Delete duplicate records in different languages

Also logs a message when duplicates have different amounts, keeping the
first partial result and deleting the others.

lib/tasks/polls.rake

@@ -116,37 +116,58 @@ namespace :polls do
     logger.info "Removing duplicate partial results in polls"
 
     Tenant.run_on_each do
-      duplicate_ids = Poll::PartialResult.where(option_id: nil)
-                                         .select(:question_id, :booth_assignment_id, :date, :answer)
-                                         .group(:question_id, :booth_assignment_id, :date, :answer)
-                                         .having("count(*) > 1")
-                                         .pluck(:question_id, :booth_assignment_id, :date, :answer)
+      Poll::Question.find_each do |question|
+        manageable_titles = PollPartialResultOptionFinder.new(question).manageable_choices.keys
 
-      duplicate_ids.each do |question_id, booth_assignment_id, date, answer|
-        partial_results = Poll::PartialResult.where(
-          question_id: question_id,
+        question.question_options.each do |option|
+          titles = option.translations.where(title: manageable_titles).select(:title).distinct
+
+          groups = question.partial_results.where(option_id: nil, answer: titles)
+                           .select(:booth_assignment_id, :date)
+                           .group(:booth_assignment_id, :date)
+                           .having("count(*) > 1")
+                           .pluck(:booth_assignment_id, :date)
+
+          groups.each do |booth_assignment_id, date|
+            partial_results = question.partial_results.where(
+              option_id: nil,
               booth_assignment_id: booth_assignment_id,
               date: date,
-          answer: answer,
-          option_id: nil
+              answer: titles
             )
 
+            tenant_info = " on tenant #{Tenant.current_schema}" unless Tenant.default?
+
+            amounts_by_id = partial_results.pluck(:id, :amount).to_h
+            if amounts_by_id.values.uniq.size > 1
+              log_message = "Found duplicate partial results with different amounts " \
+                            "for question_id #{question.id}, " \
+                            "booth_assignment_id #{booth_assignment_id} " \
+                            "and date #{date}. " \
+                            "Keeping ID #{partial_results.first.id} " \
+                            "with amount #{partial_results.first.amount}. " \
+                            "Deleting partial results with these IDs and amounts: " \
+                            "#{amounts_by_id.except(partial_results.first.id)}" + tenant_info.to_s
+              logger.info(log_message)
+              duplicate_records_logger.info(log_message)
+            end
+
             partial_results.excluding(partial_results.first).each do |partial_result|
               partial_result.delete
 
-          tenant_info = " on tenant #{Tenant.current_schema}" unless Tenant.default?
               log_message = "Deleted duplicate record with ID #{partial_result.id} " \
                             "from the #{Poll::PartialResult.table_name} table " \
-                        "with question_id #{question_id}, " \
-                        "booth_assignment_id #{booth_assignment_id}, " \
-                        "date #{date} " \
-                        "and answer #{answer}" + tenant_info.to_s
+                            "with question_id #{question.id}, " \
+                            "booth_assignment_id #{booth_assignment_id} " \
+                            "and date #{date}" + tenant_info.to_s
               logger.info(log_message)
               duplicate_records_logger.info(log_message)
             end
           end
         end
       end
+    end
+  end
 
   desc "populates the poll_partial_results option_id column"
   task populate_partial_results_option_id: :remove_duplicate_partial_results do

spec/lib/tasks/polls_spec.rb

@@ -258,6 +258,60 @@ describe "polls tasks" do
       expect(Poll::PartialResult.all).to match_array [result, other_result, other_booth_result]
     end
 
