Loading data from MusicBrainz Web Service, stability fixes, database reorganized. Closes #39

2011-06-18 01:49:32 +04:00
parent a66300c8e4
commit 17f5edf841
15 changed files with 179 additions and 170 deletions
@@ -21,23 +21,25 @@ class ArtistController < ApplicationController
    data['albums'] = []
    albums = artist.albums
    albums.each do |album|
-      tmp_album = {name: album.name, year: album.year, pic: album.pic_url}
-      album_tracks = []
-      bonus_tracks = []
-      album.tracks.each do |track|
-        tmp_track = {name: track.name, live: track.live, acoustic: track.acoustic}
-        if track.length
-          time = (track.length / 1000).round
-          time_m = (time / 60).floor
-          time_s = time - time_m * 60
-          tmp_track['duration'] = time_m.to_s + ':' + (time_s < 10 ? '0' : '') + time_s.to_s
-        else
-          tmp_track['duration'] = '0:00'
+      if album.album_type == 'Album'
+        tmp_album = {name: album.name, year: album.year, pic: album.pic_url}
+        album_tracks = []
+        bonus_tracks = []
+        album.tracks.each do |track|
+          tmp_track = {name: track.name, live: track.live, acoustic: track.acoustic}
+          if track.length
+            time = (track.length / 1000).round
+            time_m = (time / 60).floor
+            time_s = time - time_m * 60
+            tmp_track['duration'] = time_m.to_s + ':' + (time_s < 10 ? '0' : '') + time_s.to_s
+          else
+            tmp_track['duration'] = '0:00'
+          end
+          (track.bonus == 0 ? album_tracks : bonus_tracks) << tmp_track
        end
-        (track.bonus == 0 ? album_tracks : bonus_tracks) << tmp_track
+        tmp_album['tracks'] = {album: album_tracks, bonus: bonus_tracks}
+        data['albums'] << tmp_album
      end
-      tmp_album['tracks'] = {album: album_tracks, bonus: bonus_tracks}
-      data['albums'] << tmp_album
    end
    render :json => data
  end
@@ -1,82 +1,133 @@
 class ImportController < ApplicationController
+  @@user_agent = 'BeatHaven.org'
+  @@lastfm_api_key = '04fda005dbf61a50af5abc3e90f111f2'
  def self.importArtist name
-    # Get artist from MB
-    mb_artist = MusicBrainzArtist.getByName(name)
-    # # Save artist
-    artist_xml = open(
+
+    # Get artist info
+    artist_mb_xml = open(
+      'http://musicbrainz.org/ws/2/artist/?query='+ URI.escape(name) +'&limit=1',
+      'User-Agent' => @@user_agent
+    ).read
+    artist_mb_data = artist_mb_xml.scan(/<artist.*?type=\"(.+?)\"\sid=\"([a-f0-9-]+?)\">.*?<name>(.+?)<\/name>/m)
+    artist_lastfm_xml = open(
      'http://ws.audioscrobbler.com/2.0/'+
-      '?method=artist.getinfo&api_key=04fda005dbf61a50af5abc3e90f111f2'+
-      '&artist='+ URI.escape(name)).read
+      '?method=artist.getinfo&api_key='+ @@lastfm_api_key +
+      '&artist='+ URI.escape(name)
+    ).read
+    
+    # Save artist
    artist = Artist.new
-    artist.name = mb_artist.name
-    artist.desc = artist_xml.scan(/<summary><\!\[CDATA\[?(.*)\]\]><\/summary>/m)[0][0]
-    artist.pic_url = artist_xml.scan(/<image\ssize=\"extralarge\">?(.*)<\/image>/)[0][0]
+    artist.name = artist_mb_data[0][2]
+    artist.desc = artist_lastfm_xml.scan(/<summary><\!\[CDATA\[?(.*)\]\]><\/summary>/m)[0][0]
+    artist.pic_url = artist_lastfm_xml.scan(/<image\ssize=\"extralarge\">?(.*)<\/image>/)[0][0]
+    artist.artist_type = artist_mb_data[0][0]
+    artist.mbid = artist_mb_data[0][1]
    artist.save
+    
    # Get albums from MB
-    mb_albums = MusicBrainzAlbum.where('artist_id = ? AND release_type = 1', mb_artist.id).order('year ASC, id ASC')
-    mb_albums.each do |mb_album|
+    release_groups_mb_xml = open(
+      'http://musicbrainz.org/ws/2/release-group/?artist='+ artist_mb_data[0][1],
+      'User-Agent' => @@user_agent
+    ).read
+    release_groups_mb_data = release_groups_mb_xml.scan(/<release-group\stype=\"([a-zA-Z]+?)\"\sid=\"([a-f0-9-]+?)\"><title>(.+?)<\/title>/m)
+    
+    release_groups_mb_data.each do |mb_album|
+      
      # Get album releases from MB
-      mb_releases = MusicBrainzRelease.where('album_id = ? AND release_type = 1', mb_album.id).order('date ASC, id ASC')
-      unless mb_releases.empty?
+      releases_mb_xml = open(
+        'http://musicbrainz.org/ws/2/release/?release-group='+ mb_album[1],
+        'User-Agent' => @@user_agent
+      ).read
+      releases_mb_data = releases_mb_xml.scan(/<release\sid=\"([a-f0-9-]+?)\">.*?<status>([a-zA-Z]+?)<\/status>.*?<date>([0-9-]+?)<\/date>.*?<country>([A-Z]+?)<\/country>/m)
+      releases_mb_data.each do |item|
+        if item[2].length == 4
+          item[2] += '-01-01'
+        end
+      end
+      
+      unless releases_mb_data.empty?
