projects / squirrelmail.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Add MIME type parameters ("type0", "type1") to the end of the plugin arguments for...
[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 1999-2020 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, $temp1);
603 do_hook("attachment $type0/$type1", $temp);
604 /* The API for this hook has changed as of 1.5.2 so that all plugin
605 arguments are passed in an array instead of each their own plugin
606 argument, and arguments are passed by reference, so instead of
607 returning any changes, changes should simply be made to the original
608 arguments themselves. */
609 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
610 &$defaultlink, &$display_filename, &$where, &$what);
611 // Do not let a generic plugin change the default link if a more
612 // specialized one already did it...
613 if ($defaultlink != $defaultlink_orig) {
614 $dummy = '';
615 $temp[5] = &$dummy;
616 }
617 do_hook("attachment $type0/*", $temp);
618 /* The API for this hook has changed as of 1.5.2 so that all plugin
619 arguments are passed in an array instead of each their own plugin
620 argument, and arguments are passed by reference, so instead of
621 returning any changes, changes should simply be made to the original
622 arguments themselves. */
623 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
624 &$defaultlink, &$display_filename, &$where, &$what);
625 // Do not let a generic plugin change the default link if a more
626 // specialized one already did it...
627 if ($defaultlink != $defaultlink_orig) {
628 $dummy = '';
629 $temp[5] = &$dummy;
630 }
631 do_hook("attachment */*", $temp);
632
633 $this_attachment = array();
634 $this_attachment['Name'] = decodeHeader($display_filename);
635 $this_attachment['Description'] = $description;
636 $this_attachment['DefaultHREF'] = $defaultlink;
637 $this_attachment['DownloadHREF'] = $links['download link']['href'];
638 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
639
640 // base64 encoded file sizes are misleading, so approximate real size
641 if (!empty($header->encoding) && strtolower($header->encoding) == 'base64')
642 $this_attachment['Size'] = $header->size / 4 * 3;
643 else
644 $this_attachment['Size'] = $header->size;
645
646 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
647 $this_attachment['OtherLinks'] = array();
648 foreach ($links as $val) {
649 if ($val['text']==_("Download") || $val['text'] == _("View"))
650 continue;
651 if (empty($val['text']) && empty($val['extra']))
652 continue;
653
654 $temp = array();
655 $temp['HREF'] = $val['href'];
656 $temp['Text'] = (empty($val['text']) ? '' : $val['text']) . (empty($val['extra']) ? '' : $val['extra']);
657 $this_attachment['OtherLinks'][] = $temp;
658 }
659 $attachments[] = $this_attachment;
660
661 unset($links);
662 }
663
664 return $attachments;
665 }
666
667 /**
668 * Displays attachment links and information
669 *
670 * Since 1.3.0 function is not included in formatBody() call.
671 *
672 * Since 1.0.2 uses attachment $type0/$type1 hook.
673 * Since 1.2.5 uses attachment $type0/* hook.
674 * Since 1.5.0 uses attachments_bottom hook.
675 * Since 1.5.2 uses templates and does *not* return a value.
676 *
677 * @param object $message SquirrelMail message object
678 * @param array $exclude_id message parts that are not attachments.
679 * @param string $mailbox mailbox name
680 * @param integer $id message id
681 */
682 function formatAttachments($message, $exclude_id, $mailbox, $id) {
683 global $oTemplate;
684
685 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
686
687 $oTemplate->assign('attachments', $attach);
688 $oTemplate->display('read_attachments.tpl');
689 }
690
691 function sqimap_base64_decode(&$string) {
692
693 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
694 // fly decoding (to reduce memory usage) you have to check if the
695 // data has incomplete pairs
696
697 // Remove the noise in order to check if the 4 bytes pairs are complete
698 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
699
700 $sStringRem = '';
701 $iMod = strlen($string) % 4;
702 if ($iMod) {
703 $sStringRem = substr($string,-$iMod);
704 // Check if $sStringRem contains padding characters
705 if (substr($sStringRem,-1) != '=') {
706 $string = substr($string,0,-$iMod);
707 } else {
708 $sStringRem = '';
709 }
710 }
711 $string = base64_decode($string);
712 return $sStringRem;
713 }
714
715 /**
716 * Decodes encoded string (usually message body)
717 *
718 * This function decodes a string (usually the message body)
719 * depending on the encoding type. Currently quoted-printable
720 * and base64 encodings are supported.
721 *
722 * The decode_body hook was added to this function in 1.4.2/1.5.0.
723 * The $force_crlf parameter was added in 1.5.2.
724 *
725 * @param string $string The encoded string
726 * @param string $encoding used encoding
727 * @param string $force_crlf Whether or not to force CRLF or LF
728 * line endings (or to leave as is).
729 * If given as "LF", line endings will
730 * all be converted to LF; if "CRLF",
731 * line endings will all be converted
732 * to CRLF. If given as an empty value,
733 * the global $force_crlf_default will
734 * be consulted (it can be specified in
735 * config/config_local.php). Otherwise,
736 * any other value will cause the string
737 * to be left alone. Note that this will
738 * be overridden to "LF" if not using at
739 * least PHP version 4.3.0. (OPTIONAL;
740 * default is empty - consult global
741 * default value)
742 *
743 * @return string The decoded string
744 *
745 * @since 1.0
746 *
747 */
748 function decodeBody($string, $encoding, $force_crlf='') {
749
750 global $force_crlf_default;
751 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
752 $force_crlf = strtoupper($force_crlf);
753
754 // must force line endings to LF due to broken
755 // quoted_printable_decode() in PHP versions
756 // before 4.3.0 (see below)
757 //
758 if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
759 $string = str_replace("\r\n", "\n", $string);
760 else if ($force_crlf == 'CRLF')
761 $string = str_replace("\n", "\r\n", $string);
762
763 $encoding = strtolower($encoding);
764
765 $encoding_handler = do_hook('decode_body', $encoding);
766
767
768 // plugins get first shot at decoding the string
769 //
770 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
771 $string = $encoding_handler('decode', $string);
772
773 } elseif ($encoding == 'quoted-printable' ||
774 $encoding == 'quoted_printable') {
775
776 // quoted_printable_decode() function is broken in older
777 // php versions. Text with \r\n decoding was fixed only
778 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
779 // and the above call to: str_replace("\r\n", "\n", $string);
780 //
781 $string = quoted_printable_decode($string);
782
783 } elseif ($encoding == 'base64') {
784 $string = base64_decode($string);
785 }
786
787 // All other encodings are returned raw.
788 return $string;
789 }
790
791 /**
792 * Decodes headers
793 *
794 * This function decodes strings that are encoded according to
795 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
796 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
797 *
798 * @param string $string header string that has to be made readable
799 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
800 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
801 * @param boolean $decide decide if string can be utfencoded. defaults to false
802 * @return string decoded header string
803 */
804 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
805 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
806 if (is_array($string)) {
807 $string = implode("\n", $string);
808 }
809
810 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
811 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
812 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
813 // Do we need to return at this point?
814 // return $string;
815 }
816 $i = 0;
817 $iLastMatch = -2;
818 $encoded = true;
819
820 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
821 $aString = explode(' ',$string);
822 $ret = '';
823 foreach ($aString as $chunk) {
824 if ($encoded && $chunk === '') {
825 continue;
826 } elseif ($chunk === '') {
827 $ret .= ' ';
828 continue;
829 }
830 $encoded = false;
831 /* if encoded words are not separated by a linear-space-white we still catch them */
832 $j = $i-1;
833
834 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
835 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
836 if ($iLastMatch !== $j) {
837 if ($htmlsafe) {
838 $ret .= '&#32;';
839 } else {
840 $ret .= ' ';
841 }
842 }
843 $iLastMatch = $i;
844 $j = $i;
845 if ($htmlsafe) {
846 $ret .= sm_encode_html_special_chars($res[1]);
847 } else {
848 $ret .= $res[1];
849 }
850 $encoding = ucfirst($res[3]);
851
852 /* decide about valid decoding */
853 if ($decide && is_conversion_safe($res[2])) {
854 $utfencode=true;
855 $can_be_encoded=true;
856 } else {
857 $can_be_encoded=false;
858 }
859 switch ($encoding)
860 {
861 case 'B':
862 // fix broken base64-encoded strings (remove end = padding,
863 // change any = to + in middle of string, add padding back
864 // to the end)
865 if ($fix_broken_base64_encoded_messages) {
866 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
867 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
868 }
869 $replace = base64_decode($res[4]);
870 if ($utfencode) {
871 if ($can_be_encoded) {
872 /* convert string to different charset,
873 * if functions asks for it (usually in compose)
874 */
875 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
876 } else {
877 // convert string to html codes in order to display it
878 $ret .= charset_decode($res[2],$replace);
879 }
880 } else {
881 if ($htmlsafe) {
882 $replace = sm_encode_html_special_chars($replace);
883 }
884 $ret.= $replace;
885 }
886 break;
887 case 'Q':
888 $replace = str_replace('_', ' ', $res[4]);
889 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
890 create_function ('$matches', 'return chr(hexdec($matches[1]));'),
891 $replace);
892 if ($utfencode) {
893 if ($can_be_encoded) {
894 /* convert string to different charset,
895 * if functions asks for it (usually in compose)
896 */
897 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
898 } else {
899 // convert string to html codes in order to display it
900 $replace = charset_decode($res[2], $replace);
901 }
902 } else {
903 if ($htmlsafe) {
904 $replace = sm_encode_html_special_chars($replace);
905 }
906 }
907 $ret .= $replace;
908 break;
909 default:
910 break;
911 }
912 $chunk = $res[5];
913 $encoded = true;
914 }
915 if (!$encoded) {
916 if ($htmlsafe) {
917 $ret .= '&#32;';
918 } else {
919 $ret .= ' ';
920 }
921 }
922
923 if (!$encoded && $htmlsafe) {
924 $ret .= sm_encode_html_special_chars($chunk);
925 } else {
926 $ret .= $chunk;
927 }
928 ++$i;
929 }
930 /* remove the first added space */
931 if ($ret) {
932 if ($htmlsafe) {
933 $ret = substr($ret,5);
934 } else {
935 $ret = substr($ret,1);
936 }
937 }
938
939 return $ret;
940 }
941
942 /**
943 * Encodes header
944 *
945 * Function uses XTRA_CODE _encodeheader function, if such function exists.
