Expose OGC services distributions as dataservice (#3203)

Close datagouv/data.gouv.fr#1538

Inspired from the conversion in the SEMIC XSLT:
https://github.com/SEMICeu/iso-19139-to-dcat-ap/blob/f61b2921dd398b90b2dd2db14085e75687f7616b/iso-19139-to-dcat-ap.xsl#L1419

If the resource is an OGC Service, we add a `DCAT.accessService`
property on the distribution and describe the DataService accordingly.

We support resources that have a format of `(ogc:)wms` or `(ogc:)wfs`.
We don't base our decision on `request=GetCapabilities` as in the XSLT
for now.

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## Current (in progress)
 
+- Expose OGC services distributions as dataservice [#3203](https://github.com/opendatateam/udata/pull/3203)
 - Add a matomo "campaign" parameter on links in emails if `MAIL_CAMPAIGN` is configured [#3190](https://github.com/opendatateam/udata/pull/3190)
 - Add DCAT-AP HVD properties in RDF output if the dataservice or its datasets are tagged hvd [#3187](https://github.com/opendatateam/udata/pull/3187)
 

udata/core/dataset/constants.py

@@ -72,6 +72,8 @@
     ]
 )
 
+OGC_SERVICE_FORMATS = ["ogc:wms", "ogc:wfs", "wms", "wfs"]
+
 CHECKSUM_TYPES = ("sha1", "sha2", "sha256", "md5", "crc")
 DEFAULT_CHECKSUM_TYPE = "sha1"
 

udata/core/dataset/rdf.py

@@ -49,7 +49,7 @@
 from udata.uris import endpoint_for
 from udata.utils import get_by, safe_unicode
 
-from .constants import UPDATE_FREQUENCIES
+from .constants import OGC_SERVICE_FORMATS, UPDATE_FREQUENCIES
 from .models import Checksum, Dataset, License, Resource
 
 log = logging.getLogger(__name__)
@@ -126,6 +126,44 @@ def owner_to_rdf(dataset, graph=None):
     return
 
 
+def ogc_service_to_rdf(dataset, resource, graph=None, is_hvd=False):
+    """
+    Build a dataservice on the fly for OGC services distributions
+    Inspired from https://github.com/SEMICeu/iso-19139-to-dcat-ap/blob/f61b2921dd398b90b2dd2db14085e75687f7616b/iso-19139-to-dcat-ap.xsl#L1419
+    """
+    graph = graph or Graph(namespace_manager=namespace_manager)
+    service = graph.resource(BNode())
+    service.set(RDF.type, DCAT.DataService)
+    service.set(DCT.title, Literal(resource.title))
+    service.set(DCAT.endpointURL, URIRef(resource.url.split("?")[0]))
+    if "request=getcapabilities" in resource.url.lower():
+        service.set(DCAT.endpointDescription, URIRef(resource.url))
+    service.set(
+        DCT.conformsTo,
+        URIRef("http://www.opengeospatial.org/standards/" + resource.format.split(":")[-1]),
+    )
+
+    if dataset and dataset.license:
+        service.add(DCT.rights, Literal(dataset.license.title))
+        if dataset.license.url:
+            service.add(DCT.license, URIRef(dataset.license.url))
+
+    if dataset and dataset.contact_point:
+        contact_point = contact_point_to_rdf(dataset.contact_point, graph)
+        if contact_point:
+            service.set(DCAT.contactPoint, contact_point)
+
+    if is_hvd:
+        # DCAT-AP HVD applicable legislation is also expected at the distribution > accessService level
+        service.add(DCATAP.applicableLegislation, URIRef(HVD_LEGISLATION))
+        for tag in dataset.tags:
+            # Add HVD category if this dataset is tagged HVD
+            if tag in TAG_TO_EU_HVD_CATEGORIES:
+                service.add(DCATAP.hvdCategory, URIRef(TAG_TO_EU_HVD_CATEGORIES[tag]))
+
+    return service
+
+
 def resource_to_rdf(resource, dataset=None, graph=None, is_hvd=False):
     """
     Map a Resource domain model to a DCAT/RDF graph
@@ -175,6 +213,11 @@ def resource_to_rdf(resource, dataset=None, graph=None, is_hvd=False):
     if is_hvd:
         # DCAT-AP HVD applicable legislation is also expected at the distribution level
         r.add(DCATAP.applicableLegislation, URIRef(HVD_LEGISLATION))
+
+    # Add access service for known OGC service formats
+    if resource.format in OGC_SERVICE_FORMATS:
+        r.add(DCAT.accessService, ogc_service_to_rdf(dataset, resource, graph, is_hvd))
+
     return r
 
 

udata/tests/dataset/test_dataset_rdf.py

@@ -197,6 +197,34 @@ def test_all_resource_fields(self):
         assert r.graph.value(checksum.identifier, SPDX.algorithm) == SPDX.checksumAlgorithm_sha1
         assert checksum.value(SPDX.checksumValue) == Literal(resource.checksum.value)
 
+    def test_ogc_resource_access_service(self):
+        license = LicenseFactory()
+        resource = ResourceFactory(
+            format="ogc:wms",
+            url="https://services.data.shom.fr/INSPIRE/wms/r?service=WMS&request=GetCapabilities&version=1.3.0",
+        )
+        contact = ContactPointFactory()
+        dataset = DatasetFactory(resources=[resource], license=license, contact_point=contact)
+
+        r = resource_to_rdf(resource, dataset)
+
+        service = r.value(DCAT.accessService)
+        assert service.value(RDF.type).identifier == DCAT.DataService
+        assert service.value(DCT.title) == Literal(resource.title)
+        assert service.value(DCAT.endpointDescription).identifier == URIRef(
+            "https://services.data.shom.fr/INSPIRE/wms/r?service=WMS&request=GetCapabilities&version=1.3.0"
+        )
+        assert service.value(DCAT.endpointURL).identifier == URIRef(
+            "https://services.data.shom.fr/INSPIRE/wms/r"
+        )
+        assert service.value(DCT.conformsTo).identifier == URIRef(
+            "http://www.opengeospatial.org/standards/wms"
+        )
+        assert service.value(DCT.license).identifier == URIRef(license.url)
+
+        contact_rdf = service.value(DCAT.contactPoint)
+        assert contact_rdf.value(RDF.type).identifier == VCARD.Kind
+
     def test_temporal_coverage(self):
         start = faker.past_date(start_date="-30d")
         end = faker.future_date(end_date="+30d")