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
  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 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         # TODO: move to utils as a decorator
 144         duration = self.daemon_conf.getint('sima', 'history_duration')
 145         art_in_hist = list()
 146         for trk in self.sdb.get_history(duration=duration,
 147                                         artists=alist):
 148             if trk[0] not in art_in_hist:
 149                 art_in_hist.append(trk[0])
 150         art_in_hist.reverse()
 151         art_not_in_hist = [ar for ar in alist if ar not in art_in_hist]
 152         random.shuffle(art_not_in_hist)
 153         art_not_in_hist.extend(art_in_hist)
 154         self.log.info('{}'.format(
 155                        ' / '.join(art_not_in_hist)))
 156         return art_not_in_hist
 157
 158     @cache
 159     def get_artists_from_player(self, similarities):
 160         """
 161         Look in player library for availability of similar artists in
 162         similarities
 163         """
 164         dynamic = self.plugin_conf.getint('max_art')
 165         if dynamic <= 0:
 166             dynamic = 100
 167         results = list()
 168         similarities.reverse()
 169         while (len(results) < dynamic
 170             and len(similarities) > 0):
 171             art_pop = similarities.pop()
 172             results.extend(self.player.fuzzy_find_artist(art_pop))
 173         return results
 174
 175     def ws_similar_artists(self, artist=None):
 176         """
 177         Retrieve similar artists from WebServive.
 178         """
 179         if artist is None:
 180             curr = self.player.current.__dict__
 181             name = curr.get('artist')
 182             mbid = curr.get('musicbrainz_artistid', None)
 183             current = Artist(name=name, mbid=mbid)
 184         else:
 185             current = artist
 186         # initialize artists deque list to construct from DB
 187         as_art = deque()
 188         as_artists = self.ws().get_similar(artist=current)
 189         self.log.debug('Requesting {1} for "{0}"'.format(current,
 190                         self.ws.name))
 191         try:
 192             # TODO: let's propagate Artist type
 193             [as_art.append(str(art)) for art in as_artists]
 194         except WSError as err:
 195             self.log.warning('{0}: {1}'.format(self.ws.name, err))
 196         if as_art:
 197             self.log.debug('Fetched {0} artist(s)'.format(len(as_art)))
 198         return as_art
 199
 200     def get_recursive_similar_artist(self):
 201         ret_extra = list()
 202         history = deque(self.history)
 203         history.popleft()
 204         depth = 0
 205         current = self.player.current
 206         extra_arts = list()
 207         while depth < self.plugin_conf.getint('depth'):
 208             if len(history) == 0:
 209                 break
 210             trk = history.popleft()
 211             if (trk.get_artist() in extra_arts
 212                 or trk.get_artist() == current.get_artist()):
 213                 continue
 214             extra_arts.append(trk.get_artist())
 215             depth += 1
 216         self.log.info('EXTRA ARTS: {}'.format(
 217             '/'.join([art.name for art in extra_arts])))
 218         for artist in extra_arts:
 219             self.log.debug('Looking for artist similar '
 220                            'to "{}" as well'.format(artist))
 221             similar = self.ws_similar_artists(artist=artist)
 222             if not similar:
 223                 return ret_extra
 224             ret_extra.extend(self.get_artists_from_player(similar))
 225             if current.artist in ret_extra:
 226                 ret_extra.remove(current.artist)
 227         return ret_extra
 228
 229     def get_local_similar_artists(self):
 230         """Check against local player for similar artists
 231         """
 232         current = self.player.current
 233         self.log.info('Looking for artist similar to "{0.artist}"'.format(current))
 234         similar = self.ws_similar_artists()
 235         if not similar:
 236             self.log.info('Got nothing from {0}!'.format(self.ws.name))
 237             return []
 238         self.log.info('First five similar artist(s): {}...'.format(
 239                       ' / '.join([a for a in list(similar)[0:5]])))
 240         self.log.info('Looking availability in music library')
 241         ret = self.get_artists_from_player(similar)
 242         ret_extra = None
 243         if len(self.history) >= 2:
 244             if self.plugin_conf.getint('depth') > 1:
 245                 ret_extra = self.get_recursive_similar_artist()
 246         if ret_extra:
 247             ret = list(set(ret) | set(ret_extra))
 248         if not ret:
 249             self.log.warning('Got nothing from music library.')
 250             self.log.warning('Try running in debug mode to guess why...')
 251             return []
 252         self.log.info('Got {} artists in library'.format(len(ret)))
 253         # Move around similars items to get in unplayed|not recently played
 254         # artist first.
 255         return self._get_artists_list_reorg(ret)
 256
 257     def _get_album_history(self, artist=None):
 258         """Retrieve album history"""
 259         duration = self.daemon_conf.getint('sima', 'history_duration')
 260         albums_list = set()
 261         for trk in self.sdb.get_history(artist=artist, duration=duration):
 262             albums_list.add(trk[1])
 263         return albums_list
 264
 265     def find_album(self, artists):
 266         """Find albums to queue.