946 *
947 * Function uses Q encoding by default and encodes a string according to RFC
948 * 1522 for use in headers if it contains 8-bit characters or anything that
949 * looks like it should be encoded.
950 *
951 * Function switches to B encoding and encodeHeaderBase64() function, if
952 * string is 8bit and multibyte character set supported by mbstring extension
953 * is used. It can cause E_USER_NOTICE errors, if interface is used with
954 * multibyte character set unsupported by mbstring extension.
955 *
956 * @param string $string header string, that has to be encoded
957 * @return string quoted-printable encoded string
958 * @todo make $mb_charsets system wide constant
959 */
960 function encodeHeader ($string) {
961 global $default_charset, $languages, $squirrelmail_language;
962
963 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
964 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
965 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
966 }
967
968 // Use B encoding for multibyte charsets
969 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
970 if (in_array($default_charset,$mb_charsets) &&
971 in_array($default_charset,sq_mb_list_encodings()) &&
972 sq_is8bit($string)) {
973 return encodeHeaderBase64($string,$default_charset);
974 } elseif (in_array($default_charset,$mb_charsets) &&
975 sq_is8bit($string) &&
976 ! in_array($default_charset,sq_mb_list_encodings())) {
977 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
978 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
979 }
980
981 // Encode only if the string contains 8-bit characters or =?
982 $j = strlen($string);
983 $max_l = 75 - strlen($default_charset) - 7;
984 $aRet = array();
985 $ret = '';
986 $iEncStart = $enc_init = false;
987 $cur_l = $iOffset = 0;
988 for($i = 0; $i < $j; ++$i) {
989 switch($string{$i})
990 {
991 case '"':
992 case '=':
993 case '<':
994 case '>':
995 case ',':
996 case '?':
997 case '_':
998 if ($iEncStart === false) {
999 $iEncStart = $i;
1000 }
1001 $cur_l+=3;
1002 if ($cur_l > ($max_l-2)) {
1003 /* if there is an stringpart that doesn't need encoding, add it */
1004 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1005 $aRet[] = "=?$default_charset?Q?$ret?=";
1006 $iOffset = $i;
1007 $cur_l = 0;
1008 $ret = '';
1009 $iEncStart = false;
1010 } else {
1011 $ret .= sprintf("=%02X",ord($string{$i}));
1012 }
1013 break;
1014 case '(':
1015 case ')':
1016 if ($iEncStart !== false) {
1017 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1018 $aRet[] = "=?$default_charset?Q?$ret?=";
1019 $iOffset = $i;
1020 $cur_l = 0;
1021 $ret = '';
1022 $iEncStart = false;
1023 }
1024 break;
1025 case ' ':
1026 if ($iEncStart !== false) {
1027 $cur_l++;
1028 if ($cur_l > $max_l) {
1029 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1030 $aRet[] = "=?$default_charset?Q?$ret?=";
1031 $iOffset = $i;
1032 $cur_l = 0;
1033 $ret = '';
1034 $iEncStart = false;
1035 } else {
1036 $ret .= '_';
1037 }
1038 }
1039 break;
1040 default:
1041 $k = ord($string{$i});
1042 if ($k > 126) {
1043 if ($iEncStart === false) {
1044 // do not start encoding in the middle of a string, also take the rest of the word.
1045 $sLeadString = substr($string,0,$i);
1046 $aLeadString = explode(' ',$sLeadString);
1047 $sToBeEncoded = array_pop($aLeadString);
1048 $iEncStart = $i - strlen($sToBeEncoded);
1049 $ret .= $sToBeEncoded;
1050 $cur_l += strlen($sToBeEncoded);
1051 }
1052 $cur_l += 3;
1053 /* first we add the encoded string that reached it's max size */
1054 if ($cur_l > ($max_l-2)) {
1055 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1056 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1057 $cur_l = 3;
1058 $ret = '';
1059 $iOffset = $i;
1060 $iEncStart = $i;
1061 }
1062 $enc_init = true;
1063 $ret .= sprintf("=%02X", $k);
1064 } else {
1065 if ($iEncStart !== false) {
1066 $cur_l++;
1067 if ($cur_l > $max_l) {
1068 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1069 $aRet[] = "=?$default_charset?Q?$ret?=";
1070 $iEncStart = false;
1071 $iOffset = $i;
1072 $cur_l = 0;
1073 $ret = '';
1074 } else {
1075 $ret .= $string{$i};
1076 }
1077 }
1078 }
1079 break;
1080 }
1081 }
1082
1083 if ($enc_init) {
1084 if ($iEncStart !== false) {
1085 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1086 $aRet[] = "=?$default_charset?Q?$ret?=";
1087 } else {
1088 $aRet[] = substr($string,$iOffset);
1089 }
1090 $string = implode('',$aRet);
1091 }
1092 return $string;
1093 }
1094
1095 /**
1096 * Encodes string according to rfc2047 B encoding header formating rules
1097 *
1098 * It is recommended way to encode headers with character sets that store
1099 * symbols in more than one byte.
1100 *
1101 * Function requires mbstring support. If required mbstring functions are missing,
1102 * function returns false and sets E_USER_WARNING level error message.
1103 *
1104 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1105 * that mbstring functions will generate E_WARNING errors, if unsupported
1106 * character set is used. mb_encode_mimeheader function provided by php
1107 * mbstring extension is not used in order to get better control of header
1108 * encoding.
1109 *
1110 * Used php code functions - function_exists(), trigger_error(), strlen()
1111 * (is used with charset names and base64 strings). Used php mbstring
1112 * functions - mb_strlen and mb_substr.
1113 *
1114 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1115 * encoding), rfc 2822 (header folding)
1116 *
1117 * @param string $string header string that must be encoded
1118 * @param string $charset character set. Must be supported by mbstring extension.
1119 * Use sq_mb_list_encodings() to detect supported charsets.
1120 * @return string string encoded according to rfc2047 B encoding formating rules
1121 * @since 1.5.1
1122 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1123 * @todo Do we want to control max length of header?
1124 * @todo Do we want to control EOL (end-of-line) marker?
1125 * @todo Do we want to translate error message?
1126 */
1127 function encodeHeaderBase64($string,$charset) {
1128 /**
1129 * Check mbstring function requirements.
1130 */
1131 if (! function_exists('mb_strlen') ||
1132 ! function_exists('mb_substr')) {
1133 // set E_USER_WARNING
1134 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1135 // return false
1136 return false;
1137 }
1138
1139 // initial return array
1140 $aRet = array();
1141
1142 /**
1143 * header length = 75 symbols max (same as in encodeHeader)
1144 * remove $charset length
1145 * remove =? ? ?= (5 chars)
1146 * remove 2 more chars (\r\n ?)
1147 */
1148 $iMaxLength = 75 - strlen($charset) - 7;
1149
1150 // set first character position
1151 $iStartCharNum = 0;
1152
1153 // loop through all characters. count characters and not bytes.
1154 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1155 // encode string from starting character to current character.
