Fix issues with compatibility tokenization.

2 files changed, 24 insertions, 13 deletions

diff --git a/src/libstat/stat_process.c b/src/libstat/stat_process.c
index 3ec579049..141b2dfd5 100644
--- a/src/libstat/stat_process.c
+++ b/src/libstat/stat_process.c
@@ -157,6 +157,7 @@ rspamd_stat_get_tokenizer_runtime (struct rspamd_tokenizer_config *cf,
 
 	tok = rspamd_mempool_alloc (task->task_pool, sizeof (*tok));
 	tok->tokenizer = rspamd_stat_get_tokenizer (name);
+	tok->tkcf = cf;
 
 	if (tok->tokenizer == NULL) {
 		return NULL;
@@ -168,7 +169,6 @@ rspamd_stat_get_tokenizer_runtime (struct rspamd_tokenizer_config *cf,
 
 	tok->config = conf;
 	tok->conf_len = conf_len;
-	tok->tkcf = cf;
 	tok->tokens = g_tree_new (token_node_compare_func);
 	rspamd_mempool_add_destructor (task->task_pool,
 			(rspamd_mempool_destruct_t)g_tree_destroy, tok->tokens);
@@ -295,13 +295,17 @@ rspamd_stat_preprocess (struct rspamd_stat_ctx *st_ctx,
 		}
 
 		cl_runtime->clcf = clcf;
+
 		bk = rspamd_stat_get_backend (clcf->backend);
 		if (bk == NULL) {
-			msg_warn ("backend of type %s is not defined", clcf->backend);
-			cur = g_list_next (cur);
-			continue;
+			g_set_error (err, rspamd_stat_quark(), 500,
+					"backend %s is not defined", clcf->backend);
+			g_list_free (cl_runtimes);
+			return NULL;
 		}
 
+		cl_runtime->backend = bk;
+
 		curst = st_list;
 		while (curst != NULL) {
 			stcf = (struct rspamd_statfile_config *)curst->data;
@@ -348,15 +352,14 @@ rspamd_stat_preprocess (struct rspamd_stat_ctx *st_ctx,
 					return NULL;
 				}
 			}
-			else {
-				if (!cl_runtime->tok->tokenizer->compatible_config (
-						cl_runtime->tok, tok_config, conf_len)) {
-					g_set_error (err, rspamd_stat_quark(), 500,
-							"incompatible tokenizer for statfile %s", stcf->symbol);
-					g_list_free (cl_runtimes);
 
-					return NULL;
-				}
+			if (!cl_runtime->tok->tokenizer->compatible_config (
+					cl_runtime->tok, tok_config, conf_len)) {
+				g_set_error (err, rspamd_stat_quark(), 500,
+						"incompatible tokenizer for statfile %s", stcf->symbol);
+				g_list_free (cl_runtimes);
+
+				return NULL;
 			}
 
 			st_runtime = rspamd_mempool_alloc0 (task->task_pool,
diff --git a/src/libstat/tokenizers/osb.c b/src/libstat/tokenizers/osb.c
index 40dec0d82..bbb2656c2 100644
--- a/src/libstat/tokenizers/osb.c
+++ b/src/libstat/tokenizers/osb.c
@@ -347,8 +347,16 @@ rspamd_tokenizer_osb_load_config (rspamd_mempool_t *pool,
 {
 	struct rspamd_osb_tokenizer_config *osb_cf;
 
-	if (ptr == NULL) {
+	if (ptr == NULL || len == 0) {
 		osb_cf = rspamd_tokenizer_osb_config_from_ucl (pool, rt->tkcf->opts);
+
+		if (osb_cf->ht != RSPAMD_OSB_HASH_COMPAT) {
+			/* Trying to load incompatible configuration */
+			msg_err ("cannot load tokenizer configuration from a legacy statfile,"
+					" maybe you have forgotten to set 'compat' option in the "
+					"tokenizer configuration");
+			return FALSE;
+		}
 	}
 	else {
 		g_assert (len == sizeof (*osb_cf));