projects / squirrelmail.git / blob
? search:


3102fc3c17ff9d1ac67fe4dfe0cbf5bb326b4d70
[squirrelmail.git] / functions / mime.php
1 <?php
2
3 /**
4 * mime.php
5 *
6 * This contains the functions necessary to detect and decode MIME
7 * messages.
8 *
9 * @copyright 1999-2021 The SquirrelMail Project Team
10 * @license http://opensource.org/licenses/gpl-license.php GNU Public License
11 * @version $Id$
12 * @package squirrelmail
13 */
14
15 /**
16 * dependency information
17 functions dependency
18 mime_structure
19 class/mime/Message.class.php
20 Message::parseStructure
21 functions/page_header.php
22 displayPageHeader
23 functions/display_messages.php
24 plain_error_message
25 mime_fetch_body
26 functions/imap_general.php
27 sqimap_run_command
28 mime_print_body_lines
29
30
31
32 functions/imap.php
33 functions/attachment_common.php
34 functions/display_messages.php
35
36 magicHtml => url_parser
37 translateText => url_parser
38
39 */
40
41
42 /* -------------------------------------------------------------------------- */
43 /* MIME DECODING */
44 /* -------------------------------------------------------------------------- */
45
46 /**
47 * Get the MIME structure
48 *
49 * This function gets the structure of a message and stores it in the "message" class.
50 * It will return this object for use with all relevant header information and
51 * fully parsed into the standard "message" object format.
52 */
53 function mime_structure ($bodystructure, $flags=array()) {
54
55 /* Isolate the body structure and remove beginning and end parenthesis. */
56 $read = trim(substr ($bodystructure, strpos(strtolower($bodystructure), 'bodystructure') + 13));
57 $read = trim(substr ($read, 0, -1));
58 $i = 0;
59 $msg = Message::parseStructure($read,$i);
60
61 if (!is_object($msg)) {
62 global $color, $mailbox;
63 displayPageHeader( $color, $mailbox );
64 $errormessage = _("SquirrelMail could not decode the bodystructure of the message");
65 $errormessage .= '<br />'._("The bodystructure provided by your IMAP server:").'<br /><br />';
66 $errormessage .= '<pre>' . sm_encode_html_special_chars($read) . '</pre>';
67 plain_error_message( $errormessage );
68 echo '</body></html>';
69 exit;
70 }
71 if (count($flags)) {
72 foreach ($flags as $flag) {
73 //FIXME: please document why it is we have to check the first char of the flag but we then go ahead and do a full string comparison anyway. Is this a speed enhancement? If not, let's keep it simple and just compare the full string and forget the switch block.
74 $char = strtoupper($flag[1]);
75 switch ($char) {
76 case 'S':
77 if (strtolower($flag) == '\\seen') {
78 $msg->is_seen = true;
79 }
80 break;
81 case 'A':
82 if (strtolower($flag) == '\\answered') {
83 $msg->is_answered = true;
84 }
85 break;
86 case 'D':
87 if (strtolower($flag) == '\\deleted') {
88 $msg->is_deleted = true;
89 }
90 break;
91 case 'F':
92 if (strtolower($flag) == '\\flagged') {
93 $msg->is_flagged = true;
94 }
95 else if (strtolower($flag) == '$forwarded') {
96 $msg->is_forwarded = true;
97 }
98 break;
99 case 'M':
100 if (strtolower($flag) == '$mdnsent') {
101 $msg->is_mdnsent = true;
102 }
103 break;
104 default:
105 break;
106 }
107 }
108 }
109 // listEntities($msg);
110 return $msg;
111 }
112
113
114
115 /* This starts the parsing of a particular structure. It is called recursively,
116 * so it can be passed different structures. It returns an object of type
117 * $message.
118 * First, it checks to see if it is a multipart message. If it is, then it
119 * handles that as it sees is necessary. If it is just a regular entity,
120 * then it parses it and adds the necessary header information (by calling out
121 * to mime_get_elements()
122 */
123
124 function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) {
125 /* Do a bit of error correction. If we couldn't find the entity id, just guess
126 * that it is the first one. That is usually the case anyway.
127 */
128
129 if (!$ent_id) {
130 $cmd = "FETCH $id BODY[]";
131 } else {
132 $cmd = "FETCH $id BODY[$ent_id]";
133 }
134
135 if ($fetch_size!=0) $cmd .= "<0.$fetch_size>";
136
137 $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, TRUE);
138 do {
139 $topline = trim(array_shift($data));
140 } while($topline && ($topline[0] == '*') && !preg_match('/\* [0-9]+ FETCH .*BODY.*/i', $topline)) ;
141 // Matching with "BODY" above is difficult: in most cases "FETCH \(BODY" would work
142 // but some servers may put other things in the same result, perhaps something such
143 // as "* 23 FETCH (FLAGS (\Seen) BODY[1] {174}". There is some small chance that
144 // if the character sequence "BODY" appears in a response where it isn't actually
145 // a FETCH response data item name, the current regex will break things. The better
146 // way to do this would be to parse the response correctly and not use a regex.
147
148 $wholemessage = implode('', $data);
149 if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) {
150 $ret = substr($wholemessage, 0, $regs[1]);
151 /* There is some information in the content info header that could be important
152 * in order to parse html messages. Let's get them here.
153 */
154 // if ($ret[0] == '<') {
155 // $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE);
156 // }
157 } else if (preg_match('/"([^"]*)"/', $topline, $regs)) {
158 $ret = $regs[1];
159 } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) {
160 $ret = $wholemessage;
161 } else {
162 global $where, $what, $mailbox, $passed_id, $startMessage;
163 $par = 'mailbox=' . urlencode($mailbox) . '&amp;passed_id=' . $passed_id;
164 if (isset($where) && isset($what)) {
165 $par .= '&amp;where=' . urlencode($where) . '&amp;what=' . urlencode($what);
166 } else {
167 $par .= '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
168 }
169 $par .= '&amp;response=' . urlencode($response) .
170 '&amp;message=' . urlencode($message) .
171 '&amp;topline=' . urlencode($topline);
172
173 echo '<tt><br />' .
174 '<table width="80%"><tr>' .
175 '<tr><td colspan="2">' .
176 _("Body retrieval error. The reason for this is most probably that the message is malformed.") .
177 '</td></tr>' .
178 '<tr><td><b>' . _("Command:") . "</td><td>$cmd</td></tr>" .
179 '<tr><td><b>' . _("Response:") . "</td><td>$response</td></tr>" .
180 '<tr><td><b>' . _("Message:") . "</td><td>$message</td></tr>" .
181 '<tr><td><b>' . _("FETCH line:") . "</td><td>$topline</td></tr>" .
182 "</table><br /></tt></font><hr />";
183
184 $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE);
185 array_shift($data);
186 $wholemessage = implode('', $data);
187
188 $ret = $wholemessage;
189 }
190 return $ret;
191 }
192
193 // TODO: Needs documentation. $ent_id default is usually 1
194 function mime_print_body_lines ($imap_stream, $id, $ent_id, $encoding, $rStream='php://stdout', $force_crlf='') {
195
196 /* Don't kill the connection if the browser is over a dialup
197 * and it would take over 30 seconds to download it.
198 * Don't call set_time_limit in safe mode.
199 */
200
201 if (!ini_get('safe_mode')) {
202 set_time_limit(0);
203 }
204 /* in case of base64 encoded attachments, do not buffer them.
205 Instead, echo the decoded attachment directly to screen */
206 if (strtolower($encoding) == 'base64') {
207 if (!$ent_id) {
208 $query = "FETCH $id BODY[]";
209 } else {
210 $query = "FETCH $id BODY[$ent_id]";
211 }
212 sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true);
213 } else {
214 $body = mime_fetch_body ($imap_stream, $id, $ent_id);
215 if (is_resource($rStream)) {
216 fputs($rStream,decodeBody($body, $encoding, $force_crlf));
217 } else {
218 echo decodeBody($body, $encoding, $force_crlf);
219 }
220 }
221
222 /*
223 TODO, use the same method for quoted printable.
224 However, I assume that quoted printable attachments aren't that large
225 so the performancegain / memory usage drop will be minimal.
226 If we decide to add that then we need to adapt sqimap_fread because
227 we need to split te result on \n and fread doesn't stop at \n. That
228 means we also should provide $results from sqimap_fread (by ref) to
229 te function and set $no_return to false. The $filter function for
230 quoted printable should handle unsetting of $results.
231 */
232 /*
233 TODO 2: find out how we write to the output stream php://stdout. fwrite
234 doesn't work because 'php://stdout isn't a stream.
235 */
236
237 return;
238 }
239
240 /* -[ END MIME DECODING ]----------------------------------------------------------- */
241
242 /* This is here for debugging purposes. It will print out a list
243 * of all the entity IDs that are in the $message object.
244 */
245 function listEntities ($message) {
246 if ($message) {
247 echo "<tt>" . $message->entity_id . ' : ' . $message->type0 . '/' . $message->type1 . ' parent = '. $message->parent->entity_id. '<br />';
248 for ($i = 0; isset($message->entities[$i]); $i++) {
249 echo "$i : ";
250 $msg = listEntities($message->entities[$i]);
251
252 if ($msg) {
253 echo "return: ";
254 return $msg;
255 }
256 }
257 }
258 }
259
260 function getPriorityStr($priority) {
261 $priority_level = substr($priority,0,1);
262
263 switch($priority_level) {
264 /* Check for a higher then normal priority. */
265 case '1':
266 case '2':
267 $priority_string = _("High");
268 break;
269
270 /* Check for a lower then normal priority. */
271 case '4':
272 case '5':
273 $priority_string = _("Low");
274 break;
275
276 /* Check for a normal priority. */
277 case '3':
278 default:
279 $priority_level = '3';
280 $priority_string = _("Normal");
281 break;
282
283 }
284 return $priority_string;
285 }
286
287 /* returns a $message object for a particular entity id */
288 function getEntity ($message, $ent_id) {
289 return $message->getEntity($ent_id);
290 }
291
292 /* translateText
293 * Extracted from strings.php 23/03/2002
294 */
295
296 function translateText(&$body, $wrap_at, $charset) {
297 global $where, $what; /* from searching */
298 global $color; /* color theme */
299
300 // require_once(SM_PATH . 'functions/url_parser.php');
301
302 $body_ary = explode("\n", $body);
303 for ($i=0; $i < count($body_ary); $i++) {
304 $line = rtrim($body_ary[$i],"\r");
305
306 if (strlen($line) - 2 >= $wrap_at) {
307 sqWordWrap($line, $wrap_at, $charset);
308 }
309 $line = charset_decode($charset, $line);
310 $line = str_replace("\t", ' ', $line);
311
312 parseUrl ($line);
313
314 $quotes = 0;
315 $pos = 0;
316 $j = strlen($line);
317
318 while ($pos < $j) {
319 if ($line[$pos] == ' ') {
320 $pos++;
321 } else if (strpos($line, '&gt;', $pos) === $pos) {
322 $pos += 4;
323 $quotes++;
324 } else {
325 break;
326 }
327 }
328
329 if ($quotes % 2) {
330 $line = '<span class="quote1">' . $line . '</span>';
331 } elseif ($quotes) {
332 $line = '<span class="quote2">' . $line . '</span>';
333 }
334
335 $body_ary[$i] = $line;
336 }
337 $body = '<pre>' . implode("\n", $body_ary) . '</pre>';
338 }
339
340 /**
341 * This returns a parsed string called $body. That string can then
342 * be displayed as the actual message in the HTML. It contains
343 * everything needed, including HTML Tags, Attachments at the
344 * bottom, etc.
345 *
346 * Since 1.2.0 function uses message_body hook.
347 * Till 1.3.0 function included output of formatAttachments().
348 *
349 * @param resource $imap_stream imap connection resource
350 * @param object $message squirrelmail message object
351 * @param array $color squirrelmail color theme array
352 * @param integer $wrap_at number of characters per line
353 * @param string $ent_num (since 1.3.0) message part id
354 * @param integer $id (since 1.3.0) message id
355 * @param string $mailbox (since 1.3.0) imap folder name
356 * @return string html formated message text
357 */
358 function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') {
359 /* This if statement checks for the entity to show as the
360 * primary message. To add more of them, just put them in the
361 * order that is their priority.
362 */
363 global $startMessage, $languages, $squirrelmail_language,
364 $show_html_default, $sort, $has_unsafe_images, $passed_ent_id,
365 $use_iframe, $iframe_height, $download_and_unsafe_link,
366 $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text,
367 $oTemplate, $nbsp;
368
369 // workaround for not updated config.php
370 if (! isset($use_iframe)) $use_iframe = false;
371
372 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
373 // images off by default.
