]>
Commit | Line | Data |
---|---|---|
e7c14660 | 1 | # coding: utf-8 |
4c603938 MA |
2 | from __future__ import unicode_literals |
3 | ||
4 | import re | |
5 | import hashlib | |
4c603938 MA |
6 | |
7 | from .common import InfoExtractor | |
6e6bc8da | 8 | from ..compat import compat_str |
47fe42e1 | 9 | from ..utils import ( |
ae7d31af | 10 | ExtractorError, |
47fe42e1 S |
11 | int_or_none, |
12 | float_or_none, | |
0250161c | 13 | try_get, |
47fe42e1 | 14 | ) |
4c603938 | 15 | |
4c603938 | 16 | |
ae7d31af | 17 | class YandexMusicBaseIE(InfoExtractor): |
29f7c58a | 18 | _VALID_URL_BASE = r'https?://music\.yandex\.(?P<tld>ru|kz|ua|by|com)' |
19 | ||
ae7d31af S |
20 | @staticmethod |
21 | def _handle_error(response): | |
eebe6b38 S |
22 | if isinstance(response, dict): |
23 | error = response.get('error') | |
24 | if error: | |
25 | raise ExtractorError(error, expected=True) | |
197a5da1 S |
26 | if response.get('type') == 'captcha' or 'captcha' in response: |
27 | YandexMusicBaseIE._raise_captcha() | |
28 | ||
29 | @staticmethod | |
30 | def _raise_captcha(): | |
31 | raise ExtractorError( | |
cefecac1 | 32 | 'YandexMusic has considered youtube-dlc requests automated and ' |
197a5da1 S |
33 | 'asks you to solve a CAPTCHA. You can either wait for some ' |
34 | 'time until unblocked and optionally use --sleep-interval ' | |
35 | 'in future or alternatively you can go to https://music.yandex.ru/ ' | |
36 | 'solve CAPTCHA, then export cookies and pass cookie file to ' | |
cefecac1 | 37 | 'youtube-dlc with --cookies', |
197a5da1 | 38 | expected=True) |
ae7d31af | 39 | |
e5eadfa8 S |
40 | def _download_webpage_handle(self, *args, **kwargs): |
41 | webpage = super(YandexMusicBaseIE, self)._download_webpage_handle(*args, **kwargs) | |
7f776fa4 | 42 | if 'Нам очень жаль, но запросы, поступившие с вашего IP-адреса, похожи на автоматические.' in webpage: |
197a5da1 | 43 | self._raise_captcha() |
7f776fa4 YCH |
44 | return webpage |
45 | ||
ae7d31af S |
46 | def _download_json(self, *args, **kwargs): |
47 | response = super(YandexMusicBaseIE, self)._download_json(*args, **kwargs) | |
48 | self._handle_error(response) | |
49 | return response | |
50 | ||
29f7c58a | 51 | def _call_api(self, ep, tld, url, item_id, note, query): |
52 | return self._download_json( | |
53 | 'https://music.yandex.%s/handlers/%s.jsx' % (tld, ep), | |
54 | item_id, note, | |
55 | fatal=False, | |
56 | headers={ | |
57 | 'Referer': url, | |
58 | 'X-Requested-With': 'XMLHttpRequest', | |
59 | 'X-Retpath-Y': url, | |
60 | }, | |
61 | query=query) | |
62 | ||
ae7d31af S |
63 | |
64 | class YandexMusicTrackIE(YandexMusicBaseIE): | |
e4df2f98 S |
65 | IE_NAME = 'yandexmusic:track' |
66 | IE_DESC = 'Яндекс.Музыка - Трек' | |
29f7c58a | 67 | _VALID_URL = r'%s/album/(?P<album_id>\d+)/track/(?P<id>\d+)' % YandexMusicBaseIE._VALID_URL_BASE |
e4df2f98 | 68 | |
c452790a | 69 | _TESTS = [{ |
e4df2f98 | 70 | 'url': 'http://music.yandex.ru/album/540508/track/4878838', |
29f7c58a | 71 | 'md5': 'dec8b661f12027ceaba33318787fff76', |
e4df2f98 S |
72 | 'info_dict': { |
73 | 'id': '4878838', | |
74 | 'ext': 'mp3', | |
