From: Vsevolod Stakhov <vsevolod@rspamd.com>
Date: Thu, 14 Dec 2023 16:34:35 +0000 (+0000)
Subject: [Project] Try to do strings concatenation in C
X-Git-Tag: 3.8.0~33
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=65477c306436ca302709f32d15b975fbf4dc21eb;p=rspamd.git

[Project] Try to do strings concatenation in C
---

diff --git a/lualib/redis_scripts/bayes_classify.lua b/lualib/redis_scripts/bayes_classify.lua
index 1a4734cf5..e94f645fd 100644
--- a/lualib/redis_scripts/bayes_classify.lua
+++ b/lualib/redis_scripts/bayes_classify.lua
@@ -9,7 +9,6 @@ local output_ham = {}
 
 local learned_ham = tonumber(redis.call('HGET', prefix, 'learns_ham')) or 0
 local learned_spam = tonumber(redis.call('HGET', prefix, 'learns_spam')) or 0
-local prefix_underscore = prefix .. '_'
 
 -- Output is a set of pairs (token_index, token_count), tokens that are not
 -- found are not filled.
@@ -18,11 +17,11 @@ local prefix_underscore = prefix .. '_'
 if learned_ham > 0 and learned_spam > 0 then
   local input_tokens = cmsgpack.unpack(KEYS[2])
   for i, token in ipairs(input_tokens) do
-    local token_data = redis.call('HMGET', prefix_underscore .. token, 'H', 'S')
+    local token_data = redis.call('HMGET', token, 'H', 'S')
 
     if token_data then
       local ham_count = token_data[1]
-      local spam_count = tonumber(token_data[2]) or 0
+      local spam_count = token_data[2]
 
       if ham_count then
         table.insert(output_ham, { i, tonumber(ham_count) })
diff --git a/lualib/redis_scripts/bayes_learn.lua b/lualib/redis_scripts/bayes_learn.lua
index 7536f6808..244be43f6 100644
--- a/lualib/redis_scripts/bayes_learn.lua
+++ b/lualib/redis_scripts/bayes_learn.lua
@@ -12,7 +12,6 @@ local symbol = KEYS[3]
 local is_unlearn = KEYS[4] == 'true' and true or false
 local input_tokens = cmsgpack.unpack(KEYS[5])
 
-local prefix_underscore = prefix .. '_'
 local hash_key = is_spam and 'S' or 'H'
 local learned_key = is_spam and 'learns_spam' or 'learns_ham'
 
@@ -21,5 +20,5 @@ redis.call('HSET', prefix, 'version', '2') -- new schema
 redis.call('HINCRBY', prefix, learned_key, is_unlearn and -1 or 1) -- increase or decrease learned count
 
 for _, token in ipairs(input_tokens) do
-  redis.call('HINCRBY', prefix_underscore .. token, hash_key, 1)
+  redis.call('HINCRBY', token, hash_key, 1)
 end
\ No newline at end of file
diff --git a/src/libstat/backends/redis_backend.cxx b/src/libstat/backends/redis_backend.cxx
index 5e222c6cb..7bebeb1ed 100644
--- a/src/libstat/backends/redis_backend.cxx
+++ b/src/libstat/backends/redis_backend.cxx
@@ -646,11 +646,12 @@ void rspamd_redis_close(gpointer p)
  * Serialise stat tokens to message pack
  */
 static char *
-rspamd_redis_serialize_tokens(struct rspamd_task *task, GPtrArray *tokens, gsize *ser_len)
+rspamd_redis_serialize_tokens(struct rspamd_task *task, const gchar *prefix, GPtrArray *tokens, gsize *ser_len)
 {
 	/* Each token is int64_t that requires 10 bytes (2 int32_t) + 4 bytes array len + 1 byte array magic */
 	char max_int64_str[] = "18446744073709551615";
-	auto req_len = tokens->len * sizeof(max_int64_str) + 5;
+	auto prefix_len = strlen(prefix);
+	auto req_len = tokens->len * (sizeof(max_int64_str) + prefix_len + 1) + 5;
 	rspamd_token_t *tok;
 
 	auto *buf = (gchar *) rspamd_mempool_alloc(task->task_pool, req_len);
@@ -667,8 +668,8 @@ rspamd_redis_serialize_tokens(struct rspamd_task *task, GPtrArray *tokens, gsize
 	int i;
 	PTR_ARRAY_FOREACH(tokens, i, tok)
 	{
-		char numbuf[sizeof(max_int64_str)];
-		auto r = rspamd_snprintf(numbuf, sizeof(numbuf), "%uL", tok->data);
+		char *numbuf = (char *) g_alloca(sizeof(max_int64_str) + prefix_len + 1);
+		auto r = rspamd_snprintf(numbuf, sizeof(numbuf), "%s_%uL", tok->data);
 		*p++ = (gchar) ((r & 0xff) | 0xa0);
 
 		memcpy(p, &numbuf, r);