| import h5py
|
| import os
|
|
|
| def merge_hdf5_files(source_files, output_file):
|
| """
|
| 将多个 HDF5 文件中的 demo 数据合并到一个文件中，并对 demo 重新编号。
|
|
|
| Args:
|
| source_files (list): 输入的 hdf5 文件路径列表
|
| output_file (str): 输出的合并后 hdf5 文件路径
|
| """
|
|
|
| with h5py.File(output_file, 'w') as f_out:
|
|
|
| if 'data' not in f_out:
|
| out_data_group = f_out.create_group('data')
|
| else:
|
| out_data_group = f_out['data']
|
|
|
| current_demo_idx = len([k for k in out_data_group.keys() if k.startswith('demo_')])
|
|
|
| print(f"开始合并，输出文件: {output_file}")
|
|
|
| for src_path in source_files:
|
| if not os.path.exists(src_path):
|
| print(f"文件不存在，跳过: {src_path}")
|
| continue
|
|
|
| print(f"正在处理文件: {src_path} ...")
|
|
|
| with h5py.File(src_path, 'r') as f_in:
|
| if 'data' not in f_in:
|
| print(f" -> 跳过: 该文件没有 'data' 组")
|
| continue
|
|
|
| in_data_group = f_in['data']
|
|
|
| demo_keys = [k for k in in_data_group.keys() if k.startswith('demo_')]
|
| demo_keys.sort(key=lambda x: int(x.split('_')[1]))
|
|
|
| for demo_key in demo_keys:
|
| source_demo = in_data_group[demo_key]
|
|
|
| new_demo_name = f"demo_{current_demo_idx:04d}"
|
|
|
| f_in.copy(source_demo, out_data_group, name=new_demo_name)
|
|
|
| current_demo_idx += 1
|
|
|
| print("-" * 30)
|
| print(f"合并完成！")
|
| print(f"总共保存了 {current_demo_idx} 个 demo (demo_{0:04d} 到 demo_{current_demo_idx-1:04d})")
|
|
|
|
|
| if __name__ == "__main__":
|
|
|
| input_files_list = [
|
|
|
| "/home/dobot/dobot/x-trainer/datasets/1_01291036.hdf5",
|
| "/home/dobot/dobot/x-trainer/datasets/1_01291858.hdf5",
|
| "/home/dobot/dobot/x-trainer/datasets/1_02041700.hdf5"
|
|
|
| ]
|
|
|
|
|
| file_name = '1_merged_data.hdf5'
|
| target_folder = '/home/dobot/dobot/x-trainer/merged_data/merge_hdf5/'
|
|
|
| if not os.path.exists(target_folder):
|
| os.makedirs(target_folder)
|
| print(f"文件夹不存在，已自动创建 -> {target_folder}")
|
|
|
| output_full_path = os.path.join(target_folder, file_name)
|
|
|
| merge_hdf5_files(input_files_list, output_full_path) |
