Spring 5.0.0框架介绍_中文版_3.6

文章作者:Tyan
博客:noahsnail.com  |  CSDN  |  简书

3.6 定制bean特性

3.6.1 生命周期回调

为了与容器中bean生命周期的管理进行交互,你可以实现Spring的InitializingBeanDisposableBean接口。当初始化beans时容器会调用InitializingBean中的afterPropertiesSet()方法,当销毁beans时容器会调用DisposableBean中的destroy()方法,在这两个方法中bean可以执行特定的行为。

在现代Spring应用中,通常认为JSR-250的@PostConstruct@PreDestroy注解是最佳实践接收生命周期回调函数的方法。使用这些注解意味着你的bean没有耦合Spring特定的接口。更多细节请看3.9.8小节,”@PostConstruct和@PreDestroy”。

如果你不想使用JSR-250注解,但你仍要注意解耦,可以考虑使用对象定义元数据中的初始化方法和销毁方法。

在Spring内部,Spring框架使用BeanPostProcessor实现来处理任何它能发现的回调接口并调用合适的方法。如果你需要定制Spring不能提供的开箱即用的功能或其它生命周期行为,你可以自己实现BeanPostProcessor。更多信息请看3.8小节,”容器扩展点”。

除了初始化回调函数和销毁回调函数之外,Spring管理的对象也可以实现Lifecycle接口,这些对象可以参与容器自身生命周期驱动的启动和关闭过程。

本节描述了生命周期回调接口。

初始化回调函数

org.springframework.beans.factory.InitializingBean接口在容器设置了bean所有的必须属性之后,允许bean执行初始化工作。InitializingBean接口指定了一个方法:

1
void afterPropertiesSet() throws Exception;

建议你不使用InitializingBean接口,因为它对代码与Spring进行了不必要的耦合。作为一种替代方法,你可以使用@PostConstruct注解或指定一个POPJO的初始化方法。在基于XML配置元数据的情况下,你可以使用init-method特性来指定方法的名称,方法是没有返回值和参数的。如果使用Java配置,你可以使用@BeaninitMethod特性,请看”接收生命周期回调函数”小节。例如,下面的代码:

1
<bean id="exampleInitBean" class="examples.ExampleBean" init-method="init"/>
1
2
3
4
5
6
7
public class ExampleBean {

public void init() {
// do some initialization work
}

}

等价于:

1
2
3
4
5
6
7
public class AnotherExampleBean implements InitializingBean {

public void afterPropertiesSet() {
// do some initialization work
}

}

但没有与Spring代码耦合。

销毁回调函数

实现org.springframework.beans.factory.DisposableBean接口允许容器包含的bean销毁时调用回调函数。DisposableBean接口指定了一个方法:

1
void destroy() throws Exception;

建议你不使用DisposableBean回调接口,因为它对代码与Spring进行了不必要的耦合。作为一种替代方法,你可以使用@PreDestroy注解或指定一个bean定义支持的通用方法。在基于XML配置元数据的情况下,你可以使用<bean/>destroy-method特性。如果使用Java配置,你可以使用@BeandestroyMethod特性,请看”接收生命周期回调”小节。例如,下面的定义:

1
<bean id="exampleInitBean" class="examples.ExampleBean" destroy-method="cleanup"/>
1
2
3
4
5
6
7
public class ExampleBean {

public void cleanup() {
// do some destruction work (like releasing pooled connections)
}

}

等价于:

1
<bean id="exampleInitBean" class="examples.AnotherExampleBean"/>
1
2
3
4
5
6
7
public class AnotherExampleBean implements DisposableBean {

public void destroy() {
// do some destruction work (like releasing pooled connections)
}

}

但没有与Spring代码耦合。

默认初始化和销毁方法

当你编写初始化回调函数和析构回调函数时,不要使用Spring特定的InitializingBeanDisposableBean回调接口,自己编写方法,方法名通常为init()initialize()dispose()等等。理想情况下,这种生命周期回调方法的名称在整个工程中是标准化的,以便所有开发人员使用同样的方法名称,保证一致性。

