yt_dlp/extractor/byutv.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     determine_ext,
   4     merge_dicts,
   5     parse_duration,
   6     url_or_none,
   7 )
   8
   9
  10 class BYUtvIE(InfoExtractor):
  11     _WORKING = False
  12     _VALID_URL = r'https?://(?:www\.)?byutv\.org/(?:watch|player)/(?!event/)(?P<id>[0-9a-f-]+)(?:/(?P<display_id>[^/?#&]+))?'
  13     _TESTS = [{
  14         'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d/studio-c-season-5-episode-5',
  15         'info_dict': {
  16             'id': 'ZvanRocTpW-G5_yZFeltTAMv6jxOU9KH',
  17             'display_id': 'studio-c-season-5-episode-5',
  18             'ext': 'mp4',
  19             'title': 'Season 5 Episode 5',
  20             'description': 'md5:1d31dc18ef4f075b28f6a65937d22c65',
  21             'thumbnail': r're:^https?://.*',
  22             'duration': 1486.486,
  23         },
  24         'params': {
  25             'skip_download': True,
  26         },
  27     }, {
  28         # dvr
  29         'url': 'https://www.byutv.org/player/8f1dab9b-b243-47c8-b525-3e2d021a3451/byu-softball-pacific-vs-byu-41219---game-2',
  30         'info_dict': {
  31             'id': '8f1dab9b-b243-47c8-b525-3e2d021a3451',
  32             'display_id': 'byu-softball-pacific-vs-byu-41219---game-2',
  33             'ext': 'mp4',
  34             'title': 'Pacific vs. BYU (4/12/19)',
  35             'description': 'md5:1ac7b57cb9a78015910a4834790ce1f3',
  36             'duration': 11645,
  37         },
  38         'params': {
  39             'skip_download': True
  40         },
  41     }, {
  42         'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d',
  43         'only_matching': True,
  44     }, {
  45         'url': 'https://www.byutv.org/player/27741493-dc83-40b0-8420-e7ae38a2ae98/byu-football-toledo-vs-byu-93016?listid=4fe0fee5-0d3c-4a29-b725-e4948627f472&listindex=0&q=toledo',
  46         'only_matching': True,
  47     }]
  48
  49     def _real_extract(self, url):
  50         mobj = self._match_valid_url(url)
  51         video_id = mobj.group('id')
  52         display_id = mobj.group('display_id') or video_id
  53
  54         video = self._download_json(
  55             'https://api.byutv.org/api3/catalog/getvideosforcontent',
  56             display_id, query={
  57                 'contentid': video_id,
  58                 'channel': 'byutv',
  59                 'x-byutv-context': 'web$US',
  60             }, headers={
  61                 'x-byutv-context': 'web$US',
  62                 'x-byutv-platformkey': 'xsaaw9c7y5',
  63             })
  64
  65         info = {}
  66         formats = []
  67         subtitles = {}
  68         for format_id, ep in video.items():
  69             if not isinstance(ep, dict):
  70                 continue
  71             video_url = url_or_none(ep.get('videoUrl'))
  72             if not video_url:
  73                 continue
  74             ext = determine_ext(video_url)
  75             if ext == 'm3u8':
  76                 m3u8_fmts, m3u8_subs = self._extract_m3u8_formats_and_subtitles(
  77                     video_url, video_id, 'mp4', entry_protocol='m3u8_native',
  78                     m3u8_id='hls', fatal=False)
  79                 formats.extend(m3u8_fmts)
  80                 subtitles = self._merge_subtitles(subtitles, m3u8_subs)
  81             elif ext == 'mpd':
  82                 mpd_fmts, mpd_subs = self._extract_mpd_formats_and_subtitles(
  83                     video_url, video_id, mpd_id='dash', fatal=False)
  84                 formats.extend(mpd_fmts)
  85                 subtitles = self._merge_subtitles(subtitles, mpd_subs)
  86             else:
  87                 formats.append({
  88                     'url': video_url,
  89                     'format_id': format_id,
  90                 })
  91             merge_dicts(info, {
  92                 'title': ep.get('title'),
  93                 'description': ep.get('description'),
  94                 'thumbnail': ep.get('imageThumbnail'),
  95                 'duration': parse_duration(ep.get('length')),
  96             })
  97
  98         return merge_dicts(info, {
  99             'id': video_id,
 100             'display_id': display_id,
 101             'title': display_id,
 102             'formats': formats,
 103             'subtitles': subtitles,
 104         })