 267         """
 268         self.to_add = list()
 269         nb_album_add = 0
 270         target_album_to_add = self.plugin_conf.getint('album_to_add')
 271         for artist in artists:
 272             self.log.info('Looking for an album to add for "%s"...' % artist)
 273             albums = self.player.find_albums(artist)
 274             # str conversion while Album type is not propagated
 275             albums = [str(album) for album in albums]
 276             if albums:
 277                 self.log.debug('Albums candidate: {0:s}'.format(
 278                                ' / '.join(albums)))
 279             else: continue
 280             # albums yet in history for this artist
 281             albums = set(albums)
 282             albums_yet_in_hist = albums & self._get_album_history(artist=artist)
 283             albums_not_in_hist = list(albums - albums_yet_in_hist)
 284             # Get to next artist if there are no unplayed albums
 285             if not albums_not_in_hist:
 286                 self.log.info('No album found for "%s"' % artist)
 287                 continue
 288             album_to_queue = str()
 289             random.shuffle(albums_not_in_hist)
 290             for album in albums_not_in_hist:
 291                 tracks = self.player.find_album(artist, album)
 292                 # Look if one track of the album is already queued
 293                 # Good heuristic, at least enough to guess if the whole album is
 294                 # already queued.
 295                 if tracks[0] in self.player.queue:
 296                     self.log.debug('"%s" already queued, skipping!' %
 297                             tracks[0].album)
 298                     continue
 299                 album_to_queue = album
 300             if not album_to_queue:
 301                 self.log.info('No album found for "%s"' % artist)
 302                 continue
 303             self.log.info('{2} album candidate: {0} - {1}'.format(
 304                            artist, album_to_queue, self.ws.name))
 305             nb_album_add += 1
 306             self.to_add.extend(self.player.find_album(artist, album_to_queue))
 307             if nb_album_add == target_album_to_add:
 308                 return True
 309
 310     def find_top(self, artists):
 311         """
 312         find top tracks for artists in artists list.
 313         """
 314         self.to_add = list()
 315         nbtracks_target = self.plugin_conf.getint('track_to_add')
 316         webserv = self.ws()
 317         for artist in artists:
 318             artist = Artist(name=artist)
 319             if len(self.to_add) == nbtracks_target:
 320                 return True
 321             self.log.info('Looking for a top track for {0}'.format(artist))
 322             titles = deque()
 323             try:
 324                 titles = [t for t in webserv.get_toptrack(artist)]
 325             except WSError as err:
 326                 self.log.warning('{0}: {1}'.format(self.ws.name, err))
 327             if self.ws.ratelimit:
 328                 self.log.info('{0.name} ratelimit: {0.ratelimit}'.format(self.ws))
 329             for trk in titles:
 330                 found = self.player.fuzzy_find_track(artist.name, trk.title)
 331                 if found:
 332                     self.log.debug('{0}'.format(found[0]))
 333                     if self.filter_track(found):
 334                         break
 335
 336     def _track(self):
 337         """Get some tracks for track queue mode
 338         """
 339         artists = self.get_local_similar_artists()
 340         nbtracks_target = self.plugin_conf.getint('track_to_add')
 341         for artist in artists:
 342             self.log.debug('Trying to find titles to add for "{}"'.format(
 343                            artist))
 344             found = self.player.find_track(artist)
 345             if not found:
 346                 self.log.debug('Found nothing to queue for {0}'.format(artist))
 347                 continue
 348             # find tracks not in history for artist
 349             self.filter_track(found)
 350             if len(self.to_add) == nbtracks_target:
 351                 break
 352         if not self.to_add:
 353             self.log.debug('Found no tracks to queue, is your ' +
 354                             'history getting too large?')
 355             return None
 356         for track in self.to_add:
 357             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 358
 359     def _album(self):
 360         """Get albums for album queue mode
 361         """
 362         artists = self.get_local_similar_artists()
 363         self.find_album(artists)
 364
 365     def _top(self):
 366         """Get some tracks for top track queue mode
 367         """
 368         artists = self.get_local_similar_artists()
 369         self.find_top(artists)
 370         for track in self.to_add:
 371             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
 372
 373     def callback_need_track(self):
 374         self._cleanup_cache()
 375         if not self.player.current:
 376             self.log.info('No current track, cannot queue')
 377             return None
 378         if not self.player.current.artist:
 379             self.log.warning('No artist set for the current track')
 380             self.log.debug(repr(self.player.current))
 381             return None
 382         self.queue_mode()
 383         msg = ' '.join(['{0}: {1:>3d}'.format(k, v) for
 384                         k, v in sorted(self.ws.stats.items())])
 385         self.log.debug(msg)
 386         candidates = self.to_add
 387         self.to_add = list()
 388         if self.plugin_conf.get('queue_mode') != 'album':
 389             random.shuffle(candidates)
 390         return candidates
 391
 392     def callback_player_database(self):
 393         self._flush_cache()
 394
 395 # VIM MODLINE
 396 # vim: ai ts=4 sw=4 sts=4 expandtab