策略模式

定义

定义一簇算法类,将每个算法分别封装起来,让他们可以互相替换,策略模式可以使算法的变化独立于使用它们的客户端

场景

使用策略模式,可以避免冗长的if-else 或 switch分支判断

实现

  1. 策略的定义

    策略的定义需要定义一个策略接口和一组实现这个接口的策略类,因为所有的策略类都实现相同的接口

public interface Strategy{
	void algorithm();
}

public class ConcreteStrategyA implements Strategy {
 @Override
 public void algorithm() {
 //具体的算法...
 }
}
public class ConcreteStrategyB implements Strategy {
 @Override
 public void algorithm() {
 //具体的算法...
 }
}
  1. 策略的创建

    在使用的时候,一般会通过类型来判断创建哪个策略来使用,在策略上下文中,可以使用map维护好策略类

  2. 策略的使用

    策略模式包含一组可选策略,在使用策略时,一般如何确定使用哪个策略呢?最常见的是运行时动态确定使用哪种策略。程序在运行期间,根据配置、计算结果、网络等这些不确定因素,动态决定使用哪种策略

public class StrategyContext{
	private static final Map<String, Strategy> strategies = new HashMap<>();

	static {
     strategies.put("A", new ConcreteStrategyA());
     strategies.put("B", new ConcreteStrategyB());
    }

    private static Strategy getStrategy(String type) {
         if (type == null || type.isEmpty()) {
    	     throw new IllegalArgumentException("type should not be empty.");
         }
         return strategies.get(type);
	}

    public void algorithm(String type){
        Strategy strategy = this.getStrategy(type);
        strategy.algorithm();
    }
}

UML

Spring 实现策略模式--自定义注解方式解耦if...else-LMLPHP

策略模式的创建和使用--Spring和自定义注解

在介绍策略模式时,在上下文中使用了map存储好的策略实例,在根据type获取具体的策略,调用策略算法。
当需要添加一种策略时,需要修改context代码,这违反了开闭原则:对修改关闭,对扩展开放。

要实现对扩展开放,就要对type和具体的策略实现类在代码中进行关联,可以使用自定义注解的方式,在注解中指定策略的type。
策略上下文实现类实现 BeanPostProcessor 接口,在该接口中编写策略类型与bean的关系并维护到策略上下文中。

package com.masterlink.strategy;

import lombok.extern.slf4j.Slf4j;
import org.springframework.aop.support.AopUtils;
import org.springframework.beans.BeansException;
import org.springframework.beans.factory.config.BeanPostProcessor;
import org.springframework.core.Ordered;
import org.springframework.core.annotation.AnnotatedElementUtils;
import org.springframework.stereotype.Component;

import java.util.Collections;
import java.util.Set;
import java.util.concurrent.ConcurrentHashMap;

@Slf4j
@Component
public class StrategyDemoBeanPostProcessor implements BeanPostProcessor, Ordered {

    private final Set<Class<?>> nonAnnotatedClasses = Collections.newSetFromMap(new ConcurrentHashMap<>(64));

    private final StrategyContext strategyContext;

    private StrategyDemoBeanPostProcessor(StrategyContext context) {
        this.strategyContext = context;
    }

    @Override
    public int getOrder() {
        return LOWEST_PRECEDENCE;
    }

    @Override
    public Object postProcessAfterInitialization(final Object bean, final String beanName) throws BeansException {

        if (!this.nonAnnotatedClasses.contains(bean.getClass())) {
            // 获取使用 @StrategyDemo 注解的Class信息
            Class<?> targetClass = AopUtils.getTargetClass(bean);
            Class<Strategy> orderStrategyClass = (Class<Strategy>) targetClass;
            StrategyDemo ann = findAnnotation(targetClass);
            if (ann != null) {
                processListener(ann, orderStrategyClass);
            }
        }
        return bean;
    }

    @Override
    public Object postProcessBeforeInitialization(Object bean, String beanName) throws BeansException {
        return bean;
    }

    protected void processListener(StrategyDemo annotation,
                                   Class<Strategy> classes) {
        // 注册策略
        this.strategyContext
                .registerStrategy(annotation.type(), classes);
    }

    private StrategyDemo findAnnotation(Class<?> clazz) {

        StrategyDemo ann = AnnotatedElementUtils.findMergedAnnotation(clazz, StrategyDemo.class);
        return ann;
    }

