Model/epc_data/attributes/HotWaterAttributes.py

from typing import Dict, Union
from epc_data.attributes.attribute_utils import clean_description, remove_punctuation, find_keyword


class HotWaterAttributes:
    HEATER_TYPES = [
        'gas instantaneous',
        'electric heat pump',
        'electric immersion',
        'gas boiler',
        'electric instantaneous',
        'gas multipoint',
    ]

    SYSTEM_TYPES = [
        'from main system',
        'from secondary system',
        'community scheme',
    ]

    THERMOSTAT_CHARACTERISTICS = [
        'no cylinder thermostat',
    ]

    HEATING_SCOPE = [
        'water heating only',
    ]

    OTHER_CHARACTERISTICS = [
        'waste water heat recovery',
        'plus solar',
        'off-peak',
        'flue gas heat recovery',
        'standard tariff',
        'chp',
    ]

    NO_SYSTEM_PRESENT_KEYWORDS = [
        'no system present',
    ]

    def __init__(self, description: str):
        self.description: str = clean_description(description.lower())

        if not any(
            self._keyword_in_description(keywords)
            for keywords in [
                self.HEATER_TYPES,
                self.SYSTEM_TYPES,
                self.THERMOSTAT_CHARACTERISTICS,
                self.HEATING_SCOPE,
                self.OTHER_CHARACTERISTICS,
                self.NO_SYSTEM_PRESENT_KEYWORDS,
            ]
        ):
            raise ValueError('Invalid description')

    def _keyword_in_description(self, keywords):
        return any(keyword in self.description for keyword in keywords)

    def process(self) -> Dict[str, Union[str, bool]]:
        result: Dict[str, Union[str, bool]] = {
            "heater_type": find_keyword(self.description, self.HEATER_TYPES),
            "system_type": find_keyword(self.description, self.SYSTEM_TYPES),
            "thermostat_characteristics": find_keyword(self.description, self.THERMOSTAT_CHARACTERISTICS),
            "heating_scope": find_keyword(self.description, self.HEATING_SCOPE),
            "other_characteristics": [find_keyword(self.description, [keyword]) for keyword in
                                      self.OTHER_CHARACTERISTICS if find_keyword(self.description, [keyword])],
            "no_system_present": find_keyword(self.description, self.NO_SYSTEM_PRESENT_KEYWORDS),
        }

        return result