On 31/05/06, Yves Goergen <nospam.list@???> wrote:
> I'm running an Exim server (4.6x) for multiple users, with a MySQL
> database backend. The database holds local domains, mail addresses,
> passwords, maildirs and a flag whether spam scanning is enabled for that
> account. SpamAssassin is called at SMTP time to scan the message and add
> its headers. But to be able to determine whether scanning should be
> done, I need to lookup the recipient address in the database and check
> that flag. This can of course only be done with a single recipient
> address at a time. And scanning itself can only be done when the message
> has been received (DATA ACL). So my scanner currently only runs if the
> current message should be delivered to one single RCPT TO. Today I
> received several mails in my mailboxes that seem to have gone to
> multiple envelope addresses, because my scanner didn't catch them.
>
> I want to make my spam scanner more effective. Is there any way to check
> the flag for multiple addresses in the DATA ACL and control the
> processing for each recipient independently? Maybe I can deny further
> recipients to a message at RCPT ACL time when one of them has the spam
> scanner flag set. Or can I generally limit the recipients per message
> count to 1? Would that cause any problems?
There's a well-documented method of restricting recipients to 1, or
more cleverly, to 1 'class' - ie you don't mind multiple recipients if
they all need the same processing. As soon as you see a subsequent
recipient of a different class from the first, you defer it.
http://www.exim.org/eximwiki/ExiscanExamples - the whole page is worth
a read, the specific examples are down the bottom.
Peter
--
Peter Bowyer
Email: peter@???
