Add specs for google parser

alice/handlers/alpha.rb

@@ -11,6 +11,7 @@ def answer_question
     private
 
     def answer
+      # Try Wolfram Alpha first, then fall back to Google
       if results = Parser::Alpha.new(sentence).answer
         results.first
       elsif answers.any?
@@ -19,16 +20,7 @@ def answer
     end
 
     def answers
-      @answers ||= Parser::Google.new(sentence).all_answers.map do |answer|
-        if answer.include?("...")
-          answer = answer.split("...")[1] || ""
-        end
-        if answer.scan(/\. [a-zA-Z]+/).any?
-          answer.split('.')[0..-2].join(' ')
-        else
-          answer
-        end
-      end.compact
+      @answers ||= Parser::Google.new(sentence).all_answers
     end
 
     def sentence

alice/parser/google.rb

@@ -6,59 +6,52 @@ class Google
     attr_accessor :question
 
     def self.fetch(topic)
-      new(topic).answer
+      new(topic).sorted_answers.first
     end
 
     def self.fetch_all(topic)
-      new(topic).all_answers
+      new(topic).sorted_answers
     end
 
     def initialize(question)
       @question = question.gsub("+", "plus").gsub(" ", "+").encode("ASCII", invalid: :replace, undef: :replace, replace: '')
     end
 
-    def answer
-      results.first
-    end
-
-    def all_answers
-      answers
+    def sorted_answers
+      answers = sanitized_answers(full_search + reductivist_search)
+      Grammar::DeclarativeSorter.sort(query: question, corpus: answers)
     end
 
     private
 
-    def answers
-      answers = (full_search + reductivist_search).compact.map{ |a| a.encode('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '') }
-      answers.reject!{ |a| a =~ /\.\.\./}
-    rescue Exception => e
-      Alice::Util::Logger.info "*** Parser::Google: Unable to process \"#{self.question}\": #{e}"
-      Alice::Util::Logger.info e.backtrace
-      return ["Hmm, that part of my brain is returning a #{e}. Google is getting suspicious. You should probably rotate my IP address again."]
-    end
-
-    def results
-      best_answer = answers.any? ? Grammar::DeclarativeSorter.sort(query: question, corpus: answers).first : ""
-      Alice::Util::Logger.info "*** Parser::Google: Answered \"#{self.question}\" with #{best_answer}"
-      return best_answer
-    rescue Exception => e
-      Alice::Util::Logger.info "*** Parser::Google: Unable to process \"#{self.question}\": #{e}"
-      Alice::Util::Logger.info e.backtrace
-      return ["Hmm, that part of my brain is returning a #{e}"]
+    def sanitized_answers(answer_array)
+      answer_array.map do |answer|
+        answer = answer.encode('UTF-8', 'binary', invalid: :replace, undef: :replace, replace: '')
+        answer.delete!("\n")
+        if answer.include?("...")
+          answer = answer.split("...").select{ |c| c =~ /\.$/ }.first.to_s
+        end
+        answer = answer.gsub(/^ /,'')
+      end.compact.uniq#.reject{ |a| a =~ /\.\.\./}
     end
 
     def full_search
-      doc = Nokogiri::HTML(open("https://www.google.com/search?q=#{question}&hl=lang_en"))
-      doc.css("div span.st").map(&:text)
+      search(question)
     end
 
     def reductivist_search
-      doc = Nokogiri::HTML(open("https://www.google.com/search?q=#{simplified_question}&hl=lang_en"))
-      doc.css("div span.st").map(&:text)
+      parsed_question = Grammar::SentenceParser.parse(question)
+      query = "wikipedia what is #{(parsed_question.nouns + parsed_question.adjectives).join(' ')}"
+      search(query)
     end
 
