Updated SVG handling, gracefully fix broken base64-encoded messages, also close XSS...
[squirrelmail.git] / functions / mime.php
1 <?php
2
3 /**
4 * mime.php
5 *
6 * This contains the functions necessary to detect and decode MIME
7 * messages.
8 *
9 * @copyright 1999-2019 The SquirrelMail Project Team
10 * @license http://opensource.org/licenses/gpl-license.php GNU Public License
11 * @version $Id$
12 * @package squirrelmail
13 */
14
15 /**
16 * dependency information
17 functions dependency
18 mime_structure
19 class/mime/Message.class.php
20 Message::parseStructure
21 functions/page_header.php
22 displayPageHeader
23 functions/display_messages.php
24 plain_error_message
25 mime_fetch_body
26 functions/imap_general.php
27 sqimap_run_command
28 mime_print_body_lines
29
30
31
32 functions/imap.php
33 functions/attachment_common.php
34 functions/display_messages.php
35
36 magicHtml => url_parser
37 translateText => url_parser
38
39 */
40
41
42 /* -------------------------------------------------------------------------- */
43 /* MIME DECODING */
44 /* -------------------------------------------------------------------------- */
45
46 /**
47 * Get the MIME structure
48 *
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.
52 */
53 function mime_structure ($bodystructure, $flags=array()) {
54
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));
58 $i = 0;
59 $msg = Message::parseStructure($read,$i);
60
61 if (!is_object($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( $errormessage );
68 echo '</body></html>';
69 exit;
70 }
71 if (count($flags)) {
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});
75 switch ($char) {
76 case 'S':
77 if (strtolower($flag) == '\\seen') {
78 $msg->is_seen = true;
79 }
80 break;
81 case 'A':
82 if (strtolower($flag) == '\\answered') {
83 $msg->is_answered = true;
84 }
85 break;
86 case 'D':
87 if (strtolower($flag) == '\\deleted') {
88 $msg->is_deleted = true;
89 }
90 break;
91 case 'F':
92 if (strtolower($flag) == '\\flagged') {
93 $msg->is_flagged = true;
94 }
95 else if (strtolower($flag) == '$forwarded') {
96 $msg->is_forwarded = true;
97 }
98 break;
99 case 'M':
100 if (strtolower($flag) == '$mdnsent') {
101 $msg->is_mdnsent = true;
102 }
103 break;
104 default:
105 break;
106 }
107 }
108 }
109 // listEntities($msg);
110 return $msg;
111 }
112
113
114
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
117 * $message.
118 * First, it checks to see if it is a multipart message. If it is, then it
119 * handles that as it sees is necessary. If it is just a regular entity,
120 * then it parses it and adds the necessary header information (by calling out
121 * to mime_get_elements()
122 */
123
124 function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) {
125 /* Do a bit of error correction. If we couldn't find the entity id, just guess
126 * that it is the first one. That is usually the case anyway.
127 */
128
129 if (!$ent_id) {
130 $cmd = "FETCH $id BODY[]";
131 } else {
132 $cmd = "FETCH $id BODY[$ent_id]";
133 }
134
135 if ($fetch_size!=0) $cmd .= "<0.$fetch_size>";
136
137 $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, TRUE);
138 do {
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.
147
148 $wholemessage = implode('', $data);
149 if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) {
150 $ret = substr($wholemessage, 0, $regs[1]);
151 /* There is some information in the content info header that could be important
152 * in order to parse html messages. Let's get them here.
153 */
154 // if ($ret{0} == '<') {
155 // $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE);
156 // }
157 } else if (preg_match('/"([^"]*)"/', $topline, $regs)) {
158 $ret = $regs[1];
159 } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) {
160 $ret = $wholemessage;
161 } else {
162 global $where, $what, $mailbox, $passed_id, $startMessage;
163 $par = 'mailbox=' . urlencode($mailbox) . '&amp;passed_id=' . $passed_id;
164 if (isset($where) && isset($what)) {
165 $par .= '&amp;where=' . urlencode($where) . '&amp;what=' . urlencode($what);
166 } else {
167 $par .= '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
168 }
169 $par .= '&amp;response=' . urlencode($response) .
170 '&amp;message=' . urlencode($message) .
171 '&amp;topline=' . urlencode($topline);
172
173 echo '<tt><br />' .
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.") .
177 '</td></tr>' .
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 />";
183
184 $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE);
185 array_shift($data);
186 $wholemessage = implode('', $data);
187
188 $ret = $wholemessage;
189 }
190 return $ret;
191 }
192
193 function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') {
194
195 /* Don't kill the connection if the browser is over a dialup
196 * and it would take over 30 seconds to download it.
197 * Don't call set_time_limit in safe mode.
198 */
199
200 if (!ini_get('safe_mode')) {
201 set_time_limit(0);
202 }
203 /* in case of base64 encoded attachments, do not buffer them.
204 Instead, echo the decoded attachment directly to screen */
205 if (strtolower($encoding) == 'base64') {
206 if (!$ent_id) {
207 $query = "FETCH $id BODY[]";
208 } else {
209 $query = "FETCH $id BODY[$ent_id]";
210 }
211 sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
212 } else {
213 $body = mime_fetch_body ($imap_stream, $id, $ent_id);
214 if (is_resource($rStream)) {
215 fputs($rStream,decodeBody($body, $encoding, $force_crlf));
216 } else {
217 echo decodeBody($body, $encoding, $force_crlf);
218 }
219 }
220
221 /*
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.
230 */
231 /*
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.
234 */
235
236 return;
237 }
238
239 /* -[ END MIME DECODING ]----------------------------------------------------------- */
240
241 /* This is here for debugging purposes. It will print out a list
242 * of all the entity IDs that are in the $message object.
243 */
244 function listEntities ($message) {
245 if ($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]); $i++) {
248 echo "$i : ";
249 $msg = listEntities($message->entities[$i]);
250
251 if ($msg) {
252 echo "return: ";
253 return $msg;
254 }
255 }
256 }
257 }
258
259 function getPriorityStr($priority) {
260 $priority_level = substr($priority,0,1);
261
262 switch($priority_level) {
263 /* Check for a higher then normal priority. */
264 case '1':
265 case '2':
266 $priority_string = _("High");
267 break;
268
269 /* Check for a lower then normal priority. */
270 case '4':
271 case '5':
272 $priority_string = _("Low");
273 break;
274
275 /* Check for a normal priority. */
276 case '3':
277 default:
278 $priority_level = '3';
279 $priority_string = _("Normal");
280 break;
281
282 }
283 return $priority_string;
284 }
285
286 /* returns a $message object for a particular entity id */
287 function getEntity ($message, $ent_id) {
288 return $message->getEntity($ent_id);
289 }
290
291 /* translateText
292 * Extracted from strings.php 23/03/2002
293 */
294
295 function translateText(&$body, $wrap_at, $charset) {
296 global $where, $what; /* from searching */
297 global $color; /* color theme */
298
299 // require_once(SM_PATH . 'functions/url_parser.php');
300
301 $body_ary = explode("\n", $body);
302 for ($i=0; $i < count($body_ary); $i++) {
303 $line = rtrim($body_ary[$i],"\r");
304
305 if (strlen($line) - 2 >= $wrap_at) {
306 sqWordWrap($line, $wrap_at, $charset);
307 }
308 $line = charset_decode($charset, $line);
309 $line = str_replace("\t", ' ', $line);
310
311 parseUrl ($line);
312
313 $quotes = 0;
314 $pos = 0;
315 $j = strlen($line);
316
317 while ($pos < $j) {
318 if ($line[$pos] == ' ') {
319 $pos++;
320 } else if (strpos($line, '&gt;', $pos) === $pos) {
321 $pos += 4;
322 $quotes++;
323 } else {
324 break;
325 }
326 }
327
328 if ($quotes % 2) {
329 $line = '<span class="quote1">' . $line . '</span>';
330 } elseif ($quotes) {
331 $line = '<span class="quote2">' . $line . '</span>';
332 }
333
334 $body_ary[$i] = $line;
335 }
336 $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
337 }
338
339 /**
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
343 * bottom, etc.
344 *
345 * Since 1.2.0 function uses message_body hook.
346 * Till 1.3.0 function included output of formatAttachments().
347 *
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 $mailbox (since 1.3.0) imap folder name
355 * @return string html formated message text
356 */
357 function formatBody($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.
361 */
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,
366 $oTemplate, $nbsp;
367
368 // workaround for not updated config.php
369 if (! isset($use_iframe)) $use_iframe = false;
370
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);
374
375 $body = '';
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, $body_message->header->encoding);
382
383 if (isset($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);
387 }
388 }
389
390 /* As of 1.5.2, $body is passed (and modified) by reference */
391 do_hook('message_body', $body);
392
393 /* If there are other types that shouldn't be formatted, add
394 * them here.
395 */
396
397 if ($body_message->header->type1 == 'html') {
398 if ($show_html_default <> 1) {
399 $entity_conv = array('&nbsp;' => ' ',
400 '<p>' => "\n",
401 '<P>' => "\n",
402 '<br>' => "\n",
403 '<BR>' => "\n",
404 '<br />' => "\n",
405 '<BR />' => "\n",
406 '&gt;' => '>',
407 '&lt;' => '<');
408 $body = strtr($body, $entity_conv);
409 $body = strip_tags($body);
410 $body = trim($body);
411 translateText($body, $wrap_at,
412 $body_message->header->getParameter('charset'));
413 } elseif ($use_iframe) {
414 /**
415 * If we don't add html message between iframe tags,
416 * we must detect unsafe images and modify $has_unsafe_images.
417 */
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);
421
422 // creating iframe url
423 $iframeurl=sqm_baseuri().'src/view_html.php?'
424 . 'mailbox=' . $urlmailbox
425 . '&amp;passed_id=' . $id
426 . '&amp;ent_id=' . $ent_num
427 . '&amp;view_unsafe_images=' . (int) $view_unsafe_images;
428
429 global $oTemplate;
430 $oTemplate->assign('iframe_url', $iframeurl);
431 $oTemplate->assign('iframe_height', $iframe_height);
432 $oTemplate->assign('html_body', $html_body);
433
434 $body = $oTemplate->fetch('read_html_iframe.tpl');
435 } else {
436 // old way of html rendering
437 /**
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.
