On Sun, 26 Jun 2005, Marc Haber wrote:
> Subject: Re: [exim] exiscan, spamassassin, and per-domain bayes database
>
> On Tue, 14 Jun 2005 10:23:50 -0400 (EDT), Dave Lugo
> <dlugo@???> wrote:
> >On Tue, 14 Jun 2005, Peter Bowyer wrote:
> >> > Are you looking at exim's spec.txt? Section 40.2 shows as "Scanning
> >> > with SpamAssassin" in both the 4.50 and 4.51 versions I'm looking
> >> > at.
> >>
> >> OK, section 40, the Exiscan docs.
> >>
> >> So in the example above, do the %d and %l get substituted at scan time
> >> with values passed from the spam= line? Or just the %d from the username
> >> passed?
> >
> >Both %d and %l.
>
> And which values do you suggest using in the spam= line, as there are
> not recipient-based variables available?
>
The first (or only) recipient can be extracted from $recipients. You
are completely correct, and as I've acknowledged once (prolly in another
thread ;) this doesn't work when you have multiple recipients.
For my home vanity domains, it works well, as I restrict the maximum
# of rcpt_to = 1 on inbound connections.
> Maybe, it would be a good idea to have $first_domain and
> $first_local_part always initialized, even on messages that have
> multiple recipients, so that at least one recipient (and the only one
> in the common case of a message with only one recipient) gets proper
> handling?
>
I get it this way, using macros (which break if there is more than
a single recipient):
RCPTD_DOM = ${lc:${extract{2}{@}{$recipients}}}
RCPTD_LOC = ${lc:${extract{1}{+@}{$recipients}}}
...and then use the spam acl option like so:
spam = RCPTD_LOC@RCPTD_DOM
--
--------------------------------------------------------
Dave Lugo dlugo@??? LC Unit #260 TINLC
Have you hugged your firewall today? No spam, thanks.
--------------------------------------------------------
Are you the police? . . . . No ma'am, we're sysadmins.
