udata 8.0.2.dev29304__py2.py3-none-any.whl → 8.0.2.dev29339__py2.py3-none-any.whl

udata/core/dataset/rdf.py

@@ -22,9 +22,10 @@ from udata.frontend.markdown import parse_html
 from udata.core.dataset.models import HarvestDatasetMetadata, HarvestResourceMetadata
 from udata.models import db, ContactPoint
 from udata.rdf import (
-    DCAT, DCT, FREQ, SCV, SKOS, SPDX, SCHEMA, EUFREQ, EUFORMAT, IANAFORMAT, VCARD, RDFS,
-    namespace_manager, schema_from_rdf, url_from_rdf
+    DCAT, DCATAP, DCT, FREQ, SCV, SKOS, SPDX, SCHEMA, EUFREQ, EUFORMAT, IANAFORMAT, VCARD, RDFS,
+    HVD_LEGISLATION, namespace_manager, schema_from_rdf, url_from_rdf
 )
+from udata.tags import slug as slugify_tag
 from udata.utils import get_by, safe_unicode
 from udata.uris import endpoint_for
 
@@ -85,6 +86,7 @@ EU_HVD_CATEGORIES = {
     "http://data.europa.eu/bna/c_dd313021": "Observation de la terre et environnement",
     "http://data.europa.eu/bna/c_e1da4e07": "Statistiques"
 }
+TAG_TO_EU_HVD_CATEGORIES = {slugify_tag(EU_HVD_CATEGORIES[uri]): uri for uri in EU_HVD_CATEGORIES}
 
 
 class HTMLDetector(HTMLParser):
@@ -141,7 +143,7 @@ def owner_to_rdf(dataset, graph=None):
     return
 
 
-def resource_to_rdf(resource, dataset=None, graph=None):
+def resource_to_rdf(resource, dataset=None, graph=None, is_hvd=False):
     '''
     Map a Resource domain model to a DCAT/RDF graph
     '''
@@ -180,6 +182,9 @@ def resource_to_rdf(resource, dataset=None, graph=None):
         checksum.add(SPDX.algorithm, getattr(SPDX, algorithm))
         checksum.add(SPDX.checksumValue, Literal(resource.checksum.value))
         r.add(SPDX.checksum, checksum)
+    if is_hvd:
+        # DCAT-AP HVD applicable legislation is also expected at the distribution level
+        r.add(DCATAP.applicableLegislation, URIRef(HVD_LEGISLATION))
     return r
 
 
@@ -214,11 +219,20 @@ def dataset_to_rdf(dataset, graph=None):
     if dataset.acronym:
         d.set(SKOS.altLabel, Literal(dataset.acronym))
 
+    # Add DCAT-AP HVD properties if the dataset is tagged hvd.
+    # See https://semiceu.github.io/DCAT-AP/releases/2.2.0-hvd/
+    is_hvd = current_app.config['HVD_SUPPORT'] and 'hvd' in dataset.tags
+    if is_hvd:
+        d.add(DCATAP.applicableLegislation, URIRef(HVD_LEGISLATION))
+
     for tag in dataset.tags:
         d.add(DCAT.keyword, Literal(tag))
+        # Add HVD category if this dataset is tagged HVD
+        if is_hvd and tag in TAG_TO_EU_HVD_CATEGORIES:
+            d.add(DCATAP.hvdCategory, URIRef(TAG_TO_EU_HVD_CATEGORIES[tag]))
 
     for resource in dataset.resources:
-        d.add(DCAT.distribution, resource_to_rdf(resource, dataset, graph))
+        d.add(DCAT.distribution, resource_to_rdf(resource, dataset, graph, is_hvd))
 
     if dataset.temporal_coverage:
         d.set(DCT.temporal, temporal_to_rdf(dataset.temporal_coverage, graph))

