pytorch/torch/utils/cpp_extension.py

from __future__ import absolute_import, division, print_function, unicode_literals
import copy
import glob
import imp
import os
import re
import setuptools
import subprocess
import sys
import sysconfig
import tempfile
import warnings

import torch
from .file_baton import FileBaton
from ._cpp_extension_versioner import ExtensionVersioner

from setuptools.command.build_ext import build_ext


IS_WINDOWS = sys.platform == 'win32'


def _find_cuda_home():
    '''Finds the CUDA install path.'''
    # Guess #1
    cuda_home = os.environ.get('CUDA_HOME') or os.environ.get('CUDA_PATH')
    if cuda_home is None:
        # Guess #2
        if IS_WINDOWS:
            cuda_homes = glob.glob(
                'C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v*.*')
            if len(cuda_homes) == 0:
                cuda_home = ''
            else:
                cuda_home = cuda_homes[0]
        else:
            cuda_home = '/usr/local/cuda'
        if not os.path.exists(cuda_home):
            # Guess #3
            try:
                which = 'where' if IS_WINDOWS else 'which'
                nvcc = subprocess.check_output(
                    [which, 'nvcc']).decode().rstrip('\r\n')
                cuda_home = os.path.dirname(os.path.dirname(nvcc))
            except Exception:
                cuda_home = None
    if cuda_home and not torch.cuda.is_available():
        print("No CUDA runtime is found, using CUDA_HOME='{}'".format(cuda_home))
    return cuda_home


MINIMUM_GCC_VERSION = (4, 9, 0)
MINIMUM_MSVC_VERSION = (19, 0, 24215)
ABI_INCOMPATIBILITY_WARNING = '''

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler ({}) may be ABI-incompatible with PyTorch!
Please use a compiler that is ABI-compatible with GCC 4.9 and above.
See https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html.

See https://gist.github.com/goldsborough/d466f43e8ffc948ff92de7486c5216d6
for instructions on how to install GCC 4.9 or higher.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!
'''
WRONG_COMPILER_WARNING = '''

                               !! WARNING !!

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Your compiler ({user_compiler}) is not compatible with the compiler Pytorch was
built with for this platform, which is {pytorch_compiler} on {platform}. Please
use {pytorch_compiler} to to compile your extension. Alternatively, you may
compile PyTorch from source using {user_compiler}, and then you can also use
{user_compiler} to compile your extension.

See https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md for help
with compiling PyTorch from source.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

                              !! WARNING !!
'''
ACCEPTED_COMPILERS_FOR_PLATFORM = {'darwin': ['clang++', 'clang'], 'linux': ['g++', 'gcc']}
CUDA_HOME = _find_cuda_home()
CUDNN_HOME = os.environ.get('CUDNN_HOME') or os.environ.get('CUDNN_PATH')
# PyTorch releases have the version pattern major.minor.patch, whereas when
# PyTorch is built from source, we append the git commit hash, which gives
# it the below pattern.
BUILT_FROM_SOURCE_VERSION_PATTERN = re.compile(r'\d+\.\d+\.\d+\w+\+\w+')

COMMON_NVCC_FLAGS = [
    '-D__CUDA_NO_HALF_OPERATORS__',
    '-D__CUDA_NO_HALF_CONVERSIONS__',
    '-D__CUDA_NO_HALF2_OPERATORS__',
]


JIT_EXTENSION_VERSIONER = ExtensionVersioner()


def _is_binary_build():
    return not BUILT_FROM_SOURCE_VERSION_PATTERN.match(torch.version.__version__)


def get_default_build_root():
    '''
    Returns the path to the root folder under which extensions will built.

    For each extension module built, there will be one folder underneath the
    folder returned by this function. For example, if ``p`` is the path
    returned by this function and ``ext`` the name of an extension, the build
    folder for the extension will be ``p/ext``.
    '''
    # tempfile.gettempdir() will be /tmp on UNIX and \TEMP on Windows.
    return os.path.realpath(os.path.join(tempfile.gettempdir(), 'torch_extensions'))


def check_compiler_ok_for_platform(compiler):
    '''
    Verifies that the compiler is the expected one for the current platform.

    Arguments:
        compiler (str): The compiler executable to check.

    Returns:
        True if the compiler is gcc/g++ on Linux or clang/clang++ on macOS,
        and always True for Windows.
    '''
    if IS_WINDOWS:
        return True
    which = subprocess.check_output(['which', compiler], stderr=subprocess.STDOUT)
    # Use os.path.realpath to resolve any symlinks, in particular from 'c++' to e.g. 'g++'.
    compiler_path = os.path.realpath(which.decode().strip())
    accepted_compilers = ACCEPTED_COMPILERS_FOR_PLATFORM[sys.platform]
    return any(name in compiler_path for name in accepted_compilers)


def check_compiler_abi_compatibility(compiler):
    '''
    Verifies that the given compiler is ABI-compatible with PyTorch.

