X-Git-Url: https://vcs.fsf.org/?p=squirrelmail.git;a=blobdiff_plain;f=functions%2Fmime.php;h=09cdec965ff4bd4f2e8b17d8ec99f56ef3fa72f3;hp=51d84b242e5b9b24de17af3346afe66f046f097c;hb=dfc64d149b23ad2e06a10ff4f187ea9ce664bbc9;hpb=d74481266127089b66cd2bc17a0c5c3f53cbca4e diff --git a/functions/mime.php b/functions/mime.php index 51d84b24..b17eb5b2 100644 --- a/functions/mime.php +++ b/functions/mime.php @@ -3,27 +3,53 @@ /** * mime.php * - * Copyright (c) 1999-2003 The SquirrelMail Project Team - * Licensed under the GNU GPL. For full terms see the file COPYING. - * * This contains the functions necessary to detect and decode MIME * messages. * - * $Id$ + * @copyright 1999-2019 The SquirrelMail Project Team + * @license http://opensource.org/licenses/gpl-license.php GNU Public License + * @version $Id$ + * @package squirrelmail */ -require_once(SM_PATH . 'functions/imap.php'); -require_once(SM_PATH . 'functions/attachment_common.php'); +/** + * dependency information + functions dependency + mime_structure + class/mime/Message.class.php + Message::parseStructure + functions/page_header.php + displayPageHeader + functions/display_messages.php + plain_error_message + mime_fetch_body + functions/imap_general.php + sqimap_run_command + mime_print_body_lines + + + +functions/imap.php +functions/attachment_common.php +functions/display_messages.php + +magicHtml => url_parser +translateText => url_parser + +*/ + -/* --------------------------------------------------------------------------------- */ -/* MIME DECODING */ -/* --------------------------------------------------------------------------------- */ +/* -------------------------------------------------------------------------- */ +/* MIME DECODING */ +/* -------------------------------------------------------------------------- */ -/* This function gets the structure of a message and stores it in the "message" class. +/** + * Get the MIME structure + * + * This function gets the structure of a message and stores it in the "message" class. * It will return this object for use with all relevant header information and * fully parsed into the standard "message" object format. */ - function mime_structure ($bodystructure, $flags=array()) { /* Isolate the body structure and remove beginning and end parenthesis. */ @@ -31,22 +57,20 @@ function mime_structure ($bodystructure, $flags=array()) { $read = trim(substr ($read, 0, -1)); $i = 0; $msg = Message::parseStructure($read,$i); + if (!is_object($msg)) { - include_once(SM_PATH . 'functions/display_messages.php'); global $color, $mailbox; - /* removed urldecode because $_GET is auto urldecoded ??? */ displayPageHeader( $color, $mailbox ); - echo "\n\n" . - '
'; $errormessage = _("SquirrelMail could not decode the bodystructure of the message"); - $errormessage .= '
'._("the provided bodystructure by your imap-server").':

'; - $errormessage .= '
' . htmlspecialchars($read) . '
'; - plain_error_message( $errormessage, $color ); + $errormessage .= '
'._("The bodystructure provided by your IMAP server:").'

'; + $errormessage .= '
' . sm_encode_html_special_chars($read) . '
'; + plain_error_message( $errormessage ); echo ''; exit; } if (count($flags)) { foreach ($flags as $flag) { +//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. $char = strtoupper($flag{1}); switch ($char) { case 'S': @@ -68,6 +92,9 @@ function mime_structure ($bodystructure, $flags=array()) { if (strtolower($flag) == '\\flagged') { $msg->is_flagged = true; } + else if (strtolower($flag) == '$forwarded') { + $msg->is_forwarded = true; + } break; case 'M': if (strtolower($flag) == '$mdnsent') { @@ -94,33 +121,43 @@ function mime_structure ($bodystructure, $flags=array()) { * to mime_get_elements() */ -function mime_fetch_body($imap_stream, $id, $ent_id=1) { - global $uid_support; +function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) { /* Do a bit of error correction. If we couldn't find the entity id, just guess * that it is the first one. That is usually the case anyway. */ + if (!$ent_id) { $cmd = "FETCH $id BODY[]"; } else { $cmd = "FETCH $id BODY[$ent_id]"; } - $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, $uid_support); + if ($fetch_size!=0) $cmd .= "<0.$fetch_size>"; + + $data = sqimap_run_command ($imap_stream, $cmd, true, $response, $message, TRUE); do { $topline = trim(array_shift($data)); - } while($topline && ($topline[0] == '*') && !preg_match('/\* [0-9]+ FETCH.*/i', $topline)) ; + } while($topline && ($topline[0] == '*') && !preg_match('/\* [0-9]+ FETCH .*BODY.*/i', $topline)) ; + // Matching with "BODY" above is difficult: in most cases "FETCH \(BODY" would work + // but some servers may put other things in the same result, perhaps something such + // as "* 23 FETCH (FLAGS (\Seen) BODY[1] {174}". There is some small chance that + // if the character sequence "BODY" appears in a response where it isn't actually + // a FETCH response data item name, the current regex will break things. The better + // way to do this would be to parse the response correctly and not use a regex. $wholemessage = implode('', $data); - if (ereg('\\{([^\\}]*)\\}', $topline, $regs)) { + if (preg_match('/\{([^\}]*)\}/', $topline, $regs)) { $ret = substr($wholemessage, 0, $regs[1]); /* There is some information in the content info header that could be important * in order to parse html messages. Let's get them here. */ // if ($ret{0} == '<') { -// $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, $uid_support); +// $data = sqimap_run_command ($imap_stream, "FETCH $id BODY[$ent_id.MIME]", true, $response, $message, TRUE); // } - } else if (ereg('"([^"]*)"', $topline, $regs)) { + } else if (preg_match('/"([^"]*)"/', $topline, $regs)) { $ret = $regs[1]; + } else if ((stristr($topline, 'nil') !== false) && (empty($wholemessage))) { + $ret = $wholemessage; } else { global $where, $what, $mailbox, $passed_id, $startMessage; $par = 'mailbox=' . urlencode($mailbox) . '&passed_id=' . $passed_id; @@ -130,21 +167,21 @@ function mime_fetch_body($imap_stream, $id, $ent_id=1) { $par .= '&startMessage=' . $startMessage . '&show_more=0'; } $par .= '&response=' . urlencode($response) . - '&message=' . urlencode($message) . - '&topline=' . urlencode($topline); - - echo '
' . - '' . - '' . - '" . - '" . - '" . - '" . - "
' . - _("Body retrieval error. The reason for this is most probably that the message is malformed.") . - '
' . _("Command:") . "$cmd
' . _("Response:") . "$response
' . _("Message:") . "$message
' . _("FETCH line:") . "$topline


"; - - $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, $uid_support); + '&message=' . urlencode($message) . + '&topline=' . urlencode($topline); + + echo '
' . + '' . + '' . + '" . + '" . + '" . + '" . + "
' . + _("Body retrieval error. The reason for this is most probably that the message is malformed.") . + '
' . _("Command:") . "$cmd
' . _("Response:") . "$response
' . _("Message:") . "$message
' . _("FETCH line:") . "$topline


