#!/bin/bash

# crmtrain -- Script for training crm114 discriminator (spam filter)
#             with big amounts of e-mail messages, either as SPAM or
#             HAM. Use the "Export" command in Alpine mailer to store
#             SPAM as: s1, s2, s{3,4,5...} and HAM as: n1, n2, n{3,4,5...}
# Copyright (C) 2008 Adrian C. <anrxc_sysphere_org>
#           Licensed under the WTFPL


# Directories
spamdir=$HOME/spam
crmdir=$HOME/.crm114

# Mail filter
trainer=mailfilter.crm
#trainer=mailreaver.crm


if [[ -z $1 ]]; then
    echo "Usage: $0 [spam|ham] [num]"
    echo "Train crm114 with N spam or ham messages."
    exit 1
else
    # Determine the number of messages
    if [ $2 ]; then
         spamnum=$2
    else # Fallback to 1
         spamnum=1
    fi

    case "$1" in
        spam)
            for i in $(seq 1 $spamnum); do # SPAM is saved as sN
                # Learning spam, force to go around a known bug
                $crmdir/$trainer -u $crmdir --learnspam --force < $spamdir/s$i
                /bin/rm $spamdir/s$i
                sleep 2 # Don't rush things
            done
            ;;
        ham)
            for i in $(seq 1 $spamnum); do # HAM is saved as nN
                # Learning non-spam is better, don't force
                $crmdir/$trainer -u $crmdir --learnnonspam < $spamdir/n$i
                /bin/rm $spamdir/n$i
                sleep 2 # Don't rush things
            done
            ;;
        *)
            echo "Error: '$1' is not a valid e-mail tag."
            ;;
    esac
    exit 0
fi