29f7c58a | 75 | 'title': 'md5:c63e19341fdbe84e43425a30bc777856', |
76 | 'filesize': int, | |
e4df2f98 | 77 | 'duration': 193.04, |
29f7c58a | 78 | 'track': 'md5:210508c6ffdfd67a493a6c378f22c3ff', |
79 | 'album': 'md5:cd04fb13c4efeafdfa0a6a6aca36d01a', | |
80 | 'album_artist': 'md5:5f54c35462c07952df33d97cfb5fc200', | |
81 | 'artist': 'md5:e6fd86621825f14dc0b25db3acd68160', | |
796bf9de | 82 | 'release_year': 2009, |
7f776fa4 | 83 | }, |
29f7c58a | 84 | # 'skip': 'Travis CI servers blocked by YandexMusic', |
c452790a | 85 | }, { |
86 | # multiple disks | |
87 | 'url': 'http://music.yandex.ru/album/3840501/track/705105', | |
29f7c58a | 88 | 'md5': '82a54e9e787301dd45aba093cf6e58c0', |
c452790a | 89 | 'info_dict': { |
90 | 'id': '705105', | |
91 | 'ext': 'mp3', | |
29f7c58a | 92 | 'title': 'md5:f86d4a9188279860a83000277024c1a6', |
93 | 'filesize': int, | |
c452790a | 94 | 'duration': 239.27, |
29f7c58a | 95 | 'track': 'md5:40f887f0666ba1aa10b835aca44807d1', |
96 | 'album': 'md5:624f5224b14f5c88a8e812fd7fbf1873', | |
97 | 'album_artist': 'md5:dd35f2af4e8927100cbe6f5e62e1fb12', | |
98 | 'artist': 'md5:dd35f2af4e8927100cbe6f5e62e1fb12', | |
c452790a | 99 | 'release_year': 2016, |
100 | 'genre': 'pop', | |
101 | 'disc_number': 2, | |
102 | 'track_number': 9, | |
103 | }, | |
29f7c58a | 104 | # 'skip': 'Travis CI servers blocked by YandexMusic', |
105 | }, { | |
106 | 'url': 'http://music.yandex.com/album/540508/track/4878838', | |
107 | 'only_matching': True, | |
c452790a | 108 | }] |
e4df2f98 | 109 | |
58ef5e78 RA |
110 | def _real_extract(self, url): |
111 | mobj = re.match(self._VALID_URL, url) | |
29f7c58a | 112 | tld, album_id, track_id = mobj.group('tld'), mobj.group('album_id'), mobj.group('id') |
4c603938 | 113 | |
29f7c58a | 114 | track = self._call_api( |
115 | 'track', tld, url, track_id, 'Downloading track JSON', | |
116 | {'track': '%s:%s' % (track_id, album_id)})['track'] | |
58ef5e78 | 117 | track_title = track['title'] |
cf03e34a | 118 | |
58ef5e78 RA |
119 | download_data = self._download_json( |
120 | 'https://music.yandex.ru/api/v2.1/handlers/track/%s:%s/web-album_track-track-track-main/download/m' % (track_id, album_id), | |
121 | track_id, 'Downloading track location url JSON', | |
122 | headers={'X-Retpath-Y': url}) | |
4c603938 | 123 | |
58ef5e78 RA |
124 | fd_data = self._download_json( |
125 | download_data['src'], track_id, | |
126 | 'Downloading track location JSON', | |
127 | query={'format': 'json'}) | |
128 | key = hashlib.md5(('XGRlBW9FXlekgbPrRHuSiA' + fd_data['path'][1:] + fd_data['s']).encode('utf-8')).hexdigest() | |
29f7c58a | 129 | f_url = 'http://%s/get-mp3/%s/%s?track-id=%s ' % (fd_data['host'], key, fd_data['ts'] + fd_data['path'], track['id']) |
4c603938 | 130 | |
ab953c64 S |
131 | thumbnail = None |
132 | cover_uri = track.get('albums', [{}])[0].get('coverUri') | |
133 | if cover_uri: | |
134 | thumbnail = cover_uri.replace('%%', 'orig') | |
135 | if not thumbnail.startswith('http'): | |
136 | thumbnail = 'http://' + thumbnail | |
e90d1754 | 137 | |
e90d1754 | 138 | track_info = { |
58ef5e78 | 139 | 'id': track_id, |
47fe42e1 | 140 | 'ext': 'mp3', |
58ef5e78 | 141 | 'url': f_url, |
