sima/lib/webserv.py

   1 # -*- coding: utf-8 -*-
   2 # Copyright (c) 2009-2014 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
  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         history = deque(self.history)
 193         history.popleft()
 194         depth = 0
 195         if not self.player.playlist:
 196             return
 197         last_trk = self.player.playlist[-1]
 198         extra_arts = list()
 199         ret_extra = list()
 200         while depth < self.plugin_conf.getint('depth'):
 201             if len(history) == 0:
 202                 break
 203             trk = history.popleft()
 204             if (trk.Artist in extra_arts
 205                 or trk.Artist == last_trk.Artist):
 206                 continue
 207             extra_arts.append(trk.Artist)
 208             depth += 1
 209         self.log.info('EXTRA ARTS: {}'.format(
 210                       '/'.join(map(str, extra_arts))))
 211         for artist in extra_arts:
 212             self.log.debug('Looking for artist similar '
 213                            'to "{}" as well'.format(artist))
 214             similar = self.ws_similar_artists(artist=artist)
 215             if not similar:
 216                 return []
 217             ret_extra = self.get_artists_from_player(similar)
 218             if last_trk.Artist in ret_extra:
 219                 ret_extra.remove(last_trk.Artist)
 220         return ret_extra
 221
 222     def get_local_similar_artists(self):
 223         """Check against local player for similar artists
 224         """
 225         if not self.player.playlist:
 226             return []
 227         tolookfor = self.player.playlist[-1].Artist
 228         self.log.info('Looking for artist similar to "{}"'.format(tolookfor))
 229         self.log.debug(repr(tolookfor))
 230         similar = self.ws_similar_artists(tolookfor)
 231         if not similar:
 232             self.log.info('Got nothing from {0}!'.format(self.ws.name))
 233             return []
 234         self.log.info('First five similar artist(s): {}...'.format(
 235                       ' / '.join(map(str, list(similar)[:5]))))
 236         self.log.info('Looking availability in music library')
 237         ret = set(self.get_artists_from_player(similar))
 238         ret_extra = None
 239         if len(self.history) >= 2:
 240             if self.plugin_conf.getint('depth') > 1:
 241                 ret_extra = self.get_recursive_similar_artist()
 242         if ret_extra:
 243             # get them reorg to pick up best element
 244             ret_extra = self._get_artists_list_reorg(ret_extra)
 245             # pickup half the number of ret artist
 246             ret_extra = set(ret_extra[:len(ret)//2])
 247             self.log.debug('Using extra: {}'.format(
 248                            ' / '.join(map(str, ret_extra))))
 249             ret = ret | ret_extra
 250         if not ret:
 251             self.log.warning('Got nothing from music library.')
 252             self.log.warning('Try running in debug mode to guess why...')
 253             return []
 254         # WARNING:
 255         #   * operation on set will not match against aliases
 256         #   * composite set w/ mbid set and whitout won't match either
 257         queued_artists = {trk.Artist for trk in self.player.queue}
 258         if ret & queued_artists:
 259             self.log.debug('Removing already queued artists: '
 260                            '{0}'.format(ret & queued_artists))
 261             ret = ret - queued_artists
 262         if self.player.current and self.player.current.Artist in ret:
 263             self.log.debug('Removing current artist: {0}'.format(self.player.current.Artist))
 264             ret = ret - {self.player.current.Artist}
 265         # Move around similars items to get in unplayed|not recently played
 266         # artist first.
 267         self.log.info('Got {} artists in library'.format(len(ret)))
 268         candidates = self._get_artists_list_reorg(list(ret))
 269         self.log.info(' / '.join(map(str, candidates)))
 270         return candidates
 271
 272     def _get_album_history(self, artist=None):
 273         """Retrieve album history"""
 274         duration = self.daemon_conf.getint('sima', 'history_duration')
 275         albums_list = set()
 276         for trk in self.sdb.get_history(artist=artist.name, duration=duration):
 277             albums_list.add(trk[1])
 278         return albums_list
 279
 280     def find_album(self, artists):
 281         """Find albums to queue.
