]>
Commit | Line | Data |
---|---|---|
45d7bc2f JMF |
1 | from __future__ import unicode_literals |
2 | ||
70d1924f | 3 | import re |
55992530 | 4 | import json |
70d1924f JMF |
5 | |
6 | from .common import InfoExtractor | |
e0da32df | 7 | from ..compat import ( |
5c9ced95 | 8 | compat_str, |
e0da32df | 9 | compat_urlparse, |
9bccdc70 | 10 | compat_HTTPError, |
e0da32df | 11 | ) |
1cc79574 | 12 | from ..utils import ( |
70d1924f | 13 | ExtractorError, |
7d3d06a1 | 14 | int_or_none, |
9165d6ba | 15 | parse_iso8601, |
70d1924f JMF |
16 | ) |
17 | ||
88bd97e3 | 18 | |
9618c448 | 19 | class VevoBaseIE(InfoExtractor): |
9d0c08a0 | 20 | def _extract_json(self, webpage, video_id): |
9618c448 S |
21 | return self._parse_json( |
22 | self._search_regex( | |
23 | r'window\.__INITIAL_STORE__\s*=\s*({.+?});\s*</script>', | |
24 | webpage, 'initial store'), | |
9d0c08a0 | 25 | video_id) |
9618c448 S |
26 | |
27 | ||
28 | class VevoIE(VevoBaseIE): | |
2975fe1a | 29 | ''' |
0577177e | 30 | Accepts urls from vevo.com or in the format 'vevo:{id}' |
3266f0c6 | 31 | (currently used by MTVIE and MySpaceIE) |
2975fe1a | 32 | ''' |
f25571ff | 33 | _VALID_URL = r'''(?x) |
92519402 | 34 | (?:https?://(?:www\.)?vevo\.com/watch/(?!playlist|genre)(?:[^/]+/(?:[^/]+/)?)?| |
f25571ff | 35 | https?://cache\.vevo\.com/m/html/embed\.html\?video=| |
ebce53b3 | 36 | https?://videoplayer\.vevo\.com/embed/embedded\?videoId=| |
f25571ff PH |
37 | vevo:) |
38 | (?P<id>[^&?#]+)''' | |
fd5e6f7e | 39 | |
72321ead | 40 | _TESTS = [{ |
45d7bc2f | 41 | 'url': 'http://www.vevo.com/watch/hurts/somebody-to-die-for/GB1101300280', |
682f8c43 | 42 | 'md5': '95ee28ee45e70130e3ab02b0f579ae23', |
45d7bc2f | 43 | 'info_dict': { |
682f8c43 | 44 | 'id': 'GB1101300280', |
27579b9e | 45 | 'ext': 'mp4', |
881dbc86 S |
46 | 'title': 'Hurts - Somebody to Die For', |
47 | 'timestamp': 1372057200, | |
2975fe1a | 48 | 'upload_date': '20130624', |
682f8c43 | 49 | 'uploader': 'Hurts', |
881dbc86 S |
50 | 'track': 'Somebody to Die For', |
51 | 'artist': 'Hurts', | |
52 | 'genre': 'Pop', | |
2975fe1a | 53 | }, |
1db05846 | 54 | 'expected_warnings': ['Unable to download SMIL file', 'Unable to download info'], |
27579b9e PH |
55 | }, { |
56 | 'note': 'v3 SMIL format', | |
57 | 'url': 'http://www.vevo.com/watch/cassadee-pope/i-wish-i-could-break-your-heart/USUV71302923', | |
682f8c43 | 58 | 'md5': 'f6ab09b034f8c22969020b042e5ac7fc', |
27579b9e | 59 | 'info_dict': { |
682f8c43 | 60 | 'id': 'USUV71302923', |
27579b9e | 61 | 'ext': 'mp4', |
881dbc86 S |
62 | 'title': 'Cassadee Pope - I Wish I Could Break Your Heart', |
63 | 'timestamp': 1392796919, | |
27579b9e | 64 | 'upload_date': '20140219', |
682f8c43 | 65 | 'uploader': 'Cassadee Pope', |
881dbc86 S |
66 | 'track': 'I Wish I Could Break Your Heart', |
67 | 'artist': 'Cassadee Pope', | |
68 | 'genre': 'Country', | |
2975fe1a | 69 | }, |
1db05846 | 70 | 'expected_warnings': ['Unable to download SMIL file', 'Unable to download info'], |
6cadf8c8 PH |
71 | }, { |
72 | 'note': 'Age-limited video', | |
73 | 'url': 'https://www.vevo.com/watch/justin-timberlake/tunnel-vision-explicit/USRV81300282', | |