47fe42e1 S |
142 | 'filesize': int_or_none(track.get('fileSize')), |
143 | 'duration': float_or_none(track.get('durationMs'), 1000), | |
ab953c64 | 144 | 'thumbnail': thumbnail, |
e90d1754 | 145 | 'track': track_title, |
58ef5e78 RA |
146 | 'acodec': download_data.get('codec'), |
147 | 'abr': int_or_none(download_data.get('bitrate')), | |
47fe42e1 | 148 | } |
4c603938 | 149 | |
c452790a | 150 | def extract_artist_name(artist): |
151 | decomposed = artist.get('decomposed') | |
152 | if not isinstance(decomposed, list): | |
153 | return artist['name'] | |
154 | parts = [artist['name']] | |
155 | for element in decomposed: | |
156 | if isinstance(element, dict) and element.get('name'): | |
157 | parts.append(element['name']) | |
158 | elif isinstance(element, compat_str): | |
159 | parts.append(element) | |
160 | return ''.join(parts) | |
161 | ||
e90d1754 S |
162 | def extract_artist(artist_list): |
163 | if artist_list and isinstance(artist_list, list): | |
c452790a | 164 | artists_names = [extract_artist_name(a) for a in artist_list if a.get('name')] |
e90d1754 S |
165 | if artists_names: |
166 | return ', '.join(artists_names) | |
167 | ||
168 | albums = track.get('albums') | |
169 | if albums and isinstance(albums, list): | |
170 | album = albums[0] | |
171 | if isinstance(album, dict): | |
172 | year = album.get('year') | |
c452790a | 173 | disc_number = int_or_none(try_get( |
174 | album, lambda x: x['trackPosition']['volume'])) | |
175 | track_number = int_or_none(try_get( | |
176 | album, lambda x: x['trackPosition']['index'])) | |
e90d1754 S |
177 | track_info.update({ |
178 | 'album': album.get('title'), | |
179 | 'album_artist': extract_artist(album.get('artists')), | |
796bf9de | 180 | 'release_year': int_or_none(year), |
c452790a | 181 | 'genre': album.get('genre'), |
182 | 'disc_number': disc_number, | |
183 | 'track_number': track_number, | |
e90d1754 S |
184 | }) |
185 | ||
186 | track_artist = extract_artist(track.get('artists')) | |
187 | if track_artist: | |
188 | track_info.update({ | |
189 | 'artist': track_artist, | |
190 | 'title': '%s - %s' % (track_artist, track_title), | |
191 | }) | |
192 | else: | |
193 | track_info['title'] = track_title | |
e90d1754 | 194 | |
58ef5e78 | 195 | return track_info |
4c603938 | 196 | |
4c603938 | 197 | |
ae7d31af | 198 | class YandexMusicPlaylistBaseIE(YandexMusicBaseIE): |
29f7c58a | 199 | def _extract_tracks(self, source, item_id, url, tld): |
200 | tracks = source['tracks'] | |
201 | track_ids = [compat_str(track_id) for track_id in source['trackIds']] | |
202 | ||
203 | # tracks dictionary shipped with playlist.jsx API is limited to 150 tracks, | |
204 | # missing tracks should be retrieved manually. | |
205 | if len(tracks) < len(track_ids): | |
206 | present_track_ids = set([ | |
207 | compat_str(track['id']) | |
208 | for track in tracks if track.get('id')]) | |
209 | missing_track_ids = [ | |
210 | track_id for track_id in track_ids | |
211 | if track_id not in present_track_ids] | |
212 | missing_tracks = self._call_api( | |
213 | 'track-entries', tld, url, item_id, | |
214 | 'Downloading missing tracks JSON', { | |