374 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
375
376 $body = '';
377 $urlmailbox = urlencode($mailbox);
378 $body_message = getEntity($message, $ent_num);
379 if (($body_message->header->type0 == 'text') ||
380 ($body_message->header->type0 == 'rfc822')) {
381 $body = mime_fetch_body ($imap_stream, $id, $ent_num);
382 $body = decodeBody($body, $body_message->header->encoding);
383
384 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
385 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) {
386 if (mb_detect_encoding($body) != 'ASCII') {
387 $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body);
388 }
389 }
390
391 /* As of 1.5.2, $body is passed (and modified) by reference */
392 do_hook('message_body', $body);
393
394 /* If there are other types that shouldn't be formatted, add
395 * them here.
396 */
397
398 if ($body_message->header->type1 == 'html') {
399 if ($show_html_default <> 1) {
400 $entity_conv = array('&nbsp;' => ' ',
401 '<p>' => "\n",
402 '<P>' => "\n",
403 '<br>' => "\n",
404 '<BR>' => "\n",
405 '<br />' => "\n",
406 '<BR />' => "\n",
407 '&gt;' => '>',
408 '&lt;' => '<');
409 $body = strtr($body, $entity_conv);
410 $body = strip_tags($body);
411 $body = trim($body);
412 translateText($body, $wrap_at,
413 $body_message->header->getParameter('charset'));
414 } elseif ($use_iframe) {
415 /**
416 * If we don't add html message between iframe tags,
417 * we must detect unsafe images and modify $has_unsafe_images.
418 */
419 $html_body = magicHTML($body, $id, $message, $mailbox);
420 // Convert character set in order to display html mails in different character set
421 $html_body = charset_decode($body_message->header->getParameter('charset'),$html_body,false,true);
422
423 // creating iframe url
424 $iframeurl=sqm_baseuri().'src/view_html.php?'
425 . 'mailbox=' . $urlmailbox
426 . '&amp;passed_id=' . $id
427 . '&amp;ent_id=' . $ent_num
428 . '&amp;view_unsafe_images=' . (int) $view_unsafe_images;
429
430 global $oTemplate;
431 $oTemplate->assign('iframe_url', $iframeurl);
432 $oTemplate->assign('iframe_height', $iframe_height);
433 $oTemplate->assign('html_body', $html_body);
434
435 $body = $oTemplate->fetch('read_html_iframe.tpl');
436 } else {
437 // old way of html rendering
438 /**
439 * convert character set. charset_decode does not remove html special chars
440 * applied by magicHTML functions and does not sanitize them second time if
441 * fourth argument is true.
442 */
443 $charset = $body_message->header->getParameter('charset');
444 if (!empty($charset)) {
445 $body = charset_decode($charset,$body,false,true);
446 }
447 $body = magicHTML($body, $id, $message, $mailbox);
448 }
449 } else {
450 translateText($body, $wrap_at,
451 $body_message->header->getParameter('charset'));
452 }
453
454 /*
455 * Previously the links for downloading and unsafe images were printed
456 * under the mail. By putting the links in a global variable we can
457 * print it in the toolbar where it belongs. Since the original code was
458 * in this place it's left here. It might be possible to move it to some
459 * other place if that makes sense. The possibility to do so has not
460 * been evaluated yet.
461 */
462
463 // Initialize the global variable to an empty string.
464 // FIXME: To have $download_and_unsafe_link as a global variable might not be needed since the use of separate variables ($download_href, $unsafe_image_toggle_href, and $unsafe_image_toggle_text) for the templates was introduced.
465 $download_and_unsafe_link = '';
466
467 // Prepare and build a link for downloading the mail.
468 $link = 'passed_id=' . $id . '&amp;ent_id='.$ent_num.
469 '&amp;mailbox=' . $urlmailbox .'&amp;sort=' . $sort .
470 '&amp;startMessage=' . $startMessage . '&amp;show_more=0';
471 if (isset($passed_ent_id)) {
472 $link .= '&amp;passed_ent_id='.$passed_ent_id;
473 }
474 $download_href = SM_PATH . 'src/download.php?absolute_dl=true&amp;' . $link;
475
476 // Always add the link for downloading the mail as a file to the global
477 // variable.
478 $download_and_unsafe_link .= "$nbsp|$nbsp"
479 . create_hyperlink($download_href, _("Download this as a file"));
480
481 // Find out the right text to use in the link depending on the
482 // circumstances. If the unsafe images are displayed the link should
483 // hide them, if they aren't displayed the link should only appear if
484 // the mail really contains unsafe images.
485 if ($view_unsafe_images) {
486 $text = _("Hide Unsafe Images");
487 } else {
488 if (isset($has_unsafe_images) && $has_unsafe_images) {
489 $link .= '&amp;view_unsafe_images=1';
490 $text = _("View Unsafe Images");
491 } else {
492 $text = '';
493 }
494 }
495
496 // Only create a link for unsafe images if there's need for one. If so:
497 // add it to the global variable.
498 if($text != '') {
499 $unsafe_image_toggle_href = SM_PATH . 'src/read_body.php?'.$link;
500 $unsafe_image_toggle_text = $text;
501 $download_and_unsafe_link .= "$nbsp|$nbsp"
502 . create_hyperlink($unsafe_image_toggle_href, $text);
503 }
504 }
505 return $body;
506 }
507
508 /**
509 * Generate attachments array for passing to templates.
510 *
511 * @since 1.5.2
512 * @param object $message SquirrelMail message object
513 * @param array $exclude_id message parts that are not attachments.
514 * @param string $mailbox mailbox name
515 * @param integer $id message id
516 */
517 function buildAttachmentArray($message, $exclude_id, $mailbox, $id) {
518 global $where, $what, $startMessage, $color, $passed_ent_id,
519 $base_uri, $block_svg_download;
520
521 $att_ar = $message->getAttachments($exclude_id);
522 $urlMailbox = urlencode($mailbox);
523
524 $attachments = array();
525 foreach ($att_ar as $att) {
526 $ent = $att->entity_id;
527 $header = $att->header;
528 $type0 = strtolower($header->type0);
529 $type1 = strtolower($header->type1);
530 if ($block_svg_download && strpos($type1, 'svg') === 0)
531 continue;
532
533 $name = '';
534 $links = array();
535 $links['download link']['text'] = _("Download");
536 $links['download link']['href'] = $base_uri .
537 "src/download.php?absolute_dl=true&amp;passed_id=$id&amp;mailbox=$urlMailbox&amp;ent_id=$ent";
538
539 if ($type0 =='message' && $type1 == 'rfc822') {
540 $default_page = $base_uri . 'src/read_body.php';
541 $rfc822_header = $att->rfc822_header;
542 $filename = $rfc822_header->subject;
543 if (trim( $filename ) == '') {
544 $filename = 'untitled-[' . $ent . ']' ;
545 }
546 $from_o = $rfc822_header->from;
547 if (is_object($from_o)) {
548 $from_name = decodeHeader($from_o->getAddress(false));
549 } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) {
550 // something weird happens when a digest message is opened and you return to the digest
551 // now the from object is part of an array. Probably the parseHeader call overwrites the info
552 // retrieved from the bodystructure in a different way. We need to fix this later.
553 // possible starting point, do not fetch header we already have and inspect how
554 // the rfc822_header object behaves.
555 $from_name = decodeHeader($from_o[0]->getAddress(false));
556 } else {
557 $from_name = _("Unknown sender");
558 }
559 $description = _("From").': '.$from_name;
560 } else {
561 $default_page = $base_uri . 'src/download.php';
562 $filename = $att->getFilename();
563 if ($header->description) {
564 $description = decodeHeader($header->description);
565 } else {
566 $description = '';
567 }
568 }
569
570 $display_filename = $filename;
571 if (isset($passed_ent_id)) {
572 $passed_ent_id_link = '&amp;passed_ent_id='.$passed_ent_id;
573 } else {
574 $passed_ent_id_link = '';
575 }
576 $defaultlink = $default_page . "?startMessage=$startMessage"
577 . "&amp;passed_id=$id&amp;mailbox=$urlMailbox"
578 . '&amp;ent_id='.$ent.$passed_ent_id_link;
579 if ($where && $what) {
580 $defaultlink .= '&amp;where='. urlencode($where).'&amp;what='.urlencode($what);
581 }
582 // IE does make use of mime content sniffing. Forcing a download
583 // prohibit execution of XSS inside an application/octet-stream attachment
584 if ($type0 == 'application' && $type1 == 'octet-stream') {
585 $defaultlink .= '&amp;absolute_dl=true';
586 }
587
588 /* This executes the attachment hook with a specific MIME-type.
589 * It also allows plugins to run if there's a rule for a more
590 * generic type. Finally, a hook for ALL attachment types is
591 * run as well.
592 */
593 // First remember the default link.
594 $defaultlink_orig = $defaultlink;
595
596 /* The API for this hook has changed as of 1.5.2 so that all plugin
597 arguments are passed in an array instead of each their own plugin
598 argument, and arguments are passed by reference, so instead of
599 returning any changes, changes should simply be made to the original
600 arguments themselves. */
601 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
602 &$defaultlink, &$display_filename, &$where, &$what,
603 &$type0, &$type1);
604 do_hook("attachment $type0/$type1", $temp);
605 /* The API for this hook has changed as of 1.5.2 so that all plugin
606 arguments are passed in an array instead of each their own plugin
607 argument, and arguments are passed by reference, so instead of
608 returning any changes, changes should simply be made to the original
609 arguments themselves. */
610 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
611 &$defaultlink, &$display_filename, &$where, &$what,
612 &$type0, &$type1);
613 // Do not let a generic plugin change the default link if a more
614 // specialized one already did it...
615 if ($defaultlink != $defaultlink_orig) {
616 $dummy = '';
617 $temp[5] = &$dummy;
618 }
619 do_hook("attachment $type0/*", $temp);
620 /* The API for this hook has changed as of 1.5.2 so that all plugin
621 arguments are passed in an array instead of each their own plugin
622 argument, and arguments are passed by reference, so instead of
623 returning any changes, changes should simply be made to the original
624 arguments themselves. */
625 $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent,
626 &$defaultlink, &$display_filename, &$where, &$what,
627 &$type0, &$type1);
628 // Do not let a generic plugin change the default link if a more
629 // specialized one already did it...
630 if ($defaultlink != $defaultlink_orig) {
631 $dummy = '';
632 $temp[5] = &$dummy;
633 }
634 do_hook("attachment */*", $temp);
635
636 $this_attachment = array();
637 $this_attachment['Name'] = decodeHeader($display_filename);
638 $this_attachment['Description'] = $description;
639 $this_attachment['DefaultHREF'] = $defaultlink;
640 $this_attachment['DownloadHREF'] = $links['download link']['href'];
641 $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : '';
642
643 // base64 encoded file sizes are misleading, so approximate real size
644 if (!empty($header->encoding) && strtolower($header->encoding) == 'base64')
645 $this_attachment['Size'] = $header->size / 4 * 3;
646 else
647 $this_attachment['Size'] = $header->size;
648
649 $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1);
650 $this_attachment['OtherLinks'] = array();
651 foreach ($links as $val) {
652 if ($val['text']==_("Download")) {
653 $this_attachment['DownloadHREF'] = $val['href'];
654 continue;
655 }
656 if ($val['text']==_("View")) {
657 $this_attachment['ViewHREF'] = $val['href'];
658 continue;
659 }
660
661 // This makes no sense - If 'text' and 'extra' are just concatenated,
662 // there is no point in having 'extra'.... I am going to assume this
663 // was a mistake and am changing 'extra' to be what I think it was
664 // meant to be: additional tag attributes. However, I'm not checking
665 // extensively for plugins that were using this the wrong way (but why would they?)
666 if (empty($val['text']))
667 continue;
668
669 $temp = array();
670 $temp['HREF'] = $val['href'];
671 $temp['Text'] = $val['text'];
672 $temp['Extra'] = (empty($val['extra']) ? '' : $val['extra']);
673 $this_attachment['OtherLinks'][] = $temp;
674 }
675 $attachments[] = $this_attachment;
676
677 unset($links);
678 }
679
680 return $attachments;
681 }
682
683 /**
684 * Displays attachment links and information
685 *
686 * Since 1.3.0 function is not included in formatBody() call.
687 *
688 * Since 1.0.2 uses attachment $type0/$type1 hook.
689 * Since 1.2.5 uses attachment $type0/* hook.
690 * Since 1.5.0 uses attachments_bottom hook.
691 * Since 1.5.2 uses templates and does *not* return a value.
692 *
693 * @param object $message SquirrelMail message object
694 * @param array $exclude_id message parts that are not attachments.
