udata 12.0.2.dev13__py3-none-any.whl → 12.0.2.dev14__py3-none-any.whl

udata/harvest/tests/dcat/bnodes.xml

@@ -11,6 +11,7 @@
   xmlns:dcterms="http://purl.org/dc/terms/"
   xmlns:vcard="http://www.w3.org/2006/vcard/ns#"
   xmlns:schema="http://schema.org/"
+  xmlns:skos="http://www.w3.org/2004/02/skos/core#"
 >
   <dcat:Catalog rdf:about="http://data.test.org/">
     <dcat:dataset>
@@ -45,7 +46,6 @@
     <dcterms:title>Sample DCAT Catalog</dcterms:title>
     <dcat:dataset>
       <dcat:Dataset>
-        <dcat:theme>Theme 2</dcat:theme>
         <dcat:contactPoint rdf:resource="http://data.test.org/contacts/1"/>
         <dcat:landingPage>http://data.test.org/datasets/1</dcat:landingPage>
         <dcat:keyword>Tag 3</dcat:keyword>
@@ -75,6 +75,17 @@
         <dcat:distribution rdf:resource="http://data.test.org/datasets/1/resources/1"/>
         <dcterms:issued rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2016-12-14T18:59:02.737480</dcterms:issued>
         <dcterms:identifier>1</dcterms:identifier>
+        <dcat:theme>
+          <skos:Concept>
+            <skos:prefLabel xml:lang="fr">Répartition des espèces</skos:prefLabel>
+            <skos:inScheme>
+              <skos:ConceptScheme>
+                <dcterms:title xml:lang="fr">GEMET - INSPIRE themes, version 1.0</dcterms:title>
+                <dcterms:issued rdf:datatype="http://www.w3.org/2001/XMLSchema#date">2018-07-27</dcterms:issued>
+              </skos:ConceptScheme>
+            </skos:inScheme>
+          </skos:Concept>
+        </dcat:theme>
       </dcat:Dataset>
     </dcat:dataset>
     <dcat:dataset>
@@ -110,6 +121,7 @@
         </dct:spatial>
         <dcterms:identifier>2</dcterms:identifier>
         <dct:conformsTo rdf:nodeID="Ne0189e93917c4f67a412fc44883322e7"/>
+        <dcat:theme rdf:resource="http://bnode.namespace.voc/theme/hy"/>
       </dcat:Dataset>
     </dcat:dataset>
     <dcat:service>
@@ -185,4 +197,8 @@
     <dct:type rdf:resource="http://inspire.ec.europa.eu/glossary/SpatialReferenceSystem"/>
     <dct:title xml:lang="fr">RGF93 / Lambert-93 (EPSG:2154)</dct:title>
   </rdf:Description>
+  <skos:Concept rdf:about="http://bnode.namespace.voc/theme/hy">
+    <skos:inScheme rdf:resource="http://inspire.ec.europa.eu/theme"/>
+    <skos:prefLabel>Hydrographie</skos:prefLabel>
+  </skos:Concept>
 </rdf:RDF>

udata/harvest/tests/test_dcat_backend.py

@@ -445,6 +445,23 @@ class DcatBackendTest:
         assert resources_by_title["Resource 3-1"].schema.url is None
         assert resources_by_title["Resource 3-1"].schema.version == "2.2.0"
 
+    @pytest.mark.options(SCHEMA_CATALOG_URL="https://example.com/schemas")
+    def test_harvest_inspire_themese(self, rmock):
+        rmock.get("https://example.com/schemas", json=ResourceSchemaMockData.get_mock_data())
+
+        filename = "bnodes.xml"
+        url = mock_dcat(rmock, filename)
+        org = OrganizationFactory()
+        source = HarvestSourceFactory(backend="dcat", url=url, organization=org)
+
+        actions.run(source)
+
+        datasets = {d.harvest.dct_identifier: d for d in Dataset.objects}
+
+        assert set(datasets["1"].tags).issuperset(set(["repartition-des-especes", "inspire"]))
+        assert set(datasets["2"].tags).issuperset(set(["hydrographie", "inspire"]))
+        assert "inspire" not in datasets["3"].tags
+
     def test_simple_nested_attributes(self, rmock):
         filename = "nested.jsonld"
         url = mock_dcat(rmock, filename)
@@ -672,6 +689,9 @@ class DcatBackendTest:
         assert dataset.temporal_coverage is not None
         assert dataset.temporal_coverage.start == date(2004, 11, 3)
         assert dataset.temporal_coverage.end == date(2005, 3, 30)
+        assert set(dataset.tags) == set(
+            ["inspire", "biodiversity-dynamics"]
+        )  # The DCAT.theme with rdf:resource don't have labels properly defined
 
