just-hosts.rb

#!/usr/bin/env ruby
# Copyright Bouke van der Bijl
require 'nokogiri'

# This script interpretes the HTTPS everywhere rulesets and extracts a list of hosts which are eligible for a 'simple' redirect,
# e.g. where http://example.com needs to be redirected to https://example.com

def is_simple?(rule)
  rule.attributes['from'].value == "^http:" && rule.attributes['to'].value == "https:"
end

Dir["rules/*.xml"].each do |name|
  doc = Nokogiri::XML(File.open(name))
  doc.xpath('//ruleset').each do |set|
    next if set.attributes['default_off'] || set.attributes['platform']&.value == 'mixedcontent'
    next if set.xpath('exclusion').any?

    rules = set.xpath('rule')
    if rules.count == 1
      rule = rules.first
      next unless is_simple?(rule)

      set.xpath('target').each do |t|
        puts t.attributes['host'].value.sub(/\A\*\./, '*')
      end
    else
      set.xpath('target').each do |t|
        if t.attributes['host'].value.start_with?('*.')
          # Can't guarantee nothing breaks, might mean complicated regexes
          next 
        end

        host = t.attributes['host'].value
        u = "http://#{host}/"

        # Iterate over all the rules and if it matches one that isn't the simple rule, abort
        rules.each do |rule|
          r = Regexp.new(rule.attributes['from'].value)
          if r.match(u)
            if is_simple?(rule)
              puts t.attributes['host'].value
            end
            break
          end
        end
      end
    end
  end
end