sima/lib/webserv.py

   1 # -*- coding: utf-8 -*-
   2 # Copyright (c) 2009-2015 Jack Kaliko <kaliko@azylum.org>
   3 #
   4 #  This file is part of sima
   5 #
   6 #  sima is free software: you can redistribute it and/or modify
   7 #  it under the terms of the GNU General Public License as published by
   8 #  the Free Software Foundation, either version 3 of the License, or
   9 #  (at your option) any later version.
  10 #
  11 #  sima is distributed in the hope that it will be useful,
  12 #  but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 #  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 #  GNU General Public License for more details.
  15 #
  16 #  You should have received a copy of the GNU General Public License
  17 #  along with sima.  If not, see <http://www.gnu.org/licenses/>.
  18 #
  19 #
  20 """
  21 Fetching similar artists from last.fm web services
  22 """
  23
  24 # standard library import
  25 import random
  26
  27 from collections import deque
  28 from hashlib import md5
  29
  30 # third parties components
  31
  32 # local import
  33 from .plugin import Plugin
  34 from .track import Track
  35 from .meta import Artist, MetaContainer
  36 from ..utils.utils import WSError, WSNotFound
  37
  38 def cache(func):
  39     """Caching decorator"""
  40     def wrapper(*args, **kwargs):
  41         #pylint: disable=W0212,C0111
  42         cls = args[0]
  43         similarities = [art.name for art in args[1]]
  44         hashedlst = md5(''.join(similarities).encode('utf-8')).hexdigest()
  45         if hashedlst in cls._cache.get('asearch'):
  46             cls.log.debug('cached request')
  47             results = cls._cache.get('asearch').get(hashedlst)
  48         else:
  49             results = func(*args, **kwargs)
  50             cls.log.debug('caching request')
  51             cls._cache.get('asearch').update({hashedlst:list(results)})
  52         random.shuffle(results)
  53         return results
  54     return wrapper
  55
  56
  57 class WebService(Plugin):
  58     """similar artists webservice
  59     """
  60
  61     def __init__(self, daemon):
  62         Plugin.__init__(self, daemon)
  63         self.daemon_conf = daemon.config
  64         self.sdb = daemon.sdb
  65         self.history = daemon.short_history
  66         ##
  67         self.to_add = list()
  68         self._cache = None
  69         self._flush_cache()
  70         wrapper = {
  71                 'track': self._track,
  72                 'top': self._top,
  73                 'album': self._album,
  74                 }
  75         self.queue_mode = wrapper.get(self.plugin_conf.get('queue_mode'))
  76         self.ws = None
  77
  78     def _flush_cache(self):
  79         """
  80         Both flushes and instanciates _cache
  81         """
  82         name = self.__class__.__name__
  83         if isinstance(self._cache, dict):
  84             self.log.info('{0}: Flushing cache!'.format(name))
  85         else:
  86             self.log.info('{0}: Initialising cache!'.format(name))
  87         self._cache = {
  88                 'asearch': dict(),
  89                 'tsearch': dict(),
  90                 }
  91
  92     def _cleanup_cache(self):
  93         """Avoid bloated cache
  94         """
  95         for _, val in self._cache.items():
  96             if isinstance(val, dict):
  97                 while len(val) > 150:
  98                     val.popitem()
  99
 100     def get_history(self, artist):
 101         """Constructs list of Track for already played titles for an artist.
 102         """
 103         duration = self.daemon_conf.getint('sima', 'history_duration')
 104         tracks_from_db = self.sdb.get_history(duration=duration, artist=artist)
 105         # Construct Track() objects list from database history
 106         played_tracks = [Track(artist=tr[-1], album=tr[1], title=tr[2],
 107                                file=tr[3]) for tr in tracks_from_db]
 108         return played_tracks
 109
 110     def filter_track(self, tracks):
 111         """
