Boilerplate CSV analysis script

csvanalyze.rb

#!/usr/bin/env ruby

require 'csv'
require 'ostruct'
require 'optparse'
require 'i18n'
require 'set' # this should only be required until i18n updates to new rev
require 'active_support/core_ext/string/inflections'

I18n.enforce_available_locales = true # I18n spams warn unless this is set

class CSVAnalyze

  ::Version = [0,0,1]

  attr_accessor :options
  
  # $stdout.sync = true # prevents buffering of output; uncomment this only if you need it.
  Signal.trap(:INT, :EXIT) # Trap ctrl+c from users or `kill -SIGINT pid` from shell and exits from pipe chains

  def initialize(args, options = {})
    # Handle options
    @options = OpenStruct.new
    @options.format = :tsv

    opt_parser = OptionParser.new do |opt|
      opt.banner = "Usage: #{$0} [OPTION]... [FILE]..."

      opt.on("-c","--csv","Output CSV format instead of standard TSV") do
        @options.format = :csv
      end

      opt.on_tail("-h","--help","Print usage information.") do
        $stderr.puts opt_parser
        exit 1
      end

      opt.on_tail("--version", "Show version") do
        puts ::Version.join('.')
        exit 1
      end
    end

    begin 
      opt_parser.parse!
    rescue OptionParser::InvalidOption => e
      $stderr.puts "Specified #{e}"
      $stderr.puts opt_parser
      exit 64 # EX_USAGE
    end
  end

  def run!
    @data = ARGF.readlines # slurps entire file (refactor to block form if working with large files!)

    # BEGIN data pre-processing
    # Extract headers from first line
    headers = CSV.parse(data.shift).first.map { |col| col.parameterize.underscore.to_sym }
    # Parse CSV and convert to hash rows
    data.map! { |line| line.parse_csv }
    data.map! { |row| Hash[headers.zip(row)] }
    # END data pre-processing

    # BEGIN data analysis
    summaries = data.each.with_index.collect do |row, i|
      # This is where you do the work.
      row[:some_column] = "Some computed value goes here."
      row
    end
    # END data analysis

    # BEGIN report to stdout
    case @options.format
    when :tsv
      puts summaries.first.keys.join("\t")
      summaries.each do |row|
        puts row.values.join("\t")
      end
    when :csv
      puts summaries.first.keys.to_csv
      summaries.each do |row|
        puts row.values.to_csv
      end
    end
    # END report to stdout
  end

end

begin
  if $0 == __FILE__
    CSVAnalyze.new(ARGV).run!
  end
rescue Interrupt
  # Ctrl^C
  exit 130
rescue Errno::EPIPE
  # STDOUT was closed
  exit 74 # EX_IOERR
end