simhash.rb

class Simhash
  attr_reader :weights, :bits_length, :bitmasks
  def initialize(bits_length=64)
    @bits_length = bits_length
    @weights = Array.new(@bits_length, 0.0)
    @bitmasks = Array.new(@bits_length)
    (0...@bits_length).each do |i|
      @bitmasks[i] = (1 << i)
    end
  end

  def update(hashes)
    hashes.each {|h| update_hash h }
  end

  def update_hash(hash, weight=1.0)
    @bitmasks.each_with_index do |mask, i|
      @weights[i] += (hash & mask).zero? ? -weight : weight
    end
  end

  def digest(hashes=[])
    update(hashes)
    hash = 0
    @weights.each_with_index do |w, i|
      if w >= 0
        hash |= @bitmasks[i]
      end
    end
    hash
  end
end