#!/usr/bin/perl -w

sub help
{
	print "$0 [-link] files...\n\n",
		"Outputs groups of duplicate files.\n",
		"You can't have commas in filenames. Sorry.\n",
		"-link outputs bash commands to update all duplicates from the first\n",
		"one.\n",
		"-unique outputs a list of non-duplicates and first member of each set of\n",
		"duplicates\n",
		"-T reads list from a file, -T - from stdin\n",
		"-dedupe outputs bash commands to remove duplicates.\n\n";
	exit;
}

help unless @ARGV;

@options=qw(link unique dedupe);

$mode="default";

# Options are mutually exclusive
for(@options){
	if($ARGV[0] eq "-".$_){
		$mode=$_;
		shift @ARGV;
		last;
	}
}

if($ARGV[0] eq "-T"){
	open(F,"<$ARGV[1]") or die("-T must be followed by a valid filename: $!\n");
	@filenames=<F>;
	close F;
	map(chomp,@filenames);
} else{
	@filenames=@ARGV;
}

for(@filenames){
	my $x=qx{cksum $_};
	@c=split /\s+/,$x;
	my $cksum=$c[0];
	$cksums{$cksum}.="$_,";
}

for(keys %cksums){
	my $has_dupes=($cksums{$_}=~/(,.*){2}/);
	push @unique,$cksums{$_} unless $has_dupes;
	delete $cksums{$_} unless $has_dupes;
}

map(s/,$//,@unique);

for(keys %cksums){
	my @files=split/,/,$cksums{$_};
	if($mode eq "link"){
		my $src=shift @files;
		for(@files){
			print "cp $src $_\n";
		}
		print "\n";
	} elsif($mode eq "unique"){
		# Just print the first one for each
		my $src=shift @files;
		print "$src\n";
	} elsif($mode eq "dedupe"){
		shift @files;
		for(@files){
			print "rm -f $_\n";
		}
	} else{
		print join "\n",@files;
		print "\n\n";
	}
}

# Now for the ones that had no duplicates at all
if($mode eq "unique"){
	print join "\n",@unique;
	print "\n";
}