"; + + $data = sqimap_run_command ($imap_stream, "FETCH $passed_id BODY[]", true, $response, $message, TRUE); array_shift($data); $wholemessage = implode('', $data); @@ -153,54 +190,50 @@ function mime_fetch_body($imap_stream, $id, $ent_id=1) { return $ret; } -function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding) { - global $uid_support; +function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') { - $sid = sqimap_session_id($uid_support); /* Don't kill the connection if the browser is over a dialup * and it would take over 30 seconds to download it. - * DonĀ“t call set_time_limit in safe mode. + * Don't call set_time_limit in safe mode. */ if (!ini_get('safe_mode')) { set_time_limit(0); } - if ($uid_support) { - $sid_s = substr($sid,0,strpos($sid, ' ')); + /* in case of base64 encoded attachments, do not buffer them. + Instead, echo the decoded attachment directly to screen */ + if (strtolower($encoding) == 'base64') { + if (!$ent_id) { + $query = "FETCH $id BODY[]"; + } else { + $query = "FETCH $id BODY[$ent_id]"; + } + sqimap_run_command($imap_stream,$query,true,$response,$message,TRUE,'sqimap_base64_decode',$rStream,true); } else { - $sid_s = $sid; + $body = mime_fetch_body ($imap_stream, $id, $ent_id); + if (is_resource($rStream)) { + fputs($rStream,decodeBody($body, $encoding, $force_crlf)); + } else { + echo decodeBody($body, $encoding, $force_crlf); + } } - $body = mime_fetch_body ($imap_stream, $id, $ent_id); - echo decodeBody($body, $encoding); + /* + TODO, use the same method for quoted printable. + However, I assume that quoted printable attachments aren't that large + so the performancegain / memory usage drop will be minimal. + If we decide to add that then we need to adapt sqimap_fread because + we need to split te result on \n and fread doesn't stop at \n. That + means we also should provide $results from sqimap_fread (by ref) to + te function and set $no_return to false. The $filter function for + quoted printable should handle unsetting of $results. + */ + /* + TODO 2: find out how we write to the output stream php://stdout. fwrite + doesn't work because 'php://stdout isn't a stream. + */ + return; -/* - fputs ($imap_stream, "$sid FETCH $id BODY[$ent_id]\r\n"); - $cnt = 0; - $continue = true; - $read = fgets ($imap_stream,8192); - - - // This could be bad -- if the section has sqimap_session_id() . ' OK' - // or similar, it will kill the download. - while (!ereg("^".$sid_s." (OK|BAD|NO)(.*)$", $read, $regs)) { - if (trim($read) == ')==') { - $read1 = $read; - $read = fgets ($imap_stream,4096); - if (ereg("^".$sid." (OK|BAD|NO)(.*)$", $read, $regs)) { - return; - } else { - echo decodeBody($read1, $encoding) . - decodeBody($read, $encoding); - } - } else if ($cnt) { - echo decodeBody($read, $encoding); - } - $read = fgets ($imap_stream,4096); - $cnt++; -// break; - } -*/ } /* -[ END MIME DECODING ]----------------------------------------------------------- */ @@ -210,7 +243,7 @@ function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding) { */ function listEntities ($message) { if ($message) { - echo "" . $message->entity_id . ' : ' . $message->type0 . '/' . $message->type1 . ' parent = '. $message->parent->entity_id. '
'; + echo "" . $message->entity_id . ' : ' . $message->type0 . '/' . $message->type1 . ' parent = '. $message->parent->entity_id. '
'; for ($i = 0; isset($message->entities[$i]); $i++) { echo "$i : "; $msg = listEntities($message->entities[$i]); @@ -263,13 +296,14 @@ function translateText(&$body, $wrap_at, $charset) { global $where, $what; /* from searching */ global $color; /* color theme */ - require_once(SM_PATH . 'functions/url_parser.php'); + // require_once(SM_PATH . 'functions/url_parser.php'); $body_ary = explode("\n", $body); for ($i=0; $i < count($body_ary); $i++) { - $line = $body_ary[$i]; + $line = rtrim($body_ary[$i],"\r"); + if (strlen($line) - 2 >= $wrap_at) { - sqWordWrap($line, $wrap_at); + sqWordWrap($line, $wrap_at, $charset); } $line = charset_decode($charset, $line); $line = str_replace("\t", ' ', $line); @@ -292,15 +326,9 @@ function translateText(&$body, $wrap_at, $charset) { } if ($quotes % 2) { - if (!isset($color[13])) { - $color[13] = '#800000'; - } - $line = '' . $line . ''; + $line = '' . $line . ''; } elseif ($quotes) { - if (!isset($color[14])) { - $color[14] = '#FF0000'; - } - $line = '' . $line . ''; + $line = '' . $line . ''; } $body_ary[$i] = $line; @@ -308,40 +336,59 @@ function translateText(&$body, $wrap_at, $charset) { $body = '
' . implode("\n", $body_ary) . '
'; } -/* This returns a parsed string called $body. That string can then +/** + * This returns a parsed string called $body. That string can then * be displayed as the actual message in the HTML. It contains * everything needed, including HTML Tags, Attachments at the * bottom, etc. + * + * Since 1.2.0 function uses message_body hook. + * Till 1.3.0 function included output of formatAttachments(). + * + * @param resource $imap_stream imap connection resource + * @param object $message squirrelmail message object + * @param array $color squirrelmail color theme array + * @param integer $wrap_at number of characters per line + * @param string $ent_num (since 1.3.0) message part id + * @param integer $id (since 1.3.0) message id + * @param string $mailbox (since 1.3.0) imap folder name + * @return string html formated message text */ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX') { /* This if statement checks for the entity to show as the * primary message. To add more of them, just put them in the * order that is their priority. */ - global $startMessage, $username, $key, $imapServerAddress, $imapPort, - $show_html_default, $sort, $has_unsafe_images, $passed_ent_id; - global $languages, $squirrelmail_language; + global $startMessage, $languages, $squirrelmail_language, + $show_html_default, $sort, $has_unsafe_images, $passed_ent_id, + $use_iframe, $iframe_height, $download_and_unsafe_link, + $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text, + $oTemplate, $nbsp; - if( !sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET) ) { - $view_unsafe_images = false; - } + // workaround for not updated config.php + if (! isset($use_iframe)) $use_iframe = false; + + // If there's no "view_unsafe_images" variable in the URL, turn unsafe + // images off by default. + sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE); $body = ''; $urlmailbox = urlencode($mailbox); $body_message = getEntity($message, $ent_num); if (($body_message->header->type0 == 'text') || - ($body_message->header->type0 == 'rfc822')) { + ($body_message->header->type0 == 'rfc822')) { $body = mime_fetch_body ($imap_stream, $id, $ent_num); $body = decodeBody($body, $body_message->header->encoding); if (isset($languages[$squirrelmail_language]['XTRA_CODE']) && - function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) { + function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode')) { if (mb_detect_encoding($body) != 'ASCII') { - $body = $languages[$squirrelmail_language]['XTRA_CODE']('decode', $body); + $body = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decode',$body); } } - $hookResults = do_hook("message_body", $body); - $body = $hookResults[1]; + + /* As of 1.5.2, $body is passed (and modified) by reference */ + do_hook('message_body', $body); /* If there are other types that shouldn't be formatted, add * them here. @@ -351,29 +398,89 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma if ($show_html_default <> 1) { $entity_conv = array(' ' => ' ', '

' => "\n", - '
' => "\n", '

' => "\n", + '
' => "\n", '
' => "\n", + '
' => "\n", + '