74 | 'info_dict': { | |
682f8c43 | 75 | 'id': 'USRV81300282', |
6cadf8c8 | 76 | 'ext': 'mp4', |
881dbc86 | 77 | 'title': 'Justin Timberlake - Tunnel Vision (Explicit)', |
682f8c43 | 78 | 'age_limit': 18, |
682f8c43 | 79 | 'timestamp': 1372888800, |
881dbc86 S |
80 | 'upload_date': '20130703', |
81 | 'uploader': 'Justin Timberlake', | |
82 | 'track': 'Tunnel Vision (Explicit)', | |
83 | 'artist': 'Justin Timberlake', | |
84 | 'genre': 'Pop', | |
2975fe1a | 85 | }, |
1db05846 | 86 | 'expected_warnings': ['Unable to download SMIL file', 'Unable to download info'], |
ff51983e YCH |
87 | }, { |
88 | 'note': 'No video_info', | |
89 | 'url': 'http://www.vevo.com/watch/k-camp-1/Till-I-Die/USUV71503000', | |
682f8c43 | 90 | 'md5': '8b83cc492d72fc9cf74a02acee7dc1b0', |
ff51983e | 91 | 'info_dict': { |
682f8c43 | 92 | 'id': 'USUV71503000', |
ff51983e | 93 | 'ext': 'mp4', |
8b27d83e | 94 | 'title': 'K Camp ft. T.I. - Till I Die', |
682f8c43 | 95 | 'age_limit': 18, |
682f8c43 | 96 | 'timestamp': 1449468000, |
881dbc86 S |
97 | 'upload_date': '20151207', |
98 | 'uploader': 'K Camp', | |
99 | 'track': 'Till I Die', | |
100 | 'artist': 'K Camp', | |
8b27d83e AE |
101 | 'genre': 'Hip-Hop', |
102 | }, | |
1db05846 | 103 | 'expected_warnings': ['Unable to download SMIL file', 'Unable to download info'], |
8b27d83e AE |
104 | }, { |
105 | 'note': 'Featured test', | |
106 | 'url': 'https://www.vevo.com/watch/lemaitre/Wait/USUV71402190', | |
107 | 'md5': 'd28675e5e8805035d949dc5cf161071d', | |
108 | 'info_dict': { | |
109 | 'id': 'USUV71402190', | |
110 | 'ext': 'mp4', | |
111 | 'title': 'Lemaitre ft. LoLo - Wait', | |
112 | 'age_limit': 0, | |
113 | 'timestamp': 1413432000, | |
114 | 'upload_date': '20141016', | |
115 | 'uploader': 'Lemaitre', | |
116 | 'track': 'Wait', | |
117 | 'artist': 'Lemaitre', | |
118 | 'genre': 'Electronic', | |
ff51983e | 119 | }, |
1db05846 | 120 | 'expected_warnings': ['Unable to download SMIL file', 'Unable to download info'], |
8e7d0048 S |
121 | }, { |
122 | 'note': 'Only available via webpage', | |
123 | 'url': 'http://www.vevo.com/watch/GBUV71600656', | |
124 | 'md5': '67e79210613865b66a47c33baa5e37fe', | |
125 | 'info_dict': { | |
126 | 'id': 'GBUV71600656', | |
127 | 'ext': 'mp4', | |
881dbc86 | 128 | 'title': 'ABC - Viva Love', |
8e7d0048 | 129 | 'age_limit': 0, |
8e7d0048 | 130 | 'timestamp': 1461830400, |
881dbc86 S |
131 | 'upload_date': '20160428', |
132 | 'uploader': 'ABC', | |
133 | 'track': 'Viva Love', | |
134 | 'artist': 'ABC', | |
135 | 'genre': 'Pop', | |
8e7d0048 S |
136 | }, |
137 | 'expected_warnings': ['Failed to download video versions info'], | |
5c9ced95 S |
138 | }, { |
139 | # no genres available | |
140 | 'url': 'http://www.vevo.com/watch/INS171400764', | |
141 | 'only_matching': True, | |
9d0c08a0 YCH |
142 | }, { |
143 | # Another case available only via the webpage; using streams/streamsV3 formats | |
144 | # Geo-restricted to Netherlands/Germany | |
145 | 'url': 'http://www.vevo.com/watch/boostee/pop-corn-clip-officiel/FR1A91600909', | |
146 | 'only_matching': True, | |
72321ead | 147 | }] |
