Language stats for project

gistfile1.rb

require 'mongo'

client = Mongo::MongoClient.new 
db     = client["ouroboros"]

stats = {}
total = db["galaxy_zoo_classifications"].count
done  = 0 

db["galaxy_zoo_classifications"].find.each do |classification|
  done += 1
  puts "done #{done} of #{total}" if done%1000==0 
  lang = (classification["annotations"].select{|ann| ann.keys.include? "lang"}.first ||  {"lang" => "en"})["lang"]
  stats[lang] ||= {logged_in: 0 , logged_out: 0 , user_ids: Set.new, user_ips: Set.new }
  
  if classification["user_id"]
    stats[lang][:logged_in] += 1
    stats[lang][:user_ids].add classification["user_id"]
  else 
    stats[lang][:logged_out] += 1
    stats[lang][:user_ips].add classification["user_ip"]
  end
end


File.open("galaxy_zoo_lang_stats.dat","w") do |file|
  stats.each_pair do |lang, s|
    file.puts [lang, s[:logged_in], s[:logged_out], s[:user_ids].count, s[:user_ips].count].join(", ")
  end
end