improving basic typing of EpcRecord

etl/epc/Record.py

@@ -1,4 +1,4 @@
-from typing import Optional, get_origin, get_args, TypedDict, Dict
+from typing import Optional, get_origin, get_args, TypedDict, cast, TypeAlias
 from dataclasses import fields
 from datetime import datetime
 from dataclasses import dataclass
@@ -45,11 +45,15 @@ DATA_BUCKET = os.environ.get(
 
 pd.set_option("future.no_silent_downcasting", True)
 
+RawEpcRow: TypeAlias = dict[str, str | None]
+PreparedEpcValue: TypeAlias = str | int | float | bool | None
+PreparedEpcRow: TypeAlias = dict[str, PreparedEpcValue]
+
 
 class InputEpcRecords(TypedDict):
-    original_epc: Dict[str, Any]
+    original_epc: RawEpcRow
-    full_sap_epc: Dict[str, Any]
+    full_sap_epc: RawEpcRow
-    old_data: List[Dict[str, Any]]
+    old_data: list[RawEpcRow]
 
 
 @dataclass
@@ -231,22 +235,33 @@ class EPCRecord:
 
     run_mode: str = "training"
 
+    # ------------------------------------------------------------------
+    # INPUT DATA STRUCTURES
+    # ------------------------------------------------------------------
+
     epc_records: Optional[InputEpcRecords] = None
-    full_sap_epc: Optional[dict] = None
+    # Raw EPC input (immutable)
-    old_data: list[dict] = None
+    original_epc: Optional[RawEpcRow] = None
-    original_epc: Optional[dict] = None
+
-    prepared_epc: Optional[dict] = None
+    # Working dictionary that gets cleaned
+    prepared_epc: Optional[PreparedEpcRow] = None
+
+    # Supporting
+    full_sap_epc: Optional[RawEpcRow] = None
+    old_data: Optional[list[RawEpcRow]] = None
+
+    # # Metadata generated during processing
     prepared_epc_delta_metadata: pd.DataFrame = None
     cleaning_data: pd.DataFrame = None
 
     # Not used in training mod but used in newdata mode
-    age_band: str = None
+    age_band: Optional[str] = None
-    construction_age_band: str = None
+    construction_age_band: Optional[str] = None
-    year_built: int = None
+    year_built: Optional[int] = None
-    number_of_floors: int = None
+    number_of_floors: Optional[int] = None
-    number_of_open_fireplaces: int = None
+    number_of_open_fireplaces: Optional[int] = None
-    heat_loss_corridor_bool: bool = None
+    heat_loss_corridor_bool: Optional[bool] = None
-    solar_water_heating_flag_bool: bool = None
+    solar_water_heating_flag_bool: Optional[bool] = None
 
     def __post_init__(self):
         # We can have validation and cleaning steps for each of the fields
@@ -255,15 +270,18 @@ class EPCRecord:
 
         if self.run_mode == "training":
             self.validation_configuration = EPCRecordValidationConfiguration
-            # self._field_validation()
             return
 
         # We are running in newdata mode
         if self.epc_records is None:
             raise ValueError("Must provide epc records if running in newdata mode")
 
-        self.prepared_epc = self.epc_records["original_epc"]
+        # Immutable copy; raw record
         self.original_epc = self.epc_records["original_epc"].copy()
+
+        # Working copy that we will clean and manipulate
+        self.prepared_epc = self.epc_records["original_epc"].copy()
+
         self.full_sap_epc = self.epc_records["full_sap_epc"]
         self.old_data = self.epc_records["old_data"]
 
@@ -299,9 +317,12 @@ class EPCRecord:
         )
         epc_data_processor.prepare_data()
 
-        self.prepared_epc = epc_data_processor.data.to_dict(orient="records")[0]
+        record = epc_data_processor.data.to_dict(orient="records")[0]
 
-    def _cast_value(self, value, type_hint):
+        self.prepared_epc = cast(RawEpcRow, record)
+
+    @staticmethod
+    def _cast_value(value, type_hint):
 
         origin = get_origin(type_hint)
         args = get_args(type_hint)
@@ -396,14 +417,6 @@ class EPCRecord:
         self._clean_constituency()
         self._clean_new_build_descriptions()
 
-        # self._clean_potential_energy_efficiency()
-        # self._clean_environment_impact_potential()
-        # self._clean_energy_consumption_potential()
-        # self._clean_co2_emissions_potential()
-        # self._clean_current_energy_efficiency()
-        # self._clean_energy_consumption_current()
-        # self._clean_co2_emissions_current()
-
     def epc_record_as_dataframe(
         self,
         epc_type: str = "prepared_epc",
@@ -524,9 +537,7 @@ class EPCRecord:
                     cleaned_property_data["FIXED_LIGHTING_OUTLETS_COUNT"].values[0]
                 )
         else:
-            self.prepared_epc["fixed-lighting-outlets-count"] = float(
+            self.prepared_epc["fixed-lighting-outlets-count"] = float(self.prepared_epc["fixed-lighting-outlets-count"])
-                self.prepared_epc["fixed-lighting-outlets-count"]
-            )
 
     def _filter_property_dimensions(self, property_dimensions):
         """
@@ -604,15 +615,6 @@ class EPCRecord:
             self.prepared_epc["property-type"]
         )
 
-        # if self.prepared_epc["property-type"] == "House":
-        #     self.number_of_floors = 2
-        # elif self.prepared_epc["property-type"] in ["Flat", "Bungalow"]:
-        #     self.number_of_floors = 1
-        # elif self.prepared_epc["property-type"] == "Maisonette":
-        #     self.number_of_floors = 2
-        # else:
-        #     raise NotImplementedError("Implement me")
-
         if (
             self.prepared_epc["floor-height"] == ""
             or self.prepared_epc["floor-height"] in DATA_ANOMALY_MATCHES
@@ -859,9 +861,12 @@ class EPCRecord:
         This method will clean the year built, if empty or invalid
         """
         if self.full_sap_epc:
-            self.year_built = datetime.strptime(
+            lodgement_date = self.full_sap_epc["lodgement-date"]
-                self.full_sap_epc["lodgement-date"], "%Y-%m-%d"
+
-            ).year
+            if lodgement_date is None:
+                raise ValueError("full_sap_epc lodgement-date is missing")
+
+            self.year_built = datetime.strptime(str(lodgement_date), "%Y-%m-%d").year
 
             return