Stop using curly braces
[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 1999-2021 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 &$type0, &$type1);
603 do_hook("attachment $type0/$type1", $temp);
604 /* The API for this hook has changed as of 1.5.2 so that all plugin
605 arguments are passed in an array instead of each their own plugin
606 argument, and arguments are passed by reference, so instead of
607 returning any changes, changes should simply be made to the original
608 arguments themselves. */
609 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
610 &$defaultlink, &$display_filename, &$where, &$what,
611 &$type0, &$type1);
612 // Do not let a generic plugin change the default link if a more
613 // specialized one already did it...
614 if ($defaultlink != $defaultlink_orig) {
615 $dummy = '';
616 $temp[5] = &$dummy;
617 }
618 do_hook("attachment $type0/*", $temp);
619 /* The API for this hook has changed as of 1.5.2 so that all plugin
620 arguments are passed in an array instead of each their own plugin
621 argument, and arguments are passed by reference, so instead of
622 returning any changes, changes should simply be made to the original
623 arguments themselves. */
624 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
625 &$defaultlink, &$display_filename, &$where, &$what,
626 &$type0, &$type1);
627 // Do not let a generic plugin change the default link if a more
628 // specialized one already did it...
629 if ($defaultlink != $defaultlink_orig) {
630 $dummy = '';
631 $temp[5] = &$dummy;
632 }
633 do_hook("attachment */*", $temp);
634
635 $this_attachment = array();
636 $this_attachment['Name'] = decodeHeader($display_filename);
637 $this_attachment['Description'] = $description;
638 $this_attachment['DefaultHREF'] = $defaultlink;
639 $this_attachment['DownloadHREF'] = $links['download link']['href'];
640 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
641
642 // base64 encoded file sizes are misleading, so approximate real size
643 if (!empty($header->encoding) && strtolower($header->encoding) == 'base64')
644 $this_attachment['Size'] = $header->size / 4 * 3;
645 else
646 $this_attachment['Size'] = $header->size;
647
648 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
649 $this_attachment['OtherLinks'] = array();
650 foreach ($links as $val) {
651 if ($val['text']==_("Download")) {
652 $this_attachment['DownloadHREF'] = $val['href'];
653 continue;
654 }
655 if ($val['text']==_("View")) {
656 $this_attachment['ViewHREF'] = $val['href'];
657 continue;
658 }
659
660 // This makes no sense - If 'text' and 'extra' are just concatenated,
661 // there is no point in having 'extra'.... I am going to assume this
662 // was a mistake and am changing 'extra' to be what I think it was
663 // meant to be: additional tag attributes. However, I'm not checking
664 // extensively for plugins that were using this the wrong way (but why would they?)
665 if (empty($val['text']))
666 continue;
667
668 $temp = array();
669 $temp['HREF'] = $val['href'];
670 $temp['Text'] = $val['text'];
671 $temp['Extra'] = (empty($val['extra']) ? '' : $val['extra']);
672 $this_attachment['OtherLinks'][] = $temp;
673 }
674 $attachments[] = $this_attachment;
675
676 unset($links);
677 }
678
679 return $attachments;
680 }
681
682 /**
683 * Displays attachment links and information
684 *
685 * Since 1.3.0 function is not included in formatBody() call.
686 *
687 * Since 1.0.2 uses attachment $type0/$type1 hook.
688 * Since 1.2.5 uses attachment $type0/* hook.
689 * Since 1.5.0 uses attachments_bottom hook.
690 * Since 1.5.2 uses templates and does *not* return a value.
691 *
692 * @param object $message SquirrelMail message object
693 * @param array $exclude_id message parts that are not attachments.
694 * @param string $mailbox mailbox name
695 * @param integer $id message id
696 */
697 function formatAttachments($message, $exclude_id, $mailbox, $id) {
698 global $oTemplate;
699
700 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
701
702 $oTemplate->assign('attachments', $attach);
703 $oTemplate->display('read_attachments.tpl');
704 }
705
706 function sqimap_base64_decode(&$string) {
707
708 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
709 // fly decoding (to reduce memory usage) you have to check if the
710 // data has incomplete pairs
711
712 // Remove the noise in order to check if the 4 bytes pairs are complete
713 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
714
715 $sStringRem = '';
716 $iMod = strlen($string) % 4;
717 if ($iMod) {
718 $sStringRem = substr($string,-$iMod);
719 // Check if $sStringRem contains padding characters
720 if (substr($sStringRem,-1) != '=') {
721 $string = substr($string,0,-$iMod);
722 } else {
723 $sStringRem = '';
724 }
725 }
726 $string = base64_decode($string);
727 return $sStringRem;
728 }
729
730 /**
731 * Decodes encoded string (usually message body)
732 *
733 * This function decodes a string (usually the message body)
734 * depending on the encoding type. Currently quoted-printable
735 * and base64 encodings are supported.
736 *
737 * The decode_body hook was added to this function in 1.4.2/1.5.0.
738 * The $force_crlf parameter was added in 1.5.2.
739 *
740 * @param string $string The encoded string
741 * @param string $encoding used encoding
742 * @param string $force_crlf Whether or not to force CRLF or LF
743 * line endings (or to leave as is).
744 * If given as "LF", line endings will
745 * all be converted to LF; if "CRLF",
746 * line endings will all be converted
747 * to CRLF. If given as an empty value,
748 * the global $force_crlf_default will
749 * be consulted (it can be specified in
750 * config/config_local.php). Otherwise,
751 * any other value will cause the string
752 * to be left alone. Note that this will
753 * be overridden to "LF" if not using at
754 * least PHP version 4.3.0. (OPTIONAL;
755 * default is empty - consult global
756 * default value)
757 *
758 * @return string The decoded string
759 *
760 * @since 1.0
761 *
762 */
763 function decodeBody($string, $encoding, $force_crlf='') {
764
765 global $force_crlf_default;
766 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
767 $force_crlf = strtoupper($force_crlf);
768
769 // must force line endings to LF due to broken
770 // quoted_printable_decode() in PHP versions
771 // before 4.3.0 (see below)
772 //
773 if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
774 $string = str_replace("\r\n", "\n", $string);
775 else if ($force_crlf == 'CRLF')
776 $string = str_replace("\n", "\r\n", $string);
777
778 $encoding = strtolower($encoding);
779
780 $encoding_handler = do_hook('decode_body', $encoding);
781
782
783 // plugins get first shot at decoding the string
784 //
785 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
786 $string = $encoding_handler('decode', $string);
787
788 } elseif ($encoding == 'quoted-printable' ||
789 $encoding == 'quoted_printable') {
790
791 // quoted_printable_decode() function is broken in older
792 // php versions. Text with \r\n decoding was fixed only
793 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
794 // and the above call to: str_replace("\r\n", "\n", $string);
795 //
796 $string = quoted_printable_decode($string);
797
798 } elseif ($encoding == 'base64') {
799 $string = base64_decode($string);
800 }
801
802 // All other encodings are returned raw.
803 return $string;
804 }
805
806 /**
807 * Decodes headers
808 *
809 * This function decodes strings that are encoded according to
810 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
811 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
812 *
813 * @param string $string header string that has to be made readable
814 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
815 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
816 * @param boolean $decide decide if string can be utfencoded. defaults to false
817 * @return string decoded header string
818 */
819 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
820 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
821 if (is_array($string)) {
822 $string = implode("\n", $string);
823 }
824
825 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
826 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
827 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
828 // Do we need to return at this point?
