Re: [Exim] Saving rejected mail

Page principale
Supprimer ce message
Répondre à ce message
Auteur: Greg Ward
Date:  
À: exim-users
Sujet: Re: [Exim] Saving rejected mail
On 06 September 2002, Philip Hazel said:
> The RCPT ACL doesn't reject messages; it rejects addresses. If all
> addresses are rejected, the server doesn't even get to see the contents
> of the message.


Right. I think I have figured out a nicer way to do this; it involves
turning "deny" ACL statements into "warn", and then later checking for
the header added by the warning.

For example, the first statement in my RCPT ACL is this:

  # If the sender address or domain is listed in spammer-*.cdb,
  # reject all recipients.
  deny    senders = @@cdb;/etc/exim/spammer-domains.cdb : \
                    cdb;/etc/exim/spammer-addrs.cdb
          message = message from known spammer rejected


In my alternative RCPT ACL -- to be used when I want to save all
rejected mail -- I use this instead:

  warn    senders = @@cdb;/etc/exim/spammer-domains.cdb : \
                    cdb;/etc/exim/spammer-addrs.cdb
          message = X-reject: known-spammer; \
                    message from known spammer rejected


Then, my local_scan() function will look for an X-reject header. If
there, it will 1) save the message to .../known-spammer (folder name
taken from the header), and 2) return LOCAL_SCAN_REJECT with the message
"message from known spammer rejected" (also from the header).

I still have to do a fair amount of cut-n-paste coding, but at least now
it's all still in the Exim config file. Some ACL statements are trivial
to recode in Python -- eg. checking for 8-bit junk in the subject
header. But I have no desire to rewrite header syntax verification; I'm
sure it could be done, but exactly emulating Exim's behaviour is waaaay
more trouble than it's worth.

> The DATA ACL does have the message available while it is running. I
> guess you could use the ${run expansion item to sneak a copy of the -D
> file (which does exist at that point). The headers are only in main
> memory - but you can see them using $message_headers. Essentially this
> is the same trick as using local_scan().


Ooh, what an evil idea. I think looking for X-reject in local_scan() is
probably cleaner. Will keep this trick up my sleeve if I need it
though!

        Greg
--
Greg Ward <gward@???>                         http://www.gerg.ca/
Well, I didn't expect a sort of Spanish Inquisition!