X-Git-Url: https://vcs.fsf.org/?p=squirrelmail.git;a=blobdiff_plain;f=functions%2Fmime.php;h=c151ecf6f5cb2d6d30c52a7426fad33f3827ccc0;hp=98acbd543b9e13e02507ed7c7a2ef59daad0ce10;hb=515547081d52328045a174ded166ef2c3cbf55f5;hpb=bc017c1d6c07ad907a5418bcbc1fb585311bc6c7 diff --git a/functions/mime.php b/functions/mime.php index 98acbd54..c151ecf6 100644 --- a/functions/mime.php +++ b/functions/mime.php @@ -3,21 +3,41 @@ /** * mime.php * - * Copyright (c) 1999-2005 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. * + * @copyright © 1999-2007 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 */ @@ -37,20 +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 ); + 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 ($char) { case 'S': @@ -127,6 +147,8 @@ function mime_fetch_body($imap_stream, $id, $ent_id=1, $fetch_size=0) { // } } else if (ereg('"([^"]*)"', $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=' . $passed_id; @@ -265,7 +287,7 @@ function translateText(&$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); $i++) { @@ -294,15 +316,9 @@ function translateText(&$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] = $line; @@ -335,7 +351,10 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma * order that is their priority. */ global $startMessage, $languages, $squirrelmail_language, - $show_html_default, $sort, $has_unsafe_images, $passed_ent_id, $use_iframe,$iframe_height; + $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, $nbsp; // workaround for not updated config.php if (! isset($use_iframe)) $use_iframe = false; @@ -358,8 +377,9 @@ function formatBody($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', $body); /* If there are other types that shouldn't be formatted, add * them here. @@ -388,7 +408,9 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma * 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); + $html_body = magicHTML($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 $iframeurl=sqm_baseuri().'src/view_html.php?' @@ -397,35 +419,23 @@ function formatBody($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'); + global $oTemplate; + $oTemplate->assign('iframe_url', $iframeurl); + $oTemplate->assign('iframe_height', $iframe_height); + $oTemplate->assign('html_body', $html_body); - /** - * 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 + /** + * 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, $mailbox); } } else { @@ -438,14 +448,17 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma return $body; } + $download_and_unsafe_link = ''; + $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; } - $body .= '
' . _("Download this as a file") . ''; + $download_href = SM_PATH . 'src/download.php?absolute_dl=true&' . $link; + $download_and_unsafe_link .= "$nbsp|$nbsp" + . create_hyperlink($download_href, _("Download this as a file")); if ($view_unsafe_images) { $text = _("Hide Unsafe Images"); } else { @@ -457,50 +470,46 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma } } if($text != '') { - $body .= ' | ' . $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); } - $body .= '

' . "\n"; } 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. Separated from + * formatAttachments() below so that the same array can be given to the + * print-friendly version. * + * @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; $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); $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( $filename ) == '') { @@ -509,12 +518,19 @@ function formatAttachments($message, $exclude_id, $mailbox, $id) { $from_o = $rfc822_header->from; if (is_object($from_o)) { $from_name = decodeHeader($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"); } - $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); @@ -535,53 +551,105 @@ 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. + * 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", $temp); + if(count($links) <= 1 && $defaultlink == $defaultlink_orig) { + /* 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/*", $temp); } - - $links = $hookresults[1]; - $defaultlink = $hookresults[6]; - - $attachments .= '' . - ''.decodeHeader($display_filename).' ' . - '' . show_readable_size($header->size) . - '  ' . - '[ '.htmlspecialchars($type0).'/'.htmlspecialchars($type1).' ] ' . - ''; - $attachments .= '' . $description . ''; - $attachments .