441 */
442 $charset = $body_message->header->getParameter('charset');
443 if (!empty($charset)) {
444 $body = charset_decode($charset,$body,false,true);
445 }
446 $body = magicHTML($body, $id, $message, $mailbox);
447 }
448 } else {
449 translateText($body, $wrap_at,
450 $body_message->header->getParameter('charset'));
451 }
452
453 /*
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.
460 */
461
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 = '';
465
466 // Prepare and build a link for downloading the mail.
467 $link = 'passed_id=' . $id . '&amp;ent_id='.$ent_num.
468 '&amp;mailbox=' . $urlmailbox .'&amp;sort=' . $sort .
469 '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
470 if (isset($passed_ent_id)) {
471 $link .= '&amp;passed_ent_id='.$passed_ent_id;
472 }
473 $download_href = SM_PATH . 'src/download.php?absolute_dl=true&amp;' . $link;
474
475 // Always add the link for downloading the mail as a file to the global
476 // variable.
477 $download_and_unsafe_link .= "$nbsp|$nbsp"
478 . create_hyperlink($download_href, _("Download this as a file"));
479
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 $text = _("Hide Unsafe Images");
486 } else {
487 if (isset($has_unsafe_images) && $has_unsafe_images) {
488 $link .= '&amp;view_unsafe_images=1';
489 $text = _("View Unsafe Images");
490 } else {
491 $text = '';
492 }
493 }
494
495 // Only create a link for unsafe images if there's need for one. If so:
496 // add it to the global variable.
497 if($text != '') {
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);
502 }
503 }
504 return $body;
505 }
506
507 /**
508 * Generate attachments array for passing to templates.
509 *
510 * @since 1.5.2
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 $id message id
515 */
516 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
517 global $where, $what, $startMessage, $color, $passed_ent_id,
518 $base_uri, $block_svg_download;
519
520 $att_ar = $message->getAttachments($exclude_id);
521 $urlMailbox = urlencode($mailbox);
522
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)
530 continue;
531
532 $name = '';
533 $links = array();
534 $links['download link']['text'] = _("Download");
535 $links['download link']['href'] = $base_uri .
536 "src/download.php?absolute_dl=true&amp;passed_id=$id&amp;mailbox=$urlMailbox&amp;ent_id=$ent";
537
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 . ']' ;
544 }
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));
555 } else {
556 $from_name = _("Unknown sender");
557 }
558 $description = _("From").': '.$from_name;
559 } else {
560 $default_page = $base_uri . 'src/download.php';
561 $filename = $att->getFilename();
562 if ($header->description) {
563 $description = decodeHeader($header->description);
564 } else {
565 $description = '';
566 }
567 }
568
569 $display_filename = $filename;
570 if (isset($passed_ent_id)) {
571 $passed_ent_id_link = '&amp;passed_ent_id='.$passed_ent_id;
572 } else {
573 $passed_ent_id_link = '';
574 }
575 $defaultlink = $default_page . "?startMessage=$startMessage"
576 . "&amp;passed_id=$id&amp;mailbox=$urlMailbox"
577 . '&amp;ent_id='.$ent.$passed_ent_id_link;
578 if ($where && $what) {
579 $defaultlink .= '&amp;where='. urlencode($where).'&amp;what='.urlencode($what);
580 }
581 // IE does make use of mime content sniffing. Forcing a download
582 // prohibit execution of XSS inside an application/octet-stream attachment
583 if ($type0 == 'application' && $type1 == 'octet-stream') {
584 $defaultlink .= '&amp;absolute_dl=true';
585 }
586
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
590 * run as well.
591 */
592 // First remember the default link.
593 $defaultlink_orig = $defaultlink;
594
595 /* The API for this hook has changed as of 1.5.2 so that all plugin
596 arguments are passed in an array instead of each their own plugin
597 argument, and arguments are passed by reference, so instead of
598 returning any changes, changes should simply be made to the original
599 arguments themselves. */
600 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
601 &$defaultlink, &$display_filename, &$where, &$what);
602 do_hook("attachment $type0/$type1", $temp);
603 /* The API for this hook has changed as of 1.5.2 so that all plugin
604 arguments are passed in an array instead of each their own plugin
605 argument, and arguments are passed by reference, so instead of
606 returning any changes, changes should simply be made to the original
607 arguments themselves. */
608 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
609 &$defaultlink, &$display_filename, &$where, &$what);
610 // Do not let a generic plugin change the default link if a more
611 // specialized one already did it...
612 if ($defaultlink != $defaultlink_orig) {
613 $dummy = '';
614 $temp[5] = &$dummy;
615 }
616 do_hook("attachment $type0/*", $temp);
617 /* The API for this hook has changed as of 1.5.2 so that all plugin
618 arguments are passed in an array instead of each their own plugin
619 argument, and arguments are passed by reference, so instead of
620 returning any changes, changes should simply be made to the original
621 arguments themselves. */
622 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
623 &$defaultlink, &$display_filename, &$where, &$what);
624 // Do not let a generic plugin change the default link if a more
625 // specialized one already did it...
626 if ($defaultlink != $defaultlink_orig) {
627 $dummy = '';
628 $temp[5] = &$dummy;
629 }
630 do_hook("attachment */*", $temp);
631
632 $this_attachment = array();
633 $this_attachment['Name'] = decodeHeader($display_filename);
634 $this_attachment['Description'] = $description;
635 $this_attachment['DefaultHREF'] = $defaultlink;
636 $this_attachment['DownloadHREF'] = $links['download link']['href'];
637 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
638 $this_attachment['Size'] = $header->size;
639 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
640 $this_attachment['OtherLinks'] = array();
641 foreach ($links as $val) {
642 if ($val['text']==_("Download") || $val['text'] == _("View"))
643 continue;
644 if (empty($val['text']) && empty($val['extra']))
645 continue;
646
647 $temp = array();
648 $temp['HREF'] = $val['href'];
649 $temp['Text'] = (empty($val['text']) ? '' : $val['text']) . (empty($val['extra']) ? '' : $val['extra']);
650 $this_attachment['OtherLinks'][] = $temp;
651 }
652 $attachments[] = $this_attachment;
653
654 unset($links);
655 }
656
657 return $attachments;
658 }
659
660 /**
661 * Displays attachment links and information
662 *
663 * Since 1.3.0 function is not included in formatBody() call.
664 *
665 * Since 1.0.2 uses attachment $type0/$type1 hook.
666 * Since 1.2.5 uses attachment $type0/* hook.
667 * Since 1.5.0 uses attachments_bottom hook.
668 * Since 1.5.2 uses templates and does *not* return a value.
669 *
670 * @param object $message SquirrelMail message object
671 * @param array $exclude_id message parts that are not attachments.
672 * @param string $mailbox mailbox name
673 * @param integer $id message id
674 */
675 function formatAttachments($message, $exclude_id, $mailbox, $id) {
676 global $oTemplate;
677
678 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
679
680 $oTemplate->assign('attachments', $attach);
681 $oTemplate->display('read_attachments.tpl');
682 }
683
684 function sqimap_base64_decode(&$string) {
685
686 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
687 // fly decoding (to reduce memory usage) you have to check if the
688 // data has incomplete pairs
689
690 // Remove the noise in order to check if the 4 bytes pairs are complete
691 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
692
693 $sStringRem = '';
694 $iMod = strlen($string) % 4;
695 if ($iMod) {
696 $sStringRem = substr($string,-$iMod);
697 // Check if $sStringRem contains padding characters
698 if (substr($sStringRem,-1) != '=') {
699 $string = substr($string,0,-$iMod);
700 } else {
701 $sStringRem = '';
702 }
703 }
704 $string = base64_decode($string);
705 return $sStringRem;
706 }
707
708 /**
709 * Decodes encoded string (usually message body)
710 *
711 * This function decodes a string (usually the message body)
712 * depending on the encoding type. Currently quoted-printable
713 * and base64 encodings are supported.
714 *
715 * The decode_body hook was added to this function in 1.4.2/1.5.0.
716 * The $force_crlf parameter was added in 1.5.2.
717 *
718 * @param string $string The encoded string
719 * @param string $encoding used encoding
720 * @param string $force_crlf Whether or not to force CRLF or LF
721 * line endings (or to leave as is).
722 * If given as "LF", line endings will
723 * all be converted to LF; if "CRLF",
724 * line endings will all be converted
725 * to CRLF. If given as an empty value,
726 * the global $force_crlf_default will
727 * be consulted (it can be specified in
728 * config/config_local.php). Otherwise,
729 * any other value will cause the string
730 * to be left alone. Note that this will
731 * be overridden to "LF" if not using at
732 * least PHP version 4.3.0. (OPTIONAL;
733 * default is empty - consult global
734 * default value)
735 *
736 * @return string The decoded string
737 *
738 * @since 1.0
739 *
740 */
741 function decodeBody($string, $encoding, $force_crlf='') {
742
743 global $force_crlf_default;
744 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
745 $force_crlf = strtoupper($force_crlf);
746
747 // must force line endings to LF due to broken
748 // quoted_printable_decode() in PHP versions
749 // before 4.3.0 (see below)
750 //
751 if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
752 $string = str_replace("\r\n", "\n", $string);
753 else if ($force_crlf == 'CRLF')
754 $string = str_replace("\n", "\r\n", $string);
755
756 $encoding = strtolower($encoding);
757
758 $encoding_handler = do_hook('decode_body', $encoding);
759
760
761 // plugins get first shot at decoding the string
762 //
763 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
764 $string = $encoding_handler('decode', $string);
765
766 } elseif ($encoding == 'quoted-printable' ||
767 $encoding == 'quoted_printable') {
768
769 // quoted_printable_decode() function is broken in older
770 // php versions. Text with \r\n decoding was fixed only
771 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
772 // and the above call to: str_replace("\r\n", "\n", $string);
773 //
774 $string = quoted_printable_decode($string);
775
776 } elseif ($encoding == 'base64') {
777 $string = base64_decode($string);
778 }
779
780 // All other encodings are returned raw.
