Merge pull request #301 from Health-RI/add-dct-identifier-to-publisher

Add dct identifier to publisher

ckanext/dcat/converters.py

@@ -28,7 +28,15 @@ def dcat_to_ckan(dcat_dict):
         package_dict['extras'].append({'key': 'dcat_publisher_name', 'value': dcat_publisher})
     elif isinstance(dcat_publisher, dict) and dcat_publisher.get('name'):
         package_dict['extras'].append({'key': 'dcat_publisher_name', 'value': dcat_publisher.get('name')})
-        package_dict['extras'].append({'key': 'dcat_publisher_email', 'value': dcat_publisher.get('mbox')})
+
+        if dcat_publisher.get('mbox'):
+            package_dict['extras'].append({'key': 'dcat_publisher_email', 'value': dcat_publisher.get('mbox')})
+
+        if dcat_publisher.get('identifier'):
+            package_dict['extras'].append({
+                'key': 'dcat_publisher_id',
+                'value': dcat_publisher.get('identifier')  # This could be a URI like https://ror.org/05wg1m734
+            })
 
     package_dict['extras'].append({
         'key': 'language',

ckanext/dcat/processors.py

@@ -407,7 +407,8 @@ def _get_from_extra(key):
                                  ('name', Literal, FOAF.name, True,),
                                  ('email', Literal, FOAF.mbox, False,),
                                  ('url', URIRef, FOAF.homepage,False,),
-                                 ('type', Literal, DCT.type, False,))
+                                 ('type', Literal, DCT.type, False,),
+                                 ('identifier', URIRef, DCT.identifier, False,))
 
             _pub = _get_from_extra('source_catalog_publisher')
             if _pub:

ckanext/dcat/profiles/base.py

@@ -468,6 +468,8 @@ def _publisher(self, subject, predicate):
 
             publisher["type"] = self._object_value(agent, DCT.type)
 
+            publisher['identifier'] = self._object_value(agent, DCT.identifier)
+
         return publisher
 
     def _contact_details(self, subject, predicate):

ckanext/dcat/profiles/euro_dcat_ap_base.py

@@ -123,7 +123,7 @@ def _parse_dataset_base(self, dataset_dict, dataset_ref):
 
         # Publisher
         publisher = self._publisher(dataset_ref, DCT.publisher)
-        for key in ("uri", "name", "email", "url", "type"):
+        for key in ("uri", "name", "email", "url", "type", "identifier"):
             if publisher.get(key):
                 dataset_dict["extras"].append(
                     {"key": "publisher_{0}".format(key), "value": publisher.get(key)}
@@ -372,6 +372,7 @@ def _graph_from_dataset_base(self, dataset_dict, dataset_ref):
                 "email": self._get_dataset_value(dataset_dict, "publisher_email"),
                 "url": self._get_dataset_value(dataset_dict, "publisher_url"),
                 "type": self._get_dataset_value(dataset_dict, "publisher_type"),
+                "identifier": self._get_dataset_value(dataset_dict, "publisher_identifier"),
             }
         elif dataset_dict.get("organization"):
             # Fall back to dataset org
@@ -396,6 +397,7 @@ def _graph_from_dataset_base(self, dataset_dict, dataset_ref):
                     "email": org_dict.get("email"),
                     "url": org_dict.get("url"),
                     "type": org_dict.get("dcat_type"),
+                    "identifier": org_dict.get("identifier"),
                 }
         # Add to graph
         if publisher_ref:
@@ -406,6 +408,7 @@ def _graph_from_dataset_base(self, dataset_dict, dataset_ref):
                 ("email", FOAF.mbox, None, Literal),
                 ("url", FOAF.homepage, None, URIRef),
                 ("type", DCT.type, None, URIRefOrLiteral),
+                ("identifier", DCT.identifier, None, URIRefOrLiteral),
             ]
             self._add_triples_from_dict(publisher_details, publisher_ref, items)
 

ckanext/dcat/profiles/euro_dcat_ap_scheming.py

@@ -187,6 +187,13 @@ def _not_empty_dict(data_dict):
                 _type=URIRef,
                 value_modifier=self._add_mailto,
             )
+            self._add_triple_from_dict(
+                publisher,
+                publisher_ref,
+                DCT.identifier,
+                "identifier",
+                _type=URIRefOrLiteral
+            )
 
         temporal = dataset_dict.get("temporal_coverage")
         if (

ckanext/dcat/profiles/schemaorg.py

@@ -221,6 +221,10 @@ def _publisher_graph(self, dataset_ref, dataset_dict):
 
             self._add_triples_from_dict(dataset_dict, contact_point, items)
 
+            publisher_identifier = self._get_dataset_value(dataset_dict, "publisher_identifier")
+            if publisher_identifier:
+                self.g.add((publisher_details, SCHEMA.identifier, Literal(publisher_identifier)))
+
     def _temporal_graph(self, dataset_ref, dataset_dict):
         start = self._get_dataset_value(dataset_dict, "temporal_start")
         end = self._get_dataset_value(dataset_dict, "temporal_end")

ckanext/dcat/schemas/dcat_ap_full.yaml

@@ -66,6 +66,10 @@ dataset_fields:
 
     - field_name: type
       label: Type
+
+    - field_name: identifier
+      label: Identifier
+      help_text: Unique identifier for the publisher, such as a ROR ID.
   help_text: Entity responsible for making the dataset available.
 
