#! /bin/sh
# Reports groups with more than $1 articles, >90% of all articles, or
# or >$2KB of articles with expiry dates more than 10 days after their
# arrival dates.  Defaults 15 and 500.
# =()<. ${NEWSCONFIG-@<NEWSCONFIG>@}>()=
. ${NEWSCONFIG-/usr/local/libexec/cnews/config}

PATH=$NEWSCTL/bin:$NEWSBIN:$NEWSPATH ; export PATH
umask $NEWSUMASK

case "$#" in
0)	minarts=15 ; minsize=500	;;
2)	minarts=$1 ; minsize=$2		;;
*)	echo "Usage: $0 [minarts minsize]" >&2 ; exit 2	;;
esac

cd $NEWSARTS
egrep -v '~-' $NEWSCTL/history |
	awk -F'	' 'NF > 2 {
		n = split($2, f, "~")
		if (f[2] != "-" && f[2]-f[1] > 864000) {
			n = split($3, f, " ")
			for (i = 1; i <= n; i++)
				print f[i]
		}
	}' | tr . / | xargs $NEWSBIN/sizeof -i | sed 's;/[^/]* ; ;' |
	( sort ; echo dummy ) |
	awk '{
		if ($1 != newsgroup) {
			if (newsgroup != "")
				print newsgroup, n, size
			n = 0
			size = 0
			newsgroup = $1
		}
		n++
		size += $2
	}' |
	while read group narts size
	do
		npresent=`ls -f $group | egrep '^[0-9]+$' | wc -l`
		echo $group $narts $npresent $size
	done | tr / . |
	awk '	{ bad = "" }
		$2 > '$minarts' {
			bad = bad ", " $2 " articles"
		}
		$2 > 0.9*$3 {
			bad = bad ", " $2 " out of " $3 " articles"
		}
		$4 > '$minsize'*1024 {
			bad = bad ", " int(($4+512)/1024) "KB"
		}
		bad != "" {
			print $1, "has", substr(bad, 3), "on long expiry"
		}'