yt_dlp/extractor/tenplay.py

   1 import itertools
   2
   3 from .common import InfoExtractor
   4 from ..networking import HEADRequest
   5 from ..utils import int_or_none, traverse_obj, url_or_none, urljoin
   6
   7
   8 class TenPlayIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?10play\.com\.au/(?:[^/]+/)+(?P<id>tpv\d{6}[a-z]{5})'
  10     _NETRC_MACHINE = '10play'
  11     _TESTS = [{
  12         'url': 'https://10play.com.au/neighbours/web-extras/season-41/heres-a-first-look-at-mischa-bartons-neighbours-debut/tpv230911hyxnz',
  13         'info_dict': {
  14             'id': '6336940246112',
  15             'ext': 'mp4',
  16             'title': 'Here\'s A First Look At Mischa Barton\'s Neighbours Debut',
  17             'alt_title': 'Here\'s A First Look At Mischa Barton\'s Neighbours Debut',
  18             'description': 'Neighbours Premieres Monday, September 18 At 4:30pm On 10 And 10 Play And 6:30pm On 10 Peach',
  19             'duration': 74,
  20             'season': 'Season 41',
  21             'season_number': 41,
  22             'series': 'Neighbours',
  23             'thumbnail': r're:https://.*\.jpg',
  24             'uploader': 'Channel 10',
  25             'age_limit': 15,
  26             'timestamp': 1694386800,
  27             'upload_date': '20230910',
  28             'uploader_id': '2199827728001',
  29         },
  30         'params': {
  31             'skip_download': True,
  32         },
  33         'skip': 'Only available in Australia',
  34     }, {
  35         'url': 'https://10play.com.au/neighbours/episodes/season-42/episode-9107/tpv240902nzqyp',
  36         'info_dict': {
  37             'id': '9000000000091177',
  38             'ext': 'mp4',
  39             'title': 'Neighbours - S42 Ep. 9107',
  40             'alt_title': 'Thu 05 Sep',
  41             'description': 'md5:37a1f4271be34b9ee2b533426a5fbaef',
  42             'duration': 1388,
  43             'episode': 'Episode 9107',
  44             'episode_number': 9107,
  45             'season': 'Season 42',
  46             'season_number': 42,
  47             'series': 'Neighbours',
  48             'thumbnail': r're:https://.*\.jpg',
  49             'age_limit': 15,
  50             'timestamp': 1725517860,
  51             'upload_date': '20240905',
  52             'uploader': 'Channel 10',
  53             'uploader_id': '2199827728001',
  54         },
  55         'params': {
  56             'skip_download': True,
  57         },
  58         'skip': 'Only available in Australia',
  59     }, {
  60         'url': 'https://10play.com.au/how-to-stay-married/web-extras/season-1/terrys-talks-ep-1-embracing-change/tpv190915ylupc',
  61         'only_matching': True,
  62     }]
  63     _GEO_BYPASS = False
  64
  65     _AUS_AGES = {
  66         'G': 0,
  67         'PG': 15,
  68         'M': 15,
  69         'MA': 15,
  70         'MA15+': 15,
  71         'R': 18,
  72         'X': 18,
  73     }
  74
  75     def _real_extract(self, url):
  76         content_id = self._match_id(url)
  77         data = self._download_json(
  78             'https://10play.com.au/api/v1/videos/' + content_id, content_id)
  79
  80         video_data = self._download_json(
  81             f'https://vod.ten.com.au/api/videos/bcquery?command=find_videos_by_id&video_id={data["altId"]}',
  82             content_id, 'Downloading video JSON')
  83         m3u8_url = self._request_webpage(
  84             HEADRequest(video_data['items'][0]['HLSURL']),
  85             content_id, 'Checking stream URL').url
  86         if '10play-not-in-oz' in m3u8_url:
  87             self.raise_geo_restricted(countries=['AU'])
  88         # Attempt to get a higher quality stream
  89         m3u8_url = m3u8_url.replace(',150,75,55,0000', ',300,150,75,55,0000')
  90         formats = self._extract_m3u8_formats(m3u8_url, content_id, 'mp4')
  91
  92         return {
  93             'id': content_id,
  94             'formats': formats,
  95             'subtitles': {'en': [{'url': data['captionUrl']}]} if url_or_none(data.get('captionUrl')) else None,
  96             'uploader': 'Channel 10',
  97             'uploader_id': '2199827728001',
  98             **traverse_obj(data, {
  99                 'id': ('altId', {str}),
 100                 'duration': ('duration', {int_or_none}),
 101                 'title': ('subtitle', {str}),
 102                 'alt_title': ('title', {str}),
 103                 'description': ('description', {str}),
 104                 'age_limit': ('classification', {self._AUS_AGES.get}),
 105                 'series': ('tvShow', {str}),
 106                 'season_number': ('season', {int_or_none}),
 107                 'episode_number': ('episode', {int_or_none}),
 108                 'timestamp': ('published', {int_or_none}),
 109                 'thumbnail': ('imageUrl', {url_or_none}),
 110             }),
 111         }
 112
 113
 114 class TenPlaySeasonIE(InfoExtractor):
 115     _VALID_URL = r'https?://(?:www\.)?10play\.com\.au/(?P<show>[^/?#]+)/episodes/(?P<season>[^/?#]+)/?(?:$|[?#])'
 116     _TESTS = [{
 117         'url': 'https://10play.com.au/masterchef/episodes/season-14',
 118         'info_dict': {
 119             'title': 'Season 14',
 120             'id': 'MjMyOTIy',
 121         },
 122         'playlist_mincount': 64,
 123     }, {
 124         'url': 'https://10play.com.au/the-bold-and-the-beautiful-fast-tracked/episodes/season-2022',
 125         'info_dict': {
 126             'title': 'Season 2022',
 127             'id': 'Mjc0OTIw',
 128         },
 129         'playlist_mincount': 256,
 130     }]
 131
 132     def _entries(self, load_more_url, display_id=None):
 133         skip_ids = []
 134         for page in itertools.count(1):
 135             episodes_carousel = self._download_json(
 136                 load_more_url, display_id, query={'skipIds[]': skip_ids},
 137                 note=f'Fetching episodes page {page}')
 138
 139             episodes_chunk = episodes_carousel['items']
 140             skip_ids.extend(ep['id'] for ep in episodes_chunk)
 141
 142             for ep in episodes_chunk:
 143                 yield ep['cardLink']
 144             if not episodes_carousel['hasMore']:
 145                 break
 146
 147     def _real_extract(self, url):
 148         show, season = self._match_valid_url(url).group('show', 'season')
 149         season_info = self._download_json(
 150             f'https://10play.com.au/api/shows/{show}/episodes/{season}', f'{show}/{season}')
 151
 152         episodes_carousel = traverse_obj(season_info, (
 153             'content', 0, 'components', (
 154                 lambda _, v: v['title'].lower() == 'episodes',
 155                 (..., {dict}),
 156             )), get_all=False) or {}
 157
 158         playlist_id = episodes_carousel['tpId']
 159
 160         return self.playlist_from_matches(
 161             self._entries(urljoin(url, episodes_carousel['loadMoreUrl']), playlist_id),
 162             playlist_id, traverse_obj(season_info, ('content', 0, 'title', {str})),
 163             getter=urljoin(url))