jeffrey@algor2.algorists.com (Jeffrey Kegler) (09/13/89)
For those sites with a small and predictable number of news readers,
expiration on articles read based on the .newsrc of the people who
read news may make sense. The following script generates a list of
all news articles, passes it by the a .newsrc file, and removes those
not vouched for. It is freely redistributable.
The script below can easily be changed to handle multiple .newsrc
files by pulling out the stuff that checks the .newsrc and putting it
in a loop. As I am a one-person site, I have no way of testing that
result, so I leave it as an exercise to an interested party.
The location of my news spool directory is non-standard and will
probably have to be changed at your site. The method of "expiring"
articles (simply deleting them) while crude, was said by Henry Spencer
to be reasonable when I asked him about it (nonetheless any flames go
to me, not Henry). You will want to continue having date-based
expiration to keep your history file sane, but the extra space should
enable you to extend the length of time of your date-based expiration.
I have run this script for some weeks on algor2 without obvious ill
effects, but no warranty is given. It is clearly inappropriate for
many, perhaps most sites, but is distributed in the hope it will be as
worthwhile at a significant minority of them as it is at mine.
=== Start of newsclear ===
if test $# != 1
then
echo usage: $0 newsrc 1>&2
exit 1
fi
cd /news/spool
TMP=/tmp/$$.1
trap "rm -f $TMP; trap 0 1 2 3; exit 0" 0 1 2 3
sed -e '
/^options/d
/[^a-z0-9., :+-]/d
s/: */ /
s=\.=/=g' $1 |
sort -b +0 -1 > $TMP
dirlist="alt bionet biz comp control dc dr general gnu mail misc ne news pubnet rec sci soc talk u3b um unix-pc"
find $dirlist -type f -print |
sed -e 's=/\([0-9]*\)$= \1=p' |
sort -b +0 -1 |
join -a1 - $TMP |
nawk '
{
read=0
if (split($3, range, ",") == 0) read=1
for (sequence in range) {
split(range[sequence], end, "-")
if (end[2]+0 == 0 && $2+0 == end[1]) {
read=1
}
else if (end[1]+0 <= $2 && $2 <= end[2]+0) {
read=1
}
}
if (read == 0) next
printf("%s/%d\n", $1, $2)
}' | xargs rm -f
=== End of newsclear ===
--
Jeffrey Kegler, Independent UNIX Consultant, Algorists, Inc.
jeffrey@algor2.ALGORISTS.COM or uunet!algor2!jeffrey
1762 Wainwright DR, Reston VA 22090