yt_dlp/extractor/nfhsnetwork.py

   1 from .common import InfoExtractor
   2 from ..utils import try_get, unified_strdate, unified_timestamp
   3
   4
   5 class NFHSNetworkIE(InfoExtractor):
   6     IE_NAME = 'NFHSNetwork'
   7     _VALID_URL = r'https?://(?:www\.)?nfhsnetwork\.com/events/[\w-]+/(?P<id>(?:gam|evt|dd|)?[\w\d]{0,10})'
   8     _TESTS = [{
   9         # Auto-generated two-team sport (pixellot)
  10         'url': 'https://www.nfhsnetwork.com/events/rockford-high-school-rockford-mi/gamcf7e54cfbc',
  11         'info_dict': {
  12             'id': 'gamcf7e54cfbc',
  13             'ext': 'mp4',
  14             'title': 'Rockford vs Spring Lake - Girls Varsity Lacrosse 03/27/2021',
  15             'uploader': 'MHSAA - Michigan: Rockford High School, Rockford, MI',
  16             'uploader_id': 'cd2622cf76',
  17             'uploader_url': 'https://www.nfhsnetwork.com/schools/rockford-high-school-rockford-mi',
  18             'location': 'Rockford, Michigan',
  19             'timestamp': 1616859000,
  20             'upload_date': '20210327',
  21         },
  22         'params': {
  23             # m3u8 download
  24             'skip_download': True,
  25         },
  26     }, {
  27         # Non-sport activity with description
  28         'url': 'https://www.nfhsnetwork.com/events/limon-high-school-limon-co/evt4a30e3726c',
  29         'info_dict': {
  30             'id': 'evt4a30e3726c',
  31             'ext': 'mp4',
  32             'title': 'Drama Performance Limon High School vs. Limon High School - 12/13/2020',
  33             'description': 'Join the broadcast of the Limon High School Musical Performance at 2 PM.',
  34             'uploader': 'CHSAA: Limon High School, Limon, CO',
  35             'uploader_id': '7d2d121332',
  36             'uploader_url': 'https://www.nfhsnetwork.com/schools/limon-high-school-limon-co',
  37             'location': 'Limon, Colorado',
  38             'timestamp': 1607893200,
  39             'upload_date': '20201213',
  40         },
  41         'params': {
  42             # m3u8 download
  43             'skip_download': True,
  44         },
  45     }, {
  46         # Postseason game
  47         'url': 'https://www.nfhsnetwork.com/events/nfhs-network-special-events/dd8de71d45',
  48         'info_dict': {
  49             'id': 'dd8de71d45',
  50             'ext': 'mp4',
  51             'title': '2015 UA Holiday Classic Tournament: National Division  - 12/26/2015',
  52             'uploader': 'SoCal Sports Productions',
  53             'uploader_id': '063dba0150',
  54             'uploader_url': 'https://www.nfhsnetwork.com/affiliates/socal-sports-productions',
  55             'location': 'San Diego, California',
  56             'timestamp': 1451187000,
  57             'upload_date': '20151226',
  58         },
  59         'params': {
  60             # m3u8 download
  61             'skip_download': True,
  62         },
  63     }, {
  64         # Video with no broadcasts object
  65         'url': 'https://www.nfhsnetwork.com/events/wiaa-wi/9aa2f92f82',
  66         'info_dict': {
  67             'id': '9aa2f92f82',
  68             'ext': 'mp4',
  69             'title': 'Competitive Equity  - 01/21/2015',
  70             'description': 'Committee members discuss points of their research regarding a competitive equity plan',
  71             'uploader': 'WIAA - Wisconsin: Wisconsin Interscholastic Athletic Association',
  72             'uploader_id': 'a49f7d1002',
  73             'uploader_url': 'https://www.nfhsnetwork.com/associations/wiaa-wi',
  74             'location': 'Stevens Point, Wisconsin',
  75             'timestamp': 1421856000,
  76             'upload_date': '20150121',
  77         },
  78         'params': {
  79             # m3u8 download
  80             'skip_download': True,
  81         },
  82     },
  83     ]
  84
  85     def _real_extract(self, url):
  86         video_id = self._match_id(url)
  87         webpage = self._download_webpage(url, video_id)
  88         data = self._download_json(
  89             'https://cfunity.nfhsnetwork.com/v2/game_or_event/' + video_id,
  90             video_id)
  91         publisher = data.get('publishers')[0]  # always exists
  92         broadcast = (publisher.get('broadcasts') or publisher.get('vods'))[0]  # some (older) videos don't have a broadcasts object
  93         uploader = publisher.get('formatted_name') or publisher.get('name')
  94         uploader_id = publisher.get('publisher_key')
  95         pub_type = publisher.get('type')
  96         uploader_prefix = (
  97             'schools' if pub_type == 'school'
  98             else 'associations' if 'association' in pub_type
  99             else 'affiliates' if (pub_type == 'publisher' or pub_type == 'affiliate')
 100             else 'schools')
 101         uploader_page = 'https://www.nfhsnetwork.com/{}/{}'.format(uploader_prefix, publisher.get('slug'))
 102         location = '{}, {}'.format(data.get('city'), data.get('state_name'))
 103         description = broadcast.get('description')
 104         is_live = broadcast.get('on_air') or broadcast.get('status') == 'on_air' or False
 105
 106         timestamp = unified_timestamp(data.get('local_start_time'))
 107         upload_date = unified_strdate(data.get('local_start_time'))
 108
 109         title = (
 110             self._og_search_title(webpage)
 111             or self._html_search_regex(r'<h1 class="sr-hidden">(.*?)</h1>', webpage, 'title'))
 112         title = title.split('|')[0].strip()
 113
 114         video_type = 'broadcasts' if is_live else 'vods'
 115         key = broadcast.get('key') if is_live else try_get(publisher, lambda x: x['vods'][0]['key'])
 116         m3u8_url = self._download_json(
 117             f'https://cfunity.nfhsnetwork.com/v2/{video_type}/{key}/url',
 118             video_id).get('video_url')
 119
 120         formats = self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', live=is_live)
 121
 122         return {
 123             'id': video_id,
 124             'title': title,
 125             'formats': formats,
 126             'description': description,
 127             'timestamp': timestamp,
 128             'uploader': uploader,
 129             'uploader_id': uploader_id,
 130             'uploader_url': uploader_page,
 131             'location': location,
 132             'upload_date': upload_date,
 133             'is_live': is_live,
 134             '_format_sort_fields': ('res', 'tbr'),
 135         }