]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/rtlnl.py
1 from . common
import InfoExtractor
8 class RtlNlIE ( InfoExtractor
):
10 IE_DESC
= 'rtl.nl and rtlxl.nl'
11 _EMBED_REGEX
= [ r
'<iframe[^>]+?\bsrc=(?P<q1>[\' "])(?P<url>(?:https?:)?//(?:(?:www|static)\.)?rtl\.nl/(?:system/videoplayer/[^" ]+( ?
: video_
) ?
) ?embed
[ ^
"]+)(?P=q1)']
13 https?://(?:(?:www|static)\.)?
15 rtlxl\.nl/(?:[^\#]*\#!|programma)/[^/]+/|
16 rtl\.nl/(?:(?:system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html|embed) \b .+? \b uuid=|video/)|
17 embed\.rtl\.nl/\#uuid=
23 'url': 'https://www.rtlxl.nl/programma/rtl-nieuws/0bd1384d-d970-3086-98bb-5c104e10c26f',
24 'md5': '490428f1187b60d714f34e1f2e3af0b6',
26 'id': '0bd1384d-d970-3086-98bb-5c104e10c26f',
28 'title': 'RTL Nieuws',
29 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
30 'timestamp': 1593293400,
31 'upload_date': '20200627',
36 'url': 'http://www.rtlxl.nl/#!/rtl-nieuws-132237/82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
37 'md5': '473d1946c1fdd050b2c0161a4b13c373',
39 'id': '82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
41 'title': 'RTL Nieuws',
42 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
43 'timestamp': 1461951000,
44 'upload_date': '20160429',
49 # best format available a3t
50 'url': 'http://www.rtl.nl/system/videoplayer/derden/rtlnieuws/video_embed.html#uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed/autoplay=false',
51 'md5': 'dea7474214af1271d91ef332fb8be7ea',
53 'id': '84ae5571-ac25-4225-ae0c-ef8d9efb2aed',
55 'timestamp': 1424039400,
56 'title': 'RTL Nieuws - Nieuwe beelden Kopenhagen: chaos direct na aanslag',
57 'thumbnail': r're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed$',
58 'upload_date': '20150215',
59 'description': 'Er zijn nieuwe beelden vrijgegeven die vlak na de aanslag in Kopenhagen zijn gemaakt. Op de video is goed te zien hoe omstanders zich bekommeren om één van de slachtoffers, terwijl de eerste agenten ter plaatse komen.',
62 # empty synopsis and missing episodes (see https://github.com/ytdl-org/youtube-dl/issues/6275)
63 # best format available nettv
64 'url': 'http://www.rtl.nl/system/videoplayer/derden/rtlnieuws/video_embed.html#uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a/autoplay=false',
66 'id': 'f536aac0-1dc3-4314-920e-3bd1c5b3811a',
68 'title': 'RTL Nieuws - Meer beelden van overval juwelier',
69 'thumbnail': r're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a$',
70 'timestamp': 1437233400,
71 'upload_date': '20150718',
75 'skip_download': True,
78 # encrypted m3u8 streams, georestricted
79 'url': 'http://www.rtlxl.nl/#!/afl-2-257632/52a74543-c504-4cde-8aa8-ec66fe8d68a7',
80 'only_matching': True,
82 'url': 'http://www.rtl.nl/system/videoplayer/derden/embed.html#!/uuid=bb0353b0-d6a4-1dad-90e9-18fe75b8d1f0',
83 'only_matching': True,
85 'url': 'http://rtlxl.nl/?_ga=1.204735956.572365465.1466978370#!/rtl-nieuws-132237/3c487912-023b-49ac-903e-2c5d79f8410f',
86 'only_matching': True,
88 'url': 'https://www.rtl.nl/video/c603c9c2-601d-4b5e-8175-64f1e942dc7d/',
89 'only_matching': True,
91 'url': 'https://static.rtl.nl/embed/?uuid=1a2970fc-5c0b-43ff-9fdc-927e39e6d1bc&autoplay=false&publicatiepunt=rtlnieuwsnl',
