]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/rcs.py
3 from .common
import InfoExtractor
14 class RCSBaseIE(InfoExtractor
):
15 # based on VideoPlayerLoader.prototype.getVideoSrc
16 # and VideoPlayerLoader.prototype.transformSrc from
17 # https://js2.corriereobjects.it/includes2013/LIBS/js/corriere_video.sjs
19 'media2vam.corriere.it.edgesuite.net':
20 'media2vam-corriere-it.akamaized.net',
21 'media.youreporter.it.edgesuite.net':
22 'media-youreporter-it.akamaized.net',
23 'corrierepmd.corriere.it.edgesuite.net':
24 'corrierepmd-corriere-it.akamaized.net',
25 'media2vam-corriere-it.akamaized.net/fcs.quotidiani/vr/videos/':
26 'video.corriere.it/vr360/videos/',
30 'media2vam.corbologna.corriere.it.edgesuite.net':
31 'media2vam-bologna-corriere-it.akamaized.net',
32 'media2vam.corfiorentino.corriere.it.edgesuite.net':
33 'media2vam-fiorentino-corriere-it.akamaized.net',
34 'media2vam.cormezzogiorno.corriere.it.edgesuite.net':
35 'media2vam-mezzogiorno-corriere-it.akamaized.net',
36 'media2vam.corveneto.corriere.it.edgesuite.net':
37 'media2vam-veneto-corriere-it.akamaized.net',
38 'media2.oggi.it.edgesuite.net':
39 'media2-oggi-it.akamaized.net',
40 'media2.quimamme.it.edgesuite.net':
41 'media2-quimamme-it.akamaized.net',
42 'media2.amica.it.edgesuite.net':
43 'media2-amica-it.akamaized.net',
44 'media2.living.corriere.it.edgesuite.net':
45 'media2-living-corriere-it.akamaized.net',
46 'media2.style.corriere.it.edgesuite.net':
47 'media2-style-corriere-it.akamaized.net',
48 'media2.iodonna.it.edgesuite.net':
49 'media2-iodonna-it.akamaized.net',
50 'media2.leitv.it.edgesuite.net':
51 'media2-leitv-it.akamaized.net',
54 'videoamica-vh.akamaihd': 'amica',
55 'media2-amica-it.akamaized': 'amica',
56 'corrierevam-vh.akamaihd': 'corriere',
57 'media2vam-corriere-it.akamaized': 'corriere',
58 'cormezzogiorno-vh.akamaihd': 'corrieredelmezzogiorno',
59 'media2vam-mezzogiorno-corriere-it.akamaized': 'corrieredelmezzogiorno',
60 'corveneto-vh.akamaihd': 'corrieredelveneto',
61 'media2vam-veneto-corriere-it.akamaized': 'corrieredelveneto',
62 'corbologna-vh.akamaihd': 'corrieredibologna',
63 'media2vam-bologna-corriere-it.akamaized': 'corrieredibologna',
64 'corfiorentino-vh.akamaihd': 'corrierefiorentino',
65 'media2vam-fiorentino-corriere-it.akamaized': 'corrierefiorentino',
66 'corinnovazione-vh.akamaihd': 'corriereinnovazione',
67 'media2-gazzanet-gazzetta-it.akamaized': 'gazzanet',
68 'videogazzanet-vh.akamaihd': 'gazzanet',
69 'videogazzaworld-vh.akamaihd': 'gazzaworld',
70 'gazzettavam-vh.akamaihd': 'gazzetta',
71 'media2vam-gazzetta-it.akamaized': 'gazzetta',
72 'videoiodonna-vh.akamaihd': 'iodonna',
73 'media2-leitv-it.akamaized': 'leitv',
74 'videoleitv-vh.akamaihd': 'leitv',
75 'videoliving-vh.akamaihd': 'living',
76 'media2-living-corriere-it.akamaized': 'living',
77 'media2-oggi-it.akamaized': 'oggi',
78 'videooggi-vh.akamaihd': 'oggi',
79 'media2-quimamme-it.akamaized': 'quimamme',
80 'quimamme-vh.akamaihd': 'quimamme',
81 'videorunning-vh.akamaihd': 'running',
82 'media2-style-corriere-it.akamaized': 'style',
83 'style-vh.akamaihd': 'style',
84 'videostyle-vh.akamaihd': 'style',
85 'media2-stylepiccoli-it.akamaized': 'stylepiccoli',
86 'stylepiccoli-vh.akamaihd': 'stylepiccoli',
87 'doveviaggi-vh.akamaihd': 'viaggi',
88 'media2-doveviaggi-it.akamaized': 'viaggi',