 - field_name: license_id

ckanext/dcat/schemas/dcat_ap_recommended.yaml

@@ -66,6 +66,11 @@ dataset_fields:
 
     - field_name: type
       label: Type
+
+    - field_name: identifier
+      label: Identifier
+      help_text: Unique identifier for the publisher, such as a ROR ID.
+
   help_text: Entity responsible for making the dataset available.
 
 - field_name: license_id

ckanext/dcat/tests/profiles/base/test_base_profile.py

@@ -647,6 +647,7 @@ def test_publisher_foaf(self):
               <foaf:mbox>contact@some.org</foaf:mbox>
               <foaf:homepage>http://some.org</foaf:homepage>
               <dct:type rdf:resource="http://purl.org/adms/publishertype/NonProfitOrganisation"/>
+              <dct:identifier rdf:resource="https://ror.org/05wg1m734"/>
             </foaf:Organization>
           </dct:publisher>
         </rdfs:SomeClass>
@@ -666,6 +667,7 @@ def test_publisher_foaf(self):
         assert publisher['email'] == 'contact@some.org'
         assert publisher['url'] == 'http://some.org'
         assert publisher['type'] == 'http://purl.org/adms/publishertype/NonProfitOrganisation'
+        assert publisher['identifier'] == 'https://ror.org/05wg1m734'
 
     def test_publisher_ref(self):
 

ckanext/dcat/tests/profiles/dcat_ap/test_euro_dcatap_profile_parse.py

@@ -113,6 +113,7 @@ def _get_extra_value_as_list(key):
         assert _get_extra_value('publisher_email') == 'contact@some.org'
         assert _get_extra_value('publisher_url') == 'http://some.org'
         assert _get_extra_value('publisher_type') == 'http://purl.org/adms/publishertype/NonProfitOrganisation'
+        assert _get_extra_value('publisher_identifier') == 'https://ror.org/05wg1m734'
         assert _get_extra_value('contact_name') == 'Point of Contact'
         # mailto gets removed for storage and is added again on output
         assert _get_extra_value('contact_email') == 'contact@some.org'

ckanext/dcat/tests/profiles/dcat_ap_2/test_scheming_support.py

@@ -94,6 +94,7 @@ def test_e2e_ckan_to_dcat(self):
                     "email": "publisher@example.org",
                     "url": "https://example.org",
                     "type": "public_body",
+                    "identifier": "http://example.org/publisher-id",
                 },
             ],
             "temporal_coverage": [
@@ -301,6 +302,12 @@ def test_e2e_ckan_to_dcat(self):
             DCT.type,
             dataset_dict["publisher"][0]["type"],
         )
+        assert self._triple(
+            g,
+            publisher[0][2],
+            DCT.identifier,
+            URIRef(dataset_dict["publisher"][0]["identifier"])
+        )
 
         temporal = [t for t in g.triples((dataset_ref, DCT.temporal, None))]
 

ckanext/dcat/tests/profiles/schemaorg/test_schemaorg_profile_serialize.py

@@ -105,6 +105,7 @@ def test_publisher_extras(self):
                 {'key': 'publisher_email', 'value': 'publisher@example.com'},
                 {'key': 'publisher_url', 'value': 'http://example.com/publisher/home'},
                 {'key': 'publisher_type', 'value': 'http://purl.org/adms/publishertype/Company'},
+                {'key': 'publisher_identifier', 'value': 'https://ror.org/05wg1m734'},
             ]
 
 
@@ -121,6 +122,7 @@ def test_publisher_extras(self):
         assert str(publisher) == extras['publisher_uri']
         assert self._triple(g, publisher, RDF.type, SCHEMA.Organization)
         assert self._triple(g, publisher, SCHEMA.name, extras['publisher_name'])
+        assert self._triple(g, publisher, SCHEMA.identifier, extras['publisher_identifier'])
 
         contact_point = self._triple(g, publisher, SCHEMA.contactPoint, None)[2]
         assert contact_point
@@ -144,6 +146,7 @@ def test_publisher_no_uri(self):
                 {'key': 'publisher_email', 'value': 'publisher@example.com'},
                 {'key': 'publisher_url', 'value': 'http://example.com/publisher/home'},
                 {'key': 'publisher_type', 'value': 'http://purl.org/adms/publishertype/Company'},
+                {'key': 'publisher_identifier', 'value': 'https://ror.org/05wg1m734'},
             ]
         }
         extras = self._extras(dataset)
@@ -158,6 +161,7 @@ def test_publisher_no_uri(self):
         assert isinstance(publisher, BNode)
         assert self._triple(g, publisher, RDF.type, SCHEMA.Organization)
         assert self._triple(g, publisher, SCHEMA.name, extras['publisher_name'])
+        assert self._triple(g, publisher, SCHEMA.identifier, extras['publisher_identifier'])
 
         contact_point = self._triple(g, publisher, SCHEMA.contactPoint, None)[2]
         assert contact_point

ckanext/dcat/utils.py

@@ -78,6 +78,7 @@ def field_labels():
         'publisher_email': _('Publisher email'),
         'publisher_url': _('Publisher URL'),
         'publisher_type': _('Publisher type'),
+        'publisher_identifier': _('Publisher identifier'),
         'contact_name': _('Contact name'),
         'contact_email': _('Contact email'),
         'contact_uri': _('Contact URI'),