781 return $string;
782 }
783
784 /**
785 * Decodes headers
786 *
787 * This function decodes strings that are encoded according to
788 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
789 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
790 *
791 * @param string $string header string that has to be made readable
792 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
793 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
794 * @param boolean $decide decide if string can be utfencoded. defaults to false
795 * @return string decoded header string
796 */
797 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
798 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
799 if (is_array($string)) {
800 $string = implode("\n", $string);
801 }
802
803 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
804 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
805 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
806 // Do we need to return at this point?
807 // return $string;
808 }
809 $i = 0;
810 $iLastMatch = -2;
811 $encoded = true;
812
813 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
814 $aString = explode(' ',$string);
815 $ret = '';
816 foreach ($aString as $chunk) {
817 if ($encoded && $chunk === '') {
818 continue;
819 } elseif ($chunk === '') {
820 $ret .= ' ';
821 continue;
822 }
823 $encoded = false;
824 /* if encoded words are not separated by a linear-space-white we still catch them */
825 $j = $i-1;
826
827 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
828 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
829 if ($iLastMatch !== $j) {
830 if ($htmlsafe) {
831 $ret .= '&#32;';
832 } else {
833 $ret .= ' ';
834 }
835 }
836 $iLastMatch = $i;
837 $j = $i;
838 if ($htmlsafe) {
839 $ret .= sm_encode_html_special_chars($res[1]);
840 } else {
841 $ret .= $res[1];
842 }
843 $encoding = ucfirst($res[3]);
844
845 /* decide about valid decoding */
846 if ($decide && is_conversion_safe($res[2])) {
847 $utfencode=true;
848 $can_be_encoded=true;
849 } else {
850 $can_be_encoded=false;
851 }
852 switch ($encoding)
853 {
854 case 'B':
855 // fix broken base64-encoded strings (remove end = padding,
856 // change any = to + in middle of string, add padding back
857 // to the end)
858 if ($fix_broken_base64_encoded_messages) {
859 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
860 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
861 }
862 $replace = base64_decode($res[4]);
863 if ($utfencode) {
864 if ($can_be_encoded) {
865 /* convert string to different charset,
866 * if functions asks for it (usually in compose)
867 */
868 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
869 } else {
870 // convert string to html codes in order to display it
871 $ret .= charset_decode($res[2],$replace);
872 }
873 } else {
874 if ($htmlsafe) {
875 $replace = sm_encode_html_special_chars($replace);
876 }
877 $ret.= $replace;
878 }
879 break;
880 case 'Q':
881 $replace = str_replace('_', ' ', $res[4]);
882 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
883 create_function ('$matches', 'return chr(hexdec($matches[1]));'),
884 $replace);
885 if ($utfencode) {
886 if ($can_be_encoded) {
887 /* convert string to different charset,
888 * if functions asks for it (usually in compose)
889 */
890 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
891 } else {
892 // convert string to html codes in order to display it
893 $replace = charset_decode($res[2], $replace);
894 }
895 } else {
896 if ($htmlsafe) {
897 $replace = sm_encode_html_special_chars($replace);
898 }
899 }
900 $ret .= $replace;
901 break;
902 default:
903 break;
904 }
905 $chunk = $res[5];
906 $encoded = true;
907 }
908 if (!$encoded) {
909 if ($htmlsafe) {
910 $ret .= '&#32;';
911 } else {
912 $ret .= ' ';
913 }
914 }
915
916 if (!$encoded && $htmlsafe) {
917 $ret .= sm_encode_html_special_chars($chunk);
918 } else {
919 $ret .= $chunk;
920 }
921 ++$i;
922 }
923 /* remove the first added space */
924 if ($ret) {
925 if ($htmlsafe) {
926 $ret = substr($ret,5);
927 } else {
928 $ret = substr($ret,1);
929 }
930 }
931
932 return $ret;
933 }
934
935 /**
936 * Encodes header
937 *
938 * Function uses XTRA_CODE _encodeheader function, if such function exists.
939 *
940 * Function uses Q encoding by default and encodes a string according to RFC
941 * 1522 for use in headers if it contains 8-bit characters or anything that
942 * looks like it should be encoded.
943 *
944 * Function switches to B encoding and encodeHeaderBase64() function, if
945 * string is 8bit and multibyte character set supported by mbstring extension
946 * is used. It can cause E_USER_NOTICE errors, if interface is used with
947 * multibyte character set unsupported by mbstring extension.
948 *
949 * @param string $string header string, that has to be encoded
950 * @return string quoted-printable encoded string
951 * @todo make $mb_charsets system wide constant
952 */
953 function encodeHeader ($string) {
954 global $default_charset, $languages, $squirrelmail_language;
955
956 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
957 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
958 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
959 }
960
961 // Use B encoding for multibyte charsets
962 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
963 if (in_array($default_charset,$mb_charsets) &&
964 in_array($default_charset,sq_mb_list_encodings()) &&
965 sq_is8bit($string)) {
966 return encodeHeaderBase64($string,$default_charset);
967 } elseif (in_array($default_charset,$mb_charsets) &&
968 sq_is8bit($string) &&
969 ! in_array($default_charset,sq_mb_list_encodings())) {
970 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
971 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
972 }
973
974 // Encode only if the string contains 8-bit characters or =?
975 $j = strlen($string);
976 $max_l = 75 - strlen($default_charset) - 7;
977 $aRet = array();
978 $ret = '';
979 $iEncStart = $enc_init = false;
980 $cur_l = $iOffset = 0;
981 for($i = 0; $i < $j; ++$i) {
982 switch($string{$i})
983 {
984 case '"':
985 case '=':
986 case '<':
987 case '>':
988 case ',':
989 case '?':
990 case '_':
991 if ($iEncStart === false) {
992 $iEncStart = $i;
993 }
994 $cur_l+=3;
995 if ($cur_l > ($max_l-2)) {
996 /* if there is an stringpart that doesn't need encoding, add it */
997 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
998 $aRet[] = "=?$default_charset?Q?$ret?=";
999 $iOffset = $i;
1000 $cur_l = 0;
1001 $ret = '';
1002 $iEncStart = false;
1003 } else {
1004 $ret .= sprintf("=%02X",ord($string{$i}));
1005 }
1006 break;
1007 case '(':
1008 case ')':
1009 if ($iEncStart !== false) {
1010 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1011 $aRet[] = "=?$default_charset?Q?$ret?=";
1012 $iOffset = $i;
1013 $cur_l = 0;
1014 $ret = '';
1015 $iEncStart = false;
1016 }
1017 break;
1018 case ' ':
1019 if ($iEncStart !== false) {
1020 $cur_l++;
1021 if ($cur_l > $max_l) {
1022 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1023 $aRet[] = "=?$default_charset?Q?$ret?=";
1024 $iOffset = $i;
1025 $cur_l = 0;
1026 $ret = '';
1027 $iEncStart = false;
1028 } else {
1029 $ret .= '_';
1030 }
1031 }
1032 break;
1033 default:
1034 $k = ord($string{$i});
1035 if ($k > 126) {
1036 if ($iEncStart === false) {
1037 // do not start encoding in the middle of a string, also take the rest of the word.
1038 $sLeadString = substr($string,0,$i);
1039 $aLeadString = explode(' ',$sLeadString);
1040 $sToBeEncoded = array_pop($aLeadString);
1041 $iEncStart = $i - strlen($sToBeEncoded);
1042 $ret .= $sToBeEncoded;
1043 $cur_l += strlen($sToBeEncoded);
1044 }
1045 $cur_l += 3;
1046 /* first we add the encoded string that reached it's max size */
1047 if ($cur_l > ($max_l-2)) {
1048 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1049 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1050 $cur_l = 3;
1051 $ret = '';
1052 $iOffset = $i;
1053 $iEncStart = $i;
1054 }
1055 $enc_init = true;
1056 $ret .= sprintf("=%02X", $k);
1057 } else {
1058 if ($iEncStart !== false) {
1059 $cur_l++;
1060 if ($cur_l > $max_l) {
1061 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1062 $aRet[] = "=?$default_charset?Q?$ret?=";
1063 $iEncStart = false;
1064 $iOffset = $i;
1065 $cur_l = 0;
1066 $ret = '';
1067 } else {
1068 $ret .= $string{$i};
1069 }
1070 }
1071 }
1072 break;
1073 }
1074 }
1075
1076 if ($enc_init) {
1077 if ($iEncStart !== false) {
1078 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1079 $aRet[] = "=?$default_charset?Q?$ret?=";
1080 } else {
1081 $aRet[] = substr($string,$iOffset);
1082 }
1083 $string = implode('',$aRet);
1084 }
1085 return $string;
1086 }
1087
1088 /**
1089 * Encodes string according to rfc2047 B encoding header formating rules
1090 *
1091 * It is recommended way to encode headers with character sets that store
1092 * symbols in more than one byte.
1093 *
1094 * Function requires mbstring support. If required mbstring functions are missing,
1095 * function returns false and sets E_USER_WARNING level error message.
1096 *
1097 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1098 * that mbstring functions will generate E_WARNING errors, if unsupported
1099 * character set is used. mb_encode_mimeheader function provided by php
1100 * mbstring extension is not used in order to get better control of header
1101 * encoding.
1102 *
1103 * Used php code functions - function_exists(), trigger_error(), strlen()
1104 * (is used with charset names and base64 strings). Used php mbstring
1105 * functions - mb_strlen and mb_substr.
1106 *
1107 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1108 * encoding), rfc 2822 (header folding)
1109 *
1110 * @param string $string header string that must be encoded
1111 * @param string $charset character set. Must be supported by mbstring extension.
1112 * Use sq_mb_list_encodings() to detect supported charsets.
1113 * @return string string encoded according to rfc2047 B encoding formating rules
1114 * @since 1.5.1
1115 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1116 * @todo Do we want to control max length of header?
1117 * @todo Do we want to control EOL (end-of-line) marker?
1118 * @todo Do we want to translate error message?
1119 */
1120 function encodeHeaderBase64($string,$charset) {
1121 /**
1122 * Check mbstring function requirements.
1123 */
1124 if (! function_exists('mb_strlen') ||
1125 ! function_exists('mb_substr')) {
1126 // set E_USER_WARNING
1127 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1128 // return false
1129 return false;
1130 }
1131
1132 // initial return array
1133 $aRet = array();
1134
1135 /**
1136 * header length = 75 symbols max (same as in encodeHeader)
1137 * remove $charset length
1138 * remove =? ? ?= (5 chars)
1139 * remove 2 more chars (\r\n ?)
