gen.rb

require 'set'
require 'csv'

CUSTOMER_AMOUNT = 20
CUSTOMER_AMOUNT_FOCUS_HH = 50
CUSTOMER_AMOUNT_FOCUS_MU = 30
CUSTOMER_AMOUNT_FOCUS_BE = 70
PRODUCT_AMOUNT = 10


def uniq_rand_n(n, min, max)
    raise "Uniq Rand N to big" if n > (max-min)
    randoms = Set.new
    loop do
        r = rand(max-min) + min 
        randoms << r
        return randoms.to_a if randoms.size >= n
    end
end

def choose_weighted(weighted)
  sum = weighted.inject(0) do |sum, item_and_weight|
    sum += item_and_weight[1]
  end
  target = rand(sum)
  weighted.each do |item, weight|
    return item if target <= weight
    target -= weight
  end
end

customers = []
first_names = CSV.read("first_names.csv")
last_names = CSV.read("last_names.csv")
strassen = CSV.read("strassen.csv")
plz_einwohner = CSV.read("plz_einwohner.csv").map do |r|
  [r[0].to_i,r[1].to_i]
end

plz_focus_hh = plz_einwohner.select do |r|
  (r[0] >= 20095 && r[0] <= 22769)
end
plz_focus_mu = plz_einwohner.select do |r|
  (r[0] >= 80331 && r[0] <= 81929)
end
plz_focus_be = plz_einwohner.select do |r|
  (r[0] >= 10115 && r[0] <= 14199)
end

uniq_rand_n(CUSTOMER_AMOUNT, 1000000, 1999999).each do |id|
  customers << [
    id,
    first_names[rand(first_names.length)],
    last_names[rand(last_names.length)],
    "#{strassen[rand(last_names.length)].first} #{rand(132)}",
    choose_weighted(plz_einwohner)
  ]
end

uniq_rand_n(CUSTOMER_AMOUNT_FOCUS_HH, 2000000, 2999999).each do |id|
  customers << [
    id,
    first_names[rand(first_names.length)],
    last_names[rand(last_names.length)],
    "#{strassen[rand(last_names.length)].first} #{rand(132)}",
    choose_weighted(plz_focus_hh)
  ]
end

uniq_rand_n(CUSTOMER_AMOUNT_FOCUS_MU, 2000000, 2999999).each do |id|
  customers << [
    id,
    first_names[rand(first_names.length)],
    last_names[rand(last_names.length)],
    "#{strassen[rand(last_names.length)].first} #{rand(132)}",
    choose_weighted(plz_focus_mu)
  ]
end

uniq_rand_n(CUSTOMER_AMOUNT_FOCUS_BE, 2000000, 2999999).each do |id|
  customers << [
    id,
    first_names[rand(first_names.length)],
    last_names[rand(last_names.length)],
    "#{strassen[rand(last_names.length)].first} #{rand(132)}",
    choose_weighted(plz_focus_be)
  ]
end

puts "KundenId,Vorname,Nachname,Strasse,PLZ"
customers.each do |c|
  puts c.join(",")
end