mbox_extractor.rb

require 'rubygems'
require 'mail'

Dir["./**/*.mbox"].each do |file|
  puts "processing #{file}"
  dir = File.basename file
  puts "placing into #{dir}"
  messages = File.read(file).split(/^From .*$/).reject{|m| m.blank? }.map {|m| Mail.read_from_string m }
  puts "found #{messages.length} messages"
  messages.each_with_index do |m, i|
    puts "  message #{i+1} of #{messages.length}"
    filename = "#{file}.#{i}.txt"
    puts "     writing to #{filename}"

    File.open(filename, 'w+b') do |f|
      f.write "CLEAN TEXT: \n"
      f.puts "Subject: #{m.subject.to_s}"
      f.puts "From: #{m.from.to_s}"
      f.puts "To: #{m.to.to_s}"
      f.puts "Date: #{m.date.to_s}"
      f.puts
      if m.multipart?
        f.puts m.text_part.decoded.to_s
      else
        f.puts m.body.to_s
      end
      f.puts "\n\nRAW SOURCE: \n"
      f.puts m.to_s
    end

    m.attachments.each_with_index do |attach, a|
      puts "    attachment #{a+1} of #{m.attachments.length}"

      File.open("#{file}.#{i}.#{attach.filename}", 'w+b') {|f| f.write attach.body.decoded }
    end
  end
end