from glob import glob
from argparse import ArgumentParser
from os import system, path
import csv


parser = ArgumentParser(description="Prepare dataset from midi files for later use in onsets and frames.")
parser.add_argument('--dry', '-d', dest='dry', action='store_true', help="Dry run, print commands instead of executing them")
parser.add_argument('--no-convert', '-m', dest='no_convert', action='store_true', help="Only create manifest.")
parser.add_argument('--soundfont', '-f', default='/usr/share/sounds/sf2/FluidR3_GM.sf2', help="Path to soundfont.")
parser.add_argument('-n', type=int, help="Max midi files to process")

args = parser.parse_args()

execute = print if args.dry else system

processed = 0
if not args.no_convert:
    for midi in glob("dataset/midi/*.mid"):
        wav = midi.replace('.mid', '.wav').replace('/midi/', '/wav/')

        processed = processed + 1
        if args.n is not None and processed > args.n:
            break

        if path.isfile(wav):
            continue

        print(f"Converting {midi} to {wav}")
        execute(f"fluidsynth -nli -r 48000 -T wav -F {wav} {args.soundfont} {midi} > /dev/null")


manifest = []
for wav in glob("dataset/wav/*.wav"):
    midi = wav.replace('.wav', '.mid').replace('/wav/', '/midi/')

    if path.isfile(midi):
        manifest.append((path.basename(wav), path.basename(midi), 'test'))


with open('dataset/manifest.csv', 'w') as manifest_file:
    writer = csv.writer(manifest_file)
    writer.writerow(('audio_filename', 'midi_filename', 'split'))
    writer.writerows(manifest)