829 // return $string;
830 }
831 $i = 0;
832 $iLastMatch = -2;
833 $encoded = true;
834
835 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
836 $aString = explode(' ',$string);
837 $ret = '';
838 foreach ($aString as $chunk) {
839 if ($encoded && $chunk === '') {
840 continue;
841 } elseif ($chunk === '') {
842 $ret .= ' ';
843 continue;
844 }
845 $encoded = false;
846 /* if encoded words are not separated by a linear-space-white we still catch them */
847 $j = $i-1;
848
849 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
850 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
851 if ($iLastMatch !== $j) {
852 if ($htmlsafe) {
853 $ret .= '&#32;';
854 } else {
855 $ret .= ' ';
856 }
857 }
858 $iLastMatch = $i;
859 $j = $i;
860 if ($htmlsafe) {
861 $ret .= sm_encode_html_special_chars($res[1]);
862 } else {
863 $ret .= $res[1];
864 }
865 $encoding = ucfirst($res[3]);
866
867 /* decide about valid decoding */
868 if ($decide && is_conversion_safe($res[2])) {
869 $utfencode=true;
870 $can_be_encoded=true;
871 } else {
872 $can_be_encoded=false;
873 }
874 switch ($encoding)
875 {
876 case 'B':
877 // fix broken base64-encoded strings (remove end = padding,
878 // change any = to + in middle of string, add padding back
879 // to the end)
880 if ($fix_broken_base64_encoded_messages) {
881 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
882 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
883 }
884 $replace = base64_decode($res[4]);
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 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
891 } else {
892 // convert string to html codes in order to display it
893 $ret .= charset_decode($res[2],$replace);
894 }
895 } else {
896 if ($htmlsafe) {
897 $replace = sm_encode_html_special_chars($replace);
898 }
899 $ret.= $replace;
900 }
901 break;
902 case 'Q':
903 $replace = str_replace('_', ' ', $res[4]);
904 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
905 create_function ('$matches', 'return chr(hexdec($matches[1]));'),
906 $replace);
907 if ($utfencode) {
908 if ($can_be_encoded) {
909 /* convert string to different charset,
910 * if functions asks for it (usually in compose)
911 */
912 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
913 } else {
914 // convert string to html codes in order to display it
915 $replace = charset_decode($res[2], $replace);
916 }
917 } else {
918 if ($htmlsafe) {
919 $replace = sm_encode_html_special_chars($replace);
920 }
921 }
922 $ret .= $replace;
923 break;
924 default:
925 break;
926 }
927 $chunk = $res[5];
928 $encoded = true;
929 }
930 if (!$encoded) {
931 if ($htmlsafe) {
932 $ret .= '&#32;';
933 } else {
934 $ret .= ' ';
935 }
936 }
937
938 if (!$encoded && $htmlsafe) {
939 $ret .= sm_encode_html_special_chars($chunk);
940 } else {
941 $ret .= $chunk;
942 }
943 ++$i;
944 }
945 /* remove the first added space */
946 if ($ret) {
947 if ($htmlsafe) {
948 $ret = substr($ret,5);
949 } else {
950 $ret = substr($ret,1);
951 }
952 }
953
954 return $ret;
955 }
956
957 /**
958 * Encodes header
959 *
960 * Function uses XTRA_CODE _encodeheader function, if such function exists.
961 *
962 * Function uses Q encoding by default and encodes a string according to RFC
963 * 1522 for use in headers if it contains 8-bit characters or anything that
964 * looks like it should be encoded.
965 *
966 * Function switches to B encoding and encodeHeaderBase64() function, if
967 * string is 8bit and multibyte character set supported by mbstring extension
968 * is used. It can cause E_USER_NOTICE errors, if interface is used with
969 * multibyte character set unsupported by mbstring extension.
970 *
971 * @param string $string header string, that has to be encoded
972 * @return string quoted-printable encoded string
973 * @todo make $mb_charsets system wide constant
974 */
975 function encodeHeader ($string) {
976 global $default_charset, $languages, $squirrelmail_language;
977
978 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
979 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
980 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
981 }
982
983 // Use B encoding for multibyte charsets
984 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
985 if (in_array($default_charset,$mb_charsets) &&
986 in_array($default_charset,sq_mb_list_encodings()) &&
987 sq_is8bit($string)) {
988 return encodeHeaderBase64($string,$default_charset);
989 } elseif (in_array($default_charset,$mb_charsets) &&
990 sq_is8bit($string) &&
991 ! in_array($default_charset,sq_mb_list_encodings())) {
992 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
993 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
994 }
995
996 // Encode only if the string contains 8-bit characters or =?
997 $j = strlen($string);
998 $max_l = 75 - strlen($default_charset) - 7;
999 $aRet = array();
1000 $ret = '';
1001 $iEncStart = $enc_init = false;
1002 $cur_l = $iOffset = 0;
1003 for($i = 0; $i < $j; ++$i) {
1004 switch($string[$i])
1005 {
1006 case '"':
1007 case '=':
1008 case '<':
1009 case '>':
1010 case ',':
1011 case '?':
1012 case '_':
1013 if ($iEncStart === false) {
1014 $iEncStart = $i;
1015 }
1016 $cur_l+=3;
1017 if ($cur_l > ($max_l-2)) {
1018 /* if there is an stringpart that doesn't need encoding, add it */
1019 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1020 $aRet[] = "=?$default_charset?Q?$ret?=";
1021 $iOffset = $i;
1022 $cur_l = 0;
1023 $ret = '';
1024 $iEncStart = false;
1025 } else {
1026 $ret .= sprintf("=%02X",ord($string[$i]));
1027 }
1028 break;
1029 case '(':
1030 case ')':
1031 if ($iEncStart !== false) {
1032 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1033 $aRet[] = "=?$default_charset?Q?$ret?=";
1034 $iOffset = $i;
1035 $cur_l = 0;
1036 $ret = '';
1037 $iEncStart = false;
1038 }
1039 break;
1040 case ' ':
1041 if ($iEncStart !== false) {
1042 $cur_l++;
1043 if ($cur_l > $max_l) {
1044 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1045 $aRet[] = "=?$default_charset?Q?$ret?=";
1046 $iOffset = $i;
1047 $cur_l = 0;
1048 $ret = '';
1049 $iEncStart = false;
1050 } else {
1051 $ret .= '_';
1052 }
1053 }
1054 break;
1055 default:
1056 $k = ord($string[$i]);
1057 if ($k > 126) {
1058 if ($iEncStart === false) {
1059 // do not start encoding in the middle of a string, also take the rest of the word.
1060 $sLeadString = substr($string,0,$i);
1061 $aLeadString = explode(' ',$sLeadString);
1062 $sToBeEncoded = array_pop($aLeadString);
1063 $iEncStart = $i - strlen($sToBeEncoded);
1064 $ret .= $sToBeEncoded;
1065 $cur_l += strlen($sToBeEncoded);
1066 }
1067 $cur_l += 3;
1068 /* first we add the encoded string that reached it's max size */
1069 if ($cur_l > ($max_l-2)) {
1070 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1071 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1072 $cur_l = 3;
1073 $ret = '';
1074 $iOffset = $i;
1075 $iEncStart = $i;
1076 }
1077 $enc_init = true;
1078 $ret .= sprintf("=%02X", $k);
1079 } else {
1080 if ($iEncStart !== false) {
1081 $cur_l++;
1082 if ($cur_l > $max_l) {
1083 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1084 $aRet[] = "=?$default_charset?Q?$ret?=";
1085 $iEncStart = false;
1086 $iOffset = $i;
1087 $cur_l = 0;
1088 $ret = '';
1089 } else {
1090 $ret .= $string[$i];
1091 }
1092 }
1093 }
1094 break;
1095 }
1096 }
1097
1098 if ($enc_init) {
1099 if ($iEncStart !== false) {
1100 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1101 $aRet[] = "=?$default_charset?Q?$ret?=";
1102 } else {
1103 $aRet[] = substr($string,$iOffset);
1104 }
1105 $string = implode('',$aRet);
1106 }
1107 return $string;
1108 }
1109
1110 /**
1111 * Encodes string according to rfc2047 B encoding header formating rules
1112 *
1113 * It is recommended way to encode headers with character sets that store
1114 * symbols in more than one byte.
1115 *
1116 * Function requires mbstring support. If required mbstring functions are missing,
1117 * function returns false and sets E_USER_WARNING level error message.
1118 *
1119 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1120 * that mbstring functions will generate E_WARNING errors, if unsupported
1121 * character set is used. mb_encode_mimeheader function provided by php
1122 * mbstring extension is not used in order to get better control of header
1123 * encoding.
1124 *
1125 * Used php code functions - function_exists(), trigger_error(), strlen()
1126 * (is used with charset names and base64 strings). Used php mbstring
1127 * functions - mb_strlen and mb_substr.
1128 *
1129 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1130 * encoding), rfc 2822 (header folding)
1131 *
1132 * @param string $string header string that must be encoded
1133 * @param string $charset character set. Must be supported by mbstring extension.
1134 * Use sq_mb_list_encodings() to detect supported charsets.
1135 * @return string string encoded according to rfc2047 B encoding formating rules
1136 * @since 1.5.1
1137 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1138 * @todo Do we want to control max length of header?
1139 * @todo Do we want to control EOL (end-of-line) marker?
1140 * @todo Do we want to translate error message?
1141 */
1142 function encodeHeaderBase64($string,$charset) {
1143 /**
1144 * Check mbstring function requirements.
1145 */
1146 if (! function_exists('mb_strlen') ||
1147 ! function_exists('mb_substr')) {
1148 // set E_USER_WARNING
1149 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1150 // return false
1151 return false;
1152 }
1153
1154 // initial return array
1155 $aRet = array();
1156
1157 /**
1158 * header length = 75 symbols max (same as in encodeHeader)
1159 * remove $charset length
1160 * remove =? ? ?= (5 chars)
1161 * remove 2 more chars (\r\n ?)
1162 */
1163 $iMaxLength = 75 - strlen($charset) - 7;
1164
1165 // set first character position
1166 $iStartCharNum = 0;
1167
1168 // loop through all characters. count characters and not bytes.
1169 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1170 // encode string from starting character to current character.
