Filtering mail based on header contents

[Olivier Nicole]

>I have a mail system on which I have recently implemented spamassassin with
>Pyzor, DCC and Razor. I am really happy with the tagging accuracy and am
>ready to start filtering mail. I know spamassassin can be configured to drop
>all mail with a score over a certain amount, but I am concerned about

NO SA only tag the mail, it never drops anything. Droping email is a
bad attitude if you don't send a bounce, and as it it impossible to
bounce a spam, then you should not drop spam email.

>dropping false positives. I would really prefer to either drop it in a
>folder for each user, or just send them all to a mailbox.
>
>The system we use has two tiers: mail enters the filtering server running
>sendmail, spamassassin Pyzor and DCC, which then sends to a
>qmail/courier-imap server. I would prefer to have the actual mailbox server
>drop the mail into a spam folder in each user's mailbox, but I realise that
>this could be a bit ambitious. Also, not all of the users use IMAP (about
>half use POP) so differentiating between IMAP and POP users is important.
>Does anyone have any suggestions?

I solved that by putting email tagged as spam into quarantine. Once a
day I submit to the users an email summary of their quarantine and
they can recover any message by replying.

>Failing that, is there an easy way to filter based on header content
>(Spamassassin score) in Sendmail on the filtering machine, or would I have
>to implement procmail or some such to redirect all spam to one mailbox?

But I do that with procmail, at the delivery level :)

http://www.cs.ait.ac.th/laboratory/email/quarantine.shtml

Best regards,

Olivier