[Backport to 3.3.x][Fixes #9101] Keywords filter only returns the roo…

…t nodes of the tree (#9111) (#9121) (#9124)

* [Fixes #9101] Keywords filter only returns the root nodes of the tree

* [CircleCi] Testing Hierarchical Keywords tree

Co-authored-by: Alessio Fabiani <[email protected]>

geonode/base/api/tests.py

@@ -1074,27 +1074,104 @@ def test_keywords_list(self):
         # Anonymous
         response = self.client.get(url, format='json')
         self.assertEqual(response.status_code, 200)
-        self.assertEqual(response.data['total'], HierarchicalKeyword.objects.count())
+        self.assertEqual(response.data['total'], len(HierarchicalKeyword.resource_keywords_tree(None)))
 
         # Admin
         self.assertTrue(self.client.login(username='admin', password='admin'))
+        admin = get_user_model().objects.get(username='admin')
         response = self.client.get(url, format='json')
         self.assertEqual(response.status_code, 200)
-        self.assertEqual(response.data['total'], HierarchicalKeyword.objects.count())
+        self.assertEqual(response.data['total'], len(HierarchicalKeyword.resource_keywords_tree(admin)))
         # response has link to the response
-        self.assertTrue('link' in response.data['keywords'][0].keys())
+        if response.data['total'] > 0:
+            self.assertTrue('link' in response.data['keywords'][0].keys())
 
         # Authenticated user
         self.assertTrue(self.client.login(username='bobby', password='bob'))
+        bobby = get_user_model().objects.get(username='bobby')
         response = self.client.get(url, format='json')
         self.assertEqual(response.status_code, 200)
-        self.assertEqual(response.data['total'], HierarchicalKeyword.objects.count())
+        self.assertEqual(response.data['total'], len(HierarchicalKeyword.resource_keywords_tree(bobby)))
 
         # Keywords Filtering
         response = self.client.get(f"{url}?filter{{name.icontains}}=Africa", format='json')
         self.assertEqual(response.status_code, 200)
         self.assertEqual(response.data['total'], 0)
 
+        # Testing Hierarchical Keywords tree
+        try:
+            HierarchicalKeyword.objects.filter(slug__in=['a', 'a1', 'a2', 'b', 'b1']).delete()
+            HierarchicalKeyword.add_root(name='a')
+            HierarchicalKeyword.add_root(name='b')
+            a = HierarchicalKeyword.objects.get(slug='a')
+            b = HierarchicalKeyword.objects.get(slug='b')
+            a.add_child(name='a1')
+            a.add_child(name='a2')
+            b.add_child(name='b1')
+            resources = ResourceBase.objects.filter(owner__username='bobby')
+            res1 = resources.first()
+            res2 = resources.last()
+            self.assertNotEqual(res1, res2)
+            res1.keywords.add(HierarchicalKeyword.objects.get(slug='a1'))
+            res1.keywords.add(HierarchicalKeyword.objects.get(slug='a2'))
+            res2.keywords.add(HierarchicalKeyword.objects.get(slug='b1'))
+            resource_keywords = HierarchicalKeyword.resource_keywords_tree(bobby)
+            logger.error(resource_keywords)
+
+            """
+            Final api outcome will be something like
+            [
+                {
+                    'id': 116,
+                    'text': 'a',
+                    'href': 'a',
+                    'tags': [],
+                    'nodes': [
+                        {
+                            'id': 118,
+                            'text': 'a1',
+                            'href': 'a1',
+                            'tags': [1]
+                        },
+                        {
+                            'id': 119,
+                            'text': 'a2',
+                            'href': 'a2',
+                            'tags': [1]
+                        }
+                    ]
+                },
+                {
+                    'id': 117,
+                    'text': 'b',
+                    'href': 'b',
+                    'tags': [],
+                    'nodes': [
+                        {
+                            'id': 120,
+                            'text': 'b1',
+                            'href': 'b1',
+                            'tags': [1]
+                        }
+                    ]
+                },
+                ...
+            ]
+            """
+            url = reverse('keywords-list')
+            # Authenticated user
+            self.assertTrue(self.client.login(username='bobby', password='bob'))
+            response = self.client.get(url, format='json')
+            self.assertEqual(response.status_code, 200)
+            for _kw in response.data["keywords"]:
+                if _kw.get('href') in ['a', 'b']:
+                    self.assertListEqual(_kw.get('tags'), [], _kw.get('tags'))
+                    self.assertEqual(len(_kw.get('nodes')), 2)
+                    for _kw_child in _kw.get('nodes'):
+                        self.assertEqual(_kw_child.get('tags')[0], 1)
+        finally:
+            HierarchicalKeyword.objects.filter(slug__in=['a', 'a1', 'a2', 'b', 'b1']).delete()
+
     def test_tkeywords_list(self):
         """
         Ensure we can access the list of thasaurus keywords.

geonode/base/api/views.py

@@ -70,9 +70,9 @@
     TopicCategorySerializer,
     RegionSerializer,
     ThesaurusKeywordSerializer,
+    ExtraMetadataSerializer
 )
 from .pagination import GeoNodeApiPagination
-from geonode.base.api.serializers import ExtraMetadataSerializer
 from geonode.base.utils import validate_extra_metadata
 
 import logging
@@ -193,11 +193,24 @@ class HierarchicalKeywordViewSet(WithDynamicViewSetMixin, ListModelMixin, Retrie
     """
     API endpoint that lists hierarchical keywords.
     """
+
+    def get_queryset(self):
+        resource_keywords = HierarchicalKeyword.resource_keywords_tree(self.request.user)
+
+        def _get_kw_hrefs(keywords, slugs: list = []):
+            for obj in keywords:
+                if obj.get('tags', []):
+                    slugs.append(obj.get('href'))
+                _get_kw_hrefs(obj.get('nodes', []), slugs)
+            return slugs
+
+        slugs = _get_kw_hrefs(resource_keywords)
+        return HierarchicalKeyword.objects.filter(slug__in=slugs)
+
     permission_classes = [AllowAny, ]
     filter_backends = [
         DynamicFilterBackend, DynamicSortingFilter, DynamicSearchFilter
     ]
-    queryset = HierarchicalKeyword.objects.all()
     serializer_class = HierarchicalKeywordSerializer
     pagination_class = GeoNodeApiPagination
 

geonode/base/models.py

@@ -423,6 +423,7 @@ def _keywords_tree_of_a_child(cls, child, tree, newobj):
                         node = node["nodes"][-1]
                     else:
                         node = item_found
+                        node["nodes"] = getattr(node, "nodes", [])
 
         # All leaves appended but a child which is not a leaf may not be added
         # again, as a leaf, but only its tag count be updated