wiki-genres.rb

# Gets the genres for tracks in a playlist called 'Genres'in iTunes
# and sets them using data from wikipedia.
#
# Uses the album name for the search, but won't try anything fancy
# if it can't find the labum it will just ignore it.
#
# ref: http://www.apeth.com/rbappscript/05propel.html

require 'appscript'
require 'nokogiri'
require 'open-uri'
gem 'dm-core', '=0.10.2'
require 'dm-core'
require 'highline/import'

class Genre
  include DataMapper::Resource
  
  property :id, Serial
  property :name, String
end

DataMapper.setup(:default, "sqlite3://#{Dir.pwd}/genres.db")
DataMapper.auto_upgrade!

include Appscript

def get_genres( album )
  begin
  
    arg = album.gsub(' ', '_')
    url = "http://en.wikipedia.org/wiki/#{arg}"
    doc = Nokogiri::HTML( open(url) )
    
    genres = []
    doc.css('td.category').each do |a|
      a.css('a').each do |l|
        l.to_s =~ /title="([a-zA-Z0-9_ -]+)"/
        genres << $1
      end
    end
    return genres
    
  rescue OpenURI::HTTPError => e
    puts "Bad status code: #{e.message}"
    return nil
  end
end

def choose_genre(genres)
  genres.each_with_index do |g, i|
    if Genre.first(:name => g)
      puts "#{i}. #{g} (used)"
    else
      puts "#{i}. #{g}"
    end
  end
  i = ask("Choose a genre from the list above, use the index! Or type your own!")
  genre = genres[i.to_i] if Float(i) rescue genre = i
  Genre.create(:name => genre) unless Genre.first(:name => genre)
  
  genre
end

def sort_tracks(tracks)
  a = Hash.new {|h, k| h[k] = []}
  tracks.each do |t|
    a[t.album.get] << t.get
  end
  a
end

def process_album(album, tracks)
  puts "Getting genre for #{album}..."
  genres = get_genres(album)
  return nil if genres.nil? || genres.length == 0
  genre = ''
  if genres.length == 1
    genre = genres[0]
  else
    genre = choose_genre(genres)
  end
  tracks.each do |t|
    t.genre.set(genre)
  end
end

itunes = app('iTunes.app')

@albums = sort_tracks(itunes.playlists["Genres"].tracks.get)
puts "Sorted tracks..."

@albums.each do |a, ts|
  process_album(a, ts)
end

puts "Done"