+    it "removes duplicate partial results in different languages" do
+      question = create(:poll_question_multiple, max_votes: 2)
+
+      create(:poll_question_option, question: question, title_en: "Yes", title_de: "Ja")
+      create(:poll_question_option, question: question, title_en: "No",  title_de: "Nein")
+      create(:poll_question_option, question: question, title_en: "Maybe", title_de: "Vielleicht")
+
+      create(:poll_partial_result,
+             question: question,
+             booth_assignment_id: booth_assignment.id,
+             date: Date.current,
+             answer: "Yes",
+             option: nil)
+      create(:poll_partial_result,
+             question: question,
+             booth_assignment_id: booth_assignment.id,
+             date: Date.current,
+             answer: "Ja",
+             option: nil)
+
+      expect(Poll::PartialResult.count).to eq 2
+
+      Rake.application.invoke_task("polls:remove_duplicate_partial_results")
+
+      expect(Poll::PartialResult.count).to eq 1
+    end
+
+    it "does not remove duplicate partial results when many options are possible" do
+      question = create(:poll_question, title: "How do you pronounce it?")
+
+      create(:poll_question_option, question: question, title_en: "A", title_es: "EI")
+      create(:poll_question_option, question: question, title_en: "E", title_es: "I")
+      create(:poll_question_option, question: question, title_en: "I", title_es: "AI")
+
+      create(:poll_partial_result,
+             question: question,
+             booth_assignment_id: booth_assignment.id,
+             date: Date.current,
+             answer: "I",
+             option: nil)
+      create(:poll_partial_result,
+             question: question,
+             booth_assignment_id: booth_assignment.id,
+             date: Date.current,
+             answer: "AI",
+             option: nil)
+
+      expect(Poll::PartialResult.count).to eq 2
+
+      Rake.application.invoke_task("polls:remove_duplicate_partial_results")
+
+      expect(Poll::PartialResult.count).to eq 2
+    end
+
     it "does not remove partial results with the same text and different options" do
       question = create(:poll_question_multiple, :abcde, max_votes: 4)
       option_a = question.question_options.find_by(title: "Answer A")
@@ -305,6 +359,34 @@ describe "polls tasks" do
         expect(Poll::PartialResult.count).to eq 1
       end
     end
+
+    it "removes duplicates in different languages even when amounts differ" do
+      question = create(:poll_question_multiple, max_votes: 2)
+
+      create(:poll_question_option, question: question, title_en: "Yes", title_de: "Ja")
+
+      create(:poll_partial_result,
+             question: question,
+             booth_assignment_id: booth_assignment.id,
+             date: Date.current,
+             answer: "Yes",
+             option: nil,
+             amount: 3)
+
+      create(:poll_partial_result,
+             question: question,
+             booth_assignment_id: booth_assignment.id,
+             date: Date.current,
+             answer: "Ja",
+             option: nil,
+             amount: 5)
+
+      expect(Poll::PartialResult.count).to eq 2
+
+      Rake.application.invoke_task("polls:remove_duplicate_partial_results")
+
+      expect(Poll::PartialResult.count).to eq 1
+    end
   end
 
   describe "polls:populate_partial_results_option_id" do
@@ -381,6 +463,11 @@ describe "polls tasks" do
     it "removes duplicate partial results before populating the option_id column" do
       question = create(:poll_question_multiple, :abc)
 
+      localized_question = create(:poll_question_multiple)
+      create(:poll_question_option, question: localized_question, title_en: "Yes", title_de: "Ja")
+      create(:poll_question_option, question: localized_question, title_en: "No",  title_de: "Nein")
+      create(:poll_question_option, question: localized_question, title_en: "Maybe", title_de: "Vielleicht")
+
       result_attributes = {
         question_id: question.id,
         booth_assignment_id: booth_assignment.id,
@@ -391,14 +478,28 @@ describe "polls tasks" do
       result = create(:poll_partial_result, result_attributes)
       insert(:poll_partial_result, result_attributes)
 
+      localized_result_attributes = {
+        question: localized_question,
+        booth_assignment_id: booth_assignment.id,
+        date: Date.current,
+        option: nil
+      }
+      localized_result = create(:poll_partial_result, localized_result_attributes.merge(answer: "Yes"))
+      create(:poll_partial_result, localized_result_attributes.merge(answer: "Ja"))
+
       result.reload
+      localized_result.reload
+
       expect(result.option_id).to be nil
+      expect(localized_result.option_id).to be nil
 
       Rake.application.invoke_task("polls:populate_partial_results_option_id")
       result.reload
+      localized_result.reload
 
-      expect(Poll::PartialResult.count).to eq 1
+      expect(Poll::PartialResult.count).to eq 2
       expect(result.option_id).to eq question.question_options.find_by(title: "Answer A").id
+      expect(localized_result.option_id).to eq localized_question.question_options.find_by(title: "Yes").id
     end
 
     it "populates the option_id column on tenants" do