Give SexType/SpeciesType built from IRIs a name; allow dandischema 0.15.x

dandi/metadata/util.py

@@ -288,10 +288,18 @@ def timedelta2duration(delta: timedelta) -> str:
     return s
 
 
+# Names for the sex IRIs which ``extract_sex`` produces itself
+SEX_NAMES_BY_IRI = {
+    "http://purl.obolibrary.org/obo/PATO_0000384": "Male",
+    "http://purl.obolibrary.org/obo/PATO_0000383": "Female",
+}
+
+
 def extract_sex(metadata: dict) -> models.SexType | None:
     value = metadata.get("sex", None)
     if value is not None and value != "":
-        value = value.lower()
+        value_orig = value.strip()
+        value = value_orig.lower()
         if value in ["m", "male"]:
             value_id = "http://purl.obolibrary.org/obo/PATO_0000384"
             value = "Male"
@@ -305,8 +313,11 @@ def extract_sex(metadata: dict) -> models.SexType | None:
             value_id = None
             value = "Other"
         elif value.startswith("http"):
-            value_id = value
-            value = None
+            value_id = value_orig
+            # dandischema requires a name; fall back to the IRI itself
+            value = {k.lower(): v for k, v in SEX_NAMES_BY_IRI.items()}.get(
+                value, value_id
+            )
         else:
             raise ValueError(f"Cannot interpret sex field: {value}")
         return models.SexType(identifier=value_id, name=value)
@@ -597,7 +608,8 @@ def extract_species(metadata: dict) -> models.SpeciesType | None:
                         value = " - ".join(
                             [result[key] for key in lookup if key in result]
                         )
-                value_matches.append((value_id, value))
+                # dandischema requires a name; fall back to the IRI itself
+                value_matches.append((value_id, value or value_id))
         else:
             lower_value = value_orig.lower().strip()
             for record in species_map:

dandi/tests/test_metadata.py

@@ -51,6 +51,7 @@
     SpeciesRecord,
     extract_age,
     extract_cellLine,
+    extract_sex,
     extract_species,
     parse_age,
     parse_purlobourl,
@@ -887,6 +888,54 @@ def test_species_rat(species: str) -> None:
     }
 
 
+@pytest.mark.ai_generated
+def test_species_unknown_iri_lookup_fails(monkeypatch: pytest.MonkeyPatch) -> None:
+    # An NCBITaxon IRI not in species_map whose label cannot be looked up
+    # still gets a name, as dandischema requires one
+    iri = "http://purl.obolibrary.org/obo/NCBITaxon_999999999"
+
+    def fail(*_args: Any, **_kwargs: Any) -> None:
+        raise ConnectionError("no network")
+
+    monkeypatch.setattr("dandi.metadata.util.parse_purlobourl", fail)
+    species_rec = extract_species({"species": iri})
+    assert species_rec is not None
+    assert str(species_rec.identifier) == iri
+    assert species_rec.name == iri
+
+
+@pytest.mark.ai_generated
+@pytest.mark.parametrize(
+    "sex,identifier,name",
+    [
+        ("M", "http://purl.obolibrary.org/obo/PATO_0000384", "Male"),
+        ("female", "http://purl.obolibrary.org/obo/PATO_0000383", "Female"),
+        ("U", None, "Unknown"),
+        (
+            "http://purl.obolibrary.org/obo/PATO_0000384",
+            "http://purl.obolibrary.org/obo/PATO_0000384",
+            "Male",
+        ),
+        (
+            "http://purl.obolibrary.org/obo/PATO_0000383",
+            "http://purl.obolibrary.org/obo/PATO_0000383",
+            "Female",
+        ),
+        # unknown IRI: kept as is (not lowercased) and used as the name
+        (
+            "http://purl.obolibrary.org/obo/PATO_0001340",
+            "http://purl.obolibrary.org/obo/PATO_0001340",
+            "http://purl.obolibrary.org/obo/PATO_0001340",
+        ),
+    ],
+)
+def test_extract_sex(sex: str, identifier: str | None, name: str) -> None:
+    rec = extract_sex({"sex": sex})
+    assert rec is not None
+    assert (None if rec.identifier is None else str(rec.identifier)) == identifier
+    assert rec.name == name
+
+
 @pytest.mark.parametrize(
     "species",
     [

pyproject.toml

@@ -50,7 +50,7 @@ dependencies = [
     # Exclude 0.13.0: it lacks the downgrade migrations (SIMPLE_DOWNGRADES
     # entries for releaseNotes / sameAs) that the client relies on to talk
     # to older Archive servers.  0.14.0 restored them.
-    "dandischema >= 0.12.0, != 0.13.0, < 0.15.0",
+    "dandischema >= 0.12.0, != 0.13.0, < 0.16.0",
     "etelemetry >= 0.2.2",
     "fasteners >= 0.19",
     "fscacher >= 0.3.0",