oldhaven/lib/discogs.rb

require 'zlib'

class Discogs
  
  def self.artists
    self.get_nodes('tmp/data/discogs_artists.xml', 'artist') do |node|
      artist = Artist.new(
        :name => (node.css('name').first.text),
        :original_name => (node.css('name').first.text),
        :pic_url => (node.css('images > image[type="primary"]').first.attr('uri') unless node.css('images > image[type="primary"]').empty?),
        :status => 1
      )
      artist.save
      node.css('namevariations > name, aliases > name').each do |v|
        ArtistAlias.new(
          :artist_id => artist.id,
          :name => v.text
        ).save
      end
    end
  end
  
  def self.releases
    self.get_nodes('tmp/data/discogs_releases.xml', 'release') do |node|
      
      # Defining artist
      main_artist = nil
      unless node.css('artists > artist > name').empty?
        main_artist = Artist.find_or_create_by_name(node.css('artists > artist > name').first.text)
      end
      
      self.save_album main_artist, node
      
    end
  end
  
  def self.save_album main_artist, album, node
    unless node.css('master').empty?
      album.master_id = node.css('master').first.attr('id')
    end
  
    # Parsing image
    # unless node.css('images > image[type="primary"]').empty?
    #   album.pic_url = node.css('images > image[type="primary"]').first.attr('uri')
    #   album.has_pic = 1
    # else
    #   if node.css('images > image[type="secondary"]').empty?
    #     album.has_pic = 0
    #   else
    #     album.pic_url = node.css('images > image[type="secondary"]').first.attr('uri')
    #     album.has_pic = 1
    #   end
    # end
    
    album.save ## ## ##
    
    # Defining formats
    formats = node.css('formats > format > descriptions > description').each do |f|
      format = ReleaseFormat.find_or_create_by_name(f.text)
      if format.hash.nil?
        format.hash = f.text.scan(/\w/).join().downcase
        format.save
      end
      AlbumFormat.new(
        :album_id => album.id,
        :release_format_id => format.id
      ).save
    end
  
    # Defining genres
    unless node.css('genres > genre').empty?
      node.css('genres > genre').each do |g|
        genre = Genre.find_or_create_by_name(g.text)
        AlbumGenre.new(
          :album_id => album.id,
          :genre_id => genre.id
        ).save
      end
    end
    
    # Defining styles
    unless node.css('styles > style').empty?
      node.css('styles > style').each do |s|
        style = Style.find_or_create_by_name(s.text)
        AlbumStyle.new(
          :album_id => album.id,
          :style_id => style.id
        ).save
      end
    end
  
    # Writing tracklist
    unless node.css('tracklist > track').empty?
      node.css('tracklist > track').each do |t|
        track = Track.new(
          :album_id => album.id,
          :name => (t.css('title').first.text unless t.css('title').empty?),
          :position => (t.css('position').first.text.scan(/[a-zA-Z0-9]/).join('').to_i(36) unless t.css('position').empty?),
          :country => (node.css('country').first.text unless node.css('country').empty?),
          :length => (self.duration_to_length(t.css('duration').first.text) unless t.css('duration').empty?)
        )
        track.save
        
        if t.css('artists > artist').empty?
          TrackArtist.new(
            :track_id => track.id,
            :artist_id => (main_artist.id unless main_artist.nil?),
            :main => true,
            :name => main_artist.name
          ).save
        else
          t.css('artists > artist').each_with_index do |a, i|
            t_artist = Artist.find_or_create_by_name(a.css('name').first.text)
            TrackArtist.new(
              :track_id => track.id,
              :artist_id => t_artist.id,
              :main => (i == 0),
              :join => a.css('join').first.text,
              :name => a.css('name').first.text
            ).save
          end
        end
        
      end
    end
  end
  
  def self.get_master_albums url
    begin
      xml = open(url,
        'User-Agent' => 'Haven Search Bot',
        'Cookie' => 'artist_layout=med; artist_limit=500'
      ).read
    rescue => e
      xml = ""
    end
    x = Nokogiri::HTML(xml)
    albums = []
    split = 0
    x.css('table.discography > tbody > tr').each do |_|
      if _.attr('class') == 'credit_header'
        split+=1
      end
      if split == 2
        break
      end
      if _.attr('class').match(/^main.*/) and _.css('td')[3].text.scan(/\d/).join('') != '' and (
                _.css('a.mr_toggler').length == 1 or
                _.css('td')[1].text.match(/\(.*?Album.*?\)/).present?
              )
        albums << {
          :uri => (_.attr('class').match(/master/) ? 'master/' : 'release/')+_.attr('id')[1,100],
          :title => _.css('h4 > a').first.text,
          :year => _.css('td')[3].text.scan(/\d/).join('')
        }
      end
    end
    albums
  end
  
  def self.search_for_artist name
    url = 'http://api.discogs.com/search?q=' << CGI::escape(name) << '&f=xml'
    stream = open(url,
      'User-Agent' => 'Haven Import Bot',
      'Accept-Encoding' => 'gzip, deflate'
    )
    if (stream.content_encoding.empty?)
      body = stream.read
    else
      body = Zlib::GzipReader.new(stream).read
    end
    x = Nokogiri::XML(body)
    x.css('exactresults > result[type=artist] > uri, searchresults > result[type=artist] > uri').map{|n| n.text}.first
  end
  
private
  
  def self.get_nodes filename, nodename, &block
    File.open(filename) do |file|
      Nokogiri::XML::Reader.from_io(file).each do |node|
        if node.name == nodename and node.node_type == Nokogiri::XML::Reader::TYPE_ELEMENT
          yield(Nokogiri::XML(node.outer_xml).root)
        end
      end
    end
  end
  
  def self.duration_to_length duration
    duration = duration.split(':')
    duration[0].to_i * 60 + duration[1].to_i
  end
  
end