}


@Component
public class StrategyContext implements ApplicationContextAware {
    private final Map<String, Class<Strategy>> strategyClassMap = new ConcurrentHashMap<>(64);

    private final Map<String, Strategy> beanMap = new ConcurrentHashMap<>(64);

    private ApplicationContext applicationContext;

    /**
     * 注册策略
     * @param type
     * @param strategyClass
     */
    public void registerStrategy(String type, Class<Strategy> strategyClass){
        if (strategyClassMap.containsKey(type)){
            throw new RuntimeException("strategy type:"+type+" exist");
        }
        strategyClassMap.put(type, strategyClass);
    }

    /**
     * 执行策略
     * @param type
     */
    public void algorithm(String type){
        Strategy strategy = this.getStrategy(type);
        strategy.algorithm();
    }

    private Strategy getStrategy(String type) {
        if (type == null || type.isEmpty()) {
            throw new IllegalArgumentException("type should not be empty.");
        }
        Class<Strategy> strategyClass = strategyClassMap.get(type);
        return createOrGetStrategy(type, strategyClass);
    }

    private Strategy createOrGetStrategy(String type,Class<Strategy> strategyClass ){
        if (beanMap.containsKey(type)){
            return beanMap.get(type);
        }
        Strategy strategy = this.applicationContext.getBean(strategyClass);
        beanMap.put(type, strategy);
        return strategy;
    }

    @Override
    public void setApplicationContext(ApplicationContext applicationContext) throws BeansException {
        this.applicationContext = applicationContext;
    }
}

实用案例

在我们的平台中,有一部分是使用的netty框架编写的tcp服务,在服务端,需要将二进制转换为对象,在协议设计阶段,定义第一个字节表示对象类型,比如int,String等,第二三个字节,表示数据长度,后面的字节位传输内容。
比如,
0x01, 0x00, 0x04, 0x00, 0x00, 0x00, 0x09,解析出来的内容是int类型数字9。
0x02, 0x00, 0x03, 0x31, 0x32, 0x33, 解析出的内容是String类型,内容是 123。
在不使用策略模式的时候,需要将第一个字节解析出来,然会使用if--else判断类型,对后继的字节进行解析。
在实际的实现过程中,是使用了策略模式,并且使用注解的方式表示数据类型,实现过程如下。

定义策略接口和注解

定义 CodecStrategyType 注解和编码解码器的策略接口 CodecStrategy

@Target(ElementType.TYPE)
@Retention(RetentionPolicy.RUNTIME)
@Documented
public @interface CodecStrategyType {
    /**
     * 编码解码类型
     * @return
     */
    byte type();
}

public interface CodecStrategy<T> {
    T decoding(byte[] buffer);
}

/*
* 通用解码接口
 */
public interface Codec {
    Object decoding(byte[] bytes);
}


策略实现

实现两种类型的解码器: Integer  和 String

/**
 * integer解码
 */
@CodecStrategyType(type = (byte)0x01)
@Service
public class IntgerCodecStrategy implements CodecStrategy<Integer> {

    @Override
    public Integer decoding(byte[] buffer) {
        int value;
        value = (int) ((buffer[3] & 0xFF)
                | ((buffer[2] & 0xFF)<<8)
                | ((buffer[1] & 0xFF)<<16)
                | ((buffer[0] & 0xFF)<<24));
        return value;
    }
}

@CodecStrategyType(type = (byte)0x02)
@Service
public class StringCodecStrategy implements CodecStrategy<String> {

    @Override
    public String decoding(byte[] bufferr) {
        return new String(bufferr);
    }
}

策略上下文和策略注册

策略上下文类 CodecStrategyContext 提供了统一解码入口,将 byte[] 转换为 Object 类型,同时提供策略的注解接口 void registerStrategy(Byte type, Class<CodecStrategy<?>> strategyClass) ,注册解码类型对应的策略实现类。
策略上下文类同时还提供了策略Bean的创建,根据类型从Spring 的 ApplicationContext 获取策略bean,并缓存到map。
策略Bean处理类 CodecStrategyTypeBeanPostProcessor 中解析 CodecStrategyType 注解中指定的类型。


