3 from __future__ import unicode_literals
8 from .common import InfoExtractor
9 from ..compat import compat_urlparse
17 from .dailymotion import (
23 class FranceTVBaseInfoExtractor(InfoExtractor):
24 def _extract_video(self, video_id, catalogue=None):
25 info = self._download_json(
26 'https://sivideo.webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/',
27 video_id, 'Downloading video JSON', query={
28 'idDiffusion': video_id,
29 'catalogue': catalogue or '',
32 if info.get('status') == 'NOK':
34 '%s returned error: %s' % (self.IE_NAME, info['message']), expected=True)
35 allowed_countries = info['videos'][0].get('geoblocage')
38 geo_info = self._download_json(
39 'http://geo.francetv.fr/ws/edgescape.json', video_id,
40 'Downloading geo restriction info')
41 country = geo_info['reponse']['geo_info']['country_code']
42 if country not in allowed_countries:
44 'The video is not available from your location',
50 for video in info['videos']:
51 if video['statut'] != 'ONLINE':
53 video_url = video['url']
56 format_id = video['format']
57 ext = determine_ext(video_url)
60 # See https://github.com/rg3/youtube-dl/issues/3963
63 f4m_url = self._download_webpage(
64 'http://hdfauth.francetv.fr/esi/TA?url=%s' % video_url,
65 video_id, 'Downloading f4m manifest token', fatal=False)
67 formats.extend(self._extract_f4m_formats(
68 f4m_url + '&hdcore=3.7.0&plugin=aasp-3.7.0.39.44',
69 video_id, f4m_id=format_id, fatal=False))
71 formats.extend(self._extract_m3u8_formats(
72 video_url, video_id, 'mp4', entry_protocol='m3u8_native',
73 m3u8_id=format_id, fatal=False))
74 elif video_url.startswith('rtmp'):
77 'format_id': 'rtmp-%s' % format_id,
81 if self._is_valid_url(video_url, video_id, format_id):
84 'format_id': format_id,
86 self._sort_formats(formats)
89 subtitle = info.get('sous_titre')
91 title += ' - %s' % subtitle
96 'url': subformat['url'],
97 'ext': subformat.get('format'),
98 } for subformat in info.get('subtitles', []) if subformat.get('url')]
100 subtitles['fr'] = subtitles_list
105 'description': clean_html(info['synopsis']),
106 'thumbnail': compat_urlparse.urljoin('http://pluzz.francetv.fr', info['image']),
107 'duration': int_or_none(info.get('real_duration')) or parse_duration(info['duree']),
108 'timestamp': int_or_none(info['diffusion']['timestamp']),
110 'subtitles': subtitles,
114 class FranceTVIE(FranceTVBaseInfoExtractor):
115 _VALID_URL = r'https?://(?:(?:www\.)?france\.tv|mobile\.france\.tv)/(?:[^/]+/)*(?P<id>[^/]+)\.html'
118 'url': 'https://www.france.tv/france-2/13h15-le-dimanche/140921-les-mysteres-de-jesus.html',
122 'title': '13h15, le dimanche... - Les mystères de Jésus',
123 'description': 'md5:75efe8d4c0a8205e5904498ffe1e1a42',
124 'timestamp': 1494156300,
125 'upload_date': '20170507',
129 'skip_download': True,
133 'url': 'https://www.france.tv/france-3/des-chiffres-et-des-lettres/139063-emission-du-mardi-9-mai-2017.html',
134 'only_matching': True,
137 'url': 'https://www.france.tv/france-4/hero-corp/saison-1/134151-apres-le-calme.html',
138 'only_matching': True,
141 'url': 'https://www.france.tv/france-5/c-a-dire/saison-10/137013-c-a-dire.html',
142 'only_matching': True,
