#!/usr/bin/env ruby
# dupremove.rb
# Jason Jones
# 1/2/2006
# Usage: dupremove.rb DIRECTORY (optional)
#        Finds all duplicate items starting at
#		 DIRECTORY. If DIRECTORY is not given, 
#		 the script's directory is used.

require 'find'

def removeDupes(dir)
Find.find(dir) {|x|
	if File.ftype(x) == "directory"
		if File.basename(x)[0]==?.
			Find.prune
		else
			next
		end
	else
		if File.readable?(x)
			hash = `md5sum "#{x}"`
			if @h.key?(hash)
				@a << x
			else
				@h[hash] = x
			end
		end
	end	
}
end

def report()
	puts @a.length.to_s + " duplicate items found"
	if @a.length > 0
		@a.each{|x| puts x}
		puts "Remove (y/n)?"
		option = STDIN.gets().chomp!
		if option == "y"
			@a.each{|x| File.delete(x)}
		end
	end
end

@h = Hash.new
@a = Array.new
usedir = if(ARGV[0]) then ARGV[0] else Dir.getwd end

removeDupes(usedir)
report()