Add bounding box property

ckanext/dcat/profiles/__init__.py

@@ -6,6 +6,7 @@
     RDFS,
     DCAT,
     DCATAP,
+    DCATUS,
     DCT,
     ADMS,
     VCARD,

ckanext/dcat/profiles/base.py

@@ -16,6 +16,7 @@
 DCT = Namespace("http://purl.org/dc/terms/")
 DCAT = Namespace("http://www.w3.org/ns/dcat#")
 DCATAP = Namespace("http://data.europa.eu/r5r/")
+DCATUS = Namespace("http://resources.data.gov/ontology/dcat-us#")
 ADMS = Namespace("http://www.w3.org/ns/adms#")
 VCARD = Namespace("http://www.w3.org/2006/vcard/ns#")
 FOAF = Namespace("http://xmlns.com/foaf/0.1/")
@@ -30,6 +31,7 @@
     "dct": DCT,
     "dcat": DCAT,
     "dcatap": DCATAP,
+    "dcatus": DCATUS,
     "adms": ADMS,
     "vcard": VCARD,
     "foaf": FOAF,

ckanext/dcat/profiles/dcat_us_3.py

@@ -1,11 +1,15 @@
-from rdflib import Literal
+from decimal import Decimal, DecimalException
+
+from rdflib import Literal, BNode
 
 from ckanext.dcat.profiles import (
     DCAT,
+    DCATUS,
     DCT,
     FOAF,
     RDF,
     SKOS,
+    XSD,
 )
 from ckanext.dcat.utils import resource_uri
 
@@ -57,6 +61,21 @@ def graph_from_catalog(self, catalog_dict, catalog_ref):
 
     def _parse_dataset_v3_us(self, dataset_dict, dataset_ref):
 
+        g = self.g
+
+        # Bounding box
+        for bbox_ref in g.objects(dataset_ref, DCATUS.geographicBoundingBox):
+            if not dataset_dict.get("bbox"):
+                dataset_dict["bbox"] = []
+            dataset_dict["bbox"].append(
+                {
+                    "west": self._object_value(bbox_ref, DCATUS.westBoundingLongitude),
+                    "east": self._object_value(bbox_ref, DCATUS.eastBoundingLongitude),
+                    "north": self._object_value(bbox_ref, DCATUS.northBoundingLatitude),
+                    "south": self._object_value(bbox_ref, DCATUS.southBoundingLatitude),
+                }
+            )
+
         for distribution_ref in self._distributions(dataset_ref):
 
             # Distribution identifier
@@ -91,6 +110,35 @@ def _graph_from_dataset_v3_us(self, dataset_dict, dataset_ref):
             if name:
                 g.add((publisher_ref, SKOS.prefLabel, Literal(name)))
 
+        # Bounding box
+        # TODO: we could fall back to spatial or spatial_coverage's bbox/geom
+        bboxes = self._get_dataset_value(dataset_dict, "bbox")
+        if bboxes:
+            for bbox in bboxes:
+                bbox_ref = BNode()
+                g.add((dataset_ref, DCATUS.geographicBoundingBox, bbox_ref))
+                g.add((bbox_ref, RDF.type, DCATUS.geographicBoundingBox))
+
+                def add_bounding(predicate, value):
+                    try:
+                        g.add(
+                            (
+                                bbox_ref,
+                                predicate,
+                                Literal(value, datatype=XSD.decimal),
+                            )
+                        )
+                    except (ValueError, TypeError, DecimalException):
+                        g.add((bbox_ref, predicate, Literal(value)))
+
+                for item in (
+                    (DCATUS.westBoundingLongitude, bbox["west"]),
+                    (DCATUS.eastBoundingLongitude, bbox["east"]),
+                    (DCATUS.northBoundingLatitude, bbox["north"]),
+                    (DCATUS.southBoundingLatitude, bbox["south"]),
+                ):
+                    add_bounding(item[0], item[1])
+
         for resource_dict in dataset_dict.get("resources", []):
 
             distribution_ref = CleanedURIRef(resource_uri(resource_dict))

ckanext/dcat/schemas/dcat_us_full.yaml

@@ -181,6 +181,23 @@ dataset_fields:
   label: Temporal resolution
   help_text: Minimum time period resolvable in the dataset.
 
+- field_name: bbox
+  label: Geographic Bounding Box
+  repeating_subfields:
+
+    - field_name: west
+      label: West Longitude
+
+    - field_name: east
+      label: East Longitude
+
+    - field_name: north
+      label: North Latitude
+
+    - field_name: south
+      label: South Latitude
+  help_text: A geographic bounding box in WGS84 coordinate system (Lat/Long) that describes the spatial extent of the dataset.
+
 - field_name: spatial_coverage
   label: Spatial coverage
   repeating_subfields:

