Deduplicate 2022 sales logs

app/services/bulk_upload/sales/year2022/row_parser.rb

@@ -285,6 +285,7 @@ class BulkUpload::Sales::Year2022::RowParser
   validate :validate_created_by_related, on: :after_log
   validate :validate_relevant_collection_window, on: :after_log
   validate :validate_incomplete_soft_validations, on: :after_log
+  validate :validate_if_log_already_exists, on: :after_log
 
   def self.question_for_field(field)
     QUESTIONS[field]
@@ -848,6 +849,24 @@ private
     @questions ||= log.form.subsections.flat_map { |ss| ss.applicable_questions(log) }
   end
 
+  def duplicate_check_fields
+    %w[
+      saledate
+      age1
+      sex1
+      ecstat1
+      owning_organisation
+      postcode_full
+      purchid
+    ]
+  end
+
+  def log_already_exists?
+    @log_already_exists ||= SalesLog
+      .where(status: %w[not_started in_progress completed])
+      .exists?(duplicate_check_fields.index_with { |field| log.public_send(field) })
+  end
+
   def validate_owning_org_data_given
     if field_92.blank?
       block_log_creation!
@@ -985,4 +1004,21 @@ private
       end
     end
   end
+
+  def validate_if_log_already_exists
+    if log_already_exists?
+      error_message = "This is a duplicate log"
+
+      errors.add(:field_92, error_message) # Owning org
+      errors.add(:field_2, error_message) # Sale completion date
+      errors.add(:field_3, error_message) # Sale completion date
+      errors.add(:field_4, error_message) # Sale completion date
+      errors.add(:field_41, error_message) # Postcode
+      errors.add(:field_42, error_message) # Postcode
+      errors.add(:field_7, error_message) # Buyer 1 age
+      errors.add(:field_13, error_message) # Buyer 1 gender
+      errors.add(:field_24, error_message) # Buyer 1 working situation
+      errors.add(:field_1, error_message) # Purchaser code
+    end
+  end
 end

spec/services/bulk_upload/sales/year2022/row_parser_spec.rb

@@ -539,6 +539,71 @@ RSpec.describe BulkUpload::Sales::Year2022::RowParser do
         end
       end
     end
+
+    context "when the log already exists in the db" do
+      let(:attributes) { valid_attributes }
+
+      before do
+        parser.log.save!
+        parser.instance_variable_set(:@valid, nil)
+      end
+
+      it "is not a valid row" do
+        expect(parser).not_to be_valid
+      end
+
+      it "adds an error to all (and only) the fields used to determine duplicates" do
+        parser.valid?
+
+        error_message = "This is a duplicate log"
+
+        [
+          :field_92, # Owning org
+          :field_2, # Sale completion date
+          :field_3, # Sale completion date
+          :field_4, # Sale completion date
+          :field_41, # Postcode
+          :field_42, # Postcode
+          :field_7, # Buyer 1 age
+          :field_13, # Buyer 1 gender
+          :field_24, # Buyer 1 working situation
+          :field_1, # Purchaser code
+        ].each do |field|
+          expect(parser.errors[field]).to include(error_message)
+        end
+      end
+    end
+
+    context "when a hidden log already exists in db" do
+      before do
+        parser.log.status = "pending"
+        parser.log.skip_update_status = true
+        parser.log.save!
+      end
+
+      it "is a valid row" do
+        expect(parser).to be_valid
+      end
+
+      it "does not add duplicate errors" do
+        parser.valid?
+
+        [
+          :field_92, # Owning org
+          :field_2, # Sale completion date
+          :field_3, # Sale completion date
+          :field_4, # Sale completion date
+          :field_41, # Postcode
+          :field_42, # Postcode
+          :field_7, # Buyer 1 age
+          :field_13, # Buyer 1 gender
+          :field_24, # Buyer 1 working situation
+          :field_1, # Purchaser code
+        ].each do |field|
+          expect(parser.errors[field]).to be_blank
+        end
+      end
+    end
   end
 
   describe "inferences" do