你可以配置Spring容器查找每个bean的初始化方法和析构方法时的名字。这意味着,作为一个应用开发者,你可以编写应用程序类并使用名为init()的初始化回调方法,而不必在每个bean定义中配置init-method="init"特性。当bean创建时,Spring Ioc容器调用这个方法(按照前面描述的标准生命周期回调约定)。这个功能也强制了初始化方法和析构方法命名规范的一致性。

假设你的初始化回调方法名为init(),析构回调方法名为destroy()。你的类应该与下面例子中的类类似。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
public class DefaultBlogService implements BlogService {

private BlogDao blogDao;

public void setBlogDao(BlogDao blogDao) {
this.blogDao = blogDao;
}

// this is (unsurprisingly) the initialization callback method
public void init() {
if (this.blogDao == null) {
throw new IllegalStateException("The [blogDao] property must be set.");
}
}
}
1
2
3
4
5
6
7
<beans default-init-method="init">

<bean id="blogService" class="com.foo.DefaultBlogService">
<property name="blogDao" ref="blogDao" />
</bean>

</beans>

位于顶层<beans/>元素中的default-init-method特性,会让Spring IoC容器将beans中的名为init的方法识别为初始化回调方法。当一个bean创建和组装时,如果bean类有这样一个方法,它会在恰当的时间被调用。

在bean被提供了所有依赖之后,Spring容器确保会立刻调用配置的初始化回调方法。因此初始化回调会在原生bean引用上调用,这意味着AOP拦截器等仍不能应用到bean中。首先要完整的创建目标bean,然后才会应用AOP代理(例如)等拦截器链。如果分别定义了目标bean和代理,你的代码甚至能绕过代理直接与原生的目标bean进行交互。将拦截器应用到初始化方法上可能会产生不一致性,因为这样做会使目标bean的生命周期与它的代理/拦截器相耦合,当你的代码与原生目标bean直接进行交互时,语义会变的很奇怪。

组合生命周期机制

从Spring 2.5开始,在控制bean的生命周期行为时,你有三中选择:InitializingBeanDisposableBean回调接口;定制init()destroy()方法;@PostConstruct@PreDestroy`注解。在控制一个给定bean时你可以组合这些机制。

如果一个bean配置了多生命周期机制,每种机制配置了一个不同的方法名,那么每一个配置的方法会按照下面的顺序列表来执行。但是如果配置了相同的名字——例如,init()初始化方法——不止在一个生命周期机制中配置,那么这个方法只能执行一次,像之前所说的那样。

同一个bean配置了多生命周期机制,并有不同的初始化方法,那么调用顺序如下:

  • 先调用有注解@PostConstruct的方法

  • 然后调用InitializingBean回调接口定义的afterPropertiesSet()方法

  • 最好调用定制配置的init()方法

Destroy methods are called in the same order:

  • Methods annotated with @PreDestroy

  • destroy() as defined by the DisposableBean callback interface

  • A custom configured destroy() method

析构方法按同样的顺序调用:

  • 先调用有@PreDestroy注解的方法

  • 再调用DisposableBean回调接口定义的destroy()方法

  • 最好调用定制配置的destroy()方法

启动和关闭回调

Lifecycle接口定义了任何对象生命周期都需要的基本方法(例如启动和停止一些背景处理):

1
2
3
4
5
6
7
8
9
public interface Lifecycle {

void start();

void stop();

boolean isRunning();

}

任何Spring管理的对象都可以实现那个接口。当ApplicationContext本身收到启动启动和关闭信号时,例如运行时关闭/再启动场景,它将级联调用所有的上下文定义的Lifecycle实现。它通过委托LifecycleProcessor来完成这个功能:

1
2
3
4
5
6
7
public interface LifecycleProcessor extends Lifecycle {

void onRefresh();

void onClose();

}

注意LifecycleProcessor本身是Lifecycle接口的一个扩展。它也添加了两个其它的方法来响应上下文的再刷新和关闭的。

注意正规的org.springframework.context.Lifecycle接口只是一个显式启动/关闭通知的协议,并不意味着在上下文刷新时自动启动。考虑实现org.springframework.context.SmartLifecycle接口来实现对指定bean自动启动的细粒度控制(包括启动时期)。请注意停止通知不能保证在销毁之前到来:在正式关闭时,所有的Lifecycle beans在通常的析构回调传播之前首先会收到停止通知;但是,在上下文使用期间进行热刷新或尝试取消再刷新,只会调用析构方法。