    Arguments:
        compiler (str): The compiler executable name to check (e.g. ``g++``).
            Must be executable in a shell process.

    Returns:
        False if the compiler is (likely) ABI-incompatible with PyTorch,
        else True.
    '''
    if not _is_binary_build():
        return True
    if os.environ.get('TORCH_DONT_CHECK_COMPILER_ABI') in ['ON', '1', 'YES', 'TRUE', 'Y']:
        return True

    # First check if the compiler is one of the expected ones for the particular platform.
    if not check_compiler_ok_for_platform(compiler):
        warnings.warn(WRONG_COMPILER_WARNING.format(
            user_compiler=compiler,
            pytorch_compiler=ACCEPTED_COMPILERS_FOR_PLATFORM[sys.platform][0],
            platform=sys.platform))
        return False

    if sys.platform == 'darwin':
        # There is no particular minimum version we need for clang, so we're good here.
        return True
    try:
        if sys.platform == 'linux':
            minimum_required_version = MINIMUM_GCC_VERSION
            version = subprocess.check_output([compiler, '-dumpfullversion', '-dumpversion'])
            version = version.split('.')
        else:
            minimum_required_version = MINIMUM_MSVC_VERSION
            compiler_info = subprocess.check_output(compiler, stderr=subprocess.STDOUT)
            match = re.search(r'(\d+)\.(\d+)\.(\d+)', compiler_info)
            version = (0, 0, 0) if match is None else match.groups()
    except Exception:
        _, error, _ = sys.exc_info()
        warnings.warn('Error checking compiler version for {}: {}'.format(compiler, error))
        return False

    if tuple(map(int, version)) >= minimum_required_version:
        return True

    compiler = '{} {}'.format(compiler, version.group(0))
    warnings.warn(ABI_INCOMPATIBILITY_WARNING.format(compiler))

    return False


class BuildExtension(build_ext):
    '''
    A custom :mod:`setuptools` build extension .

    This :class:`setuptools.build_ext` subclass takes care of passing the
    minimum required compiler flags (e.g. ``-std=c++11``) as well as mixed
    C++/CUDA compilation (and support for CUDA files in general).

    When using :class:`BuildExtension`, it is allowed to supply a dictionary
    for ``extra_compile_args`` (rather than the usual list) that maps from
    languages (``cxx`` or ``cuda``) to a list of additional compiler flags to
    supply to the compiler. This makes it possible to supply different flags to
    the C++ and CUDA compiler during mixed compilation.
    '''

    def build_extensions(self):
        self._check_abi()
        for extension in self.extensions:
            self._add_compile_flag(extension, '-DTORCH_API_INCLUDE_EXTENSION_H')
            self._define_torch_extension_name(extension)
            self._add_gnu_abi_flag_if_binary(extension)

        # Register .cu and .cuh as valid source extensions.
        self.compiler.src_extensions += ['.cu', '.cuh']
        # Save the original _compile method for later.
        if self.compiler.compiler_type == 'msvc':
            self.compiler._cpp_extensions += ['.cu', '.cuh']
            original_compile = self.compiler.compile
            original_spawn = self.compiler.spawn
        else:
            original_compile = self.compiler._compile

        def unix_wrap_compile(obj, src, ext, cc_args, extra_postargs, pp_opts):
            # Copy before we make any modifications.
            cflags = copy.deepcopy(extra_postargs)
            try:
                original_compiler = self.compiler.compiler_so
                if _is_cuda_file(src):
                    nvcc = _join_cuda_home('bin', 'nvcc')
                    if not isinstance(nvcc, list):
                        nvcc = [nvcc]
                    self.compiler.set_executable('compiler_so', nvcc)
                    if isinstance(cflags, dict):
                        cflags = cflags['nvcc']
                    cflags = COMMON_NVCC_FLAGS + ['--compiler-options', "'-fPIC'"] + cflags
                elif isinstance(cflags, dict):
                    cflags = cflags['cxx']
                # NVCC does not allow multiple -std to be passed, so we avoid
                # overriding the option if the user explicitly passed it.
                if not any(flag.startswith('-std=') for flag in cflags):
                    cflags.append('-std=c++11')

                original_compile(obj, src, ext, cc_args, cflags, pp_opts)
            finally:
                # Put the original compiler back in place.
                self.compiler.set_executable('compiler_so', original_compiler)

        def win_wrap_compile(sources,
                             output_dir=None,
                             macros=None,
                             include_dirs=None,
                             debug=0,
                             extra_preargs=None,
                             extra_postargs=None,
                             depends=None):

            self.cflags = copy.deepcopy(extra_postargs)
            extra_postargs = None

            def spawn(cmd):
                orig_cmd = cmd
                # Using regex to match src, obj and include files

                src_regex = re.compile('/T(p|c)(.*)')
                src_list = [
                    m.group(2) for m in (src_regex.match(elem) for elem in cmd)
                    if m
                ]

