harvey/mail/lib/pipeto.lib

# create a /tmp for here documents
rfork en
bind -c /mail/tmp /tmp

# caller should set KEY

USER=`{echo $1 | sed 's/local!//;s/[+\-].*//'}
if(! ~ $#KEY 1)
	KEY=plan9

cd /mail/box/$USER
RECIP=$1
MBOX=$2
PF=/mail/box/$USER/_pattern
TMP=/mail/tmp/$pid.$sysname
BIN=/bin/upas
D=/mail/fs/mbox/1

# clean up files on exit
fn sigexit {
	rm -f $TMP.*
}

fn log {
	if(~ $#USER 1)
		echo `{date} $* >>/mail/box/$USER/_log >[2]/dev/null
}

# deliver mail to a local file
fn spool {
	if(~ $#* 0)
		_mbox=$MBOX
	if not
		_mbox=$1
	$BIN/deliver $RECIP $D/from $_mbox < $D/raw || exit $status
}

# spool but change the subject line to note spam
fn spool-tagged-spam {
	if(~ $#* 0)
		_mbox=$MBOX
	if not
		_mbox=$1
	{
		cat $D/rawheader | sed 's/^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]:/& SPAM:/'
		if(! grep -si '^subject:' $D/rawheader)
			echo 'Subject: SPAM: '
		echo
		cat $D/rawbody
	} | $BIN/deliver $RECIP $D/from $_mbox || exit $status
}


# forward mail to a list of addresses
fn forward {
	upasname=`{awk '{print $2}' $D/unixheader} cat $D/raw | upas/send $* || exit $status
}

# pipe mail through a command
fn pipe {
	if(~ $#* 0)
		exit 'bad pipe command'
	{cat $D/unixheader $D/raw; echo} | $* || exit $status
}

# add @domain to all unqualified addresses in the message
fn qualify {
	if(! ~ $#* 1){
		echo 'usage: qualify domain' >[1=2]
		exit bad-qualify
	}
	{
		sed 1q $TMP.msg
		cat $TMP.msg | sed 1d | upas/smtp -fh $1 a b
	} >$TMP.msg2 || exit $status
	mv $TMP.msg2 $TMP.msg || exit $status
	unmount /mail/fs
	upas/fs -pf $TMP.msg || exit $status
}

# classify message according to token-based white list 
fn tokenfilter {
	if($BIN/list check $PF $D/from $D/sender $D/replyto)
		echo match
	if not if(~ $status *!match*)
		echo !match
	if not if($BIN/token $KEY $D/subject)
		echo token
	if not
		echo new
}

# reject a message due to the token-based white list
fn tokenreply {
	TOKEN=`{upas/token $KEY}
	if(! ~ $#MAILTO 1)
		exit 'bad token reply: no MAILTO variable'
	{
		cat /mail/lib/token.msg |
		sed 's/TOKEN/'$TOKEN'/g;s/USER/'$USER'/g;s/MAILTO/'$MAILTO'/g'
		cat $D/raw
	} | upasname=/dev/null mail `{cat $D/replyto}
}

# add addresses in message to white list
fn listupdate {
	$BIN/list add $PF $D/from $D/to $D/cc $D/sender
}

# microsoft virus going around 9/22/2003
fn isvirus {
	virus=no
	if(grep -s '\.(exe|scr|bat|com)' $D/2/2/mimeheader >[2]/dev/null){
		s=`{ls -l $D/2/2/raw | awk '{print $6}'}
		if(~ $s 1440??) virus=yes
	}
	if(~ $virus yes)
		status=''
	if not
		status='not this virus'
}

# bayesian spam filter.  alternative to token.  see /mail/lib/setup.bayes
fn isspam {
	for(i in _prof.mbox _prof.spam _bounced){
		if(! test -f $i){
			echo 'need '^$i >[1=2]
			exit 'need '^$i
		}
	}
	{
		echo '# hash table'
		upas/msgcat $TMP.msg | upas/msgtok |
			grep -v '^....................(.*)	' |
			sed 's/$/	1/'
	} >$TMP.tok
	x=`{upas/bayes -k _prof.mbox _prof.spam ~ $TMP.tok | sed 's/_prof.//'}
	where=$x(1)
	prob=$x(2)
	prob1000=`{echo $prob '*1000' | hoc | sed 's/\..*//'}
	echo `{sed 's/^From ([^ ]+) (.*)/\2 from \1/' $D/unixheader} $x >>_bounced
	if(~ $where spam && test $prob1000 -lt 999)
		where=mbox
	upas/addhash -o _prof.$where _prof.$where 1 $TMP.tok 1
	if(~ $where spam)
		where=''
	status=$where
}


# save and parse the mail file
sed '/^$/,$ s/^From / From /' >$TMP.msg
upas/fs -pf $TMP.msg || exit $status