from bdpy import BData

# Create an empty BData instance
bdata = BData()

# Load BData from a file
bdata = BData('data_file.h5')

# Load BData from 'data_file.h5'
bdata.load('data_file.h5')

# Show 'key' and 'description' of metadata
bdata.show_meatadata()

# Get 'value' of the metadata specified by 'key'
voxel_x = bdata.get_metadata('voxel_x', where='VoxelData')

# Get an array of voxel data in V1
data_v1 = bdata.select('ROI_V1')  # shape=(M, num voxels in V1)

# `select` accepts some operators
data_v1v2 = bdata.select('ROI_V1 + ROI_V2')
data_hvc = bdata.select('ROI_LOC + ROI_FFA + ROI_PPA - LOC_LVC')

# Wildcard
data_visual = data.select('ROI_V*')

# Get labels ('image_index') in the dataset
label_a  = bdata.select('image_index')

# Add new data
x = numpy.random.rand(bdata.dataset.shape[0])
bdata.add(x, 'random_data')

# Set description of metadata
bdata.set_metadatadescription('random_data', 'Random data')

# Save data
bdata.save('output_file.h5')  # File format is selected automatically by extension. .mat, .h5,and .npy are supported.