[#56] Store geometry in spatial field for indexing

If the `spatial_coverage` field is present, store the first geometry
found so ckanext-spatial can pick it up for spatial indexing.

Added indexing tests

ckanext/dcat/plugins/__init__.py

@@ -2,6 +2,7 @@
 
 from builtins import object
 import os
+import json
 
 from ckantoolkit import config
 
@@ -146,6 +147,7 @@ def before_dataset_index(self, dataset_dict):
         except KeyError:
             pass
 
+        spatial = None
         if schema:
             for field in schema['dataset_fields']:
                 if field['field_name'] in dataset_dict and 'repeating_subfields' in field:
@@ -156,10 +158,36 @@ def before_dataset_index(self, dataset_dict):
                                 # Index a flattened version
                                 new_key = f'{field["field_name"]}__{key}'
                                 if not dataset_dict.get(new_key):
-                                    dataset_dict[new_key] = ""
-                                dataset_dict[new_key] += " " + value
-
-                    dataset_dict.pop(field['field_name'], None)
+                                    dataset_dict[new_key] = value
+                                else:
+                                    dataset_dict[new_key] += ' ' + value
+
+                    subfields = dataset_dict.pop(field['field_name'], None)
+                    if field['field_name'] == 'spatial_coverage':
+                        spatial = subfields
+
+        # Store the first geometry found so ckanext-spatial can pick it up for indexing
+        def _check_for_a_geom(spatial_dict):
+            value = None
+
+            for field in ('geom', 'bbox', 'centroid'):
+                if spatial_dict.get(field):
+                    value = spatial_dict[field]
+                    if isinstance(value, dict):
+                        try:
+                            value = json.dumps(value)
+                            break
+                        except ValueError:
+                            pass
+            return value
+
+        if spatial and not dataset_dict.get('spatial'):
+            for item in spatial:
+                value = _check_for_a_geom(item)
+                if value:
+                    dataset_dict['spatial'] = value
+                    dataset_dict['extras_spatial'] = value
+                    break
 
         return dataset_dict
 

ckanext/dcat/tests/test_scheming_support.py

@@ -1,3 +1,6 @@
+from unittest import mock
+import json
+
 import pytest
 from rdflib.namespace import RDF
 from rdflib.term import URIRef
@@ -311,9 +314,7 @@ def test_e2e_ckan_to_dcat(self):
             GEOJSON_IMT,
         )
         # Geometry in WKT
-        wkt_geom = wkt.dumps(
-            dataset["spatial_coverage"][0]["geom"], decimals=4
-        )
+        wkt_geom = wkt.dumps(dataset["spatial_coverage"][0]["geom"], decimals=4)
         assert self._triple(g, spatial[0][2], LOCN.geometry, wkt_geom, GSP.wktLiteral)
 
         distribution_ref = self._triple(g, dataset_ref, DCAT.distribution, None)[2]
@@ -606,3 +607,78 @@ def test_e2e_dcat_to_ckan(self):
         assert resource["access_services"][0]["endpoint_url"] == [
             "http://publications.europa.eu/webapi/rdf/sparql"
         ]
+
+
+@pytest.mark.usefixtures("with_plugins", "clean_db")
+@pytest.mark.ckan_config("ckan.plugins", "dcat scheming_datasets")
+@pytest.mark.ckan_config(
+    "scheming.dataset_schemas", "ckanext.dcat.schemas:dcat_ap_2.1.yaml"
+)
+@pytest.mark.ckan_config("scheming.presets", "ckanext.scheming:presets.json")
+@pytest.mark.ckan_config(
+    "ckanext.dcat.rdf.profiles", "euro_dcat_ap_2 euro_dcat_ap_scheming"
+)
+class TestSchemingIndexFields:
+    def test_repeating_subfields_index(self):
+
+        dataset_dict = {
+            # Core fields
+            "name": "test-dataset",
+            "title": "Test DCAT dataset",
+            "notes": "Some notes",
+            # Repeating subfields
+            "contact": [
+                {"name": "Contact 1", "email": "[email protected]"},
+                {"name": "Contact 2", "email": "[email protected]"},
+            ],
+        }
+
+        with mock.patch("ckan.lib.search.index.make_connection") as m:
+            call_action("package_create", **dataset_dict)
+
+            # 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["contact__email"]
+                == "[email protected] [email protected]"
+            )
+
+    def test_spatial_field(self):
+
+        dataset_dict = {
+            # Core fields
+            "name": "test-dataset",
+            "title": "Test DCAT dataset",
+            "notes": "Some notes",
+            "spatial_coverage": [
+                {
+                    "uri": "https://sws.geonames.org/6361390/",
+                    "centroid": {"type": "Point", "coordinates": [1.26639, 41.12386]},
+                },
+                {
+                    "geom": {
+                        "type": "Polygon",
+                        "coordinates": [
+                            [
+                                [11.9936, 54.0486],
+                                [11.9936, 54.2466],
+                                [12.3045, 54.2466],
+                                [12.3045, 54.0486],
+                                [11.9936, 54.0486],
+                            ]
+                        ],
+                    },
+                    "text": "Tarragona",
+                },
+            ],
+        }
+
+        with mock.patch("ckan.lib.search.index.make_connection") as m:
+            call_action("package_create", **dataset_dict)
+
+            # Dict sent to Solr
+            search_dict = m.mock_calls[1].kwargs["docs"][0]
+            assert search_dict["spatial"] == json.dumps(
+                dataset_dict["spatial_coverage"][0]["centroid"]
+            )