[americastestkitchen] Add support for downloading entire series (#31493)

Also
* support new sites and URL patterns
* back-port from yt-dlp

Co-authored-by: dirkf <fieldhouse@gmx.net>
This commit is contained in:
Brian Marks 2023-02-02 11:58:21 -05:00 committed by GitHub
parent 295736c9cb
commit 37cbdfa0e7
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -15,7 +15,7 @@ from ..utils import (
class AmericasTestKitchenIE(InfoExtractor): class AmericasTestKitchenIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?(?:americastestkitchen|cooks(?:country|illustrated))\.com/(?P<resource_type>episode|videos)/(?P<id>\d+)' _VALID_URL = r'https?://(?:www\.)?(?:americastestkitchen|cooks(?:country|illustrated))\.com/(?:cooks(?:country|illustrated)/)?(?P<resource_type>episode|videos)/(?P<id>\d+)'
_TESTS = [{ _TESTS = [{
'url': 'https://www.americastestkitchen.com/episode/582-weeknight-japanese-suppers', 'url': 'https://www.americastestkitchen.com/episode/582-weeknight-japanese-suppers',
'md5': 'b861c3e365ac38ad319cfd509c30577f', 'md5': 'b861c3e365ac38ad319cfd509c30577f',
@ -23,15 +23,20 @@ class AmericasTestKitchenIE(InfoExtractor):
'id': '5b400b9ee338f922cb06450c', 'id': '5b400b9ee338f922cb06450c',
'title': 'Japanese Suppers', 'title': 'Japanese Suppers',
'ext': 'mp4', 'ext': 'mp4',
'display_id': 'weeknight-japanese-suppers',
'description': 'md5:64e606bfee910627efc4b5f050de92b3', 'description': 'md5:64e606bfee910627efc4b5f050de92b3',
'thumbnail': r're:^https?://', 'timestamp': 1523304000,
'timestamp': 1523318400, 'upload_date': '20180409',
'upload_date': '20180410', 'release_date': '20180409',
'release_date': '20180410',
'series': "America's Test Kitchen", 'series': "America's Test Kitchen",
'season': 'Season 18',
'season_number': 18, 'season_number': 18,
'episode': 'Japanese Suppers', 'episode': 'Japanese Suppers',
'episode_number': 15, 'episode_number': 15,
'duration': 1376,
'thumbnail': r're:^https?://',
'average_rating': 0,
'view_count': int,
}, },
'params': { 'params': {
'skip_download': True, 'skip_download': True,
@ -44,15 +49,20 @@ class AmericasTestKitchenIE(InfoExtractor):
'id': '5fbe8c61bda2010001c6763b', 'id': '5fbe8c61bda2010001c6763b',
'title': 'Simple Chicken Dinner', 'title': 'Simple Chicken Dinner',
'ext': 'mp4', 'ext': 'mp4',
'display_id': 'atktv_2103_simple-chicken-dinner_full-episode_web-mp4',
'description': 'md5:eb68737cc2fd4c26ca7db30139d109e7', 'description': 'md5:eb68737cc2fd4c26ca7db30139d109e7',
'thumbnail': r're:^https?://', 'timestamp': 1610737200,
'timestamp': 1610755200, 'upload_date': '20210115',
'upload_date': '20210116', 'release_date': '20210115',
'release_date': '20210116',
'series': "America's Test Kitchen", 'series': "America's Test Kitchen",
'season': 'Season 21',
'season_number': 21, 'season_number': 21,
'episode': 'Simple Chicken Dinner', 'episode': 'Simple Chicken Dinner',
'episode_number': 3, 'episode_number': 3,
'duration': 1397,
'thumbnail': r're:^https?://',
'view_count': int,
'average_rating': 0,
}, },
'params': { 'params': {
'skip_download': True, 'skip_download': True,
@ -60,6 +70,12 @@ class AmericasTestKitchenIE(InfoExtractor):
}, { }, {
'url': 'https://www.americastestkitchen.com/videos/3420-pan-seared-salmon', 'url': 'https://www.americastestkitchen.com/videos/3420-pan-seared-salmon',
'only_matching': True, 'only_matching': True,
}, {
'url': 'https://www.americastestkitchen.com/cookscountry/episode/564-when-only-chocolate-will-do',
'only_matching': True,
}, {
'url': 'https://www.americastestkitchen.com/cooksillustrated/videos/4478-beef-wellington',
'only_matching': True,
}, { }, {
'url': 'https://www.cookscountry.com/episode/564-when-only-chocolate-will-do', 'url': 'https://www.cookscountry.com/episode/564-when-only-chocolate-will-do',
'only_matching': True, 'only_matching': True,
@ -94,7 +110,7 @@ class AmericasTestKitchenIE(InfoExtractor):
class AmericasTestKitchenSeasonIE(InfoExtractor): class AmericasTestKitchenSeasonIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?(?P<show>americastestkitchen|cookscountry)\.com/episodes/browse/season_(?P<id>\d+)' _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+)))'
_TESTS = [{ _TESTS = [{
# ATK Season # ATK Season
'url': 'https://www.americastestkitchen.com/episodes/browse/season_1', 'url': 'https://www.americastestkitchen.com/episodes/browse/season_1',
@ -105,48 +121,93 @@ class AmericasTestKitchenSeasonIE(InfoExtractor):
