Reading floats in C++ equivalent in Java

Question

I have a binary file which supposed to contain 32-bit floating point numbers. This file was written out from Objective-C code. My task is to read this file in Java.

I think i managed to do it in C++:

#include <fstream>
#include <iostream>

int main() {
    float f;
    std::ifstream fin("effect_0.cube", std::ios::binary);
    while (fin.read(reinterpret_cast<char*>(&f), sizeof(float))) {
      std::cout << f << "\n";
    }

    return 0;
}

This is how i interpret a float in Java

InputStream is = new FileInputStream("effect_0.cube");
DataInputStream dataInputStream = new DataInputStream(is);
float f = dataInputStream.readFloat();

But this returns totally wrong values.

I am running all this on OSX.

The file is here . It is very small. The values from C++ seem right, because they are in the correct range, but i got totally different values in Java. I also provide the hexdump -n 16 -s 256 :

0000100 00 00 00 00 91 90 90 3e 00 00 00 00 00 00 80 3f
0000110

Answer 1

The C code is writing in CPU endian order , which for Intel processors is little-endian .

DataInputStream is reading in network byte order, which is big-endian .

The best way to read float value in Java, where you can control the endian order, is to use a ByteBuffer .

Sample code:

try (InputStream is = new FileInputStream("effect_0.cube")) {
    ByteBuffer buf = ByteBuffer.allocate(4).order(ByteOrder.LITTLE_ENDIAN);
    for (int len; (len = is.read(buf.array())) == 4; ) {
        buf.rewind();
        float f = buf.getFloat();
        System.out.println(f);
    }
}