setup.py

"""
To build with coverage of Cython files
export SM_CYTHON_COVERAGE=1
python -m pip install -e .
pytest --cov=statsmodels statsmodels
coverage html
"""
from setuptools import Command, Extension, find_packages, setup
from setuptools.dist import Distribution

from collections import defaultdict
import fnmatch
import inspect
import os
from os.path import dirname, join as pjoin, relpath
from pathlib import Path
import shutil
import sys

SETUP_DIR = Path(__file__).parent.resolve()


try:
    # SM_FORCE_C is a testing shim to force setup to use C source files
    FORCE_C = int(os.environ.get("SM_FORCE_C", 0))
    if FORCE_C:
        raise ImportError("Force import error for testing")
    from Cython import Tempita
    from Cython.Build import cythonize
    from Cython.Distutils import build_ext

    HAS_CYTHON = True
except ImportError:
    from setuptools.command.build_ext import build_ext

    HAS_CYTHON = False

try:
    import numpy  # noqa: F401

    HAS_NUMPY = True
except ImportError:
    HAS_NUMPY = False

###############################################################################
# Key Values that Change Each Release
###############################################################################
# These are strictly installation requirements. Builds requirements are
# managed in pyproject.toml
INSTALL_REQUIRES = []
with open("requirements.txt", encoding="utf-8") as req:
    for line in req.readlines():
        INSTALL_REQUIRES.append(line.split("#")[0].strip())

DEVELOP_REQUIRES = []
with open("requirements-dev.txt", encoding="utf-8") as req:
    for line in req.readlines():
        DEVELOP_REQUIRES.append(line.split("#")[0].strip())

CYTHON_MIN_VER = "0.29.26"  # released 2020

EXTRAS_REQUIRE = {
    "build": ["cython>=" + CYTHON_MIN_VER],
    "develop": ["cython>=" + CYTHON_MIN_VER] + DEVELOP_REQUIRES,
    "docs": [
        "sphinx",
        "nbconvert",
        "jupyter_client",
        "ipykernel",
        "matplotlib",
        "nbformat",
        "numpydoc",
        "pandas-datareader",
    ],
}

###############################################################################
# Values that rarely change
###############################################################################
DISTNAME = "statsmodels"
DESCRIPTION = "Statistical computations and models for Python"
README = SETUP_DIR.joinpath("README.rst").read_text()
LONG_DESCRIPTION = README
MAINTAINER = "statsmodels Developers"
MAINTAINER_EMAIL = "pystatsmodels@googlegroups.com"
URL = "https://www.statsmodels.org/"
LICENSE = "BSD License"
DOWNLOAD_URL = ""
PROJECT_URLS = {
    "Bug Tracker": "https://github.com/statsmodels/statsmodels/issues",
    "Documentation": "https://www.statsmodels.org/stable/index.html",
    "Source Code": "https://github.com/statsmodels/statsmodels",
}

CLASSIFIERS = [
    "Development Status :: 4 - Beta",
    "Environment :: Console",
    "Programming Language :: Cython",
    "Programming Language :: Python :: 3.8",
    "Programming Language :: Python :: 3.9",
    "Programming Language :: Python :: 3.10",
    "Operating System :: OS Independent",
    "Intended Audience :: End Users/Desktop",
    "Intended Audience :: Developers",
    "Intended Audience :: Science/Research",
    "Natural Language :: English",
    "License :: OSI Approved :: BSD License",
    "Topic :: Office/Business :: Financial",
    "Topic :: Scientific/Engineering",
]

FILES_TO_INCLUDE_IN_PACKAGE = ["LICENSE.txt", "setup.cfg"]

FILES_COPIED_TO_PACKAGE = []
for filename in FILES_TO_INCLUDE_IN_PACKAGE:
    if os.path.exists(filename):
        dest = os.path.join("statsmodels", filename)
        shutil.copy2(filename, dest)
        FILES_COPIED_TO_PACKAGE.append(dest)

STATESPACE_RESULTS = "statsmodels.tsa.statespace.tests.results"

ADDITIONAL_PACKAGE_DATA = {
    "statsmodels": FILES_TO_INCLUDE_IN_PACKAGE,
    "statsmodels.datasets.tests": ["*.zip"],
    "statsmodels.iolib.tests.results": ["*.dta"],
    "statsmodels.stats.tests.results": ["*.json"],
    "statsmodels.tsa.stl.tests.results": ["*.csv"],
    "statsmodels.tsa.vector_ar.tests.results": ["*.npz", "*.dat"],
    "statsmodels.stats.tests": ["*.txt"],
    "statsmodels.stats.libqsturng": ["*.r", "*.txt", "*.dat"],
    "statsmodels.stats.libqsturng.tests": ["*.csv", "*.dat"],
    "statsmodels.sandbox.regression.tests": ["*.dta", "*.csv"],
    STATESPACE_RESULTS: ["*.pkl", "*.csv"],
    STATESPACE_RESULTS + ".frbny_nowcast": ["test*.mat"],
    STATESPACE_RESULTS + ".frbny_nowcast.Nowcasting.data.US": ["*.csv"],
}

