Handle missing CURIE validation rows gracefully (#51)

Author: ubyndr

pandasaurus/curie_validator.py

@@ -142,12 +142,14 @@ def construct_term_list(seed_list) -> List[Term]:
         term_obsoletion = CurieValidator.find_obsolete_terms(seed_list)
         term_list: List[Term] = list()
         for seed in seed_list:
+            validation_entry = term_validation.get(seed, {"label": None, "valid": False})
+            obsoletion_entry = term_obsoletion.get(seed, {})
             term = Term(
-                term_validation.get(seed).get("label"),
+                validation_entry.get("label"),
                 seed,
-                term_validation.get(seed).get("valid"),
-                term_obsoletion.get(seed).get("new_term_label") if seed in term_obsoletion else None,
-                term_obsoletion.get(seed).get("new_term") if seed in term_obsoletion else None,
+                bool(validation_entry.get("valid")),
+                obsoletion_entry.get("new_term_label"),
+                obsoletion_entry.get("new_term"),
             )
             term_list.append(term)
         return term_list

test/data/curie_validator_data.py

@@ -70,3 +70,20 @@ def get_expected_construct_term_list():
         Term("naive B cell", "CL:0000788", True),
         Term("gamma-delta T cell", "CL:0000798", True),
     ]
+
+
+def get_construct_term_list_missing_validation_data():
+    return ["CL:0000084", "CL:9999999"]
+
+
+def get_construct_term_list_missing_validation_result():
+    return [
+        {"label": "T cell", "term": "CL:0000084"},
+    ]
+
+
+def get_expected_construct_term_list_missing_validation():
+    return [
+        Term("T cell", "CL:0000084", True),
+        Term(None, "CL:9999999", False),
+    ]

test/test_curie_validator.py

@@ -1,7 +1,10 @@
 from test.data.curie_validator_data import (
     get_construct_term_list_data,
+    get_construct_term_list_missing_validation_data,
+    get_construct_term_list_missing_validation_result,
     get_construct_term_list_result,
     get_expected_construct_term_list,
+    get_expected_construct_term_list_missing_validation,
     get_expected_find_obsolete_terms,
     get_expected_validate_curie_list,
     get_find_obsolete_terms_data,
@@ -139,3 +142,18 @@ def test_construct_term_list(mocker):
     )
 
     assert CurieValidator.construct_term_list(get_construct_term_list_data()) == get_expected_construct_term_list()
+
+
+def test_construct_term_list_handles_missing_validation(mocker):
+    mocker.patch(
+        "pandasaurus.curie_validator.run_sparql_query",
+        side_effect=[
+            iter(get_construct_term_list_missing_validation_result()),
+            iter([]),
+        ],
+    )
+
+    assert (
+        CurieValidator.construct_term_list(get_construct_term_list_missing_validation_data())
+        == get_expected_construct_term_list_missing_validation()
+    )