Convert bytes to integer

Question

So I have the following functions:

void SomeClass::Read(__in uint32_t& Res)
{
    ReadBinary<uint32_t>(Res);
}

template < typename T > 
void SomeClass::ReadBinary(T& Res)
{
    size_t sBytesToRead = sizeof(T);

    Res = 0;
    std::vector<char> vcBuffer(sBytesToRead);
    m_Fstream.read(&vcBuffer[0], sBytesToRead);

    // little endian
    if (m_Endian == LITTLE_ENDIAN)
        for (int n = sBytesToRead-1; n >= 0; n--)
            Res = (Res << 8) + vcBuffer[n];

    // big endian
    else
        for (unsigned n = 0; n < sBytesToRead; n++)
            Res = (Res << 8) + vcBuffer[n];
}

void SomeClass::DetectEndian()
{
    int num = 1;
    if (*(char *)&num == 1)
        m_Endian = LITTLE_ENDIAN;
    else
        m_Endian = BIG_ENDIAN;
}

Those functions are designed to detect system endianness and read binary integers from file.

for some reason I don't get the expected values. How do I know? I've written a simple python script:

mode = 'rb'
with open(filename, mode) as f:
  print struct.unpack('i', f.read(4))[0]
  print struct.unpack('i', f.read(4))[0]

It seems like when the integer is small to be contain in a single byte the value of both programs are the same. However, once the integer consist of multiple bytes I get different values.

this leads me to think that I have a problem with those lines:

// little endian
if (m_Endian == LITTLE_ENDIAN)
   for (int n = sBytesToRead-1; n >= 0; n--)
      Res = (Res << 8) + vcBuffer[n]; 

Any ideas?

Answer 1

If char is signed in your compiler (and most likely it is) then it can be negative and so you'll substract a value instead of adding it, like demonstrated by this demo.

Use unsigned char or uint8_t instead of char.