1140 */
1141 $iMaxLength = 75 - strlen($charset) - 7;
1142
1143 // set first character position
1144 $iStartCharNum = 0;
1145
1146 // loop through all characters. count characters and not bytes.
1147 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1148 // encode string from starting character to current character.
1149 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1150
1151 // Check encoded string length
1152 if(strlen($encoded_string)>$iMaxLength) {
1153 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1154 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1155
1156 // set new starting character
1157 $iStartCharNum = $iCharNum-1;
1158
1159 // encode last char (in case it is last character in string)
1160 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1161 } // if string is shorter than max length - add next character
1162 }
1163
1164 // add last encoded string to array
1165 $aRet[] = $encoded_string;
1166
1167 // set initial return string
1168 $sRet = '';
1169
1170 // loop through encoded strings
1171 foreach($aRet as $string) {
1172 // TODO: Do we want to control EOL (end-of-line) marker
1173 if ($sRet!='') $sRet.= " ";
1174
1175 // add header tags and encoded string to return string
1176 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1177 }
1178
1179 return $sRet;
1180 }
1181
1182 /* This function trys to locate the entity_id of a specific mime element */
1183 function find_ent_id($id, $message) {
1184 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1185 if ($message->entities[$i]->header->type0 == 'multipart') {
1186 $ret = find_ent_id($id, $message->entities[$i]);
1187 } else {
1188 if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1189 // if (sq_check_save_extension($message->entities[$i])) {
1190 return $message->entities[$i]->entity_id;
1191 // }
1192 } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1193 /**
1194 * This is part of a fix for Outlook Express 6.x generating
1195 * cid URLs without creating content-id headers
1196 * @@JA - 20050207
1197 */
1198 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1199 return $message->entities[$i]->entity_id;
1200 }
1201 }
1202 }
1203 }
1204 return $ret;
1205 }
1206
1207 function sq_check_save_extension($message) {
1208 $filename = $message->getFilename();
1209 $ext = substr($filename, strrpos($filename,'.')+1);
1210 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1211 return in_array($ext, $save_extensions);
1212 }
1213
1214
1215 /**
1216 ** HTMLFILTER ROUTINES
1217 */
1218
1219 /**
1220 * This function checks attribute values for entity-encoded values
1221 * and returns them translated into 8-bit strings so we can run
1222 * checks on them.
1223 *
1224 * @param $attvalue A string to run entity check against.
1225 * @return Nothing, modifies a reference value.
1226 */
1227 function sq_defang(&$attvalue){
1228 $me = 'sq_defang';
1229 /**
1230 * Skip this if there aren't ampersands or backslashes.
1231 */
1232 if (strpos($attvalue, '&') === false
1233 && strpos($attvalue, '\\') === false){
1234 return;
1235 }
1236 $m = false;
1237 // before deent, translate the dangerous unicode characters and ... to safe values
1238 // otherwise the regular expressions do not match.
1239
1240
1241
1242 do {
1243 $m = false;
1244 $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1245 $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1246 $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1247 } while ($m == true);
1248 $attvalue = stripslashes($attvalue);
1249 }
1250
1251 /**
1252 * Kill any tabs, newlines, or carriage returns. Our friends the
1253 * makers of the browser with 95% market value decided that it'd
1254 * be funny to make "java[tab]script" be just as good as "javascript".
1255 *
1256 * @param attvalue The attribute value before extraneous spaces removed.
1257 * @return attvalue Nothing, modifies a reference value.
1258 */
1259 function sq_unspace(&$attvalue){
1260 $me = 'sq_unspace';
1261 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1262 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1263 Array('', '', '', '', ''), $attvalue);
1264 }
1265 }
1266
1267 /**
1268 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1269 * IE as regular characters.
1270 *
1271 * @param attvalue The attribute value before dangerous characters are translated.
1272 * @return attvalue Nothing, modifies a reference value.
1273 * @author Marc Groot Koerkamp.
1274 */
1275 function sq_fixIE_idiocy(&$attvalue) {
1276 // remove NUL
1277 $attvalue = str_replace("\0", "", $attvalue);
1278 // remove comments
1279 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1280
1281 // IE has the evil habit of accepting every possible value for the attribute expression.
1282 // The table below contains characters which are parsed by IE if they are used in the "expression"
1283 // attribute value.
1284 $aDangerousCharsReplacementTable = array(
1285 array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1286 '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1287 '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1288 '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1289 '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1290 '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1291 '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1292 '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1293 '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1294 '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1295 '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1296 '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1297 '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1298 '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1299 '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1300 '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1301 '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1302 '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1303 '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1304 '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1305 '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1306 '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1307 '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1308 '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1309 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1310 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1311 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1312 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1313 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1314 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1315 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1316 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1317 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1318 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1319 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1320 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1321 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1322 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1323 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1324 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1325 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1326 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1327 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1328 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1329 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1330 "\xCA\x9F", /* L UNICODE IPA Extension */
1331 "\xCA\x80", /* R UNICODE IPA Extension */
1332 "\xC9\xB4"), /* N UNICODE IPA Extension */
1333 array('l', 'l', 'r','r','n','n',
1334 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1335 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1336 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1337 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1338
1339 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1340 // used for XSS.
1341 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1342 }
1343
1344 /**
1345 * This function returns the final tag out of the tag name, an array
1346 * of attributes, and the type of the tag. This function is called by
1347 * sq_sanitize internally.
1348 *
1349 * @param $tagname the name of the tag.
1350 * @param $attary the array of attributes and their values
1351 * @param $tagtype The type of the tag (see in comments).
1352 * @return a string with the final tag representation.
1353 */
1354 function sq_tagprint($tagname, $attary, $tagtype){
1355 $me = 'sq_tagprint';
1356
1357 if ($tagtype == 2){
1358 $fulltag = '</' . $tagname . '>';
1359 } else {
1360 $fulltag = '<' . $tagname;
1361 if (is_array($attary) && sizeof($attary)){
1362 $atts = Array();
1363 while (list($attname, $attvalue) = each($attary)){
1364 array_push($atts, "$attname=$attvalue");
1365 }
1366 $fulltag .= ' ' . join(" ", $atts);
1367 }
1368 if ($tagtype == 3){
1369 $fulltag .= ' /';
1370 }
1371 $fulltag .= '>';
1372 }
1373 return $fulltag;
1374 }
1375
1376 /**
1377 * A small helper function to use with array_walk. Modifies a by-ref
1378 * value and makes it lowercase.
1379 *
1380 * @param $val a value passed by-ref.
1381 * @return void since it modifies a by-ref value.
1382 */
1383 function sq_casenormalize(&$val){
1384 $val = strtolower($val);
1385 }
1386
1387 /**
1388 * This function skips any whitespace from the current position within
1389 * a string and to the next non-whitespace value.
1390 *
1391 * @param $body the string
1392 * @param $offset the offset within the string where we should start
1393 * looking for the next non-whitespace character.
1394 * @return the location within the $body where the next
1395 * non-whitespace char is located.
1396 */
1397 function sq_skipspace($body, $offset){
1398 $me = 'sq_skipspace';
1399 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1400 if (sizeof($matches{1})){
1401 $count = strlen($matches{1});
1402 $offset += $count;
1403 }
1404 return $offset;
1405 }
1406
1407 /**
1408 * This function looks for the next character within a string. It's
1409 * really just a glorified "strpos", except it catches if failures
1410 * nicely.
1411 *
1412 * @param $body The string to look for needle in.
1413 * @param $offset Start looking from this position.
1414 * @param $needle The character/string to look for.
1415 * @return location of the next occurance of the needle, or
1416 * strlen($body) if needle wasn't found.
1417 */
1418 function sq_findnxstr($body, $offset, $needle){
1419 $me = 'sq_findnxstr';
1420 $pos = strpos($body, $needle, $offset);
1421 if ($pos === FALSE){
1422 $pos = strlen($body);
1423 }
1424 return $pos;
1425 }
1426
1427 /**
1428 * This function takes a PCRE-style regexp and tries to match it
1429 * within the string.
1430 *
1431 * @param $body The string to look for needle in.
1432 * @param $offset Start looking from here.
1433 * @param $reg A PCRE-style regex to match.
1434 * @return Returns a false if no matches found, or an array
1435 * with the following members:
1436 * - integer with the location of the match within $body
1437 * - string with whatever content between offset and the match
1438 * - string with whatever it is we matched
1439 */
1440 function sq_findnxreg($body, $offset, $reg){
1441 $me = 'sq_findnxreg';
1442 $matches = Array();
1443 $retarr = Array();
1444 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1445 if (!isset($matches{0}) || !$matches{0}){
1446 $retarr = false;
1447 } else {
1448 $retarr{0} = $offset + strlen($matches{1});
1449 $retarr{1} = $matches{1};
1450 $retarr{2} = $matches{2};
1451 }
1452 return $retarr;
1453 }
1454
1455 /**
1456 * This function looks for the next tag.
1457 *
1458 * @param $body String where to look for the next tag.
1459 * @param $offset Start looking from here.
1460 * @return false if no more tags exist in the body, or
1461 * an array with the following members:
1462 * - string with the name of the tag
1463 * - array with attributes and their values
1464 * - integer with tag type (1, 2, or 3)
1465 * - integer where the tag starts (starting "<")
1466 * - integer where the tag ends (ending ">")
1467 * first three members will be false, if the tag is invalid.
1468 */
1469 function sq_getnxtag($body, $offset){
1470 $me = 'sq_getnxtag';
1471 if ($offset > strlen($body)){
1472 return false;
1473 }
1474 $lt = sq_findnxstr($body, $offset, "<");
1475 if ($lt == strlen($body)){
1476 return false;
1477 }
1478 /**
1479 * We are here:
1480 * blah blah <tag attribute="value">
1481 * \---------^
1482 */
1483 $pos = sq_skipspace($body, $lt+1);
1484 if ($pos >= strlen($body)){
1485 return Array(false, false, false, $lt, strlen($body));
1486 }
1487 /**
1488 * There are 3 kinds of tags:
1489 * 1. Opening tag, e.g.:
1490 * <a href="blah">
1491 * 2. Closing tag, e.g.:
1492 * </a>
1493 * 3. XHTML-style content-less tag, e.g.:
1494 * <img src="blah" />
1495 */
1496 $tagtype = false;
1497 switch (substr($body, $pos, 1)){
1498 case '/':
1499 $tagtype = 2;
1500 $pos++;
1501 break;
1502 case '!':
1503 /**
1504 * A comment or an SGML declaration.