2975fe1a | 148 | _VERSIONS = { |
9165d6ba | 149 | 0: 'youtube', # only in AuthenticateVideo videoVersions |
2975fe1a | 150 | 1: 'level3', |
151 | 2: 'akamai', | |
152 | 3: 'level3', | |
153 | 4: 'amazon', | |
154 | } | |
155 | ||
682f8c43 | 156 | def _initialize_api(self, video_id): |
9165d6ba | 157 | webpage = self._download_webpage( |
b07ea5ea | 158 | 'https://accounts.vevo.com/token', None, |
9165d6ba | 159 | note='Retrieving oauth token', |
b07ea5ea S |
160 | errnote='Unable to retrieve oauth token', |
161 | data=json.dumps({ | |
162 | 'client_id': 'SPupX1tvqFEopQ1YS6SS', | |
163 | 'grant_type': 'urn:vevo:params:oauth:grant-type:anonymous', | |
164 | }).encode('utf-8'), | |
165 | headers={ | |
166 | 'Content-Type': 'application/json', | |
167 | }) | |
9165d6ba | 168 | |
621a2800 | 169 | if re.search(r'(?i)THIS PAGE IS CURRENTLY UNAVAILABLE IN YOUR REGION', webpage): |
bc7e77a0 S |
170 | self.raise_geo_restricted( |
171 | '%s said: This page is currently unavailable in your region' % self.IE_NAME) | |
9165d6ba | 172 | |
173 | auth_info = self._parse_json(webpage, video_id) | |
55992530 | 174 | self._api_url_template = self.http_scheme() + '//apiv2.vevo.com/%s?token=' + auth_info['legacy_token'] |
9165d6ba | 175 | |
516ea41a | 176 | def _call_api(self, path, *args, **kwargs): |
9bccdc70 RA |
177 | try: |
178 | data = self._download_json(self._api_url_template % path, *args, **kwargs) | |
179 | except ExtractorError as e: | |
180 | if isinstance(e.cause, compat_HTTPError): | |
181 | errors = self._parse_json(e.cause.read().decode(), None)['errors'] | |
182 | error_message = ', '.join([error['message'] for error in errors]) | |
183 | raise ExtractorError('%s said: %s' % (self.IE_NAME, error_message), expected=True) | |
184 | raise | |
185 | return data | |
9165d6ba | 186 | |
72321ead | 187 | def _real_extract(self, url): |
4b942883 | 188 | video_id = self._match_id(url) |
72321ead | 189 | |
9bccdc70 | 190 | self._initialize_api(video_id) |
682f8c43 | 191 | |
9bccdc70 RA |
192 | video_info = self._call_api( |
193 | 'video/%s' % video_id, video_id, 'Downloading api video info', | |
194 | 'Failed to download video info') | |
2975fe1a | 195 | |
9bccdc70 RA |
196 | video_versions = self._call_api( |
197 | 'video/%s/streams' % video_id, video_id, | |
198 | 'Downloading video versions info', | |
199 | 'Failed to download video versions info', | |
200 | fatal=False) | |
ff51983e | 201 | |
9bccdc70 RA |
202 | # Some videos are only available via webpage (e.g. |
203 | # https://github.com/rg3/youtube-dl/issues/9366) | |
204 | if not video_versions: | |
205 | webpage = self._download_webpage(url, video_id) | |
9d0c08a0 YCH |
206 | json_data = self._extract_json(webpage, video_id) |
207 | if 'streams' in json_data.get('default', {}): | |
208 | video_versions = json_data['default']['streams'][video_id][0] | |
209 | else: | |
210 | video_versions = [ | |
211 | value | |
212 | for key, value in json_data['apollo']['data'].items() | |
213 | if key.startswith('%s.streams' % video_id)] | |
9618c448 | 214 | |
9bccdc70 RA |
215 | uploader = None |
216 | artist = None | |
