]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/vice.py
8 from .adobepass
import AdobePassIE
9 from .common
import InfoExtractor
10 from .youtube
import YoutubeIE
11 from ..compat
import (
26 class ViceBaseIE(InfoExtractor
):
27 def _call_api(self
, resource
, resource_key
, resource_id
, locale
, fields
, args
=''):
28 return self
._download
_json
(
29 'https://video.vice.com/api/v1/graphql', resource_id
, query
={
31 %s(locale: "%s", %s: "%s"%s) {
34 }''' % (resource
, locale
, resource_key
, resource_id
, args
, fields
),
38 class ViceIE(ViceBaseIE
, AdobePassIE
):
40 _VALID_URL
= r
'https?://(?:(?:video|vms)\.vice|(?:www\.)?vice(?:land|tv))\.com/(?P<locale>[^/]+)/(?:video/[^/]+|embed)/(?P<id>[\da-f]{24})'
42 'url': 'https://video.vice.com/en_us/video/pet-cremator/58c69e38a55424f1227dc3f7',
44 'id': '58c69e38a55424f1227dc3f7',
46 'title': '10 Questions You Always Wanted To Ask: Pet Cremator',
47 'description': 'md5:fe856caacf61fe0e74fab15ce2b07ca5',
49 'uploader_id': '57a204088cb727dec794c67b',
50 'timestamp': 1489664942,
51 'upload_date': '20170316',
56 'skip_download': True,
59 # geo restricted to US
60 'url': 'https://video.vice.com/en_us/video/the-signal-from-tolva/5816510690b70e6c5fd39a56',
62 'id': '5816510690b70e6c5fd39a56',
65 'title': 'The Signal From Tölva',
66 'description': 'md5:3927e3c79f9e8094606a2b3c5b5e55d5',
67 'uploader_id': '57a204088cb727dec794c67b',
68 'timestamp': 1477941983,
69 'upload_date': '20161031',
73 'skip_download': True,
76 'url': 'https://video.vice.com/alps/video/ulfs-wien-beruchtigste-grafitti-crew-part-1/581b12b60a0e1f4c0fb6ea2f',
78 'id': '581b12b60a0e1f4c0fb6ea2f',
80 'title': 'ULFs - Wien berüchtigste Grafitti Crew - Part 1',
81 'description': 'Zwischen Hinterzimmer-Tattoos und U-Bahnschächten erzählen uns die Ulfs, wie es ist, "süchtig nach Sachbeschädigung" zu sein.',
83 'uploader_id': '57a204088cb727dec794c67b',
84 'timestamp': 1485368119,
85 'upload_date': '20170125',
90 'skip_download': True,
93 'url': 'https://video.vice.com/en_us/video/pizza-show-trailer/56d8c9a54d286ed92f7f30e4',
94 'only_matching': True,
96 'url': 'https://video.vice.com/en_us/embed/57f41d3556a0a80f54726060',
97 'only_matching': True,
99 'url': 'https://vms.vice.com/en_us/video/preplay/58c69e38a55424f1227dc3f7',
100 'only_matching': True,
102 'url': 'https://www.viceland.com/en_us/video/thursday-march-1-2018/5a8f2d7ff1cdb332dd446ec1',
103 'only_matching': True,
107 def _extract_urls(webpage
):
109 r
'<iframe\b[^>]+\bsrc=["\']((?
:https?
:)?
//video\
.vice\
.com
/[^
/]+/embed
/[\da
-f
]{24}
)',
113 def _extract_url(webpage):
114 urls = ViceIE._extract_urls(webpage)
115 return urls[0] if urls else None
117 def _real_extract(self, url):
118 locale, video_id = self._match_valid_url(url).groups()
120 video = self._call_api('videos
', 'id', video_id, locale, '''body
125 title = video['title
'].strip()
126 rating = video.get('rating
')
129 if video.get('locked
'):
130 resource = self._get_mvpd_resource(
131 'VICELAND
', title, video_id, rating)
132 query['tvetoken
'] = self._extract_mvpd_auth(
133 url, video_id, 'VICELAND
', resource)
135 # signature generation algorithm is reverse engineered from signatureGenerator in
136 # webpack:///../shared/~/vice-player/dist/js/vice-player.js in
137 # https://www.viceland.com/assets/common/js/web.vendor.bundle.js
138 # new JS is located here https://vice-web-statics-cdn.vice.com/vice-player/player-embed.js
139 exp = int(time.time()) + 1440
143 'sign
': hashlib.sha512(('%s:GET
:%d' % (video_id, exp)).encode()).hexdigest(),
144 'skipadstitching
': 1,
145 'platform
': 'desktop
',
146 'rn
': random.randint(10000, 100000),
150 preplay = self._download_json(
151 'https
://vms
.vice
.com
/%s/video
/preplay
/%s' % (locale, video_id),
152 video_id, query=query)
153 except ExtractorError as e:
154 if isinstance(e.cause, compat_HTTPError) and e.cause.code in (400, 401):
155 error = json.loads(e.cause.read().decode())
156 error_message = error.get('error_description
') or error['details
']
157 raise ExtractorError('%s said
: %s' % (
158 self.IE_NAME, error_message), expected=True)
161 video_data = preplay['video
']
162 formats = self._extract_m3u8_formats(
163 preplay['playURL
'], video_id, 'mp4
', 'm3u8_native
')
164 self._sort_formats(formats)
165 episode = video_data.get('episode
') or {}
166 channel = video_data.get('channel
') or {}
167 season = video_data.get('season
') or {}
170 for subtitle in preplay.get('subtitleURLs
', []):
171 cc_url = subtitle.get('url
')
174 language_code = try_get(subtitle, lambda x: x['languages
'][0]['language_code
'], compat_str) or 'en
'
175 subtitles.setdefault(language_code, []).append({
183 'description
': clean_html(video.get('body
')),
184 'thumbnail
': video.get('thumbnail_url
'),
185 'duration
': int_or_none(video_data.get('video_duration
')),
186 'timestamp
': int_or_none(video_data.get('created_at
'), 1000),
187 'age_limit
': parse_age_limit(video_data.get('video_rating
') or rating),
188 'series
': try_get(video_data, lambda x: x['show
']['base
']['display_title
'], compat_str),
189 'episode_number
': int_or_none(episode.get('episode_number
')),
190 'episode_id
': str_or_none(episode.get('id') or video_data.get('episode_id
')),
191 'season_number
': int_or_none(season.get('season_number
')),
192 'season_id
': str_or_none(season.get('id') or video_data.get('season_id
')),
193 'uploader
': channel.get('name
'),
194 'uploader_id
': str_or_none(channel.get('id')),
195 'subtitles
': subtitles,
199 class ViceShowIE(ViceBaseIE):
200 IE_NAME = 'vice
:show
'
201 _VALID_URL = r'https?
