yt_dlp/extractor/cybrary.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     smuggle_url,
   5     str_or_none,
   6     traverse_obj,
   7     urlencode_postdata,
   8 )
   9
  10
  11 class CybraryBaseIE(InfoExtractor):
  12     _API_KEY = 'AIzaSyCX9ru6j70PX2My1Eq6Q1zoMAhuTdXlzSw'
  13     _ENDPOINTS = {
  14         'course': 'https://app.cybrary.it/courses/api/catalog/browse/course/{}',
  15         'course_enrollment': 'https://app.cybrary.it/courses/api/catalog/{}/enrollment',
  16         'enrollment': 'https://app.cybrary.it/courses/api/enrollment/{}',
  17         'launch': 'https://app.cybrary.it/courses/api/catalog/{}/launch',
  18         'vimeo_oembed': 'https://vimeo.com/api/oembed.json?url=https://vimeo.com/{}',
  19     }
  20     _NETRC_MACHINE = 'cybrary'
  21     _TOKEN = None
  22
  23     def _perform_login(self, username, password):
  24         CybraryBaseIE._TOKEN = self._download_json(
  25             f'https://identitytoolkit.googleapis.com/v1/accounts:signInWithPassword?key={self._API_KEY}',
  26             None, data=urlencode_postdata({'email': username, 'password': password, 'returnSecureToken': True}),
  27             note='Logging in')['idToken']
  28
  29     def _real_initialize(self):
  30         if not self._TOKEN:
  31             self.raise_login_required(method='password')
  32
  33     def _call_api(self, endpoint, item_id):
  34         return self._download_json(
  35             self._ENDPOINTS[endpoint].format(item_id), item_id,
  36             note=f'Downloading {endpoint} JSON metadata',
  37             headers={'Authorization': f'Bearer {self._TOKEN}'})
  38
  39     def _get_vimeo_id(self, activity_id):
  40         launch_api = self._call_api('launch', activity_id)
  41
  42         if launch_api.get('url'):
  43             return self._search_regex(r'https?://player\.vimeo\.com/video/(?P<vimeo_id>[0-9]+)', launch_api['url'], 'vimeo_id')
  44         return traverse_obj(launch_api, ('vendor_data', 'content', ..., 'videoId'), get_all=False)
  45
  46
  47 class CybraryIE(CybraryBaseIE):
  48     _VALID_URL = r'https?://app\.cybrary\.it/immersive/(?P<enrollment>[0-9]+)/activity/(?P<id>[0-9]+)'
  49     _TESTS = [{
  50         'url': 'https://app.cybrary.it/immersive/12487950/activity/63102',
  51         'md5': '9ae12d37e555cb2ed554223a71a701d0',
  52         'info_dict': {
  53             'id': '646609770',
  54             'ext': 'mp4',
  55             'title': 'Getting Started',
  56             'thumbnail': 'https://i.vimeocdn.com/video/1301817996-76a268f0c56cff18a5cecbbdc44131eb9dda0c80eb0b3a036_1280',
  57             'series_id': '63111',
  58             'uploader_url': 'https://vimeo.com/user30867300',
  59             'duration': 88,
  60             'uploader_id': 'user30867300',
  61             'series': 'Cybrary Orientation',
  62             'uploader': 'Cybrary',
  63             'chapter': 'Cybrary Orientation Series',
  64             'chapter_id': '63110',
  65         },
  66         'expected_warnings': ['No authenticators for vimeo'],
  67     }, {
  68         'url': 'https://app.cybrary.it/immersive/12747143/activity/52686',
  69         'md5': '62f26547dccc59c44363e2a13d4ad08d',
  70         'info_dict': {
  71             'id': '445638073',
  72             'ext': 'mp4',
  73             'title': 'Azure Virtual Network IP Addressing',
  74             'thumbnail': 'https://i.vimeocdn.com/video/936667051-1647ace66c627d4a2382185e0dae8deb830309bfddd53f8b2367b2f91e92ed0e-d_1280',
  75             'series_id': '52733',
  76             'uploader_url': 'https://vimeo.com/user30867300',
  77             'duration': 426,
  78             'uploader_id': 'user30867300',
  79             'series': 'AZ-500: Microsoft Azure Security Technologies',
  80             'uploader': 'Cybrary',
  81             'chapter': 'Implement Network Security',
  82             'chapter_id': '52693',
  83         },
  84         'expected_warnings': ['No authenticators for vimeo'],
  85     }]
  86
  87     def _real_extract(self, url):
  88         activity_id, enrollment_id = self._match_valid_url(url).group('id', 'enrollment')
  89         course = self._call_api('enrollment', enrollment_id)['content']
  90         activity = traverse_obj(course, ('learning_modules', ..., 'activities', lambda _, v: int(activity_id) == v['id']), get_all=False)
  91
  92         if activity.get('type') not in ['Video Activity', 'Lesson Activity']:
  93             raise ExtractorError('The activity is not a video', expected=True)
  94
  95         module = next((m for m in course.get('learning_modules') or []
  96                       if int(activity_id) in traverse_obj(m, ('activities', ..., 'id'))), None)
  97
  98         vimeo_id = self._get_vimeo_id(activity_id)
  99
 100         return {
 101             '_type': 'url_transparent',
 102             'series': traverse_obj(course, ('content_description', 'title')),
 103             'series_id': str_or_none(traverse_obj(course, ('content_description', 'id'))),
 104             'id': vimeo_id,
 105             'chapter': module.get('title'),
 106             'chapter_id': str_or_none(module.get('id')),
 107             'title': activity.get('title'),
 108             'url': smuggle_url(f'https://player.vimeo.com/video/{vimeo_id}', {'referer': 'https://api.cybrary.it'}),
 109         }
 110
 111
 112 class CybraryCourseIE(CybraryBaseIE):
 113     _VALID_URL = r'https?://app\.cybrary\.it/browse/course/(?P<id>[\w-]+)/?(?:$|[#?])'
 114     _TESTS = [{
 115         'url': 'https://app.cybrary.it/browse/course/az-500-microsoft-azure-security-technologies',
 116         'info_dict': {
 117             'id': '898',
 118             'title': 'AZ-500: Microsoft Azure Security Technologies',
 119             'description': 'md5:69549d379c0fc1dec92926d4e8b6fbd4',
 120         },
 121         'playlist_count': 59,
 122     }, {
 123         'url': 'https://app.cybrary.it/browse/course/cybrary-orientation',
 124         'info_dict': {
 125             'id': '1245',
 126             'title': 'Cybrary Orientation',
 127             'description': 'md5:9e69ff66b32fe78744e0ad4babe2e88e',
 128         },
 129         'playlist_count': 4,
 130     }]
 131
 132     def _real_extract(self, url):
 133         course_id = self._match_id(url)
 134         course = self._call_api('course', course_id)
 135         enrollment_info = self._call_api('course_enrollment', course['id'])
 136
 137         entries = [self.url_result(
 138             f'https://app.cybrary.it/immersive/{enrollment_info["id"]}/activity/{activity["id"]}')
 139             for activity in traverse_obj(course, ('content_item', 'learning_modules', ..., 'activities', ...))]
 140
 141         return self.playlist_result(
 142             entries,
 143             traverse_obj(course, ('content_item', 'id'), expected_type=str_or_none),
 144             course.get('title'), course.get('short_description'))