695 * @param string $mailbox mailbox name
696 * @param integer $id message id
697 */
698 function formatAttachments($message, $exclude_id, $mailbox, $id) {
699 global $oTemplate;
700
701 $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id);
702
703 $oTemplate->assign('attachments', $attach);
704 $oTemplate->display('read_attachments.tpl');
705 }
706
707 function sqimap_base64_decode(&$string) {
708
709 // Base64 encoded data goes in pairs of 4 bytes. To achieve on the
710 // fly decoding (to reduce memory usage) you have to check if the
711 // data has incomplete pairs
712
713 // Remove the noise in order to check if the 4 bytes pairs are complete
714 $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string);
715
716 $sStringRem = '';
717 $iMod = strlen($string) % 4;
718 if ($iMod) {
719 $sStringRem = substr($string,-$iMod);
720 // Check if $sStringRem contains padding characters
721 if (substr($sStringRem,-1) != '=') {
722 $string = substr($string,0,-$iMod);
723 } else {
724 $sStringRem = '';
725 }
726 }
727 $string = base64_decode($string);
728 return $sStringRem;
729 }
730
731 /**
732 * Decodes encoded string (usually message body)
733 *
734 * This function decodes a string (usually the message body)
735 * depending on the encoding type. Currently quoted-printable
736 * and base64 encodings are supported.
737 *
738 * The decode_body hook was added to this function in 1.4.2/1.5.0.
739 * The $force_crlf parameter was added in 1.5.2.
740 *
741 * @param string $string The encoded string
742 * @param string $encoding used encoding
743 * @param string $force_crlf Whether or not to force CRLF or LF
744 * line endings (or to leave as is).
745 * If given as "LF", line endings will
746 * all be converted to LF; if "CRLF",
747 * line endings will all be converted
748 * to CRLF. If given as an empty value,
749 * the global $force_crlf_default will
750 * be consulted (it can be specified in
751 * config/config_local.php). Otherwise,
752 * any other value will cause the string
753 * to be left alone. Note that this will
754 * be overridden to "LF" if not using at
755 * least PHP version 4.3.0. (OPTIONAL;
756 * default is empty - consult global
757 * default value)
758 *
759 * @return string The decoded string
760 *
761 * @since 1.0
762 *
763 */
764 function decodeBody($string, $encoding, $force_crlf='') {
765
766 global $force_crlf_default;
767 if (empty($force_crlf)) $force_crlf = $force_crlf_default;
768 $force_crlf = strtoupper($force_crlf);
769
770 // must force line endings to LF due to broken
771 // quoted_printable_decode() in PHP versions
772 // before 4.3.0 (see below)
773 //
774 if (!check_php_version(4, 3, 0) || $force_crlf == 'LF')
775 $string = str_replace("\r\n", "\n", $string);
776 else if ($force_crlf == 'CRLF')
777 $string = str_replace("\n", "\r\n", $string);
778
779 $encoding = strtolower($encoding);
780
781 $encoding_handler = do_hook('decode_body', $encoding);
782
783
784 // plugins get first shot at decoding the string
785 //
786 if (!empty($encoding_handler) && function_exists($encoding_handler)) {
787 $string = $encoding_handler('decode', $string);
788
789 } elseif ($encoding == 'quoted-printable' ||
790 $encoding == 'quoted_printable') {
791
792 // quoted_printable_decode() function is broken in older
793 // php versions. Text with \r\n decoding was fixed only
794 // in php 4.3.0. Minimal code requirement is PHP 4.0.4+
795 // and the above call to: str_replace("\r\n", "\n", $string);
796 //
797 $string = quoted_printable_decode($string);
798
799 } elseif ($encoding == 'base64') {
800 $string = base64_decode($string);
801 }
802
803 // All other encodings are returned raw.
804 return $string;
805 }
806
807 /**
808 * Decodes headers
809 *
810 * This function decodes strings that are encoded according to
811 * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text).
812 * Patched by Christian Schmidt <christian@ostenfeld.dk> 23/03/2002
813 *
814 * @param string $string header string that has to be made readable
815 * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true
816 * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true
817 * @param boolean $decide decide if string can be utfencoded. defaults to false
818 * @return string decoded header string
819 */
820 function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) {
821 global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages;
822 if (is_array($string)) {
823 $string = implode("\n", $string);
824 }
825
826 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
827 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) {
828 $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string);
829 // Do we need to return at this point?
830 // return $string;
831 }
832 $i = 0;
833 $iLastMatch = -2;
834 $encoded = true;
835
836 // FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings
837 $aString = explode(' ',$string);
838 $ret = '';
839 foreach ($aString as $chunk) {
840 if ($encoded && $chunk === '') {
841 continue;
842 } elseif ($chunk === '') {
843 $ret .= ' ';
844 continue;
845 }
846 $encoded = false;
847 /* if encoded words are not separated by a linear-space-white we still catch them */
848 $j = $i-1;
849
850 while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) {
851 /* if the last chunk isn't an encoded string then put back the space, otherwise don't */
852 if ($iLastMatch !== $j) {
853 if ($htmlsafe) {
854 $ret .= '&#32;';
855 } else {
856 $ret .= ' ';
857 }
858 }
859 $iLastMatch = $i;
860 $j = $i;
861 if ($htmlsafe) {
862 $ret .= sm_encode_html_special_chars($res[1]);
863 } else {
864 $ret .= $res[1];
865 }
866 $encoding = ucfirst($res[3]);
867
868 /* decide about valid decoding */
869 if ($decide && is_conversion_safe($res[2])) {
870 $utfencode=true;
871 $can_be_encoded=true;
872 } else {
873 $can_be_encoded=false;
874 }
875 switch ($encoding)
876 {
877 case 'B':
878 // fix broken base64-encoded strings (remove end = padding,
879 // change any = to + in middle of string, add padding back
880 // to the end)
881 if ($fix_broken_base64_encoded_messages) {
882 $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+');
883 $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '=');
884 }
885 $replace = base64_decode($res[4]);
886 if ($utfencode) {
887 if ($can_be_encoded) {
888 /* convert string to different charset,
889 * if functions asks for it (usually in compose)
890 */
891 $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe);
892 } else {
893 // convert string to html codes in order to display it
894 $ret .= charset_decode($res[2],$replace);
895 }
896 } else {
897 if ($htmlsafe) {
898 $replace = sm_encode_html_special_chars($replace);
899 }
900 $ret.= $replace;
901 }
902 break;
903 case 'Q':
904 $replace = str_replace('_', ' ', $res[4]);
905 $replace = preg_replace_callback('/=([0-9a-f]{2})/i',
906 create_function ('$matches', 'return chr(hexdec($matches[1]));'),
907 $replace);
908 if ($utfencode) {
909 if ($can_be_encoded) {
910 /* convert string to different charset,
911 * if functions asks for it (usually in compose)
912 */
913 $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe);
914 } else {
915 // convert string to html codes in order to display it
916 $replace = charset_decode($res[2], $replace);
917 }
918 } else {
919 if ($htmlsafe) {
920 $replace = sm_encode_html_special_chars($replace);
921 }
922 }
923 $ret .= $replace;
924 break;
925 default:
926 break;
927 }
928 $chunk = $res[5];
929 $encoded = true;
930 }
931 if (!$encoded) {
932 if ($htmlsafe) {
933 $ret .= '&#32;';
934 } else {
935 $ret .= ' ';
936 }
937 }
938
939 if (!$encoded && $htmlsafe) {
940 $ret .= sm_encode_html_special_chars($chunk);
941 } else {
942 $ret .= $chunk;
943 }
944 ++$i;
945 }
946 /* remove the first added space */
947 if ($ret) {
948 if ($htmlsafe) {
949 $ret = substr($ret,5);
950 } else {
951 $ret = substr($ret,1);
952 }
953 }
954
955 return $ret;
956 }
957
958 /**
959 * Encodes header
960 *
961 * Function uses XTRA_CODE _encodeheader function, if such function exists.
962 *
963 * Function uses Q encoding by default and encodes a string according to RFC
964 * 1522 for use in headers if it contains 8-bit characters or anything that
965 * looks like it should be encoded.
966 *
967 * Function switches to B encoding and encodeHeaderBase64() function, if
968 * string is 8bit and multibyte character set supported by mbstring extension
969 * is used. It can cause E_USER_NOTICE errors, if interface is used with
970 * multibyte character set unsupported by mbstring extension.
971 *
972 * @param string $string header string, that has to be encoded
973 * @return string quoted-printable encoded string
974 * @todo make $mb_charsets system wide constant
975 */
976 function encodeHeader ($string) {
977 global $default_charset, $languages, $squirrelmail_language;
978
979 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
980 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) {
981 return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string);
982 }
983
984 // Use B encoding for multibyte charsets
985 $mb_charsets = array('utf-8','big5','gb2313','euc-kr');
986 if (in_array($default_charset,$mb_charsets) &&
987 in_array($default_charset,sq_mb_list_encodings()) &&
988 sq_is8bit($string)) {
989 return encodeHeaderBase64($string,$default_charset);
990 } elseif (in_array($default_charset,$mb_charsets) &&
991 sq_is8bit($string) &&
992 ! in_array($default_charset,sq_mb_list_encodings())) {
993 // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php)
994 // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE);
995 }
996
997 // Encode only if the string contains 8-bit characters or =?
998 $j = strlen($string);
999 $max_l = 75 - strlen($default_charset) - 7;
1000 $aRet = array();
1001 $ret = '';
1002 $iEncStart = $enc_init = false;
1003 $cur_l = $iOffset = 0;
1004 for($i = 0; $i < $j; ++$i) {
1005 switch($string[$i])
1006 {
1007 case '"':
1008 case '=':
1009 case '<':
1010 case '>':
1011 case ',':
1012 case '?':
1013 case '_':
1014 if ($iEncStart === false) {
1015 $iEncStart = $i;
1016 }
1017 $cur_l+=3;
1018 if ($cur_l > ($max_l-2)) {
1019 /* if there is an stringpart that doesn't need encoding, add it */
1020 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1021 $aRet[] = "=?$default_charset?Q?$ret?=";
1022 $iOffset = $i;
1023 $cur_l = 0;
1024 $ret = '';
1025 $iEncStart = false;
1026 } else {
1027 $ret .= sprintf("=%02X",ord($string[$i]));
1028 }
1029 break;
1030 case '(':
1031 case ')':
1032 if ($iEncStart !== false) {
1033 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1034 $aRet[] = "=?$default_charset?Q?$ret?=";
1035 $iOffset = $i;
1036 $cur_l = 0;
1037 $ret = '';
1038 $iEncStart = false;
1039 }
1040 break;
1041 case ' ':
1042 if ($iEncStart !== false) {
1043 $cur_l++;
1044 if ($cur_l > $max_l) {
1045 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1046 $aRet[] = "=?$default_charset?Q?$ret?=";
1047 $iOffset = $i;
1048 $cur_l = 0;
1049 $ret = '';
1050 $iEncStart = false;
1051 } else {
1052 $ret .= '_';
1053 }
1054 }
1055 break;
1056 default:
1057 $k = ord($string[$i]);
1058 if ($k > 126) {
1059 if ($iEncStart === false) {
1060 // do not start encoding in the middle of a string, also take the rest of the word.
1061 $sLeadString = substr($string,0,$i);
1062 $aLeadString = explode(' ',$sLeadString);
1063 $sToBeEncoded = array_pop($aLeadString);
1064 $iEncStart = $i - strlen($sToBeEncoded);
1065 $ret .= $sToBeEncoded;
1066 $cur_l += strlen($sToBeEncoded);
1067 }
1068 $cur_l += 3;
1069 /* first we add the encoded string that reached it's max size */
1070 if ($cur_l > ($max_l-2)) {
1071 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1072 $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */
1073 $cur_l = 3;
1074 $ret = '';
1075 $iOffset = $i;
1076 $iEncStart = $i;
1077 }
1078 $enc_init = true;
1079 $ret .= sprintf("=%02X", $k);
1080 } else {
1081 if ($iEncStart !== false) {
1082 $cur_l++;
1083 if ($cur_l > $max_l) {
1084 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1085 $aRet[] = "=?$default_charset?Q?$ret?=";
1086 $iEncStart = false;
1087 $iOffset = $i;
1088 $cur_l = 0;
1089 $ret = '';
1090 } else {
1091 $ret .= $string[$i];
1092 }
1093 }
1094 }
1095 break;
1096 }
1097 }
1098
1099 if ($enc_init) {
1100 if ($iEncStart !== false) {
1101 $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset);
1102 $aRet[] = "=?$default_charset?Q?$ret?=";
1103 } else {
1104 $aRet[] = substr($string,$iOffset);
1105 }
1106 $string = implode('',$aRet);
1107 }
1108 return $string;
1109 }
1110
1111 /**
1112 * Encodes string according to rfc2047 B encoding header formating rules
1113 *
1114 * It is recommended way to encode headers with character sets that store
1115 * symbols in more than one byte.
1116 *
1117 * Function requires mbstring support. If required mbstring functions are missing,
1118 * function returns false and sets E_USER_WARNING level error message.
1119 *
1120 * Minimal requirements - php 4.0.6 with mbstring extension. Please note,
1121 * that mbstring functions will generate E_WARNING errors, if unsupported
1122 * character set is used. mb_encode_mimeheader function provided by php
1123 * mbstring extension is not used in order to get better control of header
1124 * encoding.
