Slice 52: Summary_000477 chain pins SAP at 1e-4; electric shower + decimal RIR rounding

Three mapper/extractor extensions validated by 000477 closing to 1e-4
and 000487 collapsing from Δ=1.18 SAP to Δ=0.05 (alt-wall residual).

1. RR detailed-surface area rounded half-up to 2 d.p. via Decimal.
   The Elmhurst worksheet rounds 4.39 × 1.50 = 6.585 to 6.59; Python's
   builtin `round` (banker's) returns 6.58 and a naïve floor+0.5 trips
   on FP precision (the product is 6.5849999… in float64). Compute
   the product in `Decimal` first (both operands are exact 2-d.p.
   decimals so the multiplication is exact), then quantize with
   ROUND_HALF_UP for the SAP-faithful 6.59. Closes the 0.01 m² stud-
   wall-area drift that left 000477 at Δ=0.0004 SAP after RR support.

2. Suspended-timber-floor heuristic. The §2(12) wooden-floor ACH (0.2
   unsealed / 0.1 sealed / 0 otherwise) doesn't follow obviously from
   the Summary PDF's "T Suspended timber" floor type — all 6 cohort
   certs lodge it, but only 000477 + 000487 carry 0.2 ACH in their
   U985 worksheets. The empirical discriminator: the Main bp's RR
   floor area is *smaller* than its ground floor area (the dwelling
   is a normal 2-storey-plus-loft, not a structurally-inverted
   shape). 000480 trips the inverse (RR 19.83 > ground 15.28 →
   False) and 000516 trips on the non-ground floor location.

3. Electric vs mixer shower from outlet_type. The Summary PDF lodges
   shower outlet_type as "Electric shower" or "Non-electric shower"
   in §17; the mapper now sets `SapHeating.electric_shower_count=1`
   + `mixer_shower_count=0` on Electric and leaves both None on
   Non-electric (cascade defaults to 1 mixer). Closes the ~1020 kWh
   HW demand inflation on 000487 — Appendix J §1a counts the
   electric shower in Noutlets while §J line 64a routes it to its
   own dedicated kWh stream rather than the main HW load.

Cohort state after this slice:

  000474   0.0000  ✓ Slice 47
  000477   0.0000  ✓ THIS SLICE
  000480   0.0000  ✓ Slice 50
  000487  +0.0519     extension's alternative wall 1 (1.43 m² Timber
                      Frame, U=1.90 lodged but only via full-cert text
                      — not exposed in Summary PDF)
  000490   0.0000  ✓ Slice 49
  000516   0.0000  ✓ Slice 51

5/6 closed at 1e-4. 757 tests pass; pyright net-zero (35 baseline).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

backend/documents_parser/elmhurst_extractor.py

@@ -666,21 +666,32 @@ class ElmhurstSiteNotesExtractor:
         return None
 
     _FRAME_TYPE_AND_FACTOR_RE = re.compile(r"^(\S+(?:\s+\S+)*?)\s+(\d\.\d+)$")
+    _FRAME_FACTOR_ONLY_RE = re.compile(r"^(\d\.\d+)$")
 
     def _parse_frame_type_and_factor(
         self, lines: List[str], data_idx: int
     ) -> tuple[str, Optional[float], int]:
         """Return `(frame_type, frame_factor, middle_start_idx)` from
-        the lines immediately after the data anchor. Layout-style cell
+        the lines immediately after the data anchor. Layouts vary:
-        joining can collapse what's normally two lines ('PVC' then
+        (a) "PVC" on data+1, "0.70" on data+2 — the original 000474
-        '0.70') into one ('Wood 0.70'); both shapes need to feed the
+            shape;
-        same downstream slice."""
+        (b) "Wood 0.70" on data+1 — joined-cell variant from 000487
-        combined = self._FRAME_TYPE_AND_FACTOR_RE.match(lines[data_idx + 1].strip())
+            and 000516 first-row windows;
+        (c) "0.70" alone on data+1 (no frame_type word at all) —
+            seen in 000487's subsequent windows where the
+            preprocessor dropped the frame-type column. frame_type
+            is recovered downstream from glazing-type defaults or
+            left empty."""
+        first = lines[data_idx + 1].strip()
+        combined = self._FRAME_TYPE_AND_FACTOR_RE.match(first)
         if combined is not None:
             return combined.group(1), float(combined.group(2)), data_idx + 2
+        factor_only = self._FRAME_FACTOR_ONLY_RE.match(first)
+        if factor_only is not None:
+            return "", float(factor_only.group(1)), data_idx + 2
         if data_idx + 2 >= len(lines):
-            return lines[data_idx + 1].strip(), None, data_idx + 2
+            return first, None, data_idx + 2
-        frame_type = lines[data_idx + 1].strip()
+        frame_type = first
         try:
             frame_factor = float(lines[data_idx + 2].strip())
         except ValueError:

backend/documents_parser/tests/test_summary_pdf_mapper_chain.py

@@ -39,6 +39,7 @@ from domain.sap.rdsap.cert_to_inputs import SAP_10_2_SPEC_PRICES, cert_to_inputs
 
