import pandas as pd
import numpy as np
import json
import os

from pprint import pprint

data_dir = '../data'

# Just a munge/transform/save as feather function per file for now. 

def popularity():
    popularity = pd.read_csv(os.path.join(data_dir, 'ds_sony_track_popularity.csv'), names = ['track_id', 'date', 'popularity'])
    popularity['date'] = pd.to_datetime(popularity['date'])


def streams():
    streams = pd.read_csv(os.path.join(data_dir, 'streams_60_tracks.csv'), names = ['track_id','date','region','all_streams', 'source_album_page_streams','source_artist_page_streams','source_chart_streams', 'source_collection_streams','source_daily_mix_streams','source_discover_weekly_streams','source_other_streams','source_playlist_streams','source_radio_streams','soruce_release_radar_streams','source_search_streams','sony_playlist_streams','spotify_playlist_streams','otehr_playlist_streams','shuffle_streams'])
    streams['date'] = pd.to_datetime(streams['date'])
    streams.to_feather(os.path.join(data_dir, 'streams_60.feather'))

def track_data():
    data_dir = '../data'
    with open(os.path.join(data_dir, 'sony_tracks_data.json')) as tracks_json:
        tracks = json.load(tracks_json)
    
    # example_track = tracks['results'][1]
    # pprint(example_track)

    # track_id = example_track['track_id']
    # track_release_date = example_track['track_data']['album']['release_date']
    # track_release_date_precision = example_track['track_data']['album']['release_date_precision']

    track_ids = []
    track_release_dates = []
    track_release_date_precisions = []
    album_release_dates = []
    album_release_date_precisions = []

    for tt in tracks['results']:
        print(f'parsing {tt["track_id"]}')
        track_ids.append(tt['track_id'])
        track_release_dates.append(tt['track_data']['album']['release_date'])
        track_release_date_precisions.append(tt['track_data']['album']['release_date_precision'])

        album_release_dates.append(tt['album_data']['release_date'])
        album_release_date_precisions.append(tt['album_data']['release_date_precision'])

    track_release_dates = pd.DataFrame.from_dict( { 'track_id' : track_ids, 
        'track_release_date' : track_release_dates, 
        'track_release_date_precision': track_release_date_precisions,
        'album_release_date' : album_release_dates, 
        'album_release_date_precision' : album_release_date_precisions})

    # track_release_dates.groupby('album_release_date_precision').size()
    # track_release_dates[track_release_dates['release_date'] != track_release_dates['album_release_date']] 
    
    track_release_dates['derived_release_date'] = np.where(track_release_dates['track_release_date'] <= track_release_dates['album_release_date'], track_release_dates['track_release_date'], track_release_dates['album_release_date'])
# TODO: This covers the frequency issue too, but revisit for more data!

    track_release_dates[['track_id', 'derived_release_date']].to_feather('../data/track_release_dates.feather')