 112         Extract one unplayed track from a Track object list.
 113             * not in history
 114             * not already in the queue
 115             * not blacklisted
 116         """
 117         artist = tracks[0].artist
 118         black_list = self.player.queue + self.to_add
 119         not_in_hist = list(set(tracks) - set(self.get_history(artist=artist)))
 120         if self.plugin_conf.get('queue_mode') != 'top' and not not_in_hist:
 121             self.log.debug('All tracks already played for "{}"'.format(artist))
 122         random.shuffle(not_in_hist)
 123         candidate = []
 124         for trk in [_ for _ in not_in_hist if _ not in black_list]:
 125             # Should use albumartist heuristic as well
 126             if self.plugin_conf.getboolean('single_album'):
 127                 if (trk.album == self.player.current.album or
 128                     trk.album in [tr.album for tr in self.to_add]):
 129                     self.log.debug('Found unplayed track ' +
 130                                'but from an album already queued: %s' % (trk))
 131                     continue
 132             candidate.append(trk)
 133         if not candidate:
 134             return False
 135         self.to_add.append(random.choice(candidate))
 136         return True
 137
 138     def _get_artists_list_reorg(self, alist):
 139         """
 140         Move around items in artists_list in order to play first not recently
 141         played artists
 142         """
 143         hist = list()
 144         duration = self.daemon_conf.getint('sima', 'history_duration')
 145         for art in self.sdb.get_artists_history(alist, duration=duration):
 146             if art not in hist:
 147                 hist.insert(0, art)
 148         reorg = [art for art in alist if art not in hist]
 149         reorg.extend(hist)
 150         return reorg
 151
 152     @cache
 153     def get_artists_from_player(self, similarities):
 154         """
 155         Look in player library for availability of similar artists in
 156         similarities
 157         """
 158         dynamic = self.plugin_conf.getint('max_art')
 159         if dynamic <= 0:
 160             dynamic = 100
 161         results = list()
 162         similarities.reverse()
 163         while (len(results) < dynamic
 164                and len(similarities) > 0):
 165             art_pop = similarities.pop()
 166             res = self.player.search_artist(art_pop)
 167             if res:
 168                 results.append(res)
 169         return results
 170
 171     def ws_similar_artists(self, artist):
 172         """
 173         Retrieve similar artists from WebServive.
 174         """
 175         # initialize artists deque list to construct from DB
 176         as_art = deque()
 177         as_artists = self.ws.get_similar(artist=artist)
 178         self.log.debug('Requesting {} for {!r}'.format(self.ws.name, artist))
 179         try:
 180             [as_art.append(art) for art in as_artists]
 181         except WSNotFound as err:
 182             self.log.warning('{}: {}'.format(self.ws.name, err))
 183             if artist.mbid:
 184                 self.log.debug('Trying without MusicBrainzID')
 185                 try:
 186                     return self.ws_similar_artists(Artist(name=artist.name))
 187                 except WSNotFound as err:
 188                     self.log.debug('{}: {}'.format(self.ws.name, err))
 189         except WSError as err:
 190             self.log.warning('{}: {}'.format(self.ws.name, err))
 191         if as_art:
 192             self.log.debug('Fetched {} artist(s)'.format(len(as_art)))
 193         return as_art
 194
 195     def get_recursive_similar_artist(self):
 196         if not self.player.playlist:
 197             return
 198         history = list(self.history)
 199         history = self.player.queue + history
 200         history = deque(history)
 201         last_trk = history.popleft() # remove
 202         extra_arts = list()
 203         ret_extra = list()
 204         depth = 0
 205         while depth < self.plugin_conf.getint('depth'):
 206             if len(history) == 0:
 207                 break
 208             trk = history.popleft()
 209             if (trk.Artist in extra_arts
 210                     or trk.Artist == last_trk.Artist):
 211                 continue
 212             extra_arts.append(trk.Artist)
 213             depth += 1
 214         self.log.debug('EXTRA ARTS: {}'.format(
 215                       '/'.join(map(str, extra_arts))))
 216         for artist in extra_arts:
 217             self.log.debug('Looking for artist similar '
 218                            'to "{}" as well'.format(artist))
 219             similar = self.ws_similar_artists(artist=artist)
 220             if not similar:
 221                 continue
 222             ret_extra.extend(self.get_artists_from_player(similar))
 223
 224         if last_trk.Artist in ret_extra:
 225             ret_extra.remove(last_trk.Artist)
 226         if ret_extra:
 227             self.log.debug('similar artist(s) found: {}'.format(
 228                 ' / '.join(map(str, MetaContainer(ret_extra)))))
 229         return ret_extra
 230
 231     def get_local_similar_artists(self):
 232         """Check against local player for similar artists
 233         """
 234         if not self.player.playlist:
 235             return []
 236         tolookfor = self.player.playlist[-1].Artist
 237         self.log.info('Looking for artist similar to "{}"'.format(tolookfor))
 238         self.log.debug(repr(tolookfor))
 239         similar = self.ws_similar_artists(tolookfor)
 240         if not similar:
 241             self.log.info('Got nothing from {0}!'.format(self.ws.name))
 242             return []
 243         self.log.info('First five similar artist(s): {}...'.format(
 244                       ' / '.join(map(str, list(similar)[:5]))))
 245         self.log.info('Looking availability in music library')
 246         ret = MetaContainer(self.get_artists_from_player(similar))
 247         if ret:
 248             self.log.debug('regular found in library: {}'.format(
 249                            ' / '.join(map(str, ret))))
 250         else:
 251             self.log.debug('Got nothing similar from library!')