1505 */
1506 if (substr($body, $pos+1, 2) == "--"){
1507 $gt = strpos($body, "-->", $pos);
1508 if ($gt === false){
1509 $gt = strlen($body);
1510 } else {
1511 $gt += 2;
1512 }
1513 return Array(false, false, false, $lt, $gt);
1514 } else {
1515 $gt = sq_findnxstr($body, $pos, ">");
1516 return Array(false, false, false, $lt, $gt);
1517 }
1518 break;
1519 default:
1520 /**
1521 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1522 * later.
1523 */
1524 $tagtype = 1;
1525 break;
1526 }
1527
1528 $tag_start = $pos;
1529 $tagname = '';
1530 /**
1531 * Look for next [\W-_], which will indicate the end of the tag name.
1532 */
1533 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1534 if ($regary == false){
1535 return Array(false, false, false, $lt, strlen($body));
1536 }
1537 list($pos, $tagname, $match) = $regary;
1538 $tagname = strtolower($tagname);
1539
1540 /**
1541 * $match can be either of these:
1542 * '>' indicating the end of the tag entirely.
1543 * '\s' indicating the end of the tag name.
1544 * '/' indicating that this is type-3 xhtml tag.
1545 *
1546 * Whatever else we find there indicates an invalid tag.
1547 */
1548 switch ($match){
1549 case '/':
1550 /**
1551 * This is an xhtml-style tag with a closing / at the
1552 * end, like so: <img src="blah" />. Check if it's followed
1553 * by the closing bracket. If not, then this tag is invalid
1554 */
1555 if (substr($body, $pos, 2) == "/>"){
1556 $pos++;
1557 $tagtype = 3;
1558 } else {
1559 $gt = sq_findnxstr($body, $pos, ">");
1560 $retary = Array(false, false, false, $lt, $gt);
1561 return $retary;
1562 }
1563 case '>':
1564 return Array($tagname, false, $tagtype, $lt, $pos);
1565 break;
1566 default:
1567 /**
1568 * Check if it's whitespace
1569 */
1570 if (!preg_match('/\s/', $match)){
1571 /**
1572 * This is an invalid tag! Look for the next closing ">".
1573 */
1574 $gt = sq_findnxstr($body, $lt, ">");
1575 return Array(false, false, false, $lt, $gt);
1576 }
1577 break;
1578 }
1579
1580 /**
1581 * At this point we're here:
1582 * <tagname attribute='blah'>
1583 * \-------^
1584 *
1585 * At this point we loop in order to find all attributes.
1586 */
1587 $attname = '';
1588 $atttype = false;
1589 $attary = Array();
1590
1591 while ($pos <= strlen($body)){
1592 $pos = sq_skipspace($body, $pos);
1593 if ($pos == strlen($body)){
1594 /**
1595 * Non-closed tag.
1596 */
1597 return Array(false, false, false, $lt, $pos);
1598 }
1599 /**
1600 * See if we arrived at a ">" or "/>", which means that we reached
1601 * the end of the tag.
1602 */
1603 $matches = Array();
1604 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1605 /**
1606 * Yep. So we did.
1607 */
1608 $pos += strlen($matches{1});
1609 if ($matches{2} == "/>"){
1610 $tagtype = 3;
1611 $pos++;
1612 }
1613 return Array($tagname, $attary, $tagtype, $lt, $pos);
1614 }
1615
1616 /**
1617 * There are several types of attributes, with optional
1618 * [:space:] between members.
1619 * Type 1:
1620 * attrname[:space:]=[:space:]'CDATA'
1621 * Type 2:
1622 * attrname[:space:]=[:space:]"CDATA"
1623 * Type 3:
1624 * attr[:space:]=[:space:]CDATA
1625 * Type 4:
1626 * attrname
1627 *
1628 * We leave types 1 and 2 the same, type 3 we check for
1629 * '"' and convert to "&quot" if needed, then wrap in
1630 * double quotes. Type 4 we convert into:
1631 * attrname="yes".
1632 */
1633 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1634 if ($regary == false){
1635 /**
1636 * Looks like body ended before the end of tag.
1637 */
1638 return Array(false, false, false, $lt, strlen($body));
1639 }
1640 list($pos, $attname, $match) = $regary;
1641 $attname = strtolower($attname);
1642 /**
1643 * We arrived at the end of attribute name. Several things possible
1644 * here:
1645 * '>' means the end of the tag and this is attribute type 4
1646 * '/' if followed by '>' means the same thing as above
1647 * '\s' means a lot of things -- look what it's followed by.
1648 * anything else means the attribute is invalid.
1649 */
1650 switch($match){
1651 case '/':
1652 /**
1653 * This is an xhtml-style tag with a closing / at the
1654 * end, like so: <img src="blah" />. Check if it's followed
1655 * by the closing bracket. If not, then this tag is invalid
1656 */
1657 if (substr($body, $pos, 2) == "/>"){
1658 $pos++;
1659 $tagtype = 3;
1660 } else {
1661 $gt = sq_findnxstr($body, $pos, ">");
1662 $retary = Array(false, false, false, $lt, $gt);
1663 return $retary;
1664 }
1665 case '>':
1666 $attary{$attname} = '"yes"';
1667 return Array($tagname, $attary, $tagtype, $lt, $pos);
1668 break;
1669 default:
1670 /**
1671 * Skip whitespace and see what we arrive at.
1672 */
1673 $pos = sq_skipspace($body, $pos);
1674 $char = substr($body, $pos, 1);
1675 /**
1676 * Two things are valid here:
1677 * '=' means this is attribute type 1 2 or 3.
1678 * \w means this was attribute type 4.
1679 * anything else we ignore and re-loop. End of tag and
1680 * invalid stuff will be caught by our checks at the beginning
1681 * of the loop.
1682 */
1683 if ($char == "="){
1684 $pos++;
1685 $pos = sq_skipspace($body, $pos);
1686 /**
1687 * Here are 3 possibilities:
1688 * "'" attribute type 1
1689 * '"' attribute type 2
1690 * everything else is the content of tag type 3
1691 */
1692 $quot = substr($body, $pos, 1);
1693 if ($quot == "'"){
1694 $regary = sq_findnxreg($body, $pos+1, "\'");
1695 if ($regary == false){
1696 return Array(false, false, false, $lt, strlen($body));
1697 }
1698 list($pos, $attval, $match) = $regary;
1699 $pos++;
1700 $attary{$attname} = "'" . $attval . "'";
1701 } else if ($quot == '"'){
1702 $regary = sq_findnxreg($body, $pos+1, '\"');
1703 if ($regary == false){
1704 return Array(false, false, false, $lt, strlen($body));
1705 }
1706 list($pos, $attval, $match) = $regary;
1707 $pos++;
1708 $attary{$attname} = '"' . $attval . '"';
1709 } else {
1710 /**
1711 * These are hateful. Look for \s, or >.
1712 */
1713 $regary = sq_findnxreg($body, $pos, "[\s>]");
1714 if ($regary == false){
1715 return Array(false, false, false, $lt, strlen($body));
1716 }
1717 list($pos, $attval, $match) = $regary;
1718 /**
1719 * If it's ">" it will be caught at the top.
1720 */
1721 $attval = preg_replace("/\"/s", "&quot;", $attval);
1722 $attary{$attname} = '"' . $attval . '"';
1723 }
1724 } else if (preg_match("|[\w/>]|", $char)) {
1725 /**
1726 * That was attribute type 4.
1727 */
1728 $attary{$attname} = '"yes"';
1729 } else {
1730 /**
1731 * An illegal character. Find next '>' and return.
1732 */
1733 $gt = sq_findnxstr($body, $pos, ">");
1734 return Array(false, false, false, $lt, $gt);
1735 }
1736 break;
1737 }
1738 }
1739 /**
1740 * The fact that we got here indicates that the tag end was never
1741 * found. Return invalid tag indication so it gets stripped.
1742 */
1743 return Array(false, false, false, $lt, strlen($body));
1744 }
1745
1746 /**
1747 * Translates entities into literal values so they can be checked.
1748 *
1749 * @param $attvalue the by-ref value to check.
1750 * @param $regex the regular expression to check against.
1751 * @param $hex whether the entites are hexadecimal.
1752 * @return True or False depending on whether there were matches.
1753 */
1754 function sq_deent(&$attvalue, $regex, $hex=false){
1755 $me = 'sq_deent';
1756 $ret_match = false;
1757 // remove comments
1758 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1759 preg_match_all($regex, $attvalue, $matches);
1760 if (is_array($matches) && sizeof($matches[0]) > 0){
1761 $repl = Array();
1762 for ($i = 0; $i < sizeof($matches[0]); $i++){
1763 $numval = $matches[1][$i];
1764 if ($hex){
1765 $numval = hexdec($numval);
1766 }
1767 $repl{$matches[0][$i]} = chr($numval);
1768 }
1769 $attvalue = strtr($attvalue, $repl);
1770 return true;
1771 } else {
1772 return false;
1773 }
1774 }
1775
1776 /**
1777 * This function runs various checks against the attributes.
1778 *
1779 * @param $tagname String with the name of the tag.
1780 * @param $attary Array with all tag attributes.
1781 * @param $rm_attnames See description for sq_sanitize
1782 * @param $bad_attvals See description for sq_sanitize
1783 * @param $add_attr_to_tag See description for sq_sanitize
1784 * @param $message message object
1785 * @param $id message id
1786 * @return Array with modified attributes.
1787 */
1788 function sq_fixatts($tagname,
1789 $attary,
1790 $rm_attnames,
1791 $bad_attvals,
1792 $add_attr_to_tag,
1793 $message,
1794 $id,
1795 $mailbox
1796 ){
1797 $me = 'sq_fixatts';
1798 while (list($attname, $attvalue) = each($attary)){
1799 /**
1800 * See if this attribute should be removed.
