yt_dlp/extractor/playstuff.py

   1 from .common import InfoExtractor
   2 from ..compat import compat_str
   3 from ..utils import (
   4     smuggle_url,
   5     try_get,
   6 )
   7
   8
   9 class PlayStuffIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?play\.stuff\.co\.nz/details/(?P<id>[^/?#&]+)'
  11     _TESTS = [{
  12         'url': 'https://play.stuff.co.nz/details/608778ac1de1c4001a3fa09a',
  13         'md5': 'c82d3669e5247c64bc382577843e5bd0',
  14         'info_dict': {
  15             'id': '6250584958001',
  16             'ext': 'mp4',
  17             'title': 'Episode 1: Rotorua/Mt Maunganui/Tauranga',
  18             'description': 'md5:c154bafb9f0dd02d01fd4100fb1c1913',
  19             'uploader_id': '6005208634001',
  20             'timestamp': 1619491027,
  21             'upload_date': '20210427',
  22         },
  23         'add_ie': ['BrightcoveNew'],
  24     }, {
  25         # geo restricted, bypassable
  26         'url': 'https://play.stuff.co.nz/details/_6155660351001',
  27         'only_matching': True,
  28     }]
  29     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_default/index.html?videoId=%s'
  30
  31     def _real_extract(self, url):
  32         video_id = self._match_id(url)
  33
  34         webpage = self._download_webpage(url, video_id)
  35
  36         state = self._parse_json(
  37             self._search_regex(
  38                 r'__INITIAL_STATE__\s*=\s*({.+?})\s*;', webpage, 'state'),
  39             video_id)
  40
  41         account_id = try_get(
  42             state, lambda x: x['configurations']['accountId'],
  43             compat_str) or '6005208634001'
  44         player_id = try_get(
  45             state, lambda x: x['configurations']['playerId'],
  46             compat_str) or 'default'
  47
  48         entries = []
  49         for item_id, video in state['items'].items():
  50             if not isinstance(video, dict):
  51                 continue
  52             asset_id = try_get(
  53                 video, lambda x: x['content']['attributes']['assetId'],
  54                 compat_str)
  55             if not asset_id:
  56                 continue
  57             entries.append(self.url_result(
  58                 smuggle_url(
  59                     self.BRIGHTCOVE_URL_TEMPLATE % (account_id, player_id, asset_id),
  60                     {'geo_countries': ['NZ']}),
  61                 'BrightcoveNew', video_id))
  62
  63         return self.playlist_result(entries, video_id)