Attributes in HDF5

Attributes in HDF5 enables the dataset to be self descriptive and makes HDF5 suitable for any kind of data storage. The attrs Python attribute of H5Py classes Group and Dataset holds the attributes. attrs provide a dictionary like interface.

H5py Updated 13 Dec 2019 1 min read

Overview

  • Attributes in HDF5 allow datasets to be self-descriptive.
  • Any metadata that describe the datasets and groups can be attached to groups and datasets of HDF5 through attributes.
  • In h5py, both the Group and Dataset objects have the python attribute attrs through which attributes can be stored.
  • The attrs is an instance of AttributeManager. AttributeManager provides dictionary like access for reading and writing attributes.

Example

python
# Example Python program that adds attributes to a HDF5 group and a HDF5 dataset

import h5py

import numpy.random

# Create a HDF5 file

fileOpenMode    = "w";

hdfFileName     = "WithAttributes.HDF5";

hdf5File        = h5py.File(hdfFileName, fileOpenMode);

# Create a group at root level

aGroup  = hdf5File.create_group("/A");

datasetShape = (24,);

aDataset = aGroup.create_dataset("/A/D", datasetShape);

# Add

aDataset.attrs["SensorName"] = "FluidVelocity-X1Y5";

aDataset.attrs["SensorType"] = "Differential";

aDataset.attrs["SensorLocation"] = "40.71, -74.00";

# Assign hourly readings

for i in range(0, 24):

    aDataset[i] = numpy.random.uniform(80, 100, 1)[0];

# Print dataset along with attributes

print("(Attribute, Value:)");

for item in aDataset.attrs.items():

    print(item);

for i in range(0, aDataset.shape[0]):

    print(aDataset[i]);

Output

Output
(Attribute, Value:)

('SensorLocation', '40.71, -74.00')

('SensorName', 'FluidVelocity-X1Y5')

('SensorType', 'Differential')

86.06291

89.932724

97.62169

91.05482

85.80139

90.10574

95.83731

86.68407

94.6496

92.08971

81.00686

86.006996

97.08245

84.83142

82.231895

99.718704

90.461914

86.364334

81.27478

93.38981

91.627525

89.78677

84.98252

85.19031