' => "\n", '>' => '>', '<' => '<'); $body = strtr($body, $entity_conv); $body = strip_tags($body); $body = trim($body); translateText($body, $wrap_at, - $body_message->header->getParameter('charset')); + $body_message->header->getParameter('charset')); + } elseif ($use_iframe) { + /** + * If we don't add html message between iframe tags, + * we must detect unsafe images and modify $has_unsafe_images. + */ + $html_body = magicHTML($body, $id, $message, $mailbox); + // Convert character set in order to display html mails in different character set + $html_body = charset_decode($body_message->header->getParameter('charset'),$html_body,false,true); + + // creating iframe url + $iframeurl=sqm_baseuri().'src/view_html.php?' + . 'mailbox=' . $urlmailbox + . '&passed_id=' . $id + . '&ent_id=' . $ent_num + . '&view_unsafe_images=' . (int) $view_unsafe_images; + + global $oTemplate; + $oTemplate->assign('iframe_url', $iframeurl); + $oTemplate->assign('iframe_height', $iframe_height); + $oTemplate->assign('html_body', $html_body); + + $body = $oTemplate->fetch('read_html_iframe.tpl'); } else { + // old way of html rendering + /** + * convert character set. charset_decode does not remove html special chars + * applied by magicHTML functions and does not sanitize them second time if + * fourth argument is true. + */ + $charset = $body_message->header->getParameter('charset'); + if (!empty($charset)) { + $body = charset_decode($charset,$body,false,true); + } $body = magicHTML($body, $id, $message, $mailbox); } } else { translateText($body, $wrap_at, - $body_message->header->getParameter('charset')); + $body_message->header->getParameter('charset')); } - $link = 'read_body.php?passed_id=' . $id . '&ent_id='.$ent_num. - '&mailbox=' . $urlmailbox .'&sort=' . $sort . - '&startMessage=' . $startMessage . '&show_more=0'; + + /* + * Previously the links for downloading and unsafe images were printed + * under the mail. By putting the links in a global variable we can + * print it in the toolbar where it belongs. Since the original code was + * in this place it's left here. It might be possible to move it to some + * other place if that makes sense. The possibility to do so has not + * been evaluated yet. + */ + + // Initialize the global variable to an empty string. + // 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. + $download_and_unsafe_link = ''; + + // Prepare and build a link for downloading the mail. + $link = 'passed_id=' . $id . '&ent_id='.$ent_num. + '&mailbox=' . $urlmailbox .'&sort=' . $sort . + '&startMessage=' . $startMessage . '&show_more=0'; if (isset($passed_ent_id)) { $link .= '&passed_ent_id='.$passed_ent_id; } + $download_href = SM_PATH . 'src/download.php?absolute_dl=true&' . $link; + + // Always add the link for downloading the mail as a file to the global + // variable. + $download_and_unsafe_link .= "$nbsp|$nbsp" + . create_hyperlink($download_href, _("Download this as a file")); + + // Find out the right text to use in the link depending on the + // circumstances. If the unsafe images are displayed the link should + // hide them, if they aren't displayed the link should only appear if + // the mail really contains unsafe images. if ($view_unsafe_images) { $text = _("Hide Unsafe Images"); } else { @@ -384,37 +491,52 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma $text = ''; } } - $body .= '

'.$text. - '

' . "\n"; + + // Only create a link for unsafe images if there's need for one. If so: + // add it to the global variable. + if($text != '') { + $unsafe_image_toggle_href = SM_PATH . 'src/read_body.php?'.$link; + $unsafe_image_toggle_text = $text; + $download_and_unsafe_link .= "$nbsp|$nbsp" + . create_hyperlink($unsafe_image_toggle_href, $text); + } } return $body; } - -function formatAttachments($message, $exclude_id, $mailbox, $id) { - global $where, $what, $startMessage, $color; - static $ShownHTML = 0; +/** + * Generate attachments array for passing to templates. + * + * @since 1.5.2 + * @param object $message SquirrelMail message object + * @param array $exclude_id message parts that are not attachments. + * @param string $mailbox mailbox name + * @param integer $id message id + */ +function buildAttachmentArray($message, $exclude_id, $mailbox, $id) { + global $where, $what, $startMessage, $color, $passed_ent_id, + $base_uri, $block_svg_download; $att_ar = $message->getAttachments($exclude_id); - - if (!count($att_ar)) return ''; - - $attachments = ''; - $urlMailbox = urlencode($mailbox); + $attachments = array(); foreach ($att_ar as $att) { $ent = $att->entity_id; $header = $att->header; $type0 = strtolower($header->type0); $type1 = strtolower($header->type1); + if ($block_svg_download && strpos($type1, 'svg') === 0) + continue; + $name = ''; - $links['download link']['text'] = _("download"); - $links['download link']['href'] = SM_PATH . - "src/download.php?absolute_dl=true&passed_id=$id&mailbox=$urlMailbox&ent_id=$ent"; - $ImageURL = ''; + $links = array(); + $links['download link']['text'] = _("Download"); + $links['download link']['href'] = $base_uri . + "src/download.php?absolute_dl=true&passed_id=$id&mailbox=$urlMailbox&ent_id=$ent"; + if ($type0 =='message' && $type1 == 'rfc822') { - $default_page = SM_PATH . 'src/read_body.php'; + $default_page = $base_uri . 'src/read_body.php'; $rfc822_header = $att->rfc822_header; $filename = $rfc822_header->subject; if (trim( $filename ) == '') { @@ -422,43 +544,21 @@ function formatAttachments($message, $exclude_id, $mailbox, $id) { } $from_o = $rfc822_header->from; if (is_object($from_o)) { - $from_name = $from_o->getAddress(false); + $from_name = decodeHeader($from_o->getAddress(false)); + } elseif (is_array($from_o) && count($from_o) && is_object($from_o[0])) { + // something weird happens when a digest message is opened and you return to the digest + // now the from object is part of an array. Probably the parseHeader call overwrites the info + // retrieved from the bodystructure in a different way. We need to fix this later. + // possible starting point, do not fetch header we already have and inspect how + // the rfc822_header object behaves. + $from_name = decodeHeader($from_o[0]->getAddress(false)); } else { $from_name = _("Unknown sender"); } - $from_name = decodeHeader(($from_name)); - $description = $from_name; + $description = _("From").': '.$from_name; } else { - $default_page = SM_PATH . 'src/download.php'; - if (is_object($header->disposition)) { - $filename = $header->disposition->getProperty('filename'); - if (trim($filename) == '') { - $name = decodeHeader($header->disposition->getProperty('name')); - if (trim($name) == '') { - $name = $header->getParameter('name'); - if(trim($name) == '') { - if (trim( $header->id ) == '') { - $filename = 'untitled-[' . $ent . ']' ; - } else { - $filename = 'cid: ' . $header->id; - } - } else { - $filename = $name; - } - } else { - $filename = $name; - } - } - } else { - $filename = $header->getParameter('name'); - if (!trim($filename)) { - if (trim( $header->id ) == '') { - $filename = 'untitled-[' . $ent . ']' ; - } else { - $filename = 'cid: ' . $header->id; - } - } - } + $default_page = $base_uri . 'src/download.php'; + $filename = $att->getFilename(); if ($header->description) { $description = decodeHeader($header->description); } else { @@ -473,158 +573,348 @@ function formatAttachments($message, $exclude_id, $mailbox, $id) { $passed_ent_id_link = ''; } $defaultlink = $default_page . "?startMessage=$startMessage" - . "&passed_id=$id&mailbox=$urlMailbox" - . '&ent_id='.$ent.$passed_ent_id_link; + . "&passed_id=$id&mailbox=$urlMailbox" + . '&ent_id='.$ent.$passed_ent_id_link; if ($where && $what) { - $defaultlink .= '&where='. urlencode($where).'&what='.urlencode($what); + $defaultlink .= '&where='. urlencode($where).'&what='.urlencode($what); + } + // IE does make use of mime content sniffing. Forcing a download + // prohibit execution of XSS inside an application/octet-stream attachment + if ($type0 == 'application' && $type1 == 'octet-stream') { + $defaultlink .