= ' '; - - $skipspaces = 1; + /* 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; + $this_attachment['ContentType'] = htmlspecialchars($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 @@ -606,15 +674,23 @@ function sqimap_base64_decode(&$string) { return $sStringRem; } - -/* This function decodes the body depending on the encoding type. */ +/** + * Decodes encoded message body + * + * 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) { - global $show_html_default; $body = str_replace("\r\n", "\n", $body); $encoding = strtolower($encoding); - $encoding_handler = do_hook_function('decode_body', $encoding); + $encoding_handler = do_hook('decode_body', $encoding); // plugins get first shot at decoding the body @@ -622,15 +698,16 @@ function decodeBody($body, $encoding) { if (!empty($encoding_handler) && function_exists($encoding_handler)) { $body = $encoding_handler('decode', $body); - } else if ($encoding == 'quoted-printable' || + } 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); - - while (ereg("=\n", $body)) { - $body = ereg_replace ("=\n", '', $body); - } - - } else if ($encoding == 'base64') { + } elseif ($encoding == 'base64') { $body = base64_decode($body); } @@ -641,7 +718,7 @@ function decodeBody($body, $encoding) { /** * Decodes headers * - * This functions decode strings that is encoded according to + * This function decodes strings that are encoded according to * RFC1522 (MIME Part Two: Message Header Extensions for Non-ASCII Text). * Patched by Christian Schmidt 23/03/2002 * @@ -714,7 +791,7 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { /* 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,$htmlsave); } else { // convert string to html codes in order to display it $ret .= charset_decode($res[2],$replace); @@ -735,7 +812,7 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { /* 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,$htmlsave); } else { // convert string to html codes in order to display it $replace = charset_decode($res[2], $replace); @@ -781,14 +858,22 @@ function decodeHeader ($string, $utfencode=true,$htmlsave=true,$decide=false) { } /** - * Encodes header as quoted-printable + * Encodes header + * + * Function uses XTRA_CODE _encodeheader function, if such function exists. * - * 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. + * 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, $squirrelmail_language; @@ -798,6 +883,19 @@ function encodeHeader ($string) { return call_user_func($languages[$squirrelmail_language]['XTRA_CODE'] . '_encodeheader', $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 =? $j = strlen($string); $max_l = 75 - strlen($default_charset) - 7; @@ -911,6 +1009,100 @@ 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); $i++) { @@ -966,6 +1158,11 @@ function sq_defang(&$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'); @@ -991,6 +1188,83 @@ 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 @@ -1404,6 +1678,8 @@ 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){ $repl = Array(); @@ -1457,10 +1733,23 @@ function sq_fixatts($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); /** @@ -1483,38 +1772,34 @@ 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 == 'src' || $attname == 'background') { + sq_fix_url($attname, $attvalue, $message, $id, $mailbox); + $attary{$attname} = $attvalue; } - } /** * See if we need to append any attributes to this tag. @@ -1527,6 +1812,99 @@ function sq_fixatts($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( !sqgetGlobalVar('view_unsafe_images', $view_unsafe_images, SQ_GET) ) { + $view_unsafe_images = false; + } + $secremoveimg = '../images/' . _("sec_remove_eng.png"); + + /** + * Replace empty src tags with the blank image. src is only used + * for frames, images, and image inputs. Doing a replace should + * not affect them working as should be, however it will stop + * IE from being kicked off when src for img tags are not set + */ + if ($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'])) { + // validate image extension. + $ext = strtolower(substr($aUrl['path'],strrpos($aUrl['path'],'.'))); + if (!in_array($ext,array('.jpeg','.jpg','xjpeg','.gif','.bmp','.jpe','.png','.xbm'))) { + $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote; + } + } else { + $attvalue = $sQuote . SM_PATH . 'images/blank.png'. $sQuote; + } + } + } + break; + case 'outbind': + /** + * "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. + */ + $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox); + break; + case 'cid': + /** + * Turn cid: urls into http-friendly ones. + */ + $attvalue = sq_cid2http($message, $id, $attvalue, $mailbox); + break; + default: + $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote; + break; + } + } else { + if (!(isset($aUrl['path']) && $aUrl['path'] == $secremoveimg)) { + // parse_url did not lead to satisfying result + $attvalue = $sQuote . SM_PATH . 