                obj_regex = re.compile('/Fo(.*)')
                obj_list = [
                    m.group(1) for m in (obj_regex.match(elem) for elem in cmd)
                    if m
                ]

                include_regex = re.compile(r'((\-|\/)I.*)')
                include_list = [
                    m.group(1)
                    for m in (include_regex.match(elem) for elem in cmd) if m
                ]

                if len(src_list) >= 1 and len(obj_list) >= 1:
                    src = src_list[0]
                    obj = obj_list[0]
                    if _is_cuda_file(src):
                        nvcc = _join_cuda_home('bin', 'nvcc')
                        if isinstance(self.cflags, dict):
                            cflags = self.cflags['nvcc']
                        elif isinstance(self.cflags, list):
                            cflags = self.cflags
                        else:
                            cflags = []
                        cmd = [
                            nvcc, '-c', src, '-o', obj, '-Xcompiler',
                            '/wd4819', '-Xcompiler', '/MD'
                        ] + include_list + cflags
                    elif isinstance(self.cflags, dict):
                        cflags = self.cflags['cxx']
                        cmd += cflags
                    elif isinstance(self.cflags, list):
                        cflags = self.cflags
                        cmd += cflags

                return original_spawn(cmd)

            try:
                self.compiler.spawn = spawn
                return original_compile(sources, output_dir, macros,
                                        include_dirs, debug, extra_preargs,
                                        extra_postargs, depends)
            finally:
                self.compiler.spawn = original_spawn

        # Monkey-patch the _compile method.
        if self.compiler.compiler_type == 'msvc':
            self.compiler.compile = win_wrap_compile
        else:
            self.compiler._compile = unix_wrap_compile

        build_ext.build_extensions(self)

    def _check_abi(self):
        # On some platforms, like Windows, compiler_cxx is not available.
        if hasattr(self.compiler, 'compiler_cxx'):
            compiler = self.compiler.compiler_cxx[0]
        elif IS_WINDOWS:
            compiler = os.environ.get('CXX', 'cl')
        else:
            compiler = os.environ.get('CXX', 'c++')
        check_compiler_abi_compatibility(compiler)

    def _add_compile_flag(self, extension, flag):
        if isinstance(extension.extra_compile_args, dict):
            for args in extension.extra_compile_args.values():
                args.append(flag)
        else:
            extension.extra_compile_args.append(flag)

    def _define_torch_extension_name(self, extension):
        # pybind11 doesn't support dots in the names
        # so in order to support extensions in the packages
        # like torch._C, we take the last part of the string
        # as the library name
        names = extension.name.split('.')
        name = names[-1]
        define = '-DTORCH_EXTENSION_NAME={}'.format(name)
        self._add_compile_flag(extension, define)

    def _add_gnu_abi_flag_if_binary(self, extension):
        # If the version string looks like a binary build,
        # we know that PyTorch was compiled with gcc 4.9.2.
        # if the extension is compiled with gcc >= 5.1,
        # then we have to define _GLIBCXX_USE_CXX11_ABI=0
        # so that the std::string in the API is resolved to
        # non-C++11 symbols
        if _is_binary_build():
            self._add_compile_flag(extension, '-D_GLIBCXX_USE_CXX11_ABI=0')


def CppExtension(name, sources, *args, **kwargs):
    '''
    Creates a :class:`setuptools.Extension` for C++.

    Convenience method that creates a :class:`setuptools.Extension` with the
    bare minimum (but often sufficient) arguments to build a C++ extension.

    All arguments are forwarded to the :class:`setuptools.Extension`
    constructor.

    Example:
        >>> from setuptools import setup
        >>> from torch.utils.cpp_extension import BuildExtension, CppExtension
        >>> setup(
                name='extension',
                ext_modules=[
                    CppExtension(
                        name='extension',
                        sources=['extension.cpp'],
                        extra_compile_args=['-g'])),
                ],
                cmdclass={
                    'build_ext': BuildExtension
                })
    '''
    include_dirs = kwargs.get('include_dirs', [])
    include_dirs += include_paths()
    kwargs['include_dirs'] = include_dirs

    if IS_WINDOWS:
        library_dirs = kwargs.get('library_dirs', [])
        library_dirs += library_paths()
        kwargs['library_dirs'] = library_dirs

        libraries = kwargs.get('libraries', [])
        libraries.append('c10')
        libraries.append('caffe2')
        libraries.append('torch')
        libraries.append('_C')
        kwargs['libraries'] = libraries

    kwargs['language'] = 'c++'
    return setuptools.Extension(name, sources, *args, **kwargs)


def CUDAExtension(name, sources, *args, **kwargs):
    '''
    Creates a :class:`setuptools.Extension` for CUDA/C++.