215 | 'entries': ','.join(missing_track_ids), | |
216 | 'lang': tld, | |
217 | 'external-domain': 'music.yandex.%s' % tld, | |
218 | 'overembed': 'false', | |
219 | 'strict': 'true', | |
220 | }) | |
221 | if missing_tracks: | |
222 | tracks.extend(missing_tracks) | |
223 | ||
224 | return tracks | |
225 | ||
e7c14660 | 226 | def _build_playlist(self, tracks): |
29f7c58a | 227 | entries = [] |
228 | for track in tracks: | |
229 | track_id = track.get('id') or track.get('realId') | |
230 | if not track_id: | |
231 | continue | |
232 | albums = track.get('albums') | |
233 | if not albums or not isinstance(albums, list): | |
234 | continue | |
235 | album = albums[0] | |
236 | if not isinstance(album, dict): | |
237 | continue | |
238 | album_id = album.get('id') | |
239 | if not album_id: | |
240 | continue | |
241 | entries.append(self.url_result( | |
242 | 'http://music.yandex.ru/album/%s/track/%s' % (album_id, track_id), | |
243 | ie=YandexMusicTrackIE.ie_key(), video_id=track_id)) | |
244 | return entries | |
e7c14660 S |
245 | |
246 | ||
247 | class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE): | |
47fe42e1 S |
248 | IE_NAME = 'yandexmusic:album' |
249 | IE_DESC = 'Яндекс.Музыка - Альбом' | |
29f7c58a | 250 | _VALID_URL = r'%s/album/(?P<id>\d+)' % YandexMusicBaseIE._VALID_URL_BASE |
4c603938 | 251 | |
c452790a | 252 | _TESTS = [{ |
47fe42e1 S |
253 | 'url': 'http://music.yandex.ru/album/540508', |
254 | 'info_dict': { | |
255 | 'id': '540508', | |
29f7c58a | 256 | 'title': 'md5:7ed1c3567f28d14be9f61179116f5571', |
47fe42e1 S |
257 | }, |
258 | 'playlist_count': 50, | |
29f7c58a | 259 | # 'skip': 'Travis CI servers blocked by YandexMusic', |
c452790a | 260 | }, { |
261 | 'url': 'https://music.yandex.ru/album/3840501', | |
262 | 'info_dict': { | |
263 | 'id': '3840501', | |
29f7c58a | 264 | 'title': 'md5:36733472cdaa7dcb1fd9473f7da8e50f', |
c452790a | 265 | }, |
266 | 'playlist_count': 33, | |
29f7c58a | 267 | # 'skip': 'Travis CI servers blocked by YandexMusic', |
268 | }, { | |
269 | # empty artists | |
270 | 'url': 'https://music.yandex.ru/album/9091882', | |
271 | 'info_dict': { | |
272 | 'id': '9091882', | |
273 | 'title': 'ТЕД на русском', | |
274 | }, | |
275 | 'playlist_count': 187, | |
c452790a | 276 | }] |
4c603938 | 277 | |
29f7c58a | 278 | @classmethod |
279 | def suitable(cls, url): | |
280 | return False if YandexMusicTrackIE.suitable(url) else super(YandexMusicAlbumIE, cls).suitable(url) | |
281 | ||
4c603938 | 282 | def _real_extract(self, url): |
29f7c58a | 283 | mobj = re.match(self._VALID_URL, url) |
284 | tld = mobj.group('tld') | |
285 | album_id = mobj.group('id') | |
4c603938 | 286 | |
29f7c58a | 287 | album = self._call_api( |
288 | 'album', tld, url, album_id, 'Downloading album JSON', | |
289 | {'album': album_id}) | |
4c603938 | 290 | |
c452790a | 291 | entries = self._build_playlist([track for volume in album['volumes'] for track in volume]) |
47fe42e1 | 292 | |
29f7c58a | 293 | title = album['title'] |
294 | artist = try_get(album, lambda x: x['artists'][0]['name'], compat_str) | |
295 | if artist: | |
296 | title = '%s - %s' % (artist, title) | |
