Extract Elmhurst site notes to dataclass 🟥

This commit is contained in:
Daniel Roth 2026-04-24 13:13:24 +00:00
parent a8579db4d9
commit 1a53a8d83e

View file

@ -0,0 +1,433 @@
import json
import os
from datetime import date
import pytest
from backend.documents_parser.elmhurst_extractor import ElmhurstSiteNotesExtractor
from datatypes.epc.surveys.elmhurst_site_notes import (
BathsAndShowers,
BuildingPartDimensions,
ElmhurstSiteNotes,
FloorDetails,
FloorDimension,
Lighting,
MainHeating,
Meters,
PropertyDetails,
Renewables,
RoofDetails,
Shower,
SurveyorInfo,
VentilationAndCooling,
WallDetails,
WaterHeating,
Window,
)
FIXTURE_PATH = os.path.join(
os.path.dirname(__file__), "fixtures", "elmhurst_site_notes_1_text.json"
)
@pytest.fixture(scope="module")
def result() -> ElmhurstSiteNotes:
with open(FIXTURE_PATH) as f:
pages = json.load(f)
return ElmhurstSiteNotesExtractor(pages).extract()
class TestSurveyorInfo:
def test_surveyor_code(self, result: ElmhurstSiteNotes) -> None:
assert result.surveyor_info.surveyor_code == "P960-0001"
def test_name(self, result: ElmhurstSiteNotes) -> None:
assert result.surveyor_info.name == "Richard Matthew Ratcliff"
def test_title(self, result: ElmhurstSiteNotes) -> None:
assert result.surveyor_info.title == "Mr."
def test_tel_number(self, result: ElmhurstSiteNotes) -> None:
assert result.surveyor_info.tel_number == "07760 443 469"
def test_survey_reference(self, result: ElmhurstSiteNotes) -> None:
assert result.surveyor_info.survey_reference == "001573"
def test_my_reference_none(self, result: ElmhurstSiteNotes) -> None:
assert result.surveyor_info.my_reference is None
class TestPropertyDetails:
def test_rdsap_version(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.rdsap_version == "RdSAP10"
def test_reference_number(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.reference_number == "P960-0001-001573"
def test_lodgement_required(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.lodgement_required is False
def test_regs_region(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.regs_region == "England"
def test_epc_language(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.epc_language == "English"
def test_uprn_none(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.uprn is None
def test_postcode(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.postcode == "BB10 1XX"
def test_region(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.region == "West Pennines"
def test_house_name_none(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.house_name is None
def test_house_number(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.house_number == "19"
def test_street(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.street == "Queens Road"
def test_locality_none(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.locality is None
def test_town(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.town == "BURNLEY"
def test_county_none(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.county is None
def test_tenure(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.tenure == "Rented (social)"
def test_transaction_type(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.transaction_type == "Grant scheme"
def test_inspection_date(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.inspection_date == date(2026, 3, 6)
def test_process_date(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.process_date == date(2026, 3, 6)
def test_epc_exists(self, result: ElmhurstSiteNotes) -> None:
assert result.property_details.epc_exists is False
class TestPropertyDescription:
def test_property_type(self, result: ElmhurstSiteNotes) -> None:
assert result.property_type == "B Bungalow"
def test_attachment(self, result: ElmhurstSiteNotes) -> None:
assert result.attachment == "E End-Terrace"
def test_number_of_storeys(self, result: ElmhurstSiteNotes) -> None:
assert result.number_of_storeys == 1
def test_habitable_rooms(self, result: ElmhurstSiteNotes) -> None:
assert result.habitable_rooms == 2
def test_heated_habitable_rooms(self, result: ElmhurstSiteNotes) -> None:
assert result.heated_habitable_rooms == 2
def test_construction_age_band(self, result: ElmhurstSiteNotes) -> None:
assert result.construction_age_band == "D 1950-1966"
def test_has_conservatory(self, result: ElmhurstSiteNotes) -> None:
assert result.has_conservatory is False
class TestDimensions:
def test_dimension_type(self, result: ElmhurstSiteNotes) -> None:
assert result.dimensions.dimension_type == "Internal"
def test_floor_count(self, result: ElmhurstSiteNotes) -> None:
assert len(result.dimensions.floors) == 1
def test_floor_name(self, result: ElmhurstSiteNotes) -> None:
assert result.dimensions.floors[0].name == "Lowest Floor"
def test_floor_area(self, result: ElmhurstSiteNotes) -> None:
assert result.dimensions.floors[0].area_m2 == 44.89
def test_floor_room_height(self, result: ElmhurstSiteNotes) -> None:
