t.mh

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
#	 MH.YYYYMMDD.  This is useful when something is different.
#
#	 ./inputs/spam.mbx & ./inputs/good.mbx  - test inputs
#	 ./outputs - known correct outputs
#
#	 MH.YYYYMMDD:
#		directory containing individual output files

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

OUT="${TMPDIR}/MH.out"
CORRECT="${OUTPUTS}/MH.out"

for f in good spam ; do
    mkdir $TMPDIR/$f
    cat ${srcdir}/inputs/$f.mbx | (cd $TMPDIR/$f && splitmbox )
done

$BOGOFILTER -C -D -n $v -B $TMPDIR/good
$BOGOFILTER -C -D -s $v -B $TMPDIR/spam
wordlist=".MSG_COUNT Lotus Magazine Min Service connection more most set that unsubscribe"

$BOGOUTIL -C -w $WORDLIST $wordlist >$OUT || rm -f $OUT

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