debian-mirror-gitlab/spec/lib/csv_builder_spec.rb

# frozen_string_literal: true

require 'spec_helper'

RSpec.describe CsvBuilder do
  let(:object) { double(question: :answer) }
  let(:fake_relation) { FakeRelation.new([object]) }
  let(:subject) { described_class.new(fake_relation, 'Q & A' => :question, 'Reversed' => -> (o) { o.question.to_s.reverse }) }
  let(:csv_data) { subject.render }

  before do
    stub_const('FakeRelation', Array)

    FakeRelation.class_eval do
      def find_each(&block)
        each(&block)
      end
    end
  end

  it 'generates a csv' do
    expect(csv_data.scan(/(,|\n)/).join).to include ",\n,"
  end

  it 'uses a temporary file to reduce memory allocation' do
    expect(CSV).to receive(:new).with(instance_of(Tempfile)).and_call_original

    subject.render
  end

  it 'counts the number of rows' do
    subject.render

    expect(subject.rows_written).to eq 1
  end

  describe 'rows_expected' do
    it 'uses rows_written if CSV rendered successfully' do
      subject.render

      expect(fake_relation).not_to receive(:count)
      expect(subject.rows_expected).to eq 1
    end

    it 'falls back to calling .count before rendering begins' do
      expect(subject.rows_expected).to eq 1
    end
  end

  describe 'truncation' do
    let(:big_object) { double(question: 'Long' * 1024) }
    let(:row_size) { big_object.question.length * 2 }
    let(:fake_relation) { FakeRelation.new([big_object, big_object, big_object]) }

    it 'occurs after given number of bytes' do
      expect(subject.render(row_size * 2).length).to be_between(row_size * 2, row_size * 3)
      expect(subject).to be_truncated
      expect(subject.rows_written).to eq 2
    end

    it 'is ignored by default' do
      expect(subject.render.length).to be > row_size * 3
      expect(subject.rows_written).to eq 3
    end

    it 'causes rows_expected to fall back to .count' do
      subject.render(0)

      expect(fake_relation).to receive(:count).and_call_original
      expect(subject.rows_expected).to eq 3
    end
  end

  it 'avoids loading all data in a single query' do
    expect(fake_relation).to receive(:find_each)

    subject.render
  end

  it 'uses hash keys as headers' do
    expect(csv_data).to start_with 'Q & A'
  end

  it 'gets data by calling method provided as hash value' do
    expect(csv_data).to include 'answer'
  end

  it 'allows lamdas to look up more complicated data' do
    expect(csv_data).to include 'rewsna'
  end

  describe 'excel sanitization' do
    let(:dangerous_title) { double(title: "=cmd|' /C calc'!A0 title", description: "*safe_desc") }
    let(:dangerous_desc) { double(title: "*safe_title", description: "=cmd|' /C calc'!A0 desc") }
    let(:fake_relation) { FakeRelation.new([dangerous_title, dangerous_desc]) }
    let(:subject) { described_class.new(fake_relation, 'Title' => 'title', 'Description' => 'description') }
    let(:csv_data) { subject.render }

    it 'sanitizes dangerous characters at the beginning of a column' do
      expect(csv_data).to include "'=cmd|' /C calc'!A0 title"
      expect(csv_data).to include "'=cmd|' /C calc'!A0 desc"
    end

    it 'does not sanitize safe symbols at the beginning of a column' do
      expect(csv_data).not_to include "'*safe_desc"
      expect(csv_data).not_to include "'*safe_title"
    end

    context 'when dangerous characters are after a line break' do
      it 'does not append single quote to description' do
        fake_object = double(title: "Safe title", description: "With task list\n-[x] todo 1")
        fake_relation = FakeRelation.new([fake_object])
        builder = described_class.new(fake_relation, 'Title' => 'title', 'Description' => 'description')

        csv_data = builder.render

        expect(csv_data).to eq("Title,Description\nSafe title,\"With task list\n-[x] todo 1\"\n")
      end
    end
  end
end