how to convert python float to bytes to be interpreted in C++ programm

Question

I need to build a python script that export data to a custom file format. This file is then read by a cpp programm (I have the source code, but cannot compile it).

custome file fomat spec:

• The file must be little endian.
• The float must be 4 bytes long.

I'm failing at exporting python float to bytes. thus crashing the cpp app without any error trace. If i try to fill all the float with 0 it load perfectly fine, but if i try anything else it crashes.

This is how the cpp app read the float

double Eds2ImporterFromMemory::read4BytesAsFloat(long offset) const
{
    // Read data.
    float i = 0;
    memcpy(&i, &_data[offset], 4);

    return i;
}

And I try to export the python float as follows:

def write_float(self, num):
    # pack float 
    self.f.write(struct.pack('<f', float(num)))

And also like this as some people suggested me:

def write_float(self, num):
    # unpack as integer the previously pack as float number    
    float_hex = struct.unpack('<I', struct.pack('<f', num))[0]
    self.f.write(float_hex.to_bytes(4, byteorder='little'))

But it fails every time. I'm not a cpp guy as you can see. Do you have an idea on why my python script is not working.

Thanks in advance

Answer 1

Please excuse me for being really bad at python, but i tried to produce your desired result and it works fine for me.

The python code:

import struct

value = 13.37

ba = bytearray(struct.pack("f", value))  
for b in ba:
    print("%02x" % b)

I imagine if you would just concat that (basically write the hexadecimal representation to a file in that order), it would work just fine.

Either way, it will output

85
eb
55
41

Which i put in an array of unsigned chars and used memcpy in the same way your C++ code did:

unsigned char data[] = {0x85, 0xeb, 0x55, 0x41};

float f;
memcpy(&f, data, sizeof(float));

std::cout << f << std::endl;
std::cin.get();

Which will output 13.37 as you would expect. If you cannot reproduce the correct result this way, i suspect that the error occurs somewhere else, in which case it would be helpful to see how the format is written by python and how it's read by C++.

Also, keep in mind that there are several ways to represent the byte array, like:

\\x85eb5541 , 0x85, 0xeb, 0x55, 0x41 , 85eb5541 and so on. To me it seems very likely that you just aren't outputting the correct format and thus the "parser" crashes.