Reading binary files in C++

Question

I am trying to write a program that reads in binary files using C++. I am running into some unusual output that I was hoping for some help with.

I have a binary file that starts with these 4 bytes:

A1 B2 C3 D4 (verified using hexdump -C)

And here is the code I am using to read these 4 bytes in:

#include <iostream> // for reading and writing files
#include <fstream>

using namespace std;

char * buffer;
unsigned int LENGTH = 4;

int main(int argc, char ** argv)
{
    // open the binary file
    ifstream infile ("dump", ios::binary | ios::in);

    // create a buffer
    buffer = new char[LENGTH];

    // read LENGTH bytes from the file
    infile.read(buffer, LENGTH);

    // append the null byte to terminate the string
    buffer[LENGTH] = '\0';

    // loop over the 4 bytes read, and print
    for(int i = 0; i < LENGTH; i++)
    {
       printf("Buffer[%d] is %X\n", i, buffer[i]);
    }

    delete[] buffer;
    infile.close();


    return 0;
}

This program gives me these actual results:

Buffer[0] is FFFFFFA1
Buffer[1] is FFFFFFB2
Buffer[2] is FFFFFFC3
Buffer[3] is FFFFFFD4

BUT, I would expect these results:

Buffer[0] is A1
Buffer[1] is B2
Buffer[2] is C3
Buffer[3] is D4

Can anyone explain to me where the 3 0xFF bytes are coming from ? It only seems to be affecting the first 4 bytes of the file, the next 4 bytes print out as expected without any 0xFF bytes prepended.

Answer 1

The char buffer[i] is default promoted when passed as a variable argument. To get the right value, say static_cast<unsigned char>(buffer[i]).

Moreover, buffer[LENGTH] is out of bounds and thus undefined behaviour.

Answer 2

You can't read LENGTH bytes of content, and terminate at offset LENGTH in a buffer of size LENGTH. This is an off-by-one error.

buffer = new char[LENGTH];

This gives you space for LENGTH characters, with indices 0 through LENGTH - 1. So this:

buffer[LENGTH] = '\0';

writes outside the allocated memory, invoking undefined behavior.