Merge pull request #2708 from opensafely-core/Jongmassey/amp-desc-to-synonym

Use AMP description field as the primary name, and use the name field as the synonym - using the existing synonym mechanisms developed for SNOMED CT codelists.

Author: Jongmassey

coding_systems/dmd/coding_system.py

@@ -1,4 +1,7 @@
-from django.db.models import Q
+import collections
+
+from django.db.models import F, Q
+from django.db.models.functions import Coalesce
 
 from ..base.coding_system_base import BuilderCompatibleCodingSystem
 from .models import AMP, AMPP, VMP, VMPP, VTM, Ing
@@ -47,21 +50,38 @@ def lookup_names(self, codes):
         # these models first
         # If not found, look up VTM, VMPP, AMPPs also, in case of user-uploaded codelists that
         # might contain these
+        #
+        # If the model type is AMP, we should prefer the `descr` field over the `nm` for the name
+        # as this contains the supplier name which helps distinguish AMPs apart.
+        # It appears to be populated in 100% of cases but coalese with `nm` just in case it is not.
         codes = set(codes)
         lookup = {}
         for model_cls in [AMP, VMP, AMPP, VMPP, VTM]:
-            matched = dict(
-                model_cls.objects.using(self.database_alias)
-                .filter(id__in=codes)
-                .values_list("id", "nm")
+            model_objs = model_cls.objects.using(self.database_alias).filter(
+                id__in=codes
             )
+            name_field = Coalesce("descr", "nm") if model_cls == AMP else F("nm")
+            model_objs = model_objs.annotate(name=name_field)
+            matched = dict(model_objs.values_list("id", "name"))
             for code, name in matched.items():
                 lookup[code] = f"{name} ({model_cls.__name__})"
             codes = codes - set(matched.keys())
             if not codes:
                 break
         return lookup
 
+    def lookup_synonyms(self, codes):
+        descriptions = (
+            AMP.objects.using(self.database_alias)
+            .filter(id__in=codes)
+            .values("id", "nm")
+        )
+
+        result = collections.defaultdict(list)
+        for d in descriptions:
+            result[d["id"]].append(d["nm"])
+        return dict(result)
+
     def code_to_term(self, codes):
         lookup = self.lookup_names(codes)
         unknown = set(codes) - set(lookup)

coding_systems/dmd/tests/test_coding_system.py

@@ -10,10 +10,11 @@ def coding_system():
 
 def test_lookup_names(dmd_data, coding_system):
     assert coding_system.lookup_names(
-        ["10514511000001106", "10525011000001107", "99999"]
+        ["10514511000001106", "10525011000001107", "99999", "22503111000001109"]
     ) == {
         "10514511000001106": "Adrenaline (base) 220micrograms/dose inhaler (VMP)",
         "10525011000001107": "Adrenaline (base) 220micrograms/dose inhaler refill (VMP)",
+        "22503111000001109": "AirSalb 100micrograms/dose inhaler CFC free (Sandoz Ltd) (AMP)",
     }
 
 
@@ -215,3 +216,10 @@ def test_search_by_term_specific_fields(
 )
 def test_search_by_code(dmd_data, coding_system, code, expected_response):
     assert coding_system.search_by_code(code) == expected_response
+
+
+def test_lookup_synonyms(dmd_data, coding_system):
+    assert coding_system.lookup_synonyms(["3293111000001105", "22503111000001109"]) == {
+        "3293111000001105": ["Aerolin 100micrograms/dose Autohaler"],
+        "22503111000001109": ["AirSalb 100micrograms/dose inhaler CFC free"],
+    }