t.bogoutil

一个C语言写的快速贝叶斯垃圾邮件过滤工具

#! /bin/sh -x

. ${srcdir:=.}/t.frame

#
# test common bogoutil functions
#

# initialize
rm ${TMPDIR}/* # reset TXN detection
if [ $DB_TXN = true ] ; then addarg=--db-transaction=yes ; else addarg= ; fi
echo bacon on toast | $BOGOFILTER -C -d ${TMPDIR} -n $addarg
echo foul eggs      | $BOGOFILTER -C -d ${TMPDIR} -s $addarg

# see that these do not crash
$BOGOUTIL -C -v -V > /dev/null
$BOGOUTIL -C -h    > /dev/null

# check if bogoutil can count ;-)
cat <<_EOF >${TMPDIR}/ref
7 tokens dumped
_EOF
$BOGOUTIL -C -v -d ${WORDLIST} -D | tail -1 > ${TMPDIR}/out
cmp -s ${TMPDIR}/out ${TMPDIR}/ref

# ROBX calculation cannot work, insufficient data
if $BOGOUTIL -C -r ${WORDLIST} ; then false ; fi
if $BOGOUTIL -C -R ${WORDLIST} ; then false ; fi

# load more data as ROBX fodder
$BOGOFILTER -xd -C -n -d ${TMPDIR} -M -B ${srcdir}/inputs/good.mbx
$BOGOFILTER -xd -C -s -d ${TMPDIR} -M -B ${srcdir}/inputs/spam.mbx

# printing robx must not store the token
$BOGOUTIL -C -r ${WORDLIST} | egrep '^ ?0?.[0-9]+$' >/dev/null
if $BOGOUTIL -C -w ${WORDLIST} .ROBX | egrep '^[[:alnum:]]' ; then
    false
fi

# storing robx must store the token, quite obviously :-)
# also use -p so we test both -p and -w
$BOGOUTIL -C -R ${WORDLIST}
$BOGOUTIL -C -p ${WORDLIST} .ROBX | egrep '^\.ROBX' >/dev/null

# check if it can dump tokens from stdin
echo "head:toast" | $BOGOUTIL -C -p ${WORDLIST} \
| egrep '^head:toast' >/dev/null

if [ $DB_TYPE = db ] || [ $DB_TYPE = sqlite ] ; then
    ps=`$BOGOUTIL -C --db-print-pagesize ${WORDLIST}`
    test -n "$ps"
    test "$ps" != UNKNOWN
fi

if [ $DB_TYPE = db ] ; then
    $BOGOUTIL -C --db-verify ${WORDLIST}
    pc=`$BOGOUTIL -C --db-print-leafpage-count ${WORDLIST}`
    test "$pc" != ""
    test "$pc" != "UNKNOWN"
    test "$pc" != "0"

    if [ $DB_TXN = true ] ; then
	$BOGOUTIL -C --db-recover ${TMPDIR}/
	$BOGOUTIL -C --db-checkpoint ${TMPDIR}/
	if ls "${TMPDIR}/log."* >/dev/null 2>&1 ; then
	    $BOGOUTIL -C --db-prune ${TMPDIR}/
	    ls ${TMPDIR}/log.* >${TMPDIR}/logs.1
	    $BOGOUTIL -C --db-list-logfiles "${TMPDIR}" all >"${TMPDIR}/logs.2"
	    cmp ${TMPDIR}/logs.1 ${TMPDIR}/logs.2
	fi
	if ls "${TMPDIR}/__db."* >/dev/null 2>&1 ; then
	    $BOGOUTIL -C --db-remove-environment ${TMPDIR}/
	    if ls "${TMPDIR}/__db."* >/dev/null 2>&1 ; then
		false
	    fi
	fi
    fi
fi

#test with wordlist in current directory
BOGOUTIL="$VAL `pwd`/${relpath}/bogoutil$EXE_EXT"
WORDLIST="wordlist.${DB_EXT}"
OPC=$PRINTCORE
case $srcdir in
    /*)	PRINTCORE=${srcdir}/$INSERT/printcore ;;
    *)	PRINTCORE=`pwd`/$srcdir/$INSERT/printcore ;;
esac

unset BOGOFILTER_DIR

cd ${TMPDIR}
$BOGOUTIL -C -w $WORDLIST .MSG_COUNT >/dev/null
$BOGOUTIL -C -p $WORDLIST .MSG_COUNT >/dev/null
PRINTCORE=$OPC
cd - >/dev/null