NUM09-J. Do not rely on the default string representation of floating point values

Working with string representations of floating point values can produce incorrect conclusions about the precision of the values. For example, consider the conversion of a value from type float to type double, a widening primitive conversion. Refer to the guideline NUM10-J. Beware of precision loss when converting primitive integers to floating-point for more details about such conversions.

String representations of floating-point numbers shall not be compared or inspected. When the value of a float variable must be represented exactly using the double type, an explicit assignment is more appropriate than first converting the floating point value to a String and then to a double.

Noncompliant Code Example (String Comparison)

This noncompliant code example attempts a string-based comparison of a floating-point number.

int i = 1;
String s = Double.valueOf(i / 1000.0).toString();
if (s.equals("0.001")) {
  // ...
}

However s actually contain the string "0.0010". Consequently, the comparison unexpectedly fails.

Noncompliant Code Example (Regex)

This noncompliant code example attempts to mitigate the extra trailing zero by using a regular expression on the string before comparing it.

int i = 1;
String s = Double.valueOf(i / 1000.0).toString();
s = s.replaceFirst("[.0]*$", "");
if (s.equals("0.001")) {
  // ...
}

Thie comparison does succeed on this code.

However, the comparison fails on this similar code, which uses 1/10000.0 instead of 1/1000.0. The string produced is not 0.00010 but rather 1.0E-4.

int i = 1;
String s = Double.valueOf(i / 10000.0).toString();
s = s.replaceFirst("[.0]*$", "");
if (s.equals("0.0001")) {
  // ...
}

Compliant Solution (String Comparison)

This compliant solution uses the BigDecimal class to avoid precision loss. It then performs a numeric comparison, which passes as expected.

int i = 1;
BigDecimal d = new BigDecimal(Double.valueOf(i / 1000.0).toString())
if (d.compareTo(new BigDecimal("0.001")) == 0) {
  // ...
}

Risk Assessment

Relying on the string representation of floating point types can result in imprecise values.

Automated Detection

TODO

Related Vulnerabilities

HV-192

Bibliography

[[API 2006]]
[[JLS 2005]]

NUM08-J. Do not construct BigDecimal objects from floating-point literals      03. Floating Point (FLP)      04. Object Orientation (OBJ)