yt_dlp/extractor/blackboardcollaborate.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4
   5 from .common import InfoExtractor
   6 from ..utils import parse_iso8601
   7
   8
   9 class BlackboardCollaborateIE(InfoExtractor):
  10     _VALID_URL = r'''(?x)
  11                         https?://
  12                         (?P<region>[a-z-]+)\.bbcollab\.com/
  13                         (?:
  14                             collab/ui/session/playback/load|
  15                             recording
  16                         )/
  17                         (?P<id>[^/]+)'''
  18     _TESTS = [
  19         {
  20             'url': 'https://us-lti.bbcollab.com/collab/ui/session/playback/load/0a633b6a88824deb8c918f470b22b256',
  21             'md5': 'bb7a055682ee4f25fdb5838cdf014541',
  22             'info_dict': {
  23                 'id': '0a633b6a88824deb8c918f470b22b256',
  24                 'title': 'HESI A2 Information Session - Thursday, May 6, 2021 - recording_1',
  25                 'ext': 'mp4',
  26                 'duration': 1896000,
  27                 'timestamp': 1620331399,
  28                 'upload_date': '20210506',
  29             },
  30         },
  31         {
  32             'url': 'https://us.bbcollab.com/collab/ui/session/playback/load/76761522adfe4345a0dee6794bbcabda',
  33             'only_matching': True,
  34         },
  35         {
  36             'url': 'https://ca.bbcollab.com/collab/ui/session/playback/load/b6399dcb44df4f21b29ebe581e22479d',
  37             'only_matching': True,
  38         },
  39         {
  40             'url': 'https://eu.bbcollab.com/recording/51ed7b50810c4444a106e48cefb3e6b5',
  41             'only_matching': True,
  42         },
  43         {
  44             'url': 'https://au.bbcollab.com/collab/ui/session/playback/load/2bccf7165d7c419ab87afc1ec3f3bb15',
  45             'only_matching': True,
  46         },
  47     ]
  48
  49     def _real_extract(self, url):
  50         mobj = self._match_valid_url(url)
  51         region = mobj.group('region')
  52         video_id = mobj.group('id')
  53         info = self._download_json(
  54             'https://{}.bbcollab.com/collab/api/csa/recordings/{}/data'.format(region, video_id), video_id)
  55         duration = info.get('duration')
  56         title = info['name']
  57         upload_date = info.get('created')
  58         streams = info['streams']
  59         formats = [{'format_id': k, 'url': url} for k, url in streams.items()]
  60
  61         return {
  62             'duration': duration,
  63             'formats': formats,
  64             'id': video_id,
  65             'timestamp': parse_iso8601(upload_date),
  66             'title': title,
  67         }