OpenLanguage english tutorials downloader

download.rb

require 'httparty'  # `gem install httparty`
require 'cgi'
require 'digest/md5'
require 'nokogiri'  # `gem install nokogiri`
require 'pathname'
require 'set'

ROOT = Pathname File.expand_path(__dir__)
DOWNLOAD_PATH = ROOT.join 'Downloads'

USERNAME = 'GIVE_ME_USERNAME_HERE'
PASSWORD = 'GIVE_ME_PASSWORD_HERE'
BASE = 'http://openlanguage.com'
COOKIES = "login_status=1; email=#{CGI.escape USERNAME}; password=#{Digest::MD5.hexdigest PASSWORD}"

trap('INT') do
  $downloading.delete if $downloading && $downloading.exist?
  exit
end

def handle_index link, page: 1
  url = "#{BASE}#{link}"
  puts "Fetch #{url}?page=#{page} ..."
  response = HTTParty.get URI.encode(url), headers: {'Cookie' => COOKIES}, query: {page: page}
  html = Nokogiri::HTML response.body
  page_count = html.css('ul.pagination a').map {|a| a.text.to_i }.max if page == 1
  html.css('#list .col-xs-6 > a:first-child').each do |a|
    handle_lesson a.attributes['href'].value
  end
  return unless page == 1
  (2..page_count).each do |num|
    handle_index link, page: num
  end
end

def handle_lesson link
  url = "#{BASE}#{link}"
  puts "Fetch #{url} ..."
  response = HTTParty.get URI.encode(url), headers: {'Cookie' => COOKIES}
  html = Nokogiri::HTML response.body
  name = html.css('h1').text
  path = DOWNLOAD_PATH.join name
  path.mkpath
  html.css('#lesson-downloads a').each do |a|
    handle_download a.attributes['title'].value, a.attributes['href'].value, name
  end
rescue
  STDERR.puts "Error on #{url}"
  STDERR.puts $!.message
  [email protected] {|backtrace| STDERR.puts backtrace }
end

def handle_download title, link, name
  ext = case title
        when 'Printout' then 'pdf'
        when 'Lesson Audio', 'Dialogue', 'Vocab Review' then 'mp3'
        when 'Practice Materials' then 'ppt'
        else fail "Can't recognize this link: #{title} #{link}"
        end
  path = DOWNLOAD_PATH.join name, "#{title}.#{ext}"
  unless path.exist?
    link = "#{BASE}#{link}"
    puts "Download from #{link} ..."
    $downloading = path
    if system 'wget', '--tries=10', '-O', path.to_s, link
      $downloading = nil
    else
      path.delete if path.exist?
    end
  end
rescue
  STDERR.puts "Error on #{link}"
  STDERR.puts $!.message
  [email protected] {|backtrace| STDERR.puts backtrace }
end

handle_index '/library/learn-english/9/latest'