yt_dlp/extractor/rtlnl.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     int_or_none,
   4     parse_duration,
   5 )
   6
   7
   8 class RtlNlIE(InfoExtractor):
   9     IE_NAME = 'rtl.nl'
  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)']
  12     _VALID_URL = r'''(?x)
  13         https?://(?:(?:www|static)\.)?
  14         (?:
  15             rtlxl\.nl/(?:[^\#]*\#!|programma)/[^/]+/|
  16             rtl\.nl/(?:(?:system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html|embed)\b.+?\buuid=|video/)|
  17             embed\.rtl\.nl/\#uuid=
  18         )
  19         (?P<id>[0-9a-f-]+)'''
  20
  21     _TESTS = [{
  22         # new URL schema
  23         'url': 'https://www.rtlxl.nl/programma/rtl-nieuws/0bd1384d-d970-3086-98bb-5c104e10c26f',
  24         'md5': '490428f1187b60d714f34e1f2e3af0b6',
  25         'info_dict': {
  26             'id': '0bd1384d-d970-3086-98bb-5c104e10c26f',
  27             'ext': 'mp4',
  28             'title': 'RTL Nieuws',
  29             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  30             'timestamp': 1593293400,
  31             'upload_date': '20200627',
  32             'duration': 661.08,
  33         },
  34     }, {
  35         # old URL schema
  36         'url': 'http://www.rtlxl.nl/#!/rtl-nieuws-132237/82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
  37         'md5': '473d1946c1fdd050b2c0161a4b13c373',
  38         'info_dict': {
  39             'id': '82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
  40             'ext': 'mp4',
  41             'title': 'RTL Nieuws',
  42             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  43             'timestamp': 1461951000,
  44             'upload_date': '20160429',
  45             'duration': 1167.96,
  46         },
  47         'skip': '404',
  48     }, {
  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',
  52         'info_dict': {
  53             'id': '84ae5571-ac25-4225-ae0c-ef8d9efb2aed',
  54             'ext': 'mp4',
  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$',
  58             'upload_date': '20150215',
  59             'description': 'Er zijn nieuwe beelden vrijgegeven die vlak na de aanslag in Kopenhagen zijn gemaakt. Op de video is goed te zien hoe omstanders zich bekommeren om één van de slachtoffers, terwijl de eerste agenten ter plaatse komen.',
  60         },
  61     }, {
  62         # empty synopsis and missing episodes (see https://github.com/ytdl-org/youtube-dl/issues/6275)
  63         # best format available nettv
  64         'url': 'http://www.rtl.nl/system/videoplayer/derden/rtlnieuws/video_embed.html#uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a/autoplay=false',
  65         'info_dict': {
  66             'id': 'f536aac0-1dc3-4314-920e-3bd1c5b3811a',
  67             'ext': 'mp4',
  68             'title': 'RTL Nieuws - Meer beelden van overval juwelier',
  69             'thumbnail': r're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a$',
  70             'timestamp': 1437233400,
  71             'upload_date': '20150718',
  72             'duration': 30.474,
  73         },
  74         'params': {
  75             'skip_download': True,
  76         },
  77     }, {
  78         # encrypted m3u8 streams, georestricted
  79         'url': 'http://www.rtlxl.nl/#!/afl-2-257632/52a74543-c504-4cde-8aa8-ec66fe8d68a7',
  80         'only_matching': True,
  81     }, {
  82         'url': 'http://www.rtl.nl/system/videoplayer/derden/embed.html#!/uuid=bb0353b0-d6a4-1dad-90e9-18fe75b8d1f0',
  83         'only_matching': True,
  84     }, {
  85         'url': 'http://rtlxl.nl/?_ga=1.204735956.572365465.1466978370#!/rtl-nieuws-132237/3c487912-023b-49ac-903e-2c5d79f8410f',
  86         'only_matching': True,
  87     }, {
  88         'url': 'https://www.rtl.nl/video/c603c9c2-601d-4b5e-8175-64f1e942dc7d/',
  89         'only_matching': True,
  90     }, {
  91         'url': 'https://static.rtl.nl/embed/?uuid=1a2970fc-5c0b-43ff-9fdc-927e39e6d1bc&autoplay=false&publicatiepunt=rtlnieuwsnl',
  92         'only_matching': True,
  93     }, {
  94         # new embed URL schema