1156 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1157
1158 // Check encoded string length
1159 if(strlen($encoded_string)>$iMaxLength) {
1160 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1161 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1162
1163 // set new starting character
1164 $iStartCharNum = $iCharNum-1;
1165
1166 // encode last char (in case it is last character in string)
1167 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1168 } // if string is shorter than max length - add next character
1169 }
1170
1171 // add last encoded string to array
1172 $aRet[] = $encoded_string;
1173
1174 // set initial return string
1175 $sRet = '';
1176
1177 // loop through encoded strings
1178 foreach($aRet as $string) {
1179 // TODO: Do we want to control EOL (end-of-line) marker
1180 if ($sRet!='') $sRet.= " ";
1181
1182 // add header tags and encoded string to return string
1183 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1184 }
1185
1186 return $sRet;
1187 }
1188
1189 /* This function trys to locate the entity_id of a specific mime element */
1190 function find_ent_id($id, $message) {
1191 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1192 if ($message->entities[$i]->header->type0 == 'multipart') {
1193 $ret = find_ent_id($id, $message->entities[$i]);
1194 } else {
1195 if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1196 // if (sq_check_save_extension($message->entities[$i])) {
1197 return $message->entities[$i]->entity_id;
1198 // }
1199 } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1200 /**
1201 * This is part of a fix for Outlook Express 6.x generating
1202 * cid URLs without creating content-id headers
1203 * @@JA - 20050207
1204 */
1205 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1206 return $message->entities[$i]->entity_id;
1207 }
1208 }
1209 }
1210 }
1211 return $ret;
1212 }
1213
1214 function sq_check_save_extension($message) {
1215 $filename = $message->getFilename();
1216 $ext = substr($filename, strrpos($filename,'.')+1);
1217 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1218 return in_array($ext, $save_extensions);
1219 }
1220
1221
1222 /**
1223 ** HTMLFILTER ROUTINES
1224 */
1225
1226 /**
1227 * This function checks attribute values for entity-encoded values
1228 * and returns them translated into 8-bit strings so we can run
1229 * checks on them.
1230 *
1231 * @param $attvalue A string to run entity check against.
1232 * @return Nothing, modifies a reference value.
1233 */
1234 function sq_defang(&$attvalue){
1235 $me = 'sq_defang';
1236 /**
1237 * Skip this if there aren't ampersands or backslashes.
1238 */
1239 if (strpos($attvalue, '&') === false
1240 && strpos($attvalue, '\\') === false){
1241 return;
1242 }
1243 $m = false;
1244 // before deent, translate the dangerous unicode characters and ... to safe values
1245 // otherwise the regular expressions do not match.
1246
1247
1248
1249 do {
1250 $m = false;
1251 $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1252 $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1253 $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1254 } while ($m == true);
1255 $attvalue = stripslashes($attvalue);
1256 }
1257
1258 /**
1259 * Kill any tabs, newlines, or carriage returns. Our friends the
1260 * makers of the browser with 95% market value decided that it'd
1261 * be funny to make "java[tab]script" be just as good as "javascript".
1262 *
1263 * @param attvalue The attribute value before extraneous spaces removed.
1264 * @return attvalue Nothing, modifies a reference value.
1265 */
1266 function sq_unspace(&$attvalue){
1267 $me = 'sq_unspace';
1268 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1269 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1270 Array('', '', '', '', ''), $attvalue);
1271 }
1272 }
1273
1274 /**
1275 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1276 * IE as regular characters.
1277 *
1278 * @param attvalue The attribute value before dangerous characters are translated.
1279 * @return attvalue Nothing, modifies a reference value.
1280 * @author Marc Groot Koerkamp.
1281 */
1282 function sq_fixIE_idiocy(&$attvalue) {
1283 // remove NUL
1284 $attvalue = str_replace("\0", "", $attvalue);
1285 // remove comments
1286 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1287
1288 // IE has the evil habit of accepting every possible value for the attribute expression.
1289 // The table below contains characters which are parsed by IE if they are used in the "expression"
1290 // attribute value.
1291 $aDangerousCharsReplacementTable = array(
1292 array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1293 '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1294 '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1295 '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1296 '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1297 '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1298 '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1299 '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1300 '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1301 '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1302 '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1303 '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1304 '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1305 '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1306 '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1307 '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1308 '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1309 '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1310 '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1311 '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1312 '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1313 '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1314 '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1315 '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1316 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1317 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1318 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1319 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1320 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1321 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1322 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1323 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1324 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1325 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1326 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1327 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1328 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1329 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1330 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1331 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1332 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1333 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1334 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1335 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1336 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1337 "\xCA\x9F", /* L UNICODE IPA Extension */
1338 "\xCA\x80", /* R UNICODE IPA Extension */
1339 "\xC9\xB4"), /* N UNICODE IPA Extension */
1340 array('l', 'l', 'r','r','n','n',
1341 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1342 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1343 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1344 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1345
1346 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1347 // used for XSS.
1348 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1349 }
1350
1351 /**
1352 * This function returns the final tag out of the tag name, an array
1353 * of attributes, and the type of the tag. This function is called by
1354 * sq_sanitize internally.
1355 *
1356 * @param $tagname the name of the tag.
1357 * @param $attary the array of attributes and their values
1358 * @param $tagtype The type of the tag (see in comments).
1359 * @return a string with the final tag representation.
1360 */
1361 function sq_tagprint($tagname, $attary, $tagtype){
1362 $me = 'sq_tagprint';
1363
1364 if ($tagtype == 2){
1365 $fulltag = '</' . $tagname . '>';
1366 } else {
1367 $fulltag = '<' . $tagname;
1368 if (is_array($attary) && sizeof($attary)){
1369 $atts = Array();
1370 while (list($attname, $attvalue) = each($attary)){
1371 array_push($atts, "$attname=$attvalue");
1372 }
1373 $fulltag .= ' ' . join(" ", $atts);
1374 }
1375 if ($tagtype == 3){
1376 $fulltag .= ' /';
1377 }
1378 $fulltag .= '>';
1379 }
1380 return $fulltag;
1381 }
1382
1383 /**
1384 * A small helper function to use with array_walk. Modifies a by-ref
1385 * value and makes it lowercase.
1386 *
1387 * @param $val a value passed by-ref.
1388 * @return void since it modifies a by-ref value.
1389 */
1390 function sq_casenormalize(&$val){
1391 $val = strtolower($val);
1392 }
1393
1394 /**
1395 * This function skips any whitespace from the current position within
1396 * a string and to the next non-whitespace value.
1397 *
1398 * @param $body the string
1399 * @param $offset the offset within the string where we should start
1400 * looking for the next non-whitespace character.
1401 * @return the location within the $body where the next
1402 * non-whitespace char is located.
1403 */
1404 function sq_skipspace($body, $offset){
1405 $me = 'sq_skipspace';
1406 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1407 if (!empty($matches[1])){
1408 $offset += strlen($matches[1]);
1409 }
1410 return $offset;
1411 }
1412
1413 /**
1414 * This function looks for the next character within a string. It's
1415 * really just a glorified "strpos", except it catches if failures
1416 * nicely.
1417 *
1418 * @param $body The string to look for needle in.
1419 * @param $offset Start looking from this position.
1420 * @param $needle The character/string to look for.
1421 * @return location of the next occurance of the needle, or
1422 * strlen($body) if needle wasn't found.
1423 */
1424 function sq_findnxstr($body, $offset, $needle){
1425 $me = 'sq_findnxstr';
1426 $pos = strpos($body, $needle, $offset);
1427 if ($pos === FALSE){
1428 $pos = strlen($body);
1429 }
1430 return $pos;
1431 }
1432
1433 /**
1434 * This function takes a PCRE-style regexp and tries to match it
1435 * within the string.
1436 *
1437 * @param $body The string to look for needle in.
1438 * @param $offset Start looking from here.
1439 * @param $reg A PCRE-style regex to match.
1440 * @return Returns a false if no matches found, or an array
1441 * with the following members:
1442 * - integer with the location of the match within $body
1443 * - string with whatever content between offset and the match
1444 * - string with whatever it is we matched
1445 */
1446 function sq_findnxreg($body, $offset, $reg){
1447 $me = 'sq_findnxreg';
1448 $matches = Array();
1449 $retarr = Array();
1450 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1451 if (!isset($matches{0}) || !$matches{0}){
1452 $retarr = false;
1453 } else {
1454 $retarr{0} = $offset + strlen($matches{1});
1455 $retarr{1} = $matches{1};
1456 $retarr{2} = $matches{2};
1457 }
1458 return $retarr;
1459 }
1460
1461 /**
1462 * This function looks for the next tag.
1463 *
1464 * @param $body String where to look for the next tag.
1465 * @param $offset Start looking from here.
1466 * @return false if no more tags exist in the body, or
1467 * an array with the following members:
1468 * - string with the name of the tag
1469 * - array with attributes and their values
1470 * - integer with tag type (1, 2, or 3)
1471 * - integer where the tag starts (starting "<")
1472 * - integer where the tag ends (ending ">")
1473 * first three members will be false, if the tag is invalid.
