yt_dlp/extractor/sbs.py

   1 from .common import InfoExtractor
   2 from ..networking import HEADRequest
   3 from ..utils import (
   4     float_or_none,
   5     int_or_none,
   6     parse_duration,
   7     parse_iso8601,
   8     traverse_obj,
   9     update_url_query,
  10     url_or_none,
  11 )
  12
  13
  14 class SBSIE(InfoExtractor):
  15     IE_DESC = 'sbs.com.au'
  16     _VALID_URL = r'''(?x)
  17         https?://(?:www\.)?sbs\.com\.au/(?:
  18             ondemand(?:
  19                 /video/(?:single/)?|
  20                 /(?:movie|tv-program)/[^/]+/|
  21                 /(?:tv|news)-series/(?:[^/]+/){3}|
  22                 .*?\bplay=|/watch/
  23             )|news/(?:embeds/)?video/
  24         )(?P<id>[0-9]+)'''
  25     _EMBED_REGEX = [r'''(?x)]
  26             (?:
  27                 <meta\s+property="og:video"\s+content=|
  28                 <iframe[^>]+?src=
  29             )
  30             (["\'])(?P<url>https?://(?:www\.)?sbs\.com\.au/ondemand/video/.+?)\1''']
  31
  32     _TESTS = [{
  33         # Original URL is handled by the generic IE which finds the iframe:
  34         # http://www.sbs.com.au/thefeed/blog/2014/08/21/dingo-conservation
  35         'url': 'http://www.sbs.com.au/ondemand/video/single/320403011771/?source=drupal&vertical=thefeed',
  36         'md5': '31f84a7a19b53635db63c73f8ab0c4a7',
  37         'info_dict': {
  38             'id': '320403011771',  # '_rFBPRPO4pMR',
  39             'ext': 'mp4',
  40             'title': 'Dingo Conservation (The Feed)',
  41             'description': 'md5:f250a9856fca50d22dec0b5b8015f8a5',
  42             'thumbnail': r're:https?://.*\.jpg',
  43             'duration': 308,
  44             'timestamp': 1408613220,
  45             'upload_date': '20140821',
  46             'uploader': 'SBSC',
  47         },
  48         'expected_warnings': ['Unable to download JSON metadata'],
  49     }, {
  50         'url': 'http://www.sbs.com.au/ondemand/video/320403011771/Dingo-Conservation-The-Feed',
  51         'only_matching': True,
  52     }, {
  53         'url': 'http://www.sbs.com.au/news/video/471395907773/The-Feed-July-9',
  54         'only_matching': True,
  55     }, {
  56         'url': 'https://www.sbs.com.au/ondemand/?play=1836638787723',
  57         'only_matching': True,
  58     }, {
  59         'url': 'https://www.sbs.com.au/ondemand/program/inside-windsor-castle?play=1283505731842',
  60         'only_matching': True,
  61     }, {
  62         'url': 'https://www.sbs.com.au/news/embeds/video/1840778819866',
  63         'only_matching': True,
  64     }, {
  65         'url': 'https://www.sbs.com.au/ondemand/watch/1698704451971',
  66         'only_matching': True,
  67     }, {
  68         'url': 'https://www.sbs.com.au/ondemand/movie/coherence/1469404227931',
  69         'only_matching': True,
  70     }, {
  71         'note': 'Live stream',
  72         'url': 'https://www.sbs.com.au/ondemand/video/1726824003663/sbs-24x7-live-stream-nsw',
  73         'only_matching': True,
  74     }, {
  75         'url': 'https://www.sbs.com.au/ondemand/news-series/dateline/dateline-2022/dateline-s2022-ep26/2072245827515',
  76         'only_matching': True,
  77     }, {
  78         'url': 'https://www.sbs.com.au/ondemand/tv-series/the-handmaids-tale/season-5/the-handmaids-tale-s5-ep1/2065631811776',
  79         'only_matching': True,
  80     }, {
  81         'url': 'https://www.sbs.com.au/ondemand/tv-program/autun-romes-forgotten-sister/2116212803602',
  82         'only_matching': True,
  83     }]
  84
  85     _GEO_COUNTRIES = ['AU']
  86     _AUS_TV_PARENTAL_GUIDELINES = {
