#!/usr/local/bin/perl
#
# history database sanity checker
# David Barr <barr@math.psu.edu>
# version 1.4
# w/mods from: hucka@eecs.umich.edu
# Katsuhiro Kondou <kondou@nec.co.jp>
# version 1.1
# Throw away history entries with:
#   malformed lines (too long, contain nulls or special characters)
#
# INN Usage:
#   ctlinnd throttle 'fixing history'
#   ./fixhist <history >history.n
#   makedbz -s `wc -l <history.n` -f history.n
#      or use instructions from fixhist to avoid the `wc -l <history.n`
#   mv history.n history
#   mv history.n.dir history.dir
### if TAGGED_HASH is DO or before inn2.0
#   mv history.n.pag history.pag
### if TAGGED_HASH is DONT
#   mv history.n.hash history.hash
#   mv history.n.index history.index
### endif
#   ctlinnd reload history x
#   ctlinnd go 'fixing history'
# any malformed entries will be output to stderr.


$MAXKEYLEN=254;
$count=0;

while (<>) {
	chop;
	($msgid,$dates,$arts,$xtra) = split('\t');
	if ($xtra) {
		&tossit();		# too many fields
		next;
	}
	if (!($dates) && (($arts) || ($xtra))) {
		&tossit();		# if not date field, then the rest
		next;			# should be empty
	}
	if (length($msgid) >= $MAXKEYLEN) {
		&tossit();		# message-id too long
		next;
	}
	if ($msgid !~ /^<[^<> ]*>$/) {
		if ($msgid =~ /^\[[0-9A-F]{32}\]$/) {
			if ($arts ne "") { 
				if ($arts =~ /^\@[0-9A-F]{56}\@$/) {
					$arts =~ s/^\@([0-9A-F]{36})([0-9A-F]{20})\@$/\@${1}\@/;
					print "$msgid\t$dates\t$arts\n";
					next;
				}
				if ($arts !~ /^\@[0-9A-F]{36}\@$/) {
					&tossit();
					next;
				}
			}
		} else {
			&tossit();		# malformed msg-ids
			next;
		}
	} else {
		if ($arts ne "" && ($arts !~ /[^\/]*\/[0-9]*/)) {
			&tossit();		# malformed articles list
			next;
		}
	}
	if (/[\000-\010\012-\037\177-\237]/) { # non-control chars except tab
		&tossit();		# illegal chars
		next;
	}
	if ($dates) {
		if ($dates =~ /[^\d~\-]/) {	# rudimentary check
			&tossit();		# full check would be too slow
			next;
		}
	}
	print "$_\n";
	$count++;
	$0 = "history line $./$count" if $. % 50000 == 0;
}
print STDERR "Done.  Now run:\nmakedbz -s $count -f history.n\n";

sub tossit {
	print STDERR "$_\n";
}