    Convenience method that creates a :class:`setuptools.Extension` with the
    bare minimum (but often sufficient) arguments to build a CUDA/C++
    extension. This includes the CUDA include path, library path and runtime
    library.

    All arguments are forwarded to the :class:`setuptools.Extension`
    constructor.

    Example:
        >>> from setuptools import setup
        >>> from torch.utils.cpp_extension import BuildExtension, CUDAExtension
        >>> setup(
                name='cuda_extension',
                ext_modules=[
                    CUDAExtension(
                            name='cuda_extension',
                            sources=['extension.cpp', 'extension_kernel.cu'],
                            extra_compile_args={'cxx': ['-g'],
                                                'nvcc': ['-O2']})
                ],
                cmdclass={
                    'build_ext': BuildExtension
                })
    '''
    library_dirs = kwargs.get('library_dirs', [])
    library_dirs += library_paths(cuda=True)
    kwargs['library_dirs'] = library_dirs

    libraries = kwargs.get('libraries', [])
    libraries.append('cudart')
    if IS_WINDOWS:
        libraries.append('c10')
        libraries.append('caffe2')
        libraries.append('torch')
        libraries.append('caffe2_gpu')
        libraries.append('_C')
    kwargs['libraries'] = libraries

    include_dirs = kwargs.get('include_dirs', [])
    include_dirs += include_paths(cuda=True)
    kwargs['include_dirs'] = include_dirs

    kwargs['language'] = 'c++'

    return setuptools.Extension(name, sources, *args, **kwargs)


def include_paths(cuda=False):
    '''
    Get the include paths required to build a C++ or CUDA extension.

    Args:
        cuda: If `True`, includes CUDA-specific include paths.

    Returns:
        A list of include path strings.
    '''
    here = os.path.abspath(__file__)
    torch_path = os.path.dirname(os.path.dirname(here))
    lib_include = os.path.join(torch_path, 'lib', 'include')
    paths = [
        lib_include,
        # Remove this once torch/torch.h is officially no longer supported for C++ extensions.
        os.path.join(lib_include, 'torch', 'csrc', 'api', 'include'),
        # Some internal (old) Torch headers don't properly prefix their includes,
        # so we need to pass -Itorch/lib/include/TH as well.
        os.path.join(lib_include, 'TH'),
        os.path.join(lib_include, 'THC')
    ]
    if cuda:
        paths.append(_join_cuda_home('include'))
        if CUDNN_HOME is not None:
            paths.append(os.path.join(CUDNN_HOME, 'include'))
    return paths


def library_paths(cuda=False):
    '''
    Get the library paths required to build a C++ or CUDA extension.

    Args:
        cuda: If `True`, includes CUDA-specific library paths.

    Returns:
        A list of library path strings.
    '''
    paths = []

    if IS_WINDOWS:
        here = os.path.abspath(__file__)
        torch_path = os.path.dirname(os.path.dirname(here))
        lib_path = os.path.join(torch_path, 'lib')

        paths.append(lib_path)

    if cuda:
        lib_dir = 'lib/x64' if IS_WINDOWS else 'lib64'
        paths.append(_join_cuda_home(lib_dir))
        if CUDNN_HOME is not None:
            paths.append(os.path.join(CUDNN_HOME, lib_dir))
    return paths


def load(name,
         sources,
         extra_cflags=None,
         extra_cuda_cflags=None,
         extra_ldflags=None,
         extra_include_paths=None,
         build_directory=None,
         verbose=False,
         with_cuda=None):
    '''
    Loads a PyTorch C++ extension just-in-time (JIT).

    To load an extension, a Ninja build file is emitted, which is used to
    compile the given sources into a dynamic library. This library is
    subsequently loaded into the current Python process as a module and
    returned from this function, ready for use.

    By default, the directory to which the build file is emitted and the
    resulting library compiled to is ``<tmp>/torch_extensions/<name>``, where
    ``<tmp>`` is the temporary folder on the current platform and ``<name>``
    the name of the extension. This location can be overridden in two ways.
    First, if the ``TORCH_EXTENSIONS_DIR`` environment variable is set, it
    replaces ``<tmp>/torch_extensions`` and all extensions will be compiled
    into subfolders of this directory. Second, if the ``build_directory``
    argument to this function is supplied, it overrides the entire path, i.e.
    the library will be compiled into that folder directly.

    To compile the sources, the default system compiler (``c++``) is used,
    which can be overridden by setting the ``CXX`` environment variable. To pass
    additional arguments to the compilation process, ``extra_cflags`` or
    ``extra_ldflags`` can be provided. For example, to compile your extension
    with optimizations, pass ``extra_cflags=['-O3']``. You can also use
    ``extra_cflags`` to pass further include directories.

