calvin

calvin

# CALVIN KLEIN {source_name: 'calvinklein', batch_number: 5, request_id: 5000000042, request_name: 'CALVIN KLEIN'}
scraper_service.scrape do |browser, scraper, init_vars|
  easy_seeder = Library.lib('EasySeeder')
  easy_extractor = Library.lib('EasyExtractor')
  fetcher_agent = Library.lib('FetcherAgent')

  easy_seeder.seed(source_name: init_vars[:source_name], with_priority: true) do
    loop_states "us", radius: 10 do |short_code, full_name|
      queue_url "https://secure.gotwww.com/gotlocations.com/ck.com/ck.php##{short_code}", short_code: short_code, full_name: full_name, priority: 0
    end
  end

  easy_extractor.extract(
    source_name: init_vars[:source_name],
    scraper: scraper,
    batch_number: init_vars[:batch_number],
    request_id: init_vars[:request_id],
    request_name: init_vars[:request_name],
  ) do
    find_pages page_format: :html do |url, parser_page, page|
      action = URI.join(url, page.at("[name='search']").attr('action')).to_s
      seeder.queue_url action, {
        text:2,
        address:parser_page[:full_name],
        ip_country:'US',
        brand_array:'',
        product_categories:'',
        store_type:'',
        subcategory_array:'',
        metadata:'',
        language_user:'en_US',
        diag:'',
        Submit:'search'
      }
    end
    find_pages page_format: :html, page_type: 'search' do |url, parser_page, page|
      page.search('label.brandtextdisplay').each do |label|
        t = label.parent.inner_html.split('<br>').map(&:strip).select(&:present?).last
        
        if t =~ /FLAGSHIP/i
          t = 'CALVIN KLEIN FLAGSHIP STORE'
        elsif t =~ /RETAIL/i
          t = 'CALVIN KLEIN RETAIL STORE'
        elsif t =~ /OUTLET/i
          t = 'CALVIN KLEIN OUTLET STORE'
        elsif t =~ /ACCESSORY/i
          t = 'CALVIN KLEIN ACCESSORY STORE'
        elsif t =~ /SPECIALTY/i
          t = 'CALVIN KLEIN SPECIALTY STORE'
        elsif t =~ /SHOPPING/i
          t = 'CALVIN KLEIN SHOPPING MALL STORE'
        else
          next
        end  
        name = label.text
        address_container = label.parent.next.next.next.next.inner_html.split('<br>')
        longitude, latitude = label.parent.parent.next.next.at('a').attr('href').gsub(/[^\d,\.,\(,\,\-)]/, '').split('(')[2].split(')')[0].split(',')
        street1 = address_container[0].gsub("\t", '').gsub("\r", '').gsub("\n", '')
        street2 = address_container[1]
        tmp = address_container[2].split(',')
        city = tmp[0]
        state, zip_code = tmp[1].split(' ')
        country = ''

        doc_id = store_doc({
          # store_id: id,
          brand: 'CALVIN KLEIN',
          type: t,
          # property_id: id,
          name: name, # required
          address_1: street1,
          address_2: street2,
          city: city,
          state: state,
          zipcode: zip_code,
          country: country,
          lat: latitude,
          long: longitude,
          # map_link: page.at('a.map-it').attr('href'),
          # address_container_html: page.at('.rio-address-section').to_html,
          # location_url: location_url,
          flags: {
          }
        })
      end
    end
  end
end