yt_dlp/extractor/gamedevtv.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..networking.exceptions import HTTPError
   5 from ..utils import (
   6     ExtractorError,
   7     clean_html,
   8     int_or_none,
   9     join_nonempty,
  10     parse_iso8601,
  11     str_or_none,
  12     url_or_none,
  13 )
  14 from ..utils.traversal import traverse_obj
  15
  16
  17 class GameDevTVDashboardIE(InfoExtractor):
  18     _VALID_URL = r'https?://(?:www\.)?gamedev\.tv/dashboard/courses/(?P<course_id>\d+)(?:/(?P<lecture_id>\d+))?'
  19     _NETRC_MACHINE = 'gamedevtv'
  20     _TESTS = [{
  21         'url': 'https://www.gamedev.tv/dashboard/courses/25',
  22         'info_dict': {
  23             'id': '25',
  24             'title': 'Complete Blender Creator 3: Learn 3D Modelling for Beginners',
  25             'tags': ['blender', 'course', 'all', 'box modelling', 'sculpting'],
  26             'categories': ['Blender', '3D Art'],
  27             'thumbnail': 'https://gamedev-files.b-cdn.net/courses/qisc9pmu1jdc.jpg',
  28             'upload_date': '20220516',
  29             'timestamp': 1652694420,
  30             'modified_date': '20241027',
  31             'modified_timestamp': 1730049658,
  32         },
  33         'playlist_count': 100,
  34     }, {
  35         'url': 'https://www.gamedev.tv/dashboard/courses/63/2279',
  36         'info_dict': {
  37             'id': 'df04f4d8-68a4-4756-a71b-9ca9446c3a01',
  38             'ext': 'mp4',
  39             'modified_timestamp': 1701695752,
  40             'upload_date': '20230504',
  41             'episode': 'MagicaVoxel Community Course Introduction',
  42             'series_id': '63',
  43             'title': 'MagicaVoxel Community Course Introduction',
  44             'timestamp': 1683195397,
  45             'modified_date': '20231204',
  46             'categories': ['3D Art', 'MagicaVoxel'],
  47             'season': 'MagicaVoxel Community Course',
  48             'tags': ['MagicaVoxel', 'all', 'course'],
  49             'series': 'MagicaVoxel 3D Art Mini Course',
  50             'duration': 1405,
  51             'episode_number': 1,
  52             'season_number': 1,
  53             'season_id': '219',
  54             'description': 'md5:a378738c5bbec1c785d76c067652d650',
  55             'display_id': '63-219-2279',
  56             'alt_title': '1_CC_MVX MagicaVoxel Community Course Introduction.mp4',
  57             'thumbnail': 'https://vz-23691c65-6fa.b-cdn.net/df04f4d8-68a4-4756-a71b-9ca9446c3a01/thumbnail.jpg',
  58         },
  59     }]
  60     _API_HEADERS = {}
  61
  62     def _perform_login(self, username, password):
  63         try:
  64             response = self._download_json(
  65                 'https://api.gamedev.tv/api/students/login', None, 'Logging in',
  66                 headers={'Content-Type': 'application/json'},
  67                 data=json.dumps({
  68                     'email': username,
  69                     'password': password,
  70                     'cart_items': [],
  71                 }).encode())
  72         except ExtractorError as e:
  73             if isinstance(e.cause, HTTPError) and e.cause.status == 401:
  74                 raise ExtractorError('Invalid username/password', expected=True)
  75             raise
  76
  77         self._API_HEADERS['Authorization'] = f'{response["token_type"]} {response["access_token"]}'
  78
  79     def _real_initialize(self):
  80         if not self._API_HEADERS.get('Authorization'):
  81             self.raise_login_required(
  82                 'This content is only available with purchase', method='password')
  83
  84     def _entries(self, data, course_id, course_info, selected_lecture):
  85         for section in traverse_obj(data, ('sections', ..., {dict})):
  86             section_info = traverse_obj(section, {
  87                 'season_id': ('id', {str_or_none}),
  88                 'season': ('title', {str}),
  89                 'season_number': ('order', {int_or_none}),
  90             })
  91             for lecture in traverse_obj(section, ('lectures', lambda _, v: url_or_none(v['video']['playListUrl']))):
  92                 if selected_lecture and str(lecture.get('id')) != selected_lecture:
  93                     continue
  94                 display_id = join_nonempty(course_id, section_info.get('season_id'), lecture.get('id'))
  95                 formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  96                     lecture['video']['playListUrl'], display_id, 'mp4', m3u8_id='hls')
  97                 yield {
  98                     **course_info,
  99                     **section_info,
 100                     'id': display_id,  # fallback
 101                     'display_id': display_id,
 102                     'formats': formats,
 103                     'subtitles': subtitles,
 104                     'series': course_info.get('title'),
 105                     'series_id': course_id,
 106                     **traverse_obj(lecture, {
 107                         'id': ('video', 'guid', {str}),
 108                         'title': ('title', {str}),
 109                         'alt_title': ('video', 'title', {str}),
 110                         'description': ('description', {clean_html}),
 111                         'episode': ('title', {str}),
 112                         'episode_number': ('order', {int_or_none}),
 113                         'duration': ('video', 'duration_in_sec', {int_or_none}),
 114                         'timestamp': ('video', 'created_at', {parse_iso8601}),
 115                         'modified_timestamp': ('video', 'updated_at', {parse_iso8601}),
 116                         'thumbnail': ('video', 'thumbnailUrl', {url_or_none}),
 117                     }),
 118                 }
 119
 120     def _real_extract(self, url):
 121         course_id, lecture_id = self._match_valid_url(url).group('course_id', 'lecture_id')
 122         data = self._download_json(
 123             f'https://api.gamedev.tv/api/courses/my/{course_id}', course_id,
 124             headers=self._API_HEADERS)['data']
 125
 126         course_info = traverse_obj(data, {
 127             'title': ('title', {str}),
 128             'tags': ('tags', ..., 'name', {str}),
 129             'categories': ('categories', ..., 'title', {str}),
 130             'timestamp': ('created_at', {parse_iso8601}),
 131             'modified_timestamp': ('updated_at', {parse_iso8601}),
 132             'thumbnail': ('image', {url_or_none}),
 133         })
 134
 135         entries = self._entries(data, course_id, course_info, lecture_id)
 136         if lecture_id:
 137             lecture = next(entries, None)
 138             if not lecture:
 139                 raise ExtractorError('Lecture not found')
 140             return lecture
 141         return self.playlist_result(entries, course_id, **course_info)