]>
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, | |
13 | ) | |
4c603938 | 14 | |
4c603938 | 15 | |
ae7d31af S |
16 | class YandexMusicBaseIE(InfoExtractor): |
17 | @staticmethod | |
18 | def _handle_error(response): | |
eebe6b38 S |
19 | if isinstance(response, dict): |
20 | error = response.get('error') | |
21 | if error: | |
22 | raise ExtractorError(error, expected=True) | |
197a5da1 S |
23 | if response.get('type') == 'captcha' or 'captcha' in response: |
24 | YandexMusicBaseIE._raise_captcha() | |
25 | ||
26 | @staticmethod | |
27 | def _raise_captcha(): | |
28 | raise ExtractorError( | |
29 | 'YandexMusic has considered youtube-dl requests automated and ' | |
30 | 'asks you to solve a CAPTCHA. You can either wait for some ' | |
31 | 'time until unblocked and optionally use --sleep-interval ' | |
32 | 'in future or alternatively you can go to https://music.yandex.ru/ ' | |
33 | 'solve CAPTCHA, then export cookies and pass cookie file to ' | |
34 | 'youtube-dl with --cookies', | |
35 | expected=True) | |
ae7d31af | 36 | |
e5eadfa8 S |
37 | def _download_webpage_handle(self, *args, **kwargs): |
38 | webpage = super(YandexMusicBaseIE, self)._download_webpage_handle(*args, **kwargs) | |
7f776fa4 | 39 | if 'Нам очень жаль, но запросы, поступившие с вашего IP-адреса, похожи на автоматические.' in webpage: |
197a5da1 | 40 | self._raise_captcha() |
7f776fa4 YCH |
41 | return webpage |
42 | ||
ae7d31af S |
43 | def _download_json(self, *args, **kwargs): |
44 | response = super(YandexMusicBaseIE, self)._download_json(*args, **kwargs) | |
45 | self._handle_error(response) | |
46 | return response | |
47 | ||
48 | ||
49 | class YandexMusicTrackIE(YandexMusicBaseIE): | |
e4df2f98 S |
50 | IE_NAME = 'yandexmusic:track' |
51 | IE_DESC = 'Яндекс.Музыка - Трек' | |
52 | _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<album_id>\d+)/track/(?P<id>\d+)' | |
53 | ||
54 | _TEST = { | |
55 | 'url': 'http://music.yandex.ru/album/540508/track/4878838', | |
56 | 'md5': 'f496818aa2f60b6c0062980d2e00dc20', | |
57 | 'info_dict': { | |
58 | 'id': '4878838', | |
59 | 'ext': 'mp3', | |
796bf9de | 60 | 'title': 'Carlo Ambrosio, Carlo Ambrosio & Fabio Di Bari - Gypsy Eyes 1', |
e4df2f98 S |
61 | 'filesize': 4628061, |
62 | 'duration': 193.04, | |
e90d1754 S |
63 | 'track': 'Gypsy Eyes 1', |
64 | 'album': 'Gypsy Soul', | |
65 | 'album_artist': 'Carlo Ambrosio', | |
796bf9de S |
66 | 'artist': 'Carlo Ambrosio, Carlo Ambrosio & Fabio Di Bari', |
67 | 'release_year': 2009, | |
7f776fa4 YCH |
68 | }, |
69 | 'skip': 'Travis CI servers blocked by YandexMusic', | |
e4df2f98 S |
70 | } |
71 | ||
4c603938 | 72 | def _get_track_url(self, storage_dir, track_id): |
47fe42e1 S |
73 | data = self._download_json( |
74 | 'http://music.yandex.ru/api/v1.5/handlers/api-jsonp.jsx?action=getTrackSrc&p=download-info/%s' | |
75 | % storage_dir, | |
76 | track_id, 'Downloading track location JSON') | |
4c603938 | 77 | |
cf03e34a S |
78 | # Each string is now wrapped in a list, this is probably only temporarily thus |
79 | # supporting both scenarios (see https://github.com/rg3/youtube-dl/issues/10193) | |
80 | for k, v in data.items(): | |
81 | if v and isinstance(v, list): | |
82 | data[k] = v[0] | |
83 | ||
47fe42e1 | 84 | key = hashlib.md5(('XGRlBW9FXlekgbPrRHuSiA' + data['path'][1:] + data['s']).encode('utf-8')).hexdigest() |
4c603938 MA |
85 | storage = storage_dir.split('.') |
86 | ||
47fe42e1 S |
87 | return ('http://%s/get-mp3/%s/%s?track-id=%s&from=service-10-track&similarities-experiment=default' |
88 | % (data['host'], key, data['ts'] + data['path'], storage[1])) | |
4c603938 | 89 | |
47fe42e1 | 90 | def _get_track_info(self, track): |
ab953c64 S |
91 | thumbnail = None |