    CUDA support with mixed compilation is provided. Simply pass CUDA source
    files (``.cu`` or ``.cuh``) along with other sources. Such files will be
    detected and compiled with nvcc rather than the C++ compiler. This includes
    passing the CUDA lib64 directory as a library directory, and linking
    ``cudart``. You can pass additional flags to nvcc via
    ``extra_cuda_cflags``, just like with ``extra_cflags`` for C++. Various
    heuristics for finding the CUDA install directory are used, which usually
    work fine. If not, setting the ``CUDA_HOME`` environment variable is the
    safest option.

    Args:
        name: The name of the extension to build. This MUST be the same as the
            name of the pybind11 module!
        sources: A list of relative or absolute paths to C++ source files.
        extra_cflags: optional list of compiler flags to forward to the build.
        extra_cuda_cflags: optional list of compiler flags to forward to nvcc
            when building CUDA sources.
        extra_ldflags: optional list of linker flags to forward to the build.
        extra_include_paths: optional list of include directories to forward
            to the build.
        build_directory: optional path to use as build workspace.
        verbose: If ``True``, turns on verbose logging of load steps.
        with_cuda: Determines whether CUDA headers and libraries are added to
            the build. If set to ``None`` (default), this value is
            automatically determined based on the existence of ``.cu`` or
            ``.cuh`` in ``sources``. Set it to `True`` to force CUDA headers
            and libraries to be included.

    Returns:
        The loaded PyTorch extension as a Python module.

    Example:
        >>> from torch.utils.cpp_extension import load
        >>> module = load(
                name='extension',
                sources=['extension.cpp', 'extension_kernel.cu'],
                extra_cflags=['-O2'],
                verbose=True)
    '''
    return _jit_compile(
        name,
        [sources] if isinstance(sources, str) else sources,
        extra_cflags,
        extra_cuda_cflags,
        extra_ldflags,
        extra_include_paths,
        build_directory or _get_build_directory(name, verbose),
        verbose,
        with_cuda)


def load_inline(name,
                cpp_sources,
                cuda_sources=None,
                functions=None,
                extra_cflags=None,
                extra_cuda_cflags=None,
                extra_ldflags=None,
                extra_include_paths=None,
                build_directory=None,
                verbose=False,
                with_cuda=None):
    '''
    Loads a PyTorch C++ extension just-in-time (JIT) from string sources.

    This function behaves exactly like :func:`load`, but takes its sources as
    strings rather than filenames. These strings are stored to files in the
    build directory, after which the behavior of :func:`load_inline` is
    identical to :func:`load`.

    See `the
    tests <https://github.com/pytorch/pytorch/blob/master/test/test_cpp_extensions.py>`_
    for good examples of using this function.

    Sources may omit two required parts of a typical non-inline C++ extension:
    the necessary header includes, as well as the (pybind11) binding code. More
    precisely, strings passed to ``cpp_sources`` are first concatenated into a
    single ``.cpp`` file. This file is then prepended with ``#include
    <torch/extension.h>``.

    Furthermore, if the ``functions`` argument is supplied, bindings will be
    automatically generated for each function specified. ``functions`` can
    either be a list of function names, or a dictionary mapping from function
    names to docstrings. If a list is given, the name of each function is used
    as its docstring.

    The sources in ``cuda_sources`` are concatenated into a separate ``.cu``
    file and  prepended with ``torch/types.h``, ``cuda.h`` and
    ``cuda_runtime.h`` includes. The ``.cpp`` and ``.cu`` files are compiled
    separately, but ultimately linked into a single library. Note that no
    bindings are generated for functions in ``cuda_sources`` per  se. To bind
    to a CUDA kernel, you must create a C++ function that calls it, and either
    declare or define this C++ function in one of the ``cpp_sources`` (and
    include its name in ``functions``).

    See :func:`load` for a description of arguments omitted below.

    Args:
        cpp_sources: A string, or list of strings, containing C++ source code.
        cuda_sources: A string, or list of strings, containing CUDA source code.
        functions: A list of function names for which to generate function
            bindings. If a dictionary is given, it should map function names to
            docstrings (which are otherwise just the function names).
        with_cuda: Determines whether CUDA headers and libraries are added to
            the build. If set to ``None`` (default), this value is
            automatically determined based on whether ``cuda_sources`` is
            provided. Set it to `True`` to force CUDA headers
            and libraries to be included.

