histogram coding challenge

bigdata.rb

def simple(file)
  histogram = Hash.new.tap{|h| h.default = 0 }
  File.read(file).each_line do |line|
    histogram[line.split(',')[1]] += 1
  end
  histogram
end

def threading(file, thread_count=10)
  histogram = Hash.new.tap{|h| h.default = 0 }
  require 'thread'

  queue = Queue.new
  chunk_size = File.size(file) / thread_count

  thread_count.times do |i|
    Thread.new do
      chunk = Chunk.new(file, chunk_size, i * chunk_size)
      queue.push(chunk.parse)
    end
  end

  thread_count.times do
    histogram.merge!(queue.pop){|_,a,b| a + b}
  end

  histogram
end

def forking(file, fork_count=8)
  histogram = Hash.new.tap{|h| h.default = 0 }
  require 'rinda/tuplespace'

  uri = 'druby://localhost:12345'
  server = Rinda::TupleSpace.new
  DRb.start_service(uri, server)

  chunk_size = File.size(file) / fork_count

  fork_count.times do |i|
    Process.fork do
      chunk = Chunk.new(file, chunk_size, i * chunk_size)
      DRb.start_service
      client = DRbObject.new_with_uri(uri)
      client.write(['results', chunk.parse])
    end
  end

  fork_count.times do
    results = server.take(['results', nil])[1]
    histogram.merge!(results){|_,a,b| a + b }
  end

  histogram
end

class Chunk
  def initialize(file, size, offset)
    @handle = File.open(file)
    @handle.pos = offset
    @size = size
    @string = @handle.read(@size)
    if @string.index("\n") < @string.index(',')
      look_back(4)
    end
  end

  def parse
    results = Hash.new.tap{|h| h.default = 0}
    offset = 0
    while pos = @string.index(',', offset)
      if eol = @string.index("\n", pos)
        age = @string[pos + 1..eol - 1]
        results[age] += 1
        offset = eol + 1
      else
        offset = pos + 1
      end
    end
    results
  end

  def look_back(bytes)
    @handle.seek(- @size - bytes, IO::SEEK_CUR)
    @string = @handle.read(bytes) + @string
  end
end

# p simple(ARGV[0])        # 1.9.3: 1:38, jruby: 1:23, jruby --fast -J-Xmx1750m -J-Xms1750m: 1:07
# p threading(ARGV[0])     # 1.9.3: 2:03,              jruby --fast -J-Xmx1750m -J-Xms1750m: 0:15
p forking(ARGV[0])       # 1.9.3: 0:34