1801 */
1802 foreach ($rm_attnames as $matchtag=>$matchattrs){
1803 if (preg_match($matchtag, $tagname)){
1804 foreach ($matchattrs as $matchattr){
1805 if (preg_match($matchattr, $attname)){
1806 unset($attary{$attname});
1807 continue;
1808 }
1809 }
1810 }
1811 }
1812 /**
1813 * Workaround for IE quirks
1814 */
1815 sq_fixIE_idiocy($attvalue);
1816
1817 /**
1818 * Remove any backslashes, entities, and extraneous whitespace.
1819 */
1820
1821 $oldattvalue = $attvalue;
1822 sq_defang($attvalue);
1823 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1824 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1825 // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1826 $attvalue = "idiocy";
1827 $attary{$attname} = $attvalue;
1828 }
1829 sq_unspace($attvalue);
1830
1831 /**
1832 * Now let's run checks on the attvalues.
1833 * I don't expect anyone to comprehend this. If you do,
1834 * get in touch with me so I can drive to where you live and
1835 * shake your hand personally. :)
1836 */
1837 foreach ($bad_attvals as $matchtag=>$matchattrs){
1838 if (preg_match($matchtag, $tagname)){
1839 foreach ($matchattrs as $matchattr=>$valary){
1840 if (preg_match($matchattr, $attname)){
1841 /**
1842 * There are two arrays in valary.
1843 * First is matches.
1844 * Second one is replacements
1845 */
1846 list($valmatch, $valrepl) = $valary;
1847 $newvalue =
1848 preg_replace($valmatch, $valrepl, $attvalue);
1849 if ($newvalue != $attvalue){
1850 $attary{$attname} = $newvalue;
1851 $attvalue = $newvalue;
1852 }
1853 }
1854 }
1855 }
1856 }
1857 if ($attname == 'style') {
1858 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1859 // 8bit and control characters in style attribute values can be used for XSS, remove them
1860 $attary{$attname} = '"disallowed character"';
1861 }
1862 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1863 if (count($aMatch)) {
1864 foreach($aMatch[1] as $sMatch) {
1865 // url value
1866 $urlvalue = $sMatch;
1867 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1868 $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1869 }
1870 }
1871 }
1872 /**
1873 * Use white list based filtering on attributes which can contain url's
1874 */
1875 else if ($attname == 'href' || $attname == 'xlink:href' || $attname == 'src'
1876 || $attname == 'poster' || $attname == 'formaction'
1877 || $attname == 'background' || $attname == 'action') {
1878 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1879 $attary{$attname} = $attvalue;
1880 }
1881 }
1882 /**
1883 * See if we need to append any attributes to this tag.
1884 */
1885 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1886 if (preg_match($matchtag, $tagname)){
1887 $attary = array_merge($attary, $addattary);
1888 }
1889 }
1890 return $attary;
1891 }
1892
1893 /**
1894 * This function filters url's
1895 *
1896 * @param $attvalue String with attribute value to filter
1897 * @param $message message object
1898 * @param $id message id
1899 * @param $mailbox mailbox
1900 * @param $sQuote quoting characters around url's
1901 */
1902 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1903 $attvalue = trim($attvalue);
1904 if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1905 // remove the double quotes
1906 $sQuote = $attvalue[0];
1907 $attvalue = trim(substr($attvalue,1,-1));
1908 }
1909
1910 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1911 // images off by default.
1912 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1913
1914 global $use_transparent_security_image;
1915 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1916 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1917
1918 /**
1919 * Replace empty src tags with the blank image. src is only used
1920 * for frames, images, and image inputs. Doing a replace should
1921 * not affect them working as should be, however it will stop
1922 * IE from being kicked off when src for img tags are not set
1923 */
1924 if ($attvalue == '') {
1925 $attvalue = '"' . SM_PATH . 'images/blank.png"';
1926 } else {
1927 // first, disallow 8 bit characters and control characters
1928 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1929 switch ($attname) {
1930 case 'href':
1931 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1932 break;
1933 default:
1934 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1935 break;
1936 }
1937 } else {
1938 $aUrl = parse_url($attvalue);
1939 if (isset($aUrl['scheme'])) {
1940 switch(strtolower($aUrl['scheme'])) {
1941 case 'mailto':
1942 case 'http':
1943 case 'https':
1944 case 'ftp':
1945 if ($attname != 'href') {
1946 if ($view_unsafe_images == false) {
1947 $attvalue = $sQuote . $secremoveimg . $sQuote;
1948 } else {
1949 if (isset($aUrl['path'])) {
1950
1951 // No one has been able to show that image URIs
1952 // can be exploited, so for now, no restrictions
1953 // are made at all. If this proves to be a problem,
1954 // the commented-out code below can be of help.
1955 // (One consideration is that I see nothing in this
1956 // function that specifically says that we will
1957 // only ever arrive here when inspecting an image
1958 // tag, although that does seem to be the end
1959 // result - e.g., <script src="..."> where malicious
1960 // image URIs are in fact a problem are already
1961 // filtered out elsewhere.
1962 /* ---------------------------------
1963 // validate image extension.
1964 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1965 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1966 // If URI is to something other than
1967 // a regular image file, get the contents
1968 // and try to see if it is an image.
1969 // Don't use Fileinfo (finfo_file()) because
1970 // we'd need to make the admin configure the
1971 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1972 //
1973 $mime_type = '';
1974 if (function_exists('mime_content_type')
1975 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1976
1977 // fetch file
1978 //
1979 $file_contents = '';
1980 while (!feof($FILE)) {
1981 $file_contents .= fread($FILE, 8192);
1982 }
1983 fclose($FILE);
1984
1985 // store file locally
1986 //
1987 global $attachment_dir, $username;
1988 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1989 $localfilename = GenerateRandomString(32, '', 7);
1990 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1991 while (file_exists($full_localfilename)) {
1992 $localfilename = GenerateRandomString(32, '', 7);
1993 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1994 }
1995 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
1996 fwrite($FILE, $file_contents);
1997 fclose($FILE);
1998
1999 // get mime type and remove file
2000 //
2001 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2002 unlink("$hashed_attachment_dir/$localfilename");
2003 }
2004 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2005 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2006 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2007 }
2008 }
2009 --------------------------------- */
2010 } else {
2011 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2012 }
2013 }
2014 } else {
2015 $attvalue = $sQuote . $attvalue . $sQuote;
2016 }
2017 break;
2018 case 'outbind':
2019 /**
2020 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2021 * One day MS might actually make it match something useful, for now, falling
2022 * back to using cid2http, so we can grab the blank.png.
2023 */
2024 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2025 break;
2026 case 'cid':
2027 /**
2028 * Turn cid: urls into http-friendly ones.
2029 */
2030 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2031 break;
2032 default:
2033 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2034 break;
2035 }
2036 } else {
2037 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2038 // parse_url did not lead to satisfying result
2039 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2040 }
2041 }
2042 }
2043 }
2044 }
2045
2046 /**
2047 * This function edits the style definition to make them friendly and
2048 * usable in SquirrelMail.
2049 *
2050 * @param $message the message object
2051 * @param $id the message id
2052 * @param $content a string with whatever is between <style> and </style>
2053 * @param $mailbox the message mailbox
2054 * @return a string with edited content.
2055 */
2056 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2057 $me = 'sq_fixstyle';
2058 // workaround for </style> in between comments
2059 $iCurrentPos = $pos;
2060 $content = '';
2061 $sToken = '';
2062 $bSucces = false;
2063 $bEndTag = false;
2064 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2065 $char = $body{$i};
2066 switch ($char) {
2067 case '<':
2068 $sToken = $char;
2069 break;
2070 case '/':
2071 if ($sToken == '<') {
2072 $sToken .= $char;
2073 $bEndTag = true;
2074 } else {
2075 $content .= $char;
2076 }
2077 break;
2078 case '>':
2079 if ($bEndTag) {
2080 $sToken .= $char;
2081 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2082 $newpos = $i + 1;
2083 $bSucces = true;
2084 break 2;
2085 } else {
2086 $content .= $sToken;
2087 }
2088 $bEndTag = false;
2089 } else {
2090 $content .= $char;
2091 }
2092 break;
2093 case '!':
2094 if ($sToken == '<') {
2095 // possible comment
2096 if (isset($body{$i+2}) && substr($body,$i,3) == '!--') {
2097 $i = strpos($body,'-->',$i+3);
2098 if ($i === false) { // no end comment
2099 $i = strlen($body);
2100 }
2101 $sToken = '';
2102 }
2103 } else {
2104 $content .= $char;
2105 }
2106 break;
2107 default:
2108 if ($bEndTag) {
2109 $sToken .= $char;
2110 } else {
2111 $content .= $char;
2112 }
2113 break;
2114 }
2115 }
2116 if ($bSucces == FALSE){
2117 return array(FALSE, strlen($body));
2118 }
2119
2120
2121
2122 /**
2123 * First look for general BODY style declaration, which would be
2124 * like so:
2125 * body {background: blah-blah}
2126 * and change it to .bodyclass so we can just assign it to a <div>
2127 */
2128 $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2129
2130 global $use_transparent_security_image;
2131 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2132 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2133
2134 /**
2135 * Fix url('blah') declarations.
2136 */
2137 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2138 // "url(\\1$secremoveimg\\2)", $content);
2139
2140 // first check for 8bit sequences and disallowed control characters
2141 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2142 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2143 return array($content, $newpos);
2144 }
2145
2146 // IE Sucks hard. We have a special function for it.
2147 sq_fixIE_idiocy($content);
2148
2149 // remove @import line
2150 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2151
2152 // translate ur\l and variations (IE parses that)
2153 // TODO check if the sq_fixIE_idiocy function already handles this.
2154 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2155 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2156 if (count($aMatch)) {
2157 $aValue = $aReplace = array();
2158 foreach($aMatch[1] as $sMatch) {
2159 // url value
2160 $urlvalue = $sMatch;
2161 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2162 $aValue[] = $sMatch;
2163 $aReplace[] = $urlvalue;
2164 }
2165 $content = str_replace($aValue,$aReplace,$content);
2166 }
2167
2168 /**
2169 * Remove any backslashes, entities, and extraneous whitespace.
2170 */
2171 $contentTemp = $content;
2172 sq_defang($contentTemp);
2173 sq_unspace($contentTemp);
2174
2175 /**
2176 * Fix stupid css declarations which lead to vulnerabilities
2177 * in IE.
