setup.py

#!/usr/bin/env python

"""
setup.py file
"""

# import sys
import sysconfig
from setuptools import setup
from setuptools.extension import Extension
from setuptools.command.build_ext import build_ext

libraries = ['gsl', 'gslcblas']

if sysconfig.get_config_var("LIBM") == "-lm":
    libraries.append("m")

include_dirs = ['/usr/local/include', '/usr/include', 'c/']

library_dirs = ['/usr/local/lib', 'usr/lib']

sources = ['c/pysbrl.c', 'c/train.c', 'c/rulelib.c', 'c/save_load.c',
           'c/bit_vector.c', 'c/utils.c', 'swig/sbrl_wrap.c']

pysbrl_module = Extension('_sbrl',
                          sources=sources,
                          include_dirs=include_dirs,
                          libraries=libraries,
                          library_dirs=library_dirs,
                          swig_opts=['-keyword'],
                          # extra_link_args=["-shared"],
                          extra_compile_args=['-std=c99']
                          )


class CustomBuildExtCommand(build_ext):
    """build_ext command for use when numpy headers are needed."""

    def run(self):
        # Import numpy here, only when headers are needed
        import numpy

        # Add numpy headers to include_dirs
        self.include_dirs.append(numpy.get_include())

        # Call original build_ext command
        build_ext.run(self)


setup(
    name='pysbrl',
    version='0.4.2',
    author="Yao, Ming",
    description="""A python interface of Scalable Bayesian Rule List""",
    long_description="PySBRL is a python package that allows you to create rule lists from categorical data",
    ext_modules=[pysbrl_module],
    url='https://github.com/myaooo/pysbrl',
    packages=["pysbrl"],
    # requires=['numpy'],
    cmdclass={'build_ext': CustomBuildExtCommand},
    install_requires=[
        'numpy>=1.13'
        'fim',
    ],
    dependency_links=[
        'https://github.com/myaooo/pyfim-clone/tarball/master#egg=fim-6.28'
    ]
    #    extra_link_args = ["-bundle"],
)