Remove HFILTER_URL_ONELINE rule.

- Added functions:
 `rspamd_textpart:get_length` - get a full length of text part
 `rspamd_url:get_length` - get a full length of an URL object

- Removed `HFILTER_URL_ONELINE` as it is broken by nature
- Reworked hfilter to improve performance

diff --git a/conf/lua/hfilter.lua b/conf/lua/hfilter.lua
index f0c4748a8bc9ca868bc6a0b9002758443a37a939..0ec54f1f4771e9bf46d1fc9ecd52085dc6fe7bd3 100644 (file)
--- a/conf/lua/hfilter.lua
+++ b/conf/lua/hfilter.lua
@@ -250,28 +250,24 @@ local function hfilter(task)
     local parts = task:get_text_parts()
     if parts then
         --One text part--
-        if table.maxn(parts) > 0 and parts[1]:get_content() then
-            local part_text = trim1(parts[1]:get_content())
-            local total_part_len = string.len(part_text)
-            if total_part_len > 0 then
-                local urls = task:get_urls()
-                if urls then
-                    local total_url_len = 0
-                    for _,url in ipairs(urls) do
-                        total_url_len = total_url_len + string.len(url:get_text())
-                    end
-                    if total_url_len > 0 then
-                        if total_url_len + 7 > total_part_len then
-                            task:insert_result('HFILTER_URL_ONLY', 1.00)
-                        else
-                            if not string.find(part_text, "\n") then
-                                task:insert_result('HFILTER_URL_ONELINE', 1.00)
-                            end
-                        end
-                    end
-                end
-            end
-        end
+               total_parts_len = 0
+               for _,p in ipairs(parts) do
+                       total_parts_len = total_parts_len + p:get_length()
+               end
+               if total_parts_len > 0 then
+                       local urls = task:get_urls()
+                       if urls then
+                               local total_url_len = 0
+                               for _,url in ipairs(urls) do
+                                       total_url_len = total_url_len + url:get_length()
+                               end
+                               if total_url_len > 0 then
+                                       if total_url_len + 7 > total_part_len then
+                                               task:insert_result('HFILTER_URL_ONLY', 1.00)
+                                       end
+                               end
+                       end
+               end
     end
     
     return false
@@ -284,4 +280,4 @@ rspamd_config:register_symbols(hfilter, 1.0,
 "HFILTER_FROMHOST_NORESOLVE_MX", "HFILTER_FROMHOST_NORES_A_OR_MX", "HFILTER_FROMHOST_NOT_FQDN",  
 "HFILTER_MID_NOT_FQDN",
 "HFILTER_HOSTNAME_NOPTR",
-"HFILTER_URL_ONLY", "HFILTER_URL_ONELINE");
+"HFILTER_URL_ONLY");

diff --git a/conf/metrics.conf b/conf/metrics.conf
index 13b9796afe4f0f6ae3d51beb1b56b17b6e8fb4e0..cba0ce18cd498ad791d7104eb4aca498c88f53da 100644 (file)
--- a/conf/metrics.conf
+++ b/conf/metrics.conf
@@ -714,5 +714,4 @@ metric {
     symbol { weight = 0.50; name = "HFILTER_MID_NOT_FQDN"; description = "Message-id host not FQDN"; }
     symbol { weight = 4.00; name = "HFILTER_HOSTNAME_NOPTR"; description = "No PTR for IP"; }
     symbol { weight = 3.50; name = "HFILTER_URL_ONLY"; description = "URL only in body"; }
-    symbol { weight = 2.00; name = "HFILTER_URL_ONELINE"; description = "One line URL and text in body"; }
 }

diff --git a/src/lua/lua_task.c b/src/lua/lua_task.c
index 73da200adb22779c5da02914903d671d4539beda..eadc056623a011886bb9027d3b39c1f94ecf182a 100644 (file)
--- a/src/lua/lua_task.c
+++ b/src/lua/lua_task.c
@@ -146,6 +146,7 @@ static const struct luaL_reg    tasklib_m[] = {
 
 /* Textpart methods */
 LUA_FUNCTION_DEF (textpart, get_content);
+LUA_FUNCTION_DEF (textpart, get_length);
 LUA_FUNCTION_DEF (textpart, is_empty);
 LUA_FUNCTION_DEF (textpart, is_html);
 LUA_FUNCTION_DEF (textpart, get_fuzzy);
@@ -154,6 +155,7 @@ LUA_FUNCTION_DEF (textpart, compare_distance);
 
 static const struct luaL_reg    textpartlib_m[] = {
        LUA_INTERFACE_DEF (textpart, get_content),
+       LUA_INTERFACE_DEF (textpart, get_length),
        LUA_INTERFACE_DEF (textpart, is_empty),
        LUA_INTERFACE_DEF (textpart, is_html),
        LUA_INTERFACE_DEF (textpart, get_fuzzy),
@@ -197,6 +199,7 @@ static const struct luaL_reg    imagelib_m[] = {
 };
 
 /* URL methods */
+LUA_FUNCTION_DEF (url, get_length);
 LUA_FUNCTION_DEF (url, get_host);
 LUA_FUNCTION_DEF (url, get_user);
 LUA_FUNCTION_DEF (url, get_path);
@@ -205,6 +208,7 @@ LUA_FUNCTION_DEF (url, is_phished);
 LUA_FUNCTION_DEF (url, get_phished);
 
 static const struct luaL_reg    urllib_m[] = {
+       LUA_INTERFACE_DEF (url, get_length),
        LUA_INTERFACE_DEF (url, get_host),
        LUA_INTERFACE_DEF (url, get_user),
        LUA_INTERFACE_DEF (url, get_path),
@@ -1385,6 +1389,26 @@ lua_textpart_get_content (lua_State * L)
        return 1;
 }
 
+static gint
+lua_textpart_get_length (lua_State * L)
+{
+       struct mime_text_part          *part = lua_check_textpart (L);
+
+       if (part == NULL) {
+               lua_pushnil (L);
+               return 1;
+       }
+
+       if (part->is_empty) {
+               lua_pushnumber (L, 0);
+       }
+       else {
+               lua_pushnumber (L, part->content->len);
+       }
+
+       return 1;
+}
+
 static gint
 lua_textpart_is_empty (lua_State * L)
 {
@@ -1721,6 +1745,20 @@ lua_image_get_filename (lua_State *L)
 }
 
 /* URL part */
+static gint
+lua_url_get_length (lua_State *L)
+{
+       struct uri                      *url = lua_check_url (L);
+
+       if (url != NULL) {
+               lua_pushinteger (L, strlen (struri (url)));
+       }
+       else {
+               lua_pushnil (L);
+       }
+       return 1;
+}
+
 static gint
 lua_url_get_host (lua_State *L)
 {