1125 *
1126 * Used php code functions - function_exists(), trigger_error(), strlen()
1127 * (is used with charset names and base64 strings). Used php mbstring
1128 * functions - mb_strlen and mb_substr.
1129 *
1130 * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header
1131 * encoding), rfc 2822 (header folding)
1132 *
1133 * @param string $string header string that must be encoded
1134 * @param string $charset character set. Must be supported by mbstring extension.
1135 * Use sq_mb_list_encodings() to detect supported charsets.
1136 * @return string string encoded according to rfc2047 B encoding formating rules
1137 * @since 1.5.1
1138 * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1
1139 * @todo Do we want to control max length of header?
1140 * @todo Do we want to control EOL (end-of-line) marker?
1141 * @todo Do we want to translate error message?
1142 */
1143 function encodeHeaderBase64($string,$charset) {
1144 /**
1145 * Check mbstring function requirements.
1146 */
1147 if (! function_exists('mb_strlen') ||
1148 ! function_exists('mb_substr')) {
1149 // set E_USER_WARNING
1150 trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING);
1151 // return false
1152 return false;
1153 }
1154
1155 // initial return array
1156 $aRet = array();
1157
1158 /**
1159 * header length = 75 symbols max (same as in encodeHeader)
1160 * remove $charset length
1161 * remove =? ? ?= (5 chars)
1162 * remove 2 more chars (\r\n ?)
1163 */
1164 $iMaxLength = 75 - strlen($charset) - 7;
1165
1166 // set first character position
1167 $iStartCharNum = 0;
1168
1169 // loop through all characters. count characters and not bytes.
1170 for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) {
1171 // encode string from starting character to current character.
1172 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1173
1174 // Check encoded string length
1175 if(strlen($encoded_string)>$iMaxLength) {
1176 // if string exceeds max length, reduce number of encoded characters and add encoded string part to array
1177 $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset));
1178
1179 // set new starting character
1180 $iStartCharNum = $iCharNum-1;
1181
1182 // encode last char (in case it is last character in string)
1183 $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset));
1184 } // if string is shorter than max length - add next character
1185 }
1186
1187 // add last encoded string to array
1188 $aRet[] = $encoded_string;
1189
1190 // set initial return string
1191 $sRet = '';
1192
1193 // loop through encoded strings
1194 foreach($aRet as $string) {
1195 // TODO: Do we want to control EOL (end-of-line) marker
1196 if ($sRet!='') $sRet.= " ";
1197
1198 // add header tags and encoded string to return string
1199 $sRet.= '=?'.$charset.'?B?'.$string.'?=';
1200 }
1201
1202 return $sRet;
1203 }
1204
1205 /* This function trys to locate the entity_id of a specific mime element */
1206 function find_ent_id($id, $message) {
1207 for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) {
1208 if ($message->entities[$i]->header->type0 == 'multipart') {
1209 $ret = find_ent_id($id, $message->entities[$i]);
1210 } else {
1211 if (strcasecmp($message->entities[$i]->header->id, $id) == 0) {
1212 // if (sq_check_save_extension($message->entities[$i])) {
1213 return $message->entities[$i]->entity_id;
1214 // }
1215 } elseif (!empty($message->entities[$i]->header->parameters['name'])) {
1216 /**
1217 * This is part of a fix for Outlook Express 6.x generating
1218 * cid URLs without creating content-id headers
1219 * @@JA - 20050207
1220 */
1221 if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) {
1222 return $message->entities[$i]->entity_id;
1223 }
1224 }
1225 }
1226 }
1227 return $ret;
1228 }
1229
1230 function sq_check_save_extension($message) {
1231 $filename = $message->getFilename();
1232 $ext = substr($filename, strrpos($filename,'.')+1);
1233 $save_extensions = array('jpg','jpeg','gif','png','bmp');
1234 return in_array($ext, $save_extensions);
1235 }
1236
1237
1238 /**
1239 ** HTMLFILTER ROUTINES
1240 */
1241
1242 /**
1243 * This function checks attribute values for entity-encoded values
1244 * and returns them translated into 8-bit strings so we can run
1245 * checks on them.
1246 *
1247 * @param $attvalue A string to run entity check against.
1248 * @return Nothing, modifies a reference value.
1249 */
1250 function sq_defang(&$attvalue){
1251 $me = 'sq_defang';
1252 /**
1253 * Skip this if there aren't ampersands or backslashes.
1254 */
1255 if (strpos($attvalue, '&') === false
1256 && strpos($attvalue, '\\') === false){
1257 return;
1258 }
1259 $m = false;
1260 // before deent, translate the dangerous unicode characters and ... to safe values
1261 // otherwise the regular expressions do not match.
1262
1263
1264
1265 do {
1266 $m = false;
1267 $m = $m || sq_deent($attvalue, '/\&#0*(\d+);*/s');
1268 $m = $m || sq_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
1269 $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true);
1270 } while ($m == true);
1271 $attvalue = stripslashes($attvalue);
1272 }
1273
1274 /**
1275 * Kill any tabs, newlines, or carriage returns. Our friends the
1276 * makers of the browser with 95% market value decided that it'd
1277 * be funny to make "java[tab]script" be just as good as "javascript".
1278 *
1279 * @param attvalue The attribute value before extraneous spaces removed.
1280 * @return attvalue Nothing, modifies a reference value.
1281 */
1282 function sq_unspace(&$attvalue){
1283 $me = 'sq_unspace';
1284 if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){
1285 $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "),
1286 Array('', '', '', '', ''), $attvalue);
1287 }
1288 }
1289
1290 /**
1291 * Translate all dangerous Unicode or Shift_JIS characters which are accepted by
1292 * IE as regular characters.
1293 *
1294 * @param attvalue The attribute value before dangerous characters are translated.
1295 * @return attvalue Nothing, modifies a reference value.
1296 * @author Marc Groot Koerkamp.
1297 */
1298 function sq_fixIE_idiocy(&$attvalue) {
1299 // remove NUL
1300 $attvalue = str_replace("\0", "", $attvalue);
1301 // remove comments
1302 $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue);
1303
1304 // IE has the evil habit of accepting every possible value for the attribute expression.
1305 // The table below contains characters which are parsed by IE if they are used in the "expression"
1306 // attribute value.
1307 $aDangerousCharsReplacementTable = array(
1308 array('&#x029F;', '&#0671;' ,/* L UNICODE IPA Extension */
1309 '&#x0280;', '&#0640;' ,/* R UNICODE IPA Extension */
1310 '&#x0274;', '&#0628;' ,/* N UNICODE IPA Extension */
1311 '&#xFF25;', '&#65317;' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */
1312 '&#xFF45;', '&#65349;' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */
1313 '&#xFF38;', '&#65336;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */
1314 '&#xFF58;', '&#65368;',/* Unicode FULLWIDTH LATIN SMALL LETTER X */
1315 '&#xFF30;', '&#65328;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */
1316 '&#xFF50;', '&#65360;',/* Unicode FULLWIDTH LATIN SMALL LETTER P */
1317 '&#xFF32;', '&#65330;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */
1318 '&#xFF52;', '&#65362;',/* Unicode FULLWIDTH LATIN SMALL LETTER R */
1319 '&#xFF33;', '&#65331;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */
1320 '&#xFF53;', '&#65363;',/* Unicode FULLWIDTH LATIN SMALL LETTER S */
1321 '&#xFF29;', '&#65321;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */
1322 '&#xFF49;', '&#65353;',/* Unicode FULLWIDTH LATIN SMALL LETTER I */
1323 '&#xFF2F;', '&#65327;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */
1324 '&#xFF4F;', '&#65359;',/* Unicode FULLWIDTH LATIN SMALL LETTER O */
1325 '&#xFF2E;', '&#65326;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */
1326 '&#xFF4E;', '&#65358;',/* Unicode FULLWIDTH LATIN SMALL LETTER N */
1327 '&#xFF2C;', '&#65324;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */
1328 '&#xFF4C;', '&#65356;',/* Unicode FULLWIDTH LATIN SMALL LETTER L */
1329 '&#xFF35;', '&#65333;',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */
1330 '&#xFF55;', '&#65365;',/* Unicode FULLWIDTH LATIN SMALL LETTER U */
1331 '&#x207F;', '&#8319;' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */
1332 "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range
1333 "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */
1334 "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */
1335 "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */
1336 "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */
1337 "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */
1338 "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */
1339 "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */
1340 "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */
1341 "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */
1342 "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */
1343 "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */
1344 "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */
1345 "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */
1346 "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */
1347 "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */
1348 "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */
1349 "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */
1350 "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */
1351 "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */
1352 "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */
1353 "\xCA\x9F", /* L UNICODE IPA Extension */
1354 "\xCA\x80", /* R UNICODE IPA Extension */
1355 "\xC9\xB4"), /* N UNICODE IPA Extension */
1356 array('l', 'l', 'r','r','n','n',
1357 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I',
1358 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n',
1359 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n'));
1360 $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue);
1361
1362 // Escapes are useful for special characters like "{}[]()'&. In other cases they are
1363 // used for XSS.
1364 $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue);
1365 }
1366
1367 /**
1368 * This function returns the final tag out of the tag name, an array
1369 * of attributes, and the type of the tag. This function is called by
1370 * sq_sanitize internally.
1371 *
1372 * @param $tagname the name of the tag.
1373 * @param $attary the array of attributes and their values
1374 * @param $tagtype The type of the tag (see in comments).
1375 * @return a string with the final tag representation.
1376 */
1377 function sq_tagprint($tagname, $attary, $tagtype){
1378 $me = 'sq_tagprint';
1379
1380 if ($tagtype == 2){
1381 $fulltag = '</' . $tagname . '>';
1382 } else {
1383 $fulltag = '<' . $tagname;
1384 if (is_array($attary) && sizeof($attary)){
1385 $atts = Array();
1386 foreach ($attary as $attname => $attvalue){
1387 array_push($atts, "$attname=$attvalue");
1388 }
1389 $fulltag .= ' ' . join(" ", $atts);
1390 }
1391 if ($tagtype == 3){
1392 $fulltag .= ' /';
1393 }
1394 $fulltag .= '>';
1395 }
1396 return $fulltag;
1397 }
1398
1399 /**
1400 * A small helper function to use with array_walk. Modifies a by-ref
1401 * value and makes it lowercase.
1402 *
1403 * @param $val a value passed by-ref.
1404 * @return void since it modifies a by-ref value.
1405 */
1406 function sq_casenormalize(&$val){
1407 $val = strtolower($val);
1408 }
1409
1410 /**
1411 * This function skips any whitespace from the current position within
1412 * a string and to the next non-whitespace value.
1413 *
1414 * @param $body the string
1415 * @param $offset the offset within the string where we should start
1416 * looking for the next non-whitespace character.
1417 * @return the location within the $body where the next
1418 * non-whitespace char is located.
1419 */
1420 function sq_skipspace($body, $offset){
1421 $me = 'sq_skipspace';
1422 preg_match('/^(\s*)/s', substr($body, $offset), $matches);
1423 if (!empty($matches[1])){
1424 $offset += strlen($matches[1]);
1425 }
1426 return $offset;
1427 }
1428
1429 /**
1430 * This function looks for the next character within a string. It's
1431 * really just a glorified "strpos", except it catches if failures
1432 * nicely.
1433 *
1434 * @param $body The string to look for needle in.
1435 * @param $offset Start looking from this position.
1436 * @param $needle The character/string to look for.
1437 * @return location of the next occurance of the needle, or
1438 * strlen($body) if needle wasn't found.
1439 */
1440 function sq_findnxstr($body, $offset, $needle){
1441 $me = 'sq_findnxstr';
1442 $pos = strpos($body, $needle, $offset);
1443 if ($pos === FALSE){
1444 $pos = strlen($body);
1445 }
1446 return $pos;
1447 }
1448
1449 /**
1450 * This function takes a PCRE-style regexp and tries to match it
1451 * within the string.
1452 *
1453 * @param $body The string to look for needle in.
1454 * @param $offset Start looking from here.
1455 * @param $reg A PCRE-style regex to match.
1456 * @return Returns a false if no matches found, or an array
1457 * with the following members:
1458 * - integer with the location of the match within $body
1459 * - string with whatever content between offset and the match
1460 * - string with whatever it is we matched
1461 */
1462 function sq_findnxreg($body, $offset, $reg){
1463 $me = 'sq_findnxreg';
1464 $matches = Array();
1465 $retarr = Array();
1466 preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches);
1467 if (!isset($matches[0]) || !$matches[0]){
1468 $retarr = false;
1469 } else {
1470 $retarr[0] = $offset + strlen($matches[1]);
1471 $retarr[1] = $matches[1];
1472 $retarr[2] = $matches[2];
1473 }
1474 return $retarr;
1475 }
1476
1477 /**
1478 * This function looks for the next tag.