    Example:
        >>> from torch.utils.cpp_extension import load_inline
        >>> source = \'\'\'
        at::Tensor sin_add(at::Tensor x, at::Tensor y) {
          return x.sin() + y.sin();
        }
        \'\'\'
        >>> module = load_inline(name='inline_extension',
                                 cpp_sources=[source],
                                 functions=['sin_add'])
    '''
    build_directory = build_directory or _get_build_directory(name, verbose)

    if isinstance(cpp_sources, str):
        cpp_sources = [cpp_sources]
    cuda_sources = cuda_sources or []
    if isinstance(cuda_sources, str):
        cuda_sources = [cuda_sources]

    cpp_sources.insert(0, '#include <torch/extension.h>')

    # If `functions` is supplied, we create the pybind11 bindings for the user.
    # Here, `functions` is (or becomes, after some processing) a map from
    # function names to function docstrings.
    if functions is not None:
        cpp_sources.append('PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {')
        if isinstance(functions, str):
            functions = [functions]
        if isinstance(functions, list):
            # Make the function docstring the same as the function name.
            functions = dict((f, f) for f in functions)
        elif not isinstance(functions, dict):
            raise ValueError(
                "Expected 'functions' to be a list or dict, but was {}".format(
                    type(functions)))
        for function_name, docstring in functions.items():
            cpp_sources.append('m.def("{0}", &{0}, "{1}");'.format(
                function_name, docstring))
        cpp_sources.append('}')

    cpp_source_path = os.path.join(build_directory, 'main.cpp')
    with open(cpp_source_path, 'w') as cpp_source_file:
        cpp_source_file.write('\n'.join(cpp_sources))

    sources = [cpp_source_path]

    if cuda_sources:
        cuda_sources.insert(0, '#include <torch/types.h>')
        cuda_sources.insert(1, '#include <cuda.h>')
        cuda_sources.insert(2, '#include <cuda_runtime.h>')

        cuda_source_path = os.path.join(build_directory, 'cuda.cu')
        with open(cuda_source_path, 'w') as cuda_source_file:
            cuda_source_file.write('\n'.join(cuda_sources))

        sources.append(cuda_source_path)

    return _jit_compile(
        name,
        sources,
        extra_cflags,
        extra_cuda_cflags,
        extra_ldflags,
        extra_include_paths,
        build_directory,
        verbose,
        with_cuda)


def _jit_compile(name,
                 sources,
                 extra_cflags,
                 extra_cuda_cflags,
                 extra_ldflags,
                 extra_include_paths,
                 build_directory,
                 verbose,
                 with_cuda=None):
    old_version = JIT_EXTENSION_VERSIONER.get_version(name)
    version = JIT_EXTENSION_VERSIONER.bump_version_if_changed(
        name,
        sources,
        build_arguments=[extra_cflags, extra_cuda_cflags, extra_ldflags, extra_include_paths],
        build_directory=build_directory,
        with_cuda=with_cuda
    )
    if version > 0:
        if version != old_version and verbose:
            print('The input conditions for extension module {} have changed. '.format(name) +
                  'Bumping to version {0} and re-building as {1}_v{0}...'.format(version, name))
        name = '{}_v{}'.format(name, version)

    if version != old_version:
        baton = FileBaton(os.path.join(build_directory, 'lock'))
        if baton.try_acquire():
            try:
                _write_ninja_file_and_build(
                    name=name,
                    sources=sources,
                    extra_cflags=extra_cflags or [],
                    extra_cuda_cflags=extra_cuda_cflags or [],
                    extra_ldflags=extra_ldflags or [],
                    extra_include_paths=extra_include_paths or [],
                    build_directory=build_directory,
                    verbose=verbose,
                    with_cuda=with_cuda)
            finally:
                baton.release()
        else:
            baton.wait()
    elif verbose:
        print('No modifications detected for re-loaded extension '
              'module {}, skipping build step...'.format(name))

    if verbose:
        print('Loading extension module {}...'.format(name))
    return _import_module_from_library(name, build_directory)


def _write_ninja_file_and_build(name,
                                sources,
                                extra_cflags,
                                extra_cuda_cflags,
                                extra_ldflags,
                                extra_include_paths,
                                build_directory,
                                verbose,
                                with_cuda):
    verify_ninja_availability()
    check_compiler_abi_compatibility(os.environ.get('CXX', 'c++'))
    if with_cuda is None:
        with_cuda = any(map(_is_cuda_file, sources))
    extra_ldflags = _prepare_ldflags(
        extra_ldflags or [],
        with_cuda,
        verbose)
    build_file_path = os.path.join(build_directory, 'build.ninja')
    if verbose:
        print(
            'Emitting ninja build file {}...'.format(build_file_path))
    # NOTE: Emitting a new ninja build file does not cause re-compilation if
    # the sources did not change, so it's ok to re-emit (and it's fast).
    _write_ninja_file(
        path=build_file_path,
        name=name,
        sources=sources,
        extra_cflags=extra_cflags or [],
        extra_cuda_cflags=extra_cuda_cflags or [],
        extra_ldflags=extra_ldflags or [],
        extra_include_paths=extra_include_paths or [],
        with_cuda=with_cuda)

    if verbose:
        print('Building extension module {}...'.format(name))
    _build_extension_module(name, build_directory, verbose)


def verify_ninja_availability():
    '''
    Returns ``True`` if the `ninja <https://ninja-build.org/>`_ build system is
    available on the system.
    '''
    with open(os.devnull, 'wb') as devnull:
        try:
            subprocess.check_call('ninja --version'.split(), stdout=devnull)
        except OSError:
            raise RuntimeError("Ninja is required to load C++ extensions")
        else:
            return True


