[extractor] Support default implicit graph in JSON-LD (#1983)

Original PR: https://github.com/ytdl-org/youtube-dl/pull/30229

Per W3C JSON-LD v1.1 §4.9 (non-normative ref):

    When a JSON-LD document's top-level structure is a map that contains
    no other keys than @graph and optionally @context (properties that
    are not mapped to an IRI or a keyword are ignored), @graph is
    considered to express the otherwise implicit default graph.

Authored by: zmousm

diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py
index cbca22c91d3059094eeb32fb24dd10ab8ff6de6c..cf06dbde46d2130869e8a6d8e6cd0bcbc7701bb3 100644 (file)
--- a/test/test_InfoExtractor.py
+++ b/test/test_InfoExtractor.py
@@ -99,10 +99,10 @@ def test_html_search_meta(self):
         self.assertRaises(RegexNotFoundError, ie._html_search_meta, ('z', 'x'), html, None, fatal=True)
 
     def test_search_json_ld_realworld(self):
-        # https://github.com/ytdl-org/youtube-dl/issues/23306
-        expect_dict(
-            self,
-            self.ie._search_json_ld(r'''<script type="application/ld+json">
+        _TESTS = [
+            # https://github.com/ytdl-org/youtube-dl/issues/23306
+            (
+                r'''<script type="application/ld+json">
 {
 "@context": "http://schema.org/",
 "@type": "VideoObject",
@@ -135,17 +135,86 @@ def test_search_json_ld_realworld(self):
 "name": "Kleio Valentien",
 "url": "https://www.eporner.com/pornstar/kleio-valentien/"
 }]}
-</script>''', None),
-            {
-                'title': '1 On 1 With Kleio',
-                'description': 'Kleio Valentien',
-                'url': 'https://gvideo.eporner.com/xN49A1cT3eB/xN49A1cT3eB.mp4',
-                'timestamp': 1449347075,
-                'duration': 743.0,
-                'view_count': 1120958,
-                'width': 1920,
-                'height': 1080,
-            })
+                </script>''',
+                {
+                    'title': '1 On 1 With Kleio',
+                    'description': 'Kleio Valentien',
+                    'url': 'https://gvideo.eporner.com/xN49A1cT3eB/xN49A1cT3eB.mp4',
+                    'timestamp': 1449347075,
+                    'duration': 743.0,
+                    'view_count': 1120958,
+                    'width': 1920,
+                    'height': 1080,
+                },
+                {},
+            ),
+            (
+                r'''<script type="application/ld+json">
+      {
+      "@context": "https://schema.org",
+      "@graph": [
+      {
+      "@type": "NewsArticle",
+      "mainEntityOfPage": {
+      "@type": "WebPage",
+      "@id": "https://www.ant1news.gr/Society/article/620286/symmoria-anilikon-dikigoros-thymaton-ithelan-na-toys-apoteleiosoyn"
+      },
+      "headline": "Συμμορία ανηλίκων – δικηγόρος θυμάτων: ήθελαν να τους αποτελειώσουν",
+      "name": "Συμμορία ανηλίκων – δικηγόρος θυμάτων: ήθελαν να τους αποτελειώσουν",
+      "description": "Τα παιδιά δέχθηκαν την επίθεση επειδή αρνήθηκαν να γίνουν μέλη της συμμορίας, ανέφερε ο Γ. Ζαχαρόπουλος.",
+      "image": {
+      "@type": "ImageObject",
+      "url": "https://ant1media.azureedge.net/imgHandler/1100/a635c968-be71-447c-bf9c-80d843ece21e.jpg",
+      "width": 1100,
+      "height": 756            },
+      "datePublished": "2021-11-10T08:50:00+03:00",
+      "dateModified": "2021-11-10T08:52:53+03:00",
+      "author": {
+      "@type": "Person",
+      "@id": "https://www.ant1news.gr/",
+      "name": "Ant1news",
+      "image": "https://www.ant1news.gr/images/logo-e5d7e4b3e714c88e8d2eca96130142f6.png",
+      "url": "https://www.ant1news.gr/"
+      },
+      "publisher": {
+      "@type": "Organization",
+      "@id": "https://www.ant1news.gr#publisher",
+      "name": "Ant1news",
+      "url": "https://www.ant1news.gr",
+      "logo": {
+      "@type": "ImageObject",
+      "url": "https://www.ant1news.gr/images/logo-e5d7e4b3e714c88e8d2eca96130142f6.png",
+      "width": 400,
+      "height": 400                },
+      "sameAs": [
+      "https://www.facebook.com/Ant1news.gr",
+      "https://twitter.com/antennanews",
+      "https://www.youtube.com/channel/UC0smvAbfczoN75dP0Hw4Pzw",
+      "https://www.instagram.com/ant1news/"
+      ]
+      },
+
+      "keywords": "μαχαίρωμα,συμμορία ανηλίκων,ΕΙΔΗΣΕΙΣ,ΕΙΔΗΣΕΙΣ ΣΗΜΕΡΑ,ΝΕΑ,Κοινωνία - Ant1news",
+
+
+      "articleSection": "Κοινωνία"
+      }
+      ]
+      }
+                </script>''',
+                {
+                    'timestamp': 1636523400,
+                    'title': 'md5:91fe569e952e4d146485740ae927662b',
+                },
+                {'expected_type': 'NewsArticle'},
+            ),
+        ]
+        for html, expected_dict, search_json_ld_kwargs in _TESTS:
+            expect_dict(
+                self,
+                self.ie._search_json_ld(html, None, **search_json_ld_kwargs),
+                expected_dict
+            )
 
     def test_download_json(self):
         uri = encode_data_uri(b'{"foo": "blah"}', 'application/json')

diff --git a/yt_dlp/extractor/common.py b/yt_dlp/extractor/common.py
index ebf2e3ceacc41273d1c1ecdbf9e76e549e022fce..52099b4b428a82e74b5ba807798eb4a1ea0c96e7 100644 (file)
--- a/yt_dlp/extractor/common.py
+++ b/yt_dlp/extractor/common.py
@@ -1451,8 +1451,13 @@ def extract_video_object(e):
             })
             extract_interaction_statistic(e)
 
-        for e in json_ld:
-            if '@context' in e:
+        def traverse_json_ld(json_ld, at_top_level=True):
+            for e in json_ld:
+                if at_top_level and '@context' not in e:
+                    continue
+                if at_top_level and set(e.keys()) == {'@context', '@graph'}:
+                    traverse_json_ld(variadic(e['@graph'], allowed_types=(dict,)), at_top_level=False)
+                    break
                 item_type = e.get('@type')
                 if expected_type is not None and expected_type != item_type:
                     continue
@@ -1488,7 +1493,7 @@ def extract_video_object(e):
                     info.update({
                         'timestamp': parse_iso8601(e.get('datePublished')),
                         'title': unescapeHTML(e.get('headline')),
-                        'description': unescapeHTML(e.get('articleBody')),
+                        'description': unescapeHTML(e.get('articleBody') or e.get('description')),
                     })
                 elif item_type == 'VideoObject':
                     extract_video_object(e)
@@ -1503,6 +1508,8 @@ def extract_video_object(e):
                     continue
                 else:
                     break
+        traverse_json_ld(json_ld)
+
         return dict((k, v) for k, v in info.items() if v is not None)
 
     def _search_nextjs_data(self, webpage, video_id, **kw):