Thrift - 快速入门

Getting Started

如果有homebrew的话,直接执行以下命令即可,brew会处理相关依赖(https://thrift.apache.org/docs/install/)。

brew install thrift

 

或者可以从源码安装。
下载tar包 https://thrift.apache.org/download
参考 https://thrift.apache.org/docs/BuildingFromSource

 

先写一个例子,目录结构如下:

├── pom.xml
├── src
│   ├── main
│   │   ├── java
│   │   └── resources
│   └── test
│       └── java
└── thrift
    ├── Common.thrift
    └── ShopService.thrift

 

pom.xml中添加以下依赖:

<dependency>
    <groupId>org.apache.thrift</groupId>
    <artifactId>libthrift</artifactId>
    <version>0.10.0</version>
</dependency>
<dependency>
    <groupId>org.projectlombok</groupId>
    <artifactId>lombok</artifactId>
    <version>1.16.18</version>
    <scope>provided</scope>
</dependency>
<dependency>
    <groupId>org.slf4j</groupId>
    <artifactId>slf4j-simple</artifactId>
    <version>1.7.25</version>
</dependency>

 

thrift目录下创建两个thrift文件:

Common.thrift

namespace java me.kavlez.thrift.service

service BaseService {
    string echoServiceName()
}

 

ShopService.thrift

include "Common.thrift"

namespace java me.kavlez.thrift.service

struct Shop {
    1: required i32 id,
    2: required string name
}

struct Item {
    1: required i32 id,
    2: required string name = "unknown",
    3: required string detail,
    4: required Shop shop
}

service ShopService extends Common.BaseService {

  Shop queryShopInfo(1: i32 id),
  bool isValidShop(1: Shop shop),
  set<Item> queryItems(1: i32 shopId),
}

 

Thrift提供了多个语言的生成器实现,按照thrift文件生成java类,生成代码命令的用法如下:

thrift -r --gen <language> <Thrift filename>

 

其中-r即recursive,如果在文件中通过include关键字引用了其他文件,-r选项可以一并生成被引用的文件。

例如上面ShopService.thrift中的:

include Common.thrift

 

默认情况下,代码会在gen-<language>目录下生成,生成目录可以通过--out指定。

生成后再拷贝有点麻烦,直接生成到代码目录下,在工程目录下执行以下命令:

thrift -r --gen java --out src/main/java thrift/ShopService.thrift

 

执行后src/main/java/目录下生成me/kavlez/thrift/service/目录,以及4个java文件。

 

在service目录下创建impl,提供接口实现:

package me.kavlez.thrift.service.impl;

import lombok.extern.slf4j.Slf4j;
import me.kavlez.thrift.service.Item;
import me.kavlez.thrift.service.Shop;
import me.kavlez.thrift.service.ShopService;
import org.apache.thrift.TException;

import java.util.Collections;
import java.util.HashSet;
import java.util.Set;

/**
 * Created by Kavlez.Kim@gmail.com
 */
@Slf4j
public class ShopServiceImpl implements ShopService.Iface {
    @Override
    public Shop queryShopInfo(int id) throws TException {
        return new Shop(id, "DMC_".concat(String.valueOf(id)));
    }

    @Override
    public boolean isValidShop(Shop shop) throws TException {
        return shop != null;
    }

    @Override
    public Set<Item> queryItems(int shopId) throws TException {

        if (shopId < 1) {
            return Collections.emptySet();
        }

        Set<Item> items = new HashSet<>();
        Shop shop = new Shop(1101, "DMC");
        for (int i = 0; i < 8; i++) {
            Item item = new Item(shopId + i, "sample_".concat(String.valueOf(shopId + i))
                    , "this is sample_".concat(String.valueOf(i))
                    , shop);
            items.add(item);
        }
        return items;
    }

    @Override
    public String echoServiceName() throws TException {
        return "alo! this is shop service!";
    }
}

 

除了业务实现,我们需要额外做两件事情——构建Server和Client。

构建Server,也就是为Server指定Transparent、Protocol、Processor:

package me.kavlez.thrift.server;

import lombok.extern.slf4j.Slf4j;
import me.kavlez.thrift.service.ShopService;
import me.kavlez.thrift.service.impl.ShopServiceImpl;
import org.apache.thrift.TProcessor;
import org.apache.thrift.protocol.TCompactProtocol;
import org.apache.thrift.server.TServer;
import org.apache.thrift.server.TSimpleServer;
import org.apache.thrift.transport.TServerSocket;
import org.apache.thrift.transport.TTransportException;

