[Fix] Improve OMOGRAPH_URL rule

- Calculate omographs in each dot component separately
- Normalize omographs
- Count utf8 characters properly

2 files changed, 20 insertions, 4 deletions

diff --git a/rules/misc.lua b/rules/misc.lua
index 56de79a6b..75f31f39c 100644
--- a/rules/misc.lua
+++ b/rules/misc.lua
@@ -767,10 +767,21 @@ rspamd_config.OMOGRAPH_URL = {
         local h = u:get_host()
 
         if h then
-          local non_latin,total = util.count_non_ascii(h)
+          local parts = rspamd_str_split(h, '.')
 
-          if non_latin ~= total and non_latin > 0 then
-            return true, 1.0, h
+          local bad_omographs = 0
+
+          for _,p in ipairs(parts) do
+            local cnlat,ctot = util.count_non_ascii(p)
+
+            if cnlat > 0 and cnlat ~= ctot then
+              bad_omographs = bad_omographs + 1.0 / cnlat
+            end
+          end
+
+          if bad_omographs > 0 then
+            if bad_omographs > 1 then bad_omographs = 1.0 end
+            return true, bad_omographs, h
           end
         end
       end
diff --git a/src/lua/lua_util.c b/src/lua/lua_util.c
index 220467664..669ac5bd9 100644
--- a/src/lua/lua_util.c
+++ b/src/lua/lua_util.c
@@ -1891,7 +1891,7 @@ lua_util_count_non_ascii (lua_State *L)
 {
 	gsize len;
 	const gchar *str = lua_tolstring (L, 1, &len);
-	const gchar *p, *end;
+	const gchar *p, *end, *np;
 	gint ret = 0, total = 0;
 
 	if (str != NULL) {
@@ -1900,8 +1900,13 @@ lua_util_count_non_ascii (lua_State *L)
 
 		while (p < end) {
 			if (*p & 0x80) {
+				np = g_utf8_find_next_char (p, end);
 				ret ++;
 				total ++;
+
+				p = (np != p) ? np : p + 1;
+
+				continue;
 			}
 			else if (g_ascii_isalpha (*p)) {
 				total ++;