]>
Commit | Line | Data |
---|---|---|
5d8afe69 | 1 | # encoding: utf-8 |
3c1b4669 PH |
2 | |
3 | from __future__ import unicode_literals | |
4 | ||
5d8afe69 | 5 | import re |
5b333c1c | 6 | import json |
5d8afe69 PR |
7 | |
8 | from .common import InfoExtractor | |
1d18e26e | 9 | from ..compat import compat_urlparse |
1cc79574 | 10 | from ..utils import ( |
64892c0b | 11 | clean_html, |
1cc79574 | 12 | ExtractorError, |
64892c0b | 13 | int_or_none, |
1cc79574 | 14 | parse_duration, |
bc03228a | 15 | determine_ext, |
5d8afe69 | 16 | ) |
6f96e308 | 17 | from .dailymotion import DailymotionCloudIE |
5d8afe69 PR |
18 | |
19 | ||
648d25d4 | 20 | class FranceTVBaseInfoExtractor(InfoExtractor): |
64892c0b S |
21 | def _extract_video(self, video_id, catalogue): |
22 | info = self._download_json( | |
23 | 'http://webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/?idDiffusion=%s&catalogue=%s' | |
24 | % (video_id, catalogue), | |
25 | video_id, 'Downloading video JSON') | |
26 | ||
27 | if info.get('status') == 'NOK': | |
28 | raise ExtractorError( | |
29 | '%s returned error: %s' % (self.IE_NAME, info['message']), expected=True) | |
00e9d396 JMF |
30 | allowed_countries = info['videos'][0].get('geoblocage') |
31 | if allowed_countries: | |
32 | georestricted = True | |
33 | geo_info = self._download_json( | |
34 | 'http://geo.francetv.fr/ws/edgescape.json', video_id, | |
35 | 'Downloading geo restriction info') | |
36 | country = geo_info['reponse']['geo_info']['country_code'] | |
37 | if country not in allowed_countries: | |
38 | raise ExtractorError( | |
39 | 'The video is not available from your location', | |
40 | expected=True) | |
41 | else: | |
42 | georestricted = False | |
43 | ||
64892c0b S |
44 | formats = [] |
45 | for video in info['videos']: | |
46 | if video['statut'] != 'ONLINE': | |
47 | continue | |
48 | video_url = video['url'] | |
49 | if not video_url: | |
50 | continue | |
51 | format_id = video['format'] | |
bc03228a S |
52 | ext = determine_ext(video_url) |
53 | if ext == 'f4m': | |
00e9d396 JMF |
54 | if georestricted: |
55 | # See https://github.com/rg3/youtube-dl/issues/3963 | |
56 | # m3u8 urls work fine | |
57 | continue | |
64892c0b | 58 | f4m_url = self._download_webpage( |
ecdbe09e | 59 | 'http://hdfauth.francetv.fr/esi/TA?url=%s' % video_url, |
64892c0b S |
60 | video_id, 'Downloading f4m manifest token', fatal=False) |
61 | if f4m_url: | |
632cbb8e | 62 | formats.extend(self._extract_f4m_formats( |
3c20208e S |
63 | f4m_url + '&hdcore=3.7.0&plugin=aasp-3.7.0.39.44', |
64 | video_id, f4m_id=format_id, fatal=False)) | |
bc03228a | 65 | elif ext == 'm3u8': |
3c20208e S |
66 | formats.extend(self._extract_m3u8_formats( |
67 | video_url, video_id, 'mp4', entry_protocol='m3u8_native', | |
68 | m3u8_id=format_id, fatal=False)) | |
64892c0b S |
69 | elif video_url.startswith('rtmp'): |
70 | formats.append({ | |
71 | 'url': video_url, | |
72 | 'format_id': 'rtmp-%s' % format_id, | |
73 | 'ext': 'flv', | |
64892c0b S |
74 | }) |
75 | else: | |
3c20208e S |
76 | if self._is_valid_url(video_url, video_id, format_id): |
77 | formats.append({ | |
78 | 'url': video_url, | |
79 | 'format_id': format_id, | |
80 | }) | |
64892c0b | 81 | self._sort_formats(formats) |
648d25d4 | 82 | |
36c15522 S |
83 | title = info['titre'] |
84 | subtitle = info.get('sous_titre') | |
85 | if subtitle: | |
86 | title += ' - %s' % subtitle | |
db264e3c | 87 | title = title.strip() |
36c15522 | 88 | |
5dadae07 | 89 | subtitles = {} |
6e4b8b28 | 90 | subtitles_list = [{ |
7ccb2b84 JMF |
91 | 'url': subformat['url'], |
92 | 'ext': subformat.get('format'), | |
93 | } for subformat in info.get('subtitles', []) if subformat.get('url')] | |
6e4b8b28 S |
94 | if subtitles_list: |
95 | subtitles['fr'] = subtitles_list | |
5dadae07 | 96 | |
7e8d73c1 JMF |
97 | return { |
98 | 'id': video_id, | |
36c15522 | 99 | 'title': title, |
64892c0b S |
100 | 'description': clean_html(info['synopsis']), |
101 | 'thumbnail': compat_urlparse.urljoin('http://pluzz.francetv.fr', info['image']), | |
5705ee6e | 102 | 'duration': int_or_none(info.get('real_duration')) or parse_duration(info['duree']), |
64892c0b | 103 | 'timestamp': int_or_none(info['diffusion']['timestamp']), |
7e8d73c1 | 104 | 'formats': formats, |
5dadae07 | 105 | 'subtitles': subtitles, |
7e8d73c1 | 106 | } |
648d25d4 JMF |
107 | |
108 | ||
109 | class PluzzIE(FranceTVBaseInfoExtractor): | |
3c1b4669 | 110 | IE_NAME = 'pluzz.francetv.fr' |
0a192fbe | 111 | _VALID_URL = r'https?://(?:m\.)?pluzz\.francetv\.fr/videos/(?P<id>.+?)\.html' |
5d8afe69 | 112 | |
5d13df79 | 113 | # Can't use tests, videos expire in 7 days |
5d8afe69 PR |
114 | |
115 | def _real_extract(self, url): | |
0a192fbe S |
116 | display_id = self._match_id(url) |
117 | ||
118 | webpage = self._download_webpage(url, display_id) | |
119 | ||
120 | video_id = self._html_search_meta( | |
121 | 'id_video', webpage, 'video id', default=None) | |
122 | if not video_id: | |
123 | video_id = self._search_regex( | |
7ccb2b84 | 124 | r'data-diffusion=["\'](\d+)', webpage, 'video id') |
0a192fbe | 125 | |
64892c0b | 126 | return self._extract_video(video_id, 'Pluzz') |
5d8afe69 | 127 | |
5d8afe69 | 128 | |
648d25d4 | 129 | class FranceTvInfoIE(FranceTVBaseInfoExtractor): |
3c1b4669 | 130 | IE_NAME = 'francetvinfo.fr' |
db264e3c | 131 | _VALID_URL = r'https?://(?:www|mobile|france3-regions)\.francetvinfo\.fr/.*/(?P<title>.+)\.html' |
5d8afe69 | 132 | |
5c30b268 | 133 | _TESTS = [{ |
3c1b4669 | 134 | 'url': 'http://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-lundi-26-aout-2013_393427.html', |
3c1b4669 | 135 | 'info_dict': { |
5c30b268 | 136 | 'id': '84981923', |
3c20208e | 137 | 'ext': 'mp4', |
3c1b4669 | 138 | 'title': 'Soir 3', |
64892c0b S |
139 | 'upload_date': '20130826', |
140 | 'timestamp': 1377548400, | |
c137cc0d S |
141 | 'subtitles': { |
142 | 'fr': 'mincount:2', | |
143 | }, | |
648d25d4 | 144 | }, |
3c20208e S |
145 | 'params': { |
146 | # m3u8 downloads | |
147 | 'skip_download': True, | |
148 | }, | |
5c30b268 PH |
149 | }, { |
150 | 'url': 'http://www.francetvinfo.fr/elections/europeennes/direct-europeennes-regardez-le-debat-entre-les-candidats-a-la-presidence-de-la-commission_600639.html', | |
151 | 'info_dict': { | |
152 | 'id': 'EV_20019', | |
153 | 'ext': 'mp4', | |
154 | 'title': 'Débat des candidats à la Commission européenne', | |
155 | 'description': 'Débat des candidats à la Commission européenne', | |
156 | }, | |
157 | 'params': { | |
158 | 'skip_download': 'HLS (reqires ffmpeg)' | |
64892c0b S |
159 | }, |
160 | 'skip': 'Ce direct est terminé et sera disponible en rattrapage dans quelques minutes.', | |
6f96e308 YCH |
161 | }, { |
162 | 'url': 'http://www.francetvinfo.fr/economie/entreprises/les-entreprises-familiales-le-secret-de-la-reussite_933271.html', | |
163 | 'md5': 'f485bda6e185e7d15dbc69b72bae993e', | |
164 | 'info_dict': { | |
3c20208e | 165 | 'id': 'NI_173343', |
6f96e308 YCH |
166 | 'ext': 'mp4', |
167 | 'title': 'Les entreprises familiales : le secret de la réussite', | |
168 | 'thumbnail': 're:^https?://.*\.jpe?g$', | |
3c20208e S |
169 | 'timestamp': 1433273139, |
170 | 'upload_date': '20150602', | |
171 | }, | |
172 | 'params': { | |
173 | # m3u8 downloads | |
174 | 'skip_download': True, | |
175 | }, | |
db264e3c S |
176 | }, { |
177 | 'url': 'http://france3-regions.francetvinfo.fr/bretagne/cotes-d-armor/thalassa-echappee-breizh-ce-venredi-dans-les-cotes-d-armor-954961.html', | |
178 | 'md5': 'f485bda6e185e7d15dbc69b72bae993e', | |
179 | 'info_dict': { | |
180 | 'id': 'NI_657393', | |
3c20208e | 181 | 'ext': 'mp4', |
db264e3c S |
182 | 'title': 'Olivier Monthus, réalisateur de "Bretagne, le choix de l’Armor"', |
183 | 'description': 'md5:a3264114c9d29aeca11ced113c37b16c', | |
184 | 'thumbnail': 're:^https?://.*\.jpe?g$', | |
185 | 'timestamp': 1458300695, | |
186 | 'upload_date': '20160318', | |
187 | }, | |
188 | 'params': { | |
189 | 'skip_download': True, | |
190 | }, | |
5c30b268 | 191 | }] |
648d25d4 JMF |
192 | |
193 | def _real_extract(self, url): | |
194 | mobj = re.match(self._VALID_URL, url) | |
195 | page_title = mobj.group('title') | |
196 | webpage = self._download_webpage(url, page_title) | |
6f96e308 YCH |
197 | |
198 | dmcloud_url = DailymotionCloudIE._extract_dmcloud_url(webpage) | |
199 | if dmcloud_url: | |
200 | return self.url_result(dmcloud_url, 'DailymotionCloud') | |
201 | ||
64892c0b | 202 | video_id, catalogue = self._search_regex( |
db264e3c S |
203 | (r'id-video=([^@]+@[^"]+)', |
204 | r'<a[^>]+href="(?:https?:)?//videos\.francetv\.fr/video/([^@]+@[^"]+)"'), | |
205 | webpage, 'video id').split('@') | |
64892c0b | 206 | return self._extract_video(video_id, catalogue) |
a825f330 JMF |
207 | |
208 | ||
9e606020 | 209 | class FranceTVIE(FranceTVBaseInfoExtractor): |
3c1b4669 PH |
210 | IE_NAME = 'francetv' |
211 | IE_DESC = 'France 2, 3, 4, 5 and Ô' | |
3f5c6d0c S |
212 | _VALID_URL = r'''(?x) |
213 | https?:// | |
214 | (?: | |
215 | (?:www\.)?france[2345o]\.fr/ | |
216 | (?: | |
308c505c S |
217 | emissions/[^/]+/(?:videos|diffusions)| |
218 | emission/[^/]+| | |
789a12aa S |
219 | videos| |
220 | jt | |
3f5c6d0c S |
221 | ) |
222 | /| | |
223 | embed\.francetv\.fr/\?ue= | |
224 | ) | |
225 | (?P<id>[^/?]+) | |
226 | ''' | |
a825f330 | 227 | |
9e606020 JMF |
228 | _TESTS = [ |
229 | # france2 | |
230 | { | |
3c1b4669 | 231 | 'url': 'http://www.france2.fr/emissions/13h15-le-samedi-le-dimanche/videos/75540104', |
64892c0b | 232 | 'md5': 'c03fc87cb85429ffd55df32b9fc05523', |
3c1b4669 | 233 | 'info_dict': { |
64892c0b S |
234 | 'id': '109169362', |
235 | 'ext': 'flv', | |
236 | 'title': '13h15, le dimanche...', | |
237 | 'description': 'md5:9a0932bb465f22d377a449be9d1a0ff7', | |
238 | 'upload_date': '20140914', | |
239 | 'timestamp': 1410693600, | |
9e606020 | 240 | }, |
a825f330 | 241 | }, |
9e606020 JMF |
242 | # france3 |
243 | { | |
3c1b4669 | 244 | 'url': 'http://www.france3.fr/emissions/pieces-a-conviction/diffusions/13-11-2013_145575', |
64892c0b | 245 | 'md5': '679bb8f8921f8623bd658fa2f8364da0', |
3c1b4669 PH |
246 | 'info_dict': { |
247 | 'id': '000702326_CAPP_PicesconvictionExtrait313022013_120220131722_Au', | |
64892c0b | 248 | 'ext': 'mp4', |
3c1b4669 PH |
249 | 'title': 'Le scandale du prix des médicaments', |
250 | 'description': 'md5:1384089fbee2f04fc6c9de025ee2e9ce', | |
64892c0b S |
251 | 'upload_date': '20131113', |
252 | 'timestamp': 1384380000, | |
9e606020 | 253 | }, |
a825f330 | 254 | }, |
9e606020 JMF |
255 | # france4 |
256 | { | |
3c1b4669 | 257 | 'url': 'http://www.france4.fr/emissions/hero-corp/videos/rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', |
64892c0b | 258 | 'md5': 'a182bf8d2c43d88d46ec48fbdd260c1c', |
3c1b4669 PH |
259 | 'info_dict': { |
260 | 'id': 'rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', | |
64892c0b | 261 | 'ext': 'mp4', |
3c1b4669 PH |
262 | 'title': 'Hero Corp Making of - Extrait 1', |
263 | 'description': 'md5:c87d54871b1790679aec1197e73d650a', | |
64892c0b S |
264 | 'upload_date': '20131106', |
265 | 'timestamp': 1383766500, | |
9e606020 JMF |
266 | }, |
267 | }, | |
268 | # france5 | |
269 | { | |
77306e8b S |
270 | 'url': 'http://www.france5.fr/emissions/c-a-dire/videos/quels_sont_les_enjeux_de_cette_rentree_politique__31-08-2015_908948?onglet=tous&page=1', |
271 | 'md5': 'f6c577df3806e26471b3d21631241fd0', | |
3c1b4669 | 272 | 'info_dict': { |
77306e8b | 273 | 'id': '123327454', |
64892c0b | 274 | 'ext': 'flv', |
6917d2a2 | 275 | 'title': 'C à dire ?! - Quels sont les enjeux de cette rentrée politique ?', |
77306e8b S |
276 | 'description': 'md5:4a0d5cb5dce89d353522a84462bae5a4', |
277 | 'upload_date': '20150831', | |
278 | 'timestamp': 1441035120, | |
9e606020 JMF |
279 | }, |
280 | }, | |
281 | # franceo | |
282 | { | |
3bc9fb58 S |
283 | 'url': 'http://www.franceo.fr/jt/info-soir/18-07-2015', |
284 | 'md5': '47d5816d3b24351cdce512ad7ab31da8', | |
3c1b4669 | 285 | 'info_dict': { |
3bc9fb58 | 286 | 'id': '125377621', |
64892c0b | 287 | 'ext': 'flv', |
3bc9fb58 S |
288 | 'title': 'Infô soir', |
289 | 'description': 'md5:01b8c6915a3d93d8bbbd692651714309', | |
290 | 'upload_date': '20150718', | |
291 | 'timestamp': 1437241200, | |
292 | 'duration': 414, | |
9e606020 | 293 | }, |
9e606020 | 294 | }, |
3f5c6d0c S |
295 | { |
296 | # francetv embed | |
297 | 'url': 'http://embed.francetv.fr/?ue=8d7d3da1e3047c42ade5a5d7dfd3fc87', | |
298 | 'info_dict': { | |
299 | 'id': 'EV_30231', | |
300 | 'ext': 'flv', | |
301 | 'title': 'Alcaline, le concert avec Calogero', | |
ac4b8df5 | 302 | 'description': 'md5:61f08036dcc8f47e9cfc33aed08ffaff', |
3f5c6d0c S |
303 | 'upload_date': '20150226', |
304 | 'timestamp': 1424989860, | |
5705ee6e | 305 | 'duration': 5400, |
3f5c6d0c S |
306 | }, |
307 | }, | |
308 | { | |
309 | 'url': 'http://www.france4.fr/emission/highlander/diffusion-du-17-07-2015-04h05', | |
310 | 'only_matching': True, | |
311 | }, | |
312 | { | |
313 | 'url': 'http://www.franceo.fr/videos/125377617', | |
314 | 'only_matching': True, | |
315 | } | |
9e606020 | 316 | ] |
a825f330 JMF |
317 | |
318 | def _real_extract(self, url): | |
3f5c6d0c S |
319 | video_id = self._match_id(url) |
320 | webpage = self._download_webpage(url, video_id) | |
64892c0b | 321 | video_id, catalogue = self._html_search_regex( |
f3f9cd92 | 322 | r'(?:href=|player\.setVideo\(\s*)"http://videos?\.francetv\.fr/video/([^@]+@[^"]+)"', |
64892c0b S |
323 | webpage, 'video ID').split('@') |
324 | return self._extract_video(video_id, catalogue) | |
5b333c1c JMF |
325 | |
326 | ||
327 | class GenerationQuoiIE(InfoExtractor): | |
3c1b4669 | 328 | IE_NAME = 'france2.fr:generation-quoi' |
c4e817ce | 329 | _VALID_URL = r'https?://generation-quoi\.france2\.fr/portrait/(?P<id>[^/?#]+)' |
5b333c1c JMF |
330 | |
331 | _TEST = { | |
3c1b4669 | 332 | 'url': 'http://generation-quoi.france2.fr/portrait/garde-a-vous', |
3c1b4669 | 333 | 'info_dict': { |
c4e817ce PH |
334 | 'id': 'k7FJX8VBcvvLmX4wA5Q', |
335 | 'ext': 'mp4', | |
3c1b4669 PH |
336 | 'title': 'Génération Quoi - Garde à Vous', |
337 | 'uploader': 'Génération Quoi', | |
5b333c1c | 338 | }, |
3c1b4669 | 339 | 'params': { |
5b333c1c | 340 | # It uses Dailymotion |
3c1b4669 | 341 | 'skip_download': True, |
5b333c1c JMF |
342 | }, |
343 | } | |
344 | ||
345 | def _real_extract(self, url): | |
c4e817ce PH |
346 | display_id = self._match_id(url) |
347 | info_url = compat_urlparse.urljoin(url, '/medias/video/%s.json' % display_id) | |
348 | info_json = self._download_webpage(info_url, display_id) | |
5b333c1c JMF |
349 | info = json.loads(info_json) |
350 | return self.url_result('http://www.dailymotion.com/video/%s' % info['id'], | |
9e1a5b84 | 351 | ie='Dailymotion') |
469ec941 JMF |
352 | |
353 | ||
354 | class CultureboxIE(FranceTVBaseInfoExtractor): | |
3c1b4669 | 355 | IE_NAME = 'culturebox.francetvinfo.fr' |
23d3c422 | 356 | _VALID_URL = r'https?://(?:m\.)?culturebox\.francetvinfo\.fr/(?P<name>.*?)(\?|$)' |
469ec941 JMF |
357 | |
358 | _TEST = { | |
aed2d4b3 | 359 | 'url': 'http://culturebox.francetvinfo.fr/live/musique/musique-classique/le-livre-vermeil-de-montserrat-a-la-cathedrale-delne-214511', |
ac651e97 | 360 | 'md5': '9b88dc156781c4dbebd4c3e066e0b1d6', |
3c1b4669 | 361 | 'info_dict': { |
aed2d4b3 | 362 | 'id': 'EV_50111', |
ac651e97 | 363 | 'ext': 'flv', |
aed2d4b3 S |
364 | 'title': "Le Livre Vermeil de Montserrat à la Cathédrale d'Elne", |
365 | 'description': 'md5:f8a4ad202e8fe533e2c493cc12e739d9', | |
366 | 'upload_date': '20150320', | |
367 | 'timestamp': 1426892400, | |
368 | 'duration': 2760.9, | |
369 | }, | |
469ec941 JMF |
370 | } |
371 | ||
372 | def _real_extract(self, url): | |
373 | mobj = re.match(self._VALID_URL, url) | |
374 | name = mobj.group('name') | |
184a1974 | 375 | |
469ec941 | 376 | webpage = self._download_webpage(url, name) |
184a1974 S |
377 | |
378 | if ">Ce live n'est plus disponible en replay<" in webpage: | |
379 | raise ExtractorError('Video %s is not available' % name, expected=True) | |
380 | ||
64892c0b S |
381 | video_id, catalogue = self._search_regex( |
382 | r'"http://videos\.francetv\.fr/video/([^@]+@[^"]+)"', webpage, 'video id').split('@') | |
383 | ||
384 | return self._extract_video(video_id, catalogue) |