Simple last.fm ripper

lastfm_rip.rb

#!/usr/bin/env ruby
#
# Usage:
#  ./lastfm_rip.rb "http://www.last.fm/music/Various+Artists/Razor%2B1911%2BChipdisk%2B%25233" "Razor Chipdisk #3"
# Requires: em-http-request, mechanize
require 'rubygems'
require 'mechanize'
require 'fileutils'
require 'em-http-request'
require 'cgi'

url = ARGV[0]
output_dir = ARGV[1]

if !url || !output_dir
  puts "At least two arguments required, lastfm page URL and output directory"
  exit
end

FileUtils.mkdir_p(output_dir)

em_thread = Thread.new { EventMachine.run }
em_thread.abort_on_exception = true

agent = Mechanize.new
page = agent.get(url)

multi_request = EM::MultiRequest.new

filenames = {}
page.links.each do |link|
  if link.text =~ /free download/i
    filename = CGI.unescape(CGI.unescape(File.basename(link.href))) # I have no slightest idea why they encode the filenames twice
    request = EM::HttpRequest.new(link.href).get
    filenames[request] = filename
    multi_request.add(request)
  end
end

multi_request.callback do
  responses = multi_request.responses
  if responses[:succeeded].size > 0
    puts "Succeded:"
    responses[:succeeded].each do |req|
      filename = filenames[req]
      puts "\t#{filename}"
      File.open(File.join(output_dir, filename), "w+") {|f| f.write(req.response) }
    end
  end

  if responses[:failed].size > 0
    puts "Failed: "
    responses[:failed].each do |req|
      puts "\t#{filenames[req]}"
    end
  end

  EM.stop do
    exit
  end

end


em_thread.join