#!/bin/bash

# sa-harvest
# Copyright 2006, Faisal N. Jawdat, faisal@faisal.com
# Version 0.71
# License
# Permission is granted to use and modify under the terms of the Apache
# License, Version 2.0, # as of January 2004.  Please read the license
# here:
# http://www.apache.org/licenses/LICENSE-2.0
#
# Special thanks to The Obscure Organization, http://www.obscure.org/
# for supporting this project.  Additional thanks to Chris Heiser, who
# helped with the autogenerated whitelist code.

# you really want this line.  or your sysadmin does.  unless you're the
# only user.  you might have to tweak it for your os.
alias sa-learn='/bin/nice -19 /usr/bin/sa-learn'

cd

echo 'Gathering configuration Information.'
echo ''

echo 'Ham:'
export HAMDIRS=''
for ham in `cat ~/.spamassassin/mail.ham`; do
	export HAMDIRS="$HAMDIRS $ham";
done
export HAMDIRS=`echo $HAMDIRS|sed -s 's/^ //'`
echo "$HAMDIRS"
echo ''

echo 'Sent:'
export SENTDIRS=''
for sent in `cat ~/.spamassassin/mail.sent`; do
	export SENTDIRS="$SENTDIRS $sent";
done
export SENTDIRS=`echo $SENTDIRS|sed -s 's/^ //'`
echo "$SENTDIRS"
echo ''

echo 'Spam:'
export SPAMDIRS=''
for spam in `cat ~/.spamassassin/mail.spam`; do
	export SPAMDIRS="$SPAMDIRS $spam";
done
export SPAMDIRS=`echo $SPAMDIRS|sed -s 's/^ //'`
echo "$SPAMDIRS"
echo ''

echo ''

# Sync once to clear any remaining issues

sa-learn --sync
echo ''

# Learn

echo 'Learning...'
echo ''
echo 'Ham:'
sa-learn --no-sync --ham $HAMDIRS
echo ''
echo 'Spam:  '
sa-learn --no-sync --spam $SPAMDIRS
echo ''


sa-learn --sync
echo ''

sa-learn --dump magic
echo ''


# Strip mailbox config hack so we don't try to grep flags.

export HAMDIRS="`echo $HAMDIRS | sed -e 's/ --mbox//' | sed -e 's/ --mbx//'`"
export SENTDIRS="`echo $SENTDIRS | sed -e 's/ --mbox//' | sed -e 's/ --mbx//'`"

# Get whitelist
echo -n 'Recalculating whitelist...'

# start building user_prefs
cp ~/.spamassassin/user_prefs.base ~/.spamassassin/user_prefs.tmp

# build transient whitelists for current ham and sent inboxes
grep -a -h -r -E '^From:.*@.*' $HAMDIRS | sed -n "s/^.*<\(.*\)>.*/\1/p" > ~/.spamassassin/tmp.whitelist.ham
grep -a -h -r -E '^To:.*@.*' $SENTDIRS | sed -n "s/^.*<\(.*\)>.*/\1/p" > ~/.spamassassin/tmp.whitelist.sent

# add the manual and cached addressbooks and the transient whitelist to create the candidate new addressbook
cat ~/.spamassassin/addressbook ~/.spamassassin/addressbook.cache ~/.spamassassin/tmp.whitelist.ham ~/.spamassassin/tmp.whitelist.sent | tr A-Z a-z | sort | uniq > ~/.spamassassin/tmp.candidate

# build whitelist.negative from current spam inboxes and existing addressbook.negative
grep -a -h -r -E '^From:.*@.*' $SPAMDIRS | sed -n "s/^.*<\(.*\)>.*/\1/p" >> ~/.spamassassin/tmp.negative.spam
cat  ~/.spamassassin/addressbook.negative  ~/.spamassassin/tmp.negative.spam | tr A-Z a-z | sort | uniq > ~/.spamassassin/tmp.negative

# subtract tmp.whitelist.negative from tmp.whitelist to create new cached whitelist
cat ~/.spamassassin/tmp.candidate | grep -v -f ~/.spamassassin/tmp.negative | tr A-Z a-z | sort | uniq >> ~/.spamassassin/addressbook.cache

# add the cached addressbook to the whitelist in user_prefs
cat ~/.spamassassin/addressbook.cache | sed -n "s/\(.*\)/whitelist_from \1/p" >> ~/.spamassassin/user_prefs.tmp

# new user prefs is built:  activate it
mv ~/.spamassassin/user_prefs.tmp ~/.spamassassin/user_prefs

# clean up
rm ~/.spamassassin/tmp.*

echo ' done.'
echo ''

echo 'Ham:'
echo
grep -h -r ^X-Spam-Level $HAMDIRS | sort | uniq -c
echo ''

echo 'Spam:'
echo
grep -h -r ^X-Spam-Level $SPAMDIRS | sort | uniq -c
echo ''

uptime