'images/blank.png' . $sQuote; + } + } + } + } +} + /** * This function edits the style definition to make them friendly and * usable in SquirrelMail. @@ -1538,14 +1916,71 @@ function sq_fixatts($tagname, * @return a string with edited content. */ function sq_fixstyle($body, $pos, $message, $id, $mailbox){ - global $view_unsafe_images; $me = 'sq_fixstyle'; - $ret = sq_findnxreg($body, $pos, ''); - if ($ret == FALSE){ + // workaround for in between comments + $iCurrentPos = $pos; + $content = ''; + $sToken = ''; + $bSucces = false; + $bEndTag = false; + for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) { + $char = $body{$i}; + switch ($char) { + case '<': + $sToken .= $char; + break; + case '/': + if ($sToken == '<') { + $sToken .= $char; + $bEndTag = true; + } else { + $content .= $char; + } + break; + case '>': + if ($bEndTag) { + $sToken .= $char; + if (preg_match('/\<\/\s*style\s*\>/i',$sToken,$aMatch)) { + $newpos = $i + 1; + $bSucces = true; + break 2; + } else { + $content .= $sToken; + } + $bEndTag = false; + } else { + $content .= $char; + } + break; + case '!': + if ($sToken == '<') { + // possible comment + if (isset($body{$i+2}) && substr($body,$i,3) == '!--') { + $i = strpos($body,'-->',$i+3); + if ($i === false) { // no end comment + $i = strlen($body); + } + $sToken = ''; + } + } else { + $content .= $char; + } + break; + default: + if ($bEndTag) { + $sToken .= $char; + } else { + $content .= $char; + } + break; + } + } + if ($bSucces == FALSE){ return array(FALSE, strlen($body)); } - $newpos = $ret[0] + strlen($ret[2]); - $content = $ret[1]; + + + /** * First look for general BODY style declaration, which would be * like so: @@ -1559,45 +1994,34 @@ function sq_fixstyle($body, $pos, $message, $id, $mailbox){ */ // $content = preg_replace("|url\s*\(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\)|si", // "url(\\1$secremoveimg\\2)", $content); - // remove NUL - $content = str_replace("\0", "", $content); - // NB I insert NUL characters to keep to avoid an infinite loop. They are removed after the loop. - while (preg_match("/url\s*\(\s*[\'\"]?([^:]+):(.*)?[\'\"]?\s*\)/si", $content, $matches)) { - $sProto = strtolower($matches[1]); - switch ($sProto) { - /** - * Fix url('https*://.*) declarations but only if $view_unsafe_images - * is false. - */ - case 'https': - case 'http': - if (!$view_unsafe_images){ - $sExpr = "/url\s*\(\s*([\'\"])\s*$sProto*:.*?([\'\"])\s*\)/si"; - $content = preg_replace($sExpr, "u\0r\0l(\\1$secremoveimg\\2)", $content); - } - break; - /** - * Fix urls that refer to cid: - */ - case 'cid': - $cidurl = 'cid:'. $matches[2]; - $httpurl = sq_cid2http($message, $id, $cidurl, $mailbox); - $content = preg_replace("|url\s*\(\s*$cidurl\s*\)|si", - "u\0r\0l($httpurl)", $content); - break; - default: - /** - * replace url with protocol other then the white list - * http,https and cid by an empty string. - */ - $content = preg_replace("/url\s*\(\s*[\'\"]?([^:]+):(.*)?[\'\"]?\s*\)/si", - "", $content); - break; + + // first check for 8bit sequences and disallowed control characters + if (preg_match('/[\16-\37\200-\377]+/',$content)) { + $content = ''; + return array($content, $newpos); + } + + // IE Sucks hard. We have a special function for it. + sq_fixIE_idiocy($content); + + // remove @import line + $content = preg_replace("/^\s*(@import.*)$/mi","\n\n",$content); + + // translate ur\l and variations (IE parses that) + // TODO check if the sq_fixIE_idiocy function already handles this. + $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content); + preg_match_all("/url\s*\((.+)\)/si",$content,$aMatch); + if (count($aMatch)) { + $aValue = $aReplace = array(); + foreach($aMatch[1] as $sMatch) { + // url value + $urlvalue = $sMatch; + sq_fix_url('style',$urlvalue, $message, $id, $mailbox,"'"); + $aValue[] = $sMatch; + $aReplace[] = $urlvalue; } - break; + $content = str_replace($aValue,$aReplace,$content); } - // remove NUL - $content = str_replace("\0", "", $content); /** * Remove any backslashes, entities, and extraneous whitespace. @@ -1610,11 +2034,14 @@ function sq_fixstyle($body, $pos, $message, $id, $mailbox){ * Fix stupid css declarations which lead to vulnerabilities * in IE. */ - $match = Array('/expression/i', + $match = Array('/\/\*.*\*\//', + '/expression/i', '/behaviou*r/i', '/binding/i', - '/include-source/i'); - $replace = Array('idiocy', 'idiocy', 'idiocy', 'idiocy'); + '/include-source/i', + '/javascript/i', + '/script/i'); + $replace = Array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy'); $contentNew = preg_replace($match, $replace, $contentTemp); if ($contentNew !