Hello Will,
Will Smith <undertakingyou@???> (Di 25 Mär 2014 00:03:05 CET):
> I want to tune the Bayesian filter by training it using sa-learn. To do
> that I need emails that are valid and invalid. The way that I have it set
> up now is that exim receives all emails for a specific domain. It then
> scans using clam-av and spamassassin. It then sends the email off to a
> corporate exchange box (via a transport rule).
>
> What I am hoping to do, is have a rule that if an email is marked as spam
> that it cc's it to a local delivery box (I will make something like
> spam@???) and then I can run the sa-learn off of that box.
>
> So, questions are:
> Can I have adding a CC be part of a ACL rule when something is marked as
> spam?
> Is there a better way to do what I am looking for?
> Maybe I should BCC all email to an account, and then I get the ham and
> spam? Not sure.
your data acl:
…
warn spam = nobody:true
one of your first routers:
salearn:
driver = redirect
condition = ${if >{$spam_score_int}{50}}
data = spam@???, $local_part@$domain
repeat_use = no
verify = no
If your spam@… is a local destination, you might check
the shadow_transport feature, about that way:
begin transports
local:
driver = appendfile
…
shadow_transport = salearn
remote_smtp:
driver = smtp
…
shadow_transport = salearn
salearn:
driver = pipe
command = sa-learn …
Beware, I didn't cross check this with the documentation, it's completly
untested and might just be considered as a hint :)
Best regards from Dresden/Germany
Viele Grüße aus Dresden
Heiko Schlittermann
--
SCHLITTERMANN.de ---------------------------- internet & unix support -
Heiko Schlittermann, Dipl.-Ing. (TU) - {fon,fax}: +49.351.802998{1,3} -
gnupg encrypted messages are welcome --------------- key ID: 7CBF764A -
gnupg fingerprint: 9288 F17D BBF9 9625 5ABC 285C 26A9 687E 7CBF 764A -
(gnupg fingerprint: 3061 CFBF 2D88 F034 E8D2 7E92 EE4E AC98 48D0 359B)-