udata/core/site/api.py

@@ -105,7 +105,10 @@ class SiteRdfCatalogFormat(API):
         params = multi_to_dict(request.args)
         page = int(params.get('page', 1))
         page_size = int(params.get('page_size', 100))
-        datasets = Dataset.objects.visible().paginate(page, page_size)
+        datasets = Dataset.objects.visible()
+        if 'tag' in params:
+            datasets = datasets.filter(tags=params.get('tag', ''))
+        datasets = datasets.paginate(page, page_size)
         catalog = build_catalog(current_site, datasets, format=format)
         # bypass flask-restplus make_response, since graph_response
         # is handling the content negociation directly

udata/rdf.py

@@ -21,6 +21,7 @@ log = logging.getLogger(__name__)
 # Extra Namespaces
 ADMS = Namespace('http://www.w3.org/ns/adms#')
 DCAT = Namespace('http://www.w3.org/ns/dcat#')
+DCATAP = Namespace('http://data.europa.eu/r5r/')
 HYDRA = Namespace('http://www.w3.org/ns/hydra/core#')
 SCHEMA = Namespace('http://schema.org/')
 SCV = Namespace('http://purl.org/NET/scovo#')
@@ -35,6 +36,7 @@ VCARD = Namespace('http://www.w3.org/2006/vcard/ns#')
 
 namespace_manager = NamespaceManager(Graph())
 namespace_manager.bind('dcat', DCAT)
+namespace_manager.bind('dcatap', DCATAP)
 namespace_manager.bind('dct', DCT)
 namespace_manager.bind('foaf', FOAF)
 namespace_manager.bind('foaf', FOAF)
@@ -98,6 +100,8 @@ RDF_EXTENSIONS = {
 # Includes control characters, unicode surrogate characters and unicode end-of-plane non-characters
 ILLEGAL_XML_CHARS = '[\x00-\x08\x0b\x0c\x0e-\x1F\uD800-\uDFFF\uFFFE\uFFFF]'
 
+HVD_LEGISLATION = 'http://data.europa.eu/eli/reg_impl/2023/138/oj'
+
 
 def guess_format(string):
     '''Guess format given an extension or a mime-type'''

udata/tests/dataset/test_dataset_rdf.py

@@ -18,11 +18,11 @@ from udata.core.dataset.factories import (
 from udata.core.dataset.rdf import (
     dataset_to_rdf, dataset_from_rdf, resource_to_rdf, resource_from_rdf,
     temporal_from_rdf, frequency_to_rdf, frequency_from_rdf,
-    EU_RDF_REQUENCIES
+    EU_RDF_REQUENCIES, TAG_TO_EU_HVD_CATEGORIES
 )
 from udata.core.organization.factories import OrganizationFactory
 from udata.i18n import gettext as _
-from udata.rdf import DCAT, DCT, FREQ, SPDX, SCHEMA, SKOS
+from udata.rdf import DCAT, DCATAP, DCT, FREQ, SPDX, SCHEMA, SKOS, HVD_LEGISLATION
 from udata.utils import faker
 from udata.tests.helpers import assert200, assert_redirects
 
@@ -181,6 +181,21 @@ class DatasetToRdfTest:
         assert str(d.identifier) == 'https://somewhere.org/dataset'
         assert d.value(DCT.identifier) == Literal('an-identifier')
 
+    def test_hvd_dataset(self):
+        '''Test that a dataset tagged hvd has appropriate DCAT-AP HVD properties'''
+        dataset = DatasetFactory(
+            resources=ResourceFactory.build_batch(3),
+            tags=['hvd', 'mobilite', 'test']
+        )
+        d = dataset_to_rdf(dataset)
+
+        assert d.value(DCATAP.applicableLegislation).identifier == URIRef(HVD_LEGISLATION)
+        assert d.value(DCATAP.hvdCategory).identifier == URIRef(
+            TAG_TO_EU_HVD_CATEGORIES['mobilite']
+        )
+        for distrib in d.objects(DCAT.distribution):
+            assert distrib.value(DCATAP.applicableLegislation).identifier == URIRef(HVD_LEGISLATION)
+
 