/**
 * Created by Kavlez.Kim@gmail.com
 */
@Slf4j
public class SimpleServerHolder {

    public static TServer buildServer() {
        TServerSocket serverSocket = null;
        try {
            serverSocket = new TServerSocket(8081);
        } catch (TTransportException e) {
            e.printStackTrace();
        }
        TProcessor tprocessor = new ShopService.Processor<ShopService.Iface>(new ShopServiceImpl());

        TServer.Args tArgs = new TServer.Args(serverSocket);
        tArgs.protocolFactory(new TCompactProtocol.Factory());
        tArgs.processor(tprocessor);

        TServer server = new TSimpleServer(tArgs);
        return server;
    }

    public static void main(String[] args) {
        TServer server = SimpleServerHolder.buildServer();
        log.info("server ready...");
        server.serve();
    }
}

 

相应地,构建Client:

package me.kavlez.thrift.client;

import lombok.extern.slf4j.Slf4j;
import me.kavlez.thrift.service.Item;
import me.kavlez.thrift.service.ShopService;
import org.apache.thrift.TException;
import org.apache.thrift.protocol.TCompactProtocol;
import org.apache.thrift.protocol.TProtocol;
import org.apache.thrift.transport.TSocket;
import org.apache.thrift.transport.TTransport;

import java.util.Set;

/**
 * Created by Kavlez.Kim@gmail.com
 */
@Slf4j
public class SimpleClientHolder {

    private TTransport transport;

    public ShopService.Client buildClient(String serverAddr, int serverPort, int timeout) throws TException {

        this.transport = new TSocket(serverAddr, serverPort, timeout);
        TProtocol protocol = new TCompactProtocol(transport);
        transport.open();

        ShopService.Client client = new ShopService.Client(protocol);
        return client;
    }

    public static void main(String[] args) {
        SimpleClientHolder simpleClientHolder = new SimpleClientHolder();
        ShopService.Client client = null;
        try {
            client = simpleClientHolder.buildClient("localhost", 8081, 1000);
            Set<Item> items = client.queryItems(666);
            log.info("return items = {}", String.valueOf(items));
        } catch (TException e) {
            e.printStackTrace();
        }

        if (null != simpleClientHolder.transport) {
            simpleClientHolder.transport.close();
        }
    }
}

依次运行Server和Client,输出正常。

 

IDL (Interface Description Language)

提供服务的第一步是用IDL编写Thrift文件,IDL几乎可以描述接口所需的所有元素,接口定义中包括以下内容:

 

namespace

每个thrift文件都在自己的命名空间中,多个thrift文件可以用同一个命名空间作为标识,并指定要使用的语言的generator。

例如:

namespace java me.kavlez.thrift.service
namespace php tutorial

 

基本类型

类型说明
bool布尔类型
i8 (byte)8-bit 有符号整型,对应java的byte
i1616-bit 有符号整型,对应java的short
i3232-bit 有符号整型,对应java的int
i6464-bit 有符号整型,对应java的long
double64-bit 浮点类型,对应java的double
string字符串
binaryBlob (byte array)

 

结构体

用于定义一个对象类型。

字段默认为optional,可以声明required。
字段可以设置默认值。
结构体之间可以互相引用。
0.9.2开始可以引用自身。

struct Shop {
  1: required i32 id,
  2: required string name
}

struct Item {
  1: required i32 id,
  2: required string name = "unknown",
  3: required string detail,
  4: required Shop shop
}

 

枚举

值是可选项,枚举不能嵌套;基本上就是K、V的形式,不能描述太复杂的枚举类。

enum Numberz {
  ONE = 1,
  TWO,
  THREE,
  FIVE = 5,
  SIX,
  EIGHT = 8
}

 

常量

可以自定义常量,像Map、List这样的复杂结构可以用json表示。

const i32 INT_CONST = 1234;    // a
const map<string,string> MAP_CONST = {"hello": "world", "goodnight": "moon"}
const list<string> LIST_CONST = ["a","b","c"]

 

容器类型

不支持异构容器,容器的元素类型必须一致。
元素类型可以是service以外的任何类型。

类型说明
map<t1,t2>Map from one type to another
list<t1>Ordered list of one type
set<t1>Set of unique elements of one type

