fix ontology look-up that failed with empty values

scirexport/airr_export.py

@@ -450,25 +450,20 @@ def format_id_val_ontologies(meta_dict: dict, ontologies_dic: dict) -> None:
         if ontologies_dic[x]["preferred_label"] == 1
     }
 
-    missing = [
-        (x, meta_dict[x])
-        for x in meta_dict
-        if meta_dict[x] not in preferred_labels
-        and x in validate_airr_dic
-        and "Ontology" in validate_airr_dic[x]["Format"]
-    ]
-
-    if missing:
-        logger.warning(f"WARNING {missing} not in ontologies table")
-        fix_ontos = {missing[0]: {"id": {}, "label": {}}}
-        # for x in missing}
+    fix_ontos = {}
 
-    else:
-        fix_ontos = {
-            x: {"id": meta_dict[x], "label": preferred_labels[meta_dict[x]]}
-            for x in meta_dict
-            if x in validate_airr_dic and "Ontology" in validate_airr_dic[x]["Format"]
-        }
+    for x in meta_dict:
+        if x in validate_airr_dic and "Ontology" in validate_airr_dic[x]["Format"]:
+            if meta_dict[x] not in preferred_labels:
+                logger.warning(f"WARNING {x} not in ontologies table")
+                fix_ontos[x] = {"id": None, "label": None}
+            elif meta_dict[x] in preferred_labels:
+                fix_ontos[x] = {
+                    "id": meta_dict[x],
+                    "label": preferred_labels[meta_dict[x]],
+                }
+            else:
+                print(x)
 
     for x in meta_dict:
         if x in fix_ontos:
@@ -584,14 +579,21 @@ def process_repertoires(
             else repertoire_dic[rep][x]
             for x in repertoire_dic[rep]
         }
-        list_reps.append(
-            sr.dump_repertoire(
-                format_id_val_ontologies(repertoire_dic[rep], replace_ontos)
-            )
+        rept = sr.dump_repertoire(
+            format_id_val_ontologies(repertoire_dic[rep], replace_ontos)
         )
 
+        list_reps.append(rept)
+
     essential_validation(repertoire_dic[rep], rep)
 
+    import airr
+
+    try:
+        airr.RepertoireSchema.validate_object(rept)
+    except airr.ValidationError as e:
+        print("has repertoire at array position with validation error: %s\n" % e)
+
     repertoires_dic = {"Repertoire": list_reps}
     return repertoires_dic