STIW3 - IMDB's Top 20 Worst Actors

(I'll release webcache later, you can get this working just fine without caching--not that you want to)

require 'rubygems'
require 'hpricot'
require 'webcache'

$max = (ARGV.shift || 10).to_i    class IMDB < WebCache
  def initialize
    super ".imdbcache", "ryand-ruby@zenspider.com Ruby/#{RUBY_VERSION}", 999
    @base = "http://www.imdb.com"
  end

  def bottom
    c = cache("imdb", "bottom") do
      get File.join(@base, "chart/bottom")
    end

    doc = Hpricot(c)

    (doc/"a").select { |a| a.attributes['href'] =~ /title.tt/ }.map { |a|
      [a.attributes['href'], a.children.first.content]
    }
  end

  def movie(url)
    n = url[/\d+/]
    c = cache("movie", n) do
      t = get File.join(@base, url)
      sleep 1
      t
    end

    fragment = c[/^.*(?:Cast overview|[Cc]redited cast).*fullcredits.*$/]

    begin
      doc = Hpricot(fragment)
    rescue => e
      p e.message, fragment
      puts c
      exit 1
    end

    (doc/"a").select { |a| a.attributes['href'] =~ /name.nm/ and Hpricot::Text === a.children.first }.map { |a|
      [a.attributes['href'], a.children.first.content]
    }
  end

  def worst
    movie_names = {}
    actor_names = {}
    movies = Hash.new { |h,k| h[k] = [] }
    scores = Hash.new(0)

    bottom.each_with_index do |(movie_url, movie_name), i|
      movie_names[movie_url] = movie_name
      score = 100 - i
      m = movie(movie_url)
      m.each do |actor_url, actor_name|
        movies[actor_url] << movie_url
        actor_names[actor_url] = actor_name unless actor_names.has_key? actor_url
        scores[actor_url] += score
      end
    end

    by_scores = Hash.new { |h,k| h[k] = [] }

    scores.each do |actor_url, score|
      s = "%s (%s)" % [actor_names[actor_url], movies[actor_url].map { |url| movie_names[url]}.join(", ")]
      by_scores[score] << s
    end

    puts "Top #{$max} worst actors"
    puts

    by_scores.sort_by {|score,x| -score}.first($max).each_with_index do |(score, actors), i|
      n = i + 1
      puts "%2d) %3d %s" % [n, score, actors.join("\n        ")]
    end
  end
end

imdb = IMDB.new
imdb.worst