yt_dlp/extractor/s4c.py

   1 from .common import InfoExtractor
   2 from ..utils import traverse_obj, url_or_none
   3
   4
   5 class S4CIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?s4c\.cymru/clic/programme/(?P<id>\d+)'
   7     _TESTS = [{
   8         'url': 'https://www.s4c.cymru/clic/programme/861362209',
   9         'info_dict': {
  10             'id': '861362209',
  11             'ext': 'mp4',
  12             'title': 'Y Swn',
  13             'description': 'md5:f7681a30e4955b250b3224aa9fe70cf0',
  14             'duration': 5340,
  15             'thumbnail': 'https://www.s4c.cymru/amg/1920x1080/Y_Swn_2023S4C_099_ii.jpg'
  16         },
  17     }, {
  18         'url': 'https://www.s4c.cymru/clic/programme/856636948',
  19         'info_dict': {
  20             'id': '856636948',
  21             'ext': 'mp4',
  22             'title': 'Am Dro',
  23             'duration': 2880,
  24             'description': 'md5:100d8686fc9a632a0cb2db52a3433ffe',
  25             'thumbnail': 'https://www.s4c.cymru/amg/1920x1080/Am_Dro_2022-23S4C_P6_4005.jpg'
  26         },
  27     }]
  28
  29     def _real_extract(self, url):
  30         video_id = self._match_id(url)
  31         details = self._download_json(
  32             f'https://www.s4c.cymru/df/full_prog_details?lang=e&programme_id={video_id}',
  33             video_id, fatal=False)
  34
  35         player_config = self._download_json(
  36             'https://player-api.s4c-cdn.co.uk/player-configuration/prod', video_id, query={
  37                 'programme_id': video_id,
  38                 'signed': '0',
  39                 'lang': 'en',
  40                 'mode': 'od',
  41                 'appId': 'clic',
  42                 'streamName': '',
  43             }, note='Downloading player config JSON')
  44         subtitles = {}
  45         for sub in traverse_obj(player_config, ('subtitles', lambda _, v: url_or_none(v['0']))):
  46             subtitles.setdefault(sub.get('3', 'en'), []).append({
  47                 'url': sub['0'],
  48                 'name': sub.get('1'),
  49             })
  50         m3u8_url = self._download_json(
  51             'https://player-api.s4c-cdn.co.uk/streaming-urls/prod', video_id, query={
  52                 'mode': 'od',
  53                 'application': 'clic',
  54                 'region': 'WW',
  55                 'extra': 'false',
  56                 'thirdParty': 'false',
  57                 'filename': player_config['filename'],
  58             }, note='Downloading streaming urls JSON')['hls']
  59
  60         return {
  61             'id': video_id,
  62             'formats': self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', m3u8_id='hls'),
  63             'subtitles': subtitles,
  64             'thumbnail': url_or_none(player_config.get('poster')),
  65             **traverse_obj(details, ('full_prog_details', 0, {
  66                 'title': (('programme_title', 'series_title'), {str}),
  67                 'description': ('full_billing', {str.strip}),
  68                 'duration': ('duration', {lambda x: int(x) * 60}),
  69             }), get_all=False),
  70         }
  71
  72
  73 class S4CSeriesIE(InfoExtractor):
  74     _VALID_URL = r'https?://(?:www\.)?s4c\.cymru/clic/series/(?P<id>\d+)'
  75     _TESTS = [{
  76         'url': 'https://www.s4c.cymru/clic/series/864982911',
  77         'playlist_mincount': 6,
  78         'info_dict': {
  79             'id': '864982911',
  80             'title': 'Iaith ar Daith',
  81         },
  82     }, {
  83         'url': 'https://www.s4c.cymru/clic/series/866852587',
  84         'playlist_mincount': 8,
  85         'info_dict': {
  86             'id': '866852587',
  87             'title': 'FFIT Cymru',
  88         },
  89     }]
  90
  91     def _real_extract(self, url):
  92         series_id = self._match_id(url)
  93         series_details = self._download_json(
  94             'https://www.s4c.cymru/df/series_details', series_id, query={
  95                 'lang': 'e',
  96                 'series_id': series_id,
  97                 'show_prog_in_series': 'Y'
  98             }, note='Downloading series details JSON')
  99
 100         return self.playlist_result(
 101             [self.url_result(f'https://www.s4c.cymru/clic/programme/{episode_id}', S4CIE, episode_id)
 102              for episode_id in traverse_obj(series_details, ('other_progs_in_series', ..., 'id'))],
 103             series_id, traverse_obj(series_details, ('full_prog_details', 0, 'series_title', {str})))