Skip to content

Hprose 过滤器

Jump to bottom
小马哥 edited this page Jul 2, 2016 · 9 revisions

简介

有时候，我们可能会希望在远程过程调用中对通讯的一些细节有更多的控制，比如对传输中的数据进行加密、压缩、签名、跟踪、协议转换等等，但是又希望这些工作能够跟服务函数/方法本身可以解耦。这个时候，Hprose 过滤器就是一个不错的选择。

Hprose 过滤器是一个接口，它定义在 hprose.common 包中，它有两个方法，该接口定义如下：

public interface HproseFilter {
    ByteBuffer inputFilter(ByteBuffer data, HproseContext context);
    ByteBuffer outputFilter(ByteBuffer data, HproseContext context);
}

其中 inputFilter 的作用是对输入数据进行处理，outputFilter 的作用是对输出数据进行处理。

data 参数就是输入输出数据。这两个方法的返回值表示已经处理过的数据，如果你不打算对数据进行修改，你可以直接将 data 参数作为返回值返回。

context 参数是调用的上下文对象，我们在服务器和客户端的介绍中已经多次提到过它。

执行顺序

不论是客户端，还是服务器，都可以添加多个过滤器。假设我们按照添加的顺序把它们叫做 filter1, filter2, ... filterN。那么它们的执行顺序是这样的。

在客户端的执行顺序

+------------------- outputFilter -------------------+
| +-------+      +-------+                 +-------+ |
| |filter1|----->|filter2|-----> ... ----->|filterN| |---------+
| +-------+      +-------+                 +-------+ |         v
+----------------------------------------------------+ +---------------+
                                                       | Hprose Server |
+-------------------- inputFilter -------------------+ +---------------+
| +-------+      +-------+                 +-------+ |         |
| |filter1|<-----|filter2|<----- ... <-----|filterN| |<--------+
| +-------+      +-------+                 +-------+ |
+----------------------------------------------------+

在服务器端的执行顺序

                  +-------------------- inputFilter -------------------+
                  | +-------+                 +-------+      +-------+ |
        +-------->| |filterN|-----> ... ----->|filter2|----->|filter1| |
        |         | +-------+                 +-------+      +-------+ |
+---------------+ +----------------------------------------------------+
| Hprose Client |                                                     
+---------------+ +------------------- outputFilter -------------------+
        ^         | +-------+                 +-------+      +-------+ |
        +---------| |filterN|<----- ... <-----|filter2|<-----|filter1| |
                  | +-------+                 +-------+      +-------+ |
                  +----------------------------------------------------+

跟踪调试

有时候我们在调试过程中，可能会需要查看输入输出数据。用抓包工具抓取数据当然是一个办法，但是使用 过滤器可以更方便更直接的显示出输入输出数据。

LogFilter.java

package hprose.example.filter.log;

import hprose.common.HproseContext;
import hprose.common.HproseFilter;
import hprose.util.StrUtil;
import java.nio.ByteBuffer;
import java.util.logging.Level;
import java.util.logging.Logger;

public class LogFilter implements HproseFilter {
    private static final Logger logger = Logger.getLogger(LogFilter.class.getName());
    @Override
    public ByteBuffer inputFilter(ByteBuffer data, HproseContext context) {
        logger.log(Level.INFO, StrUtil.toString(data));
        return data;
    }
    @Override
    public ByteBuffer outputFilter(ByteBuffer data, HproseContext context) {
        logger.log(Level.INFO, StrUtil.toString(data));
        return data;
    }
}

Server.java

package hprose.example.filter.log;

import hprose.server.HproseTcpServer;
import java.io.IOException;
import java.net.URISyntaxException;

public class Server {
    public static String hello(String name) {
        return "Hello " + name + "!";
    }
    public static void main(String[] args) throws URISyntaxException, IOException {
        HproseTcpServer server = new HproseTcpServer("tcp://0.0.0.0:8082");
        server.add("hello", Server.class);
        server.addFilter(new LogFilter());
        server.start();
        System.out.println("START");
        System.in.read();
        server.stop();
        System.out.println("STOP");
    }
}

Client.java

package hprose.example.filter.log;

import hprose.client.HproseClient;
import java.io.IOException;
import java.net.URISyntaxException;

interface IHello {
    String hello(String name);
}
public class Client {
    public static void main(String[] args) throws URISyntaxException, IOException {
        HproseClient client = HproseClient.create("tcp://127.0.0.1:8082");
        client.addFilter(new LogFilter());
        IHello h = client.useService(IHello.class);
        System.out.println(h.hello("World"));
    }
}

然后分别启动服务器和客户端，就会看到如下输出：

服务器输出

START
七月 02, 2016 11:32:19 上午 hprose.example.filter.log.LogFilter inputFilter
信息: Cs5"hello"a1{s5"World"}z
七月 02, 2016 11:32:19 上午 hprose.example.filter.log.LogFilter outputFilter
信息: Rs12"Hello World!"z

客户端输出

七月 02, 2016 11:32:19 上午 hprose.example.filter.log.LogFilter outputFilter
信息: Cs5"hello"a1{s5"World"}z
七月 02, 2016 11:32:19 上午 hprose.example.filter.log.LogFilter inputFilter
信息: Rs12"Hello World!"z
Hello World!

