adjusted invenio.py and its test a bit

Author: notactuallyfinn

src/hermes/commands/deposit/invenio.py

@@ -513,6 +513,8 @@ def _codemeta_to_invenio_deposition(self) -> dict:
 
         creators = []
         for author in metadata.get("author", []):
+            if not "Person" in author.get("@type", []):
+                continue
             creator = {}
             if len(
                 affils := [

src/hermes/commands/deposit/invenio_rdm.py

@@ -6,9 +6,8 @@
 # SPDX-FileContributor: Oliver Bertuch
 # SPDX-FileContributor: Michael Meinel
 
-import typing as t
-
 from requests import HTTPError
+from typing import Union
 
 from hermes.commands.deposit.invenio import InvenioClient, InvenioDepositPlugin, InvenioResolver
 
@@ -27,7 +26,7 @@ def get_licenses(self):
 class InvenioRDMResolver(InvenioResolver):
     invenio_client_class = InvenioRDMClient
 
-    def resolve_license_id(self, license_url: t.Optional[str]) -> t.Optional[dict]:
+    def resolve_license_id(self, license_url: Union[str, None]) -> Union[dict, None]:
         """Deliberately try to resolve the license URL to a valid InvenioRDM license information record from the
         vocabulary.
 
@@ -47,6 +46,12 @@ def resolve_license_id(self, license_url: t.Optional[str]) -> t.Optional[dict]:
         except HTTPError:
             pass
 
+        # FIXME: Why not get all license_cross_refs and then use a query parameter like this:
+        # ?q=props.url:("license_url" OR "license_cross_ref[1]" OR ...)&size=1000
+        # That would be able to replace _search_license_info.
+        # FIXME: Some licenses in valid_licenses["hits"]["hits"]["props"]["url"] are only http although
+        # https://spdx.org/licenses/license.json lists them in crossRef as https
+
         # If the easy "mapping" did not work, we really need to "search" for the correct license ID.
         response = self.client.get_licenses()
         response.raise_for_status()
@@ -65,6 +70,7 @@ def resolve_license_id(self, license_url: t.Optional[str]) -> t.Optional[dict]:
                 if license_info is not None:
                     break
             else:
+                # FIXME: Why is this only raised here and not always when license_info is None?
                 raise RuntimeError(f"Could not resolve license URL {license_url} to a valid identifier.")
 
         return license_info
@@ -73,7 +79,7 @@ def resolve_license_id(self, license_url: t.Optional[str]) -> t.Optional[dict]:
     def _extract_license_id_from_response(data: dict) -> str:
         return data["id"]
 
-    def _search_license_info(self, _url: str, valid_licenses: dict) -> t.Optional[dict]:
+    def _search_license_info(self, _url: str, valid_licenses: dict) -> Union[dict, None]:
         for license_info in valid_licenses['hits']['hits']:
             try:
                 if license_info['props']['url'] == _url:

test/hermes_test/model/test_api_e2e.py

@@ -172,7 +172,7 @@ def sandbox_auth():
                 "http://schema.org/license": [{"@id": "https://spdx.org/licenses/Apache-2.0"}],
                 "http://schema.org/name": [{"@value": "Test"}],
                 "http://schema.org/url": [
-                    {"@id": 'https://arxiv.org/abs/2201.09015'},
+                    {"@id": "https://arxiv.org/abs/2201.09015"},
                     {"@id": "https://docs.software-metadata.pub/en/latest"}
                 ],
                 "http://schema.org/version": [{"@value": "9.0.1"}]
@@ -398,30 +398,45 @@ def test_file_deposit(tmp_path, monkeypatch, metadata):
         if e.code != 0:
             raise e
     finally:
-        with open('codemeta.json', 'r') as cache:
+        with open("codemeta.json", "r") as cache:
             result = SoftwareMetadata(json.load(cache))
         sys.argv = orig_argv
 
     assert result == metadata
 
 
 @pytest.mark.parametrize(
-    "metadata",
+    "metadata, invenio_metadata",
     [
-        SoftwareMetadata({
-            "@type": ["http://schema.org/SoftwareSourceCode"],
-            "http://schema.org/description": [{"@value": "for testing"}],
-            "http://schema.org/name": [{"@value": "Test"}],
-            "http://schema.org/author": [{
-                "@type": "http://schema.org/Person",
-                "http://schema.org/familyName": [{"@value": "Test"}],
-                "http://schema.org/givenName": [{"@value": "Testi"}]
-            }],
-            "http://schema.org/license": [{"@id": "https://spdx.org/licenses/apache-2.0"}]
-        }),
+        (
+            SoftwareMetadata({
+                "@type": ["http://schema.org/SoftwareSourceCode"],
+                "http://schema.org/description": [{"@value": "for testing"}],
+                "http://schema.org/name": [{"@value": "Test"}],
+                "http://schema.org/author": [{
+                    "@type": "http://schema.org/Person",
+                    "http://schema.org/familyName": [{"@value": "Test"}],
+                    "http://schema.org/givenName": [{"@value": "Testi"}]
+                }],
+                "http://schema.org/license": [{"@id": "https://spdx.org/licenses/Apache-2.0"}]
+            }),
+            {
+                "upload_type": "software",
+                "publication_date": "2026-02-02",
+                "title": "Test",
+                "creators": [{"name": "Test, Testi"}],
+                "description": "for testing",
+                "access_right": "closed",
+                "license": "apache-2.0",
+                "prereserve_doi": True,
+                "related_identifiers": [
+                    {"identifier": "10.5281/zenodo.13311079", "relation": "isCompiledBy", "scheme": "doi"}
+                ]
+            }
+        )
     ]
 )
-def test_invenio_deposit(tmp_path, monkeypatch, sandbox_auth, metadata):
+def test_invenio_deposit(tmp_path, monkeypatch, sandbox_auth, metadata, invenio_metadata):
     monkeypatch.chdir(tmp_path)
 
     manager = context_manager.HermesContext(tmp_path)
@@ -453,10 +468,10 @@ def test_invenio_deposit(tmp_path, monkeypatch, sandbox_auth, metadata):
             raise e
     finally:
         manager.prepare_step("deposit")
-        with manager["deposit"] as cache:
-            result = cache["result"]
+        with manager["invenio"] as cache:
+            result = cache["deposit"]
         manager.finalize_step("deposit")
         sys.argv = orig_argv
 
     # TODO: compare to actually expected value
-    assert result == {}
+    assert result == invenio_metadata