possibilities:
1694 * "'" attribute type 1
1695 * '"' attribute type 2
1696 * everything else is the content of tag type 3
1698 $quot = substr($body, $pos, 1);
1700 $regary = sq_findnxreg($body, $pos+
1, "\'");
1701 if ($regary == false){
1702 return Array(false, false, false, $lt, strlen($body));
1704 list($pos, $attval, $match) = $regary;
1706 $attary{$attname} = "'" . $attval . "'";
1707 } else if ($quot == '"'){
1708 $regary = sq_findnxreg($body, $pos+
1, '\"');
1709 if ($regary == false){
1710 return Array(false, false, false, $lt, strlen($body));
1712 list($pos, $attval, $match) = $regary;
1714 $attary{$attname} = '"' . $attval . '"';
1717 * These are hateful. Look for \s, or >.
1719 $regary = sq_findnxreg($body, $pos, "[\s>]");
1720 if ($regary == false){
1721 return Array(false, false, false, $lt, strlen($body));
1723 list($pos, $attval, $match) = $regary;
1725 * If it's ">" it will be caught at the top.
1727 $attval = preg_replace("/\"/s", """, $attval);
1728 $attary{$attname} = '"' . $attval . '"';
1730 } else if (preg_match("|[\w/>]|", $char)) {
1732 * That was attribute type 4.
1734 $attary{$attname} = '"yes"';
1737 * An illegal character. Find next '>' and return.
1739 $gt = sq_findnxstr($body, $pos, ">");
1740 return Array(false, false, false, $lt, $gt);
1746 * The fact that we got here indicates that the tag end was never
1747 * found. Return invalid tag indication so it gets stripped.
1749 return Array(false, false, false, $lt, strlen($body));
1753 * Translates entities into literal values so they can be checked.
1755 * @param $attvalue the by-ref value to check.
1756 * @param $regex the regular expression to check against.
1757 * @param $hex whether the entites are hexadecimal.
1758 * @return True or False depending on whether there were matches.
1760 function sq_deent(&$attvalue, $regex, $hex=false){
1764 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1765 preg_match_all($regex, $attvalue, $matches);
1766 if (is_array($matches) && sizeof($matches[0]) > 0){
1768 for ($i = 0; $i < sizeof($matches[0]); $i++
){
1769 $numval = $matches[1][$i];
1771 $numval = hexdec($numval);
1773 $repl{$matches[0][$i]} = chr($numval);
1775 $attvalue = strtr($attvalue, $repl);
1783 * This function runs various checks against the attributes.
1785 * @param $tagname String with the name of the tag.
1786 * @param $attary Array with all tag attributes.
1787 * @param $rm_attnames See description for sq_sanitize
1788 * @param $bad_attvals See description for sq_sanitize
1789 * @param $add_attr_to_tag See description for sq_sanitize
1790 * @param $message message object
1791 * @param $id message id
1792 * @return Array with modified attributes.
1794 function sq_fixatts($tagname,
1804 while (list($attname, $attvalue) = each($attary)){
1806 * See if this attribute should be removed.
1808 foreach ($rm_attnames as $matchtag=>$matchattrs){
1809 if (preg_match($matchtag, $tagname)){
1810 foreach ($matchattrs as $matchattr){
1811 if (preg_match($matchattr, $attname)){
1812 unset($attary{$attname});
1819 * Workaround for IE quirks
1821 sq_fixIE_idiocy($attvalue);
1824 * Remove any backslashes, entities, and extraneous whitespace.
1827 $oldattvalue = $attvalue;
1828 sq_defang($attvalue);
1829 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1830 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1831 // i.e.<div style="{ left:expʀessioɴ( alert('XSS') ) }">
1832 $attvalue = "idiocy";
1833 $attary{$attname} = $attvalue;
1835 sq_unspace($attvalue);
1838 * Now let's run checks on the attvalues.
