Map PasHub walls_insulation_type labels to SAP10 codes

`from_site_notes` copied the raw `walls_insulation_type` survey label onto
the building part; the calculator's `_int_or_none` read it as `None`, so
the ~101 filled-cavity walls in the cohort lost their lower-U credit and
were over-counted.

Add `_pashub_wall_insulation_type_int` (mirroring the wall-construction
helper and the Elmhurst `_ELMHURST_WALL_INSULATION_TO_SAP10` sibling)
mapping the surveyed labels to the SAP10 wall-insulation codes `u_wall`
consumes ("As built" -> 4 assumed/default, "Filled Cavity" -> 2,
"External" -> 1), strict-raising `UnmappedPasHubLabel` on an unknown label.

Stacked on #1560, this is the fabric fix that removes the cohort's
systematic SAP bias: mean signed -0.79 -> -0.04, MAE 3.11 -> 2.79,
within-0.5 8.5% -> 11.4%.

Closes #1561

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

datatypes/epc/domain/mapper.py

@@ -5939,7 +5939,9 @@ def _map_main_building_part(
         wall_construction=_pashub_wall_construction_int(
             main.walls_construction_type
         ),
-        wall_insulation_type=main.walls_insulation_type,
+        wall_insulation_type=_pashub_wall_insulation_type_int(
+            main.walls_insulation_type
+        ),
         wall_thickness_measured=main.wall_thickness_mm is not None,
         party_wall_construction=_pashub_party_wall_construction_int(
             main.party_wall_construction_type
@@ -5967,7 +5969,9 @@ def _map_extension_building_part(
         wall_construction=_pashub_wall_construction_int(
             ext_c.walls_construction_type
         ),
-        wall_insulation_type=ext_c.walls_insulation_type,
+        wall_insulation_type=_pashub_wall_insulation_type_int(
+            ext_c.walls_insulation_type
+        ),
         wall_thickness_measured=ext_c.wall_thickness_mm is not None,
         party_wall_construction=_pashub_party_wall_construction_int(
             ext_c.party_wall_construction_type
@@ -7160,6 +7164,33 @@ def _pashub_wall_construction_int(label: Optional[str]) -> Union[int, str]:
     return _PASHUB_WALL_CONSTRUCTION_TO_SAP10[key]
 
 
+# PasHub surveyed `walls_insulation_type` label → SAP10 wall-insulation code that
+# `u_wall` consumes. Same target code-space as the Elmhurst sibling
+# (`_ELMHURST_WALL_INSULATION_TO_SAP10`) — PasHub just lodges the full human
+# label. "As built" is the assumed/default cascade code (Elmhurst "A"=4).
+_PASHUB_WALL_INSULATION_TYPE_TO_SAP10: Dict[str, int] = {
+    "As built": 4,  # as built / assumed (default cascade)
+    "Filled Cavity": 2,  # WALL_INSULATION_FILLED_CAVITY
+    "External": 1,  # external wall insulation
+}
+
+
+def _pashub_wall_insulation_type_int(label: Optional[str]) -> Union[int, str]:
+    """Resolve a PasHub surveyed `walls_insulation_type` label to its SAP10
+    wall-insulation code at the mapper boundary. A blank label is "no lodging" and
+    passes through as the empty string (the field is `Union[int, str]`; the
+    calculator's `_int_or_none` resolves a non-numeric string to None); a non-empty
+    label the lookup does not cover strict-raises `UnmappedPasHubLabel` so the
+    coverage gap is fixed here rather than silently dropping the filled-cavity
+    U-value credit downstream (#1561)."""
+    if label is None or not label.strip():
+        return ""
+    key = label.strip()
+    if key not in _PASHUB_WALL_INSULATION_TYPE_TO_SAP10:
+        raise UnmappedPasHubLabel("wall insulation type", label)
+    return _PASHUB_WALL_INSULATION_TYPE_TO_SAP10[key]
+
+
 def _resolve_elmhurst_underfloor_subtype(
     main_floor: ElmhurstFloorDetails,
     main_age_band: str,

datatypes/epc/domain/tests/test_from_site_notes.py

@@ -113,6 +113,44 @@ class TestWallConstructionCoding:
             EpcPropertyDataMapper.from_site_notes(survey)
 
 
+class TestWallInsulationTypeCoding:
+    """`from_site_notes` must int-code the PAS Hub `walls_insulation_type` label
+    to the SAP10 wall-insulation code `u_wall` consumes (matching the Elmhurst
+    `_ELMHURST_WALL_INSULATION_TO_SAP10` sibling), not copy the raw label. A raw
+    label reads as `None`, so a filled cavity loses its lower-U credit and the
+    wall is over-counted (Guinness cohort, issue #1561).
+    """
+
+    @pytest.mark.parametrize(
+        "label, expected_code",
+        [
+            ("As built", 4),  # as built / assumed (default cascade)
+            ("Filled Cavity", 2),  # WALL_INSULATION_FILLED_CAVITY
+            ("External", 1),  # external wall insulation
+            ("", ""),  # no lodging — passes through (calc nulls it via _int_or_none)
+        ],
+    )
+    def test_label_maps_to_sap_code(
+        self, label: str, expected_code: "int | str"
+    ) -> None:
+        data = load("pashub_rdsap_site_notes_example1.json")
+        data["building_construction"]["main_building"][
+            "walls_insulation_type"
+        ] = label
+        survey = from_dict(PasHubRdSapSiteNotes, data)
+        result = EpcPropertyDataMapper.from_site_notes(survey)
+        assert result.sap_building_parts[0].wall_insulation_type == expected_code
+
+    def test_unknown_label_strict_raises(self) -> None:
+        data = load("pashub_rdsap_site_notes_example1.json")
+        data["building_construction"]["main_building"][
+            "walls_insulation_type"
+        ] = "Aerogel Blanket"
+        survey = from_dict(PasHubRdSapSiteNotes, data)
+        with pytest.raises(UnmappedPasHubLabel):
+            EpcPropertyDataMapper.from_site_notes(survey)
+
+
 class TestFromSiteNotesExample1:
     """
     Fixture: pashub_rdsap_site_notes_example1.json
@@ -302,8 +340,8 @@ class TestFromSiteNotesExample1:
         assert result.sap_building_parts[0].wall_construction == 4
 
     def test_wall_insulation_type(self, result: EpcPropertyData) -> None:
-        # main_building.walls_insulation_type: "As built"
-        assert result.sap_building_parts[0].wall_insulation_type == "As built"
+        # main_building.walls_insulation_type: "As built" → SAP10 code 4 (as built/assumed)
+        assert result.sap_building_parts[0].wall_insulation_type == 4
 
     def test_wall_thickness_measured(self, result: EpcPropertyData) -> None:
         # main_building.wall_thickness_mm: 280 → thickness was measured
@@ -549,7 +587,7 @@ class TestFromSiteNotesExample1:
                     identifier=BuildingPartIdentifier.MAIN,
                     construction_age_band="I",
                     wall_construction=4,  # "Cavity" → SAP10 code 4 (WALL_CAVITY)
-                    wall_insulation_type="As built",
+                    wall_insulation_type=4,  # "As built" → SAP10 code 4 (as built/assumed)
                     wall_thickness_measured=True,
                     party_wall_construction=4,  # "Cavity Masonry, Unfilled" → SAP10 code 4 (U=0.5)
                     sap_floor_dimensions=[