diff options
| author | Vsevolod Stakhov <vsevolod@highsecure.ru> | 2021-04-06 15:39:19 +0100 |
|---|---|---|
| committer | Vsevolod Stakhov <vsevolod@highsecure.ru> | 2021-04-06 15:39:19 +0100 |
| commit | d07edfa195c7e166cee4b30e7dea6c17a12e169b (patch) | |
| tree | 0f5b8f779effd2225eda0433bae6962b566970f9 /lualib/lua_stat.lua | |
| parent | ebc1286073fb291de96adc97e5305234cfb3b6c0 (diff) | |
| download | rspamd-d07edfa195c7e166cee4b30e7dea6c17a12e169b.tar.gz rspamd-d07edfa195c7e166cee4b30e7dea6c17a12e169b.zip | |
[Minor] Disable headers hash for statistics
Discussed with: @moisseev
Diffstat (limited to 'lualib/lua_stat.lua')
| -rw-r--r-- | lualib/lua_stat.lua | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/lualib/lua_stat.lua b/lualib/lua_stat.lua index 2606a8d22..920f20205 100644 --- a/lualib/lua_stat.lua +++ b/lualib/lua_stat.lua @@ -678,6 +678,10 @@ local function get_mime_stat_tokens(task, res, i) end local function get_headers_stat_tokens(task, cf, res, i) + --[[ + -- As discussed with Alexander Moisseev, this feature can skew statistics + -- especially when learning is separated from scanning, so learning + -- has a different set of tokens where this token can have too high weight local hdrs_cksum = task:get_mempool():get_variable("headers_hash") if hdrs_cksum then @@ -686,6 +690,7 @@ local function get_headers_stat_tokens(task, cf, res, i) res[i]) i = i + 1 end + ]]-- for k,hdrs in pairs(cf.classify_headers_parsed) do for _,hname in ipairs(hdrs) do |