This package is used to load a shared library in Python. It is basically just a thin wrapper around ctypes (for libraries that use the __cdecl or __stdcall calling convention), Python for .NET (for libraries that use Microsoft's .NET Framework, CLR) and Py4J (for Java .jar or .class files). However, the primary advantage is that it is possible to communicate with a 32-bit shared library from 64-bit Python.

Tested in Python 2.7, 3.3+. The examples provided are currently only supported in Windows and Linux. The 32-bit server has not yet been created for OSX.

MSL-LoadLib is a pure-python package, but Python for .NET depends on the .NET Common Language Runtime (CLR) on Windows and Mono Runtime on Linux and OSX and Py4J depends on having a Java Virtual Machine installed.

To install MSL-LoadLib run:

pip install msl-loadlib

Alternatively, using the MSL Package Manager run:

msl install loadlib

To install the dependencies on Linux, please follow the instructions on the prerequisites section of the documentation.

If you are loading a 64-bit library in 64-bit Python, or a 32-bit library in 32-bit Python, then you can directly load the library using LoadLibrary.

The following examples load a 64-bit library in a 64-bit Python interpreter. If you are using a 32-bit Python interpreter then replace the 64 with 32 in the filename.

Import the LoadLibrary class and the directory where the example libraries are located

>>> from msl.loadlib import LoadLibrary
>>> from msl.examples.loadlib import EXAMPLES_DIR

If the file extension is not included then a default extension, .dll (Windows) or .so (Linux), is used.

Load a C++ library and call the add function

>>> cpp = LoadLibrary(EXAMPLES_DIR + '/cpp_lib64')
>>> cpp.lib.add(1, 2)
3

Load a FORTRAN library and call the factorial function

>>> fortran = LoadLibrary(EXAMPLES_DIR + '/fortran_lib64')

With a FORTRAN library you must pass values by reference using ctypes, and, since the returned value is not of type int we must configure ctypes for a value of type double to be returned

>>> from ctypes import byref, c_int, c_double
>>> fortran.lib.factorial.restype = c_double
>>> fortran.lib.factorial(byref(c_int(37)))
1.3763753091226343e+43

Load a .NET library and call the reverse_string function, we must specify that the library type is a .NET library by passing in the 'net' argument

>>> net = LoadLibrary(EXAMPLES_DIR + '/dotnet_lib64.dll', 'net')
>>> net.lib.StringManipulation.reverse_string('abcdefghijklmnopqrstuvwxyz')
'zyxwvutsrqponmlkjihgfedcba'

Load Java byte code and call the cos function

>>> java = LoadLibrary(EXAMPLES_DIR + '/Trig.class')
>>> java.lib.Trig.cos(1.234)
0.33046510807172985

Python interacts with the Java Virtual Machine via a local network socket and therefore the connection needs to be closed when you are done using the Java library

>>> java.gateway.shutdown()

Inter-process communication is used to access a 32-bit shared library from a module that is running within a 64-bit Python interpreter. The procedure uses a client-server protocol where the client is a subclass of msl.loadlib.Client64 and the server is a subclass of msl.loadlib.Server32. See the tutorials for examples on how to implement inter-process communication.

The documentation for MSL-LoadLib can be found here.

MSL-LoadLib uses pytest for testing the source code and sphinx for creating the documentation.

Run the tests (a coverage report is generated in the htmlcov/index.html file):

python setup.py test

Build the documentation, which can be viewed by opening the docs/_build/html/index.html file:

python setup.py docs

Automatically create the API documentation from the docstrings in the source code (uses sphinx-apidoc):

python setup.py apidoc

NOTE: By default, the docs/_autosummary folder that is created by running the apidoc command is automatically generated (it will overwrite existing files). As such, it is excluded from the repository (i.e., this folder is specified in the .gitignore file). If you want to keep the files located in docs/_autosummary you can rename the folder to be, for example, docs/_api and then the changes made to the files in the docs/_api folder will be kept and will be included in the repository.