Add deduplication to the 23/24 sales

app/services/bulk_upload/sales/year2023/row_parser.rb

@@ -410,6 +410,7 @@ class BulkUpload::Sales::Year2023::RowParser
   validate :validate_address_line_1, on: :after_log
   validate :validate_town_or_city, on: :after_log
   validate :validate_postcode, on: :after_log
+  validate :validate_if_log_already_exists, on: :after_log, if: -> { FeatureToggle.bulk_upload_duplicate_log_check_enabled? }
 
   def self.question_for_field(field)
     QUESTIONS[field]
@@ -466,6 +467,12 @@ class BulkUpload::Sales::Year2023::RowParser
     "#<BulkUpload::Sales::Year2023::RowParser:#{object_id}>"
   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
+
 private
 
   def prevtenbuy2
@@ -786,7 +793,7 @@ private
 
     attributes["othtype"] = field_11
 
-    attributes["owning_organisation_id"] = owning_organisation_id
+    attributes["owning_organisation"] = owning_organisation
     attributes["created_by"] = created_by || bulk_upload.user
     attributes["hhregres"] = hhregres
     attributes["hhregresstill"] = hhregresstill
@@ -1000,10 +1007,6 @@ private
     Organisation.find_by_id_on_multiple_fields(field_1)
   end
 
-  def owning_organisation_id
-    owning_organisation&.id
-  end
-
   def created_by
     @created_by ||= User.find_by(email: field_2)
   end
@@ -1035,6 +1038,18 @@ 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 validate_owning_org_data_given
     if field_1.blank?
       block_log_creation!
@@ -1156,4 +1171,21 @@ private
       errors.add(:field_5, I18n.t("validations.date.outside_collection_window", year_combo: bulk_upload.year_combo, start_year: bulk_upload.year, end_year: bulk_upload.end_year), category: :setup)
     end
   end
+
+  def validate_if_log_already_exists
+    if log_already_exists?
+      error_message = "This is a duplicate log"
+
+      errors.add(:field_1, error_message) # Owning org
+      errors.add(:field_3, error_message) # Sale completion date
+      errors.add(:field_4, error_message) # Sale completion date
+      errors.add(:field_5, error_message) # Sale completion date
+      errors.add(:field_24, error_message) # Postcode
+      errors.add(:field_25, error_message) # Postcode
+      errors.add(:field_30, error_message) # Buyer 1 age
+      errors.add(:field_31, error_message) # Buyer 1 gender
+      errors.add(:field_35, error_message) # Buyer 1 working situation
+      errors.add(:field_6, error_message) # Purchaser code
+    end
+  end
 end

spec/services/bulk_upload/sales/year2023/row_parser_spec.rb

@@ -467,6 +467,71 @@ RSpec.describe BulkUpload::Sales::Year2023::RowParser do
       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_1, # Owning org
+          :field_3, # Sale completion date
+          :field_4, # Sale completion date
+          :field_5, # Sale completion date
+          :field_24, # Postcode
+          :field_25, # Postcode
+          :field_30, # Buyer 1 age
+          :field_31, # Buyer 1 gender
+          :field_35, # Buyer 1 working situation
+          :field_6, # 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_1, # Owning org
+          :field_3, # Sale completion date
+          :field_4, # Sale completion date
+          :field_5, # Sale completion date
+          :field_24, # Postcode
+          :field_25, # Postcode
+          :field_30, # Buyer 1 age
+          :field_31, # Buyer 1 gender
+          :field_35, # Buyer 1 working situation
+          :field_6, # Purchaser code
+        ].each do |field|
+          expect(parser.errors[field]).to be_blank
+        end
+      end
+    end
+
     describe "#field_19" do # UPRN
       context "when UPRN known" do
         let(:attributes) { setup_section_params.merge({ field_19: "100023336956" }) }