Copy 23 BU files to 24

app/services/bulk_upload/lettings/year2024/csv_parser.rb

@@ -0,0 +1,113 @@
+require "csv"
+
+class BulkUpload::Lettings::Year2024::CsvParser
+  include CollectionTimeHelper
+
+  FIELDS = 134
+  MAX_COLUMNS = 142
+  FORM_YEAR = 2024
+
+  attr_reader :path
+
+  def initialize(path:)
+    @path = path
+  end
+
+  def row_offset
+    if with_headers?
+      rows.find_index { |row| row[0].match(/field number/i) } + 1
+    else
+      0
+    end
+  end
+
+  def col_offset
+    with_headers? ? 1 : 0
+  end
+
+  def cols
+    @cols ||= ("A".."EL").to_a
+  end
+
+  def row_parsers
+    @row_parsers ||= body_rows.map do |row|
+      stripped_row = row[col_offset..]
+      hash = Hash[field_numbers.zip(stripped_row)]
+
+      BulkUpload::Lettings::Year2024::RowParser.new(hash)
+    end
+  end
+
+  def body_rows
+    rows[row_offset..]
+  end
+
+  def rows
+    @rows ||= CSV.parse(normalised_string, row_sep:)
+  end
+
+  def column_for_field(field)
+    cols[field_numbers.find_index(field) + col_offset]
+  end
+
+  def correct_field_count?
+    valid_field_numbers_count = field_numbers.count { |f| f != "field_blank" }
+
+    valid_field_numbers_count == FIELDS
+  end
+
+  def too_many_columns?
+    return if with_headers?
+
+    max_columns_count = body_rows.map(&:size).max - col_offset
+
+    max_columns_count > MAX_COLUMNS
+  end
+
+  def wrong_template_for_year?
+    collection_start_year_for_date(first_record_start_date) != FORM_YEAR
+  rescue Date::Error
+    false
+  end
+
+private
+
+  def default_field_numbers
+    [5, nil, nil, 15, 16, nil, 13, 40, 41, 42, 43, 46, 52, 56, 60, 64, 68, 72, 76, 47, 53, 57, 61, 65, 69, 73, 77, 51, 55, 59, 63, 67, 71, 75, 50, 54, 58, 62, 66, 70, 74, 78, 48, 49, 79, 81, 82, 123, 124, 122, 120, 102, 103, nil, 83, 84, 85, 86, 87, 88, 104, 109, 107, 108, 106, 100, 101, 105, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 126, 128, 129, 130, 131, 132, 127, 125, 133, 134, 33, 34, 35, 36, 37, 38, nil, 7, 8, 9, 28, 14, 32, 29, 30, 31, 26, 27, 25, 23, 24, nil, 1, 3, 2, 80, nil, 121, 44, 89, 98, 92, 95, 90, 91, 93, 94, 97, 96, 99, 10, 11, 12, 45, 39, 6, 4, 17, 18, 19, 20, 21, 22].map { |h| h.present? && h.to_s.match?(/^[0-9]+$/) ? "field_#{h}" : "field_blank" }
+  end
+
+  def field_numbers
+    @field_numbers ||= if with_headers?
+                         rows[row_offset - 1][col_offset..].map { |h| h.present? && h.match?(/^[0-9]+$/) ? "field_#{h}" : "field_blank" }
+                       else
+                         default_field_numbers
+                       end
+  end
+
+  def with_headers?
+    rows.map { |r| r[0] }.any? { |cell| cell&.match?(/field number/i) }
+  end
+
+  def row_sep
+    "\n"
+  end
+
+  def normalised_string
+    return @normalised_string if @normalised_string
+
+    @normalised_string = File.read(path, encoding: "bom|utf-8")
+    @normalised_string.gsub!("\r\n", "\n")
+    @normalised_string.scrub!("")
+    @normalised_string.tr!("\r", "\n")
+
+    @normalised_string
+  end
+
+  def first_record_start_date
+    if with_headers?
+      Date.new(row_parsers.first.field_98.to_i + 2000, row_parsers.first.field_97.to_i, row_parsers.first.field_96.to_i)
+    else
+      Date.new(rows.first[97].to_i + 2000, rows.first[96].to_i, rows.first[95].to_i)
+    end
+  end
+end