ckanext/dcat/tests/profiles/dcat_us_3/test_dcat_us_3_profile_parse.py

@@ -151,3 +151,39 @@ def test_e2e_dcat_to_ckan(self):
         assert resource["access_services"][0]["endpoint_url"] == [
             "http://publications.europa.eu/webapi/rdf/sparql"
         ]
+
+    def test_bbox(self):
+
+        data = """
+        @prefix dcat: <http://www.w3.org/ns/dcat#> .
+        @prefix dcat-us: <http://resources.data.gov/ontology/dcat-us#> .
+        @prefix dcterms: <http://purl.org/dc/terms/> .
+        @prefix locn: <http://www.w3.org/ns/locn#> .
+        @prefix gsp: <http://www.opengis.net/ont/geosparql#> .
+
+        <https://example.com/dataset1>
+          a dcat:Dataset ;
+          dcterms:title "Dataset 1" ;
+          dcterms:description "This is a dataset" ;
+          dcterms:publisher <https://example.com/publisher1> ;
+          dcat-us:geographicBoundingBox [
+              a dcat-us:GeographicBoundingBox ;
+              dcat-us:westBoundingLongitude 22.3;
+              dcat-us:eastBoundingLongitude 10.3;
+              dcat-us:northBoundingLatitude 50.2;
+              dcat-us:southBoundingLatitude 20.2;
+          ]
+        .
+        """
+        p = RDFParser()
+
+        p.parse(data, _format="ttl")
+
+        datasets = [d for d in p.datasets()]
+
+        dataset = datasets[0]
+
+        assert dataset["bbox"][0]["west"] == "22.3"
+        assert dataset["bbox"][0]["east"] == "10.3"
+        assert dataset["bbox"][0]["north"] == "50.2"
+        assert dataset["bbox"][0]["south"] == "20.2"

ckanext/dcat/tests/profiles/dcat_us_3/test_dcat_us_3_profile_serialize.py

@@ -11,9 +11,8 @@
 
 from ckanext.dcat import utils
 from ckanext.dcat.profiles import (
-    RDF,
     DCAT,
-    DCATAP,
+    DCATUS,
     DCT,
     ADMS,
     XSD,
@@ -317,7 +316,7 @@ def test_e2e_ckan_to_dcat(self):
 
         # Resources: statements
         statement = [s for s in g.objects(distribution_ref, DCT.rights)][0]
-        assert self._triple(g, statement, RDFS.label, resource['rights'])
+        assert self._triple(g, statement, RDFS.label, resource["rights"])
 
     def test_distribution_identifier(self):
 
@@ -366,6 +365,48 @@ def test_distribution_identifier_falls_back_to_id(self):
         distribution_ref = self._triple(g, dataset_ref, DCAT.distribution, None)[2]
         resource = dataset_dict["resources"][0]
 
+        assert self._triple(g, distribution_ref, DCT.identifier, resource["id"])
+
+    def test_bbox(self):
+        dataset_dict = {
+            "name": "test-dcat-us",
+            "description": "Test",
+            "bbox": [
+                {"west": -179.15, "east": -129.98, "north": 71.54, "south": 51.21}
+            ],
+        }
+
+        s = RDFSerializer(profiles=DCAT_AP_PROFILES)
+        g = s.g
+
+        dataset_ref = s.graph_from_dataset(dataset_dict)
+
+        bbox_ref = [s for s in g.objects(dataset_ref, DCATUS.geographicBoundingBox)][0]
+        assert self._triple(
+            g,
+            bbox_ref,
+            DCATUS.westBoundingLongitude,
+            dataset_dict["bbox"][0]["west"],
+            data_type=XSD.decimal,
+        )
+        assert self._triple(
+            g,
+            bbox_ref,
+            DCATUS.eastBoundingLongitude,
+            dataset_dict["bbox"][0]["east"],
+            data_type=XSD.decimal,
+        )
         assert self._triple(
-            g, distribution_ref, DCT.identifier, resource["id"]
+            g,
+            bbox_ref,
+            DCATUS.northBoundingLatitude,
+            dataset_dict["bbox"][0]["north"],
+            data_type=XSD.decimal,
+        )
+        assert self._triple(
+            g,
+            bbox_ref,
+            DCATUS.southBoundingLatitude,
+            dataset_dict["bbox"][0]["south"],
+            data_type=XSD.decimal,
         )

ckanext/dcat/tests/shacl/test_shacl.py

@@ -215,6 +215,7 @@ def test_validate_dcat_us_3_graph():
 
     graph = graph_from_dataset("ckan_full_dataset_dcat_us_vocabularies.json")
 
+    graph.serialize(destination="graph.ttl")
     path = _get_shacl_file_path("dcat-us_3.0_shacl_shapes.ttl")
     r = validate(graph, shacl_graph=path)
     conforms, results_graph, results_text = r

examples/ckan/ckan_full_dataset_dcat_us_vocabularies.json

@@ -76,9 +76,22 @@
             "start": "2024-04-10",
             "end": "2024-05-29"
         }
-
     ],
     "temporal_resolution": "PT15M",
+    "bbox": [
+        {
+            "west": -179.15,
+            "east": -129.98,
+            "north": 71.54,
+            "south": 51.21
+        },
+        {
+            "west": -125.0,
+            "east": -66.9,
+            "north": 49.4,
+            "south": 24.4
+        }
+    ],
     "spatial_coverage": [
         {
             "geom": {