1839 * I don't expect anyone to comprehend this. If you do,
1840 * get in touch with me so I can drive to where you live and
1841 * shake your hand personally. :)
1843 foreach ($bad_attvals as $matchtag=>$matchattrs){
1844 if (preg_match($matchtag, $tagname)){
1845 foreach ($matchattrs as $matchattr=>$valary){
1846 if (preg_match($matchattr, $attname)){
1848 * There are two arrays in valary.
1850 * Second one is replacements
1852 list($valmatch, $valrepl) = $valary;
1854 preg_replace($valmatch, $valrepl, $attvalue);
1855 if ($newvalue != $attvalue){
1856 $attary{$attname} = $newvalue;
1857 $attvalue = $newvalue;
1863 if ($attname == 'style') {
1864 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1865 // 8bit and control characters in style attribute values can be used for XSS, remove them
1866 $attary{$attname} = '"disallowed character"';
1868 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1869 if (count($aMatch)) {
1870 foreach($aMatch[1] as $sMatch) {
1872 $urlvalue = $sMatch;
1873 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1874 $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1879 * Use white list based filtering on attributes which can contain url's
1881 else if ($attname == 'href' ||
$attname == 'xlink:href' ||
$attname == 'src'
1882 ||
$attname == 'poster' ||
$attname == 'formaction'
1883 ||
$attname == 'background' ||
$attname == 'action') {
1884 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1885 $attary{$attname} = $attvalue;
1889 * See if we need to append any attributes to this tag.
1891 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1892 if (preg_match($matchtag, $tagname)){
1893 $attary = array_merge($attary, $addattary);
1900 * This function filters url's
1902 * @param $attvalue String with attribute value to filter
1903 * @param $message message object
1904 * @param $id message id
1905 * @param $mailbox mailbox
1906 * @param $sQuote quoting characters around url's
1908 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1909 $attvalue = trim($attvalue);
1910 if ($attvalue && ($attvalue[0] =='"'||
$attvalue[0] == "'")) {
1911 // remove the double quotes
1912 $sQuote = $attvalue[0];
1913 $attvalue = trim(substr($attvalue,1,-1));
1916 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1917 // images off by default.
1918 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
1920 global $use_transparent_security_image;
1921 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1922 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1925 * Replace empty src tags with the blank image. src is only used
1926 * for frames, images, and image inputs. Doing a replace should
1927 * not affect them working as should be, however it will stop
1928 * IE from being kicked off when src for img tags are not set
1930 if ($attvalue == '') {
1931 $attvalue = '"' . SM_PATH
. 'images/blank.png"';
1933 // first, disallow 8 bit characters and control characters
1934 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1937 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1940 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
1944 $aUrl = parse_url($attvalue);
1945 if (isset($aUrl['scheme'])) {
1946 switch(strtolower($aUrl['scheme'])) {
1951 if ($attname != 'href') {
1952 if ($view_unsafe_images == false) {
1953 $attvalue = $sQuote . $secremoveimg . $sQuote;
1955 if (isset($aUrl['path'])) {
1957 // No one has been able to show that image URIs
1958 // can be exploited, so for now, no restrictions
1959 // are made at all. If this proves to be a problem,
1960 // the commented-out code below can be of help.
1961 // (One consideration is that I see nothing in this
1962 // function that specifically says that we will
1963 // only ever arrive here when inspecting an image
1964 // tag, although that does seem to be the end
1965 // result - e.g., <script src="..."> where malicious
1966 // image URIs are in fact a problem are already
1967 // filtered out elsewhere.
1968 /* ---------------------------------
1969 // validate image extension.
1970 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1971 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1972 // If URI is to something other than
1973 // a regular image file, get the contents
1974 // and try to see if it is an image.