assert result.dimensions.floors[0].room_height_m == 2.24
def test_floor_heat_loss_perimeter(self, result: ElmhurstSiteNotes) -> None:
assert result.dimensions.floors[0].heat_loss_perimeter_m == 20.10
def test_floor_party_wall_length(self, result: ElmhurstSiteNotes) -> None:
assert result.dimensions.floors[0].party_wall_length_m == 6.70
class TestWalls:
def test_wall_type(self, result: ElmhurstSiteNotes) -> None:
assert result.walls.wall_type == "CA Cavity"
def test_insulation(self, result: ElmhurstSiteNotes) -> None:
assert result.walls.insulation == "F Filled Cavity"
def test_thickness_unknown(self, result: ElmhurstSiteNotes) -> None:
assert result.walls.thickness_unknown is False
def test_thickness_mm(self, result: ElmhurstSiteNotes) -> None:
assert result.walls.thickness_mm == 300
def test_u_value_known(self, result: ElmhurstSiteNotes) -> None:
assert result.walls.u_value_known is False
def test_party_wall_type(self, result: ElmhurstSiteNotes) -> None:
assert result.walls.party_wall_type == "U Unable to determine"
class TestRoof:
def test_roof_type(self, result: ElmhurstSiteNotes) -> None:
assert result.roof.roof_type == "PA Pitched (slates/tiles), access to loft"
def test_insulation(self, result: ElmhurstSiteNotes) -> None:
assert result.roof.insulation == "J Joists"
def test_insulation_thickness_mm(self, result: ElmhurstSiteNotes) -> None:
assert result.roof.insulation_thickness_mm == 270
def test_u_value_known(self, result: ElmhurstSiteNotes) -> None:
assert result.roof.u_value_known is False
class TestFloor:
def test_location(self, result: ElmhurstSiteNotes) -> None:
assert result.floor.location == "G Ground floor"
def test_floor_type(self, result: ElmhurstSiteNotes) -> None:
assert result.floor.floor_type == "N Suspended, not timber"
def test_insulation(self, result: ElmhurstSiteNotes) -> None:
assert result.floor.insulation == "A As built"
def test_default_u_value(self, result: ElmhurstSiteNotes) -> None:
assert result.floor.default_u_value == 0.69
def test_u_value_known(self, result: ElmhurstSiteNotes) -> None:
assert result.floor.u_value_known is False
class TestDoors:
def test_door_count(self, result: ElmhurstSiteNotes) -> None:
assert result.door_count == 0
def test_insulated_door_count(self, result: ElmhurstSiteNotes) -> None:
assert result.insulated_door_count == 0
class TestWindows:
def test_window_count(self, result: ElmhurstSiteNotes) -> None:
assert len(result.windows) == 4
def test_draught_proofing_percent(self, result: ElmhurstSiteNotes) -> None:
assert result.draught_proofing_percent == 100
def test_first_window_dimensions(self, result: ElmhurstSiteNotes) -> None:
w = result.windows[0]
assert w.width_m == 1.30
assert w.height_m == 1.10
assert w.area_m2 == 1.43
def test_first_window_glazing(self, result: ElmhurstSiteNotes) -> None:
w = result.windows[0]
assert w.glazing_type == "Double post or during 2022"
assert w.frame_factor == 0.70
def test_first_window_location(self, result: ElmhurstSiteNotes) -> None:
w = result.windows[0]
assert w.building_part == "Main"
assert w.location == "External wall"
assert w.orientation == "North"
def test_first_window_performance(self, result: ElmhurstSiteNotes) -> None:
w = result.windows[0]
assert w.data_source == "Manufacturer"
assert w.u_value == 1.40
assert w.g_value == 0.72
assert w.draught_proofed is True
assert w.permanent_shutters == "None"
def test_third_window_orientation(self, result: ElmhurstSiteNotes) -> None:
assert result.windows[2].orientation == "South"
def test_fourth_window_dimensions(self, result: ElmhurstSiteNotes) -> None:
w = result.windows[3]
assert w.width_m == 0.70
assert w.height_m == 1.30
assert w.area_m2 == 0.91
class TestVentilation:
def test_open_chimneys(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.open_chimneys_count == 0
def test_open_flues(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.open_flues_count == 0
def test_open_chimneys_closed_fire(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.open_chimneys_closed_fire_count == 0
def test_solid_fuel_boiler_flues(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.solid_fuel_boiler_flues_count == 0
def test_other_heater_flues(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.other_heater_flues_count == 0
def test_blocked_chimneys(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.blocked_chimneys_count == 0
def test_extract_fans(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.extract_fans_count == 2
def test_passive_vents(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.passive_vents_count == 0
def test_flueless_gas_fires(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.flueless_gas_fires_count == 0
def test_fixed_space_cooling(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.fixed_space_cooling is False
def test_draught_lobby(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.draught_lobby == "Not present"
