fix: disallows ark identifier modifications

sonar/config.py

@@ -379,10 +379,14 @@ def _(x):
             "application/json": (
                 "sonar.modules.documents.serializers" ":json_v1_response"
             ),
+            "application/rero+json": (
+                "sonar.modules.documents.serializers" ":json_doc_response"
+            ),
             "text/xml": ("sonar.modules.documents.serializers" ":dc_v1_response"),
         },
         record_serializers_aliases={
             "json": "application/json",
+            "rero": "application/rero+json",
             "dc": "text/xml",
         },
         search_serializers={

sonar/modules/documents/marshmallow/__init__.py

@@ -15,6 +15,11 @@
 
 """Schemas for marshmallow."""
 
-from .json import DocumentListSchemaV1, DocumentMetadataSchemaV1, DocumentSchemaV1
+from .json import (
+    DocumentListSchemaV1,
+    DocumentMetadataSchemaV1,
+    DocumentReroSchemaV1,
+    DocumentSchemaV1,
+)
 
-__all__ = ("DocumentListSchemaV1", "DocumentMetadataSchemaV1", "DocumentSchemaV1")
+__all__ = ("DocumentListSchemaV1", "DocumentMetadataSchemaV1", "DocumentReroSchemaV1", "DocumentSchemaV1")

sonar/modules/documents/marshmallow/json.py

@@ -300,6 +300,22 @@ class DocumentMetadataSchemaV1(DocumentListMetadataSchemaV1):
 
     _files = Nested(FileSchemaV1, many=True)
 
+    @pre_load
+    def restore_ark_identifier(self, data, **kwargs):
+        """Remove ark identifier if any.
+
+        :param item: Dict representing the record.
+        :returns: Modified dict.
+        """
+        item = data.get("identifiedBy", [])
+        data["identifiedBy"] = [identifier for identifier in item if identifier.get("type") != "ark"]
+        from sonar.modules.documents.api import DocumentRecord
+
+        if doc := DocumentRecord.get_record_by_pid(data.get("pid")):
+            item = doc.get("identifiedBy", [])
+            data["identifiedBy"].extend([identifier for identifier in item if identifier.get("type") == "ark"])
+        return data
+
 
 class DocumentSchemaV1(StrictKeysMixin):
     """Document schema."""
@@ -310,6 +326,21 @@ class DocumentSchemaV1(StrictKeysMixin):
     explanation = fields.Raw(dump_only=True)
 
 
+class DocumentReroSchemaV1(DocumentSchemaV1):
+    """Document schema."""
+
+    @pre_dump
+    def remove_ark_identifier(self, data, **kwargs):
+        """Remove ark identifier if any.
+
+        :param item: Dict representing the record.
+        :returns: Modified dict.
+        """
+        item = data["metadata"].get("identifiedBy", [])
+        data["metadata"]["identifiedBy"] = [identifier for identifier in item if identifier.get("type") != "ark"]
+        return data
+
+
 class DocumentListSchemaV1(StrictKeysMixin):
     """Document schema."""
 

sonar/modules/documents/serializers/__init__.py

@@ -22,7 +22,7 @@
 
 from sonar.modules.documents.serializers.schemas.dc import DublinCoreSchema
 
-from ..marshmallow import DocumentListSchemaV1, DocumentSchemaV1
+from ..marshmallow import DocumentListSchemaV1, DocumentReroSchemaV1, DocumentSchemaV1
 from .dc import DublinCoreSerializer
 from .google_scholar import SonarGoogleScholarSerializer
 from .json import JSONSerializer
@@ -34,6 +34,7 @@
 # ===========
 #: JSON serializer definition.
 json_v1 = JSONSerializer(DocumentSchemaV1)
+json_doc = JSONSerializer(DocumentReroSchemaV1)
 json_list_v1 = JSONSerializer(DocumentListSchemaV1)
 #: schema.org serializer
 schemaorg_v1 = SonarSchemaOrgSerializer(SchemaOrgV1, replace_refs=True)
@@ -46,6 +47,7 @@
 # ========================
 #: JSON record serializer for individual records.
 json_v1_response = record_responsify(json_v1, "application/json")
+json_doc_response = record_responsify(json_doc, "application/rero+json")
 #: JSON record serializer for search results.
 json_v1_search = search_responsify(json_list_v1, "application/json")
 

tests/api/documents/test_documents_rest.py

@@ -16,16 +16,20 @@
 """Test REST endpoint for documents."""
 
