[Dovecot] How to use filters with the LDA?

[Peter Fern]

martin f krafft wrote:
> also sprach Peter Fern <dovecot at obfusc8.org> [2006.06.28.1101 +0200]:
>   
>>> I am trying to deploy dovecot LDA and liking it so far. However,
>>> I need to pass the message through some filters before it's
>>> delivered to the user (e.g. spamc). I realise I can just set up
>>> a chain (postfix delivers to spamc, pipes to foo, pipes to bar,
>>> pipes to deliver), but there's no reliable error handling with that
>>> approach.
>>>   
>>>       
>> I do this using postfix, dspam via LMTP and deliver, and it does indeed
>> have reliable error handling.
>>     
>
> Yeah, but this is not per-user, right? dspam does not need per-user
> because of it's cookie support, but I have some filters here that
> aren't this "smart".
>   

Correct, dspam is a little smarter than your average filter, if you're
attached to your current filters, perhaps look into amavis and see if it
will do the job for you?

>>> What would be really sweet is an LDA plugin that can call arbitrary
>>> filters for each message it processes. Has something (obvious) like
>>> this been done?
>>>   
>>>       
>> I don't believe this has been coded, but the new deliver uses sieve as a
>> plugin, so I don't see why this shouldn't be possible.  You could always
>> write one ;)
>>     
>
> Ha! Thanks for the confidence in my coding skills.
>
> Does the sieve plugin use libsieve or the sieve binary?
>   

It's a version of libsieve, but there's no reason a plugin couldn't call
an external tool, like the original dspam plugin for imap if you're
looking for an example.