 282         """
 283         self.to_add = list()
 284         nb_album_add = 0
 285         target_album_to_add = self.plugin_conf.getint('album_to_add')
 286         for artist in artists:
 287             self.log.info('Looking for an album to add for "%s"...' % artist)
 288             albums = self.player.search_albums(artist)
 289             # str conversion while Album type is not propagated
 290             albums = [str(album) for album in albums]
 291             if albums:
 292                 self.log.debug('Albums candidate: {0:s}'.format(
 293                                ' / '.join(albums)))
 294             else: continue
 295             # albums yet in history for this artist
 296             albums = set(albums)
 297             albums_yet_in_hist = albums & self._get_album_history(artist=artist)
 298             albums_not_in_hist = list(albums - albums_yet_in_hist)
 299             # Get to next artist if there are no unplayed albums
 300             if not albums_not_in_hist:
 301                 self.log.info('No album found for "%s"' % artist)
 302                 continue
 303             album_to_queue = str()
 304             random.shuffle(albums_not_in_hist)
 305             for album in albums_not_in_hist:
 306                 tracks = self.player.find_album(artist, album)
 307                 # Look if one track of the album is already queued
 308                 # Good heuristic, at least enough to guess if the whole album is
 309                 # already queued.
 310                 if tracks[0] in self.player.queue:
 311                     self.log.debug('"%s" already queued, skipping!' %
 312                             tracks[0].album)
 313                     continue
 314                 album_to_queue = album
 315             if not album_to_queue:
 316                 self.log.info('No album found for "%s"' % artist)
 317                 continue
 318             self.log.info('{2} album candidate: {0} - {1}'.format(
 319                            artist, album_to_queue, self.ws.name))
 320             nb_album_add += 1
 321             self.to_add.extend(self.player.find_album(artist, album_to_queue))
 322             if nb_album_add == target_album_to_add:
 323                 return True
 324
 325     def find_top(self, artists):
 326         """
 327         find top tracks for artists in artists list.
 328         """
 329         self.to_add = list()
 330         nbtracks_target = self.plugin_conf.getint('track_to_add')
 331         for artist in artists:
 332             artist = Artist(name=artist)
 333             if len(self.to_add) == nbtracks_target:
 334                 return True
 335             self.log.info('Looking for a top track for {0}'.format(artist))
 336             titles = deque()
 337             try:
 338                 titles = [t for t in self.ws.get_toptrack(artist)]
 339             except WSError as err:
 340                 self.log.warning('{0}: {1}'.format(self.ws.name, err))
 341             for trk in titles:
 342                 found = self.player.fuzzy_find_track(artist, trk.title)
 343                 random.shuffle(found)
 344                 if found:
 345                     self.log.debug('{0}'.format(found[0]))
 346                     if self.filter_track(found):
 347                         break
 348
 349     def _track(self):
 350         """Get some tracks for track queue mode
 351         """
 352         artists = self.get_local_similar_artists()
 353         nbtracks_target = self.plugin_conf.getint('track_to_add')
 354         for artist in artists:
 355             self.log.debug('Trying to find titles to add for "{}"'.format(
 356                            artist))
 357             found = self.player.find_track(artist)
 358             random.shuffle(found)
 359             if not found:
 360                 self.log.debug('Found nothing to queue for {0}'.format(artist))
 361                 continue
 362             # find tracks not in history for artist
 363             self.filter_track(found)
 364             if len(self.to_add) == nbtracks_target:
 365                 break
 366         if not self.to_add:
 367             self.log.debug('Found no tracks to queue!')
 368             return None
 369         for track in self.to_add:
 370             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 371
 372     def _album(self):
 373         """Get albums for album queue mode
 374         """
 375         artists = self.get_local_similar_artists()
 376         self.find_album(artists)
 377
 378     def _top(self):
 379         """Get some tracks for top track queue mode
 380         """
 381         artists = self.get_local_similar_artists()
 382         self.find_top(artists)
 383         for track in self.to_add:
 384             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 385
 386     def callback_need_track(self):
 387         self._cleanup_cache()
 388         if len(self.player.playlist) == 0:
 389             self.log.info('No last track, cannot queue')
 390             return None
 391         if not self.player.playlist[-1].artist:
 392             self.log.warning('No artist set for the last track in queue')
 393             self.log.debug(repr(self.player.current))
 394             return None
 395         self.queue_mode()
 396         msg = ' '.join(['{0}: {1:>3d}'.format(k, v) for
 397                         k, v in sorted(self.ws.stats.items())])
 398         self.log.debug('http stats: ' + msg)
 399         candidates = self.to_add
 400         self.to_add = list()
 401         if self.plugin_conf.get('queue_mode') != 'album':
 402             random.shuffle(candidates)
 403         return candidates
 404
 405     def callback_player_database(self):
 406         self._flush_cache()
 407
 408 # VIM MODLINE
 409 # vim: ai ts=4 sw=4 sts=4 expandtab