pdf json to RdSapSiteNotes full extract 🟩

backend/documents_parser/extractor.py

@@ -82,7 +82,22 @@ class PasHubRdSapSiteNotesExtractor:
     # --- public extract methods ---
 
     def extract(self) -> PasHubRdSapSiteNotes:
-        raise NotImplementedError
+        return PasHubRdSapSiteNotes(
+            inspection_metadata=None,
+            general=self.extract_general(),
+            building_construction=self.extract_building_construction(),
+            building_measurements=self.extract_building_measurements(),
+            roof_space=self.extract_roof_space(),
+            windows=self.extract_windows(),
+            heating_and_hot_water=self.extract_heating_and_hot_water(),
+            ventilation=self.extract_ventilation(),
+            conservatories=self.extract_conservatories(),
+            renewables=self.extract_renewables(),
+            room_count_elements=self.extract_room_count_elements(),
+            water_use=self.extract_water_use(),
+            customer_response=self.extract_customer_response(),
+            addendum=self.extract_addendum(),
+        )
 
     def extract_general(self) -> General:
         inspection_date_raw = self._get("Inspection Date:")

datatypes/epc/surveys/pashub_rdsap_site_notes.py

@@ -275,7 +275,7 @@ class SurveyAddendum:
 
 @dataclass
 class PasHubRdSapSiteNotes:
-    inspection_metadata: InspectionMetadata
+    inspection_metadata: Optional[InspectionMetadata]
     general: General
     building_construction: BuildingConstruction
     building_measurements: BuildingMeasurements