://(?
:video\
.vice|
(?
:www\
.)?
vice(?
:land|tv
))\
.com
/(?P
<locale
>[^
/]+)/show
/(?P
<id>[^
/?
#&]+)'
204 'url': 'https://video.vice.com/en_us/show/fck-thats-delicious',
206 'id': '57a2040c8cb727dec794c901',
207 'title': 'F*ck, That’s Delicious',
208 'description': 'The life and eating habits of rap’s greatest bon vivant, Action Bronson.',
210 'playlist_mincount': 64,
212 'url': 'https://www.vicetv.com/en_us/show/fck-thats-delicious',
213 'only_matching': True,
216 def _fetch_page(self
, locale
, show_id
, page
):
217 videos
= self
._call
_api
('videos', 'show_id', show_id
, locale
, '''body
219 url''', ', page: %d, per_page: %d' % (page
+ 1, self
._PAGE
_SIZE
))
221 yield self
.url_result(
222 video
['url'], ViceIE
.ie_key(), video
.get('id'))
224 def _real_extract(self
, url
):
225 locale
, display_id
= self
._match
_valid
_url
(url
).groups()
226 show
= self
._call
_api
('shows', 'slug', display_id
, locale
, '''dek
231 entries
= OnDemandPagedList(
232 functools
.partial(self
._fetch
_page
, locale
, show_id
),
235 return self
.playlist_result(
236 entries
, show_id
, show
.get('title'), show
.get('dek'))
239 class ViceArticleIE(ViceBaseIE
):
240 IE_NAME
= 'vice:article'
241 _VALID_URL
= r
'https://(?:www\.)?vice\.com/(?P<locale>[^/]+)/article/(?:[0-9a-z]{6}/)?(?P<id>[^?#]+)'
244 'url': 'https://www.vice.com/en_us/article/on-set-with-the-woman-making-mormon-porn-in-utah',
246 'id': '58dc0a3dee202d2a0ccfcbd8',
248 'title': 'Mormon War on Porn',
249 'description': 'md5:1c5d91fe25fa8aa304f9def118b92dbf',
251 'uploader_id': '57a204088cb727dec794c67b',
252 'timestamp': 1491883129,
253 'upload_date': '20170411',
258 'skip_download': True,
260 'add_ie': [ViceIE
.ie_key()],
262 'url': 'https://www.vice.com/en_us/article/how-to-hack-a-car',
263 'md5': '13010ee0bc694ea87ec40724397c2349',
267 'title': 'How to Hack a Car: Phreaked Out (Episode 2)',
268 'description': 'md5:ee95453f7ff495db8efe14ae8bf56f30',
269 'uploader': 'Motherboard',
270 'uploader_id': 'MotherboardTV',
271 'upload_date': '20140529',
273 'add_ie': [YoutubeIE
.ie_key()],
275 'url': 'https://www.vice.com/en_us/article/znm9dx/karley-sciortino-slutever-reloaded',
276 'md5': 'a7ecf64ee4fa19b916c16f4b56184ae2',
278 'id': '57f41d3556a0a80f54726060',
280 'title': "Making The World's First Male Sex Doll",
281 'description': 'md5:19b00b215b99961cf869c40fbe9df755',
283 'uploader_id': '57a204088cb727dec794c67b',
284 'timestamp': 1476919911,
285 'upload_date': '20161019',
289 'skip_download': True,
291 'add_ie': [ViceIE
.ie_key()],
293 'url': 'https://www.vice.com/en_us/article/cowboy-capitalists-part-1',
294 'only_matching': True,
296 'url': 'https://www.vice.com/ru/article/big-night-out-ibiza-clive-martin-229',
297 'only_matching': True,
300 def _real_extract(self
, url
):
301 locale
, display_id
= self
._match
_valid
_url
(url
).groups()
303 article
= self
._call
_api
('articles', 'slug', display_id
, locale
, '''body
305 body
= article
['body']
307 def _url_res(video_url
, ie_key
):
309 '_type': 'url_transparent',
311 'display_id': display_id
,
315 vice_url
= ViceIE
._extract
_url
(body
)
317 return _url_res(vice_url
, ViceIE
.ie_key())
319 embed_code
= self
._search
_regex
(
320 r
'embedCode=([^&\'"]+)', body,
321 'ooyala embed code', default=None)
323 return _url_res('ooyala:%s' % embed_code, 'Ooyala')
325 youtube_url = YoutubeIE._extract_url(body)
327 return _url_res(youtube_url, YoutubeIE.ie_key())
329 video_url = self._html_search_regex(
330 r'data-video-url="([^
"]+)"',
331 article['embed_code
'], 'video URL
')
333 return _url_res(video_url, ViceIE.ie_key())