1474 */
1475 function sq_getnxtag($body, $offset){
1476 $me = 'sq_getnxtag';
1477 if ($offset > strlen($body)){
1478 return false;
1479 }
1480 $lt = sq_findnxstr($body, $offset, "<");
1481 if ($lt == strlen($body)){
1482 return false;
1483 }
1484 /**
1485 * We are here:
1486 * blah blah <tag attribute="value">
1487 * \---------^
1488 */
1489 $pos = sq_skipspace($body, $lt+1);
1490 if ($pos >= strlen($body)){
1491 return Array(false, false, false, $lt, strlen($body));
1492 }
1493 /**
1494 * There are 3 kinds of tags:
1495 * 1. Opening tag, e.g.:
1496 * <a href="blah">
1497 * 2. Closing tag, e.g.:
1498 * </a>
1499 * 3. XHTML-style content-less tag, e.g.:
1500 * <img src="blah" />
1501 */
1502 $tagtype = false;
1503 switch (substr($body, $pos, 1)){
1504 case '/':
1505 $tagtype = 2;
1506 $pos++;
1507 break;
1508 case '!':
1509 /**
1510 * A comment or an SGML declaration.
1511 */
1512 if (substr($body, $pos+1, 2) == "--"){
1513 $gt = strpos($body, "-->", $pos);
1514 if ($gt === false){
1515 $gt = strlen($body);
1516 } else {
1517 $gt += 2;
1518 }
1519 return Array(false, false, false, $lt, $gt);
1520 } else {
1521 $gt = sq_findnxstr($body, $pos, ">");
1522 return Array(false, false, false, $lt, $gt);
1523 }
1524 break;
1525 default:
1526 /**
1527 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1528 * later.
1529 */
1530 $tagtype = 1;
1531 break;
1532 }
1533
1534 $tag_start = $pos;
1535 $tagname = '';
1536 /**
1537 * Look for next [\W-_], which will indicate the end of the tag name.
1538 */
1539 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1540 if ($regary == false){
1541 return Array(false, false, false, $lt, strlen($body));
1542 }
1543 list($pos, $tagname, $match) = $regary;
1544 $tagname = strtolower($tagname);
1545
1546 /**
1547 * $match can be either of these:
1548 * '>' indicating the end of the tag entirely.
1549 * '\s' indicating the end of the tag name.
1550 * '/' indicating that this is type-3 xhtml tag.
1551 *
1552 * Whatever else we find there indicates an invalid tag.
1553 */
1554 switch ($match){
1555 case '/':
1556 /**
1557 * This is an xhtml-style tag with a closing / at the
1558 * end, like so: <img src="blah" />. Check if it's followed
1559 * by the closing bracket. If not, then this tag is invalid
1560 */
1561 if (substr($body, $pos, 2) == "/>"){
1562 $pos++;
1563 $tagtype = 3;
1564 } else {
1565 $gt = sq_findnxstr($body, $pos, ">");
1566 $retary = Array(false, false, false, $lt, $gt);
1567 return $retary;
1568 }
1569 case '>':
1570 return Array($tagname, false, $tagtype, $lt, $pos);
1571 break;
1572 default:
1573 /**
1574 * Check if it's whitespace
1575 */
1576 if (!preg_match('/\s/', $match)){
1577 /**
1578 * This is an invalid tag! Look for the next closing ">".
1579 */
1580 $gt = sq_findnxstr($body, $lt, ">");
1581 return Array(false, false, false, $lt, $gt);
1582 }
1583 break;
1584 }
1585
1586 /**
1587 * At this point we're here:
1588 * <tagname attribute='blah'>
1589 * \-------^
1590 *
1591 * At this point we loop in order to find all attributes.
1592 */
1593 $attname = '';
1594 $atttype = false;
1595 $attary = Array();
1596
1597 while ($pos <= strlen($body)){
1598 $pos = sq_skipspace($body, $pos);
1599 if ($pos == strlen($body)){
1600 /**
1601 * Non-closed tag.
1602 */
1603 return Array(false, false, false, $lt, $pos);
1604 }
1605 /**
1606 * See if we arrived at a ">" or "/>", which means that we reached
1607 * the end of the tag.
1608 */
1609 $matches = Array();
1610 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1611 /**
1612 * Yep. So we did.
1613 */
1614 $pos += strlen($matches{1});
1615 if ($matches{2} == "/>"){
1616 $tagtype = 3;
1617 $pos++;
1618 }
1619 return Array($tagname, $attary, $tagtype, $lt, $pos);
1620 }
1621
1622 /**
1623 * There are several types of attributes, with optional
1624 * [:space:] between members.
1625 * Type 1:
1626 * attrname[:space:]=[:space:]'CDATA'
1627 * Type 2:
1628 * attrname[:space:]=[:space:]"CDATA"
1629 * Type 3:
1630 * attr[:space:]=[:space:]CDATA
1631 * Type 4:
1632 * attrname
1633 *
1634 * We leave types 1 and 2 the same, type 3 we check for
1635 * '"' and convert to "&quot" if needed, then wrap in
1636 * double quotes. Type 4 we convert into:
1637 * attrname="yes".
1638 */
1639 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1640 if ($regary == false){
1641 /**
1642 * Looks like body ended before the end of tag.
1643 */
1644 return Array(false, false, false, $lt, strlen($body));
1645 }
1646 list($pos, $attname, $match) = $regary;
1647 $attname = strtolower($attname);
1648 /**
1649 * We arrived at the end of attribute name. Several things possible
1650 * here:
1651 * '>' means the end of the tag and this is attribute type 4
1652 * '/' if followed by '>' means the same thing as above
1653 * '\s' means a lot of things -- look what it's followed by.
1654 * anything else means the attribute is invalid.
1655 */
1656 switch($match){
1657 case '/':
1658 /**
1659 * This is an xhtml-style tag with a closing / at the
1660 * end, like so: <img src="blah" />. Check if it's followed
1661 * by the closing bracket. If not, then this tag is invalid
1662 */
1663 if (substr($body, $pos, 2) == "/>"){
1664 $pos++;
1665 $tagtype = 3;
1666 } else {
1667 $gt = sq_findnxstr($body, $pos, ">");
1668 $retary = Array(false, false, false, $lt, $gt);
1669 return $retary;
1670 }
1671 case '>':
1672 $attary{$attname} = '"yes"';
1673 return Array($tagname, $attary, $tagtype, $lt, $pos);
1674 break;
1675 default:
1676 /**
1677 * Skip whitespace and see what we arrive at.
1678 */
1679 $pos = sq_skipspace($body, $pos);
1680 $char = substr($body, $pos, 1);
1681 /**
1682 * Two things are valid here:
1683 * '=' means this is attribute type 1 2 or 3.
1684 * \w means this was attribute type 4.
1685 * anything else we ignore and re-loop. End of tag and
1686 * invalid stuff will be caught by our checks at the beginning
1687 * of the loop.
1688 */
1689 if ($char == "="){
1690 $pos++;
1691 $pos = sq_skipspace($body, $pos);
1692 /**
1693 * Here are 3 possibilities:
1694 * "'" attribute type 1
1695 * '"' attribute type 2
1696 * everything else is the content of tag type 3
1697 */
1698 $quot = substr($body, $pos, 1);
1699 if ($quot == "'"){
1700 $regary = sq_findnxreg($body, $pos+1, "\'");
1701 if ($regary == false){
1702 return Array(false, false, false, $lt, strlen($body));
1703 }
1704 list($pos, $attval, $match) = $regary;
1705 $pos++;
1706 $attary{$attname} = "'" . $attval . "'";
1707 } else if ($quot == '"'){
1708 $regary = sq_findnxreg($body, $pos+1, '\"');
1709 if ($regary == false){
1710 return Array(false, false, false, $lt, strlen($body));
1711 }
1712 list($pos, $attval, $match) = $regary;
1713 $pos++;
1714 $attary{$attname} = '"' . $attval . '"';
1715 } else {
1716 /**
1717 * These are hateful. Look for \s, or >.
1718 */
1719 $regary = sq_findnxreg($body, $pos, "[\s>]");
1720 if ($regary == false){
1721 return Array(false, false, false, $lt, strlen($body));
1722 }
1723 list($pos, $attval, $match) = $regary;
1724 /**
1725 * If it's ">" it will be caught at the top.
1726 */
1727 $attval = preg_replace("/\"/s", "&quot;", $attval);
1728 $attary{$attname} = '"' . $attval . '"';
1729 }
1730 } else if (preg_match("|[\w/>]|", $char)) {
1731 /**
1732 * That was attribute type 4.
1733 */
1734 $attary{$attname} = '"yes"';
1735 } else {
1736 /**
1737 * An illegal character. Find next '>' and return.
1738 */
1739 $gt = sq_findnxstr($body, $pos, ">");
1740 return Array(false, false, false, $lt, $gt);
1741 }
1742 break;
1743 }
1744 }
1745 /**
1746 * The fact that we got here indicates that the tag end was never
1747 * found. Return invalid tag indication so it gets stripped.
1748 */
1749 return Array(false, false, false, $lt, strlen($body));
1750 }
1751
1752 /**
1753 * Translates entities into literal values so they can be checked.
1754 *
1755 * @param $attvalue the by-ref value to check.
1756 * @param $regex the regular expression to check against.
