Java有选择地使用字节读取序列化的对象

Question

I am trying to write a program which reads serialized objects selectively from a file given their byte offsets. I started by serializing three objects of same type in a file and then reading them using FileInputStream, ByteArrayInputStream and ObjectInputStream. But everytime I try to read a specific object it always returns me the first object. Here is the small program :

   public class TestObject implements Serializable {

    String term;
    double value;

    public TestObject(String term, double value) {
        this.term =term;
        this.value = value;
    }       

    public String toString() {
        String str = term + " : " + value;
        return str;
    }
}

   public class ObjectReader {

public static void main(String[] args) throws IOException,ClassNotFoundException {

            TestObject t1 = new TestObject("abc", 1.0);
    TestObject t2 = new TestObject("xyz", 1.0);
    TestObject t3 = new TestObject("123", 1.0);

            //Writing 3 objects to file and displaying offsets

        ByteArrayOutputStream baos = new ByteArrayOutputStream();
    ObjectOutputStream objOut = new ObjectOutputStream(baos);
    FileOutputStream fos = new FileOutputStream("data.dat");
    objOut.writeObject(t1);     

    byte[] arr = baos.toByteArray();
    System.out.println(arr.length);  //displays 81
    fos.write(arr);

    objOut.writeObject(t2);
    arr= baos.toByteArray();
            System.out.println(arr.length);  //displays 101 
    fos.write(arr);

    objOut.writeObject(t3);     
    arr= baos.toByteArray();
            System.out.println(arr.length);  //displays 121
    fos.write(arr);

    fos.close();
    objOut.close();

    //Reading a specific object back using offset

    FileInputStream fis = new FileInputStream("data.dat");
    byte[] inArr = new byte[101];
    fis.skip(81);    //skip to second object
    fis.read(inArr);  

    System.out.println(fis.available());  //displays 121 which is correct

    ByteArrayInputStream bain = new ByteArrayInputStream(inArr);
    ObjectInputStream objIn = new ObjectInputStream(bain);
    TestObject t4 =(TestObject)objIn.readObject();
    System.out.println(t4);
     }
    }

However everytime I run this program it displays only the first object (abc: 1.0). What could be the reason for this ? Is it not possible to read serialized objects like this ? Please ignore the efficiency issues as I am just trying to understand the concept and experiment.

Answer 1

ObjectOutputStream is a state machine it writes only increment data when you write mutiple objects, in order to write all data you can do ObjectOutputStream.reset() it will forget about previous writes and write fresh data with all information included. Or you can create a new ObjectOutputStream for each object as Peter said.

From javadoc:

The objects must be read back from the corresponding ObjectInputstream with the same types and in the same order as they were written.

Answer 2

如果不关闭流，请确保在writeObject之后进行刷新。

Answer 3

Also, you keep writing the entire byte[] to fos . So over your three calls you're doing:

1. Write object 1
2. Write object 1 & 2
3. Write object 1 & 2 & 3

Instead, use write(byte[], int, int) .

Answer 4

To read each object independently you need to write each object in its own ObjectOutputStream. You can't start an ObjectStream anywhere except the start of the stream. However you can write multiple stream to a file and provided you know the start and end of the stream reconstitute the object.