Add top-level descriptions response field (#20)

Author: shuchenliu

src/nodenorm/handlers/normalized_nodes.py

@@ -201,7 +201,7 @@ async def get_normalized_nodes(
 
 async def create_normalized_node(
     aggregate_node: NormalizedNode,
-    include_descriptions: bool = True,
+    include_descriptions: bool = False,
     include_individual_types: bool = False,
     conflations: dict = None,
 ) -> dict:
@@ -254,16 +254,19 @@ async def create_normalized_node(
         else:
             normal_node = {"id": {"identifier": aggregate_node.canonical_identifier}}
 
-    # if descriptions are enabled, look for the first available description and use that
+    # if descriptions are enabled, collect all available descriptions and use the first as the preferred one
     if include_descriptions:
-        descriptions = list(
-            map(
-                lambda x: x[0],
-                filter(lambda x: len(x) > 0, [eid["d"] for eid in aggregate_node.identifiers if "d" in eid]),
-            )
+        descriptions = unique_list(
+            [
+                description
+                for identifier in aggregate_node.identifiers
+                for description in identifier.get("d", [])
+                if description
+            ]
         )
         if len(descriptions) > 0:
             normal_node["id"]["description"] = descriptions[0]
+            normal_node["descriptions"] = descriptions
 
     # now need to reformat the identifier keys.  It could be cleaner but we have to worry about if there is a label
     normal_node["equivalent_identifiers"] = []

tests/test_normalized_nodes_lookup.py

@@ -41,6 +41,8 @@ def load_normalized_nodes_module():
 normalized_nodes = load_normalized_nodes_module()
 _lookup_curie_metadata = normalized_nodes._lookup_curie_metadata
 _lookup_equivalent_identifiers = normalized_nodes._lookup_equivalent_identifiers
+create_normalized_node = normalized_nodes.create_normalized_node
+NormalizedNode = normalized_nodes.NormalizedNode
 
 
 class FakeAsyncElasticsearch:
@@ -78,6 +80,50 @@ def no_hit_response():
     return {"hits": {"total": {"value": 0}, "hits": []}}
 
 
+@pytest.mark.asyncio
+async def test_create_normalized_node_aggregates_descriptions_when_requested():
+    node = NormalizedNode(
+        curie="NCIT:C34373",
+        canonical_identifier="MONDO:0004976",
+        preferred_label="amyotrophic lateral sclerosis",
+        information_content=74.9,
+        identifiers=[
+            {"i": "MONDO:0004976", "l": "amyotrophic lateral sclerosis", "d": ["first description"]},
+            {"i": "NCIT:C34373", "l": "Amyotrophic Lateral Sclerosis", "d": ["second description"]},
+            {"i": "UMLS:C0002736", "l": "Amyotrophic Lateral Sclerosis", "d": ["first description", ""]},
+            {"i": "MESH:D000690", "l": "Amyotrophic Lateral Sclerosis"},
+        ],
+        types=["biolink:Disease"],
+        taxa=[],
+    )
+
+    response = await create_normalized_node(node, include_descriptions=True)
+
+    assert response["id"]["description"] == "first description"
+    assert response["descriptions"] == ["first description", "second description"]
+    assert response["equivalent_identifiers"][0]["description"] == "first description"
+    assert response["equivalent_identifiers"][1]["description"] == "second description"
+
+
+@pytest.mark.asyncio
+async def test_create_normalized_node_hides_descriptions_by_default():
+    node = NormalizedNode(
+        curie="NCIT:C34373",
+        canonical_identifier="MONDO:0004976",
+        preferred_label="amyotrophic lateral sclerosis",
+        information_content=74.9,
+        identifiers=[{"i": "MONDO:0004976", "l": "amyotrophic lateral sclerosis", "d": ["first description"]}],
+        types=["biolink:Disease"],
+        taxa=[],
+    )
+
+    response = await create_normalized_node(node)
+
+    assert "description" not in response["id"]
+    assert "descriptions" not in response
+    assert "description" not in response["equivalent_identifiers"][0]
+
+
 @pytest.mark.asyncio
 async def test_lookup_equivalent_identifiers_uses_shared_msearch_index():
     namespace = fake_namespace([[hit_response("CHEBI:17310"), no_hit_response()]])