run-masses   [plain text]

#!/bin/sh

# run-masses, Theo Van Dinter <felicity@kluge.net> (c) 2002
# $Id: run-masses,v 1.1 2004/11/29 21:55:37 dasenbro Exp $

#
# This script will run a mass-check against all mbox files in a given
# set of directories.  The expected directory structure is:
#
# spamassassin				SA version to use
# corpus				location of all corpus files
#  run-masses				this script
#  spam					directory of spam
#  ham					directory of ham
#
# By default, the script looks for "spamassassin-head" which on my
# machine is the latest version from CVS.  You can easily change this
# by passing parameters to this script as such:
#
# run-masses ../spamassassin-garun --progress
#
# The first parameter is the directory to use for mass-check, and all
# other parameters are passed to mass-check itself.
#
# At the end, you will have a ham.log, spam.log, and results.log in the
# corpus directory.
#
# BTW: cpucount is a small script to figure out the number
# of CPUs on the current machine.  It is available from
# http://www.kluge.net/~felicity/random/
#

# Setup the path as necessary
PATH=/bin:/usr/bin:/usr/local/bin:${HOME}/bin:.
if [ -d /sw/bin ]; then
	PATH=${PATH}:/sw/bin
fi
export PATH

# Use the specified directory for tests
if [ -z "$1" ]; then
	DIR=../spamassassin-head
else
	DIR="$1"
	shift
fi

# How many processes should we run at once?
PROCS=`cpucount`

# Where are our files located?
MASS=$DIR/masses
RULES=$DIR/rules

# do the mass-check
# class:format:path
# class = ham | spam
# format = file (file w/ single message) | mbox (file w/ multiple messsages) | dir (of 'file's)
$MASS/mass-check --all -c $RULES -j $PROCS "$@" `mbox-to-check`

echo "Generating hit frequency results"
$MASS/hit-frequencies -c $RULES -x -p -a > results.log

# remove the parse-rules-for-masses tmp directory
echo "Removing temporary files"
rm -rf ./tmp