[Feature] Add rule to detect broken content type

rules/misc.lua

@@ -168,7 +168,7 @@ rspamd_config.R_SUSPICIOUS_URL = {
     end
     return false
   end,
-  score = 6.0,
+  score = 5.0,
   one_shot = true,
   description = 'Obfusicated or suspicious URL has been found in a message',
   group = 'url'
@@ -183,6 +183,16 @@ rspamd_config.BROKEN_HEADERS = {
   description = 'Headers structure is likely broken'
 }
 
+rspamd_config.BROKEN_CONTENT_TYPE = {
+  callback = function(task)
+    return fun.any(function(p) return p:is_broken() end,
+      task:get_parts())
+  end,
+  score = 3.0,
+  group = 'header',
+  description = 'Message has part with broken content type'
+}
+
 rspamd_config.HEADER_RCONFIRM_MISMATCH = {
   callback = function (task)
     local header_from = nil

src/lua/lua_mimepart.c

@@ -288,6 +288,12 @@ LUA_FUNCTION_DEF (mimepart, get_text);
  * @return {string} 128 characters hex string with digest of the part
  */
 LUA_FUNCTION_DEF (mimepart, get_digest);
+/***
+ * @method mime_part:is_broken()
+ * Returns true if mime part has incorrectly specified content type
+ * @return {bool} true if a part has bad content type
+ */
+LUA_FUNCTION_DEF (mimepart, is_broken);
 
 static const struct luaL_reg mimepartlib_m[] = {
 	LUA_INTERFACE_DEF (mimepart, get_content),
@@ -302,6 +308,7 @@ static const struct luaL_reg mimepartlib_m[] = {
 	LUA_INTERFACE_DEF (mimepart, is_archive),
 	LUA_INTERFACE_DEF (mimepart, get_archive),
 	LUA_INTERFACE_DEF (mimepart, is_text),
+	LUA_INTERFACE_DEF (mimepart, is_broken),
 	LUA_INTERFACE_DEF (mimepart, get_text),
 	LUA_INTERFACE_DEF (mimepart, get_digest),
 	{"__tostring", rspamd_lua_class_tostring},
@@ -738,6 +745,26 @@ lua_mimepart_is_text (lua_State * L)
 	return 1;
 }
 
+static gint
+lua_mimepart_is_broken (lua_State * L)
+{
+	struct rspamd_mime_part *part = lua_check_mimepart (L);
+
+	if (part == NULL) {
+		return luaL_error (L, "invalid arguments");
+	}
+
+	if (part->ct) {
+		lua_pushboolean (L, (part->ct->flags & RSPAMD_CONTENT_TYPE_BROKEN) ?
+				true : false);
+	}
+	else {
+		lua_pushboolean (L, true);
+	}
+
+	return 1;
+}
+
 static gint
 lua_mimepart_get_image (lua_State * L)
 {