count-commits

#!/usr/bin/env ruby

if ARGV.include? '-h' or ARGV.include? '--help' then puts <<-HELP
Count git commits across multiple projects.
http://benalman.com/

Usage: #{File.basename $0} [directories] > counts.txt

If directories aren't specified, all child directories of the current directory
will be searched.

Arbitrary name and email changes that can be accounted for will be reflected
in the output. Counts/names/emails are sent to STDOUT in the following format:

commitcount email1, email2, ... (name1, name2, ...)

Eg:

1843 cowboy@rj3.net (Ben Alman, cowboy)
1202 tyler@sleekcode.net (Tyler Kellen, tkellen)
 695 kyle@dontkry.com, kyle@kyletyoung.com (Kyle Robinson Young, shama)
 402 vlad.filippov@gmail.com (vladikoff, Vlad Filippov)
 388 sindresorhus@gmail.com (Sindre Sorhus, sindresorhus)

All other messages are sent to STDERR.

Copyright (c) 2014 "Cowboy" Ben Alman
Licensed under the MIT license.
http://benalman.com/about/license/
HELP
exit; end

require 'open3'

def warn(msg) $stderr.puts msg end
def fatal(msg) $stderr.puts msg; exit 1 end
def debug(msg) $stderr.puts "> #{msg}" end

# Validate specified dirs.
def get_dirs(dirs)
  # Use all child dirs of the current dir if none were specified.
  if dirs.empty?
    dirs = Dir.glob('*/')
  end
  # Remove non-dirs
  dirs.select! {|d| File.directory? d}
  # Remove trailing slashes
  dirs.map! {|d| d.sub %r{/$}, ''}
  # Fail if no dirs were found.
  if dirs.empty?
    fatal 'Error: No directories found.'
  end
  dirs
end

# Get computed total number of commits.
def get_computed_total()
  $committers.reduce(0) {|sum, committer| sum + committer[:commit_count]}
end

# Add a commit.
def add_commit(email, name)
  # debug "#{email} #{name}"
  emails = [email]
  names = [name]
  commit_count = 1

  # Get the corresponding committer object(s) indexed by name and email.
  committer_by_email = $committers_by_email[email]
  committer_by_name = $committers_by_name[name]

  # Two committer objects exist, but they're not the same object. Combine them!
  if committer_by_email and committer_by_name and committer_by_email != committer_by_name
    emails = committer_by_name[:emails]
    names = committer_by_name[:names]
    commit_count += committer_by_name[:commit_count]
    # We save committer_by_email and kill off committer_by_name.
    $committers.reject! {|committer| committer == committer_by_name}
    committer = committer_by_email

  # One committer object exists, use it.
  elsif committer_by_email
    committer = committer_by_email
  elsif committer_by_name
    committer = committer_by_name

  # No committer object exists, create one!
  else
    committer = Hash.new
    committer[:emails] = []
    committer[:names] = []
    committer[:commit_count] = 0
    $committers << committer
  end
  # Update committers_by_name/committers_by_email maps.
  names.each {|name| $committers_by_name[name] = committer}
  emails.each {|email| $committers_by_email[email] = committer}
  # Update committer object with new data.
  committer[:names].concat(names).uniq!
  committer[:emails].concat(emails).uniq!
  committer[:commit_count] += commit_count
end

# Our data store.
$committers = []
$committers_by_email = Hash.new
$committers_by_name = Hash.new

# Raw counters based on lines parsed.
$commit_count = 0
$project_count = 0

# Get commits for each specified dir.
get_dirs(ARGV).each do |dir|
  dir.sub! %r{/$}, ''
  stdin, stdout, stderr = Open3.popen3('git log --format="%aE%x09%aN"', :chdir => dir)
  if !stderr.readlines.empty?
    warn %Q{Error running "git log" in "#{dir}" directory.}
  else
    $project_count += 1
    stdout.readlines.each do |line|
      $commit_count += 1
      email, name = line.chomp.split "\t"
      add_commit(email, name)
    end
  end
end

# Sort committer names and email fields.
$committers.each do |committer|
  committer[:names].sort! {|a, b| a.downcase <=> b.downcase}
  committer[:emails].sort! {|a, b| a.downcase <=> b.downcase}
end

# Sort committers first by commit count (descending) then by name.
$committers.sort! do |a, b|
  comp = b[:commit_count] <=> a[:commit_count]
  comp.zero? ? (a[:names][0].downcase <=> b[:names][0].downcase) : comp
end

# Whoops?
$total = get_computed_total
if $total != $commit_count
  fatal "Error: Parsed #{$commit_count} commits, but for some reason only #{$total} appear."
end

# Useful output to STDOUT
$committers.each do |c|
  puts "%#{$total.to_s.length}d %s (%s)" % [c[:commit_count], c[:emails].join(", "), c[:names].join(", ")]
end

# Totals.
def pl(n, str) "#{n} #{str}#{n == 1 ? '' : ?s}" end
warn "#{pl $commit_count, 'commit'} by #{pl $committers.length, 'author'} in #{pl $project_count, 'project'}."