Add ability to ignore certain symbols in metric when validating cache

4 files changed, 83 insertions, 30 deletions

diff --git a/src/libserver/cfg_file.h b/src/libserver/cfg_file.h
index 6263705ac..05946a79e 100644
--- a/src/libserver/cfg_file.h
+++ b/src/libserver/cfg_file.h
@@ -89,6 +89,9 @@ struct rspamd_symbols_group {
 	gboolean one_shot;
 };
 
+#define RSPAMD_SYMBOL_FLAG_ONESHOT (1 << 0)
+#define RSPAMD_SYMBOL_FLAG_IGNORE (1 << 1)
+
 /**
  * Symbol definition
  */
@@ -99,7 +102,7 @@ struct rspamd_symbol_def {
 	gdouble score;
 	struct rspamd_symbols_group *gr;
 	GList *groups;
-	gboolean one_shot;
+	guint flags;
 };
 
 
diff --git a/src/libserver/cfg_rcl.c b/src/libserver/cfg_rcl.c
index e8e0818ca..bbe241cb1 100644
--- a/src/libserver/cfg_rcl.c
+++ b/src/libserver/cfg_rcl.c
@@ -321,6 +321,7 @@ rspamd_rcl_symbol_handler (rspamd_mempool_t *pool, const ucl_object_t *obj,
 	struct rspamd_symbol_def *sym_def;
 	struct metric *metric;
 	struct rspamd_config *cfg;
+	const ucl_object_t *elt;
 	GList *metric_list;
 
 	g_assert (key != NULL);
@@ -359,6 +360,18 @@ rspamd_rcl_symbol_handler (rspamd_mempool_t *pool, const ucl_object_t *obj,
 		msg_warn_config ("redefining symbol '%s' in metric '%s'", key, metric->name);
 	}
 
+	if ((elt = ucl_object_find_key (obj, "one_shot")) != NULL) {
+		if (ucl_object_toboolean (elt)) {
+			sym_def->flags |= RSPAMD_SYMBOL_FLAG_ONESHOT;
+		}
+	}
+
+	if ((elt = ucl_object_find_key (obj, "ignore")) != NULL) {
+		if (ucl_object_toboolean (elt)) {
+			sym_def->flags |= RSPAMD_SYMBOL_FLAG_IGNORE;
+		}
+	}
+
 	if (!rspamd_rcl_section_parse_defaults (section, pool, obj,
 			sym_def, err)) {
 		return FALSE;
@@ -408,11 +421,13 @@ rspamd_rcl_metric_handler (rspamd_mempool_t *pool, const ucl_object_t *obj,
 		const gchar *key, gpointer ud,
 		struct rspamd_rcl_section *section, GError **err)
 {
-	const ucl_object_t *val, *cur;
+	const ucl_object_t *val, *cur, *elt;
+	ucl_object_iter_t it;
 	struct rspamd_config *cfg = ud;
 	struct metric *metric;
 	struct rspamd_rcl_section *subsection;
 	struct rspamd_rcl_symbol_data sd;
+	struct rspamd_symbol_def *sym_def;
 
 	g_assert (key != NULL);
 
@@ -483,6 +498,29 @@ rspamd_rcl_metric_handler (rspamd_mempool_t *pool, const ucl_object_t *obj,
 		}
 	}
 
+	/* Handle ignored symbols */
+	val = ucl_object_find_key (obj, "ignore");
+	if (val != NULL && ucl_object_type (val) == UCL_ARRAY) {
+		LL_FOREACH (val, cur) {
+			it = NULL;
+
+			while ((elt = ucl_iterate_object (cur, &it, true)) != NULL) {
+				if (ucl_object_type (elt) == UCL_STRING) {
+					sym_def = g_hash_table_lookup (metric->symbols,
+							ucl_object_tostring (elt));
+
+					if (sym_def != NULL) {
+						sym_def->flags |= RSPAMD_SYMBOL_FLAG_IGNORE;
+					}
+					else {
+						msg_warn ("cannot find symbol %s to set ignore flag",
+								ucl_object_tostring (elt));
+					}
+				}
+			}
+		}
+	}
+
 	return TRUE;
 }
 
