使用 F2PY 创建 Fortran 扩展模块，自定义签名文件和子例程存储在单独的 Fortran 文件中？

Question

As an exercise, I am trying to create a python module fibadd1 which contains a single method fibadd . This method will eat an integer and spit out a numpy array of the first n fibonacci numbers with 1 added to them (ie 1, 2, 2, 3,...). The Fortran code containing this routine is fib2.f , and uses a subroutine FIB contained in fib1.f . This subroutine calculates the first N fibonacci numbers. With just these two files I have been able to successfully compile the given routine straight from the source, using a python script called test2.py . However, in trying to create a custom .pyf signature file as in this tutorial , I have run into a road block. The signature file is called fib2.pyf .

I was able to successfully compile and run everything using the command python -m numpy.f2py -c fib2.pyf fib2.f fib1.f --compiler=mingw32 from the command line. However, I would like to be able to use the numpy.f2py.run_main(...) and numpy.f2py.compile(...) functions to be able to do this in a python script. Based on this documentation it seems like they ought to have the same (or similar) functionalities. The script I tried is called test3.py .

It just gives a 1 on exit. Is it possible to do this in a python script? And also is there further documentation available than what I have linked? I have been having trouble parsing the numpy.f2py.compile function.

The source code is as follows:

fib2.f

C FILE: FIB2.F
      SUBROUTINE FIBADD(A,N)
C
C     ADD 1 TO THE FIBONACCI ROUTINE
C
      INTEGER N
      REAL*8 A(N)
      CALL FIB(A, N)
      DO I=1,N
          A(I) = A(I) + 1.0D0
      ENDDO
      END
C END FILE FIB2.F

fib1.f

C FILE: FIB1.F
      SUBROUTINE FIB(A,N)
C
C     CALCULATE FIRST N FIBONACCI NUMBERS
C
      INTEGER N
      REAL*8 A(N)
      DO I=1,N
         IF (I.EQ.1) THEN
            A(I) = 0.0D0
         ELSEIF (I.EQ.2) THEN
            A(I) = 1.0D0
         ELSE 
            A(I) = A(I-1) + A(I-2)
         ENDIF
      ENDDO
      RETURN
      END
C END FILE FIB1.F

test2.py

import numpy.f2py
import numpy as np

with open('fib2.f') as file:
    source = file.read()
module = 'fibadd1'
args = ['fib1.f', '--compiler=mingw32']
failure = numpy.f2py.compile(source, modulename=module, 
                             extra_args=args, verbose=False)
print(failure)

import fibadd1
a = np.zeros(8, 'd')
fibadd1.fibadd(a)
print(a)

fib2.pyf

!    -*- f90 -*-
! Note: the context of this file is case sensitive.

python module fib2
    interface
        subroutine fibadd(a,n)
            real*8 dimension(n),intent(out),depend(n) :: a
            integer intent(in) :: n
        end subroutine fibadd
    end interface 
end python module fib2

! This file was auto-generated with f2py (version:2).
! See http://cens.ioc.ee/projects/f2py2e/

test3.py

import numpy.f2py
import numpy as np

with open('fib2.f') as file:
    source = file.read()
module = 'fibadd1'
args = ['fib2.pyf','fib1.f', '--compiler=mingw32']
failure = numpy.f2py.compile(source, modulename=module, 
                             extra_args=args, verbose=False)
print(failure)

Answer 1

It turns out that, in the modified .pyf file there are two lines python module fib2 and end python module fib2 which dictate what the imported python module name will be (go figure). In order to successfully compile, the modulename keyword of the numpy.f2py.compile function must match that name. Thus, the corrected .py compilation script is:

import numpy.f2py

with open('fib2.f') as file:
    source = file.read()
    
module = 'fib2'
args = ['fib2.pyf','fib1.f', '--compiler=mingw32']

failure = numpy.f2py.compile(source, extra_args=args, 
                             modulename=module, verbose=False)
print(failure)

import fib2

a = fib2.fibadd(8)
print(a)