def test_mechanical_ventilation(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.mechanical_ventilation is False
def test_pressure_test_method(self, result: ElmhurstSiteNotes) -> None:
assert result.ventilation.pressure_test_method == "Not available"
class TestLighting:
def test_total_bulbs(self, result: ElmhurstSiteNotes) -> None:
assert result.lighting.total_bulbs == 8
def test_led_cfl_count_known(self, result: ElmhurstSiteNotes) -> None:
assert result.lighting.led_cfl_count_known is True
def test_led_count(self, result: ElmhurstSiteNotes) -> None:
assert result.lighting.led_count == 4
def test_cfl_count(self, result: ElmhurstSiteNotes) -> None:
assert result.lighting.cfl_count == 4
def test_incandescent_count(self, result: ElmhurstSiteNotes) -> None:
assert result.lighting.incandescent_count == 0
class TestMainHeating:
def test_pcdf_boiler_reference(self, result: ElmhurstSiteNotes) -> None:
assert (
result.main_heating.pcdf_boiler_reference
== "17742 Potterton, Promax 33 Combi ErP, 88.30%"
)
def test_heat_emitter(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.heat_emitter == "Radiators"
def test_heat_pump_age(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.heat_pump_age == "Unknown"
def test_fuel_type(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.fuel_type == "Mains gas"
def test_flue_type(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.flue_type == "Balanced"
def test_fan_assisted_flue(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.fan_assisted_flue is True
def test_design_flow_temperature(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.design_flow_temperature == "Unknown"
def test_heating_controls_ees(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.heating_controls_ees == "CBE"
def test_heating_controls_sap(self, result: ElmhurstSiteNotes) -> None:
assert (
result.main_heating.heating_controls_sap
== "SAP code 2106, Programmer, room thermostat and TRVs"
)
def test_percentage_of_heat(self, result: ElmhurstSiteNotes) -> None:
assert result.main_heating.percentage_of_heat == 100
class TestMeters:
def test_electricity_meter_type(self, result: ElmhurstSiteNotes) -> None:
assert result.meters.electricity_meter_type == "Single"
def test_main_gas(self, result: ElmhurstSiteNotes) -> None:
assert result.meters.main_gas is True
def test_electricity_smart_meter(self, result: ElmhurstSiteNotes) -> None:
assert result.meters.electricity_smart_meter is False
def test_gas_smart_meter(self, result: ElmhurstSiteNotes) -> None:
assert result.meters.gas_smart_meter is False
class TestWaterHeating:
def test_water_heating_code(self, result: ElmhurstSiteNotes) -> None:
assert result.water_heating.water_heating_code == "HWP"
def test_water_heating_sap_code(self, result: ElmhurstSiteNotes) -> None:
assert result.water_heating.water_heating_sap_code == 901
def test_water_heating_fuel_type(self, result: ElmhurstSiteNotes) -> None:
assert result.water_heating.water_heating_fuel_type == "Mains gas"
def test_hot_water_cylinder_present(self, result: ElmhurstSiteNotes) -> None:
assert result.water_heating.hot_water_cylinder_present is False
class TestBathsAndShowers:
def test_number_of_baths(self, result: ElmhurstSiteNotes) -> None:
assert result.baths_and_showers.number_of_baths == 0
def test_number_of_baths_connected(self, result: ElmhurstSiteNotes) -> None:
assert result.baths_and_showers.number_of_baths_connected == 0
def test_shower_count(self, result: ElmhurstSiteNotes) -> None:
assert len(result.baths_and_showers.showers) == 1
def test_shower_number(self, result: ElmhurstSiteNotes) -> None:
assert result.baths_and_showers.showers[0].shower_number == 1
def test_shower_outlet_type(self, result: ElmhurstSiteNotes) -> None:
assert result.baths_and_showers.showers[0].outlet_type == "Electric shower"
def test_shower_connected(self, result: ElmhurstSiteNotes) -> None:
assert result.baths_and_showers.showers[0].connected == "None"
class TestRenewables:
def test_solar_water_heating(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.solar_water_heating is False
def test_wwhrs_present(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.wwhrs_present is False
def test_flue_gas_heat_recovery_present(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.flue_gas_heat_recovery_present is False
def test_photovoltaic_panel(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.photovoltaic_panel == "None"
def test_export_capable_meter(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.export_capable_meter is False
def test_wind_turbine_present(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.wind_turbine_present is False
def test_wind_turbines_terrain_type(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.wind_turbines_terrain_type == "Suburban"
def test_hydro_electricity_generated_kwh(self, result: ElmhurstSiteNotes) -> None:
assert result.renewables.hydro_electricity_generated_kwh == 0.0