Merge pull request #5043 from rspamd/vstakhov-gpt-tunes

Improve some functions in the GPT plugin

2 files changed, 37 insertions, 14 deletions

diff --git a/src/lua/lua_task.c b/src/lua/lua_task.c
index dc41d4ab7..61aac63f2 100644
--- a/src/lua/lua_task.c
+++ b/src/lua/lua_task.c
@@ -2642,7 +2642,7 @@ struct rspamd_url_query_to_inject_cbd {
 
 static gboolean
 inject_url_query_callback(struct rspamd_url *url, gsize start_offset,
-						gsize end_offset, gpointer ud)
+						  gsize end_offset, gpointer ud)
 {
 	struct rspamd_url_query_to_inject_cbd *cbd =
 		(struct rspamd_url_query_to_inject_cbd *) ud;
@@ -2661,7 +2661,7 @@ inject_url_query_callback(struct rspamd_url *url, gsize start_offset,
 
 static void
 inject_url_query(struct rspamd_task *task, struct rspamd_url *url,
-					   GPtrArray *part_urls)
+				 GPtrArray *part_urls)
 {
 	if (url->querylen > 0) {
 		struct rspamd_url_query_to_inject_cbd cbd;
@@ -2692,11 +2692,11 @@ lua_task_inject_url(lua_State *L)
 	if (lua_isuserdata(L, 3)) {
 		/* We also have a mime part there */
 		mpart = *((struct rspamd_mime_part **)
-							   rspamd_lua_check_udata_maybe(L, 3, rspamd_mimepart_classname));
+					  rspamd_lua_check_udata_maybe(L, 3, rspamd_mimepart_classname));
 	}
 	if (task && task->message && url && url->url) {
 		if (rspamd_url_set_add_or_increase(MESSAGE_FIELD(task, urls), url->url, false)) {
-			if(mpart && mpart->urls) {
+			if (mpart && mpart->urls) {
 				inject_url_query(task, url->url, mpart->urls);
 			}
 		}
@@ -4682,7 +4682,7 @@ lua_push_symbol_result(lua_State *L,
 	struct rspamd_symbol_option *opt;
 	struct rspamd_symbols_group *sym_group;
 	unsigned int i;
-	int j = 1, table_fields_cnt = 4;
+	int j = 1, table_fields_cnt = 5;
 
 	if (!metric_res) {
 		metric_res = task->result;
@@ -4710,10 +4710,22 @@ lua_push_symbol_result(lua_State *L,
 			lua_pushstring(L, symbol);
 			lua_settable(L, -3);
 		}
+
 		lua_pushstring(L, "score");
 		lua_pushnumber(L, s->score);
 		lua_settable(L, -3);
 
+		/* Dynamic weight of the symbol */
+		if (s->sym->score != 0) {
+			lua_pushstring(L, "weight");
+			lua_pushnumber(L, s->score / s->sym->score);
+		}
+		else {
+			lua_pushstring(L, "weight");
+			lua_pushnumber(L, 0.0);
+		}
+		lua_settable(L, -3);
+
 		if (s->sym && s->sym->gr) {
 			lua_pushstring(L, "group");
 			lua_pushstring(L, s->sym->gr->name);
diff --git a/src/plugins/lua/gpt.lua b/src/plugins/lua/gpt.lua
index ddd2f0186..6adbce3bf 100644
--- a/src/plugins/lua/gpt.lua
+++ b/src/plugins/lua/gpt.lua
@@ -59,13 +59,13 @@ local fun = require "fun"
 
 -- Exclude checks if one of those is found
 local default_symbols_to_except = {
-  'BAYES_SPAM', -- We already know that it is a spam, so we can safely skip it, but no same logic for HAM!
-  'WHITELIST_SPF',
-  'WHITELIST_DKIM',
-  'WHITELIST_DMARC',
-  'FUZZY_DENIED',
-  'REPLY',
-  'BOUNCE',
+  BAYES_SPAM = 0.9, -- We already know that it is a spam, so we can safely skip it, but no same logic for HAM!
+  WHITELIST_SPF = -1,
+  WHITELIST_DKIM = -1,
+  WHITELIST_DMARC = -1,
+  FUZZY_DENIED = -1,
+  REPLY = -1,
+  BOUNCE = -1,
 }
 
 local settings = {
@@ -105,9 +105,20 @@ local function default_condition(task)
     end
   end
   -- We also exclude some symbols
-  for _, s in ipairs(settings.symbols_to_except) do
+  for s, required_weight in pairs(settings.symbols_to_except) do
     if task:has_symbol(s) then
-      return false, 'skip as "' .. s .. '" is found'
+      if required_weight > 0 then
+        -- Also check score
+        local sym = task:get_symbol(s)
+        -- Must exist as we checked it before with `has_symbol`
+        if math.abs(sym.weight) >= required_weight then
+          return false, 'skip as "' .. s .. '" is found (weight: ' .. sym.weight .. ')'
+        end
+        lua_util.debugm(N, task, 'symbol %s has weight %s, but required %s', s,
+            sym.weight, required_weight)
+      else
+        return false, 'skip as "' .. s .. '" is found'
+      end
     end
   end