NUM03-J. Provide mechanisms to handle unsigned data when required

The only unsigned primitive integer type in Java is the 16 bit char data type; all of the other primitive integer types are signed. To interoperate with native languages, such as C or C++, that use unsigned types extensively, any unsigned values must be read and stored into a Java integer type that can represent all of the values of the signed type. For example, use the Java long type to hold an unsigned 32-bit integer value obtained from native code.

Noncompliant Code Example

This noncompliant code example uses a generic method for reading integer data without considering the signedness of the source. It assumes that the values read are always signed and treats the most significant bit as the sign bit. When the data being read is unsigned, this causes misinterpretations of the actual sign and magnitude of the value.

public static int getInteger(DataInputStream is) throws IOException {
  return is.readInt();	
}

Compliant Solution

This compliant solution assumes that the values being read are 32-bit unsigned integers. It reads an unsigned integer value into a long variable using the readInt() method. The readInt() method assumes signed values and returns a signed Java int; the return value is converted to a long with sign extension. The code uses an & operation to mask off the upper 32-bits of the long; this produces a value in the range of a 32-bit unsigned integer, as intended. The mask size should be chosen to match the size of the unsigned integer values being read.

public static long getInteger(DataInputStream is) throws IOException {
  return is.readInt() & 0xFFFFFFFFL; // mask with 32 one-bits
}

As a general rule, you must be aware of the signedness of the data you are reading.

Risk Assessment

Treating unsigned data as though it were signed will produce incorrect values and can lead to lost or misinterpreted data.

Automated Detection

Automated detection is infeasible in the general case.

Bibliography

NUM02-J. Ensure that division and modulo operations do not result in divide-by-zero errors      03. Numeric Types and Operations (NUM)