1757 * @param $hex whether the entites are hexadecimal.
1758 * @return True or False depending on whether there were matches.
1759 */
1760 function sq_deent(&$attvalue, $regex, $hex=false){
1761 $me = 'sq_deent';
1762 $ret_match = false;
1763 // remove comments
1764 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1765 preg_match_all($regex, $attvalue, $matches);
1766 if (is_array($matches) && sizeof($matches[0]) > 0){
1767 $repl = Array();
1768 for ($i = 0; $i < sizeof($matches[0]); $i++){
1769 $numval = $matches[1][$i];
1770 if ($hex){
1771 $numval = hexdec($numval);
1772 }
1773 $repl{$matches[0][$i]} = chr($numval);
1774 }
1775 $attvalue = strtr($attvalue, $repl);
1776 return true;
1777 } else {
1778 return false;
1779 }
1780 }
1781
1782 /**
1783 * This function runs various checks against the attributes.
1784 *
1785 * @param $tagname String with the name of the tag.
1786 * @param $attary Array with all tag attributes.
1787 * @param $rm_attnames See description for sq_sanitize
1788 * @param $bad_attvals See description for sq_sanitize
1789 * @param $add_attr_to_tag See description for sq_sanitize
1790 * @param $message message object
1791 * @param $id message id
1792 * @return Array with modified attributes.
1793 */
1794 function sq_fixatts($tagname,
1795 $attary,
1796 $rm_attnames,
1797 $bad_attvals,
1798 $add_attr_to_tag,
1799 $message,
1800 $id,
1801 $mailbox
1802 ){
1803 $me = 'sq_fixatts';
1804 while (list($attname, $attvalue) = each($attary)){
1805 /**
1806 * See if this attribute should be removed.
1807 */
1808 foreach ($rm_attnames as $matchtag=>$matchattrs){
1809 if (preg_match($matchtag, $tagname)){
1810 foreach ($matchattrs as $matchattr){
1811 if (preg_match($matchattr, $attname)){
1812 unset($attary{$attname});
1813 continue;
1814 }
1815 }
1816 }
1817 }
1818 /**
1819 * Workaround for IE quirks
1820 */
1821 sq_fixIE_idiocy($attvalue);
1822
1823 /**
1824 * Remove any backslashes, entities, and extraneous whitespace.
1825 */
1826
1827 $oldattvalue = $attvalue;
1828 sq_defang($attvalue);
1829 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1830 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1831 // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1832 $attvalue = "idiocy";
1833 $attary{$attname} = $attvalue;
1834 }
1835 sq_unspace($attvalue);
1836
1837 /**
1838 * Now let's run checks on the attvalues.
1839 * I don't expect anyone to comprehend this. If you do,
1840 * get in touch with me so I can drive to where you live and
1841 * shake your hand personally. :)
1842 */
1843 foreach ($bad_attvals as $matchtag=>$matchattrs){
1844 if (preg_match($matchtag, $tagname)){
1845 foreach ($matchattrs as $matchattr=>$valary){
1846 if (preg_match($matchattr, $attname)){
1847 /**
1848 * There are two arrays in valary.
1849 * First is matches.
1850 * Second one is replacements
1851 */
1852 list($valmatch, $valrepl) = $valary;
1853 $newvalue =
1854 preg_replace($valmatch, $valrepl, $attvalue);
1855 if ($newvalue != $attvalue){
1856 $attary{$attname} = $newvalue;
1857 $attvalue = $newvalue;
1858 }
1859 }
1860 }
1861 }
1862 }
1863 if ($attname == 'style') {
1864 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1865 // 8bit and control characters in style attribute values can be used for XSS, remove them
1866 $attary{$attname} = '"disallowed character"';
1867 }
1868 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1869 if (count($aMatch)) {
1870 foreach($aMatch[1] as $sMatch) {
1871 // url value
1872 $urlvalue = $sMatch;
1873 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1874 $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue);
1875 }
1876 }
1877 }
1878 /**
1879 * Use white list based filtering on attributes which can contain url's
1880 */
1881 else if ($attname == 'href' || $attname == 'xlink:href' || $attname == 'src'
1882 || $attname == 'poster' || $attname == 'formaction'
1883 || $attname == 'background' || $attname == 'action') {
1884 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1885 $attary{$attname} = $attvalue;
1886 }
1887 }
1888 /**
1889 * See if we need to append any attributes to this tag.
1890 */
1891 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1892 if (preg_match($matchtag, $tagname)){
1893 $attary = array_merge($attary, $addattary);
1894 }
1895 }
1896 return $attary;
1897 }
1898
1899 /**
1900 * This function filters url's
1901 *
1902 * @param $attvalue String with attribute value to filter
1903 * @param $message message object
1904 * @param $id message id
1905 * @param $mailbox mailbox
1906 * @param $sQuote quoting characters around url's
1907 */
1908 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1909 $attvalue = trim($attvalue);
1910 if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1911 // remove the double quotes
1912 $sQuote = $attvalue[0];
1913 $attvalue = trim(substr($attvalue,1,-1));
1914 }
1915
1916 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1917 // images off by default.
1918 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1919
1920 global $use_transparent_security_image;
1921 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1922 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1923
1924 /**
1925 * Replace empty src tags with the blank image. src is only used
1926 * for frames, images, and image inputs. Doing a replace should
1927 * not affect them working as should be, however it will stop
1928 * IE from being kicked off when src for img tags are not set
1929 */
1930 if ($attvalue == '') {
1931 $attvalue = '"' . SM_PATH . 'images/blank.png"';
1932 } else {
1933 // first, disallow 8 bit characters and control characters
1934 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1935 switch ($attname) {
1936 case 'href':
1937 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1938 break;
1939 default:
1940 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1941 break;
1942 }
1943 } else {
1944 $aUrl = parse_url($attvalue);
1945 if (isset($aUrl['scheme'])) {
1946 switch(strtolower($aUrl['scheme'])) {
1947 case 'mailto':
1948 case 'http':
1949 case 'https':
1950 case 'ftp':
1951 if ($attname != 'href') {
1952 if ($view_unsafe_images == false) {
1953 $attvalue = $sQuote . $secremoveimg . $sQuote;
1954 } else {
1955 if (isset($aUrl['path'])) {
1956
1957 // No one has been able to show that image URIs
1958 // can be exploited, so for now, no restrictions
1959 // are made at all. If this proves to be a problem,
1960 // the commented-out code below can be of help.
1961 // (One consideration is that I see nothing in this
1962 // function that specifically says that we will
1963 // only ever arrive here when inspecting an image
1964 // tag, although that does seem to be the end
1965 // result - e.g., <script src="..."> where malicious
1966 // image URIs are in fact a problem are already
1967 // filtered out elsewhere.
1968 /* ---------------------------------
1969 // validate image extension.
1970 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1971 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1972 // If URI is to something other than
1973 // a regular image file, get the contents
1974 // and try to see if it is an image.
1975 // Don't use Fileinfo (finfo_file()) because
1976 // we'd need to make the admin configure the
1977 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1978 //
1979 $mime_type = '';
1980 if (function_exists('mime_content_type')
1981 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1982
1983 // fetch file
1984 //
1985 $file_contents = '';
1986 while (!feof($FILE)) {
1987 $file_contents .= fread($FILE, 8192);
1988 }
1989 fclose($FILE);
1990
1991 // store file locally
1992 //
1993 global $attachment_dir, $username;
1994 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
1995 $localfilename = GenerateRandomString(32, '', 7);
1996 $full_localfilename = "$hashed_attachment_dir/$localfilename";
1997 while (file_exists($full_localfilename)) {
1998 $localfilename = GenerateRandomString(32, '', 7);
1999 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2000 }
2001 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2002 fwrite($FILE, $file_contents);
2003 fclose($FILE);
2004
2005 // get mime type and remove file
2006 //
2007 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2008 unlink("$hashed_attachment_dir/$localfilename");
2009 }
2010 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2011 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2012 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2013 }
2014 }
2015 --------------------------------- */
2016 } else {
2017 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2018 }
2019 }
2020 } else {
2021 $attvalue = $sQuote . $attvalue . $sQuote;
2022 }
2023 break;
2024 case 'outbind':
2025 /**
2026 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2027 * One day MS might actually make it match something useful, for now, falling
2028 * back to using cid2http, so we can grab the blank.png.
2029 */
2030 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2031 break;
2032 case 'cid':
2033 /**
2034 * Turn cid: urls into http-friendly ones.
2035 */
2036 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2037 break;
2038 default:
2039 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2040 break;
2041 }
2042 } else {
2043 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2044 // parse_url did not lead to satisfying result
2045 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2046 }
2047 }
2048 }
2049 }
2050 }
2051
2052 /**
2053 * This function edits the style definition to make them friendly and
2054 * usable in SquirrelMail.
2055 *
2056 * @param $message the message object
2057 * @param $id the message id
2058 * @param $content a string with whatever is between <style> and </style>
2059 * @param $mailbox the message mailbox
2060 * @return a string with edited content.
