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