import json


artistTracksPlayedCounter = {}

artistUniqueTracksPlayedCounter = {}

artistCumulativeTimePlayed = {}

songPlayCounter = {}

REPORT_SIZE = 30

with open('StreamingHistory_music_0.json') as f:
    data = json.load(f)

    for song in data:

        # example record
        # "endTime" : "2023-12-01 13:36",
        # "artistName" : "The Vaccines",
        # "trackName" : "Teenage Icon",
        # "msPlayed" : 184240

        artist = song['artistName']
        track = song['trackName']
        timePlayed = song['msPlayed']
        
        datePlayed = song['endTime']

        # only consider songs that were played after Jan 1 2024
        if datePlayed < "2024-01-01":
            continue

        if artist in artistTracksPlayedCounter:
            artistTracksPlayedCounter[artist] += 1
        else:
            artistTracksPlayedCounter[artist] = 1

        if artist in artistCumulativeTimePlayed:
            artistCumulativeTimePlayed[artist] += timePlayed
        else:
            artistCumulativeTimePlayed[artist] = timePlayed

        if track in songPlayCounter:
            songPlayCounter[track] += 1
        else:
            songPlayCounter[track] = 1

        if artist in artistUniqueTracksPlayedCounter:
            if track not in artistUniqueTracksPlayedCounter[artist]:
                artistUniqueTracksPlayedCounter[artist].append(track)
        else:
            artistUniqueTracksPlayedCounter[artist] = [track]

    
    # print the top 10 artists by number of tracks played
    print("Top %d artists by number of tracks played" % REPORT_SIZE)

    sortedArtists = sorted(artistTracksPlayedCounter.items(), key=lambda x: x[1], reverse=True)

    for i in range(REPORT_SIZE):
        print(sortedArtists[i])

    # print the top 10 artists by cumulative time played
    print("\n\n\nTop %d artists by cumulative time played" % REPORT_SIZE)

    sortedArtists = sorted(artistCumulativeTimePlayed.items(), key=lambda x: x[1], reverse=True)

    for i in range(REPORT_SIZE):

        # convert time to minutes
        time = sortedArtists[i][1] / 60000
        # round to 2 decimal places
        time = round(time, 2)
        print(sortedArtists[i][0], time)

    # print the top 10 songs by number of times played
    print("\n\n\nTop %d songs by number of times played" % REPORT_SIZE)

    sortedSongs = sorted(songPlayCounter.items(), key=lambda x: x[1], reverse=True)

    for i in range(REPORT_SIZE):

        print(sortedSongs[i])

    # print the top 10 artists by number of unique tracks played

    print("\n\n\nTop %d artists by number of unique tracks played" % REPORT_SIZE)

    sortedArtists = sorted(artistUniqueTracksPlayedCounter.items(), key=lambda x: len(x[1]), reverse=True)

    for i in range(REPORT_SIZE):

        # just print the artist name and the count of  unique tracks
        
        print(sortedArtists[i][0], len(sortedArtists[i][1]))