#!/usr/bin/ruby

# Author: SkyOut
# Date: June 2008
# Website: http://wired-security.net/

# This little helper script compares two directory with each
# other recursively. If two files are the same (based on their
# md5 checksum) this gets echoed to the user (if "v" is set) and
# written to a output file (result.txt).

# How can this help?
# Imagine: You have found a vulnerability in a software of a hardware
# utility, such as a router (so: firmware) and you want to know if this
# part (for example a *.cgi script) is used in other firmwares as well, but
# might differ in the name or location. Then this helper utility comes in:
# Just download both firmwares and all scripts and check them against each
# other. If something is totally equal you will get to know it and can confirm
# the bug in more then only one firmware... Handy, isn't it? Sometimes yes!

# The required modules we will use later
# 'find' gets used for directory indexing
# 'digest/md5' gets used for checksum calculation
require 'find'
require 'digest/md5'

# Print the usage dialogue and quit the program
def usage()
  puts ""
  puts "ruby(.exe) dir2md5.rb [directory] [directory] [v]"
  exit(0)
end

# Make sure at least two arguments have been set
if (ARGV[0] == nil) || (ARGV[1] == nil)
  usage()
end

# Write the arguments given by the user to local variables (dir1, dir2, verbose (optional))
dir1 = ARGV[0]
dir2 = ARGV[1]
verbose = ARGV[2]

# Create the file for the first directory to be indexed
dir1file = File.open("#{dir1}/../dir1file.txt", File::WRONLY|File::CREAT|File::TRUNC, 0777)
# Create the file for the second directory to be indexed
dir2file = File.open("#{dir2}/../dir2file.txt", File::WRONLY|File::CREAT|File::TRUNC, 0777)
# Create the file, that will hold our result later on
outfile = File.open("result.txt", File::WRONLY|File::CREAT|File::TRUNC, 0777)

puts ""
puts "Indexing directory #{dir1}..."
puts ""
# Jump over . and ..
excludes = [".", ".."]
  # Go through the first directory
  Find.find(dir1) do |path|
    # Test if it is a directory
    if FileTest.directory?(path)
      # If the directory equals . or .. don't go deeper into it
      if excludes.include?(File.basename(path))
        Find.prune
      else
        # Go on...
        next
      end
    else
      # Calculate a checksum with md5 of the file found
      digest = Digest::MD5.hexdigest(File.read("#{path}"))
      if (verbose == "v")
        puts "#{path}:#{digest}"
      end
      # Write it to the indexing file of the first directory (dir1file.txt)
      dir1file.puts "#{path} #{digest}"
    end
  end
# Close the file handle
dir1file.close

puts ""
puts "Indexing directory #{dir2}..."
puts ""
# Jump over . and ..
excludes = [".", ".."]
  # Go through the second directory
  Find.find(dir2) do |path|
    # Test if it is a directory
    if FileTest.directory?(path)
      # If the directory equals . or .. don't go deeper into it
      if excludes.include?(File.basename(path))
        Find.prune
      else
        # Go on...
        next
      end
    else
      # Calculate a checksum with md5 of the file found
      digest = Digest::MD5.hexdigest(File.read("#{path}"))
      if (verbose == "v")
        puts "#{path}:#{digest}"
      end
      # Write it to the indexing file of the second directory (dir2file.txt)
      dir2file.puts "#{path} #{digest}"
    end
  end
# Close the file handle
dir2file.close

puts ""
puts "Comparing #{dir1} with #{dir2}..."
puts ""
# Open the first indexing file
dir1file2 = File.open("#{dir1}/../dir1file.txt", File::RDONLY)
begin
  # Go through the file line by line
  while (line = dir1file2.readline)
    # Get the filename
    fp = line
    fp = fp.split(" ")
    fp = fp.fetch(0)
    
    # Get the md5 hash value
    hash = line
    hash = hash.split(" ")
    hash = hash.fetch(1)
    # Open the second indexing file
    dir2file2 = File.open("#{dir2}/../dir2file.txt", File::RDONLY)
    begin
      # Go through the file line by line
      while (line2 = dir2file2.readline)
        # Get the filename
        fp2 = line2
        fp2 = fp2.split(" ")
        fp2 = fp2.fetch(0)
        
        # Get the md5 hash value
        hash2 = line2
        hash2 = hash2.split(" ")
        hash2 = hash2.fetch(1)
        
        # Check if the hash from dir1file equals the one of dir2file
        if (hash == hash2)
          if (verbose == "v")
            puts ""
            puts "#{fp} == #{fp2}"
          end
          # Write it to the resulting file!
          outfile.puts "#{fp} == #{fp2}"
        else
          if (verbose == "v")
            puts "Comparing #{fp}:#{hash} to #{fp2}:#{hash2}"
          end
          next
        end
      end
    # End of file reached, close it...
    rescue EOFError
      dir2file2.close
    end
  end
# End of file reached, close it...
rescue EOFError
  dir1file2.close
end

puts ""
puts "Finished... Result written to - result.txt -..."
puts ""

# JOB DONE!
exit(0)