Fixes in example and profile to get validation passing

ckan · Feb 7, 2025 · 36afa7d · 36afa7d
1 parent a1390fa
commit 36afa7d
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 13 deletions.
diff --git a/ckanext/dcat/profiles/croissant.py b/ckanext/dcat/profiles/croissant.py
@@ -351,10 +351,24 @@ def _resource_basic_fields_graph(self, resource_ref, resource_dict):
         items = [
             ("name", SCHEMA.name, None, Literal),
             ("description", SCHEMA.description, None, Literal),
-            ("hash", SCHEMA.sha256, None, Literal),
         ]
         self._add_triples_from_dict(resource_dict, resource_ref, items)
 
+        if resource_dict.get("hash"):
+            predicate = None
+            if len(resource_dict["hash"]) == 32:
+                predicate = SCHEMA.md5
+            elif len(resource_dict["hash"]) == 64:
+                predicate = SCHEMA.sha256
+            if predicate:
+                self._add_triple_from_dict(
+                    resource_dict,
+                    resource_ref,
+                    predicate,
+                    "hash",
+                    _type=Literal
+                )
+
     def _resource_list_fields_graph(self, resource_ref, resource_dict):
         items = [
             ("same_as", SCHEMA.sameAs, None, Literal),
@@ -402,13 +416,16 @@ def _resource_subresources_graph(self, dataset_ref, resource_ref, resource_dict)
                 self.g.add((dataset_ref, SCHEMA.distribution, subresource_ref)) # Note that this is added to the dataset_ref node, not to the resource_ref node
                 self.g.add((subresource_ref, RDF.type, subresource_type_specific))
 
-                items = [
-                    ("description", SCHEMA.description, None, Literal),
-                    ("format", SCHEMA.encodingFormat, None, Literal),
-                ]
-                self._add_triples_from_dict(subresource_dict, subresource_ref, items)
+                # Basic fields
+                self._resource_basic_fields_graph(subresource_ref, subresource_dict)
+
+                # Format
+                self._resource_format_graph(subresource_ref, subresource_dict)
+
+                # URL
+                self._resource_url_graph(subresource_ref, subresource_dict)
 
-                if resource_dict.get("type") == "fileSet":
+                if subresource_dict.get("type") == "fileSet":
                     items = [
                         ("includes", CR.includes, None, Literal),
                         ("excludes", CR.excludes, None, Literal),

diff --git a/ckanext/dcat/schemas/croissant.yaml b/ckanext/dcat/schemas/croissant.yaml
@@ -390,6 +390,12 @@ resource_fields:
       required: false
       preset: resource_format_autocomplete
 
+    # schema:sha256 | schema:Text | ONE
+    - field_name: hash
+      label: Hash
+      help_text: Checksum for the file contents.
+      required: false
+
     # cr:includes | schema:Text | MANY
     - field_name: includes
       label: Includes
@@ -411,4 +417,4 @@ resource_fields:
     # - field_name: contained_in
     #   label: Contained in
     #   help_text: Another FileObject or FileSet that this one is contained in, e.g., in the case of a file extracted from an archive. When this property is present, the contentUrl is evaluated as a relative path within the container object.
-    #   required: false
+    #   required: false
diff --git a/ckanext/dcat/tests/profiles/croissant/test_validate.py b/ckanext/dcat/tests/profiles/croissant/test_validate.py
@@ -10,7 +10,10 @@
 from ckanext.dcat.profiles.croissant import JSONLD_CONTEXT
 from ckanext.dcat.tests.utils import get_file_contents
 
-@pytest.mark.skipif(sys.version_info < (3, 10), reason="croissant is not available in py<3.10")
+
+@pytest.mark.skipif(
+    sys.version_info < (3, 10), reason="croissant is not available in py<3.10"
+)
 def test_valid_output():
 
     dataset_dict = json.loads(

diff --git a/examples/ckan/ckan_full_dataset_croissant.json b/examples/ckan/ckan_full_dataset_croissant.json
@@ -57,19 +57,27 @@
             "description": "Some description",
             "url": "https://example.com/data.csv",
             "format": "CSV",
-            "type": "fileSet",
             "id_given": "my-custom-resource-id",
-            "size": 12323,
-            "hash": "4304cf2e751e6053c90b1804c89c0ebb758f395a",
+            "size": "12323",
+            "hash": "b221d9dbb083a7f33428d7c2a3c3198ae925614d70210e28716ccaa7cd4ddb79",
             "subresources": [
                 {
                     "type": "fileObject",
-                    "id_given": "my-custom-subresource-id",
+                    "id_given": "my-custom-subresource-id-1",
+                    "url": "https://example.com/data.csv",
                     "description": "Test subresource 1",
+                    "hash": "b221d9dbb083a7f33428d7c2a3c3198ae925614d70210e28716ccaa7cd4ddb79",
+                    "format": "CSV"
+                },
+                {
+                    "type": "fileSet",
+                    "id_given": "my-custom-subresource-id-2",
+                    "description": "Test subresource 2",
                     "format": "CSV",
                     "includes": "**.csv",
                     "excludes": "**.txt"
                 }
+
             ]
         }
     ]