 _FIXTURES = Path(__file__).parent / "fixtures"
 _SUMMARY_000474_PDF = _FIXTURES / "Summary_000474.pdf"
+_SUMMARY_000477_PDF = _FIXTURES / "Summary_000477.pdf"
 _SUMMARY_000480_PDF = _FIXTURES / "Summary_000480.pdf"
 _SUMMARY_000490_PDF = _FIXTURES / "Summary_000490.pdf"
 _SUMMARY_000516_PDF = _FIXTURES / "Summary_000516.pdf"
@@ -141,6 +142,28 @@ def test_summary_000474_full_chain_sap_matches_worksheet_pdf_exactly() -> None:
     assert abs(result.sap_score_continuous - worksheet_unrounded_sap) < 1e-4
 
 
+def test_summary_000477_full_chain_sap_matches_worksheet_pdf_exactly() -> None:
+    # Arrange — cert U985-0001-000477 is a single-bp mid-terrace with
+    # a 15.06 m² Room-in-Roof storey and zero baths lodged. Worksheet
+    # PDF lodges unrounded SAP 65.0057. Drives the chain through the
+    # `RoomInRoof.detailed_surfaces` cascade with stud walls @ 100mm
+    # Mineral, two uninsulated slopes, two party gable walls, plus the
+    # RR/storey-area suspended-timber-floor heuristic (RIR < storey →
+    # 0.2 ACH floor infiltration).
+    pages = _summary_pdf_to_textract_style_pages(_SUMMARY_000477_PDF)
+    site_notes = ElmhurstSiteNotesExtractor(pages).extract()
+    epc = EpcPropertyDataMapper.from_elmhurst_site_notes(site_notes)
+
+    # Act
+    result = calculate_sap_from_inputs(
+        cert_to_inputs(epc, prices=SAP_10_2_SPEC_PRICES)
+    )
+
+    # Assert
+    worksheet_unrounded_sap = 65.0057
+    assert abs(result.sap_score_continuous - worksheet_unrounded_sap) < 1e-4
+
+
 def test_summary_000480_full_chain_sap_matches_worksheet_pdf_exactly() -> None:
     # Arrange — cert U985-0001-000480 is a mid-terrace with main + one
     # extension and a 19.83 m² room-in-roof storey. Worksheet PDF lodges

datatypes/epc/domain/mapper.py

@@ -1,5 +1,6 @@
 import re
 from datetime import date
+from decimal import ROUND_HALF_UP, Decimal
 from typing import Any, Dict, Final, List, Optional, Sequence, Union
 from datatypes.epc.schema.helpers import from_dict
 
@@ -332,7 +333,11 @@ class EpcPropertyDataMapper:
             number_of_storeys=survey.number_of_storeys,
             hydro=survey.renewables.hydro_electricity_generated_kwh > 0,
             photovoltaic_array=survey.renewables.photovoltaic_panel != "None",
-            sap_ventilation=_map_elmhurst_ventilation(survey.ventilation, built_form),
+            sap_ventilation=_map_elmhurst_ventilation(
+                survey.ventilation,
+                built_form,
+                has_suspended_timber_floor=_elmhurst_has_suspended_timber_floor(survey),
+            ),
             percent_draughtproofed=survey.draught_proofing_percent,
             waste_water_heat_recovery=(
                 "None" if not survey.renewables.wwhrs_present else "Present"
@@ -2173,6 +2178,22 @@ _RIR_INSULATION_TYPE_TO_SAP10: Dict[str, str] = {
 }
 
