#!/usr/bin/perl -l

use strict;
use warnings;
use File::Find;
use Digest::MD5;
use Getopt::Std;

sub takemd5($);

my %opt;
getopts 'i:o:', \%opt;

@ARGV=grep { -d or !warn "`$_': not a directory!\n" } @ARGV;
die <<"EOD" unless @ARGV;
Usage: $0 [options] <dir> [<dirs>]

  -i <file>  read chached info from <file>
  -o <file>  write cached info to <file>
EOD

my %files;

if ($opt{i}) {
    for ($opt{i}) {
        -f or next;
        open my $fh, '<', $_ or
          die "Can't open `$_': $!\n";
        while (<$fh>) {
            chomp;
            /(\d+)\s{2}([0-9a-z]{32})\s{2}(.*)/ or
              warn "`$opt{i}': line $. not in the correct format\n" and
              next;
            next if -f $3;
            $files{$1}{$2}=1;
        }
    }
}

find { no_chdir => 1,
       preprocess => sub {
           sort {lc $a cmp lc $b} @_;
       },
       wanted => sub {
           return unless -f and my $sz=-s;
           for my $f ($files{$sz}) {
               $f=$_ and return unless $f;
               $f={takemd5 $f => $f} unless ref $f;
               my $md5=takemd5 $_;
               if ($f->{$md5}) {
                   unlink $_ and
                     print "Removing `$_'" or
                     warn "Can't remove `$_': $!\n";
               } else {
                   $f->{$md5}=$_;
               }
           }
       } }, @ARGV;

if ($opt{o}) {
    open my $fh, '>>', $opt{o} or
      die "Can't open `$opt{o}' for updating: $!\n";
    for my $k (keys %files) {
        next unless ref $files{$k};
        print $fh $k, '  ',
          $_, '  ', $files{$k}{$_} for
          keys %{$files{$k}};
    }
}

sub takemd5($) {
    my $f=shift;
    open my $fh, '<:raw', $f or
      warn "Couldn't open `$f': $!\n";
    Digest::MD5->new->addfile($fh)->hexdigest;
}

__END__