1975 // Don't use Fileinfo (finfo_file()) because
1976 // we'd need to make the admin configure the
1977 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1980 if (function_exists('mime_content_type')
1981 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1985 $file_contents = '';
1986 while (!feof($FILE)) {
1987 $file_contents .= fread($FILE, 8192);
1991 // store file locally
1993 global $attachment_dir, $username;
1994 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1995 $localfilename = GenerateRandomString(32, '', 7);
1996 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1997 while (file_exists($full_localfilename)) {
1998 $localfilename = GenerateRandomString(32, '', 7);
1999 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2001 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2002 fwrite($FILE, $file_contents);
2005 // get mime type and remove file
2007 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2008 unlink("$hashed_attachment_dir/$localfilename");
2010 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2011 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2012 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2015 --------------------------------- */
2017 $attvalue = $sQuote . SM_PATH
. 'images/blank.png'. $sQuote;
2021 $attvalue = $sQuote . $attvalue . $sQuote;
2026 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2027 * One day MS might actually make it match something useful, for now, falling
2028 * back to using cid2http, so we can grab the blank.png.
2030 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2034 * Turn cid: urls into http-friendly ones.
2036 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2039 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2043 if (!isset($aUrl['path']) ||
$aUrl['path'] != $secremoveimg) {
2044 // parse_url did not lead to satisfying result
2045 $attvalue = $sQuote . SM_PATH
. 'images/blank.png' . $sQuote;
2053 * This function edits the style definition to make them friendly and
2054 * usable in SquirrelMail.
2056 * @param $message the message object
2057 * @param $id the message id
2058 * @param $content a string with whatever is between <style> and </style>
2059 * @param $mailbox the message mailbox
2060 * @return a string with edited content.
2062 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2063 $me = 'sq_fixstyle';
2064 // workaround for </style> in between comments
2065 $iCurrentPos = $pos;
2070 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++
$i) {
2077 if ($sToken == '<') {
2087 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2092 $content .= $sToken;
2100 if ($sToken == '<') {
2102 if (isset($body{$i+
2}) && substr($body,$i,3) == '!--') {
2103 $i = strpos($body,'-->',$i+
3);
2104 if ($i === false) { // no end comment
2122 if ($bSucces == FALSE){
2123 return array(FALSE, strlen($body));
2129 * First look for general BODY style declaration, which would be
2131 * body {background: blah-blah}
2132 * and change it to .bodyclass so we can just assign it to a <div>
2134 // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2135 // Nah, this is even better - try to preface all CSS selectors with
2136 // our <div> class ID "bodyclass" then correct generic "body" selectors
2137 // TODO: this works pretty good but breaks stuff like this:
2138 // @media print { body { font-size: 10pt; } }
2139 // but there isn't an easy way to make this regex skip @media
2140 // definitions... though lots of the ones in the wild will be
2141 // correctly handled because they tend to end with a parenthesis, like:
2142 // @media screen and (max-width:480px) { ...
2143 $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2144 $content = str_replace('.bodyclass body', '.bodyclass', $content);
2146 global $use_transparent_security_image;
2147 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2148 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2151 * Fix url('blah') declarations.
2153 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2154 // "url(\\1$secremoveimg\\2)", $content);
2156 // first check for 8bit sequences and disallowed control characters
2157 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2158 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2159 return array($content, $newpos);
2162 // IE Sucks hard. We have a special function for it.
2163 sq_fixIE_idiocy($content);
2165 // remove @import line
2166 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2168 // translate ur\l and variations (IE parses that)
2169 // TODO check if the sq_fixIE_idiocy function already handles this.
2170 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2171 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2172 if (count($aMatch)) {
2173 $aValue = $aReplace = array();
2174 foreach($aMatch[1] as $sMatch) {
2176 $urlvalue = $sMatch;
2177 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2178 $aValue[] = $sMatch;
2179 $aReplace[] = $urlvalue;
2181 $content = str_replace($aValue,$aReplace,$content);
2185 * Remove any backslashes, entities, and extraneous whitespace.
2187 $contentTemp = $content;
2188 sq_defang($contentTemp);
2189 sq_unspace($contentTemp);
2192 * Fix stupid css declarations which lead to vulnerabilities
2195 * Also remove "position" attribute, as it can easily be set
2196 * to "fixed" or "absolute" with "left" and "top" attributes
2197 * of zero, taking over the whole content frame. It can also
2198 * be set to relative and move itself anywhere it wants to,
2199 * displaying content in areas it shouldn't be allowed to touch.
