extraction finished now to do load

etl/condition_report_etl.py

@@ -1,3 +1,6 @@
 from etl.surveyedData.surveryedData import surveyedDataProcessor
 condition_report_file_path = "/workspaces/survey-extractor/etl/files/osmosis_condition_report.pdf"
-sdp = surveyedDataProcessor("123 Fake Street", [condition_report_file_path])
+sdp = surveyedDataProcessor("123 Fake Street", [condition_report_file_path])
+
+
+# TODO: add the ability to add document type, and sharepoint or s3 link so we can process access it again

etl/pdfReader/sitenotes.py

@@ -14,7 +14,7 @@ from etl.transform.conditionReportTypes import (
     ExternalElevationGableOne, ExternalElevationGableTwo, ExternalElevationRear, ConservatoryOrOutbuilding,
     AccessAndElevations, Hallway, RoomInfo, WindowsInfo, VentilationInfo, LivingRoom, DiningRoom, Kitchen, Rooms,
     Utility, WC, Landing, Bedroom, Bathroom, LoftSpace, RoomInRoof, HeatingSystem, GeneralConditionHeatingSystem,
-    MainHeatingOne, MainHeatingTwo, SecondaryHeating, HeatingByRoom, Renewables
+    MainHeatingOne, MainHeatingTwo, SecondaryHeating, HeatingByRoom, Renewables, Occupant, EnergyUse, Heating, ShowerAndBath, FridgeAndFreezers, Cooker, TumbleDryer
 )
 from datetime import datetime
 from pprint import pprint
@@ -99,6 +99,7 @@ class ConditionReport(SiteNotesExtractor):
         self.get_section_2()
         self.get_section_3()
         self.get_section_4()
+        self.get_section_5()
         self.get_section_0()
 
