/libmatpy

C++ wrapper for libmat, MATLAB C API to Read MAT-File Data

Primary LanguagePython

libmatpy: C++ wrapper for libmat, MATLAB C API to Read MAT-File Data

NOTE: This wrapper requires locally installed Matlab (Release 2018a or later) to access the required shared libraries

TODO:

  • Complete the test script
  • Test on OSX and Linux
  • Add support for writing

Setup

  1. Download libmat.py into the appropriate place in your project folder

  2. Add path to the Matlab folder containing libmat and libmx libraries to os.environ['PATH']

    For Matlab R2019a on Windows,

    import os
matlabdir = 'C:\\Program Files\\MATLAB\\R2019a\\bin\\win64'
os.environ['PATH'] = matlabdir+os.pathsep+os.environ['PATH']

    Here, it's important to prepend the Matlab path to avoid libmat.dll to load incorrect dependent DLLs.

Alternate way

This Python module could also be included to a project by using this repository as a whole as a git submodule of your project. In that case, the submodule directory becomes Python package, and you can access the 2 classes in the module as libmatpy.matfile and libmatpy.matlab_array assuming the submodule directory is named libmatpy

Basic Use

  1. Import the module (assuming the module is placed in one of the sys.path paths):

    import libmat

    then create libmat.matfile class object

  2. Open the MAT files

    matfilename = 'mydata.mat' # assume contains variables x, y, and z
FILE = libmat.matfile(matfilename)

  3. Get the variable list

    varnames = FILE.getDir()
print(varnames) # prints ['x', 'y', 'z']

  4. libmat.matfile is a custom sequential class, so its variables can be accessed via self[key] notation. Its items are libmat.matlab_array objects and their basic attributes are obtained via mtype(), size(), shape(), and data() methods:

        var = file['x']     # libmat.matlab_array object for the stored variable x
    sz = var.size()     # number of elements (integer)
    dims = var.shape()  # array dimensions (tuple of numbers)
    mtype = var.mtype() # matlab data type (string)
    data = var.data()   # data content (depends on mtype)

    NOTE: For non-scalar data, data() returns a numpy.ndarary object. It uses the raw data pointer returned from the dynamic library. So, make sure that the originating libmat.matlab_array object remain in the memory while data is being use.