yt_dlp/extractor/sbs.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     smuggle_url,
   4     ExtractorError,
   5 )
   6
   7
   8 class SBSIE(InfoExtractor):
   9     IE_DESC = 'sbs.com.au'
  10     _VALID_URL = r'''(?x)
  11         https?://(?:www\.)?sbs\.com\.au/(?:
  12             ondemand(?:
  13                 /video/(?:single/)?|
  14                 /movie/[^/]+/|
  15                 /(?:tv|news)-series/(?:[^/]+/){3}|
  16                 .*?\bplay=|/watch/
  17             )|news/(?:embeds/)?video/
  18         )(?P<id>[0-9]+)'''
  19     _EMBED_REGEX = [r'''(?x)]
  20             (?:
  21                 <meta\s+property="og:video"\s+content=|
  22                 <iframe[^>]+?src=
  23             )
  24             (["\'])(?P<url>https?://(?:www\.)?sbs\.com\.au/ondemand/video/.+?)\1''']
  25
  26     _TESTS = [{
  27         # Original URL is handled by the generic IE which finds the iframe:
  28         # http://www.sbs.com.au/thefeed/blog/2014/08/21/dingo-conservation
  29         'url': 'http://www.sbs.com.au/ondemand/video/single/320403011771/?source=drupal&vertical=thefeed',
  30         'md5': '3150cf278965eeabb5b4cea1c963fe0a',
  31         'info_dict': {
  32             'id': '_rFBPRPO4pMR',
  33             'ext': 'mp4',
  34             'title': 'Dingo Conservation (The Feed)',
  35             'description': 'md5:f250a9856fca50d22dec0b5b8015f8a5',
  36             'thumbnail': r're:http://.*\.jpg',
  37             'duration': 308,
  38             'timestamp': 1408613220,
  39             'upload_date': '20140821',
  40             'uploader': 'SBSC',
  41         },
  42     }, {
  43         'url': 'http://www.sbs.com.au/ondemand/video/320403011771/Dingo-Conservation-The-Feed',
  44         'only_matching': True,
  45     }, {
  46         'url': 'http://www.sbs.com.au/news/video/471395907773/The-Feed-July-9',
  47         'only_matching': True,
  48     }, {
  49         'url': 'https://www.sbs.com.au/ondemand/?play=1836638787723',
  50         'only_matching': True,
  51     }, {
  52         'url': 'https://www.sbs.com.au/ondemand/program/inside-windsor-castle?play=1283505731842',
  53         'only_matching': True,
  54     }, {
  55         'url': 'https://www.sbs.com.au/news/embeds/video/1840778819866',
  56         'only_matching': True,
  57     }, {
  58         'url': 'https://www.sbs.com.au/ondemand/watch/1698704451971',
  59         'only_matching': True,
  60     }, {
  61         'url': 'https://www.sbs.com.au/ondemand/movie/coherence/1469404227931',
  62         'only_matching': True,
  63     }, {
  64         'note': 'Live stream',
  65         'url': 'https://www.sbs.com.au/ondemand/video/1726824003663/sbs-24x7-live-stream-nsw',
  66         'only_matching': True,
  67     }, {
  68         'url': 'https://www.sbs.com.au/ondemand/news-series/dateline/dateline-2022/dateline-s2022-ep26/2072245827515',
  69         'only_matching': True,
  70     }, {
  71         'url': 'https://www.sbs.com.au/ondemand/tv-series/the-handmaids-tale/season-5/the-handmaids-tale-s5-ep1/2065631811776',
  72         'only_matching': True,
  73     }]
  74
  75     def _real_extract(self, url):
  76         video_id = self._match_id(url)
  77         player_params = self._download_json(
  78             'http://www.sbs.com.au/api/video_pdkvars/id/%s?form=json' % video_id, video_id)
  79
  80         error = player_params.get('error')
  81         if error:
  82             error_message = 'Sorry, The video you are looking for does not exist.'
  83             video_data = error.get('results') or {}
  84             error_code = error.get('errorCode')
  85             if error_code == 'ComingSoon':
  86                 error_message = '%s is not yet available.' % video_data.get('title', '')
  87             elif error_code in ('Forbidden', 'intranetAccessOnly'):
  88                 error_message = 'Sorry, This video cannot be accessed via this website'
  89             elif error_code == 'Expired':
  90                 error_message = 'Sorry, %s is no longer available.' % video_data.get('title', '')
  91             raise ExtractorError('%s said: %s' % (self.IE_NAME, error_message), expected=True)
  92
  93         urls = player_params['releaseUrls']
  94         theplatform_url = (urls.get('progressive') or urls.get('html')
  95                            or urls.get('standard') or player_params['relatedItemsURL'])
  96
  97         return {
  98             '_type': 'url_transparent',
  99             'ie_key': 'ThePlatform',
 100             'id': video_id,
 101             'url': smuggle_url(self._proto_relative_url(theplatform_url), {'force_smil_url': True}),
 102             'is_live': player_params.get('streamType') == 'live',
 103         }