yt_dlp/extractor/nfhsnetwork.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5
   6
   7 from ..utils import (
   8     try_get,
   9     unified_strdate,
  10     unified_timestamp
  11 )
  12
  13
  14 class NFHSNetworkIE(InfoExtractor):
  15     IE_NAME = 'NFHSNetwork'
  16     _VALID_URL = r'https?://(?:www\.)?nfhsnetwork\.com/events/[\w-]+/(?P<id>(?:gam|evt|dd|)?[\w\d]{0,10})'
  17     _TESTS = [{
  18         # Auto-generated two-team sport (pixellot)
  19         'url': 'https://www.nfhsnetwork.com/events/rockford-high-school-rockford-mi/gamcf7e54cfbc',
  20         'info_dict': {
  21             'id': 'gamcf7e54cfbc',
  22             'ext': 'mp4',
  23             'title': 'Rockford vs Spring Lake - Girls Varsity Lacrosse 03/27/2021',
  24             'uploader': 'MHSAA - Michigan: Rockford High School, Rockford, MI',
  25             'uploader_id': 'cd2622cf76',
  26             'uploader_url': 'https://www.nfhsnetwork.com/schools/rockford-high-school-rockford-mi',
  27             'location': 'Rockford, Michigan',
  28             'timestamp': 1616859000,
  29             'upload_date': '20210327'
  30         },
  31         'params': {
  32             # m3u8 download
  33             'skip_download': True,
  34         }
  35     }, {
  36         # Non-sport activity with description
  37         'url': 'https://www.nfhsnetwork.com/events/limon-high-school-limon-co/evt4a30e3726c',
  38         'info_dict': {
  39             'id': 'evt4a30e3726c',
  40             'ext': 'mp4',
  41             'title': 'Drama Performance Limon High School vs. Limon High School - 12/13/2020',
  42             'description': 'Join the broadcast of the Limon High School Musical Performance at 2 PM.',
  43             'uploader': 'CHSAA: Limon High School, Limon, CO',
  44             'uploader_id': '7d2d121332',
  45             'uploader_url': 'https://www.nfhsnetwork.com/schools/limon-high-school-limon-co',
  46             'location': 'Limon, Colorado',
  47             'timestamp': 1607893200,
  48             'upload_date': '20201213'
  49         },
  50         'params': {
  51             # m3u8 download
  52             'skip_download': True,
  53         }
  54     }, {
  55         # Postseason game
  56         'url': 'https://www.nfhsnetwork.com/events/nfhs-network-special-events/dd8de71d45',
  57         'info_dict': {
  58             'id': 'dd8de71d45',
  59             'ext': 'mp4',
  60             'title': '2015 UA Holiday Classic Tournament: National Division  - 12/26/2015',
  61             'uploader': 'SoCal Sports Productions',
  62             'uploader_id': '063dba0150',
  63             'uploader_url': 'https://www.nfhsnetwork.com/affiliates/socal-sports-productions',
  64             'location': 'San Diego, California',
  65             'timestamp': 1451187000,
  66             'upload_date': '20151226'
  67         },
  68         'params': {
  69             # m3u8 download
  70             'skip_download': True,
  71         }
  72     }, {
  73         # Video with no broadcasts object
  74         'url': 'https://www.nfhsnetwork.com/events/wiaa-wi/9aa2f92f82',
  75         'info_dict': {
  76             'id': '9aa2f92f82',
  77             'ext': 'mp4',
  78             'title': 'Competitive Equity  - 01/21/2015',
  79             'description': 'Committee members discuss points of their research regarding a competitive equity plan',
  80             'uploader': 'WIAA - Wisconsin: Wisconsin Interscholastic Athletic Association',
  81             'uploader_id': 'a49f7d1002',
  82             'uploader_url': 'https://www.nfhsnetwork.com/associations/wiaa-wi',
  83             'location': 'Stevens Point, Wisconsin',
  84             'timestamp': 1421856000,
  85             'upload_date': '20150121'
  86         },
  87         'params': {
  88             # m3u8 download
  89             'skip_download': True,
  90         }
  91     }
  92     ]
  93
  94     def _real_extract(self, url):
  95         video_id = self._match_id(url)
  96         webpage = self._download_webpage(url, video_id)
  97         data = self._download_json(
  98             'https://cfunity.nfhsnetwork.com/v2/game_or_event/' + video_id,
  99             video_id)
 100         publisher = data.get('publishers')[0]  # always exists
 101         broadcast = (publisher.get('broadcasts') or publisher.get('vods'))[0]  # some (older) videos don't have a broadcasts object
 102         uploader = publisher.get('formatted_name') or publisher.get('name')
 103         uploaderID = publisher.get('publisher_key')
 104         pubType = publisher.get('type')
 105         uploaderPrefix = (
 106             "schools" if pubType == "school"
 107             else "associations" if "association" in pubType
 108             else "affiliates" if (pubType == "publisher" or pubType == "affiliate")
 109             else "schools")
 110         uploaderPage = 'https://www.nfhsnetwork.com/%s/%s' % (uploaderPrefix, publisher.get('slug'))
 111         location = '%s, %s' % (data.get('city'), data.get('state_name'))
 112         description = broadcast.get('description')
 113         isLive = broadcast.get('on_air') or broadcast.get('status') == 'on_air' or False
 114
 115         timestamp = unified_timestamp(data.get('local_start_time'))
 116         upload_date = unified_strdate(data.get('local_start_time'))
 117
 118         title = (
 119             self._og_search_title(webpage)
 120             or self._html_search_regex(r'<h1 class="sr-hidden">(.*?)</h1>', webpage, 'title'))
 121         title = title.split('|')[0].strip()
 122
 123         video_type = 'broadcasts' if isLive else 'vods'
 124         key = broadcast.get('key') if isLive else try_get(publisher, lambda x: x['vods'][0]['key'])
 125         m3u8_url = self._download_json(
 126             'https://cfunity.nfhsnetwork.com/v2/%s/%s/url' % (video_type, key),
 127             video_id).get('video_url')
 128
 129         formats = self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', live=isLive)
 130         self._sort_formats(formats, ['res', 'tbr'])
 131
 132         return {
 133             'id': video_id,
 134             'title': title,
 135             'formats': formats,
 136             'description': description,
 137             'timestamp': timestamp,
 138             'uploader': uploader,
 139             'uploader_id': uploaderID,
 140             'uploader_url': uploaderPage,
 141             'location': location,
 142             'upload_date': upload_date,
 143             'is_live': isLive
 144         }