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=None):
 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             if artist.mbid:
 183                 return self.ws_similar_artists(Artist(name=artist.name))
 184             self.log.warning('{}: {}'.format(self.ws.name, err))
 185         except WSError as err:
 186             self.log.warning('{}: {}'.format(self.ws.name, err))
 187         if as_art:
 188             self.log.debug('Fetched {} artist(s)'.format(len(as_art)))
 189         return as_art
 190
 191     def get_recursive_similar_artist(self):
 192         if not self.player.playlist:
 193             return
 194         history = list(self.history)
 195         history = self.player.queue + history
 196         history = deque(history)
 197         last_trk = history.popleft() # remove
 198         extra_arts = list()
 199         ret_extra = list()
 200         depth = 0
 201         while depth < self.plugin_conf.getint('depth'):
 202             if len(history) == 0:
 203                 break
 204             trk = history.popleft()
 205             if (trk.Artist in extra_arts
 206                     or trk.Artist == last_trk.Artist):
 207                 continue
 208             extra_arts.append(trk.Artist)
 209             depth += 1
 210         self.log.debug('EXTRA ARTS: {}'.format(
 211                       '/'.join(map(str, extra_arts))))
 212         for artist in extra_arts:
 213             self.log.debug('Looking for artist similar '
 214                            'to "{}" as well'.format(artist))
 215             similar = self.ws_similar_artists(artist=artist)
 216             if not similar:
 217                 continue
 218             ret_extra.extend(self.get_artists_from_player(similar))
 219
 220         if last_trk.Artist in ret_extra:
 221             ret_extra.remove(last_trk.Artist)
 222         if ret_extra:
 223             self.log.debug('similar artist(s) found: {}'.format(
 224                 ' / '.join(map(str, MetaContainer(ret_extra)))))
 225         return ret_extra
 226
 227     def get_local_similar_artists(self):
 228         """Check against local player for similar artists
 229         """
 230         if not self.player.playlist:
 231             return []
 232         tolookfor = self.player.playlist[-1].Artist
 233         self.log.info('Looking for artist similar to "{}"'.format(tolookfor))
 234         self.log.debug(repr(tolookfor))
 235         similar = self.ws_similar_artists(tolookfor)
 236         if not similar:
 237             self.log.info('Got nothing from {0}!'.format(self.ws.name))
 238             return []
 239         self.log.info('First five similar artist(s): {}...'.format(
 240                       ' / '.join(map(str, list(similar)[:5]))))
 241         self.log.info('Looking availability in music library')
 242         ret = MetaContainer(self.get_artists_from_player(similar))
 243         if ret:
 244             self.log.debug('regular found in library: {}'.format(
 245                            ' / '.join(map(str, ret))))
 246         else:
 247             self.log.debug('Got nothing similar from library!')
 248         ret_extra = None
 249         if len(self.history) >= 2:
 250             if self.plugin_conf.getint('depth') > 1:
 251                 ret_extra = self.get_recursive_similar_artist()
 252         if ret_extra:
 253             # get them reorg to pick up best element
 254             ret_extra = self._get_artists_list_reorg(ret_extra)
 255             # tries to pickup less artist from extra art
 256             if len(ret) < 4:
 257                 ret_extra = MetaContainer(ret_extra)
 258             else:
 259                 ret_extra = MetaContainer(ret_extra[:max(4, len(ret))//2])
 260             if ret_extra:
 261                 self.log.debug('extra found in library: {}'.format(
 262                                ' / '.join(map(str, ret_extra))))
 263             ret = ret | ret_extra
 264         if not ret:
 265             self.log.warning('Got nothing from music library.')
 266             return []
 267         # WARNING:
 268         #   * operation on set will not match against aliases
 269         #   * composite set w/ mbid set and whitout won't match either
 270         queued_artists = MetaContainer([trk.Artist for trk in self.player.queue])
 271         if ret & queued_artists:
 272             self.log.debug('Removing already queued artists: '
 273                            '{0}'.format('/'.join(map(str, ret & queued_artists))))
 274             ret = ret - queued_artists
 275         if self.player.current and self.player.current.Artist in ret:
 276             self.log.debug('Removing current artist: {0}'.format(self.player.current.Artist))
 277             ret = ret -  MetaContainer([self.player.current.Artist])
 278         # Move around similars items to get in unplayed|not recently played
 279         # artist first.
 280         self.log.info('Got {} artists in library'.format(len(ret)))
 281         candidates = self._get_artists_list_reorg(list(ret))
 282         if candidates:
 283             self.log.info(' / '.join(map(str, candidates)))
 284         return candidates
 285
 286     def _get_album_history(self, artist=None):
 287         """Retrieve album history"""
 288         duration = self.daemon_conf.getint('sima', 'history_duration')
 289         albums_list = set()
 290         for trk in self.sdb.get_history(artist=artist.name, duration=duration):
 291             albums_list.add(trk[1])
 292         return albums_list
 293
 294     def find_album(self, artists):
 295         """Find albums to queue.