     def test_sigoreme_xml_catalog(self, rmock):
         LicenseFactory(id="fr-lo", title="Licence ouverte / Open Licence")
@@ -911,6 +931,7 @@ class CswDcatBackendTest:
                 "oise",
                 "somme",
                 "aisne",
+                # "inspire",  TODO: the geonetwork v4 examples use broken URI as theme resources, check if this is still a problem or not
             ]
         )
         assert dataset.harvest.issued_at.date() == date(2017, 1, 1)
@@ -1085,6 +1106,7 @@ class CswIso19139DcatBackendTest:
                 "donnees-ouvertes",
                 "plu",
                 "usage-des-sols",
+                "inspire",
             ]
         )
         assert dataset.harvest.issued_at.date() == date(2017, 10, 7)
@@ -1195,3 +1217,6 @@ class CswIso19139DcatBackendTest:
         assert dataset.extras["dcat"].get("rights") is None
         for resource in dataset.resources:
             assert resource.extras["dcat"].get("rights") is None
+
+        # Additional INSPIRE tag due to the dataset having a GEMET INSPIRE theme
+        assert "inspire" in dataset.tags

udata/rdf.py

@@ -128,6 +128,12 @@ EU_HVD_CATEGORIES = {
 HVD_LEGISLATION = "http://data.europa.eu/eli/reg_impl/2023/138/oj"
 TAG_TO_EU_HVD_CATEGORIES = {slugify_tag(EU_HVD_CATEGORIES[uri]): uri for uri in EU_HVD_CATEGORIES}
 
+INSPIRE_GEMET_THEME_NAMESPACE = "http://inspire.ec.europa.eu/theme"
+INSPIRE_GEMET_SCHEME_URIS = [
+    INSPIRE_GEMET_THEME_NAMESPACE,
+    "http://www.eionet.europa.eu/gemet/inspire_themes",
+]
+
 AGENT_ROLE_TO_RDF_PREDICATE = {
     "contact": DCAT.contactPoint,
     "publisher": DCT.publisher,
@@ -303,16 +309,37 @@ def theme_labels_from_rdf(rdf):
     """
     Get theme labels to use as keywords.
     Map HVD keywords from known URIs resources if HVD support is activated.
+    Map INSPIRE keyword from known themes if INSPIRE support is activated.
+    - An INSPIRE dataset is a dataset with a theme INSPIRE encoded with gmd:descriptiveKeywords/gmd:MD_Keywords.
+      In DCAT, it is shown as a DCAT.theme with a SKOS.inScheme pointing to to the INSPIRE thesaurus.
+      We filter on this thesaurus based on its name (expecting "GEMET - INSPIRE themes, version 1.0") or its uri.
     """
     for theme in rdf.objects(DCAT.theme):
         if isinstance(theme, RdfResource):
+            label = rdf_value(theme, SKOS.prefLabel)
             uri = theme.identifier.toPython()
             if current_app.config["HVD_SUPPORT"] and uri in EU_HVD_CATEGORIES:
+                # Map label from EU HVD categories
                 label = EU_HVD_CATEGORIES[uri]
                 # Additionnally yield hvd keyword
                 yield "hvd"
-            else:
-                label = rdf_value(theme, SKOS.prefLabel)
+            if current_app.config["INSPIRE_SUPPORT"]:
+                if uri.startswith(INSPIRE_GEMET_THEME_NAMESPACE):
+                    yield "inspire"
+                else:
+                    # Check if the theme belongs to the GEMET INSPIRE scheme
+                    if scheme := theme.value(SKOS.inScheme):
+                        scheme_title = (
+                            rdf_value(scheme, DCT.title)
+                            or rdf_value(scheme, SKOS.prefLabel)
+                            or rdf_value(scheme, RDFS.label)
+                        )
+                        scheme_uri = scheme.identifier.toPython()
+                        if (
+                            scheme_title
+                            and scheme_title.lower() == "gemet - inspire themes, version 1.0"
+                        ) or scheme_uri in INSPIRE_GEMET_SCHEME_URIS:
+                            yield "inspire"
         else:
             label = theme.toPython()
         if label:

udata/settings.py

@@ -324,9 +324,14 @@ class Defaults(object):
     S3_ACCESS_KEY_ID = None
     S3_SECRET_ACCESS_KEY = None
 