= '&absolute_dl=true'; } + /* This executes the attachment hook with a specific MIME-type. - * If that doesn't have results, it tries if there's a rule - * for a more generic type. + * It also allows plugins to run if there's a rule for a more + * generic type. Finally, a hook for ALL attachment types is + * run as well. */ - $hookresults = do_hook("attachment $type0/$type1", $links, - $startMessage, $id, $urlMailbox, $ent, $defaultlink, - $display_filename, $where, $what); - if(count($hookresults[1]) <= 1) { - $hookresults = do_hook("attachment $type0/*", $links, - $startMessage, $id, $urlMailbox, $ent, $defaultlink, - $display_filename, $where, $what); + // First remember the default link. + $defaultlink_orig = $defaultlink; + + /* The API for this hook has changed as of 1.5.2 so that all plugin + arguments are passed in an array instead of each their own plugin + argument, and arguments are passed by reference, so instead of + returning any changes, changes should simply be made to the original + arguments themselves. */ + $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent, + &$defaultlink, &$display_filename, &$where, &$what); + do_hook("attachment $type0/$type1", $temp); + /* The API for this hook has changed as of 1.5.2 so that all plugin + arguments are passed in an array instead of each their own plugin + argument, and arguments are passed by reference, so instead of + returning any changes, changes should simply be made to the original + arguments themselves. */ + $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent, + &$defaultlink, &$display_filename, &$where, &$what); + // Do not let a generic plugin change the default link if a more + // specialized one already did it... + if ($defaultlink != $defaultlink_orig) { + $dummy = ''; + $temp[5] = &$dummy; } - - $links = $hookresults[1]; - $defaultlink = $hookresults[6]; - - $attachments .= '' . - ''.decodeHeader($display_filename).' ' . - '' . show_readable_size($header->size) . - '  ' . - "[ $type0/$type1 ] " . - ''; - $attachments .= '' . $description . ''; - $attachments .= ' '; - - $skipspaces = 1; + do_hook("attachment $type0/*", $temp); + /* The API for this hook has changed as of 1.5.2 so that all plugin + arguments are passed in an array instead of each their own plugin + argument, and arguments are passed by reference, so instead of + returning any changes, changes should simply be made to the original + arguments themselves. */ + $temp = array(&$links, &$startMessage, &$id, &$urlMailbox, &$ent, + &$defaultlink, &$display_filename, &$where, &$what); + // Do not let a generic plugin change the default link if a more + // specialized one already did it... + if ($defaultlink != $defaultlink_orig) { + $dummy = ''; + $temp[5] = &$dummy; + } + do_hook("attachment */*", $temp); + + $this_attachment = array(); + $this_attachment['Name'] = decodeHeader($display_filename); + $this_attachment['Description'] = $description; + $this_attachment['DefaultHREF'] = $defaultlink; + $this_attachment['DownloadHREF'] = $links['download link']['href']; + $this_attachment['ViewHREF'] = isset($links['attachment_common']) ? $links['attachment_common']['href'] : ''; + $this_attachment['Size'] = $header->size; + $this_attachment['ContentType'] = sm_encode_html_special_chars($type0 .'/'. $type1); + $this_attachment['OtherLinks'] = array(); foreach ($links as $val) { - if ($skipspaces) { - $skipspaces = 0; - } else { - $attachments .= '  |  '; - } - $attachments .= '' . $val['text'] . ''; + if ($val['text']==_("Download") || $val['text'] == _("View")) + continue; + if (empty($val['text']) && empty($val['extra'])) + continue; + + $temp = array(); + $temp['HREF'] = $val['href']; + $temp['Text'] = (empty($val['text']) ? '' : $val['text']) . (empty($val['extra']) ? '' : $val['extra']); + $this_attachment['OtherLinks'][] = $temp; } + $attachments[] = $this_attachment; + unset($links); - $attachments .= "\n"; } + return $attachments; } -/* This function decodes the body depending on the encoding type. */ -function decodeBody($body, $encoding) { - global $show_html_default; +/** + * Displays attachment links and information + * + * Since 1.3.0 function is not included in formatBody() call. + * + * Since 1.0.2 uses attachment $type0/$type1 hook. + * Since 1.2.5 uses attachment $type0/* hook. + * Since 1.5.0 uses attachments_bottom hook. + * Since 1.5.2 uses templates and does *not* return a value. + * + * @param object $message SquirrelMail message object + * @param array $exclude_id message parts that are not attachments. + * @param string $mailbox mailbox name + * @param integer $id message id + */ +function formatAttachments($message, $exclude_id, $mailbox, $id) { + global $oTemplate; + + $attach = buildAttachmentArray($message, $exclude_id, $mailbox, $id); + + $oTemplate->assign('attachments', $attach); + $oTemplate->display('read_attachments.tpl'); +} + +function sqimap_base64_decode(&$string) { - $body = str_replace("\r\n", "\n", $body); - $encoding = strtolower($encoding); + // Base64 encoded data goes in pairs of 4 bytes. To achieve on the + // fly decoding (to reduce memory usage) you have to check if the + // data has incomplete pairs - if ($encoding == 'quoted-printable' || - $encoding == 'quoted_printable') { - $body = quoted_printable_decode($body); + // Remove the noise in order to check if the 4 bytes pairs are complete + $string = str_replace(array("\r\n","\n", "\r", " "),array('','','',''),$string); - while (ereg("=\n", $body)) { - $body = ereg_replace ("=\n", '', $body); + $sStringRem = ''; + $iMod = strlen($string) % 4; + if ($iMod) { + $sStringRem = substr($string,-$iMod); + // Check if $sStringRem contains padding characters + if (substr($sStringRem,-1) != '=') { + $string = substr($string,0,-$iMod); + } else { + $sStringRem = ''; } + } + $string = base64_decode($string); + return $sStringRem; +} + +/** + * Decodes encoded string (usually message body) + * + * This function decodes a string (usually the message body) + * depending on the encoding type. Currently quoted-printable + * and base64 encodings are supported. + * + * The decode_body hook was added to this function in 1.4.2/1.5.0. + * The $force_crlf parameter was added in 1.5.2. + * + * @param string $string The encoded string + * @param string $encoding used encoding + * @param string $force_crlf Whether or not to force CRLF or LF + * line endings (or to leave as is). + * If given as "LF", line endings will + * all be converted to LF; if "CRLF", + * line endings will all be converted + * to CRLF. If given as an empty value, + * the global $force_crlf_default will + * be consulted (it can be specified in + * config/config_local.php). Otherwise, + * any other value will cause the string + * to be left alone. Note that this will + * be overridden to "LF" if not using at + * least PHP version 4.3.0. (OPTIONAL; + * default is empty - consult global + * default value) + * + * @return string The decoded string + * + * @since 1.0 + * + */ +function decodeBody($string, $encoding, $force_crlf='') { + + global $force_crlf_default; + if (empty($force_crlf)) $force_crlf = $force_crlf_default; + $force_crlf = strtoupper($force_crlf); + + // must force line endings to LF due to broken + // quoted_printable_decode() in PHP versions + // before 4.3.0 (see below) + // + if (!check_php_version(4, 3, 0) || $force_crlf == 'LF') + $string = str_replace("\r\n", "\n", $string); + else if ($force_crlf == 'CRLF') + $string = str_replace("\n", "\r\n", $string); + + $encoding = strtolower($encoding); - } else if ($encoding == 'base64') { - $body = base64_decode($body); + $encoding_handler = do_hook('decode_body', $encoding); + + + // plugins get first shot at decoding the string + // + if (!empty($encoding_handler) && function_exists($encoding_handler)) { + $string = $encoding_handler('decode', $string); + + } elseif ($encoding == 'quoted-printable' || + $encoding == 'quoted_printable') { + + // quoted_printable_decode() function is broken in older + // php versions. Text with \r\n decoding was fixed only + // in php 4.3.0. Minimal code requirement is PHP 4.0.4+ + // and the above call to: str_replace("\r\n", "\n", $string); + // + $string = quoted_printable_decode($string); + + } elseif ($encoding == 'base64') { + $string = base64_decode($string); } // All other encodings are returned raw. - return $body; + return $string; } -/* - * This functions decode strings that is encoded according to +/** + * Decodes headers + * + * This function decodes strings that are encoded according to * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text). * Patched by Christian Schmidt 23/03/2002 + * + * @param string $string header string that has to be made readable + * @param boolean $utfencode change message in order to be readable on user's charset. defaults to true + * @param boolean $htmlsafe preserve spaces and sanitize html special characters. defaults to true + * @param boolean $decide decide if string can be utfencoded. defaults to false + * @return string decoded header string */ -function decodeHeader ($string, $utfencode=true,$htmlsave=true) { - global $languages, $squirrelmail_language; +function decodeHeader ($string, $utfencode=true,$htmlsafe=true,$decide=false) { + global $languages, $squirrelmail_language,$default_charset, $fix_broken_base64_encoded_messages; if (is_array($string)) { $string = implode("\n", $string); } - + if (isset($languages[$squirrelmail_language]['XTRA_CODE']) && - function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) { - $string = $languages[$squirrelmail_language]['XTRA_CODE']('decodeheader', $string); + function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader')) { + $string = call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_decodeheader', $string); // Do we need to return at this point? // return $string; } $i = 0; $iLastMatch = -2; - $encoded = false; + $encoded = true; +// FIXME: spaces are allowed inside quoted-printable encoding, but the following line will bust up any such encoded strings $aString = explode(' ',$string); $ret = ''; foreach ($aString as $chunk) { - if ($encoded && !$chunk) { + if ($encoded && $chunk === '') { continue; - } elseif (!$chunk) { + } elseif ($chunk === '') { $ret .= ' '; continue; } $encoded = false; /* if encoded words are not separated by a linear-space-white we still catch them */ $j = $i-1; + while ($match = preg_match('/^(.*)=\?([^?]*)\?(Q|B)\?([^?]*)\?=(.*)$/Ui',$chunk,$res)) { /* if the last chunk isn't an encoded string then put back the space, otherwise don't */ if ($iLastMatch !== $j) { - if ($htmlsave) { - $ret .= ' '; + if ($htmlsafe) { + $ret .= ' '; } else { $ret .= ' '; } } $iLastMatch = $i; $j = $i; - $ret .= $res[1]; + if ($htmlsafe) { + $ret .= sm_encode_html_special_chars($res[1]); + } else { + $ret .= $res[1]; + } $encoding = ucfirst($res[3]); + + /* decide about valid decoding */ + if ($decide && is_conversion_safe($res[2])) { + $utfencode=true; + $can_be_encoded=true; + } else { + $can_be_encoded=false; + } switch ($encoding) { - case 'B': - $replace = base64_decode($res[4]); - $ret .= charset_decode($res[2],$replace); - break; - case 'Q': - $replace = str_replace('_', ' ', $res[4]); - $replace = preg_replace('/=([0-9a-f]{2})/ie', 'chr(hexdec("\1"))', - $replace); - /* Only encode into entities by default. Some places - * don't need the encoding, like the compose form. - */ - if ($utfencode) { - $replace = charset_decode($res[2], $replace); - } else { - if ($htmlsave) { - $replace = htmlspecialchars($replace); + case 'B': + // fix broken base64-encoded strings (remove end = padding, + // change any = to + in middle of string, add padding back + // to the end) + if ($fix_broken_base64_encoded_messages) { + $encoded_string_minus_padding = strtr(rtrim($res[4], '='), '=', '+'); + $res[4] = str_pad($encoded_string_minus_padding, strlen($res[4]), '='); } - } - $ret .= $replace; - break; - default: - break; + $replace = base64_decode($res[4]); + if ($utfencode) { + if ($can_be_encoded) { + /* convert string to different charset, + * if functions asks for it (usually in compose) + */ + $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe); + } else { + // convert string to html codes in order to display it + $ret .= charset_decode($res[2],$replace); + } + } else { + if ($htmlsafe) { + $replace = sm_encode_html_special_chars($replace); + } + $ret.= $replace; + } + break; + case 'Q': + $replace = str_replace('_', ' ', $res[4]); + $replace = preg_replace_callback('/=([0-9a-f]{2})/i', + create_function ('$matches', 'return chr(hexdec($matches[1]));'), + $replace); + if ($utfencode) { + if ($can_be_encoded) { + /* convert string to different charset, + * if functions asks for it (usually in compose) + */ + $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe); + } else { + // convert string to html codes in order to display it + $replace = charset_decode($res[2], $replace); + } + } else { + if ($htmlsafe) { + $replace = sm_encode_html_special_chars($replace); + } + } + $ret .= $replace; + break; + default: + break; } $chunk = $res[5]; $encoded = true; } if (!$encoded) { - if ($htmlsave) { - $ret .= ' '; + if ($htmlsafe) { + $ret .= ' '; } else { $ret .= ' '; - } + } } - if (!$encoded && $htmlsave) { - $ret .= htmlspecialchars($chunk); + if (!$encoded && $htmlsafe) { + $ret .= sm_encode_html_special_chars($chunk); } else { $ret .= $chunk; } @@ -632,30 +922,53 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true) { } /* remove the first added space */ if ($ret) { - if ($htmlsave) { - $ret = substr($ret,6); + if ($htmlsafe) { + $ret = substr($ret,5); } else { $ret = substr($ret,1); } } - + return $ret; } -/* - * Encode a string according to RFC 1522 for use in headers if it - * contains 8-bit characters or anything that looks like it should - * be encoded. +/** + * Encodes header + * + * Function uses XTRA_CODE _encodeheader function, if such function exists. + * + * Function uses Q encoding by default and encodes a string according to RFC + * 1522 for use in headers if it contains 8-bit characters or anything that + * looks like it should be encoded. + * + * Function switches to B encoding and encodeHeaderBase64() function, if + * string is 8bit and multibyte character set supported by mbstring extension + * is used. It can cause E_USER_NOTICE errors, if interface is used with + * multibyte character set unsupported by mbstring extension. + * + * @param string $string header string, that has to be encoded + * @return string quoted-printable encoded string + * @todo make $mb_charsets system wide constant */ function encodeHeader ($string) { global $default_charset, $languages, $squirrelmail_language; if (isset($languages[$squirrelmail_language]['XTRA_CODE']) && - function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) { - return $languages[$squirrelmail_language]['XTRA_CODE']('encodeheader', $string); + function_exists($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader')) { + return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $string); } - if (strtolower($default_charset) == 'iso-8859-1') { - $string = str_replace("\240",' ',$string); + + // Use B encoding for multibyte charsets + $mb_charsets = array('utf-8','big5','gb2313','euc-kr'); + if (in_array($default_charset,$mb_charsets) && + in_array($default_charset,sq_mb_list_encodings()) && + sq_is8bit($string)) { + return encodeHeaderBase64($string,$default_charset); + } elseif (in_array($default_charset,$mb_charsets) && + sq_is8bit($string) && + ! in_array($default_charset,sq_mb_list_encodings())) { + // Add E_USER_NOTICE error here (can cause 'Cannot add header information' warning in compose.php) + // trigger_error('encodeHeader: Multibyte character set unsupported by mbstring extension.',E_USER_NOTICE); } // Encode only if the string contains 8-bit characters or =? @@ -668,43 +981,19 @@ function encodeHeader ($string) { for($i = 0; $i < $j; ++$i) { switch($string{$i}) { - case '=': - case '<': - case '>': - case ',': - case '?': - case '_': - if ($iEncStart === false) { - $iEncStart = $i; - } - $cur_l+=3; - if ($cur_l > ($max_l-2)) { - /* if there is an stringpart that doesn't need encoding, add it */ - $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset); - $aRet[] = "=?