import pandas as pd
from model_data.utils import setup_logger

logger = setup_logger()


class Need:
    """
    Contains methods to read and interface with the NEED dataset.

    Current iterations of this data is the 2021 anonymised dataset, which can be found here:
    https://www.gov.uk/government/statistics/national-energy-efficiency-data-framework-need-anonymised-data-2021
    """

    def __init__(self, local_authorities, path):
        self.local_authorities = local_authorities
        self.path = path

    def read(self) -> pd.DataFrame:
        """
        Reads the NEED dataset from a csv file.
        :param path:    path to the csv file
        :return:        pandas dataframe containing the data
        """
        logger.info("Reading NEED data - could take a moment")
        df = pd.read_csv(self.path)
        df = df[df["REGION"].isin(self.local_authorities)]

        z = df[df["REGION"].str.contains("E9")]

        type(df["REGION"].values[0])
        return df