47fe42e1 S |
297 | year = album.get('year') |
298 | if year: | |
299 | title += ' (%s)' % year | |
300 | ||
301 | return self.playlist_result(entries, compat_str(album['id']), title) | |
4c603938 | 302 | |
4c603938 | 303 | |
e7c14660 | 304 | class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE): |
47fe42e1 S |
305 | IE_NAME = 'yandexmusic:playlist' |
306 | IE_DESC = 'Яндекс.Музыка - Плейлист' | |
29f7c58a | 307 | _VALID_URL = r'%s/users/(?P<user>[^/]+)/playlists/(?P<id>\d+)' % YandexMusicBaseIE._VALID_URL_BASE |
47fe42e1 | 308 | |
baf510bf | 309 | _TESTS = [{ |
47fe42e1 | 310 | 'url': 'http://music.yandex.ru/users/music.partners/playlists/1245', |
4c603938 | 311 | 'info_dict': { |
47fe42e1 | 312 | 'id': '1245', |
29f7c58a | 313 | 'title': 'md5:841559b3fe2b998eca88d0d2e22a3097', |
47fe42e1 S |
314 | 'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9', |
315 | }, | |
29f7c58a | 316 | 'playlist_count': 5, |
317 | # 'skip': 'Travis CI servers blocked by YandexMusic', | |
baf510bf | 318 | }, { |
baf510bf | 319 | 'url': 'https://music.yandex.ru/users/ya.playlist/playlists/1036', |
29f7c58a | 320 | 'only_matching': True, |
321 | }, { | |
322 | # playlist exceeding the limit of 150 tracks (see | |
323 | # https://github.com/ytdl-org/youtube-dl/issues/6666) | |
324 | 'url': 'https://music.yandex.ru/users/mesiaz/playlists/1364', | |
baf510bf | 325 | 'info_dict': { |
29f7c58a | 326 | 'id': '1364', |
327 | 'title': 'md5:b3b400f997d3f878a13ae0699653f7db', | |
baf510bf | 328 | }, |
29f7c58a | 329 | 'playlist_mincount': 437, |
330 | # 'skip': 'Travis CI servers blocked by YandexMusic', | |
baf510bf | 331 | }] |
4c603938 MA |
332 | |
333 | def _real_extract(self, url): | |
e960c3c2 S |
334 | mobj = re.match(self._VALID_URL, url) |
335 | tld = mobj.group('tld') | |
336 | user = mobj.group('user') | |
337 | playlist_id = mobj.group('id') | |
338 | ||
29f7c58a | 339 | playlist = self._call_api( |
340 | 'playlist', tld, url, playlist_id, 'Downloading playlist JSON', { | |
e960c3c2 S |
341 | 'owner': user, |
342 | 'kinds': playlist_id, | |
343 | 'light': 'true', | |
344 | 'lang': tld, | |
345 | 'external-domain': 'music.yandex.%s' % tld, | |
346 | 'overembed': 'false', | |
347 | })['playlist'] | |
baf510bf | 348 | |
29f7c58a | 349 | tracks = self._extract_tracks(playlist, playlist_id, url, tld) |
47fe42e1 | 350 | |
47fe42e1 | 351 | return self.playlist_result( |
baf510bf | 352 | self._build_playlist(tracks), |
e7c14660 | 353 | compat_str(playlist_id), |
203a3c0e | 354 | playlist.get('title'), playlist.get('description')) |
29f7c58a | 355 | |
356 | ||
357 | class YandexMusicArtistBaseIE(YandexMusicPlaylistBaseIE): | |
358 | def _call_artist(self, tld, url, artist_id): | |
359 | return self._call_api( | |
360 | 'artist', tld, url, artist_id, | |
361 | 'Downloading artist %s JSON' % self._ARTIST_WHAT, { | |
362 | 'artist': artist_id, | |
363 | 'what': self._ARTIST_WHAT, | |
364 | 'sort': self._ARTIST_SORT or '', | |
365 | 'dir': '', | |
366 | 'period': '', | |
367 | 'lang': tld, | |
368 | 'external-domain': 'music.yandex.%s' % tld, | |
369 | 'overembed': 'false', | |
370 | }) | |
