attr_stat

attr_stat.rb

#!/usr/bin/env ruby
require 'csv'

@file = ARGV[0]
@attr = ARGV[1]
if ARGV[2]
  @pattern = eval(ARGV[2]) # >:)
end

if @file && @attr
  data = []
  CSV.foreach(@file, headers: true) do |row|
    data << Hash[row.headers.zip(row.fields)]
  end

  summaries = {}
  data.each do |hsh|
    hsh.each do |k,v|
      summaries[k] ||=[]
      summaries[k] << v
    end
  end

  set = summaries[@attr]

  if @pattern
    set = set.select!{|x| x =~ @pattern}
  end

  counters = Hash.new(0)

  set.each do |match|
    counters[match] += 1
  end

  counters.to_a
          .sort_by {|_,count| count}
          .reverse
          .each do |val, count|
            puts "%-5d: %s" % [count, val]
          end

else
  puts "usage: #{__FILE__} something.csv attr /somepattern/i"
end