[Minor] Add more diagnostic for fuzzy hashes lookup

2 files changed, 46 insertions, 18 deletions

diff --git a/src/libserver/fuzzy_backend_redis.c b/src/libserver/fuzzy_backend_redis.c
index b77addea4..d9d1cc0a7 100644
--- a/src/libserver/fuzzy_backend_redis.c
+++ b/src/libserver/fuzzy_backend_redis.c
@@ -86,6 +86,7 @@ struct rspamd_fuzzy_redis_session {
 	gchar **argv;
 	gsize *argv_lens;
 	struct upstream *up;
+	guchar found_digest[rspamd_cryptobox_HASHBYTES];
 };
 
 static inline void
@@ -416,10 +417,11 @@ rspamd_fuzzy_redis_shingles_callback (redisAsyncContext *c, gpointer r,
 					session->argv_lens[2] = 1;
 					session->argv[3] = g_strdup ("F");
 					session->argv_lens[3] = 1;
-					session->argv[3] = g_strdup ("C");
-					session->argv_lens[3] = 1;
+					session->argv[4] = g_strdup ("C");
+					session->argv_lens[4] = 1;
 					g_string_free (key, FALSE); /* Do not free underlying array */
-					memcpy (rep.digest, sel->digest, sizeof (rep.digest));
+					memcpy (session->found_digest, sel->digest,
+							sizeof (session->cmd->digest));
 
 					g_assert (session->ctx != NULL);
 					if (redisAsyncCommandArgv (session->ctx,
@@ -476,7 +478,7 @@ rspamd_fuzzy_backend_check_shingles (struct rspamd_fuzzy_redis_session *session)
 	struct rspamd_fuzzy_reply rep;
 	const struct rspamd_fuzzy_shingle_cmd *shcmd;
 	GString *key;
-	guint i;
+	guint i, init_len;
 
 	rspamd_fuzzy_redis_session_free_args (session);
 	/* First of all check digest */
@@ -487,10 +489,13 @@ rspamd_fuzzy_backend_check_shingles (struct rspamd_fuzzy_redis_session *session)
 
 	session->argv[0] = g_strdup ("MGET");
 	session->argv_lens[0] = 4;
+	init_len = strlen (session->backend->redis_object);
 
 	for (i = 0; i < RSPAMD_SHINGLE_SIZE; i ++) {
-		key = g_string_new (session->backend->redis_object);
-		rspamd_printf_gstring (key, "_%d_%uL", i, shcmd->sgl.hashes[i]);
+
+		key = g_string_sized_new (init_len + 2 + 2 + sizeof ("18446744073709551616"));
+		rspamd_printf_gstring (key, "%s_%d_%uL", session->backend->redis_object,
+				i, shcmd->sgl.hashes[i]);
 		session->argv[i + 1] = key->str;
 		session->argv_lens[i + 1] = key->len;
 		g_string_free (key, FALSE); /* Do not free underlying array */
@@ -557,7 +562,7 @@ rspamd_fuzzy_redis_check_callback (redisAsyncContext *c, gpointer r,
 
 			if (found_elts >= 2) {
 				rep.v1.prob = session->prob;
-				memcpy (rep.digest, session->cmd->digest, sizeof (rep.digest));
+				memcpy (rep.digest, session->found_digest, sizeof (rep.digest));
 			}
 
 			rep.ts = 0;
@@ -630,6 +635,8 @@ rspamd_fuzzy_backend_check_redis (struct rspamd_fuzzy_backend *bk,
 	session->command = RSPAMD_FUZZY_REDIS_COMMAND_CHECK;
 	session->cmd = cmd;
 	session->prob = 1.0;
+	memcpy (rep.digest, session->cmd->digest, sizeof (rep.digest));
+	memcpy (session->found_digest, session->cmd->digest, sizeof (rep.digest));
 	session->ev_base = rspamd_fuzzy_backend_event_base (bk);
 
 	/* First of all check digest */
diff --git a/src/plugins/fuzzy_check.c b/src/plugins/fuzzy_check.c
index c504e7603..0a3d3803a 100644
--- a/src/plugins/fuzzy_check.c
+++ b/src/plugins/fuzzy_check.c
@@ -1804,6 +1804,7 @@ fuzzy_insert_result (struct fuzzy_client_session *session,
 	guchar buf[2048];
 	const gchar *type = "bin";
 	struct fuzzy_client_result *res;
+	gboolean is_fuzzy = FALSE;
 
 	/* Get mapping by flag */
 	if ((map =
@@ -1848,23 +1849,43 @@ fuzzy_insert_result (struct fuzzy_client_session *session,
 	}
 
 	res->score = nval;
-	msg_info_task (
-			"found fuzzy hash(%s) %*xs with weight: "
-			"%.2f, probability %.2f, in list: %s:%d%s",
-			type,
-			(gint)sizeof (rep->digest), rep->digest,
-			nval,
-			(gdouble)rep->v1.prob,
-			symbol,
-			rep->v1.flag,
-			map == NULL ? "(unknown)" : "");
+
+	if (memcmp (rep->digest, cmd->digest, sizeof (rep->digest)) != 0) {
+		is_fuzzy = TRUE;
+	}
+
+	if (is_fuzzy) {
+		msg_info_task (
+				"found fuzzy hash(%s) %*xs (%*xs requested) with weight: "
+						"%.2f, probability %.2f, in list: %s:%d%s",
+				type,
+				(gint) sizeof (rep->digest), rep->digest,
+				(gint) sizeof (cmd->digest), cmd->digest,
+				nval,
+				(gdouble) rep->v1.prob,
+				symbol,
+				rep->v1.flag,
+				map == NULL ? "(unknown)" : "");
+	}
+	else {
+		msg_info_task (
+				"found exact fuzzy hash(%s) %*xs with weight: "
+						"%.2f, probability %.2f, in list: %s:%d%s",
+				type,
+				(gint) sizeof (rep->digest), rep->digest,
+				nval,
+				(gdouble) rep->v1.prob,
+				symbol,
+				rep->v1.flag,
+				map == NULL ? "(unknown)" : "");
+	}
 
 	if (map != NULL || !session->rule->skip_unknown) {
 		rspamd_snprintf (buf,
 				sizeof (buf),
 				"%d:%*xs:%.2f:%s",
 				rep->v1.flag,
-				(gint)sizeof (rep->digest), rep->digest,
+				(gint)MIN(rspamd_fuzzy_hash_len, sizeof (rep->digest)), rep->digest,
 				rep->v1.prob,
 				type);
 		res->option = rspamd_mempool_strdup (task->task_pool, buf);