drupal -> rails db example

migrator.rake

# encoding: utf-8

namespace :migrator do

  def db
    connection_string         = ENV['DB']
    default_connection_string = 'mysql://root:***@localhost:3306/***'
    
    @connection ||= Sequel.connect(connection_string || default_connection_string)
    @connection
  end

  def drupal_root
    root_dir = ENV['DRUPAL_ROOT']
    root_dir || '/Users/rbdev/db_backup/site/2013.04.01.01.00.02/sites_backup/archives/home/rbdev/sites/apache/sail-cruise.ru/www'
  end

  def yachts_pdf_root
    root_dir = ENV['ROOT_DIR']
    root_dir || '/Users/rbdev/db_backup/site/pdf/yachts.sail-cruise.ru/www'
  end

  def term(id)
    row = db[:term_data].where(:tid => id.to_i).first
    
    row ? row[:name] : nil
  end

  def pdf_field_file_path(nid)
    field = db[:content_field_pdf].where(:nid => nid).first

    if field and field[:field_pdf_fid].present?
      file_data = db[:files].where(:fid => field[:field_pdf_fid]).first

      if file_data
        return File.join(drupal_root, file_data[:filepath])
      end
    end
  end

  def pdf_file_path_in_nid_body?(nid)
    !!pdf_file_path_from_nid_body(nid)
  end

  def pdf_file_path_from_nid_body(nid)
    body       = db[:node_revisions].where(:nid => nid).first[:body]
    match_data = body.match(/\[pdf=.*yachts\.sail\-cruise\.ru(.*)\]/)

    if match_data and match_data[1]
      return File.join(yachts_pdf_root, match_data[1])
    end
  end

  def error_on(field, record, meta = nil)
    raise Exception, "error on :#{field}\n #{record.inspect} \n #{meta.inspect}"
  end

  YACHT_TYPES = {
    1 => 'sail',
    2 => 'motor',
    3 => 'gulet',
    38 => 'catamaran'
  }

  # таблица node
  #   type  -> 'yachtpage' (нужен только этот тип)
  #
  # связь данных:
  #   content_node_field_instance -> content_type_yachtpage   -> term data
  #   берем имя поля              -> добавляем к имени _value -> ищем тут по vid
  #
  # поля node
  #   - * title
  #     node_revisions.title
  #
  #   - body
  #     node_revisions.body
  #
  #   - Производитель
  #     content_type_yachtpage.field_creator_value
  #
  #   - * тип яхты
  #     content_type_yachtpage.field_type_value
  #
  #   - * мега
  #     content_type_yachtpage.field_mega_value
  #     vid:16 -> яхта
  #     vid:17 -> мега
  #
  #   - год выпуска
  #     content_type_yachtpage.field_year_value -> to_i
  #
  #   - страна
  #     content_field_country.field_country_value
  #
  #   - длинна
  #     content_type_yachtpage.field_long_value
  #
  #   - кабины
  #     content_type_yachtpage.field_cabins_value
  #
  #   - спальные места
  #     content_type_yachtpage.field_places_value
  #
  #   - цена
  #     content_type_yachtpage.field_price_value
  #
  #   - PDF
  #     некоторые ноды не имеют залитого пдф-а
  #     эти пдфы лежат на yachts.sail-cruise.ru (и ссылки на них есть в body)
  #     
  #     связующая таблица - content_field_pdf(nid, field_pdf_fid)
  #     таблица файлов    - files(nid, filename, filepath)
  #
  #   - Главная картинка
  #     content_field_yacht_icon(nid, field_yacht_icon_imceimage_path)
  #
  #     
  # таблица vocabulary
  #   нужны только: (vid, name)
  #     * 1, Тип яхты
  #     * 2, Производитель
  #     * 3, Длинна
  #     * 4, Мега
  #     * 5, Год выпуска (относится к мега яхтам)
  #     * 6, Страна (TODO: посмотреть, нет ли там трешевых данных)
  #     * 7, Каюты
  #     * 8, Спальные места
  #     * 9, Цена
  #
  #   все значения словарей лежат в `term_data` (tid, vid, name)
  #   соотношения между нодой -> словарем -> термом лежат в `term_node` (nid, vid, tid)

  task :do => :environment do
    db[:node].where(:type => 'yachtpage').each do |node|
      yacht = Yacht.new

      revision  = db[:node_revisions].where(:nid => node[:nid]).first
      meta      = db[:content_type_yachtpage].where(:nid => node[:nid]).first


