git-top-trailers

git-top-trailers

#!/usr/bin/env ruby

ROLES_REGEX = /^([^ \t\n\[\]\(\):]+): ([^<>]+) <(\S+?)>$/

$roles = Hash.new(0)

log_args = [ '--no-merges', '--format=%b%x00' ]
IO.popen(%w[git log] + log_args + ARGV) do |pipe|
  pipe.each("\0\n", chomp: true) do |commit|
    trailer = commit.scrub.split("\n\n").last
    next unless trailer
    trailer.scan(ROLES_REGEX) do
      role = $1.downcase
      $roles[role] += 1 if role != 'signed-off-by'
    end
  end
end

def quantiles(data, probs)
  values = data.sort

  probs.map do |prob|
    x = 1 + (values.count - 1) * prob
    mod = x % 1
    (1 - mod) * values[x.floor - 1] + (mod) * values[x.ceil - 1]
  end
end

def show_group(title, threshold, top: true)
  puts [nil, title, '-' * title.size]

  quantile_sum = 0
  $roles.sort_by(&:last).reverse.each_with_index do |(role, count), i|
    next if top ? (count <= threshold) : count > threshold
    puts "%3d. %s (%d)" % [i + 1, role, count]
    quantile_sum += count
  end

  puts [nil, 'Total wealth: %d%%' % [(100 * quantile_sum.to_f / $total).round]]
end

$total = $roles.values.sum
$q = quantiles($roles.values, [ 0.99, 0.90, 0.75, 0.50 ])

puts '1%%: %d, 10%% %d, 25%% %d, 50%% %d' % $q

top_1, top_10, top_25, median = $q

show_group('Top 1%', top_1)
show_group('Top 10%', top_10)
show_group('Top 25%', top_25)
show_group('Bottom 50%', median, top: false)