Added the ability to build a project with PyTorch 2.0. (open-mmlab#2553)

* Added the ability to build a project with PyTorch 2.0.
Namely, I added the flag -std=c++17 to extra_compile_args
 depending on the version of Torch.

* Lost the condition for the presence of nvcc

* Lost the condition for the presence of nvcc

* Add parse_version

* fix lint

---------

Co-authored-by: Xin Chen <[email protected]>

Author: Danil328

setup.py

@@ -1,9 +1,11 @@
 import os
 
+from pkg_resources import parse_version
 from setuptools import find_packages, setup
 
 EXT_TYPE = ''
 try:
+    import torch
     from torch.utils.cpp_extension import BuildExtension
     cmd_class = {'build_ext': BuildExtension}
     EXT_TYPE = 'torch'
@@ -139,7 +141,10 @@ def get_extensions():
         # to compile those cpp files, so there is no need to add the
         # argument
         if platform.system() != 'Windows':
-            extra_compile_args['cxx'] = ['-std=c++14']
+            if parse_version(torch.__version__) <= parse_version('1.12.1'):
+                extra_compile_args['cxx'] = ['-std=c++14']
+            else:
+                extra_compile_args['cxx'] = ['-std=c++17']
 
         include_dirs = []
 
@@ -159,7 +164,10 @@ def get_extensions():
         # to compile those cpp files, so there is no need to add the
         # argument
         if 'nvcc' in extra_compile_args and platform.system() != 'Windows':
-            extra_compile_args['nvcc'] += ['-std=c++14']
+            if parse_version(torch.__version__) <= parse_version('1.12.1'):
+                extra_compile_args['nvcc'] += ['-std=c++14']
+            else:
+                extra_compile_args['nvcc'] += ['-std=c++17']
 
         ext_ops = extension(
             name=ext_name,