這一次搞懂Spring代理建立及AOP鏈式調用過程

時間 2020-06-28

標籤這一次 spring 代理建立 aop 鏈式調用過程欄目 Spring 简体版

原文原文鏈接

@java

前言

AOP，也就是面向切面編程，它能夠將公共的代碼抽離出來，動態的織入到目標類、目標方法中，大大提升咱們編程的效率，也使程序變得更加優雅。如事務、操做日誌等均可以使用AOP實現。這種織入能夠是在運行期動態生成代理對象實現，也能夠在編譯期、類加載時期靜態織入到代碼中。而Spring正是經過第一種方法實現，且在代理類的生成上也有兩種方式：JDK Proxy和CGLIB，默認當類實現了接口時使用前者，不然使用後者；另外Spring AOP只能實現對方法的加強。spring

正文

基本概念

AOP的術語不少，雖然不清楚術語咱們也能很熟練地使用AOP，可是要理解分析源碼，術語就須要深入體會其含義。express

加強（Advice）：就是咱們想要額外增長的功能
目標對象（Target）：就是咱們想要加強的目標類，若是沒有AOP，咱們須要在每一個目標對象中實現日誌、事務管理等非業務邏輯
鏈接點（JoinPoint）：程序執行時的特定時機，如方法執行前、後以及拋出異常後等等。
切點（Pointcut）：鏈接點的導航，咱們如何找到目標對象呢？切點的做用就在於此，在Spring中就是匹配表達式。
引介（Introduction）：引介是一種特殊的加強，它爲類添加一些屬性和方法。這樣，即便一個業務類本來沒有實現某個接口，經過AOP的引介功能，咱們能夠動態地爲該業務類添加接口的實現邏輯，讓業務類成爲這個接口的實現類。
織入（Weaving）：即如何將加強添加到目標對象的鏈接點上，有動態（運行期生成代理）、靜態（編譯期、類加載時期）兩種方式。
代理（Proxy）：目標對象被織入加強後，就會產生一個代理對象，該對象多是和原對象實現了一樣的一個接口（JDK），也多是原對象的子類（CGLIB）。
切面（Aspect、Advisor）：切面由切點和加強組成，包含了這二者的定義。

代理對象的建立

在熟悉了AOP術語後，下面就來看看Spring是如何建立代理對象的，是否還記得上一篇提到的AOP的入口呢？在AbstractAutowireCapableBeanFactory類的applyBeanPostProcessorsAfterInitialization方法中循環調用了BeanPostProcessor的postProcessAfterInitialization方法，其中一個就是咱們建立代理對象的入口。這裏是Bean實例化完成去建立代理對象，理所固然應該這樣，但實際上在Bean實例化以前調用了一個resolveBeforeInstantiation方法，這裏實際上咱們也是有機會能夠提早建立代理對象的，這裏放到最後來分析，先來看主入口，進入到AbstractAutoProxyCreator類中：編程

public Object postProcessAfterInitialization(@Nullable Object bean, String beanName) {
		if (bean != null) {
			Object cacheKey = getCacheKey(bean.getClass(), beanName);
			if (!this.earlyProxyReferences.contains(cacheKey)) {
				return wrapIfNecessary(bean, beanName, cacheKey);
			}
		}
		return bean;
	}

	protected Object wrapIfNecessary(Object bean, String beanName, Object cacheKey) {
		//建立當前bean的代理，若是這個bean有advice的話，重點看
		// Create proxy if we have advice.
		Object[] specificInterceptors = getAdvicesAndAdvisorsForBean(bean.getClass(), beanName, null);
		//若是有切面，則生成該bean的代理
		if (specificInterceptors != DO_NOT_PROXY) {
			this.advisedBeans.put(cacheKey, Boolean.TRUE);
			//把被代理對象bean實例封裝到SingletonTargetSource對象中
			Object proxy = createProxy(
					bean.getClass(), beanName, specificInterceptors, new SingletonTargetSource(bean));
			this.proxyTypes.put(cacheKey, proxy.getClass());
			return proxy;
		}