92 | cover_uri = track.get('albums', [{}])[0].get('coverUri') | |
93 | if cover_uri: | |
94 | thumbnail = cover_uri.replace('%%', 'orig') | |
95 | if not thumbnail.startswith('http'): | |
96 | thumbnail = 'http://' + thumbnail | |
e90d1754 S |
97 | |
98 | track_title = track['title'] | |
99 | track_info = { | |
47fe42e1 S |
100 | 'id': track['id'], |
101 | 'ext': 'mp3', | |
102 | 'url': self._get_track_url(track['storageDir'], track['id']), | |
47fe42e1 S |
103 | 'filesize': int_or_none(track.get('fileSize')), |
104 | 'duration': float_or_none(track.get('durationMs'), 1000), | |
ab953c64 | 105 | 'thumbnail': thumbnail, |
e90d1754 | 106 | 'track': track_title, |
47fe42e1 | 107 | } |
4c603938 | 108 | |
e90d1754 S |
109 | def extract_artist(artist_list): |
110 | if artist_list and isinstance(artist_list, list): | |
111 | artists_names = [a['name'] for a in artist_list if a.get('name')] | |
112 | if artists_names: | |
113 | return ', '.join(artists_names) | |
114 | ||
115 | albums = track.get('albums') | |
116 | if albums and isinstance(albums, list): | |
117 | album = albums[0] | |
118 | if isinstance(album, dict): | |
119 | year = album.get('year') | |
120 | track_info.update({ | |
121 | 'album': album.get('title'), | |
122 | 'album_artist': extract_artist(album.get('artists')), | |
796bf9de | 123 | 'release_year': int_or_none(year), |
e90d1754 S |
124 | }) |
125 | ||
126 | track_artist = extract_artist(track.get('artists')) | |
127 | if track_artist: | |
128 | track_info.update({ | |
129 | 'artist': track_artist, | |
130 | 'title': '%s - %s' % (track_artist, track_title), | |
131 | }) | |
132 | else: | |
133 | track_info['title'] = track_title | |
134 | return track_info | |
135 | ||
4c603938 | 136 | def _real_extract(self, url): |
47fe42e1 S |
137 | mobj = re.match(self._VALID_URL, url) |
138 | album_id, track_id = mobj.group('album_id'), mobj.group('id') | |
4c603938 | 139 | |
47fe42e1 S |
140 | track = self._download_json( |
141 | 'http://music.yandex.ru/handlers/track.jsx?track=%s:%s' % (track_id, album_id), | |
142 | track_id, 'Downloading track JSON')['track'] | |
4c603938 | 143 | |
47fe42e1 | 144 | return self._get_track_info(track) |
4c603938 | 145 | |
4c603938 | 146 | |
ae7d31af | 147 | class YandexMusicPlaylistBaseIE(YandexMusicBaseIE): |
e7c14660 S |
148 | def _build_playlist(self, tracks): |
149 | return [ | |
150 | self.url_result( | |
151 | 'http://music.yandex.ru/album/%s/track/%s' % (track['albums'][0]['id'], track['id'])) | |
6d53cdd6 | 152 | for track in tracks if track.get('albums') and isinstance(track.get('albums'), list)] |
e7c14660 S |
153 | |
154 | ||
155 | class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE): | |
47fe42e1 S |
156 | IE_NAME = 'yandexmusic:album' |
157 | IE_DESC = 'Яндекс.Музыка - Альбом' | |
29171bc2 | 158 | _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<id>\d+)/?(\?|$)' |
4c603938 | 159 | |
47fe42e1 S |
160 | _TEST = { |
161 | 'url': 'http://music.yandex.ru/album/540508', | |
162 | 'info_dict': { | |
163 | 'id': '540508', | |
164 | 'title': 'Carlo Ambrosio - Gypsy Soul (2009)', | |
165 | }, | |
166 | 'playlist_count': 50, | |
7f776fa4 | 167 | 'skip': 'Travis CI servers blocked by YandexMusic', |
47fe42e1 | 168 | } |
4c603938 MA |
169 | |
170 | def _real_extract(self, url): | |
47fe42e1 | 171 | album_id = self._match_id(url) |
4c603938 | 172 | |
47fe42e1 S |
173 | album = self._download_json( |
174 | 'http://music.yandex.ru/handlers/album.jsx?album=%s' % album_id, | |
175 | album_id, 'Downloading album JSON') | |
4c603938 | 176 | |
e7c14660 | 177 | entries = self._build_playlist(album['volumes'][0]) |
47fe42e1 S |
178 | |
179 | title = '%s - %s' % (album['artists'][0]['name'], album['title']) | |
180 | year = album.get('year') | |