217 | featured_artist = None | |
218 | artists = video_info.get('artists') | |
219 | for curr_artist in artists: | |
220 | if curr_artist.get('role') == 'Featured': | |
221 | featured_artist = curr_artist['name'] | |
222 | else: | |
223 | artist = uploader = curr_artist['name'] | |
9165d6ba | 224 | |
9bccdc70 RA |
225 | formats = [] |
226 | for video_version in video_versions: | |
9d0c08a0 | 227 | version = self._VERSIONS.get(video_version.get('version'), 'generic') |
9bccdc70 RA |
228 | version_url = video_version.get('url') |
229 | if not version_url: | |
230 | continue | |
9165d6ba | 231 | |
9bccdc70 RA |
232 | if '.ism' in version_url: |
233 | continue | |
234 | elif '.mpd' in version_url: | |
235 | formats.extend(self._extract_mpd_formats( | |
236 | version_url, video_id, mpd_id='dash-%s' % version, | |
237 | note='Downloading %s MPD information' % version, | |
238 | errnote='Failed to download %s MPD information' % version, | |
239 | fatal=False)) | |
240 | elif '.m3u8' in version_url: | |
241 | formats.extend(self._extract_m3u8_formats( | |
242 | version_url, video_id, 'mp4', 'm3u8_native', | |
243 | m3u8_id='hls-%s' % version, | |
244 | note='Downloading %s m3u8 information' % version, | |
245 | errnote='Failed to download %s m3u8 information' % version, | |
246 | fatal=False)) | |
247 | else: | |
248 | m = re.search(r'''(?xi) | |
249 | _(?P<width>[0-9]+)x(?P<height>[0-9]+) | |
250 | _(?P<vcodec>[a-z0-9]+) | |
251 | _(?P<vbr>[0-9]+) | |
252 | _(?P<acodec>[a-z0-9]+) | |
253 | _(?P<abr>[0-9]+) | |
254 | \.(?P<ext>[a-z0-9]+)''', version_url) | |
255 | if not m: | |
9165d6ba | 256 | continue |
9165d6ba | 257 | |
9bccdc70 RA |
258 | formats.append({ |
259 | 'url': version_url, | |
260 | 'format_id': 'http-%s-%s' % (version, video_version['quality']), | |
261 | 'vcodec': m.group('vcodec'), | |
262 | 'acodec': m.group('acodec'), | |
263 | 'vbr': int(m.group('vbr')), | |
264 | 'abr': int(m.group('abr')), | |
265 | 'ext': m.group('ext'), | |
266 | 'width': int(m.group('width')), | |
267 | 'height': int(m.group('height')), | |
268 | }) | |
2975fe1a | 269 | self._sort_formats(formats) |
27579b9e | 270 | |
881dbc86 | 271 | track = video_info['title'] |
9508738f S |
272 | if featured_artist: |
273 | artist = '%s ft. %s' % (artist, featured_artist) | |
274 | title = '%s - %s' % (artist, track) if artist else track | |
5c9ced95 S |
275 | |
276 | genres = video_info.get('genres') | |
277 | genre = ( | |
278 | genres[0] if genres and isinstance(genres, list) and | |
279 | isinstance(genres[0], compat_str) else None) | |
9165d6ba | 280 | |
6cadf8c8 PH |
281 | is_explicit = video_info.get('isExplicit') |
282 | if is_explicit is True: | |
283 | age_limit = 18 | |
284 | elif is_explicit is False: | |
285 | age_limit = 0 | |
286 | else: | |
287 | age_limit = None | |
288 | ||
45d7bc2f | 289 | return { |
88bd97e3 | 290 | 'id': video_id, |
ff51983e | 291 | 'title': title, |
88bd97e3 | 292 | 'formats': formats, |
9165d6ba | 293 | 'thumbnail': video_info.get('imageUrl') or video_info.get('thumbnailUrl'), |
9bccdc70 | 294 | 'timestamp': parse_iso8601(video_info.get('releaseDate')), |
