lastfm.rb

# vim:fileencoding=utf-8
# http://www.smokeymonkey.net/2009/03/lastfmrankruby.html
require 'rubygems' if RUBY_VERSION < '1.9'
require 'open-uri'
require 'nokogiri'
require 'date'
require 'activesupport'

def extract_from(url, regexp)
  id = $1 if regexp =~ url['href']
  username = url.inner_text
  return id, username
end

def parse(track, element)
  case element
  when 'date'
    Date.parse(track.at('date').inner_text)
  else
    track.at(element).inner_text
  end
end

def check_enabled?
  timestamp = 'timestamp.yaml'
  last_checked = File.exist?(timestamp) ? YAML.load(File.open(timestamp)) : 30.minutes.ago
  Time.now - last_checked > 30.minutes
end

unless check_enabled?
  puts 'too early'
  exit
end

doc = Nokogiri::HTML(open('http://twitter.g.hatena.ne.jp/keyword/Last.fm%E9%83%A8'))
header, *users = doc.xpath('//div[@class="section"][3]/table/tr')
users = users.map do |user|
  twitter, lastfm = user.xpath('td/a')
  t_id, t_name = extract_from(twitter, %r|http://twitter\.com/(.*)|)
  l_id, = extract_from(lastfm, %r!http://www\.last(?:fm\.jp|\.fm)/user/(.*[^/])!)
  {:twitter => {:id => t_id, :name => t_name}, :lastfm => l_id}
end

tracks = users.inject([]) do |list, user|
  url = "http://ws.audioscrobbler.com/1.0/user/%s/recenttracks.xml" % user[:lastfm]
  result = begin
    doc = Nokogiri::XML(open(url))
    track = doc.at('//recenttracks/track')
    %w(date artist name).map{|element| parse(track, element)} << user[:lastfm] << user[:twitter][:id]
  rescue
    []
  end
  list + [result]
end.reject(&:empty?).sort{|a,b| b[0] <=> a[0]}    # sort by date

YAML.dump(Time.now, File.open('timestamp.yaml', 'w'))
YAML.dump(tracks, File.open('lastfm.yaml', 'w'))
recent_music = tracks.first.each{|info| puts info}