1171 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1172
1173 // Check encoded string length
1174 if(strlen($encoded_string)>$iMaxLength) {
1175 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1176 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1177
1178 // set new starting character
1179 $iStartCharNum = $iCharNum-1;
1180
1181 // encode last char (in case it is last character in string)
1182 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1183 } // if string is shorter than max length - add next character
1184 }
1185
1186 // add last encoded string to array
1187 $aRet[] = $encoded_string;
1188
1189 // set initial return string
1190 $sRet = '';
1191
1192 // loop through encoded strings
1193 foreach($aRet as $string) {
1194 // TODO: Do we want to control EOL (end-of-line) marker
1195 if ($sRet!='') $sRet.= " ";
1196
1197 // add header tags and encoded string to return string
1198 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1199 }
1200
1201 return $sRet;
1202 }
1203
1204 /* This function trys to locate the entity_id of a specific mime element */
1205 function find_ent_id($id, $message) {
1206 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1207 if ($message->entities[$i]->header->type0 == 'multipart') {
1208 $ret = find_ent_id($id, $message->entities[$i]);
1209 } else {
1210 if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1211 // if (sq_check_save_extension($message->entities[$i])) {
1212 return $message->entities[$i]->entity_id;
1213 // }
1214 } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1215 /**
1216 * This is part of a fix for Outlook Express 6.x generating
1217 * cid URLs without creating content-id headers
1218 * @@JA - 20050207
1219 */
1220 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1221 return $message->entities[$i]->entity_id;
1222 }
1223 }
1224 }
1225 }
1226 return $ret;
1227 }
1228
1229 function sq_check_save_extension($message) {
1230 $filename = $message->getFilename();
1231 $ext = substr($filename, strrpos($filename,'.')+1);
1232 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1233 return in_array($ext, $save_extensions);
1234 }
1235
1236
1237 /**
1238 ** HTMLFILTER ROUTINES
1239 */
1240
1241 /**
1242 * This function checks attribute values for entity-encoded values
1243 * and returns them translated into 8-bit strings so we can run
1244 * checks on them.
1245 *
1246 * @param $attvalue A string to run entity check against.
1247 * @return Nothing, modifies a reference value.
1248 */
1249 function sq_defang(&$attvalue){
1250 $me = 'sq_defang';
1251 /**
1252 * Skip this if there aren't ampersands or backslashes.
1253 */
1254 if (strpos($attvalue, '&') === false
1255 && strpos($attvalue, '\\') === false){
1256 return;
1257 }
1258 $m = false;
1259 // before deent, translate the dangerous unicode characters and ... to safe values
1260 // otherwise the regular expressions do not match.
1261
1262
1263
1264 do {
1265 $m = false;
1266 $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1267 $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1268 $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1269 } while ($m == true);
1270 $attvalue = stripslashes($attvalue);
1271 }
1272
1273 /**
1274 * Kill any tabs, newlines, or carriage returns. Our friends the
1275 * makers of the browser with 95% market value decided that it'd
1276 * be funny to make "java[tab]script" be just as good as "javascript".
1277 *
1278 * @param attvalue The attribute value before extraneous spaces removed.
1279 * @return attvalue Nothing, modifies a reference value.
1280 */
1281 function sq_unspace(&$attvalue){
1282 $me = 'sq_unspace';
1283 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1284 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1285 Array('', '', '', '', ''), $attvalue);
1286 }
1287 }
1288
1289 /**
1290 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1291 * IE as regular characters.
1292 *
1293 * @param attvalue The attribute value before dangerous characters are translated.
1294 * @return attvalue Nothing, modifies a reference value.
1295 * @author Marc Groot Koerkamp.
1296 */
1297 function sq_fixIE_idiocy(&$attvalue) {
1298 // remove NUL
1299 $attvalue = str_replace("\0", "", $attvalue);
1300 // remove comments
1301 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1302
1303 // IE has the evil habit of accepting every possible value for the attribute expression.
1304 // The table below contains characters which are parsed by IE if they are used in the "expression"
1305 // attribute value.
1306 $aDangerousCharsReplacementTable = array(
1307 array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1308 '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1309 '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1310 '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1311 '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1312 '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1313 '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1314 '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1315 '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1316 '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1317 '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1318 '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1319 '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1320 '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1321 '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1322 '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1323 '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1324 '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1325 '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1326 '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1327 '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1328 '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1329 '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1330 '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1331 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1332 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1333 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1334 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1335 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1336 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1337 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1338 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1339 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1340 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1341 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1342 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1343 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1344 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1345 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1346 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1347 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1348 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1349 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1350 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1351 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1352 "\xCA\x9F", /* L UNICODE IPA Extension */
1353 "\xCA\x80", /* R UNICODE IPA Extension */
1354 "\xC9\xB4"), /* N UNICODE IPA Extension */
1355 array('l', 'l', 'r','r','n','n',
1356 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1357 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1358 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1359 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1360
1361 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1362 // used for XSS.
1363 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1364 }
1365
1366 /**
1367 * This function returns the final tag out of the tag name, an array
1368 * of attributes, and the type of the tag. This function is called by
1369 * sq_sanitize internally.
1370 *
1371 * @param $tagname the name of the tag.
1372 * @param $attary the array of attributes and their values
1373 * @param $tagtype The type of the tag (see in comments).
1374 * @return a string with the final tag representation.
1375 */
1376 function sq_tagprint($tagname, $attary, $tagtype){
1377 $me = 'sq_tagprint';
1378
1379 if ($tagtype == 2){
1380 $fulltag = '</' . $tagname . '>';
1381 } else {
1382 $fulltag = '<' . $tagname;
1383 if (is_array($attary) && sizeof($attary)){
1384 $atts = Array();
1385 foreach ($attary as $attname => $attvalue){
1386 array_push($atts, "$attname=$attvalue");
1387 }
1388 $fulltag .= ' ' . join(" ", $atts);
1389 }
1390 if ($tagtype == 3){
1391 $fulltag .= ' /';
1392 }
1393 $fulltag .= '>';
1394 }
1395 return $fulltag;
1396 }
1397
1398 /**
1399 * A small helper function to use with array_walk. Modifies a by-ref
1400 * value and makes it lowercase.
1401 *
1402 * @param $val a value passed by-ref.
1403 * @return void since it modifies a by-ref value.
1404 */
1405 function sq_casenormalize(&$val){
1406 $val = strtolower($val);
1407 }
1408
1409 /**
1410 * This function skips any whitespace from the current position within
1411 * a string and to the next non-whitespace value.
1412 *
1413 * @param $body the string
1414 * @param $offset the offset within the string where we should start
1415 * looking for the next non-whitespace character.
1416 * @return the location within the $body where the next
1417 * non-whitespace char is located.
1418 */
1419 function sq_skipspace($body, $offset){
1420 $me = 'sq_skipspace';
1421 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1422 if (!empty($matches[1])){
1423 $offset += strlen($matches[1]);
1424 }
1425 return $offset;
1426 }
1427
1428 /**
1429 * This function looks for the next character within a string. It's
1430 * really just a glorified "strpos", except it catches if failures
1431 * nicely.
1432 *
1433 * @param $body The string to look for needle in.
1434 * @param $offset Start looking from this position.
1435 * @param $needle The character/string to look for.
1436 * @return location of the next occurance of the needle, or
1437 * strlen($body) if needle wasn't found.
1438 */
1439 function sq_findnxstr($body, $offset, $needle){
1440 $me = 'sq_findnxstr';
1441 $pos = strpos($body, $needle, $offset);
1442 if ($pos === FALSE){
1443 $pos = strlen($body);
1444 }
1445 return $pos;
1446 }
1447
1448 /**
1449 * This function takes a PCRE-style regexp and tries to match it
1450 * within the string.
1451 *
1452 * @param $body The string to look for needle in.
1453 * @param $offset Start looking from here.
1454 * @param $reg A PCRE-style regex to match.
1455 * @return Returns a false if no matches found, or an array
1456 * with the following members:
1457 * - integer with the location of the match within $body
1458 * - string with whatever content between offset and the match
1459 * - string with whatever it is we matched
1460 */
1461 function sq_findnxreg($body, $offset, $reg){
1462 $me = 'sq_findnxreg';
1463 $matches = Array();
1464 $retarr = Array();
1465 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1466 if (!isset($matches[0]) || !$matches[0]){
1467 $retarr = false;
1468 } else {
1469 $retarr[0] = $offset + strlen($matches[1]);
1470 $retarr[1] = $matches[1];
1471 $retarr[2] = $matches[2];
1472 }
1473 return $retarr;
1474 }
1475
1476 /**
1477 * This function looks for the next tag.
1478 *
1479 * @param $body String where to look for the next tag.
1480 * @param $offset Start looking from here.
1481 * @return false if no more tags exist in the body, or
1482 * an array with the following members:
1483 * - string with the name of the tag
1484 * - array with attributes and their values
1485 * - integer with tag type (1, 2, or 3)
1486 * - integer where the tag starts (starting "<")
1487 * - integer where the tag ends (ending ">")
1488 * first three members will be false, if the tag is invalid.