2201 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2205 '/include-source/i',
2209 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2210 $contentNew = preg_replace($match, $replace, $contentTemp);
2211 if ($contentNew !== $contentTemp) {
2212 // insecure css declarations are used. From now on we don't care
2213 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2214 $content = $contentNew;
2216 return array($content, $newpos);
2221 * This function converts cid: url's into the ones that can be viewed in
2224 * @param $message the message object
2225 * @param $id the message id
2226 * @param $cidurl the cid: url.
2227 * @param $mailbox the message mailbox
2228 * @return a string with a http-friendly url
2230 function sq_cid2http($message, $id, $cidurl, $mailbox){
2232 * Get rid of quotes.
2234 $quotchar = substr($cidurl, 0, 1);
2235 if ($quotchar == '"' ||
$quotchar == "'"){
2236 $cidurl = str_replace($quotchar, "", $cidurl);
2240 $cidurl = substr(trim($cidurl), 4);
2242 $match_str = '/\{.*?\}\//';
2244 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2246 $linkurl = find_ent_id($cidurl, $message);
2247 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2248 unsafe link image */
2252 * This is part of a fix for Outlook Express 6.x generating
2253 * cid URLs without creating content-id headers. These images are
2254 * not part of the multipart/related html mail. The html contains
2255 * <img src="cid:{some_id}/image_filename.ext"> references to
2256 * attached images with as goal to render them inline although
2257 * the attachment disposition property is not inline.
2260 if (empty($linkurl)) {
2261 if (preg_match('/{.*}\//', $cidurl)) {
2262 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2263 if (!empty($cidurl)) {
2264 $linkurl = find_ent_id($cidurl, $message);
2269 if (!empty($linkurl)) {
2270 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&' .
2271 "passed_id=$id&mailbox=" . urlencode($mailbox) .
2272 '&ent_id=' . $linkurl . $quotchar;
2275 * If we couldn't generate a proper img url, drop in a blank image
2276 * instead of sending back empty, otherwise it causes unusual behaviour
2278 $httpurl = $quotchar . SM_PATH
. 'images/blank.png' . $quotchar;
2285 * This function changes the <body> tag into a <div> tag since we
2286 * can't really have a body-within-body.
2288 * @param $attary an array of attributes and values of <body>
2289 * @param $mailbox mailbox we're currently reading (for cid2http)
2290 * @param $message current message (for cid2http)
2291 * @param $id current message id (for cid2http)
2292 * @return a modified array of attributes to be set for <div>
2294 function sq_body2div($attary, $mailbox, $message, $id){
2295 $me = 'sq_body2div';
2296 $divattary = Array('class' => "'bodyclass'");
2298 $has_bgc_stl = $has_txt_stl = false;
2300 if (is_array($attary) && sizeof($attary) > 0){
2301 foreach ($attary as $attname=>$attvalue){
2302 $quotchar = substr($attvalue, 0, 1);
2303 $attvalue = str_replace($quotchar, "", $attvalue);
2306 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2307 $styledef .= "background-image: url('$attvalue'); ";
2310 $has_bgc_stl = true;
2311 $styledef .= "background-color: $attvalue; ";
2314 $has_txt_stl = true;
2315 $styledef .= "color: $attvalue; ";
2319 // Outlook defines a white bgcolor and no text color. This can lead to
2320 // white text on a white bg with certain themes.
2321 if ($has_bgc_stl && !$has_txt_stl) {
2322 $styledef .= "color: $text; ";
2324 if (strlen($styledef) > 0){
2325 $divattary{"style"} = "\"$styledef\"";
2332 * This is the main function and the one you should actually be calling.
2333 * There are several variables you should be aware of an which need
2334 * special description.
2336 * Since the description is quite lengthy, see it here:
2337 * http://linux.duke.edu/projects/mini/htmlfilter/
2339 * @param $body the string with HTML you wish to filter
2340 * @param $tag_list see description above
2341 * @param $rm_tags_with_content see description above
2342 * @param $self_closing_tags see description above
2343 * @param $force_tag_closing see description above
2344 * @param $rm_attnames see description above
2345 * @param $bad_attvals see description above
2346 * @param $add_attr_to_tag see description above
2347 * @param $message message object
2348 * @param $id message id
2349 * @param $recursively_called boolean flag for recursive calls into this function (optional; default FALSE)
2350 * @return sanitized html safe to show on your pages.