  95         'url': 'https://embed.rtl.nl/#uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed/autoplay=false',
  96         'only_matching': True,
  97     }]
  98
  99     def _real_extract(self, url):
 100         uuid = self._match_id(url)
 101         info = self._download_json(
 102             f'http://www.rtl.nl/system/s4m/vfd/version=2/uuid={uuid}/fmt=adaptive/',
 103             uuid)
 104
 105         material = info['material'][0]
 106         title = info['abstracts'][0]['name']
 107         subtitle = material.get('title')
 108         if subtitle:
 109             title += f' - {subtitle}'
 110         description = material.get('synopsis')
 111
 112         meta = info.get('meta', {})
 113
 114         videopath = material['videopath']
 115         m3u8_url = meta.get('videohost', 'http://manifest.us.rtl.nl') + videopath
 116
 117         formats = self._extract_m3u8_formats(
 118             m3u8_url, uuid, 'mp4', m3u8_id='hls', fatal=False)
 119
 120         thumbnails = []
 121
 122         for p in ('poster_base_url', '"thumb_base_url"'):
 123             if not meta.get(p):
 124                 continue
 125
 126             thumbnails.append({
 127                 'url': self._proto_relative_url(meta[p] + uuid),
 128                 'width': int_or_none(self._search_regex(
 129                     r'/sz=([0-9]+)', meta[p], 'thumbnail width', fatal=False)),
 130                 'height': int_or_none(self._search_regex(
 131                     r'/sz=[0-9]+x([0-9]+)',
 132                     meta[p], 'thumbnail height', fatal=False)),
 133             })
 134
 135         return {
 136             'id': uuid,
 137             'title': title,
 138             'formats': formats,
 139             'timestamp': material['original_date'],
 140             'description': description,
 141             'duration': parse_duration(material.get('duration')),
 142             'thumbnails': thumbnails,
 143         }
 144
 145
 146 class RTLLuBaseIE(InfoExtractor):
 147     _MEDIA_REGEX = {
 148         'video': r'<rtl-player\s[^>]*\bhls\s*=\s*"([^"]+)',
 149         'audio': r'<rtl-audioplayer\s[^>]*\bsrc\s*=\s*"([^"]+)',
 150         'thumbnail': r'<rtl-player\s[^>]*\bposter\s*=\s*"([^"]+)',
 151     }
 152
 153     def get_media_url(self, webpage, video_id, media_type):
 154         return self._search_regex(self._MEDIA_REGEX[media_type], webpage, f'{media_type} url', default=None)
 155
 156     def get_formats_and_subtitles(self, webpage, video_id):
 157         video_url, audio_url = self.get_media_url(webpage, video_id, 'video'), self.get_media_url(webpage, video_id, 'audio')
 158
 159         formats, subtitles = [], {}
 160         if video_url is not None:
 161             formats, subtitles = self._extract_m3u8_formats_and_subtitles(video_url, video_id)
 162         if audio_url is not None:
 163             formats.append({'url': audio_url, 'ext': 'mp3', 'vcodec': 'none'})
 164
 165         return formats, subtitles
 166
 167     def _real_extract(self, url):
 168         video_id = self._match_id(url)
 169         is_live = video_id in ('live', 'live-2', 'lauschteren')
 170
 171         # TODO: extract comment from https://www.rtl.lu/comments?status=1&order=desc&context=news|article|<video_id>
 172         # we can context from <rtl-comments context=<context> in webpage
 173         webpage = self._download_webpage(url, video_id)
 174
 175         formats, subtitles = self.get_formats_and_subtitles(webpage, video_id)
 176
 177         return {
 178             'id': video_id,
 179             'title': self._og_search_title(webpage),
 180             'description': self._og_search_description(webpage, default=None),
 181             'formats': formats,
 182             'subtitles': subtitles,
 183             'thumbnail': self.get_media_url(webpage, video_id, 'thumbnail') or self._og_search_thumbnail(webpage, default=None),
 184             'is_live': is_live,
 185         }
 186
 187
 188 class RTLLuTeleVODIE(RTLLuBaseIE):
 189     IE_NAME = 'rtl.lu:tele-vod'
 190     _VALID_URL = r'https?://(?:www\.)?rtl\.lu/(tele/(?P<slug>[\w-]+)/v/|video/)(?P<id>\d+)(\.html)?'