1489 */
1490 function sq_getnxtag($body, $offset){
1491 $me = 'sq_getnxtag';
1492 if ($offset > strlen($body)){
1493 return false;
1494 }
1495 $lt = sq_findnxstr($body, $offset, "<");
1496 if ($lt == strlen($body)){
1497 return false;
1498 }
1499 /**
1500 * We are here:
1501 * blah blah <tag attribute="value">
1502 * \---------^
1503 */
1504 $pos = sq_skipspace($body, $lt+1);
1505 if ($pos >= strlen($body)){
1506 return Array(false, false, false, $lt, strlen($body));
1507 }
1508 /**
1509 * There are 3 kinds of tags:
1510 * 1. Opening tag, e.g.:
1511 * <a href="blah">
1512 * 2. Closing tag, e.g.:
1513 * </a>
1514 * 3. XHTML-style content-less tag, e.g.:
1515 * <img src="blah" />
1516 */
1517 $tagtype = false;
1518 switch (substr($body, $pos, 1)){
1519 case '/':
1520 $tagtype = 2;
1521 $pos++;
1522 break;
1523 case '!':
1524 /**
1525 * A comment or an SGML declaration.
1526 */
1527 if (substr($body, $pos+1, 2) == "--"){
1528 $gt = strpos($body, "-->", $pos);
1529 if ($gt === false){
1530 $gt = strlen($body);
1531 } else {
1532 $gt += 2;
1533 }
1534 return Array(false, false, false, $lt, $gt);
1535 } else {
1536 $gt = sq_findnxstr($body, $pos, ">");
1537 return Array(false, false, false, $lt, $gt);
1538 }
1539 break;
1540 default:
1541 /**
1542 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1543 * later.
1544 */
1545 $tagtype = 1;
1546 break;
1547 }
1548
1549 $tag_start = $pos;
1550 $tagname = '';
1551 /**
1552 * Look for next [\W-_], which will indicate the end of the tag name.
1553 */
1554 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1555 if ($regary == false){
1556 return Array(false, false, false, $lt, strlen($body));
1557 }
1558 list($pos, $tagname, $match) = $regary;
1559 $tagname = strtolower($tagname);
1560
1561 /**
1562 * $match can be either of these:
1563 * '>' indicating the end of the tag entirely.
1564 * '\s' indicating the end of the tag name.
1565 * '/' indicating that this is type-3 xhtml tag.
1566 *
1567 * Whatever else we find there indicates an invalid tag.
1568 */
1569 switch ($match){
1570 case '/':
1571 /**
1572 * This is an xhtml-style tag with a closing / at the
1573 * end, like so: <img src="blah" />. Check if it's followed
1574 * by the closing bracket. If not, then this tag is invalid
1575 */
1576 if (substr($body, $pos, 2) == "/>"){
1577 $pos++;
1578 $tagtype = 3;
1579 } else {
1580 $gt = sq_findnxstr($body, $pos, ">");
1581 $retary = Array(false, false, false, $lt, $gt);
1582 return $retary;
1583 }
1584 case '>':
1585 return Array($tagname, false, $tagtype, $lt, $pos);
1586 break;
1587 default:
1588 /**
1589 * Check if it's whitespace
1590 */
1591 if (!preg_match('/\s/', $match)){
1592 /**
1593 * This is an invalid tag! Look for the next closing ">".
1594 */
1595 $gt = sq_findnxstr($body, $lt, ">");
1596 return Array(false, false, false, $lt, $gt);
1597 }
1598 break;
1599 }
1600
1601 /**
1602 * At this point we're here:
1603 * <tagname attribute='blah'>
1604 * \-------^
1605 *
1606 * At this point we loop in order to find all attributes.
1607 */
1608 $attname = '';
1609 $atttype = false;
1610 $attary = Array();
1611
1612 while ($pos <= strlen($body)){
1613 $pos = sq_skipspace($body, $pos);
1614 if ($pos == strlen($body)){
1615 /**
1616 * Non-closed tag.
1617 */
1618 return Array(false, false, false, $lt, $pos);
1619 }
1620 /**
1621 * See if we arrived at a ">" or "/>", which means that we reached
1622 * the end of the tag.
1623 */
1624 $matches = Array();
1625 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1626 /**
1627 * Yep. So we did.
1628 */
1629 $pos += strlen($matches[1]);
1630 if ($matches[2] == "/>"){
1631 $tagtype = 3;
1632 $pos++;
1633 }
1634 return Array($tagname, $attary, $tagtype, $lt, $pos);
1635 }
1636
1637 /**
1638 * There are several types of attributes, with optional
1639 * [:space:] between members.
1640 * Type 1:
1641 * attrname[:space:]=[:space:]'CDATA'
1642 * Type 2:
1643 * attrname[:space:]=[:space:]"CDATA"
1644 * Type 3:
1645 * attr[:space:]=[:space:]CDATA
1646 * Type 4:
1647 * attrname
1648 *
1649 * We leave types 1 and 2 the same, type 3 we check for
1650 * '"' and convert to "&quot" if needed, then wrap in
1651 * double quotes. Type 4 we convert into:
1652 * attrname="yes".
1653 */
1654 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1655 if ($regary == false){
1656 /**
1657 * Looks like body ended before the end of tag.
1658 */
1659 return Array(false, false, false, $lt, strlen($body));
1660 }
1661 list($pos, $attname, $match) = $regary;
1662 $attname = strtolower($attname);
1663 /**
1664 * We arrived at the end of attribute name. Several things possible
1665 * here:
1666 * '>' means the end of the tag and this is attribute type 4
1667 * '/' if followed by '>' means the same thing as above
1668 * '\s' means a lot of things -- look what it's followed by.
1669 * anything else means the attribute is invalid.
1670 */
1671 switch($match){
1672 case '/':
1673 /**
1674 * This is an xhtml-style tag with a closing / at the
1675 * end, like so: <img src="blah" />. Check if it's followed
1676 * by the closing bracket. If not, then this tag is invalid
1677 */
1678 if (substr($body, $pos, 2) == "/>"){
1679 $pos++;
1680 $tagtype = 3;
1681 } else {
1682 $gt = sq_findnxstr($body, $pos, ">");
1683 $retary = Array(false, false, false, $lt, $gt);
1684 return $retary;
1685 }
1686 case '>':
1687 $attary[$attname] = '"yes"';
1688 return Array($tagname, $attary, $tagtype, $lt, $pos);
1689 break;
1690 default:
1691 /**
1692 * Skip whitespace and see what we arrive at.
1693 */
1694 $pos = sq_skipspace($body, $pos);
1695 $char = substr($body, $pos, 1);
1696 /**
1697 * Two things are valid here:
1698 * '=' means this is attribute type 1 2 or 3.
1699 * \w means this was attribute type 4.
1700 * anything else we ignore and re-loop. End of tag and
1701 * invalid stuff will be caught by our checks at the beginning
1702 * of the loop.
1703 */
1704 if ($char == "="){
1705 $pos++;
1706 $pos = sq_skipspace($body, $pos);
1707 /**
1708 * Here are 3 possibilities:
1709 * "'" attribute type 1
1710 * '"' attribute type 2
1711 * everything else is the content of tag type 3
1712 */
1713 $quot = substr($body, $pos, 1);
1714 if ($quot == "'"){
1715 $regary = sq_findnxreg($body, $pos+1, "\'");
1716 if ($regary == false){
1717 return Array(false, false, false, $lt, strlen($body));
1718 }
1719 list($pos, $attval, $match) = $regary;
1720 $pos++;
1721 $attary[$attname] = "'" . $attval . "'";
1722 } else if ($quot == '"'){
1723 $regary = sq_findnxreg($body, $pos+1, '\"');
1724 if ($regary == false){
1725 return Array(false, false, false, $lt, strlen($body));
1726 }
1727 list($pos, $attval, $match) = $regary;
1728 $pos++;
1729 $attary[$attname] = '"' . $attval . '"';
1730 } else {
1731 /**
1732 * These are hateful. Look for \s, or >.
1733 */
1734 $regary = sq_findnxreg($body, $pos, "[\s>]");
1735 if ($regary == false){
1736 return Array(false, false, false, $lt, strlen($body));
1737 }
1738 list($pos, $attval, $match) = $regary;
1739 /**
1740 * If it's ">" it will be caught at the top.
1741 */
1742 $attval = preg_replace("/\"/s", "&quot;", $attval);
1743 $attary[$attname] = '"' . $attval . '"';
1744 }
1745 } else if (preg_match("|[\w/>]|", $char)) {
1746 /**
1747 * That was attribute type 4.
1748 */
1749 $attary[$attname] = '"yes"';
1750 } else {
1751 /**
1752 * An illegal character. Find next '>' and return.
1753 */
1754 $gt = sq_findnxstr($body, $pos, ">");
1755 return Array(false, false, false, $lt, $gt);
1756 }
1757 break;
1758 }
1759 }
1760 /**
1761 * The fact that we got here indicates that the tag end was never
1762 * found. Return invalid tag indication so it gets stripped.
1763 */
1764 return Array(false, false, false, $lt, strlen($body));
1765 }
1766
1767 /**
1768 * Translates entities into literal values so they can be checked.
