PoC for parce browser_history dump

browser_history.rb

#!/usr/bin/env ruby

require "addressable/uri"
require 'nokogiri'
require 'string_monkey_path'

def item_field_by_name(item, name)
  item.css(name).first.children.map(&:to_s).first || ""
end

def get_host_by_uri(url)
  site = Addressable::URI.parse(url).site
  return unless site.include?(".")
  return site if site.include?("http")
  "http://#{site}"
rescue Addressable::URI::InvalidURIError => e
  raise "Unable to parse url '#{url}'. \n #{e.class}: #{e}."
end

def get_host(url)
  host = url[7..-1].downcase.split(".")
  site = host.last.is_number? ? host.join(".") : host.reverse[0, 2].reverse.join(".")
  return unless site.include?(".")
  return site if site.include?("http")
  "http://#{site}"
end

def collect_browsers
  parced_xml_file = ::Nokogiri::XML(::File.read(ARGV[0]))
  items = parced_xml_file.xpath("//browsing_history_items/item")

  items.map do |item|
    url = item_field_by_name(item, "url")
    case
    when (url.include?(":Host:") && !url.include?("My Computer"))
      get_host(url)
    when url.includes?(::URI.scheme_list.keys.map(&:downcase))
      get_host_by_uri(url)
    else
      next
    end
  end.compact.uniq
end

puts collect_browsers.inspect