import pandas as pd


def apply_adj_to_dod(data_df: pd.DataFrame):
    """
    Add adjusted columns to each day-over-day column.
    The adjustment dependent on day of week.
    """

    # replicated from DEV_ENGINEERING.TSTOWE.DAY_OF_WEEK_MODEL
    # WEEKDAY_NUM, WEEKDAY, PCT_CHANGE_VS_PREV
    # 0	Monday	9.7992323305187380
    # 1	Tuesday	3.4008245639724453
    # 2	Wednesday	0.6553378172314275
    # 3	Thursday	2.3678775205019287
    # 4	Friday	3.8720233434785910
    # 5	Saturday	-6.4442725833266910
    # 6	Sunday	-12.0355455758962520
    dow_data = [
        [0, 'Monday', 9.7992323305187380],
        [1, 'Tuesday', 3.4008245639724453],
        [2, 'Wednesday', 0.6553378172314275],
        [3, 'Thursday', 2.3678775205019287],
        [4, 'Friday', 3.8720233434785910],
        [5, 'Saturday', -6.4442725833266910],
        [6, 'Sunday', -12.0355455758962520],
    ]
    dow_df = pd.DataFrame(dow_data, columns=['weekday_num', 'weekday', 'pct_change_vs_prev'])

    data_df['weekday_num'] = data_df['report_date'].dt.dayofweek  # 0=Mon, ..., 6=Sun
    data_df = data_df.merge(dow_df, on='weekday_num', how='left')

    # Add columns that adjust the DOD columns to index against the expected.
    for x in data_df.columns:
        if x[0:4] == 'dod_':
            adj_col = 'adj_' + x
            # Index vs expected (ratio; handle divide by zero)
            data_df[adj_col] = data_df[x] / data_df['pct_change_vs_prev'].replace(0, pd.NA)

    return data_df


def add_is_any(df):
    try:
        del df['is_any']
    except KeyError:
        pass

    iscols = [x for x in df.columns if x[0:3] == 'is_']

    df['is_any'] = 0

    for col in iscols:
        df['is_any'][df[col] == 1] = 1

    return df


def combine_data_sources(data_df: pd.DataFrame, days_trending_df: pd.DataFrame, report_date: str):
    ironed_out_df_mini = days_trending_df[days_trending_df['report_date'] == pd.to_datetime(report_date)]
    ironed_out_df_mini = ironed_out_df_mini[['pfn_geo', 'combined_forecast', 'consecutive_trend', 'vs_forecast_lift']]

    combined_df = data_df.merge(ironed_out_df_mini, how='left', on=['pfn_geo'])

    # ensure dataframe is ready for predictions.
    combined_df = combined_df.fillna(0.0)

    return combined_df