MlabPy - Mathlab frontend for Python

With this project I just tried to see whether it would be possible to run your matlab scripts on the Python engine - now you can!

Requirements

MlabPy needs the following to run:

• Python 3.4
• numpy 1.9
• ply 3.4

The following packages are also recommended:

• scipy 1.9

Installation

Either install mlabpy using pip:

pip3 install mplabpy

You can also download a packaged version. To install it, run:

python3 setup.py install

Quickstart

MlabPy can operation in three modes: module loader, stand alone and interactive.

Module loader

MlabPy can be configured to automatically import matlab-Files just as it were Python modules. To enable this feature, you have to call mlabpy.loader.enable_matlab_import before you use this feature.

You can add this call to the __init__.py of the package containing your matlab-Files. This way, you do not have to bother about when it's called. However, as the creation of the parser takes some time (see parsetab issue below), this might lead to slow downs of your program.

Stand alone

You can execute matlab-Files using the mlabpy command line tool:

mlabpy file.m

If there is a function main defined, it will be executed.

Interactive

If you do not pass any files to the command line tool, an interactive shell will be started. The shell is currently VERY rudimentary.

Fine tuning and debugging

There are several possibilities to adjust the behavior of MlabPy. Most important is the automatic module import.

Automatic module import

MlabPy comes with an extensible runtime. It mainly enables you to import the complete namespace of a module into your matlab environment. There are some runtime extensions which are required and automatically imported. Those are mlabpy.runtime.core which provides basic functionality such as data types and NumPy bindings. mlabpy.runtime.scipy is responsible for the optional SciPy bindings.

You can extend the list modules that should be imported by simply adding their full dotted name to mlabpy.autoload. If you run the command line tool, you can use the -u or --use option to add modules to that list. Additionally, you can store a comma separated list of modules in the envrionment variable MLABPY_AUTOLOAD.

Debug output

If you set mlabpy.loader.OUTPUT_TREE to True, the Python AST will be written to a .ast file next to the matlab-File.

History

I used the great work of Victor Leikehman he has done with smop as a starting point. Especially the lexer and paser modules were of great help. I changed the parser to produce and Python AST which was surprisingly straight forward.

AFter successfully parsing my first matlab files, I started to collect some "patches" in mlabpy.runtime to account for all the small differences... Ideally, this wrapper layer should be replaced by patching the AST.

TODO

• Parse tab is generated every startup, very slow!
• Non-conforming syntax trees and grammar defects.
  • Copy-on-assign?
  • Slices.
• Some glitches in the docs.