feat(conservatory): read §6.1 geometry through extractor + mapper

RdSAP 10 §6.1 (PDF p.49) models a non-separated (heated) conservatory as
part of the dwelling. Until now the Summary §5 block was reduced to an
inert `has_conservatory` bool and the geometry (floor area, glazed
perimeter, glazing, storey height) was dropped on both paths.

Plumbing only — no cascade consumer yet (Slices B/C/D wire §3/§6):
- ElmhurstSiteNotesExtractor reads the §5 Conservatory block into a new
  `Conservatory` site-notes record (scoped to §5 so the generic
  "Floor Area"/"Room Height" labels can't collide with §4 dimensions);
- domain gains a frozen `SapConservatory` (floor area, glazed perimeter,
  double/single glazing, thermally-separated guard, equivalent storey
  count) on `EpcPropertyData.sap_conservatory`;
- the Elmhurst mapper threads it through, dropping SEPARATED
  conservatories per §6.2 ("A separated conservatory ... is disregarded").

Verified against the simulated case-44 Summary (RefNo 001431): extracts
floor_area=12.0, glazed_perimeter=9.0, double_glazed=True, 1 storey.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

backend/documents_parser/elmhurst_extractor.py

@@ -7,6 +7,7 @@ from datatypes.epc.surveys.elmhurst_site_notes import (
     BathsAndShowers,
     BuildingPartDimensions,
     CommunityHeating,
+    Conservatory,
     ElmhurstSiteNotes,
     ExtensionPart,
     FloorDetails,
@@ -30,6 +31,17 @@ from datatypes.epc.surveys.elmhurst_site_notes import (
 )
 
 
+def _parse_conservatory_storeys(raw: Optional[str]) -> float:
+    """Parse the §5 "Room Height" lodgement ("1 Storey", "1.5 Storey",
+    "1½ Storey") into the equivalent-storey count RdSAP 10 §6.1 translates
+    to a metre height. Defaults to 1.0 (single storey) when unparseable."""
+    if not raw:
+        return 1.0
+    text = raw.replace("½", ".5")
+    m = re.search(r"\d+(?:\.\d+)?", text)
+    return float(m.group(0)) if m else 1.0
+
+
 def _parse_solar_pitch_deg(raw: Optional[str]) -> Optional[int]:
     """Parse the §16.0 "Collector elevation" lodgement (e.g. "30°", "60°",
     or a bare integer). Returns None when absent or unparseable."""
@@ -81,6 +93,13 @@ class ElmhurstSiteNotesExtractor:
         except (ValueError, IndexError):
             return 0
 
+    def _float_val(self, label: str) -> Optional[float]:
+        v = self._next_val(label)
+        if not v:
+            return None
+        m = re.search(r"-?\d+(?:\.\d+)?", v)
+        return float(m.group(0)) if m else None
+
     def _date_val(self, label: str) -> date:
         v = self._next_val(label)
         if not v:
@@ -179,8 +198,39 @@ class ElmhurstSiteNotesExtractor:
         v = self._local_val(lines, label)
         return v is not None and v.lower() == "yes"
 
+    def _local_float(self, lines: List[str], label: str) -> Optional[float]:
+        v = self._local_val(lines, label)
+        if not v:
+            return None
+        m = re.search(r"-?\d+(?:\.\d+)?", v)
+        return float(m.group(0)) if m else None
+
     # --- section extractors ---
 
+    def _extract_conservatory(self) -> Optional[Conservatory]:
+        """Summary §5.0 — geometry of a conservatory (RdSAP 10 §6, PDF
+        p.49). Returns None when none is lodged. Scoped to the §5 block
+        so the generic labels ("Floor Area", "Room Height") can't collide
+        with §4 dimensions. A separated conservatory is still returned
+        (with `thermally_separated=True`); the mapper drops it per §6.2."""
+        if not self._bool_val("Is there a conservatory?"):
+            return None
+        lines = self._section_lines_first_end(
+            "5.0 Conservatory", ("7.0 Walls", "6.0 ", "Summary Information"),
+        )
+        return Conservatory(
+            thermally_separated=self._local_bool(
+                lines, "Is it thermally separated?"
+            ),
+            floor_area_m2=self._local_float(lines, "Floor Area [m2]") or 0.0,
+            double_glazed=self._local_bool(lines, "Double Glazed"),
+            glazed_perimeter_m=self._local_float(lines, "Glazed Perimeter [m]")
+            or 0.0,
+            room_height_storeys=_parse_conservatory_storeys(
+                self._local_val(lines, "Room Height")
+            ),
+        )
+
     def _extract_surveyor_info(self) -> SurveyorInfo:
         return SurveyorInfo(
             surveyor_code=self._str_val("Surveyor"),
@@ -1820,6 +1870,7 @@ class ElmhurstSiteNotesExtractor:
             construction_age_band=self._extract_main_age_band(),
             dimensions=self._extract_dimensions(),
             has_conservatory=self._bool_val("Is there a conservatory?"),
+            conservatory=self._extract_conservatory(),
             walls=self._extract_walls(),
             roof=self._extract_roof(),
             floor=self._extract_floor(),

datatypes/epc/domain/epc_property_data.py

@@ -261,6 +261,33 @@ class SapRoofWindow:
     window_location: Union[int, str] = 0
 
