From 648fff29afb18112da1448a3730467bf41484df3 Mon Sep 17 00:00:00 2001
From: Daniel Roth <daniel_roth@hotmail.co.uk>
Date: Tue, 14 Apr 2026 16:37:49 +0000
Subject: [PATCH] parse to strongly typed dict. will map to excel row later

---
 .../ecmk_fetcher/tests/test_xml_processor.py  | 40 ++++++++++++-------
 backend/ecmk_fetcher/xml_processor.py         | 31 +++++++++++++-
 2 files changed, 54 insertions(+), 17 deletions(-)

diff --git a/backend/ecmk_fetcher/tests/test_xml_processor.py b/backend/ecmk_fetcher/tests/test_xml_processor.py
index 10381198..7e19181b 100644
--- a/backend/ecmk_fetcher/tests/test_xml_processor.py
+++ b/backend/ecmk_fetcher/tests/test_xml_processor.py
@@ -49,7 +49,7 @@ SAMPLE_XML = """<RdSAP-Report xmlns="https://epbr.digital.communities.gov.uk/xsd
 """
 
 
-def test_parse_rdsap_wide_flat_contract():
+def test_parse_rdsap_contract():
     # arrange + act
     result = parse_rdsap(SAMPLE_XML)
 
@@ -57,18 +57,28 @@ def test_parse_rdsap_wide_flat_contract():
     assert result == {
         "address": "1, Fake Avenue, Random, AB24 5CD",
         "property_type": "House",
-        # Main Dwelling - floor 0
-        "main_dwelling_floor_index_0_area_m2": 43.61,
-        "main_dwelling_floor_index_0_height_m": 2.46,
-        "main_dwelling_floor_index_0_heat_loss_perimeter_m": 25.31,
-        "main_dwelling_floor_index_0_party_wall_length_m": 0.0,
-        # Main Dwelling - floor 1
-        "main_dwelling_floor_index_1_area_m2": 42.33,
-        "main_dwelling_floor_index_1_height_m": 2.44,
-        "main_dwelling_floor_index_1_heat_loss_perimeter_m": 26.16,
-        "main_dwelling_floor_index_1_party_wall_length_m": 0.0,
-        # Roof (building-level, repeated across floors or stored once)
-        "main_dwelling_roof_construction": "4",
-        "main_dwelling_roof_insulation_location": "2",
-        "main_dwelling_roof_insulation_thickness": "100mm",
+        "building_parts": [
+            {
+                "identifier": "Main Dwelling",
+                "floors": [
+                    {
+                        "area_m2": 43.61,
+                        "height_m": 2.46,
+                        "heat_loss_perimeter_m": 25.31,
+                        "party_wall_length_m": 0.0,
+                    },
+                    {
+                        "area_m2": 42.33,
+                        "height_m": 2.44,
+                        "heat_loss_perimeter_m": 26.16,
+                        "party_wall_length_m": 0.0,
+                    },
+                ],
+                "roof": {
+                    "construction": 4,
+                    "insulation_location": 2,
+                    "insulation_thickness_mm": 100.0,
+                },
+            }
+        ],
     }
diff --git a/backend/ecmk_fetcher/xml_processor.py b/backend/ecmk_fetcher/xml_processor.py
index 5a392d38..03f435f5 100644
--- a/backend/ecmk_fetcher/xml_processor.py
+++ b/backend/ecmk_fetcher/xml_processor.py
@@ -1,8 +1,35 @@
-from typing import Any
+from typing import Any, List, Optional, TypedDict
 
 from etl.xml_survey_extraction.XmlParser import PROPERTY_TYPE_LOOKUP
 
 
+class Floor(TypedDict):
+    area_m2: float
+    height_m: float
+    heat_loss_perimeter_m: float
+    party_wall_length_m: float
+
+
+class Roof(TypedDict, total=False):
+    construction: int  # TODO: map to str
+    insulation_location: int  # TODO: map to str
+    insulation_thickness_mm: float
+
+
+class BuildingPart(TypedDict):
+    identifier: str  # e.g. "Main Dwelling", "Extension"
+    floors: List[Floor]
+    roof: Optional[Roof]
+
+
+class SapPropertyDetails(TypedDict):
+    address: str
+    property_type: str
+    building_parts: List[BuildingPart]
+
+
 # This file should ultimately live somewhere different, probably
-def parse_rdsap(xml_string: str) -> Any:  # TODO: define shape of return object
+def parse_rdsap(
+    xml_string: str,
+) -> SapPropertyDetails:
     raise NotImplementedError