+        
+        # Sorting releases by date
+        releases_mb_data.sort! { |a, b| a[2].to_date <=> b[2].to_date }
+        main_release = releases_mb_data.shift
+        
        # Get tracks from the first release and then exclude release
-        mb_main_tracks = MusicBrainzTrack.where('release_id IN (?)', mb_releases.shift.id).order('position ASC')
-        unless mb_main_tracks.empty?
+        main_tracks_mb_xml = open(
+          'http://musicbrainz.org/ws/2/release/'+ main_release[0] +'/?inc=recordings',
+          'User-Agent' => @@user_agent
+        ).read
+        mb_tracks = main_tracks_mb_xml.scan(/<track>.*?<position>(\d+?)<\/position><recording\sid=\"([a-f0-9-]+?)\"><title>(.*?)<\/title><length>(.*?)<\/length>/m)
+        mb_tracks.map! do |item|
+          item << main_release[3]
+        end
+        
+        unless mb_tracks.empty?
+          
          # Saving album
-          album_xml = open(
-            'http://ws.audioscrobbler.com/2.0/'+
-            '?method=album.getinfo&api_key=04fda005dbf61a50af5abc3e90f111f2'+
-            '&artist='+ URI.escape(name).gsub(/\&/, '%26').gsub(/\?/, '%3F')+
-            '&album='+ URI.escape(mb_album.name).gsub(/\&/, '%26').gsub(/\?/, '%3F')).read
+          begin
+            album_lastm_xml = open(
+              'http://ws.audioscrobbler.com/2.0/'+
+              '?method=album.getinfo&api_key='+ @@lastfm_api_key +
+              '&artist='+ URI.escape(name).gsub(/\&/, '%26').gsub(/\?/, '%3F')+
+              '&album='+ URI.escape(mb_album[2]).gsub(/\&/, '%26').gsub(/\?/, '%3F')
+            ).read
+            album_image = album_lastm_xml.scan(/<image\ssize=\"extralarge\">?(.*)<\/image>/)[0][0]
+          rescue
+            album_image = ''
+          end
          album = Album.new
-          album.name = mb_album.name
-          album.year = mb_album.year
+          album.name = mb_album[2]
+          album.year = main_release[2].to_date.year
          album.artist_id = artist.id
-          album.status = 1
-          album.pic_url = album_xml.scan(/<image\ssize=\"extralarge\">?(.*)<\/image>/)[0][0]
+          album.mbid = mb_album[1]
+          album.album_type = mb_album[0]
+          album.pic_url = album_image
          album.save
+          
          # Creating hashed array for matching
          mb_main_tracks_hashed = []
-          mb_main_tracks.each do |mb_track|
-            mb_main_tracks_hashed << mb_track.name.downcase.scan(/[a-z0-9]*/).join
-          end
-          # Getting other releases ids
-          release_ids = []
-          mb_releases.each do |mb_release|
-            release_ids << mb_release.id
+          mb_tracks.each do |mb_track|
+            mb_main_tracks_hashed << mb_track[2].downcase.scan(/[a-z0-9]*/).join
+            mb_track << main_release[3] # Release country
+            mb_track << 0 # Bonus flag
          end
+          
          # Searching for tracks in other releases
-          mb_other_tracks = MusicBrainzTrack.where('release_id IN (?)', release_ids).order('position ASC')
-          mb_bonus_tracks = []
-          mb_other_tracks.each do |mb_track|
-            unless mb_main_tracks_hashed.include? mb_track.name.downcase.scan(/[a-z0-9]*/).join
-              mb_bonus_tracks << mb_track
+          releases_mb_data.each do |mb_release|
+            other_tracks_mb_xml = open(
+              'http://musicbrainz.org/ws/2/release/'+ mb_release[0] +'/?inc=recordings',
+              'User-Agent' => @@user_agent
+            ).read
+            mb_other_tracks = other_tracks_mb_xml.scan(/<track>.*?<position>(\d+?)<\/position><recording\sid=\"([a-f0-9-]+?)\"><title>(.*?)<\/title><length>(.*?)<\/length>/m)
+            mb_other_tracks.each do |item|
+              unless mb_main_tracks_hashed.include? item[2].downcase.scan(/[a-z0-9]*/).join
+                item << mb_release[3] # Release country
+                item << 1 # Bonus flag
+                mb_tracks << item
+              end
            end
-          end # mb_other_tracks.each do |mb_track|
+          end
+          
          # Saving tracks
-          mb_main_tracks.each do |mb_track|
+          mb_tracks.each do |mb_track|
            track = Track.new
-            track.name = mb_track.name
+            track.name = mb_track[2]
            track.album_id = album.id
-            track.position = mb_track.position
-            track.length = mb_track.length
-            track.bonus = 0
-            track.live = mb_track.name.downcase.include? 'live'
-            track.acoustic = mb_track.name.downcase.include? 'acoustic'
+            track.position = mb_track[0]
+            track.length = mb_track[3]
+            track.country = mb_track[4]
+            track.bonus = mb_track[5]
+            track.live = mb_track[2].downcase.include? 'live'
+            track.acoustic = mb_track[2].downcase.include? 'acoustic'
            track.save
-          end #mb_main_tracks.each do |mb_track|
-          mb_bonus_tracks.each do |mb_track|
-            track = Track.new
-            track.name = mb_track.name
-            track.album_id = album.id
-            track.position = mb_track.position
-            track.length = mb_track.length
-            track.bonus = 1
-            track.live = mb_track.name.downcase.include? 'live'
-            track.acoustic = mb_track.name.downcase.include? 'acoustic'
-            track.save
-          end # mb_bonus_tracks.each do |mb_track|
+          end
+          
        end # unless mb_main_tracks.empty?
+        
      end # unless mb_releases.empty?
+      
    end # mb_albums.each do |mb_album|
+    
  end # def self.importArtist name
+  
 end