Develop Reference

Using cython and gcc compiler [duplicate]

Is it possible (and how) to use MinGW-w64 for building of C-extensions for Python or embeding Python on Windows?
Let's take as example the following cython-extension foo.pyx:
print("foo loaded")
from which the C-code can be generated either via cython -3 foo.pyx or cython -3 --embed foo.pyx if interpreter should be embedded.

While mingw-w64-compiler is not really supported (the only supported windows compiler is MSVC), it can be used to create C-extensions or to embed Python. There are however no guarantee, this won't break in the future versions.
distutils does not support mingw-w64, so there is no gain in setting up a setup.py-file - the steps must be performed manually.
First we need some information usually provided by distutils:
Headers: We need the path to the Python includes. For a way to find them see this SO-post.
DLL: mingw-w64's linker works differently than MSVC's: python-dll and not python-lib is needed. So we need the path to the pythonXY.dll which is usually next the the python.exe.
Once the C-code is created/generated, the extension can be build via
x86_64-w64-mingw32-gcc -shared foo.c -DMS_WIN64 -O2 <other_options> -I <path_to_python_include> -L <path_to_python_dll> -lpython37 -o foo.pyd
The important details are:
it is probably Ok to use only use -O2 for optimization and leave <other_options> empty-
It is important to define MS_WIN64-macro (e.g. via -DMS_WIN64). In order to build for x64 on windows it must be set, but it works out of the box only for MSVC (defining _WIN64 could have slightly different outcomes):
#ifdef _WIN64
#define MS_WIN64
#endif
if it is not done, at least for files generated by Cython the following error message will be generated by the compiler:
error: enumerator value for ‘__pyx_check_sizeof_voidp’ is not an integer constant
201 | enum { __pyx_check_sizeof_voidp = 1 / (int)(SIZEOF_VOID_P == sizeof(void*)) };
pyd is just a dll in disguise, thus we need the -shared option, which means a dynamic library (i.e. shared-object in Linux-world) will be created.
It is important, that the python-library (pythonXY) should be the dll itself and not the lib (see this SO-post). Thua we use the path to pythonXY.dll (in my case python37) and not pythonXY.lib, as it would be the case for MSVC.
One probably should add the proper suffix to the resulting pyd-file, I use the old convention for simplicity here.
Embeded Python:
In this case an executable should be build (e.g. the C-file is generated by Cython with --embed option: cython -3 --embed foo.pyx) and thus the command line looks as follows:
x86_64-w64-mingw32-gcc foo.c -DMS_WIN64 -O2 <other_options> -I <path_to_python_include> -L <path_to_python_dll> -lpython37 -o foo.exe -municode
There are two important differences:
-shared should no longer be used, as the result is no longer a dynamic library (that is what *.pyd-file is after all) but an executable.
-municode is needed, because for Windows, Cython defines int wmain(int argc, wchar_t **argv) instead of int main(int argc, char** argv). Without this option, an error message like
/build/mingw-w64-_1w3Xm/mingw-w64-4.0.4/mingw-w64-crt/crt/crt0_c.c:18: undefined reference to 'WinMain'
collect2: error: ld returned 1 exit status
would appear (see this SO-post for more information).
Note: for the resulting executable to run, a whole python-distribution (and not only the dll) is needed (see also this SO-post), otherwise the resulting executable will abort with error (either the python dll wasn't found or the python installation or the site packages - depending on the configuration of the machine on which the exe has to run).
mingw-w64 can also be used on Linux for cross-compilation for Windows, see this SO-post.

f2py is not able to link with NetCDF library -

I am trying install some Python modules in my system using f2py, which are related to an ocean model (in Fortran 90). I am facing some problems with f2py. To be specific, the f2py is unable to link with the NetCDF library even though I have the required library and include files installed. I am using Python 2.7 with Anaconda 2 on Ubuntu16.04 on a 64-bit machine. I'm using gfortran.
To test it's working, I wrote a small code - an f90 module that contains a small subroutine. The subroutine performs a basic math task, and calls a NetCDF routine that prints the NetCDF version installed. The module (testsub.f90) is as follows:
module testsub
implicit none
contains
subroutine f_sum(a, b, s)
!#include 'netcdf.inc'
use netcdf
real(8) :: a, b, s
s = a+b;
!Calls a function that prints the netcdf version
write(*,*) trim(nf90_inq_libvers())
end subroutine f_sum
end module
The makefile for testsub is:
#Fortran compiler
FC=gfortran
NCLIB = -L/home/sonaljit/anaconda2/lib -lnetcdf -lnetcdff -L/usr/lib/python2.7/config-x86_64-linux-gnu -lpython2.7
NCINC = -I/home/sonaljit/anaconda2/include
#f2py and flags
F2PY = /home/sonaljit/anaconda2/bin/f2py
PYFLAGS = "-fPIC -g -O2 -fdefault-real-8"
pytest : testsub.f90
$(F2PY) --fcompiler=$(FC) --f90flags=$(PYFLAGS) -c $(NCINC) -m testpymod testsub.f90 $(NCLIB)
clean :
rm testpymod.so
I have the NetCDF library and include files installed in the given paths. When I run the makefile using make pytest, I get the following error:
/usr/bin/ld: /home/sonaljit/anaconda2/lib/libnetcdf.a(netcdf.o): relocation R_X86_64_32 against `.rodata' can not be used when making a shared object; recompile with -fPIC
/home/sonaljit/anaconda2/lib/libnetcdf.a: error adding symbols: Bad value
collect2: error: ld returned 1 exit status
However, I am not seeing this error when I am commenting out the NetCDF lines in the module. It seems f2py is not able to link to NetCDF routines. What can be the error here? Is it due to the structure of the code? Or, do I need to include some other library?