def _prepare_ldflags(extra_ldflags, with_cuda, verbose):
    if IS_WINDOWS:
        python_path = os.path.dirname(sys.executable)
        python_lib_path = os.path.join(python_path, 'libs')

        here = os.path.abspath(__file__)
        torch_path = os.path.dirname(os.path.dirname(here))
        lib_path = os.path.join(torch_path, 'lib')

        extra_ldflags.append('c10.lib')
        extra_ldflags.append('caffe2.lib')
        extra_ldflags.append('torch.lib')
        if with_cuda:
            extra_ldflags.append('caffe2_gpu.lib')
        extra_ldflags.append('_C.lib')
        extra_ldflags.append('/LIBPATH:{}'.format(python_lib_path))
        extra_ldflags.append('/LIBPATH:{}'.format(lib_path))

    if with_cuda:
        if verbose:
            print('Detected CUDA files, patching ldflags')
        if IS_WINDOWS:
            extra_ldflags.append('/LIBPATH:{}'.format(
                _join_cuda_home('lib/x64')))
            extra_ldflags.append('cudart.lib')
            if CUDNN_HOME is not None:
                extra_ldflags.append(os.path.join(CUDNN_HOME, 'lib/x64'))
        else:
            extra_ldflags.append('-L{}'.format(_join_cuda_home('lib64')))
            extra_ldflags.append('-lcudart')
            if CUDNN_HOME is not None:
                extra_ldflags.append('-L{}'.format(os.path.join(CUDNN_HOME, 'lib64')))

    return extra_ldflags


def _get_build_directory(name, verbose):
    root_extensions_directory = os.environ.get('TORCH_EXTENSIONS_DIR')
    if root_extensions_directory is None:
        root_extensions_directory = get_default_build_root()

    if verbose:
        print('Using {} as PyTorch extensions root...'.format(
            root_extensions_directory))

    build_directory = os.path.join(root_extensions_directory, name)
    if not os.path.exists(build_directory):
        if verbose:
            print('Creating extension directory {}...'.format(build_directory))
        # This is like mkdir -p, i.e. will also create parent directories.
        os.makedirs(build_directory)

    return build_directory


def _build_extension_module(name, build_directory, verbose):
    try:
        sys.stdout.flush()
        sys.stderr.flush()
        if sys.version_info >= (3, 5):
            subprocess.run(
                ['ninja', '-v'],
                stdout=None if verbose else subprocess.PIPE,
                stderr=subprocess.STDOUT,
                cwd=build_directory,
                check=True)
        else:
            subprocess.check_output(
                ['ninja', '-v'],
                stderr=subprocess.STDOUT,
                cwd=build_directory)
    except subprocess.CalledProcessError:
        # Python 2 and 3 compatible way of getting the error object.
        _, error, _ = sys.exc_info()
        # error.output contains the stdout and stderr of the build attempt.
        message = "Error building extension '{}'".format(name)
        if hasattr(error, 'output') and error.output:
            message += ": {}".format(error.output.decode())
        raise RuntimeError(message)


def _import_module_from_library(module_name, path):
    # https://stackoverflow.com/questions/67631/how-to-import-a-module-given-the-full-path
    file, path, description = imp.find_module(module_name, [path])
    # Close the .so file after load.
    with file:
        return imp.load_module(module_name, file, path, description)


def _write_ninja_file(path,
                      name,
                      sources,
                      extra_cflags,
                      extra_cuda_cflags,
                      extra_ldflags,
                      extra_include_paths,
                      with_cuda):
    extra_cflags = [flag.strip() for flag in extra_cflags]
    extra_cuda_cflags = [flag.strip() for flag in extra_cuda_cflags]
    extra_ldflags = [flag.strip() for flag in extra_ldflags]
    extra_include_paths = [flag.strip() for flag in extra_include_paths]

    # Version 1.3 is required for the `deps` directive.
    config = ['ninja_required_version = 1.3']
    config.append('cxx = {}'.format(os.environ.get('CXX', 'c++')))
    if with_cuda:
        config.append('nvcc = {}'.format(_join_cuda_home('bin', 'nvcc')))

    # Turn into absolute paths so we can emit them into the ninja build
    # file wherever it is.
    sources = [os.path.abspath(file) for file in sources]
    user_includes = [os.path.abspath(file) for file in extra_include_paths]

    # include_paths() gives us the location of torch/extension.h
    system_includes = include_paths(with_cuda)
    # sysconfig.get_paths()['include'] gives us the location of Python.h
    system_includes.append(sysconfig.get_paths()['include'])

