hdf5matfile 0.3.1

Load data from v7.3 *.mat files.

Load data from v7.3 *.mat files. Only reading is supported, no writing.

Usage

Provides a class Hdf5Matfile and a convience function load_hdf5mat.

To load all the variables from the file, use Hdf5Matfile.load_file:

with Hdf5Matfile(filename) as file:
    data = file.load_file()

To load a specific variable, use Hdf5Matfile.load_variable:

with Hdf5Matfile(filename) as file:
    results = file.load_variable('results')

A mapping/dict-like interface is also supported:

with Hdf5Matfile(filename) as file:
    results = file['results']
    variables = file.keys()
    values = file.values()
    for var, value in file.items():
        ...

If you’re not using a context manager, make sure to close the file after you’re done:

file = Hdf5Matfile(filename)
data = file.load_file()
...
file.close()

By default, arrays are not squeezed; since MATLAB represents even scalars as 2-D arrays, this means that something you expect to be a scalar will in fact be a 1-by-1 np.ndarray. You can change this by passing squeeze=True to the constructor:

with Hdf5Matfile(filename, squeeze=True) as file:
    data = file.load_file()

Supported data types

Data type support is pretty limited; this isn’t a terribly fancy class. Supported MATLAB data types, and the Python objects or NumPy dtypes they map to:

MATLAB type	Python object	NumPy dtype
cell	np.ndarray	object
char	str	n/a
double	np.ndarray	np.double
int8	np.ndarray	np.int8
int16	np.ndarray	np.int16
int32	np.ndarray	np.int32
int64	np.ndarray	np.int64
logical	np.ndarray	np.bool8
single	np.ndarray	np.single
struct (scalar)	dict	n/a
struct (array)	np.ndarray	object (dict)
uint8	np.ndarray	np.uint8
uint16	np.ndarray	np.uint16
uint32	np.ndarray	np.uint32
uint64	np.ndarray	np.uint64