[Rework] Use hash tables for symbols options

5 files changed, 65 insertions, 51 deletions

diff --git a/src/libmime/filter.c b/src/libmime/filter.c
index e3ee3cb63..baeff9fb9 100644
--- a/src/libmime/filter.c
+++ b/src/libmime/filter.c
@@ -75,11 +75,12 @@ insert_metric_result (struct rspamd_task *task,
 	struct metric *metric,
 	const gchar *symbol,
 	double flag,
-	GList * opts,
+	const gchar *opt,
 	gboolean single)
 {
 	struct metric_result *metric_res;
 	struct symbol *s;
+	char *opt_cpy;
 	gdouble w, *gr_score = NULL;
 	struct rspamd_symbol_def *sdef;
 	struct rspamd_symbols_group *gr = NULL;
@@ -142,19 +143,24 @@ insert_metric_result (struct rspamd_task *task,
 			 */
 			single = TRUE;
 		}
-		if (s->options && opts && opts != s->options &&
-				!(sdef && (sdef->flags & RSPAMD_SYMBOL_FLAG_ONEPARAM))) {
-			/* Append new options */
-			s->options = g_list_concat (s->options, g_list_copy (opts));
-			/*
-			 * Note that there is no need to add new destructor of GList as elements of appended
-			 * GList are used directly, so just free initial GList
-			 */
-		}
-		else if (opts) {
-			s->options = g_list_copy (opts);
-			rspamd_mempool_add_destructor (task->task_pool,
-				(rspamd_mempool_destruct_t) g_list_free, s->options);
+		if (opt) {
+			if (s->options && !(sdef &&
+					(sdef->flags & RSPAMD_SYMBOL_FLAG_ONEPARAM))) {
+				/* Append new options */
+				if (!g_hash_table_lookup (s->options, opt)) {
+					opt_cpy = rspamd_mempool_strdup (task->task_pool, opt);
+					g_hash_table_insert (s->options, opt_cpy, opt_cpy);
+				}
+			}
+			else {
+				s->options = g_hash_table_new (rspamd_strcase_hash,
+						rspamd_strcase_equal);
+				rspamd_mempool_add_destructor (task->task_pool,
+						(rspamd_mempool_destruct_t)g_hash_table_unref,
+						s->options);
+				opt_cpy = rspamd_mempool_strdup (task->task_pool, opt);
+				g_hash_table_insert (s->options, opt_cpy, opt_cpy);
+			}
 		}
 		if (!single) {
 			/* Handle grow factor */
@@ -190,10 +196,14 @@ insert_metric_result (struct rspamd_task *task,
 		s->def = sdef;
 		metric_res->score += w;
 
-		if (opts) {
-			s->options = g_list_copy (opts);
+		if (opt) {
+			s->options = g_hash_table_new (rspamd_strcase_hash,
+					rspamd_strcase_equal);
 			rspamd_mempool_add_destructor (task->task_pool,
-				(rspamd_mempool_destruct_t) g_list_free, s->options);
+					(rspamd_mempool_destruct_t)g_hash_table_unref,
+					s->options);
+			opt_cpy = rspamd_mempool_strdup (task->task_pool, opt);
+			g_hash_table_insert (s->options, opt_cpy, opt_cpy);
 		}
 		else {
 			s->options = NULL;
@@ -201,6 +211,7 @@ insert_metric_result (struct rspamd_task *task,
 
 		g_hash_table_insert (metric_res->symbols, (gpointer) symbol, s);
 	}
+
 	msg_debug ("symbol %s, score %.2f, metric %s, factor: %f",
 		symbol,
 		s->score,
@@ -212,7 +223,7 @@ static void
 insert_result_common (struct rspamd_task *task,
 	const gchar *symbol,
 	double flag,
-	GList * opts,
+	const gchar *opt,
 	gboolean single)
 {
 	struct metric *metric;
@@ -224,7 +235,7 @@ insert_result_common (struct rspamd_task *task,
 
 		while (cur) {
 			metric = cur->data;
-			insert_metric_result (task, metric, symbol, flag, opts, single);
+			insert_metric_result (task, metric, symbol, flag, opt, single);
 			cur = g_list_next (cur);
 		}
 	}
@@ -234,7 +245,7 @@ insert_result_common (struct rspamd_task *task,
 			task->cfg->default_metric,
 			symbol,
 			flag,
-			opts,
+			opt,
 			single);
 	}
 
@@ -242,11 +253,6 @@ insert_result_common (struct rspamd_task *task,
 	if (task->cfg->cache) {
 		rspamd_symbols_cache_inc_frequency (task->cfg->cache, symbol);
 	}
-
-	if (opts != NULL) {
-		/* XXX: it is not wise to destroy them here */
-		g_list_free (opts);
-	}
 }
 
 /* Insert result that may be increased on next insertions */
@@ -254,9 +260,9 @@ void
 rspamd_task_insert_result (struct rspamd_task *task,
 	const gchar *symbol,
 	double flag,
-	GList * opts)
+	const gchar *opt)
 {
-	insert_result_common (task, symbol, flag, opts, task->cfg->one_shot_mode);
+	insert_result_common (task, symbol, flag, opt, task->cfg->one_shot_mode);
 }
 
