yt_dlp/extractor/nhl.py

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