2352 function sq_sanitize($body,
2354 $rm_tags_with_content,
2363 $recursively_called=FALSE
2365 $me = 'sq_sanitize';
2368 * See if tag_list is of tags to remove or tags to allow.
2369 * false means remove these tags
2370 * true means allow these tags
2372 $orig_tag_list = $tag_list;
2373 $rm_tags = array_shift($tag_list);
2376 * Normalize rm_tags and rm_tags_with_content.
2378 @array_walk
($tag_list, 'sq_casenormalize');
2379 @array_walk
($rm_tags_with_content, 'sq_casenormalize');
2380 @array_walk
($self_closing_tags, 'sq_casenormalize');
2383 $open_tags = Array();
2384 $trusted = "\n<!-- begin sanitized html -->\n";
2385 $skip_content = false;
2387 * Take care of netscape's stupid javascript entities like
2390 $body = preg_replace("/&(\{.*?\};)/si", "&\\1", $body);
2392 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2393 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2396 * RCDATA and RAWTEXT tags are handled differently:
2397 * next instance of closing tag is used, whether or not
2398 * the HTML is well formed before that
2400 global $rcdata_rawtext_tags;
2401 if (!$recursively_called
2402 && in_array($tagname, $rcdata_rawtext_tags)
2404 $closing_tag = false;
2405 $closing_tag_offset = $curpos;
2406 // seek out the closing tag for the current RCDATA/RAWTEXT tag
2408 // first we need to move forward to next available closing tag
2409 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2410 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2411 if ($next_tag === false) {
2412 $closing_tag = false;
2415 // but then we have to make sure it's a well-formed tag
2416 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2417 if ($closing_tag === false)
2419 else if ($closing_tag[0] !== false
2420 // these should be redundant
2421 && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2422 $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos +
1),
2423 $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2424 $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2425 $message, $id, $mailbox, true);
2426 $curpos = $closing_tag[4] +
1;
2429 $closing_tag_offset = $next_tag[0] +
1;
2431 if ($closing_tag === false)
2432 { /* 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 */ }
2435 $free_content = substr($body, $curpos, $lt-$curpos);
2437 * Take care of <style>
2439 if ($tagname == "style" && $tagtype == 1){
2440 list($free_content, $curpos) =
2441 sq_fixstyle($body, $gt+
1, $message, $id, $mailbox);
2442 if ($free_content != FALSE){
2443 if ( !empty($attary) ) {
2444 $attary = sq_fixatts($tagname,
2454 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2455 $trusted .= $free_content;
2456 $trusted .= sq_tagprint($tagname, false, 2);
2460 if ($skip_content == false){
2461 $trusted .= $free_content;
2463 if ($tagname != FALSE){
2465 if ($skip_content == $tagname){
2467 * Got to the end of tag we needed to remove.
2470 $skip_content = false;
2472 if ($skip_content == false){
2473 if ($tagname == "body"){
2476 if (isset($open_tags{$tagname}) &&
2477 $open_tags{$tagname} > 0){
2478 $open_tags{$tagname}--;
2486 * $rm_tags_with_content
2488 if ($skip_content == false){
2490 * See if this is a self-closing type and change
2491 * tagtype appropriately.
2494 && in_array($tagname, $self_closing_tags)){
2498 * See if we should skip this tag and any content
2501 if ($tagtype == 1 &&
2502 in_array($tagname, $rm_tags_with_content)){
2503 $skip_content = $tagname;
2505 if (($rm_tags == false
2506 && in_array($tagname, $tag_list)) ||
2507 ($rm_tags == true &&
2508 !in_array($tagname, $tag_list))){
2512 * Convert body into div.
2514 if ($tagname == "body"){
2516 $attary = sq_body2div($attary, $mailbox,
2520 if (isset($open_tags{$tagname})){
2521 $open_tags{$tagname}++
;
2523 $open_tags{$tagname}=1;
2527 * This is where we run other checks.