 @pytest.mark.usefixtures('clean_db')
 class RdfToDatasetTest:

udata/tests/site/test_site_rdf.py

@@ -228,3 +228,19 @@ class SiteRdfViewsTest:
         url = url_for('api.site_rdf_catalog_format', format='unknown')
         response = client.get(url)
         assert404(response)
+
+    def test_catalog_rdf_filter_tag(self, client):
+        DatasetFactory.create_batch(4, tags=['my-tag'])
+        DatasetFactory.create_batch(3)
+        url = url_for('api.site_rdf_catalog_format', format='xml', tag='my-tag')
+
+        response = client.get(url, headers={'Accept': 'application/xml'})
+        assert200(response)
+
+        graph = Graph().parse(data=response.data, format='xml')
+
+        datasets = list(graph.subjects(RDF.type, DCAT.Dataset))
+        assert len(datasets) == 4
+
+        for dat in datasets:
+            assert graph.value(dat, DCAT.keyword) == Literal('my-tag')

{udata-8.0.2.dev29304.dist-info → udata-8.0.2.dev29339.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: udata
-Version: 8.0.2.dev29304
+Version: 8.0.2.dev29339
 Summary: Open data portal
 Home-page: https://github.com/opendatateam/udata
 Author: Opendata Team
@@ -138,6 +138,7 @@ It is collectively taken care of by members of the
 ## Current (in progress)
 
 - **breaking change** Harvest backend is now sync [#3030](https://github.com/opendatateam/udata/pull/3030)
+- Add DCAT-AP HVD properties in RDF output if the dataset is tagged hvd [#3050](https://github.com/opendatateam/udata/pull/3050)
 - Allow dataservices to be discussed and followed [#3049](https://github.com/opendatateam/udata/pull/3049)
 - Add purge-dataservices job [#3049](https://github.com/opendatateam/udata/pull/3049)
 - Harvest all the available polygons from a spatial coverage [#3039](https://github.com/opendatateam/udata/pull/3039)

