Add task to delete duplicate poll partial results

Adds rake task "polls:remove_duplicate_partial_results" to delete duplicated
rows in "poll_partial_results" made before the DB was strict about duplicates.

Duplicates are considered only for records without "option_id", grouping by:
(question_id, booth_assignment_id, date, answer). We keep the first one and
delete the rest, per tenant.

The controller use:
  Poll::PartialResult.find_or_initialize_by(booth_assignment_id, date, question_id, answer)
which is not a strong protection against race conditions. Without a unique
index at the DB level, duplicates could be created. This task cleans up any
existing duplicates.

lib/tasks/consul.rake

@@ -3,8 +3,10 @@ namespace :consul do
   task execute_release_tasks: ["settings:rename_setting_keys",
                                "settings:add_new_settings",
                                "cache:clear",
-                               "execute_release_2.3.0_tasks"]
+                               "execute_release_2.4.0_tasks"]
 
-  desc "Runs tasks needed to upgrade from 2.2.2 to 2.3.0"
+  desc "Runs tasks needed to upgrade from 2.3.1 to 2.4.0"
-  task "execute_release_2.3.0_tasks": []
+  task "execute_release_2.4.0_tasks": [
+    "polls:remove_duplicate_partial_results"
+  ]
 end

lib/tasks/polls.rake

@@ -107,4 +107,44 @@ namespace :polls do
       end
     end
   end
+
+  desc "Removes duplicate poll partial results"
+  task remove_duplicate_partial_results: :environment do
+    logger = ApplicationLogger.new
+    duplicate_records_logger = DuplicateRecordsLogger.new
+
+    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)
+
+      duplicate_ids.each do |question_id, booth_assignment_id, date, answer|
+        partial_results = Poll::PartialResult.where(
+          question_id: question_id,
+          booth_assignment_id: booth_assignment_id,
+          date: date,
+          answer: answer,
+          option_id: nil
+        )
+
+        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
+          logger.info(log_message)
+          duplicate_records_logger.info(log_message)
+        end
+      end
+    end
+  end
 end

spec/lib/tasks/polls_spec.rb

@@ -223,4 +223,88 @@ describe "polls tasks" do
       end
     end
   end
+
+  describe "polls:remove_duplicate_partial_results" do
+    before { Rake::Task["polls:remove_duplicate_partial_results"].reenable }
+
+    let(:booth_assignment) { create(:poll_booth_assignment) }
+    let(:other_booth_assignment) { create(:poll_booth_assignment) }
+
+    it "removes duplicate partial results" do
+      question = create(:poll_question_multiple, :abcde, poll: poll, max_votes: 4)
+
+      result_attributes = {
+        question_id: question.id,
+        booth_assignment_id: booth_assignment.id,
+        date: Date.current,
+        answer: "Answer A",
+        option_id: nil
+      }
+      other_result_attributes = result_attributes.merge(answer: "Answer B")
+
+      result = create(:poll_partial_result, result_attributes)
+      other_result = create(:poll_partial_result, other_result_attributes)
+      other_booth_result = create(:poll_partial_result,
+                                  result_attributes.merge(booth_assignment_id: other_booth_assignment.id))
+
+      2.times { insert(:poll_partial_result, result_attributes) }
+      insert(:poll_partial_result, other_result_attributes)
+      insert(:poll_partial_result, result_attributes.merge(booth_assignment_id: other_booth_assignment.id))
+
+      expect(Poll::PartialResult.count).to eq 7
+
+      Rake.application.invoke_task("polls:remove_duplicate_partial_results")
+
+      expect(Poll::PartialResult.count).to eq 3
+      expect(Poll::PartialResult.all).to match_array [result, other_result, other_booth_result]
+    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")
+      option_b = question.question_options.find_by(title: "Answer B")
+
+      result_attributes = {
+        question: question,
+        booth_assignment_id: booth_assignment.id,
+        date: Date.current,
+        answer: "Answer A"
+      }
+      create(:poll_partial_result, result_attributes.merge(option: option_a))
+      insert(:poll_partial_result, result_attributes.merge(option_id: option_b.id))
+
+      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 "removes duplicate partial results on tenants" do
+      create(:tenant, schema: "partial_results")
+
+      Tenant.switch("partial_results") do
+        question = create(:poll_question_multiple, :abc)
+        booth_assignment = create(:poll_booth_assignment)
+
+        result_attributes = {
+          question_id: question.id,
+          booth_assignment_id: booth_assignment.id,
+          date: Date.current,
+          answer: "Answer A",
+          option_id: nil
+        }
+        create(:poll_partial_result, result_attributes)
+        insert(:poll_partial_result, result_attributes)
+
+        expect(Poll::PartialResult.count).to eq 2
+      end
+
+      Rake.application.invoke_task("polls:remove_duplicate_partial_results")
+
+      Tenant.switch("partial_results") do
+        expect(Poll::PartialResult.count).to eq 1
+      end
+    end
+  end
 end