1769 *
1770 * @param $attvalue the by-ref value to check.
1771 * @param $regex the regular expression to check against.
1772 * @param $hex whether the entites are hexadecimal.
1773 * @return True or False depending on whether there were matches.
1774 */
1775 function sq_deent(&$attvalue, $regex, $hex=false){
1776 $me = 'sq_deent';
1777 $ret_match = false;
1778 // remove comments
1779 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1780 preg_match_all($regex, $attvalue, $matches);
1781 if (is_array($matches) && sizeof($matches[0]) > 0){
1782 $repl = Array();
1783 for ($i = 0; $i < sizeof($matches[0]); $i++){
1784 $numval = $matches[1][$i];
1785 if ($hex){
1786 $numval = hexdec($numval);
1787 }
1788 $repl[$matches[0][$i]] = chr($numval);
1789 }
1790 $attvalue = strtr($attvalue, $repl);
1791 return true;
1792 } else {
1793 return false;
1794 }
1795 }
1796
1797 /**
1798 * This function runs various checks against the attributes.
1799 *
1800 * @param $tagname String with the name of the tag.
1801 * @param $attary Array with all tag attributes.
1802 * @param $rm_attnames See description for sq_sanitize
1803 * @param $bad_attvals See description for sq_sanitize
1804 * @param $add_attr_to_tag See description for sq_sanitize
1805 * @param $message message object
1806 * @param $id message id
1807 * @return Array with modified attributes.
1808 */
1809 function sq_fixatts($tagname,
1810 $attary,
1811 $rm_attnames,
1812 $bad_attvals,
1813 $add_attr_to_tag,
1814 $message,
1815 $id,
1816 $mailbox
1817 ){
1818 $me = 'sq_fixatts';
1819 foreach ($attary as $attname => $attvalue){
1820 /**
1821 * See if this attribute should be removed.
1822 */
1823 foreach ($rm_attnames as $matchtag=>$matchattrs){
1824 if (preg_match($matchtag, $tagname)){
1825 foreach ($matchattrs as $matchattr){
1826 if (preg_match($matchattr, $attname)){
1827 unset($attary[$attname]);
1828 continue;
1829 }
1830 }
1831 }
1832 }
1833 /**
1834 * Workaround for IE quirks
1835 */
1836 sq_fixIE_idiocy($attvalue);
1837
1838 /**
1839 * Remove any backslashes, entities, and extraneous whitespace.
1840 */
1841
1842 $oldattvalue = $attvalue;
1843 sq_defang($attvalue);
1844 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1845 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1846 // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1847 $attvalue = "idiocy";
1848 $attary[$attname] = $attvalue;
1849 }
1850 sq_unspace($attvalue);
1851
1852 /**
1853 * Now let's run checks on the attvalues.
1854 * I don't expect anyone to comprehend this. If you do,
1855 * get in touch with me so I can drive to where you live and
1856 * shake your hand personally. :)
1857 */
1858 foreach ($bad_attvals as $matchtag=>$matchattrs){
1859 if (preg_match($matchtag, $tagname)){
1860 foreach ($matchattrs as $matchattr=>$valary){
1861 if (preg_match($matchattr, $attname)){
1862 /**
1863 * There are two arrays in valary.
1864 * First is matches.
1865 * Second one is replacements
1866 */
1867 list($valmatch, $valrepl) = $valary;
1868 $newvalue =
1869 preg_replace($valmatch, $valrepl, $attvalue);
1870 if ($newvalue != $attvalue){
1871 $attary[$attname] = $newvalue;
1872 $attvalue = $newvalue;
1873 }
1874 }
1875 }
1876 }
1877 }
1878 if ($attname == 'style') {
1879 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1880 // 8bit and control characters in style attribute values can be used for XSS, remove them
1881 $attary[$attname] = '"disallowed character"';
1882 }
1883 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1884 if (count($aMatch)) {
1885 foreach($aMatch[1] as $sMatch) {
1886 // url value
1887 $urlvalue = $sMatch;
1888 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1889 $attary[$attname] = str_replace($sMatch,$urlvalue,$attvalue);
1890 }
1891 }
1892 }
1893 /**
1894 * Use white list based filtering on attributes which can contain url's
1895 */
1896 else if ($attname == 'href' || $attname == 'xlink:href' || $attname == 'src'
1897 || $attname == 'poster' || $attname == 'formaction'
1898 || $attname == 'background' || $attname == 'action') {
1899 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1900 $attary[$attname] = $attvalue;
1901 }
1902 }
1903 /**
1904 * See if we need to append any attributes to this tag.
1905 */
1906 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1907 if (preg_match($matchtag, $tagname)){
1908 $attary = array_merge($attary, $addattary);
1909 }
1910 }
1911 return $attary;
1912 }
1913
1914 /**
1915 * This function filters url's
1916 *
1917 * @param $attvalue String with attribute value to filter
1918 * @param $message message object
1919 * @param $id message id
1920 * @param $mailbox mailbox
1921 * @param $sQuote quoting characters around url's
1922 */
1923 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1924 $attvalue = trim($attvalue);
1925 if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1926 // remove the double quotes
1927 $sQuote = $attvalue[0];
1928 $attvalue = trim(substr($attvalue,1,-1));
1929 }
1930
1931 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1932 // images off by default.
1933 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1934
1935 global $use_transparent_security_image;
1936 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1937 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1938
1939 /**
1940 * Replace empty src tags with the blank image. src is only used
1941 * for frames, images, and image inputs. Doing a replace should
1942 * not affect them working as should be, however it will stop
1943 * IE from being kicked off when src for img tags are not set
1944 */
1945 if ($attvalue == '') {
1946 $attvalue = '"' . SM_PATH . 'images/blank.png"';
1947 } else {
1948 // first, disallow 8 bit characters and control characters
1949 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1950 switch ($attname) {
1951 case 'href':
1952 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1953 break;
1954 default:
1955 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1956 break;
1957 }
1958 } else {
1959 $aUrl = parse_url($attvalue);
1960 if (isset($aUrl['scheme'])) {
1961 switch(strtolower($aUrl['scheme'])) {
1962 case 'mailto':
1963 case 'http':
1964 case 'https':
1965 case 'ftp':
1966 if ($attname != 'href') {
1967 if ($view_unsafe_images == false) {
1968 $attvalue = $sQuote . $secremoveimg . $sQuote;
1969 } else {
1970 if (isset($aUrl['path'])) {
1971
1972 // No one has been able to show that image URIs
1973 // can be exploited, so for now, no restrictions
1974 // are made at all. If this proves to be a problem,
1975 // the commented-out code below can be of help.
1976 // (One consideration is that I see nothing in this
1977 // function that specifically says that we will
1978 // only ever arrive here when inspecting an image
1979 // tag, although that does seem to be the end
1980 // result - e.g., <script src="..."> where malicious
1981 // image URIs are in fact a problem are already
1982 // filtered out elsewhere.
1983 /* ---------------------------------
1984 // validate image extension.
1985 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1986 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1987 // If URI is to something other than
1988 // a regular image file, get the contents
1989 // and try to see if it is an image.
1990 // Don't use Fileinfo (finfo_file()) because
1991 // we'd need to make the admin configure the
1992 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1993 //
1994 $mime_type = '';
1995 if (function_exists('mime_content_type')
1996 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1997
1998 // fetch file
1999 //
2000 $file_contents = '';
2001 while (!feof($FILE)) {
2002 $file_contents .= fread($FILE, 8192);
2003 }
2004 fclose($FILE);
2005
2006 // store file locally
2007 //
2008 global $attachment_dir, $username;
2009 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
2010 $localfilename = GenerateRandomString(32, '', 7);
2011 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2012 while (file_exists($full_localfilename)) {
2013 $localfilename = GenerateRandomString(32, '', 7);
2014 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2015 }
2016 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2017 fwrite($FILE, $file_contents);
2018 fclose($FILE);
2019
2020 // get mime type and remove file
2021 //
2022 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2023 unlink("$hashed_attachment_dir/$localfilename");
2024 }
2025 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2026 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2027 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2028 }
2029 }
2030 --------------------------------- */
2031 } else {
2032 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2033 }
2034 }
2035 } else {
2036 $attvalue = $sQuote . $attvalue . $sQuote;
2037 }
2038 break;
2039 case 'outbind':
2040 /**
2041 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2042 * One day MS might actually make it match something useful, for now, falling
2043 * back to using cid2http, so we can grab the blank.png.
2044 */
2045 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2046 break;
2047 case 'cid':
2048 /**
2049 * Turn cid: urls into http-friendly ones.
2050 */
2051 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2052 break;
2053 default:
2054 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2055 break;
2056 }
2057 } else {
2058 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2059 // parse_url did not lead to satisfying result
2060 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2061 }
2062 }
2063 }
2064 }
2065 }
2066
2067 /**
2068 * This function edits the style definition to make them friendly and
2069 * usable in SquirrelMail.