{udata-8.0.2.dev29304.dist-info → udata-8.0.2.dev29339.dist-info}/RECORD

@@ -9,7 +9,7 @@ udata/errors.py,sha256=chc1880TPGteIL-Rfb9hdeAyZI4vdWZQU3JQ_5VF5t0,116
 udata/factories.py,sha256=6en2qdHHoR6rEvNEUeH_0-T-vR7x7GUBGgNEm9vuylE,492
 udata/i18n.py,sha256=E-JoY57Cv4vgPvChkWjpwZAbuYVI8e6BO6raeu_3_Pw,8352
 udata/mail.py,sha256=dAMcbEtk5e54alpQezvF5adDrRPgdaT36QEdHD_5v50,2145
-udata/rdf.py,sha256=QiHcK2iklbtPkl6UFdBxTj0oaiY41en9k-kmbIVJ8Xw,9893
+udata/rdf.py,sha256=0G-orRN5L3E338iXWSih6yZqlj10r-rqBqQN3f38A1U,10051
 udata/routing.py,sha256=Qhpf5p97fs1SoJXtDctc1FPk0MeOKLn_C0Z1dP4ZNJA,7234
 udata/sentry.py,sha256=KiZz0PpmYpZMvykH9UAbHpF4xBY0Q-8DeiEbXEHDUdw,2683
 udata/settings.py,sha256=hoyo2YffL6eEKn90tjtraTMmEMRKfFxiIaDyMv69HMo,17390
@@ -99,7 +99,7 @@ udata/core/dataset/forms.py,sha256=VJCsGtgzhQgLW-M-J4ObpQ8o6c_saC3TTc1Js33m3sQ,6
 udata/core/dataset/models.py,sha256=q6AknSHYNjeUyVKoFgxEA0r9lSvMa_a4tJ3fLOzL03M,36053
 udata/core/dataset/permissions.py,sha256=3F2J7le3_rEYNhh88o3hSRWHAAt01_yHJM6RPmvCrRo,1090
 udata/core/dataset/preview.py,sha256=puPKT3fBD7ezAcT6owh0JK1_rGNDFZOqgT223qGn3LY,2597
-udata/core/dataset/rdf.py,sha256=KOIZLqQ085CiqaPds7CA0u481zJhQHmO527CaxNXEGA,25281
+udata/core/dataset/rdf.py,sha256=KGYAyZSEuPh_rz2jRsRC0VlNFn_wSNyOOTLhxL1Rvvk,26116
 udata/core/dataset/search.py,sha256=Ca23ljX7TZiKZKiAZQIH5xwS2tjXQcTDzieAJd-E22c,5314
 udata/core/dataset/signals.py,sha256=TK6dfrOUitZZkGGOh6XmhYqYvIjzZpI70JTLV4k-JRM,161
 udata/core/dataset/tasks.py,sha256=UYdm_O9hA0yo7cCNQFvZE9pGzUshkc3Wr4Gpel2aupU,8434
@@ -175,7 +175,7 @@ udata/core/reuse/search.py,sha256=I3FX-9YP_cEqw5GFwFv2iqLZExeaQXsak13RMSCN7t8,28
 udata/core/reuse/signals.py,sha256=2vv668u8miJ9t6H-FwRgXcWwsYMQem3oLNXp36rKtno,155
 udata/core/reuse/tasks.py,sha256=FgkUN6s1uTDFQgiImiOEfIMYYW984bxxDtlGK0uZ0yY,1530
 udata/core/site/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-udata/core/site/api.py,sha256=8sXpE_zRAe4LFYL-R58k4VwB-qQubrE9QhNselcq1HQ,4260
+udata/core/site/api.py,sha256=UXsMV8klzP5wQJazk_HfQGpckvkYddbuF8UJ5u0OBc0,4383
 udata/core/site/factories.py,sha256=8-RwP6QRCxXqWF4UHIkm2NfWRP2p8Ktc5RSY-4jgnfk,378
 udata/core/site/forms.py,sha256=lPnxm0MDhK7_ORzPgJLcCAPMj59hDrJ80gwRSW-jtIo,469
 udata/core/site/models.py,sha256=96IZHtPaMio7VMNCdP1Km_SL1GlMErwBuDqgUXfzKVY,5981
@@ -617,7 +617,7 @@ udata/tests/dataset/test_dataset_actions.py,sha256=4_0F_TCSCfHpzqDt659c8hG4GpkWj
 udata/tests/dataset/test_dataset_commands.py,sha256=TAgVdimIcBVmcSNKXKXbhmCGYRo8Keh_InBNiS6UlkM,802
 udata/tests/dataset/test_dataset_events.py,sha256=zt6F4j_vOBfMxDW6vBVQ5PGwr0d4s6LGw0cGkt9BlvI,2835
 udata/tests/dataset/test_dataset_model.py,sha256=_rvlp3j8ItvF4Dsacx0ymtTbbulrTXBafUcTVo2-N0o,29959
