feat: extract ratings

Author: kjellberg

lib/serp_parser/elements/google/organic_result.rb

@@ -31,6 +31,10 @@ def root_domain
           PublicSuffix.parse(domain).domain
         end
 
+        def rating
+          nil
+        end
+
         def to_h
           {
             "position" => position,
@@ -39,7 +43,8 @@ def to_h
             "url" => url,
             "domain" => domain,
             "root_domain" => root_domain,
-            "_serp_parser_meta" => meta
+            "_serp_parser_meta" => meta,
+            "rating" => rating
           }
         end
       end

lib/serp_parser/elements/google/variants/organic_result1.rb

@@ -11,6 +11,20 @@ def description
             remove_span_elements(match)
           end
 
+          def rating
+            element = element_from_selector(".r0bn4c.rQMQod.tP9Zud")
+            text = element.text.strip # "4,0  (90)"
+            return nil if text.empty?
+
+            matches = text.match(/(\d+,\d+)\s+\((\d+)\)/)
+
+            {
+              "value" => matches[1].gsub(",", ".").to_f,
+              "max_value" => 5,
+              "votes" => matches[2].to_i
+            }
+          end
+
           private
 
           # Find the text node that contains the description text

lib/serp_parser/elements/google/variants/organic_result2.rb

@@ -9,6 +9,29 @@ def description
             element = element_from_selector(".VwiC3b")
             element.text
           end
+
+          # Returns the rating of the result
+          # @return [String]
+          def rating
+            element = element_from_selector(".fG8Fp.uo4vr .ChPIuf")
+
+            text = element&.text&.strip
+            matches = text.match(/(\d+\.\d+)\/(\d+) · ‎([\d,]+)/)
+
+            return nil if matches.nil?
+
+            value = matches[1].to_f
+            max_value = matches[2].to_i
+
+            # Use split with a regular expression to divide the string at each comma or dot, then join to concatenate the parts
+            votes = matches[3].split(/[,\.]/).join.to_i
+
+            {
+              "value" => value,
+              "max_value" => max_value,
+              "votes" => votes
+            }
+          end
         end
       end
     end

lib/serp_parser/helpers.rb

@@ -4,7 +4,7 @@ module Helpers
     # @param selector [String]
     # @return [Nokogiri::XML::Element]
     def element_from_selector(selector)
-      @element.css(selector)
+      elements = @element.css(selector)
     end
 
     # Extract text from selector

spec/files/google/1/organic_results/organic_result.json

@@ -5,6 +5,7 @@
   "url": "https://dalaavfall.se/sorteringsguide/plastforpackning/plastkasse/",
   "domain": "dalaavfall.se",
   "root_domain": "dalaavfall.se",
+  "rating": null,
   "_serp_parser_meta": {
     "template": "SerpParser::Elements::Google::Variants::OrganicResult1"
   }

spec/files/google/1/organic_results/organic_result_with_date.json

@@ -5,6 +5,11 @@
   "url": "https://www.testin.se/hushall/matkasse",
   "domain": "www.testin.se",
   "root_domain": "testin.se",
+  "rating": {
+    "value": 4.7,
+    "max_value": 5,
+    "votes": 52
+  },
   "_serp_parser_meta": {
     "template": "SerpParser::Elements::Google::Variants::OrganicResult1"
   }

spec/files/google/1/organic_results/organic_result_with_rating.json

@@ -5,6 +5,11 @@
   "url": "https://www.jordklok.se/sv/articles/2.113.6157/omom-ekologisk-matkasse",
   "domain": "www.jordklok.se",
   "root_domain": "jordklok.se",
+  "rating": {
+    "value": 4.0,
+    "max_value": 5,
+    "votes": 90
+  },
   "_serp_parser_meta": {
     "template": "SerpParser::Elements::Google::Variants::OrganicResult1"
   }

spec/files/google/1/organic_results/organic_result_with_sitelinks.json

@@ -5,6 +5,7 @@
   "url": "https://www.linasmatkasse.se/",
   "domain": "www.linasmatkasse.se",
   "root_domain": "linasmatkasse.se",
+  "rating": null,
   "_serp_parser_meta": {
     "template": "SerpParser::Elements::Google::Variants::OrganicResult1"
   }

spec/files/google/2/organic_results/organic_result.json

@@ -5,6 +5,7 @@
   "url": "https://www.whitehouse.gov/administration/president-biden/",
   "domain": "www.whitehouse.gov",
   "root_domain": "whitehouse.gov",
+  "rating": null,
   "_serp_parser_meta": {
     "template": "SerpParser::Elements::Google::Variants::OrganicResult2"
   }

spec/files/google/2/organic_results/organic_result_with_rating.json

@@ -5,6 +5,11 @@
   "url": "https://www.imdb.com/title/tt0112346/",
   "domain": "www.imdb.com",
   "root_domain": "imdb.com",
+  "rating": {
+    "value": 6.8,
+    "max_value": 10,
+    "votes": 60953
+  },
   "_serp_parser_meta": {
     "template": "SerpParser::Elements::Google::Variants::OrganicResult2"
   }

spec/files/google/2/organic_results/top_result.json

@@ -5,6 +5,7 @@
   "url": "https://en.wikipedia.org/wiki/President_of_the_United_States#:~:text=In%20all%2C%2045%20individuals%20have,office%20on%20January%2020%2C%202021.",
   "domain": "en.wikipedia.org",
   "root_domain": "wikipedia.org",
+  "rating": null,
   "_serp_parser_meta": {
     "template": "SerpParser::Elements::Google::Variants::OrganicTopResult1"
   }