[#56] Index subfields as extras_ Solr field

As this is a `text` field that allows free text search
ckan · Jun 13, 2024 · b600493 · b600493
1 parent ad35359
commit b600493
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 3 deletions.
diff --git a/ckanext/dcat/plugins/__init__.py b/ckanext/dcat/plugins/__init__.py
@@ -173,7 +173,7 @@ def before_dataset_index(self, dataset_dict):
                             value = item[key]
                             if not isinstance(value, dict):
                                 # Index a flattened version
-                                new_key = f'{field["field_name"]}__{key}'
+                                new_key = f'extras_{field["field_name"]}__{key}'
                                 if not dataset_dict.get(new_key):
                                     dataset_dict[new_key] = value
                                 else:

diff --git a/ckanext/dcat/tests/test_scheming_support.py b/ckanext/dcat/tests/test_scheming_support.py
@@ -769,12 +769,36 @@ def test_repeating_subfields_index(self):
 
             # Dict sent to Solr
             search_dict = m.mock_calls[1].kwargs["docs"][0]
-            assert search_dict["contact__name"] == "Contact 1 Contact 2"
+            assert search_dict["extras_contact__name"] == "Contact 1 Contact 2"
             assert (
-                search_dict["contact__email"]
+                search_dict["extras_contact__email"]
                 == "contact1@example.org contact2@example.org"
             )
 
+    def test_repeating_subfields_search(self):
+
+        dataset_dict = {
+            # Core fields
+            "name": "test-dataset",
+            "title": "Test DCAT dataset",
+            "notes": "Some notes",
+            # Repeating subfields
+            "contact": [
+                {"name": "Contact 1", "email": "contact1@example.org"},
+                {"name": "Contact 2", "email": "contact2@example.org"},
+            ],
+        }
+
+        dataset = call_action("package_create", **dataset_dict)
+
+        result = call_action("package_search", q="Contact 2")
+
+        assert result["results"][0]["id"] == dataset["id"]
+
+        result = call_action("package_search", q="Contact 3")
+
+        assert result["count"] == 0
+
     def test_spatial_field(self):
 
         dataset_dict = {