blame-stats

#!/usr/bin/env ruby
# Show how many lines people are "responsible" for, according to `git-blame`.

# optionally take a pathspec to use as a filter
pathspec = ARGV.any? ? "-- #{ARGV.join(' ')}" : ''

# grab all the files in the source tree, filtered by the pathspec
files = `git ls-tree -r HEAD #{pathspec}`.split("\n").map do |line|
  metadata, path = line.split("\t")
end

# look only at blobs
paths = files.select { |metadata, _| metadata =~ /blob/ }.
              map { |_, path| path.chomp }

global_stats = Hash.new(0)

BLAME_FORMAT_REGEX = Regexp.compile /^author-mail <(.+)>$/

paths.each do |path|
  print "Processing : #{path}"

  begin
    file_stats = Hash.new(0)
    blame = `git blame --line-porcelain HEAD -- #{path}`
    blame.split("\n").each do |line|
      next unless line.match(BLAME_FORMAT_REGEX)
      author = $~[1]
      global_stats[author] += 1
      file_stats[author] += 1
    end

    most_guilty = file_stats.sort_by { |_, count| -count }.first(3)
    if most_guilty.any?
      authors = most_guilty.map { |author, count| "#{author} [#{count}]" }
      print "; (most guilty: #{authors.join '; '})"
    end
  rescue StandardError => e
    # don't choke on binary files or other unexpected input
    print "; [***] rescued #{e}"
  end

  puts
end

# calculate appropriate width for display
longest_author = global_stats.keys.max { |a, b| a.length <=> b.length }
width = longest_author ? longest_author.length : 0

puts "\nOverall stats:"
global_stats.
  sort_by { |_, count| -count }.
  each { |author, count| puts "  %#{width}s : %d" % [author, count] }