JRuby script to pull iOS app ranking data from AppAnnie.com. Must be passed a file containing an app name, internal id (optional), and the store id per new line. Using System.out to print results. Example usage: ruby ranking.rb seed_file.txt > ranking.txt

ranking.rb

require 'java'
require 'net/http'
require 'uri'
require 'nokogiri'

System = java.lang.System

java_import java.util.concurrent.Executors
@executor = Executors.newFixedThreadPool 3

class TooManyRedirects < StandardError; end

def ratings_from_body(body)
    ratings = {:current => {}, :all => {}} 
    document = Nokogiri::HTML(body)
    rating_container = document.at_css('div.app_slide_container')
    unless rating_container.nil?
        current_el = rating_container.at_css('div#r_current')
        unless current_el.nil?
          strongs = current_el.css('strong')
          rating = strongs[0].content.to_f
          count = strongs[1].content.to_i
          ratings[:current] = {:rating => rating, :count => count}
        end
        all_el = rating_container.at_css('div#r_all')
        unless all_el.nil?
          strongs = all_el.css('strong')
          rating = strongs[0].content.to_f
          count = strongs[1].content.to_i
          ratings[:all] = {:rating => rating, :count => count}
        end       
    end
    ratings
end

def get_body(url, redirects = 5)
  uri = URI(url)
  response = Net::HTTP.get_response(uri)
  if response.kind_of? Net::HTTPRedirection
    raise TooManyRedirects if redirects == 0
    return get_body(response['location'], redirects - 1)
  end
  response.body
end

def request_ratings(store_id, details = {})
    app_id = details[:app_id] || ""
    app_name = details[:name] || ""
 
    @executor.submit do
       body = nil
        begin
          url = "http://www.appannie.com/app/ios/#{store_id}"         
          #System.out.println("#Request! #{url}")
          body = get_body url
        rescue TooManyRedirects
          System.err.println("#Timeout! #{store_id}")
        end
        unless body.nil?
          ratings = ratings_from_body(body)
          current_rating = ratings[:current][:rating] || 0
          current_count = ratings[:current][:count] || 0
          all_rating = ratings[:all][:rating] || 0
          all_count = ratings[:all][:count] || 0

          System.out.printf("%s,%s,%s,%s,%s,%s\n", app_name, app_id, current_rating, current_count, all_rating, all_count)
        end
    end
end

lines = File.open(ARGV[0], 'rb').read.split("\n")
apps = lines.map { |v| v.split(",")}

apps.each do |details|
    name = details[1]#['name']
    app_id = details[2]#['app_id']
    id = details[0]#['store_id']
    request_ratings id, {:name => name, :app_id => app_id}
end
#request.headers['User-Agent'] = 'Mozilla/5.0 (Windows NT 6.1; WOW64; rv:7.0.1) Gecko/20100101 Firefox/7.0.12011-10-16 20:23:00'