A Ruby script to compute complexity/churn stats on a codebase

complexity_churn.rb

#!/usr/bin/env ruby

require "flog"
require "open3"

def capture_output(cmd, *args)
  stdout, stderr, status = Open3.capture3(cmd, *args)
  return stdout if status.success?

  STDERR.puts("Failed to run git #{args.join(" ")}")
  STDERR.puts(stderr)
  ''
end

def flog(file)
  flog = Flog.new
  flog.flog(file)
  flog.total_score
end

def normalize(hash)
  return hash if hash.empty?

  min = hash.each_value.min
  max = hash.each_value.max
  range = [1, max - min].max

  hash.transform_values { |v| (v - min) / range }
end

def complexity(files)
  files.each_with_object({}) do |file, h|
    h[file] = flog(file)
  end
end

def churn
  capture_output("git", "log", "--name-only", '--since="6 months ago"', '--format=')
    .each_line(chomp: true)
    .select { |name| name.end_with?('.rb') }
    .reject { |name| name.end_with?('_test.rb') }
    .select { |name| File.readable?(name) }
    .tally
end

def filter_out_subdir(hash, subdir)
  return hash unless subdir
  hash.select { |file, _| file.start_with?("#{subdir}/") }
end

def complexity_and_churn(subdir)
  churns = normalize(filter_out_subdir(churn, subdir))
  complexities = normalize(complexity(churns.keys))
  churns.map do |file, churn|
    [0.5 * churn + 0.5 * complexities[file], file]
  end
end

complexity_and_churn(ARGV.first).sort.reverse.take(10).each do |score, file|
  puts "%.3f #{file}" % score
end