[extractor] Improve json+ld extraction

Related #5035

diff --git a/yt_dlp/extractor/common.py b/yt_dlp/extractor/common.py
index 87660bb23bd8f63e1dfe06478c389ed8b803120f..d36f025ab823a5675ad1e428dd4ddc94925efc46 100644 (file)
--- a/yt_dlp/extractor/common.py
+++ b/yt_dlp/extractor/common.py
@@ -1536,10 +1536,10 @@ def extract_chapter_information(e):
                 info['chapters'] = chapters
 
         def extract_video_object(e):
-            assert is_type(e, 'VideoObject')
             author = e.get('author')
             info.update({
                 'url': url_or_none(e.get('contentUrl')),
+                'ext': mimetype2ext(e.get('encodingFormat')),
                 'title': unescapeHTML(e.get('name')),
                 'description': unescapeHTML(e.get('description')),
                 'thumbnails': [{'url': unescapeHTML(url)}
@@ -1552,12 +1552,19 @@ def extract_video_object(e):
                 # however some websites are using 'Text' type instead.
                 # 1. https://schema.org/VideoObject
                 'uploader': author.get('name') if isinstance(author, dict) else author if isinstance(author, str) else None,
+                'artist': traverse_obj(e, ('byArtist', 'name'), expected_type=str),
                 'filesize': int_or_none(float_or_none(e.get('contentSize'))),
                 'tbr': int_or_none(e.get('bitrate')),
                 'width': int_or_none(e.get('width')),
                 'height': int_or_none(e.get('height')),
                 'view_count': int_or_none(e.get('interactionCount')),
+                'tags': try_call(lambda: e.get('keywords').split(',')),
             })
+            if is_type(e, 'AudioObject'):
+                info.update({
+                    'vcodec': 'none',
+                    'abr': int_or_none(e.get('bitrate')),
+                })
             extract_interaction_statistic(e)
             extract_chapter_information(e)
 
@@ -1608,7 +1615,7 @@ def traverse_json_ld(json_ld, at_top_level=True):
                         extract_video_object(e['video'][0])
                     elif is_type(traverse_obj(e, ('subjectOf', 0)), 'VideoObject'):
                         extract_video_object(e['subjectOf'][0])
-                elif is_type(e, 'VideoObject'):
+                elif is_type(e, 'VideoObject', 'AudioObject'):
                     extract_video_object(e)
                     if expected_type is None:
                         continue

diff --git a/yt_dlp/extractor/generic.py b/yt_dlp/extractor/generic.py
index 672034c6d73af2fa137af4a8d7ba875028098bf7..73aefc78297b1a916802be71972f40700787de56 100644 (file)
--- a/yt_dlp/extractor/generic.py
+++ b/yt_dlp/extractor/generic.py
@@ -2910,7 +2910,7 @@ def _real_extract(self, url):
         if json_ld.get('url') not in (url, None):
             self.report_detected('JSON LD')
             return merge_dicts({
-                '_type': 'url_transparent',
+                '_type': 'video' if json_ld.get('ext') else 'url_transparent',
                 'url': smuggle_url(json_ld['url'], {
                     'force_videoid': video_id,
                     'to_generic': True,

diff --git a/yt_dlp/utils.py b/yt_dlp/utils.py
index d655bfdd038cb47bd39ddfa19bd1f778f0977224..724e34ef7d84efd9f16c1274359c61175aaab301 100644 (file)
--- a/yt_dlp/utils.py
+++ b/yt_dlp/utils.py
@@ -232,7 +232,7 @@ def random_user_agent():
 ])
 
 PACKED_CODES_RE = r"}\('(.+)',(\d+),(\d+),'([^']+)'\.split\('\|'\)"
-JSON_LD_RE = r'(?is)<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>\s*(?P<json_ld>{.+?})\s*</script>'
+JSON_LD_RE = r'(?is)<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>\s*(?P<json_ld>{.+?}|\[.+?\])\s*</script>'
 
 NUMBER_RE = r'\d+(?:\.\d+)?'