89 'media2-vivimilano-corriere-it.akamaized': 'vivimilano',
90 'vivimilano-vh.akamaihd': 'vivimilano',
91 'media2-youreporter-it.akamaized': 'youreporter'
94 'advrcs-vh.akamaihd': '',
95 'corriere-f.akamaihd': '',
96 'corrierepmd-corriere-it.akamaized': '',
97 'corrprotetto-vh.akamaihd': '',
98 'gazzetta-f.akamaihd': '',
99 'gazzettapmd-gazzetta-it.akamaized': '',
100 'gazzprotetto-vh.akamaihd': '',
101 'periodici-f.akamaihd': '',
102 'periodicisecure-vh.akamaihd': '',
103 'videocoracademy-vh.akamaihd': ''
106 def _get_video_src(self
, video
):
107 mediaFiles
= video
.get('mediaProfile').get('mediaFile')
110 if video
.get('mediaType') == 'AUDIO':
111 for aud
in mediaFiles
:
113 src
['mp3'] = aud
.get('value')
116 for vid
in mediaFiles
:
117 if vid
.get('mimeType') == 'application/vnd.apple.mpegurl':
118 src
['m3u8'] = vid
.get('value')
119 if vid
.get('mimeType') == 'video/mp4':
120 src
['mp4'] = vid
.get('value')
124 for s
, r
in self
._ALL
_REPLACE
.items():
125 src
[t
] = src
[t
].replace(s
, r
)
126 for s
, r
in self
._MP
4_REPLACE
.items():
127 src
[t
] = src
[t
].replace(s
, r
)
130 if 'mp4' in src
and 'm3u8' in src
:
131 if ('-lh.akamaihd' not in src
.get('m3u8')
132 and 'akamai' in src
.get('mp4')):
134 matches
= re
.search(r
'(?:https*:)?\/\/(?P<host>.*)\.net\/i(?P<path>.*)$', src
.get('m3u8'))
135 src
['m3u8'] = 'https://vod.rcsobjects.it/hls/%s%s' % (
136 self
._MIGRATION
_MAP
[matches
.group('host')],
137 matches
.group('path').replace(
144 matches
= re
.search(r
'(?:https*:)?\/\/(?P<host>.*)\.net\/i(?P<path>.*)$', src
.get('mp4'))
146 if matches
.group('host') in self
._MIGRATION
_MEDIA
:
147 vh_stream
= 'https://media2.corriereobjects.it'
148 if src
.get('mp4').find('fcs.quotidiani_!'):
149 vh_stream
= 'https://media2-it.corriereobjects.it'
150 src
['mp4'] = '%s%s' % (
152 matches
.group('path').replace(
155 '/fcs.quotidiani/mediacenter', '').replace(
156 '/fcs.quotidiani_!/mediacenter', '').replace(
157 'corriere/content/mediacenter/', '').replace(
158 'gazzetta/content/mediacenter/', '')
161 src
['mp4'] = 'https://vod.rcsobjects.it/%s%s' % (
162 self
._MIGRATION
_MAP
[matches
.group('host')],
163 matches
.group('path').replace('///', '/').replace('//', '/')
167 src
['mp3'] = src
.get('mp3').replace(
168 'media2vam-corriere-it.akamaized.net',
169 'vod.rcsobjects.it/corriere')
171 if src
.get('mp4').find('fcs.quotidiani_!'):
172 src
['mp4'] = src
.get('mp4').replace('vod.rcsobjects', 'vod-it.rcsobjects')
174 if src
.get('m3u8').find('fcs.quotidiani_!'):
175 src
['m3u8'] = src
.get('m3u8').replace('vod.rcsobjects', 'vod-it.rcsobjects')
177 if 'geoblocking' in video
.get('mediaProfile'):
179 src
['m3u8'] = src
.get('m3u8').replace('vod.rcsobjects', 'vod-it.rcsobjects')
181 src
['mp4'] = src
.get('mp4').replace('vod.rcsobjects', 'vod-it.rcsobjects')
183 if src
.get('m3u8').find('csmil') and src
.get('m3u8').find('vod'):
184 src
['m3u8'] = src
.get('m3u8').replace('.csmil', '.urlset')
188 def _create_formats(self
, urls
, video_id
):
190 formats
= self
._extract
_m
3u8_formats
(
191 urls
.get('m3u8'), video_id
, 'mp4', entry_protocol
='m3u8_native',
192 m3u8_id
='hls', fatal
=False)
196 'format_id': 'http-mp4',
199 self
._sort
_formats
(formats
)
202 def _real_extract(self
, url
):
203 mobj
= self
._match
_valid
_url
(url
)
204 video_id
= mobj
.group('id')
206 if 'cdn' not in mobj
.groupdict():
