yt_dlp/extractor/craftsy.py

   1 import json
   2
   3 from .brightcove import BrightcoveNewIE
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     extract_attributes,
   7     get_element_html_by_class,
   8     get_element_text_and_html_by_tag,
   9 )
  10 from ..utils.traversal import traverse_obj
  11
  12
  13 class CraftsyIE(InfoExtractor):
  14     _VALID_URL = r'https?://www\.craftsy\.com/class/(?P<id>[\w-]+)'
  15     _TESTS = [{
  16         'url': 'https://www.craftsy.com/class/the-midnight-quilt-show-season-5/',
  17         'info_dict': {
  18             'id': 'the-midnight-quilt-show-season-5',
  19             'title': 'The Midnight Quilt Show Season 5',
  20             'description': 'md5:113eda818e985d1a566625fb2f833b7a',
  21         },
  22         'playlist_count': 10,
  23     }, {
  24         'url': 'https://www.craftsy.com/class/sew-your-own-designer-handbag/',
  25         'info_dict': {
  26             'id': 'sew-your-own-designer-handbag',
  27             'title': 'Sew Your Own Designer Handbag',
  28             'description': 'md5:8270d0ef5427d3c895a27351aeaac276',
  29         },
  30         'playlist_mincount': 1,
  31     }, {
  32         'url': 'https://www.craftsy.com/class/all-access-estes-park-wool-market/',
  33         'info_dict': {
  34             'id': 'all-access-estes-park-wool-market',
  35             'title': 'All Access: Estes Park Wool Market',
  36             'description': 'md5:aded1bd8d38ae2fae4dae936c0ae01e7',
  37         },
  38         'playlist_count': 6,
  39     }]
  40
  41     def _real_extract(self, url):
  42         video_id = self._match_id(url)
  43         webpage = self._download_webpage(url, video_id)
  44
  45         video_player = get_element_html_by_class('class-video-player', webpage)
  46         video_data = traverse_obj(video_player, (
  47             {extract_attributes}, 'wire:snapshot', {json.loads}, 'data', {dict})) or {}
  48         video_js = traverse_obj(video_player, (
  49             {lambda x: get_element_text_and_html_by_tag('video-js', x)}, 1, {extract_attributes})) or {}
  50
  51         has_access = video_data.get('userHasAccess')
  52         lessons = traverse_obj(video_data, ('lessons', ..., ..., lambda _, v: v['video_id']))
  53
  54         preview_id = video_js.get('data-video-id')
  55         if preview_id and preview_id not in traverse_obj(lessons, (..., 'video_id')):
  56             if not lessons and not has_access:
  57                 self.report_warning(
  58                     'Only extracting preview. For the full class, pass cookies '
  59                     f'from an account that has access. {self._login_hint()}')
  60             lessons.append({'video_id': preview_id})
  61
  62         if not lessons and not has_access:
  63             self.raise_login_required('You do not have access to this class')
  64
  65         account_id = video_data.get('accountId') or video_js['data-account']
  66
  67         def entries(lessons):
  68             for lesson in lessons:
  69                 yield self.url_result(
  70                     f'http://players.brightcove.net/{account_id}/default_default/index.html?videoId={lesson["video_id"]}',
  71                     BrightcoveNewIE, lesson['video_id'], lesson.get('title'))
  72
  73         return self.playlist_result(
  74             entries(lessons), video_id, self._html_search_meta(('og:title', 'twitter:title'), webpage),
  75             self._html_search_meta(('og:description', 'description'), webpage, default=None))