2061 */
2062 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2063 $me = 'sq_fixstyle';
2064 // workaround for </style> in between comments
2065 $iCurrentPos = $pos;
2066 $content = '';
2067 $sToken = '';
2068 $bSucces = false;
2069 $bEndTag = false;
2070 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2071 $char = $body{$i};
2072 switch ($char) {
2073 case '<':
2074 $sToken = $char;
2075 break;
2076 case '/':
2077 if ($sToken == '<') {
2078 $sToken .= $char;
2079 $bEndTag = true;
2080 } else {
2081 $content .= $char;
2082 }
2083 break;
2084 case '>':
2085 if ($bEndTag) {
2086 $sToken .= $char;
2087 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2088 $newpos = $i + 1;
2089 $bSucces = true;
2090 break 2;
2091 } else {
2092 $content .= $sToken;
2093 }
2094 $bEndTag = false;
2095 } else {
2096 $content .= $char;
2097 }
2098 break;
2099 case '!':
2100 if ($sToken == '<') {
2101 // possible comment
2102 if (isset($body{$i+2}) && substr($body,$i,3) == '!--') {
2103 $i = strpos($body,'-->',$i+3);
2104 if ($i === false) { // no end comment
2105 $i = strlen($body);
2106 }
2107 $sToken = '';
2108 }
2109 } else {
2110 $content .= $char;
2111 }
2112 break;
2113 default:
2114 if ($bEndTag) {
2115 $sToken .= $char;
2116 } else {
2117 $content .= $char;
2118 }
2119 break;
2120 }
2121 }
2122 if ($bSucces == FALSE){
2123 return array(FALSE, strlen($body));
2124 }
2125
2126
2127
2128 /**
2129 * First look for general BODY style declaration, which would be
2130 * like so:
2131 * body {background: blah-blah}
2132 * and change it to .bodyclass so we can just assign it to a <div>
2133 */
2134 // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2135 // Nah, this is even better - try to preface all CSS selectors with
2136 // our <div> class ID "bodyclass" then correct generic "body" selectors
2137 // TODO: this works pretty good but breaks stuff like this:
2138 // @media print { body { font-size: 10pt; } }
2139 // but there isn't an easy way to make this regex skip @media
2140 // definitions... though lots of the ones in the wild will be
2141 // correctly handled because they tend to end with a parenthesis, like:
2142 // @media screen and (max-width:480px) { ...
2143 $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2144 $content = str_replace('.bodyclass body', '.bodyclass', $content);
2145
2146 global $use_transparent_security_image;
2147 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2148 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2149
2150 /**
2151 * Fix url('blah') declarations.
2152 */
2153 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2154 // "url(\\1$secremoveimg\\2)", $content);
2155
2156 // first check for 8bit sequences and disallowed control characters
2157 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2158 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2159 return array($content, $newpos);
2160 }
2161
2162 // IE Sucks hard. We have a special function for it.
2163 sq_fixIE_idiocy($content);
2164
2165 // remove @import line
2166 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2167
2168 // translate ur\l and variations (IE parses that)
2169 // TODO check if the sq_fixIE_idiocy function already handles this.
2170 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2171 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2172 if (count($aMatch)) {
2173 $aValue = $aReplace = array();
2174 foreach($aMatch[1] as $sMatch) {
2175 // url value
2176 $urlvalue = $sMatch;
2177 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2178 $aValue[] = $sMatch;
2179 $aReplace[] = $urlvalue;
2180 }
2181 $content = str_replace($aValue,$aReplace,$content);
2182 }
2183
2184 /**
2185 * Remove any backslashes, entities, and extraneous whitespace.
2186 */
2187 $contentTemp = $content;
2188 sq_defang($contentTemp);
2189 sq_unspace($contentTemp);
2190
2191 /**
2192 * Fix stupid css declarations which lead to vulnerabilities
2193 * in IE.
2194 *
2195 * Also remove "position" attribute, as it can easily be set
2196 * to "fixed" or "absolute" with "left" and "top" attributes
2197 * of zero, taking over the whole content frame. It can also
2198 * be set to relative and move itself anywhere it wants to,
2199 * displaying content in areas it shouldn't be allowed to touch.
2200 */
2201 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2202 '/expression/i',
2203 '/behaviou*r/i',
2204 '/binding/i',
2205 '/include-source/i',
2206 '/javascript/i',
2207 '/script/i',
2208 '/position/i');
2209 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2210 $contentNew = preg_replace($match, $replace, $contentTemp);
2211 if ($contentNew !== $contentTemp) {
2212 // insecure css declarations are used. From now on we don't care
2213 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2214 $content = $contentNew;
2215 }
2216 return array($content, $newpos);
2217 }
2218
2219
2220 /**
2221 * This function converts cid: url's into the ones that can be viewed in
2222 * the browser.
2223 *
2224 * @param $message the message object
2225 * @param $id the message id
2226 * @param $cidurl the cid: url.
2227 * @param $mailbox the message mailbox
2228 * @return a string with a http-friendly url
2229 */
2230 function sq_cid2http($message, $id, $cidurl, $mailbox){
2231 /**
2232 * Get rid of quotes.
2233 */
2234 $quotchar = substr($cidurl, 0, 1);
2235 if ($quotchar == '"' || $quotchar == "'"){
2236 $cidurl = str_replace($quotchar, "", $cidurl);
2237 } else {
2238 $quotchar = '';
2239 }
2240 $cidurl = substr(trim($cidurl), 4);
2241
2242 $match_str = '/\{.*?\}\//';
2243 $str_rep = '';
2244 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2245
2246 $linkurl = find_ent_id($cidurl, $message);
2247 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2248 unsafe link image */
2249 $httpurl = '';
2250
2251 /**
2252 * This is part of a fix for Outlook Express 6.x generating
2253 * cid URLs without creating content-id headers. These images are
2254 * not part of the multipart/related html mail. The html contains
2255 * <img src="cid:{some_id}/image_filename.ext"> references to
2256 * attached images with as goal to render them inline although
2257 * the attachment disposition property is not inline.
2258 */
2259
2260 if (empty($linkurl)) {
2261 if (preg_match('/{.*}\//', $cidurl)) {
2262 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2263 if (!empty($cidurl)) {
2264 $linkurl = find_ent_id($cidurl, $message);
2265 }
2266 }
2267 }
2268
2269 if (!empty($linkurl)) {
2270 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2271 "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2272 '&amp;ent_id=' . $linkurl . $quotchar;
2273 } else {
2274 /**
2275 * If we couldn't generate a proper img url, drop in a blank image
2276 * instead of sending back empty, otherwise it causes unusual behaviour
2277 */
2278 $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2279 }
2280
2281 return $httpurl;
2282 }
2283
2284 /**
2285 * This function changes the <body> tag into a <div> tag since we
2286 * can't really have a body-within-body.
2287 *
2288 * @param $attary an array of attributes and values of <body>
2289 * @param $mailbox mailbox we're currently reading (for cid2http)
2290 * @param $message current message (for cid2http)
2291 * @param $id current message id (for cid2http)
2292 * @return a modified array of attributes to be set for <div>
2293 */
2294 function sq_body2div($attary, $mailbox, $message, $id){
2295 $me = 'sq_body2div';
2296 $divattary = Array('class' => "'bodyclass'");
2297 $text = '#000000';
2298 $has_bgc_stl = $has_txt_stl = false;
2299 $styledef = '';
2300 if (is_array($attary) && sizeof($attary) > 0){
2301 foreach ($attary as $attname=>$attvalue){
2302 $quotchar = substr($attvalue, 0, 1);
2303 $attvalue = str_replace($quotchar, "", $attvalue);
2304 switch ($attname){
2305 case 'background':
2306 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2307 $styledef .= "background-image: url('$attvalue'); ";
2308 break;
2309 case 'bgcolor':
2310 $has_bgc_stl = true;
2311 $styledef .= "background-color: $attvalue; ";
2312 break;
2313 case 'text':
2314 $has_txt_stl = true;
2315 $styledef .= "color: $attvalue; ";
2316 break;
2317 }
2318 }
2319 // Outlook defines a white bgcolor and no text color. This can lead to
2320 // white text on a white bg with certain themes.
2321 if ($has_bgc_stl && !$has_txt_stl) {
2322 $styledef .= "color: $text; ";
2323 }
2324 if (strlen($styledef) > 0){
2325 $divattary{"style"} = "\"$styledef\"";
2326 }
2327 }
2328 return $divattary;
2329 }
2330
2331 /**
2332 * This is the main function and the one you should actually be calling.
2333 * There are several variables you should be aware of an which need
2334 * special description.
