Load BuildingConstruction from SiteNotes JSON 🟩

backend/documents_parser/extractor.py

@@ -1,7 +1,12 @@
 from datetime import datetime
+from typing import List, Optional
 
 from datatypes.epc.surveys.pashub_rdsap_site_notes import (
+    BuildingConstruction,
+    ExtensionConstruction,
+    FloorConstruction,
     General,
+    MainBuildingConstruction,
     PasHubRdSapSiteNotes,
 )
 
@@ -10,7 +15,9 @@ class PasHubRdSapSiteNotesExtractor:
     def __init__(self, text_list: list[str]) -> None:
         self.text_list = text_list
 
-    def _get(self, key: str, offset: int = 1) -> str | None:
+    # --- generic helpers ---
+
+    def _get(self, key: str, offset: int = 1) -> Optional[str]:
         try:
             idx = self.text_list.index(key)
             return self.text_list[idx + offset].strip() or None
@@ -21,6 +28,35 @@ class PasHubRdSapSiteNotesExtractor:
         val = self._get(key, offset)
         return val is not None and val.lower() == "yes"
 
+    def _get_in(self, lst: List[str], key: str, offset: int = 1) -> Optional[str]:
+        try:
+            idx = lst.index(key)
+            return lst[idx + offset].strip() or None
+        except (ValueError, IndexError):
+            return None
+
+    def _bool_in(self, lst: List[str], key: str) -> bool:
+        val = self._get_in(lst, key)
+        return val is not None and val.lower() == "yes"
+
+    def _is_known_in(self, lst: List[str], key: str) -> bool:
+        val = self._get_in(lst, key)
+        return val is not None and val.lower() != "not known"
+
+    def _wall_thickness_in(self, lst: List[str]) -> int:
+        val = self._get_in(lst, "Wall thickness:")
+        return int(val.split()[0]) if val else 0
+
+    def _section(self, start: str, end: str) -> List[str]:
+        try:
+            start_idx = self.text_list.index(start)
+            end_idx = self.text_list.index(end, start_idx)
+            return self.text_list[start_idx:end_idx]
+        except ValueError:
+            return []
+
+    # --- public extract methods ---
+
     def extract(self) -> PasHubRdSapSiteNotes:
         raise NotImplementedError
 
@@ -58,3 +94,99 @@ class PasHubRdSapSiteNotesExtractor:
             gas_meter_accessible=self._bool("Is the gas meter accessible?"),
             measurements_location=self._get("Select Measurements Location:") or "",
         )
+
+    def extract_building_construction(self) -> BuildingConstruction:
+        bc_section = self._section("Building Construction", "Building Measurements")
+
+        # Find extension markers within this section
+        extension_markers = []
+        i = 1
+        while f"Extension {i}" in bc_section:
+            extension_markers.append(f"Extension {i}")
+            i += 1
+
+        # Slice main building data: from "Main Building" to first extension or end
+        main_start = bc_section.index("Main Building")
+        main_end = (
+            bc_section.index(extension_markers[0])
+            if extension_markers
+            else len(bc_section)
+        )
+        main_data = bc_section[main_start:main_end]
+
+        # Slice each extension's data
+        extensions = []
+        for n, marker in enumerate(extension_markers):
+            ext_start = bc_section.index(marker)
+            ext_end = (
+                bc_section.index(extension_markers[n + 1])
+                if n + 1 < len(extension_markers)
+                else len(bc_section)
+            )
+            ext_data = bc_section[ext_start:ext_end]
+            extensions.append(self._parse_extension_construction(n + 1, ext_data))
+
+        return BuildingConstruction(
+            main_building=self._parse_main_building_construction(main_data),
+            floor=self._parse_floor_construction(main_data),
+            extensions=extensions if extensions else None,
+        )
+
+    # --- private parsing helpers ---
+
+    def _parse_main_building_construction(
+        self, data: List[str]
+    ) -> MainBuildingConstruction:
+        return MainBuildingConstruction(
+            age_range=self._get_in(data, "Age Range:") or "",
+            age_indicators=self._get_in(data, "Record indicators of property age:") or "",
+            walls_construction_type=self._get_in(data, "Walls - Construction Type:") or "",
+            cavity_construction_indicators=self._get_in(
+                data, "Record external indicators of Cavity Construction:"
+            ) or "",
+            walls_insulation_type=self._get_in(data, "Walls - Insulation Type:") or "",
+            filled_cavity_indicators=self._get_in(
+                data, "Record indicators of filled cavity:"
+            ),
+            thermal_conductivity_of_wall_insulation=self._get_in(
+                data, "Thermal conductivity of wall insulation:"
+            ) or "",
+            wall_u_value_known=self._is_known_in(data, "Wall U-Value known?"),
+            wall_thickness_mm=self._wall_thickness_in(data),
+            party_wall_construction_type=self._get_in(
+                data, "Party wall construction type:"
+            ) or "",
+        )
+
+    def _parse_extension_construction(
+        self, ext_id: int, data: List[str]
+    ) -> ExtensionConstruction:
+        return ExtensionConstruction(
+            id=ext_id,
+            age_range=self._get_in(data, "Age Range:") or "",
+            age_indicators=self._get_in(data, "Record indicators of property age:") or "",
+            walls_construction_type=self._get_in(data, "Walls - Construction Type:") or "",
+            cavity_construction_indicators=self._get_in(
+                data, "Record external indicators of Cavity Construction:"
+            ) or "",
+            walls_insulation_type=self._get_in(data, "Walls - Insulation Type:") or "",
+            filled_cavity_indicators=self._get_in(
+                data, "Record indicators of filled cavity:"
+            ),
+            thermal_conductivity_of_wall_insulation=self._get_in(
+                data, "Thermal conductivity of wall insulation:"
+            ) or "",
+            wall_u_value_known=self._is_known_in(data, "Wall U-Value known?"),
+            wall_thickness_mm=self._wall_thickness_in(data),
+            party_wall_construction_type=self._get_in(
+                data, "Party wall construction type:"
+            ) or "",
+        )
+
+    def _parse_floor_construction(self, data: List[str]) -> FloorConstruction:
+        return FloorConstruction(
+            floor_type=self._get_in(data, "Floor type:") or "",
+            floor_construction=self._get_in(data, "Floor Construction:") or "",
+            floor_insulation_type=self._get_in(data, "Floor Insulation Type:") or "",
+            floor_u_value_known=self._is_known_in(data, "Floor U-Value known?"),
+        )