 191     _TESTS = [{
 192         'url': 'https://www.rtl.lu/tele/de-journal-vun-der-tele/v/3266757.html',
 193         'info_dict': {
 194             'id': '3266757',
 195             'title': 'Informatiounsversammlung Héichwaasser',
 196             'ext': 'mp4',
 197             'thumbnail': 'https://replay-assets.rtl.lu/2021/11/16/d3647fc4-470d-11ec-adc2-3a00abd6e90f_00008.jpg',
 198             'description': 'md5:b1db974408cc858c9fd241812e4a2a14',
 199         },
 200     }, {
 201         'url': 'https://www.rtl.lu/video/3295215',
 202         'info_dict': {
 203             'id': '3295215',
 204             'title': 'Kulturassisen iwwer d\'Bestandsopnam vum Lëtzebuerger Konscht',
 205             'ext': 'mp4',
 206             'thumbnail': 'https://replay-assets.rtl.lu/2022/06/28/0000_3295215_0000.jpg',
 207             'description': 'md5:85bcd4e0490aa6ec969d9bf16927437b',
 208         },
 209     }]
 210
 211
 212 class RTLLuArticleIE(RTLLuBaseIE):
 213     IE_NAME = 'rtl.lu:article'
 214     _VALID_URL = r'https?://(?:(www|5minutes|today)\.)rtl\.lu/(?:[\w-]+)/(?:[\w-]+)/a/(?P<id>\d+)\.html'
 215     _TESTS = [{
 216         # Audio-only
 217         'url': 'https://www.rtl.lu/sport/news/a/1934360.html',
 218         'info_dict': {
 219             'id': '1934360',
 220             'ext': 'mp3',
 221             'thumbnail': 'https://static.rtl.lu/rtl2008.lu/nt/p/2022/06/28/19/e4b37d66ddf00bab4c45617b91a5bb9b.jpeg',
 222             'description': 'md5:5eab4a2a911c1fff7efc1682a38f9ef7',
 223             'title': 'md5:40aa85f135578fbd549d3c9370321f99',
 224         },
 225     }, {
 226         # 5minutes
 227         'url': 'https://5minutes.rtl.lu/espace-frontaliers/frontaliers-en-questions/a/1853173.html',
 228         'info_dict': {
 229             'id': '1853173',
 230             'ext': 'mp4',
 231             'description': 'md5:ac031da0740e997a5cf4633173634fee',
 232             'title': 'md5:87e17722ed21af0f24be3243f4ec0c46',
 233             'thumbnail': 'https://replay-assets.rtl.lu/2022/01/26/screenshot_20220126104933_3274749_12b249833469b0d6e4440a1dec83cdfa.jpg',
 234         },
 235     }, {
 236         # today.lu
 237         'url': 'https://today.rtl.lu/entertainment/news/a/1936203.html',
 238         'info_dict': {
 239             'id': '1936203',
 240             'ext': 'mp4',
 241             'title': 'Once Upon A Time...zu Lëtzebuerg: The Three Witches\' Tower',
 242             'description': 'The witchy theme continues in the latest episode of Once Upon A Time...',
 243             'thumbnail': 'https://replay-assets.rtl.lu/2022/07/02/screenshot_20220702122859_3290019_412dc5185951b7f6545a4039c8be9235.jpg',
 244         },
 245     }]
 246
 247
 248 class RTLLuLiveIE(RTLLuBaseIE):
 249     _VALID_URL = r'https?://www\.rtl\.lu/(?:tele|radio)/(?P<id>live(?:-\d+)?|lauschteren)'
 250     _TESTS = [{
 251         # Tele:live
 252         'url': 'https://www.rtl.lu/tele/live',
 253         'info_dict': {
 254             'id': 'live',
 255             'ext': 'mp4',
 256             'live_status': 'is_live',
 257             'title': r're:RTL - Télé LIVE \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
 258             'thumbnail': 'https://static.rtl.lu/livestream/channel1.jpg',
 259         },
 260     }, {
 261         # Tele:live-2
 262         'url': 'https://www.rtl.lu/tele/live-2',
 263         'info_dict': {
 264             'id': 'live-2',
 265             'ext': 'mp4',
 266             'live_status': 'is_live',
 267             'title': r're:RTL - Télé LIVE \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
 268             'thumbnail': 'https://static.rtl.lu/livestream/channel2.jpg',
 269         },
 270     }, {
 271         # Radio:lauschteren
 272         'url': 'https://www.rtl.lu/radio/lauschteren',
 273         'info_dict': {
 274             'id': 'lauschteren',
 275             'ext': 'mp4',
 276             'live_status': 'is_live',
 277             'title': r're:RTL - Radio LIVE \d{4}-\d{2}-\d{2} \d{2}:\d{2}',
 278             'thumbnail': 'https://static.rtl.lu/livestream/rtlradiowebtv.jpg',
 279         },
 280     }]
 281
 282
 283 class RTLLuRadioIE(RTLLuBaseIE):
 284     _VALID_URL = r'https?://www\.rtl\.lu/radio/(?:[\w-]+)/s/(?P<id>\d+)(\.html)?'
 285     _TESTS = [{
 286         'url': 'https://www.rtl.lu/radio/5-vir-12/s/4033058.html',
 287         'info_dict': {
 288             'id': '4033058',
 289             'ext': 'mp3',
 290             'description': 'md5:f855a4f3e3235393ae47ed1db5d934b9',
 291             'title': '5 vir 12 - Stau um Stau',
 292             'thumbnail': 'https://static.rtl.lu/rtlg//2022/06/24/c9c19e5694a14be46a3647a3760e1f62.jpg',
 293         },
 294     }]