145 'url': 'https://www.france.tv/france-o/archipels/132249-mon-ancetre-l-esclave.html',
146 'only_matching': True,
149 'url': 'https://www.france.tv/france-2/direct.html',
150 'only_matching': True,
152 'url': 'https://www.france.tv/documentaires/histoire/136517-argentine-les-500-bebes-voles-de-la-dictature.html',
153 'only_matching': True,
155 'url': 'https://www.france.tv/jeux-et-divertissements/divertissements/133965-le-web-contre-attaque.html',
156 'only_matching': True,
158 'url': 'https://mobile.france.tv/france-5/c-dans-l-air/137347-emission-du-vendredi-12-mai-2017.html',
159 'only_matching': True,
161 'url': 'https://www.france.tv/142749-rouge-sang.html',
162 'only_matching': True,
165 def _real_extract(self, url):
166 display_id = self._match_id(url)
168 webpage = self._download_webpage(url, display_id)
171 video_id = self._search_regex(
172 r'data-main-video=(["\'])(?P<id>(?:(?!\1).)+)\1',
173 webpage, 'video id', default=None, group='id')
176 video_id, catalogue = self._html_search_regex(
177 r'(?:href=|player\.setVideo\(\s*)"http://videos?\.francetv\.fr/video/([^@]+@[^"]+)"',
178 webpage, 'video ID').split('@')
179 return self._extract_video(video_id, catalogue)
182 class FranceTVEmbedIE(FranceTVBaseInfoExtractor):
183 _VALID_URL = r'https?://embed\.francetv\.fr/*\?.*?\bue=(?P<id>[^&]+)'
186 'url': 'http://embed.francetv.fr/?ue=7fd581a2ccf59d2fc5719c5c13cf6961',
190 'title': 'Le Pen Reims',
191 'upload_date': '20170505',
192 'timestamp': 1493981780,
197 def _real_extract(self, url):
198 video_id = self._match_id(url)
200 video = self._download_json(
201 'http://api-embed.webservices.francetelevisions.fr/key/%s' % video_id,
204 return self._extract_video(video['video_id'], video.get('catalog'))
207 class FranceTVInfoIE(FranceTVBaseInfoExtractor):
208 IE_NAME = 'francetvinfo.fr'
209 _VALID_URL = r'https?://(?:www|mobile|france3-regions)\.francetvinfo\.fr/(?:[^/]+/)*(?P<title>[^/?#&.]+)'
212 'url': 'http://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-lundi-26-aout-2013_393427.html',
217 'upload_date': '20130826',
218 'timestamp': 1377548400,
225 'skip_download': True,
228 'url': 'http://www.francetvinfo.fr/elections/europeennes/direct-europeennes-regardez-le-debat-entre-les-candidats-a-la-presidence-de-la-commission_600639.html',
232 'title': 'Débat des candidats à la Commission européenne',
233 'description': 'Débat des candidats à la Commission européenne',
236 'skip_download': 'HLS (reqires ffmpeg)'
238 'skip': 'Ce direct est terminé et sera disponible en rattrapage dans quelques minutes.',
240 'url': 'http://www.francetvinfo.fr/economie/entreprises/les-entreprises-familiales-le-secret-de-la-reussite_933271.html',
241 'md5': 'f485bda6e185e7d15dbc69b72bae993e',
245 'title': 'Les entreprises familiales : le secret de la réussite',
246 'thumbnail': r're:^https?://.*\.jpe?g$',
247 'timestamp': 1433273139,
248 'upload_date': '20150602',
252 'skip_download': True,
255 'url': 'http://france3-regions.francetvinfo.fr/bretagne/cotes-d-armor/thalassa-echappee-breizh-ce-venredi-dans-les-cotes-d-armor-954961.html',
256 'md5': 'f485bda6e185e7d15dbc69b72bae993e',
260 'title': 'Olivier Monthus, réalisateur de "Bretagne, le choix de l’Armor"',
261 'description': 'md5:a3264114c9d29aeca11ced113c37b16c',
262 'thumbnail': r're:^https?://.*\.jpe?g$',
263 'timestamp': 1458300695,
264 'upload_date': '20160318',
267 'skip_download': True,
271 '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',
272 'md5': 'ee7f1828f25a648addc90cb2687b1f12',
276 'title': 'NDDL, référendum, Brexit : Cécile Duflot répond à Patrick Cohen',
277 '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',
278 'timestamp': 1467011958,
279 'upload_date': '20160627',
280 'uploader': 'France Inter',
281 'uploader_id': 'x2q2ez',
283 'add_ie': ['Dailymotion'],
285 'url': 'http://france3-regions.francetvinfo.fr/limousin/emissions/jt-1213-limousin',
286 'only_matching': True,
289 def _real_extract(self, url):
290 mobj = re.match(self._VALID_URL, url)
291 page_title = mobj.group('title')
292 webpage = self._download_webpage(url, page_title)
294 dmcloud_url = DailymotionCloudIE._extract_dmcloud_url(webpage)
296 return self.url_result(dmcloud_url, DailymotionCloudIE.ie_key())
298 dailymotion_urls = DailymotionIE._extract_urls(webpage)
300 return self.playlist_result([
301 self.url_result(dailymotion_url, DailymotionIE.ie_key())
302 for dailymotion_url in dailymotion_urls])
304 video_id, catalogue = self._search_regex(
305 (r'id-video=([^@]+@[^"]+)',
306 r'<a[^>]+href="(?:https?:)?//videos\.francetv\.fr/video/([^@]+@[^"]+)"'),
307 webpage, 'video id').split('@')
308 return self._extract_video(video_id, catalogue)
311 class GenerationQuoiIE(InfoExtractor):
312 IE_NAME = 'france2.fr:generation-quoi'
313 _VALID_URL = r'https?://generation-quoi\.france2\.fr/portrait/(?P<id>[^/?#]+)'
316 'url': 'http://generation-quoi.france2.fr/portrait/garde-a-vous',
318 'id': 'k7FJX8VBcvvLmX4wA5Q',
320 'title': 'Génération Quoi - Garde à Vous',
321 'uploader': 'Génération Quoi',
324 # It uses Dailymotion
325 'skip_download': True,
329 def _real_extract(self, url):
330 display_id = self._match_id(url)
331 info_url = compat_urlparse.urljoin(url, '/medias/video/%s.json' % display_id)
332 info_json = self._download_webpage(info_url, display_id)
333 info = json.loads(info_json)
334 return self.url_result('http://www.dailymotion.com/video/%s' % info['id'],
338 class CultureboxIE(FranceTVBaseInfoExtractor):
339 IE_NAME = 'culturebox.francetvinfo.fr'
340 _VALID_URL = r'https?://(?:m\.)?culturebox\.francetvinfo\.fr/(?P<name>.*?)(\?|$)'
343 'url': 'http://culturebox.francetvinfo.fr/live/musique/musique-classique/le-livre-vermeil-de-montserrat-a-la-cathedrale-delne-214511',
344 'md5': '9b88dc156781c4dbebd4c3e066e0b1d6',
348 'title': "Le Livre Vermeil de Montserrat à la Cathédrale d'Elne",
349 'description': 'md5:f8a4ad202e8fe533e2c493cc12e739d9',
350 'upload_date': '20150320',
351 'timestamp': 1426892400,
356 def _real_extract(self, url):
357 mobj = re.match(self._VALID_URL, url)
358 name = mobj.group('name')
360 webpage = self._download_webpage(url, name)
362 if ">Ce live n'est plus disponible en replay<" in webpage:
363 raise ExtractorError('Video %s is not available' % name, expected=True)
365 video_id, catalogue = self._search_regex(
366 r'"https?://videos\.francetv\.fr/video/([^@]+@[^"]+)"', webpage, 'video id').split('@')
368 return self._extract_video(video_id, catalogue)