3 from __future__
import unicode_literals
6 from .common
import InfoExtractor
14 from .dailymotion
import DailymotionIE
17 class FranceTVBaseInfoExtractor(InfoExtractor
):
18 def _make_url_result(self
, video_or_full_id
, catalog
=None):
19 full_id
= 'francetv:%s' % video_or_full_id
20 if '@' not in video_or_full_id
and catalog
:
21 full_id
+= '@%s' % catalog
22 return self
.url_result(
23 full_id
, ie
=FranceTVIE
.ie_key(),
24 video_id
=video_or_full_id
.split('@')[0])
27 class FranceTVIE(InfoExtractor
):
31 sivideo\.webservices\.francetelevisions\.fr/tools/getInfosOeuvre/v2/\?
32 .*?\bidDiffusion=[^&]+|
34 https?://videos\.francetv\.fr/video/|
37 (?P<id>[^@]+)(?:@(?P<catalog>.+))?
43 'url': 'https://sivideo.webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/?idDiffusion=162311093&callback=_jsonp_loader_callback_request_0',
44 'md5': 'c2248a8de38c4e65ea8fae7b5df2d84f',
48 'title': '13h15, le dimanche... - Les mystères de Jésus',
49 'description': 'md5:75efe8d4c0a8205e5904498ffe1e1a42',
50 'timestamp': 1502623500,
51 'upload_date': '20170813',
55 'url': 'https://sivideo.webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/?idDiffusion=NI_1004933&catalogue=Zouzous&callback=_jsonp_loader_callback_request_4',
56 'only_matching': True,
58 'url': 'http://videos.francetv.fr/video/NI_657393@Regions',
59 'only_matching': True,
61 'url': 'francetv:162311093',
62 'only_matching': True,
64 'url': 'francetv:NI_1004933@Zouzous',
65 'only_matching': True,
67 'url': 'francetv:NI_983319@Info-web',
68 'only_matching': True,
70 'url': 'francetv:NI_983319',
71 'only_matching': True,
73 'url': 'francetv:NI_657393@Regions',
74 'only_matching': True,
77 'url': 'francetv:SIM_France3',
78 'only_matching': True,
81 def _extract_video(self
, video_id
, catalogue
=None):
82 # Videos are identified by idDiffusion so catalogue part is optional.
83 # However when provided, some extra formats may be returned so we pass
94 for device_type
in ('desktop', 'mobile'):
95 dinfo
= self
._download
_json
(
96 'https://player.webservices.francetelevisions.fr/v1/videos/%s' % video_id
,
97 video_id
, 'Downloading %s video JSON' % device_type
, query
={
98 'device_type': device_type
,
105 video
= dinfo
.get('video')
109 duration
= video
.get('duration')
111 is_live
= video
.get('is_live')
112 if spritesheets
is None:
113 spritesheets
= video
.get('spritesheets')
115 meta
= dinfo
.get('meta')
118 title
= meta
.get('title')
119 # XXX: what is meta['pre_title']?
121 subtitle
= meta
.get('additional_title')
123 image
= meta
.get('image_url')
124 if timestamp
is None:
125 timestamp
= parse_iso8601(meta
.get('broadcasted_at'))
130 format_id
= video
.get('format')
133 if video
.get('workflow') == 'token-akamai':
134 token_url
= video
.get('token')
136 token_json
= self
._download
_json
(
138 'Downloading signed %s manifest URL' % format_id
)
140 video_url
= token_json
.get('url')
142 video_url
= video
.get('url')
144 ext
= determine_ext(video_url
)
146 formats
.extend(self
._extract
_f
4m
_formats
(
147 video_url
, video_id
, f4m_id
=format_id
, fatal
=False))
149 fmts
, subs
= self
._extract
_m
3u8_formats
_and
_subtitles
(
150 video_url
, video_id
, 'mp4',
151 entry_protocol
='m3u8_native', m3u8_id
=format_id
,
154 self
._merge
_subtitles
(subs
, target
=subtitles
)
156 fmts
, subs
= self
._extract
_mpd
_formats
_and
_subtitles
(
157 video_url
, video_id
, mpd_id
=format_id
, fatal
=False)
159 self
._merge
_subtitles
(subs
, target
=subtitles
)
160 elif video_url
.startswith('rtmp'):
163 'format_id': 'rtmp-%s' % format_id
,
167 if self
._is
_valid
_url
(video_url
, video_id
, format_id
):
170 'format_id': format_id
,
173 # XXX: what is video['captions']?
