Prints the top users who contribute to the most civic tech repositories outside their home organization(s).

stefan_baack.rb

# Prints the top 10 users who contribute to the most civic tech repositories
# outside their home organization(s). If you'd like to print the top 15, run:
#
# ruby stefan_baack.rb 15
#
# To see the top 3 organizations to which each user contributed, run:
#
# ruby stefan_baack.rb 10 3
#
# @see http://sbaack.com/2015/11/19/scraping-the-global-civic-tech-community-on-github-part-2.html
# @see https://www.dropbox.com/sh/y2dcr9dy96xg7uc/AAAobZh8G36C4gtuJfCt7z08a?dl=0
require 'set'

require 'nokogiri'

# Some real organizations have multiple GitHub organizations, and some users
# have been employed at multiple organizations over time.
secondary = {
  'chrismytton' => ['everypolitician'],
  'danmelton' => ['Granicus'],
  'davewhiteland' => ['everypolitician'],
  'duncanparkes' => ['everypolitician'],
  'pudo' => ['CodeForAfrica', 'okfde'],
  'sebbacon' => ['mysociety'],
  'stefanw' => ['okfn'],
  'tmtmtmtm' => ['mysociety'],
  'zarino' => ['everypolitician'],
}

data = File.read('github_scrape_civic-tech_2015-11-19/contributor-network_2015-11-19_11:38:55.gexf')
document = Nokogiri::XML(data)
document.remove_namespaces!

# Get the number of repositories contributed to per user by organization.
counts_by_organization = {}
document.xpath('//edge').each do |edge|
  source = edge[:source]
  organization = edge.at_xpath('.//attvalue/@value').value

  counts_by_organization[source] ||= Hash.new(0)
  counts_by_organization[source][organization] += 1
end

# Eliminate the organization to which the user has contributed the most (e.g. day job).
counts_excluding_home_organization = {}
counts_by_organization.each do |source,organizations|
  organizations.delete(organizations.max_by{|_,v| v}[0])
  if secondary.key?(source)
    secondary[source].each do |organization|
      organizations.delete(organization)
    end
  end
  counts_excluding_home_organization[source] = organizations.values.reduce(0, :+)
end

# Print the top 10 contributors.
puts 'By the number of repositories contributed to outside their home organization(s):'
counts_excluding_home_organization.sort_by{|_,v| -v}.first(Integer(ARGV[0] || 10)).each do |source,count|
  puts '%3d %s' % [count, source]
  counts_by_organization[source].sort_by{|_,v| -v}.first(Integer(ARGV[1] || 0)).each do |organization,count|
    puts '  %3d %s' % [count, organization]
  end
end

puts '---'

puts 'By the of organizations contributed to besides their home organization(s):'
counts_by_organization.sort_by{|_,v| -v.size}.first(Integer(ARGV[0] || 10)).each do |source,organizations|
  puts '%3d %s' % [organizations.size, source]
end