Check for synched off-vocal versions

synched-off-vocal.py

#!/usr/bin/env python3
'''Check for synched off-vocal versions'''

import os
import sys
import hashlib
import json
BUF_SIZE = 65535

OFF_VOCAL_TAG = 'c0cc87b9-55b9-40f0-878a-fbb9e34c151e'

def find_off_vocal(directory):
    '''Returns list of .ass for off vocal karaokes'''
    file_list = []
    karajson_directory = os.path.join(directory, 'karaokes')
    for filename in os.listdir(karajson_directory):
        if not filename.endswith('.kara.json'):
            continue
        with open(os.path.join(karajson_directory, filename), 'r', encoding='utf-8') as file:
            karajson = json.load(file)
            if not 'versions' in karajson['data']['tags']:
                continue
            if OFF_VOCAL_TAG in karajson['data']['tags']['versions']:
                file_list.append(karajson['medias'][0]['lyrics'][0]['filename'])
    print(f'Found {len(file_list)} off-vocal files.')
    return file_list

def compute_hashes(directory):
    '''Computes hashes of all .ass'''
    out = {}
    ass_directory = os.path.join(directory, 'lyrics')
    for filename in os.listdir(ass_directory):
        with open(os.path.join(ass_directory, filename), 'rb') as file:
            sha512 = hashlib.sha512()
            while True:
                data = file.read(BUF_SIZE)
                if not data:
                    break
                sha512.update(data)
        out[filename] = sha512.hexdigest()
    return out


def main(directory):
    '''Main function'''
    off_vocals = find_off_vocal(directory)
    hashes = compute_hashes(directory)
    total_count = 0
    sync = 0
    out_sync = 0
    trio = 0
    for off_vocal_kara in off_vocals:
        off_vocal_hash = hashes[off_vocal_kara]
        found = False
        total_count += 1
        for file, file_hash in hashes.items():
            if file_hash == off_vocal_hash and off_vocal_kara != file:
                print(f'Found synchronized lyrics for {off_vocal_kara}: {file_hash}')
                if found:
                    trio += 1
                else:
                    found = True
                    sync += 1
        if not found:
            out_sync += 1
            print(f'No Synchronized lyric for {off_vocal_kara}')
    print(f'Total: {total_count}, sync: {sync}, out of sync: {out_sync}, trio: {trio}')

if __name__ == '__main__':
    if len(sys.argv) < 2:
        sys.exit(1)
    main(sys.argv[1])