207 raise ExtractorError('CDN not found in url: %s' % url
)
209 # for leitv/youreporter/viaggi don't use the embed page
210 if ((mobj
.group('cdn') not in ['leitv.it', 'youreporter.it'])
211 and (mobj
.group('vid') == 'video')):
212 url
= 'https://video.%s/video-embed/%s' % (mobj
.group('cdn'), video_id
)
214 page
= self
._download
_webpage
(url
, video_id
)
217 # look for json video data url
218 json
= self
._search
_regex
(
219 r
'''(?x)url\s*=\s*(["'])
221 (?:https?:)?//video\.rcs\.it
222 /fragment-includes/video-includes/.+?\.json
224 page
, video_id
, group
='url', default
=None)
226 if json
.startswith('//'):
227 json
= 'https:%s' % json
228 video_data
= self
._download
_json
(json
, video_id
)
230 # if json url not found, look for json video data directly in the page
232 # RCS normal pages and most of the embeds
233 json
= self
._search
_regex
(
234 r
'[\s;]video\s*=\s*({[\s\S]+?})(?:;|,playlist=)',
235 page
, video_id
, default
=None)
236 if not json
and 'video-embed' in url
:
237 page
= self
._download
_webpage
(url
.replace('video-embed', 'video-json'), video_id
)
238 json
= self
._search
_regex
(
239 r
'##start-video##({[\s\S]+?})##end-video##',
240 page
, video_id
, default
=None)
242 # if no video data found try search for iframes
243 emb
= RCSEmbedsIE
._extract
_url
(page
)
246 '_type': 'url_transparent',
248 'ie_key': RCSEmbedsIE
.ie_key()
251 video_data
= self
._parse
_json
(
252 json
, video_id
, transform_source
=js_to_json
)
255 raise ExtractorError('Video data not found in the page')
257 formats
= self
._create
_formats
(
258 self
._get
_video
_src
(video_data
), video_id
)
260 description
= (video_data
.get('description')
261 or clean_html(video_data
.get('htmlDescription'))
262 or self
._html
_search
_meta
('description', page
))
263 uploader
= video_data
.get('provider') or mobj
.group('cdn')
267 'title': video_data
.get('title'),
268 'description': description
,
269 'uploader': uploader
,
274 class RCSEmbedsIE(RCSBaseIE
):
275 _VALID_URL
= r
'''(?x)
276 https?://(?P<vid>video)\.
280 (?:corriere\w+\.)?corriere|
281 (?:gazzanet\.)?gazzetta
283 /video-embed/(?P<id>[^/=&\?]+?)(?:$|\?)'''
285 'url': 'https://video.rcs.it/video-embed/iodonna-0001585037',
286 'md5': '623ecc8ffe7299b2d0c1046d8331a9df',
288 'id': 'iodonna-0001585037',
290 'title': 'Sky Arte racconta Madonna nella serie "Artist to icon"',
291 'description': 'md5:65b09633df9ffee57f48b39e34c9e067',
292 'uploader': 'rcs.it',
295 # redownload the page changing 'video-embed' in 'video-json'
296 'url': 'https://video.gazzanet.gazzetta.it/video-embed/gazzanet-mo05-0000260789',
297 'md5': 'a043e3fecbe4d9ed7fc5d888652a5440',
299 'id': 'gazzanet-mo05-0000260789',
301 'title': 'Valentino Rossi e papà Graziano si divertono col drifting',
302 'description': 'md5:a8bf90d6adafd9815f70fc74c0fc370a',
306 'url': 'https://video.corriere.it/video-embed/b727632a-f9d0-11ea-91b0-38d50a849abb?player',
309 'url': 'https://video.gazzetta.it/video-embed/49612410-00ca-11eb-bcd8-30d4253e0140',
314 def _sanitize_urls(urls
):
315 # add protocol if missing
316 for i
, e
in enumerate(urls
):
317 if e
.startswith('//'):
318 urls
[i
] = 'https:%s' % e
320 for i
, e
in enumerate(urls
):
321 urls
[i
] = urljoin(base_url(e
), url_basename(e
))
325 def _extract_urls(webpage
):
328 for mobj
in re
.finditer(r
'''(?x)
334 (?P<url>(?:https?:)?//video\.