2335 *
2336 * Since the description is quite lengthy, see it here:
2337 * http://linux.duke.edu/projects/mini/htmlfilter/
2338 *
2339 * @param $body the string with HTML you wish to filter
2340 * @param $tag_list see description above
2341 * @param $rm_tags_with_content see description above
2342 * @param $self_closing_tags see description above
2343 * @param $force_tag_closing see description above
2344 * @param $rm_attnames see description above
2345 * @param $bad_attvals see description above
2346 * @param $add_attr_to_tag see description above
2347 * @param $message message object
2348 * @param $id message id
2349 * @param $recursively_called boolean flag for recursive calls into this function (optional; default FALSE)
2350 * @return sanitized html safe to show on your pages.
2351 */
2352 function sq_sanitize($body,
2353 $tag_list,
2354 $rm_tags_with_content,
2355 $self_closing_tags,
2356 $force_tag_closing,
2357 $rm_attnames,
2358 $bad_attvals,
2359 $add_attr_to_tag,
2360 $message,
2361 $id,
2362 $mailbox,
2363 $recursively_called=FALSE
2364 ){
2365 $me = 'sq_sanitize';
2366
2367 /**
2368 * See if tag_list is of tags to remove or tags to allow.
2369 * false means remove these tags
2370 * true means allow these tags
2371 */
2372 $orig_tag_list = $tag_list;
2373 $rm_tags = array_shift($tag_list);
2374
2375 /**
2376 * Normalize rm_tags and rm_tags_with_content.
2377 */
2378 @array_walk($tag_list, 'sq_casenormalize');
2379 @array_walk($rm_tags_with_content, 'sq_casenormalize');
2380 @array_walk($self_closing_tags, 'sq_casenormalize');
2381
2382 $curpos = 0;
2383 $open_tags = Array();
2384 $trusted = "\n<!-- begin sanitized html -->\n";
2385 $skip_content = false;
2386 /**
2387 * Take care of netscape's stupid javascript entities like
2388 * &{alert('boo')};
2389 */
2390 $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2391
2392 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2393 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2394
2395 /**
2396 * RCDATA and RAWTEXT tags are handled differently:
2397 * next instance of closing tag is used, whether or not
2398 * the HTML is well formed before that
2399 */
2400 global $rcdata_rawtext_tags;
2401 if (!$recursively_called
2402 && in_array($tagname, $rcdata_rawtext_tags)
2403 && $tagtype === 1){
2404 $closing_tag = false;
2405 $closing_tag_offset = $curpos;
2406 // seek out the closing tag for the current RCDATA/RAWTEXT tag
2407 while (1) {
2408 // first we need to move forward to next available closing tag
2409 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2410 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2411 if ($next_tag === false) {
2412 $closing_tag = false;
2413 break;
2414 }
2415 // but then we have to make sure it's a well-formed tag
2416 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2417 if ($closing_tag === false)
2418 break;
2419 else if ($closing_tag[0] !== false
2420 // these should be redundant
2421 && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2422 $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos + 1),
2423 $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2424 $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2425 $message, $id, $mailbox, true);
2426 $curpos = $closing_tag[4] + 1;
2427 continue 2;
2428 }
2429 $closing_tag_offset = $next_tag[0] + 1;
2430 }
2431 if ($closing_tag === false)
2432 { /* no-op... there was no closing tag for this RCDATA/RAWTEXT tag - we could probably set $curpos to the end of $body, but this HTML is malformed anyway and should just fall apart on its own */ }
2433 }
2434
2435 $free_content = substr($body, $curpos, $lt-$curpos);
2436 /**
2437 * Take care of <style>
2438 */
2439 if ($tagname == "style" && $tagtype == 1){
2440 list($free_content, $curpos) =
2441 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2442 if ($free_content != FALSE){
2443 if ( !empty($attary) ) {
2444 $attary = sq_fixatts($tagname,
2445 $attary,
2446 $rm_attnames,
2447 $bad_attvals,
2448 $add_attr_to_tag,
2449 $message,
2450 $id,
2451 $mailbox
2452 );
2453 }
2454 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2455 $trusted .= $free_content;
2456 $trusted .= sq_tagprint($tagname, false, 2);
2457 }
2458 continue;
2459 }
2460 if ($skip_content == false){
2461 $trusted .= $free_content;
2462 }
2463 if ($tagname != FALSE){
2464 if ($tagtype == 2){
2465 if ($skip_content == $tagname){
2466 /**
2467 * Got to the end of tag we needed to remove.
2468 */
2469 $tagname = false;
2470 $skip_content = false;
2471 } else {
2472 if ($skip_content == false){
2473 if ($tagname == "body"){
2474 $tagname = "div";
2475 }
2476 if (isset($open_tags{$tagname}) &&
2477 $open_tags{$tagname} > 0){
2478 $open_tags{$tagname}--;
2479 } else {
2480 $tagname = false;
2481 }
2482 }
2483 }
2484 } else {
2485 /**
2486 * $rm_tags_with_content
2487 */
2488 if ($skip_content == false){
2489 /**
2490 * See if this is a self-closing type and change
2491 * tagtype appropriately.
2492 */
2493 if ($tagtype == 1
2494 && in_array($tagname, $self_closing_tags)){
2495 $tagtype = 3;
2496 }
2497 /**
2498 * See if we should skip this tag and any content
2499 * inside it.
2500 */
2501 if ($tagtype == 1 &&
2502 in_array($tagname, $rm_tags_with_content)){
2503 $skip_content = $tagname;
2504 } else {
2505 if (($rm_tags == false
2506 && in_array($tagname, $tag_list)) ||
2507 ($rm_tags == true &&
2508 !in_array($tagname, $tag_list))){
2509 $tagname = false;
2510 } else {
2511 /**
2512 * Convert body into div.
2513 */
2514 if ($tagname == "body"){
2515 $tagname = "div";
2516 $attary = sq_body2div($attary, $mailbox,
2517 $message, $id);
2518 }
2519 if ($tagtype == 1){
2520 if (isset($open_tags{$tagname})){
2521 $open_tags{$tagname}++;
2522 } else {
2523 $open_tags{$tagname}=1;
2524 }
2525 }
2526 /**
2527 * This is where we run other checks.
2528 */
2529 if (is_array($attary) && sizeof($attary) > 0){
2530 $attary = sq_fixatts($tagname,
2531 $attary,
2532 $rm_attnames,
2533 $bad_attvals,
2534 $add_attr_to_tag,
2535 $message,
2536 $id,
2537 $mailbox
2538 );
2539 }
2540 }
2541 }
2542 }
2543 }
2544 if ($tagname != false && $skip_content == false){
2545 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2546 }
2547 }
2548 $curpos = $gt+1;
2549 }
2550 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2551 if ($force_tag_closing == true){
2552 foreach ($open_tags as $tagname=>$opentimes){
2553 while ($opentimes > 0){
2554 $trusted .= '</' . $tagname . '>';
2555 $opentimes--;
2556 }
2557 }
2558 $trusted .= "\n";
2559 }
2560 $trusted .= "<!-- end sanitized html -->\n";
2561 return $trusted;
2562 }
2563
2564 /**
2565 * This is a wrapper function to call html sanitizing routines.
2566 *
2567 * @param $body the body of the message
2568 * @param $id the id of the message
2569
2570 * @param $message
2571 * @param $mailbox
2572 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2573 * into internal SM compose links
2574 * (optional; default = TRUE)
2575 * @return a string with html safe to display in the browser.
2576 */
2577 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2578
2579 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2580
2581 global $attachment_common_show_images, $view_unsafe_images,
2582 $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2583 $remove_rcdata_rawtext_tags_and_content;
2584
2585 $rcdata_rawtext_tags = array(
2586 "noscript",
2587 "noframes",
2588 "noembed",
2589 "textarea",
2590 // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2591 );
2592
2593 /**
2594 * Don't display attached images in HTML mode.
2595 *
2596 * SB: why?
