chase

chase

# CHASE BANK {source_name: 'chase', batch_number: 7, request_id: 7000000068, request_name: 'CHASE BANK'}

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]) do
    loop_states "us", radius: 10 do |short_code, full_name|
      queue_url "https://locator.chase.com/search/#{short_code}/?q=#{full_name}"
    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|

      page.search('.result.clearfix').each do |detail|
        next if detail.at('.resultName').nil?

        max_page = t_body.at('.clearfix.pagination').search('li').map(&:text).map(&:to_i).max rescue nil

        if max_page && max_page > 1
  
          (2..max_page).each do |p|
    
            seeder.queue_url "#{url}&page=#{p}"
          end
        end    
        
        name = detail.at('.resultName').text.gsub("\t", '').gsub("\r", '').gsub("\n", '')
        href = detail.at('.titleLeft').at('a').attr('href')
        location_url = URI.join(url, href).to_s

        street1 = detail.at('.address').at('.line.street-address').text
        city = detail.at('.address').at('.line.last').at('.locality').text
        state = detail.at('.address').at('.line.last').at('.region').text
        zip_code = detail.at('.address').at('.line.last').at('.postal-code').text

        loc_type = detail.at('.locationType').text

        if loc_type =~ /ONLY/i
          t = "CHASE ATM ONLY"
        else
          t = "CHASE BRANCHES"
        end  

        doc_id = store_doc({
          # store_id: id,
          brand: "CHASE",,
          type: t,
          # property_id: id,
          name: name, # required
          address_1: street1,
          address_2: '',
          city: city,
          state: state,
          zipcode: zip_code,
          country: '',
          # lat: lat,
          # long: long,
          # address_container_html: address_container,
          # map_link: store.parent.parent.at('.googlemap').attr('name'),
          location_url: location_url,
          flags: {
          }
        })

        seeder.queue_url location_url, {
          page_type: 'profile',
          doc_id: doc_id
        }
      end    
    end
    
    # Find profile pages
    find_pages page_format: :html, page_type: 'profile' do |url, parser_page, page|
      doc = find_location(parser_page[:doc_id])

      if doc.present?
        atm_count = 0
        
        unless page.at('.atmInformation').nil?
          
          unless page.at('.atmInformation').search('.halfMain.floatLeft').nil?
            
            page.at('.atmInformation').search('.halfMain.floatLeft').each do |info|
              tmp = info.text.split(' ')
              condition = Float(tmp.first) != nil rescue false

              if condition
                atm_count = tmp.first
                break
              end
            end
          end
        end    

        bs = ""
        page.search(".sectionTitle").each do |st|
          next unless st.text == "Branch Services"

          divs = st.parent.search('div:not([class])')
          tmp = []
          divs.each do |div|
            next if div.text.nil? 

            tmp.push(div.text.split("\n")[0])
          end  
          bs = tmp.join(", ")
        end  
        
        lat = page.at('[property="place:location:latitude"]').attr('content')
        long = page.at('[property="place:location:longitude"]').attr('content')
        
        frmt = page.at('.branchType').text
        frmt = frmt.gsub("\t", '').gsub("\r", '').gsub("\n", '') unless frmt.nil?
        doc[:lat] = lat
        doc[:long] = long
        doc[:flags]['ATM COUNT'] = atm_count
        doc[:flags]['BRANCH SERVICES'] = bs
        doc[:flags]['FORMAT'] = frmt
        
        store_doc doc
      end  
    end
  end
end