iTunes Receipt Email Parser

parser.rb

class ReceiptParser
  attr_reader :counts, :prices

  def initialize(currency_symbol)
    if (currency_symbol == "") then currency_symbol = "£" end
    @currency_symbol = currency_symbol 
    @all_items = []
    @counts = { :ios_apps => 0, :iaps => 0 }
    @prices = { :ios_apps => 0, :iaps => 0 }
  end

  def parse_email(file) 
    lines = slice_out_relevant_section(file)

    if (lines) then

      lines.map! { |line|
        line.gsub!(/\Fr=\r\n/, "Free")        
        line.gsub!(/\=\r\n/, "")
        line.gsub!(/\=C2=A3/, @currency_symbol)   
        line.gsub!(/\=A3=\r\n/, @currency_symbol)
        line.gsub!(/\=A3/, @currency_symbol)
        line
      }

      new_lines = []
      lines.each_with_index do |line, index|
        if (line.end_with? @currency_symbol ) then
          new_lines << (line + lines[index+1])
        end
        if (line =~ /   #{@currency_symbol}(\.|\d)+\r\n/) then
          new_lines << (lines[index-1] + line)
        end
        if (line.end_with? "Free") then
          new_lines << line
        end
      end

      new_lines.select! { |line| 
        (line.include?("Seller:") || line.include?("App"))
      }

      new_lines.reject! { |line|
        line.include?("Mac App")  # Filter out all Mac Apps
      }

      # Clean up the data
      new_lines.map! { |line| 
        line.gsub!(/^\d{1,2}\s+/, "")
        line.gsub!(/^Q\d{4}\s+/, "")
        line.gsub!(/\ {3,}/, "\t")
        line.gsub!(/, Seller: /, "\t")
        if (line.count("\t") == 2) then
          line.gsub!(/\tFree$/, "\t\tFree")
          line.gsub!(/\t#{@currency_symbol}/, "\t\t#{@currency_symbol}")        
        end
        line
      }

      new_lines
    end
  end

  def slice_out_relevant_section(file)
    lines = File.readlines(file)

    item_index = lines.index { |line| line =~ /^Item/ }
    if (!item_index) then return end
    lines.slice!(0, item_index+2)

    line_index = lines.index { |line| line !~ /^-----------------------------------------------/ }
    lines.slice!(0, line_index+1)

    next_line_index = lines.index { |line| line =~ /^-----------------------------------------------/ }
    lines.slice!(next_line_index, lines.size - next_line_index)

    return lines
  end

  def price_for_part(part)
    return 0 if (part == "")

    return part.gsub(/#{@currency_symbol}/, "").to_f
  end

  def parse
    Dir.glob("**/*.eml") do |file| 
      email_items = parse_email(file)
      @all_items += email_items || []
    end

    @all_items.each do |item|
      parts = item.split("\t")
      case parts[2]
        when ""
          @counts[:ios_apps] += 1
          @prices[:ios_apps] += price_for_part(parts[3])
        when "In-App Pu"
          @counts[:iaps] += 1
          @prices[:iaps] += price_for_part(parts[3])
        when "In App Pu"
          @counts[:iaps] += 1
          @prices[:iaps] += price_for_part(parts[3])
        when "App"
          @counts[:ios_apps] += 1
          @prices[:ios_apps] += price_for_part(parts[3])
        when "iOS App"
          @counts[:ios_apps] += 1
          @prices[:ios_apps] += price_for_part(parts[3])
      end
    end

    File.open("App.tsv", "w") do |file|
      @all_items.each do |item|
        file.write item
      end
    end
  end

  def price_for_type(item_type)
    format_price(@prices[item_type])
  end

  def format_price(price)
    "#{@currency_symbol}#{price.round(2)}"
  end
end

currency_symbol = [(print 'What currency symbol do you use? (default £) '), gets.rstrip][1]

parser = ReceiptParser.new(currency_symbol)
parser.parse

puts
puts "iOS Apps: #{parser.counts[:ios_apps]} (#{parser.price_for_type(:ios_apps)})"
puts "In-App Purchases: #{parser.counts[:iaps]} (#{parser.price_for_type(:iaps)})"
puts
puts "Total Purchases: #{parser.counts.values.inject(:+)} (#{parser.format_price(parser.prices.values.inject(:+))})"