92 'only_matching': True,
94 # new embed URL schema
95 'url': 'https://embed.rtl.nl/#uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed/autoplay=false',
96 'only_matching': True,
99 def _real_extract(self, url):
100 uuid = self._match_id(url)
101 info = self._download_json(
102 'http://www.rtl.nl/system/s4m/vfd/version=2/uuid= %s /fmt=adaptive/' % uuid,
105 material = info['material'][0]
106 title = info['abstracts'][0]['name']
107 subtitle = material.get('title')
109 title += ' - %s ' % subtitle
110 description = material.get('synopsis')
112 meta = info.get('meta', {})
114 videopath = material['videopath']
115 m3u8_url = meta.get('videohost', 'http://manifest.us.rtl.nl') + videopath
117 formats = self._extract_m3u8_formats(
118 m3u8_url, uuid, 'mp4', m3u8_id='hls', fatal=False)
119 self._sort_formats(formats)
123 for p in ('poster_base_url', '" thumb_base_url
"'):
128 'url': self._proto_relative_url(meta[p] + uuid),
129 'width': int_or_none(self._search_regex(
130 r'/sz=([0-9]+)', meta[p], 'thumbnail width', fatal=False)),
131 'height': int_or_none(self._search_regex(
132 r'/sz=[0-9]+x([0-9]+)',
133 meta[p], 'thumbnail height', fatal=False))
140 'timestamp': material['original_date'],
141 'description': description,
142 'duration': parse_duration(material.get('duration')),
143 'thumbnails': thumbnails,
147 class RTLLuBaseIE(InfoExtractor):
149 'video': r'<rtl-player\s[^>]* \b hls\s*=\s*" ([ ^
"]+)',
150 'audio': r'<rtl-audioplayer\s[^>]* \b src\s*=\s*" ([ ^
"]+)',
151 'thumbnail': r'<rtl-player\s[^>]* \b poster\s*=\s*" ([ ^
"]+)',
154 def get_media_url(self, webpage, video_id, media_type):
155 return self._search_regex(self._MEDIA_REGEX[media_type], webpage, f' {media_type} url', default=None)
157 def get_formats_and_subtitles(self, webpage, video_id):
158 video_url, audio_url = self.get_media_url(webpage, video_id, 'video'), self.get_media_url(webpage, video_id, 'audio')
160 formats, subtitles = [], {}
161 if video_url is not None:
162 formats, subtitles = self._extract_m3u8_formats_and_subtitles(video_url, video_id)
163 if audio_url is not None:
164 formats.append( {'url': audio_url, 'ext': 'mp3', 'vcodec': 'none'} )
166 return formats, subtitles
168 def _real_extract(self, url):
169 video_id = self._match_id(url)
170 is_live = video_id in ('live', 'live-2', 'lauschteren')
172 # TODO: extract comment from https://www.rtl.lu/comments?status=1&order=desc&context=news|article|<video_id>
173 # we can context from <rtl-comments context=<context> in webpage
174 webpage = self._download_webpage(url, video_id)
176 formats, subtitles = self.get_formats_and_subtitles(webpage, video_id)
177 self._sort_formats(formats)
181 'title': self._og_search_title(webpage),
182 'description': self._og_search_description(webpage, default=None),
184 'subtitles': subtitles,
185 'thumbnail': self.get_media_url(webpage, video_id, 'thumbnail') or self._og_search_thumbnail(webpage, default=None),
190 class RTLLuTeleVODIE(RTLLuBaseIE):
191 IE_NAME = 'rtl.lu:tele-vod'
192 _VALID_URL = r'https?://(?:www\.)?rtl\.lu/(tele/(?P<slug>[\w-]+)/v/|video/)(?P<id>\d+)(\.html)?'
