t.multiple.wordlists

一个C语言写的快速贝叶斯垃圾邮件过滤工具

#!/bin/sh

. ${srcdir=.}/t.frame

OUT=${TMPDIR}/test.out
CORRECT=${SYSTEST}/outputs/multiple.wordlists.ref

CF_DEFAULT=default.cf
CF_IGN_REG=ign_reg.cf
CF_IGN_USR_SYS=ign_usr_sys.cf
CF_MRG_USR_SYS=mrg_usr_sys.cf

# single wordlist
cat <<EOF > ${TMPDIR}/$CF_DEFAULT
bogofilter_dir=${TMPDIR}
EOF

# ignore list, single wordlist
cat <<EOF > ${TMPDIR}/$CF_IGN_REG
bogofilter_dir=${TMPDIR}
wordlist i,ign,ignore.${DB_EXT},1
wordlist r,wrd,wordlist.${DB_EXT},2
EOF

# ignore list, user & system wordlists
cat <<EOF > ${TMPDIR}/$CF_IGN_USR_SYS
bogofilter_dir=${TMPDIR}
wordlist r,sys,system.${DB_EXT},6
wordlist i,ign,ignore.${DB_EXT},4
wordlist r,usr,user.${DB_EXT},5
EOF

# merged user & system wordlists
cat <<EOF > ${TMPDIR}/$CF_MRG_USR_SYS
bogofilter_dir=${TMPDIR}
wordlist r,usr,user.${DB_EXT},7
wordlist r,sys,system.${DB_EXT},7
EOF

# files for building databases
cat <<EOF > ${TMPDIR}/ignore.txt
ignore
user_low
system_hi
EOF

cat <<EOF > ${TMPDIR}/wordlist.txt
.MSG_COUNT 1 1
.WORDLIST_VERSION 20040500 0
ignore 2 7
common 2 7
word_low 2 7
word_hi 7 2
EOF

cat <<EOF > ${TMPDIR}/user.txt
.MSG_COUNT 1 1
.WORDLIST_VERSION 20040500 0
ignore 2 8
common 2 8
user_low 2 8
user_hi 8 2
EOF

cat <<EOF > ${TMPDIR}/system.txt
.MSG_COUNT 10 10
.WORDLIST_VERSION 20040500 0
ignore 2 9
common 2 9
system_low 2 9
system_hi 9 2
EOF

cat <<EOF > ${TMPDIR}/test.ref
.MSG_COUNT 1 1
.WORDLIST_VERSION 20040500 0
system
EOF

# test message
cat <<EOF > ${TMPDIR}/message
ignore
common
user_low
system_hi
word_low
word_hi
user_low
user_hi
system_low
system_hi
message
EOF

# suppress version info in output
BFOPTS="-D -H -e -m0.1 -y 0 --stats-in-header=no"
BFHEAD='--header-format=%h: %c, tests=bogofilter, spamicity=%p'

# build wordlists
$BOGOUTIL -l ${TMPDIR}/ignore.${DB_EXT}   < ${TMPDIR}/ignore.txt
$BOGOUTIL -l ${TMPDIR}/wordlist.${DB_EXT} < ${TMPDIR}/wordlist.txt
$BOGOUTIL -l ${TMPDIR}/user.${DB_EXT}     < ${TMPDIR}/user.txt
$BOGOUTIL -l ${TMPDIR}/system.${DB_EXT}   < ${TMPDIR}/system.txt

(echo "#### 1: score message using each wordlist ####"; echo) >> $OUT

for DB in ${TMPDIR}/*.${DB_EXT} ; do
    N=`basename $DB .${DB_EXT}`
    printf "%-12s" $N: >> $OUT
    $BOGOFILTER $BFOPTS "$BFHEAD" -v -C --wordlist=r,wordlist,$N.${DB_EXT},1 -I ${TMPDIR}/message >> $OUT
done
echo >> $OUT

(echo "#### 1: score message using each config file ####"; echo) >> $OUT

for CFG in ${TMPDIR}/*.cf ; do
    N=`basename $CFG`
    echo Using $N: >> $OUT
    $BOGOFILTER $BFOPTS "$BFHEAD" -vvv -c $CFG -I ${TMPDIR}/message >> $OUT
    echo >> $OUT
done

(echo "#### Update user using $CF_IGN_USR_SYS ####"; echo) >> $OUT

# Add message to user.${DB_EXT} (first regular wordlist in CF_IGN_USR_SYS)

echo >> $OUT
echo common message ignore | $BOGOFILTER $BFOPTS "$BFHEAD" -n -c ${TMPDIR}/$CF_IGN_USR_SYS

(echo "#### 2: score message using each wordlist ####"; echo) >> $OUT

for DB in ${TMPDIR}/*.${DB_EXT} ; do
    N=`basename $DB .${DB_EXT}`
    printf "%-12s" $N: >> $OUT
    $BOGOFILTER $BFOPTS "$BFHEAD" -v -C --wordlist=r,wordlist,$N.${DB_EXT},1 -I ${TMPDIR}/message >> $OUT
done
echo >> $OUT

(echo "#### 2: score message using each config file ####"; echo) >> $OUT

for CFG in ${TMPDIR}/*.cf ; do
    N=`basename $CFG`
    echo Using $N: >> $OUT
    $BOGOFILTER $BFOPTS "$BFHEAD" -vvv -c $CFG -I ${TMPDIR}/message >> $OUT
    echo >> $OUT
done

# compare test output to reference output

if  [ $verbose -eq 0 ]; then
    diff $CORRECT $OUT
    cmp $CORRECT $OUT
else
    diff $DIFF_BRIEF $CORRECT $OUT
fi