X-Git-Url: https://vcs.fsf.org/?p=squirrelmail.git;a=blobdiff_plain;f=functions%2Fmime.php;h=002e78892f76890a381be629e00c9ed92698d808;hp=bed523a7d1c3cb1782dbeb8fe034dd1ef67fca73;hb=55521a0717830d38714bdc4177fc31df8089f8dc;hpb=23a8174a694766290509e1a7ffc0aa4d64b6ecb2 diff --git a/functions/mime.php b/functions/mime.php index bed523a7..002e7889 100644 --- a/functions/mime.php +++ b/functions/mime.php @@ -6,17 +6,38 @@ * This contains the functions necessary to detect and decode MIME * messages. * - * @copyright © 1999-2006 The SquirrelMail Project Team + * @copyright 1999-2020 The SquirrelMail Project Team * @license http://opensource.org/licenses/gpl-license.php GNU Public License * @version $Id$ * @package squirrelmail */ -/** The typical includes... */ -require_once(SM_PATH . 'functions/imap.php'); -require_once(SM_PATH . 'functions/attachment_common.php'); -/** add sqm_baseuri()*/ -include_once(SM_PATH . 'functions/display_messages.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 */ @@ -36,20 +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 ); $errormessage = _("SquirrelMail could not decode the bodystructure of the message"); $errormessage .= '
'._("The bodystructure provided by your IMAP server:").'

'; - $errormessage .= '
' . htmlspecialchars($read) . '
'; - plain_error_message( $errormessage, $color ); + $errormessage .= '
' . sm_encode_html_special_chars($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 ($char) { case 'S': @@ -71,6 +92,9 @@ function mime_structure ($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) == '$mdnsent') { @@ -113,10 +137,16 @@ function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) { $data = sqimap_run_command ($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, $regs[1]); /* There is some information in the content info header that could be important * in order to parse html messages. Let's get them here. @@ -124,8 +154,10 @@ function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) { // if ($ret{0} == '<') { // $data = sqimap_run_command ($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=' . $passed_id; @@ -158,7 +190,7 @@ function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) { return $ret; } -function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout') { +function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStream='php://stdout', $force_crlf='') { /* Don't kill the connection if the browser is over a dialup * and it would take over 30 seconds to download it. @@ -180,9 +212,9 @@ function mime_print_body_lines ($imap_stream, $id, $ent_id=1, $encoding, $rStrea } else { $body = mime_fetch_body ($imap_stream, $id, $ent_id); if (is_resource($rStream)) { - fputs($rStream,decodeBody($body,$encoding)); + fputs($rStream,decodeBody($body, $encoding, $force_crlf)); } else { - echo decodeBody($body, $encoding); + echo decodeBody($body, $encoding, $force_crlf); } } @@ -264,11 +296,12 @@ function translateText(&$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, $charset); } @@ -293,15 +326,9 @@ function translateText(&$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] = $line; @@ -325,24 +352,25 @@ function translateText(&$body, $wrap_at, $charset) { * @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 - * @param boolean $clean (since 1.5.1) Do not output stuff that's irrelevant for the printable version. * @return string html formated message text */ -function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $mailbox='INBOX', $clean=FALSE) { +function formatBody($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, $languages, $squirrelmail_language, $show_html_default, $sort, $has_unsafe_images, $passed_ent_id, - $use_iframe, $iframe_height, $download_and_unsafe_link; + $use_iframe, $iframe_height, $download_and_unsafe_link, + $download_href, $unsafe_image_toggle_href, $unsafe_image_toggle_text, + $oTemplate, $nbsp; // workaround for not updated config.php if (! isset($use_iframe)) $use_iframe = false; - if( !sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET) ) { - $view_unsafe_images = 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); @@ -358,8 +386,9 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma $body = call_user_func($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', $body); /* If there are other types that shouldn't be formatted, add * them here. @@ -381,9 +410,7 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma $body = trim($body); translateText($body, $wrap_at, $body_message->header->getParameter('charset')); - } elseif ($use_iframe && ! $clean) { - // $clean is used to remove iframe in printable view. - + } elseif ($use_iframe) { /** * If we don't add html message between iframe tags, * we must detect unsafe images and modify $has_unsafe_images. @@ -399,63 +426,61 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma . '&ent_id=' . $ent_num . '&view_unsafe_images=' . (int) $view_unsafe_images; - // adding warning message - $body = html_tag('div',_("Viewing HTML formatted email"),'center'); - - /** - * height can't be set to 100%, because it does not work as expected when - * iframe is inside the table. Browsers