 296         """
 297         self.to_add = list()
 298         nb_album_add = 0
 299         target_album_to_add = self.plugin_conf.getint('album_to_add')
 300         for artist in artists:
 301             self.log.info('Looking for an album to add for "%s"...' % artist)
 302             albums = self.player.search_albums(artist)
 303             # str conversion while Album type is not propagated
 304             albums = [str(album) for album in albums]
 305             if albums:
 306                 self.log.debug('Albums candidate: {0:s}'.format(
 307                                ' / '.join(albums)))
 308             else: continue
 309             # albums yet in history for this artist
 310             albums = set(albums)
 311             albums_yet_in_hist = albums & self._get_album_history(artist=artist)
 312             albums_not_in_hist = list(albums - albums_yet_in_hist)
 313             # Get to next artist if there are no unplayed albums
 314             if not albums_not_in_hist:
 315                 self.log.info('No album found for "%s"' % artist)
 316                 continue
 317             album_to_queue = str()
 318             random.shuffle(albums_not_in_hist)
 319             for album in albums_not_in_hist:
 320                 tracks = self.player.find_album(artist, album)
 321                 # Look if one track of the album is already queued
 322                 # Good heuristic, at least enough to guess if the whole album is
 323                 # already queued.
 324                 if tracks[0] in self.player.queue:
 325                     self.log.debug('"%s" already queued, skipping!' %
 326                             tracks[0].album)
 327                     continue
 328                 album_to_queue = album
 329             if not album_to_queue:
 330                 self.log.info('No album found for "%s"' % artist)
 331                 continue
 332             self.log.info('{2} album candidate: {0} - {1}'.format(
 333                            artist, album_to_queue, self.ws.name))
 334             nb_album_add += 1
 335             self.to_add.extend(self.player.find_album(artist, album_to_queue))
 336             if nb_album_add == target_album_to_add:
 337                 return True
 338
 339     def find_top(self, artists):
 340         """
 341         find top tracks for artists in artists list.
 342         """
 343         self.to_add = list()
 344         nbtracks_target = self.plugin_conf.getint('track_to_add')
 345         for artist in artists:
 346             if len(self.to_add) == nbtracks_target:
 347                 return True
 348             self.log.info('Looking for a top track for {0}'.format(artist))
 349             titles = deque()
 350             try:
 351                 titles = [t for t in self.ws.get_toptrack(artist)]
 352             except WSError as err:
 353                 self.log.warning('{0}: {1}'.format(self.ws.name, err))
 354             for trk in titles:
 355                 found = self.player.fuzzy_find_track(artist, trk.title)
 356                 random.shuffle(found)
 357                 if found:
 358                     self.log.debug('{0}'.format(found[0]))
 359                     if self.filter_track(found):
 360                         break
 361
 362     def _track(self):
 363         """Get some tracks for track queue mode
 364         """
 365         artists = self.get_local_similar_artists()
 366         nbtracks_target = self.plugin_conf.getint('track_to_add')
 367         for artist in artists:
 368             self.log.debug('Trying to find titles to add for "{}"'.format(
 369                            artist))
 370             found = self.player.find_track(artist)
 371             random.shuffle(found)
 372             if not found:
 373                 self.log.debug('Found nothing to queue for {0}'.format(artist))
 374                 continue
 375             # find tracks not in history for artist
 376             self.filter_track(found)
 377             if len(self.to_add) == nbtracks_target:
 378                 break
 379         if not self.to_add:
 380             self.log.debug('Found no tracks to queue!')
 381             return None
 382         for track in self.to_add:
 383             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 384
 385     def _album(self):
 386         """Get albums for album queue mode
 387         """
 388         artists = self.get_local_similar_artists()
 389         self.find_album(artists)
 390
 391     def _top(self):
 392         """Get some tracks for top track queue mode
 393         """
 394         artists = self.get_local_similar_artists()
 395         self.find_top(artists)
 396         for track in self.to_add:
 397             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 398
 399     def callback_need_track(self):
 400         self._cleanup_cache()
 401         if len(self.player.playlist) == 0:
 402             self.log.info('No last track, cannot queue')
 403             return None
 404         if not self.player.playlist[-1].artist:
 405             self.log.warning('No artist set for the last track in queue')
 406             self.log.debug(repr(self.player.current))
 407             return None
 408         self.queue_mode()
 409         msg = ' '.join(['{0}: {1:>3d}'.format(k, v) for
 410                         k, v in sorted(self.ws.stats.items())])
 411         self.log.debug('http stats: ' + msg)
 412         candidates = self.to_add
 413         self.to_add = list()
 414         if self.plugin_conf.get('queue_mode') != 'album':
 415             random.shuffle(candidates)
 416         return candidates
 417
 418     def callback_player_database(self):
 419         self._flush_cache()
 420
 421 # VIM MODLINE
 422 # vim: ai ts=4 sw=4 sts=4 expandtab