Ruby と tumugi によるデータパイプライン構築

Gemfile

source "https://rubygems.org"
gem "tumugi", "~> 0.6.3"

run_tumugi.rb

# bundle exec tumugi run -f run_tumugi.rb squared_numbers

task :print_numbers do
  output target(:local_file, "numbers_up_to_10.txt")
  
  run {
    output.open("w") do |f|
      (1..10).each {|i| f.puts(i)}
    end
  }
end

task :squared_numbers do
  requires :print_numbers
  
  output target(:local_file, "squares.txt")
  
  run {
    output.open("w") do |fout|
      input.open("r") do |fin|
        fin.each_line do |line|
          n = line.to_i
          out = n * n
          fout.puts("#{n}:#{out}")
        end
      end
    end
  }
end

run_tumugi2.rb

# bundle exec tumugi run -f run_tumugi.rb squared_numbers -p max_num:5

task :print_numbers do
  # max_num という名前のパラメータを使うことを宣言
  param :max_num, type: :integer, auto_bind: true, required: true

  output target(:local_file, "numbers_up_to_10.txt")
  
  run {
    output.open("w") do |f|
      (1..max_num).each {|i| f.puts(i)}
    end
  }
end

task :squared_numbers do
  requires :print_numbers
  
  output target(:local_file, "squares.txt")
  
  run {
    output.open("w") do |fout|
      input.open("r") do |fin|
        fin.each_line do |line|
          n = line.to_i
          out = n * n
          fout.puts("#{n}:#{out}")
        end
      end
    end
  }
end