$default_charset?Q?$ret?="; - $iOffset = $i; - $cur_l = 0; - $ret = ''; - $iEncStart = false; - } else { - $ret .= sprintf("=%02X",ord($string{$i})); - } - break; - case '(': - case ')': - if ($iEncStart !== false) { - $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset); - $aRet[] = "=?$default_charset?Q?$ret?="; - $iOffset = $i; - $cur_l = 0; - $ret = ''; - $iEncStart = false; - } - break; - case ' ': - if ($iEncStart !== false) { - $cur_l++; - if ($cur_l > $max_l) { + case '"': + case '=': + case '<': + case '>': + case ',': + case '?': + case '_': + if ($iEncStart === false) { + $iEncStart = $i; + } + $cur_l+=3; + if ($cur_l > ($max_l-2)) { + /* if there is an stringpart that doesn't need encoding, add it */ $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset); $aRet[] = "=?$default_charset?Q?$ret?="; $iOffset = $i; @@ -712,44 +1001,75 @@ function encodeHeader ($string) { $ret = ''; $iEncStart = false; } else { - $ret .= '_'; - } - } - break; - default: - $k = ord($string{$i}); - if ($k > 126) { - if ($iEncStart === false) { - $iEncStart = $i; + $ret .= sprintf("=%02X",ord($string{$i})); } - $cur_l += 3; - /* first we add the encoded string that reached it's max size */ - if ($cur_l > ($max_l-2)) { + break; + case '(': + case ')': + if ($iEncStart !== false) { $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset); - $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */ - $cur_l = 3; - $ret = ''; + $aRet[] = "=?$default_charset?Q?$ret?="; $iOffset = $i; - $iEncStart = $i; + $cur_l = 0; + $ret = ''; + $iEncStart = false; } - $enc_init = true; - $ret .= sprintf("=%02X", $k); - } else { + break; + case ' ': if ($iEncStart !== false) { $cur_l++; if ($cur_l > $max_l) { $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset); $aRet[] = "=?$default_charset?Q?$ret?="; - $iEncStart = false; $iOffset = $i; $cur_l = 0; $ret = ''; + $iEncStart = false; } else { - $ret .= $string{$i}; + $ret .= '_'; } } - } - break; + break; + default: + $k = ord($string{$i}); + if ($k > 126) { + if ($iEncStart === false) { + // do not start encoding in the middle of a string, also take the rest of the word. + $sLeadString = substr($string,0,$i); + $aLeadString = explode(' ',$sLeadString); + $sToBeEncoded = array_pop($aLeadString); + $iEncStart = $i - strlen($sToBeEncoded); + $ret .= $sToBeEncoded; + $cur_l += strlen($sToBeEncoded); + } + $cur_l += 3; + /* first we add the encoded string that reached it's max size */ + if ($cur_l > ($max_l-2)) { + $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset); + $aRet[] = "=?$default_charset?Q?$ret?= "; /* the next part is also encoded => separate by space */ + $cur_l = 3; + $ret = ''; + $iOffset = $i; + $iEncStart = $i; + } + $enc_init = true; + $ret .= sprintf("=%02X", $k); + } else { + if ($iEncStart !== false) { + $cur_l++; + if ($cur_l > $max_l) { + $aRet[] = substr($string,$iOffset,$iEncStart-$iOffset); + $aRet[] = "=?$default_charset?Q?$ret?="; + $iEncStart = false; + $iOffset = $i; + $cur_l = 0; + $ret = ''; + } else { + $ret .= $string{$i}; + } + } + } + break; } } @@ -765,6 +1085,100 @@ function encodeHeader ($string) { return $string; } +/** + * Encodes string according to rfc2047 B encoding header formating rules + * + * It is recommended way to encode headers with character sets that store + * symbols in more than one byte. + * + * Function requires mbstring support. If required mbstring functions are missing, + * function returns false and sets E_USER_WARNING level error message. + * + * Minimal requirements - php 4.0.6 with mbstring extension. Please note, + * that mbstring functions will generate E_WARNING errors, if unsupported + * character set is used. mb_encode_mimeheader function provided by php + * mbstring extension is not used in order to get better control of header + * encoding. + * + * Used php code functions - function_exists(), trigger_error(), strlen() + * (is used with charset names and base64 strings). Used php mbstring + * functions - mb_strlen and mb_substr. + * + * Related documents: rfc 2045 (BASE64 encoding), rfc 2047 (mime header + * encoding), rfc 2822 (header folding) + * + * @param string $string header string that must be encoded + * @param string $charset character set. Must be supported by mbstring extension. + * Use sq_mb_list_encodings() to detect supported charsets. + * @return string string encoded according to rfc2047 B encoding formating rules + * @since 1.5.1 + * @todo First header line can be wrapped to $iMaxLength - $HeaderFieldLength - 1 + * @todo Do we want to control max length of header? + * @todo Do we want to control EOL (end-of-line) marker? + * @todo Do we want to translate error message? + */ +function encodeHeaderBase64($string,$charset) { + /** + * Check mbstring function requirements. + */ + if (! function_exists('mb_strlen') || + ! function_exists('mb_substr')) { + // set E_USER_WARNING + trigger_error('encodeHeaderBase64: Required mbstring functions are missing.',E_USER_WARNING); + // return false + return false; + } + + // initial return array + $aRet = array(); + + /** + * header length = 75 symbols max (same as in encodeHeader) + * remove $charset length + * remove =? ? ?= (5 chars) + * remove 2 more chars (\r\n ?) + */ + $iMaxLength = 75 - strlen($charset) - 7; + + // set first character position + $iStartCharNum = 0; + + // loop through all characters. count characters and not bytes. + for ($iCharNum=1; $iCharNum<=mb_strlen($string,$charset); $iCharNum++) { + // encode string from starting character to current character. + $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset)); + + // Check encoded string length + if(strlen($encoded_string)>$iMaxLength) { + // if string exceeds max length, reduce number of encoded characters and add encoded string part to array + $aRet[] = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum-1,$charset)); + + // set new starting character + $iStartCharNum = $iCharNum-1; + + // encode last char (in case it is last character in string) + $encoded_string = base64_encode(mb_substr($string,$iStartCharNum,$iCharNum-$iStartCharNum,$charset)); + } // if string is shorter than max length - add next character + } + + // add last encoded string to array + $aRet[] = $encoded_string; + + // set initial return string + $sRet = ''; + + // loop through encoded strings + foreach($aRet as $string) { + // TODO: Do we want to control EOL (end-of-line) marker + if ($sRet!='') $sRet.= " "; + + // add header tags and encoded string to return string + $sRet.= '=?'.$charset.'?B?'.$string.'?