#!/bin/sh
TESTROOT=$HOME/sa;			# Root of test heirarchy
SA=$TESTROOT/Mail-SpamAssassin-2.61;	# SA Tarball, extracted
SALIB=$SA/lib/Mail/SpamAssassin;	# Where Bayes.pm lives
MASSES=$SA/masses			# Where masses scripts live
REALHOME=$HOME/.spamassassin;		# My real home directory
					# (I tested on my workstation which doesn't actually 
					# process any mail so the script clobbers stuff in there)
MASSESHOME=$MASSES/spamassassin;	# masses "home" directory, where mass-check's bayes_* lives

# Try three bayes expiry sizes
for bayessize in 150000 250000 500000 ; do 

	# set bayes max tokens.
	echo bayes_expiry_max_db_size $bayessize > $HOME/.spamassassin/user_prefs

	# Three different bayes.pm files
	# dist is the Bayes.pm from the original SA tarball.
	# custom transforms all body tokens by sorting the letters in the words 
	#     and learning the resulting token, then removing any duplicate
	#     (consecutive) letters and learning the resulting token.
	# custom2 transforms all body tokens by sorting the letters in the words 
	#     then removing any duplicate (consecutive) letters and learning 
	#     the resulting token. (custom2 doesn't learn the in-between token
	#     which contains duplicate letters)

	for customtype in dist custom custom2 ; do
		
		# learn all mail older than 3 months ("old").
		
		# Then test against all mail 3 months old or newer ("new").
		# Second pass, test against all mail 30 days old or newer ("verynew")

		for mailage in new verynew ; do
			# Set Bayes implementation (with or without a customization)
			cp $SALIB/Bayes.$customtype.pm $SALIB/Bayes.pm;
	
			# pre-train bayes db with old spam and ham
			rm -f $REALHOME/bayes*;
			rm -f $MASSESHOME/bayes*;
			sa-learn --ham $TESTROOT/ham/old;
			sa-learn --spam $TESTROOT/spam/old;

			# Hack to learn mail with age between "old" and "verynew"
			if [ "verynew" = $mailage ] ; then
				sa-learn --ham $TESTROOT/ham/new_not_verynew;
				sa-learn --spam $TESTROOT/spam/new_not_verynew;
			fi

			sa-learn --force-expire; 	# try to trim the bayes DB down to $bayessize
							# (doesn't seem to do anything... whatever.)

			# Put the bayes databases into MASSESHOME/ for mass-check to use
			cp -R $REALHOME/bayes* $MASSESHOME/;
			# mass-check the mail newer than "$mailage"
			$MASSES/mass-check spam:dir:$TESTROOT/spam/$mailage ham:dir:$TESTROOT/ham/$mailage;
			# Do one hitfreq with percents and one with absolute values.
			$MASSES/hit-frequencies -x -p \
				| grep "\(OVERALL\|all messages\|BAYES\)" \
				| sort -k 7 \
				> freq.percent.${bayessize}.${mailage}.${customtype}.txt;
			$MASSES/hit-frequencies -x \
				| grep "\(OVERALL\|all messages\|BAYES\)" \
				| sort -k 7 \
				> freq.absolute.${bayessize}.${mailage}.${customtype}.txt;
		done
	done
done