读取h5文件

Question

I am a new python user and I want to read the data from a h5 file. The code that I have used to read the data is given below:

>>> import numpy as np
>>> import h5py
>>> f = h5py.File('file.h5', 'r')
>>> list(f.keys())
[u'data']
>>> dset = f[u'data']
>>> dset.shape
(64, 64, 64)
>>> dset.dtype
dtype(('<f8', (3,)))

Can anyone help me to understand these commands? My questions are: 1. What can I understand by the terms [u'data'] and dtype(('

Answer 1

As @hpaulj mentions, the h5py doc is a good reference. You also need to understand basic HDF5 file concepts. It's a big topic. To get started, review the Learning HDF5 pages from The HDF Group. I know (from personal experience) that you will struggle if you don't understand how to navigate the hierarchy. Learn the differences between Group and Dataset objects.

Here's an explanation of the output in your OP. Note that these are all h5py functions (not numpy).

>>> list(f.keys())
[u'data']

The “keys” are the names of group members, and the “values” are the members (Group and Dataset objects). This is a list of the Node names at the root level of the file. Nodes can be Groups or Datasets. In your case, you have one dataset named data . (The only group in this HDF5 file is the root group: '/', there are no groups below the root.)

The next step accesses the data in the data dataset.

>>> dset = f[u'data']

The shape attribute gives you the dimensions of the dataset:

>>> dset.shape
(64, 64, 64)

The dtype attribute gives you the data types of the dataset (just like numpy):

>>> dset.dtype
type(('<f8', (3,)))

So, you have an array of floats. If you want to see the data, you can enter this (I would only do this for small datasets, or slice to print a few rows):

>>> for row in dset:
        print (row)

From here, what you do with the data is up to you.