t.grftest

一个C语言写的快速贝叶斯垃圾邮件过滤工具

#!/bin/sh

# Note:  When run via "make check", test output files are automatically deleted.
#	 When run from the command line, output files are left in directory
#	 grftest.YYYYMMDD.  This is useful when something is different.
#
#	 ./inputs  - test inputs
#	 ./outputs - known correct outputs
#		     esp. grftest.out
#
#	 grftest.YYYYMMDD:
#		directory containing individual output files

NODB=1 . ${srcdir=.}/t.frame

# Fisher config
cat <<EOF > ${TMPDIR}/cfg.f
robx=0.415
min_dev=0.1
ham_cutoff=0.0
EOF

# Fisher tristate
cat <<EOF > ${TMPDIR}/cfg.t
robx=0.415
min_dev=0.1
ham_cutoff=0.1
terse_format = %1.1c %d
spamicity_tags = s,h,u
spamicity_formats = %0.6f %0.6f %0.6f
EOF

V="v"	# verbosity level

map_rc()
{
    (
	set +e
	eval "$@"
	a=$?
	[ $a -eq 0 ] && exit 0
	[ $a -eq 1 ] && exit 0
	[ $a -eq 2 ] && exit 0
	exit $a
    )
}

run_test()
{
    id="$1"
    fil="$2"
    ver="$3"
    OUT="${TMPDIR}/tests/msg.$fil.$mth.$ver"
    OPTS="-t -$ver $CFG"
    map_rc $BOGOFILTER $OPTS <${SYSTEST}/inputs/msg.$id.txt >${TMPDIR}/tests-$alg 
    sed s/,.version=.*// < ${TMPDIR}/tests-$alg > $OUT
}

if [   -d ${TMPDIR}/tests ] ; then rm -f ${TMPDIR}/tests/* ; else : ; fi
mkdir -p ${TMPDIR}/tests

for mth in f t ; do
    BOGOFILTER_DIR="${TMPDIR}/words.$mth"
    CFG="-c ${TMPDIR}/cfg.$mth -y 0"
    export BOGOFILTER_DIR
    if [   -d $BOGOFILTER_DIR ] ; then rm -f $BOGOFILTER_DIR/* ; else : ; fi
    mkdir -p $BOGOFILTER_DIR
    $BOGOFILTER $CFG -s < ${SYSTEST}/inputs/spam.mbx
    $BOGOFILTER $CFG -n < ${SYSTEST}/inputs/good.mbx
    for f in $BOGOFILTER_DIR/*.${DB_EXT} ; do
	n=`basename $f .${DB_EXT}`
	$BOGOUTIL -d $BOGOFILTER_DIR/$n.${DB_EXT} > $BOGOFILTER_DIR/$n.txt
    done
#   [ $verbose -gt 0 ] && $BOGOUTIL -w $BOGOFILTER_DIR .MSG_COUNT
    #
    # run tests for msg.[1-8].txt
    #
    for msg in ${SYSTEST}/inputs/msg.?.txt ; do
	tst=`echo $msg | sed s@${SYSTEST}/inputs/msg.@@ | sed s@.txt@@`
	args="$tst $tst $V"
	run_test $args
    done
done

OUT="grftest.out"

printf "%2s     %-10s   %-10s\n" id fisher tristate > ${TMPDIR}/tests/$OUT
for out in ${TMPDIR}/tests/msg.?.t.$V ; do
    id=`echo $out | sed s@${TMPDIR}/tests/msg.@@ | sed s@.t.$V@@`
    f=`head -1 ${TMPDIR}/tests/msg.$id.f.$V | tr "HS" "hs"`
    t=`head -1 ${TMPDIR}/tests/msg.$id.t.$V`
    printf "%2s   %c %s   %c %s\n" $id $f $t >> ${TMPDIR}/tests/$OUT
done

if [ "$BF_SAVEDIR" ] ; then . ${srcdir}/t.save ; fi


if  [ $verbose -eq 0 ]; then
    cmp ${OUTPUTS}/$OUT ${TMPDIR}/tests/$OUT
else
    diff $DIFF_BRIEF ${OUTPUTS}/$OUT ${TMPDIR}/tests/$OUT
fi