     def get_section_0(self):
@@ -562,8 +563,87 @@ class ConditionReport(SiteNotesExtractor):
             are_the_gas_and_electricity_meters_accessible=True if self.get_next_value(data, "Are the Gas and Electricity Meters accessible?") else False,
             dual_or_single_electric_meter=self.get_next_value(data, "Dual or single electric meter?"),
         )
+    
+    def get_section_5(self):
+        _ = self.get_occupants()
+        _ = self.get_energy_use()
+        _ = self.get_heating()
+        _ = self.get_shower_and_bath()
+        _ = self.get_appliances()
+        _ = self.get_fridge_and_freezers()
+        _ = self.get_cooker()
+        _ = self.get_tumble_dryer()
+        pprint(_)
+
+    def get_occupants(self):
+        data = self.get_data_between("Occupants", "Energy use")
+        second_data = self.get_data_between("Tumble dryer", "Media summary")
+        return Occupant(
+            name=self.get_next_value(second_data, "Name of the occupant:"),
+            have_evidence_of_12_months_of_fuel_bill_data= True if self.get_next_value(second_data, "Have you evidenced 12 months of fuel bill data?").lower() == "yes" else False,
+            total_number_of_occupants=int(self.get_next_value(data, "Total number of occupants:")),
+            no_of_adult_occupants=int(self.get_next_value(data, "No. of Adult Occupants (18+)")),
+            no_of_child_occupants=int(self.get_next_value(data, "No. of Child Occupants (Under 18)")),
+            no_of_occupant_of_a_pensionable_age=int(self.get_next_value(data, "No. of occupant of a pensionable age")),
+            are_there_any_vulnerable_people=True if self.get_next_value(data, "Are there any vulnerable people?").lower() == "yes" else False,
+            is_there_anyone_with_a_disability=True if self.get_next_value(data, "Is there anyone with a disability?").lower() == "yes" else False,
+            status_of_occupant=self.get_next_value(data, "Status of the occupant:"),
+            landlord_has_written_confirmation_that_the_tenent_agrees_to_the_assessment_been_supplied=True if self.get_next_value(data, "the assessment been supplied").lower() == "yes" else False,
+        )
+    
+    def get_energy_use(self):
+        data = self.get_data_between("Energy use", "Heating")
+        return EnergyUse(
+            property_tenure=self.get_next_value(data, "Property tenure:"),
+            who_is_the_electricity_payer=self.get_next_value(data, "Who is the electricity bill payer?")
+        )
+
+    def get_heating(self):
+        data = self.get_data_between("Heating", "Shower & bath")
+        return Heating(
+            room_stat_in_temperature_in_celsius=self.get_next_value(data, "Room Stat Temperature (in °C)", avoid=["Room Stat Location", '\xa0']),
+            room_stat_location=self.get_next_value(data, "Room Stat Location", avoid = ["Is the heating pattern known?", '\xa0']),
+            is_the_heating_pattern_known=self.get_next_value(data, "Is the heating pattern known?", avoid=["Shower & bath", '\xa0']),
+        )
+    
+    def get_shower_and_bath(self):
+        data = self.get_data_between("Shower & bath", "Appliances")
+        return ShowerAndBath(
+            shower_type=self.get_next_value(data, "Shower Type:"),
+            do_you_know_the_no_of_showers_per_day_per_week=True if self.get_next_value(data, "Do you know the number of showers per day or per week?").lower() == "yes" else False,
+            please_input_no_of_showers_and_specify_a_day_or_a_week=self.get_next_value(data, '"per week"'),
+            do_you_know_the_number_of_baths_per_day_or_per_week=self.get_next_value(data, "Do you know the number of baths per day or per week?"),
+        ) 
+    
+    def get_appliances(self):
+        print("Skipped appliances due to not having this example yet")
+
+    def get_fridge_and_freezers(self):
+        data = self.get_data_between("Fridge & freezers", "Cooker")
+        return FridgeAndFreezers(
+            no_of_stand_alone_seperate_fridges=int(self.get_next_value(data,"No. of Standalone (separate) Fridges:")),
+            no_of_stand_alone_seperate_freezers=int(self.get_next_value(data, "No. of Standalone (separate) Freezers:")),
+            no_of_stand_alone_or_integrated_fridge_freezers=int(self.get_next_value(data, "No. of Standalone or Integrated Fridge Freezers:"))
+        )        
+    
+    def get_cooker(self):
+        data = self.get_data_between("Cooker", "Tumble dryer")
+        return Cooker(
+            cooker_type=self.get_next_value(data, "Cooker Type:"),
+            normal_large_range=self.get_next_value(data, "Normal - Large - Range"),
+            range_fuel=self.get_next_value(data, "Range Fuel:")
+        )
+    
+    def get_tumble_dryer(self):
+        data = self.get_data_between("Tumble dryer", "Have you evidenced 12 months of fuel bill data?")
+        return TumbleDryer(
+            percentage_of_annual_use=int(self.get_next_value(data, "Percentage of annual use:")),
+            space_for_outdoor_drying=True if self.get_next_value(data,"Space for outdoor drying?").lower() == "yes" else False,
+        )
 
 
+
+    
 class QuidosSiteNotesExtractor(SiteNotesExtractor):
     def __init__(self, data_list):
         super().__init__(data_list) 

etl/transform/conditionReportTypes.py

@@ -230,6 +230,7 @@ class Occupant(BaseModel):
 
 class TumbleDryer(BaseModel):
     percentage_of_annual_use: int
+    space_for_outdoor_drying: bool
 
 class Cooker(BaseModel):
     range_fuel: str
@@ -248,12 +249,12 @@ class ShowerAndBath(BaseModel):
     shower_type: str
     do_you_know_the_no_of_showers_per_day_per_week: bool
     please_input_no_of_showers_and_specify_a_day_or_a_week: str
-
+    do_you_know_the_number_of_baths_per_day_or_per_week: str
 class Heating(BaseModel):
     # TODO find one with an example of this one
-    room_stat_in_temperature_in_celsius: str
-    room_stat_location: str
-    is_the_heating_pattern_known: str
+    room_stat_in_temperature_in_celsius: Optional[str] = None
+    room_stat_location: Optional[str] = None
+    is_the_heating_pattern_known: Optional[str] = None
 
 class EnergyUse(BaseModel):
     property_tenure: str