youtube_dl/extractor/americastestkitchen.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5 import re
   6
   7 from .common import InfoExtractor
   8 from ..utils import (
   9     clean_html,
  10     int_or_none,
  11     try_get,
  12     unified_strdate,
  13     unified_timestamp,
  14 )
  15
  16
  17 class AmericasTestKitchenIE(InfoExtractor):
  18     _VALID_URL = r'https?://(?:www\.)?(?:americastestkitchen|cooks(?:country|illustrated))\.com/(?P<resource_type>episode|videos)/(?P<id>\d+)'
  19     _TESTS = [{
  20         'url': 'https://www.americastestkitchen.com/episode/582-weeknight-japanese-suppers',
  21         'md5': 'b861c3e365ac38ad319cfd509c30577f',
  22         'info_dict': {
  23             'id': '5b400b9ee338f922cb06450c',
  24             'title': 'Japanese Suppers',
  25             'ext': 'mp4',
  26             'description': 'md5:64e606bfee910627efc4b5f050de92b3',
  27             'thumbnail': r're:^https?://',
  28             'timestamp': 1523318400,
  29             'upload_date': '20180410',
  30             'release_date': '20180410',
  31             'series': "America's Test Kitchen",
  32             'season_number': 18,
  33             'episode': 'Japanese Suppers',
  34             'episode_number': 15,
  35         },
  36         'params': {
  37             'skip_download': True,
  38         },
  39     }, {
  40         # Metadata parsing behaves differently for newer episodes (705) as opposed to older episodes (582 above)
  41         'url': 'https://www.americastestkitchen.com/episode/705-simple-chicken-dinner',
  42         'md5': '06451608c57651e985a498e69cec17e5',
  43         'info_dict': {
  44             'id': '5fbe8c61bda2010001c6763b',
  45             'title': 'Simple Chicken Dinner',
  46             'ext': 'mp4',
  47             'description': 'md5:eb68737cc2fd4c26ca7db30139d109e7',
  48             'thumbnail': r're:^https?://',
  49             'timestamp': 1610755200,
  50             'upload_date': '20210116',
  51             'release_date': '20210116',
  52             'series': "America's Test Kitchen",
  53             'season_number': 21,
  54             'episode': 'Simple Chicken Dinner',
  55             'episode_number': 3,
  56         },
  57         'params': {
  58             'skip_download': True,
  59         },
  60     }, {
  61         'url': 'https://www.americastestkitchen.com/videos/3420-pan-seared-salmon',
  62         'only_matching': True,
  63     }, {
  64         'url': 'https://www.cookscountry.com/episode/564-when-only-chocolate-will-do',
  65         'only_matching': True,
  66     }, {
  67         'url': 'https://www.cooksillustrated.com/videos/4478-beef-wellington',
  68         'only_matching': True,
  69     }]
  70
  71     def _real_extract(self, url):
  72         resource_type, video_id = re.match(self._VALID_URL, url).groups()
  73         is_episode = resource_type == 'episode'
  74         if is_episode:
  75             resource_type = 'episodes'
  76
  77         resource = self._download_json(
  78             'https://www.americastestkitchen.com/api/v6/%s/%s' % (resource_type, video_id), video_id)
  79         video = resource['video'] if is_episode else resource
  80         episode = resource if is_episode else resource.get('episode') or {}
  81
  82         return {
  83             '_type': 'url_transparent',
  84             'url': 'https://player.zype.com/embed/%s.js?api_key=jZ9GUhRmxcPvX7M3SlfejB6Hle9jyHTdk2jVxG7wOHPLODgncEKVdPYBhuz9iWXQ' % video['zypeId'],
  85             'ie_key': 'Zype',
  86             'description': clean_html(video.get('description')),
  87             'timestamp': unified_timestamp(video.get('publishDate')),
  88             'release_date': unified_strdate(video.get('publishDate')),
  89             'episode_number': int_or_none(episode.get('number')),
  90             'season_number': int_or_none(episode.get('season')),
  91             'series': try_get(episode, lambda x: x['show']['title']),
  92             'episode': episode.get('title'),
  93         }
  94
  95
  96 class AmericasTestKitchenSeasonIE(InfoExtractor):
  97     _VALID_URL = r'https?://(?:www\.)?(?P<show>americastestkitchen|cookscountry)\.com/episodes/browse/season_(?P<id>\d+)'
  98     _TESTS = [{
  99         # ATK Season
 100         'url': 'https://www.americastestkitchen.com/episodes/browse/season_1',
 101         'info_dict': {
 102             'id': 'season_1',
 103             'title': 'Season 1',
 104         },
 105         'playlist_count': 13,
 106     }, {
 107         # Cooks Country Season
 108         'url': 'https://www.cookscountry.com/episodes/browse/season_12',
 109         'info_dict': {
 110             'id': 'season_12',
 111             'title': 'Season 12',
 112         },
 113         'playlist_count': 13,
 114     }]
 115
 116     def _real_extract(self, url):
 117         show_name, season_number = re.match(self._VALID_URL, url).groups()
 118         season_number = int(season_number)
 119
 120         slug = 'atk' if show_name == 'americastestkitchen' else 'cco'
 121
 122         season = 'Season %d' % season_number
 123
 124         season_search = self._download_json(
 125             'https://y1fnzxui30-dsn.algolia.net/1/indexes/everest_search_%s_season_desc_production' % slug,
 126             season, headers={
 127                 'Origin': 'https://www.%s.com' % show_name,
 128                 'X-Algolia-API-Key': '8d504d0099ed27c1b73708d22871d805',
 129                 'X-Algolia-Application-Id': 'Y1FNZXUI30',
 130             }, query={
 131                 'facetFilters': json.dumps([
 132                     'search_season_list:' + season,
 133                     'search_document_klass:episode',
 134                     'search_show_slug:' + slug,
 135                 ]),
 136                 'attributesToRetrieve': 'description,search_%s_episode_number,search_document_date,search_url,title' % slug,
 137                 'attributesToHighlight': '',
 138                 'hitsPerPage': 1000,
 139             })
 140
 141         def entries():
 142             for episode in (season_search.get('hits') or []):
 143                 search_url = episode.get('search_url')
 144                 if not search_url:
 145                     continue
 146                 yield {
 147                     '_type': 'url',
 148                     'url': 'https://www.%s.com%s' % (show_name, search_url),
 149                     'id': try_get(episode, lambda e: e['objectID'].split('_')[-1]),
 150                     'title': episode.get('title'),
 151                     'description': episode.get('description'),
 152                     'timestamp': unified_timestamp(episode.get('search_document_date')),
 153                     'season_number': season_number,
 154                     'episode_number': int_or_none(episode.get('search_%s_episode_number' % slug)),
 155                     'ie_key': AmericasTestKitchenIE.ie_key(),
 156                 }
 157
 158         return self.playlist_result(
 159             entries(), 'season_%d' % season_number, season)