do not create full height objects - * even when iframe is not nested. Maybe there is some way to get full size - * with CSS. Tested in firefox 1.02 and opera 7.53 - * - * width="100%" does not work as expected, when table width is not set (automatic) - * - * tokul: I think \n"; + $body = $oTemplate->fetch('read_html_iframe.tpl'); } else { // old way of html rendering - $body = magicHTML($body, $id, $message, $mailbox); /** * 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. */ - $body = charset_decode($body_message->header->getParameter('charset'),$body,false,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')); } - // if this is the clean display (i.e. printer friendly), stop here. - if ( $clean ) { - return $body; - } + /* + * 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_and_unsafe_link .= ' | ' . _("Download this as a file") . ''; + $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) { $text = _("Hide Unsafe Images"); } else { @@ -466,50 +491,52 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma $text = ''; } } + + // Only create a link for unsafe images if there's need for one. If so: + // add it to the global variable. if($text != '') { - $download_and_unsafe_link .= ' | ' . $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; } /** - * 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. + * 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 $id message id - * @return string html formated attachment information. */ -function formatAttachments($message, $exclude_id, $mailbox, $id) { - global $where, $what, $startMessage, $color, $passed_ent_id; +function buildAttachmentArray($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 = array(); $links['download link']['text'] = _("Download"); - $links['download link']['href'] = sqm_baseuri() . + $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 = sqm_baseuri() . 'src/read_body.php'; + $default_page = $base_uri . 'src/read_body.php'; $rfc822_header = $att->rfc822_header; $filename = $rfc822_header->subject; if (trim( $filename ) == '') { @@ -528,9 +555,9 @@ function formatAttachments($message, $exclude_id, $mailbox, $id) { } else { $from_name = _("Unknown sender"); } - $description = $from_name; + $description = _("From").': '.$from_name; } else { - $default_page = sqm_baseuri() . 'src/download.php'; + $default_page = $base_uri . 'src/download.php'; $filename = $att->getFilename(); if ($header->description) { $description = decodeHeader($header->description); @@ -551,53 +578,109 @@ function formatAttachments($message, $exclude_id, $mailbox, $id) { if ($where && $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($header->size) . - '  ' . - '[ '.htmlspecialchars($type0).'/'.htmlspecialchars($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 .= '  |  '; - } - $attachments .= '' - . (isset($val['text']) && !empty($val['text']) ? $val['text'] : '') - . (isset($val['extra']) && !empty($val['extra']) ? $val['extra'] : '') - . ''; + if ($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"; } - $attachmentadd = do_hook_function('attachments_bottom',$attachments); - if ($attachmentadd != '') - $attachments = $attachmentadd; + return $attachments; } +/** + * 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) { // Base64 encoded data goes in pairs of 4 bytes. To achieve on the @@ -623,61 +706,96 @@ function sqimap_base64_decode(&$string) { } /** - * Decodes encoded message body + * 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 * - * This function decodes the body depending on the encoding type. - * Currently quoted-printable and base64 encodings are supported. - * decode_body hook was added to this function in 1.4.2/1.5.0 - * @param string $body encoded message body - * @param string $encoding used encoding - * @return string decoded string * @since 1.0 + * */ -function decodeBody($body, $encoding) { +function decodeBody($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); - $body = str_replace("\r\n", "\n", $body); $encoding = strtolower($encoding); - $encoding_handler = do_hook_function('decode_body', $encoding); + $encoding_handler = do_hook('decode_body', $encoding); - // plugins get first shot at decoding the body + // plugins get first shot at decoding the string // if (!empty($encoding_handler) && function_exists($encoding_handler)) { - $body = $encoding_handler('decode', $body); + $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 4.0.4 + - * str_replace("\r\n", "\n", $body); call. - */ - $body = quoted_printable_decode($body); + + // 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') { - $body = base64_decode($body); + $string = base64_decode($string); } // All other encodings are returned raw. - return $body; + return $string; } /** * Decodes headers * - * This functions decode strings that is encoded according to + * 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 $htmlsave preserve spaces and sanitize html special characters. 