Write语句不能在用户定义的格式化I / O过程中为派生类型生成新行

时间:2017-12-26 11:33:18

标签: io fortran newline derived-types formatted

我想在Fortran代码中为派生类型实现用户定义的I / O过程。但是,这些过程中的write语句无法在两个连续的write语句之间生成新行。派生类型和程序定义如下。

模块:

module station_module
    implicit none

    character(8), parameter :: FmtFloat = '(5E15.7)'

    type :: station
        integer, private :: ns = 0
        real, public, allocatable :: xloc(:), yloc(:), zloc(:)
    contains
        procedure, public :: import_station
        procedure, public :: export_station
        procedure, private :: read_station
        generic, public :: read (formatted) => read_station
        procedure, private :: write_station
        generic, public :: write (formatted) => write_station
        final :: destruct_station
    end type station

    interface station
        module procedure new_station
    end interface station

contains

    function new_station(n) result(t)
        implicit none
        integer, intent(in) :: n
        type(station) :: t

        if (n > 0) then
            allocate (t%zloc(n))
            allocate (t%yloc(n))
            allocate (t%xloc(n))
            t%ns = n
        end if
    end function new_station

    subroutine read_station(dtv, unit, iotype, vlist, iostat, iomsg)
        implicit none
        class(station), intent(inout) :: dtv
        integer, intent(in) :: unit
        character(*), intent(in) :: iotype
        integer, intent(in) :: vlist(:)
        integer, intent(out) :: iostat
        character(*), intent(inout) :: iomsg

        call dtv%import_station(unit)

        iostat = 0
    end subroutine read_station

    subroutine import_station(this, unit)
        implicit none
        class(station), intent(inout) :: this
        integer, intent(in) :: unit
        character(256) :: header, footer
        integer ns

        read (unit, '(A)') header !> Header
        read (unit, *) ns
        if (ns > 0) then
            if (allocated(this%zloc)) then
                deallocate (this%zloc)
            end if
            allocate (this%zloc(ns))
            read (unit, *) this%zloc

            if (allocated(this%yloc)) then
                deallocate (this%yloc)
            end if
            allocate (this%yloc(ns))
            read (unit, *) this%yloc

            if (allocated(this%xloc)) then
                deallocate (this%xloc)
            end if
            allocate (this%xloc(ns))
            read (unit, *) this%xloc

            this%ns = ns
        end if
        read (unit, '(A)') footer !> Footer
    end subroutine import_station

    subroutine export_station(this, unit)
        implicit none
        class(station), intent(in) :: this
        integer, intent(in) :: unit

        write (unit, '(A)') ">STATION INFO"
        write (unit, '(I6)') this%ns
        write (unit, *) "Z:"
        write (unit, FmtFloat) this%zloc
        write (unit, *) "Y:"
        write (unit, FmtFloat) this%yloc
        write (unit, *) "X:"
        write (unit, FmtFloat) this%xloc
        write (unit, '(A)') ">END STATION"
    end subroutine export_station

    subroutine write_station(dtv, unit, iotype, vlist, iostat, iomsg)
        implicit none
        class(station), intent(in) :: dtv
        integer, intent(in) :: unit
        character(*), intent(in) :: iotype
        integer, intent(in) :: vlist(:)
        integer, intent(out) :: iostat
        character(*), intent(inout) :: iomsg

        call dtv%export_station(unit)

        iostat = 0
    end subroutine write_station

    subroutine destruct_station(this)
        implicit none
        type(station), intent(inout) :: this

        if (allocated(this%xloc)) then
            deallocate (this%xloc)
        end if
        if (allocated(this%yloc)) then
            deallocate (this%yloc)
        end if
        if (allocated(this%zloc)) then
            deallocate (this%zloc)
        end if
        this%ns = 0
    end subroutine destruct_station

end module station_module

我们可以看到用户定义的格式化的写语句只调用一个名为export_station的常规子例程,我希望通过这两个方法得到相同的结果。

这是我的测试计划:

program Test
    use station_module
    implicit none
    type(station) :: pt, pt1, pt2

    pt = station(4)

    write(*, *) pt

    call pt%export_station(6)

end program Test

输出:

 >STATION INFO     4Z:  0.0000000E+00  0.0000000E+00  0.0000000E+00  0.0000000E+00
 Y:  0.0000000E+00  0.0000000E+00  0.0000000E+00  0.0000000E+00X:  0.0000000E+00  0.0000000E+00  0.0000000E+00  0.0000000E+00>END STATION
>STATION INFO
     4
 Z:
  0.0000000E+00  0.0000000E+00  0.0000000E+00  0.0000000E+00
 Y:
  0.0000000E+00  0.0000000E+00  0.0000000E+00  0.0000000E+00
 X:
  0.0000000E+00  0.0000000E+00  0.0000000E+00  0.0000000E+00
>END STATION

常规子例程export_station产生了我的期望。在两个write语句之间生成新行,而派生类型的write语句则不生成。

2 个答案:

答案 0 :(得分:6)

此问题也在英特尔论坛上提出。我在那里回答。"用户定义的派生类型I / O都是非推进的(你不能改变它)。如果你想要换行,你必须明确地写它们(例如,使用/ format。)"

答案 1 :(得分:2)

这里有两类输出语句:父语句和子语句。第一种情况下的父输出语句是write (*,*) pt

当第一个是父项时,对export_stationwrite_station的调用会导致写语句出现子输出语句。当用户直接调用export_station时,这些写语句本身就是父输出语句。

子数据传输语句和父数据传输语句之间的一个显着区别是父语句在数据传输之前和之后定位文件。也就是说,当write (unit,*) "Z:"完成时,文件位于仅在传输语句为父级时才写入的记录之后。

因此,您会看到新的行:这只是放在书面记录之后。

子数据传输语句(不是在完成时定位文件)不会影响新行。

我目前无法访问测试机器,所以这部分是推测性的。您可以显式写入从new_line('')返回的换行符,作为子转移语句输出的一部分。由于advance='no'将在子语句中被忽略,您可以在两种情况下使用它,明确控制新行的写入位置,而不是依赖于当前存在的拆分记录方法。