1479 *
1480 * @param $body String where to look for the next tag.
1481 * @param $offset Start looking from here.
1482 * @return false if no more tags exist in the body, or
1483 * an array with the following members:
1484 * - string with the name of the tag
1485 * - array with attributes and their values
1486 * - integer with tag type (1, 2, or 3)
1487 * - integer where the tag starts (starting "<")
1488 * - integer where the tag ends (ending ">")
1489 * first three members will be false, if the tag is invalid.
1490 */
1491 function sq_getnxtag($body, $offset){
1492 $me = 'sq_getnxtag';
1493 if ($offset > strlen($body)){
1494 return false;
1495 }
1496 $lt = sq_findnxstr($body, $offset, "<");
1497 if ($lt == strlen($body)){
1498 return false;
1499 }
1500 /**
1501 * We are here:
1502 * blah blah <tag attribute="value">
1503 * \---------^
1504 */
1505 $pos = sq_skipspace($body, $lt+1);
1506 if ($pos >= strlen($body)){
1507 return Array(false, false, false, $lt, strlen($body));
1508 }
1509 /**
1510 * There are 3 kinds of tags:
1511 * 1. Opening tag, e.g.:
1512 * <a href="blah">
1513 * 2. Closing tag, e.g.:
1514 * </a>
1515 * 3. XHTML-style content-less tag, e.g.:
1516 * <img src="blah" />
1517 */
1518 $tagtype = false;
1519 switch (substr($body, $pos, 1)){
1520 case '/':
1521 $tagtype = 2;
1522 $pos++;
1523 break;
1524 case '!':
1525 /**
1526 * A comment or an SGML declaration.
1527 */
1528 if (substr($body, $pos+1, 2) == "--"){
1529 $gt = strpos($body, "-->", $pos);
1530 if ($gt === false){
1531 $gt = strlen($body);
1532 } else {
1533 $gt += 2;
1534 }
1535 return Array(false, false, false, $lt, $gt);
1536 } else {
1537 $gt = sq_findnxstr($body, $pos, ">");
1538 return Array(false, false, false, $lt, $gt);
1539 }
1540 break;
1541 default:
1542 /**
1543 * Assume tagtype 1 for now. If it's type 3, we'll switch values
1544 * later.
1545 */
1546 $tagtype = 1;
1547 break;
1548 }
1549
1550 $tag_start = $pos;
1551 $tagname = '';
1552 /**
1553 * Look for next [\W-_], which will indicate the end of the tag name.
1554 */
1555 $regary = sq_findnxreg($body, $pos, "[^\w\-_]");
1556 if ($regary == false){
1557 return Array(false, false, false, $lt, strlen($body));
1558 }
1559 list($pos, $tagname, $match) = $regary;
1560 $tagname = strtolower($tagname);
1561
1562 /**
1563 * $match can be either of these:
1564 * '>' indicating the end of the tag entirely.
1565 * '\s' indicating the end of the tag name.
1566 * '/' indicating that this is type-3 xhtml tag.
1567 *
1568 * Whatever else we find there indicates an invalid tag.
1569 */
1570 switch ($match){
1571 case '/':
1572 /**
1573 * This is an xhtml-style tag with a closing / at the
1574 * end, like so: <img src="blah" />. Check if it's followed
1575 * by the closing bracket. If not, then this tag is invalid
1576 */
1577 if (substr($body, $pos, 2) == "/>"){
1578 $pos++;
1579 $tagtype = 3;
1580 } else {
1581 $gt = sq_findnxstr($body, $pos, ">");
1582 $retary = Array(false, false, false, $lt, $gt);
1583 return $retary;
1584 }
1585 case '>':
1586 return Array($tagname, false, $tagtype, $lt, $pos);
1587 break;
1588 default:
1589 /**
1590 * Check if it's whitespace
1591 */
1592 if (!preg_match('/\s/', $match)){
1593 /**
1594 * This is an invalid tag! Look for the next closing ">".
1595 */
1596 $gt = sq_findnxstr($body, $lt, ">");
1597 return Array(false, false, false, $lt, $gt);
1598 }
1599 break;
1600 }
1601
1602 /**
1603 * At this point we're here:
1604 * <tagname attribute='blah'>
1605 * \-------^
1606 *
1607 * At this point we loop in order to find all attributes.
1608 */
1609 $attname = '';
1610 $atttype = false;
1611 $attary = Array();
1612
1613 while ($pos <= strlen($body)){
1614 $pos = sq_skipspace($body, $pos);
1615 if ($pos == strlen($body)){
1616 /**
1617 * Non-closed tag.
1618 */
1619 return Array(false, false, false, $lt, $pos);
1620 }
1621 /**
1622 * See if we arrived at a ">" or "/>", which means that we reached
1623 * the end of the tag.
1624 */
1625 $matches = Array();
1626 if (preg_match("%^(\s*)(>|/>)%s", substr($body, $pos), $matches)) {
1627 /**
1628 * Yep. So we did.
1629 */
1630 $pos += strlen($matches[1]);
1631 if ($matches[2] == "/>"){
1632 $tagtype = 3;
1633 $pos++;
1634 }
1635 return Array($tagname, $attary, $tagtype, $lt, $pos);
1636 }
1637
1638 /**
1639 * There are several types of attributes, with optional
1640 * [:space:] between members.
1641 * Type 1:
1642 * attrname[:space:]=[:space:]'CDATA'
1643 * Type 2:
1644 * attrname[:space:]=[:space:]"CDATA"
1645 * Type 3:
1646 * attr[:space:]=[:space:]CDATA
1647 * Type 4:
1648 * attrname
1649 *
1650 * We leave types 1 and 2 the same, type 3 we check for
1651 * '"' and convert to "&quot" if needed, then wrap in
1652 * double quotes. Type 4 we convert into:
1653 * attrname="yes".
1654 */
1655 $regary = sq_findnxreg($body, $pos, "[^:\w\-_]");
1656 if ($regary == false){
1657 /**
1658 * Looks like body ended before the end of tag.
1659 */
1660 return Array(false, false, false, $lt, strlen($body));
1661 }
1662 list($pos, $attname, $match) = $regary;
1663 $attname = strtolower($attname);
1664 /**
1665 * We arrived at the end of attribute name. Several things possible
1666 * here:
1667 * '>' means the end of the tag and this is attribute type 4
1668 * '/' if followed by '>' means the same thing as above
1669 * '\s' means a lot of things -- look what it's followed by.
1670 * anything else means the attribute is invalid.
1671 */
1672 switch($match){
1673 case '/':
1674 /**
1675 * This is an xhtml-style tag with a closing / at the
1676 * end, like so: <img src="blah" />. Check if it's followed
1677 * by the closing bracket. If not, then this tag is invalid
1678 */
1679 if (substr($body, $pos, 2) == "/>"){
1680 $pos++;
1681 $tagtype = 3;
1682 } else {
1683 $gt = sq_findnxstr($body, $pos, ">");
1684 $retary = Array(false, false, false, $lt, $gt);
1685 return $retary;
1686 }
1687 case '>':
1688 $attary[$attname] = '"yes"';
1689 return Array($tagname, $attary, $tagtype, $lt, $pos);
1690 break;
1691 default:
1692 /**
1693 * Skip whitespace and see what we arrive at.
1694 */
1695 $pos = sq_skipspace($body, $pos);
1696 $char = substr($body, $pos, 1);
1697 /**
1698 * Two things are valid here:
1699 * '=' means this is attribute type 1 2 or 3.
1700 * \w means this was attribute type 4.
1701 * anything else we ignore and re-loop. End of tag and
1702 * invalid stuff will be caught by our checks at the beginning
1703 * of the loop.
1704 */
1705 if ($char == "="){
1706 $pos++;
1707 $pos = sq_skipspace($body, $pos);
1708 /**
1709 * Here are 3 possibilities:
1710 * "'" attribute type 1
1711 * '"' attribute type 2
1712 * everything else is the content of tag type 3
1713 */
1714 $quot = substr($body, $pos, 1);
1715 if ($quot == "'"){
1716 $regary = sq_findnxreg($body, $pos+1, "\'");
1717 if ($regary == false){
1718 return Array(false, false, false, $lt, strlen($body));
1719 }
1720 list($pos, $attval, $match) = $regary;
1721 $pos++;
1722 $attary[$attname] = "'" . $attval . "'";
1723 } else if ($quot == '"'){
1724 $regary = sq_findnxreg($body, $pos+1, '\"');
1725 if ($regary == false){
1726 return Array(false, false, false, $lt, strlen($body));
1727 }
1728 list($pos, $attval, $match) = $regary;
1729 $pos++;
1730 $attary[$attname] = '"' . $attval . '"';
1731 } else {
1732 /**
1733 * These are hateful. Look for \s, or >.
1734 */
1735 $regary = sq_findnxreg($body, $pos, "[\s>]");
1736 if ($regary == false){
1737 return Array(false, false, false, $lt, strlen($body));
1738 }
1739 list($pos, $attval, $match) = $regary;
1740 /**
1741 * If it's ">" it will be caught at the top.
1742 */
1743 $attval = preg_replace("/\"/s", "&quot;", $attval);
1744 $attary[$attname] = '"' . $attval . '"';
1745 }
1746 } else if (preg_match("|[\w/>]|", $char)) {
1747 /**
1748 * That was attribute type 4.
1749 */
1750 $attary[$attname] = '"yes"';
1751 } else {
1752 /**
1753 * An illegal character. Find next '>' and return.
1754 */
1755 $gt = sq_findnxstr($body, $pos, ">");
1756 return Array(false, false, false, $lt, $gt);
1757 }
1758 break;
1759 }
1760 }
1761 /**
1762 * The fact that we got here indicates that the tag end was never
1763 * found. Return invalid tag indication so it gets stripped.
1764 */
1765 return Array(false, false, false, $lt, strlen($body));
1766 }
1767
1768 /**
1769 * Translates entities into literal values so they can be checked.
1770 *
1771 * @param $attvalue the by-ref value to check.
1772 * @param $regex the regular expression to check against.
1773 * @param $hex whether the entites are hexadecimal.
1774 * @return True or False depending on whether there were matches.
1775 */
1776 function sq_deent(&$attvalue, $regex, $hex=false){
1777 $me = 'sq_deent';
1778 $ret_match = false;
1779 // remove comments
1780 //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue);
1781 preg_match_all($regex, $attvalue, $matches);
1782 if (is_array($matches) && sizeof($matches[0]) > 0){
1783 $repl = Array();
1784 for ($i = 0; $i < sizeof($matches[0]); $i++){
1785 $numval = $matches[1][$i];
1786 if ($hex){
1787 $numval = hexdec($numval);
1788 }
1789 $repl[$matches[0][$i]] = chr($numval);
1790 }
1791 $attvalue = strtr($attvalue, $repl);
1792 return true;
1793 } else {
1794 return false;
1795 }
1796 }
1797
1798 /**
1799 * This function runs various checks against the attributes.
1800 *
1801 * @param $tagname String with the name of the tag.
1802 * @param $attary Array with all tag attributes.
1803 * @param $rm_attnames See description for sq_sanitize
1804 * @param $bad_attvals See description for sq_sanitize
1805 * @param $add_attr_to_tag See description for sq_sanitize
1806 * @param $message message object
1807 * @param $id message id
1808 * @return Array with modified attributes.
1809 */
1810 function sq_fixatts($tagname,
1811 $attary,
1812 $rm_attnames,
1813 $bad_attvals,
1814 $add_attr_to_tag,
1815 $message,
1816 $id,
1817 $mailbox
1818 ){
1819 $me = 'sq_fixatts';
1820 foreach ($attary as $attname => $attvalue){
1821 /**
1822 * See if this attribute should be removed.
1823 */
1824 foreach ($rm_attnames as $matchtag=>$matchattrs){
1825 if (preg_match($matchtag, $tagname)){
1826 foreach ($matchattrs as $matchattr){
1827 if (preg_match($matchattr, $attname)){
1828 unset($attary[$attname]);
1829 continue;
1830 }
1831 }
1832 }
1833 }
1834 /**
1835 * Workaround for IE quirks
1836 */
1837 sq_fixIE_idiocy($attvalue);
1838
1839 /**
1840 * Remove any backslashes, entities, and extraneous whitespace.
1841 */
1842
1843 $oldattvalue = $attvalue;
1844 sq_defang($attvalue);
1845 if ($attname == 'style' && $attvalue !== $oldattvalue) {
1846 // entities are used in the attribute value. In 99% of the cases it's there as XSS
1847 // i.e.<div style="{ left:exp&#x0280;essio&#x0274;( alert('XSS') ) }">
1848 $attvalue = "idiocy";
1849 $attary[$attname] = $attvalue;
1850 }
1851 sq_unspace($attvalue);
1852
1853 /**
1854 * Now let's run checks on the attvalues.