启动和关闭的调用顺序是很重要的。如果任何两个对象间存在一个”depends-on”关系,那么依赖关系将在它的依赖之后开始,在它的依赖之前停止。然而有时直接的依赖关系是未知的。你可能只知道某个类型的对象应该在另一个类型的对象之前启动。在那种情况下,SmartLifecycle接口定义了另一种选择,也就是说getPhase()定义在它的父接口Phased中。

1
2
3
public interface Phased {
int getPhase();
}
1
2
3
4
public interface SmartLifecycle extends Lifecycle, Phased {
boolean isAutoStartup();
void stop(Runnable callback);
}

当开始时,最低相位的对象先启动,当停止时,最高相位的对象先停止。因此,实现了SmartLifecycle接口,getPhase()方法返回值为Integer.MIN_VALUE的对象将最先启动并最后停止。另一方面,相位值Integer.MAX_VALUE表明对象应该最后启动,最先停止(可能是因为它依赖其它运行的进程)。当考虑相位值时,知道任何没有实现SmartLifecycle接口的Lifecycle对象的默认值为0是很重要的。因此,任何负相位值表示对象应该在那么标准组件之前启动(在它们之后停止),反之为任何正相位值。

正如你看到的,在SmartLifecycle中定义的停止方法接收一个回调函数。任何实现在关闭进程完成之后都必须调用回调的run()方法。当需要时这可以进行异步关闭,因为LifecycleProcessor接口、DefaultLifecycleProcessor接口的默认实现会等待每个阶段的对象组直到达到超时值,然后调用回调函数。默认每个阶段的超时值为30秒。你可以在上下文中通过定义名为”lifecycleProcessor”的bean来覆盖默认的生命周期处理器实例。如果你只想修改超时值,如下定义是足够的:

1
2
3
4
<bean id="lifecycleProcessor" class="org.springframework.context.support.DefaultLifecycleProcessor">
<!-- timeout value in milliseconds -->
<property name="timeoutPerShutdownPhase" value="10000"/>
</bean>

像上面提到的那样,LifecycleProcessor接口为再刷新和上下文的关闭也定义了回调方法。后者会简单的驱动关闭进程就像显式的调用了stop()方法一样,但当上下文关闭时它才会发生。另一方面refresh回调能使SmartLifecycle beans的另一个功能可用。当上下文再刷新时(所有对象已经实例化并初始化),回调函数将被调用,那时默认的生命周期处理器将会检查每个SmartLifecycle对象的isAutoStartup()方法返回的布尔值。如果为true,对象将会在那时启动而不是等待上下文的显式调用或它自己的start()方法(不像上下文再刷新,对于一个标准的上下文实现上下启动不会自动发生)。”phase”值以及”depends-on”关系将决定启动顺序,像上面描述的一样。

在非web应用中妥善的关闭Spring IoC容器

这一节只应用于非web应用。Spring的基于web的ApplicationContext实现已经有代码来处理当相关的web应用关闭时,妥善关闭Spring IoC容器的问题。

如果你在非web应用环境使用Spring的IoC容器;例如,在一个富桌面客户端环境中,你在JVM中注册一个关闭钩子。这样做确保了妥善的关闭,为了释放所有资源需要调用与单例beans相关的析构方法。当然,你仍然必须正确的配置和实现这些销毁回调函数。

为了注册一个关闭钩子,你可以调用ConfigurableApplicationContext接口中声明的registerShutdownHook()方法:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
import org.springframework.context.ConfigurableApplicationContext;
import org.springframework.context.support.ClassPathXmlApplicationContext;

public final class Boot {

public static void main(final String[] args) throws Exception {

ConfigurableApplicationContext ctx = new ClassPathXmlApplicationContext(
new String []{"beans.xml"});

// add a shutdown hook for the above context...
ctx.registerShutdownHook();

// app runs here...

// main method exits, hook is called prior to the app shutting down...

