Due to it's convenience, I was thinking on reading and writing file headers by creating structs mirroring the layouts of actual headers I would need. I've seen many examples of this in C, however I' struggling using the same methods through the use of code.stdc.stdio, especially as I can't really trace bugs from fread.

On Saturday, 22 July 2017 at 01:45:29 UTC, solidstate1991 wrote:
> Due to it's convenience, I was thinking on reading and writing file headers by creating structs mirroring the layouts of actual headers I would need. I've seen many examples of this in C, however I' struggling using the same methods through the use of code.stdc.stdio, especially as I can't really trace bugs from fread.

struct Data {
    int x;
    float y;
    ubyte z;
}

void main() {
    import core.stdc.stdio;

    Data od = Data(10, 3.0f, 5);

    FILE* fp = fopen("data.dat", "wb");
    size_t ret = fwrite(&od, od.sizeof, 1, fp);
    fclose(fp);

    assert(ret == 1);

    Data id;
    fp = fopen("data.dat", "rb");
    ret = fread(&id, id.sizeof, 1, fp);
    fclose(fp);

    assert(ret == 1);

    assert(id.x == 10);
    assert(id.y == 3.0f);
    assert(id.z == 5);
}

On Saturday, 22 July 2017 at 02:11:27 UTC, Mike Parker wrote:
> On Saturday, 22 July 2017 at 01:45:29 UTC, solidstate1991 wrote:
>> Due to it's convenience, I was thinking on reading and writing file headers by creating structs mirroring the layouts of actual headers I would need. I've seen many examples of this in C, however I' struggling using the same methods through the use of code.stdc.stdio, especially as I can't really trace bugs from fread.
>
> struct Data {
>     int x;
>     float y;
>     ubyte z;
> }
>
> void main() {
>     import core.stdc.stdio;
>
>     Data od = Data(10, 3.0f, 5);
>
>     FILE* fp = fopen("data.dat", "wb");
>     size_t ret = fwrite(&od, od.sizeof, 1, fp);
>     fclose(fp);
>
>     assert(ret == 1);
>
>     Data id;
>     fp = fopen("data.dat", "rb");
>     ret = fread(&id, id.sizeof, 1, fp);
>     fclose(fp);
>
>     assert(ret == 1);
>
>     assert(id.x == 10);
>     assert(id.y == 3.0f);
>     assert(id.z == 5);
> }

I should add, though, that you're better off using either std.stdio.File or std.file. Use the former if you need to make multiple reads/writes to a file, the latter if you can pull it in or push it out all in one go. They take arrays as arguments, so if you have something like Data[], you can pass it directly to the appropriate functions. To write a single instance, you'll have to take the pointer and slice it. Either way, it's less code, less error prone, and more idiomatic than using the C API.

On Saturday, 22 July 2017 at 02:22:53 UTC, Mike Parker wrote:
>
> I should add, though, that you're better off using either std.stdio.File or std.file. Use the former if you need to make multiple reads/writes to a file, the latter if you can pull it in or push it out all in one go. They take arrays as arguments, so if you have something like Data[], you can pass it directly to the appropriate functions. To write a single instance, you'll have to take the pointer and slice it. Either way, it's less code, less error prone, and more idiomatic than using the C API.

Well, it seems the toStringz() function adds a few garbage characters to the end of the filename, I might look into a way to read the data with Phobos instead of the C API.

On Sunday, 23 July 2017 at 02:07:45 UTC, solidstate1991 wrote:
> Well, it seems the toStringz() function adds a few garbage characters to the end of the filename

How are you using it? The only character it should be adding is the zero terminator.

I suspect you might be using it on a static array without slicing it down to size first...

Hi !
I have a dub package that doing this.

https://github.com/Temtaime/tt-utils/blob/master/source/tt/binary/tests.d

Have a look at the tests.
Currently it has no documentation, but feel free to ask questions