Quick and ugly script to generate 1,000,000 fake accounts in MongoDB

one_million_records.rb

#!/usr/bin/env ruby

require 'mongo'
require 'faker'
require 'pp'
require 'benchmark'

include Mongo

class Exception
  def errmsg
    "%s: %s\n%s" % [self.class, message, (backtrace || []).join("\n") << "\n"]
  end
end

host = 'localhost'
port = Connection::DEFAULT_PORT

db = Connection.new(host, port).db('test')

puts ">> Getting collection: mongo_test_index"
begin
  coll = db.collection('mongo_test_index')
  puts "created: #{coll.inspect}"
rescue => e
  puts "Error: #{e.errmsg}"
end

OBJS_EACH  = 500
OBJS_TIMES = 2000
TOTAL = OBJS_EACH * OBJS_TIMES

$TOTAL_TIME = 0
$AVERAGE_TIME = 0

arr = nil
rndm = ("a".."f").to_a
rndm << ("0".."9").to_a
rndm.flatten!

puts ">> Generating test data (#{OBJS_EACH*OBJS_TIMES} records, #{OBJS_EACH} per)..."

OBJS_TIMES.times do |x|
  individual_time = Benchmark.realtime do
    percent = ((x/OBJS_TIMES.to_f)*100).to_i
    time_left = ($AVERAGE_TIME)*(OBJS_TIMES - (x+1))
    mins_left = (time_left / 60).to_i
    secs_left = (time_left % 60).to_i
    puts "#{x}/#{OBJS_TIMES} (#{percent}%) #{mins_left} m #{secs_left} s left"
    arr = (0...OBJS_EACH).collect do
      first = Faker::Name.first_name
      last  = Faker::Name.last_name
      user1 = Faker::Internet.user_name
      user2 = Faker::Internet.user_name
      key1  = ""
      key2  = ""
      16.times do
        key1 += rndm[rand(16)]
        key2 += rndm[rand(16)]
      end
      {
        :first => first,
        :last  => last,
        :email => Faker::Internet.email,
        :accounts => [
          {
            :type => "facebook",
            :username => user1,
            :key => key1
          },
          {
            :type => "twitter",
            :username => user2,
            :key => key2
          }
        ]
      }
    end
    coll.insert(arr)
  end
  $TOTAL_TIME += individual_time
  $AVERAGE_TIME = $TOTAL_TIME / (x+1)
  iters_left = OBJS_TIMES - (x+1)
end

last_record = arr.last
pp last_record

puts " Done in #{$TOTAL_TIME}"

Not pretty at all (and even a couple errors now that I look) but it got the job done. :)