 

自定义异常

语法上和struct相似,生成后的代码,不同语言各有各的实现方式。

exception IllegalShopException {
  1: i32 errorCode,
  2: string message,
  3: Shop shop
}

 

service

一个函数集合,语法和java定义接口的语法类似,下面是一些例子。

service ThriftTest {
    
  /**
   * 无返回,空参数列表
   */
  void testVoid(),
    
  /**
   * 声明返回类型、参数
   */
  string testString(1: string thing),
  
  /**
   * 返回结构体
   */
  Shop queryShopInfo(1: i32 id),
    
  /**
   * 结构体作为参数
   */
  bool isValidShop(1: Shop shop),
    
  /**
   * ...
   */
  set<Item> queryItems(1: i32 shopId),
    
  /**
   * 抛出异常
   */
  bool changeShopStatus(1: i32 shopId) throws(1: IllegalShopException err),
  
  /**
   * 多异常
   */
  bool changeItemStatus(1: i32 itemId) throws(1: IllegalShopException shopErr,2:IllegalItemException itemErr),
    
  /**
   * oneway表示该方法在客户端发起请求后不会等待响应,返回类型必须为void
   */
  oneway void sendMessage(1:i32 shopId,2:string message)
}

 

thrift working stack

用Thrift构建服务和客户端,架构如下:

+-------------------+       +-------------------+
|      Server       |       |      Client       |
|                   |       |                   |
| +---------------+ |       | +---------------+ |
| |               | |       | |               | |
| |   your code   | |       | |   your code   | |
| +---------------+ |       | +---------------+ |
| |   Service     | |       | |   Service     | |
| |   processor   | |       | |   Client      | |
| +---------------+ |       | +---------------+ |
| |               | |       | |               | |
| |   Protocol    | |       | |   Protocol    | |
| +---------------+ |       | +---------------+ |
| |               | |       | |               | |
| |   Transport   |<--------->|   Transport   | |
| +---------------+ |       | +---------------+ |
+-------------------+       +-------------------+

 

生成的接口类中大致包括三样,分别是Iface、Client、Processor。
另外还有Server、Transport、Protocol。

Transport

在RPC框架的语境下谈传输层很容易只想到网络通信,但Transport表述的并不只是网络通信。

不如说Transport是多种IO的抽象,其不仅限于网络IO。

比如,基础的TIOStreamTransport,以及其两个子类,TSocket和TZlibTransport。

TSocket在上面的例子中作为TBinaryProtocol依赖的transport类型,与Server的TServerSocket进行通信。 

但后者是封装了InflaterInputStream和DeflaterOutputStream,其InputStream并不要求是SocketInputStream。

从开发角度来讲,如果将一个TMemoryBuffer对象传入Protocol,并以此创建某个service对应的Client,再调用相应接口。

整个过程在代码上并没有什么限制,只是运行时抛出org.apache.thrift.TApplicationException。

 

 

Protocol

protocol依赖transport,决定双方以什么协议通信,同时也是通信内容的载体。

org.apache.thrift.protocol.TProtocol中的方法声明里,一系列readXX和writeXX,在具体实现中通常都是通过transport来完成。

以TJSONProtocol为例,其实现的TProtocol的所有write方法都是以几个私有的write方法组织起来。

比如,writeI32和writeI64都是通过私有方法writeJSONInteger,而writeJSONInteger则是由实例化时传入的trasnport进行write。

 

 

Processor

构建自己的server时需要在tArgs提供一个Processor,比如本文中的ShopService.Processor。
(p.s. 如果需要提供多个Processor,比如再加一个ItemService,则使用TMultiplexedProcessor即可。)

Server通过Processor执行业务逻辑代码,文件中描述的每个函数作为ProcessFunction子类进行实例化,放入Processor的processMap中。

Server收到请求,从输入的protocol中读取方法名,根据方法名从processMap中拿到对应的ProcessFunction;
通过ProcessFunction的process方法执行业务逻辑,过程大体分为3步:

  • 从protocol读入请求参数,构建参数对象;
  • 传入参数,本地执行业务方法。假设方法名为"getItems",调用结果则为getItems_success;
  • 将结果写入protocol,调用protocol.writeXX;