@Component
public class CodecStrategyContext implements ApplicationContextAware, Codec {
    private final Map<Byte, Class<CodecStrategy<?>>> strategyClassMap = new ConcurrentHashMap<>(64);

    private final Map<Byte, CodecStrategy<?>> beanMap = new ConcurrentHashMap<>(64);

    private ApplicationContext applicationContext;

    /**
     * 注册策略
     * @param type
     * @param strategyClass
     */
    public void registerStrategy(Byte type, Class<CodecStrategy<?>> strategyClass){
        if (strategyClassMap.containsKey(type)){
            throw new RuntimeException("strategy type:"+type+" exist");
        }
        strategyClassMap.put(type, strategyClass);
    }

    /**
     * 执行策略
     */
    @Override
    public Object decoding(byte[] bytes){
        Byte type = bytes[0];
        CodecStrategy<?> strategy =this.getStrategy(type);
        byte l1 = bytes[1];
        byte l2= bytes[2];
        short length =  (short) ((l2 & 0xFF)
                | ((l1 & 0xFF)<<8));
        byte[] contentBytes = new byte[length];
        arraycopy(bytes,3,contentBytes,0, length);
        return strategy.decoding(contentBytes);
    }

    private CodecStrategy<?> getStrategy(Byte type) {
        Class<CodecStrategy<?>> strategyClass = strategyClassMap.get(type);
        return createOrGetStrategy(type, strategyClass);
    }

    private CodecStrategy<?> createOrGetStrategy(Byte type, Class<CodecStrategy<?>> strategyClass ){
        if (beanMap.containsKey(type)){
            return beanMap.get(type);
        }
        CodecStrategy<?> strategy = this.applicationContext.getBean(strategyClass);
        beanMap.put(type, strategy);
        return strategy;
    }

    @Override
    public void setApplicationContext(ApplicationContext applicationContext) throws BeansException {
        this.applicationContext = applicationContext;
    }
}

@Component
public class CodecStrategyTypeBeanPostProcessor implements BeanPostProcessor, Ordered {

    private final Set<Class<?>> nonAnnotatedClasses = Collections.newSetFromMap(new ConcurrentHashMap<>(64));

    private final CodecStrategyContext strategyContext;

    private CodecStrategyTypeBeanPostProcessor(CodecStrategyContext context) {
        this.strategyContext = context;
    }

    @Override
    public int getOrder() {
        return LOWEST_PRECEDENCE;
    }

    @Override
    public Object postProcessAfterInitialization(final Object bean, final String beanName) throws BeansException {

        if (!this.nonAnnotatedClasses.contains(bean.getClass())) {
            // 获取使用 @StrategyDemo 注解的Class信息
            Class<?> targetClass = AopUtils.getTargetClass(bean);
            Class<CodecStrategy<?>> orderStrategyClass = (Class<CodecStrategy<?>>) targetClass;
            CodecStrategyType ann = findAnnotation(targetClass);
            if (ann != null) {
                processListener(ann, orderStrategyClass);
            }
        }
        return bean;
    }

    @Override
    public Object postProcessBeforeInitialization(Object bean, String beanName) throws BeansException {
        return bean;
    }

    protected void processListener(CodecStrategyType annotation,
                                   Class<CodecStrategy<?>> classes) {
        // 注册策略
        this.strategyContext
                .registerStrategy(annotation.type(), classes);
    }

    private CodecStrategyType findAnnotation(Class<?> clazz) {

        CodecStrategyType ann = AnnotatedElementUtils.findMergedAnnotation(clazz, CodecStrategyType.class);
        return ann;
    }

}

使用和测试

测试Integer和String类型的策略:

  1. 0x01, 0x00, 0x04, 0x00, 0x00, 0x00, 0x09,解析出来的内容是int类型数字9。
  2. 0x02, 0x00, 0x03, 0x31, 0x32, 0x33, 解析出的内容是String类型,内容是 123。

@ExtendWith(SpringExtension.class)
@ContextConfiguration(classes = {CodecStrategyTest.CodecStrategyTestConfig.class})
public class CodecStrategyTest {

    @Resource
    Codec codec;

