def build_catalog(site, datasets, format=None):
'''Build the DCAT catalog for this site'''
site_url = url_for('site.home_redirect', _external=True)
catalog_url = url_for('site.rdf_catalog', _external=True)
graph = Graph(namespace_manager=namespace_manager)
catalog = graph.resource(URIRef(catalog_url))
catalog.set(RDF.type, DCAT.Catalog)
catalog.set(DCT.title, Literal(site.title))
catalog.set(DCT.language, Literal(current_app.config['DEFAULT_LANGUAGE']))
catalog.set(FOAF.homepage, URIRef(site_url))
publisher = graph.resource(BNode())
publisher.set(RDF.type, FOAF.Organization)
publisher.set(FOAF.name, Literal(current_app.config['SITE_AUTHOR']))
catalog.set(DCT.publisher, publisher)
for dataset in datasets:
rdf_dataset = dataset_to_rdf(dataset, graph)
if dataset.owner:
rdf_dataset.add(DCT.publisher, user_to_rdf(dataset.owner, graph))
elif dataset.organization:
rdf_dataset.add(DCT.publisher,
organization_to_rdf(dataset.organization, graph))
catalog.add(DCAT.dataset, rdf_dataset)
if isinstance(datasets, Paginable):
paginate_catalog(catalog, graph, datasets, format,
'site.rdf_catalog_format')
return catalog
def dataset_to_rdf(dataset, graph=None):
'''
Map a dataset domain model to a DCAT/RDF graph
'''
# Use the unlocalized permalink to the dataset as URI when available
# unless there is already an upstream URI
if 'uri' in dataset.extras:
id = URIRef(dataset.extras['uri'])
elif dataset.id:
id = URIRef(
url_for('datasets.show_redirect',
dataset=dataset.id,
_external=True))
else:
id = BNode()
# Expose upstream identifier if present
if 'dct:identifier' in dataset.extras:
identifier = dataset.extras['dct:identifier']
else:
identifier = dataset.id
graph = graph or Graph(namespace_manager=namespace_manager)
d = graph.resource(id)
d.set(RDF.type, DCAT.Dataset)
d.set(DCT.identifier, Literal(identifier))
d.set(DCT.title, Literal(dataset.title))
d.set(DCT.description, Literal(dataset.description))
d.set(DCT.issued, Literal(dataset.created_at))
d.set(DCT.modified, Literal(dataset.last_modified))
if dataset.acronym:
d.set(SKOS.altLabel, Literal(dataset.acronym))
for tag in dataset.tags:
d.add(DCAT.keyword, Literal(tag))
for resource in dataset.resources:
d.add(DCAT.distribution, resource_to_rdf(resource, dataset, graph))
if dataset.owner:
d.add(DCT.publisher, user_to_rdf(dataset.owner, graph))
elif dataset.organization:
d.add(DCT.publisher, organization_to_rdf(dataset.organization, graph))
if dataset.temporal_coverage:
d.set(DCT.temporal, temporal_to_rdf(dataset.temporal_coverage, graph))
frequency = frequency_to_rdf(dataset.frequency)
if frequency:
d.set(DCT.accrualPeriodicity, frequency)
return d
def test_minimal(self):
org = OrganizationFactory.build() # Does not have an URL
o = organization_to_rdf(org)
g = o.graph
self.assertIsInstance(o, RdfResource)
self.assertEqual(len(list(g.subjects(RDF.type, FOAF.Organization))), 1)
self.assertEqual(o.value(RDF.type).identifier, FOAF.Organization)
self.assertIsInstance(o.identifier, BNode)
self.assertEqual(o.value(FOAF.name), Literal(org.name))
self.assertEqual(o.value(RDFS.label), Literal(org.name))
def test_minimal(self):
org = OrganizationFactory.build() # Does not have an URL
o = organization_to_rdf(org)
g = o.graph
self.assertIsInstance(o, RdfResource)
self.assertEqual(len(list(g.subjects(RDF.type, FOAF.Organization))), 1)
self.assertEqual(o.value(RDF.type).identifier, FOAF.Organization)
self.assertIsInstance(o.identifier, BNode)
self.assertEqual(o.value(FOAF.name), Literal(org.name))
self.assertEqual(o.value(RDFS.label), Literal(org.name))
def dataset_to_rdf(dataset, graph=None):
'''
Map a dataset domain model to a DCAT/RDF graph
'''
# Use the unlocalized permalink to the dataset as URI when available
# unless there is already an upstream URI
if 'uri' in dataset.extras:
id = URIRef(dataset.extras['uri'])
elif dataset.id:
id = URIRef(url_for('datasets.show_redirect',
dataset=dataset.id,
_external=True))
else:
id = BNode()
# Expose upstream identifier if present
if 'dct:identifier' in dataset.extras:
identifier = dataset.extras['dct:identifier']
else:
identifier = dataset.id
graph = graph or Graph(namespace_manager=namespace_manager)
d = graph.resource(id)
d.set(RDF.type, DCAT.Dataset)
d.set(DCT.identifier, Literal(identifier))
d.set(DCT.title, Literal(dataset.title))
d.set(DCT.description, Literal(dataset.description))
d.set(DCT.issued, Literal(dataset.created_at))
d.set(DCT.modified, Literal(dataset.last_modified))
if dataset.acronym:
d.set(SKOS.altLabel, Literal(dataset.acronym))
for tag in dataset.tags:
d.add(DCAT.keyword, Literal(tag))
for resource in dataset.resources:
d.add(DCAT.distribution, resource_to_rdf(resource, dataset, graph))
if dataset.owner:
d.add(DCT.publisher, user_to_rdf(dataset.owner, graph))
elif dataset.organization:
d.add(DCT.publisher, organization_to_rdf(dataset.organization, graph))
if dataset.temporal_coverage:
d.set(DCT.temporal, temporal_to_rdf(dataset.temporal_coverage, graph))
frequency = frequency_to_rdf(dataset.frequency)
if frequency:
d.set(DCT.accrualPeriodicity, frequency)
return d
def test_all_fields(self):
org = OrganizationFactory(url=faker.uri())
org_url = url_for('organizations.show_redirect',
org=org.id,
_external=True)
o = organization_to_rdf(org)
g = o.graph
self.assertIsInstance(o, RdfResource)
self.assertEqual(len(list(g.subjects(RDF.type, FOAF.Organization))), 1)
self.assertEqual(o.value(RDF.type).identifier, FOAF.Organization)
self.assertIsInstance(o.identifier, URIRef)
self.assertEqual(o.identifier.toPython(), org_url)
self.assertEqual(o.value(FOAF.name), Literal(org.name))
self.assertEqual(o.value(RDFS.label), Literal(org.name))
self.assertEqual(o.value(FOAF.homepage).identifier, URIRef(org.url))
def test_all_fields(self):
org = OrganizationFactory(url=faker.uri())
org_url = url_for('organizations.show_redirect',
org=org.id,
_external=True)
o = organization_to_rdf(org)
g = o.graph
self.assertIsInstance(o, RdfResource)
self.assertEqual(len(list(g.subjects(RDF.type, FOAF.Organization))), 1)
self.assertEqual(o.value(RDF.type).identifier, FOAF.Organization)
self.assertIsInstance(o.identifier, URIRef)
self.assertEqual(o.identifier.toPython(), org_url)
self.assertEqual(o.value(FOAF.name), Literal(org.name))
self.assertEqual(o.value(RDFS.label), Literal(org.name))
self.assertEqual(o.value(FOAF.homepage).identifier, URIRef(org.url))
