parser.rb

#!/usr/bin/env ruby
require 'csv'
require 'optparse'
require 'ostruct'

# Opt Parser
options = OpenStruct.new
OptionParser.new do |opts|
  opts.banner = "Usage: ./test.rb [options]"

  opts.on('-c', '--csv [CSV File]', 'Path of the CSV to be parsed') do |n|
    options.csv = n || nil
  end

  opts.on('-a', '--array_one [ARRAY]', 'First Array') do |array|
    array_one = array.split(",").to_a
    options.array_one = array_one || nil
  end

  opts.on('-b', '--array_two [ARRAY]', 'Second Array') do |array|
    array_two = array.split(",").to_a
    options.array_two = array_two || nil
  end

  opts.on('-e', '--export [FILENAME]', 'Export CSV to File') do |filename|
    options.export = true
    options.export_filename = filename
  end

  opts.on('-v', '--verbose', 'Verbose Mode') do
    options.verbose = true
  end
end.parse!

# Inputs
array_one = options.array_one || ["teen_star"]
array_two = options.array_two || ["tv_stars", "old_stars"]
csv = options.csv || 'test.csv'

# CSV Parse
table = CSV.table(csv, headers: true)
# Enumerator to make the script performant
rows = table.each

# Case I
rows.each do |master_row|
  if array_one.include? master_row[:source]
    rows.delete_if { |row| row[:email] == master_row[:email] }
  end
end

# Case II
emails = []
index = {}
prev_email = nil
rows.each do |master_row|
  prev_email = master_row[:email]
  email_index = emails.index master_row[:email]
  index[master_row[:source]] = array_two.index(master_row[:source]) if array_two.include? master_row[:source]
  if email_index.nil?
    emails << master_row[:email]
  else
    rows.delete_if do |row|
      index[row[:source]] != index.values.min if prev_email == row[:email] and emails.include? row[:email]
    end
  end
end

# Packaging Options
if options.verbose
  rows.each { |r| puts r.inspect }
end

if options.export
  puts "Exporting.."
  CSV.open(options.export_filename, "wb") do |csv|
    rows.each do |row|
      csv << row
    end
  end
  puts "File #{options.export_filename} has been created."
end