Does not work well on files with IPv4 and IPv6 addresses

[konfusator]

On a dual stack machine a typical log file contains ipv4 and ipv6 addresses. Anonymizing with the algorithm "IPV4_NETWORK" will work as Ruby's IPAddr works with ipv4 and ipv6 adressess, but the result is not useful (with a key "24" 193.99.144.85 will become 193.99.144.0 (good) and 2a02:2e0:3fe:1001:7777:772e:2:85 will become 2a02:200:: (useless)).

A solution would be a new algorithm (called e.g. IP_NETWORK) using different prefix lengths for IPv4 and IPv6.

This could be done like

  def anonymize_ip_network(ip_string)
    # in JRuby 1.7.11 outputs as US-ASCII
    addr = IPAddr.new(ip_string)
    if addr.ipv4?
       addr.mask(@key.split(':')[0].to_i).to_s.force_encoding(Encoding::UTF_8)
    else
       addr.mask(@key.split(':')[1].to_i).to_s.force_encoding(Encoding::UTF_8)
    end
  end

expecting a key in the format "v4prefix:v6prefix" or

  def anonymize_ip_network(ip_string)
    # in JRuby 1.7.11 outputs as US-ASCII
    addr = IPAddr.new(ip_string)
    if addr.ipv4?
       addr.mask(@prefix_length[0]).to_s.force_encoding(Encoding::UTF_8)
    else
       addr.mask(@prefix_length[1]).to_s.force_encoding(Encoding::UTF_8)
    end
  end

with a new array parameter prefix_length.