[#7] Add all other parsing methods

app/services/google_service/parser_service.rb

@@ -4,9 +4,17 @@ module GoogleService
   class ParserService
     require 'nokogiri'
 
+    @@non_ads_result_selector = 'a[data-ved]:not([role]):not([jsaction]):not(.adwords):not(.footer-links)'
+
     def initialize(html)
       @html = html
       @document = Nokogiri::HTML.parse(html)
+
+      # Add a class to all AdWords link for easier manipulation
+      @document.css('div[data-text-ad] a[data-ved]').add_class('adwords')
+
+      # Mark footer links to identify them
+      @document.css('#footcnt a').add_class('footer-links')
     end
 
     def parse_into(keyword)
@@ -20,28 +28,34 @@ def parse_into(keyword)
     end
 
     def ads_top_count
-      @document.css('#tads div[data-text-ad]').count
+      @document.css('#tads .adwords').count
     end
 
     def ads_page_count
-      @document.css('div[data-text-ad]').count
+      @document.css('.adwords').count
     end
 
     def ads_top_url
       # data-ved enables to filter "role=list" (sub links) items
-      @document.css('#tads div[data-text-ad] a[data-ved]').map { |a_tag| a_tag['href'] }
+      @document.css('#tads .adwords').map { |a_tag| a_tag['href'] }
     end
 
     def ads_page_url
-
+      @document.css('.adwords').map { |a_tag| a_tag['href'] }
     end
 
     def non_ads_result_count
+      @document.css(@@non_ads_result_selector).count
+    end
 
+    def non_ads_url
+      @document.css(@@non_ads_result_selector).map { |a_tag| a_tag['href'] }
     end
 
     def total_link_count
-
+      Rails.logger.info 'Counter HERE!!!'
+      @document.css('a').map { | a_tag | Rails.logger.info a_tag['href'] }
+      @document.css('a').count
     end
   end
 end

spec/services/google_service/parser_service_spec.rb

@@ -14,7 +14,7 @@
     end
   end
 
-  context 'when parsing a page having 3 top ads and 3 bottom ads' do
+  context 'when parsing a page having 3 top ads, 3 bottom ads and 14 non ads links' do
     it 'counts exactly 3 top ads' do
       result = nil
       VCR.use_cassette('google_search_top_ads_6') do
@@ -33,14 +33,40 @@
       expect(described_class.new(result).ads_page_count).to eq(6)
     end
 
-    it 'find 3 different top ads url' do
+    it 'finds exactly the 3 top ads urls' do
       result = nil
       VCR.use_cassette('google_search_top_ads_6') do
         result = GoogleService::ClientService.query('vpn')
       end
 
-      expect(described_class.new(result).ads_top_url).to contain_exactly('https://cloud.google.com/free','https://www.expressvpn.com/', 'https://www.top10vpn.com/best-vpn-for-vietnam/')
+      expect(described_class.new(result).ads_top_url).to contain_exactly('https://cloud.google.com/free', 'https://www.expressvpn.com/', 'https://www.top10vpn.com/best-vpn-for-vietnam/')
     end
 
+    it 'counts exactly 14 non ads results' do
+      result = nil
+      VCR.use_cassette('google_search_top_ads_6') do
+        result = GoogleService::ClientService.query('vpn')
+      end
+
+      expect(described_class.new(result).non_ads_result_count).to eq(14)
+    end
+
+    it 'gets 14 results' do
+      result = nil
+      VCR.use_cassette('google_search_top_ads_6') do
+        result = GoogleService::ClientService.query('vpn')
+      end
+
+      expect(described_class.new(result).non_ads_url.count).to eq(14)
+    end
+
+    it 'gets exactly 113 links' do # Counted from cassette html raw code
+      result = nil
+      VCR.use_cassette('google_search_top_ads_6') do
+        result = GoogleService::ClientService.query('vpn')
+      end
+
+      expect(described_class.new(result).total_link_count).to eq(113)
+    end
   end
 end