    @Test
    public void testInterDecoding(){
        byte[] buffer = new byte[]{
                0x01,0x00,  0x04, 0x00, 0x00,0x00, 0x09
        };
        Integer decoding = (Integer)codec.decoding(buffer);
        assertThat(decoding)
                .isNotNull()
                .isEqualTo(9);
    }

    @Test
    public void testStringDecoding(){
        byte[] buffer = new byte[]{
                0x02, 0x00, 0x03, 0x31, 0x32,0x33
        };
        String decoding = (String)codec.decoding(buffer);
        assertThat(decoding)
                .isNotNull()
                .isEqualTo("123");
    }

    @ComponentScan({"com.masterlink.strategy"})
    @Configuration
    public static class CodecStrategyTestConfig {
    }
}

扩展复杂类型

自定义复杂类型User类,对应协议类型为 0xA0, 第2 、3 字节表示整个对象的字段长度,紧接着是 Integer 类型的age 和 String 类型的name,
比如 0xA0, 0x00 0x10 0x00, 0x04, 0x00, 0x00, 0x00, 0x17, 0x00, 0x08, 0x5A,0x68,0x61,0x6E,0x67,0x53, 0x61,0x6E, 对应的user对象是

{
  "age": 23,
  "name": "ZhangSan"
}
@Data
public class User {
    private Integer age;
    private String name;
}

实现解码策略类

已知 User 中的基础类型依赖了 Integer 和 String ,所以在User的解码策略类中,依赖了 IntgerCodecStrategy 和 StringCodecStrategy


@CodecStrategyType(type = (byte) (0xA0))
@Service
public class UserCodeStrategy implements CodecStrategy<User> {
    private final StringCodecStrategy stringCodecStrategy;
    private final IntgerCodecStrategy intgerCodecStrategy;

    public UserCodeStrategy(StringCodecStrategy stringCodecStrategy, IntgerCodecStrategy intgerCodecStrategy) {
        this.stringCodecStrategy = stringCodecStrategy;
        this.intgerCodecStrategy = intgerCodecStrategy;
    }

    @Override
    public User decoding(byte[] buffer) {
        byte ageL1 = buffer[0];
        byte ageL2 = buffer[1];
        short ageLength =  (short) ((ageL2 & 0xFF)
                | ((ageL1 & 0xFF)<<8));
        byte[] ageBytes = new byte[ageLength];
        System.arraycopy(buffer,2, ageBytes,0,ageLength);

        byte nameL1 = buffer[0+ageLength];
        byte nameL2 = buffer[1+ageLength];

        short nameLength =  (short) ((nameL2 & 0xFF)
                | ((nameL1 & 0xFF)<<8));

        byte[] nameBytes = new byte[nameLength];
        System.arraycopy(buffer,2+ageLength+2, nameBytes,0,nameLength);

        User user = new User();
        user.setAge(intgerCodecStrategy.decoding(ageBytes));
        user.setName(stringCodecStrategy.decoding(nameBytes));
        return user;
    }
}

测试

通过测试可以发现很轻松的就扩展了一个复杂类型的解码算法,这样随着协议的增加,可以做到对修改代码关闭,对扩展代码开放,符合开闭原则。


    @Test
    public void testUserDecoding(){
        byte[] buffer = new byte[]{
                (byte)0xA0, (byte)0x00 ,(byte)0x10 ,(byte)0x00, (byte)0x04,
                (byte)0x00, (byte)0x00, (byte)0x00, (byte)0x17, (byte)0x00,
                (byte)0x08, (byte)0x5A, (byte)0x68, (byte)0x61, (byte)0x6E,
                (byte)0x67, (byte)0x53, (byte)0x61, (byte)0x6E
        };
        User user = (User)codec.decoding(buffer);
        assertThat(user)
                .isNotNull();
        assertThat(user.getAge()).isEqualTo(23);
        assertThat(user.getName()).isEqualTo("ZhangSan");
    }

总结

  1. 使用策略模式,可以避免冗长的if-else 或 switch分支判断
  2. 掌握自定义注解的是使用方式
  3. 与使用 @Service("name") 注解相比,自定义注解方式支撑和扩展的类型或更灵活

关注我的公众号,一起探索新知识新技术

Spring 实现策略模式--自定义注解方式解耦if...else-LMLPHP

05-23 04:58