Merge pull request #1417 from dandi/gh-1308

Convert DOI URLs in `related_publications` to related resources

dandi/metadata/util.py

@@ -11,6 +11,7 @@
 from dandischema import models
 import requests
 import tenacity
+from yarl import URL
 
 from .. import __version__
 from ..utils import ensure_datetime
@@ -583,6 +584,29 @@ def extract_digest(metadata: dict) -> dict[models.DigestType, str] | None:
         return None
 
 
+def extract_related_resource(metadata: dict) -> list[models.Resource] | None:
+    pubs = metadata.get("related_publications")
+    if not isinstance(pubs, (list, tuple)):
+        return None
+    related = []
+    for v in pubs:
+        if not isinstance(v, str):
+            continue
+        try:
+            u = URL(v)
+        except ValueError:
+            continue
+        if u.scheme not in ("http", "https") or u.host != "doi.org":
+            continue
+        related.append(
+            models.Resource(
+                identifier=v,
+                relation=models.RelationType.IsDescribedBy,
+            )
+        )
+    return related
+
+
 FIELD_EXTRACTORS: dict[str, Callable[[dict], Any]] = {
     "wasDerivedFrom": extract_wasDerivedFrom,
     "wasAttributedTo": extract_wasAttributedTo,
@@ -595,6 +619,7 @@ def extract_digest(metadata: dict) -> dict[models.DigestType, str] | None:
     "anatomy": extract_anatomy,
     "digest": extract_digest,
     "species": extract_species,
+    "relatedResource": extract_related_resource,
 }
 
 

dandi/tests/data/metadata/metadata2asset_3.json

@@ -92,5 +92,12 @@
         "name": "Cyperus bulbosus"
       }
     }
+  ],
+  "relatedResource": [
+    {
+      "schemaKey": "Resource",
+      "identifier": "https://doi.org/10.48324/dandi.000027/0.210831.2033",
+      "relation": "dcite:IsDescribedBy"
+    }
   ]
 }

dandi/tests/data/metadata/metadata2asset_simple1.json

@@ -42,5 +42,6 @@
       "schemaKey": "Participant",
       "identifier": "sub-01"
     }
-  ]
+  ],
+  "relatedResource": []
 }

dandi/tests/test_metadata.py

@@ -332,7 +332,9 @@ def test_timedelta2duration(td: timedelta, duration: str) -> None:
                 "institution": "University College",
                 "keywords": ["test", "sample", "example", "test-case"],
                 "lab": "Retriever Laboratory",
-                "related_publications": "A Brief History of Test Cases",
+                "related_publications": [
+                    "https://doi.org/10.48324/dandi.000027/0.210831.2033"
+                ],
                 "session_description": "Some test data",
                 "session_id": "XYZ789",
                 "session_start_time": "2020-08-31T15:58:28-04:00",
@@ -871,6 +873,7 @@ def test_nwb2asset(simple2_nwb: Path) -> None:
         variableMeasured=[],
         measurementTechnique=[],
         approach=[],
+        relatedResource=[],
     )
 
 
@@ -950,4 +953,5 @@ def test_nwb2asset_remote_asset(nwb_dandiset: SampleDandiset) -> None:
         variableMeasured=[],
         measurementTechnique=[],
         approach=[],
+        relatedResource=[],
     )