2070 *
2071 * @param $message the message object
2072 * @param $id the message id
2073 * @param $content a string with whatever is between <style> and </style>
2074 * @param $mailbox the message mailbox
2075 * @return a string with edited content.
2076 */
2077 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2078 $me = 'sq_fixstyle';
2079 // workaround for </style> in between comments
2080 $iCurrentPos = $pos;
2081 $content = '';
2082 $sToken = '';
2083 $bSucces = false;
2084 $bEndTag = false;
2085 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2086 $char = $body[$i];
2087 switch ($char) {
2088 case '<':
2089 $sToken = $char;
2090 break;
2091 case '/':
2092 if ($sToken == '<') {
2093 $sToken .= $char;
2094 $bEndTag = true;
2095 } else {
2096 $content .= $char;
2097 }
2098 break;
2099 case '>':
2100 if ($bEndTag) {
2101 $sToken .= $char;
2102 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2103 $newpos = $i + 1;
2104 $bSucces = true;
2105 break 2;
2106 } else {
2107 $content .= $sToken;
2108 }
2109 $bEndTag = false;
2110 } else {
2111 $content .= $char;
2112 }
2113 break;
2114 case '!':
2115 if ($sToken == '<') {
2116 // possible comment
2117 if (isset($body[$i+2]) && substr($body,$i,3) == '!--') {
2118 $i = strpos($body,'-->',$i+3);
2119 if ($i === false) { // no end comment
2120 $i = strlen($body);
2121 }
2122 $sToken = '';
2123 }
2124 } else {
2125 $content .= $char;
2126 }
2127 break;
2128 default:
2129 if ($bEndTag) {
2130 $sToken .= $char;
2131 } else {
2132 $content .= $char;
2133 }
2134 break;
2135 }
2136 }
2137 if ($bSucces == FALSE){
2138 return array(FALSE, strlen($body));
2139 }
2140
2141
2142
2143 /**
2144 * First look for general BODY style declaration, which would be
2145 * like so:
2146 * body {background: blah-blah}
2147 * and change it to .bodyclass so we can just assign it to a <div>
2148 */
2149 // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2150 // Nah, this is even better - try to preface all CSS selectors with
2151 // our <div> class ID "bodyclass" then correct generic "body" selectors
2152 // TODO: this works pretty good but breaks stuff like this:
2153 // @media print { body { font-size: 10pt; } }
2154 // but there isn't an easy way to make this regex skip @media
2155 // definitions... though lots of the ones in the wild will be
2156 // correctly handled because they tend to end with a parenthesis, like:
2157 // @media screen and (max-width:480px) { ...
2158 $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2159 $content = str_replace('.bodyclass body', '.bodyclass', $content);
2160
2161 global $use_transparent_security_image;
2162 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2163 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2164
2165 /**
2166 * Fix url('blah') declarations.
2167 */
2168 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2169 // "url(\\1$secremoveimg\\2)", $content);
2170
2171 // first check for 8bit sequences and disallowed control characters
2172 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2173 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2174 return array($content, $newpos);
2175 }
2176
2177 // IE Sucks hard. We have a special function for it.
2178 sq_fixIE_idiocy($content);
2179
2180 // remove @import line
2181 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2182
2183 // translate ur\l and variations (IE parses that)
2184 // TODO check if the sq_fixIE_idiocy function already handles this.
2185 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2186 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2187 if (count($aMatch)) {
2188 $aValue = $aReplace = array();
2189 foreach($aMatch[1] as $sMatch) {
2190 // url value
2191 $urlvalue = $sMatch;
2192 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2193 $aValue[] = $sMatch;
2194 $aReplace[] = $urlvalue;
2195 }
2196 $content = str_replace($aValue,$aReplace,$content);
2197 }
2198
2199 /**
2200 * Remove any backslashes, entities, and extraneous whitespace.
2201 */
2202 $contentTemp = $content;
2203 sq_defang($contentTemp);
2204 sq_unspace($contentTemp);
2205
2206 /**
2207 * Fix stupid css declarations which lead to vulnerabilities
2208 * in IE.
2209 *
2210 * Also remove "position" attribute, as it can easily be set
2211 * to "fixed" or "absolute" with "left" and "top" attributes
2212 * of zero, taking over the whole content frame. It can also
2213 * be set to relative and move itself anywhere it wants to,
2214 * displaying content in areas it shouldn't be allowed to touch.
2215 */
2216 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2217 '/expression/i',
2218 '/behaviou*r/i',
2219 '/binding/i',
2220 '/include-source/i',
2221 '/javascript/i',
2222 '/script/i',
2223 '/position/i');
2224 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2225 $contentNew = preg_replace($match, $replace, $contentTemp);
2226 if ($contentNew !== $contentTemp) {
2227 // insecure css declarations are used. From now on we don't care
2228 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2229 $content = $contentNew;
2230 }
2231 return array($content, $newpos);
2232 }
2233
2234
2235 /**
2236 * This function converts cid: url's into the ones that can be viewed in
2237 * the browser.
2238 *
2239 * @param $message the message object
2240 * @param $id the message id
2241 * @param $cidurl the cid: url.
2242 * @param $mailbox the message mailbox
2243 * @return a string with a http-friendly url
2244 */
2245 function sq_cid2http($message, $id, $cidurl, $mailbox){
2246 /**
2247 * Get rid of quotes.
2248 */
2249 $quotchar = substr($cidurl, 0, 1);
2250 if ($quotchar == '"' || $quotchar == "'"){
2251 $cidurl = str_replace($quotchar, "", $cidurl);
2252 } else {
2253 $quotchar = '';
2254 }
2255 $cidurl = substr(trim($cidurl), 4);
2256
2257 $match_str = '/\{.*?\}\//';
2258 $str_rep = '';
2259 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2260
2261 $linkurl = find_ent_id($cidurl, $message);
2262 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2263 unsafe link image */
2264 $httpurl = '';
2265
2266 /**
2267 * This is part of a fix for Outlook Express 6.x generating
2268 * cid URLs without creating content-id headers. These images are
2269 * not part of the multipart/related html mail. The html contains
2270 * <img src="cid:{some_id}/image_filename.ext"> references to
2271 * attached images with as goal to render them inline although
2272 * the attachment disposition property is not inline.
2273 */
2274
2275 if (empty($linkurl)) {
2276 if (preg_match('/{.*}\//', $cidurl)) {
2277 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2278 if (!empty($cidurl)) {
2279 $linkurl = find_ent_id($cidurl, $message);
2280 }
2281 }
2282 }
2283
2284 if (!empty($linkurl)) {
2285 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2286 "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2287 '&amp;ent_id=' . $linkurl . $quotchar;
2288 } else {
2289 /**
2290 * If we couldn't generate a proper img url, drop in a blank image
2291 * instead of sending back empty, otherwise it causes unusual behaviour
2292 */
2293 $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2294 }
2295
2296 return $httpurl;
2297 }
2298
2299 /**
2300 * This function changes the <body> tag into a <div> tag since we
2301 * can't really have a body-within-body.
2302 *
2303 * @param $attary an array of attributes and values of <body>
2304 * @param $mailbox mailbox we're currently reading (for cid2http)
2305 * @param $message current message (for cid2http)
2306 * @param $id current message id (for cid2http)
2307 * @return a modified array of attributes to be set for <div>
2308 */
2309 function sq_body2div($attary, $mailbox, $message, $id){
2310 $me = 'sq_body2div';
2311 $divattary = Array('class' => "'bodyclass'");
2312 $text = '#000000';
2313 $has_bgc_stl = $has_txt_stl = false;
2314 $styledef = '';
2315 if (is_array($attary) && sizeof($attary) > 0){
2316 foreach ($attary as $attname=>$attvalue){
2317 $quotchar = substr($attvalue, 0, 1);
2318 $attvalue = str_replace($quotchar, "", $attvalue);
2319 switch ($attname){
2320 case 'background':
2321 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2322 $styledef .= "background-image: url('$attvalue'); ";
2323 break;
2324 case 'bgcolor':
2325 $has_bgc_stl = true;
2326 $styledef .= "background-color: $attvalue; ";
2327 break;
2328 case 'text':
2329 $has_txt_stl = true;
2330 $styledef .= "color: $attvalue; ";
2331 break;
2332 }
2333 }
2334 // Outlook defines a white bgcolor and no text color. This can lead to
2335 // white text on a white bg with certain themes.
2336 if ($has_bgc_stl && !$has_txt_stl) {
2337 $styledef .= "color: $text; ";
2338 }
2339 if (strlen($styledef) > 0){
2340 $divattary["style"] = "\"$styledef\"";
2341 }
2342 }
2343 return $divattary;
2344 }
2345
2346 /**
2347 * This is the main function and the one you should actually be calling.
2348 * There are several variables you should be aware of an which need
2349 * special description.
