#!/usr/bin/perl
use strict;
use warnings;
use Digest::MD5;
use FileHandle;
use Getopt::Std;
use Cwd;

my $context=new Digest::MD5;
my %filelist;

sub md5sum ($);
sub processdirectory ($);

sub md5sum ($)
{
	my $testfile = shift;
	my $filehandler=new FileHandle "<$testfile";
	binmode($filehandler);
	$context->reset();
	$context->addfile($filehandler);
	return $context->hexdigest();
}

sub processdirectory ($)
{
  my $targetdir = shift;
  my $startdir = cwd;
  my $dirhandle;

  printf "processing $targetdir\n";
  chdir ($targetdir);

  opendir($dirhandle, ".") || die "couldn't open directory";
  my @direntries = readdir($dirhandle);

  foreach my $filename (@direntries)
  {
	if (($filename ne "\.") && ($filename ne "\.\."))
	{
	  	if (-d($filename))
	        {
	        	processdirectory("$filename");
	        }
		elsif (not -l($filename))
		{
			my $fullfilename = cwd . "/" . $filename;
			$filelist{$fullfilename} = (stat($filename))[7];
		}
	}
  } #end foreach

  closedir($dirhandle);
  chdir ($startdir);
} # processdirectory

# main
my %potentialduplicates;
my $lastfilesize = -1;
my $lastfilehash = "none";
my $lastfilename;
my $file;

foreach my $direntry (@ARGV)
{
        processdirectory($direntry);
}

foreach $file (sort {$filelist{$b} <=> $filelist{$a} } keys %filelist)
{
	if ($lastfilesize == $filelist{$file})
	{
		$potentialduplicates{$file} = md5sum($file);
		$potentialduplicates{$lastfilename} = md5sum($lastfilename);
	}
	$lastfilename = $file;
	$lastfilesize = $filelist{$file};
}

foreach $file (sort {$potentialduplicates{$b} cmp $potentialduplicates{$a} } keys %potentialduplicates)
{
	if ($lastfilehash eq $potentialduplicates{$file})
	{
		print $file . " and " . $lastfilename . " are duplicates with a hash of " . $lastfilehash . "\n";
	}

	$lastfilename = $file;
	$lastfilehash = $potentialduplicates{$file};
}