yt_dlp/extractor/nhl.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     determine_ext,
   4     int_or_none,
   5     join_nonempty,
   6     parse_duration,
   7     parse_iso8601,
   8 )
   9
  10
  11 class NHLBaseIE(InfoExtractor):
  12     def _real_extract(self, url):
  13         site, tmp_id = self._match_valid_url(url).groups()
  14         video_data = self._download_json(
  15             'https://{}/{}/{}id/v1/{}/details/web-v1.json'.format(
  16                 self._CONTENT_DOMAIN, site[:3], 'item/' if site == 'mlb' else '', tmp_id), tmp_id)
  17         if video_data.get('type') != 'video':
  18             video_data = video_data['media']
  19             video = video_data.get('video')
  20             if video:
  21                 video_data = video
  22             else:
  23                 videos = video_data.get('videos')
  24                 if videos:
  25                     video_data = videos[0]
  26
  27         video_id = str(video_data['id'])
  28         title = video_data['title']
  29
  30         formats = []
  31         for playback in video_data.get('playbacks', []):
  32             playback_url = playback.get('url')
  33             if not playback_url:
  34                 continue
  35             ext = determine_ext(playback_url)
  36             if ext == 'm3u8':
  37                 m3u8_formats = self._extract_m3u8_formats(
  38                     playback_url, video_id, 'mp4', 'm3u8_native',
  39                     m3u8_id=playback.get('name', 'hls'), fatal=False)
  40                 self._check_formats(m3u8_formats, video_id)
  41                 formats.extend(m3u8_formats)
  42             else:
  43                 height = int_or_none(playback.get('height'))
  44                 formats.append({
  45                     'format_id': playback.get('name') or join_nonempty('http', height and f'{height}p'),
  46                     'url': playback_url,
  47                     'width': int_or_none(playback.get('width')),
  48                     'height': height,
  49                     'tbr': int_or_none(self._search_regex(r'_(\d+)[kK]', playback_url, 'bitrate', default=None)),
  50                 })
  51
  52         thumbnails = []
  53         cuts = video_data.get('image', {}).get('cuts') or []
  54         if isinstance(cuts, dict):
  55             cuts = cuts.values()
  56         for thumbnail_data in cuts:
  57             thumbnail_url = thumbnail_data.get('src')
  58             if not thumbnail_url:
  59                 continue
  60             thumbnails.append({
  61                 'url': thumbnail_url,
  62                 'width': int_or_none(thumbnail_data.get('width')),
  63                 'height': int_or_none(thumbnail_data.get('height')),
  64             })
  65
  66         return {
  67             'id': video_id,
  68             'title': title,
  69             'description': video_data.get('description'),
  70             'timestamp': parse_iso8601(video_data.get('date')),
  71             'duration': parse_duration(video_data.get('duration')),
  72             'thumbnails': thumbnails,
  73             'formats': formats,
  74         }
  75
  76
  77 class NHLIE(NHLBaseIE):
  78     IE_NAME = 'nhl.com'
  79     _VALID_URL = r'https?://(?:www\.)?(?P<site>nhl|wch2016)\.com/(?:[^/]+/)*c-(?P<id>\d+)'
  80     _CONTENT_DOMAIN = 'nhl.bamcontent.com'
  81     _TESTS = [{
  82         # type=video
  83         'url': 'https://www.nhl.com/video/anisimov-cleans-up-mess/t-277752844/c-43663503',
  84         'md5': '0f7b9a8f986fb4b4eeeece9a56416eaf',
  85         'info_dict': {
  86             'id': '43663503',
  87             'ext': 'mp4',
  88             'title': 'Anisimov cleans up mess',
  89             'description': 'md5:a02354acdfe900e940ce40706939ca63',
  90             'timestamp': 1461288600,
  91             'upload_date': '20160422',
  92         },
  93     }, {
  94         # type=article
  95         'url': 'https://www.nhl.com/news/dennis-wideman-suspended/c-278258934',
  96         'md5': '1f39f4ea74c1394dea110699a25b366c',
  97         'info_dict': {
  98             'id': '40784403',
  99             'ext': 'mp4',
 100             'title': 'Wideman suspended by NHL',
 101             'description': 'Flames defenseman Dennis Wideman was banned 20 games for violation of Rule 40 (Physical Abuse of Officials)',
 102             'upload_date': '20160204',
 103             'timestamp': 1454544904,
 104         },
 105     }, {
 106         # Some m3u8 URLs are invalid (https://github.com/ytdl-org/youtube-dl/issues/10713)
 107         'url': 'https://www.nhl.com/predators/video/poile-laviolette-on-subban-trade/t-277437416/c-44315003',
 108         'md5': '50b2bb47f405121484dda3ccbea25459',
 109         'info_dict': {
 110             'id': '44315003',
 111             'ext': 'mp4',
 112             'title': 'Poile, Laviolette on Subban trade',
 113             'description': 'General manager David Poile and head coach Peter Laviolette share their thoughts on acquiring P.K. Subban from Montreal (06/29/16)',
 114             'timestamp': 1467242866,
 115             'upload_date': '20160629',
 116         },
 117     }, {
 118         'url': 'https://www.wch2016.com/video/caneur-best-of-game-2-micd-up/t-281230378/c-44983703',
 119         'only_matching': True,
 120     }, {
 121         'url': 'https://www.wch2016.com/news/3-stars-team-europe-vs-team-canada/c-282195068',
 122         'only_matching': True,
 123     }]