压缩传输

上面的例子，我们只使用了一个过滤器。在本例中，我们展示多个过滤器组合使用的效果。

CompressFilter.java

package hprose.example.filter.compress;

import hprose.common.HproseContext;
import hprose.common.HproseFilter;
import hprose.io.ByteBufferStream;
import java.io.IOException;
import java.nio.ByteBuffer;
import java.util.logging.Level;
import java.util.logging.Logger;
import java.util.zip.GZIPInputStream;
import java.util.zip.GZIPOutputStream;

public class CompressFilter implements HproseFilter {
    private static final Logger logger = Logger.getLogger(CompressFilter.class.getName());
    @Override
    public ByteBuffer inputFilter(ByteBuffer data, HproseContext context) {
        ByteBufferStream is = new ByteBufferStream(data);
        ByteBufferStream os = new ByteBufferStream();
        try {
            GZIPInputStream gis = new GZIPInputStream(is.getInputStream());
            os.readFrom(gis);
        }
        catch (IOException ex) {
            logger.log(Level.SEVERE, null, ex);
        }
        return os.buffer;
    }
    @Override
    public ByteBuffer outputFilter(ByteBuffer data, HproseContext context) {
        ByteBufferStream is = new ByteBufferStream(data);
        ByteBufferStream os = new ByteBufferStream();
        try {
            GZIPOutputStream gos = new GZIPOutputStream(os.getOutputStream());
            is.writeTo(gos);
            gos.finish();
        }
        catch (IOException ex) {
            logger.log(Level.SEVERE, null, ex);
        }
        return os.buffer;
    }
}

SizeFilter.java

package hprose.example.filter.compress;

import hprose.common.HproseContext;
import hprose.common.HproseFilter;
import java.nio.ByteBuffer;
import java.util.logging.Level;
import java.util.logging.Logger;

public class SizeFilter implements HproseFilter {
    private static final Logger logger = Logger.getLogger(SizeFilter.class.getName());
    private String message = "";
    public SizeFilter(String message) {
        this.message = message;
    }
    @Override
    public ByteBuffer inputFilter(ByteBuffer data, HproseContext context) {
        logger.log(Level.INFO, message + " input size: {0}", data.remaining());
        return data;
    }
    @Override
    public ByteBuffer outputFilter(ByteBuffer data, HproseContext context) {
        logger.log(Level.INFO, message + " output size: {0}", data.remaining());
        return data;
    }
}

Server.java

package hprose.example.filter.compress;

import hprose.server.HproseTcpServer;
import java.io.IOException;
import java.net.URISyntaxException;

public class Server {
    public static Object echo(Object obj) {
        return obj;
    }
    public static void main(String[] args) throws URISyntaxException, IOException {
        HproseTcpServer server = new HproseTcpServer("tcp://0.0.0.0:8083");
        server.add("echo", Server.class);
        server.addFilter(new SizeFilter("Non compressed"));
        server.addFilter(new CompressFilter());
        server.addFilter(new SizeFilter("Compressed"));
        server.start();
        System.out.println("START");
        System.in.read();
        server.stop();
        System.out.println("STOP");
    }
}

Client.java

package hprose.example.filter.compress;

import hprose.client.HproseClient;
import java.io.IOException;
import java.net.URISyntaxException;

interface IEcho {
    int[] echo(int[] obj);
}
public class Client {
    public static void main(String[] args) throws URISyntaxException, IOException {
        HproseClient client = HproseClient.create("tcp://127.0.0.1:8083");
        client.addFilter(new SizeFilter("Non compressed"));
        client.addFilter(new CompressFilter());
        client.addFilter(new SizeFilter("Compressed"));
        IEcho h = client.useService(IEcho.class);
        int n = 100000;
        int[] value = new int[n];
        for (int i = 0; i < n; ++i) {
            value[i] = i;
        }
        System.out.println(h.echo(value).length);
    }
}

然后分别启动服务器和客户端，就会看到如下输出：

服务器输出

START
七月 02, 2016 1:30:06 下午 hprose.example.filter.compress.SizeFilter inputFilter
信息: Compressed input size: 213,178
七月 02, 2016 1:30:06 下午 hprose.example.filter.compress.SizeFilter inputFilter
信息: Non compressed input size: 688,893
七月 02, 2016 1:30:06 下午 hprose.example.filter.compress.SizeFilter outputFilter
信息: Non compressed output size: 688,881
七月 02, 2016 1:30:06 下午 hprose.example.filter.compress.SizeFilter outputFilter
信息: Compressed output size: 213,154

客户端输出

七月 02, 2016 1:30:05 下午 hprose.example.filter.compress.SizeFilter outputFilter
信息: Non compressed output size: 688,893
七月 02, 2016 1:30:05 下午 hprose.example.filter.compress.SizeFilter outputFilter
信息: Compressed output size: 213,178
七月 02, 2016 1:30:06 下午 hprose.example.filter.compress.SizeFilter inputFilter
信息: Compressed input size: 213,154
七月 02, 2016 1:30:06 下午 hprose.example.filter.compress.SizeFilter inputFilter
信息: Non compressed input size: 688,881
100000

Hprose for Java 用户手册

Clone this wiki locally