Crediting the full-SAP lodged PV closes the 2-point under-rating on uprn
10023444324 — the engine now reproduces the accredited lodged value exactly.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Full-SAP certs lodge measured PV under sap_energy_source.pv_arrays, but the
SAP-Schema-17.1 SapEnergySource modelled only wind, so from_dict dropped it at
parse and the mapper hardcoded no PV. A new SapPvArray schema field + the
_sap_17_1_pv_arrays mapper helper carry it to the domain photovoltaic_arrays the
Appendix-M generation credit reads — without it an all-electric dwelling the
array lifts to A/B is mis-modelled down a band or more.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Full-SAP energy_tariff (1=standard/2=7hr/3=10hr/4=24hr) and RdSAP meter_type
(1=dual/2=single/3=unknown/4=24hr) are different code spaces; the mapper passed
the code through untranslated, reading standard-tariff full-SAP certs as Economy 7
(over-rated) and Economy-7 certs as single (under-rated). Map onto the RdSAP word
aliases so the resolved Table 12a tariff is correct; absent/ND -> unknown.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
0 = no PV, 1 = not connected to dwelling meter, 2 = connected. Decoded
empirically from the corpus + accredited Elmhurst (the gov RdSAP schema
types it as a bare int). Points readers to the crediting gate.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
_is_elmhurst_roof_window routed a vertical window (Location "External wall")
to sap_roof_windows whenever its building part's roof type was a party
ceiling (A "Another dwelling above" / NR "Non-residential space above").
That is right for an EXTENSION (cert 000565 Ext2/Ext4 score on worksheet
(27a)) but wrong for the MAIN BP: a Main "A"/"NR" roof is the dwelling's own
party ceiling — no external roof — so its windows are vertical.
Gate the BP-roof-type rooflight rule to non-Main building parts. On a mid-
floor flat (case 53 / cert 000565 re-keyed) the Main External-wall window is
now vertical: its area deducts from the wall (was over-counted ~7 W/K) and it
prices as a window. Summary-path SAP 74.0 -> 75.17, matching Elmhurst's
accredited worksheet (75; space-heat 3408.6 vs 3408.58) — confirming the
calculator is correct; the gap was the misclassification.
000565's extension rooflights, the "Roof of Room" branch, and the U>3.0
room-in-roof backstop are unchanged. gov-API path unaffected (it routes roof
windows via window_wall_type==4).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Both passed no sap_ventilation= → EpcPropertyData default None, dropping the
whole §2 block so MEV/MVHR certs scored as natural. Mirror 21.0.1's block for
the only fields each schema lodges: mechanical_ventilation_kind (§2 MV-kind
dispatch) + sheltered_sides (§S5, from built_form). Neither schema lodges the
fan/flue/vent counts (RdSAP Table-5 age defaults), so those stay unset.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
19.0 + 21.0.0 pass no sap_ventilation= at all → EpcPropertyData default None,
dropping the entire §2 block (sheltered_sides + mechanical_ventilation_kind).
Extend the MVHR mapper test to both and add a sheltered_sides regression,
mirroring 21.0.1 (remaining ADR-0037 ventilation follow-up).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
17.1/18.0/20.0.0 built SapVentilation(sheltered_sides=…) and dropped the cert's
mechanical_ventilation, mis-scoring MEV/MVHR dwellings as natural. Mirror 21.0.1's
_api_mechanical_ventilation_kind. Natural certs (code 0/5 → None) unchanged
(corpus 73.3%/0.774, mapper corpus green, 726605 SAP 68.058 unchanged).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Ports the §6.1 conservatory split (declare the four glazed fields; exclude the
glazed BP from the fabric loop; carry it as sap_conservatory) to the remaining
gov mappers that lacked it. Fixes the modelling_e2e construction_age_band
NOT-NULL failures: 5 type-4 conservatory certs on RdSAP-Schema-20.0.0 (e.g.
property 725213) mapped the conservatory to a null-age fabric part that crashed
on persist. 17.0/17.1/18.0 ported preventively (no-op for non-conservatory
certs). 21.0.0 deferred — it never adopted the ADR-0027 all-Optional
SapBuildingPart, so a sparse conservatory part needs that broader change first.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per review (Khalim): lift the full-SAP detection out of the SAP-Schema-16.x
branch into a single top-level gate in from_api_response, and key it on the
gov-API's own assessment_type declaration rather than the structural
sap_opening_types proxy.
