Merge a2fd63ce22 into c5098961b0

Let JSON-LD extract description from articleBody and fall back to
description field when processing (News)Article typed nodes

test/test_InfoExtractor.py

@@ -187,10 +187,10 @@ class TestInfoExtractor(unittest.TestCase):
         self.assertEqual(search['track']['id'], 'testid')
 
     def test_search_json_ld_realworld(self):
+        _TESTS = [
             # https://github.com/ytdl-org/youtube-dl/issues/23306
-        expect_dict(
-            self,
-            self.ie._search_json_ld(r'''<script type="application/ld+json">
+            (
+                r'''<script type="application/ld+json">
 {
 "@context": "http://schema.org/",
 "@type": "VideoObject",
@@ -223,7 +223,7 @@ class TestInfoExtractor(unittest.TestCase):
 "name": "Kleio Valentien",
 "url": "https://www.eporner.com/pornstar/kleio-valentien/"
 }]}
-</script>''', None),
+                </script>''',
                 {
                     'title': '1 On 1 With Kleio',
                     'description': 'Kleio Valentien',
@@ -233,7 +233,76 @@ class TestInfoExtractor(unittest.TestCase):
                     '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')

youtube_dl/extractor/common.py

@@ -1438,8 +1438,16 @@ class InfoExtractor(object):
             })
             extract_interaction_statistic(e)
 
+        def traverse_json_ld(json_ld, info, at_top_level=True):
             for e in json_ld:
-            if '@context' in e:
+                if at_top_level and '@context' not in e:
+                    continue
+                if at_top_level and all(k in ('@context', '@graph') for k in e):
+                    graph = e['@graph']
+                    if isinstance(graph, dict):
+                        graph = [graph]
+                    traverse_json_ld(graph, info, at_top_level=False)
+                    break
                 item_type = e.get('@type')
                 if expected_type is not None and expected_type != item_type:
                     continue
@@ -1472,7 +1480,7 @@ class InfoExtractor(object):
                     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)
@@ -1487,6 +1495,8 @@ class InfoExtractor(object):
                     continue
                 else:
                     break
+        traverse_json_ld(json_ld, info)
+
         return dict((k, v) for k, v in info.items() if v is not None)
 
     def _search_nextjs_data(self, webpage, video_id, **kw):