-    # Specific support for hvd (map HVD categories URIs to keywords)
+    # Specific support for hvd:
+    # - map HVD categories URIs to keywords
     HVD_SUPPORT = True
 
+    # Specific support for inspire:
+    # - add inspire keyword during harvest if GEMETE INSPIRE thesaurus is used in DCAT.theme
+    INSPIRE_SUPPORT = True
+
     ACTIVATE_TERRITORIES = False
     # The order is important to compute parents/children, smaller first.
     HANDLED_LEVELS = tuple()

{udata-12.0.2.dev13.dist-info → udata-12.0.2.dev14.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: udata
-Version: 12.0.2.dev13
+Version: 12.0.2.dev14
 Summary: Open data portal
 Author-email: Opendata Team <opendatateam@data.gouv.fr>
 Maintainer-email: Opendata Team <opendatateam@data.gouv.fr>

{udata-12.0.2.dev13.dist-info → udata-12.0.2.dev14.dist-info}/RECORD

@@ -7,10 +7,10 @@ udata/errors.py,sha256=E8W7b4PH7c5B85g_nsUMt8fHqMVpDFOZFkO6wMPl6bA,117
 udata/factories.py,sha256=MoklZnU8iwNL25dm3JsoXhoQs1PQWSVYL1WvcUBtJqM,492
 udata/i18n.py,sha256=bC9ajf66YgcYoJffvresLZLa32rb6NsY-JGMtFiVsG4,8163
 udata/mail.py,sha256=Huhx_1QthJkLvuRUuP6jqb5Qq5R4iSmqeEpLVO9ZkQ4,2671
-udata/rdf.py,sha256=4SruNYSJzqwureb8no6ghh0ZXypG7wCahqSpSrBo8K0,19435
+udata/rdf.py,sha256=dQSYVS2NrMRE6fsrbGwXl_roM0k6jtt1JPtMhHxeVGg,20970
 udata/routing.py,sha256=Hnc1ktmKVS-RUHNKw2zYTft2HJ903FhjtlcenQ9igwI,8044
 udata/sentry.py,sha256=ekcxqUSqxfM98TtvCsPaOoX5i2l6PEcYt7kb4l3od-Q,3223
-udata/settings.py,sha256=1gDu1fnorsgzRzkMIEzFjWiVZ_7UPe1kRW-GQulb7O0,21686
+udata/settings.py,sha256=99A0b87QX6AukPlodoaBBd45RpyBINshimnH_3LvJMg,21850
 udata/sitemap.py,sha256=oRRWoPI7ZsFFnUAOqGT1YuXFFKHBe8EcRnUCNHD7xjM,979
 udata/tags.py,sha256=ydq4uokd6bzdeGVSpEXASVtGvDfO2LfQs9mptvvKJCM,631
 udata/tasks.py,sha256=Sv01dhvATtq_oHOBp3J1j1VT1HQe0Pab7zxwIeIdKoo,5122
@@ -324,7 +324,7 @@ udata/harvest/tests/person.jsonld,sha256=I7Ynh-PQlNeD51I1LrCgYOEjhL-WBeb65xzIE_s
 udata/harvest/tests/test_actions.py,sha256=d5TTFTbs4PdBydWICqDtfoeo3zLyzcNDzv4aMH9spxo,25881
 udata/harvest/tests/test_api.py,sha256=gSuICkPy3KVRUhHAyudXVf_gLwiB7SoriUp3DLXWDdA,21611
 udata/harvest/tests/test_base_backend.py,sha256=ow8ecGtD836mUqyPWYjkS5nx0STyT5RMLgBdDyOhts4,19233
-udata/harvest/tests/test_dcat_backend.py,sha256=ekr-cgD1HxteYsZe2_iugKxNnKEZF8V-CRe05Frlt_w,50636
+udata/harvest/tests/test_dcat_backend.py,sha256=WB7jSjRuFvSAjskKVhW5lUfiJHomL3hwebSJ3f8ubEI,51842
 udata/harvest/tests/test_filters.py,sha256=PT2qopEIoXsqi8MsNDRuhNH7jGXiQo8r0uJrCOUd4aM,2465