371 | ||
372 | def _real_extract(self, url): | |
373 | mobj = re.match(self._VALID_URL, url) | |
374 | tld = mobj.group('tld') | |
375 | artist_id = mobj.group('id') | |
376 | data = self._call_artist(tld, url, artist_id) | |
377 | tracks = self._extract_tracks(data, artist_id, url, tld) | |
378 | title = try_get(data, lambda x: x['artist']['name'], compat_str) | |
379 | return self.playlist_result( | |
380 | self._build_playlist(tracks), artist_id, title) | |
381 | ||
382 | ||
383 | class YandexMusicArtistTracksIE(YandexMusicArtistBaseIE): | |
384 | IE_NAME = 'yandexmusic:artist:tracks' | |
385 | IE_DESC = 'Яндекс.Музыка - Артист - Треки' | |
386 | _VALID_URL = r'%s/artist/(?P<id>\d+)/tracks' % YandexMusicBaseIE._VALID_URL_BASE | |
387 | ||
388 | _TESTS = [{ | |
389 | 'url': 'https://music.yandex.ru/artist/617526/tracks', | |
390 | 'info_dict': { | |
391 | 'id': '617526', | |
392 | 'title': 'md5:131aef29d45fd5a965ca613e708c040b', | |
393 | }, | |
394 | 'playlist_count': 507, | |
395 | # 'skip': 'Travis CI servers blocked by YandexMusic', | |
396 | }] | |
397 | ||
398 | _ARTIST_SORT = '' | |
399 | _ARTIST_WHAT = 'tracks' | |
400 | ||
401 | def _real_extract(self, url): | |
402 | mobj = re.match(self._VALID_URL, url) | |
403 | tld = mobj.group('tld') | |
404 | artist_id = mobj.group('id') | |
405 | data = self._call_artist(tld, url, artist_id) | |
406 | tracks = self._extract_tracks(data, artist_id, url, tld) | |
407 | artist = try_get(data, lambda x: x['artist']['name'], compat_str) | |
408 | title = '%s - %s' % (artist or artist_id, 'Треки') | |
409 | return self.playlist_result( | |
410 | self._build_playlist(tracks), artist_id, title) | |
411 | ||
412 | ||
413 | class YandexMusicArtistAlbumsIE(YandexMusicArtistBaseIE): | |
414 | IE_NAME = 'yandexmusic:artist:albums' | |
415 | IE_DESC = 'Яндекс.Музыка - Артист - Альбомы' | |
416 | _VALID_URL = r'%s/artist/(?P<id>\d+)/albums' % YandexMusicBaseIE._VALID_URL_BASE | |
417 | ||
418 | _TESTS = [{ | |
419 | 'url': 'https://music.yandex.ru/artist/617526/albums', | |
420 | 'info_dict': { | |
421 | 'id': '617526', | |
422 | 'title': 'md5:55dc58d5c85699b7fb41ee926700236c', | |
423 | }, | |
424 | 'playlist_count': 8, | |
425 | # 'skip': 'Travis CI servers blocked by YandexMusic', | |
426 | }] | |
427 | ||
428 | _ARTIST_SORT = 'year' | |
429 | _ARTIST_WHAT = 'albums' | |
430 | ||
431 | def _real_extract(self, url): | |
432 | mobj = re.match(self._VALID_URL, url) | |
433 | tld = mobj.group('tld') | |
434 | artist_id = mobj.group('id') | |
435 | data = self._call_artist(tld, url, artist_id) | |
436 | entries = [] | |
437 | for album in data['albums']: | |
438 | if not isinstance(album, dict): | |
439 | continue | |
440 | album_id = album.get('id') | |
441 | if not album_id: | |
442 | continue | |
443 | entries.append(self.url_result( | |
444 | 'http://music.yandex.ru/album/%s' % album_id, | |
445 | ie=YandexMusicAlbumIE.ie_key(), video_id=album_id)) | |
446 | artist = try_get(data, lambda x: x['artist']['name'], compat_str) | |
447 | title = '%s - %s' % (artist or artist_id, 'Альбомы') | |
448 | return self.playlist_result(entries, artist_id, title) |