澄澈的眼|这Spring注入对象处理过程也太细了，收藏了，什么神仙( 三 ) 1.1自动装配

不同的后置处理器的postProcessProperties()方法对应的是不同的处理逻辑。
4、@Autowired注解属性4.1处理过程
首先，通过findAutowiringMetadata()方法获取被注入的元数据。以上述：「情形一」，为例：
publicvoidinject(Objecttarget,@NullableStringbeanName,@NullablePropertyValuespvs)throwsThrowable{CollectioncheckedElements=this.checkedElements;//获取被注入的元素CollectionelementsToIterate=(checkedElements!=null?checkedElements:this.injectedElements);if(!elementsToIterate.isEmpty()){//循环被注入的元素，调用inject方法for(InjectedElementelement:elementsToIterate){if(logger.isTraceEnabled()){logger.trace("Processinginjectedelementofbean'"+beanName+"':"+element);}//调用注入方法element.inject(target,beanName,pvs);}}}在这个方法中首先会检查注入的对象，这里需要指出，在@Autowired注解构造器的方式下，最终得到的elementsToIterate是空。
对于@Autowired注解的其他使用方式，最终都会调用element.inject(target,beanName,pvs);
在AutowiredFieldElement中类中的最终通过beanFactory.resolveDependency(desc,beanName,autowiredBeanNames,typeConverter)返回注入的对象，然后通过field.set(bean,value);的方式来维护对象与属性之间的关系。
接下来将分析resolveDependency()方法，通过该方法可以发现，最终调用的是doResolveDependency() 。看到了doXXXX()的方法，就又到了Spring惯用的套路了，这里这方法就是真正做事的方式。下面就来看看在doResolveDependency()中做了什么事情...
publicObjectdoResolveDependency(DependencyDescriptordescriptor,@NullableStringbeanName,@NullableSetautowiredBeanNames,@NullableTypeConvertertypeConverter)throwsBeansException{InjectionPointpreviousInjectionPoint=ConstructorResolver.setCurrentInjectionPoint(descriptor);try{Objectshortcut=descriptor.resolveShortcut(this);if(shortcut!=null){returnshortcut;}/***根据类型获取， @Autowired默认根据type*/Class>type=descriptor.getDependencyType();/***支持@value注解*/Objectvalue=https://pcff.toutiao.jxnews.com.cn/p/20200906/getAutowireCandidateResolver().getSuggestedValue(descriptor);if(value!=null){if(valueinstanceofString){StringstrVal=resolveEmbeddedValue((String)value);BeanDefinitionbd=(beanName!=null&&containsBean(beanName)?getMergedBeanDefinition(beanName):null);value=evaluateBeanDefinitionString(strVal,bd);}TypeConverterconverter=(typeConverter!=null?typeConverter:getTypeConverter());try{/**通过转换器将Bean的值转换为对应的type类型*/returnconverter.convertIfNecessary(value,type,descriptor.getTypeDescriptor());}catch(UnsupportedOperationExceptionex){//AcustomTypeConverterwhichdoesnotsupportTypeDescriptorresolution...return(descriptor.getField()!=null?converter.convertIfNecessary(value,type,descriptor.getField()):converter.convertIfNecessary(value,type,descriptor.getMethodParameter()));}}ObjectmultipleBeans=resolveMultipleBeans(descriptor,beanName,autowiredBeanNames,typeConverter);if(multipleBeans!=null){returnmultipleBeans;}/***根据属性类型找到beanFactory中所有类型匹配的Bean*返回值的结构为：*key：匹配的BeanName；*value：beanName对应的实例化后的bean通过getBean(beanName)返回*/MapmatchingBeans=findAutowireCandidates(beanName,type,descriptor);if(matchingBeans.isEmpty()){/***如果autowire的require属性为true*找到的匹配项为空则抛出异常*/if(isRequired(descriptor)){raiseNoMatchingBeanFound(type,descriptor.getResolvableType(),descriptor);}returnnull;}StringautowiredBeanName;ObjectinstanceCandidate;//根据类型匹配到的数量大于1个if(matchingBeans.size()>1){//确定自动注入的beanNameautowiredBeanName=determineAutowireCandidate(matchingBeans,descriptor);if(autowiredBeanName==null){if(isRequired(descriptor)||!indicatesMultipleBeans(type)){returndescriptor.resolveNotUnique(descriptor.getResolvableType(),matchingBeans);}else{returnnull;}}instanceCandidate=matchingBeans.get(autowiredBeanName);}else{//Wehaveexactlyonematch.Map.Entryentry=matchingBeans.entrySet().iterator().next();autowiredBeanName=entry.getKey();instanceCandidate=entry.getValue();}if(autowiredBeanNames!=null){autowiredBeanNames.add(autowiredBeanName);}if(instanceCandidateinstanceofClass){//实例化对象instanceCandidate=descriptor.resolveCandidate(autowiredBeanName,type,this);}Objectresult=instanceCandidate;if(resultinstanceofNullBean){if(isRequired(descriptor)){raiseNoMatchingBeanFound(type,descriptor.getResolvableType(),descriptor);}result=null;}if(!ClassUtils.isAssignableValue(type,result)){thrownewBeanNotOfRequiredTypeException(autowiredBeanName,type,instanceCandidate.getClass());}returnresult;}finally{ConstructorResolver.setCurrentInjectionPoint(previousInjectionPoint);}}分析上述代码，可以发现其中主要做了如下的几件事：