1855 * I don't expect anyone to comprehend this. If you do,
1856 * get in touch with me so I can drive to where you live and
1857 * shake your hand personally. :)
1858 */
1859 foreach ($bad_attvals as $matchtag=>$matchattrs){
1860 if (preg_match($matchtag, $tagname)){
1861 foreach ($matchattrs as $matchattr=>$valary){
1862 if (preg_match($matchattr, $attname)){
1863 /**
1864 * There are two arrays in valary.
1865 * First is matches.
1866 * Second one is replacements
1867 */
1868 list($valmatch, $valrepl) = $valary;
1869 $newvalue =
1870 preg_replace($valmatch, $valrepl, $attvalue);
1871 if ($newvalue != $attvalue){
1872 $attary[$attname] = $newvalue;
1873 $attvalue = $newvalue;
1874 }
1875 }
1876 }
1877 }
1878 }
1879 if ($attname == 'style') {
1880 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1881 // 8bit and control characters in style attribute values can be used for XSS, remove them
1882 $attary[$attname] = '"disallowed character"';
1883 }
1884 preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch);
1885 if (count($aMatch)) {
1886 foreach($aMatch[1] as $sMatch) {
1887 // url value
1888 $urlvalue = $sMatch;
1889 sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'");
1890 $attary[$attname] = str_replace($sMatch,$urlvalue,$attvalue);
1891 }
1892 }
1893 }
1894 /**
1895 * Use white list based filtering on attributes which can contain url's
1896 */
1897 else if ($attname == 'href' || $attname == 'xlink:href' || $attname == 'src'
1898 || $attname == 'poster' || $attname == 'formaction'
1899 || $attname == 'background' || $attname == 'action') {
1900 sq_fix_url($attname, $attvalue, $message, $id, $mailbox);
1901 $attary[$attname] = $attvalue;
1902 }
1903 }
1904 /**
1905 * See if we need to append any attributes to this tag.
1906 */
1907 foreach ($add_attr_to_tag as $matchtag=>$addattary){
1908 if (preg_match($matchtag, $tagname)){
1909 $attary = array_merge($attary, $addattary);
1910 }
1911 }
1912 return $attary;
1913 }
1914
1915 /**
1916 * This function filters url's
1917 *
1918 * @param $attvalue String with attribute value to filter
1919 * @param $message message object
1920 * @param $id message id
1921 * @param $mailbox mailbox
1922 * @param $sQuote quoting characters around url's
1923 */
1924 function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') {
1925 $attvalue = trim($attvalue);
1926 if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
1927 // remove the double quotes
1928 $sQuote = $attvalue[0];
1929 $attvalue = trim(substr($attvalue,1,-1));
1930 }
1931
1932 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
1933 // images off by default.
1934 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
1935
1936 global $use_transparent_security_image;
1937 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
1938 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
1939
1940 /**
1941 * Replace empty src tags with the blank image. src is only used
1942 * for frames, images, and image inputs. Doing a replace should
1943 * not affect them working as should be, however it will stop
1944 * IE from being kicked off when src for img tags are not set
1945 */
1946 if ($attvalue == '') {
1947 $attvalue = '"' . SM_PATH . 'images/blank.png"';
1948 } else {
1949 // first, disallow 8 bit characters and control characters
1950 if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
1951 switch ($attname) {
1952 case 'href':
1953 $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
1954 break;
1955 default:
1956 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
1957 break;
1958 }
1959 } else {
1960 $aUrl = parse_url($attvalue);
1961 if (isset($aUrl['scheme'])) {
1962 switch(strtolower($aUrl['scheme'])) {
1963 case 'mailto':
1964 case 'http':
1965 case 'https':
1966 case 'ftp':
1967 if ($attname != 'href') {
1968 if ($view_unsafe_images == false) {
1969 $attvalue = $sQuote . $secremoveimg . $sQuote;
1970 } else {
1971 if (isset($aUrl['path'])) {
1972
1973 // No one has been able to show that image URIs
1974 // can be exploited, so for now, no restrictions
1975 // are made at all. If this proves to be a problem,
1976 // the commented-out code below can be of help.
1977 // (One consideration is that I see nothing in this
1978 // function that specifically says that we will
1979 // only ever arrive here when inspecting an image
1980 // tag, although that does seem to be the end
1981 // result - e.g., <script src="..."> where malicious
1982 // image URIs are in fact a problem are already
1983 // filtered out elsewhere.
1984 /* ---------------------------------
1985 // validate image extension.
1986 $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.')));
1987 if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) {
1988 // If URI is to something other than
1989 // a regular image file, get the contents
1990 // and try to see if it is an image.
1991 // Don't use Fileinfo (finfo_file()) because
1992 // we'd need to make the admin configure the
1993 // location of the magic.mime file (FIXME: add finfo_file() support later?)
1994 //
1995 $mime_type = '';
1996 if (function_exists('mime_content_type')
1997 && ($FILE = @fopen($attvalue, 'rb', FALSE))) {
1998
1999 // fetch file
2000 //
2001 $file_contents = '';
2002 while (!feof($FILE)) {
2003 $file_contents .= fread($FILE, 8192);
2004 }
2005 fclose($FILE);
2006
2007 // store file locally
2008 //
2009 global $attachment_dir, $username;
2010 $hashed_attachment_dir = getHashedDir($username, $attachment_dir);
2011 $localfilename = GenerateRandomString(32, '', 7);
2012 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2013 while (file_exists($full_localfilename)) {
2014 $localfilename = GenerateRandomString(32, '', 7);
2015 $full_localfilename = "$hashed_attachment_dir/$localfilename";
2016 }
2017 $FILE = fopen("$hashed_attachment_dir/$localfilename", 'wb');
2018 fwrite($FILE, $file_contents);
2019 fclose($FILE);
2020
2021 // get mime type and remove file
2022 //
2023 $mime_type = mime_content_type("$hashed_attachment_dir/$localfilename");
2024 unlink("$hashed_attachment_dir/$localfilename");
2025 }
2026 // debug: echo "$attvalue FILE TYPE IS $mime_type<HR>";
2027 if (substr(strtolower($mime_type), 0, 5) != 'image') {
2028 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2029 }
2030 }
2031 --------------------------------- */
2032 } else {
2033 $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote;
2034 }
2035 }
2036 } else {
2037 $attvalue = $sQuote . $attvalue . $sQuote;
2038 }
2039 break;
2040 case 'outbind':
2041 /**
2042 * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags.
2043 * One day MS might actually make it match something useful, for now, falling
2044 * back to using cid2http, so we can grab the blank.png.
2045 */
2046 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2047 break;
2048 case 'cid':
2049 /**
2050 * Turn cid: urls into http-friendly ones.
2051 */
2052 $attvalue = $sQuote . sq_cid2http($message, $id, $attvalue, $mailbox) . $sQuote;
2053 break;
2054 default:
2055 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2056 break;
2057 }
2058 } else {
2059 if (!isset($aUrl['path']) || $aUrl['path'] != $secremoveimg) {
2060 // parse_url did not lead to satisfying result
2061 $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote;
2062 }
2063 }
2064 }
2065 }
2066 }
2067
2068 /**
2069 * This function edits the style definition to make them friendly and
2070 * usable in SquirrelMail.
2071 *
2072 * @param $message the message object
2073 * @param $id the message id
2074 * @param $content a string with whatever is between <style> and </style>
2075 * @param $mailbox the message mailbox
2076 * @return a string with edited content.
2077 */
2078 function sq_fixstyle($body, $pos, $message, $id, $mailbox){
2079 $me = 'sq_fixstyle';
2080 // workaround for </style> in between comments
2081 $iCurrentPos = $pos;
2082 $content = '';
2083 $sToken = '';
2084 $bSucces = false;
2085 $bEndTag = false;
2086 for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
2087 $char = $body[$i];
2088 switch ($char) {
2089 case '<':
2090 $sToken = $char;
2091 break;
2092 case '/':
2093 if ($sToken == '<') {
2094 $sToken .= $char;
2095 $bEndTag = true;
2096 } else {
2097 $content .= $char;
2098 }
2099 break;
2100 case '>':
2101 if ($bEndTag) {
2102 $sToken .= $char;
2103 if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) {
2104 $newpos = $i + 1;
2105 $bSucces = true;
2106 break 2;
2107 } else {
2108 $content .= $sToken;
2109 }
2110 $bEndTag = false;
2111 } else {
2112 $content .= $char;
2113 }
2114 break;
2115 case '!':
2116 if ($sToken == '<') {
2117 // possible comment
2118 if (isset($body[$i+2]) && substr($body,$i,3) == '!--') {
2119 $i = strpos($body,'-->',$i+3);
2120 if ($i === false) { // no end comment
2121 $i = strlen($body);
2122 }
2123 $sToken = '';
2124 }
2125 } else {
2126 $content .= $char;
2127 }
2128 break;
2129 default:
2130 if ($bEndTag) {
2131 $sToken .= $char;
2132 } else {
2133 $content .= $char;
2134 }
2135 break;
2136 }
2137 }
2138 if ($bSucces == FALSE){
2139 return array(FALSE, strlen($body));
2140 }
2141
2142
2143
2144 /**
2145 * First look for general BODY style declaration, which would be
2146 * like so:
2147 * body {background: blah-blah}
2148 * and change it to .bodyclass so we can just assign it to a <div>
2149 */
2150 // $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
2151 // Nah, this is even better - try to preface all CSS selectors with
2152 // our <div> class ID "bodyclass" then correct generic "body" selectors
2153 // TODO: this works pretty good but breaks stuff like this:
2154 // @media print { body { font-size: 10pt; } }
2155 // but there isn't an easy way to make this regex skip @media
2156 // definitions... though lots of the ones in the wild will be
2157 // correctly handled because they tend to end with a parenthesis, like:
2158 // @media screen and (max-width:480px) { ...
2159 $content = preg_replace('/([a-z0-9._-][a-z0-9 >+~|:._-]*\s*(?:,|{.*?}))/si', '.bodyclass $1', $content);
2160 $content = str_replace('.bodyclass body', '.bodyclass', $content);
2161
2162 global $use_transparent_security_image;
2163 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2164 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2165
2166 /**
2167 * Fix url('blah') declarations.
2168 */
2169 // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si",
2170 // "url(\\1$secremoveimg\\2)", $content);
2171
2172 // first check for 8bit sequences and disallowed control characters
2173 if (preg_match('/[\16-\37\200-\377]+/',$content)) {
2174 $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
2175 return array($content, $newpos);
2176 }
2177
2178 // IE Sucks hard. We have a special function for it.
2179 sq_fixIE_idiocy($content);
2180
2181 // remove @import line
2182 $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
2183
2184 // translate ur\l and variations (IE parses that)
2185 // TODO check if the sq_fixIE_idiocy function already handles this.
2186 $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
2187 preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch);
2188 if (count($aMatch)) {
2189 $aValue = $aReplace = array();
2190 foreach($aMatch[1] as $sMatch) {
2191 // url value
2192 $urlvalue = $sMatch;
2193 sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'");
2194 $aValue[] = $sMatch;
2195 $aReplace[] = $urlvalue;
2196 }
2197 $content = str_replace($aValue,$aReplace,$content);
2198 }
2199
2200 /**
2201 * Remove any backslashes, entities, and extraneous whitespace.
2202 */
2203 $contentTemp = $content;
2204 sq_defang($contentTemp);
2205 sq_unspace($contentTemp);
2206
2207 /**
2208 * Fix stupid css declarations which lead to vulnerabilities
2209 * in IE.
2210 *
2211 * Also remove "position" attribute, as it can easily be set
2212 * to "fixed" or "absolute" with "left" and "top" attributes
2213 * of zero, taking over the whole content frame. It can also
2214 * be set to relative and move itself anywhere it wants to,
2215 * displaying content in areas it shouldn't be allowed to touch.
2216 */
2217 $match = Array('/\/\*.*\*\//', // removes /* blah blah */
2218 '/expression/i',
2219 '/behaviou*r/i',
2220 '/binding/i',
2221 '/include-source/i',
2222 '/javascript/i',
2223 '/script/i',
2224 '/position/i');
2225 $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
2226 $contentNew = preg_replace($match, $replace, $contentTemp);
2227 if ($contentNew !== $contentTemp) {
2228 // insecure css declarations are used. From now on we don't care
2229 // anymore if the css is destroyed by sq_deent, sq_unspace or sq_unbackslash
2230 $content = $contentNew;
2231 }
2232 return array($content, $newpos);
2233 }
2234
2235
2236 /**
2237 * This function converts cid: url's into the ones that can be viewed in
2238 * the browser.
2239 *
2240 * @param $message the message object
2241 * @param $id the message id
2242 * @param $cidurl the cid: url.