- _is_full_sap_assessment(data): assessment_type == "SAP" — the authoritative
SAP-vs-RdSAP classification. Verified to separate the entire fixture corpus:
every full-SAP schema (SAP-Schema-17.x/18.x + the broken LIG 16.x) is "SAP";
every reduced cert (RdSAP-Schema-* and reduced SAP-Schema-16.x, incl.
sap_16_0.json) is "RdSAP". data_type / sap_opening_types agree but are derived
shape artifacts; assessment_type is the meaning.
- A cert that is full-SAP by assessment_type but whose schema_type LABEL is not a
recognised full-SAP schema is a *broken schema type* (label disagrees with the
assessment). _record_broken_schema_type logs it — visible, not silently
rerouted — so unreliable labels surface and coverage grows as new mislabels
appear. Generalises beyond 16.x to any future mislabel.
- _from_full_sap maps it via the full-SAP 17.1 mapper (real measured openings,
no defaulting; only `tenure` defaulted). Correctly-labelled full-SAP certs keep
their dedicated branches (one-mapper-per-schema convention); reduced certs are
unchanged.
Tests: broken cert routed AND recorded; correctly-labelled full-SAP not recorded.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A full-SAP cert (assessment_type=SAP, the as-designed LIG-* new-builds) lodged
under a SAP-Schema-16.x version failed to map with:
ValueError: RdSapSchema17_1: missing required field 'door_count'
— crashing the property's modelling_e2e subtask. 89 distinct properties in the
2026-06-24 run hit this; ~22 of every 30 sampled are this full-SAP shape.
Root cause: from_api_response dispatches on the schema_type STRING only, and the
SAP-Schema-16.x branch assumed a single shape — "reduced-field (RdSAP-shaped)" —
sending every 16.x cert through _normalize_sap_schema_16_x -> RdSapSchema17_1.
But the SAP-Schema-16.x name covers two structurally different certs:
* reduced RdSAP (assessment_type=RdSAP): top-level door_count / glazed_area
band / construction-code building parts.
* full SAP (assessment_type=SAP): measured shape — sap_opening_types +
structured sap_building_parts carrying measured U-values and door/window
OPENINGS, with NO top-level door_count (the door is an opening). These omit
the reduced-only count fields, so the reduced normaliser failed loud on the
first one it checks (door_count) — and 14 others behind it.
Force-fitting a full-SAP cert to the reduced RdSAP schema was the bug: the data
was never missing, it was being validated against the wrong schema.
Fix: discriminate on shape (_is_full_sap_cert: assessment_type=SAP AND
sap_opening_types present) and route full-SAP 16.x certs to the existing
full-SAP 17.1 mapper, which reads the real measured openings (the lodged
0.9x2.1m door -> door_count 1) — no reduced-field defaulting. The only field the
16.x full-SAP shape omits that SapSchema17_1 needs is `tenure` (register
metadata, no SAP effect), defaulted. Reduced 16.x certs (assessment_type=RdSAP,
no opening types) are untouched — all pinned reduced 16.x fixtures stay on the
RdSAP path.
Regression test pins a real full-SAP-16.0 cert (0293-...-8795, lodged 83) mapping
via the full-SAP path + the reduced 16.2 cert staying on the RdSAP path.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Mirror the merged 21.0.1 fix (d501535c): declare the four glazed conservatory
fields on the 19.0 SapBuildingPart so from_dict stops dropping them, exclude the
glazed BP from the fabric loop, and carry it as EpcPropertyData.sap_conservatory
(§6.1). Fixes cert 718138 (conservatory_type=4) mis-scoring as a fabric part and
failing to persist on the NOT-NULL construction_age_band.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