181 | if year: | |
182 | title += ' (%s)' % year | |
183 | ||
184 | return self.playlist_result(entries, compat_str(album['id']), title) | |
4c603938 | 185 | |
4c603938 | 186 | |
e7c14660 | 187 | class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE): |
47fe42e1 S |
188 | IE_NAME = 'yandexmusic:playlist' |
189 | IE_DESC = 'Яндекс.Музыка - Плейлист' | |
e960c3c2 | 190 | _VALID_URL = r'https?://music\.yandex\.(?P<tld>ru|kz|ua|by)/users/(?P<user>[^/]+)/playlists/(?P<id>\d+)' |
47fe42e1 | 191 | |
baf510bf | 192 | _TESTS = [{ |
47fe42e1 | 193 | 'url': 'http://music.yandex.ru/users/music.partners/playlists/1245', |
4c603938 | 194 | 'info_dict': { |
47fe42e1 S |
195 | 'id': '1245', |
196 | 'title': 'Что слушают Enter Shikari', | |
197 | 'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9', | |
198 | }, | |
199 | 'playlist_count': 6, | |
7f776fa4 | 200 | 'skip': 'Travis CI servers blocked by YandexMusic', |
baf510bf S |
201 | }, { |
202 | # playlist exceeding the limit of 150 tracks shipped with webpage (see | |
203 | # https://github.com/rg3/youtube-dl/issues/6666) | |
204 | 'url': 'https://music.yandex.ru/users/ya.playlist/playlists/1036', | |
205 | 'info_dict': { | |
206 | 'id': '1036', | |
207 | 'title': 'Музыка 90-х', | |
208 | }, | |
9da526aa | 209 | 'playlist_mincount': 300, |
7f776fa4 | 210 | 'skip': 'Travis CI servers blocked by YandexMusic', |
baf510bf | 211 | }] |
4c603938 MA |
212 | |
213 | def _real_extract(self, url): | |
e960c3c2 S |
214 | mobj = re.match(self._VALID_URL, url) |
215 | tld = mobj.group('tld') | |
216 | user = mobj.group('user') | |
217 | playlist_id = mobj.group('id') | |
218 | ||
219 | playlist = self._download_json( | |
220 | 'https://music.yandex.%s/handlers/playlist.jsx' % tld, | |
221 | playlist_id, 'Downloading missing tracks JSON', | |
222 | fatal=False, | |
223 | headers={ | |
224 | 'Referer': url, | |
225 | 'X-Requested-With': 'XMLHttpRequest', | |
226 | 'X-Retpath-Y': url, | |
227 | }, | |
228 | query={ | |
229 | 'owner': user, | |
230 | 'kinds': playlist_id, | |
231 | 'light': 'true', | |
232 | 'lang': tld, | |
233 | 'external-domain': 'music.yandex.%s' % tld, | |
234 | 'overembed': 'false', | |
235 | })['playlist'] | |
baf510bf | 236 | |
a4c81e49 S |
237 | tracks = playlist['tracks'] |
238 | track_ids = [compat_str(track_id) for track_id in playlist['trackIds']] | |
baf510bf | 239 | |
e960c3c2 | 240 | # tracks dictionary shipped with playlist.jsx API is limited to 150 tracks, |
baf510bf S |
241 | # missing tracks should be retrieved manually. |
242 | if len(tracks) < len(track_ids): | |
2a48e6f0 S |
243 | present_track_ids = set([ |
244 | compat_str(track['id']) | |
245 | for track in tracks if track.get('id')]) | |
246 | missing_track_ids = [ | |
247 | track_id for track_id in track_ids | |
248 | if track_id not in present_track_ids] | |
15fc0658 S |
249 | missing_tracks = self._download_json( |
250 | 'https://music.yandex.%s/handlers/track-entries.jsx' % tld, | |
251 | playlist_id, 'Downloading missing tracks JSON', | |
252 | fatal=False, | |
253 | headers={ | |
254 | 'Referer': url, | |
255 | 'X-Requested-With': 'XMLHttpRequest', | |
256 | }, | |
257 | query={ | |
baf510bf | 258 | 'entries': ','.join(missing_track_ids), |
e960c3c2 S |
259 | 'lang': tld, |
260 | 'external-domain': 'music.yandex.%s' % tld, | |
baf510bf | 261 | 'overembed': 'false', |
baf510bf | 262 | 'strict': 'true', |
15fc0658 | 263 | }) |
baf510bf S |
264 | if missing_tracks: |
265 | tracks.extend(missing_tracks) | |
47fe42e1 | 266 | |
47fe42e1 | 267 | return self.playlist_result( |
baf510bf | 268 | self._build_playlist(tracks), |
e7c14660 | 269 | compat_str(playlist_id), |
203a3c0e | 270 | playlist.get('title'), playlist.get('description')) |