I currently process sa-learn against two mailboxes in each user's
mailspool, sal-ham and sal-spam using a line like this:
/usr/local/bin/sa-learn --spam --mbox -u $i $S_PATH
where $i is the user name and $S_PATH is /home/${i}/Mail/sal-spam
Now, I want to process maildirs that are stored for virtual users
where the path will be to a maildir
/usr/local/virtual/[EMAIL PROTECTED]/.sal-spam/
I THINK that simply running
sa-learn --spam -u virtual $S_PATH
will do the trick, but I also want to MOVE those messages to a sal-
spam-learned folder so that the user knows that a message has been
learned (and a sal-ham-learned as well). I currently do this with
mbox files quite easily:
/usr/local/bin/sudo -u $i sh -c "cat $S_PATH >> ${S_PATH}-learned
&& \
cat /dev/null > $S_PATH";
But, naturally, that doesn't work with maildirs.
I also want to remove the markup from the messages marked as ham, so
I think I need to run spamassasin -d as well, but I'm unclear if that
will work over a maildir or if I have to feed it individual
messages. I currently do THIS via a procmailrc that trawls the sal-
ham mbox file and runs spamassassin-r over each message in it.
Anyone have any suggestions?
--
*** AgentSmith sets mode: +m