2597 */
2598 $attachment_common_show_images = false;
2599 $tag_list = Array(
2600 false, // remove these tags
2601 "meta",
2602 "html",
2603 "head",
2604 "base",
2605 "link",
2606 "frame",
2607 "iframe",
2608 "plaintext",
2609 "marquee",
2610 );
2611
2612 $rm_tags_with_content = Array(
2613 "script",
2614 "object",
2615 "applet",
2616 "embed",
2617 "title",
2618 "frameset",
2619 "xmp",
2620 "xml",
2621 );
2622 if (!$allow_svg_display)
2623 $rm_tags_with_content[] = 'svg';
2624 /**
2625 * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2626 * case that they are, but if you prefer to remove them and their contents entirely
2627 * (in most cases, should be a safe thing with minimal impact), you can add the
2628 * following to config/config_local.php
2629 * $remove_rcdata_rawtext_tags_and_content = TRUE;
2630 */
2631 if ($remove_rcdata_rawtext_tags_and_content)
2632 $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2633
2634 $self_closing_tags = Array(
2635 "img",
2636 "br",
2637 "hr",
2638 "input",
2639 "outbind",
2640 );
2641
2642 $force_tag_closing = true;
2643
2644 $rm_attnames = Array(
2645 "/.*/" =>
2646 Array(
2647 "/target/i",
2648 "/^on.*/i",
2649 "/^dynsrc/i",
2650 "/^data.*/i",
2651 "/^lowsrc.*/i",
2652 )
2653 );
2654
2655 global $use_transparent_security_image;
2656 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2657 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2658
2659 $bad_attvals = Array(
2660 "/.*/" =>
2661 Array(
2662 "/^src|background/i" =>
2663 Array(
2664 Array(
2665 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2666 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2667 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2668 ),
2669 Array(
2670 "\\1$secremoveimg\\2",
2671 "\\1$secremoveimg\\2",
2672 "\\1$secremoveimg\\2",
2673 )
2674 ),
2675 "/^href|action/i" =>
2676 Array(
2677 Array(
2678 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2679 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2680 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2681 ),
2682 Array(
2683 "\\1#\\1",
2684 "\\1#\\1",
2685 "\\1#\\1"
2686 )
2687 ),
2688 "/^style/i" =>
2689 Array(
2690 Array(
2691 "/\/\*.*\*\//",
2692 "/expression/i",
2693 "/binding/i",
2694 "/behaviou*r/i",
2695 "/include-source/i",
2696
2697 // position:relative can also be exploited
2698 // to put content outside of email body area
2699 // and position:fixed is similarly exploitable
2700 // as position:absolute, so we'll remove it
2701 // altogether....
2702 //
2703 // Does this screw up legitimate HTML messages?
2704 // If so, the only fix I see is to allow position
2705 // attributes (any values? I think we still have
2706 // to block static and fixed) only if $use_iframe
2707 // is enabled (1.5.0+)
2708 //
2709 // was: "/position\s*:\s*absolute/i",
2710 //
2711 "/position\s*:/i",
2712
2713 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2714 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2715 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2716 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2717 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2718 ),
2719 Array(
2720 "",
2721 "idiocy",
2722 "idiocy",
2723 "idiocy",
2724 "idiocy",
2725 "idiocy",
2726 "url",
2727 "url(\\1#\\1)",
2728 "url(\\1#\\1)",
2729 "url(\\1#\\1)",
2730 "\\1:url(\\2#\\3)"
2731 )
2732 )
2733 )
2734 );
2735
2736 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2737 // images off by default.
2738 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2739
2740 if (!$view_unsafe_images){
2741 /**
2742 * Remove any references to http/https if view_unsafe_images set
2743 * to false.
2744 */
2745 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[0],
2746 '/^([\'\"])\s*https*:.*([\'\"])/si');
2747 array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1],
2748 "\\1$secremoveimg\\1");
2749 array_push($bad_attvals{'/.*/'}{'/^style/i'}[0],
2750 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2751 array_push($bad_attvals{'/.*/'}{'/^style/i'}[1],
2752 "url(\\1$secremoveimg\\1)");
2753 }
2754
2755 $add_attr_to_tag = Array(
2756 "/^a$/i" =>
2757 Array('target'=>'"_blank"',
2758 'title'=>'"'._("This external link will open in a new window").'"'
2759 )
2760 );
2761 $trusted = sq_sanitize($body,
2762 $tag_list,
2763 $rm_tags_with_content,
2764 $self_closing_tags,
2765 $force_tag_closing,
2766 $rm_attnames,
2767 $bad_attvals,
2768 $add_attr_to_tag,
2769 $message,
2770 $id,
2771 $mailbox
2772 );
2773 if (strpos($trusted,$secremoveimg)){
2774 $has_unsafe_images = true;
2775 }
2776
2777 // we want to parse mailto's in HTML output, change to SM compose links
2778 // this is a modified version of code from url_parser.php... but Marc is
2779 // right: we need a better filtering implementation; adding this randomly
2780 // here is not a great solution
2781 //
2782 if ($take_mailto_links) {
2783 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2784 global $MailTo_PReg_Match;
2785 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2786 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2787 // quote before mailto, and we'll also try to convert the non-links back into links
2788 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2789 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2790 foreach ($regs[0] as $i => $mailto_before) {
2791 $mailto_params = $regs[11][$i];
2792
2793 // get rid of any leading quote we may have captured but don't care about
2794 //
2795 $mailto_before = ltrim($mailto_before, '"\'');
2796
2797 // get rid of any tailing quote since we have to add send_to to the end
2798 //
2799 $mailto_before = rtrim($mailto_before, '"\'');
2800 $mailto_params = rtrim($mailto_params, '"\'');
2801
2802 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2803 $to = 'to=' . $regs[2][$i];
2804 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2805 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2806 else {
2807 if ($mailto_params) //already some params, append to them
2808 $mailto_params .= '&amp;' . $to;
2809 else
2810 $mailto_params .= '?' . $to;
2811 }
2812 }
2813
2814 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2815
2816 // we'll already have target=_blank, no need to allow comp_in_new
2817 // here (which would be a lot more work anyway)
2818 //
2819 global $compose_new_win;
2820 $temp_comp_in_new = $compose_new_win;
2821 $compose_new_win = 0;
2822 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2823 $compose_new_win = $temp_comp_in_new;
2824
2825 // remove <a href=" and anything after the next quote (we only
2826 // need the uri, not the link HTML) in compose uri
2827 //
2828 // but only do this if the original mailto was in a real anchor tag
2829 //
2830 if (!empty($regs[1][$i])) {
2831 $comp_uri = substr($comp_uri, 9);
2832 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2833 }
2834 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2835 }
2836 }
2837 }
2838
2839 return $trusted;
2840 }
2841
2842 /**
2843 * function SendDownloadHeaders - send file to the browser
2844 *
2845 * Original Source: SM core src/download.php
2846 * moved here to make it available to other code, and separate
2847 * front end from back end functionality.
2848 *
2849 * @param string $type0 first half of mime type
2850 * @param string $type1 second half of mime type
2851 * @param string $filename filename to tell the browser for downloaded file
2852 * @param boolean $force whether to force the download dialog to pop
2853 * @param optional integer $filesize send the Content-Header and length to the browser
2854 * @return void
2855 */
2856 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2857 global $languages, $squirrelmail_language;
2858 $isIE = $isIE6plus = false;
2859
2860 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2861
2862 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2863 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2864 $isIE = true;
2865 }
2866
2867 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2868 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2869 $isIE6plus = true;
2870 }
2871
2872 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2873 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2874 $filename =
2875 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2876 } else {
2877 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2878 }
2879
2880 // A Pox on Microsoft and it's Internet Explorer!
2881 //
2882 // IE has lots of bugs with file downloads.
2883 // It also has problems with SSL. Both of these cause problems
2884 // for us in this function.
2885 //
2886 // See this article on Cache Control headers and SSL
2887 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2888 //
2889 // The best thing you can do for IE is to upgrade to the latest
2890 // version
2891 //set all the Cache Control Headers for IE
2892 if ($isIE) {
2893 $filename=rawurlencode($filename);
2894 header ("Pragma: public");
2895 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2896 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2897 // header ("Cache-Control: post-check=0, pre-check=0", false);
2898 header ("Cache-Control: private");
2899
2900 //set the inline header for IE, we'll add the attachment header later if we need it
2901 header ("Content-Disposition: inline; filename=$filename");
2902 }
2903
2904 if (!$force) {
2905 // Try to show in browser window
2906 header ("Content-Disposition: inline; filename=\"$filename\"");
2907 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2908 } else {
2909 // Try to pop up the "save as" box
2910
2911 // IE makes this hard. It pops up 2 save boxes, or none.
2912 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2913 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2914 // But, according to Microsoft, it is "RFC compliant but doesn't
2915 // take into account some deviations that allowed within the
2916 // specification." Doesn't that mean RFC non-compliant?
2917 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2918
2919 // all browsers need the application/octet-stream header for this
2920 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2921
2922 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2923 // Do not have quotes around filename, but that applied to
2924 // "attachment"... does it apply to inline too?
2925 header ("Content-Disposition: attachment; filename=\"$filename\"");
2926
2927 if ($isIE && !$isIE6plus) {
2928 // This combination seems to work mostly. IE 5.5 SP 1 has
2929 // known issues (see the Microsoft Knowledge Base)
2930
2931 // This works for most types, but doesn't work with Word files
2932 header ("Content-Type: application/download; name=\"$filename\"");
2933 header ("Content-Type: application/force-download; name=\"$filename\"");
2934 // These are spares, just in case. :-)
2935 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2936 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2937 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2938 } else if ($isIE) {
2939 // This is to prevent IE for MIME sniffing and auto open a file in IE
2940 header ("Content-Type: application/force-download; name=\"$filename\"");
2941 } else {
2942 // another application/octet-stream forces download for Netscape
2943 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2944 }
2945 }
2946
2947 //send the content-length header if the calling function provides it
2948 if ($filesize > 0) {
2949 header("Content-Length: $filesize");
2950 }
2951
2952 } // end fn SendDownloadHeaders
Unnamed repository; edit this file 'description' to name the repository.