import os
import json
from huggingface_hub import login
from datasets import Dataset, Audio

def load_audio_data(root_folder_path):
    """从指定文件夹加载音频和对应的JSON数据。"""
    data = {
        "Rebuttal Subject": [],
        "Audio Name": [],
        "audio": [],
        "text": [],
        "json": [],
    }
    for subfolder in os.listdir(root_folder_path):
        subfolder_path = os.path.join(root_folder_path, subfolder)

        if os.path.isdir(subfolder_path):
            for file_name in os.listdir(subfolder_path):
                if file_name.endswith('.json'):
                    json_path = os.path.join(subfolder_path, file_name)

                    with open(json_path, 'r', encoding='utf-8') as json_file:
                        json_data = json.load(json_file)
                        text = json_data.get('text', '')

                    audio_file = file_name.replace('.json', '.wav')
                    audio_path = os.path.join(subfolder_path, audio_file)

                    if os.path.exists(audio_path):
                        data["Audio Name"].append(audio_file)
                        data["audio"].append(audio_path)
                        data["text"].append(text)
                        data["json"].append(json_data)
                        data["Rebuttal Subject"].append(subfolder.replace('、', '/'))

    return data

def main():
    login("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")

    root_folder_path = 'data/'  # 替换为你的文件夹路径
    data = load_audio_data(root_folder_path)

    audio_dataset = Dataset.from_dict(data)
    audio_dataset = audio_dataset.cast_column("audio", Audio())

    audio_dataset.push_to_hub("amphion/Debatts-Data")
    print("数据集已上传到 Hugging Face Hub，并可播放音频。")

if __name__ == "__main__":
    main()