Client

像本文中,指定Transport和Protocol,构建ShopService.Client,客户端通过Client对象像调用本地方法一样调用queryItems;
在ShopService中,Client类同样实现了ShopService.Iface中的方法,以queryItems为例,其实现如下:

public Shop queryShopInfo(int id) throws org.apache.thrift.TException {
  send_queryShopInfo(id);
  return recv_queryShopInfo();
}

 

在send_queryShopInfo,构建该函数对应的xx_args对象,将其写入oprot,并通过oprot.tranport进行flush;

相应地,recv_queryShopInfo就是从iport中读取函数的返回值,构建该函数对应的queryShopInfo_result对象。

 

Server

将Transport、Protocol和Processor集合在一起就是一个完整的Server,父类TServer提供了唯一的抽象方法——serve()。

以TSimpleServer为例,serve中通过java.lang.ServerSocket的accept获取client Socket并转为client Transport,以此获取相应的Processor、创建相应的inputTransport、outputTransport和iProt、oProt。

(p.s. 默认的TProcessorFactory没有子类,其getProcessor(Transport)和并没有通过transport来获取processor。可以用来扩展,比如用一个server提供多版本服务之类的。)

剩下的工作由Processor进行处理,从iPort读入请求信息并构造TMessage,找到相应的ProcessFunction并执行其process方法,这个在上面说过。

Thrift为TServer提供了3种实现:

  • TSimpleServer: 单线程ServerSocket实现,仅用于测试;

  • TThreadPoolServer: 封装了ThreadPoolExecutor,用内部类WorkerProcess表示单个请求,通过每个WorkerProcess对象的transport获取相应的Processor和Protocol,调用业务代码并返回;

  • AbstractNonblockingServer: 非阻塞server抽象类,其serve()方法即整个过程的skeleton,serve()中调用的方法交给其子类提供具体实现。

    public void serve() {
        // start any IO threads
        if (!startThreads()) {
          return;
        }
    
        // start listening, or exit
        if (!startListening()) {
          return;
        }
    
        setServing(true);
    
        // this will block while we serve
        waitForShutdown();
    
        setServing(false);
    
        // do a little cleanup
        stopListening();
    }

 

AbstractNonblockingServer的3个子类,分别为:

  • TNonblockingServer: 实现父类的startThreads(),启动selector线程(也就是SelectAcceptThread,父类声明了protected final Selector selector),开始轮询SelectedKeys,检查状态并进行相应处理:

    if (key.isAcceptable()) {
        handleAccept();
    } else if (key.isReadable()) {
    handleRead(key);
    } else if (key.isWritable()) {
    handleWrite(key);
    } else {
    LOGGER.warn("Unexpected state in select! " + key.interestOps());
    }

    另外,使用TNonblockingServer时transport必须为TFramedTransport,以此保证能正确读取单次方法调用。

  • THsHaServer: "HsHa",即"Half-Sync/Half-Async",是TNonblockingServer的子类。

    工作流程和TNonblockingServer相似,主要区别在与handleRead()
    handleRead中完成读取后,另外一项重要的工作就是requestInvoke(buffer),也就是执行processor.process(iProt,oProt)。

    不过,TNonblockingServer是单线程执行,而THsHaServer则是通过线程池。
    将FrameBuffer装进Invocation(其run方法即frameBuffer.invoke()),提交给线程池处理。

    线程池参数的默认值如下:

    corePoolSize = 5;
    maximumPoolSize = Integer.MAX_VALUE;
    keepAliveTime = 60;
    workQueue = new LinkedBlockingQueue<Runnable>();
  • TThreadedSelectorServer: 进一步加强HsHaServer,用一个AcceptThread接收所有连接请求,并担任负载均衡的角色。

    负载均衡的工作由构造器参数中的SelectorThreadLoadBalancer进行,该类只提供了一种实现——对已注册的selector线程列表进行round robin。
    AcceptThread处理连接时,通过SelectorThreadLoadBalancer选出selector线程,将接收到的socketChannel放入selector线程的队列中。

    虽然TThreadedSelectorServer的requestInvoke也是使用线程池进行,但线程池的默认配置和THsHaServer不同,默认时为corePoolSize为5的FixedThreadPool。
    如果corePoolSize小为0,则由caller线程执行。

 

