gogaku2016.rb

#!bundle exec ruby
# -*- coding: utf-8 -*-
require 'open-uri'
require 'openssl'
require 'rexml/document'
require 'rubygems'
require 'taglib'

OP_NO_TLSv1_2 = 0x08000000
OpenSSL::SSL::SSLContext::DEFAULT_PARAMS[:options] |= OP_NO_TLSv1_2

def make_path file
  "https://nhk-vh.akamaihd.net/i/gogaku-stream/mp4/#{file}/master.m3u8"
end

def listdataxml(lang, course)
  "https://www2.nhk.or.jp/gogaku/st/xml/#{lang}/#{course}/listdataflv.xml"
end

def download path, outfile
  system "ffmpeg -i #{path} -absf aac_adtstoasc -acodec copy #{outfile}"
end

def proc_xml url
  open(url) do |f|
    doc = REXML::Document.new(f)
    doc.elements.each("musicdata/music") do |e|
      title = e.attributes["title"]
      hdate = e.attributes["hdate"]
      kouza = e.attributes["kouza"]
      code = e.attributes["code"]
      file = e.attributes["file"]
      path = make_path(file)
      outfile = "data/" + file.sub("mp4", "m4a")
      if FileTest.exist?(outfile) && FileTest.size?(outfile) >= 1000000
        puts "Skipped . . . #{outfile}"
      else
        puts "Downloading . . . #{outfile}"
        download path, outfile
        set_title outfile, title, hdate, kouza, code
      end
    end
  end
end

def proc_url_list(list)
  list.each do |url|
    proc_xml url
  end
end

def language(lang)
  listdataxml(lang, "kouza")
end

def levelup(lang)
  listdataxml(lang, "levelup")
end

def english(course)
  listdataxml("english", course)
end

def set_title file, title, hdate, kouza, code
  TagLib::MP4::File.open(file) do |mp4|
    tag = mp4.tag
    tag.genre = "Education"
    tag.artist = 'NHK'
    tag.album = title
    tag.title = hdate
    tag.year = 2013
    tag.track = code.to_i
    tag.comment = code
    mp4.save
  end
end


list = [
  language("german"),
  language("french"),
  language("italian"),
  language("spanish"),
  language("russian"),
  language("chinese"),
  levelup("chinese"),
  language("hangeul"),
  levelup("hangeul"),
  english("business1"),
  english("business2"),
  english("yomu"),
]

# list = [ english("business1") ]
# list = [ english("yomu") ]

proc_url_list list