knife-ec-backup data validation

check-data.rb

#!/opt/chef/embedded/bin/ruby
require 'json'

class Checker
  def check
    check_cookbook_metadata
    check_node_sizes
    check_bad_platform_metadata
    check_bad_invitations
    check_duplicates_in_groups
    check_self_referential_groups
    check_user_client_conflict
  end

  def check_cookbook_metadata
    puts "\nCookbooks with missing metadata fields"
    puts "---------------------------"
    Dir["organizations/*/cookbooks/metadata.rb"].each do |filename|
      metadata = File.readlines(filename)
      puts filename if metadata.grep(/^name|^maintainer\s+|^maintainer_email/).count < 3
    end
  end

  def check_node_sizes
    puts "\nNodes that may be too large"
    puts "---------------------------"
    Dir["organizations/*/nodes/*.json"].each do |filename|
      puts filename if File.stat(filename).size > 1_000_000
    end
  end

  def check_bad_platform_metadata
    puts "\nBad platform metadata in cookbooks"
    puts "-----------------------------------"
    Dir["organizations/*/cookbooks/*/metadata.json"].each do |filename|
      md = JSON.parse(File.read(filename))
      bad_json = false
      md["platforms"].each {|k,v| bad_json = true if v == [] }
      puts filename if bad_json
    end
  end

  def check_bad_invitations
    puts "\nNull username in invitations"
    puts "----------------------------"
    Dir["organizations/*/invitations.json"].each do |filename|
      if JSON.parse(File.read(filename)).any? {|i| i["username"].nil?}
        puts filename
      end
    end
  end


  def check_duplicates_in_groups
    puts "\nDuplicates in groups"
    puts "-------------------"
    Dir["organizations/*/groups/*.json"].each do |filename|
      group = JSON.parse(File.read(filename))
      dup_clients = ! ( group["clients"].nil? || (group["clients"].uniq.size == group["clients"].size))
      dup_users = ! ( group["users"].nil? || (group["users"].uniq.size == group["users"].size))
      if dup_clients || dup_users
        puts "", "* #{filename}"
      end

      if dup_clients
        puts "Duplicate Clients:"
        puts group["clients"].uniq.map {|c| [group["clients"].count(c), c] }.select {|i| i[0] > 1}.map {|i| i[1] }
      end

      if dup_users
        puts "\tFollowing users are duplicated:"
        puts group["users"].uniq.map {|c| [group["users"].count(c), c] }.select {|i| i[0] > 1}.map {|i| i[1] }
      end
    end
  end

  def check_self_referential_groups
    puts "\nGroups that contain themselves"
    puts "--------------------------------"
    Dir["organizations/*/groups/*.json"].each do |filename|
      group_name = File.basename(filename).gsub(".json", "")
      group = JSON.parse(File.read(filename))
      if group["groups"] && group["groups"].include?(group_name)
        puts filename
      end
    end
  end

  def check_user_client_conflict
    puts "\nClients that conflict with users"
    puts "--------------------------------"
    Dir["organizations/*/members.json"]. each do |mfile|
      org_name = mfile.split("/")[1]
      members = JSON.parse(File.read(mfile))
      members.each do |member|
        uname = member["user"]["username"]
        if File.exists?("organizations/#{org_name}/clients/#{uname}.json")
          puts "organizations/#{org_name}/clients/#{uname}.json"
        end
      end
    end
  end
end

if ! File.exists?("organizations")
  puts "No organizations folder found. Please run this from an ec-export"
  exit 1
else
  Checker.new.check
end