 udata/harvest/tests/test_models.py,sha256=f9NRR2_S4oZFgF8qOumg0vv-lpnEBJbI5vNtcwFdSqM,831
 udata/harvest/tests/test_notifications.py,sha256=MMzTzkv-GXMNFeOwAi31rdTsAXyLCLOSna41zOtaJG0,816
@@ -344,7 +344,7 @@ udata/harvest/tests/csw_dcat/geonetworkv4-page-1.xml,sha256=k2pKidlQvJpoltGFm9HN
 udata/harvest/tests/csw_dcat/geonetworkv4-page-3.xml,sha256=fsN0E4TVd_ts-sYA612yBP-gRAwpyQWqJdNm7ohczbs,20945
 udata/harvest/tests/csw_dcat/geonetworkv4-page-5.xml,sha256=0VmPp1kspik7YAmOFyr-3yJLzWGA6kuQp_x_w-W385o,21213
 udata/harvest/tests/dcat/bnodes.jsonld,sha256=Leqny-ccp30564yojQYYckw_HKbhR0f5qUCaavc2ruE,7964
-udata/harvest/tests/dcat/bnodes.xml,sha256=bjG-pE2jDuJ7ZNDzQV4JEiMeAHCeX5eMQyUcDecVQ08,11333
+udata/harvest/tests/dcat/bnodes.xml,sha256=k1AGGPD8P4qSIxzS2vorJMNnf1sw4Deq3Kwm_rDJMaM,12137
 udata/harvest/tests/dcat/catalog.xml,sha256=7bXxQDAu-BRppqIq4WAu6QAdOGOSbf900zBzJttP30k,12785
 udata/harvest/tests/dcat/evian.json,sha256=R3RxP5azUuf9aZ9fU7n6iJkfbJ6oj-Zej2cjOtkYr8M,16647
 udata/harvest/tests/dcat/flat.jsonld,sha256=BAw08MDhtW9Px3q6RAoTIqO_OwJmAwBS9EpC8BY_x98,8459
@@ -625,9 +625,9 @@ udata/translations/pt/LC_MESSAGES/udata.mo,sha256=nv80xZLfIfUsSOMBcr29L268FDc_Gt
 udata/translations/pt/LC_MESSAGES/udata.po,sha256=bUp-7Ray8t8ALgJk3Icw1jmiGIc9_pEJQHiGw_2EU2o,50989
 udata/translations/sr/LC_MESSAGES/udata.mo,sha256=Y_XpUxD074wXc63oJTnoVOyOQ2lmBxl-MrgluZ0Qdw4,27961
 udata/translations/sr/LC_MESSAGES/udata.po,sha256=qh8mrz9AFuVQtXYSSP4QWsXLM_Lv3EHVifHT1NflWXY,57529
-udata-12.0.2.dev13.dist-info/licenses/LICENSE,sha256=V8j_M8nAz8PvAOZQocyRDX7keai8UJ9skgmnwqETmdY,34520
-udata-12.0.2.dev13.dist-info/METADATA,sha256=0p4YaygMwNbbY_fnRD4PmOzXTqOb40rc6-mwEAzrgls,5174
-udata-12.0.2.dev13.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-udata-12.0.2.dev13.dist-info/entry_points.txt,sha256=v2u12qO11i2lyLNIp136WmLJ-NHT-Kew3Duu8J-AXPM,614
-udata-12.0.2.dev13.dist-info/top_level.txt,sha256=EF6CE6YSHd_og-8LCEA4q25ALUpWVe8D0okOLdMAE3A,6
-udata-12.0.2.dev13.dist-info/RECORD,,
+udata-12.0.2.dev14.dist-info/licenses/LICENSE,sha256=V8j_M8nAz8PvAOZQocyRDX7keai8UJ9skgmnwqETmdY,34520
+udata-12.0.2.dev14.dist-info/METADATA,sha256=y3BF3EBBTZacX22DKY4alYa20nE8CbbU_tKLwv6iz6s,5174
+udata-12.0.2.dev14.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+udata-12.0.2.dev14.dist-info/entry_points.txt,sha256=v2u12qO11i2lyLNIp136WmLJ-NHT-Kew3Duu8J-AXPM,614
+udata-12.0.2.dev14.dist-info/top_level.txt,sha256=EF6CE6YSHd_og-8LCEA4q25ALUpWVe8D0okOLdMAE3A,6
+udata-12.0.2.dev14.dist-info/RECORD,,