2243 * @param $mailbox the message mailbox
2244 * @return a string with a http-friendly url
2245 */
2246 function sq_cid2http($message, $id, $cidurl, $mailbox){
2247 /**
2248 * Get rid of quotes.
2249 */
2250 $quotchar = substr($cidurl, 0, 1);
2251 if ($quotchar == '"' || $quotchar == "'"){
2252 $cidurl = str_replace($quotchar, "", $cidurl);
2253 } else {
2254 $quotchar = '';
2255 }
2256 $cidurl = substr(trim($cidurl), 4);
2257
2258 $match_str = '/\{.*?\}\//';
2259 $str_rep = '';
2260 $cidurl = preg_replace($match_str, $str_rep, $cidurl);
2261
2262 $linkurl = find_ent_id($cidurl, $message);
2263 /* in case of non-safe cid links $httpurl should be replaced by a sort of
2264 unsafe link image */
2265 $httpurl = '';
2266
2267 /**
2268 * This is part of a fix for Outlook Express 6.x generating
2269 * cid URLs without creating content-id headers. These images are
2270 * not part of the multipart/related html mail. The html contains
2271 * <img src="cid:{some_id}/image_filename.ext"> references to
2272 * attached images with as goal to render them inline although
2273 * the attachment disposition property is not inline.
2274 */
2275
2276 if (empty($linkurl)) {
2277 if (preg_match('/{.*}\//', $cidurl)) {
2278 $cidurl = preg_replace('/{.*}\//','', $cidurl);
2279 if (!empty($cidurl)) {
2280 $linkurl = find_ent_id($cidurl, $message);
2281 }
2282 }
2283 }
2284
2285 if (!empty($linkurl)) {
2286 $httpurl = $quotchar . sqm_baseuri() . 'src/download.php?absolute_dl=true&amp;' .
2287 "passed_id=$id&amp;mailbox=" . urlencode($mailbox) .
2288 '&amp;ent_id=' . $linkurl . $quotchar;
2289 } else {
2290 /**
2291 * If we couldn't generate a proper img url, drop in a blank image
2292 * instead of sending back empty, otherwise it causes unusual behaviour
2293 */
2294 $httpurl = $quotchar . SM_PATH . 'images/blank.png' . $quotchar;
2295 }
2296
2297 return $httpurl;
2298 }
2299
2300 /**
2301 * This function changes the <body> tag into a <div> tag since we
2302 * can't really have a body-within-body.
2303 *
2304 * @param $attary an array of attributes and values of <body>
2305 * @param $mailbox mailbox we're currently reading (for cid2http)
2306 * @param $message current message (for cid2http)
2307 * @param $id current message id (for cid2http)
2308 * @return a modified array of attributes to be set for <div>
2309 */
2310 function sq_body2div($attary, $mailbox, $message, $id){
2311 $me = 'sq_body2div';
2312 $divattary = Array('class' => "'bodyclass'");
2313 $text = '#000000';
2314 $has_bgc_stl = $has_txt_stl = false;
2315 $styledef = '';
2316 if (is_array($attary) && sizeof($attary) > 0){
2317 foreach ($attary as $attname=>$attvalue){
2318 $quotchar = substr($attvalue, 0, 1);
2319 $attvalue = str_replace($quotchar, "", $attvalue);
2320 switch ($attname){
2321 case 'background':
2322 $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox);
2323 $styledef .= "background-image: url('$attvalue'); ";
2324 break;
2325 case 'bgcolor':
2326 $has_bgc_stl = true;
2327 $styledef .= "background-color: $attvalue; ";
2328 break;
2329 case 'text':
2330 $has_txt_stl = true;
2331 $styledef .= "color: $attvalue; ";
2332 break;
2333 }
2334 }
2335 // Outlook defines a white bgcolor and no text color. This can lead to
2336 // white text on a white bg with certain themes.
2337 if ($has_bgc_stl && !$has_txt_stl) {
2338 $styledef .= "color: $text; ";
2339 }
2340 if (strlen($styledef) > 0){
2341 $divattary["style"] = "\"$styledef\"";
2342 }
2343 }
2344 return $divattary;
2345 }
2346
2347 /**
2348 * This is the main function and the one you should actually be calling.
2349 * There are several variables you should be aware of an which need
2350 * special description.
2351 *
2352 * Since the description is quite lengthy, see it here:
2353 * http://linux.duke.edu/projects/mini/htmlfilter/
2354 *
2355 * @param $body the string with HTML you wish to filter
2356 * @param $tag_list see description above
2357 * @param $rm_tags_with_content see description above
2358 * @param $self_closing_tags see description above
2359 * @param $force_tag_closing see description above
2360 * @param $rm_attnames see description above
2361 * @param $bad_attvals see description above
2362 * @param $add_attr_to_tag see description above
2363 * @param $message message object
2364 * @param $id message id
2365 * @param $recursively_called boolean flag for recursive calls into this function (optional; default FALSE)
2366 * @return sanitized html safe to show on your pages.
2367 */
2368 function sq_sanitize($body,
2369 $tag_list,
2370 $rm_tags_with_content,
2371 $self_closing_tags,
2372 $force_tag_closing,
2373 $rm_attnames,
2374 $bad_attvals,
2375 $add_attr_to_tag,
2376 $message,
2377 $id,
2378 $mailbox,
2379 $recursively_called=FALSE
2380 ){
2381 $me = 'sq_sanitize';
2382
2383 /**
2384 * See if tag_list is of tags to remove or tags to allow.
2385 * false means remove these tags
2386 * true means allow these tags
2387 */
2388 $orig_tag_list = $tag_list;
2389 $rm_tags = array_shift($tag_list);
2390
2391 /**
2392 * Normalize rm_tags and rm_tags_with_content.
2393 */
2394 @array_walk($tag_list, 'sq_casenormalize');
2395 @array_walk($rm_tags_with_content, 'sq_casenormalize');
2396 @array_walk($self_closing_tags, 'sq_casenormalize');
2397
2398 $curpos = 0;
2399 $open_tags = Array();
2400 $trusted = "\n<!-- begin sanitized html -->\n";
2401 $skip_content = false;
2402 /**
2403 * Take care of netscape's stupid javascript entities like
2404 * &{alert('boo')};
2405 */
2406 $body = preg_replace("/&(\{.*?\};)/si", "&amp;\\1", $body);
2407
2408 while (($curtag = sq_getnxtag($body, $curpos)) != FALSE){
2409 list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
2410
2411 /**
2412 * RCDATA and RAWTEXT tags are handled differently:
2413 * next instance of closing tag is used, whether or not
2414 * the HTML is well formed before that
2415 */
2416 global $rcdata_rawtext_tags;
2417 if (!$recursively_called
2418 && in_array($tagname, $rcdata_rawtext_tags)
2419 && $tagtype === 1){
2420 $closing_tag = false;
2421 $closing_tag_offset = $curpos;
2422 // seek out the closing tag for the current RCDATA/RAWTEXT tag
2423 while (1) {
2424 // first we need to move forward to next available closing tag
2425 // (intentionally leave off the closing > and let sq_getnxtag() validate a proper tag syntax)
2426 $next_tag = sq_findnxreg($body, $closing_tag_offset, "</\s*$tagname");
2427 if ($next_tag === false) {
2428 $closing_tag = false;
2429 break;
2430 }
2431 // but then we have to make sure it's a well-formed tag
2432 $closing_tag = sq_getnxtag($body, $next_tag[0]);
2433 if ($closing_tag === false)
2434 break;
2435 else if ($closing_tag[0] !== false
2436 // these should be redundant
2437 && $closing_tag[0] === $tagname && $closing_tag[2] === 2) {
2438 $trusted .= sq_sanitize(substr($body, $curpos, $closing_tag[4] - $curpos + 1),
2439 $orig_tag_list, $rm_tags_with_content, $self_closing_tags,
2440 $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag,
2441 $message, $id, $mailbox, true);
2442 $curpos = $closing_tag[4] + 1;
2443 continue 2;
2444 }
2445 $closing_tag_offset = $next_tag[0] + 1;
2446 }
2447 if ($closing_tag === false)
2448 { /* 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 */ }
2449 }
2450
2451 $free_content = substr($body, $curpos, $lt-$curpos);
2452 /**
2453 * Take care of <style>
2454 */
2455 if ($tagname == "style" && $tagtype == 1){
2456 list($free_content, $curpos) =
2457 sq_fixstyle($body, $gt+1, $message, $id, $mailbox);
2458 if ($free_content != FALSE){
2459 if ( !empty($attary) ) {
2460 $attary = sq_fixatts($tagname,
2461 $attary,
2462 $rm_attnames,
2463 $bad_attvals,
2464 $add_attr_to_tag,
2465 $message,
2466 $id,
2467 $mailbox
2468 );
2469 }
2470 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2471 $trusted .= $free_content;
2472 $trusted .= sq_tagprint($tagname, false, 2);
2473 }
2474 continue;
2475 }
2476 if ($skip_content == false){
2477 $trusted .= $free_content;
2478 }
2479 if ($tagname != FALSE){
2480 if ($tagtype == 2){
2481 if ($skip_content == $tagname){
2482 /**
2483 * Got to the end of tag we needed to remove.
2484 */
2485 $tagname = false;
2486 $skip_content = false;
2487 } else {
2488 if ($skip_content == false){
2489 if ($tagname == "body"){
2490 $tagname = "div";
2491 }
2492 if (isset($open_tags[$tagname]) &&
2493 $open_tags[$tagname] > 0){
2494 $open_tags[$tagname]--;
2495 } else {
2496 $tagname = false;
2497 }
2498 }
2499 }
2500 } else {
2501 /**
2502 * $rm_tags_with_content
2503 */
2504 if ($skip_content == false){
2505 /**
2506 * See if this is a self-closing type and change
2507 * tagtype appropriately.
2508 */
2509 if ($tagtype == 1
2510 && in_array($tagname, $self_closing_tags)){
2511 $tagtype = 3;
2512 }
2513 /**
2514 * See if we should skip this tag and any content
2515 * inside it.
2516 */
2517 if ($tagtype == 1 &&
2518 in_array($tagname, $rm_tags_with_content)){
2519 $skip_content = $tagname;
2520 } else {
2521 if (($rm_tags == false
2522 && in_array($tagname, $tag_list)) ||
2523 ($rm_tags == true &&
2524 !in_array($tagname, $tag_list))){
2525 $tagname = false;
2526 } else {
2527 /**
2528 * Convert body into div.
2529 */
2530 if ($tagname == "body"){
2531 $tagname = "div";
2532 $attary = sq_body2div($attary, $mailbox,
2533 $message, $id);
2534 }
2535 if ($tagtype == 1){
2536 if (isset($open_tags[$tagname])){
2537 $open_tags[$tagname]++;
2538 } else {
2539 $open_tags[$tagname]=1;
2540 }
2541 }
2542 /**
2543 * This is where we run other checks.
2544 */
2545 if (is_array($attary) && sizeof($attary) > 0){
2546 $attary = sq_fixatts($tagname,
2547 $attary,
2548 $rm_attnames,
2549 $bad_attvals,
2550 $add_attr_to_tag,
2551 $message,
2552 $id,
2553 $mailbox
2554 );
2555 }
2556 }
2557 }
2558 }
2559 }
2560 if ($tagname != false && $skip_content == false){
2561 $trusted .= sq_tagprint($tagname, $attary, $tagtype);
2562 }
2563 }
2564 $curpos = $gt+1;
2565 }
2566 $trusted .= substr($body, $curpos, strlen($body)-$curpos);
2567 if ($force_tag_closing == true){
2568 foreach ($open_tags as $tagname=>$opentimes){
2569 while ($opentimes > 0){
2570 $trusted .= '</' . $tagname . '>';
2571 $opentimes--;
2572 }
2573 }
2574 $trusted .= "\n";
2575 }
2576 $trusted .= "<!-- end sanitized html -->\n";
2577 return $trusted;
2578 }
2579
2580 /**
2581 * This is a wrapper function to call html sanitizing routines.
2582 *
2583 * @param $body the body of the message
2584 * @param $id the id of the message
2585
2586 * @param $message
2587 * @param $mailbox
2588 * @param boolean $take_mailto_links When TRUE, converts mailto: links
2589 * into internal SM compose links
2590 * (optional; default = TRUE)
2591 * @return a string with html safe to display in the browser.
2592 */
2593 function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) {
2594
2595 // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match
2596
2597 global $attachment_common_show_images, $view_unsafe_images,
2598 $has_unsafe_images, $allow_svg_display, $rcdata_rawtext_tags,
2599 $remove_rcdata_rawtext_tags_and_content;
2600
2601 $rcdata_rawtext_tags = array(
2602 "noscript",
2603 "noframes",
2604 "noembed",
2605 "textarea",
2606 // also "title", "xmp", "script", "iframe", "plaintext" which we already remove below
2607 );
2608
2609 /**
2610 * Don't display attached images in HTML mode.