='; + } + + return $sRet; +} + /* This function trys to locate the entity_id of a specific mime element */ function find_ent_id($id, $message) { for ($i = 0, $ret = ''; $ret == '' && $i < count($message->entities); $i++) { @@ -773,8 +1187,17 @@ function find_ent_id($id, $message) { } else { if (strcasecmp($message->entities[$i]->header->id, $id) == 0) { // if (sq_check_save_extension($message->entities[$i])) { + return $message->entities[$i]->entity_id; +// } + } elseif (!empty($message->entities[$i]->header->parameters['name'])) { + /** + * This is part of a fix for Outlook Express 6.x generating + * cid URLs without creating content-id headers + * @@JA - 20050207 + */ + if (strcasecmp($message->entities[$i]->header->parameters['name'], $id) == 0) { return $message->entities[$i]->entity_id; -// } + } } } } @@ -793,9 +1216,134 @@ function sq_check_save_extension($message) { ** HTMLFILTER ROUTINES */ +/** + * This function checks attribute values for entity-encoded values + * and returns them translated into 8-bit strings so we can run + * checks on them. + * + * @param $attvalue A string to run entity check against. + * @return Nothing, modifies a reference value. + */ +function sq_defang(&$attvalue){ + $me = 'sq_defang'; + /** + * Skip this if there aren't ampersands or backslashes. + */ + if (strpos($attvalue, '&') === false + && strpos($attvalue, '\\') === false){ + return; + } + $m = false; + // before deent, translate the dangerous unicode characters and ... to safe values + // otherwise the regular expressions do not match. + + + + do { + $m = false; + $m = $m || sq_deent($attvalue, '/\�*(\d+);*/s'); + $m = $m || sq_deent($attvalue, '/\�*((\d|[a-f])+);*/si', true); + $m = $m || sq_deent($attvalue, '/\\\\(\d+)/s', true); + } while ($m == true); + $attvalue = stripslashes($attvalue); +} + +/** + * Kill any tabs, newlines, or carriage returns. Our friends the + * makers of the browser with 95% market value decided that it'd + * be funny to make "java[tab]script" be just as good as "javascript". + * + * @param attvalue The attribute value before extraneous spaces removed. + * @return attvalue Nothing, modifies a reference value. + */ +function sq_unspace(&$attvalue){ + $me = 'sq_unspace'; + if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)){ + $attvalue = str_replace(Array("\t", "\r", "\n", "\0", " "), + Array('', '', '', '', ''), $attvalue); + } +} + +/** + * Translate all dangerous Unicode or Shift_JIS characters which are accepted by + * IE as regular characters. + * + * @param attvalue The attribute value before dangerous characters are translated. + * @return attvalue Nothing, modifies a reference value. + * @author Marc Groot Koerkamp. + */ +function sq_fixIE_idiocy(&$attvalue) { + // remove NUL + $attvalue = str_replace("\0", "", $attvalue); + // remove comments + $attvalue = preg_replace("/(\/\*.*?\*\/)/","",$attvalue); + + // IE has the evil habit of accepting every possible value for the attribute expression. + // The table below contains characters which are parsed by IE if they are used in the "expression" + // attribute value. + $aDangerousCharsReplacementTable = array( + array('ʟ', 'ʟ' ,/* L UNICODE IPA Extension */ + 'ʀ', 'ʀ' ,/* R UNICODE IPA Extension */ + 'ɴ', 'ɴ' ,/* N UNICODE IPA Extension */ + 'E', 'E' ,/* Unicode FULLWIDTH LATIN CAPITAL LETTER E */ + 'e', 'e' ,/* Unicode FULLWIDTH LATIN SMALL LETTER E */ + 'X', 'X',/* Unicode FULLWIDTH LATIN CAPITAL LETTER X */ + 'x', 'x',/* Unicode FULLWIDTH LATIN SMALL LETTER X */ + 'P', 'P',/* Unicode FULLWIDTH LATIN CAPITAL LETTER P */ + 'p', 'p',/* Unicode FULLWIDTH LATIN SMALL LETTER P */ + 'R', 'R',/* Unicode FULLWIDTH LATIN CAPITAL LETTER R */ + 'r', 'r',/* Unicode FULLWIDTH LATIN SMALL LETTER R */ + 'S', 'S',/* Unicode FULLWIDTH LATIN CAPITAL LETTER S */ + 's', 's',/* Unicode FULLWIDTH LATIN SMALL LETTER S */ + 'I', 'I',/* Unicode FULLWIDTH LATIN CAPITAL LETTER I */ + 'i', 'i',/* Unicode FULLWIDTH LATIN SMALL LETTER I */ + 'O', 'O',/* Unicode FULLWIDTH LATIN CAPITAL LETTER O */ + 'o', 'o',/* Unicode FULLWIDTH LATIN SMALL LETTER O */ + 'N', 'N',/* Unicode FULLWIDTH LATIN CAPITAL LETTER N */ + 'n', 'n',/* Unicode FULLWIDTH LATIN SMALL LETTER N */ + 'L', 'L',/* Unicode FULLWIDTH LATIN CAPITAL LETTER L */ + 'l', 'l',/* Unicode FULLWIDTH LATIN SMALL LETTER L */ + 'U', 'U',/* Unicode FULLWIDTH LATIN CAPITAL LETTER U */ + 'u', 'u',/* Unicode FULLWIDTH LATIN SMALL LETTER U */ + 'ⁿ', 'ⁿ' ,/* Unicode SUPERSCRIPT LATIN SMALL LETTER N */ + "\xEF\xBC\xA5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER E */ // in unicode this is some Chinese char range + "\xEF\xBD\x85", /* Shift JIS FULLWIDTH LATIN SMALL LETTER E */ + "\xEF\xBC\xB8", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER X */ + "\xEF\xBD\x98", /* Shift JIS FULLWIDTH LATIN SMALL LETTER X */ + "\xEF\xBC\xB0", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER P */ + "\xEF\xBD\x90", /* Shift JIS FULLWIDTH LATIN SMALL LETTER P */ + "\xEF\xBC\xB2", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER R */ + "\xEF\xBD\x92", /* Shift JIS FULLWIDTH LATIN SMALL LETTER R */ + "\xEF\xBC\xB3", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER S */ + "\xEF\xBD\x93", /* Shift JIS FULLWIDTH LATIN SMALL LETTER S */ + "\xEF\xBC\xA9", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER I */ + "\xEF\xBD\x89", /* Shift JIS FULLWIDTH LATIN SMALL LETTER I */ + "\xEF\xBC\xAF", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER O */ + "\xEF\xBD\x8F", /* Shift JIS FULLWIDTH LATIN SMALL LETTER O */ + "\xEF\xBC\xAE", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER N */ + "\xEF\xBD\x8E", /* Shift JIS FULLWIDTH LATIN SMALL LETTER N */ + "\xEF\xBC\xAC", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER L */ + "\xEF\xBD\x8C", /* Shift JIS FULLWIDTH LATIN SMALL LETTER L */ + "\xEF\xBC\xB5", /* Shift JIS FULLWIDTH LATIN CAPITAL LETTER U */ + "\xEF\xBD\x95", /* Shift JIS FULLWIDTH LATIN SMALL LETTER U */ + "\xE2\x81\xBF", /* Shift JIS FULLWIDTH SUPERSCRIPT N */ + "\xCA\x9F", /* L UNICODE IPA Extension */ + "\xCA\x80", /* R UNICODE IPA Extension */ + "\xC9\xB4"), /* N UNICODE IPA Extension */ + array('l', 'l', 'r','r','n','n', + 'E','E','e','e','X','X','x','x','P','P','p','p','R','R','r','r','S','S','s','s','I','I', + 'i','i','O','O','o','o','N','N','n','n','L','L','l','l','U','U','u','u','n','n', + 'E','e','X','x','P','p','R','r','S','s','I','i','O','o','N','n','L','l','U','u','n','l','r','n')); + $attvalue = str_replace($aDangerousCharsReplacementTable[0],$aDangerousCharsReplacementTable[1],$attvalue); + + // Escapes are useful for special characters like "{}[]()'&. In other cases they are + // used for XSS. + $attvalue = preg_replace("/(\\\\)([a-zA-Z]{1})/",'$2',$attvalue); +} + /** * This function returns the final tag out of the tag name, an array - * of attributes, and the type of the tag. This function is called by + * of attributes, and the type of the tag. This function is called by * sq_sanitize internally. * * @param $tagname the name of the tag. @@ -839,7 +1387,7 @@ function sq_casenormalize(&$val){ /** * This function skips any whitespace from the current position within * a string and to the next non-whitespace value. - * + * * @param $body the string * @param $offset the offset within the string where we should start * looking for the next non-whitespace character. @@ -849,9 +1397,8 @@ function sq_casenormalize(&$val){ function sq_skipspace($body, $offset){ $me = 'sq_skipspace'; preg_match('/^(\s*)/s', substr($body, $offset), $matches); - if (sizeof($matches{1})){ - $count = strlen($matches{1}); - $offset += $count; + if (!empty($matches[1])){ + $offset += strlen($matches[1]); } return $offset; } @@ -893,8 +1440,8 @@ function sq_findnxreg($body, $offset, $reg){ $me = 'sq_findnxreg'; $matches = Array(); $retarr = Array(); - preg_match("%^(.*?)($reg)%s", substr($body, $offset), $matches); - if (!$matches{0}){ + preg_match("%^(.*?)($reg)%si", substr($body, $offset), $matches); + if (!isset($matches{0}) || !