audio_file_downloader.rb

#!/usr/bin/env ruby

require 'nokogiri'
require 'open-uri'
require 'progressbar'

def parse_url(url, formats)
  response = open(url, 'User-Agent' => 'Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1;Trident/5.0)')
  page = Nokogiri.parse(response.read)
  links = page.css('a')
  urls = Array.new
  links.each do |anchor|
    if anchor.attributes['href']
      value = anchor.attributes['href'].value
      if value.match(/.*(#{formats.join('|')})$/) != nil
        urls << value
      end
    end
  end
  puts %[Got #{urls.length} audio links.]
  return urls
end

def dowload_url_content(output_path, urls)
  count = 0
  length = urls.length
  urls.each do |uri|
    file_name = uri.split('/').last
    referer = uri.split('/')[0..1].join('//')
    unless File.file? "#{output_path}/#{file_name}"
      puts %[Downloading #{uri} (#{count + 1} of #{length})]
      File.open(%[#{output_path}/#{file_name}], 'wb') do |file|
        pbar = nil
        open(%[#{uri}], 
             'User-Agent' => 'Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/5.0)', 
             'Referer' => %[#{referer}],
             :content_length_proc => lambda { |t|
               if t && 0 < t
                 pbar = ProgressBar.new("Saving...", t)
                 pbar.file_transfer_mode
               end
             },
             :progress_proc => lambda {|s|
               pbar.set s if pbar
             }) do |response|
          file.write(response.read)
        end
      end
      count += 1
      puts %[Saved #{file_name}]
    else
      puts %[#{file_name}: file exists in destination directory.]
    end
  end
  puts %[Downloaded #{count} file(s).]
end

if __FILE__ == $0
  media_file_formats = ['mp3', 'ogg', 'aac']
  unless ARGV.first.nil? and ARGV[1].nil?
    if ARGV.first.match(/^http|https:\/\/.*$/)
      begin
        urls = parse_url(ARGV.first, media_file_formats)
        if File.directory? File.absolute_path(ARGV[1])
          dowload_url_content(File.absolute_path(ARGV[1]), urls)
        else
          Dir.mkdir(File.absolute_path(ARGV[1]))
          dowload_url_content(File.absolute_path(ARGV[1]), urls)
        end
      rescue Interrupt
        puts %[halting...]
      end
    else
      puts 'Enter a url starting with http or https.'
    end
  else
    puts 'Input url, followed by output directory.'
  end
end