]> kaliko git repositories - mpd-sima.git/blob - sima/lib/webserv.py
Fixed http caching for Lastfm.
[mpd-sima.git] / 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         if ret & queued_artists:
273             self.log.debug('Removing already queued artists: '
274                            '{0}'.format('/'.join(map(str, ret & queued_artists))))
275             ret = ret - queued_artists
276         if self.player.current and self.player.current.Artist in ret:
277             self.log.debug('Removing current artist: {0}'.format(self.player.current.Artist))
278             ret = ret -  MetaContainer([self.player.current.Artist])
279         # Move around similars items to get in unplayed|not recently played
280         # artist first.
281         self.log.info('Got {} artists in library'.format(len(ret)))
282         candidates = self._get_artists_list_reorg(list(ret))
283         if candidates:
284             self.log.info(' / '.join(map(str, candidates)))
285         return candidates
286
287     def _get_album_history(self, artist=None):
288         """Retrieve album history"""
289         duration = self.daemon_conf.getint('sima', 'history_duration')
290         albums_list = set()
291         for trk in self.sdb.get_history(artist=artist.name, duration=duration):
292             albums_list.add(trk[1])
293         return albums_list
294
295     def find_album(self, artists):
296         """Find albums to queue.
297         """
298         self.to_add = list()
299         nb_album_add = 0
300         target_album_to_add = self.plugin_conf.getint('album_to_add')
301         for artist in artists:
302             self.log.info('Looking for an album to add for "%s"...' % artist)
303             albums = self.player.search_albums(artist)
304             # str conversion while Album type is not propagated
305             albums = [str(album) for album in albums]
306             if albums:
307                 self.log.debug('Albums candidate: {0:s}'.format(
308                                ' / '.join(albums)))
309             else: continue
310             # albums yet in history for this artist
311             albums = set(albums)
312             albums_yet_in_hist = albums & self._get_album_history(artist=artist)
313             albums_not_in_hist = list(albums - albums_yet_in_hist)
314             # Get to next artist if there are no unplayed albums
315             if not albums_not_in_hist:
316                 self.log.info('No album found for "%s"' % artist)
317                 continue
318             album_to_queue = str()
319             random.shuffle(albums_not_in_hist)
320             for album in albums_not_in_hist:
321                 tracks = self.player.find_album(artist, album)
322                 # Look if one track of the album is already queued
323                 # Good heuristic, at least enough to guess if the whole album is
324                 # already queued.
325                 if tracks[0] in self.player.queue:
326                     self.log.debug('"%s" already queued, skipping!' %
327                             tracks[0].album)
328                     continue
329                 album_to_queue = album
330             if not album_to_queue:
331                 self.log.info('No album found for "%s"' % artist)
332                 continue
333             self.log.info('{2} album candidate: {0} - {1}'.format(
334                            artist, album_to_queue, self.ws.name))
335             nb_album_add += 1
336             self.to_add.extend(self.player.find_album(artist, album_to_queue))
337             if nb_album_add == target_album_to_add:
338                 return True
339
340     def find_top(self, artists):
341         """
342         find top tracks for artists in artists list.
343         """
344         self.to_add = list()
345         nbtracks_target = self.plugin_conf.getint('track_to_add')
346         for artist in artists:
347             if len(self.to_add) == nbtracks_target:
348                 return True
349             self.log.info('Looking for a top track for {0}'.format(artist))
350             titles = deque()
351             try:
352                 titles = [t for t in self.ws.get_toptrack(artist)]
353             except WSError as err:
354                 self.log.warning('{0}: {1}'.format(self.ws.name, err))
355             for trk in titles:
356                 found = self.player.fuzzy_find_track(artist, trk.title)
357                 random.shuffle(found)
358                 if found:
359                     self.log.debug('{0}'.format(found[0]))
360                     if self.filter_track(found):
361                         break
362
363     def _track(self):
364         """Get some tracks for track queue mode
365         """
366         artists = self.get_local_similar_artists()
367         nbtracks_target = self.plugin_conf.getint('track_to_add')
368         for artist in artists:
369             self.log.debug('Trying to find titles to add for "{}"'.format(
370                            artist))
371             found = self.player.find_track(artist)
372             random.shuffle(found)
373             if not found:
374                 self.log.debug('Found nothing to queue for {0}'.format(artist))
375                 continue
376             # find tracks not in history for artist
377             self.filter_track(found)
378             if len(self.to_add) == nbtracks_target:
379                 break
380         if not self.to_add:
381             self.log.debug('Found no tracks to queue!')
382             return None
383         for track in self.to_add:
384             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
385
386     def _album(self):
387         """Get albums for album queue mode
388         """
389         artists = self.get_local_similar_artists()
390         self.find_album(artists)
391
392     def _top(self):
393         """Get some tracks for top track queue mode
394         """
395         artists = self.get_local_similar_artists()
396         self.find_top(artists)
397         for track in self.to_add:
398             self.log.info('{1} candidates: {0!s}'.format(track, self.ws.name))
399
400     def callback_need_track(self):
401         self._cleanup_cache()
402         if len(self.player.playlist) == 0:
403             self.log.info('No last track, cannot queue')
404             return None
405         if not self.player.playlist[-1].artist:
406             self.log.warning('No artist set for the last track in queue')
407             self.log.debug(repr(self.player.current))
408             return None
409         self.queue_mode()
410         msg = ' '.join(['{0}: {1:>3d}'.format(k, v) for
411                         k, v in sorted(self.ws.stats.items())])
412         self.log.debug('http stats: ' + msg)
413         candidates = self.to_add
414         self.to_add = list()
415         if self.plugin_conf.get('queue_mode') != 'album':
416             random.shuffle(candidates)
417         return candidates
418
419     def callback_player_database(self):
420         self._flush_cache()
421
422 # VIM MODLINE
423 # vim: ai ts=4 sw=4 sts=4 expandtab