comedydl.rb

#!/usr/bin/env ruby

require 'net/http'
require 'digest/sha2'
require 'fileutils'
require 'date'

show = ARGV[0]

case show
when 'colbert'
  base_url = 'http://www.colbertnation.com/full-episodes/'
  base_filename = 'The.Colbert.Report.'
when 'dailyshow'
  base_url = 'http://www.thedailyshow.com/full-episodes/'
  base_filename = 'The.Daily.Show.with.Jon.Stewart.'
else
  raise "Unknown show: #{show}"
end

response = Net::HTTP.get_response(URI(base_url))
episode_url = response['Location']

date_part = episode_url.match(/\/full-episodes\/([a-z]+-[a-z]+-[0-9]+-[0-9]+).*/i)[1].gsub('-', ' ')
date = Date.parse(date_part)

hash = Digest::SHA256.hexdigest(episode_url)

if (date.nil?)
  output_filename = "#{base_filename}#{hash}.mp4"
else
  output_filename = "#{base_filename}#{date.strftime('%Y.%m.%d')}.mp4"
end

if (File.exists?(output_filename))
  raise "File already exists: #{output_filename}"
end

begin
  FileUtils.mkdir(hash)
rescue Errno::EEXIST
end

FileUtils.cd(hash) do
  system("youtube-dl \"#{episode_url}\"")
  
  mp4s = Dir.entries('.').reject do |file|
    !file.match(/\.mp4$/)
  end
  
  if (mp4s.count == 0)
    raise "No videos..."
  end
  
  mp4s.sort!
  
  tsses = []
  mp4s.each do |mp4|
    ts = mp4.gsub(/.mp4$/, '.ts')
    
    command = "ffmpeg -y -i #{mp4} -c copy -bsf h264_mp4toannexb #{ts}"
    print "--> #{command}\n"
    system(command)
    
    tsses << ts
  end
  
  tsses_flat = tsses.join('|')
  
  command = "ffmpeg -y -i \"concat:#{tsses_flat}\" -c copy -absf aac_adtstoasc \"../#{output_filename}\""
  print "--> #{command}\n"
  system (command)
  
  FileUtils.rm(mp4s)
  FileUtils.rm(tsses)
end

FileUtils.rmdir(hash)