car data.tsv | ruby stacked-graph-transformer.rb

stacked-graph-transformer.rb

class StackedGraphTransformer
  # @param [IO] io
  def initialize(io)
    @io = io
  end

  def gets_tsv
    (line = @io.gets) ? line.chomp.split("\t") : nil
  end

  def each(&block)
    group_title, *_ = gets_tsv

    num_one_group = nil
    first_group = nil
    titles = [group_title]

    cols = []
    loop do
      group, main, *subs, val = gets_tsv
      break if group.nil?

      first_group ||= group
      if first_group === group
        fmt = '%s'
        fmt << (subs.empty? ? '' : ' (%s)')
        titles << (fmt % [main, subs.join(", ")])
      elsif num_one_group.nil?
        block.call(titles)
        num_one_group = cols.size
      end

      if num_one_group === cols.size
        block.call(cols)
        cols = []
      end

      cols << group if cols.empty?
      cols << val
    end
    block.call(cols) unless cols.empty?
  end
end


if ARGV[0] == "--test"
  require 'test/unit/assertions'
  include Test::Unit::Assertions
  result = []
  StackedGraphTransformer.new(Object::DATA).each {|cols| result << cols}
  assert_equal [
    ['date', '_ (PC)', '_ (SP)', 'facebook (PC)', 'facebook (SP)', 'twitter (PC)', 'twitter (SP)'],
    ['2016-02-01', 1, 2, 3, 4, 5, 6].map(&:to_s),
    ['2016-02-02', 11, 12, 13, 14, 15, 16].map(&:to_s),
  ], result
  puts "OK"
else
  StackedGraphTransformer.new($stdin).each do |cols|
    puts cols.join("\t")
  end
end

__END__
date	sns	device	num
2016-02-01	_	PC	1
2016-02-01	_	SP	2
2016-02-01	facebook	PC	3
2016-02-01	facebook	SP	4
2016-02-01	twitter	PC	5
2016-02-01	twitter	SP	6
2016-02-02	_	PC	11
2016-02-02	_	SP	12
2016-02-02	facebook	PC	13
2016-02-02	facebook	SP	14
2016-02-02	twitter	PC	15
2016-02-02	twitter	SP	16