H5py Quick Start Guide

Hello everyone , I meet you again , I'm your friend, Quan Jun .

h5py yes Python Language is used to operate HDF5 Module . The following article mainly introduces h5py Quick start guide for , Translated from h5py Official documents of ：http://docs.h5py.org/en/latest/quick.html . This translation is for personal learning only h5py For the purpose , If there is any improper translation , Please contact the author or provide the correct translation , Thank you very much ！

install

Use Anaconda perhaps Miniconda:

conda install h5py

Copy

use Enthought Canopy, have access to GUI Install the package or use

enpkg h5py

Copy

install . use pip or setup.py install , Please refer to Installation mode .

The core concept

One HDF5 A file is a container , Used to store two types of objects ：datasets, A data set similar to an array ;groups, Folder like container , Can store datasets And other groups. When using h5py when , The most basic rule is ：

groups It's like a dictionary （dictionaries）,dataset Be similar to Numpy In the array （arrays）.

Suppose someone sends you a HDF5 file , mytestfile.hdf5（ How to create this file , Please refer to ： appendix ： Create a file ）. The first thing you need to do is open this file to read data ：

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

Copy

This File Object is your starting point . So what is stored in this file ？ remember ,h5py.File Like a Python Dictionaries , So we can check these key values ,

>>> list(f.keys())
['mydataset']

Copy

According to our observations , There is a... In this file dataset, namely mydataset. Let's take this dataset As Dataset Object to verify

>>> dset = f['mydataset']

Copy

The object we get is not an array , It is a HDF5 dataset. It's like Numpy The data in ,datasets There is shape （shape） And data types （data type）

>>> dset.shape
(100,)
>>> dset.dtype
dtype('int32')

Copy

They also support array style slicing . Here is how you complete one of the files dataset Reading and writing methods

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

Copy

Want more references , Please go to File Objects and Datasets.

appendix ： Create a file

At the moment , You may be curious mytestdata.hdf5 How to create . When File After object initialization , We use patterns （mode） Set to w To create a file . Other modes （mode） by a（ Used to read 、 Write 、 newly build ） and r+（ Used to read 、 Write ）. A complete File A list of patterns and their meanings can be found File object .

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

Copy

File object There are several ways that look interesting . One is create_dataset, seeing the name of a thing one thinks of its function , It is to create a given shape and data type dataset

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

Copy

File The object is the context manager , therefore , The following code can also be run

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

Copy

Groups And hierarchical structure

“HDF” yes “Hierarchical Data Format” Abbreviation . Every HDF5 All objects in the file have a name （name）, They are similar to POSIX Style layered storage , use / The separator separates

>>> dset.name
u'/mydataset'

Copy

In this system “ Folder ”（folders） Was named groups. We created File The object itself is also a group, In this case, it is root group（root group）, The name is /:

>>> f.name
u'/'

Copy

Create a child group（subgroup） You can use a clever command create_group To complete . however , We first need to read / Open file in write mode

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

Copy

all Group object , Like File object , Also have create_* Method ：

>>> dset2 = grp.create_dataset("another_dataset", (50,), dtype='f')
>>> dset2.name
u'/subgroup/another_dataset'

Copy

By the way , You don't need to manually create all the intermediate groups. It is also possible to specify a complete path

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

Copy

Groups Support most of Python Dictionary style interface . You can use entries to get （item-retrieval） To get the objects in this file ：

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

Copy

Iterate one group, Will produce the names of its members ：

>>> for name in f:
...     print name
mydataset
subgroup
subgroup2

Copy

Membership detection can also be achieved by using names ：

>>> "mydataset" in f
True
>>> "somethingelse" in f
False

Copy

You can even use the full path name ：

>>> "subgroup/another_dataset" in f
True

Copy

It also has something you are familiar with keys(), values(), items() and iter() Methods , as well as get() Method .

Because iteration one group Only its immediate members will be generated , So I want to iterate over a complete file , have access to Group Methods visit() and visititems(), They pass through a call （callable） To achieve :

>>> def printname(name):
...     print name
>>> f.visit(printname)
mydataset
subgroup
subgroup/another_dataset
subgroup2
subgroup2/dataset_three

Copy

Want more references , Please go to Groups.

attribute

HDF5 One of the best features of is that you can store metadata after the data described （metadata）. be-all groups and datasets All support the subsidiary naming of several data bits , Called attribute .（All groups and datasets support attached named bits of data called attributes.）

Properties can be accessed through attrs This proxy object to get , This will execute the dictionary interface again ：

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

Copy

Want more references , Please go to Attributes.

Publisher ： Full stack programmer stack length , Reprint please indicate the source ：https://javaforall.cn/124968.html Link to the original text ：https://javaforall.cn