import datetime as dt
import logging
import os

import pandas as pd

from tadas.domain.trending import prep_timeseries, iron_out_trends
from tadas.platform import config
from tadas.snowflake import client as snowflake_utils
from tadas.platform.caching import cached_df

logger = logging.getLogger(__name__)


@cached_df()
def version_v3(report_date, group_by_country, select_tracks_table):
    source_schema = config.get('SOURCE_SCHEMA')
    days_back: int = config.get('TRENDING_DAYS_LOOKBACK_PERIOD_DAYS')

    logger.info(f"--- {dt.datetime.now().strftime('%H:%M:%S')} pulling streams_active...")
    q = f"""
    select
        f.download_activity_date as report_date,
        t.isrc_cd,
        t.geo_country,
        sum(
          // lean_forward aka streams_active (spotify + apple + others...)
          // expected: spotify: 'Artist','Album','Search','Play_Queue', 'Collection'
          // streams_active has only: ('artist', 'album', 'search', 'play_queue')
          // so we add collections
          streams_active +
          streams_sos_spotify_collection
          ) AS streams
    from {source_schema}v_streams_by_track_country_feed_distributor_daily f
    inner join {select_tracks_table} t on f.isrc = t.isrc_cd {'and f.country_code = t.geo_country' if group_by_country else ''}
    where f.download_activity_date between %(report_date)s::date - %(days_back)s and %(report_date)s::date
    group by all
"""
    params = {
        'report_date': report_date,
        'days_back': days_back,
    }
    with snowflake_utils.snowflake_connection() as conn:
        big_raw_df = snowflake_utils.query_snowflake_to_df(query=q, params=params, connection=conn)

    big_raw_df = big_raw_df.groupby(['report_date', 'isrc_cd', 'geo_country']).sum().reset_index()
    big_raw_df['pfn_geo'] = big_raw_df['isrc_cd'].astype(str) + '_' + big_raw_df['geo_country'].astype(str)

    logger.info(f"--- {dt.datetime.now().strftime('%H:%M:%S')} : got big_raw_df! {len(big_raw_df)} rows of data.  Adjusting for Lag.")

    logger.info(f"--- {dt.datetime.now().strftime('%H:%M:%S')} : prepping the time series - {len(big_raw_df)} rows of data.")
    timeseries_df = prep_timeseries(big_raw_df)

    logger.info(f"--- {dt.datetime.now().strftime('%H:%M:%S')} : ironing out trends")
    ironed_out_intl_df = iron_out_trends(timeseries_df)

    ironed_out_intl_df['vs_forecast_lift'] = ironed_out_intl_df['streams'] / ironed_out_intl_df[
        'combined_forecast'] - 1.0

    return ironed_out_intl_df


def save_days_trending(df, table_name):
    assert 'days_trending' in table_name, 'table name must contain "days_trending"'
    df['report_date'] = pd.to_datetime(df['report_date']).dt.date

    snowflake_utils.saveto_snowflake(
        df=df,
        myschema=os.getenv('SNOWFLAKE_SCHEMA'),
        table=table_name,
        mode='replace',
    )