#!/bin/sh
#
# $Header: distribute,v 1.18 88/09/18 22:26:23 lamy Exp $
#
# Distribute stdin (presumed to be a digest or direct-mail article from
# a mailing list) according to the following flags
#
# -n newsgroup		post article to specified newsgroup.
# -a file		append to the file specified.
# -m people		remail to people, people may be a filename
#			in which case the contents will be assumed
#			to be a list of recipients, one to a line.
#			note that in the case of a file all recipients
#			will appear on the headers, and therefore
#			an alias is best used for large lists.
# -d directory		store into a file named by volume and
#			issue under "directory".  Useful for digests.
# -am name		shorthand for '-a name -m /local/share/mail/lists/name'
# -dm name		shorthand for '-d name -m /local/share/mail/lists/name'
#
# Notes:
# - When using -a or -m alone, a full path is required, which does differ
#   with what -am and -dm expect (the name of the mailing list only).
# - If invoked from sendmail, all arguments will be lowercase only!
#
# Rayan Zachariassen - rayan@ai.toronto.edu
# touch-ups by Jean-Francois Lamy - lamy@ai.toronto.edu

#exec 2>&1
#set -x
umask 022

USAGE='distribute [-n newsgroup] [-a archivefile] [-d archivedir] [-m mailto]'

# could be the same site, though not necessarily
thissite=`domainname`.toronto.edu
newsserver=jarvis.csri.toronto.edu

# we keep the archives under our anonymous FTP directory so other people
# can get at them.
archdir=/local/ftp
cd $archdir

# this person receives bounces and such like.  Make it an alias as it
# will appear on the sender: line of messages.
admin=list-admin

PEOPLE=/local/share/mail/lists
LOG=/var/log/distribute

Mail=/usr/ucb/Mail
sendmail=/usr/lib/sendmail
cat=/bin/cat
rm=/bin/rm
# This program should be suid daemon, so that sensitive files can be
# protected.
blankcat=/local/lib/mail/bin/appendfile

#
# Parse arguments
#
ARTICLE=/tmp/dist$$
newsgroup=""
file=""
directory=""
mailto=""
case $# in
0|1)	echo Usage: $USAGE
	exit 1 ;;
esac
state=x
for i in $@
do
	case "$i" in
	-*)	state=$i ;;
	*)	case $state in
		-n)	newsgroup="$i"
			distribution="`expr $newsgroup : '\([a-z]*\)\.*'`"
			;;
		-n*)	newsgroup="$i"
			distribution="`expr $state : '-n\(.*\)'`"
			;;
		-a)	file="$i" ;;
		-d)	directory="$i" ;;
		-m)	mailto="$i" ;;
		-am)	file="$i"
			mailto="$PEOPLE/$i" ;;
		-dm)	directory="$i"
			mailto="$PEOPLE/$i" ;;
		x)	echo Usage: $USAGE
			exit 2 ;;
		esac
		state=x ;;
	esac
done
#
case "${newsgroup}${file}${directory}${mailto}" in
?*)	$cat - > $ARTICLE ;;
*)	exec $Mail -s "No options to /local/lib/mail/distribute" rayan
	exec /bin/mail rayan
	exit 4 ;;
esac
case $file in
?*)	$blankcat -lists/${file} $ARTICLE ;;
esac
case $directory in
?*)	eval `/usr/ucb/head -40 $ARTICLE \
	| /usr/bin/fgrep -i digest \
	| /bin/sed -n \
	-e 's/.*Vol[^0-9]*\([0-9][0-9]*\)[^0-9][^0-9]*\([0-9][0-9]*\)[^0-9]*/VOL=\1 NUM=\2/p'`
	$blankcat "-$directory/V${VOL}.${NUM}" $ARTICLE ;;
*)	directory="$archdir" ;;
esac

# extract return path from article
/local/bin/ed - $ARTICLE <<EOF
1s/From \\([^ ]*\\) .*/\\1/
1w $ARTICLE.from
1d
w
q
EOF

# forwarding via mail
if [ -s $ARTICLE.from ]; then
	path="`cat $ARTICLE.from`"
	case $mailto in
	*/*)	( cd $directory ;
		if [ -s $mailto ]; then
			$sendmail -f $path `cat $mailto` < $ARTICLE
		fi
		)
		;;
	?*)	$sendmail -f $path $mailto < $ARTICLE
		;;
	esac
else
	path="$newsserver!news"
fi

# forwarding to newsgroups
case $newsgroup in
?*)	case $path in
	*!*!*)	path="`expr $path : '.*!\([^!]*![^!]*\)$'`"
		case $path in
		*.*!*)	;;
		*)	path="`echo $path | sed 's/!/.uucp&/'`" ;;
		esac ;;
	*!*)	;;
	*)	path="`hostname`.toronto.edu!$path" 
	esac
# All this does is massage the headers so they look like what news
# software expects.  To:, Cc: and Resent-*: headers are masked.
# Reply-To: is turned into references, which is questionable (could
# just as well be dropped.
#
# The From: line is rewritten to use the "address (comments)" form
# instead of "phrase <route>" form our mailer uses.  Also, addresses
# with no "@domainname" are assumed to originate locally, and so are
# given a domain.
#
# The Sender: field below reflects the address of the person who
# maintains our mailing lists.  The Approved: field is in a special
# form, so that we can do bidirectional gatewaying.  Any message
# in a newsgroup that bears this stamp will not be fed into the
# matching mailing list.
	sed -n -e "1{i\\
		Path: $path
		}" \
	    -e ":a
		/^[Rr]eceived:/b r
		/^[Tt][Oo]:/s/^/Original-/
		/^[Cc][Cc]:/s/^/Original-/
		/^[Rr][Ee][Ss][Ee][Nn][Tt]-.*/s/^/Original-/
		s/^[Ii]n-[Rr]eply-[Tt]o:/References:/
		/^From:/{
			s/<\([^@]*\)>\$/<\1@$thissite>/
			s/^From:[ 	][	]*\(.*\)  *<\(.*\)>\$/From: \2 (\1)/
			}
		s/-[Ii]d:/-ID:/
		s/^\([^:]*:\)[	 ]*/\1 /
		/^\$/{i\\
			Newsgroups: $newsgroup\\
			Distribution: $distribution\\
			Sender: $admin@$thissite\\
			Approved: $newsgroup@mail.ai.toronto.edu
			b e
		}
		p
		n
		b a
		:r
		s/.*//g
		n
		/^[ 	]/b r
		b a
		:e
		p
		n
		b e" < $ARTICLE > ${ARTICLE}.news
# ${ARTICLE}.news is the article to be posted.  Any method could
# be used to do it, normally invoking "inews" or even "relaynews" 
# directly would work.  Our setup is peculiar in that the news server
# is another machine, for historical (hysterical?) reasons.
# It turns out that invoking nntp on each article puts more load on
# the news server than making up a fake batch and mailing it to them,
# so we do just that.
		set - `wc -c ${ARTICLE}.news`
		(echo "#! rnews $1" ; cat ${ARTICLE}.news) |
			sed -e 's/^/N/' |
			$sendmail -f"$admin" recnews@jarvis.csri
#
# update the logs
#
		messageid=`egrep -i '^message-id:[ 	]' $ARTICLE`
		messageid=`expr "$messageid" : '..........:[ 	]\(.*\)'`
		time=`date | awk '{print $2,$3,$4}'`
		echo $time $newsgroup $messageid >> $LOG
esac
$rm -f $ARTICLE $ARTICLE.from $ARTICLE.news
exit 0