Getting a Java field through reflection, but not from its String name

Question

Is it possible to get a Field through Java reflection if I have the field itself? It's a primitive float (public, no problem). I don't want to use its name as a String.

Example:

public class TVset {
  public float voltageA;
  public float voltageB;
  public float voltageC;
  public TVset(...) {...} // constructor
  public void function() {...} // it changes voltages
}

class Voltmeter{
  Object theObject;
  Field theField;

  Voltmeter(Object obj) {
    theObject = obj;
    Class theFieldClass = obj.getClass();
    Class theContainerClass = theFieldClass.getDeclaringClass();
    Field theField = ??? // <-- here I don't want to use a String
  }

  float getVoltage() {
    return theField.getFloat(theObject);
  }
}

TVset tv1 = new TVset(...);
TVset tv2 = new TVset(...);

Voltmeter meter = new Voltmeter(tv1.voltageB);
meter.getVoltage();
tv1.function();
meter.getVoltage(); <- should reflect the changed voltage
tv1.function();
meter.getVoltage(); <- should reflect the changed voltage
...

The effect is similar to passing the float by reference, but without wrapping it into a wrapper class.

I need to measure different voltages on different TV sets, just by changing the line:

Voltmeter meter = new Voltmeter(tv1.voltageB);

to something else, like:

Voltmeter meter = new Voltmeter(tv2.voltageA);

Is it possible to do it with reflection?

Thx

Answer 1

To use reflection you have to use a String. Instead of using a float you can use an object to wrap mutable float or a simple float[1];

BTW I wouldn't use float unless you have a really good reason, double suffers far less rounding error.

public class TVset {
  public double[] voltageA = { 0.0 };
  public double[] voltageB = { 0.0 };
  public double[] voltageC = { 0.0 };
}

class Voltmeter{
  final double[] theField;

  Voltmeter(double[] theField) {
    this.theField = theField;
  }

  double getVoltage() {
    return theField[0];
  }
}
// works just fine.
Voltmeter meter = new Voltmeter(tv1.voltageB);

EDIT: Using an abstract accessor. This is the fastest way to do this. AFAIK,the difference is less than 10 nano-seconds.

public abstract class Voltmeter{ // or use an interface
  public abstract double get();
  public abstract void set(double voltage);
}

public class TVset {
  private double _voltageA = 0.0;
  private double _voltageB = 0.0;
  private double _voltageC = 0.0;
  public final Voltmeter voltageA = new Voltmeter() {
     public double get() { return _voltageA; }
     public void set(double voltage) { _voltageA = voltage; }
  }
  public final Voltmeter voltageB = new Voltmeter() {
     public double get() { return _voltageB; }
     public void set(double voltage) { _voltageB = voltage; }
  }
  public final Voltmeter voltageC = new Voltmeter() {
     public double get() { return _voltageC; }
     public void set(double voltage) { _voltageC = voltage; }
  }
}

Personally, if speed is critical, I would just use the fields directly by name. You won't get simpler or faster than that.

Answer 2

Just for completeness I've included the delegate way of solving this. I would also not recommend having your floats with public access.

public class stackoverflow_5383947 {

    public static class Tvset {

        public float voltageA;
        public float voltageB;
        public float voltageC;

        public Tvset() {
        }

        public void function() {
            voltageA++;
        }
    };

    public static class Voltmeter {

        private VoltageDelegate _delegate;

        public Voltmeter(VoltageDelegate delegate) {
            _delegate = delegate;
        }

        float getVoltage() {
            return _delegate.getVoltage();
        }
    };

    public static interface VoltageDelegate {

        public float getVoltage();
    }   

    public static void main(String[] args) {
        final Tvset tv1 = new Tvset();
        Voltmeter meter = new Voltmeter(new VoltageDelegate()   {
            public float getVoltage() {
                return tv1.voltageA;
            }
        });

        System.out.println(meter.getVoltage());
        tv1.function();
        System.out.println(meter.getVoltage());
        tv1.function();
        System.out.println(meter.getVoltage());
    }
}

Answer 3

If you control the TVSet but need to use reflection for some reason, a good way to avoid errors is to write the method/field names that you need as String Constants in the TVSet class.

However if your concern is performance, reflection is not the way to go because accessing a field or method through reflection can be much slower than accessing through getters or directly.

Answer 4

Here a variant where you can give your float value instead of a string.

class Voltmeter{
  Object container;
  Field theField;

  Voltmeter(Object obj, float currentValue) {
    container = obj;
    Class<?> containerClass = obj.getClass();
    Field[] fields = containerClass.getFields();
    for(Field f : fields) {
       if (f.getType() == float.class &&
           f.getFloat(container) == currentValue) {
          this.theField = f;
          break;
       }
    }
  }

  float getVoltage() {
    return theField.getFloat(container);
  }
}

Then call it like this:

Voltmeter meter = new Voltmeter(tv1, tv1.voltageB);

It works only if the voltages in the moment of Voltmeter creation are different (and not NaN), as it takes the first Field with the right value. And it is not really more efficient, I think.

I wouldn't really recommend this.