194 'url': 'https://www.rtl.lu/tele/de-journal-vun-der-tele/v/3266757.html',
197 'title': 'Informatiounsversammlung Héichwaasser',
199 'thumbnail': 'https://replay-assets.rtl.lu/2021/11/16/d3647fc4-470d-11ec-adc2-3a00abd6e90f_00008.jpg',
200 'description': 'md5:b1db974408cc858c9fd241812e4a2a14',
203 'url': 'https://www.rtl.lu/video/3295215',
206 'title': 'Kulturassisen iwwer d \' Bestandsopnam vum Lëtzebuerger Konscht',
208 'thumbnail': 'https://replay-assets.rtl.lu/2022/06/28/0000_3295215_0000.jpg',
209 'description': 'md5:85bcd4e0490aa6ec969d9bf16927437b',
214 class RTLLuArticleIE(RTLLuBaseIE):
215 IE_NAME = 'rtl.lu:article'
216 _VALID_URL = r'https?://(?:(www|5minutes|today)\.)rtl\.lu/(?:[\w-]+)/(?:[\w-]+)/a/(?P<id>\d+)\.html'
219 'url': 'https://www.rtl.lu/sport/news/a/1934360.html',
223 'thumbnail': 'https://static.rtl.lu/rtl2008.lu/nt/p/2022/06/28/19/e4b37d66ddf00bab4c45617b91a5bb9b.jpeg',
224 'description': 'md5:5eab4a2a911c1fff7efc1682a38f9ef7',
225 'title': 'md5:40aa85f135578fbd549d3c9370321f99',
229 'url': 'https://5minutes.rtl.lu/espace-frontaliers/frontaliers-en-questions/a/1853173.html',
233 'description': 'md5:ac031da0740e997a5cf4633173634fee',
234 'title': 'md5:87e17722ed21af0f24be3243f4ec0c46',
235 'thumbnail': 'https://replay-assets.rtl.lu/2022/01/26/screenshot_20220126104933_3274749_12b249833469b0d6e4440a1dec83cdfa.jpg',
239 'url': 'https://today.rtl.lu/entertainment/news/a/1936203.html',
243 'title': 'Once Upon A Time...zu Lëtzebuerg: The Three Witches \' Tower',
244 'description': 'The witchy theme continues in the latest episode of Once Upon A Time...',
245 'thumbnail': 'https://replay-assets.rtl.lu/2022/07/02/screenshot_20220702122859_3290019_412dc5185951b7f6545a4039c8be9235.jpg',
250 class RTLLuLiveIE(RTLLuBaseIE):
251 _VALID_URL = r'https?://www\.rtl\.lu/(?:tele|radio)/(?P<id>live(?:-\d+)?|lauschteren)'
254 'url': 'https://www.rtl.lu/tele/live',
258 'live_status': 'is_live',
259 'title': r're:RTL - Télé LIVE \d {4} -\d {2} -\d {2} \d {2} :\d {2} ',
260 'thumbnail': 'https://static.rtl.lu/livestream/channel1.jpg',
264 'url': 'https://www.rtl.lu/tele/live-2',
268 'live_status': 'is_live',
269 'title': r're:RTL - Télé LIVE \d {4} -\d {2} -\d {2} \d {2} :\d {2} ',
270 'thumbnail': 'https://static.rtl.lu/livestream/channel2.jpg',
274 'url': 'https://www.rtl.lu/radio/lauschteren',
278 'live_status': 'is_live',
279 'title': r're:RTL - Radio LIVE \d {4} -\d {2} -\d {2} \d {2} :\d {2} ',
280 'thumbnail': 'https://static.rtl.lu/livestream/rtlradiowebtv.jpg',
285 class RTLLuRadioIE(RTLLuBaseIE):
286 _VALID_URL = r'https?://www\.rtl\.lu/radio/(?:[\w-]+)/s/(?P<id>\d+)(\.html)?'
288 'url': 'https://www.rtl.lu/radio/5-vir-12/s/4033058.html',
292 'description': 'md5:f855a4f3e3235393ae47ed1db5d934b9',
293 'title': '5 vir 12 - Stau um Stau',
294 'thumbnail': 'https://static.rtl.lu/rtlg//2022/06/24/c9c19e5694a14be46a3647a3760e1f62.jpg',