Compare Substack exported subscribers with Beehiiv exported subscribers

compare.rb

require 'csv'
require 'time'
require 'set'

class Diff
  MissingData = Data.define(:rows, :headers)

  def initialize(source_file:, second_file:)
    @source_file = source_file
    @second_file = second_file
    @output_filename = generate_output_filename
  end

  def compare
    validate
    missing_data = find_missing_data
    write_output(missing_data)
    report_results(missing_data.rows.size)
  end

  private

  def validate
    unless File.exist?(@source_file) && File.exist?(@source_file)
      raise ArgumentError, "One or both input files do not exist."
    end
  end

  def find_missing_data
    first_data = CSV.read(@source_file, headers: true)
    second_data = CSV.read(@second_file, headers: true)

    first_emails = Set.new(first_data.map { _1['email'].to_s.downcase })
    second_emails = Set.new(second_data.map { _1['email'].to_s.downcase })

    missing_emails = first_emails - second_emails

    missing_lines = first_data.select { missing_emails.include?(_1['email'].to_s.downcase) }

    MissingData.new(rows: missing_lines, headers: first_data.headers)
  end

  def write_output(missing_data)
    CSV.open(@output_filename, "wb") do |csv|
      csv << missing_data.headers
      missing_data.rows.each { csv << _1 }
    end
  end

  def report_results(missing_count)
    puts "Lines from the first file not present in the second file have been written to: #{@output_filename}"
    puts "Total missing lines: #{missing_count}"
  end

  def generate_output_filename
    "diff_#{Time.now.strftime('%Y%m%d_%H_%M_%S')}.csv"
  end
end

if ARGV.length != 2
  puts "Usage: ruby script.rb <first_file.csv> <second_file.csv>"
  exit
end

begin
  diff = Diff.new(source_file: ARGV[0], second_file: ARGV[1])
  diff.compare
rescue ArgumentError => e
  puts "Error: #{e.message}"
  exit
end