this is obsolete doc -- see http://doc.nethence.com/ instead

Setting up server-side Bayesian mail filtering with procmail and QSF 

http://pbraun.nethence.com/unix/mail/procmail.html 

http://pbraun.nethence.com/unix/mail/procmail-bmf.html 

http://pbraun.nethence.com/unix/mail/procmail-qsf.html 

Configuring Procmail and QSF 

Now we are going to setup the Local Delivery Agent for Sendmail, namely the good old procmail, and with the QSF spam filter on top of it. I prefer QSF over BMF because it handles MIME and (hopefully) base64 encoded messages too. I also prefer QSF over bogofilter because it doesn't complain about non-existing databases at first run and lets the messages pass through anyway. 

Make sure those packages are installed, 

  echo $PACKAGESITE 

  pkg_add -r procmail qsf 

Prepare the Maildir folders, 

mkdir -p ~MAILUSER/Maildir/

chown MAILUSER:mail ~MAILUSER/Maildir/

Configure Procmail system-wide, 

cd /etc/

ftp -a http://pbraun.nethence.com/configs/mail/procmailrc

I actually use QSF on a per-user basis only for now. So for QSF to work you also configure Procmail per-user e.g., 

  cd ~MAILUSER/ 

  ftp -a http://pbraun.nethence.com/configs/mail/.procmailrc 

  vi .procmailrc 

add this, 

#

# Bayesian spam filter

#

:0 wf

| /usr/local/bin/qsf -ra

:0 H

* X-Spam: YES

.spam/

Note. you don't need to enable MAILUSER's shell, procmail SHELL=/bin/ksh and qsf still works without it. 

Daily QSF learning & unlearning 

Create the mailboxes for that, 

cd ~MAILUSER/Maildir/

mkdir -p .spam/

mkdir -p .spam.learn/

mkdir -p .spam.unlearn/

cd /home/

chown -R MAILUSER:mail MAILUSER/

chmod 700 MAILUSER/

Fetch the scripts to reprocess a Maildir format mailbox, 

mkdir -p ~MAILUSER/bin/

cd ~MAILUSER/bin/

ftp -a http://pbraun.nethence.com/scripts/mail/reprocess_maildir

ftp -a http://pbraun.nethence.com/scripts/mail/reprocess_maildir.learn

chmod +x reprocess_maildir*

Provide the procmail configuration needed by reprocess_maildir.learn, ".procmailrc.learn", 

SHELL=/bin/ksh

DROPPRIVS=yes

VERBOSE=no

ORGMAIL=$HOME/Maildir/

MAILDIR=$HOME/Maildir

DEFAULT=$ORGMAIL

SYSYEAR=`date +%Y`

LOGFILE=$HOME/.procmailrc.log.$SYSYEAR

#

# QSF learn

#

:0

| qsf -a -m

and ".procmailrc.unlearn", 

SHELL=/bin/ksh

DROPPRIVS=yes

VERBOSE=no

ORGMAIL=$HOME/Maildir/

MAILDIR=$HOME/Maildir

DEFAULT=$ORGMAIL

SYSYEAR=`date +%Y`

LOGFILE=$HOME/.procmailrc.log.$SYSYEAR

#

# QSF unlearn

#

:0

| qsf -a -M

Configure conrtab for user MAILDIR, 

cd /etc/

vi crontab

add, 

0       4       *       *       *       MAILUSER  $HOME/bin/reprocess_maildir.learn

Note that if you evern need to re-process messages by yourself you need to enable MAILUSER's shell, at least temporarily, 

  chsh -s /bin/ksh MAILUSER 

and to disable it back again, 

  chsh -s /sbin/nologin MAILUSER