  87         'P': 0,
  88         'C': 7,
  89         'G': 0,
  90         'PG': 0,
  91         'M': 14,
  92         'MA15+': 15,
  93         'MAV15+': 15,
  94         'R18+': 18,
  95     }
  96     _PLAYER_API = 'https://www.sbs.com.au/api/v3'
  97
  98     def _real_extract(self, url):
  99         video_id = self._match_id(url)
 100         formats, subtitles = self._extract_smil_formats_and_subtitles(
 101             update_url_query(f'{self._PLAYER_API}/video_smil', {'id': video_id}), video_id)
 102
 103         if not formats:
 104             urlh = self._request_webpage(
 105                 HEADRequest('https://sbs-vod-prod-01.akamaized.net/'), video_id,
 106                 note='Checking geo-restriction', fatal=False, expected_status=403)
 107             if urlh:
 108                 error_reasons = urlh.headers.get_all('x-error-reason') or []
 109                 if 'geo-blocked' in error_reasons:
 110                     self.raise_geo_restricted(countries=['AU'])
 111             self.raise_no_formats('No formats are available', video_id=video_id)
 112
 113         media = traverse_obj(self._download_json(
 114             f'{self._PLAYER_API}/video_stream', video_id, fatal=False,
 115             query={'id': video_id, 'context': 'tv'}), ('video_object', {dict})) or {}
 116
 117         media.update(self._download_json(
 118             f'https://catalogue.pr.sbsod.com/mpx-media/{video_id}',
 119             video_id, fatal=not media) or {})
 120
 121         # For named episodes, use the catalogue's title to set episode, rather than generic 'Episode N'.
 122         if traverse_obj(media, ('partOfSeries', {dict})):
 123             media['epName'] = traverse_obj(media, ('title', {str}))
 124
 125         return {
 126             'id': video_id,
 127             **traverse_obj(media, {
 128                 'title': ('name', {str}),
 129                 'description': ('description', {str}),
 130                 'channel': ('taxonomy', 'channel', 'name', {str}),
 131                 'series': ((('partOfSeries', 'name'), 'seriesTitle'), {str}),
 132                 'series_id': ((('partOfSeries', 'uuid'), 'seriesID'), {str}),
 133                 'season_number': ('seasonNumber', {int_or_none}),
 134                 'episode': ('epName', {str}),
 135                 'episode_number': ('episodeNumber', {int_or_none}),
 136                 'timestamp': (('datePublished', ('publication', 'startDate')), {parse_iso8601}),
 137                 'release_year': ('releaseYear', {int_or_none}),
 138                 'duration': ('duration', ({float_or_none}, {parse_duration})),
 139                 'is_live': ('liveStream', {bool}),
 140                 'age_limit': (('classificationID', 'contentRating'), {str.upper}, {
 141                     lambda x: self._AUS_TV_PARENTAL_GUIDELINES.get(x)}),  # dict.get is unhashable in py3.7
 142             }, get_all=False),
 143             **traverse_obj(media, {
 144                 'categories': (('genres', ...), ('taxonomy', ('genre', 'subgenre'), 'name'), {str}),
 145                 'tags': (('consumerAdviceTexts', ('sbsSubCertification', 'consumerAdvice')), ..., {str}),
 146                 'thumbnails': ('thumbnails', lambda _, v: url_or_none(v['contentUrl']), {
 147                     'id': ('name', {str}),
 148                     'url': 'contentUrl',
 149                     'width': ('width', {int_or_none}),
 150                     'height': ('height', {int_or_none}),
 151                 }),
 152             }),
 153             'formats': formats,
 154             'subtitles': subtitles,
 155             'uploader': 'SBSC',
 156         }