Menu +

Search Posts

h5py : Quick Start Guide

1. Install

conda install h5py

2. Core concepts

An HDF5 file is a container for two kinds of objects: datasets, which are array-like collections of data, and groups, which are folder-like containers that hold datasets and other groups. The most fundamental thing to remember when using h5py is:
  Groups work like dictionaries, and datasets work like NumPy arrays

Suppose someone has sent you a HDF5 file, mytestfile.hdf5. (To create this file, read Appendix: Creating a file.) The very first thing you’ll need to do is to open the file for reading:

>>> import h5py
>>> f = h5py.File('mytestfile.hdf5', 'r')

The File object is your starting point. What is stored in this file? Remember h5py.File acts like a Python dictionary, thus we can check the keys,

>>> list(f.keys())

Based on our observation, there is one data set, mydatasetin the file. Let us examine the data set as a Dataset object

>>> dset = f['mydataset']

The object we obtained isn’t an array, but an HDF5 dataset. Like NumPy arrays, datasets have both a shape and a data type:

>>> dset.shape
>>> dset.dtype

They also support array-style slicing. This is how you read and write data from a dataset in the file:

>>> dset[...] = np.arange(100)
>>> dset[0]
>>> dset[10]
>>> dset[0:100:10]
array([ 0, 10, 20, 30, 40, 50, 60, 70, 80, 90])

For more, see File Objects and Datasets.

Appendix: Creating a file

At this point, you may wonder how mytestdata.hdf5 is created. We can create a file by setting the mode to wwhen the File object is initialized. Some other modes are a (for read/write/create access), and r+ (for read/write access). A full list of file access modes and their meanings is at File Objects.

>>> import h5py
>>> import numpy as np
>>> f = h5py.File("mytestfile.hdf5", "w")

The File object has a couple of methods which look interesting. One of them is create_dataset, which as the name suggests, creates a data set of given shape and dtype

>>> dset = f.create_dataset("mydataset", (100,), dtype='i')

The File object is a context manager; so the following code works too

>>> import h5py
>>> import numpy as np
>>> with h5py.File("mytestfile.hdf5", "w") as f:
>>>     dset = f.create_dataset("mydataset", (100,), dtype='i')

Groups and hierarchical organization

“HDF” stands for “Hierarchical Data Format”. Every object in an HDF5 file has a name, and they’re arranged in a POSIX-style hierarchy with /-separators:


The “folders” in this system are called groups. The Fileobject we created is itself a group, in this case the root group, named /:


Creating a subgroup is accomplished via the aptly-named create_group. But we need to open the file in the “append” mode first (Read/write if exists, create otherwise)

>>> f = h5py.File('mydataset.hdf5', 'a')
>>> grp = f.create_group("subgroup")

All Group objects also have the create_* methods like File:

>>> dset2 = grp.create_dataset("another_dataset", (50,), dtype='f')

By the way, you don’t have to create all the intermediate groups manually. Specifying a full path works just fine:

>>> dset3 = f.create_dataset('subgroup2/dataset_three', (10,), dtype='i')

Groups support most of the Python dictionary-style interface. You retrieve objects in the file using the item-retrieval syntax:

>>> dataset_three = f['subgroup2/dataset_three']

Iterating over a group provides the names of its members:

>>> for name in f:
...     print name

Membership testing also uses names:

>>> "mydataset" in f
>>> "somethingelse" in f

You can even use full path names:

>>> "subgroup/another_dataset" in f

There are also the familiar keys()values()items()and iter() methods, as well as get().

Since iterating over a group only yields its directly-attached members, iterating over an entire file is accomplished with the Group methods visit() and visititems(), which take a callable:

>>> def printname(name):
...     print name
>>> f.visit(printname)

For more, see Groups.


One of the best features of HDF5 is that you can store metadata right next to the data it describes. All groups and datasets support attached named bits of data called attributes.

Attributes are accessed through the attrs proxy object, which again implements the dictionary interface:

>>> dset.attrs['temperature'] = 99.5
>>> dset.attrs['temperature']
>>> 'temperature' in dset.attrs

For more, see Attributes.

Leave a Reply

Your email address will not be published. Required fields are marked *