java和python在逻辑运算符中给出不同的结果

Question

I am trying to get broadcast ip address coding in Java and Python for practice. Here is my code in java which gives me correct output:

public IPv4Address getBroadcastAddress() throws IllegalArgumentException {
    long bits = 0;
    bits = this.address.decimalIP() ^ (~(0xffffffff ^ (1L << 32 - getMaskLength()) - 1));
    IPv4Address broadcast = new IPv4Address(bits);

    return broadcast; 
}

Here decimalIP is the number representing IP address, getMaskLength - number representing its mask. Here I got expected broadcast address. By in python using the same logic I got unexpected results:

def broadcastaddress(self):
    return IPv4Address(self.address.decimalip ^ (~(0xffffffff ^ (1 << 32 - self.mask) - 1)))

Here all components represent the same entries as in java. After debugging I got that number (~(0xffffffff ^ (1 << 32 - self.mask) - 1)) is negative. In documents operator ~ gives inversion in bits but I don't understand why this number is negative in Python and not in Java?

Answer 1

Python integers are not bounded and are not signed, so ~ creates a negative number:

>>> hex(~(0xffffffff ^ (1 << 32 - 24) - 1))
'-0xffffff01'
>>> ~(0xffffffff ^ (1 << 32 - 24) - 1)
-4294967041

However, ~ in Java would give you the two's complement , while the operation on the subnet mask requires a one's complement , which can be achieved by using XOR on the netmask. Since that'd undo the other XOR operation you already applied, you don't need to use the 1's complement here at all :

return IPv4Address(self.address.decimalip ^ ((1 << 32 - self.mask) - 1))

I suspect that you could just use IPv4Network.broadcast_address here:

network = IPv4Network('{}/{}'.format(self.address.decimalip, self.mask)
return network.broadcast_address`