 252         ret_extra = None
 253         if len(self.history) >= 2:
 254             if self.plugin_conf.getint('depth') > 1:
 255                 ret_extra = self.get_recursive_similar_artist()
 256         if ret_extra:
 257             # get them reorg to pick up best element
 258             ret_extra = self._get_artists_list_reorg(ret_extra)
 259             # tries to pickup less artist from extra art
 260             if len(ret) < 4:
 261                 ret_extra = MetaContainer(ret_extra)
 262             else:
 263                 ret_extra = MetaContainer(ret_extra[:max(4, len(ret))//2])
 264             if ret_extra:
 265                 self.log.debug('extra found in library: {}'.format(
 266                                ' / '.join(map(str, ret_extra))))
 267             ret = ret | ret_extra
 268         if not ret:
 269             self.log.warning('Got nothing from music library.')
 270             return []
 271         queued_artists = MetaContainer([trk.Artist for trk in self.player.queue])
 272         self.log.trace('Already queued: {}'.format(queued_artists))
 273         self.log.trace('Candidate: {}'.format(ret))
 274         if ret & queued_artists:
 275             self.log.debug('Removing already queued artists: '
 276                            '{0}'.format('/'.join(map(str, ret & queued_artists))))
 277             ret = ret - queued_artists
 278         if self.player.current and self.player.current.Artist in ret:
 279             self.log.debug('Removing current artist: {0}'.format(self.player.current.Artist))
 280             ret = ret -  MetaContainer([self.player.current.Artist])
 281         # Move around similars items to get in unplayed|not recently played
 282         # artist first.
 283         self.log.info('Got {} artists in library'.format(len(ret)))
 284         candidates = self._get_artists_list_reorg(list(ret))
 285         if candidates:
 286             self.log.info(' / '.join(map(str, candidates)))
 287         return candidates
 288
 289     def _get_album_history(self, artist=None):
 290         """Retrieve album history"""
 291         duration = self.daemon_conf.getint('sima', 'history_duration')
 292         albums_list = set()
 293         for trk in self.sdb.get_history(artist=artist.name, duration=duration):
 294             albums_list.add(trk[1])
 295         return albums_list
 296
 297     def find_album(self, artists):
 298         """Find albums to queue.
 299         """
 300         self.to_add = list()
 301         nb_album_add = 0
 302         target_album_to_add = self.plugin_conf.getint('album_to_add')
 303         for artist in artists:
 304             self.log.info('Looking for an album to add for "%s"...' % artist)
 305             albums = self.player.search_albums(artist)
 306             # str conversion while Album type is not propagated
 307             albums = [str(album) for album in albums]
 308             if albums:
 309                 self.log.debug('Albums candidate: {0:s}'.format(
 310                                ' / '.join(albums)))
 311             else: continue
 312             # albums yet in history for this artist
 313             albums = set(albums)
 314             albums_yet_in_hist = albums & self._get_album_history(artist=artist)
 315             albums_not_in_hist = list(albums - albums_yet_in_hist)
 316             # Get to next artist if there are no unplayed albums
 317             if not albums_not_in_hist:
 318                 self.log.info('No album found for "%s"' % artist)
 319                 continue
 320             album_to_queue = str()
 321             random.shuffle(albums_not_in_hist)
 322             for album in albums_not_in_hist:
 323                 tracks = self.player.find_album(artist, album)
 324                 # Look if one track of the album is already queued
 325                 # Good heuristic, at least enough to guess if the whole album is
 326                 # already queued.
