Richard Freeman wrote:
Richard Freeman wrote:
I figured that a content filter would be the best approach, but I wasn't
sure that this wouldn't cause confusion with all the aliases. If it
will just magically figure everything out I can just give it a try.
Thanks to those on this list who gave me tips and to those on irc as
well - things are mostly working. I've run into a different problem though.
I'd like to be able to retrain by sending mail to [EMAIL PROTECTED] (and not
parse To lines). However, I also want to bounce my mail and not have to
put IDs in the footers. This means that neither the sender nor the
recipient of an email is going to contain the dspam user. Is there any
way to easily handle this? I think I actually have the [EMAIL PROTECTED]
address successfully going to dspam - but it looks like it just doesn't
do anything since it is trying to look for a user based on the from line
of the spam itself.
I'd have actually thought that the signature in the email would uniquely
identify the user. It looks like it does in the dspam_signature_data
table. I guess if I were desperate a script could do a query to figure
out who the user for an email actually is, but I'm surprised that dspam
doesn't already do this. Ie - if you're retraining with --source=error
why do you even need to specify --user?
Richard,
DSpam requires the signature be in the message somewhere, if you don't
want to see it in the footer then put it in the message headers. You can
also identify the user by turning on the MySQLUIDInSignature option.
Then you won't have to parse the To: headers. In addition the --user
option is pretty much null and void if dspam finds the UID in the
Signature. Regardless of what --user you use DSpam will train based upon
the UID it finds in the Signature. So set these options in your
dspam.conf and use any valid user for the --user like, --user globaluser.
Preference "signatureLocation=headers"
"MySQLUIDInSignature on"
--Jeff Harris
(LedHed)
