commit 44c6c56: [Project] Add classify redis script

[Vsevolod Stakhov]

Author: Vsevolod Stakhov
Date: 2023-12-02 12:36:17 +0000
URL: https://github.com/rspamd/rspamd/commit/44c6c563c2faf5ca97646e92e8382b972d2d513f

[Project] Add classify redis script

---
 .luacheckrc                             |  1 +
 lualib/redis_scripts/bayes_classify.lua | 26 ++++++++++++++++++++++++++
 2 files changed, 27 insertions(+)

diff --git a/.luacheckrc b/.luacheckrc
index 353bee41d..7e48b8ee0 100644
--- a/.luacheckrc
+++ b/.luacheckrc
@@ -64,6 +64,7 @@ files['/**/lualib/redis_scripts/**'].globals = {
   'redis',
   'KEYS',
   'cjson',
+  'cmsgpack',
 }
 
 files['/**/src/rspamadm/*'].globals = {
diff --git a/lualib/redis_scripts/bayes_classify.lua b/lualib/redis_scripts/bayes_classify.lua
new file mode 100644
index 000000000..c2654e476
--- /dev/null
+++ b/lualib/redis_scripts/bayes_classify.lua
@@ -0,0 +1,26 @@
+-- Lua script to perform bayes classification
+-- This script accepts the following parameters:
+-- key1 - prefix for bayes tokens (e.g. for per-user classification)
+-- key2 - set of tokens encoded in messagepack array of int64_t
+
+local prefix = KEYS[1]
+local input_tokens = cmsgpack.unpack(KEYS[2])
+local output_spam = {}
+local output_ham = {}
+
+for i, token in ipairs(input_tokens) do
+  local token_data = redis.call('HMGET', prefix .. tostring(token), 'H', 'S')
+
+  if token_data then
+    local ham_count = tonumber(token_data[1]) or 0
+    local spam_count = tonumber(token_data[2]) or 0
+
+    output_ham[i] = ham_count
+    output_spam[i] = spam_count
+  else
+    output_ham[i] = 0
+    output_spam[i] = 0
+  end
+end
+
+return cmsgpack.pack({ output_ham, output_spam })
\ No newline at end of file