From 2be0c24897cf99557a8f969c160077159184b8b1 Mon Sep 17 00:00:00 2001 From: Michael Skyba Date: Mon, 4 Nov 2024 23:54:19 -0500 Subject: [PATCH] [ie/suno] add fallbacks for basic page metadata --- yt_dlp/extractor/suno.py | 39 +++++++++++++++++++++++++++++---------- 1 file changed, 29 insertions(+), 10 deletions(-) diff --git a/yt_dlp/extractor/suno.py b/yt_dlp/extractor/suno.py index 881cfb66df..d06736c7a0 100644 --- a/yt_dlp/extractor/suno.py +++ b/yt_dlp/extractor/suno.py @@ -4,7 +4,23 @@ from ..utils import ExtractorError, unescapeHTML -class SunoIE(InfoExtractor): +class SunoBaseIE(InfoExtractor): + def _get_title(self, webpage): + return self._html_search_meta( + ['og:title', 'twitter:title'], webpage, 'title', + default=None) or self._html_extract_title(webpage) + + def _get_description(self, webpage): + return self._html_search_meta( + ['og:description', 'description', 'twitter:description'], + webpage, 'description', default=None) + + def _get_thumbnail(self, webpage): + return self._html_search_meta( + ['og:image', 'twitter:image'], webpage, 'thumbnail', default=None) + + +class SunoIE(SunoBaseIE): _VALID_URL = r'https?://(?:www\.)?suno\.com/song/(?P[-a-f0-9]+)' _TESTS = [ { @@ -35,16 +51,19 @@ def _real_extract(self, url): video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) + url = self._html_search_meta( + ['og:audio', 'twitter:player:stream'], webpage, 'url', default=None) + return { 'id': video_id, - 'title': self._og_search_title(webpage), - 'description': self._og_search_description(webpage), - 'thumbnail': self._og_search_thumbnail(webpage), - 'url': self._og_search_property('audio', webpage), + 'title': self._get_title(webpage), + 'description': self._get_description(webpage), + 'thumbnail': self._get_thumbnail(webpage), + 'url': url, } -class SunoPlaylistIE(InfoExtractor): +class SunoPlaylistIE(SunoBaseIE): _VALID_URL = r'https?://(?:www\.)?suno\.com/playlist/(?P[-a-f0-9]+)' _TESTS = [ { @@ -69,7 +88,7 @@ class SunoPlaylistIE(InfoExtractor): 'info_dict': { 'id': '568eeaab-dfbf-4da6-aa0a-0fb1a32330de', 'title': 'Piano by @kunal | Suno', - 'description': 'Here are some good piano ', + 'description': 'Here are some good piano', 'thumbnail': r're:https?://.*0ecc0956-3b17-4d4b-8504-55849dd75e22.*\.jpeg$', }, 'playlist': [ @@ -120,9 +139,9 @@ def _real_extract(self, url): return { '_type': 'playlist', 'id': playlist_id, - 'title': self._og_search_title(webpage), - 'description': self._og_search_description(webpage), - 'thumbnail': self._og_search_thumbnail(webpage), + 'title': self._get_title(webpage), + 'description': self._get_description(webpage), + 'thumbnail': self._get_thumbnail(webpage), 'entries': [{ 'id': song_tuple[0],