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,$decide=false) { - global $languages, $squirrelmail_language,$default_charset; +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", $string); } @@ -692,6 +810,7 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { $iLastMatch = -2; $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 $chunk) { @@ -708,7 +827,7 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { while ($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) { + if ($htmlsafe) { $ret .= ' '; } else { $ret .= ' '; @@ -716,8 +835,8 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { } $iLastMatch = $i; $j = $i; - if ($htmlsave) { - $ret .= htmlspecialchars($res[1]); + if ($htmlsafe) { + $ret .= sm_encode_html_special_chars($res[1]); } else { $ret .= $res[1]; } @@ -733,41 +852,49 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { switch ($encoding) { 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]), '='); + } $replace = base64_decode($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); + $ret .= charset_convert($res[2],$replace,$default_charset,$htmlsafe); } else { // convert string to html codes in order to display it $ret .= charset_decode($res[2],$replace); } } else { - if ($htmlsave) { - $replace = htmlspecialchars($replace); + if ($htmlsafe) { + $replace = sm_encode_html_special_chars($replace); } $ret.= $replace; } break; case 'Q': $replace = str_replace('_', ' ', $res[4]); - $replace = preg_replace('/=([0-9a-f]{2})/ie', 'chr(hexdec("\1"))', + $replace = preg_replace_callback('/=([0-9a-f]{2})/i', + create_function ('$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); + $replace = charset_convert($res[2], $replace,$default_charset,$htmlsafe); } else { // convert string to html codes in order to display it $replace = charset_decode($res[2], $replace); } } else { - if ($htmlsave) { - $replace = htmlspecialchars($replace); + if ($htmlsafe) { + $replace = sm_encode_html_special_chars($replace); } } $ret .= $replace; @@ -779,15 +906,15 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { $encoded = true; } if (!$encoded) { - if ($htmlsave) { + if ($htmlsafe) { $ret .= ' '; } else { $ret .= ' '; } } - if (!$encoded && $htmlsave) { - $ret .= htmlspecialchars($chunk); + if (!$encoded && $htmlsafe) { + $ret .= sm_encode_html_special_chars($chunk); } else { $ret .= $chunk; } @@ -795,7 +922,7 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { } /* remove the first added space */ if ($ret) { - if ($htmlsave) { + if ($htmlsafe) { $ret = substr($ret,5); } else { $ret = substr($ret,1); @@ -854,6 +981,7 @@ function encodeHeader ($string) { for($i = 0; $i < $j; ++$i) { switch($string{$i}) { + case '"': case '=': case '<': case '>': @@ -1106,6 +1234,11 @@ function sq_defang(&$attvalue){ 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'); @@ -1131,6 +1264,83 @@ function sq_unspace(&$attvalue){ } } +/** + * Translate all dangerous Unicode or Shift_JIS characters which are accepted by + * IE as regular characters. + * + * @param attvalue The attribute value before dangerous characters are translated. + * @return attvalue Nothing, modifies a reference value. + * @author Marc Groot Koerkamp. + */ +function sq_fixIE_idiocy(&$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 @@ -1187,9 +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 $offset; } @@ -1544,6 +1753,8 @@ function sq_getnxtag($body, $offset){ function sq_deent(&$attvalue, $regex, $hex=false){ $me = 'sq_deent'; $ret_match = false; + // remove comments + //$attvalue = preg_replace("/(\/\*.*\*\/)/","",$attvalue); preg_match_all($regex, $attvalue, $matches); if (is_array($matches) && sizeof($matches[0]) > 0){ $repl = Array(); @@ -1597,10 +1808,23 @@ function sq_fixatts($tagname, } } } + /** + * Workaround for IE quirks + */ + sq_fixIE_idiocy($attvalue); + /** * Remove any backslashes, entities, and extraneous whitespace. */ + + $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); /** @@ -1623,38 +1847,36 @@ function sq_fixatts($tagname, preg_replace($valmatch, $valrepl, $attvalue); if ($newvalue != $attvalue){ $attary{$attname} = $newvalue; + $attvalue = $newvalue; } } } } } - - /** - * Replace empty src tags with the blank image. src is only used - * for frames, images, and image inputs. Doing a replace should - * not affect them working as should be, however it will stop - * IE from being kicked off when src for img tags are not set - */ - if (($attname == 'src') && ($attvalue == '""')) { - $attary{$attname} = '"' . SM_PATH . 'images/blank.png"'; - } - - /** - * Turn cid: urls into http-friendly ones. - */ - if (preg_match("/^[\'\"]\s*cid:/si", $attvalue)){ - $attary{$attname} = sq_cid2http($message, $id, $attvalue, $mailbox); + if ($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); + } + } } - /** - * "Hack" fix for Outlook using propriatary outbind:// protocol in img tags. - * One day MS might actually make it match something useful, for now, falling - * back to using cid2http, so we can grab the blank.png. + * Use white list based filtering on attributes which can contain url's */ - if (preg_match("/^[\'\"]\s*outbind:\/\//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} = $attvalue; } - } /** * See if we need to append any attributes to this tag. @@ -1667,6 +1889,159 @@ function sq_fixatts($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.,