yt_dlp/extractor/americastestkitchen.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     clean_html,
   6     int_or_none,
   7     try_get,
   8     unified_strdate,
   9     unified_timestamp,
  10 )
  11
  12
  13 class AmericasTestKitchenIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?americastestkitchen\.com/(?:cooks(?:country|illustrated)/)?(?P<resource_type>episode|videos)/(?P<id>\d+)'
  15     _TESTS = [{
  16         'url': 'https://www.americastestkitchen.com/episode/582-weeknight-japanese-suppers',
  17         'md5': 'b861c3e365ac38ad319cfd509c30577f',
  18         'info_dict': {
  19             'id': '5b400b9ee338f922cb06450c',
  20             'title': 'Japanese Suppers',
  21             'ext': 'mp4',
  22             'display_id': 'weeknight-japanese-suppers',
  23             'description': 'md5:64e606bfee910627efc4b5f050de92b3',
  24             'timestamp': 1523304000,
  25             'upload_date': '20180409',
  26             'release_date': '20180409',
  27             'series': 'America\'s Test Kitchen',
  28             'season': 'Season 18',
  29             'episode': 'Japanese Suppers',
  30             'season_number': 18,
  31             'episode_number': 15,
  32             'duration': 1376,
  33             'thumbnail': r're:^https?://',
  34             'average_rating': 0,
  35             'view_count': int,
  36         },
  37         'params': {
  38             'skip_download': True,
  39         },
  40     }, {
  41         # Metadata parsing behaves differently for newer episodes (705) as opposed to older episodes (582 above)
  42         'url': 'https://www.americastestkitchen.com/episode/705-simple-chicken-dinner',
  43         'md5': '06451608c57651e985a498e69cec17e5',
  44         'info_dict': {
  45             'id': '5fbe8c61bda2010001c6763b',
  46             'title': 'Simple Chicken Dinner',
  47             'ext': 'mp4',
  48             'display_id': 'atktv_2103_simple-chicken-dinner_full-episode_web-mp4',
  49             'description': 'md5:eb68737cc2fd4c26ca7db30139d109e7',
  50             'timestamp': 1610737200,
  51             'upload_date': '20210115',
  52             'release_date': '20210115',
  53             'series': 'America\'s Test Kitchen',
  54             'season': 'Season 21',
  55             'episode': 'Simple Chicken Dinner',
  56             'season_number': 21,
  57             'episode_number': 3,
  58             'duration': 1397,
  59             'thumbnail': r're:^https?://',
  60             'view_count': int,
  61             'average_rating': 0,
  62         },
  63         'params': {
  64             'skip_download': True,
  65         },
  66     }, {
  67         'url': 'https://www.americastestkitchen.com/videos/3420-pan-seared-salmon',
  68         'only_matching': True,
  69     }, {
  70         'url': 'https://www.americastestkitchen.com/cookscountry/episode/564-when-only-chocolate-will-do',
  71         'only_matching': True,
  72     }, {
  73         'url': 'https://www.americastestkitchen.com/cooksillustrated/videos/4478-beef-wellington',
  74         'only_matching': True,
  75     }]
  76
  77     def _real_extract(self, url):
  78         resource_type, video_id = self._match_valid_url(url).groups()
  79         is_episode = resource_type == 'episode'
  80         if is_episode:
  81             resource_type = 'episodes'
  82
  83         resource = self._download_json(
  84             'https://www.americastestkitchen.com/api/v6/%s/%s' % (resource_type, video_id), video_id)
  85         video = resource['video'] if is_episode else resource
  86         episode = resource if is_episode else resource.get('episode') or {}
  87
  88         return {
  89             '_type': 'url_transparent',
  90             'url': 'https://player.zype.com/embed/%s.js?api_key=jZ9GUhRmxcPvX7M3SlfejB6Hle9jyHTdk2jVxG7wOHPLODgncEKVdPYBhuz9iWXQ' % video['zypeId'],
  91             'ie_key': 'Zype',
  92             'description': clean_html(video.get('description')),
  93             'timestamp': unified_timestamp(video.get('publishDate')),
  94             'release_date': unified_strdate(video.get('publishDate')),
  95             'episode_number': int_or_none(episode.get('number')),
  96             'season_number': int_or_none(episode.get('season')),
  97             'series': try_get(episode, lambda x: x['show']['title']),
  98             'episode': episode.get('title'),
  99         }
 100
 101
 102 class AmericasTestKitchenSeasonIE(InfoExtractor):
 103     _VALID_URL = r'https?://(?:www\.)?americastestkitchen\.com(?P<show>/cookscountry)?/episodes/browse/season_(?P<id>\d+)'
 104     _TESTS = [{
 105         # ATK Season
 106         'url': 'https://www.americastestkitchen.com/episodes/browse/season_1',
 107         'info_dict': {
 108             'id': 'season_1',
 109             'title': 'Season 1',
 110         },
 111         'playlist_count': 13,
 112     }, {
 113         # Cooks Country Season
 114         'url': 'https://www.americastestkitchen.com/cookscountry/episodes/browse/season_12',
 115         'info_dict': {
 116             'id': 'season_12',
 117             'title': 'Season 12',
 118         },
 119         'playlist_count': 13,
 120     }]
 121
 122     def _real_extract(self, url):
 123         show_path, season_number = self._match_valid_url(url).group('show', 'id')
 124         season_number = int(season_number)
 125
 126         slug = 'cco' if show_path == '/cookscountry' else 'atk'
 127
 128         season = 'Season %d' % season_number
 129
 130         season_search = self._download_json(
 131             'https://y1fnzxui30-dsn.algolia.net/1/indexes/everest_search_%s_season_desc_production' % slug,
 132             season, headers={
 133                 'Origin': 'https://www.americastestkitchen.com',
 134                 'X-Algolia-API-Key': '8d504d0099ed27c1b73708d22871d805',
 135                 'X-Algolia-Application-Id': 'Y1FNZXUI30',
 136             }, query={
 137                 'facetFilters': json.dumps([
 138                     'search_season_list:' + season,
 139                     'search_document_klass:episode',
 140                     'search_show_slug:' + slug,
 141                 ]),
 142                 'attributesToRetrieve': 'description,search_%s_episode_number,search_document_date,search_url,title' % slug,
 143                 'attributesToHighlight': '',
 144                 'hitsPerPage': 1000,
 145             })
 146
 147         def entries():
 148             for episode in (season_search.get('hits') or []):
 149                 search_url = episode.get('search_url')  # always formatted like '/episode/123-title-of-episode'
 150                 if not search_url:
 151                     continue
 152                 yield {
 153                     '_type': 'url',
 154                     'url': f'https://www.americastestkitchen.com{show_path or ""}{search_url}',
 155                     'id': try_get(episode, lambda e: e['objectID'].split('_')[-1]),
 156                     'title': episode.get('title'),
 157                     'description': episode.get('description'),
 158                     'timestamp': unified_timestamp(episode.get('search_document_date')),
 159                     'season_number': season_number,
 160                     'episode_number': int_or_none(episode.get('search_%s_episode_number' % slug)),
 161                     'ie_key': AmericasTestKitchenIE.ie_key(),
 162                 }
 163
 164         return self.playlist_result(
 165             entries(), 'season_%d' % season_number, season)