any sugestions ?

one of my ideas is when copying a file, create another file and add the sizeof the file the name and the position of the pointer.

What does your archive file format look like? A simple format would consist of records like so:

struct record {
    size_t size;
    unsigned char *bytes;
};

void write_record(FILE *out, struct record *rec)
{
    fwrite(rec->size, sizeof rec->size, 1, out);
    fwrite(rec->bytes, 1, rec->size, out);
}

int read_record(FILE *in, struct record *rec)
{
    if (fread(rec->size, sizeof rec->size, 1, in) == 1) {
        rec->bytes = malloc(rec->size);

        if (rec->bytes != NULL) {
            fread(rec->bytes, 1, rec->size, in);
            return 1;
        }
    }

    return 0;
}

From there it's a simple matter of appending each file to the archive file:

int archive_file(FILE *archive, FILE *source)
{
    struct record rec = {0};

    // Assuming get_file_size has been written
    rec->size = get_file_size(source);
    rec->bytes = malloc(rec->size);

    if (rec->bytes != NULL) {
        fread(rec->bytes, 1, rec->size, in);
        write_record(archive, &rec);
        free(rec->bytes);
        return 1;
    }

    return 0;
}

And pulling each archived file back in sequence:

void extract_all(FILE *archive)
{
    struct record rec;

    while (read_record(archive, &rec)) {
        FILE *dest = fopen(tmpnam(), "wb");

        if (dest != NULL) {
            fwrite(rec.bytes, rec.size, 1, dest);
            free(rec.bytes);
            fclose(dest);
        }
    }
}

All of this code is completely off the top of my head and written in the post text editor, so don't expect perfection. ;)

so basically you store the size of the file and the bytes, then you write this within the archive and when you extract the archive you just write out the size and the record of each file plus the destination

so you do not even need to store the record in a file?

in your extraction example would everything be extracted into "tmpnam"

so basically you store the size of the file and the bytes, then you write this within the archive and when you extract the archive you just write out the size and the record of each file plus the destination

All correct except the size isn't written when extracting. It's only used to determine how much actual data to extract.

so you do not even need to store the record in a file?

Records are stored in the archive file.

in your extraction example would everything be extracted into "tmpnam"

No, the tmpnam function returns a unique file name. In a more realistic program I'd include the name and permissions of the original file so that extraction produces the same thing that was archived.