used to see how much outside contribution we have from gitstats output

author_stats.rb

#! /usr/bin/env ruby
require 'rubygems'
require 'nokogiri'
require 'pp'

exclude = ["Author", "defunkt", "Chris Wanstrath", "Kyle Neath", "PJ Hyett", "Scott Chacon", "Tom Preston-Werner", "tekkub", "Tekkub", "Ryan Tomayko"]
Dir.glob("*").each do |d|
  if !File.file?(d)
    puts d
    stats = {:authors => 0, :percent => 0.0, :commits => 0, :lines => 0}
    a = File.read("#{d}/authors.html")
    doc = Nokogiri::HTML(a)
    doc.search('table#authors > tr').each do |tr|
      author = tr.children.first.content
      if !exclude.include?(author)
        commits_raw = tr.children[1].content
        m = /(\d+) \((.*?)%\)/.match(commits_raw)
        stats[:authors] += 1
        stats[:commits] += m[1].to_i
        stats[:percent] += m[2].to_f
        add = tr.children[2].content.to_i
        remove = tr.children[3].content.to_i
        if (add + remove) < 2000
          stats[:lines] += add
          stats[:lines] += remove
        end
      end
    end
    pp stats
    puts "", ""
  end
end