@@ -1516,11 +1554,6 @@ rspamd_rcl_config_init (void)
 			G_STRUCT_OFFSET (struct rspamd_symbol_def, description),
 			0);
 	rspamd_rcl_add_default_handler (ssub,
-			"one_shot",
-			rspamd_rcl_parse_struct_boolean,
-			G_STRUCT_OFFSET (struct rspamd_symbol_def, one_shot),
-			0);
-	rspamd_rcl_add_default_handler (ssub,
 			"score",
 			rspamd_rcl_parse_struct_double,
 			G_STRUCT_OFFSET (struct rspamd_symbol_def, score),
@@ -1575,11 +1608,6 @@ rspamd_rcl_config_init (void)
 			G_STRUCT_OFFSET (struct rspamd_symbol_def, description),
 			0);
 	rspamd_rcl_add_default_handler (sssub,
-			"one_shot",
-			rspamd_rcl_parse_struct_boolean,
-			G_STRUCT_OFFSET (struct rspamd_symbol_def, one_shot),
-			0);
-	rspamd_rcl_add_default_handler (sssub,
 			"score",
 			rspamd_rcl_parse_struct_double,
 			G_STRUCT_OFFSET (struct rspamd_symbol_def, score),
diff --git a/src/libserver/cfg_utils.c b/src/libserver/cfg_utils.c
index 9447eaf32..042658f8a 100644
--- a/src/libserver/cfg_utils.c
+++ b/src/libserver/cfg_utils.c
@@ -1246,7 +1246,10 @@ rspamd_config_add_metric_symbol (struct rspamd_config *cfg,
 	sym_def->score = score;
 	sym_def->weight_ptr = score_ptr;
 	sym_def->name = rspamd_mempool_strdup (cfg->cfg_pool, symbol);
-	sym_def->one_shot = one_shot;
+
+	if (one_shot) {
+		sym_def->flags |= RSPAMD_SYMBOL_FLAG_ONESHOT;
+	}
 
 	if (description) {
 		sym_def->description = rspamd_mempool_strdup (cfg->cfg_pool, description);
diff --git a/src/libserver/symbols_cache.c b/src/libserver/symbols_cache.c
index 94a3302e8..e64c35e2a 100644
--- a/src/libserver/symbols_cache.c
+++ b/src/libserver/symbols_cache.c
@@ -848,7 +848,12 @@ rspamd_symbols_cache_validate (struct symbols_cache *cache,
 	gboolean strict)
 {
 	struct cache_item *item;
-	GList *cur, *metric_symbols;
+	GHashTableIter it;
+	GList *cur;
+	gpointer k, v;
+	struct rspamd_symbol_def *sym_def;
+	struct metric *metric;
+	gboolean ignore_symbol = FALSE, ret = TRUE;
 
 	if (cache == NULL) {
 		msg_err_cache ("empty cache is invalid");
@@ -866,28 +871,42 @@ rspamd_symbols_cache_validate (struct symbols_cache *cache,
 			rspamd_symbols_cache_validate_cb,
 			cache);
 	/* Now check each metric item and find corresponding symbol in a cache */
-	metric_symbols = g_hash_table_get_keys (cfg->metrics_symbols);
-	cur = metric_symbols;
-	while (cur) {
-		item = g_hash_table_lookup (cache->items_by_symbol, cur->data);
-
-		if (item == NULL) {
-			msg_warn_cache (
-				"symbol '%s' has its score defined but there is no "
-				"corresponding rule registered",
-				cur->data);
-			if (strict) {
-				g_list_free (metric_symbols);
-				return FALSE;
+	g_hash_table_iter_init (&it, cfg->metrics_symbols);
+
+	while (g_hash_table_iter_next (&it, &k, &v)) {
+		ignore_symbol = FALSE;
+		cur = v;
+
+		while (cur) {
+			metric = cur->data;
+			sym_def = g_hash_table_lookup (metric->symbols, k);
+
+			if (sym_def && (sym_def->flags & RSPAMD_SYMBOL_FLAG_IGNORE)) {
+				ignore_symbol = TRUE;
+				break;
+			}
+
+			cur = g_list_next (cur);
+		}
+
+		if (!ignore_symbol) {
+			item = g_hash_table_lookup (cache->items_by_symbol, k);
+
+			if (item == NULL) {
+				msg_warn_cache (
+						"symbol '%s' has its score defined but there is no "
+								"corresponding rule registered",
+						k);
+				if (strict) {
+					ret = FALSE;
+				}
 			}
 		}
-		cur = g_list_next (cur);
 	}
-	g_list_free (metric_symbols);
 
 	post_cache_init (cache);
 
-	return TRUE;
+	return ret;
 }
 
 static gboolean