 327                 if tracks[0] in self.player.queue:
 328                     self.log.debug('"%s" already queued, skipping!' %
 329                             tracks[0].album)
 330                     continue
 331                 album_to_queue = album
 332             if not album_to_queue:
 333                 self.log.info('No album found for "%s"' % artist)
 334                 continue
 335             self.log.info('{2} album candidate: {0} - {1}'.format(
 336                            artist, album_to_queue, self.ws.name))
 337             nb_album_add += 1
 338             self.to_add.extend(self.player.find_album(artist, album_to_queue))
 339             if nb_album_add == target_album_to_add:
 340                 return True
 341
 342     def find_top(self, artists):
 343         """
 344         find top tracks for artists in artists list.
 345         """
 346         self.to_add = list()
 347         nbtracks_target = self.plugin_conf.getint('track_to_add')
 348         for artist in artists:
 349             if len(self.to_add) == nbtracks_target:
 350                 return True
 351             self.log.info('Looking for a top track for {0}'.format(artist))
 352             titles = deque()
 353             try:
 354                 titles = [t for t in self.ws.get_toptrack(artist)]
 355             except WSError as err:
 356                 self.log.warning('{0}: {1}'.format(self.ws.name, err))
 357             for trk in titles:
 358                 found = self.player.fuzzy_find_track(artist, trk.title)
 359                 random.shuffle(found)
 360                 if found:
 361                     self.log.debug('{0}'.format(found[0]))
 362                     if self.filter_track(found):
 363                         break
 364
 365     def _track(self):
 366         """Get some tracks for track queue mode
 367         """
 368         artists = self.get_local_similar_artists()
 369         nbtracks_target = self.plugin_conf.getint('track_to_add')
 370         for artist in artists:
 371             self.log.debug('Trying to find titles to add for "{}"'.format(
 372                            artist))
 373             found = self.player.find_track(artist)
 374             random.shuffle(found)
 375             if not found:
 376                 self.log.debug('Found nothing to queue for {0}'.format(artist))
 377                 continue
 378             # find tracks not in history for artist
 379             self.filter_track(found)
 380             if len(self.to_add) == nbtracks_target:
 381                 break
 382         if not self.to_add:
 383             self.log.debug('Found no tracks to queue!')
 384             return None
 385         for track in self.to_add:
 386             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 387
 388     def _album(self):
 389         """Get albums for album queue mode
 390         """
 391         artists = self.get_local_similar_artists()
 392         self.find_album(artists)
 393
 394     def _top(self):
 395         """Get some tracks for top track queue mode
 396         """
 397         artists = self.get_local_similar_artists()
 398         self.find_top(artists)
 399         for track in self.to_add:
 400             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 401
 402     def callback_need_track(self):
 403         self._cleanup_cache()
 404         if len(self.player.playlist) == 0:
 405             self.log.info('No last track, cannot queue')
 406             return None
 407         if not self.player.playlist[-1].artist:
 408             self.log.warning('No artist set for the last track in queue')
 409             self.log.debug(repr(self.player.current))
 410             return None
 411         self.queue_mode()
 412         msg = ' '.join(['{0}: {1:>3d}'.format(k, v) for
 413                         k, v in sorted(self.ws.stats.items())])
 414         self.log.debug('http stats: ' + msg)
 415         candidates = self.to_add
 416         self.to_add = list()
 417         if self.plugin_conf.get('queue_mode') != 'album':
 418             random.shuffle(candidates)
 419         return candidates
 420
 421     def callback_player_database(self):
 422         self._flush_cache()
 423
 424 # VIM MODLINE
 425 # vim: ai ts=4 sw=4 sts=4 expandtab