我应该如何将POJO表示形式写入RandomAccessFile？

Question

I have a POJO which has a number of String fields. I want to write instances of this object (sorted to natural ordering) to a RandomAccessFile and then eventually search this file for the objects I need.

The problem is I'm not too sure how to go about this - for example, in a plain text file I would delimit a new object instance with, say, a new line. With RandomAccessFile should I be using the the size of the object in bytes to indicate a new object in the file - and if I know I want the second object, I would look at 2 object's worth of bytes into the file?

I've followed tutorials that write sorted integers to file - and if I want to find the second occurrence I would look 8 bytes into the file. Should I be applying the same logic with my POJO - or could I just delimit using a new line ?

Edit* these are my object's fields:

import java.io.FileWriter;
import java.io.IOException;
import java.io.Serializable;

/**
 * A RegradeRule is an individual rule found in the Regrade .dat file (one
 * line).
 * 
 * @author E Rowlands
 */
public class RegradeRule implements Serializable,  Comparable<RegradeRule> { 

    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;

    // getters & setters
}

Answer 1

You have several general approaches to implementing this:

• Use fixed-length strings - decide that your strings have a certain maximum length, and pad all strings that you write to that length with some unused character (say, zero). You will need to deal with encoding of your strings if you are to allow Unicode code points past 127.
• Write an index at the top of the file or in a separate file - Add a separate block of data at the top of the file to help you navigate to the correct item. This approach makes it difficult to grow the file beyond a certain limit, because the content needs to get copied or read into memory for expansion.
• Do not use RandomAccessFile if you always get the whole file at once - this is the simplest approach to implement, and you get a human-viewable/editable text file. The drawback to this approach is that you can no longer navigate to a specific item without reading the rest of the items.

Answer 2

There are two main strategies for putting variable-length records into random-access files.

1. Choose a fixed maximum length for all records and multiply by that value to locate the string you want. You will need to choose a null character to fill the unused section of each record, remove it on read and add it as padding on write.
2. Maintain an index to the start and length of each string (either in the file or in memory) and pass every file access through that index. Maintaining the index in the file can be tricky, especially when you need to add more records. You may prefer using a second file as an index or rebuilding it every time you open the file.

Each has it's advantages. (1) requires no index but can significantly increase the file size. (2) does not inflate the file size but requires maintenance of the index.

Answer 3

You could give your POJO a one-line csv-formatted toString() method and always write it as a single line to your random access file.

public class RegradeRule implements Serializable,  Comparable<RegradeRule> { 

    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;
    private String privateData;

    // getters & setters

    @override
    public String toString() {
        StringBuilder sb = new StringBuilder();
        sb.append(privateData);
        sb.append(";");
        sb.append(privateData);
        sb.append(";");
        // do that until your last String attribute and then the last one without a semicolon
        sb.append(privateData);
        return sb.toString();
    }
}

Doing so would write a csv file (where you could provide a header row).