Map to RdSapSiteNotes from site notes JSON 🟩

backend/documents_parser/extractor.py

@@ -1,9 +1,60 @@
-from datatypes.epc.surveys.pashub_rdsap_site_notes import PasHubRdSapSiteNotes
+from datetime import datetime
+
+from datatypes.epc.surveys.pashub_rdsap_site_notes import (
+    General,
+    PasHubRdSapSiteNotes,
+)
 
 
 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:
+        try:
+            idx = self.text_list.index(key)
+            return self.text_list[idx + offset].strip() or None
+        except (ValueError, IndexError):
+            return None
+
+    def _bool(self, key: str, offset: int = 1) -> bool:
+        val = self._get(key, offset)
+        return val is not None and val.lower() == "yes"
+
     def extract(self) -> PasHubRdSapSiteNotes:
         raise NotImplementedError
+
+    def extract_general(self) -> General:
+        inspection_date_raw = self._get("Inspection Date:")
+        inspection_date = (
+            datetime.strptime(inspection_date_raw, "%d/%m/%Y").strftime("%Y-%m-%d")
+            if inspection_date_raw
+            else ""
+        )
+
+        storeys_raw = self._get("Number of storeys:") or "0"
+        extensions_raw = self._get("Number of Extensions:") or "0"
+
+        return General(
+            epc_checked_before_assessment=self._bool(
+                "Confirm you have checked for the existence of an", offset=2
+            ),
+            epc_exists_at_point_of_assessment=self._bool(
+                "Does an EPC exist at the point of carrying out this", offset=2
+            ),
+            inspection_date=inspection_date,
+            transaction_type=self._get("Transaction Type:") or "",
+            tenure=self._get("Tenure:") or "",
+            property_type=self._get("Type of Property:") or "",
+            detachment_type=self._get("Detachment Type:") or "",
+            number_of_storeys=int(storeys_raw.split()[0]),
+            terrain_type=self._get("Terrain Type:") or "",
+            number_of_extensions=int(extensions_raw.split()[0]),
+            electricity_smart_meter=self._bool("Is an electricity smart meter present?"),
+            electric_meter_type=self._get("Electric meter type:") or "",
+            dwelling_export_capable=self._bool("Is the dwelling export-capable?"),
+            mains_gas_available=self._bool("Is mains gas available?"),
+            gas_smart_meter=self._bool("Is there a gas smart meter?"),
+            gas_meter_accessible=self._bool("Is the gas meter accessible?"),
+            measurements_location=self._get("Select Measurements Location:") or "",
+        )

backend/documents_parser/tests/test_extractor.py

@@ -17,7 +17,7 @@ def load_text_fixture() -> list[str]:
 class TestGeneral:
     @pytest.fixture
     def general(self) -> General:
-        return PasHubRdSapSiteNotesExtractor(load_text_fixture()).extract().general
+        return PasHubRdSapSiteNotesExtractor(load_text_fixture()).extract_general()
 
     def test_epc_checked_before_assessment(self, general: General) -> None:
         assert general.epc_checked_before_assessment is True