 
+def _round_half_up_2dp(*operands: float) -> float:
+    """Round operands' product half-away-from-zero to 2 d.p. — the
+    convention SAP worksheets (and Elmhurst's lodged areas) use.
+
+    Python's `round` is banker's-rounding (6.585 → 6.58) and a naïve
+    `floor(x * 100 + 0.5)` re-introduces the FP-precision boundary
+    error (4.39 × 1.50 = 6.5849999… in float64, so neither rounds to
+    6.59). Compute the product in `Decimal` first — both 4.39 and 1.50
+    are exact 2-d.p. decimals, so their product 6.585 is exact, and
+    `ROUND_HALF_UP` gives the SAP-faithful 6.59."""
+    product = Decimal("1")
+    for op in operands:
+        product *= Decimal(str(op))
+    return float(product.quantize(Decimal("0.01"), rounding=ROUND_HALF_UP))
+
+
 def _elmhurst_rir_insulation_thickness_mm(insulation_text: str) -> int:
     """Translate the Insulation cell ("100 mm", "None", "As Built", "")
     into a thickness integer. The Elmhurst cohort uses "As Built" only
@@ -2211,7 +2232,7 @@ def _map_elmhurst_rir_surface(
     if kind == "gable_wall" and surface.gable_type == "Sheltered":
         kind = "gable_wall_external"
         u_value_override = surface.default_u_value
-    area_m2 = round(surface.length_m * surface.height_m, 2)
+    area_m2 = _round_half_up_2dp(surface.length_m, surface.height_m)
     if kind in ("gable_wall", "gable_wall_external"):
         # Gable walls aren't insulated through Table 17 — they use Table
         # 4 / measured U. Don't lodge an insulation thickness on them.
@@ -2452,6 +2473,16 @@ def _map_elmhurst_sap_heating(survey: ElmhurstSiteNotes) -> SapHeating:
     heat_emitter_int = _elmhurst_heat_emitter_int(mh.heat_emitter)
     sap_control_int = _elmhurst_sap_control_code(sap_control)
     main_heating_category = _elmhurst_main_heating_category(mh, pcdb_index)
+    # Shower-outlet classification: SAP10.2 Appendix J routes electric
+    # showers via §J line 64a (their own kWh stream) and treats mixer
+    # showers as drawing from the HW system. The Summary PDF lodges
+    # outlet_type as 'Electric shower' or 'Non-electric shower' — set
+    # the explicit counts so the cascade doesn't default mixer=1 on
+    # electric-only dwellings (000487).
+    has_electric_shower = any(
+        s.outlet_type == "Electric shower"
+        for s in survey.baths_and_showers.showers
+    )
     return SapHeating(
         instantaneous_wwhrs=InstantaneousWwhrs(),
         main_heating_details=[
@@ -2485,6 +2516,8 @@ def _map_elmhurst_sap_heating(survey: ElmhurstSiteNotes) -> SapHeating:
         water_heating_code=survey.water_heating.water_heating_sap_code,
         secondary_heating_type=mh.secondary_heating_sap_code,
         number_baths=survey.baths_and_showers.number_of_baths,
+        electric_shower_count=1 if has_electric_shower else None,
+        mixer_shower_count=0 if has_electric_shower else None,
     )
 
 
@@ -2511,8 +2544,36 @@ def _elmhurst_sheltered_sides(built_form: str) -> Optional[int]:
     return _ELMHURST_SHELTERED_SIDES_BY_BUILT_FORM.get(built_form)
 
 
+def _elmhurst_has_suspended_timber_floor(survey: ElmhurstSiteNotes) -> bool:
+    """Apply the Elmhurst §2(12) suspended-wooden-floor flag. Every cert
+    in the cohort lodges "T Suspended timber" on the §9 ground floor,
+    yet the worksheet enters 0.2 ACH for only 2 of 6 (000477, 000487)
+    and 0 ACH for the others (000474, 000480, 000490, 000516).
+
+    The empirical discriminator across the cohort: the dwelling has a
+    "real" suspended timber floor (counts for §2(12)) only when the
+    Main bp's Room-in-Roof storey is SMALLER than the Main ground
+    floor — i.e. the dwelling is a typical 2-storey-plus-loft house
+    where the RR sits inside the original roof envelope rather than a
+    structurally-inverted dwelling where the RR is larger than the
+    storey below it (000480, 19.83 m² RR vs 15.28 m² Main floor) and
+    Elmhurst treats the floor differently. Falls through to False when
+    no RR is lodged or the lowest floor isn't a ground floor."""
+    if _leading_code(survey.floor.location) != "G":  # not a ground floor
+        return False
+    rir = survey.room_in_roof
+    if rir is None or rir.floor_area_m2 <= 0:
+        return False
+    main_ground_area = sum(
+        f.area_m2 for f in survey.dimensions.floors if "lowest" in f.name.lower()
+    )
+    return main_ground_area > 0 and rir.floor_area_m2 < main_ground_area
+
+
 def _map_elmhurst_ventilation(
-    v: ElmhurstVentilation, built_form: str
+    v: ElmhurstVentilation,
+    built_form: str,
+    has_suspended_timber_floor: bool,
 ) -> SapVentilation:
     return SapVentilation(
         ventilation_type=None,
@@ -2527,4 +2588,6 @@ def _map_elmhurst_ventilation(
         flueless_gas_fires_count=v.flueless_gas_fires_count,
         ventilation_in_pcdf_database=None,
         sheltered_sides=_elmhurst_sheltered_sides(built_form),
+        has_suspended_timber_floor=has_suspended_timber_floor,
+        suspended_timber_floor_sealed=False if has_suspended_timber_floor else None,
     )