本文节选自《Spring 5核心原理》
在之前的源码分析中我们已经了解到,依赖注入(DI)的入口是getBean()方法,前面的IoC手写部分基本流程已通。先在GPApplicationContext中定义好IoC容器,然后将GPBeanWrapper对象保存到Map中。在GPApplicationContext中设计两个Map:factoryBeanObjectCache保存单例对象的缓存,factoryBeanInstanceCache保存GPBeanWrapper的缓存,变量命名也和原生Spring一致,这两个对象的设计其实就是注册式单例模式的经典应用。
public class GPApplicationContext extends GPDefaultListableBeanFactory implements GPBeanFactory {
private String [] configLocations;
private GPBeanDefinitionReader reader;
//用来保证注册式单例的容器
private Map<String,Object> factoryBeanObjectCache = new HashMap<String, Object>();
//用来存储所有的被代理过的对象
private Map<String,GPBeanWrapper> factoryBeanInstanceCache = new ConcurrentHashMap<String, GPBeanWrapper>();
...
}
下面我们从完善getBean()方法开始:
@Override
public Object getBean(String beanName) {
GPBeanDefinition beanDefinition = super.beanDefinitionMap.get(beanName);
try{
//生成通知事件
GPBeanPostProcessor beanPostProcessor = new GPBeanPostProcessor();
Object instance = instantiateBean(beanDefinition);
if(null == instance){ return null;}
//在实例初始化以前调用一次
beanPostProcessor.postProcessBeforeInitialization(instance,beanName);
GPBeanWrapper beanWrapper = new GPBeanWrapper(instance);
this.factoryBeanInstanceCache.put(beanName,beanWrapper);
//在实例初始化以后调用一次
beanPostProcessor.postProcessAfterInitialization(instance,beanName);
populateBean(beanName,instance);
//通过这样调用,相当于给我们自己留有了可操作的空间
return this.factoryBeanInstanceCache.get(beanName).getWrappedInstance();
}catch (Exception e){
// e.printStackTrace();
return null;
}
}
//传一个BeanDefinition,就返回一个实例Bean
private Object instantiateBean(GPBeanDefinition beanDefinition){
Object instance = null;
String className = beanDefinition.getBeanClassName();
try{
//因为根据Class才能确定一个类是否有实例
if(this.factoryBeanObjectCache.containsKey(className)){
instance = this.factoryBeanObjectCache.get(className);
}else{
Class<?> clazz = Class.forName(className);
instance = clazz.newInstance();
this.factoryBeanObjectCache.put(beanDefinition.getFactoryBeanName(),instance);
}
return instance;
}catch (Exception e){
e.printStackTrace();
}
return null;
}
private void populateBean(String beanName,Object instance){
Class clazz = instance.getClass();
if(!(clazz.isAnnotationPresent(GPController.class) ||
clazz.isAnnotationPresent(GPService.class))){
return;
}
Field [] fields = clazz.getDeclaredFields();
for (Field field : fields) {
if (!field.isAnnotationPresent(GPAutowired.class)){ continue; }
GPAutowired autowired = field.getAnnotation(GPAutowired.class);
String autowiredBeanName = autowired.value().trim();
if("".equals(autowiredBeanName)){
autowiredBeanName = field.getType().getName();
}
field.setAccessible(true);
try {
field.set(instance,this.factoryBeanInstanceCache.get(autowiredBeanName). getWrappedInstance());
} catch (IllegalAccessException e) {
// e.printStackTrace();
}
}
}
原生Spring中的BeanPostProcessor是为对象初始化事件设置的一种回调机制。这个Mini版本中只做说明,不做具体实现,感兴趣的“小伙伴”可以继续深入研究Spring源码。
package com.tom.spring.formework.beans.config;
public class GPBeanPostProcessor {
//为在Bean的初始化之前提供回调入口
public Object postProcessBeforeInitialization(Object bean, String beanName) throws Exception {
return bean;
}
//为在Bean的初始化之后提供回调入口
public Object postProcessAfterInitialization(Object bean, String beanName) throws Exception {
return bean;
}
}
至此,DI部分就手写完成了,也就是说完成了Spring的核心部分。“小伙伴们”是不是发现其实还是很简单的?
本文为“Tom弹架构”原创,转载请注明出处。