176 if f
.get('acodec') != 'none' and f
.get('language') in ('qtz', 'qad'):
177 f
['language_preference'] = -10
178 f
['format_note'] = 'audio description%s' % format_field(f
, 'format_note', ', %s')
182 'format_id': 'spritesheets',
183 'format_note': 'storyboard',
188 'url': 'about:dummy',
191 # XXX: not entirely accurate; each spritesheet seems to be
192 # a 10×10 grid of thumbnails corresponding to approximately
193 # 2 seconds of the video; the last spritesheet may be shorter
195 } for sheet
in spritesheets
]
198 self
._sort
_formats
(formats
)
201 title
+= ' - %s' % subtitle
202 title
= title
.strip()
206 'title': self
._live
_title
(title
) if is_live
else title
,
208 'duration': duration
,
209 'timestamp': timestamp
,
212 'subtitles': subtitles
,
215 def _real_extract(self
, url
):
216 mobj
= self
._match
_valid
_url
(url
)
217 video_id
= mobj
.group('id')
218 catalog
= mobj
.group('catalog')
222 video_id
= qs
.get('idDiffusion', [None])[0]
223 catalog
= qs
.get('catalogue', [None])[0]
225 raise ExtractorError('Invalid URL', expected
=True)
227 return self
._extract
_video
(video_id
, catalog
)
230 class FranceTVSiteIE(FranceTVBaseInfoExtractor
):
231 _VALID_URL
= r
'https?://(?:(?:www\.)?france\.tv|mobile\.france\.tv)/(?:[^/]+/)*(?P<id>[^/]+)\.html'
234 'url': 'https://www.france.tv/france-2/13h15-le-dimanche/140921-les-mysteres-de-jesus.html',
236 'id': 'ec217ecc-0733-48cf-ac06-af1347b849d1',
238 'title': '13h15, le dimanche... - Les mystères de Jésus',
239 'description': 'md5:75efe8d4c0a8205e5904498ffe1e1a42',
240 'timestamp': 1502623500,
241 'upload_date': '20170813',
244 'skip_download': True,
246 'add_ie': [FranceTVIE
.ie_key()],
249 'url': 'https://www.france.tv/france-3/des-chiffres-et-des-lettres/139063-emission-du-mardi-9-mai-2017.html',
250 'only_matching': True,
253 'url': 'https://www.france.tv/france-4/hero-corp/saison-1/134151-apres-le-calme.html',
254 'only_matching': True,
257 'url': 'https://www.france.tv/france-5/c-a-dire/saison-10/137013-c-a-dire.html',
258 'only_matching': True,
261 'url': 'https://www.france.tv/france-o/archipels/132249-mon-ancetre-l-esclave.html',
262 'only_matching': True,
265 'url': 'https://www.france.tv/france-2/direct.html',
266 'only_matching': True,
268 'url': 'https://www.france.tv/documentaires/histoire/136517-argentine-les-500-bebes-voles-de-la-dictature.html',
269 'only_matching': True,
271 'url': 'https://www.france.tv/jeux-et-divertissements/divertissements/133965-le-web-contre-attaque.html',
272 'only_matching': True,
274 'url': 'https://mobile.france.tv/france-5/c-dans-l-air/137347-emission-du-vendredi-12-mai-2017.html',
275 'only_matching': True,
277 'url': 'https://www.france.tv/142749-rouge-sang.html',
278 'only_matching': True,
281 'url': 'https://www.france.tv/france-3/direct.html',
282 'only_matching': True,
285 def _real_extract(self
, url
):
286 display_id
= self
._match
_id
(url
)
288 webpage
= self
._download
_webpage
(url
, display_id
)
291 video_id
= self
._search
_regex
(
292 r
'(?:data-main-video\s*=|videoId["\']?\s
*[:=])\s
*(["\'])(?P<id>(?:(?!\1).)+)\1',
293 webpage, 'video id', default=None, group='id')
296 video_id, catalogue = self._html_search_regex(
297 r'(?:href=|player\.setVideo\(\s*)"http
://videos?\
.francetv\
.fr
/video
/([^
@]+@[^
"]+)"',
298 webpage, 'video ID
').split('@')
300 return self._make_url_result(video_id, catalogue)
303 class FranceTVInfoIE(FranceTVBaseInfoExtractor):
304 IE_NAME = 'francetvinfo
.fr
'
305 _VALID_URL = r'https?