 /* Insert result as a single option */
@@ -264,9 +270,9 @@ void
 rspamd_task_insert_result_single (struct rspamd_task *task,
 	const gchar *symbol,
 	double flag,
-	GList * opts)
+	const gchar *opt)
 {
-	insert_result_common (task, symbol, flag, opts, TRUE);
+	insert_result_common (task, symbol, flag, opt, TRUE);
 }
 
 gboolean
diff --git a/src/libmime/filter.h b/src/libmime/filter.h
index 6c61f5d6f..99e9c78ae 100644
--- a/src/libmime/filter.h
+++ b/src/libmime/filter.h
@@ -19,7 +19,7 @@ struct rspamd_classifier_config;
  */
 struct symbol {
 	double score;                                   /**< symbol's score							*/
-	GList *options;                                 /**< list of symbol's options				*/
+	GHashTable *options;                            /**< list of symbol's options				*/
 	const gchar *name;
 	struct rspamd_symbol_def *def;					/**< symbol configuration					*/
 };
@@ -57,7 +57,7 @@ struct metric_result * rspamd_create_metric_result (struct rspamd_task *task,
 void rspamd_task_insert_result (struct rspamd_task *task,
 	const gchar *symbol,
 	double flag,
-	GList *opts);
+	const gchar *opts);
 
 /**
  * Insert a single result to task
@@ -70,7 +70,7 @@ void rspamd_task_insert_result (struct rspamd_task *task,
 void rspamd_task_insert_result_single (struct rspamd_task *task,
 	const gchar *symbol,
 	double flag,
-	GList *opts);
+	const gchar *opts);
 
 /**
  * Default consolidation function for metric, it get all symbols and multiply symbol
diff --git a/src/libserver/protocol.c b/src/libserver/protocol.c
index f33422dfe..ace286dbf 100644
--- a/src/libserver/protocol.c
+++ b/src/libserver/protocol.c
@@ -804,17 +804,21 @@ make_rewritten_subject (struct metric *metric, struct rspamd_task *task)
 }
 
 static ucl_object_t *
-rspamd_str_list_ucl (GList *str_list)
+rspamd_str_hash_ucl (GHashTable *ht)
 {
+	GHashTableIter it;
+	gpointer k, v;
 	ucl_object_t *top = NULL, *obj;
-	GList *cur;
 
 	top = ucl_object_typed_new (UCL_ARRAY);
-	cur = str_list;
-	while (cur) {
-		obj = ucl_object_fromstring (cur->data);
-		ucl_array_append (top, obj);
-		cur = g_list_next (cur);
+
+	if (ht) {
+		g_hash_table_iter_init (&it, ht);
+
+		while (g_hash_table_iter_next (&it, &k, &v)) {
+			obj = ucl_object_fromstring ((const char *)v);
+			ucl_array_append (top, obj);
+		}
 	}
 
 	return top;
@@ -841,8 +845,8 @@ rspamd_metric_symbol_ucl (struct rspamd_task *task, struct metric *m,
 				description), "description", 0, false);
 	}
 	if (sym->options != NULL) {
-		ucl_object_insert_key (obj, rspamd_str_list_ucl (
-				sym->options), "options", 0, false);
+		ucl_object_insert_key (obj, rspamd_str_hash_ucl (sym->options),
+				"options", 0, false);
 	}
 
 	return obj;
diff --git a/src/libserver/task.c b/src/libserver/task.c
index b9fca4d4a..6cf7532fd 100644
--- a/src/libserver/task.c
+++ b/src/libserver/task.c
@@ -1043,14 +1043,17 @@ rspamd_task_log_metric_res (struct rspamd_task *task,
 				}
 
 				if (lf->flags & RSPAMD_LOG_FLAG_SYMBOLS_PARAMS) {
-					GList *cur;
+					GHashTableIter it;
+					gpointer k, v;
 
 					rspamd_printf_fstring (&symbuf, "{");
-
 					j = 0;
+					g_hash_table_iter_init (&it, sym->options);
+
+					while (g_hash_table_iter_next (&it, &k, &v)) {
+						const char *opt = v;
 
-					for (cur = sym->options; cur != NULL; cur = g_list_next (cur)) {
-						rspamd_printf_fstring (&symbuf, "%s;", cur->data);
+						rspamd_printf_fstring (&symbuf, "%s;", opt);
 
 						if (j >= max_log_elts) {
 							rspamd_printf_fstring (&symbuf, "...;");
diff --git a/src/lua/lua_task.c b/src/lua/lua_task.c
index b43cbfeb5..8f28bb97b 100644
--- a/src/lua/lua_task.c
+++ b/src/lua/lua_task.c
@@ -2322,7 +2322,6 @@ lua_push_symbol_result (lua_State *L,
 	struct metric_result *metric_res;
 	struct symbol *s;
 	gint j;
-	GList *opt;
 
 	metric_res = g_hash_table_lookup (task->results, metric->name);
 	if (metric_res) {
@@ -2348,14 +2347,16 @@ lua_push_symbol_result (lua_State *L,
 			}
 
 			if (s->options) {
-				opt = s->options;
+				GHashTableIter it;
+				gpointer k, v;
+
 				lua_pushstring (L, "options");
-				lua_newtable (L);
+				lua_createtable (L, g_hash_table_size (s->options), 0);
+				g_hash_table_iter_init (&it, s->options);
 
-				while (opt) {
-					lua_pushstring (L, opt->data);
+				while (g_hash_table_iter_next (&it, &k, &v)) {
+					lua_pushstring (L, (const char*)v);
 					lua_rawseti (L, -2, j++);
-					opt = g_list_next (opt);
 				}
 
 				lua_settable (L, -3);