You are compiling a shared (dynamic) library and you should use a shared library version of NetCDF.
If you installed NetCDF yourself (as suggested by the path in your /home/sonaljit), you should install the .so version and link with this version.

Compile file .c with embedded Python/C functions

I'm starting the study of Python/C API and I make the first code to test some functions, I write this:
file: test.c
#include "Python.h"
int main() {
PyObject* none = Py_BuildValue("");
}
I compile with command:
gcc -I/usr/include/python2.7 test.c
I've the error undefined reference to `Py_BuildValue'
After I run:
gcc -I/usr/include/python2.7 --shared -fPIC hashmem.c
this compile without errors, but when I run the compiled file I've a
Segmentation fault (core dumped)
How do I set the gcc parameters?
I've ubuntu 12.04, python 2.7.3, gcc 4.6.3 and I installed python-dev.
Thanks.

In the comments #Pablo has provided the solution
gcc -I/usr/include/python2.7 test.c -lpython2.7
I forgot to link the python library with the "-l" parameter.
-llibrary
-l library
Search the library named library when linking. (The second alternative with the library as a separate argument is only for POSIX) compliance and is not recommended.)It makes a difference where in the command you write this option; the linker searches and processes libraries and object files in the order they are specified. Thus, foo.o -lz bar.o' searches libraryz' after file foo.o but before bar.o. If bar.o refers to functions in z', those functions may not be loaded.The linker searches a standard list of directories for the library, which is actually a file named liblibrary.a. The linker then uses this file as if it had been specified precisely by name.The directories searched include several standard system directories plus any that you specify with -L.Normally the files found this way are library files—archive files whose members are object files. The linker handles an archive file by scanning through it for members which define symbols that have so far been referenced but not defined. But if the file that is found is an ordinary object file, it is linked in the usual fashion. The only difference between using an -l option and specifying a file name is that - l surrounds library withlib' and `.a' and searches several directories. 
Parameter description source

How can I build my C extensions with MinGW-w64 in Python?

So I have a few Python C extensions I have previously built for and used in 32 bit Python running in Win7. I have now however switched to 64 bit Python, and I am having issues building the C extension with MinGW-w64.
I made the changes to distutils as per this post, but I am getting some weird errors suggesting something is wrong:
$ python setup.py build
running build
running build_ext
building 'MyLib' extension
c:\MinGW64\bin\x86_64-w64-mingw32-gcc.exe -mdll -O -Wall -Ic:\Python27\lib\site-packages\numpy\core\include -Ic:\Python27\include -Ic:\Python27\PC -c MyLib.c -o build\temp.win-amd64-2.7\Release\mylib.o
MyLib.c: In function 'initMyLib':
MyLib.c:631:5: warning: implicit declaration of function 'Py_InitModule4_64' [-Wimplicit-function-declaration]
writing build\temp.win-amd64-2.7\Release\MyLib.def
c:\MinGW64\bin\x86_64-w64-mingw32-gcc.exe -shared -s build\temp.win-amd64-2.7\Release\mylib.o build\temp.win-amd64-2.7\Release\MyLib.def -Lc:\Python27\libs -Lc:\Python27\PCbuild\amd64 -lpython27 -o build\lib.win-amd64-2.7\MyLib.pyd
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x13d): undefined reference to `__imp_PyExc_ValueError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x1275): undefined reference to `__imp_PyExc_ValueError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x1eef): undefined reference to `__imp_PyExc_ImportError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x1f38): undefined reference to `__imp_PyExc_AttributeError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x1f4d): undefined reference to `__imp_PyCObject_Type'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x1f61): undefined reference to `__imp_PyExc_RuntimeError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x1fc7): undefined reference to `__imp_PyExc_RuntimeError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x1ffe): undefined reference to `__imp_PyExc_RuntimeError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x2042): undefined reference to `__imp_PyExc_RuntimeError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x206c): undefined reference to `__imp_PyExc_RuntimeError'
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x208a): more undefined references to `__imp_PyExc_RuntimeError' follow
build\temp.win-amd64-2.7\Release\mylib.o:MyLib.c:(.text+0x20a7): undefined reference to `__imp_PyExc_ImportError'
collect2.exe: error: ld returned 1 exit status
error: command 'x86_64-w64-mingw32-gcc' failed with exit status 1
I have googled around quite a bit to find information, but it's not easy to find a definite answer. Could someone shed some light on this? What further changes should I do to be able to successfully build C extensions for 64 bit Python in Win7?
EDIT:
After some helpful pointers in cgohlke's comments below I managed to generate libpython27.a. However after following the advice on this post (2nd to last) I still had a the __imp_Py_InitModule4_64 error. After some serious Google-fu I managed to trip over this post telling me to rename the Py_InitModule4 line to Py_InitModule4_64. After that everything worked swimmingly.