9165d6ba | 295 | 'uploader': uploader, |
9bccdc70 RA |
296 | 'duration': int_or_none(video_info.get('duration')), |
297 | 'view_count': int_or_none(video_info.get('views', {}).get('total')), | |
6cadf8c8 | 298 | 'age_limit': age_limit, |
881dbc86 S |
299 | 'track': track, |
300 | 'artist': uploader, | |
301 | 'genre': genre, | |
88bd97e3 | 302 | } |
e0da32df S |
303 | |
304 | ||
9618c448 | 305 | class VevoPlaylistIE(VevoBaseIE): |
92519402 | 306 | _VALID_URL = r'https?://(?:www\.)?vevo\.com/watch/(?P<kind>playlist|genre)/(?P<id>[^/?#&]+)' |
e0da32df S |
307 | |
308 | _TESTS = [{ | |
309 | 'url': 'http://www.vevo.com/watch/playlist/dadbf4e7-b99f-4184-9670-6f0e547b6a29', | |
310 | 'info_dict': { | |
311 | 'id': 'dadbf4e7-b99f-4184-9670-6f0e547b6a29', | |
312 | 'title': 'Best-Of: Birdman', | |
313 | }, | |
314 | 'playlist_count': 10, | |
e2bd301c S |
315 | }, { |
316 | 'url': 'http://www.vevo.com/watch/genre/rock', | |
317 | 'info_dict': { | |
318 | 'id': 'rock', | |
319 | 'title': 'Rock', | |
320 | }, | |
321 | 'playlist_count': 20, | |
e0da32df S |
322 | }, { |
323 | 'url': 'http://www.vevo.com/watch/playlist/dadbf4e7-b99f-4184-9670-6f0e547b6a29?index=0', | |
324 | 'md5': '32dcdfddddf9ec6917fc88ca26d36282', | |
325 | 'info_dict': { | |
326 | 'id': 'USCMV1100073', | |
327 | 'ext': 'mp4', | |
881dbc86 | 328 | 'title': 'Birdman - Y.U. MAD', |
e0da32df S |
329 | 'timestamp': 1323417600, |
330 | 'upload_date': '20111209', | |
331 | 'uploader': 'Birdman', | |
881dbc86 S |
332 | 'track': 'Y.U. MAD', |
333 | 'artist': 'Birdman', | |
334 | 'genre': 'Rap/Hip-Hop', | |
e0da32df S |
335 | }, |
336 | 'expected_warnings': ['Unable to download SMIL file'], | |
e0da32df S |
337 | }, { |
338 | 'url': 'http://www.vevo.com/watch/genre/rock?index=0', | |
339 | 'only_matching': True, | |
340 | }] | |
341 | ||
342 | def _real_extract(self, url): | |
e2bd301c S |
343 | mobj = re.match(self._VALID_URL, url) |
344 | playlist_id = mobj.group('id') | |
345 | playlist_kind = mobj.group('kind') | |
e0da32df S |
346 | |
347 | webpage = self._download_webpage(url, playlist_id) | |
348 | ||
349 | qs = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query) | |
350 | index = qs.get('index', [None])[0] | |
351 | ||
352 | if index: | |
353 | video_id = self._search_regex( | |
354 | r'<meta[^>]+content=(["\'])vevo://video/(?P<id>.+?)\1[^>]*>', | |
355 | webpage, 'video id', default=None, group='id') | |
356 | if video_id: | |
357 | return self.url_result('vevo:%s' % video_id, VevoIE.ie_key()) | |
358 | ||
9d0c08a0 | 359 | playlists = self._extract_json(webpage, playlist_id)['default']['%ss' % playlist_kind] |
e0da32df | 360 | |
e2bd301c S |
361 | playlist = (list(playlists.values())[0] |
362 | if playlist_kind == 'playlist' else playlists[playlist_id]) | |
e0da32df S |
363 | |
364 | entries = [ | |
365 | self.url_result('vevo:%s' % src, VevoIE.ie_key()) | |
366 | for src in playlist['isrcs']] | |
367 | ||
368 | return self.playlist_result( | |
78a3ff33 | 369 | entries, playlist.get('playlistId') or playlist_id, |
e0da32df | 370 | playlist.get('name'), playlist.get('description')) |