dupes.rb

#!/usr/bin/ruby
require 'digest/md5'
require 'find'

@hashfun = Digest::MD5.new
@files = Hash.new()
@filenames = Hash.new()
@hashes = Array.[]
def hashsum(filename)
	file = File.new(filename)
	contents = file.read()	
	return @hashfun.hexdigest(contents)
end


def checkdupe(filename)
	filehash  = @files[filename]
	for hash in @hashes
		if filehash == hash
			puts filename + " and " + @filenames[hash] + " are duplicate."
		end
	end
end

puts "Enter dir name to search:"
dir = gets.strip()

dirs = [dir]
for dir in dirs
	Find.find(dir) do |path|
		if FileTest.directory?(path)
		else
			hash = hashsum(path)
			@files[path] = hash
			@filenames[hash] = path
			checkdupe(path)
			@hashes << hash
		#	puts @files[path]
		end
	end
end