Reading a binary file with Julia

Question

Hi im new in Julia and im trying to do some proprocessing with it. I have a lot of files in binary. Im using F90 for the processing. The file that i write are like this

open(11,file="ring.grid",form="unformatted")

write(11)nv
write(11)Nx,Ny,Nzt
write(11) xx
write(11) yy
write(11) zz
close(11)

where, nv,nx,ny,nzt are integers and xx,yy,zz are 3D matrix. In F90 i use this simple block to read the file

open(11,file='ring.grid',form='unformatted')
read(11)nd
read(11)n1,n2,n3
read(11)xx
read(11)yy
read(11)zz
close(11)

But i have troubles to read it with Julia. Someone can help me with this? Thank you

I tried using

io=read(NAMEFILE)
y=Array{UInt8,1}(undef,1)
read!(io,y)

But i dont know how i can separe the nv,n1,n2,n3,xx,yy,zz in differente arrays

Answer 1

Suppose that the complete F90 code to write the data looks like this:

program SaveGrid
    implicit none

    integer :: nv
    integer :: Nx, Ny, Nzt
    real, dimension(:), allocatable :: xx, yy, zz
    integer :: i

    nv = 3
    Nx = 5
    Ny = 7
    Nzt = 10

    allocate(xx(Nx), yy(Ny), zz(Nzt))

    do i = 1, Nx
        xx(i) = float(i)
    end do
    do i = 1, Ny
        yy(i) = float(i) * 2.0
    end do
    do i = 1, Nzt
        zz(i) = float(i) * 3.0
    end do

    open(unit=11, file="ring.grid", form="unformatted")

    write(11) nv
    write(11) Nx, Ny, Nzt
    write(11) xx
    write(11) yy
    write(11) zz

    close(11)

    deallocate(xx, yy, zz)
end program SaveGrid

To read this into Julia you need to handle the way that F90 writes unformatted data.

function read_fortran_binary(filename)
    open(filename, "r") do file
        # Read record marker.
        record_size = read(file, Int32)
        # Read data.
        nv = read(file, Int32)
        # Read record marker.
        end_record_size = read(file, Int32)

        # Read record marker.
        read(file, Int32)
        # Read data.
        Nx = read(file, Int32)
        Ny = read(file, Int32)
        Nzt = read(file, Int32)
        # Read record marker.
        read(file, Int32)

        return nv, Nx, Ny, Nzt
    end
end

nv, Nx, Ny, Nzt = read_fortran_binary("ring.grid")

println("nv:  ", nv)
println("Nx:  ", Nx)
println("Ny:  ", Ny)
println("Nzt: ", Nzt)

There are record markers both before and after each chunk of data written to the file. You need to read these too. Obviously you don't need to ignore these data and the record sizes can be used to make your Julia code more robust and flexible by catering for changes in the unformatted F90 data.

nv:  3
Nx:  5
Ny:  7
Nzt: 10