]>
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 | |
1cc79574 PH |
9 | from ..compat import ( |
10 | compat_urllib_parse_urlparse, | |
5d8afe69 | 11 | compat_urlparse, |
1cc79574 PH |
12 | ) |
13 | from ..utils import ( | |
64892c0b | 14 | clean_html, |
1cc79574 | 15 | ExtractorError, |
64892c0b | 16 | int_or_none, |
f05d0e73 | 17 | float_or_none, |
1cc79574 | 18 | parse_duration, |
bc03228a | 19 | determine_ext, |
5d8afe69 | 20 | ) |
6f96e308 | 21 | from .dailymotion import DailymotionCloudIE |
5d8afe69 PR |
22 | |
23 | ||
648d25d4 | 24 | class FranceTVBaseInfoExtractor(InfoExtractor): |
64892c0b S |
25 | def _extract_video(self, video_id, catalogue): |
26 | info = self._download_json( | |
27 | 'http://webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/?idDiffusion=%s&catalogue=%s' | |
28 | % (video_id, catalogue), | |
29 | video_id, 'Downloading video JSON') | |
30 | ||
31 | if info.get('status') == 'NOK': | |
32 | raise ExtractorError( | |
33 | '%s returned error: %s' % (self.IE_NAME, info['message']), expected=True) | |
00e9d396 JMF |
34 | allowed_countries = info['videos'][0].get('geoblocage') |
35 | if allowed_countries: | |
36 | georestricted = True | |
37 | geo_info = self._download_json( | |
38 | 'http://geo.francetv.fr/ws/edgescape.json', video_id, | |
39 | 'Downloading geo restriction info') | |
40 | country = geo_info['reponse']['geo_info']['country_code'] | |
41 | if country not in allowed_countries: | |
42 | raise ExtractorError( | |
43 | 'The video is not available from your location', | |
44 | expected=True) | |
45 | else: | |
46 | georestricted = False | |
47 | ||
64892c0b S |
48 | formats = [] |
49 | for video in info['videos']: | |
50 | if video['statut'] != 'ONLINE': | |
51 | continue | |
52 | video_url = video['url'] | |
53 | if not video_url: | |
54 | continue | |
55 | format_id = video['format'] | |
bc03228a S |
56 | ext = determine_ext(video_url) |
57 | if ext == 'f4m': | |
00e9d396 JMF |
58 | if georestricted: |
59 | # See https://github.com/rg3/youtube-dl/issues/3963 | |
60 | # m3u8 urls work fine | |
61 | continue | |
64892c0b S |
62 | video_url_parsed = compat_urllib_parse_urlparse(video_url) |
63 | f4m_url = self._download_webpage( | |
99e6833c | 64 | 'http://hdfauth.francetv.fr/esi/TA?url=%s' % video_url_parsed.path, |
64892c0b S |
65 | video_id, 'Downloading f4m manifest token', fatal=False) |
66 | if f4m_url: | |
bc03228a S |
67 | formats.extend(self._extract_f4m_formats(f4m_url, video_id, 1, format_id)) |
68 | elif ext == 'm3u8': | |
69 | formats.extend(self._extract_m3u8_formats(video_url, video_id, 'mp4', m3u8_id=format_id)) | |
64892c0b S |
70 | elif video_url.startswith('rtmp'): |
71 | formats.append({ | |
72 | 'url': video_url, | |
73 | 'format_id': 'rtmp-%s' % format_id, | |
74 | 'ext': 'flv', | |
75 | 'preference': 1, | |
76 | }) | |
77 | else: | |
78 | formats.append({ | |
79 | 'url': video_url, | |
80 | 'format_id': format_id, | |
2399535f | 81 | 'preference': -1, |
64892c0b S |
82 | }) |
83 | self._sort_formats(formats) | |
648d25d4 | 84 | |
7e8d73c1 JMF |
85 | return { |
86 | 'id': video_id, | |
64892c0b S |
87 | 'title': info['titre'], |
88 | 'description': clean_html(info['synopsis']), | |
89 | 'thumbnail': compat_urlparse.urljoin('http://pluzz.francetv.fr', info['image']), | |
f05d0e73 | 90 | 'duration': float_or_none(info.get('real_duration'), 1000) or parse_duration(info['duree']), |
64892c0b | 91 | 'timestamp': int_or_none(info['diffusion']['timestamp']), |
7e8d73c1 | 92 | 'formats': formats, |
7e8d73c1 | 93 | } |
648d25d4 JMF |
94 | |
95 | ||
96 | class PluzzIE(FranceTVBaseInfoExtractor): | |
3c1b4669 | 97 | IE_NAME = 'pluzz.francetv.fr' |
5d8afe69 PR |
98 | _VALID_URL = r'https?://pluzz\.francetv\.fr/videos/(.*?)\.html' |
99 | ||
5d13df79 | 100 | # Can't use tests, videos expire in 7 days |
5d8afe69 PR |
101 | |
102 | def _real_extract(self, url): | |
103 | title = re.match(self._VALID_URL, url).group(1) | |
104 | webpage = self._download_webpage(url, title) | |
105 | video_id = self._search_regex( | |
106 | r'data-diffusion="(\d+)"', webpage, 'ID') | |
64892c0b | 107 | return self._extract_video(video_id, 'Pluzz') |
5d8afe69 | 108 | |
5d8afe69 | 109 | |
648d25d4 | 110 | class FranceTvInfoIE(FranceTVBaseInfoExtractor): |
3c1b4669 | 111 | IE_NAME = 'francetvinfo.fr' |
23d3c422 | 112 | _VALID_URL = r'https?://(?:www|mobile)\.francetvinfo\.fr/.*/(?P<title>.+)\.html' |
5d8afe69 | 113 | |
5c30b268 | 114 | _TESTS = [{ |
3c1b4669 | 115 | 'url': 'http://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-lundi-26-aout-2013_393427.html', |
3c1b4669 | 116 | 'info_dict': { |
5c30b268 | 117 | 'id': '84981923', |
64892c0b | 118 | 'ext': 'flv', |
3c1b4669 | 119 | 'title': 'Soir 3', |
64892c0b S |
120 | 'upload_date': '20130826', |
121 | 'timestamp': 1377548400, | |
648d25d4 | 122 | }, |
5c30b268 PH |
123 | }, { |
124 | 'url': 'http://www.francetvinfo.fr/elections/europeennes/direct-europeennes-regardez-le-debat-entre-les-candidats-a-la-presidence-de-la-commission_600639.html', | |
125 | 'info_dict': { | |
126 | 'id': 'EV_20019', | |
127 | 'ext': 'mp4', | |
128 | 'title': 'Débat des candidats à la Commission européenne', | |
129 | 'description': 'Débat des candidats à la Commission européenne', | |
130 | }, | |
131 | 'params': { | |
132 | 'skip_download': 'HLS (reqires ffmpeg)' | |
64892c0b S |
133 | }, |
134 | 'skip': 'Ce direct est terminé et sera disponible en rattrapage dans quelques minutes.', | |
6f96e308 YCH |
135 | }, { |
136 | 'url': 'http://www.francetvinfo.fr/economie/entreprises/les-entreprises-familiales-le-secret-de-la-reussite_933271.html', | |
137 | 'md5': 'f485bda6e185e7d15dbc69b72bae993e', | |
138 | 'info_dict': { | |
139 | 'id': '556e03339473995ee145930c', | |
140 | 'ext': 'mp4', | |
141 | 'title': 'Les entreprises familiales : le secret de la réussite', | |
142 | 'thumbnail': 're:^https?://.*\.jpe?g$', | |
143 | } | |
5c30b268 | 144 | }] |
648d25d4 JMF |
145 | |
146 | def _real_extract(self, url): | |
147 | mobj = re.match(self._VALID_URL, url) | |
148 | page_title = mobj.group('title') | |
149 | webpage = self._download_webpage(url, page_title) | |
6f96e308 YCH |
150 | |
151 | dmcloud_url = DailymotionCloudIE._extract_dmcloud_url(webpage) | |
152 | if dmcloud_url: | |
153 | return self.url_result(dmcloud_url, 'DailymotionCloud') | |
154 | ||
64892c0b S |
155 | video_id, catalogue = self._search_regex( |
156 | r'id-video=([^@]+@[^"]+)', webpage, 'video id').split('@') | |
157 | return self._extract_video(video_id, catalogue) | |
a825f330 JMF |
158 | |
159 | ||
9e606020 | 160 | class FranceTVIE(FranceTVBaseInfoExtractor): |
3c1b4669 PH |
161 | IE_NAME = 'francetv' |
162 | IE_DESC = 'France 2, 3, 4, 5 and Ô' | |
9e606020 | 163 | _VALID_URL = r'''(?x)https?://www\.france[2345o]\.fr/ |
da0a5d2d | 164 | (?: |
9e606020 | 165 | emissions/.*?/(videos|emissions)/(?P<id>[^/?]+) |
3514813d | 166 | | (emissions?|jt)/(?P<key>[^/?]+) |
da0a5d2d | 167 | )''' |
a825f330 | 168 | |
9e606020 JMF |
169 | _TESTS = [ |
170 | # france2 | |
171 | { | |
3c1b4669 | 172 | 'url': 'http://www.france2.fr/emissions/13h15-le-samedi-le-dimanche/videos/75540104', |
64892c0b | 173 | 'md5': 'c03fc87cb85429ffd55df32b9fc05523', |
3c1b4669 | 174 | 'info_dict': { |
64892c0b S |
175 | 'id': '109169362', |
176 | 'ext': 'flv', | |
177 | 'title': '13h15, le dimanche...', | |
178 | 'description': 'md5:9a0932bb465f22d377a449be9d1a0ff7', | |
179 | 'upload_date': '20140914', | |
180 | 'timestamp': 1410693600, | |
9e606020 | 181 | }, |
a825f330 | 182 | }, |
9e606020 JMF |
183 | # france3 |
184 | { | |
3c1b4669 | 185 | 'url': 'http://www.france3.fr/emissions/pieces-a-conviction/diffusions/13-11-2013_145575', |
64892c0b | 186 | 'md5': '679bb8f8921f8623bd658fa2f8364da0', |
3c1b4669 PH |
187 | 'info_dict': { |
188 | 'id': '000702326_CAPP_PicesconvictionExtrait313022013_120220131722_Au', | |
64892c0b | 189 | 'ext': 'mp4', |
3c1b4669 PH |
190 | 'title': 'Le scandale du prix des médicaments', |
191 | 'description': 'md5:1384089fbee2f04fc6c9de025ee2e9ce', | |
64892c0b S |
192 | 'upload_date': '20131113', |
193 | 'timestamp': 1384380000, | |
9e606020 | 194 | }, |
a825f330 | 195 | }, |
9e606020 JMF |
196 | # france4 |
197 | { | |
3c1b4669 | 198 | 'url': 'http://www.france4.fr/emissions/hero-corp/videos/rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', |
64892c0b | 199 | 'md5': 'a182bf8d2c43d88d46ec48fbdd260c1c', |
3c1b4669 PH |
200 | 'info_dict': { |
201 | 'id': 'rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', | |
64892c0b | 202 | 'ext': 'mp4', |
3c1b4669 PH |
203 | 'title': 'Hero Corp Making of - Extrait 1', |
204 | 'description': 'md5:c87d54871b1790679aec1197e73d650a', | |
64892c0b S |
205 | 'upload_date': '20131106', |
206 | 'timestamp': 1383766500, | |
9e606020 JMF |
207 | }, |
208 | }, | |
209 | # france5 | |
210 | { | |
3c1b4669 | 211 | 'url': 'http://www.france5.fr/emissions/c-a-dire/videos/92837968', |
64892c0b | 212 | 'md5': '78f0f4064f9074438e660785bbf2c5d9', |
3c1b4669 | 213 | 'info_dict': { |
64892c0b S |
214 | 'id': '108961659', |
215 | 'ext': 'flv', | |
3c1b4669 | 216 | 'title': 'C à dire ?!', |
64892c0b S |
217 | 'description': 'md5:1a4aeab476eb657bf57c4ff122129f81', |
218 | 'upload_date': '20140915', | |
219 | 'timestamp': 1410795000, | |
9e606020 JMF |
220 | }, |
221 | }, | |
222 | # franceo | |
223 | { | |
3c1b4669 | 224 | 'url': 'http://www.franceo.fr/jt/info-afrique/04-12-2013', |
64892c0b | 225 | 'md5': '52f0bfe202848b15915a2f39aaa8981b', |
3c1b4669 | 226 | 'info_dict': { |
64892c0b S |
227 | 'id': '108634970', |
228 | 'ext': 'flv', | |
229 | 'title': 'Infô Afrique', | |
3c1b4669 | 230 | 'description': 'md5:ebf346da789428841bee0fd2a935ea55', |
64892c0b S |
231 | 'upload_date': '20140915', |
232 | 'timestamp': 1410822000, | |
9e606020 | 233 | }, |
9e606020 JMF |
234 | }, |
235 | ] | |
a825f330 JMF |
236 | |
237 | def _real_extract(self, url): | |
238 | mobj = re.match(self._VALID_URL, url) | |
64892c0b S |
239 | webpage = self._download_webpage(url, mobj.group('key') or mobj.group('id')) |
240 | video_id, catalogue = self._html_search_regex( | |
241 | r'href="http://videos\.francetv\.fr/video/([^@]+@[^"]+)"', | |
242 | webpage, 'video ID').split('@') | |
243 | return self._extract_video(video_id, catalogue) | |
5b333c1c JMF |
244 | |
245 | ||
246 | class GenerationQuoiIE(InfoExtractor): | |
3c1b4669 | 247 | IE_NAME = 'france2.fr:generation-quoi' |
c4e817ce | 248 | _VALID_URL = r'https?://generation-quoi\.france2\.fr/portrait/(?P<id>[^/?#]+)' |
5b333c1c JMF |
249 | |
250 | _TEST = { | |
3c1b4669 | 251 | 'url': 'http://generation-quoi.france2.fr/portrait/garde-a-vous', |
3c1b4669 | 252 | 'info_dict': { |
c4e817ce PH |
253 | 'id': 'k7FJX8VBcvvLmX4wA5Q', |
254 | 'ext': 'mp4', | |
3c1b4669 PH |
255 | 'title': 'Génération Quoi - Garde à Vous', |
256 | 'uploader': 'Génération Quoi', | |
5b333c1c | 257 | }, |
3c1b4669 | 258 | 'params': { |
5b333c1c | 259 | # It uses Dailymotion |
3c1b4669 | 260 | 'skip_download': True, |
5b333c1c JMF |
261 | }, |
262 | } | |
263 | ||
264 | def _real_extract(self, url): | |
c4e817ce PH |
265 | display_id = self._match_id(url) |
266 | info_url = compat_urlparse.urljoin(url, '/medias/video/%s.json' % display_id) | |
267 | info_json = self._download_webpage(info_url, display_id) | |
5b333c1c JMF |
268 | info = json.loads(info_json) |
269 | return self.url_result('http://www.dailymotion.com/video/%s' % info['id'], | |
9e1a5b84 | 270 | ie='Dailymotion') |
469ec941 JMF |
271 | |
272 | ||
273 | class CultureboxIE(FranceTVBaseInfoExtractor): | |
3c1b4669 | 274 | IE_NAME = 'culturebox.francetvinfo.fr' |
23d3c422 | 275 | _VALID_URL = r'https?://(?:m\.)?culturebox\.francetvinfo\.fr/(?P<name>.*?)(\?|$)' |
469ec941 JMF |
276 | |
277 | _TEST = { | |
aed2d4b3 | 278 | 'url': 'http://culturebox.francetvinfo.fr/live/musique/musique-classique/le-livre-vermeil-de-montserrat-a-la-cathedrale-delne-214511', |
ac651e97 | 279 | 'md5': '9b88dc156781c4dbebd4c3e066e0b1d6', |
3c1b4669 | 280 | 'info_dict': { |
aed2d4b3 | 281 | 'id': 'EV_50111', |
ac651e97 | 282 | 'ext': 'flv', |
aed2d4b3 S |
283 | 'title': "Le Livre Vermeil de Montserrat à la Cathédrale d'Elne", |
284 | 'description': 'md5:f8a4ad202e8fe533e2c493cc12e739d9', | |
285 | 'upload_date': '20150320', | |
286 | 'timestamp': 1426892400, | |
287 | 'duration': 2760.9, | |
288 | }, | |
469ec941 JMF |
289 | } |
290 | ||
291 | def _real_extract(self, url): | |
292 | mobj = re.match(self._VALID_URL, url) | |
293 | name = mobj.group('name') | |
184a1974 | 294 | |
469ec941 | 295 | webpage = self._download_webpage(url, name) |
184a1974 S |
296 | |
297 | if ">Ce live n'est plus disponible en replay<" in webpage: | |
298 | raise ExtractorError('Video %s is not available' % name, expected=True) | |
299 | ||
64892c0b S |
300 | video_id, catalogue = self._search_regex( |
301 | r'"http://videos\.francetv\.fr/video/([^@]+@[^"]+)"', webpage, 'video id').split('@') | |
302 | ||
303 | return self._extract_video(video_id, catalogue) |