$matches{0}){ $retarr = false; } else { $retarr{0} = $offset + strlen($matches{1}); @@ -943,7 +1490,7 @@ function sq_getnxtag($body, $offset){ * 2. Closing tag, e.g.: * * 3. XHTML-style content-less tag, e.g.: - * + * */ $tagtype = false; switch (substr($body, $pos, 1)){ @@ -994,14 +1541,14 @@ function sq_getnxtag($body, $offset){ * '>' indicating the end of the tag entirely. * '\s' indicating the end of the tag name. * '/' indicating that this is type-3 xhtml tag. - * + * * Whatever else we find there indicates an invalid tag. */ switch ($match){ case '/': /** * This is an xhtml-style tag with a closing / at the - * end, like so: . Check if it's followed + * end, like so: . Check if it's followed * by the closing bracket. If not, then this tag is invalid */ if (substr($body, $pos, 2) == "/>"){ @@ -1023,7 +1570,7 @@ function sq_getnxtag($body, $offset){ /** * This is an invalid tag! Look for the next closing ">". */ - $gt = sq_findnxstr($body, $offset, ">"); + $gt = sq_findnxstr($body, $lt, ">"); return Array(false, false, false, $lt, $gt); } break; @@ -1082,7 +1629,7 @@ function sq_getnxtag($body, $offset){ * double quotes. Type 4 we convert into: * attrname="yes". */ - $regary = sq_findnxreg($body, $pos, "[^\w\-_]"); + $regary = sq_findnxreg($body, $pos, "[^:\w\-_]"); if ($regary == false){ /** * Looks like body ended before the end of tag. @@ -1103,7 +1650,7 @@ function sq_getnxtag($body, $offset){ case '/': /** * This is an xhtml-style tag with a closing / at the - * end, like so: . Check if it's followed + * end, like so: . Check if it's followed * by the closing bracket. If not, then this tag is invalid */ if (substr($body, $pos, 2) == "/>"){ @@ -1196,50 +1743,33 @@ function sq_getnxtag($body, $offset){ } /** - * This function checks attribute values for entity-encoded values - * and returns them translated into 8-bit strings so we can run - * checks on them. + * Translates entities into literal values so they can be checked. * - * @param $attvalue A string to run entity check against. - * @return Translated value. + * @param $attvalue the by-ref value to check. + * @param $regex the regular expression to check against. + * @param $hex whether the entites are hexadecimal. + * @return True or False depending on whether there were matches. */ -function sq_deent($attvalue){ +function sq_deent(&$attvalue, $regex, $hex=false){ $me = 'sq_deent'; - /** - * See if we have to run the checks first. All entities must start - * with "&". - */ - if (strpos($attvalue, "&") === false){ - return $attvalue; - } - /** - * Check named entities first. - */ - $trans = get_html_translation_table(HTML_ENTITIES); - /** - * Leave " in, as it can mess us up. - */ - $trans = array_flip($trans); - unset($trans{"""}); - while (list($ent, $val) = each($trans)){ - $attvalue = preg_replace("/$ent*(\W)/si", "$val\\1", $attvalue); - } - /** - * Now translate numbered entities from 1 to 255 if needed. - */ - if (strpos($attvalue, "#") !== false){ - $omit = Array(34, 39); - for ($asc=1; $asc<256; $asc++){ - if (!in_array($asc, $omit)){ - $chr = chr($asc); - $attvalue = preg_replace("/\�*$asc;*(\D)/si", "$chr\\1", - $attvalue); - $attvalue = preg_replace("/\�*".dechex($asc).";*(\W)/si", - "$chr\\1", $attvalue); + $ret_match = false; + // remove comments + //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue); + preg_match_all($regex, $attvalue, $matches); + if (is_array($matches) && sizeof($matches[0]) > 0){ + $repl = Array(); + for ($i = 0; $i < sizeof($matches[0]); $i++){ + $numval = $matches[1][$i]; + if ($hex){ + $numval = hexdec($numval); } + $repl{$matches[0][$i]} = chr($numval); } + $attvalue = strtr($attvalue, $repl); + return true; + } else { + return false; } - return $attvalue; } /** @@ -1254,8 +1784,8 @@ function sq_deent($attvalue){ * @param $id message id * @return Array with modified attributes. */ -function sq_fixatts($tagname, - $attary, +function sq_fixatts($tagname, + $attary, $rm_attnames, $bad_attvals, $add_attr_to_tag, @@ -1279,9 +1809,23 @@ function sq_fixatts($tagname, } } /** - * Remove any entities. + * Workaround for IE quirks + */ + sq_fixIE_idiocy($attvalue); + + /** + * Remove any backslashes, entities, and extraneous whitespace. */ - $attvalue = sq_deent($attvalue); + + $oldattvalue = $attvalue; + sq_defang($attvalue); + if ($attname == 'style' && $attvalue !== $oldattvalue) { + // entities are used in the attribute value. In 99% of the cases it's there as XSS + // i.e.
+ $attvalue = "idiocy"; + $attary{$attname} = $attvalue; + } + sq_unspace($attvalue); /** * Now let's run checks on the attvalues. @@ -1299,20 +1843,39 @@ function sq_fixatts($tagname, * Second one is replacements */ list($valmatch, $valrepl) = $valary; - $newvalue = + $newvalue = preg_replace($valmatch, $valrepl, $attvalue); if ($newvalue != $attvalue){ $attary{$attname} = $newvalue; + $attvalue = $newvalue; } } } } } + if ($attname == 'style') { + if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) { + // 8bit and control characters in style attribute values can be used for XSS, remove them + $attary{$attname} = '"disallowed character"'; + } + preg_match_all("/url\s*\((.+)\)/si",$attvalue,$aMatch); + if (count($aMatch)) { + foreach($aMatch[1] as $sMatch) { + // url value + $urlvalue = $sMatch; + sq_fix_url($attname, $urlvalue, $message, $id, $mailbox,"'"); + $attary{$attname} = str_replace($sMatch,$urlvalue,$attvalue); + } + } + } /** - * Turn cid: urls into http-friendly ones. + * Use white list based filtering on attributes which can contain url's */ - if (preg_match("/^[\'\"]\s*cid:/si", $attvalue)){ - $attary{$attname} = sq_cid2http($message, $id, $attvalue, $mailbox); + else if ($attname == 'href' || $attname == 'xlink:href' || $attname == 'src' + || $attname == 'poster' || $attname == 'formaction' + || $attname == 'background' || $attname == 'action') { + sq_fix_url($attname, $attvalue, $message, $id, $mailbox); + $attary{$attname} = $attvalue; } } /** @@ -1326,18 +1889,235 @@ function sq_fixatts($tagname, return $attary; } +/** + * This function filters url's + * + * @param $attvalue String with attribute value to filter + * @param $message message object + * @param $id message id + * @param $mailbox mailbox + * @param $sQuote quoting characters around url's + */ +function sq_fix_url($attname, &$attvalue, $message, $id, $mailbox,$sQuote = '"') { + $attvalue = trim($attvalue); + if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) { + // remove the double quotes + $sQuote = $attvalue[0]; + $attvalue = trim(substr($attvalue,1,-1)); + } + + // If there's no "view_unsafe_images" variable in the URL, turn unsafe + // images off by default. + sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET, FALSE); + + global $use_transparent_security_image; + if ($use_transparent_security_image) $secremoveimg = '../images/spacer.png'; + else $secremoveimg = '../images/' . _("sec_remove_eng.png"); + + /** + * Replace empty src tags with the blank image. src is only used + * for frames, images, and image inputs. Doing a replace should + * not affect them working as should be, however it will stop + * IE from being kicked off when src for img tags are not set + */ + if ($attvalue == '') { + $attvalue = '"' . SM_PATH . 'images/blank.png"'; + } else { + // first, disallow 8 bit characters and control characters + if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) { + switch ($attname) { + case 'href': + $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote; + break; + default: + $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote; + break; + } + } else { + $aUrl = parse_url($attvalue); + if (isset($aUrl['scheme'])) { + switch(strtolower($aUrl['scheme'])) { + case 'mailto': + case 'http': + case 'https': + case 'ftp': + if ($attname != 'href') { + if ($view_unsafe_images == false) { + $attvalue = $sQuote . $secremoveimg . $sQuote; + } else { + if (isset($aUrl['path'])) { + + // No one has been able to show that image URIs + // can be exploited, so for now, no restrictions + // are made at all. If this proves to be a problem, + // the commented-out code below can be of help. + // (One consideration is that I see nothing in this + // function that specifically says that we will + // only ever arrive here when inspecting an image + // tag, although that does seem to be the end + // result - e.g.,