2611 *
2612 * SB: why?
2613 */
2614 $attachment_common_show_images = false;
2615 $tag_list = Array(
2616 false, // remove these tags
2617 "meta",
2618 "html",
2619 "head",
2620 "base",
2621 "link",
2622 "frame",
2623 "iframe",
2624 "plaintext",
2625 "marquee",
2626 );
2627
2628 $rm_tags_with_content = Array(
2629 "script",
2630 "object",
2631 "applet",
2632 "embed",
2633 "title",
2634 "frameset",
2635 "xmp",
2636 "xml",
2637 );
2638 if (!$allow_svg_display)
2639 $rm_tags_with_content[] = 'svg';
2640 /**
2641 * SquirrelMail will parse RCDATA and RAWTEXT tags and handle them as the special
2642 * case that they are, but if you prefer to remove them and their contents entirely
2643 * (in most cases, should be a safe thing with minimal impact), you can add the
2644 * following to config/config_local.php
2645 * $remove_rcdata_rawtext_tags_and_content = TRUE;
2646 */
2647 if ($remove_rcdata_rawtext_tags_and_content)
2648 $rm_tags_with_content = array_merge($rm_tags_with_content, $rcdata_rawtext_tags);
2649
2650 $self_closing_tags = Array(
2651 "img",
2652 "br",
2653 "hr",
2654 "input",
2655 "outbind",
2656 );
2657
2658 $force_tag_closing = true;
2659
2660 $rm_attnames = Array(
2661 "/.*/" =>
2662 Array(
2663 "/target/i",
2664 "/^on.*/i",
2665 "/^dynsrc/i",
2666 "/^data.*/i",
2667 "/^lowsrc.*/i",
2668 )
2669 );
2670
2671 global $use_transparent_security_image;
2672 if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png';
2673 else $secremoveimg = '../images/' . _("sec_remove_eng.png");
2674
2675 $bad_attvals = Array(
2676 "/.*/" =>
2677 Array(
2678 "/^src|background/i" =>
2679 Array(
2680 Array(
2681 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2682 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2683 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2684 ),
2685 Array(
2686 "\\1$secremoveimg\\2",
2687 "\\1$secremoveimg\\2",
2688 "\\1$secremoveimg\\2",
2689 )
2690 ),
2691 "/^href|action/i" =>
2692 Array(
2693 Array(
2694 "/^([\'\"])\s*\S+script\s*:.*([\'\"])/si",
2695 "/^([\'\"])\s*mocha\s*:*.*([\'\"])/si",
2696 "/^([\'\"])\s*about\s*:.*([\'\"])/si"
2697 ),
2698 Array(
2699 "\\1#\\1",
2700 "\\1#\\1",
2701 "\\1#\\1"
2702 )
2703 ),
2704 "/^style/i" =>
2705 Array(
2706 Array(
2707 "/\/\*.*\*\//",
2708 "/expression/i",
2709 "/binding/i",
2710 "/behaviou*r/i",
2711 "/include-source/i",
2712
2713 // position:relative can also be exploited
2714 // to put content outside of email body area
2715 // and position:fixed is similarly exploitable
2716 // as position:absolute, so we'll remove it
2717 // altogether....
2718 //
2719 // Does this screw up legitimate HTML messages?
2720 // If so, the only fix I see is to allow position
2721 // attributes (any values? I think we still have
2722 // to block static and fixed) only if $use_iframe
2723 // is enabled (1.5.0+)
2724 //
2725 // was: "/position\s*:\s*absolute/i",
2726 //
2727 "/position\s*:/i",
2728
2729 "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i",
2730 "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si",
2731 "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si",
2732 "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si",
2733 "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si",
2734 ),
2735 Array(
2736 "",
2737 "idiocy",
2738 "idiocy",
2739 "idiocy",
2740 "idiocy",
2741 "idiocy",
2742 "url",
2743 "url(\\1#\\1)",
2744 "url(\\1#\\1)",
2745 "url(\\1#\\1)",
2746 "\\1:url(\\2#\\3)"
2747 )
2748 )
2749 )
2750 );
2751
2752 // If there's no "view_unsafe_images" variable in the URL, turn unsafe
2753 // images off by default.
2754 sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE);
2755
2756 if (!$view_unsafe_images){
2757 /**
2758 * Remove any references to http/https if view_unsafe_images set
2759 * to false.
2760 */
2761 array_push($bad_attvals['/.*/']['/^src|background/i'][0],
2762 '/^([\'\"])\s*https*:.*([\'\"])/si');
2763 array_push($bad_attvals['/.*/']['/^src|background/i'][1],
2764 "\\1$secremoveimg\\1");
2765 array_push($bad_attvals['/.*/']['/^style/i'][0],
2766 '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si');
2767 array_push($bad_attvals['/.*/']['/^style/i'][1],
2768 "url(\\1$secremoveimg\\1)");
2769 }
2770
2771 $add_attr_to_tag = Array(
2772 "/^a$/i" =>
2773 Array('target'=>'"_blank"',
2774 'title'=>'"'._("This external link will open in a new window").'"'
2775 )
2776 );
2777 $trusted = sq_sanitize($body,
2778 $tag_list,
2779 $rm_tags_with_content,
2780 $self_closing_tags,
2781 $force_tag_closing,
2782 $rm_attnames,
2783 $bad_attvals,
2784 $add_attr_to_tag,
2785 $message,
2786 $id,
2787 $mailbox
2788 );
2789 if (strpos($trusted,$secremoveimg)){
2790 $has_unsafe_images = true;
2791 }
2792
2793 // we want to parse mailto's in HTML output, change to SM compose links
2794 // this is a modified version of code from url_parser.php... but Marc is
2795 // right: we need a better filtering implementation; adding this randomly
2796 // here is not a great solution
2797 //
2798 if ($take_mailto_links) {
2799 // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive
2800 global $MailTo_PReg_Match;
2801 // some mailers (Microsoft, surprise surprise) produce mailto strings without being
2802 // inside an anchor (link) tag, so we have to make sure the regex looks for the
2803 // quote before mailto, and we'll also try to convert the non-links back into links
2804 $MailTo_PReg_Match = '/([\'"])?mailto:' . substr($MailTo_PReg_Match, 1) ;
2805 if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) {
2806 foreach ($regs[0] as $i => $mailto_before) {
2807 $mailto_params = $regs[11][$i];
2808
2809 // get rid of any leading quote we may have captured but don't care about
2810 //
2811 $mailto_before = ltrim($mailto_before, '"\'');
2812
2813 // get rid of any tailing quote since we have to add send_to to the end
2814 //
2815 $mailto_before = rtrim($mailto_before, '"\'');
2816 $mailto_params = rtrim($mailto_params, '"\'');
2817
2818 if ($regs[2][$i]) { //if there is an email addr before '?', we need to merge it with the params
2819 $to = 'to=' . $regs[2][$i];
2820 if (strpos($mailto_params, 'to=') > -1) //already a 'to='
2821 $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
2822 else {
2823 if ($mailto_params) //already some params, append to them
2824 $mailto_params .= '&amp;' . $to;
2825 else
2826 $mailto_params .= '?' . $to;
2827 }
2828 }
2829
2830 $url_str = preg_replace(array('/to=/i', '/(?<!b)cc=/i', '/bcc=/i'), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
2831
2832 // we'll already have target=_blank, no need to allow comp_in_new
2833 // here (which would be a lot more work anyway)
2834 //
2835 global $compose_new_win;
2836 $temp_comp_in_new = $compose_new_win;
2837 $compose_new_win = 0;
2838 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
2839 $compose_new_win = $temp_comp_in_new;
2840
2841 // remove <a href=" and anything after the next quote (we only
2842 // need the uri, not the link HTML) in compose uri
2843 //
2844 // but only do this if the original mailto was in a real anchor tag
2845 //
2846 if (!empty($regs[1][$i])) {
2847 $comp_uri = substr($comp_uri, 9);
2848 $comp_uri = substr($comp_uri, 0, strpos($comp_uri, '"', 1));
2849 }
2850 $trusted = str_replace($mailto_before, $comp_uri, $trusted);
2851 }
2852 }
2853 }
2854
2855 return $trusted;
2856 }
2857
2858 /**
2859 * function SendDownloadHeaders - send file to the browser
2860 *
2861 * Original Source: SM core src/download.php
2862 * moved here to make it available to other code, and separate
2863 * front end from back end functionality.
2864 *
2865 * @param string $type0 first half of mime type
2866 * @param string $type1 second half of mime type
2867 * @param string $filename filename to tell the browser for downloaded file
2868 * @param boolean $force whether to force the download dialog to pop
2869 * @param optional integer $filesize send the Content-Header and length to the browser
2870 * @return void
2871 */
2872 function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) {
2873 global $languages, $squirrelmail_language;
2874 $isIE = $isIE6plus = false;
2875
2876 sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER);
2877
2878 if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false &&
2879 strstr($HTTP_USER_AGENT, 'Opera') === false) {
2880 $isIE = true;
2881 }
2882
2883 if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) &&
2884 ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) {
2885 $isIE6plus = true;
2886 }
2887
2888 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
2889 function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename')) {
2890 $filename =
2891 call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_downloadfilename', $filename, $HTTP_USER_AGENT);
2892 } else {
2893 $filename = preg_replace('/[\\\\\/:*?"<>|;]/', '_', str_replace('&nbsp;', ' ', $filename));
2894 }
2895
2896 // A Pox on Microsoft and it's Internet Explorer!
2897 //
2898 // IE has lots of bugs with file downloads.
2899 // It also has problems with SSL. Both of these cause problems
2900 // for us in this function.
2901 //
2902 // See this article on Cache Control headers and SSL
2903 // http://support.microsoft.com/default.aspx?scid=kb;en-us;323308
2904 //
2905 // The best thing you can do for IE is to upgrade to the latest
2906 // version
2907 //set all the Cache Control Headers for IE
2908 if ($isIE) {
2909 $filename=rawurlencode($filename);
2910 header ("Pragma: public");
2911 header ("Cache-Control: no-store, max-age=0, no-cache, must-revalidate"); // HTTP/1.1
2912 // does nothing - see: https://blogs.msdn.microsoft.com/ieinternals/2009/07/20/internet-explorers-cache-control-extensions/
2913 // header ("Cache-Control: post-check=0, pre-check=0", false);
2914 header ("Cache-Control: private");
2915
2916 //set the inline header for IE, we'll add the attachment header later if we need it
2917 header ("Content-Disposition: inline; filename=$filename");
2918 }
2919
2920 if (!$force) {
2921 // Try to show in browser window
2922 header ("Content-Disposition: inline; filename=\"$filename\"");
2923 header ("Content-Type: $type0/$type1; name=\"$filename\"");
2924 } else {
2925 // Try to pop up the "save as" box
2926
2927 // IE makes this hard. It pops up 2 save boxes, or none.
2928 // http://support.microsoft.com/support/kb/articles/Q238/5/88.ASP
2929 // http://support.microsoft.com/default.aspx?scid=kb;EN-US;260519
2930 // But, according to Microsoft, it is "RFC compliant but doesn't
2931 // take into account some deviations that allowed within the
2932 // specification." Doesn't that mean RFC non-compliant?
2933 // http://support.microsoft.com/support/kb/articles/Q258/4/52.ASP
2934
2935 // all browsers need the application/octet-stream header for this
2936 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2937
2938 // http://support.microsoft.com/support/kb/articles/Q182/3/15.asp
2939 // Do not have quotes around filename, but that applied to
2940 // "attachment"... does it apply to inline too?
2941 header ("Content-Disposition: attachment; filename=\"$filename\"");
2942
2943 if ($isIE && !$isIE6plus) {
2944 // This combination seems to work mostly. IE 5.5 SP 1 has
2945 // known issues (see the Microsoft Knowledge Base)
2946
2947 // This works for most types, but doesn't work with Word files
2948 header ("Content-Type: application/download; name=\"$filename\"");
2949 header ("Content-Type: application/force-download; name=\"$filename\"");
2950 // These are spares, just in case. :-)
2951 //header("Content-Type: $type0/$type1; name=\"$filename\"");
2952 //header("Content-Type: application/x-msdownload; name=\"$filename\"");
2953 //header("Content-Type: application/octet-stream; name=\"$filename\"");
2954 } else if ($isIE) {
2955 // This is to prevent IE for MIME sniffing and auto open a file in IE
2956 header ("Content-Type: application/force-download; name=\"$filename\"");
2957 } else {
2958 // another application/octet-stream forces download for Netscape
2959 header ("Content-Type: application/octet-stream; name=\"$filename\"");
2960 }
2961 }
2962
2963 //send the content-length header if the calling function provides it
2964 if ($filesize > 0) {
2965 header("Content-Length: $filesize");
2966 }
2967
2968 } // end fn SendDownloadHeaders
Unnamed repository; edit this file 'description' to name the repository.