Thread: How covert unsigned char[4] to unsigned integer?

Originally Posted by Al3

Only if his system is B/E.

Oh, I didn't mean it that way. I was referring to the look & feel of the code, not its functionality. But you were correct to point that out.
( If I understood correctly, you meant Big Endian )

The code worked when I needed get number of header offset. It was small number. So when tried to load bigger numbers greater then 255 so it did not work. I am not sure if there is not some other problem. For example planes should be 1 and not 8 so I am not sure if the error is in the conversion or elsewhere.

Are you sure that you're reading everything in the right order?

But your read*Bytes() functions are wrong, you don't need those ambersands(&) in there.

Thanks for the notice. I removed the ampersands from within the functions and most of the numbers are correct not. At least bitmap size and width and height is correct. I must find why planes and palColors are not correct.

If you're still using post #10, then it's wrong.

&item is a pointer to the parameter, not what it points to.

Also, passing the 2 or 4 byte buffer n seems a waste of effort, unless you use the result elsewhere.

Good notice, I fixed both functions. Now all values are fine. Thanks

Notice:
When I declare local variable

Code:

unsigned char n[4];

is this declared every time when function is called? Maybe better use

Code:

static  unsigned char n[4];

?

I have found this

Code:

#define UINT16_SWAP_LE_BE_CONSTANT(val)        \
  ((uint16_t)                    \
   (                        \
    (uint16_t) ((uint16_t) (val) >> 8) |    \
    (uint16_t) ((uint16_t) (val) << 8)))

#define UINT32_SWAP_LE_BE_CONSTANT(val)              \
  ((uint32_t)                          \
   (                              \
    (((uint32_t) (val) & (uint32_t) 0x000000ffU) << 24) | \
    (((uint32_t) (val) & (uint32_t) 0x0000ff00U) <<  8) | \
    (((uint32_t) (val) & (uint32_t) 0x00ff0000U) >>  8) | \
    (((uint32_t) (val) & (uint32_t) 0xff000000U) >> 24)))

I expect that it converts uint32_t to little endian. Author of this code uses it to write data to binary file.

> is this declared every time when function is called? Maybe better use
> static unsigned char n[4];
No.

On most machines, creating the local stack frame is a single instruction regardless of the number of local variables.

Could you yet help me how to get first 2 bytes of the bitmap "BM" characters to member char magicNo[2]?

readChars(fp, &bmpfile->header.magicNo, 2);

I tried:

Code:

static int readChars(FILE * fp, char** item, size_t size)
{
  size_t in;
  char n[size];
  in = fread(n, size, 1, fp);
  memcpy(item,n,size);
  if (!in) return -2;
  return in;
}

and

Code:

static int readChars(FILE * fp, char** item, size_t size)
{
  size_t in;
  char n[size];
  in = fread(item, size, 1, fp);
if (!in) return -2;
  return in;
}

None works

If I delete one star, then I got this message:
note: expected 'char *' but argument is of type 'char (*)[2]'|

Hence is not clear to me how to pass the char array by reference

> readChars(fp, &bmpfile->header.magicNo, 2);
Drop the &, and all those double stars should be single stars.

Originally Posted by Al3

Code:

const WORD id = 19778;

But this works only on little endians. On big endians the magic number will be 16973.

A better way is to compare the ID byte by byte.

Code:

typedef union int_2byte_t {
    uint16_t full;
    uint8_t part[2];
} int_2byte;

…
    int_2byte bmp_id;
    if (fread(&bmp_id, sizeof(bmp_id), 1, filepointer) != 1) return -1;
    if (bmp_id.part[0] == 'B' && bmp_id.part[1] == 'M') {
        /* work with BMP */
    }
    else {
        fclose(filepointer);
        return -2;
    }
…

(not tested, free typed from my head)