Re: Controlling endian-ness?

["Steve" <nospam_steved94@comcast.net>]

"Peter C. Chapin" <pcc482719@gmail.com> wrote in message 
news:4897b7f5$0$19705$4d3efbfe@news.sover.net...
> I'm trying to read/write 32 bit quantities from/to a binary file. The file 
> has a format defined by a specification that is outside my control. Some 
> of the quantities are stored in the file in big endian order. However, my 
> machine is a little endian machine. The file in question also contains 
> various values that are most naturally represented with different data 
> types. Thus I'm looking at Stream_IO as a way to deal with it.
>
> As an experiment I wrote a small program that defines a 32 bit unsigned 
> integer type and then overrides the 'Write attribute to write the value in 
> big endian form. Here is what I have:
>
> with Ada.Streams;              use Ada.Streams;
> with Ada.Streams.Stream_IO;    use Ada.Streams.Stream_IO;
> with Interfaces;               use Interfaces;
>
> procedure Stream_Check is
>    pragma Assert(Stream_Element'Size = 8);
>
>    type Word is mod 2**32;
>    procedure Word_Write
>      (Stream : access Root_Stream_Type'Class; Item : in Word);
>    for Word'Size  use 32;
>    for Word'Write use Word_Write;
>
>    procedure Word_Write
>      (Stream : access Root_Stream_Type'Class; Item : in Word) is
>       Buffer    : Stream_Element_Array(0..3);
>       Workspace : Unsigned_32 := Unsigned_32(Word);  -- ERROR HERE!
>    begin
>       Buffer(0) :=
>           Stream_Element(Shift_Right((Workspace and 16#FF000000#), 24));
>       Buffer(1) :=
>           Stream_Element(Shift_Right((Workspace and 16#00FF0000#), 16));
>       Buffer(2) :=
>           Stream_Element(Shift_Right((Workspace and 16#0000FF00#),  8));
>       Buffer(3) :=
>           Stream_Element(Shift_Right((Workspace and 16#000000FF#),  0));
>       Write(Stream.all, Buffer);
>    end Word_Write;
>
>    Output_File    : File_Type;
>    Stream_Pointer : Stream_Access;
>
>    W : Word := 16#0000FFFF#;
> begin
>    Create(Output_File, Out_File, "test.bin");
>    Stream_Pointer := Stream(Output_File);
>    Word'Write(Stream_Pointer, W);
>    Close(Output_File);
> end Stream_Check;
>
> I'm using GNAT GPL 2008. It produces an error on the indicated line 
> saying, "invalid use of subtype mark in expression or call." Apparently it 
> doesn't like me trying to convert a Word to an Unsigned_32, but I don't 
> understand why (am I doing that conversion right?).
>
> I want to use Unsigned_32 so that I can use Shift_Right. I can't override 
> the 'Write attribute for Unsigned_32 directly because it's in a different 
> package (right?).
>
> Overall I have a feeling that there is probably a much better way to do 
> this.
>
> Peter

I don't know if this will be useful to you, but here is the technique I used 
for reversing the byte order on a 32 bit integer.  Actually I was going for 
"network byte order" of a float (s_float is a 32 bit float, u_long is a 
32bit unsigned integer).

      TYPE aByte IS MOD 256;
      FOR aByte'SIZE USE 8;
      TYPE aByteArray IS ARRAY( Positive RANGE <> ) OF aByte;

      FUNCTION nltohf( value : u_long ) RETURN s_float IS
         TYPE aFourBytes IS NEW aByteArray(1..4);
         FUNCTION Conv IS
           NEW Ada.Unchecked_Conversion( aFourBytes, s_float );
         FUNCTION Conv IS
           NEW Ada.Unchecked_Conversion( u_long, aFourBytes );
         temp : aFourBytes := Conv( value );
      BEGIN
         RETURN Conv( aFourBytes'( temp(4), temp(3), temp(2), temp(1) ) );
      END nltohf;

It isn't endian neutral (or particularly pretty), but it got the job done.

Regards,
Steve