Ruby Tapas Episode fetcher

.envrc.sample

export TAPAS_HTTP_COOKIE=
export HTTP_AUTH=

Gemfile

source "https://rubygems.org"

gem "typhoeus"
gem "nokogiri"

Gemfile.lock

GEM
  remote: https://rubygems.org/
  specs:
    ethon (0.7.1)
      ffi (>= 1.3.0)
    ffi (1.9.6)
    mini_portile (0.6.0)
    nokogiri (1.6.3.1)
      mini_portile (= 0.6.0)
    typhoeus (0.6.9)
      ethon (>= 0.7.1)

PLATFORMS
  ruby

DEPENDENCIES
  nokogiri
  typhoeus

tapas.rb

require 'byebug'
require 'typhoeus'
require 'nokogiri'
require 'pathname'

# TapasDownloader class
class TapasDownloader
  HTTP_COOKIE   = ENV['TAPAS_HTTP_COOKIE'] || fail("Set a valid TAPAS_HTTP_COOKIE, e.g.\n`export TAPAS_HTTP_COOKIE=\"..\"`")
  AUTHORIZATION = ENV['HTTP_AUTH'] || fail("Set a valid HTTP_AUTH, e.g.\n`export HTTP_AUTH=\"Basic Z3JlZ29yeS5pZ2VsbXVuZEBnbWFpbC5jb..\"`")

  HEADERS = {
    'Accept'          => 'text/html',
    'Accept-Language' => 'en-US,en;q=0.8',
    'Connection'      => 'keep-alive',
    'Host'            => 'rubytapas.dpdcart.com',
    'Cookie'          => HTTP_COOKIE,
    'Authorization'   => AUTHORIZATION,
    'User-Agent'      => 'Tapas Downloader'
  }

  RSS_URL = 'https://rubytapas.dpdcart.com/feed'

  def initialize(download_dir: 'downloads')
    @hydra = Typhoeus::Hydra.new(max_concurrency: 5)
    @download_path = Pathname.new(download_dir)
  end

  def feed
    @feed ||= begin
      response = request(RSS_URL).run
      response.code == 200 || fail("#{RSS_URL} returned #{response.code}")
      response.body
    end
  end

  def doc
    @doc ||= Nokogiri::XML(feed)
  end

  def episode_urls
    @urls ||= doc.xpath('//item/enclosure/@url').map(&:value)
  end

  def fetch_all
    requests = episode_urls.map { |url| request(url) }
    prepared_requests = filter_and_prepare_requests requests
    prepared_requests.each { |request| @hydra.queue(request) }
    if prepared_requests.size > 0
      puts "Starting download process for #{prepared_requests.size} files"
      @hydra.run
    else
      puts "No new files to download"
    end
  end

  def filter_and_prepare_requests(requests)
    requests.map do |request|
      target_file = @download_path + File.basename(request.base_url)
      if File.exists? target_file
        puts "Skipping #{target_file} since the file already exists."
        next
      end
      request.on_complete do |response|
        on_complete_handler(request, response, target_file)
      end
      request
    end.compact.flatten
  end

  def on_complete_handler(request, response, target_file)
    fail "Boo #{request.base_url}" unless response.code == 200
    puts "Finished: #{request.base_url}"
    IO.binwrite target_file, response.body
    puts ".. written to #{target_file}.\n"
  end

  private

  def setup_dir
    FileUtils.mkdir absolute_path_for('downloads')
  end

  def request(url, opts = {})
    Typhoeus::Request.new url, headers: HEADERS
  end

  def absolute_path_for(path)
    File.expand_path("../#{path}" || '..', __FILE__)
  end
end

downloader = TapasDownloader.new download_dir: ARGV.first
downloader.fetch_all

update

ruby tapas.rb "/Users/gregoryigelmund/Movies/Educational/ruby-tapas"