yt_dlp/extractor/craftsy.py

   1 from .brightcove import BrightcoveNewIE
   2 from .common import InfoExtractor
   3
   4 from ..utils import (
   5     dict_get,
   6     get_element_by_id,
   7     js_to_json,
   8     traverse_obj,
   9 )
  10
  11
  12 class CraftsyIE(InfoExtractor):
  13     _VALID_URL = r'https?://www\.craftsy\.com/class/(?P<id>[\w-]+)'
  14     _TESTS = [{
  15         'url': 'https://www.craftsy.com/class/the-midnight-quilt-show-season-5/',
  16         'info_dict': {
  17             'id': 'the-midnight-quilt-show-season-5',
  18             'title': 'The Midnight Quilt Show Season 5',
  19             'description': 'md5:113eda818e985d1a566625fb2f833b7a',
  20         },
  21         'playlist_count': 10,
  22     }, {
  23         'url': 'https://www.craftsy.com/class/sew-your-own-designer-handbag/',
  24         'info_dict': {
  25             'id': 'sew-your-own-designer-handbag',
  26             'title': 'Sew Your Own Designer Handbag',
  27             'description': 'md5:8270d0ef5427d3c895a27351aeaac276',
  28         },
  29         'playlist_mincount': 1,
  30     }, {
  31         'url': 'https://www.craftsy.com/class/all-access-estes-park-wool-market/',
  32         'info_dict': {
  33             'id': 'all-access-estes-park-wool-market',
  34             'title': 'All Access: Estes Park Wool Market',
  35             'description': 'md5:aded1bd8d38ae2fae4dae936c0ae01e7',
  36         },
  37         'playlist_count': 6,
  38     }]
  39
  40     def _real_extract(self, url):
  41         video_id = self._match_id(url)
  42         webpage = self._download_webpage(url, video_id)
  43
  44         video_data = self._parse_json(self._search_regex(
  45             r'class_video_player_vars\s*=\s*({.*})\s*;',
  46             get_element_by_id('vidstore-classes_class-video-player-js-extra', webpage),
  47             'video data'), video_id, transform_source=js_to_json)
  48
  49         account_id = traverse_obj(video_data, ('video_player', 'bc_account_id'))
  50
  51         entries = []
  52         class_preview = traverse_obj(video_data, ('video_player', 'class_preview'))
  53         if class_preview:
  54             v_id = class_preview.get('video_id')
  55             entries.append(self.url_result(
  56                 f'http://players.brightcove.net/{account_id}/default_default/index.html?videoId={v_id}',
  57                 BrightcoveNewIE, v_id, class_preview.get('title')))
  58
  59         if dict_get(video_data, ('is_free', 'user_has_access')):
  60             entries += [
  61                 self.url_result(
  62                     f'http://players.brightcove.net/{account_id}/default_default/index.html?videoId={lesson["video_id"]}',
  63                     BrightcoveNewIE, lesson['video_id'], lesson.get('title'))
  64                 for lesson in video_data['lessons']]
  65
  66         return self.playlist_result(
  67             entries, video_id, video_data.get('class_title'),
  68             self._html_search_meta(('og:description', 'description'), webpage, default=None))