TOUZEAU DAVID skrev, on 27-10-2007 01:29:
I try to implement dspam on a gateway/postfix.
This gateway does not store mailboxes but store a quarantine area.
Messages that not follow RFC SMTP or tagged by SpamAssassin or infected
by clamav will be stopped and moved automatically into a local folder on
the server.
I would like to use dspam and force it to learn all messages stored in
the quarantine folder in order to increase the detection rate..
But the dspam_learn tool use only mailbox format and not messages flat
files format .
I suspect that you mean that dspam uses individual messages for
retraining (thus including maildir format files) and can't act on
mailbox format messages.
There is no way to change this, AFAIK.
Is there a way or a tips to force dspam_learn to parse this kind of
messages structure...??
Use your scripting skills and a utility such as mb2md
(http://batleth.sapienti-sat.org/projects/mb2md/) to separate individual
messages and move these to a separate directory for dspam to learn.
At present dspam_train (3.8.0) needs non-spam (at least one message and
probably more) as well as spam. Previous versions (e.g. 3.6.2) had the
utility dspam_corpus which didn't need non-spam.
Best,
--Tonni
--
Tony Earnshaw
Email: tonni at hetnet dot nl