2178 *
2179 * Also remove "position" attribute, as it can easily be set
2180 * to "fixed" or "absolute" with "left" and "top" attributes
2181 * of zero, taking over the whole content frame. It can also
2182 * be set to relative and move itself anywhere it wants to,
2183 * displaying content in areas it shouldn't be allowed to touch.
2184 */
2185 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2186 '/expression/i',
2187 '/behaviou*r/i',
2188 '/binding/i',
2189 '/include-source/i',
2190 '/javascript/i',
2191 '/script/i',
2192 '/position/i');
2193 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2194 $contentNew = preg_replace($match, $replace, $contentTemp);
2195 if ($contentNew !== $contentTemp) {
2196 // insecure css declarations are used. From now on we don't care
2197 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2198 $content = $contentNew;
2199 }
2200 return array($content, $newpos);
2201 }
2202
2203
2204 /**
2205 * This function converts cid: url's into the ones that can be viewed in
2206 * the browser.
2207 *
2208 * @param $message the message object
2209 * @param $id the message id
2210 * @param $cidurl the cid: url.
2211 * @param $mailbox the message mailbox
2212 * @return a string with a http-friendly url
2213 */
2214 function sq_cid2http($message, $id, $cidurl, $mailbox){
2215 /**
2216 * Get rid of quotes.
2217 */
2218 $quotchar = substr($cidurl, 0, 1);
2219 if ($quotchar == '"' || $quotchar == "'"){
2220 $cidurl = str_replace($quotchar, "", $cidurl);
2221 } else {
2222 $quotchar = '';
2223 }
2224 $cidurl = substr(trim($cidurl), 4);
2225
2226 $match_str = '/\{.*?\}\//';
2227 $str_rep = '';
2228 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2229
2230 $linkurl = find_ent_id($cidurl, $message);
2231 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2232 unsafe link image */
2233 $httpurl = '';
2234
2235 /**
2236 * This is part of a fix for Outlook Express 6.x generating
2237 * cid URLs without creating content-id headers. These images are
2238 * not part of the multipart/related html mail. The html contains
2239 * <img src="cid:{some_id}/image_filename.ext"> references to
2240 * attached images with as goal to render them inline although
2241 * the attachment disposition property is not inline.
2242 */
2243
2244 if (empty($linkurl)) {
2245 if (preg_match('/{.*}\//', $cidurl)) {
2246 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2247 if (!empty($cidurl)) {
2248 $linkurl = find_ent_id($cidurl, $message);
2249 }
2250 }
2251 }
2252
2253 if (!empty($linkurl)) {
2254 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2255 "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2256 '&amp;ent_id=' . $linkurl . $quotchar;
2257 } else {
2258 /**
2259 * If we couldn't generate a proper img url, drop in a blank image
2260 * instead of sending back empty, otherwise it causes unusual behaviour
2261 */
2262 $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2263 }
2264
2265 return $httpurl;
2266 }
2267
2268 /**
2269 * This function changes the <body> tag into a <div> tag since we
2270 * can't really have a body-within-body.
2271 *
2272 * @param $attary an array of attributes and values of <body>
2273 * @param $mailbox mailbox we're currently reading (for cid2http)
2274 * @param $message current message (for cid2http)
2275 * @param $id current message id (for cid2http)
2276 * @return a modified array of attributes to be set for <div>
2277 */
2278 function sq_body2div($attary, $mailbox, $message, $id){
2279 $me = 'sq_body2div';
2280 $divattary = Array('class' => "'bodyclass'");
2281 $text = '#000000';
2282 $has_bgc_stl = $has_txt_stl = false;
2283 $styledef = '';
2284 if (is_array($attary) && sizeof($attary) > 0){
2285 foreach ($attary as $attname=>$attvalue){
2286 $quotchar = substr($attvalue, 0, 1);
2287 $attvalue = str_replace($quotchar, "", $attvalue);
2288 switch ($attname){
2289 case 'background':
2290 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2291 $styledef .= "background-image: url('$attvalue'); ";
2292 break;
2293 case 'bgcolor':
2294 $has_bgc_stl = true;
2295 $styledef .= "background-color: $attvalue; ";
2296 break;
2297 case 'text':
2298 $has_txt_stl = true;
2299 $styledef .= "color: $attvalue; ";
2300 break;
2301 }
2302 }
2303 // Outlook defines a white bgcolor and no text color. This can lead to
2304 // white text on a white bg with certain themes.
2305 if ($has_bgc_stl && !$has_txt_stl) {
2306 $styledef .= "color: $text; ";
2307 }
2308 if (strlen($styledef) > 0){
2309 $divattary{"style"} = "\"$styledef\"";
2310 }
2311 }
2312 return $divattary;
2313 }
2314
2315 /**
2316 * This is the main function and the one you should actually be calling.
2317 * There are several variables you should be aware of an which need
2318 * special description.
2319 *
2320 * Since the description is quite lengthy, see it here:
2321 * http://linux.duke.edu/projects/mini/htmlfilter/
2322 *
2323 * @param $body the string with HTML you wish to filter
2324 * @param $tag_list see description above
2325 * @param $rm_tags_with_content see description above
2326 * @param $self_closing_tags see description above
2327 * @param $force_tag_closing see description above
2328 * @param $rm_attnames see description above
2329 * @param $bad_attvals see description above
2330 * @param $add_attr_to_tag see description above
2331 * @param $message message object
2332 * @param $id message id
2333 * @return sanitized html safe to show on your pages.
2334 */
2335 function sq_sanitize($body,
2336 $tag_list,
2337 $rm_tags_with_content,
2338 $self_closing_tags,
2339 $force_tag_closing,
2340 $rm_attnames,
2341 $bad_attvals,
2342 $add_attr_to_tag,
2343 $message,
2344 $id,
2345 $mailbox
2346 ){
2347 $me = 'sq_sanitize';
2348 $rm_tags = array_shift($tag_list);
2349 /**
2350 * Normalize rm_tags and rm_tags_with_content.
2351 */
2352 @array_walk($tag_list, 'sq_casenormalize');
2353 @array_walk($rm_tags_with_content, 'sq_casenormalize');
2354 @array_walk($self_closing_tags, 'sq_casenormalize');
2355 /**
2356 * See if tag_list is of tags to remove or tags to allow.
2357 * false means remove these tags
2358 * true means allow these tags
2359 */
2360 $curpos = 0;
2361 $open_tags = Array();
2362 $trusted = "\n<!-- begin sanitized html -->\n";
2363 $skip_content = false;
2364 /**
2365 * Take care of netscape's stupid javascript entities like
2366 * &{alert('boo')};
2367 */
2368 $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2369
2370 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2371 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2372 $free_content = substr($body, $curpos, $lt-$curpos);
2373 /**
2374 * Take care of <style>
2375 */
2376 if ($tagname == "style" && $tagtype == 1){
2377 list($free_content, $curpos) =
2378 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2379 if ($free_content != FALSE){
2380 if ( !empty($attary) ) {
2381 $attary = sq_fixatts($tagname,
2382 $attary,
2383 $rm_attnames,
2384 $bad_attvals,
2385 $add_attr_to_tag,
2386 $message,
2387 $id,
2388 $mailbox
2389 );
2390 }
2391 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2392 $trusted .= $free_content;
2393 $trusted .= sq_tagprint($tagname, false, 2);
2394 }
2395 continue;
2396 }
2397 if ($skip_content == false){
2398 $trusted .= $free_content;
2399 }
2400 if ($tagname != FALSE){
2401 if ($tagtype == 2){
2402 if ($skip_content == $tagname){
2403 /**
2404 * Got to the end of tag we needed to remove.
2405 */
2406 $tagname = false;
2407 $skip_content = false;
2408 } else {
2409 if ($skip_content == false){
2410 if ($tagname == "body"){
2411 $tagname = "div";
2412 }
2413 if (isset($open_tags{$tagname}) &&
2414 $open_tags{$tagname} > 0){
2415 $open_tags{$tagname}--;
2416 } else {
2417 $tagname = false;
2418 }
2419 }
2420 }
2421 } else {
2422 /**
2423 * $rm_tags_with_content
2424 */
2425 if ($skip_content == false){
2426 /**
2427 * See if this is a self-closing type and change
2428 * tagtype appropriately.
2429 */
2430 if ($tagtype == 1
2431 && in_array($tagname, $self_closing_tags)){
2432 $tagtype = 3;
2433 }
2434 /**
2435 * See if we should skip this tag and any content
2436 * inside it.
2437 */
2438 if ($tagtype == 1 &&
2439 in_array($tagname, $rm_tags_with_content)){
2440 $skip_content = $tagname;
2441 } else {
2442 if (($rm_tags == false
2443 && in_array($tagname, $tag_list)) ||
2444 ($rm_tags == true &&
2445 !in_array($tagname, $tag_list))){
2446 $tagname = false;
2447 } else {
2448 /**
2449 * Convert body into div.
2450 */
2451 if ($tagname == "body"){
2452 $tagname = "div";
2453 $attary = sq_body2div($attary, $mailbox,
2454 $message, $id);
2455 }
2456 if ($tagtype == 1){
2457 if (isset($open_tags{$tagname})){
2458 $open_tags{$tagname}++;
2459 } else {
2460 $open_tags{$tagname}=1;
2461 }
2462 }
2463 /**
2464 * This is where we run other checks.
2465 */
2466 if (is_array($attary) && sizeof($attary) > 0){
2467 $attary = sq_fixatts($tagname,
2468 $attary,
2469 $rm_attnames,
2470 $bad_attvals,
2471 $add_attr_to_tag,
2472 $message,
2473 $id,
2474 $mailbox
2475 );
2476 }
2477 }
2478 }
2479 }
2480 }
2481 if ($tagname != false && $skip_content == false){
2482 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2483 }
2484 }
2485 $curpos = $gt+1;
2486 }
2487 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2488 if ($force_tag_closing == true){
2489 foreach ($open_tags as $tagname=>$opentimes){
2490 while ($opentimes > 0){
2491 $trusted .= '</' . $tagname . '>';
2492 $opentimes--;
2493 }
2494 }
2495 $trusted .= "\n";
2496 }
2497 $trusted .= "<!-- end sanitized html -->\n";
2498 return $trusted;
2499 }
2500
2501 /**
2502 * This is a wrapper function to call html sanitizing routines.