 
+@dataclass(frozen=True)
+class SapConservatory:
+    """RdSAP 10 §6.1 (PDF p.49) — a NON-SEPARATED (heated) conservatory.
+
+    Its floor area and volume are added to the dwelling total (TFA (4),
+    volume (5)); its fully-glazed walls bill as a window (27) and its
+    fully-glazed roof as a rooflight (27a); the floor adds a ground-loss
+    term (28a). U-values come from RdSAP 10 Table 25 (p.51): double 6 mm
+    window 3.1 / roof 3.4 / g 0.76; single window 4.8 / roof 5.3 / g 0.85.
+
+    `room_height_storeys` is the equivalent number of storey heights of
+    the dwelling to the nearest half (Summary §5 "Room Height", gov-API
+    glazed building part), translated to a metre height per §6.1:
+    1 storey → ground-floor room height; 1½ → ground + 0.25 + 0.5×first;
+    2 → ground + 0.25 + first; etc.
+
+    A SEPARATED conservatory (§6.2) is disregarded entirely and is never
+    represented here (`thermally_separated` stays a guard for the cascade).
+    """
+
+    floor_area_m2: float
+    glazed_perimeter_m: float
+    double_glazed: bool
+    thermally_separated: bool
+    room_height_storeys: float
+
+
 @dataclass
 class SapWindow:
     frame_material: Optional[str]
@@ -774,6 +801,10 @@ class EpcPropertyData:
     # has no roof windows; for cert-cascade fixtures the bootstrap path
     # lodges per-window area + raw U.
     sap_roof_windows: Optional[List[SapRoofWindow]] = None
+    # RdSAP 10 §6.1 — geometry of a non-separated (heated) conservatory.
+    # None when no conservatory is lodged or it is thermally separated
+    # (§6.2 disregards separated conservatories).
+    sap_conservatory: Optional[SapConservatory] = None
     calculation_software_version: Optional[str] = None  # Do we care about this?
     mechanical_vent_duct_placement: Optional[int] = None
     mechanical_vent_duct_insulation: Optional[int] = None

datatypes/epc/domain/mapper.py

@@ -24,6 +24,7 @@ from datatypes.epc.domain.epc_property_data import (
     SapEnergySource,
     SapFlatDetails,
     SapFloorDimension,
+    SapConservatory,
     SapHeating,
     SapRoofWindow,
     SapRoomInRoof,
@@ -68,6 +69,7 @@ from datatypes.epc.surveys.elmhurst_site_notes import (
     AlternativeWall as ElmhurstAlternativeWall,
     BuildingPartDimensions as ElmhurstBuildingPartDimensions,
     CommunityHeating,
+    Conservatory as ElmhurstConservatory,
     ElmhurstSiteNotes,
     FloorDetails as ElmhurstFloorDetails,
     MainHeating as ElmhurstMainHeating,
@@ -420,6 +422,7 @@ class EpcPropertyDataMapper:
             built_form=built_form,
             property_type=property_type,
             has_conservatory=survey.has_conservatory,
+            sap_conservatory=_map_elmhurst_conservatory(survey.conservatory),
             blocked_chimneys_count=survey.ventilation.blocked_chimneys_count,
             number_of_storeys=survey.number_of_storeys,
             hydro=survey.renewables.hydro_electricity_generated_kwh > 0,
@@ -5226,6 +5229,24 @@ def _elmhurst_roof_window_u_value(w: ElmhurstWindow) -> float:
     return w.u_value + _ELMHURST_ROOF_WINDOW_INCLINATION_ADJUSTMENT_W_PER_M2K
 
 
+def _map_elmhurst_conservatory(
+    cons: Optional[ElmhurstConservatory],
+) -> Optional[SapConservatory]:
+    """RdSAP 10 §6 — translate the Summary §5 conservatory geometry into
+    the domain `SapConservatory`. A SEPARATED conservatory (§6.2, PDF
+    p.49) is disregarded entirely, so it maps to None (the cascade adds
+    nothing). Returns None when no conservatory is lodged."""
+    if cons is None or cons.thermally_separated:
+        return None
+    return SapConservatory(
+        floor_area_m2=cons.floor_area_m2,
+        glazed_perimeter_m=cons.glazed_perimeter_m,
+        double_glazed=cons.double_glazed,
+        thermally_separated=cons.thermally_separated,
+        room_height_storeys=cons.room_height_storeys,
+    )
+
+
 def _map_elmhurst_roof_window(w: ElmhurstWindow) -> SapRoofWindow:
     return SapRoofWindow(
         area_m2=w.area_m2,

datatypes/epc/surveys/elmhurst_site_notes.py

@@ -478,6 +478,21 @@ class ExtensionPart:
     room_in_roof: Optional[RoomInRoof] = None
 
 
+@dataclass
+class Conservatory:
+    """Summary §5 geometry of a NON-SEPARATED conservatory (RdSAP 10
+    §6.1). `room_height_storeys` is the lodged equivalent-storey count
+    ("1 Storey" → 1.0, "1.5 Storey" → 1.5); the mapper/cascade translate
+    it to a metre height. A SEPARATED conservatory (§6.2) is disregarded,
+    so `thermally_separated=True` records are dropped before the cascade."""
+
+    thermally_separated: bool
+    floor_area_m2: float
+    double_glazed: bool
+    glazed_perimeter_m: float
+    room_height_storeys: float
+
+
 @dataclass
 class ElmhurstSiteNotes:
     surveyor_info: SurveyorInfo
@@ -560,3 +575,9 @@ class ElmhurstSiteNotes:
     # cold loft instead of a room-in-roof). The mapper translates the
     # surface table into a `SapRoomInRoof` attached to the Main bp.
     room_in_roof: Optional[RoomInRoof] = None
+
+    # §5.0 Conservatory geometry — None when the dwelling has no
+    # conservatory (`has_conservatory=False`). Populated (incl. for
+    # separated conservatories) so the mapper can apply the §6.1/§6.2
+    # rule; the mapper drops separated ones.
+    conservatory: Optional[Conservatory] = None