    # Windoze does not understand `-isystem`.
    if IS_WINDOWS:
        user_includes += system_includes
        system_includes.clear()

    common_cflags = ['-DTORCH_EXTENSION_NAME={}'.format(name)]
    common_cflags.append('-DTORCH_API_INCLUDE_EXTENSION_H')
    common_cflags += ['-I{}'.format(include) for include in user_includes]
    common_cflags += ['-isystem {}'.format(include) for include in system_includes]

    if _is_binary_build():
        common_cflags += ['-D_GLIBCXX_USE_CXX11_ABI=0']

    cflags = common_cflags + ['-fPIC', '-std=c++11'] + extra_cflags
    if IS_WINDOWS:
        from distutils.spawn import _nt_quote_args
        cflags = _nt_quote_args(cflags)
    flags = ['cflags = {}'.format(' '.join(cflags))]

    if with_cuda:
        cuda_flags = common_cflags + COMMON_NVCC_FLAGS
        if IS_WINDOWS:
            cuda_flags = _nt_quote_args(cuda_flags)
        else:
            cuda_flags += ['--compiler-options', "'-fPIC'"]
            cuda_flags += extra_cuda_cflags
            if not any(flag.startswith('-std=') for flag in cuda_flags):
                cuda_flags.append('-std=c++11')

        flags.append('cuda_flags = {}'.format(' '.join(cuda_flags)))

    if IS_WINDOWS:
        ldflags = ['/DLL'] + extra_ldflags
    else:
        ldflags = ['-shared'] + extra_ldflags
    # The darwin linker needs explicit consent to ignore unresolved symbols.
    if sys.platform == 'darwin':
        ldflags.append('-undefined dynamic_lookup')
    elif IS_WINDOWS:
        ldflags = _nt_quote_args(ldflags)
    flags.append('ldflags = {}'.format(' '.join(ldflags)))

    # See https://ninja-build.org/build.ninja.html for reference.
    compile_rule = ['rule compile']
    if IS_WINDOWS:
        compile_rule.append(
            '  command = cl /showIncludes $cflags -c $in /Fo$out')
        compile_rule.append('  deps = msvc')
    else:
        compile_rule.append(
            '  command = $cxx -MMD -MF $out.d $cflags -c $in -o $out')
        compile_rule.append('  depfile = $out.d')
        compile_rule.append('  deps = gcc')

    if with_cuda:
        cuda_compile_rule = ['rule cuda_compile']
        cuda_compile_rule.append(
            '  command = $nvcc $cuda_flags -c $in -o $out')

    link_rule = ['rule link']
    if IS_WINDOWS:
        cl_paths = subprocess.check_output(['where',
                                            'cl']).decode().split('\r\n')
        if len(cl_paths) >= 1:
            cl_path = os.path.dirname(cl_paths[0]).replace(':', '$:')
        else:
            raise RuntimeError("MSVC is required to load C++ extensions")
        link_rule.append(
            '  command = "{}/link.exe" $in /nologo $ldflags /out:$out'.format(
                cl_path))
    else:
        link_rule.append('  command = $cxx $in $ldflags -o $out')

    # Emit one build rule per source to enable incremental build.
    object_files = []
    build = []
    for source_file in sources:
        # '/path/to/file.cpp' -> 'file'
        file_name = os.path.splitext(os.path.basename(source_file))[0]
        if _is_cuda_file(source_file) and with_cuda:
            rule = 'cuda_compile'
            # Use a different object filename in case a C++ and CUDA file have
            # the same filename but different extension (.cpp vs. .cu).
            target = '{}.cuda.o'.format(file_name)
        else:
            rule = 'compile'
            target = '{}.o'.format(file_name)
        object_files.append(target)
        if IS_WINDOWS:
            source_file = source_file.replace(':', '$:')
        source_file = source_file.replace(" ", "$ ")
        build.append('build {}: {} {}'.format(target, rule, source_file))

    ext = 'pyd' if IS_WINDOWS else 'so'
    library_target = '{}.{}'.format(name, ext)

    link = ['build {}: link {}'.format(library_target, ' '.join(object_files))]

    default = ['default {}'.format(library_target)]

    # 'Blocks' should be separated by newlines, for visual benefit.
    blocks = [config, flags, compile_rule]
    if with_cuda:
        blocks.append(cuda_compile_rule)
    blocks += [link_rule, build, link, default]
    with open(path, 'w') as build_file:
        for block in blocks:
            lines = '\n'.join(block)
            build_file.write('{}\n\n'.format(lines))


def _join_cuda_home(*paths):
    '''
    Joins paths with CUDA_HOME, or raises an error if it CUDA_HOME is not set.

    This is basically a lazy way of raising an error for missing $CUDA_HOME
    only once we need to get any CUDA-specific path.
    '''
    if CUDA_HOME is None:
        raise EnvironmentError('CUDA_HOME environment variable is not set. '
                               'Please set it to your CUDA install root.')
    return os.path.join(CUDA_HOME, *paths)


def _is_cuda_file(path):
    return os.path.splitext(path)[1] in ['.cu', '.cuh']