rails_contributors.rb

# Originally by Xavier Noria and posted to the Rails Core list here:
# http://groups.google.com/group/rubyonrails-core/browse_thread/thread/7554c8c5a25b4765?hl=en

# Output has been aggregated at:
# http://gist.github.com/20721

require 'rubygems'
require 'grit'

$KCODE = 'u'
RAILS_DIR = '/Users/dancroak/dev/rails'

# Some people appear in Rails logs under different names, there are nicks,
# typos, email addresses, shortenings, etc. This is a hand-made list to map
# them in order to be able to aggregate commits from the same real author.
SEEN_IN_LOG_ALSO_AS = {
  'Aliaksey Kandratsenka'    => 'Aleksey Kondratenko',
  'Austin Ziegler'           => 'Thanks to Austin Ziegler for Transaction::Simple',
  'Bob Silva'                => 'BobSilva',
  'Cheah Chu Yeow'           => ['Chu Yeow', 'chuyeow'],
  'court3nay'                => 'courtenay',
  'Damian Janowski'          => 'djanowski',
  'Dan Manges'               => 'dcmanges',
  'Dave Thomas'              => ['dave@...', 'After much pestering from Dave Thomas'],
  'David Heinemeier Hansson' => 'DHH',
  'Ernesto Jimenez'          => 'ernesto.jimenez',
  'Frederick Cheung'         => 'fcheung',
  'Geoff Buesing'            => ['gbuesing', 'Geoffrey Buesing'],
  'Hongli Lai (Phusion)'     => ['Hongli Lai (Phusion', 'FooBarWidget'],
  'Isaac Feliu'              => 'isaacfeliu',
  'Jeremy Kemper'            => 'bitsweat',
  'John Barnette'            => 'jbarnette',
  'José Valim'               => 'josevalim',
  'Josh Peek'                => ['josh', 'Josh', 'Joshua Peek', '[email protected]'],
  'Josh Susser'              => %w(hasmanyjosh josh@...),
  'Juanjo Bazan'             => 'juanjo.bazan',
  'Kevin Clark'              => 'Kevin Clark kevin.clark@...',
  'Marcel Molina'            => ['Marcel Molina Jr.', 'Marcel'],
  'Michael Koziarski'        => %w(Koz nzkoz),
  'Michael Schoen'           => 'Michael A. Schoen',
  'Mislav Marohnić'          => ['mislav', 'mislav@...'],
  'Pratik Naik'              => %w(Pratik lifofifo lifo),
  'Rick Olson'               => ['rick', 'Rick', 'Rick Olsen', 'technoweenie@...'],
  'Ryan Bates'               => 'ryanb',
  'Steve Purcell'            => 'stephen_purcell@...',
  'Tarmo Tänav'              => 'tarmo',
  'Tim Pope'                 => 'tpope',
  'Xavier Noria'             => 'fxn',
}

# Reverse SEEN_IN_LOG_ALSO_AS as NAME_NORMALIZER.
NAME_NORMALIZER = {}
SEEN_IN_LOG_ALSO_AS.each do |name, also_as|
  [*also_as].each { |alt| NAME_NORMALIZER[alt] = name }
end

def normalize_name(name)
  name = name.sub(/\s*<[^>]+>/, '') # remove any email address in angles
  name.strip!
  NAME_NORMALIZER[name] || name
end

def commit_from_svn?(commit)
  commit.message.include?('git-svn-id:')
end

# When Rails had a svn repo there was a convention for authors: the committer
# put their name at the end of the commit message between brackets. For example:
#
#   Fix case-sensitive validates_uniqueness_of. Closes #11366 [miloops]
#
# I think there were a few exceptions but this is the best we can get to figure
# out authors from svn.
def extract_svn_authors_from_message(message)
  svn_authors = []
  if message =~ /\[([^\]]+)\]\s*$/ # end-of-line anchor on purpose
    # [Adam Milligan, Pratik]
    # [Rick Olson/Nicholas Seckar]
    # [Kevin Clark & Jeremy Hopple]
    $1.split(%r{\s*[,/&]\s*}).each do |c|
      if looks_like_an_author_name(c)
        svn_authors << c
      end
    end
  end
  svn_authors
end

def extract_svn_authors(commit)
  svn_authors = extract_svn_authors_from_message(commit.message)
  if svn_authors.empty?
    # Sometimes the name was just added to the CHANGELOG, we fallback to the diff
    Dir.chdir(RAILS_DIR) do
      in_changelog = false
      `git show #{commit.id}`.each do |line|
        if line =~ /^diff --git/
          in_changelog = false
          next
        end
        if line =~ /^\+\+\+.*changelog$/i
          in_changelog = true
          next
        end
        if in_changelog && line =~ /^\+\s*\*/
          svn_authors += extract_svn_authors_from_message(line)
        end
      end
    end
  end
  svn_authors = [commit.author.name] if svn_authors.empty?
  svn_authors.map {|a| normalize_name(a)}.uniq
end

# Author name extraction in svn commits returns a few strings we just ignore.
def looks_like_an_author_name(str)
  str !~ /\A\d+\z/ && # Remove side effects of [5684]
  str !~ /\A\s*\z/ &&
  str != 'See rails ML' &&
  str != 'subject "Text::Format Licence Exception" on Oct 15' &&
  str != 'RubyConf 2005' &&
  str != "RubyConf '05'" &&
  str !~ /^Includes duplicates of changes/ # example: Includes duplicates of changes from 1.1.4 - 1.2.3
end

def print_authors(authors)
  authors.sort {|a,b| b[1] <=> a[1]}.each do |name, count|
    puts "%4d %s" % [count, name]
  end
end

def process_commits!
  authors = Hash.new(0)
  repo = Grit::Repo.new(RAILS_DIR)
  i = 0
  page_size = 300

  loop do
    commits = repo.commits('master', page_size, i)
    break if commits.empty?
    commits.each do |commit|
      if commit_from_svn?(commit)
        extract_svn_authors(commit).each {|a| authors[normalize_name(a)] += 1}  
      else
        authors[normalize_name(commit.author.name)] += 1
      end
      i += 1    
    end
  end
  
  print_authors(authors)
end

process_commits!