misc tool to edit csv columns

csv_contractor.rb

# coding: utf-8
require 'csv'

class Contractors
  HEAD = BasicObject.new
  class << HEAD
    def calc(values)
        values[0]
    end
  end

  SUM = BasicObject.new
  class << SUM
    def calc(values)
        values.map{|v| v.to_i}.sum.to_s
    end
  end
end

class CSVColumn
  def initialize(pattern, calculator)
    @pattern = pattern
    @calculator = calculator
  end

  def matches?(name)
    @pattern === name
  end

  def calc(values)
    @calculator.calc(values)
  end

  def inspect
    @pattern
  end  
end

class CSVContractor
  def initialize(output_columns)
    @output_columns = output_columns
    @indices_of_columns = Hash.new { |hash, key| hash[key] = [] }
  end

  def decide_column_indices(headers)
    # ヘッダ行を読んで各カラムのインデックスを決める
    headers.each_with_index do |header, index|
      @output_columns.each do |column|
        if column.matches?(header)
          @indices_of_columns[column].push(index)
          break
        end
      end
    end
  end

  def process_single_line(line)
    # 与えられた 1 行を処理して標準出力する
    values = Hash.new { |hash, key| hash[key] = [] }
    line.each_with_index do |val, index|
      @output_columns.each do |column|
        if @indices_of_columns[column].include?(index)
          values[column].push(val)
          break
        end
      end
    end

    puts @output_columns.map{ |col| col.calc(values[col])}.join(",")
  end

  def process(path2csv)
    # CSV を処理して標準出力する
    csv = CSV.read(path2csv)    
    decide_column_indices(csv.first)
    csv.drop(1).each do |line|
      process_single_line(line)
    end
  end
end

test.csv

test.rb

require './csv_contractor.rb'

output_columns = [
  as_cpu = CSVColumn.new("Archive Server CPU", Contractors::HEAD),
  as_mem = CSVColumn.new("Archive Server Mem", Contractors::HEAD),
  web_cpu = CSVColumn.new("Web Server CPU", Contractors::HEAD),
  web_mem = CSVColumn.new("Web Server Mem", Contractors::HEAD),
  solr_cpu = CSVColumn.new("Solr CPU", Contractors::HEAD),
  solr_mem = CSVColumn.new("Solr Mem", Contractors::HEAD),
  pgsql_cpu = CSVColumn.new(/PostgreSQL\d* CPU/, Contractors::SUM),
  pgsql_mem = CSVColumn.new(/PostgreSQL\d* Mem/, Contractors::SUM)
]

csv_contractor = CSVContractor.new(output_columns)
csv_contractor.process("./test.csv")