fix: Fixes sameAs property to follow schema.org spec (#944)

* fix: Fixes sameAs property to follow schema.org spec

* Adds SHACL validation

* Adds explicit sameAs migration

Co-authored-by: Rok Roškar <rokroskar@gmail.com>

renku/cli/log.py

@@ -91,7 +91,11 @@
 
 
 @click.command()
-@click.option('--revision', default='HEAD')
+@click.option(
+    '--revision',
+    default='HEAD',
+    help='The git revision to generate the log for, default: HEAD'
+)
 @click.option(
     '--format',
     type=click.Choice(FORMATS),

renku/core/commands/dataset.py

@@ -38,7 +38,8 @@
     MigrationRequired, OperationError, ParameterError, UsageError
 from renku.core.management.datasets import DATASET_METADATA_PATHS
 from renku.core.management.git import COMMIT_DIFF_STRATEGY
-from renku.core.models.datasets import Dataset, generate_default_short_name
+from renku.core.models.datasets import Dataset, Url, \
+    generate_default_short_name
 from renku.core.models.provenance.agents import Person
 from renku.core.models.refs import LinkReference
 from renku.core.models.tabulate import tabulate
@@ -239,8 +240,9 @@ def add_to_dataset(
                 with_metadata.files = dataset.files
 
                 if is_doi(with_metadata.identifier):
-                    dataset.same_as = urllib.parse.urljoin(
-                        'https://doi.org', with_metadata.identifier
+                    dataset.same_as = Url(
+                        url=urllib.parse.
+                        urljoin('https://doi.org', with_metadata.identifier)
                     )
 
                 dataset.update_metadata(with_metadata)

renku/core/models/datasets.py

@@ -46,6 +46,30 @@
 )
 
 
+@jsonld.s(
+    type='schema:URL',
+    context={
+        'schema': 'http://schema.org/',
+    },
+)
+class Url:
+    """Represents a schema URL reference."""
+
+    url = jsonld.ib(default=None, kw_only=True, context='schema:url')
+
+    _id = jsonld.ib(kw_only=True, context='@id')
+
+    @_id.default
+    def default_id(self):
+        """Define default value for id field."""
+        if self.url:
+            parsed_result = urllib.parse.urlparse(self.url)
+            id_ = urllib.parse.ParseResult('', *parsed_result[1:]).geturl()
+        else:
+            id_ = str(uuid.uuid4())
+        return '_:URL@{0}'.format(id_)
+
+
 def _convert_creators(value):
     """Convert creators."""
     if isinstance(value, dict):  # compatibility with previous versions
@@ -349,7 +373,9 @@ class Dataset(Entity, CreatorMixin):
         kw_only=True
     )
 
-    same_as = jsonld.ib(context='schema:sameAs', default=None, kw_only=True)
+    same_as = jsonld.ib(
+        context='schema:sameAs', default=None, kw_only=True, type=Url
+    )
 
     short_name = jsonld.ib(
         default=None, context='schema:alternateName', kw_only=True

renku/core/models/migrations/__init__.py

@@ -18,9 +18,13 @@
 """Renku JSON-LD migrations."""
 
 from renku.core.models.migrations.dataset import migrate_absolute_paths, \
-    migrate_dataset_schema, migrate_doi_identifier
+    migrate_dataset_schema, migrate_doi_identifier, \
+    migrate_same_as_structure
 
 JSONLD_MIGRATIONS = {
     'dctypes:Dataset': [migrate_dataset_schema, migrate_absolute_paths],
-    'schema:Dataset': [migrate_absolute_paths, migrate_doi_identifier],
+    'schema:Dataset': [
+        migrate_absolute_paths, migrate_doi_identifier,
+        migrate_same_as_structure
+    ],
 }

renku/core/models/migrations/dataset.py

@@ -80,8 +80,41 @@ def migrate_doi_identifier(data):
         if not is_uuid(identifier):
             data['identifier'] = str(uuid.uuid4())
         if is_doi(data.get('_id', '')):
-            data['same_as'] = data['_id']
+            data['same_as'] = {'@type': ['schema:URL'], 'url': data['_id']}
             if data.get('@context'):
-                data['@context'].setdefault('same_as', 'schema:sameAs')
+                data['@context'].setdefault(
+                    'same_as', {
+                        '@id': 'schema:sameAs',
+                        '@type': 'schema:URL',
+                        '@context': {
+                            '@version': '1.1',
+                            'url': 'schema:url',
+                            'schema': 'http://schema.org/'
+                        }
+                    }
+                )
         data['_id'] = data['identifier']
     return data
+
+
+def migrate_same_as_structure(data):
+    """Changes sameAs string to schema:URL object."""
+    same_as = data.get('same_as')
+
+    if same_as and isinstance(same_as, str):
+        data['same_as'] = {'@type': ['schema:URL'], 'url': same_as}
+
+        if data.get('@context'):
+            data['@context'].setdefault(
+                'same_as', {
+                    '@id': 'schema:sameAs',
+                    '@type': 'schema:URL',
+                    '@context': {
+                        '@version': '1.1',
+                        'url': 'schema:url',
+                        'schema': 'http://schema.org/'
+                    }
+                }
+            )
+
+    return data

renku/data/shacl_shape.json

@@ -323,8 +323,8 @@
             {
                "nodeKind": "sh:Literal",
                "path": "schema:sameAs",
-               "datatype": {
-                  "@id": "xsd:string"
+               "sh:class": {
+                  "@id": "schema:URL"
                }
             },
             {
@@ -390,6 +390,34 @@
             }
          ]
       },
+      {
+         "@id": "_:URLShape",
+         "@type": "sh:NodeShape",
+         "ignoredProperties": [
+            {
+               "@id": "rdf:type"
+            }
+         ],
+         "closed": true,
+         "targetClass": "schema:URL",
+         "property": [
+            {
+               "path": "schema:url",
+               "or": [
+                  {
+                     "nodeKind": "sh:Literal",
+                     "datatype": {
+                        "@id": "xsd:string"
+                     }
+                  },
+                  {
+                     "nodeKind": "sh:IRI"
+                  }
+               ],
+               "maxCount": 1
+            }
+         ]
+      },
       {
          "@id": "_:inLanguageShape",
          "@type": "sh:NodeShape",

tests/core/commands/test_serialization.py

@@ -98,7 +98,8 @@ def test_uuid_migration(dataset_metadata, client):
     assert urljoin(
         'https://localhost/datasets/', dataset.identifier
     ) == dataset._id
-    assert dataset.same_as.startswith('https://doi.org')
+
+    assert dataset.same_as.url.startswith('https://doi.org')
 
 
 def test_dataset_creator_email(dataset_metadata):