#!/usr/bin/perl -w

# $Id: ufile,v 1.2 2002/05/21 20:29:12 jason Exp jason $

# This takes sha1sum(1), sha(1) or md5sum(1) input and prints filenames of
# unique files to stdout.  Use -v (like grep(1)) to get non-unique files
# instead, and -a to get all fields, not just the filename.

# FUTURE:  accept md5(1), openssl(1), gpg(1), and digest(1) input?

use Getopt::Std;
require "flush.pl";

$count = 0;
$count_unique = 0;

$print_dupes = 0;
$print_all = 0;

%opts = ();

getopts ('va', \%opts) || die "$!";

if (defined ($opts{'v'})) {
    $print_dupes = 1;
}
if (defined ($opts{'a'})) {
    $print_all = 1;
}


while (<STDIN>) {

    next if (/^#/); # skip comments

    $count++;

    @vals = split;
    if ($#vals == 2) { # sha1sum(1) format, hash + size + filename
	# NB:  We might miss some matches if the filesizes aren't _properly_
	#      reported, but at least we won't blindly fall prey to preimage
	#      attacks.
	$hash = $vals[0] . "\t" . $vals[1];
	$file = $vals[2];
    } else { # lesser formats, hash + filename
	$hash = $vals[0];
	$file = $vals[1];
    }

    if ($print_all) {
	chomp;
	$output = $_;
    } else {
	$output = $file;
    }

    if (! exists ($hashes{$hash})) { # use first file, not last...
	$hashes{$hash} = $file;
	$count_unique++;
	print $output, "\n" if (!$print_dupes);
    } elsif ($print_dupes) {
	print $output, "\n";
    }

}

flush (STDOUT);
print STDERR "info:  $count_unique unique file(s), $count total (",
    $count - $count_unique, " duplicate(s)).\n";