spec/services/bulk_upload/lettings/year2024/csv_parser_spec.rb

@@ -0,0 +1,226 @@
+require "rails_helper"
+
+RSpec.describe BulkUpload::Lettings::Year2024::CsvParser do
+  subject(:service) { described_class.new(path:) }
+
+  let(:file) { Tempfile.new }
+  let(:path) { file.path }
+  let(:log) { build(:lettings_log, :completed) }
+
+  context "when parsing csv with headers" do
+    before do
+      file.write("Question\n")
+      file.write("Additional info\n")
+      file.write("Values\n")
+      file.write("Can be empty?\n")
+      file.write("Type of letting the question applies to\n")
+      file.write("Duplicate check field?\n")
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).default_2024_field_numbers_row)
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).to_2024_csv_row)
+      file.rewind
+    end
+
+    it "returns correct offsets" do
+      expect(service.row_offset).to eq(7)
+      expect(service.col_offset).to eq(1)
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+  end
+
+  context "when parsing csv with headers with extra rows" do
+    before do
+      file.write("Section\n")
+      file.write("Question\n")
+      file.write("Additional info\n")
+      file.write("Values\n")
+      file.write("Can be empty?\n")
+      file.write("Type of letting the question applies to\n")
+      file.write("Duplicate check field?\n")
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).default_2024_field_numbers_row)
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).to_2024_csv_row)
+      file.rewind
+    end
+
+    it "returns correct offsets" do
+      expect(service.row_offset).to eq(8)
+      expect(service.col_offset).to eq(1)
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+  end
+
+  context "when parsing csv with headers in arbitrary order" do
+    let(:seed) { rand }
+
+    before do
+      file.write("Question\n")
+      file.write("Additional info\n")
+      file.write("Values\n")
+      file.write("Can be empty?\n")
+      file.write("Type of letting the question applies to\n")
+      file.write("Duplicate check field?\n")
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).default_2024_field_numbers_row(seed:))
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).to_2024_csv_row(seed:))
+      file.rewind
+    end
+
+    it "returns correct offsets" do
+      expect(service.row_offset).to eq(7)
+      expect(service.col_offset).to eq(1)
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+  end
+
+  context "when parsing csv with extra invalid headers" do
+    let(:seed) { rand }
+    let(:log_to_csv) { BulkUpload::LettingsLogToCsv.new(log:) }
+    let(:field_numbers) { log_to_csv.default_2024_field_numbers + %w[invalid_field_number] }
+    let(:field_values) { log_to_csv.to_2024_row + %w[value_for_invalid_field_number] }
+
+    before do
+      file.write("Question\n")
+      file.write("Additional info\n")
+      file.write("Values\n")
+      file.write("Can be empty?\n")
+      file.write("Type of letting the question applies to\n")
+      file.write("Duplicate check field?\n")
+      file.write(log_to_csv.custom_field_numbers_row(seed:, field_numbers:))
+      file.write(log_to_csv.to_custom_csv_row(seed:, field_values:))
+      file.rewind
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+
+    it "counts the number of valid field numbers correctly" do
+      expect(service).to be_correct_field_count
+    end
+  end
+
+  context "when parsing csv without headers" do
+    before do
+      file.write(BulkUpload::LettingsLogToCsv.new(log:, col_offset: 0).to_2024_csv_row)
+      file.rewind
+    end
+
+    it "returns correct offsets" do
+      expect(service.row_offset).to eq(0)
+      expect(service.col_offset).to eq(0)
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+  end
+
+  context "when parsing with BOM aka byte order mark" do
+    let(:bom) { "\uFEFF" }
+
+    before do
+      file.write(bom)
+      file.write(BulkUpload::LettingsLogToCsv.new(log:, col_offset: 0).to_2024_csv_row)
+      file.rewind
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+  end
+
+  context "when an invalid byte sequence" do
+    let(:invalid_sequence) { "\x81" }
+
+    before do
+      file.write(invalid_sequence)
+      file.write(BulkUpload::LettingsLogToCsv.new(log:, col_offset: 0).to_2024_csv_row)
+      file.rewind
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+  end
+
+  context "when parsing csv with carriage returns" do
+    before do
+      file.write("Question\r\n")
+      file.write("Additional info\r")
+      file.write("Values\r\n")
+      file.write("Can be empty?\r")
+      file.write("Type of letting the question applies to\r\n")
+      file.write("Duplicate check field?\r")
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).default_2024_field_numbers_row)
+      file.write(BulkUpload::LettingsLogToCsv.new(log:).to_2024_csv_row)
+      file.rewind
+    end
+
+    it "parses csv correctly" do
+      expect(service.row_parsers[0].field_13).to eql(log.tenancycode)
+    end
+  end
+
+  describe "#column_for_field", aggregate_failures: true do
+    context "when with headers using default ordering" do
+      before do
+        file.write("Question\n")
+        file.write("Additional info\n")
+        file.write("Values\n")
+        file.write("Can be empty?\n")
+        file.write("Type of letting the question applies to\n")
+        file.write("Duplicate check field?\n")
+        file.write(BulkUpload::LettingsLogToCsv.new(log:).default_2024_field_numbers_row)
+        file.write(BulkUpload::LettingsLogToCsv.new(log:).to_2024_csv_row)
+        file.rewind
+      end
+
+      it "returns correct column" do
+        expect(service.column_for_field("field_5")).to eql("B")
+        expect(service.column_for_field("field_22")).to eql("EL")
+      end
+    end
+
+    context "when without headers using default ordering" do
+      before do
+        file.write(BulkUpload::LettingsLogToCsv.new(log:, col_offset: 0).to_2024_csv_row)
+        file.rewind
+      end
+
+      it "returns correct column" do
+        expect(service.column_for_field("field_5")).to eql("A")
+        expect(service.column_for_field("field_22")).to eql("EK")
+      end
+    end
+
+    context "when with headers using custom ordering" do
+      let(:seed) { 123 }
+
+      before do
+        file.write("Question\n")
+        file.write("Additional info\n")
+        file.write("Values\n")
+        file.write("Can be empty?\n")
+        file.write("Type of letting the question applies to\n")
+        file.write("Duplicate check field?\n")
+        file.write(BulkUpload::LettingsLogToCsv.new(log:).default_2024_field_numbers_row(seed:))
+        file.write(BulkUpload::LettingsLogToCsv.new(log:).to_2024_csv_row(seed:))
+        file.rewind
+      end
+
+      it "returns correct column" do
+        expect(service.column_for_field("field_5")).to eql("N")
+        expect(service.column_for_field("field_22")).to eql("O")
+        expect(service.column_for_field("field_26")).to eql("B")
+        expect(service.column_for_field("field_25")).to eql("EF")
+      end
+    end
+  end
+end