import os
from tqdm import tqdm

if __name__ == "__main__":
    manifest_filepath = "/app/suno/data/audio_2ch_24khz_lg/val_manifest.txt"

    with open(manifest_filepath, "r") as fp:
        filepaths = fp.readlines()
        filepaths = [filepath.strip("\n") for filepath in filepaths]

    datasets = {}
    for filepath in tqdm(filepaths):
        dataset = filepath.split("/")[6]
        if dataset not in datasets:
            datasets[dataset] = 0
        datasets[dataset] += 1

    total_count = 0
    for dataset, count in datasets.items():
        total_count += count

    for dataset, count in datasets.items():
        print(f"{dataset:15s} {count: 8d} ({(count/total_count)*100:.1f}%)")