2529 if (is_array($attary) && sizeof($attary) > 0){
2530 $attary = sq_fixatts($tagname,
2544 if ($tagname != false && $skip_content == false){
2545 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2550 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2551 if ($force_tag_closing == true){
2552 foreach ($open_tags as $tagname=>$opentimes){
2553 while ($opentimes > 0){
2554 $trusted .= '</' . $tagname . '>';
2560 $trusted .= "<!-- end sanitized html -->\n";
2565 * This is a wrapper function to call html sanitizing routines.
2567 * @param $body the body of the message
2568 * @param $id the id of the message
2572 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2573 * into internal SM compose links
2574 * (optional; default = TRUE)
2575 * @return a string with html safe to display in the browser.
2577 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2579 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2581 global $attachment_common_show_images, $view_unsafe_images,
2582 $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2583 $remove_rcdata_rawtext_tags_and_content;
2585 $rcdata_rawtext_tags = array(
2590 // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2594 * Don't display attached images in HTML mode.
2598 $attachment_common_show_images = false;
2600 false, // remove these tags
2612 $rm_tags_with_content = Array(
2622 if (!$allow_svg_display)
2623 $rm_tags_with_content[] = 'svg';
2625 * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2626 * case that they are, but if you prefer to remove them and their contents entirely
2627 * (in most cases, should be a safe thing with minimal impact), you can add the
2628 * following to config/config_local.php
2629 * $remove_rcdata_rawtext_tags_and_content = TRUE;
2631 if ($remove_rcdata_rawtext_tags_and_content)
2632 $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2634 $self_closing_tags = Array(
2642 $force_tag_closing = true;
2644 $rm_attnames = Array(
2655 global $use_transparent_security_image;
2656 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2657 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2659 $bad_attvals = Array(
2662 "/^src|background/i" =>
2665 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2666 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2667 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2670 "\\1$secremoveimg\\2",
2671 "\\1$secremoveimg\\2",
2672 "\\1$secremoveimg\\2",
2675 "/^href|action/i" =>
2678 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2679 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2680 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2695 "/include-source/i",
2697 // position:relative can also be exploited
2698 // to put content outside of email body area
2699 // and position:fixed is similarly exploitable
2700 // as position:absolute, so we'll remove it
2703 // Does this screw up legitimate HTML messages?
2704 // If so, the only fix I see is to allow position
2705 // attributes (any values? I think we still have
2706 // to block static and fixed) only if $use_iframe
2707 // is enabled (1.5.0+)
2709 // was: "/position\s*:\s*absolute/i",
2713 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2714 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2715 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2716 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2717 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2736 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2737 // images off by default.
2738 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET
, FALSE);
2740 if (!$view_unsafe_images){
2742 * Remove any references to http/https if view_unsafe_images set
2745 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2746 '/^([\'\"])\s*https*:.*([\'\"])/si');
2747 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2748 "\\1$secremoveimg\\1");
2749 array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2750 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2751 array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2752 "url(\\1$secremoveimg\\1)");
2755 $add_attr_to_tag = Array(
2757 Array('target'=>'"_blank"',
2758 'title'=>'"'._("This external link will open in a new window").'"'
