From 460da07439718d9af1e3661da2a23e05a913a2e6 Mon Sep 17 00:00:00 2001 From: bashonly <88596187+bashonly@users.noreply.github.com> Date: Thu, 16 Mar 2023 14:54:25 -0500 Subject: [PATCH] [extractor/genius] Add support for articles (#6474) Closes #6465 Authored by: bashonly --- yt_dlp/extractor/genius.py | 34 ++++++++++++++++++++++++++-------- 1 file changed, 26 insertions(+), 8 deletions(-) diff --git a/yt_dlp/extractor/genius.py b/yt_dlp/extractor/genius.py index 62f5a28ff..57c25e71e 100644 --- a/yt_dlp/extractor/genius.py +++ b/yt_dlp/extractor/genius.py @@ -10,7 +10,7 @@ class GeniusIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?genius\.com/videos/(?P[^?/#]+)' + _VALID_URL = r'https?://(?:www\.)?genius\.com/(?:videos|(?P
a))/(?P[^?/#]+)' _TESTS = [{ 'url': 'https://genius.com/videos/Vince-staples-breaks-down-the-meaning-of-when-sparks-fly', 'md5': '64c2ad98cfafcfda23bfa0ad0c512f4c', @@ -41,19 +41,37 @@ class GeniusIE(InfoExtractor): 'timestamp': 1631209167, 'thumbnail': r're:^https?://.*\.jpg$', }, + }, { + 'url': 'https://genius.com/a/cordae-anderson-paak-break-down-the-meaning-of-two-tens', + 'md5': 'f98a4e03b16b0a2821bd6e52fb3cc9d7', + 'info_dict': { + 'id': '6321509903112', + 'ext': 'mp4', + 'title': 'Cordae & Anderson .Paak Breaks Down The Meaning Of “Two Tens”', + 'description': 'md5:1255f0e1161d07342ce56a8464ac339d', + 'tags': ['song id: 5457554'], + 'uploader_id': '4863540648001', + 'duration': 361.813, + 'upload_date': '20230301', + 'timestamp': 1677703908, + 'thumbnail': r're:^https?://.*\.jpg$', + }, }] def _real_extract(self, url): - display_id = self._match_id(url) + display_id, is_article = self._match_valid_url(url).group('id', 'article') webpage = self._download_webpage(url, display_id) metadata = self._search_json( - r'[^?/#]+)-lyrics[?/#]?' + _VALID_URL = r'https?://(?:www\.)?genius\.com/(?P[^?/#]+)-lyrics(?:[?/#]|$)' _TESTS = [{ 'url': 'https://genius.com/Lil-baby-heyy-lyrics', 'playlist_mincount': 2,