 import json
+from copy import deepcopy
 
 from flask import url_for
 from invenio_accounts.testutils import login_user_via_session
 
+from sonar.modules.documents.api import DocumentRecord
+
 
 def test_get(client, document_with_file):
     """Get REST methods."""
     res = client.get(url_for("invenio_records_rest.doc_list", view="global"))
     assert res.status_code == 200
     assert res.json["hits"]["total"]["value"] == 1
+
     # the search results does not contains permissions
     fdata = res.json["hits"]["hits"][0]["metadata"]["_files"][0]
     assert list(fdata.keys()) == [
@@ -47,28 +51,72 @@ def test_get(client, document_with_file):
         "read": False,
         "update": False,
     }
+    assert "ark" in [r["type"] for r in res.json["metadata"]["identifiedBy"]]
+
+    # ark identifier is removed
+    res = client.get(url_for("invenio_records_rest.doc_item", pid_value=document_with_file["pid"], format="rero"))
+    assert res.status_code == 200
+    assert "ark" not in [r["type"] for r in res.json["metadata"]["identifiedBy"]]
 
 
-def test_put(app, client, document_with_file):
+def test_post_put_delete(app, client, document_json, organisation):
     """Test putting metadata on existing file."""
     # Disable configuration
     app.config.update(SONAR_APP_DISABLE_PERMISSION_CHECKS=True)
     headers = [("Content-Type", "application/json")]
+    data = deepcopy(document_json)
+    data["organisation"] = [{"$ref": f"https://sonar.ch/api/organisations/{organisation['code']}"}]
 
-    # Retrieve document by doing a get request.
-    response = client.get(
-        url_for("invenio_records_rest.doc_item", pid_value=document_with_file["pid"]),
-        headers=headers,
-    )
+    data["identifiedBy"].append({"type": "ark", "value": "ark:/99999/foo"})
+    response = client.post(url_for("invenio_records_rest.doc_list"), headers=headers, data=json.dumps(data))
+    assert response.status_code == 201
+    data = response.json["metadata"]
+    assert [r for r in data["identifiedBy"] if r["type"] == "ark"]
+    doc = DocumentRecord.get_record_by_pid(data["pid"])
+    assert doc.get_ark() != "ark:/99999/foo"
 
-    # Put data to document
     response = client.put(
-        url_for("invenio_records_rest.doc_item", pid_value=document_with_file["pid"]),
+        url_for("invenio_records_rest.doc_item", pid_value=data["pid"]),
         headers=headers,
-        data=json.dumps(response.json["metadata"]),
+        data=json.dumps(data),
     )
     assert response.status_code == 200
 
+    response = client.delete(url_for("invenio_records_rest.doc_item", pid_value=data["pid"]))
+    assert response.status_code == 204
+    response = client.get(url_for("invenio_records_rest.doc_item", pid_value=data["pid"]))
+
+    assert response.status_code == 410
+
+    # assert document_with_file.get_ark()
+
+    # # Retrieve document by doing a get request.
+    # response = client.get(
+    #     url_for("invenio_records_rest.doc_item", pid_value=document_with_file["pid"]),
+    #     headers=headers,
+    # )
+    # data = response.json["metadata"]
+    # # Put data to document
+    # response = client.put(
+    #     url_for("invenio_records_rest.doc_item", pid_value=document_with_file["pid"]),
+    #     headers=headers,
+    #     data=json.dumps(data),
+    # )
+    # assert response.status_code == 200
+
+    # data["identifiedBy"].append({"type": "ark", "value": "ark:/99999/foo"})
+
+    # # Put data to document
+    # response = client.put(
+    #     url_for("invenio_records_rest.doc_item", pid_value=document_with_file["pid"]),
+    #     headers=headers,
+    #     data=json.dumps(data),
+    # )
+    # from sonar.modules.documents.api import DocumentRecord
+    # doc = DocumentRecord.get_record_by_pid(document_with_file["pid"])
+    # assert doc.get_ark() == document_with_file.get_ark()
+    # assert response.status_code == 200
+
 
 def test_aggregations(app, client, document, superuser, admin):
     """Test aggregations."""