[ie/MLBArticle] Fix extractor (#9021)

Closes #8682
Authored by: HobbyistDev
This commit is contained in:
HobbyistDev 2024-01-20 05:31:06 +09:00 committed by GitHub
parent 4310b6650e
commit 50e06e21a6
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -355,11 +355,11 @@ class MLBArticleIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '36db7394-343c-4ea3-b8ca-ead2e61bca9a', 'id': '36db7394-343c-4ea3-b8ca-ead2e61bca9a',
'title': 'Machado\'s grab draws hilarious irate reaction', 'title': 'Machado\'s grab draws hilarious irate reaction',
'modified_timestamp': 1650130737, 'modified_timestamp': 1675888370,
'description': 'md5:a19d4eb0487b2cb304e9a176f6b67676', 'description': 'md5:a19d4eb0487b2cb304e9a176f6b67676',
'modified_date': '20220416', 'modified_date': '20230208',
}, },
'playlist_count': 2, 'playlist_mincount': 2,
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -367,15 +367,13 @@ def _real_extract(self, url):
webpage = self._download_webpage(url, display_id) webpage = self._download_webpage(url, display_id)
apollo_cache_json = self._search_json(r'window\.initState\s*=', webpage, 'window.initState', display_id)['apolloCache'] apollo_cache_json = self._search_json(r'window\.initState\s*=', webpage, 'window.initState', display_id)['apolloCache']
content_data_id = traverse_obj( content_real_info = traverse_obj(
apollo_cache_json, ('ROOT_QUERY', lambda k, _: k.startswith('getForgeContent'), 'id'), get_all=False) apollo_cache_json, ('ROOT_QUERY', lambda k, _: k.startswith('getArticle')), get_all=False)
content_real_info = apollo_cache_json[content_data_id]
return self.playlist_from_matches( return self.playlist_from_matches(
traverse_obj(content_real_info, ('parts', lambda _, v: v['typename'] == 'Video', 'id')), traverse_obj(content_real_info, ('parts', lambda _, v: v['__typename'] == 'Video' or v['type'] == 'video')),
getter=lambda x: f'https://www.mlb.com/video/{apollo_cache_json[x]["slug"]}', getter=lambda x: f'https://www.mlb.com/video/{x["slug"]}',
ie=MLBVideoIE, playlist_id=content_real_info.get('_translationId'), ie=MLBVideoIE, playlist_id=content_real_info.get('translationId'),
title=self._html_search_meta('og:title', webpage), title=self._html_search_meta('og:title', webpage),
description=content_real_info.get('summary'), description=content_real_info.get('summary'),
modified_timestamp=parse_iso8601(content_real_info.get('lastUpdatedDate'))) modified_timestamp=parse_iso8601(content_real_info.get('lastUpdatedDate')))