== $contentTemp) { // insecure css declarations are used. From now on we don't care @@ -1926,12 +2353,14 @@ function sq_sanitize($body, */ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links =true) { - require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match + // require_once(SM_PATH . 'functions/url_parser.php'); // for $MailTo_PReg_Match global $attachment_common_show_images, $view_unsafe_images, $has_unsafe_images; /** * Don't display attached images in HTML mode. + * + * SB: why? */ $attachment_common_show_images = false; $tag_list = Array( @@ -1994,7 +2423,6 @@ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links "\\1$secremoveimg\\2", "\\1$secremoveimg\\2", "\\1$secremoveimg\\2", - "\\1$secremoveimg\\2" ) ), "/^href|action/i" => @@ -2005,7 +2433,6 @@ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links "/^([\'\"])\s*about\s*:.*([\'\"])/si" ), Array( - "\\1#\\1", "\\1#\\1", "\\1#\\1", "\\1#\\1" @@ -2014,24 +2441,26 @@ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links "/^style/i" => Array( Array( + "/\/\*.*\*\//", "/expression/i", "/binding/i", "/behaviou*r/i", "/include-source/i", "/position\s*:\s*absolute/i", + "/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", "/url\s*\(\s*([\'\"])\s*\S+script\s*:.*([\'\"])\s*\)/si", "/url\s*\(\s*([\'\"])\s*mocha\s*:.*([\'\"])\s*\)/si", "/url\s*\(\s*([\'\"])\s*about\s*:.*([\'\"])\s*\)/si", "/(.*)\s*:\s*url\s*\(\s*([\'\"]*)\s*\S+script\s*:.*([\'\"]*)\s*\)/si" ), Array( + "", "idiocy", "idiocy", "idiocy", "idiocy", - "", - "url(\\1#\\1)", - "url(\\1#\\1)", + "idiocy", + "url", "url(\\1#\\1)", "url(\\1#\\1)", "url(\\1#\\1)", @@ -2053,7 +2482,7 @@ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links array_push($bad_attvals{'/.*/'}{'/^src|background/i'}[1], "\\1$secremoveimg\\1"); array_push($bad_attvals{'/.*/'}{'/^style/i'}[0], - '/url\(([\'\"])\s*https*:.*([\'\"])\)/si'); + '/url\([\'\"]?https?:[^\)]*[\'\"]?\)/si'); array_push($bad_attvals{'/.*/'}{'/^style/i'}[1], "url(\\1$secremoveimg\\1)"); } @@ -2076,7 +2505,7 @@ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links $id, $mailbox ); - if (preg_match("|$secremoveimg|i", $trusted)){ + if (strpos($trusted,$secremoveimg)){ $has_unsafe_images = true; } @@ -2088,7 +2517,7 @@ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links if ($take_mailto_links) { // parseUrl($trusted); // this even parses URLs inside of tags... too aggressive global $MailTo_PReg_Match; - $MailTo_PReg_Match = '/mailto:' . substr($MailTo_PReg_Match, 1); + $MailTo_PReg_Match = '/mailto:' . substr($MailTo_PReg_Match, 1) ; if ((preg_match_all($MailTo_PReg_Match, $trusted, $regs)) && ($regs[0][0] != '')) { foreach ($regs[0] as $i => $mailto_before) { $mailto_params = $regs[10][$i]; @@ -2151,18 +2580,18 @@ function magicHTML($body, $id, $message, $mailbox = 'INBOX', $take_mailto_links */ function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) { global $languages, $squirrelmail_language; - $isIE = $isIE6 = 0; + $isIE = $isIE6plus = false; sqgetGlobalVar('HTTP_USER_AGENT', $HTTP_USER_AGENT, SQ_SERVER); if (strstr($HTTP_USER_AGENT, 'compatible; MSIE ') !== false && strstr($HTTP_USER_AGENT, 'Opera') === false) { - $isIE = 1; + $isIE = true; } - if (strstr($HTTP_USER_AGENT, 'compatible; MSIE 6') !== false && - strstr($HTTP_USER_AGENT, 'Opera') === false) { - $isIE6 = 1; + if (preg_match('/compatible; MSIE ([0-9]+)/', $HTTP_USER_AGENT, $match) && + ((int)$match[1]) >= 6 && strstr($HTTP_USER_AGENT, 'Opera') === false) { + $isIE6plus = true; } if (isset($languages[$squirrelmail_language]['XTRA_CODE']) && @@ -2219,17 +2648,20 @@ function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) { // "attachment"... does it apply to inline too? header ("Content-Disposition: attachment; filename=\"$filename\""); - if ($isIE && !$isIE6) { + if ($isIE && !$isIE6plus) { // This combination seems to work mostly. IE 5.5 SP 1 has // known issues (see the Microsoft Knowledge Base) // This works for most types, but doesn't work with Word files header ("Content-Type: application/download; name=\"$filename\""); - + header ("Content-Type: application/force-download; name=\"$filename\""); // These are spares, just in case. :-) //header("Content-Type: $type0/$type1; name=\"$filename\""); //header("Content-Type: application/x-msdownload; name=\"$filename\""); //header("Content-Type: application/octet-stream; name=\"$filename\""); + } else if ($isIE) { + // This is to prevent IE for MIME sniffing and auto open a file in IE + header ("Content-Type: application/force-download; name=\"$filename\""); } else { // another application/octet-stream forces download for Netscape header ("Content-Type: application/octet-stream; name=\"$filename\""); @@ -2242,5 +2674,3 @@ function SendDownloadHeaders($type0, $type1, $filename, $force, $filesize=0) { } } // end fn SendDownloadHeaders - -?>