booksamillion

booksamillion

# BOOKS A MILLION {source_name: 'booksamillion', batch_number: 5, request_id: 5000000041, request_name: 'BOOKS A MILLION'}
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_postal_codes "us", radius: 10 do |zip, lat, long|
      queue_url "http://www.bullseyelocations.com/pages/BAMStoreFinder?f=1##{zip}", zip: zip, lat: lat, long: long, 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('#form1').attr('action')).to_s
      seeder.queue_url action, {
        method: :post,
        page_type: 'search',
        body: {
          '__EVENTTARGET' => '',
          '__EVENTARGUMENT' => '',
          '__LASTFOCUS' => '',
          '__SCROLLPOSITIONX' => 0,
          '__SCROLLPOSITIONY' => 0,
          'ctl00$ContentPlaceHolder1$txtCityStateZip' => parser_page[:zip],
          'ctl00$ToolkitScriptManager1' =>'ctl00$ContentPlaceHolder1$upLocator|ctl00$ContentPlaceHolder1$searchButton',
          'ctl00$ContentPlaceHolder1$hfLocation' => "#{parser_page[:zip]}&country=United States",
          'ctl00$ContentPlaceHolder1$hfOrigCoords' => "#{parser_page[:lat]},#{parser_page[:long]}",
          'ctl00$ContentPlaceHolder1$hfMobile' => 'false',
          'ctl00$ContentPlaceHolder1$hfSearch' => "&PostalCode=#{parser_page[:zip]}&CountryId=1&CategoryIds=87387,87389,87388&Radius=50&InterfaceID=5308&FindNearestForNoResults=true&GetHoursForUpcomingWeek=true&LanguageCode=ru",
          'ctl00$ContentPlaceHolder1$hfNext' => '',
          'ctl00$ContentPlaceHolder1$hfPrev' => '',
          'ctl00$ContentPlaceHolder1$searchButton' => 'Search'
        },
        priority: 100,
        headers: fetcher_agent.get_headers(parser_page),
      }
    end
    find_pages page_format: :html, page_type: 'search' do |url, parser_page, page|
      page.search('.resultsDetails').each do |el|
        name = el.css('[itemprop="name"]').children.text
        t = el.next.next.next.next.next.next.children[3].children.text
        id = el.css('[itemprop="url"]').attribute('href').value.split('/')[2]
        city = el.css('[itemprop="addressLocality"]').children.text.gsub(", ", '')
        state = el.css('[itemprop="addressRegion"]').children.text
        zip_code = el.css('[itemprop="postalCode"]').children.text
        location_url = ('http://www.bullseyelocations.com' + el.css('[itemprop="url"]').attribute('href').value).split('&PostalCode')[0]

        latitude = el.css('[itemprop="latitude"]').attribute('content').value
        longitude = el.css('[itemprop="geo"]').first.inner_html.gsub("\r", '').gsub("\n", '').split("meta itemprop=")[2].gsub(/[^\d,\.,\(,\,\-)]/, '')
        street1 = el.css('[itemprop="streetAddress"]').children.text
        country = ''
        
        if name =~ /MILLION/i
          brand = "BOOKS A MILLION"
        else

          brand = "BOOKS AND CO"
        end

        doc_id = store_doc({
          store_id: id,
          brand: brand,
          type: brand,
          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: {
            'JOE MUGS' => (t =~ /JOE MUGS/i).present?,
            'KIDS-A-MILLION' => (t =~ /KIDS-A-MILLION/i).present?,
            'YOGURT MOUNTAIN' => (t =~ /YOGURT MOUNTAIN/i).present?
          }
        })
      end
    end
  end
end