yt_dlp/extractor/getcourseru.py

   1 import re
   2 import time
   3 import urllib.parse
   4
   5 from .common import InfoExtractor
   6 from ..utils import ExtractorError, int_or_none, url_or_none, urlencode_postdata
   7 from ..utils.traversal import traverse_obj
   8
   9
  10 class GetCourseRuPlayerIE(InfoExtractor):
  11     _VALID_URL = r'https?://player02\.getcourse\.ru/sign-player/?\?(?:[^#]+&)?json=[^#&]+'
  12     _EMBED_REGEX = [rf'<iframe[^>]+\bsrc=[\'"](?P<url>{_VALID_URL}[^\'"]*)']
  13     _TESTS = [{
  14         'url': 'http://player02.getcourse.ru/sign-player/?json=eyJ2aWRlb19oYXNoIjoiMTkwYmRmOTNmMWIyOTczNTMwOTg1M2E3YTE5ZTI0YjMiLCJ1c2VyX2lkIjozNTk1MjUxODMsInN1Yl9sb2dpbl91c2VyX2lkIjpudWxsLCJsZXNzb25faWQiOm51bGwsImlwIjoiNDYuMTQyLjE4Mi4yNDciLCJnY19ob3N0IjoiYWNhZGVteW1lbC5vbmxpbmUiLCJ0aW1lIjoxNzA1NDQ5NjQyLCJwYXlsb2FkIjoidV8zNTk1MjUxODMiLCJ1aV9sYW5ndWFnZSI6InJ1IiwiaXNfaGF2ZV9jdXN0b21fc3R5bGUiOnRydWV9&s=354ad2c993d95d5ac629e3133d6cefea&vh-static-feature=zigzag',
  15         'info_dict': {
  16             'id': '513573381',
  17             'title': '190bdf93f1b29735309853a7a19e24b3',
  18             'ext': 'mp4',
  19             'thumbnail': 'https://preview-htz.kinescopecdn.net/preview/190bdf93f1b29735309853a7a19e24b3/preview.jpg?version=1702370546&host=vh-80',
  20             'duration': 1693
  21         },
  22         'skip': 'JWT expired',
  23     }]
  24
  25     def _real_extract(self, url):
  26         webpage = self._download_webpage(url, None, 'Downloading player page')
  27         window_configs = self._search_json(
  28             r'window\.configs\s*=', webpage, 'config', None)
  29         video_id = str(window_configs['gcFileId'])
  30         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  31             window_configs['masterPlaylistUrl'], video_id)
  32
  33         return {
  34             **traverse_obj(window_configs, {
  35                 'title': ('videoHash', {str}),
  36                 'thumbnail': ('previewUrl', {url_or_none}),
  37                 'duration': ('videoDuration', {int_or_none}),
  38             }),
  39             'id': video_id,
  40             'formats': formats,
  41             'subtitles': subtitles
  42         }
  43
  44
  45 class GetCourseRuIE(InfoExtractor):
  46     _NETRC_MACHINE = 'getcourseru'
  47     _DOMAINS = [
  48         'academymel.online',
  49         'marafon.mani-beauty.com',
  50         'on.psbook.ru'
  51     ]
  52     _BASE_URL_RE = rf'https?://(?:(?!player02\.)[^.]+\.getcourse\.(?:ru|io)|{"|".join(map(re.escape, _DOMAINS))})'
  53     _VALID_URL = [
  54         rf'{_BASE_URL_RE}/(?!pl/|teach/)(?P<id>[^?#]+)',
  55         rf'{_BASE_URL_RE}/(:?pl/)?teach/control/lesson/view\?(?:[^#]+&)?id=(?P<id>\d+)',
  56     ]
  57     _TESTS = [{
  58         'url': 'http://academymel.online/3video_1',
  59         'info_dict': {
  60             'id': '3059742',
  61             'display_id': '3video_1',
  62             'title': 'Промоуроки Академии МЕЛ',
  63         },
  64         'playlist_count': 1,
  65         'playlist': [{
  66             'info_dict': {
  67                 'id': '513573381',
  68                 'ext': 'mp4',
  69                 'title': 'Промоуроки Академии МЕЛ',
  70                 'thumbnail': 'https://preview-htz.kinescopecdn.net/preview/190bdf93f1b29735309853a7a19e24b3/preview.jpg?version=1702370546&host=vh-80',
  71                 'duration': 1693
  72             },
  73         }]
  74     }, {
  75         'url': 'https://academymel.getcourse.ru/3video_1',
  76         'info_dict': {
  77             'id': '3059742',
  78             'display_id': '3video_1',
  79             'title': 'Промоуроки Академии МЕЛ',
  80         },
  81         'playlist_count': 1,
  82         'playlist': [{
  83             'info_dict': {
  84                 'id': '513573381',
  85                 'ext': 'mp4',
  86                 'title': 'Промоуроки Академии МЕЛ',
  87                 'thumbnail': 'https://preview-htz.kinescopecdn.net/preview/190bdf93f1b29735309853a7a19e24b3/preview.jpg?version=1702370546&host=vh-80',
  88                 'duration': 1693
