gogakuondemand.rb

#!/usr/bin/env ruby
# coding: utf-8

# http://d.hatena.ne.jp/riocampos+tech/20130725/p1
# 全講座DLで約20分、約557MB（mp4からmp3への変換はPC能力に依存）

require 'nokogiri'
require 'net/https'
require 'fileutils'

def load_pref
  set_script_dir
  load_pref_file
  load_subjects_file
  set_save_file_dest_dir
end

def set_script_dir
  @script_dir = File.expand_path(File.dirname(__FILE__))
  push_dir(@script_dir)
end

def load_pref_file
  load './pref.rb'
  @select_subjects_jp = []
  pref.each do |k,v|
    @select_subjects_jp << k if v
  end
end

def load_subjects_file
  load './subjects.rb'
  @select_subjects = []
  @subject_urls = subject_urls
  @subjects_jp  = subjects_jp
  @select_subjects_jp.each do |sub_jp|
    @select_subjects << @subjects_jp[sub_jp]
  end
end

def set_save_file_dest_dir
  working_dir = save_file_dest_dir ?
                File.expand_path(File.path(save_file_dest_dir)) :
                File.expand_path(File.dirname(__FILE__))
  Dir.mkdir(working_dir) unless File.directory?(working_dir)
  push_dir(working_dir)
end

def print_download_subjects
  download_subjects = @select_subjects_jp * "、"
  print "ダウンロードする語学講座：#{download_subjects}\n\n"
end

def https_body(url)
  uri = URI.parse(url)
  https = Net::HTTP.new(uri.host, uri.port)
  https.use_ssl = true
  https.ssl_version = :TLSv1
  https.verify_mode = OpenSSL::SSL::VERIFY_PEER
  https.get(uri.path).body
end

def get_source_urls
  data_hash ={}
  @select_subjects.each do |sub|
    url = "https://cgi2.nhk.or.jp/gogaku/#{@subject_urls[sub]}/listdataflv.xml"
    doc = Nokogiri.XML(https_body(url))
    data_array = doc.xpath("/musicdata/music")
    date = data_array.map {|d| d["hdate"].scan(/(\d+)月(\d+)日/)} \
          .map{|dt| "2013_%02d_%02d" % dt.flatten}
    data_hash[sub] = {kouza: data_array[0].values[2]}
    data_array.each_with_index do |d, i|
      data_hash[sub][{date: date[i]}] = data_array[i].values[4]
    end
  end
  data_hash
end

def prepare_download_each_subject(key, value)
  @subject = value[:kouza]
  puts "講座名：#{@subject}"
  subject_dir = File.join(current_dir, key.to_s + "/")
  Dir.mkdir(subject_dir) unless File.directory?(subject_dir)
  subject_dir
end

def connect_directory
  `cd #{@script_dir}; phantomjs #{File.join(@script_dir, "loadConnectDirectory.js")}`.strip
end

def each_date_process(value)
  value.each do |k, v|
    next if k == :kouza
    @metadata = {}
    date = k[:date]
    @metadata[:subject] = @subject
    @metadata[:title]   = "#{@subject}_#{date}"
    @metadata[:genre]   = "Speech"
    @metadata[:create]  = "NHK"
    @metadata[:year]    = "2013"
    puts "日付：#{date}"
    audio_source_url = "https://nhkmovs-i.akamaihd.net/i/gogaku/streaming/mp4/#{connect_directory}/#{v}/index_0_a.m3u8"
    source_urls = https_body(audio_source_url).lines.grep(/http/).map{|url| url.strip }
    crypt_key_url = source_urls.shift[/https:\/\/.*.key/]
    @decrypt_key = https_body(crypt_key_url).unpack('C'*16).map{|x| x.to_s(16)}.pack('a2'*16)
    merge_list = make_merge_list(source_urls)
    print "ファイル変換中...\n"
    merge_mp4(merge_list)
    convert_mp4_mp3
    delete_temp_files
    print "\n"
  end
end

def make_merge_list(urls)
  temp_dir = File.join(current_dir, "temp/")
  Dir.mkdir(temp_dir) unless File.directory?(temp_dir)
  download_and_decrypt_gogaku_files(urls, temp_dir)
end

def progress_bar(progress, max, unit)
  max_digits = max.to_s.size
  print "\r" + "[#{'%-50s' % ('*' * (progress.to_f / max * 50).to_i)}] #{'%*d' % [max_digits, progress]} / #{max} #{unit}"
end

def download_and_decrypt_gogaku_files(urls, temp_dir)
  merge_list = ""
  max = urls.size
  urls.each_with_index do |url, index|
    uri = URI.parse(url)
    http = Net::HTTP.new(uri.host, uri.port)
    ts = http.get(uri.path).body
    iv = '%032x' % (index + 1)
    temp_crypt_ts_path = File.join(temp_dir, "temp_#{index + 1}.crypt.ts")
    temp_mp4_path = temp_crypt_ts_path.sub(/\.crypt\.ts/, '.mp4')
    merge_list << "#{temp_mp4_path} "
    open(temp_crypt_ts_path, "w"){ |f| f.write(ts) }
    `openssl aes-128-cbc -d -in #{temp_crypt_ts_path} -out #{temp_mp4_path} -p -nosalt -iv #{iv} -K #{@decrypt_key}`
    progress_bar(index + 1, max, "ファイル")
  end
  print "\n"
  merge_list
end

def merge_mp4(merge_list)
  `cat #{merge_list} > #{title_path}.mp4`
end

def convert_mp4_mp3
  command_ffmpeg = %Q[ffmpeg -y -i #{title_path}.mp4 -ab 64k -metadata album="#{@metadata[:subject]}" -metadata title="#{@metadata[:title]}" -metadata genre="#{@metadata[:genre]}" -metadata artist="#{@metadata[:create]}" -metadata date="#{@metadata[:year]}" -id3v2_version 3 #{title_path}.mp3 2>&1]
  IO.popen(command_ffmpeg) do |pipe|
    duration = nil
    progress = 0
    pipe.each("r") do |line|
      if line =~ /: (\d{2}):(\d{2}):(\d{2}).(\d{2}),/
        duration = (($1.to_i * 360000 + $2.to_i * 6000 + $3.to_i * 100 + $4.to_i) / 100.0).round
      end
      if line =~ /time=(\d{2}):(\d{2}):(\d{2}).(\d{2})/
        progress = (($1.to_i * 360000 + $2.to_i * 6000 + $3.to_i * 100 + $4.to_i) / 100.0).round
        progress = duration if duration && progress > duration
      end
      if duration
        progress_bar(progress, duration, "秒")
      end
    end
    print "\n"
  end
end

def delete_temp_files
  temp_dir = File.join(current_dir, "temp/")
  temp_mp4 = Dir.glob(File.join(current_dir, "*.mp4"))
  FileUtils.rm_rf(temp_dir)
  FileUtils.rm_f(temp_mp4)
end

def current_dir
  @pwd
end

def title_path
  File.join(current_dir, @metadata[:title])
end

def push_dir(dir)
  @pwds ||= []
  @pwds.push(dir)
  @pwd = @pwds.last
  Dir::chdir(@pwd)
  @pwd
end

def pop_dir
  @pwds.pop
  @pwd = @pwds.last
  Dir::chdir(@pwd)
  @pwd
end

def gogaku_on_demand
  load_pref
  print_download_subjects
  data_hash = get_source_urls
  data_hash.each do |key, value|
    subject_dir = prepare_download_each_subject(key, value)
    push_dir(subject_dir)
    each_date_process(value)
    pop_dir
  end
  puts "作業終了"
end

gogaku_on_demand