      # -- NID
      yacht.drupal_nid = node[:nid]


      # -- NAME && BODY
      yacht.name = revision[:title].strip
      yacht.body = revision[:body].strip


      # -- Manufacture
      manufacture = Manufacture.find_or_create_by_name( term(meta[:field_creator_value]) )
      yacht.manufacture = manufacture


      # -- is_mega
      # 16 - обычная
      # 17 - мега
      if meta[:field_mega_value].blank?
        error_on(:field_mega_value, node, meta)
      else
        if meta[:field_mega_value] == 17
          yacht.is_mega = true
        else
          yacht.is_mega = false
        end
      end


      # -- boat type
      # 1 - sail
      # 2 - motor
      # 3 - gulet
      # 38 - catamaran
      if meta[:field_type_value].blank?
        error_on(:field_type_value, node, meta)
      else
        yacht.boat_type = YACHT_TYPES[meta[:field_type_value]]
      end


      # -- creation year
      year_value = term(meta[:field_year_value])
      if year_value.present?
        yacht.creation_year = year_value
      end

    
      # -- country
      # drupal_country_field = db[:content_field_country].where(:nid => node[:nid]).first
      # country_name = term(drupal_country_field[:field_country_value])
      # if country_name.present?
      #   country = Country.find_or_create_by_name(country_name)
      #   yacht.country = country
      # end


      # -- boat length
      length_value = term(meta[:field_long_value])
      if length_value.blank?
        error_on(:length_value, node, meta)
      else
        boat_length = BoatLength.find_or_create_by_name(length_value)
        yacht.boat_length = boat_length
      end


      # -- cabins_amount
      cabins_amount_value = term(meta[:field_cabins_value])
      if cabins_amount_value.present?
        yacht.cabins_amount = cabins_amount_value
      end


      # -- sleepers_amount
      sleepers_amount_value = term(meta[:field_places_value])
      if sleepers_amount_value.present?
        yacht.sleepers_amount = sleepers_amount_value
      end


      # -- price
      price_value = term(meta[:field_price_value])
      if price_value.present?
        yacht.price_string = price_value
      end


      # -- content logo
      logo_metadata = db[:content_field_yacht_icon].where(:nid => node[:nid]).first
      if logo_metadata and logo_metadata[:field_yacht_icon_imceimage_path].present?
        drupal_file_path = File.join(drupal_root, logo_metadata[:field_yacht_icon_imceimage_path])
        drupal_file_path = drupal_file_path.gsub('%20', ' ')

        if File.exists?(drupal_file_path)
          file = File.new(drupal_file_path, 'r')

          logo = ContentLogo.new
          logo.image = file
          logo.save!

          yacht.content_logo = logo

          file.close
        end
      end


      # -- PDF
      pdf_path = pdf_field_file_path(node[:nid]) || pdf_file_path_from_nid_body(node[:nid])
      if pdf_path
        if File.exists?(pdf_path)
          file = File.open(pdf_path, 'r')
          yacht.pdf = file
          file.close

          if pdf_file_path_in_nid_body?(node[:nid])
            yacht.body = nil
          end
        else
          raise Exception, "not exists -> #{pdf_path}"
        end

      end

      if yacht.valid?
        yacht.save!
        puts "yachts created with NID => #{yacht.drupal_nid}"
      else
        raise Exception, yacht.errors.inspect
      end

    end
  end

end