2503 *
2504 * @param $body the body of the message
2505 * @param $id the id of the message
2506
2507 * @param $message
2508 * @param $mailbox
2509 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2510 * into internal SM compose links
2511 * (optional; default = TRUE)
2512 * @return a string with html safe to display in the browser.
2513 */
2514 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2515
2516 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2517
2518 global $attachment_common_show_images, $view_unsafe_images,
2519 $has_unsafe_images, $block_svg_display;
2520 /**
2521 * Don't display attached images in HTML mode.
2522 *
2523 * SB: why?
2524 */
2525 $attachment_common_show_images = false;
2526 $tag_list = Array(
2527 false,
2528 "meta",
2529 "html",
2530 "head",
2531 "base",
2532 "link",
2533 "frame",
2534 "iframe",
2535 "plaintext",
2536 "marquee",
2537 );
2538
2539 $rm_tags_with_content = Array(
2540 "script",
2541 "object",
2542 "applet",
2543 "embed",
2544 "title",
2545 "frameset",
2546 "xmp",
2547 "xml",
2548 );
2549 if ($block_svg_display)
2550 $rm_tags_with_content[] = 'svg';
2551
2552 $self_closing_tags = Array(
2553 "img",
2554 "br",
2555 "hr",
2556 "input",
2557 "outbind",
2558 );
2559
2560 $force_tag_closing = true;
2561
2562 $rm_attnames = Array(
2563 "/.*/" =>
2564 Array(
2565 "/target/i",
2566 "/^on.*/i",
2567 "/^dynsrc/i",
2568 "/^data.*/i",
2569 "/^lowsrc.*/i",
2570 )
2571 );
2572
2573 global $use_transparent_security_image;
2574 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2575 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2576
2577 $bad_attvals = Array(
2578 "/.*/" =>
2579 Array(
2580 "/^src|background/i" =>
2581 Array(
2582 Array(
2583 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2584 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2585 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2586 ),
2587 Array(
2588 "\\1$secremoveimg\\2",
2589 "\\1$secremoveimg\\2",
2590 "\\1$secremoveimg\\2",
2591 )
2592 ),
2593 "/^href|action/i" =>
2594 Array(
2595 Array(
2596 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2597 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2598 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2599 ),
2600 Array(
2601 "\\1#\\1",
2602 "\\1#\\1",
2603 "\\1#\\1"
2604 )
2605 ),
2606 "/^style/i" =>
2607 Array(
2608 Array(
2609 "/\/\*.*\*\//",
2610 "/expression/i",
2611 "/binding/i",
2612 "/behaviou*r/i",
2613 "/include-source/i",
2614
2615 // position:relative can also be exploited
2616 // to put content outside of email body area
2617 // and position:fixed is similarly exploitable
2618 // as position:absolute, so we'll remove it
2619 // altogether....
2620 //
2621 // Does this screw up legitimate HTML messages?
2622 // If so, the only fix I see is to allow position
2623 // attributes (any values? I think we still have
2624 // to block static and fixed) only if $use_iframe
2625 // is enabled (1.5.0+)
2626 //
2627 // was: "/position\s*:\s*absolute/i",
2628 //
2629 "/position\s*:/i",
2630
2631 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2632 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2633 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2634 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2635 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2636 ),
2637 Array(
2638 "",
2639 "idiocy",
2640 "idiocy",
2641 "idiocy",
2642 "idiocy",
2643 "idiocy",
2644 "url",
2645 "url(\\1#\\1)",
2646 "url(\\1#\\1)",
2647 "url(\\1#\\1)",
2648 "\\1:url(\\2#\\3)"
2649 )
2650 )
2651 )
2652 );
2653
2654 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2655 // images off by default.
2656 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2657
2658 if (!$view_unsafe_images){
2659 /**
2660 * Remove any references to http/https if view_unsafe_images set
2661 * to false.
2662 */
2663 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2664 '/^([\'\"])\s*https*:.*([\'\"])/si');
2665 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2666 "\\1$secremoveimg\\1");
2667 array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2668 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2669 array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2670 "url(\\1$secremoveimg\\1)");
2671 }
2672
2673 $add_attr_to_tag = Array(
2674 "/^a$/i" =>
2675 Array('target'=>'"_blank"',
2676 'title'=>'"'._("This external link will open in a new window").'"'
2677 )
2678 );
2679 $trusted = sq_sanitize($body,
2680 $tag_list,
2681 $rm_tags_with_content,
2682 $self_closing_tags,
2683 $force_tag_closing,
2684 $rm_attnames,
2685 $bad_attvals,
2686 $add_attr_to_tag,
2687 $message,
2688 $id,
2689 $mailbox
2690 );
2691 if (strpos($trusted,$secremoveimg)){
2692 $has_unsafe_images = true;
2693 }
2694
2695 // we want to parse mailto's in HTML output, change to SM compose links
2696 // this is a modified version of code from url_parser.php... but Marc is
2697 // right: we need a better filtering implementation; adding this randomly
2698 // here is not a great solution
2699 //
2700 if ($take_mailto_links) {
2701 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2702 global $MailTo_PReg_Match;
2703 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2704 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2705 // quote before mailto, and we'll also try to convert the non-links back into links
2706 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2707 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2708 foreach ($regs[0] as $i => $mailto_before) {
2709 $mailto_params = $regs[11][$i];
2710
2711 // get rid of any leading quote we may have captured but don't care about
2712 //
2713 $mailto_before = ltrim($mailto_before, '"\'');
2714
2715 // get rid of any tailing quote since we have to add send_to to the end
2716 //
2717 $mailto_before = rtrim($mailto_before, '"\'');
2718 $mailto_params = rtrim($mailto_params, '"\'');
2719
2720 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2721 $to = 'to=' . $regs[2][$i];
2722 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2723 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2724 else {
2725 if ($mailto_params) //already some params, append to them
2726 $mailto_params .= '&amp;' . $to;
2727 else
2728 $mailto_params .= '?' . $to;
2729 }
2730 }
2731
2732 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2733
2734 // we'll already have target=_blank, no need to allow comp_in_new
2735 // here (which would be a lot more work anyway)
2736 //
2737 global $compose_new_win;
2738 $temp_comp_in_new = $compose_new_win;
2739 $compose_new_win = 0;
2740 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2741 $compose_new_win = $temp_comp_in_new;
2742
2743 // remove <a href=" and anything after the next quote (we only
2744 // need the uri, not the link HTML) in compose uri
2745 //
2746 // but only do this if the original mailto was in a real anchor tag
2747 //
2748 if (!empty($regs[1][$i])) {
2749 $comp_uri = substr($comp_uri, 9);
2750 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2751 }
2752 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2753 }
2754 }
2755 }
2756
2757 return $trusted;
2758 }
2759
2760 /**
2761 * function SendDownloadHeaders - send file to the browser
2762 *
2763 * Original Source: SM core src/download.php
2764 * moved here to make it available to other code, and separate
2765 * front end from back end functionality.
2766 *
2767 * @param string $type0 first half of mime type
2768 * @param string $type1 second half of mime type
2769 * @param string $filename filename to tell the browser for downloaded file
2770 * @param boolean $force whether to force the download dialog to pop
2771 * @param optional integer $filesize send the Content-Header and length to the browser
2772 * @return void
2773 */
2774 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2775 global $languages, $squirrelmail_language;
2776 $isIE = $isIE6plus = false;
2777
2778 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2779
2780 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2781 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2782 $isIE = true;
2783 }
2784
2785 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2786 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2787 $isIE6plus = true;
2788 }
2789
2790 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2791 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2792 $filename =
2793 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2794 } else {
2795 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2796 }
2797
2798 // A Pox on Microsoft and it's Internet Explorer!
2799 //
2800 // IE has lots of bugs with file downloads.
2801 // It also has problems with SSL. Both of these cause problems
2802 // for us in this function.
2803 //
2804 // See this article on Cache Control headers and SSL
2805 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2806 //
2807 // The best thing you can do for IE is to upgrade to the latest
2808 // version
2809 //set all the Cache Control Headers for IE
2810 if ($isIE) {
2811 $filename=rawurlencode($filename);
2812 header ("Pragma: public");
2813 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2814 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2815 // header ("Cache-Control: post-check=0, pre-check=0", false);
2816 header ("Cache-Control: private");
2817
2818 //set the inline header for IE, we'll add the attachment header later if we need it
2819 header ("Content-Disposition: inline; filename=$filename");
2820 }
2821
2822 if (!$force) {
2823 // Try to show in browser window
2824 header ("Content-Disposition: inline; filename=\"$filename\"");
2825 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2826 } else {
2827 // Try to pop up the "save as" box
2828
2829 // IE makes this hard. It pops up 2 save boxes, or none.
2830 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2831 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2832 // But, according to Microsoft, it is "RFC compliant but doesn't
2833 // take into account some deviations that allowed within the
2834 // specification." Doesn't that mean RFC non-compliant?
2835 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2836
2837 // all browsers need the application/octet-stream header for this
2838 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2839
2840 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2841 // Do not have quotes around filename, but that applied to
2842 // "attachment"... does it apply to inline too?
2843 header ("Content-Disposition: attachment; filename=\"$filename\"");
2844
2845 if ($isIE && !$isIE6plus) {
2846 // This combination seems to work mostly. IE 5.5 SP 1 has
2847 // known issues (see the Microsoft Knowledge Base)
2848
2849 // This works for most types, but doesn't work with Word files
2850 header ("Content-Type: application/download; name=\"$filename\"");
2851 header ("Content-Type: application/force-download; name=\"$filename\"");
2852 // These are spares, just in case. :-)
2853 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2854 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2855 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2856 } else if ($isIE) {
2857 // This is to prevent IE for MIME sniffing and auto open a file in IE
2858 header ("Content-Type: application/force-download; name=\"$filename\"");
2859 } else {
2860 // another application/octet-stream forces download for Netscape
2861 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2862 }
2863 }
2864
2865 //send the content-length header if the calling function provides it
2866 if ($filesize > 0) {
2867 header("Content-Length: $filesize");
2868 }
2869
2870 } // end fn SendDownloadHeaders