# (c) Ben Richeson 2022
# github.com/benricheson101

# USAGE:
#   python3 spotify_stats.py [year]
#
# SETUP:
#   1. place this file in the same directory as `endsong_N.json` files in your spotify data package
#   2. run the file, optionally appending a year to the end of the command

from collections import Counter
from typing import TypedDict, Optional
from glob import glob
from sys import argv
import json

top_n = 5

ms = 0
count = 0

top_tracks = Counter()
top_artists = Counter()

class StreamedTrack(TypedDict):
    ts: str
    username: str
    platform: str
    ms_played: int
    conn_country: str
    ip_addr_decrypted: str
    user_agent_decrypted: Optional[str]
    master_metadata_track_name: Optional[str]
    master_metadata_album_artist_name: Optional[str]
    master_metadata_album_album_name: Optional[str]
    spotify_track_uri: Optional[str]
    episode_name: Optional[str]
    episode_show_name: Optional[str]
    spotify_episode_uri: Optional[str]
    reason_start: str
    reason_end: str
    shuffle: bool
    skipped: Optional[bool]
    offline: bool
    offline_timestamp: str
    incognito_mode: bool

filter_year = argv[1] if len(argv) > 1 else None

paths = glob("endsong_*.json")

if len(paths) == 0:
    print("could not find streaming history data files. make sure this file is in the same folder as `endsong_N.json` files")
    exit(1)

for filename in paths:
    with open(filename, "r") as file:
        j: list[StreamedTrack] = json.loads(file.read())
        for track in j:
            if filter_year and not track["ts"].startswith(filter_year):
                continue

            name = track["master_metadata_track_name"] or track["episode_name"] or ""
            artist = track["master_metadata_album_artist_name"] or ""

            top_tracks[(name, artist)] += 1
            top_artists[artist] += 1

            ms += int(track["ms_played"])
            count += 1

seconds = ms / 1000
minutes = seconds / 60
hours = minutes / 60
days = hours / 24
months = days / 30
years = months / 12

if count == 0:
    print("no stats available")
    exit(0)

if filter_year:
    print(f"-- spotify playback stats {filter_year} --")
else:
    print("-- spotify playback stats --")
print()
print(f"  ms     : {ms:,.2f}")
print(f"  seconds: {seconds:,.2f}")
print(f"  minutes: {minutes:,.2f}")
print(f"  hours  : {hours:,.2f}")
print(f"  days   : {days:,.2f}")
print(f"  months : {months:,.2f}")
print(f"  years  : {years:,.2f}")
print(f"  plays  : {count:,}")
print()
print("  top tracks:")
for t in top_tracks.most_common(top_n):
    print(f"    {t[1]}: {t[0][1]} - {t[0][0]}")
print()
print("  top artists:")
for a in top_artists.most_common(top_n):
    print(f"    {a[1]}: {a[0]}")