Ticket #25: cleannewsgroups

#!/bin/perl
# cleannewsgroups.pl
# Copyright 1997-1999 Arthur Hagen
# This program is Freeware.  It's not GPL or Public Domain.

require 5.0;
$DEBUG=1;

# Change the path below!
$dbdir = '/usr/local/news/db';

chdir($dbdir)
  or die "Can't cd to '$dbdir'\n";

my %ng;
my @groups;

print "Reading active ...\n" if $DEBUG;
open(INFILE,'<active')
  or die "Cant find active file in '$dbdir'\n";
while (<INFILE>) {
    my ($group,@rest) = split;
    push @groups,$group;
}

print "Reading newsgroups ...\n" if $DEBUG;
open(INFILE,'<newsgroups')
  or die "Can't find newsgroups file in '$dbdir'\n";
while (<INFILE>) {
    chomp;
    my ($group,@rest) = split /\t/;
    my $desc = $rest[-1]; # Only the last entry, no need for extra tabs.
    # Remove duplicate (Moderated) comments
    while ($desc =~ s/\(Moderated\)(\s)+\(Moderated\)/(Moderated)/g)
      {};
    # Strip trailing spaces too
    $desc =~ s/(\s)+$//;
    $ng{$group} = $desc;
}
close INFILE;

print "Reading newsgroups.full ...\n" if $DEBUG;
if (open(INFILE,'<newsgroups.full')) {
    # No harm done if there aren't any newsgroups.full file.  It will
    # be created later.
    while (<INFILE>) {
        chomp;
        my ($group,@rest) = split /\t/;
        # Let newsgroups take precedence over newsgroups.full
        unless ($ng{$group}) {
            my $desc = $rest[-1];
            while ($desc =~ s/\(Moderated\)(\s)+\(Moderated\)/(Moderated)/g)
              {};
            $desc =~ s/(\s)+$//;
            $ng{$group} = $desc;
        }
    }
    close INFILE;
}

# Remove this part if you want your newsgroups file to be in the same
# order as your active file.  I prefer both to be sorted.
print "Sorting groups ...\n" if $DEBUG;
my @groupss = sort @groups;

print "Writing newsgroups.full.n ...\n" if $DEBUG;
open(OUTFILE,'>newsgroups.full.n');
foreach $group (sort keys %ng) {
    # One tab is enough.  Pretty-printing is for humans, not machines.
    print OUTFILE $group,"\t",$ng{$group},"\n";
}
close OUTFILE;

print "Writing newsgroups.n ...\n" if $DEBUG;
open(OUTFILE,'>newsgroups.n');
my $numgroups = scalar @groupss;
for ($i=0;$i<$numgroups;$i++) {
    my $group = $groupss[$i];
    if (my $desc = $ng{$group}) {
        print OUTFILE $group,"\t",$desc,"\n";
    }
}
close OUTFILE;

print "Done.\n" if $DEBUG;

exit 0;