Compute a canonical disease display label in the ETL

src/igh_data_transform/transformations/diseases.py

@@ -181,6 +181,36 @@ def transform_diseases(
         mask = has_dash & suffix_matches_secondary
         df.loc[mask, "disease_filter"] = parent_candidate[mask]
 
+    # =========================================================
+    # Display label (single source of truth)
+    # =========================================================
+    #
+    # The portal prints ONE disease label in tables, slide-ins and the
+    # per-column table filter. The agreed rule:
+    #   * default       -> the secondary disease (COVID-19, Lassa fever,
+    #                       Gonorrhea, …);
+    #   * no secondary  -> the primary disease group (Tuberculosis,
+    #                       Buruli ulcer);
+    #   * Malaria (the  -> "<primary> – <secondary>", because the strains
+    #     only outlier)     (P. falciparum, P. vivax) never stand alone.
+    # Computing it here keeps the rule in one tested place; downstream
+    # (Gold, GraphQL, the React tables/slide-ins) just passes it through.
+    # Both inputs are already normalized above, so "no secondary" is
+    # exactly `secondary_disease_name IS NULL` here.
+    if "disease_filter" in df.columns and "secondary_disease_name" in df.columns:
+        primary = df["disease_filter"]
+        secondary = df["secondary_disease_name"]
+        has_secondary = secondary.notna()
+
+        # Default branch: the secondary when present, else the primary group.
+        label = secondary.where(has_secondary, primary)
+
+        # Malaria outlier: prefix the primary so the strain never stands alone.
+        malaria = has_secondary & primary.eq("Malaria")
+        label = label.mask(malaria, primary.str.cat(secondary, sep=" – "))
+
+        df["disease_label"] = label
+
     cleaned_option_sets: dict[str, pd.DataFrame] = {}
 
     if option_sets and "_optionset_new_globalhealtharea" in option_sets:

src/igh_data_transform/transformations/silver_to_gold/config/schema_map.py

@@ -83,6 +83,10 @@
         # any new joins to filter by either.
         "disease_filter": "disease_filter",
         "secondary_disease_name": "secondary_disease_name",
+        # Pre-computed canonical display label (primary/secondary/Malaria
+        # rule). Tables, slide-ins and the per-column table filter all read
+        # this so there is one definition of how a disease is named.
+        "disease_label": "disease_label",
     },
     "dim_phase": {
         "_source_table": "vin_rdstages",

tests/e2e/test_silver_to_gold_e2e.py

@@ -191,6 +191,7 @@ def test_dim_disease_has_expected_columns(self, gold_conn):
             "disease_name",
             "disease_group_name",
             "global_health_area",
+            "disease_label",
         }
         missing = expected - cols
         assert not missing, f"Missing columns: {missing}"

tests/unit/test_diseases.py

@@ -309,3 +309,76 @@ def test_normalizes_sti_primary_when_suffix_matches_secondary(self):
         # Secondary unchanged.
         assert result["secondary_disease_name"].iloc[0] == "Gonorrhea"
         assert result["secondary_disease_name"].iloc[1] == "Different"
+
+    def test_disease_label_prefers_secondary(self):
+        # Default rule: when a secondary disease exists, it is the label
+        # (the primary group is implied by context).
+        df = self._make_input_df(
+            overrides={
+                "new_diseasefilter": ["Coronaviral diseases", "Filoviral diseases"],
+                "new_secondary_diseae_choice_text": ["COVID-19", "Ebola"],
+            }
+        )
+        result, _ = transform_diseases(df)
+        assert result["disease_label"].iloc[0] == "COVID-19"
+        assert result["disease_label"].iloc[1] == "Ebola"
+
+    def test_disease_label_falls_back_to_primary_without_secondary(self):
+        # No secondary -> show the primary disease group. The sentinel
+        # "No secondary disease" has already been collapsed to NULL.
+        df = self._make_input_df(
+            overrides={
+                "new_diseasefilter": ["Tuberculosis", "Buruli ulcer"],
+                "new_secondary_diseae_choice_text": [None, "No secondary disease"],
+            }
+        )
+        result, _ = transform_diseases(df)
+        assert result["disease_label"].iloc[0] == "Tuberculosis"
+        assert result["disease_label"].iloc[1] == "Buruli ulcer"
+
+    def test_disease_label_combines_for_malaria(self):
+        # Malaria is the only outlier: its strains never stand alone, so
+        # the label is "<primary> – <secondary>" with a spaced en dash.
+        df = self._make_input_df(
+            overrides={
+                "new_diseasefilter": ["Malaria", "Malaria"],
+                "new_secondary_diseae_choice_text": ["P. falciparum", "P. vivax"],
+            }
+        )
+        result, _ = transform_diseases(df)
+        assert result["disease_label"].iloc[0] == "Malaria – P. falciparum"
+        assert result["disease_label"].iloc[1] == "Malaria – P. vivax"
+
+    def test_disease_label_for_sti_uses_secondary(self):
+        # STIs arrive parent-collapsed with the specific infection in the
+        # secondary field, so the default branch already prints it.
+        df = self._make_input_df(
+            overrides={
+                "new_diseasefilter": [
+                    "Sexually transmitted infections (STIs) - Gonorrhea",
+                    "Sexually transmitted infections (STIs)",
+                ],
+                "new_secondary_diseae_choice_text": [
+                    "Gonorrhea",
+                    "Trichomoniasis",
+                ],
+            }
+        )
+        result, _ = transform_diseases(df)
+        assert result["disease_label"].iloc[0] == "Gonorrhea"
+        assert result["disease_label"].iloc[1] == "Trichomoniasis"
+
+    def test_disease_label_is_null_when_both_inputs_missing(self):
+        # Cross-cutting "R&D for all global health areas" rows carry
+        # neither a primary filter nor a secondary -> no label. The
+        # second row keeps a secondary populated so the column survives
+        # `drop_empty_columns` (in production it always has values).
+        df = self._make_input_df(
+            overrides={
+                "new_diseasefilter": [None, "Dengue"],
+                "new_secondary_diseae_choice_text": [None, "Severe dengue"],
+            }
+        )
+        result, _ = transform_diseases(df)
+        assert pd.isna(result["disease_label"].iloc[0])
+        assert result["disease_label"].iloc[1] == "Severe dengue"