This worked for me with Python 3.3 :
create static python lib from dll
python dll is usually in C:/Windows/System32; in msys shell:
gendef.exe python33.dll
dlltool.exe --dllname python33.dll --def python33.def --output-lib libpython33.a
mv libpython33.a C:/Python33/libs
use swig to generate wrappers
e.g., swig -c++ -python myExtension.i
wrapper MUST be compiled with MS_WIN64, or your computer will crash when you import the class in Python
g++ -c myExtension.cpp -I/other/includes
g++ -DMS_WIN64 -c myExtension_wrap.cxx -IC:/Python33/include
shared library
g++ -shared -o _myExtension.pyd myExtension.o myExtension_wrap.o -lPython33 -lOtherSharedLibs -LC:/Python33/libs -LC:/path/to/other/shared/libs
make sure all shared libs (gdal, OtherSharedLibs) are in your PATH
(windows does not use LD_LIBRARY_PATH or PYTHONPATH)
in Python, just: import myExtension
voila!

I realize this is an old question, but it is still the top search result. Today, in 2019, I was able to do this:
https://github.com/PetterS/quickjs/commit/67bc2428b8c0716538b4583f4f2b0a2a5a49106c
In short:
Make sure a 64-bit version of mingw-w64 is in the PATH.
Monkey-patch distutils:
import distutils.cygwinccompiler
distutils.cygwinccompiler.get_msvcr = lambda: []
Some differences in the shell w.r.t. escaping.
extra_link_args = ["-Wl,-Bstatic", "-lpthread"] in order to link statically and not have extra runtime deps.
pipenv run python setup.py build -c mingw32 now works.

Here is a example code for VC++ Build Tools
https://github.com/starnight/python-c-extension/tree/master/00-HelloWorld
You could try:
python setup.py -c mingw32
However this is not work for me.
My Solution is:
install Anaconda 64bit python 3.6
install mingw64
add mingw64/bin to PATH
compile dll from c file by
gcc -c libmypy.c -IC:\Users\{user_name}\Anaconda3\pkgs\python-3.6.4-h6538335_1\include
gcc -shared -o libmypy.dll libmypy.o -LC:\Users\{user_name}\Anaconda3\pkgs\python-3.6.4-h6538335_1\libs -lPython36
load dll file in .py script
from ctypes import *
m = cdll.LoadLibrary(r"C:\{path_to_dll}\libmypy.dll")
print(m.hello())

I created a monkey-patch for setuptools to let you to build_ext with mingw64 on Windows easily. See https://github.com/imba-tjd/mingw64ccompiler

