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|cooks(?:country|illustrated))\.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         'url': 'https://www.cookscountry.com/episode/564-when-only-chocolate-will-do',
  77         'only_matching': True,
  78     }, {
  79         'url': 'https://www.cooksillustrated.com/videos/4478-beef-wellington',
  80         'only_matching': True,
  81     }]
  82
  83     def _real_extract(self, url):
  84         resource_type, video_id = self._match_valid_url(url).groups()
  85         is_episode = resource_type == 'episode'
  86         if is_episode:
  87             resource_type = 'episodes'
  88
  89         resource = self._download_json(
  90             'https://www.americastestkitchen.com/api/v6/%s/%s' % (resource_type, video_id), video_id)
  91         video = resource['video'] if is_episode else resource
  92         episode = resource if is_episode else resource.get('episode') or {}
  93
  94         return {
  95             '_type': 'url_transparent',
  96             'url': 'https://player.zype.com/embed/%s.js?api_key=jZ9GUhRmxcPvX7M3SlfejB6Hle9jyHTdk2jVxG7wOHPLODgncEKVdPYBhuz9iWXQ' % video['zypeId'],
  97             'ie_key': 'Zype',
  98             'description': clean_html(video.get('description')),
  99             'timestamp': unified_timestamp(video.get('publishDate')),
 100             'release_date': unified_strdate(video.get('publishDate')),
 101             'episode_number': int_or_none(episode.get('number')),
 102             'season_number': int_or_none(episode.get('season')),
 103             'series': try_get(episode, lambda x: x['show']['title']),
 104             'episode': episode.get('title'),
 105         }
 106
 107
 108 class AmericasTestKitchenSeasonIE(InfoExtractor):
 109     _VALID_URL = r'https?://(?:www\.)?(?P<show>americastestkitchen|(?P<cooks>cooks(?:country|illustrated)))\.com(?:(?:/(?P<show2>cooks(?:country|illustrated)))?(?:/?$|(?<!ated)(?<!ated\.com)/episodes/browse/season_(?P<season>\d+)))'
 110     _TESTS = [{
 111         # ATK Season
 112         'url': 'https://www.americastestkitchen.com/episodes/browse/season_1',
 113         'info_dict': {
 114             'id': 'season_1',
 115             'title': 'Season 1',
 116         },
 117         'playlist_count': 13,
 118     }, {
 119         # Cooks Country Season
 120         'url': 'https://www.americastestkitchen.com/cookscountry/episodes/browse/season_12',
 121         'info_dict': {
 122             'id': 'season_12',
 123             'title': 'Season 12',
 124         },
 125         'playlist_count': 13,
 126     }, {
 127         # America's Test Kitchen Series
 128         'url': 'https://www.americastestkitchen.com/',
 129         'info_dict': {
 130             'id': 'americastestkitchen',
 131             'title': 'America\'s Test Kitchen',
 132         },
 133         'playlist_count': 558,
 134     }, {
 135         # Cooks Country Series
 136         'url': 'https://www.americastestkitchen.com/cookscountry',
 137         'info_dict': {
 138             'id': 'cookscountry',
 139             'title': 'Cook\'s Country',
 140         },
 141         'playlist_count': 199,
 142     }, {
 143         'url': 'https://www.americastestkitchen.com/cookscountry/',
 144         'only_matching': True,
 145     }, {
 146         'url': 'https://www.cookscountry.com/episodes/browse/season_12',
 147         'only_matching': True,
 148     }, {
 149         'url': 'https://www.cookscountry.com',
 150         'only_matching': True,
 151     }, {
 152         'url': 'https://www.americastestkitchen.com/cooksillustrated/',
 153         'only_matching': True,
 154     }, {
 155         'url': 'https://www.cooksillustrated.com',
 156         'only_matching': True,
 157     }]
 158
 159     def _real_extract(self, url):
 160         season_number, show1, show = self._match_valid_url(url).group('season', 'show', 'show2')
 161         show_path = ('/' + show) if show else ''
 162         show = show or show1
 163         season_number = int_or_none(season_number)
 164
 165         slug, title = {
 166             'americastestkitchen': ('atk', 'America\'s Test Kitchen'),
 167             'cookscountry': ('cco', 'Cook\'s Country'),
 168             'cooksillustrated': ('cio', 'Cook\'s Illustrated'),
 169         }[show]
 170
 171         facet_filters = [
 172             'search_document_klass:episode',
 173             'search_show_slug:' + slug,
 174         ]
 175
 176         if season_number:
 177             playlist_id = 'season_%d' % season_number
 178             playlist_title = 'Season %d' % season_number
 179             facet_filters.append('search_season_list:' + playlist_title)
 180         else:
 181             playlist_id = show
 182             playlist_title = title
 183
 184         season_search = self._download_json(
 185             'https://y1fnzxui30-dsn.algolia.net/1/indexes/everest_search_%s_season_desc_production' % slug,
 186             playlist_id, headers={
 187                 'Origin': 'https://www.americastestkitchen.com',
 188                 'X-Algolia-API-Key': '8d504d0099ed27c1b73708d22871d805',
 189                 'X-Algolia-Application-Id': 'Y1FNZXUI30',
 190             }, query={
 191                 'facetFilters': json.dumps(facet_filters),
 192                 'attributesToRetrieve': 'description,search_%s_episode_number,search_document_date,search_url,title,search_atk_episode_season' % slug,
 193                 'attributesToHighlight': '',
 194                 'hitsPerPage': 1000,
 195             })
 196
 197         def entries():
 198             for episode in (season_search.get('hits') or []):
 199                 search_url = episode.get('search_url')  # always formatted like '/episode/123-title-of-episode'
 200                 if not search_url:
 201                     continue
 202                 yield {
 203                     '_type': 'url',
 204                     'url': f'https://www.americastestkitchen.com{show_path or ""}{search_url}',
 205                     'id': try_get(episode, lambda e: e['objectID'].split('_')[-1]),
 206                     'title': episode.get('title'),
 207                     'description': episode.get('description'),
 208                     'timestamp': unified_timestamp(episode.get('search_document_date')),
 209                     'season_number': season_number,
 210                     'episode_number': int_or_none(episode.get('search_%s_episode_number' % slug)),
 211                     'ie_key': AmericasTestKitchenIE.ie_key(),
 212                 }
 213
 214         return self.playlist_result(
 215             entries(), playlist_id, playlist_title)