a.rb

# encoding: utf-8


#
  companies = Company.where(:locale => 'ko')
  domains = Domain.where(:company_id.in => companies.map(&:id))
  domains.destroy_all
  companies.destroy_all


#
  ko_csv = "#{ Rails.root }/data/ko.csv"

  buf = File.read(ko_csv)

#
  require 'csv'
  csv = CSV.parse(buf)

#
  headers = %w[
    title url cei _a energy_transparency _b renewable_energy_commitment _c renewable_energy_champion re oil hydro amount_gas amount_coal amount_nuclear _d _e grade locale twitter_handle_en service_host_en twitter_handle_ko service_host_ko competitor_1 competitor_2 competitor_3 domains
  ]

#
  header = nil
  rows = []

#
  csv.each do |row|
    unless header
      header = row
    else
      hash = Hash[*headers.zip(row).flatten]
      rows.push(Map.for(hash))
    end
  end

#
  companies = []
  competitors = {}

  rows.each do |row|
  #
    companies.push( company = Company.new )

  #
    %w[ title url cei energy_transparency grade locale domains ].each do |key|
      company[key] = Coerce.string row[key]
    end

    %w[ amount_gas amount_coal amount_nuclear ].each do |key|
      company[key] = row[key].to_i
    end

    %w[ renewable_energy_commitment renewable_energy_champion ].each do |key|
      company[key] = Coerce.boolean row[key]
    end

  #
    company[:twitter_handle] = row["twitter_handle_ko"]
    company[:service_host] = row["service_host_ko"]

  #
    translation = company.translations.build :locale => :en
    translation.twitter_handle = row["twitter_handle_en"]
    translation.service_host = row["service_host_en"]

  #
    names = []
    Coerce.list_of_strings(company[:domains]).each do |domain|
      names.push(domain.gsub('*.', ''))
      names.push(domain)
    end
    names.uniq!
    names.each do |name|
      domain = company.domains.build(:name => name)
    end

  #
    company.raw_score ||= 0

  #
    unless company.url.blank?
      company.url = company.url.downcase
    end

  #
    if company.twitter_handle.blank?
      company.twitter_handle = company.url
    end

  #
    competitors[company.id] =
      Coerce.list_of_strings(
        row['competitor_1'],
        row['competitor_2'],
        row['competitor_3']
      )

  end

#
  companies.each do |company|
    puts('='*79)
    p company.title
    p company.as_document
    p company.translations.map(&:as_document)
    p company.domains.map(&:as_document)
    company.save!
    company.domains.each{|domain| domain.save!}
  end

#
  competitors.each do |company_id, list|
    company = Company.find(company_id)
    competitor_ids = []

    list.each do |url|
      domain = Domain.best_match_query_for(url) 
      if domain
        competitor_ids.push(domain.company_id)
      end
    end

    company.set(:competitor_ids, competitor_ids)
  end


%w[

  title
  url
  cei
  energy_transparency
  grade
  locale
  domains

  amount_gas
  amount_coal
  amount_nuclear

  renewable_energy_commitment
  renewable_energy_champion


  twitter_handle_en
  service_host_en

  twitter_handle_ko
  service_host_ko

  competitor_1
  competitor_2
  competitor_3

]



<<-__


{"id"=>"560ac3570fa604968800016f",
 "_id"=>"560ac3570fa604968800016f",
 "name"=>"equinix_com",
 "title"=>"Equinix.com",
 "slug"=>"equinix-com",
 "description"=>nil,
 "url"=>"http://equinix.com",
 "energy_transparency"=>true,
 "renewable_energy_commitment"=>true,
 "renewable_energy_champion"=>false,
 "cei"=>15,
 "amount_gas"=>29,
 "amount_coal"=>29,
 "amount_nuclear"=>29,
 "raw_score"=>nil,
 "grade"=>"C",
 "twitter_handle"=>"equinix",
 "service_host"=>"Equinix.com",
 "competitor_ids"=>
  ["560ac3570fa6049688000169",
   "560ac3530fa6049688000043",
   "560ac3520fa6049688000010"],
 "locale"=>"en",
 "created_at"=>Tue, 29 Sep 2015 16:59:03 UTC +00:00,
 "updated_at"=>Mon, 20 Jul 2015 15:43:51 UTC +00:00}


{"title"=>"nuclearwebsite",
 "url"=>"http://greenpeacekorea.org/nonuke",
 "cei"=>"100",
 "_a"=>"25",
 "energy_transparency"=>"True",
 "_b"=>"25",
 "renewable_energy_commitment"=>"TRUE",
 "_c"=>"25",
 "renewable_energy_champion"=>"TRUE",
 "re"=>"100",
 "oil"=>nil,
 "hydro"=>nil,
 "amount_gas"=>nil,
 "amount_coal"=>nil,
 "amount_nuclear"=>nil,
 "_d"=>nil,
 "_e"=>nil,
 "grade"=>"A",
 "lang"=>"ko",
 "twitter_handle_en"=>"@greenpeacekorea",
 "service_host_en"=>"Google",
 "twitter_handle_ko"=>"@greenpeacekorea",
 "service_host_ko"=>"구글",
 "competitor_1"=>nil,
 "competitor_2"=>nil,
 "competitor_3"=>nil,
 "domains"=>"*.greenpeacekorea.org"}


__