2761 $trusted = sq_sanitize($body,
2763 $rm_tags_with_content,
2773 if (strpos($trusted,$secremoveimg)){
2774 $has_unsafe_images = true;
2777 // we want to parse mailto's in HTML output, change to SM compose links
2778 // this is a modified version of code from url_parser.php... but Marc is
2779 // right: we need a better filtering implementation; adding this randomly
2780 // here is not a great solution
2782 if ($take_mailto_links) {
2783 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2784 global $MailTo_PReg_Match;
2785 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2786 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2787 // quote before mailto, and we'll also try to convert the non-links back into links
2788 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2789 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2790 foreach ($regs[0] as $i => $mailto_before) {
2791 $mailto_params = $regs[11][$i];
2793 // get rid of any leading quote we may have captured but don't care about
2795 $mailto_before = ltrim($mailto_before, '"\'');
2797 // get rid of any tailing quote since we have to add send_to to the end
2799 $mailto_before = rtrim($mailto_before, '"\'');
2800 $mailto_params = rtrim($mailto_params, '"\'');
2802 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2803 $to = 'to=' . $regs[2][$i];
2804 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2805 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2807 if ($mailto_params) //already some params, append to them
2808 $mailto_params .= '&' . $to;
2810 $mailto_params .= '?' . $to;
2814 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2816 // we'll already have target=_blank, no need to allow comp_in_new
2817 // here (which would be a lot more work anyway)
2819 global $compose_new_win;
2820 $temp_comp_in_new = $compose_new_win;
2821 $compose_new_win = 0;
2822 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2823 $compose_new_win = $temp_comp_in_new;
2825 // remove <a href=" and anything after the next quote (we only
2826 // need the uri, not the link HTML) in compose uri
2828 // but only do this if the original mailto was in a real anchor tag
2830 if (!empty($regs[1][$i])) {
2831 $comp_uri = substr($comp_uri, 9);
2832 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2834 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2843 * function SendDownloadHeaders - send file to the browser
2845 * Original Source: SM core src/download.php
2846 * moved here to make it available to other code, and separate
2847 * front end from back end functionality.
2849 * @param string $type0 first half of mime type
2850 * @param string $type1 second half of mime type
2851 * @param string $filename filename to tell the browser for downloaded file
2852 * @param boolean $force whether to force the download dialog to pop
2853 * @param optional integer $filesize send the Content-Header and length to the browser
2856 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2857 global $languages, $squirrelmail_language;
2858 $isIE = $isIE6plus = false;
2860 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER
);
2862 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2863 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2867 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2868 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2872 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2873 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2875 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2877 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace(' ', ' ', $filename));
2880 // A Pox on Microsoft and it's Internet Explorer!
2882 // IE has lots of bugs with file downloads.
2883 // It also has problems with SSL. Both of these cause problems
2884 // for us in this function.
2886 // See this article on Cache Control headers and SSL
2887 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2889 // The best thing you can do for IE is to upgrade to the latest
2891 //set all the Cache Control Headers for IE
2893 $filename=rawurlencode($filename);
2894 header ("Pragma: public");
2895 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2896 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2897 // header ("Cache-Control: post-check=0, pre-check=0", false);
2898 header ("Cache-Control: private");
2900 //set the inline header for IE, we'll add the attachment header later if we need it
2901 header ("Content-Disposition: inline; filename=$filename");
2905 // Try to show in browser window
2906 header ("Content-Disposition: inline; filename=\"$filename\"");
2907 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2909 // Try to pop up the "save as" box
2911 // IE makes this hard. It pops up 2 save boxes, or none.
2912 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2913 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2914 // But, according to Microsoft, it is "RFC compliant but doesn't
2915 // take into account some deviations that allowed within the
2916 // specification." Doesn't that mean RFC non-compliant?
2917 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2919 // all browsers need the application/octet-stream header for this
2920 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2922 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2923 // Do not have quotes around filename, but that applied to
2924 // "attachment"... does it apply to inline too?
2925 header ("Content-Disposition: attachment; filename=\"$filename\"");
2927 if ($isIE && !$isIE6plus) {
2928 // This combination seems to work mostly. IE 5.5 SP 1 has
2929 // known issues (see the Microsoft Knowledge Base)
2931 // This works for most types, but doesn't work with Word files
2932 header ("Content-Type: application/download; name=\"$filename\"");
2933 header ("Content-Type: application/force-download; name=\"$filename\"");
2934 // These are spares, just in case. :-)
2935 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2936 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2937 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2939 // This is to prevent IE for MIME sniffing and auto open a file in IE
2940 header ("Content-Type: application/force-download; name=\"$filename\"");
2942 // another application/octet-stream forces download for Netscape
2943 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2947 //send the content-length header if the calling function provides it
2948 if ($filesize > 0) {
2949 header("Content-Length: $filesize");
2952 } // end fn SendDownloadHeaders