Getting and Setting Arrays and Their Components

Just as in non-reflective code, an array field may be set or retrieved in its entirety or component by component. To set the entire array at once, use java.lang.reflect.Field.set(Object obj, Object value) . To retrieve the entire array, use Field.get(Object) . Individual components can be set or retrieved using methods in java.lang.reflect.Array .

Array provides methods of the form setFoo() and getFoo() for setting and getting components of any primitive type. For example, the component of an int array may be set with Array.setInt(Object array, int index, int value) and may be retrieved with Array.getInt(Object array, int index) .

These methods support automatic widening of data types. Therefore, Array.getShort() may be used to set the values of an int array since a 16-bit short may be widened to a 32-bit int without loss of data; on the other hand, invoking Array.setLong() on an array of int will cause an IllegalArgumentException to be thrown because a 64-bit long can not be narrowed to for storage in a 32-bit int with out loss of information. This is true regardless of whether the actual values being passed could be accurately represented in the target data type. The Java Language Specification, Java SE 7 Edition , sections Widening Primitive Conversion and Narrowing Primitive Conversion contains a complete discussion of widening and narrowing conversions.

The components of arrays of reference types (including arrays of arrays) are set and retrieved using Array.set(Object array, int index, int value) and Array.get(Object array, int index) .

Setting a Field of Type Array

The GrowBufferedReader example illustrates how to replace the value of a field of type array. In this case, the code replaces the backing array for a java.io.BufferedReader with a larger one. (This assumes that the creation of the original BufferedReader is in code that is not modifiable; otherwise, it would be trivial to simply use the alternate constructor BufferedReader(java.io.Reader in, int size) which accepts an input buffer size.)

import java.io.BufferedReader;
import java.io.CharArrayReader;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.lang.reflect.Field;
import java.util.Arrays;
import static java.lang.System.out;

public class GrowBufferedReader {
    private static final int srcBufSize = 10 * 1024;
    private static char[] src = new char[srcBufSize];
    static {
	src[srcBufSize - 1] = 'x';
    }
    private static CharArrayReader car = new CharArrayReader(src);

    public static void main(String... args) {
	try {
	    BufferedReader br = new BufferedReader(car);

	    Class<?> c = br.getClass();
	    Field f = c.getDeclaredField("cb");

	    // cb is a private field
	    f.setAccessible(true);
	    char[] cbVal = char[].class.cast(f.get(br));

	    char[] newVal = Arrays.copyOf(cbVal, cbVal.length * 2);
	    if (args.length > 0 && args[0].equals("grow"))
		f.set(br, newVal);

	    for (int i = 0; i < srcBufSize; i++)
		br.read();

	    // see if the new backing array is being used
	    if (newVal[srcBufSize - 1] == src[srcBufSize - 1])
		out.format("Using new backing array, size=%d%n", newVal.length);
	    else
		out.format("Using original backing array, size=%d%n", cbVal.length);

        // production code should handle these exceptions more gracefully
	} catch (FileNotFoundException x) {
	    x.printStackTrace();
	} catch (NoSuchFieldException x) {
	    x.printStackTrace();
	} catch (IllegalAccessException x) {
	    x.printStackTrace();
	} catch (IOException x) {
	    x.printStackTrace();
	}
    }
}
$ java GrowBufferedReader grow
Using new backing array, size=16384
$ java GrowBufferedReader
Using original backing array, size=8192

Note that the above example makes use of the array utility method java.util.Arrays.copyOf) . java.util.Arrays contains many methods which are convenient when operating on arrays.

Accessing Elements of a Multidimensional Array

Multi-dimensional arrays are simply nested arrays. A two-dimensional array is an array of arrays. A three-dimensional array is an array of two-dimensional arrays, and so on. The CreateMatrix example illustrates how to create and initialize a multi-dimensional array using reflection.

import java.lang.reflect.Array;
import static java.lang.System.out;

public class CreateMatrix {
    public static void main(String... args) {
        Object matrix = Array.newInstance(int.class, 2, 2);
        Object row0 = Array.get(matrix, 0);
        Object row1 = Array.get(matrix, 1);

        Array.setInt(row0, 0, 1);
        Array.setInt(row0, 1, 2);
        Array.setInt(row1, 0, 3);
        Array.setInt(row1, 1, 4);

        for (int i = 0; i < 2; i++)
            for (int j = 0; j < 2; j++)
                out.format("matrix[%d][%d] = %d%n", i, j, ((int[][])matrix)[i][j]);
    }
}
$ java CreateMatrix
matrix[0][0] = 1
matrix[0][1] = 2
matrix[1][0] = 3
matrix[1][1] = 4

The same result could be obtained by using the following code fragment:

Object matrix = Array.newInstance(int.class, 2);
Object row0 = Array.newInstance(int.class, 2);
Object row1 = Array.newInstance(int.class, 2);

Array.setInt(row0, 0, 1);
Array.setInt(row0, 1, 2);
Array.setInt(row1, 0, 3);
Array.setInt(row1, 1, 4);

Array.set(matrix, 0, row0);
Array.set(matrix, 1, row1);

The variable argument Array.newInstance(Class<?> componentType, int... dimensions) provides a convenient way to create multi-dimensional arrays, but the components still need to initialized using the principle that that multi-dimensional arrays are nested arrays. (Reflection does not provide multiple indexed get/set methods for this purpose.)