  89             },
  90         }]
  91     }, {
  92         'url': 'https://academymel.getcourse.ru/pl/teach/control/lesson/view?id=319141781&editMode=0',
  93         'info_dict': {
  94             'id': '319141781',
  95             'title': '1. Разминка у стены',
  96         },
  97         'playlist_count': 1,
  98         'playlist': [{
  99             'info_dict': {
 100                 'id': '4919601',
 101                 'ext': 'mp4',
 102                 'title': '1. Разминка у стены',
 103                 'thumbnail': 'https://preview-htz.vhcdn.com/preview/5a521788e7dc25b4f70c3dff6512d90e/preview.jpg?version=1703223532&host=vh-81',
 104                 'duration': 704
 105             },
 106         }],
 107         'skip': 'paid lesson'
 108     }, {
 109         'url': 'https://manibeauty.getcourse.ru/pl/teach/control/lesson/view?id=272499894',
 110         'info_dict': {
 111             'id': '272499894',
 112             'title': 'Мотивация к тренировкам',
 113         },
 114         'playlist_count': 1,
 115         'playlist': [{
 116             'info_dict': {
 117                 'id': '447479687',
 118                 'ext': 'mp4',
 119                 'title': 'Мотивация к тренировкам',
 120                 'thumbnail': 'https://preview-htz.vhcdn.com/preview/70ed5b9f489dd03b4aff55bfdff71a26/preview.jpg?version=1685115787&host=vh-71',
 121                 'duration': 30
 122             },
 123         }],
 124         'skip': 'paid lesson'
 125     }, {
 126         'url': 'https://gaismasmandalas.getcourse.io/ATLAUTSEVBUT',
 127         'only_matching': True,
 128     }]
 129
 130     _LOGIN_URL_PATH = '/cms/system/login'
 131
 132     def _login(self, hostname, username, password):
 133         if self._get_cookies(f'https://{hostname}').get('PHPSESSID5'):
 134             return
 135         login_url = f'https://{hostname}{self._LOGIN_URL_PATH}'
 136         webpage = self._download_webpage(login_url, None)
 137
 138         self._request_webpage(
 139             login_url, None, 'Logging in', 'Failed to log in',
 140             data=urlencode_postdata({
 141                 'action': 'processXdget',
 142                 'xdgetId': self._html_search_regex(
 143                     r'<form[^>]+\bclass="[^"]*\bstate-login[^"]*"[^>]+\bdata-xdget-id="([^"]+)"',
 144                     webpage, 'xdgetId'),
 145                 'params[action]': 'login',
 146                 'params[url]': login_url,
 147                 'params[object_type]': 'cms_page',
 148                 'params[object_id]': -1,
 149                 'params[email]': username,
 150                 'params[password]': password,
 151                 'requestTime': int(time.time()),
 152                 'requestSimpleSign': self._html_search_regex(
 153                     r'window.requestSimpleSign\s*=\s*"([\da-f]+)"', webpage, 'simple sign'),
 154             }))
 155
 156     def _real_extract(self, url):
 157         hostname = urllib.parse.urlparse(url).hostname
 158         username, password = self._get_login_info(netrc_machine=hostname)
 159         if username:
 160             self._login(hostname, username, password)
 161
 162         display_id = self._match_id(url)
 163         webpage, urlh = self._download_webpage_handle(url, display_id)
 164         if self._LOGIN_URL_PATH in urlh.url:
 165             raise ExtractorError(
 166                 f'This video is only available for registered users. {self._login_hint("any", netrc=hostname)}',
 167                 expected=True)
 168
 169         playlist_id = self._search_regex(
 170             r'window\.(?:lessonId|gcsObjectId)\s*=\s*(\d+)', webpage, 'playlist id', default=display_id)
 171         title = self._og_search_title(webpage) or self._html_extract_title(webpage)
 172
 173         return self.playlist_from_matches(
 174             re.findall(GetCourseRuPlayerIE._EMBED_REGEX[0], webpage),
 175             playlist_id, title, display_id=display_id, ie=GetCourseRuPlayerIE, video_kwargs={
 176                 'url_transparent': True,
 177                 'title': title,
 178             })