RailsCasts download

download_railscasts.sh

#!/usr/bin/env ruby

# A script to download the latest episodes from railscasts.com
#
# modifications: sgnn7 <sgnn7@sgnn7.org>
# original author: modsaid   <mahmoud@modsaid.com>

require 'rubygems'
require 'json'
require 'open-uri'

SUBSCRIPTION_ID = 'xxx'
PRO_MEDIA_LINK_PREFIX = "http://media.railscasts.com/assets/subscriptions/#{SUBSCRIPTION_ID}/videos/"
MEDIA_LINK_PREFIX = 'http://media.railscasts.com/assets/episodes/videos/'

EPISODES_API_ENDPOINT = 'http://www.railscasts.com/episodes.json'
MEDIA_TYPE = "webm"
TARGET_DIRECTORY = '.'

def parse_json_uri uri
  result = nil
  open(uri, "rb") do | file |
    result = JSON.parse file.read
  end

  result
end

puts "Checking railscasts feeds for new episodes..."
episodes = parse_json_uri EPISODES_API_ENDPOINT
puts "Total episodes: #{episodes.count}"

downloaded_ids = Dir.new(TARGET_DIRECTORY).entries.select{ |f| f =~ Regexp.new("^(\\d){3}.*.#{MEDIA_TYPE}") }.collect{ |f| f.split('-').first.to_i }
puts "Downloaded #{downloaded_ids.count} episode(s)"

new_id_count = 0

episodes.reject!{ |episode| downloaded_ids.include? episode['position'] }
puts "To download: #{episodes.count} episode(s)"

episodes.each_with_index do |episode, index|
  filename = "#{"%03d" % episode['position']}-#{episode['permalink']}.#{MEDIA_TYPE}"

  if episode['pro'] == true
    media_link = "#{PRO_MEDIA_LINK_PREFIX}#{filename}"
  else
    media_link = "#{MEDIA_LINK_PREFIX}#{filename}"
  end

  puts "Downloading #{media_link}... (#{index + 1}/#{episodes.count})"
  begin
    open("#{TARGET_DIRECTORY}/#{filename}", "wb") do |output_file|
      open(media_link, "rb") do |input_file|
        output_file << input_file.read
      end
    end
  rescue
    File.remove "#{TARGET_DIRECTORY}/#{filename}"
    exit 1
  end

  new_id_count = new_id_count + 1
end

puts "Done. #{new_id_count} new episodes were downloaded"