geolite2_country_ipv4_mask_blocks.rb

#!/usr/bin/env ruby
# coding: utf-8

require 'set'

module GeoLite2_Country
  class IPv4Block < Struct.new(:start_ip,
                           :mask_length,
                           :geoname_id)
    def self.from_line(line)
      c = line.strip.split(",")

      network_start_ip = c[0]
      network_mask_length = c[1].to_i
      geoname_id = c[2].to_i

      if network_start_ip =~ /^::ffff:(\d+)\.(\d+)\.(\d+)\.(\d+)$/
        start_ip = ($1.to_i << 24) | ($2.to_i << 16) | ($3.to_i << 8) | $4.to_i
        mask_length = network_mask_length - 96
      else
        raise 'Bad IP'
      end

      self.new(start_ip, mask_length, geoname_id)
    end
  end

  class IPv4BlockLister
    def initialize(mask, in_fp = $stdin, out_fp = $stdout)
      @keep = []
      @in = in_fp
      @out = out_fp
      @mask = mask
    end

    def run
      header = @in.gets
      while line = @in.gets
        next if line[0..6] != '::ffff:'
        cur = IPv4Block.from_line(line)

        if !@keep.empty? && @keep.last.start_ip >> (32 - @mask) != cur.start_ip >> (32 - @mask)
          flush
        end

        if cur.mask_length == @mask
          @out.puts [cur.start_ip, cur.geoname_id].join("\t")
        elsif cur.mask_length < @mask
          (0...(1 << (@mask - cur.mask_length))).each do |n|
            @out.puts [cur.start_ip | (n << (32 - @mask)), cur.geoname_id].join("\t")
          end
        else
          @keep.push(cur)
        end
      end

      if !@keep.empty?
        flush
      end
    end

    private
    def flush
      geoname_ids = @keep.inject(Set.new){|ids, b| ids << b.geoname_id}
      geoname_ids.delete(0)
      if geoname_ids.length == 1
        @out.puts [(@keep.last.start_ip >> (32 - @mask)) << (32 - @mask), geoname_ids.first].join("\t")
      else
        # ambiguous
        @out.puts [(@keep.last.start_ip >> (32 - @mask)) << (32 - @mask), 0].join("\t")
      end
      @keep = []
    end
  end
end

GeoLite2_Country::IPv4BlockLister.new(24).run