#!/usr/bin/perl

#
# Nasty script for finding bits of office documents in a filesystem
# image. Assumes that the files are 256k or smaller. Increase $BLKS
# for larger files.
#
# Usage: cd scratchdir; docfind.pl < /dev/diskdev

$BSIZE = 8192;
$BLKS = 32;

while (defined($dp = nextblk())) {
	next unless (substr($$dp, 0, 4) eq "\xd0\xcf\x11\xe0");

	$dat = $$dp;
	@newblks = ();
	for (1..($BLKS - 1)) {
		$dp = nextblk();
		last unless defined($dp);
		push(@newblks, $dp);
		$dat .= $$dp;
	}
	unshift(@blks, @newblks);

	$n++;
	print "Writing $n.doc\n";
	open(FILE, ">$n.doc") || die "$n.doc: $!\n";
	print FILE $dat;
	close(FILE);
}

sub nextblk {
	my $blk = shift(@blks);
	return $blk if (defined($blk));

	my $data, $nread;
	$nread = read(STDIN, $data, $BSIZE);
	die "read: $!\n" unless (defined $nread);
	return undef if ($nread != $BSIZE);
	return \$data;
}