asciiwwdc-scraper.rb

require 'open-uri'
require 'nokogiri'

year = 2014
doc = Nokogiri::HTML(open("https://developer.apple.com/videos/wwdc/#{year}/"))

doc.search("p.download").each do |download|
  href = download.at("a").attr('href')
  uri = URI(href)
  uri.path = uri.path.split('/')[0...-1].join('/') + '/'
  uri.query = nil

  p session = uri.path.split('/').last.to_i

  begin
    subtitles = []
    m3u8 = open(URI.join(uri, "subtitles/eng/prog_index.m3u8"))
    m3u8.each_line do |line|
      next unless /\.webvtt$/ === line
      webvtt = open(URI.join(uri, "subtitles/eng/", line.strip))
      subtitles << webvtt.read
    end

    File.open("#{session}.vtt", "w") do |f|
      f.write subtitles.join("\n")
    end
  rescue => e
    p "! #{session} (#{e})"
    next
  end
end