handle spreadsheet dupes in sales

app/services/bulk_upload/sales/validator.rb

@@ -12,9 +12,11 @@ class BulkUpload::Sales::Validator
   end
 
   def call
-    row_parsers.each_with_index do |row_parser, index|
+    row_parsers.each(&:valid?)
-      row_parser.valid?
+
+    validate_duplicate_rows
 
+    row_parsers.each_with_index do |row_parser, index|
       row = index + row_offset + 1
 
       row_parser.errors.each do |error|
@@ -66,6 +68,19 @@ class BulkUpload::Sales::Validator
 
 private
 
+  # n^2 algo
+  def validate_duplicate_rows
+    row_parsers.each do |rp|
+      dupe = row_parsers.reject { |r| r.object_id.equal?(rp.object_id) }.any? do |rp_counter|
+        rp.spreadsheet_duplicate_hash == rp_counter.spreadsheet_duplicate_hash
+      end
+
+      if dupe
+        rp.add_duplicate_found_in_spreadsheet_errors
+      end
+    end
+  end
+
   def any_logs_invalid?
     row_parsers.any? { |row_parser| row_parser.log.invalid? }
   end

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

@@ -411,6 +411,27 @@ class BulkUpload::Sales::Year2022::RowParser
     field_1
   end
 
+  def spreadsheet_duplicate_hash
+    attributes.slice(
+      "field_1",   # purcahser code
+      "field_2",   # saledate
+      "field_3",   # saledate
+      "field_4",   # saledate
+      "field_7",   # age1
+      "field_13",  # sex1
+      "field_24",  # ecstat1
+      "field_54",  # postcode
+      "field_55",  # postcode
+      "field_92",  # owning org
+    )
+  end
+
+  def add_duplicate_found_in_spreadsheet_errors
+    spreadsheet_duplicate_hash.each_key do |field|
+      errors.add(field, "Duplicate row found in spreadsheet", category: :setup)
+    end
+  end
+
 private
 
   def validate_data_protection_answered

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

@@ -533,6 +533,27 @@ class BulkUpload::Sales::Year2023::RowParser
     field_6
   end
 
+  def spreadsheet_duplicate_hash
+    attributes.slice(
+      "field_1",  # owning org
+      "field_3",  # saledate
+      "field_4",  # saledate
+      "field_5",  # saledate
+      "field_6",  # purchaser_code
+      "field_24", # postcode
+      "field_25", # postcode
+      "field_30", # age1
+      "field_31", # sex1
+      "field_35", # ecstat1
+    )
+  end
+
+  def add_duplicate_found_in_spreadsheet_errors
+    spreadsheet_duplicate_hash.each_key do |field|
+      errors.add(field, "Duplicate row found in spreadsheet", category: :setup)
+    end
+  end
+
 private
 
   def validate_data_protection_answered

spec/factories/sales_log.rb

@@ -36,6 +36,7 @@ FactoryBot.define do
       jointpur { 2 }
     end
     trait :completed do
+      purchid { rand(999_999_999).to_s }
       ownershipsch { 2 }
       type { 8 }
       saledate { Time.utc(2023, 2, 2, 10, 36, 49) }

spec/services/bulk_upload/sales/validator_spec.rb

@@ -83,6 +83,22 @@ RSpec.describe BulkUpload::Sales::Validator do
         expect { validator.call }.to change(BulkUploadError, :count)
       end
     end
+
+    context "when duplicate rows present" do
+      let(:file) { Tempfile.new }
+      let(:path) { file.path }
+      let(:log) { build(:sales_log, :completed) }
+
+      before do
+        file.write(BulkUpload::SalesLogToCsv.new(log:, col_offset: 0).to_2022_csv_row)
+        file.write(BulkUpload::SalesLogToCsv.new(log:, col_offset: 0).to_2022_csv_row)
+        file.close
+      end
+
+      it "creates errors" do
+        expect { validator.call }.to change(BulkUploadError.where(category: :setup, error: "Duplicate row found in spreadsheet"), :count).by(20)
+      end
+    end
   end
 
   describe "#create_logs?" do

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

@@ -804,4 +804,16 @@ RSpec.describe BulkUpload::Sales::Year2022::RowParser do
       end
     end
   end
+
+  describe "#spreadsheet_duplicate_hash" do
+    it "returns a hash" do
+      expect(parser.spreadsheet_duplicate_hash).to be_a(Hash)
+    end
+  end
+
+  describe "#add_duplicate_found_in_spreadsheet_errors" do
+    it "adds errors" do
+      expect { parser.add_duplicate_found_in_spreadsheet_errors }.to change(parser.errors, :size)
+    end
+  end
 end

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

@@ -1045,4 +1045,16 @@ RSpec.describe BulkUpload::Sales::Year2023::RowParser do
       end
     end
   end
+
+  describe "#spreadsheet_duplicate_hash" do
+    it "returns a hash" do
+      expect(parser.spreadsheet_duplicate_hash).to be_a(Hash)
+    end
+  end
+
+  describe "#add_duplicate_found_in_spreadsheet_errors" do
+    it "adds errors" do
+      expect { parser.add_duplicate_found_in_spreadsheet_errors }.to change(parser.errors, :size)
+    end
+  end
 end

spec/services/csv/sales_log_csv_service_spec.rb

@@ -3,7 +3,7 @@ require "rails_helper"
 RSpec.describe Csv::SalesLogCsvService do
   let(:form_handler_mock) { instance_double(FormHandler) }
   let(:organisation) { create(:organisation) }
-  let!(:log) { create(:sales_log, :completed, owning_organisation: organisation) }
+  let!(:log) { create(:sales_log, :completed, owning_organisation: organisation, purchid: nil) }
   let(:service) { described_class.new(export_type: "labels") }
   let(:csv) { CSV.parse(service.prepare_csv(SalesLog.all)) }