#!/bin/sh

#
# tool to train both spam classifiers:
# a) SpamAssasin's bayes filter
# b) CRM114
# with input mails from the filter.good and filter.spam IMAP folders
#

# source folders (in IMAP)
imap_spam="Spam.train_spam"
imap_good="Spam.train_good"
imap_user="mschuett"
username="vscan"

# working folders
train_dir_good=/var/amavis/crm114/train_good
train_dir_spam=/var/amavis/crm114/train_spam

for dir in $train_dir_good $train_dir_spam; do
        if [ -d $dir ]; then
                rm -f ${dir}/*
        else
                mkdir $dir
        fi
done

# find files and link into training dirs
echo -n "Linking Spam msgs: "
doveadm search -u $imap_user mailbox $imap_spam |
while read guid uid; do
        cacheid=`doveadm fetch -u $imap_user hdr mailbox-guid $guid uid $uid |
                fgrep X-CRM114-CacheID: | head -1 | sed -e 's/^X-CRM114-CacheID: sfid-\(2[^ ]*\) *$/\1/'`
        file=/var/amavis/crm114/reaver_cache/texts/$cacheid
        if [ -f $file ]; then
                ln -f $file $train_dir_spam && echo -n .
                # echo S $cacheid
        else
                #echo "warning: cannot find mail with guid '$guid' uid '$uid' and CacheID '$cacheid'"
                tempfoo=`mktemp $train_dir_spam/mail.${uid}.XXXXXXX`
                if [ $? -ne 0 ]; then
                        echo "$0: Can't create temp file, exiting..."
                        exit 1
                fi
                #echo "warning: cannot link mail with guid '$guid' uid '$uid' and CacheID '$cacheid' --> create $tempfoo"
                echo -n o
                doveadm fetch -u $imap_user text mailbox-guid $guid uid $uid \
                        | sed -e '1d' -e '$d' > $tempfoo
        fi
done
echo " "

echo -n "Linking Ham msgs: "
doveadm search -u $imap_user mailbox $imap_good |
while read guid uid; do
        cacheid=`doveadm fetch -u $imap_user hdr mailbox-guid $guid uid $uid |
                fgrep X-CRM114-CacheID: | head -1 | sed -e 's/^X-CRM114-CacheID: sfid-\(2[^ ]*\) *$/\1/'`
        file=/var/amavis/crm114/reaver_cache/texts/$cacheid
        if [ -f $file ]; then
                ln -f $file $train_dir_good && echo -n .
                #echo G $cacheid
        else
                #echo "warning: cannot find mail with guid '$guid' uid '$uid' and CacheID '$cacheid'"
                tempfoo=`mktemp $train_dir_good/mail.${uid}.XXXXXXX`
                if [ $? -ne 0 ]; then
                        echo "$0: Can't create temp file, exiting..."
                        exit 1
                fi
                echo -n o
                #echo "warning: cannot link mail with guid '$guid' uid '$uid' and CacheID '$cacheid' --> create $tempfoo"
                doveadm fetch -u $imap_user text mailbox-guid $guid uid $uid \
                        | sed -e '1d' -e '$d' > $tempfoo
        fi
done
echo " "

# ensure no mail is in both training folders
for goodfile in $train_dir_good/*; do
        cacheid=`basename $goodfile`
        if [ -e $train_dir_spam/$cacheid ]; then
                echo "Same file in CRM spam-dir and good-dir: $cacheid"
                echo "E-Mail info:"
                egrep '^(Date|From|To): ' $goodfile
                echo
                exit 1
        fi
done
echo linked files.

chown -R ${username}:${username} $train_dir_spam $train_dir_good

echo "run sa-learn --spam:"
su ${username} -c "sa-learn --local --progress --spam $train_dir_spam"
echo "run sa-learn --ham:"
su ${username} -c "sa-learn --local --progress --ham $train_dir_good"

# CRM114 mailtrainer.crm
cd /var/amavis/crm114
echo "run mailtrainer.crm:"
su ${username} -c "crm -u /var/amavis/crm114 mailtrainer.crm --spam=$train_dir_spam/ --good=$train_dir_good/ --repeat=2 --random --worst=3 --thick=10"

echo
echo "clean dirs..."
rm $train_dir_good/* $train_dir_spam/*