grab.rb

require 'rubygems'
require 'nokogiri'
require 'open-uri'
require 'json'
require 'cgi'
require 'find'
require 'net/http'
require 'progressbar'

class Movie
  
  def initialize(name, imdb_link, torrent_link=nil, year=nil, downloaded=nil)
    @name = name
    @imdb_link = imdb_link
    @torrent_link = torrent_link
    @year = year
    @downloaded = downloaded
  end  
  
  attr_reader :name, :imdb_link, :torrent_link, :year, :downloaded
  
  attr_writer :year, :torrent_link, :downloaded
  
  def get_year
    imdb_data = Nokogiri::HTML(open(@imdb_link))
    imdb_data.css("a[href*='/year/']").each do |link|
      @year = link.content
    end
  end
  
  def to_json(*a)
    {
      "title" => @name,
      "info"  => {
        "imdb_link"    => @imdb_link,
        "torrent_link" => @torrent_link,
        "year"         => @year,
        "downloaded"   => @downloaded }
    }.to_json(*a)
  end
  
  def find_torrent
    if @torrent_link.nil? 
      if @year.nil? 
        self.get_year
      end
    
      btjunkie_url = "http://btjunkie.org/search?q=" + CGI.escape(@name) + "+" + @year
      btjunkie_url = btjunkie_url.to_str.gsub("%E2%80%99","%27")
      html_data = Nokogiri::HTML(open(btjunkie_url))
      torrents = []
    
      html_data.css("a[href^='http://dl.btjunkie.org/torrent/']").each do |torrent|
        rank = torrent.parent.next.at("td>div[style*='vote_up.png']>font")  
        if !rank.nil?
          #must have five positive reports
          if rank.content.to_i > 4
    
            #easiest way to grab all the torrent details (pretty nasty)
            detail_str = torrent.parent.parent.parent.parent.parent.content.to_str
            details = detail_str.split
            file_size = details[-4].to_i
            seeders = details[-2].to_i
            leechers = details[-1].to_i
    
            if (650 < file_size) && (file_size < 1401) && (seeders > 1)
              #stick them in an array, not really altogether sure why, cuz btjunkie sorts 'em pretty well ¬_¬
              torrents << [torrent["href"], file_size, seeders, leechers]
            end
          end
        end
      end
    
      #possibly some crazy sorting, but for now quick and simplez
      if torrents.length > 0
        @torrent_link = torrents[0][0]
      end
      #puts torrents.length
    end
  end
  
  def download_torrent(folder)
    file_name = folder + @name.gsub(/[^[:alnum:]]/, '') + ".torrent"
    
    if !@torrent_link.nil?
      tor = @torrent_link
      File.open(file_name, 'wb') { |f| f.write Net::HTTP.get(URI.parse tor) }
      @downloaded = true
    end
  end

end

class MovieList

  def initialize
    @movies = []
    @file = 'movies.js'
    @dir = 'torrents/'
    puts "\n\nOpening save file..."
    self.open_from_file
    print self.length
    puts " Movies loaded OK","Getting new film list..."
    self.parse_feed
    puts "OK", "Finding torrent files..."
    self.find_torrents
    puts "OK","Downloading Torrents..."
    self.grab_torrents
    puts "All done."
    self.write_to_file
  end
  
  def <<(aMovie)
    findit = @movies.find { |bMovie| bMovie.name == aMovie.name }
    if findit.nil?
      @movies << aMovie
      self
    end
  end
  
  def [](key)
    return @movies[key] if key.kind_of?(Integer)
    return @movies.find { |aMovie| aMovie.imdb_link == key }
  end
  
  def parse_feed
    feed_url = "http://torrentfreak.com/category/dvdrip/feed/"
    xml_data = Nokogiri::HTML(open(feed_url))
    xml_data.css("a[href^='http://www.imdb.com/']").each do |link|
      #check regex for (TC), (CAM) or (TS)  
      if !link.parent.content.match("\(TS\)|\(TC\)|\(CAM\)")
        cont = link.content
        self << Movie.new(cont, link.attribute("href"))
      end
    end
  end
  
  def length
    @movies.length
  end
  
  def list
    str = ""
    @movies.each do |mov|
      str += mov.name + " (" + mov.year + ") " + mov.imdb_link + "\n"
    end
    str
  end
  
  def find_torrents
    pb = ProgressBar.new("finding", @movies.length)
    @movies.each do |mov|
      if mov.torrent_link.nil?
        mov.find_torrent
        self.write_to_file
      end
      pb.inc
    end
    pb.finish
  end
  
  def grab_torrents
    pb = ProgressBar.new("grabbing", @movies.length)
    @movies.each do |mov|
      if mov.downloaded.nil?
        mov.download_torrent(@dir)
      end
      pb.inc
    end
    pb.finish
  end
  
  def write_to_file
    File.open(@file, 'w') { |f| f.write(JSON.pretty_generate(@movies)) }
  end
  
  def open_from_file
    JSON.parse(File.read(@file)).each do |mov|
        self << Movie.new(mov["title"], mov["info"]["imdb_link"], mov["info"]["torrent_link"], mov["info"]["year"], mov["info"]["downloaded"])
    end
  end
  
end

list = MovieList.new