yt_dlp/extractor/nhl.py

   1 from __future__ import unicode_literals
   2
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_str
   6 from ..utils import (
   7     determine_ext,
   8     int_or_none,
   9     parse_iso8601,
  10     parse_duration,
  11 )
  12
  13
  14 class NHLBaseIE(InfoExtractor):
  15     def _real_extract(self, url):
  16         site, tmp_id = self._match_valid_url(url).groups()
  17         video_data = self._download_json(
  18             'https://%s/%s/%sid/v1/%s/details/web-v1.json'
  19             % (self._CONTENT_DOMAIN, site[:3], 'item/' if site == 'mlb' else '', tmp_id), tmp_id)
  20         if video_data.get('type') != 'video':
  21             video_data = video_data['media']
  22             video = video_data.get('video')
  23             if video:
  24                 video_data = video
  25             else:
  26                 videos = video_data.get('videos')
  27                 if videos:
  28                     video_data = videos[0]
  29
  30         video_id = compat_str(video_data['id'])
  31         title = video_data['title']
  32
  33         formats = []
  34         for playback in video_data.get('playbacks', []):
  35             playback_url = playback.get('url')
  36             if not playback_url:
  37                 continue
  38             ext = determine_ext(playback_url)
  39             if ext == 'm3u8':
  40                 m3u8_formats = self._extract_m3u8_formats(
  41                     playback_url, video_id, 'mp4', 'm3u8_native',
  42                     m3u8_id=playback.get('name', 'hls'), fatal=False)
  43                 self._check_formats(m3u8_formats, video_id)
  44                 formats.extend(m3u8_formats)
  45             else:
  46                 height = int_or_none(playback.get('height'))
  47                 formats.append({
  48                     'format_id': playback.get('name', 'http' + ('-%dp' % height if height else '')),
  49                     'url': playback_url,
  50                     'width': int_or_none(playback.get('width')),
  51                     'height': height,
  52                     'tbr': int_or_none(self._search_regex(r'_(\d+)[kK]', playback_url, 'bitrate', default=None)),
  53                 })
  54         self._sort_formats(formats)
  55
  56         thumbnails = []
  57         cuts = video_data.get('image', {}).get('cuts') or []
  58         if isinstance(cuts, dict):
  59             cuts = cuts.values()
  60         for thumbnail_data in cuts:
  61             thumbnail_url = thumbnail_data.get('src')
  62             if not thumbnail_url:
  63                 continue
  64             thumbnails.append({
  65                 'url': thumbnail_url,
  66                 'width': int_or_none(thumbnail_data.get('width')),
  67                 'height': int_or_none(thumbnail_data.get('height')),
  68             })
  69
  70         return {
  71             'id': video_id,
  72             'title': title,
  73             'description': video_data.get('description'),
  74             'timestamp': parse_iso8601(video_data.get('date')),
  75             'duration': parse_duration(video_data.get('duration')),
  76             'thumbnails': thumbnails,
  77             'formats': formats,
  78         }
  79
  80
  81 class NHLIE(NHLBaseIE):
  82     IE_NAME = 'nhl.com'
  83     _VALID_URL = r'https?://(?:www\.)?(?P<site>nhl|wch2016)\.com/(?:[^/]+/)*c-(?P<id>\d+)'
  84     _CONTENT_DOMAIN = 'nhl.bamcontent.com'
  85     _TESTS = [{
  86         # type=video
  87         'url': 'https://www.nhl.com/video/anisimov-cleans-up-mess/t-277752844/c-43663503',
  88         'md5': '0f7b9a8f986fb4b4eeeece9a56416eaf',
  89         'info_dict': {
  90             'id': '43663503',
  91             'ext': 'mp4',
  92             'title': 'Anisimov cleans up mess',
  93             'description': 'md5:a02354acdfe900e940ce40706939ca63',
  94             'timestamp': 1461288600,
  95             'upload_date': '20160422',
  96         },
  97     }, {
  98         # type=article
  99         'url': 'https://www.nhl.com/news/dennis-wideman-suspended/c-278258934',
 100         'md5': '1f39f4ea74c1394dea110699a25b366c',
 101         'info_dict': {
 102             'id': '40784403',
 103             'ext': 'mp4',
 104             'title': 'Wideman suspended by NHL',
 105             'description': 'Flames defenseman Dennis Wideman was banned 20 games for violation of Rule 40 (Physical Abuse of Officials)',
 106             'upload_date': '20160204',
 107             'timestamp': 1454544904,
 108         },
 109     }, {
 110         # Some m3u8 URLs are invalid (https://github.com/ytdl-org/youtube-dl/issues/10713)
 111         'url': 'https://www.nhl.com/predators/video/poile-laviolette-on-subban-trade/t-277437416/c-44315003',
 112         'md5': '50b2bb47f405121484dda3ccbea25459',
 113         'info_dict': {
 114             'id': '44315003',
 115             'ext': 'mp4',
 116             'title': 'Poile, Laviolette on Subban trade',
 117             'description': 'General manager David Poile and head coach Peter Laviolette share their thoughts on acquiring P.K. Subban from Montreal (06/29/16)',
 118             'timestamp': 1467242866,
 119             'upload_date': '20160629',
 120         },
 121     }, {
 122         'url': 'https://www.wch2016.com/video/caneur-best-of-game-2-micd-up/t-281230378/c-44983703',
 123         'only_matching': True,
 124     }, {
 125         'url': 'https://www.wch2016.com/news/3-stars-team-europe-vs-team-canada/c-282195068',
 126         'only_matching': True,
 127     }]