##############################################################################
# Extension Building
##############################################################################
CYTHON_COVERAGE = os.environ.get("SM_CYTHON_COVERAGE", False)
CYTHON_COVERAGE = CYTHON_COVERAGE in ("1", "true", '"true"')
CYTHON_TRACE_NOGIL = str(int(CYTHON_COVERAGE))
if CYTHON_COVERAGE:
    print("Building with coverage for Cython code")
COMPILER_DIRECTIVES = {"linetrace": CYTHON_COVERAGE}
DEFINE_MACROS = [
    ("CYTHON_TRACE_NOGIL", CYTHON_TRACE_NOGIL),
    ("NPY_NO_DEPRECATED_API", "NPY_1_7_API_VERSION"),
]


exts = dict(
    _stl={"source": "statsmodels/tsa/stl/_stl.pyx"},
    _exponential_smoothers={
        "source": "statsmodels/tsa/holtwinters/_exponential_smoothers.pyx"
    },  # noqa: E501
    _ets_smooth={
        "source": "statsmodels/tsa/exponential_smoothing/_ets_smooth.pyx"
    },  # noqa: E501
    _innovations={"source": "statsmodels/tsa/_innovations.pyx"},
    _hamilton_filter={
        "source": "statsmodels/tsa/regime_switching/_hamilton_filter.pyx.in"
    },  # noqa: E501
    _kim_smoother={
        "source": "statsmodels/tsa/regime_switching/_kim_smoother.pyx.in"
    },  # noqa: E501
    _arma_innovations={
        "source": "statsmodels/tsa/innovations/_arma_innovations.pyx.in"
    },  # noqa: E501
    linbin={"source": "statsmodels/nonparametric/linbin.pyx"},
    _qn={"source": "statsmodels/robust/_qn.pyx"},
    _smoothers_lowess={
        "source": "statsmodels/nonparametric/_smoothers_lowess.pyx"
    },  # noqa: E501
)

statespace_exts = [
    "statsmodels/tsa/statespace/_initialization.pyx.in",
    "statsmodels/tsa/statespace/_representation.pyx.in",
    "statsmodels/tsa/statespace/_kalman_filter.pyx.in",
    "statsmodels/tsa/statespace/_filters/_conventional.pyx.in",
    "statsmodels/tsa/statespace/_filters/_inversions.pyx.in",
    "statsmodels/tsa/statespace/_filters/_univariate.pyx.in",
    "statsmodels/tsa/statespace/_filters/_univariate_diffuse.pyx.in",
    "statsmodels/tsa/statespace/_kalman_smoother.pyx.in",
    "statsmodels/tsa/statespace/_smoothers/_alternative.pyx.in",
    "statsmodels/tsa/statespace/_smoothers/_classical.pyx.in",
    "statsmodels/tsa/statespace/_smoothers/_conventional.pyx.in",
    "statsmodels/tsa/statespace/_smoothers/_univariate.pyx.in",
    "statsmodels/tsa/statespace/_smoothers/_univariate_diffuse.pyx.in",
    "statsmodels/tsa/statespace/_simulation_smoother.pyx.in",
    "statsmodels/tsa/statespace/_cfa_simulation_smoother.pyx.in",
    "statsmodels/tsa/statespace/_tools.pyx.in",
]


class CleanCommand(Command):
    user_options = []

    def initialize_options(self) -> None:
        pass

    def finalize_options(self) -> None:
        pass

    def run(self) -> None:
        msg = """

python setup.py clean is not supported.

Use one of:

* `git clean -xdf` to clean all untracked files
* `git clean -Xdf` to clean untracked files ignored by .gitignore
"""
        print(msg)
        sys.exit(1)


def update_extension(extension, requires_math=True):
    import numpy as np

    numpy_includes = [np.get_include()]
    extra_incl = pjoin(dirname(inspect.getfile(np.core)), "include")
    numpy_includes += [extra_incl]
    numpy_includes = list(set(numpy_includes))
    numpy_math_libs = {
        "include_dirs": [np.get_include()],
        "library_dirs": [os.path.join(np.get_include(), '..', 'lib')],
        "libraries": ["npymath"]
    }

    if not hasattr(extension, "include_dirs"):
        return
    extension.include_dirs = list(set(extension.include_dirs + numpy_includes))
    if requires_math:
        extension.include_dirs += numpy_math_libs["include_dirs"]
        extension.libraries += numpy_math_libs["libraries"]
        extension.library_dirs += numpy_math_libs["library_dirs"]


