Question

我试图在Fortran中读取我用Python编写的二进制文件。我知道如何做相反的事情（写Fortran并阅读Python）

这就是我用Python编写文件（.dat）的方法.txt。是检查数字

ph1 = np.linspace(-pi, pi, num=7200)

f_ph = open('phi.dat', 'w')
f_ph.write(ph1.tobytes('F'))
f_ph.close()

f_ph = open('phi.txt', 'w')
for aaa in ph1:
    ts = str(aaa) + '\n'
    f_ph.write(ts)
f_ph.close()

而我的Fortran代码看起来像这样：

       program reading

          real    realvalue
          integer i

        i=1

        open(unit=8,file='phi.dat',form='UNFORMATTED',status='OLD')


        do 
          read(8,END=999,ERR=1000)  realvalue
          write(*,'(1PE13.6)') realvalue
          i = i + 1
        enddo

999     write(*,'(/"End-of-file when i = ",I5)') i
        stop

1000    write(*,'(/"ERROR reading when i = ",I5)') i
        stop 

       end program reading

我在此示例http://numerical.recipes/forum/showthread.php?t=1697

上对此程序进行了建模

但如果我运行它，我会得到这个：

[gs66-stumbras:~/Desktop/fortran_exp] gbrambil% ./reading
-2.142699E+00

End-of-file when i =     2

Answer 1

关于Python，您必须添加二进制选项才能打开，即

import numpy as np
pi = np.pi
ph1 = np.linspace(-pi, pi, num=7200)

f_ph = open('phi.dat', 'wb')
f_ph.write(ph1.tobytes('F'))
f_ph.close()

f_ph = open('phi.txt', 'w')
for aaa in ph1:
    ts = str(aaa) + '\n'
    f_ph.write(ts)
f_ph.close()

关于Fortran，您必须考虑：

numpy默认基类型（很可能）Float64对应Fortran real(kind(1.d0))
由于Fortran通常会在读/写之前和之后跳过/添加记录标记，因此必须禁用此行为，将access="stream"添加到open语句

program reading
real(kind(1.d0)) :: realvalue
integer :: i

i=1

open(unit=8,file='phi.dat',form='UNFORMATTED',status='OLD', access="stream")

do
    read(8,END=999,ERR=1000)  realvalue
    write(*,'(1PE13.6)') realvalue
    i = i + 1
enddo

999     write(*,'(/"End-of-file when i = ",I5)') i-1
stop

1000    write(*,'(/"ERROR reading when i = ",I5)') i-1
stop

end program reading

阅读用Python编写的Fortran二进制文件

1 个答案: