dataset.py

import datasets
import tarfile
import io
import json
import soundfile as sf
import os
from pathlib import Path

class TestDS1(datasets.GeneratorBasedBuilder):
    VERSION = datasets.Version("1.0.0")

    def _info(self):
        return datasets.DatasetInfo(
            description="Dataset of audiobook chunks with audio and metadata from multiple books.",
            features=datasets.Features(
                {
                    "audio": datasets.Audio(sampling_rate=16000),  # Use datasets.Audio
                    "audio_path": datasets.Value("string"),
                    "start_seconds": datasets.Value("float32"),
                    "end_seconds": datasets.Value("float32"),
                    "duration_seconds": datasets.Value("float32"),
                    "text": datasets.Value("string"),
                    "book_id": datasets.Value("string"),
                }
            ),
            supervised_keys=None,
            homepage="https://huggingface.co/datasets/jtarjanyi/testds1",
            citation="",
        )

    def _split_generators(self, dl_manager):
        base_url = "https://huggingface.co/datasets/jtarjanyi/testds1/resolve/main"
        book_folders = ["Buszkeseg_1"]  # Update with actual folder names
        tar_paths = {}

        for book in book_folders:
            tar_url = f"{base_url}/{book}/chunks.tar"
            print(f"Downloading tar file: {tar_url}")
            tar_paths[book] = dl_manager.download(tar_url)

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={"tar_paths": tar_paths},
            ),
        ]

    def _generate_examples(self, tar_paths):
        idx = 0
        for book_id, tar_path in tar_paths.items():
            print(f"Processing tar file for book: {book_id} at {tar_path}")
            try:
                with tarfile.open(tar_path, "r") as tar:
                    json_files = sorted([f.name for f in tar.getmembers() if f.name.endswith(".json")])
                    print(f"Found {len(json_files)} JSON files in {book_id}")

                    for json_filename in json_files:
                        try:
                            json_file = tar.extractfile(json_filename)
                            if json_file is None:
                                print(f"Skipping {json_filename}: Could not extract from {tar_path}")
                                continue
                            json_data = json.load(json_file)
                            print(f"Loaded JSON: {json_filename}")

                            wav_filename = json_data["audio_path"]
                            if not wav_filename.endswith(".wav"):
                                print(f"Skipping {wav_filename}: Invalid audio path in {json_filename}")
                                continue

                            try:
                                wav_member = tar.getmember(wav_filename)
                            except KeyError:
                                print(f"Skipping {wav_filename}: Not found in {tar_path}")
                                continue

                            wav_file = tar.extractfile(wav_member)
                            if wav_file is None:
                                print(f"Skipping {wav_filename}: Could not extract from {tar_path}")
                                continue

                            wav_bytes = wav_file.read()
                            print(f"Extracted {wav_filename}: {len(wav_bytes)} bytes")

                            wav_io = io.BytesIO(wav_bytes)
                            _, sampling_rate = sf.read(wav_io)
                            print(f"Sampling rate for {wav_filename}: {sampling_rate}")

                            yield idx, {
                                "audio": {"bytes": wav_bytes, "sampling_rate": sampling_rate},
                                "audio_path": wav_filename,
                                "start_seconds": json_data["start_seconds"],
                                "end_seconds": json_data["end_seconds"],
                                "duration_seconds": json_data["duration_seconds"],
                                "text": json_data["text"],
                                "book_id": book_id,
                            }
                            idx += 1

                        except Exception as e:
                            print(f"Error processing {json_filename} in {tar_path}: {str(e)}")
                            continue
            except Exception as e:
                print(f"Error opening tar file {tar_path}: {str(e)}")
                continue