sima/plugins/internal/echonest.py

   1 # -*- coding: utf-8 -*-
   2 """
   3 Fetching similar artists from echonest web services
   4 """
   5
   6 # standard library import
   7 import random
   8
   9 from collections import deque
  10 from hashlib import md5
  11
  12 # third parties components
  13
  14 # local import
  15 from ...lib.plugin import Plugin
  16 from ...lib.simaecho import SimaEch, EchoError, EchoNotFound
  17 from ...lib.track import Track
  18 from ...lib.meta import Artist
  19
  20
  21 def cache(func):
  22     """Caching decorator"""
  23     def wrapper(*args, **kwargs):
  24         #pylint: disable=W0212,C0111
  25         cls = args[0]
  26         similarities = [art for art in args[1]]
  27         hashedlst = md5(''.join(similarities).encode('utf-8')).hexdigest()
  28         if hashedlst in cls._cache.get('asearch'):
  29             cls.log.debug('cached request')
  30             results = cls._cache.get('asearch').get(hashedlst)
  31         else:
  32             results = func(*args, **kwargs)
  33             cls.log.debug('caching request')
  34             cls._cache.get('asearch').update({hashedlst:list(results)})
  35         random.shuffle(results)
  36         return results
  37     return wrapper
  38
  39
  40 class EchoNest(Plugin):
  41     """Echonest autoqueue plugin http://the.echonest.com/
  42     """
  43
  44     def __init__(self, daemon):
  45         Plugin.__init__(self, daemon)
  46         self.daemon_conf = daemon.config
  47         self.sdb = daemon.sdb
  48         self.history = daemon.short_history
  49         ##
  50         self.to_add = list()
  51         self._cache = None
  52         self._flush_cache()
  53         wrapper = {
  54                 'track': self._track,
  55                 'top': self._top,
  56                 'album': self._album,
  57                 }
  58         self.queue_mode = wrapper.get(self.plugin_conf.get('queue_mode'))
  59
  60     def _flush_cache(self):
  61         """
  62         Both flushes and instanciates _cache
  63         """
  64         name = self.__class__.__name__
  65         if isinstance(self._cache, dict):
  66             self.log.info('{0}: Flushing cache!'.format(name))
  67         else:
  68             self.log.info('{0}: Initialising cache!'.format(name))
  69         self._cache = {
  70                 'asearch': dict(),
  71                 'tsearch': dict(),
  72                 }
  73
  74     def _cleanup_cache(self):
  75         """Avoid bloated cache
  76         """
  77         for _ , val in self._cache.items():
  78             if isinstance(val, dict):
  79                 while len(val) > 150:
  80                     val.popitem()
  81
  82     def get_history(self, artist):
  83         """Constructs list of Track for already played titles for an artist.
  84         """
  85         duration = self.daemon_conf.getint('sima', 'history_duration')
  86         tracks_from_db = self.sdb.get_history(duration=duration, artist=artist)
  87         # Construct Track() objects list from database history
  88         played_tracks = [Track(artist=tr[-1], album=tr[1], title=tr[2],
  89                                file=tr[3]) for tr in tracks_from_db]
  90         return played_tracks
  91
  92     def filter_track(self, tracks):
  93         """