337 (?:corriere\w+\.)?corriere|
338 (?:gazzanet\.)?gazzetta
340 \.it/video-embed/.+?)
342 return RCSEmbedsIE
._sanitize
_urls
(entries
)
345 def _extract_url(webpage
):
346 urls
= RCSEmbedsIE
._extract
_urls
(webpage
)
347 return urls
[0] if urls
else None
350 class RCSIE(RCSBaseIE
):
351 _VALID_URL
= r
'''(?x)https?://(?P<vid>video|viaggi)\.
354 corrieredelmezzogiorno\.
357 |corrierefiorentino\.
359 |(?:gazzanet\.)?gazzetta\.it)
360 /(?!video-embed/).+?/(?P<id>[^/\?]+)(?=\?|/$|$)'''
362 'url': 'https://video.corriere.it/sport/formula-1/vettel-guida-ferrari-sf90-mugello-suo-fianco-c-elecrerc-bendato-video-esilarante/b727632a-f9d0-11ea-91b0-38d50a849abb',
363 'md5': '0f4ededc202b0f00b6e509d831e2dcda',
365 'id': 'b727632a-f9d0-11ea-91b0-38d50a849abb',
367 'title': 'Vettel guida la Ferrari SF90 al Mugello e al suo fianco c\'è Leclerc (bendato): il video è esilarante',
368 'description': 'md5:93b51c9161ac8a64fb2f997b054d0152',
369 'uploader': 'Corriere Tv',
372 # video data inside iframe
373 'url': 'https://viaggi.corriere.it/video/norvegia-il-nuovo-ponte-spettacolare-sopra-la-cascata-di-voringsfossen/',
374 'md5': 'da378e4918d2afbf7d61c35abb948d4c',
376 'id': '5b7cd134-e2c1-11ea-89b3-b56dd0df2aa2',
378 'title': 'La nuova spettacolare attrazione in Norvegia: il ponte sopra Vøringsfossen',
379 'description': 'md5:18b35a291f6746c0c8dacd16e5f5f4f8',
380 'uploader': 'DOVE Viaggi',
383 'url': 'https://video.gazzetta.it/video-motogp-catalogna-cadute-dovizioso-vale-rossi/49612410-00ca-11eb-bcd8-30d4253e0140?vclk=Videobar',
384 'md5': 'eedc1b5defd18e67383afef51ff7bdf9',
386 'id': '49612410-00ca-11eb-bcd8-30d4253e0140',
388 'title': 'Dovizioso, il contatto con Zarco e la caduta. E anche Vale finisce a terra',
389 'description': 'md5:8c6e905dc3b9413218beca11ebd69778',
390 'uploader': 'AMorici',
393 'url': 'https://video.corriere.it/video-360/metro-copenaghen-tutta-italiana/a248a7f0-e2db-11e9-9830-af2de6b1f945',
398 class RCSVariousIE(RCSBaseIE
):
399 _VALID_URL
= r
'''(?x)https?://www\.
403 )/(?:[^/]+/)?(?P<id>[^/]+?)(?:$|\?|/)'''
405 'url': 'https://www.leitv.it/benessere/mal-di-testa-come-combatterlo-ed-evitarne-la-comparsa/',
406 'md5': '92b4e63667b8f95acb0a04da25ae28a1',
408 'id': 'mal-di-testa-come-combatterlo-ed-evitarne-la-comparsa',
410 'title': 'Cervicalgia e mal di testa, il video con i suggerimenti dell\'esperto',
411 'description': 'md5:ae21418f34cee0b8d02a487f55bcabb5',
412 'uploader': 'leitv.it',
415 'url': 'https://www.youreporter.it/fiume-sesia-3-ottobre-2020/',
416 'md5': '8dccd436b47a830bab5b4a88232f391a',
418 'id': 'fiume-sesia-3-ottobre-2020',
420 'title': 'Fiume Sesia 3 ottobre 2020',
421 'description': 'md5:0070eef1cc884d13c970a4125063de55',
422 'uploader': 'youreporter.it',