2350 *
2351 * Since the description is quite lengthy, see it here:
2352 * http://linux.duke.edu/projects/mini/htmlfilter/
2353 *
2354 * @param $body the string with HTML you wish to filter
2355 * @param $tag_list see description above
2356 * @param $rm_tags_with_content see description above
2357 * @param $self_closing_tags see description above
2358 * @param $force_tag_closing see description above
2359 * @param $rm_attnames see description above
2360 * @param $bad_attvals see description above
2361 * @param $add_attr_to_tag see description above
2362 * @param $message message object
2363 * @param $id message id
2364 * @param $recursively_called boolean flag for recursive calls into this function (optional; default FALSE)
2365 * @return sanitized html safe to show on your pages.
2366 */
2367 function sq_sanitize($body,
2368 $tag_list,
2369 $rm_tags_with_content,
2370 $self_closing_tags,
2371 $force_tag_closing,
2372 $rm_attnames,
2373 $bad_attvals,
2374 $add_attr_to_tag,
2375 $message,
2376 $id,
2377 $mailbox,
2378 $recursively_called=FALSE
2379 ){
2380 $me = 'sq_sanitize';
2381
2382 /**
2383 * See if tag_list is of tags to remove or tags to allow.
2384 * false means remove these tags
2385 * true means allow these tags
2386 */
2387 $orig_tag_list = $tag_list;
2388 $rm_tags = array_shift($tag_list);
2389
2390 /**
2391 * Normalize rm_tags and rm_tags_with_content.
2392 */
2393 @array_walk($tag_list, 'sq_casenormalize');
2394 @array_walk($rm_tags_with_content, 'sq_casenormalize');
2395 @array_walk($self_closing_tags, 'sq_casenormalize');
2396
2397 $curpos = 0;
2398 $open_tags = Array();
2399 $trusted = "\n<!-- begin sanitized html -->\n";
2400 $skip_content = false;
2401 /**
2402 * Take care of netscape's stupid javascript entities like
2403 * &{alert('boo')};
2404 */
2405 $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2406
2407 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2408 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2409
2410 /**
2411 * RCDATA and RAWTEXT tags are handled differently:
2412 * next instance of closing tag is used, whether or not
2413 * the HTML is well formed before that
2414 */
2415 global $rcdata_rawtext_tags;
2416 if (!$recursively_called
2417 && in_array($tagname, $rcdata_rawtext_tags)
2418 && $tagtype === 1){
2419 $closing_tag = false;
2420 $closing_tag_offset = $curpos;
2421 // seek out the closing tag for the current RCDATA/RAWTEXT tag
2422 while (1) {
2423 // first we need to move forward to next available closing tag
2424 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2425 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2426 if ($next_tag === false) {
2427 $closing_tag = false;
2428 break;
2429 }
2430 // but then we have to make sure it's a well-formed tag
2431 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2432 if ($closing_tag === false)
2433 break;
2434 else if ($closing_tag[0] !== false
2435 // these should be redundant
2436 && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2437 $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos + 1),
2438 $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2439 $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2440 $message, $id, $mailbox, true);
2441 $curpos = $closing_tag[4] + 1;
2442 continue 2;
2443 }
2444 $closing_tag_offset = $next_tag[0] + 1;
2445 }
2446 if ($closing_tag === false)
2447 { /* no-op... there was no closing tag for this RCDATA/RAWTEXT tag - we could probably set $curpos to the end of $body, but this HTML is malformed anyway and should just fall apart on its own */ }
2448 }
2449
2450 $free_content = substr($body, $curpos, $lt-$curpos);
2451 /**
2452 * Take care of <style>
2453 */
2454 if ($tagname == "style" && $tagtype == 1){
2455 list($free_content, $curpos) =
2456 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2457 if ($free_content != FALSE){
2458 if ( !empty($attary) ) {
2459 $attary = sq_fixatts($tagname,
2460 $attary,
2461 $rm_attnames,
2462 $bad_attvals,
2463 $add_attr_to_tag,
2464 $message,
2465 $id,
2466 $mailbox
2467 );
2468 }
2469 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2470 $trusted .= $free_content;
2471 $trusted .= sq_tagprint($tagname, false, 2);
2472 }
2473 continue;
2474 }
2475 if ($skip_content == false){
2476 $trusted .= $free_content;
2477 }
2478 if ($tagname != FALSE){
2479 if ($tagtype == 2){
2480 if ($skip_content == $tagname){
2481 /**
2482 * Got to the end of tag we needed to remove.
2483 */
2484 $tagname = false;
2485 $skip_content = false;
2486 } else {
2487 if ($skip_content == false){
2488 if ($tagname == "body"){
2489 $tagname = "div";
2490 }
2491 if (isset($open_tags[$tagname]) &&
2492 $open_tags[$tagname] > 0){
2493 $open_tags[$tagname]--;
2494 } else {
2495 $tagname = false;
2496 }
2497 }
2498 }
2499 } else {
2500 /**
2501 * $rm_tags_with_content
2502 */
2503 if ($skip_content == false){
2504 /**
2505 * See if this is a self-closing type and change
2506 * tagtype appropriately.
2507 */
2508 if ($tagtype == 1
2509 && in_array($tagname, $self_closing_tags)){
2510 $tagtype = 3;
2511 }
2512 /**
2513 * See if we should skip this tag and any content
2514 * inside it.
2515 */
2516 if ($tagtype == 1 &&
2517 in_array($tagname, $rm_tags_with_content)){
2518 $skip_content = $tagname;
2519 } else {
2520 if (($rm_tags == false
2521 && in_array($tagname, $tag_list)) ||
2522 ($rm_tags == true &&
2523 !in_array($tagname, $tag_list))){
2524 $tagname = false;
2525 } else {
2526 /**
2527 * Convert body into div.
2528 */
2529 if ($tagname == "body"){
2530 $tagname = "div";
2531 $attary = sq_body2div($attary, $mailbox,
2532 $message, $id);
2533 }
2534 if ($tagtype == 1){
2535 if (isset($open_tags[$tagname])){
2536 $open_tags[$tagname]++;
2537 } else {
2538 $open_tags[$tagname]=1;
2539 }
2540 }
2541 /**
2542 * This is where we run other checks.
2543 */
2544 if (is_array($attary) && sizeof($attary) > 0){
2545 $attary = sq_fixatts($tagname,
2546 $attary,
2547 $rm_attnames,
2548 $bad_attvals,
2549 $add_attr_to_tag,
2550 $message,
2551 $id,
2552 $mailbox
2553 );
2554 }
2555 }
2556 }
2557 }
2558 }
2559 if ($tagname != false && $skip_content == false){
2560 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2561 }
2562 }
2563 $curpos = $gt+1;
2564 }
2565 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2566 if ($force_tag_closing == true){
2567 foreach ($open_tags as $tagname=>$opentimes){
2568 while ($opentimes > 0){
2569 $trusted .= '</' . $tagname . '>';
2570 $opentimes--;
2571 }
2572 }
2573 $trusted .= "\n";
2574 }
2575 $trusted .= "<!-- end sanitized html -->\n";
2576 return $trusted;
2577 }
2578
2579 /**
2580 * This is a wrapper function to call html sanitizing routines.
2581 *
2582 * @param $body the body of the message
2583 * @param $id the id of the message
2584
2585 * @param $message
2586 * @param $mailbox
2587 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2588 * into internal SM compose links
2589 * (optional; default = TRUE)
2590 * @return a string with html safe to display in the browser.
2591 */
2592 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2593
2594 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2595
2596 global $attachment_common_show_images, $view_unsafe_images,
2597 $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2598 $remove_rcdata_rawtext_tags_and_content;
2599
2600 $rcdata_rawtext_tags = array(
2601 "noscript",
2602 "noframes",
2603 "noembed",
2604 "textarea",
2605 // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2606 );
2607
2608 /**
2609 * Don't display attached images in HTML mode.
2610 *
2611 * SB: why?
