dupes

#!/usr/bin/perl

my %files=();
my $DUPES_DIR="++dupes";

if(-d $DUPES_DIR) {
	warn "Directory already exists: $DUPES_DIR\n";
	exit(1)
}

mkdir($DUPES_DIR);

# sort files by date
my @files = sort {(stat $a)[10] <=> (stat $b)[10]} <*>;
my $fileCount = scalar @files;
# for display
my $count=0;
my $lineCount=0;
my $dupeCount=0;
my $sectionCount = 160;

foreach my $file (@files) {
	next if -d $file;
	next unless -r $file;
	my $md5 =`md5 -q "$file"`;
	chomp($md5);
	if(defined($files{$md5})) {
		if(scalar @{ $files{$md5} } == 1) {
			$dupeCount++;
		}
		push @{ $files{$md5} }, $file;
		# print "\nln ${file} ${DUPES_DIR}/${file}\n";
		# print "pushed $file onto [ @{ $files{$md5} } ] for $md5\n";
	} else {
		$files{$md5} = [$file];
		# print "set files{$md5} = [ @{ $files{$md5} } ] ($file)\n";
	}
	print "." if $count % 2;
	$count++;
	if($count > ${sectionCount}) {
		$count=0;
		$lineCount++;
		print " - ", int($fileCount/${sectionCount}) - $lineCount, "\n";
	}
}

print("\n\n");

my $moveCount=$dupeCount;
foreach my $md5 (keys %files) {
	my @arr = @{ $files{$md5} };
	if(scalar @arr > 1) {
		print "${key}: @{ $files{$md5} }\n";
		foreach $file (@arr) {
			link "${file}", "${DUPES_DIR}/${md5}-${file}";
			print "${DUPES_DIR}/${md5}-${file}\n";
		}
	}
	if(scalar @arr > 1) {
		# skip the first element
		shift @arr;
		# move the rest to the trash
		# system "trash", @arr;
		print "rm @arr\n";
	}
}