  94         Extract one unplayed track from a Track object list.
  95             * not in history
  96             * not already in the queue
  97             * not blacklisted
  98         """
  99         artist = tracks[0].artist
 100         black_list = self.player.queue + self.to_add
 101         not_in_hist = list(set(tracks) - set(self.get_history(artist=artist)))
 102         if not not_in_hist:
 103             self.log.debug('All tracks already played for "{}"'.format(artist))
 104         random.shuffle(not_in_hist)
 105         #candidate = [ trk for trk in not_in_hist if trk not in black_list
 106                       #if not self.sdb.get_bl_track(trk, add_not=True)]
 107         candidate = []
 108         for trk in [_ for _ in not_in_hist if _ not in black_list]:
 109             if self.sdb.get_bl_track(trk, add_not=True):
 110                 self.log.info('Blacklisted: {0}: '.format(trk))
 111                 continue
 112             if self.sdb.get_bl_album(trk, add_not=True):
 113                 self.log.info('Blacklisted album: {0}: '.format(trk))
 114                 continue
 115             # Should use albumartist heuristic as well
 116             if self.plugin_conf.getboolean('single_album'):
 117                 if (trk.album == self.player.current.album or
 118                     trk.album in [tr.album for tr in self.to_add]):
 119                     self.log.debug('Found unplayed track ' +
 120                                'but from an album already queued: %s' % (trk))
 121                     continue
 122             candidate.append(trk)
 123         if not candidate:
 124             self.log.debug('Unable to find title to add' +
 125                            ' for "%s".' % artist)
 126             return None
 127         self.to_add.append(random.choice(candidate))
 128
 129     def _get_artists_list_reorg(self, alist):
 130         """
 131         Move around items in artists_list in order to play first not recently
 132         played artists
 133         """
 134         # TODO: move to utils as a decorator
 135         duration = self.daemon_conf.getint('sima', 'history_duration')
 136         art_in_hist = list()
 137         for trk in self.sdb.get_history(duration=duration,
 138                                         artists=alist):
 139             if trk[0] not in art_in_hist:
 140                 art_in_hist.append(trk[0])
 141         art_in_hist.reverse()
 142         art_not_in_hist = [ ar for ar in alist if ar not in art_in_hist ]
 143         random.shuffle(art_not_in_hist)
 144         art_not_in_hist.extend(art_in_hist)
 145         self.log.debug('history ordered: {}'.format(
 146                        ' / '.join(art_not_in_hist)))
 147         return art_not_in_hist
 148
 149     @cache
 150     def get_artists_from_player(self, similarities):
 151         """
 152         Look in player library for availability of similar artists in
 153         similarities
 154         """
 155         dynamic = self.plugin_conf.getint('dynamic')
 156         if dynamic <= 0:
 157             dynamic = 100
 158         similarity = self.plugin_conf.getint('similarity')
 159         results = list()
 160         while (len(results) < dynamic
 161             and len(similarities) > 0):
 162             art_pop = similarities.pop()
 163             results.extend(self.player.fuzzy_find_artist(art_pop))
 164         return results
 165
 166     def lfm_similar_artists(self, artist=None):
 167         """
 168         Retrieve similar artists on echonest server.
 169         """
 170         if artist is None:
 171             curr = self.player.current.__dict__
 172             name = curr.get('artist')
 173             mbid = curr.get('musicbrainz_artistid', None)
 174             current = Artist(name=name, mbid=mbid)
 175         else:
 176             current = artist
 177         simaech = SimaEch()
 178         # initialize artists deque list to construct from DB
 179         as_art = deque()
 180         as_artists = simaech.get_similar(artist=current)
 181         self.log.debug('Requesting EchoNest for "{0}"'.format(current))
 182         try:
 183             # TODO: let's propagate Artist type
 184             [as_art.append(str(art)) for art in as_artists]
 185         except EchoNotFound as err:
 186             self.log.warning(err)
 187         except EchoError as err:
 188             self.log.warning('EchoNest: {0}'.format(err))
 189         if as_art:
 190             self.log.debug('Fetched  {0} artist(s) from echonest'.format(
 191                             len(as_art)))
 192         self.log.debug('x-ratelimit-remaining: {}'.format(SimaEch.ratelimit))
 193         return as_art
 194
 195     def get_recursive_similar_artist(self):
 196         ret_extra = list()
 197         history = deque(self.history)
 198         history.popleft()
 199         depth = 0
 200         current = self.player.current
 201         extra_arts = list()
 202         while depth < self.plugin_conf.getint('depth'):
 203             if len(history) == 0:
 204                 break
 205             trk = history.popleft()
 206             if (trk.artist in [trk.artist for trk in extra_arts]
 207                 or trk.artist == current.artist):
 208                 continue
 209             extra_arts.append(trk)
 210             depth += 1
 211         self.log.info('EXTRA ARTS: {}'.format(
 212             '/'.join([trk.artist for trk in extra_arts])))
 213         for artist in extra_arts:
 214             self.log.debug('Looking for artist similar to "{0.artist}" as well'.format(artist))
 215             similar = self.lfm_similar_artists(artist=artist)
 216             if not similar:
 217                 return ret_extra
 218             ret_extra.extend(self.get_artists_from_player(similar))
 219             if current.artist in ret_extra:
 220                 ret_extra.remove(current.artist)
 221         return ret_extra
 222
 223     def get_local_similar_artists(self):
 224         """Check against local player for similar artists fetched from echonest
 225         """
 226         current = self.player.current
 227         self.log.info('Looking for artist similar to "{0.artist}"'.format(current))
 228         similar = list(self.lfm_similar_artists())
 229         if not similar:
 230             self.log.info('Got nothing from echonest!')
 231             return []
 232         self.log.info('First five similar artist(s): {}...'.format(
 233                       ' / '.join([a for a in similar[0:5]])))
 234         self.log.info('Looking availability in music library')
 235         ret = self.get_artists_from_player(similar)
 236         ret_extra = None
 237         if len(self.history) >= 2:
 238             if self.plugin_conf.getint('depth') > 1:
 239                 ret_extra = self.get_recursive_similar_artist()
 240         if ret_extra:
 241             ret = list(set(ret) | set(ret_extra))
 242         if not ret:
 243             self.log.warning('Got nothing from music library.')
