yt_dlp/extractor/screencastify.py

   1 import urllib.parse
   2
   3 from .common import InfoExtractor
   4 from ..utils import traverse_obj, update_url_query
   5
   6
   7 class ScreencastifyIE(InfoExtractor):
   8     _VALID_URL = r'https?://watch\.screencastify\.com/v/(?P<id>[^/?#]+)'
   9     _TESTS = [{
  10         'url': 'https://watch.screencastify.com/v/sYVkZip3quLKhHw4Ybk8',
  11         'info_dict': {
  12             'id': 'sYVkZip3quLKhHw4Ybk8',
  13             'ext': 'mp4',
  14             'title': 'Inserting and Aligning the Case Top and Bottom',
  15             'description': '',
  16             'uploader': 'Paul Gunn',
  17             'extra_param_to_segment_url': str,
  18         },
  19         'params': {
  20             'skip_download': 'm3u8',
  21         },
  22     }]
  23
  24     def _real_extract(self, url):
  25         video_id = self._match_id(url)
  26         info = self._download_json(
  27             f'https://umbrella.svc.screencastify.com/api/umbrellaService/watch/{video_id}', video_id)
  28
  29         query_string = traverse_obj(info, ('manifest', 'auth', 'query'))
  30         query = urllib.parse.parse_qs(query_string)
  31         formats = []
  32         dash_manifest_url = traverse_obj(info, ('manifest', 'url'))
  33         if dash_manifest_url:
  34             formats.extend(
  35                 self._extract_mpd_formats(
  36                     dash_manifest_url, video_id, mpd_id='dash', query=query, fatal=False))
  37         hls_manifest_url = traverse_obj(info, ('manifest', 'hlsUrl'))
  38         if hls_manifest_url:
  39             formats.extend(
  40                 self._extract_m3u8_formats(
  41                     hls_manifest_url, video_id, ext='mp4', m3u8_id='hls', query=query, fatal=False))
  42         for f in formats:
  43             f['url'] = update_url_query(f['url'], query)
  44
  45         return {
  46             'id': video_id,
  47             'title': info.get('title'),
  48             'description': info.get('description'),
  49             'uploader': info.get('userName'),
  50             'formats': formats,
  51             'extra_param_to_segment_url': query_string,
  52         }