I used this thread to wade through learning how to make a C extension, and since most of what I learned is in it, I thought I'd put the final discovery here too, so that someone else can find it if they are looking.
I wasn't trying to compile something big, just the example in Hetland's Beginning Python. Here is what I did (the example C pgm is called palindrome.c). I'm using Anaconda with python 3.7 in it, and the TDM-GCC version of MinGW64. I put all of the tools used into my Path, and all of the paths needed in PYTHONPATH, and the ..\Anaconda3 directory into PYTHON_HOME. I still ended up using explicit paths on some things.
I created the libpython37.a library with gendef.exe and dlltool.exe as Mark said above, and put it in ..\Anaconda3\libs.
I followed the prescription in Hetland:
gcc -c palindrome.c
gcc -I$PYTHON_HOME -I$PYTHON_HOME/Include -c palindrome_wrap.c
The second failed, the compiler couldn't find Python.h, the following worked:
gcc -I[somedirectories]\Anaconda3\Include -c palindrome_wrap.c
I then did, as many have said, including Hetland 3rd ed.,
gcc -shared palindrome.o palindrome_wrap.o [somedirectories]/Anaconda3/libs/libpython37.a -o _palindrome.dll
This did not work. Even with the Load Library cswu used (which I found elsewhere, too).
So I gendef'd _palindrome.dll and couldn't find the function in it, "is_palindrome" in the exports. I went through some of the SWIG documentation, and declared the function both in the %{ %} section and below it, both extern, that finally got the function extern'd in palindrome_wrap.c as it should have been. But no export, so I went back into palindrome.c and redeclared the function as:
declspec(dllexport) extern int __stdcall is_palindrome(char* text)
and redeclared it in palindrome.i in both places as above with this signature.
Partial success! It got listed in the Export section when I gendef'd _palindrome.dll and I could do cswu's call using Load Library. But still not do what Hetland says and do
import _palindrome
in Python.
Going back to all the sources again, I could not figure this out. I finally started reading the SWIG documentation from the beginning leaving no stone unturned -- Searching through the manual doesn't produce the place found.
At the end of Introduction Sec. 2.7 Incorporating Into a Build System, under the sample Make process, it says:
"The above example will generate native build files such as makefiles, nmake files and Visual Studio projects which will invoke SWIG and compile the generated C++ files into _example.so (UNIX) or _example.pyd (Windows). For other target languages on Windows a dll, instead of a .pyd file, is usually generated."
And that's the answer to the last problem:
The compile step for the dll should read:
gcc -shared palindrome.o palindrome_wrap.o [somedirectories]/Anaconda3/libs/libpython37.a -o _palindrome.pyd
(I didn't go back and change out my declspec declarations so I don't know whether they were necessary, so they were still there too).
I got a file, _palindrome.pyd
Which if in the PYTHONPATH (mine was local) works, and one can then do
import _palindrome
from _palindrome import is_palindrome
and use the exported, properly wrapped and packaged C function, compiled with TDM-GCC, in python as promised. gcc, which is MinGW64 in a different installation, knows how to do the .pyd file. I diffed the dll and pyd since they were the same byte length. They are not the same at hundreds of points.
Hope this helps someone else.

How to get program to not throw "Error: Can't load Floyd's algorithm library"

another question for all of you-
So i am trying to get a program called Pysomap to work (its basically ISOMAP but for python[http://web.vscht.cz/spiwokv/pysomap/]), i follow the directions best as i can, building it on Ubuntu, Windows, and Fedora (prebuilt libraries), but cant seem to get it to work. On windows (which is the preferred implementation platform), every time i go to python and import pysomap, it gives me the above error. Anybody know how to solve this?
Thanks
-J

In the pysomap directory, create a setup.py file with the following content:
from distutils.core import setup, Extension
setup(name="pysomap",
version="2007.07",
url ="http://web.vscht.cz/spiwokv/pysomap/",
description="A library for isometric feature mapping.",
license='Unknown',
author="Vojtech Spiwok",
py_modules=["floyd", "pysomap"],
ext_modules=[Extension('_floyd', ["floyd_wrap.c", "floyd.c"])]
)
Build the _floyd extension (assuming you have a Python distutils compatible C compiler installed):
python setup.py build_ext --inplace
Run demo.py:
python demo.py
Calculating distance matrix ............ 2.24 s
Using epsilon-isomap, epsilon = 0.500000, calculating 2-dimensional embedding
Calculating graph matrix ............... 1.21 s
Calculating the shortest path matrix ... 2.22 s
Multidimensionally scalling ............ 7.11 s
--------------------------------------------------
Total procedure ........................ 12.79 s

I had a look at the code because Isomap is a cool algorithm. The code doesn't look like it was written by someone familiar with Python, and the whole floyd extension module compilation is sort of shoddy - it actually didn't build for me, and I'm pretty sure that's your problem (they catch the import exception for the module and print out the error that you give).
I changed some hard-coded stuff in build_floyd.sh to different hard-coded values for my system. (A proper build system would make this unnecessary). Here's what worked in my Ubuntu system. I tested and it ran ok.
#!/bin/sh
# run this script to generate library for Floyd's
# algorithm library
echo "generating input files using SWIG ..."
swig -python floyd.i
echo "compiling ..."
# change compiler if you use other than gcc
gcc -c floyd.c floyd_wrap.c -I/usr/include/python2.6 -fPIC
echo "linking ..."
ld -shared floyd.o floyd_wrap.o -o _floyd.so
#echo "for SELinux you must run chcon ..."
#chcon -t textrel_shlib_t _floyd.so
You may also have to change the first line in demo.py to #!/usr/bin/env python.

This source code is over 5 years old and the build script for floyd looks to assume hard-coded python2.4.
It seems pretty clear that your floyd module did not build. You will most likely have to go back to the build step and ensure that you are properly generating a _floyd.so.
If you built it correctly, then this should not fail for you:
python -c "import _floyd"