-    def simplified_question
-      parsed_question = Grammar::SentenceParser.parse(question)
-      "wikipedia what is #{(parsed_question.nouns + parsed_question.adjectives).join(' ')}"
+    def search(query)
+      doc = Nokogiri::HTML(open("https://www.google.com/search?q=#{query}&hl=lang_en"))
+      doc.css("div span.st").map(&:text)
+    rescue Exception => e
+      Alice::Util::Logger.info "*** Parser::Google: Unable to process \"#{self.question}\": #{e}"
+      Alice::Util::Logger.info e.backtrace
+      return ["Hmm, I can't really answer that since Google is getting suspicious of me. You should probably rotate my IP address again."]
     end
 
   end

spec/parsers/google_spec.rb

@@ -0,0 +1,45 @@
+require 'spec_helper'
+
+describe "Parser::Google" do
+
+  let(:results) {
+    [
+      "The latest Tweets from Coraline Ada Ehmke (@CoralineAda). Code witch. Ruby \nHero. Speaker, writer, podcaster, activist. Creator of the Contributor Covenant.",
+      "Coraline Ada Ehmke. Coraline Ada Ehmke is a software developer and open source advocate based in Chicago, Illinois. She began her career as a web developer in 1994 and has worked in a variety of industries, including engineering, consulting, education, advertising, healthcare, and software development infrastructure.",
+      "Coraline Ada Ehmke. Notable Rubyist, Nefarious Code Witch, And Notorious \nSocial Justice Warrior. Coraline Ada Ehmke. I am a well-known speaker, writer, ...",
+      "Coraline Ada Ehmke | November 30, 2015. On June 18 of this year a friend on \nIRC expressed his frustration with tweets by a person named Elia (@elia).",
+      "Code witch. Ruby Hero. Speaker, writer, activist. @CoralineAda on Twitter. - \nCoralineAda.",
+      "Jul 6, 2017 ... Well-known programmer Coraline Ada Ehmke was fired from her job at GitHub \nand she had a lot to say about how, and why, she was let go.",
+      "Feb 26, 2016 ... Deconstructing Coraline Ada Ehmke's “Contributor Covenant”, And Why It's \nFoolish. Recently, this happened, where the subject of this article ...",
+      "May 17, 2018 ... CORALINE ADA EHMKE. Principal Engineer, Stitch Fix. Coraline Ada Ehmke is \nan open-source advocate and developer with over 20 years of ...",
+      "Apr 14, 2016 ... Coraline Ada Ehmke is known for the creation of the “Contributor Covenant,” an \nSJW Code ... GITHUB hires Ehmke to work on anti-harassment.",
+      "Dec 29, 2016 - 27 min - Uploaded by Trans  HackathonCoraline keynote's Trans*H4CK's first online conference with a talk on diversity in  tech and ..."
+    ]
+  }
+  let(:parser) { Parser::Google.new("who is coraline ehmke") }
+
+  before do
+    allow(Grammar::DeclarativeSorter).to receive(:sort).and_return(results)
+    allow(parser).to receive(:full_search).and_return(results[0..4])
+    allow(parser).to receive(:reductivist_search).and_return(results[5..9])
+  end
+
+  describe "#sanitized_answers" do
+
+    it "strips out newlines" do
+      answer = parser.send(:sanitized_answers, [results[0]]).first
+      expect(answer).to eq("The latest Tweets from Coraline Ada Ehmke (@CoralineAda). Code witch. Ruby Hero. Speaker, writer, podcaster, activist. Creator of the Contributor Covenant.")
+    end
+
+    it "splits at ellipses" do
+      answer = parser.send(:sanitized_answers, [results[5]]).first
+      expect(answer).to eq("Well-known programmer Coraline Ada Ehmke was fired from her job at GitHub and she had a lot to say about how, and why, she was let go.")
+    end
+
+    it "grabs a complete sentence when there are ellipses" do
+      answer = parser.send(:sanitized_answers, [results[8]]).first
+      expect(answer).to eq("GITHUB hires Ehmke to work on anti-harassment.")
+    end
+
+  end
+end