From ea206d3d9299ce32996dcd24d49790e36f2b89a2 Mon Sep 17 00:00:00 2001
From: thomascube <thomas@roundcube.net>
Date: Fri, 1 Sep 2006 10:01:31 +0000
Subject: Improved message parsing and HTML validation

---
 program/steps/mail/func.inc | 31 ++++++++++++++-----------------
 program/steps/mail/get.inc  | 14 ++++++++------
 program/steps/mail/show.inc |  2 +-
 3 files changed, 23 insertions(+), 24 deletions(-)

(limited to 'program/steps/mail')

diff --git a/program/steps/mail/func.inc b/program/steps/mail/func.inc
index bae14a299..58da0ca45 100644
--- a/program/steps/mail/func.inc
+++ b/program/steps/mail/func.inc
@@ -676,18 +676,18 @@ function rcmail_print_body($part, $safe=FALSE, $plain=FALSE)
 
     if (!$safe)  // remove remote images and scripts
       {
-      $remote_patterns = array('/(src|background)=(["\']?)([hftps]{3,5}:\/{2}[^"\'\s]+)(\2|\s|>)/Ui',
-                           //  '/(src|background)=(["\']?)([\.\/]+[^"\'\s]+)(\2|\s|>)/Ui',
+      $remote_patterns = array('/<img\s+(.*)src=(["\']?)([hftps]{3,5}:\/{2}[^"\'\s]+)(\2|\s|>)/Ui',
+                               '/(src|background)=(["\']?)([hftps]{3,5}:\/{2}[^"\'\s]+)(\2|\s|>)/Ui',
                                '/(<base.*href=["\']?)([hftps]{3,5}:\/{2}[^"\'\s]+)([^<]*>)/i',
                                '/(<link.*href=["\']?)([hftps]{3,5}:\/{2}[^"\'\s]+)([^<]*>)/i',
                                '/url\s*\(["\']?([hftps]{3,5}:\/{2}[^"\'\s]+)["\']?\)/i',
                                '/url\s*\(["\']?([\.\/]+[^"\'\s]+)["\']?\)/i',
                                '/<script.+<\/script>/Umis');
 
-      $remote_replaces = array('',  // '\\1=\\2#\\4',
-                            // '\\1=\\2#\\4',
+      $remote_replaces = array('<img \\1src=\\2./program/blank.gif\\4',
+                               '',
+                               '',
                                '',
-                               '',  // '\\1#\\3',
                                'none',
                                'none',
                                '');
@@ -766,7 +766,7 @@ function rcmail_print_body($part, $safe=FALSE, $plain=FALSE)
     // insert the links for urls and mailtos
     $body = preg_replace("/##string_replacement\{([0-9]+)\}##/e", "\$replace_strings[\\1]", join("\n", $a_lines));
     
-    return "<div class=\"pre\">\n".$body."\n</div>";
+    return "<div class=\"pre\">".$body."\n</div>";
     }
   }
 
@@ -947,7 +947,7 @@ function rcmail_parse_message(&$structure, $arg=array(), $recursive=FALSE)
       $a_replaces = array();
         
       foreach ($sa_inline_objects as $inline_object)
-        $a_replaces['cid:'.$inline_object->content_id] = sprintf($get_url, $inline_object->mime_id);
+        $a_replaces['cid:'.$inline_object->content_id] = htmlspecialchars(sprintf($get_url, $inline_object->mime_id));
       
       // add replace array to each content part
       // (will be applied later when part body is available)
@@ -1092,8 +1092,8 @@ function rcmail_message_body($attrib)
     foreach ($MESSAGE['attachments'] as $attach_prop)
       {
       if (strpos($attach_prop->mimetype, 'image/')===0)
-        $out .= sprintf("\n<hr />\n<p align=\"center\"><img src=\"%s&_part=%s\" alt=\"%s\" title=\"%s\" /></p>\n",
-                        $GET_URL, $attach_prop->mime_id,
+        $out .= sprintf("\n<hr />\n<p align=\"center\"><img src=\"%s&amp;_part=%s\" alt=\"%s\" title=\"%s\" /></p>\n",
+                        htmlspecialchars($GET_URL), $attach_prop->mime_id,
                         $attach_prop->filename,
                         $attach_prop->filename);
       }
@@ -1121,15 +1121,12 @@ function rcmail_mod_html_body($body, $container_id)
   // find STYLE tags
   while (($pos = strpos($body_lc, '<style', $last_style_pos)) && ($pos2 = strpos($body_lc, '</style>', $pos)))
     {
-    $pos2 += 8;
-    $body_pre = substr($body, 0, $pos);
-    $styles = substr($body, $pos, $pos2-$pos);
-    $body_post = substr($body, $pos2, strlen($body)-$pos2);
-    
+    $pos = strpos($body_lc, '>', $pos)+1;
+
     // replace all css definitions with #container [def]
-    $styles = rcmail_mod_css_styles($styles, $container_id);
-    
-    $body = $body_pre . $styles . $body_post;
+    $styles = rcmail_mod_css_styles(substr($body, $pos, $pos2-$pos), $container_id);
+
+    $body = substr($body, 0, $pos) . $styles . substr($body, $pos2);    
     $last_style_pos = $pos2;
     }
 
diff --git a/program/steps/mail/get.inc b/program/steps/mail/get.inc
index f05bbbd94..083de8664 100644
--- a/program/steps/mail/get.inc
+++ b/program/steps/mail/get.inc
@@ -87,17 +87,19 @@ else if ($pid = get_input_value('_part', RCUBE_INPUT_GET))
     // deliver part content
     if ($ctype_primary=='text' && $ctype_secondary=='html')
       {
-      // get part body if not available
-      if (!$part->body)
-        $part->body = $IMAP->get_message_part($MESSAGE['UID'], $part->mime_id, $part);      
-
-      list($MESSAGE['parts']) = rcmail_parse_message($part,
+      // we have to analyze the whole structure again to find inline objects
+      list($MESSAGE['parts']) = rcmail_parse_message($MESSAGE['structure'],
                                                      array('safe' => (bool)$_GET['_safe'],
                                                            'prefer_html' => TRUE,
                                                            'get_url' => $GET_URL.'&_part=%s'));
+      $part = &$MESSAGE['parts'][0];
+
+      // get part body if not available
+      if (!$part->body)
+        $part->body = $IMAP->get_message_part($MESSAGE['UID'], $part->mime_id, $part);      
 
       $OUTPUT = new rcube_html_page();
-      $OUTPUT->write(rcmail_print_body($MESSAGE['parts'][0], (bool)$_GET['_safe']));
+      $OUTPUT->write(rcmail_print_body($part, (bool)$_GET['_safe']));
       }
     else
       {
diff --git a/program/steps/mail/show.inc b/program/steps/mail/show.inc
index f381818ed..51633fa7b 100644
--- a/program/steps/mail/show.inc
+++ b/program/steps/mail/show.inc
@@ -108,7 +108,7 @@ function rcmail_message_attachments($attrib)
                         show_bytes($attach_prop->size));
       else
         $out .= sprintf('<li><a href="%s&amp;_part=%s" onclick="return %s.command(\'load-attachment\',{part:\'%s\', mimetype:\'%s\'},this)">%s</a></li>'."\n",
-                        htmlentities($GET_URL),
+                        htmlspecialchars($GET_URL),
                         $attach_prop->mime_id,
                         $JS_OBJECT_NAME,
                         $attach_prop->mime_id,
-- 
cgit v1.2.3