2612 */
2613 $attachment_common_show_images = false;
2614 $tag_list = Array(
2615 false, // remove these tags
2616 "meta",
2617 "html",
2618 "head",
2619 "base",
2620 "link",
2621 "frame",
2622 "iframe",
2623 "plaintext",
2624 "marquee",
2625 );
2626
2627 $rm_tags_with_content = Array(
2628 "script",
2629 "object",
2630 "applet",
2631 "embed",
2632 "title",
2633 "frameset",
2634 "xmp",
2635 "xml",
2636 );
2637 if (!$allow_svg_display)
2638 $rm_tags_with_content[] = 'svg';
2639 /**
2640 * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2641 * case that they are, but if you prefer to remove them and their contents entirely
2642 * (in most cases, should be a safe thing with minimal impact), you can add the
2643 * following to config/config_local.php
2644 * $remove_rcdata_rawtext_tags_and_content = TRUE;
2645 */
2646 if ($remove_rcdata_rawtext_tags_and_content)
2647 $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2648
2649 $self_closing_tags = Array(
2650 "img",
2651 "br",
2652 "hr",
2653 "input",
2654 "outbind",
2655 );
2656
2657 $force_tag_closing = true;
2658
2659 $rm_attnames = Array(
2660 "/.*/" =>
2661 Array(
2662 "/target/i",
2663 "/^on.*/i",
2664 "/^dynsrc/i",
2665 "/^data.*/i",
2666 "/^lowsrc.*/i",
2667 )
2668 );
2669
2670 global $use_transparent_security_image;
2671 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2672 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2673
2674 $bad_attvals = Array(
2675 "/.*/" =>
2676 Array(
2677 "/^src|background/i" =>
2678 Array(
2679 Array(
2680 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2681 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2682 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2683 ),
2684 Array(
2685 "\\1$secremoveimg\\2",
2686 "\\1$secremoveimg\\2",
2687 "\\1$secremoveimg\\2",
2688 )
2689 ),
2690 "/^href|action/i" =>
2691 Array(
2692 Array(
2693 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2694 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2695 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2696 ),
2697 Array(
2698 "\\1#\\1",
2699 "\\1#\\1",
2700 "\\1#\\1"
2701 )
2702 ),
2703 "/^style/i" =>
2704 Array(
2705 Array(
2706 "/\/\*.*\*\//",
2707 "/expression/i",
2708 "/binding/i",
2709 "/behaviou*r/i",
2710 "/include-source/i",
2711
2712 // position:relative can also be exploited
2713 // to put content outside of email body area
2714 // and position:fixed is similarly exploitable
2715 // as position:absolute, so we'll remove it
2716 // altogether....
2717 //
2718 // Does this screw up legitimate HTML messages?
2719 // If so, the only fix I see is to allow position
2720 // attributes (any values? I think we still have
2721 // to block static and fixed) only if $use_iframe
2722 // is enabled (1.5.0+)
2723 //
2724 // was: "/position\s*:\s*absolute/i",
2725 //
2726 "/position\s*:/i",
2727
2728 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2729 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2730 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2731 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2732 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2733 ),
2734 Array(
2735 "",
2736 "idiocy",
2737 "idiocy",
2738 "idiocy",
2739 "idiocy",
2740 "idiocy",
2741 "url",
2742 "url(\\1#\\1)",
2743 "url(\\1#\\1)",
2744 "url(\\1#\\1)",
2745 "\\1:url(\\2#\\3)"
2746 )
2747 )
2748 )
2749 );
2750
2751 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2752 // images off by default.
2753 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2754
2755 if (!$view_unsafe_images){
2756 /**
2757 * Remove any references to http/https if view_unsafe_images set
2758 * to false.
2759 */
2760 array_push($bad_attvals['/.*/']['/^src|background/i'][0],
2761 '/^([\'\"])\s*https*:.*([\'\"])/si');
2762 array_push($bad_attvals['/.*/']['/^src|background/i'][1],
2763 "\\1$secremoveimg\\1");
2764 array_push($bad_attvals['/.*/']['/^style/i'][0],
2765 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2766 array_push($bad_attvals['/.*/']['/^style/i'][1],
2767 "url(\\1$secremoveimg\\1)");
2768 }
2769
2770 $add_attr_to_tag = Array(
2771 "/^a$/i" =>
2772 Array('target'=>'"_blank"',
2773 'title'=>'"'._("This external link will open in a new window").'"'
2774 )
2775 );
2776 $trusted = sq_sanitize($body,
2777 $tag_list,
2778 $rm_tags_with_content,
2779 $self_closing_tags,
2780 $force_tag_closing,
2781 $rm_attnames,
2782 $bad_attvals,
2783 $add_attr_to_tag,
2784 $message,
2785 $id,
2786 $mailbox
2787 );
2788 if (strpos($trusted,$secremoveimg)){
2789 $has_unsafe_images = true;
2790 }
2791
2792 // we want to parse mailto's in HTML output, change to SM compose links
2793 // this is a modified version of code from url_parser.php... but Marc is
2794 // right: we need a better filtering implementation; adding this randomly
2795 // here is not a great solution
2796 //
2797 if ($take_mailto_links) {
2798 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2799 global $MailTo_PReg_Match;
2800 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2801 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2802 // quote before mailto, and we'll also try to convert the non-links back into links
2803 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2804 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2805 foreach ($regs[0] as $i => $mailto_before) {
2806 $mailto_params = $regs[11][$i];
2807
2808 // get rid of any leading quote we may have captured but don't care about
2809 //
2810 $mailto_before = ltrim($mailto_before, '"\'');
2811
2812 // get rid of any tailing quote since we have to add send_to to the end
2813 //
2814 $mailto_before = rtrim($mailto_before, '"\'');
2815 $mailto_params = rtrim($mailto_params, '"\'');
2816
2817 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2818 $to = 'to=' . $regs[2][$i];
2819 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2820 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2821 else {
2822 if ($mailto_params) //already some params, append to them
2823 $mailto_params .= '&amp;' . $to;
2824 else
2825 $mailto_params .= '?' . $to;
2826 }
2827 }
2828
2829 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2830
2831 // we'll already have target=_blank, no need to allow comp_in_new
2832 // here (which would be a lot more work anyway)
2833 //
2834 global $compose_new_win;
2835 $temp_comp_in_new = $compose_new_win;
2836 $compose_new_win = 0;
2837 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2838 $compose_new_win = $temp_comp_in_new;
2839
2840 // remove <a href=" and anything after the next quote (we only
2841 // need the uri, not the link HTML) in compose uri
2842 //
2843 // but only do this if the original mailto was in a real anchor tag
2844 //
2845 if (!empty($regs[1][$i])) {
2846 $comp_uri = substr($comp_uri, 9);
2847 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2848 }
2849 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2850 }
2851 }
2852 }
2853
2854 return $trusted;
2855 }
2856
2857 /**
2858 * function SendDownloadHeaders - send file to the browser
2859 *
2860 * Original Source: SM core src/download.php
2861 * moved here to make it available to other code, and separate
2862 * front end from back end functionality.
2863 *
2864 * @param string $type0 first half of mime type
2865 * @param string $type1 second half of mime type
2866 * @param string $filename filename to tell the browser for downloaded file
2867 * @param boolean $force whether to force the download dialog to pop
2868 * @param optional integer $filesize send the Content-Header and length to the browser
2869 * @return void
2870 */
2871 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2872 global $languages, $squirrelmail_language;
2873 $isIE = $isIE6plus = false;
2874
2875 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2876
2877 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2878 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2879 $isIE = true;
2880 }
2881
2882 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2883 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2884 $isIE6plus = true;
2885 }
2886
2887 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2888 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2889 $filename =
2890 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2891 } else {
2892 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2893 }
2894
2895 // A Pox on Microsoft and it's Internet Explorer!
2896 //
2897 // IE has lots of bugs with file downloads.
2898 // It also has problems with SSL. Both of these cause problems
2899 // for us in this function.
2900 //
2901 // See this article on Cache Control headers and SSL
2902 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2903 //
2904 // The best thing you can do for IE is to upgrade to the latest
2905 // version
2906 //set all the Cache Control Headers for IE
2907 if ($isIE) {
2908 $filename=rawurlencode($filename);
2909 header ("Pragma: public");
2910 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2911 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2912 // header ("Cache-Control: post-check=0, pre-check=0", false);
2913 header ("Cache-Control: private");
2914
2915 //set the inline header for IE, we'll add the attachment header later if we need it
2916 header ("Content-Disposition: inline; filename=$filename");
2917 }
2918
2919 if (!$force) {
2920 // Try to show in browser window
2921 header ("Content-Disposition: inline; filename=\"$filename\"");
2922 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2923 } else {
2924 // Try to pop up the "save as" box
2925
2926 // IE makes this hard. It pops up 2 save boxes, or none.
2927 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2928 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2929 // But, according to Microsoft, it is "RFC compliant but doesn't
2930 // take into account some deviations that allowed within the
2931 // specification." Doesn't that mean RFC non-compliant?
2932 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2933
2934 // all browsers need the application/octet-stream header for this
2935 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2936
2937 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2938 // Do not have quotes around filename, but that applied to
2939 // "attachment"... does it apply to inline too?
2940 header ("Content-Disposition: attachment; filename=\"$filename\"");
2941
2942 if ($isIE && !$isIE6plus) {
2943 // This combination seems to work mostly. IE 5.5 SP 1 has
2944 // known issues (see the Microsoft Knowledge Base)
2945
2946 // This works for most types, but doesn't work with Word files
2947 header ("Content-Type: application/download; name=\"$filename\"");
2948 header ("Content-Type: application/force-download; name=\"$filename\"");
2949 // These are spares, just in case. :-)
2950 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2951 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2952 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2953 } else if ($isIE) {
2954 // This is to prevent IE for MIME sniffing and auto open a file in IE
2955 header ("Content-Type: application/force-download; name=\"$filename\"");
2956 } else {
2957 // another application/octet-stream forces download for Netscape
2958 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2959 }
2960 }
2961
2962 //send the content-length header if the calling function provides it
2963 if ($filesize > 0) {
2964 header("Content-Length: $filesize");
2965 }
2966
2967 } // end fn SendDownloadHeaders