 244             self.log.warning('Try running in debug mode to guess why...')
 245             return []
 246         self.log.info('Got {} artists in library'.format(len(ret)))
 247         self.log.info(' / '.join(ret))
 248         # Move around similars items to get in unplayed|not recently played
 249         # artist first.
 250         return self._get_artists_list_reorg(ret)
 251
 252     def _get_album_history(self, artist=None):
 253         """Retrieve album history"""
 254         duration = self.daemon_conf.getint('sima', 'history_duration')
 255         albums_list = set()
 256         for trk in self.sdb.get_history(artist=artist, duration=duration):
 257             albums_list.add(trk[1])
 258         return albums_list
 259
 260     def find_album(self, artists):
 261         """Find albums to queue.
 262         """
 263         self.to_add = list()
 264         nb_album_add = 0
 265         target_album_to_add = self.plugin_conf.getint('album_to_add')
 266         for artist in artists:
 267             self.log.info('Looking for an album to add for "%s"...' % artist)
 268             albums = self.player.find_albums(artist)
 269             # str conversion while Album type is not propagated
 270             albums = [ str(album) for album in albums]
 271             if albums:
 272                 self.log.debug('Albums candidate: {0:s}'.format(' / '.join(albums)))
 273             else: continue
 274             # albums yet in history for this artist
 275             albums = set(albums)
 276             albums_yet_in_hist = albums & self._get_album_history(artist=artist)
 277             albums_not_in_hist = list(albums - albums_yet_in_hist)
 278             # Get to next artist if there are no unplayed albums
 279             if not albums_not_in_hist:
 280                 self.log.info('No album found for "%s"' % artist)
 281                 continue
 282             album_to_queue = str()
 283             random.shuffle(albums_not_in_hist)
 284             for album in albums_not_in_hist:
 285                 tracks = self.player.find_album(artist, album)
 286                 # Look if one track of the album is already queued
 287                 # Good heuristic, at least enough to guess if the whole album is
 288                 # already queued.
 289                 if tracks[0] in self.player.queue:
 290                     self.log.debug('"%s" already queued, skipping!' %
 291                             tracks[0].album)
 292                     continue
 293                 album_to_queue = album
 294             if not album_to_queue:
 295                 self.log.info('No album found for "%s"' % artist)
 296                 continue
 297             self.log.info('echonest album candidates: {0} - {1}'.format(
 298                            artist, album_to_queue))
 299             nb_album_add += 1
 300             self.to_add.extend(self.player.find_album(artist, album_to_queue))
 301             if nb_album_add == target_album_to_add:
 302                 return True
 303
 304     def _track(self):
 305         """Get some tracks for track queue mode
 306         """
 307         artists = self.get_local_similar_artists()
 308         nbtracks_target = self.plugin_conf.getint('track_to_add')
 309         for artist in artists:
 310             self.log.debug('Trying to find titles to add for "{}"'.format(
 311                            artist))
 312             found = self.player.find_track(artist)
 313             # find tracks not in history for artist
 314             self.filter_track(found)
 315             if len(self.to_add) == nbtracks_target:
 316                 break
 317         if not self.to_add:
 318             self.log.debug('Found no tracks to queue, is your ' +
 319                             'history getting too large?')
 320             return None
 321         for track in self.to_add:
 322             self.log.info('echonest candidates: {0!s}'.format(track))
 323
 324     def _album(self):
 325         """Get albums for album queue mode
 326         """
 327         artists = self.get_local_similar_artists()
 328         self.find_album(artists)
 329
 330     def _top(self):
 331         """Get some tracks for top track queue mode
 332         """
 333         #artists = self.get_local_similar_artists()
 334         pass
 335
 336     def callback_need_track(self):
 337         self._cleanup_cache()
 338         if not self.player.current:
 339             self.log.info('No current track, cannot queue')
 340             return None
 341         if not self.player.current.artist:
 342             self.log.warning('No artist set for the current track')
 343             self.log.debug(repr(self.player.current))
 344             return None
 345         self.queue_mode()
 346         candidates = self.to_add
 347         self.to_add = list()
 348         if self.plugin_conf.get('queue_mode') != 'album':
 349             random.shuffle(candidates)
 350         return candidates
 351
 352     def callback_player_database(self):
 353         self._flush_cache()
 354
 355 # VIM MODLINE
 356 # vim: ai ts=4 sw=4 sts=4 expandtab