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

    def build_extensions(self):
        self._update_extensions()
        build_ext.build_extensions(self)

    def _update_extensions(self):
        for extension in self.extensions:
            requires_math = extension.name in EXT_REQUIRES_NUMPY_MATH_LIBS
            update_extension(extension, requires_math=requires_math)


cmdclass = {"clean": CleanCommand}
if not HAS_NUMPY:
    cmdclass["build_ext"] = DeferredBuildExt


def check_source(source_name):
    """Chooses C or pyx source files, and raises if C is needed but missing"""
    source_ext = ".pyx"
    if not HAS_CYTHON:
        source_name = source_name.replace(".pyx.in", ".c")
        source_name = source_name.replace(".pyx", ".c")
        source_ext = ".c"
        if not os.path.exists(source_name):
            msg = (
                "C source not found.  You must have Cython installed to "
                "build if the C source files have not been generated."
            )
            raise IOError(msg)
    return source_name, source_ext


def process_tempita(source_name):
    """Runs pyx.in files through tempita is needed"""
    if source_name.endswith("pyx.in"):
        with open(source_name, "r", encoding="utf-8") as templated:
            pyx_template = templated.read()
        pyx = Tempita.sub(pyx_template)
        pyx_filename = source_name[:-3]
        with open(pyx_filename, "w", encoding="utf-8") as pyx_file:
            pyx_file.write(pyx)
        file_stats = os.stat(source_name)
        try:
            os.utime(
                pyx_filename,
                ns=(file_stats.st_atime_ns, file_stats.st_mtime_ns),
            )
        except AttributeError:
            os.utime(pyx_filename, (file_stats.st_atime, file_stats.st_mtime))
        source_name = pyx_filename
    return source_name


EXT_REQUIRES_NUMPY_MATH_LIBS = []
extensions = []
for config in exts.values():
    uses_blas = True
    source, ext = check_source(config["source"])
    source = process_tempita(source)
    name = source.replace("/", ".").replace(ext, "")
    include_dirs = config.get("include_dirs", [])
    depends = config.get("depends", [])
    libraries = config.get("libraries", [])
    library_dirs = config.get("library_dirs", [])

    uses_numpy_libraries = config.get("numpy_libraries", False)
    if uses_blas or uses_numpy_libraries:
        EXT_REQUIRES_NUMPY_MATH_LIBS.append(name)

    ext = Extension(
        name,
        [source],
        include_dirs=include_dirs,
        depends=depends,
        libraries=libraries,
        library_dirs=library_dirs,
        define_macros=DEFINE_MACROS,
    )
    extensions.append(ext)

for source in statespace_exts:
    source, ext = check_source(source)
    source = process_tempita(source)
    name = source.replace("/", ".").replace(ext, "")

    EXT_REQUIRES_NUMPY_MATH_LIBS.append(name)
    ext = Extension(
        name,
        [source],
        include_dirs=["statsmodels/src"],
        depends=[],
        libraries=[],
        library_dirs=[],
        define_macros=DEFINE_MACROS,
    )
    extensions.append(ext)

if HAS_NUMPY:
    for extension in extensions:
        requires_math = extension.name in EXT_REQUIRES_NUMPY_MATH_LIBS
        update_extension(extension, requires_math=requires_math)

if HAS_CYTHON:
    extensions = cythonize(
        extensions,
        compiler_directives=COMPILER_DIRECTIVES,
        language_level=3,
        force=CYTHON_COVERAGE,
    )

##############################################################################
# Construct package data
##############################################################################
package_data = defaultdict(list)
filetypes = ["*.csv", "*.txt", "*.dta"]
for root, _, filenames in os.walk(
    pjoin(os.getcwd(), "statsmodels", "datasets")
):  # noqa: E501
    matches = []
    for filetype in filetypes:
        for filename in fnmatch.filter(filenames, filetype):
            matches.append(filename)
    if matches:
        package_data[".".join(relpath(root).split(os.path.sep))] = filetypes
for root, _, _ in os.walk(pjoin(os.getcwd(), "statsmodels")):
    if root.endswith("results"):
        package_data[".".join(relpath(root).split(os.path.sep))] = filetypes

for path, filetypes in ADDITIONAL_PACKAGE_DATA.items():
    package_data[path].extend(filetypes)

if os.path.exists("MANIFEST"):
    os.unlink("MANIFEST")


class BinaryDistribution(Distribution):
    def is_pure(self):
        return False


setup(
    name=DISTNAME,
    maintainer=MAINTAINER,
    ext_modules=extensions,
    maintainer_email=MAINTAINER_EMAIL,
    description=DESCRIPTION,
    license=LICENSE,
    url=URL,
    download_url=DOWNLOAD_URL,
    project_urls=PROJECT_URLS,
    long_description=LONG_DESCRIPTION,
    classifiers=CLASSIFIERS,
    platforms="any",
    cmdclass=cmdclass,
    packages=find_packages(),
    package_data=package_data,
    distclass=BinaryDistribution,
    include_package_data=False,  # True will install all files in repo
    install_requires=INSTALL_REQUIRES,
    extras_require=EXTRAS_REQUIRE,
    zip_safe=False,
    python_requires=">=3.8",
)

# Clean-up copied files
for copy in FILES_COPIED_TO_PACKAGE:
    os.unlink(copy)