[Rspamd-Users] Rspamd statistic settings: How to set learned_ids when users_enabled is true?

[Gabriele Nencioni]

Hi,
we have set the statistic module in order to have the gloabl and peruser
classifier and we noticed that the keys name into learned_ids redis
field (hash) are the same for global and peruser classifier.

Here an example of taht problem:
~# cat 1644312420.15363.eml | rspamc -c peruser -d
hari at int.billibotton.net -t 90  learn_ham
Results for file: stdin (0.008 seconds)
success = true;
filename = "stdin";
scan_time = 0.008000;

~# cat 1644312420.15363.eml | rspamc -c global -t 90  learn_ham
Results for file: stdin (0.004 seconds)
error = "<20220207080348.008504 at rebelwaltz.it> has been already learned
as ham, ignore it";
filename = "stdin";
scan_time = 0.004000;

~# redis-cli monitor | grep learned_ids
1644505170.215501 [0 127.0.0.1:43620] "HGET" "learned_ids"
"1mpwf7m975idffxpnh9dmzd1qx5y1445kxtynogpyscu1isa457kn3fhmcrq15b7iqkgnf6485knpfcd7hkkkpq7wpmgxbuboaw9tun"
1644505170.217240 [0 127.0.0.1:43622] "HSET" "learned_ids"
"1mpwf7m975idffxpnh9dmzd1qx5y1445kxtynogpyscu1isa457kn3fhmcrq15b7iqkgnf6485knpfcd7hkkkpq7wpmgxbuboaw9tun"
"-1"

1644505182.578004 [0 127.0.0.1:43714] "HGET" "learned_ids"
"1mpwf7m975idffxpnh9dmzd1qx5y1445kxtynogpyscu1isa457kn3fhmcrq15b7iqkgnf6485knpfcd7hkkkpq7wpmgxbuboaw9tun"



So is there a way to set a different key name for each classifier?

Moreover how is the string of that key name generate? Or is there a
command to know it (given a certain message/email)?

(In our example we are talking about the string:
1mpwf7m975idffxpnh9dmzd1qx5y1445kxtynogpyscu1isa457kn3fhmcrq15b7iqkgnf6485knpfcd7hkkkpq7wpmgxbuboaw9tun,
as you can see it is the same for both classifier)


It follows our statistic module configuration
classifier {
    bayes {
        tokenizer {
            name = "osb";
        }
        name = "global";
        backend = 'redis';
        read_servers =
"172.20.235.35,172.20.235.36,172.20.235.37,172.20.235.38";
        write_servers = '172.20.235.32';
        timeout = 30s;
        new_schema = true;
        expire = 10368000;
        lazy = true;
        min_tokens = 11;
        min_learns = 200;
        languages_enabled = true;
        statfile {
            spam = false;
            symbol = "BAYES_HAM";
        }
        statfile {
            spam = true;
            symbol = "BAYES_SPAM";
        }
        learn_condition = <<EOD
return function(task, is_spam, is_unlearn)
  local learn_type = task:get_request_header('Learn-Type')

  if not (learn_type and tostring(learn_type) == 'bulk') then
    local prob = task:get_mempool():get_variable('bayes_prob', 'double')

    if prob then
      local in_class = false
      local cl
      if is_spam then
        cl = 'spam'
        in_class = prob >= 0.95
      else
        cl = 'ham'
        in_class = prob <= 0.05
      end

      if in_class then
        return false,string.format('already in class %s; probability
%.2f%%',
          cl, math.abs((prob - 0.5) * 200.0))
      end
    end
  end

  return true
end
EOD;
  .include(try=true; priority=1)
"$LOCAL_CONFDIR/local.d/classifier-bayes.conf"
  .include(try=true; priority=10)
"$LOCAL_CONFDIR/override.d/classifier-bayes.conf"

    }
}

classifier {
    bayes {
        tokenizer {
            name = "osb";
        }
        name = "peruser";
        backend = 'redis';
        read_servers =
"172.20.235.35,172.20.235.36,172.20.235.37,172.20.235.38";
        write_servers = '172.20.235.32';
        timeout = 30s;
        new_schema = true;
        expire = 10368000;
        lazy = true;
        min_tokens = 11;
        min_learns = 0;
        languages_enabled = true;
        users_enabled = true;
        per_user = true;
        statfile {
            symbol = "BAYES_HAM_USER";
            spam = false;
        }
        statfile {
            symbol = "BAYES_SPAM_USER";
            spam = true;
        }
        learn_condition = <<EOD
return function(task, is_spam, is_unlearn)
  local learn_type = task:get_request_header('Learn-Type')

  if not (learn_type and tostring(learn_type) == 'bulk') then
    local prob = task:get_mempool():get_variable('bayes_prob', 'double')

    if prob then
      local in_class = false
      local cl
      if is_spam then
        cl = 'spam'
        in_class = prob >= 0.95
      else
        cl = 'ham'
        in_class = prob <= 0.05
      end

      if in_class then
        return false,string.format('already in class %s; probability
%.2f%%',
          cl, math.abs((prob - 0.5) * 200.0))
      end
    end
  end

  return true
end
EOD;
  .include(try=true; priority=1)
"$LOCAL_CONFDIR/local.d/classifier-bayes.conf"
  .include(try=true; priority=10)
"$LOCAL_CONFDIR/override.d/classifier-bayes.conf"

    }


Thanks in advance
-- 
Gabriele Nencioni