最后,把之前的例子修改一下,看看效果。

AbstractTServerHolder.java

package me.kavlez.thrift.server;

import org.apache.thrift.server.TServer;
import org.apache.thrift.transport.TTransportException;

public abstract class AbstractTServerHolder {
    private TServer tServer;

    public abstract TServer build() throws TTransportException;
}

ThreadedSelectorServerHolder.java

package me.kavlez.thrift.server;

import me.kavlez.thrift.service.ShopService;
import me.kavlez.thrift.service.impl.ShopServiceImpl;
import org.apache.thrift.protocol.TBinaryProtocol;
import org.apache.thrift.server.TServer;
import org.apache.thrift.server.TThreadedSelectorServer;
import org.apache.thrift.transport.TFramedTransport;
import org.apache.thrift.transport.TNonblockingServerSocket;
import org.apache.thrift.transport.TNonblockingServerTransport;
import org.apache.thrift.transport.TTransportException;

public class ThreadedSelectorServerHolder extends AbstractTServerHolder {
    @Override
    public TServer build() throws TTransportException {
        TNonblockingServerTransport transport = new TNonblockingServerSocket(8090);
        TThreadedSelectorServer.Args args = new TThreadedSelectorServer.Args(transport);

        ShopService.Processor<ShopService.Iface> shopServiceProcessor
                = new ShopService.Processor<ShopService.Iface>(new ShopServiceImpl());
        args.processor(shopServiceProcessor)
                .protocolFactory(new TBinaryProtocol.Factory())
                .transportFactory(new TFramedTransport.Factory());

        TServer server = new TThreadedSelectorServer(args);
        return server;

    }
}

Launcher.java

package me.kavlez.thrift;

import lombok.extern.slf4j.Slf4j;
import me.kavlez.thrift.client.AbstractShopServiceClientHolder;
import me.kavlez.thrift.client.NonBlockingClientHolder;
import me.kavlez.thrift.client.ShopServiceClientAgent;
import me.kavlez.thrift.server.AbstractTServerHolder;
import me.kavlez.thrift.server.ThreadedSelectorServerHolder;
import me.kavlez.thrift.service.Item;
import me.kavlez.thrift.service.ShopService;
import org.apache.thrift.TException;
import org.apache.thrift.server.TServer;
import org.apache.thrift.transport.TTransportException;

import java.io.FileNotFoundException;
import java.util.Set;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.Future;

@Slf4j
public class Launcher {

    static class TServerClientHolderPair {
        private AbstractTServerHolder tServerHolder;
        private Class<? extends AbstractShopServiceClientHolder> clientHolderClass;

        public TServerClientHolderPair(AbstractTServerHolder tServerHolder, Class<? extends AbstractShopServiceClientHolder> clientHolderClass) {
            this.tServerHolder = tServerHolder;
            this.clientHolderClass = clientHolderClass;
        }
    }

    public static void main(String[] args) throws InterruptedException, TTransportException, FileNotFoundException {

        final AbstractTServerHolder serverHolder = new ThreadedSelectorServerHolder();
        final TServer tServer = serverHolder.build();

        ExecutorService executorService = Executors.newCachedThreadPool();
        Future<?> serverFuture = executorService.submit(new Runnable() {
            @Override
            public void run() {
                tServer.serve();
            }
        });

        Thread.sleep(100);

        int times = 10;
        final CountDownLatch countDownLatch = new CountDownLatch(times);

        class ShopServiceClientTask implements Runnable {

            @Override
            public void run() {
                AbstractShopServiceClientHolder clientHolder = null;
                clientHolder = new NonBlockingClientHolder();

                try {
                    ShopService.Iface shopService = new ShopServiceClientAgent(clientHolder.build());
                    for (int i = 0; i < 1000; i++) {
                        Set<Item> items = shopService.queryItems(666);
                        log.info("return items = {}", String.valueOf(items));
                    }

                } catch (TException e) {
                    log.info("thread name={} get TException", Thread.currentThread().getName(), e);
                } finally {
                    clientHolder.close();
                    countDownLatch.countDown();
                }
            }
        }

        long start = System.currentTimeMillis();

        for (int i = 0; i < times; i++) {
            executorService.submit(new ShopServiceClientTask());
        }

        countDownLatch.await();
        log.info("used {} ms ", System.currentTimeMillis() - start);
        tServer.setShouldStop(true);
        tServer.stop();
        executorService.shutdown();

    }
}

转载于:https://www.cnblogs.com/kavlez/p/thrift-tutorial.html

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值