}
}

3.6.2 ApplicationContextAware和BeanNameAware

ApplicationContext创建一个实现org.springframework.context.ApplicationContextAware接口的对象实例时,这个实例会提供一个ApplicationContext的引用。

1
2
3
4
5
public interface ApplicationContextAware {

void setApplicationContext(ApplicationContext applicationContext) throws BeansException;

}

因此beans可以以编程方式操纵创建它们的ApplicationContext,通过ApplicationContext接口,或通过将引用抛给这个接口的一个已知子类,例如ConfigurableApplicationContext,它暴露了额外的功能。一个方法是编程式检索其他的bean。有时这个能力是很有用的,但是通常你应该避免使用它,因为它耦合了代码和Spring,不能遵循控制反转的风格,在控制反转中协作者是作为属性提供给beans的。ApplicationContext的其它方法提供了对文件资源的访问,发布应用事件,访问MessageSource的功能。这些额外的特性将在3.15小节『ApplicationContext”的额外能力』中描述。

从Spring 2.5起,自动装配是另一种可替代的获得ApplicationContext引用的方法。『传统的』constructorbyType自动装配模式(如3.4.5小节所述,『自动装配协作者』)可以分别为构造函数参数或setter方法参数提供ApplicationContext类型的依赖。更多的灵活性包括自动装配变量的能力和多参数方法,使用新的基于注解的自动装配特性。如果你这一做的话,ApplicationContext可以被自动装配到变量中,构造函数参数中或方法参数中,如果讨论的变量,构造函数或方法有@Autowired注解,那么可以期望它是ApplicationContext类型。更多信息请看3.9.2小节,@autowired

ApplicationContext创建一个实现了org.springframework.beans.factory.BeanNameAware接口的类时,类中有相关的对象定义中定义的名称的引用。

1
2
3
4
5
public interface BeanNameAware {

void setBeanName(String name) throws BeansException;

}

在正常的bean属性填入之后,回调方法调用,但在初始化回调方法之前,例如InitializingBean的afterPropertiesSet或一个定制的初始化方法。

3.6.3 其它的Aware接口

除了上面讨论的ApplicationContextAwareBeanNameAware之外,Spring给予了一系列Aware接口来允许beans向容器表明它们需要一个确定的基础结构依赖。最重要的Aware接口总结如下——作为一个通用规则,名字是依赖类型的一个很好暗示:

表3.4. Aware接口

Name Injected Dependency Explained in
ApplicationContextAware 声明ApplicationContext Section 3.6.2, “ApplicationContextAware and BeanNameAware”
ApplicationEventPublisherAware 封装事件发布的ApplicationContext Section 3.15, “Additional Capabilities of the ApplicationContext”
BeanClassLoaderAware 用来加载bean的类加载器 Section 3.3.2, “Instantiating beans”
BeanFactoryAware 声明BeanFactory Section 3.6.2, “ApplicationContextAware and BeanNameAware”
BeanNameAware 声明的bean的名字 Section 3.6.2, “ApplicationContextAware and BeanNameAware”
BootstrapContextAware 容器运行的资源自适应BootstrapContext. 通常只在JCA aware ApplicationContexts可获得 Chapter 28, JCA CCI
LoadTimeWeaverAware 加载时为处理类定义定义的weaver Section 7.8.4, “Load-time weaving with AspectJ in the Spring Framework”
MessageSourceAware 解析消息配置策略 (支持参数化和国际化) Section 3.15, “Additional Capabilities of the ApplicationContext”
NotificationPublisherAware Spring JMX通知发布器 Section 27.7, “Notifications”
ResourceLoaderAware 为底层访问资源配置的加载器 Chapter 4, Resources
ServletConfigAware 容器运行的当前ServletConfig。 仅在web感知的Spring ApplicationContext中有效 Chapter 18, Web MVC framework
ServletContextAware 容器运行的当前ServletContext。 仅在web感知的Spring ApplicationContext中有效 Chapter 18, Web MVC framework

注意这些接口的用法将你的代码与Spring进行了捆绑,不符合控制反转的风格。因此,它们是为那么需要以编程方式访问容器的基础结构beans推荐的。

如果有收获,可以请我喝杯咖啡!