HDF5文件中单个(连续)数据集的字节偏移量可以使用作为HDF5 library的一部分分发的h5ls tool来确定，但我一直无法确定如何使用Python(具体地说，h5py package)以编程方式完成这项任务.
我找到了R的rhdf5 package，其中包括h5ls function，但据我所知，h5py没有任何类似的东西.我也找到了How can I get raw byte offsets and bytes read from that offset using the HDF5 C API?，但它没有答案，而且也是在询问C而不是Python.
我并不一定要寻找一行解决方案，我只是想找一种使用Python访问这些信息的方法.我也没有硬性规定要用h5py美元.如果有其他方案或解决方案可以做得更好，我会很乐意考虑的.
下面是一个使用低级h5py API的示例..get_offset()方法适用于任何DataSet对象.下面的示例显示了如何通过两种方式获取低级id(和偏移量):1)通过对象引用；2)通过名称引用.
# first create a simple file with 3 datasets
with h5py.File('SO_76068934.h5','w') as h5f:
data = [float(i) for i in range (10)]
h5f.create_dataset('ds_1', data=data)
data = [float(i) for i in range (10,20)]
h5f.create_dataset('ds_2', data=data)
data = [float(i) for i in range (20,30)]
h5f.create_dataset('ds_3', data=data)
# open the file in read mode and get dataset objects, IDs and Offsets
with h5py.File('SO_76068934.h5','r') as h5f:
# loop over root level dataset names and objects
for ds_name, ds_obj in h5f.items():
# referencing dataset object:
ds_id = ds_obj.id # get low level dataset ID from object
ds_off = ds_id.get_offset() # Low-level method
print(f'For: {ds_name}; Offset: {ds_off}')
# referencing dataset name:
ds_off = h5f[ds_name].id.get_offset() # get ID and offset in 1 step
print(f'For: {ds_name}; Offset: {ds_off}\n')