-udata/tests/dataset/test_dataset_rdf.py,sha256=2Wt4TQsJcH4w1CAFSW8H2O40Fh8hO--r3H9eSC89z7o,29388
+udata/tests/dataset/test_dataset_rdf.py,sha256=KgtljY7S55oJiBLYrHG6eMxPdNpDZik7M0z28Ec9Z0U,30105
 udata/tests/dataset/test_dataset_tasks.py,sha256=BN5qcvq6753WcpPOZd7vL1iIEW36ZBZrb2Lszb4eWEM,2283
 udata/tests/dataset/test_resource_preview.py,sha256=q1C9W9dO3nLWaR3-hpUOPYW17hCmQ3raty9Bl4FX3hg,3947
 udata/tests/features/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -661,7 +661,7 @@ udata/tests/site/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,
 udata/tests/site/test_site_api.py,sha256=XIjRH-iiwKdwmX145bQflxOatfIChi5RLikIMwJSbjQ,2276
 udata/tests/site/test_site_metrics.py,sha256=Sn9dQORwe-fvoyA7ZAnF63Cq5CwUGq1myH8Xe0EtfDA,2263
 udata/tests/site/test_site_model.py,sha256=nAx9JjEKdfjdw1Kj5Rs7P5FEePoATtCuOYPiSXEnVD0,1313
-udata/tests/site/test_site_rdf.py,sha256=M360uq2A6hl1WhmyRmRjvzuSqz_YFIUQNPR4vqWnYnE,9687
+udata/tests/site/test_site_rdf.py,sha256=NR09qJo8vmyAmtn0qQb-_nat7OyNV2fvwZTqW12BvQA,10283
 udata/tests/user/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 udata/tests/user/test_user_rdf.py,sha256=HrKirMURUXS9N3If_NMb8qnfJ4kE9IZymR1SPcNvlF0,1851
 udata/tests/workers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -684,9 +684,9 @@ udata/translations/pt/LC_MESSAGES/udata.mo,sha256=uttB2K8VsqzkEQG-5HfTtFms_3LtV9
 udata/translations/pt/LC_MESSAGES/udata.po,sha256=8Ql1Lp7Z9KLnvp-qRxw-NhFu1p35Xj-q6Jg9JHsYhcw,43733
 udata/translations/sr/LC_MESSAGES/udata.mo,sha256=US8beNIMPxP5h-zD_jfP1TheDDd4DdRVS5UIiY5XVZ8,28553
 udata/translations/sr/LC_MESSAGES/udata.po,sha256=TM0yMDvKRljyOzgZZMlTX6OfpF6OC4Ngf_9Zc8n6ayA,50313
-udata-8.0.2.dev29304.dist-info/LICENSE,sha256=V8j_M8nAz8PvAOZQocyRDX7keai8UJ9skgmnwqETmdY,34520
-udata-8.0.2.dev29304.dist-info/METADATA,sha256=rufDqdUYqfG6w_v_xL9iHJtWli-g988Tg3va4gAqY5o,123860
-udata-8.0.2.dev29304.dist-info/WHEEL,sha256=DZajD4pwLWue70CAfc7YaxT1wLUciNBvN_TTcvXpltE,110
-udata-8.0.2.dev29304.dist-info/entry_points.txt,sha256=3SKiqVy4HUqxf6iWspgMqH8d88Htk6KoLbG1BU-UddQ,451
-udata-8.0.2.dev29304.dist-info/top_level.txt,sha256=39OCg-VWFWOq4gCKnjKNu-s3OwFlZIu_dVH8Gl6ndHw,12
-udata-8.0.2.dev29304.dist-info/RECORD,,
+udata-8.0.2.dev29339.dist-info/LICENSE,sha256=V8j_M8nAz8PvAOZQocyRDX7keai8UJ9skgmnwqETmdY,34520
+udata-8.0.2.dev29339.dist-info/METADATA,sha256=Fhk0bn2S4psXBb_HgCnqhrcG6kTl9lXBYBoRNcu3kLA,123989
+udata-8.0.2.dev29339.dist-info/WHEEL,sha256=DZajD4pwLWue70CAfc7YaxT1wLUciNBvN_TTcvXpltE,110
+udata-8.0.2.dev29339.dist-info/entry_points.txt,sha256=3SKiqVy4HUqxf6iWspgMqH8d88Htk6KoLbG1BU-UddQ,451
+udata-8.0.2.dev29339.dist-info/top_level.txt,sha256=39OCg-VWFWOq4gCKnjKNu-s3OwFlZIu_dVH8Gl6ndHw,12
+udata-8.0.2.dev29339.dist-info/RECORD,,