'playlist_count': 13, 'playlist_count': 13,
}, { }, {
# Cooks Country Season # Cooks Country Season
'url': 'https://www.cookscountry.com/episodes/browse/season_12', 'url': 'https://www.americastestkitchen.com/cookscountry/episodes/browse/season_12',
'info_dict': { 'info_dict': {
'id': 'season_12', 'id': 'season_12',
'title': 'Season 12', 'title': 'Season 12',
}, },
'playlist_count': 13, 'playlist_count': 13,
}, {
# America's Test Kitchen Series
'url': 'https://www.americastestkitchen.com/',
'info_dict': {
'id': 'americastestkitchen',
'title': 'America\'s Test Kitchen',
},
'playlist_count': 558,
}, {
# Cooks Country Series
'url': 'https://www.americastestkitchen.com/cookscountry',
'info_dict': {
'id': 'cookscountry',
'title': 'Cook\'s Country',
},
'playlist_count': 199,
}, {
'url': 'https://www.americastestkitchen.com/cookscountry/',
'only_matching': True,
}, {
'url': 'https://www.cookscountry.com/episodes/browse/season_12',
'only_matching': True,
}, {
'url': 'https://www.cookscountry.com',
'only_matching': True,
}, {
'url': 'https://www.americastestkitchen.com/cooksillustrated/',
'only_matching': True,
}, {
'url': 'https://www.cooksillustrated.com',
'only_matching': True,
}] }]
def _real_extract(self, url): def _real_extract(self, url):
show_name, season_number = re.match(self._VALID_URL, url).groups() match = re.match(self._VALID_URL, url).groupdict()
season_number = int(season_number) show = match.get('show2')
show_path = ('/' + show) if show else ''
show = show or match['show']
season_number = int_or_none(match.get('season'))
slug = 'atk' if show_name == 'americastestkitchen' else 'cco' slug, title = {
'americastestkitchen': ('atk', 'America\'s Test Kitchen'),
'cookscountry': ('cco', 'Cook\'s Country'),
'cooksillustrated': ('cio', 'Cook\'s Illustrated'),
}[show]
season = 'Season %d' % season_number facet_filters = [
'search_document_klass:episode',
'search_show_slug:' + slug,
]
if season_number:
playlist_id = 'season_%d' % season_number
playlist_title = 'Season %d' % season_number
facet_filters.append('search_season_list:' + playlist_title)
else:
playlist_id = show
playlist_title = title
season_search = self._download_json( season_search = self._download_json(
'https://y1fnzxui30-dsn.algolia.net/1/indexes/everest_search_%s_season_desc_production' % slug, 'https://y1fnzxui30-dsn.algolia.net/1/indexes/everest_search_%s_season_desc_production' % slug,
season, headers={ playlist_id, headers={
'Origin': 'https://www.%s.com' % show_name, 'Origin': 'https://www.americastestkitchen.com',
'X-Algolia-API-Key': '8d504d0099ed27c1b73708d22871d805', 'X-Algolia-API-Key': '8d504d0099ed27c1b73708d22871d805',
'X-Algolia-Application-Id': 'Y1FNZXUI30', 'X-Algolia-Application-Id': 'Y1FNZXUI30',
}, query={ }, query={
'facetFilters': json.dumps([ 'facetFilters': json.dumps(facet_filters),
'search_season_list:' + season, 'attributesToRetrieve': 'description,search_%s_episode_number,search_document_date,search_url,title,search_atk_episode_season' % slug,
'search_document_klass:episode',
'search_show_slug:' + slug,
]),
'attributesToRetrieve': 'description,search_%s_episode_number,search_document_date,search_url,title' % slug,
'attributesToHighlight': '', 'attributesToHighlight': '',
'hitsPerPage': 1000, 'hitsPerPage': 1000,
}) })
def entries(): def entries():
for episode in (season_search.get('hits') or []): for episode in (season_search.get('hits') or []):
search_url = episode.get('search_url') search_url = episode.get('search_url') # always formatted like '/episode/123-title-of-episode'
if not search_url: if not search_url:
continue continue
yield { yield {
'_type': 'url', '_type': 'url',
'url': 'https://www.%s.com%s' % (show_name, search_url), 'url': 'https://www.americastestkitchen.com%s%s' % (show_path, search_url),
'id': try_get(episode, lambda e: e['objectID'].split('_')[-1]), 'id': try_get(episode, lambda e: e['objectID'].rsplit('_', 1)[-1]),
'title': episode.get('title'), 'title': episode.get('title'),
'description': episode.get('description'), 'description': episode.get('description'),
'timestamp': unified_timestamp(episode.get('search_document_date')), 'timestamp': unified_timestamp(episode.get('search_document_date')),
@ -156,4 +217,4 @@ class AmericasTestKitchenSeasonIE(InfoExtractor):
} }
return self.playlist_result( return self.playlist_result(
entries(), 'season_%d' % season_number, season) entries(), playlist_id, playlist_title)