://(?
:www|mobile|france3
-regions
)\
.francetvinfo\
.fr
/(?
:[^
/]+/)*(?P
<id>[^
/?
#&.]+)'
308 'url': 'https://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-jeudi-22-aout-2019_3561461.html',
310 'id': 'd12458ee-5062-48fe-bfdd-a30d6a01b793',
313 'upload_date': '20190822',
314 'timestamp': 1566510900,
315 'description': 'md5:72d167097237701d6e8452ff03b83c00',
321 'skip_download': True,
323 'add_ie': [FranceTVIE
.ie_key()],
325 'note': 'Only an image exists in initial webpage instead of the video',
326 'url': 'https://www.francetvinfo.fr/sante/maladie/coronavirus/covid-19-en-inde-une-situation-catastrophique-a-new-dehli_4381095.html',
328 'id': '7d204c9e-a2d3-11eb-9e4c-000d3a23d482',
330 'title': 'Covid-19 : une situation catastrophique à New Dehli',
333 'timestamp': 1619028518,
334 'upload_date': '20210421',
337 'skip_download': True,
339 'add_ie': [FranceTVIE
.ie_key()],
341 'url': 'http://www.francetvinfo.fr/elections/europeennes/direct-europeennes-regardez-le-debat-entre-les-candidats-a-la-presidence-de-la-commission_600639.html',
342 'only_matching': True,
344 'url': 'http://www.francetvinfo.fr/economie/entreprises/les-entreprises-familiales-le-secret-de-la-reussite_933271.html',
345 'only_matching': True,
347 'url': 'http://france3-regions.francetvinfo.fr/bretagne/cotes-d-armor/thalassa-echappee-breizh-ce-venredi-dans-les-cotes-d-armor-954961.html',
348 'only_matching': True,
351 'url': 'http://www.francetvinfo.fr/politique/notre-dame-des-landes/video-sur-france-inter-cecile-duflot-denonce-le-regard-meprisant-de-patrick-cohen_1520091.html',
352 'md5': 'ee7f1828f25a648addc90cb2687b1f12',
356 'title': 'NDDL, référendum, Brexit : Cécile Duflot répond à Patrick Cohen',
357 'description': 'Au lendemain de la victoire du "oui" au référendum sur l\'aéroport de Notre-Dame-des-Landes, l\'ancienne ministre écologiste est l\'invitée de Patrick Cohen. Plus d\'info : https://www.franceinter.fr/emissions/le-7-9/le-7-9-27-juin-2016',
358 'timestamp': 1467011958,
359 'upload_date': '20160627',
360 'uploader': 'France Inter',
361 'uploader_id': 'x2q2ez',
363 'add_ie': ['Dailymotion'],
365 'url': 'http://france3-regions.francetvinfo.fr/limousin/emissions/jt-1213-limousin',
366 'only_matching': True,
368 # "<figure id=" pattern (#28792)
369 'url': 'https://www.francetvinfo.fr/culture/patrimoine/incendie-de-notre-dame-de-paris/notre-dame-de-paris-de-l-incendie-de-la-cathedrale-a-sa-reconstruction_4372291.html',
370 'only_matching': True,
373 def _real_extract(self
, url
):
374 display_id
= self
._match
_id
(url
)
376 webpage
= self
._download
_webpage
(url
, display_id
)
378 dailymotion_urls
= DailymotionIE
._extract
_urls
(webpage
)
380 return self
.playlist_result([
381 self
.url_result(dailymotion_url
, DailymotionIE
.ie_key())
382 for dailymotion_url
in dailymotion_urls
])
384 video_id
= self
._search
_regex
(
385 (r
'player\.load[^;]+src:\s*["\']([^
"\']+)',
386 r'id-video=([^@]+@[^"]+)',
387 r'<a
[^
>]+href
="(?:https?:)?//videos\.francetv\.fr/video/([^@]+@[^"]+)"',
388 r'(?:data-id|<figure[^<]+\bid)=["\']([\da
-f
]{8}
-[\da
-f
]{4}
-[\da
-f
]{4}
-[\da
-f
]{4}
-[\da
-f
]{12}
)'),
391 return self._make_url_result(video_id)