used_files.rb

#!/usr/bin/ruby

require 'puppetdb'
require 'puppet'
require 'json'

Puppet.initialize_settings
key = Puppet.settings['hostprivkey']
cert = Puppet.settings['hostcert']
ca_file = Puppet.settings['localcacert']

server = 'puppetdb'  # where do I get this?

client = PuppetDB::Client.new({
  :server => "https://#{server}:8081",
  :pem    => {
    'key'     => key,
    'cert'    => cert,
    'ca_file' => ca_file
  }
})

response = client.request(
  'resources',
  [:and, [:'~', 'type', '[A-Za-z0-9_-]+']],
  { }
)

resources = response.data

# Initialize output
output = { 'modules' => {}, 'site-modules' => {} }


# Get list of environments
environment_paths = resources.map { |r|
  if r['file']
    match = r['file'].match(%r{^(/.*)/(?:site-)?modules/})
    match.captures[0] if match
  end
}.compact.uniq

resources.each do |r|
  if r['file']
    f = r['file']
    match = f.match(%r{^(/.*)/((?:site-)?modules)/([^/]+)/(.*)$})
    if match
      lead, type, mod, path = match.captures
      # Add used files
      output[type][mod] ||= {}
      output[type][mod]['used'] ||= []
      output[type][mod]['used'] |= [path]

      # Add unused files
      mod_path = "#{lead}/#{type}/#{mod}/"
      output[type][mod]['unused'] ||= environment_paths.map { |e|
        env_mod_path = "#{e}/#{type}/#{mod}/"
        Dir.glob("#{env_mod_path}manifests/**/*.pp").map { |p| p.gsub(env_mod_path, '') }
      }.flatten.compact.uniq
      output[type][mod]['unused'].delete(f.gsub(mod_path, ''))
    end
  end
end


puts JSON.dump output