#!/bin/bash
#
# Teach SpamAssassin from Maildirs
# Junk and Inbox folders.
#

MYSQL_HOST="localhost"
MYSQL_DATABASE="mail"
MYSQL_USERNAME="username"
MYSQL_PASSWORD="password"
MYSQL_BIN="/usr/bin/mysql"
SALEARNBIN="/usr/bin/sa-learn"
MAILDIR="/var/mail/virtual"
JUNKDIR=".Junk"

echo "Updating SpamAssassin Bayesian spam/ham filter"

# Get all enabled domains from the MySQL database
${MYSQL_BIN} --skip-column-names -u${MYSQL_USERNAME} -p${MYSQL_PASSWORD} -h${MYSQL_HOST} ${MYSQL_DATABASE} -e "SELECT domain FROM domain WHERE enabled='1'" | while read DOMAIN; do
    # Check if the domain exists and hold useraccounts.
    if [[ "$(ls -A ${MAILDIR}/${DOMAIN} 2>/dev/null)" ]]; then
        echo "Will go through all accounts in ${MAILDIR}/${DOMAIN}."
        ACCOUNTS=(`ls -A ${MAILDIR}/${DOMAIN}`)
        for USER in "${ACCOUNTS[@]}"; do
            # Check if ${USER} have a ${JUNKDIR} folder.
            if [[ -e "${MAILDIR}/${DOMAIN}/${USER}/${JUNKDIR}" ]]; then
                # Learning HAM and SPAM!
                echo "Will teach SpamAssassin from \"${USER}\" ${JUNKDIR}-folder!"
                HAM=${MAILDIR}/${DOMAIN}/${USER}/{cur,new}
                SPAM=${MAILDIR}/${DOMAIN}/${USER}/${JUNKDIR}/{cur,new}

                ${SALEARNBIN} --no-sync --spam ${SPAM} | while read LINE; do
                    echo "${LINE}"
                done

                ${SALEARNBIN} --no-sync --ham ${HAM} | while read LINE; do
                    echo "${LINE}"
                done
            else
                # Learning HAM only!
                echo "User \"${USER}\" does not have any ${JUNKDIR}-folder yet."
                HAM=${MAILDIR}/${DOMAIN}/${USER}/{cur,new}

                ${SALEARNBIN} --no-sync --ham ${HAM} | while read LINE; do
                    echo "${LINE}"
                done
            fi
        done
    else
        echo "Domain ${DOMAIN} exist, but doesn't seem to hold any useraccounts."
    fi
done
echo ""
echo "Syncing..."
${SALEARNBIN} --sync | while read LINE; do
    echo "${LINE}"
done

echo ""
echo "Statistics:"
sa-learn --dump magic | while read LINE; do
    echo "${LINE}"
done