summarize gatlins runs

summarize_sim.rb

#!/usr/bin/env ruby                                                                                                                                                                                [23/9866]
             
require 'csv'
                         
inputfile = ARGV[0]
intermediate = {         
  'node'                 => {'KOs' => 0, 'values' => [], 'ok' => 0},
  'filemeta plugins'     => {'KOs' => 0, 'values' => [], 'ok' => 0},
  'filemeta pluginfacts' => {'KOs' => 0, 'values' => [], 'ok' => 0},
  'catalog'              => {'KOs' => 0, 'values' => [], 'ok' => 0},
  'filemeta'             => {'KOs' => 0, 'values' => [], 'ok' => 0},
  'report'               => {'KOs' => 0, 'values' => [], 'ok' => 0}
}             
                      
CSV.foreach(inputfile, col_sep: "\t") do |row|
  if row[0] == 'REQUEST'
    type = row[4]
   
    if row[7] != 'OK'
      intermediate[type]['KOs'] += 1                                       
    else            
      intermediate[type]['values'] << row[6].to_i - row[5].to_i
      intermediate[type]['ok'] += 1
    end
  end                  
end

output = []
intermediate.each_pair do |type, info|
  length = info['ok']
  first_quarti = (length * 0.25).to_i
  fourth_quarti = (length * 0.75).to_i
  mediani = (length * 0.5).to_i
  p90i = (length * 0.9).to_i
  p95i = (length * 0.95).to_i
  p99i = (length * 0.99).to_i
  p999i = (length * 0.999).to_i
  p9999i = (length * 0.9999).to_i

  # puts "Sorting values for #{type}"
  info['values'].sort!

  # puts "Finding mean for #{type}"
  mean = (info['values'].reduce(:+) / length)


  # puts "Finding interquartile mean for #{type}"
  rough_iqm = (info['values'][first_quarti..fourth_quarti].reduce(:+) / (length / 2))

  # puts "Finding median of medians for #{type}"
  median = info['values'][mediani]
  distances_from_median = info['values'].map {|v| (v - median).abs }.sort
  median_of_medians = distances_from_median[mediani]

  output << [
    type,
    info['values'].first,
    median,
    info['values'][p90i],
    info['values'][p95i],
    info['values'][p99i],
    info['values'][p999i],
    info['values'][p9999i],
    info['values'].last,
    mean,
    rough_iqm,
    median_of_medians,
    info['ok'] + info['KOs'],
    info['KOs']
  ]
end

puts 'endpoint,min,median,p90,p95,p99,p999,p9999,max,mean,iqm,mm,count,kos'
output.each do |row|
  puts row.join(',')
end