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


class MainheatControlAttributes:
    # These systems allow for the automatic regulation of temperature
    THERMOSTATIC_CONTROL_KEYWORDS = [
        'room thermostats',
        'appliance thermostats',
        'trvs',
        'room thermostat',
        'temperature zone control',
        'time and temperature zone control',
        'no thermostatic control'
    ]

    # These systems provide a way to store and manage energy usage
    CHARGING_SYSTEM_KEYWORDS = [
        'charging system',
        'automatic charge control',
        'manual charge control',
        'flat rate charging',
        'high heat retention storage heaters'
    ]

    # These systems are mainly focused on on/off operations
    SWITCH_SYSTEM_KEYWORDS = [
        'programmer',
        'flow switch',
        'bypass'
    ]

    # These keywords indicate a lack of control system
    NO_CONTROL_SYSTEM_KEYWORDS = [
        'no time or thermostatic control',
        'none'
    ]

    # These systems provide control specifically for Domestic Hot Water
    DHW_CONTROL_KEYWORDS = [
        'dhw only'
    ]

    # This system indicates the use of community heating
    COMMUNITY_HEATING_KEYWORDS = [
        'use of community heating'
    ]

    MULTIPLE_ROOM_THERMOSTATS_PHRASES = [
        'at least two room stats',
        'at least two room thermostats'
    ]

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

        self.nodata = not description or any(keyword in self.description for keyword in self.NO_CONTROL_SYSTEM_KEYWORDS)

        if not self.nodata and not any(
            self._keyword_in_description(keywords)
            for keywords in [
                self.THERMOSTATIC_CONTROL_KEYWORDS,
                self.CHARGING_SYSTEM_KEYWORDS,
                self.SWITCH_SYSTEM_KEYWORDS,
                self.DHW_CONTROL_KEYWORDS,
                self.COMMUNITY_HEATING_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]] = {
            "thermostatic_control": self._find_keyword(self.THERMOSTATIC_CONTROL_KEYWORDS),
            "charging_system": self._find_keyword(self.CHARGING_SYSTEM_KEYWORDS),
            "switch_system": self._find_keyword(self.SWITCH_SYSTEM_KEYWORDS),
            "no_control": self._find_keyword(self.NO_CONTROL_SYSTEM_KEYWORDS),
            "dhw_control": self._find_keyword(self.DHW_CONTROL_KEYWORDS),
            "community_heating": self._find_keyword(self.COMMUNITY_HEATING_KEYWORDS),
            "multiple_room_thermostats": any(
                phrase in self.description for phrase in self.MULTIPLE_ROOM_THERMOSTATS_PHRASES),
        }

        return result

    def _find_keyword(self, keywords):
        description_words = set(self.description.split())

        # Sort keywords by length, longest first. This ensures that 'time and temperature zone control'
        # will be checked before 'temperature zone control' if both are present in the keywords list
        keywords.sort(key=len, reverse=True)

        for keyword in keywords:
            keyword_words = set(keyword.split())
            if keyword_words.issubset(description_words):
                return keyword

        # If no keyword is found, try again after removing punctuation
        description_without_punct = remove_punctuation(self.description)
        description_words_without_punct = set(description_without_punct.split())

        for keyword in keywords:
            keyword_words = set(keyword.split())
            if keyword_words.issubset(description_words_without_punct):
                return keyword

        return None