yt_dlp/extractor/brilliantpala.py

   1 import hashlib
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     traverse_obj,
   7     urlencode_postdata,
   8 )
   9
  10
  11 class BrilliantpalaBaseIE(InfoExtractor):
  12     _NETRC_MACHINE = 'brilliantpala'
  13     _DOMAIN = '{subdomain}.brilliantpala.org'
  14
  15     def _initialize_pre_login(self):
  16         self._HOMEPAGE = f'https://{self._DOMAIN}'
  17         self._LOGIN_API = f'{self._HOMEPAGE}/login/'
  18         self._LOGOUT_DEVICES_API = f'{self._HOMEPAGE}/logout_devices/?next=/'
  19         self._CONTENT_API = f'{self._HOMEPAGE}/api/v2.4/contents/{{content_id}}/'
  20         self._HLS_AES_URI = f'{self._HOMEPAGE}/api/v2.5/video_contents/{{content_id}}/key/'
  21
  22     def _get_logged_in_username(self, url, video_id):
  23         webpage, urlh = self._download_webpage_handle(url, video_id)
  24         if urlh.url.startswith(self._LOGIN_API):
  25             self.raise_login_required()
  26         return self._html_search_regex(
  27             r'"username"\s*:\s*"(?P<username>[^"]+)"', webpage, 'logged-in username')
  28
  29     def _perform_login(self, username, password):
  30         login_form = self._hidden_inputs(self._download_webpage(
  31             self._LOGIN_API, None, 'Downloading login page'))
  32         login_form.update({
  33             'username': username,
  34             'password': password,
  35         })
  36         self._set_cookie(self._DOMAIN, 'csrftoken', login_form['csrfmiddlewaretoken'])
  37
  38         logged_page = self._download_webpage(
  39             self._LOGIN_API, None, note='Logging in', headers={'Referer': self._LOGIN_API},
  40             data=urlencode_postdata(login_form))
  41
  42         if self._html_search_regex(
  43                 r'(Your username / email and password)', logged_page, 'auth fail', default=None):
  44             raise ExtractorError('wrong username or password', expected=True)
  45
  46         # the maximum number of logins is one
  47         if self._html_search_regex(
  48                 r'(Logout Other Devices)', logged_page, 'logout devices button', default=None):
  49             logout_device_form = self._hidden_inputs(logged_page)
  50             self._download_webpage(
  51                 self._LOGOUT_DEVICES_API, None, headers={'Referer': self._LOGIN_API},
  52                 note='Logging out other devices', data=urlencode_postdata(logout_device_form))
  53
  54     def _real_extract(self, url):
  55         course_id, content_id = self._match_valid_url(url).group('course_id', 'content_id')
  56         video_id = f'{course_id}-{content_id}'
  57
  58         username = self._get_logged_in_username(url, video_id)
  59
  60         content_json = self._download_json(
  61             self._CONTENT_API.format(content_id=content_id), video_id,
  62             note='Fetching content info', errnote='Unable to fetch content info')
  63
  64         entries = []
  65         for stream in traverse_obj(content_json, ('video', 'streams', lambda _, v: v['id'] and v['url'])):
  66             formats = self._extract_m3u8_formats(stream['url'], video_id, fatal=False)
  67             if not formats:
  68                 continue
  69             entries.append({
  70                 'id': str(stream['id']),
  71                 'title': content_json.get('title'),
  72                 'formats': formats,
  73                 'hls_aes': {'uri': self._HLS_AES_URI.format(content_id=content_id)},
  74                 'http_headers': {'X-Key': hashlib.sha256(username.encode('ascii')).hexdigest()},
  75                 'thumbnail': content_json.get('cover_image'),
  76             })
  77
  78         return self.playlist_result(
  79             entries, playlist_id=video_id, playlist_title=content_json.get('title'))
  80
  81
  82 class BrilliantpalaElearnIE(BrilliantpalaBaseIE):
  83     IE_NAME = 'Brilliantpala:Elearn'
  84     IE_DESC = 'VoD on elearn.brilliantpala.org'
  85     _VALID_URL = r'https?://elearn\.brilliantpala\.org/courses/(?P<course_id>\d+)/contents/(?P<content_id>\d+)/?'
  86     _TESTS = [{
  87         'url': 'https://elearn.brilliantpala.org/courses/42/contents/12345/',
  88         'only_matching': True,
  89     }, {
  90         'url': 'https://elearn.brilliantpala.org/courses/98/contents/36683/',
  91         'info_dict': {
  92             'id': '23577',
  93             'ext': 'mp4',
  94             'title': 'Physical World, Units and Measurements  - 1',
  95             'thumbnail': 'https://d1j3vi2u94ebt0.cloudfront.net/institute/brilliantpalalms/chapter_contents/26237/e657f81b90874be19795c7ea081f8d5c.png',
  96             'live_status': 'not_live',
  97         },
  98         'params': {
  99             'skip_download': True,
 100         },
 101     }]
 102
 103     _DOMAIN = BrilliantpalaBaseIE._DOMAIN.format(subdomain='elearn')
 104
 105
 106 class BrilliantpalaClassesIE(BrilliantpalaBaseIE):
 107     IE_NAME = 'Brilliantpala:Classes'
 108     IE_DESC = 'VoD on classes.brilliantpala.org'
 109     _VALID_URL = r'https?://classes\.brilliantpala\.org/courses/(?P<course_id>\d+)/contents/(?P<content_id>\d+)/?'
 110     _TESTS = [{
 111         'url': 'https://classes.brilliantpala.org/courses/42/contents/12345/',
 112         'only_matching': True,
 113     }, {
 114         'url': 'https://classes.brilliantpala.org/courses/416/contents/25445/',
 115         'info_dict': {
 116             'id': '9128',
 117             'ext': 'mp4',
 118             'title': 'Motion in a Straight Line - Class 1',
 119             'thumbnail': 'https://d3e4y8hquds3ek.cloudfront.net/institute/brilliantpalaelearn/chapter_contents/ff5ba838d0ec43419f67387fe1a01fa8.png',
 120             'live_status': 'not_live',
 121         },
 122         'params': {
 123             'skip_download': True,
 124         },
 125     }]
 126
 127     _DOMAIN = BrilliantpalaBaseIE._DOMAIN.format(subdomain='classes')