t.lexer.mbx

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

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

OUT="lexer.mbx.out"

for f in good spam ; do
    $BOGOLEXER -C -D -p $v < ${srcdir}/inputs/$f.mbx > $TMPDIR/$f.1
    sort < $TMPDIR/$f.1 | uniq > $TMPDIR/$f.2
    if  [ $verbose -ne 0 ]; then
	$BOGOLEXER -C -D -x ml -vvv -p < ${srcdir}/inputs/$f.mbx > $TMPDIR/$f.v
    fi
done

RESULT=`cat $TMPDIR/spam.2 | wc -l`.`cat $TMPDIR/good.2 | wc -l`
RESULT=`echo $RESULT | sed s@\ @@g`

WANT="1787.4046"

if [ "$RESULT" != "$WANT" ] || [ $verbose -ne 0 ] ; then
    echo want: $WANT, have: $RESULT | tee $TMPDIR/$OUT
fi

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

test "$RESULT" = "$WANT"

### Test binary attachment in first message followed by text attachment in second message
### Starting with v0.96.3 the second message is incorrectly parsed
### The problem is fixed in v0.96.6

cat >$TMPDIR/msg.1.txt <<EOF
From [e-mail] Wed Nov 09 13:30:28 2005
Content-Type: multipart/mixed;
	boundary="----=_NextPart"

------=_NextPart
Content-Type: application/octet-stream;
	name="data.zip"

UEsDBAoAAAAAAHVjaTOjiB3egHMAAIBzAABUAAAAZGV0YWlscy50eHQgICAgICAgICAgICAg
LnBpZlBLBQYAAAAAAQABAIIAAADycwAAAAA=

------=_NextPart--

EOF

cat >$TMPDIR/msg.2.txt <<EOF
From [e-mail] Wed Nov 09 19:49:24 2005
Content-Type: text/plain;
	charset="iso-8859-1"

Ein hei遝 geile
Ein hei遝 geile

EOF

cat $TMPDIR/msg.1.txt $TMPDIR/msg.2.txt >  $TMPDIR/msg.12.mbx

(   $BOGOLEXER -C -D -p < ${TMPDIR}/msg.1.txt ; \
    $BOGOLEXER -C -D -p < ${TMPDIR}/msg.2.txt  ) > $TMPDIR/txt.out
(   $BOGOLEXER -C -D -p < ${TMPDIR}/msg.12.mbx ) > $TMPDIR/mbx.out

cmp ${TMPDIR}/txt.out ${TMPDIR}/mbx.out \
    || diff ${TMPDIR}/txt.out ${TMPDIR}/mbx.out