#!/usr/bin/perl

use strict;
use File::Spec;
use File::Compare;
use File::Copy;
use File::Basename;
use Cwd;

our $DEBUG=0;
if ($DEBUG){
 use Data::Dump qw(dump);
}

our $TMP="DUPFILES";
our $order=0;
our %files=();
our %dirs=();

sub find_dir {
 my ($a, $level) = @_;
 opendir(DIR, $a) or warn "Can't open directory '$a'";
 my @dl=readdir(DIR);
 closedir(DIR);
 foreach my $d (@dl){
  next if $d=~m/^\./o;
  next if $d eq $TMP;
  my $canon=File::Spec->canonpath(
   Cwd::realpath(
    File::Spec->catfile($a,$d)
   )
  );
  return if $dirs{$canon}++;
  if (-f $canon){
   unless (defined($files{$canon})){
    $files{$canon}=$level<<8 | $order++;
   }
  } elsif (-d $canon){
   find_dir($canon, $level+1);
  }
 }
}


foreach my $a (@ARGV){
 find_dir($a,0);
}

my @flist=sort {$files{$a} <=> $files{$b}} keys %files;
my %S;

foreach my $f (@flist){
 push @{$S{(stat($f))[7]}}, $f;
}

dump(%S) if $DEBUG;

while ( my ($size, $refnames) = each %S ){
 my @n=@{$refnames};
 my $ns=@n;
 for (my $j=0; $j < $ns-1; $j++){
  next unless $n[$j];
  for (my $i=1+$j; $i < $ns; $i++){
   next unless $n[$i];
   print "$n[0] : $n[$i]\n" if $DEBUG;;
   if (!File::Compare::compare($n[0],$n[$i])){
    print "dup: $n[$i] == $n[0]\n";
    $n[$i]=undef if myunlink($n[$i]);
   }
  }
 }
}

sub myunlink {
 my ($fn, $dir, $suff) = File::Basename::fileparse($_[0]);
 my $tmpdir="$dir$TMP";
 if (-d $tmpdir || mkdir($tmpdir)){
  return File::Copy::move($_[0], $tmpdir);
 } else {
  warn "mkdir($tmpdir): $!";
 }
 return 0;
}

__END__

=head1 NAME

dedup - Remove duplicate files from directories list

=head1 SYNOPSIS

    dedup.pl DIR1 DIR2 ..

=head1 DESCRIPTION

  This utility recursively passes through directories list hierarchy
  DIR1, DIR2, .. and leaves only one copy from identical files was found.
  Duplicate files are being moved to special directories
  DUPFILES for further revision and easy manual removing from