		this.advisedBeans.put(cacheKey, Boolean.FALSE);
		return bean;
	}

先從緩存中拿，沒有則調用wrapIfNecessary方法建立。在這個方法裏面主要看兩個地方：getAdvicesAndAdvisorsForBean和createProxy。簡單一句話歸納就是先掃描後建立，問題是掃描什麼呢？你能夠先結合上面的概念思考下，換你會怎麼作。進入到子類AbstractAdvisorAutoProxyCreator的getAdvicesAndAdvisorsForBean方法中：數組

protected Object[] getAdvicesAndAdvisorsForBean(
			Class<?> beanClass, String beanName, @Nullable TargetSource targetSource) {

		//找到合格的切面
		List<Advisor> advisors = findEligibleAdvisors(beanClass, beanName);
		if (advisors.isEmpty()) {
			return DO_NOT_PROXY;
		}
		return advisors.toArray();
	}

	protected List<Advisor> findEligibleAdvisors(Class<?> beanClass, String beanName) {
		//找到候選的切面,其實就是一個尋找有@Aspectj註解的過程，把工程中全部有這個註解的類封裝成Advisor返回
		List<Advisor> candidateAdvisors = findCandidateAdvisors();

		//判斷候選的切面是否做用在當前beanClass上面，就是一個匹配過程。如今就是一個匹配
		List<Advisor> eligibleAdvisors = findAdvisorsThatCanApply(candidateAdvisors, beanClass, beanName);
		extendAdvisors(eligibleAdvisors);
		if (!eligibleAdvisors.isEmpty()) {
			//對有@Order@Priority進行排序
			eligibleAdvisors = sortAdvisors(eligibleAdvisors);
		}
		return eligibleAdvisors;
	}

在findEligibleAdvisors方法中能夠看到有兩個步驟，第一先找到全部的切面，即掃描全部帶有@Aspect註解的類，並將其中的切點（表達式）和加強封裝爲切面，掃描完成後，天然是要判斷哪些切面可以鏈接到當前Bean實例上。下面一步步來分析，首先是掃描過程，進入到AnnotationAwareAspectJAutoProxyCreator類中：緩存

protected List<Advisor> findCandidateAdvisors() {
		// 先經過父類AbstractAdvisorAutoProxyCreator掃描，這裏不重要
		List<Advisor> advisors = super.findCandidateAdvisors();
		// 主要看這裏
		if (this.aspectJAdvisorsBuilder != null) {
			advisors.addAll(this.aspectJAdvisorsBuilder.buildAspectJAdvisors());
		}
		return advisors;
	}

這裏委託給了BeanFactoryAspectJAdvisorsBuilderAdapter類，並調用其父類的buildAspectJAdvisors方法建立切面對象：app

public List<Advisor> buildAspectJAdvisors() {
		List<String> aspectNames = this.aspectBeanNames;

		if (aspectNames == null) {
			synchronized (this) {
				aspectNames = this.aspectBeanNames;
				if (aspectNames == null) {
					List<Advisor> advisors = new ArrayList<>();
					aspectNames = new ArrayList<>();
					//獲取spring容器中的全部bean的名稱BeanName
					String[] beanNames = BeanFactoryUtils.beanNamesForTypeIncludingAncestors(
							this.beanFactory, Object.class, true, false);
					for (String beanName : beanNames) {
						if (!isEligibleBean(beanName)) {
							continue;
						}
						Class<?> beanType = this.beanFactory.getType(beanName);
						if (beanType == null) {
							continue;
						}
						//判斷類上是否有@Aspect註解
						if (this.advisorFactory.isAspect(beanType)) {
							aspectNames.add(beanName);
							AspectMetadata amd = new AspectMetadata(beanType, beanName);
							if (amd.getAjType().getPerClause().getKind() == PerClauseKind.SINGLETON) {
								// 當@Aspect的value屬性爲""時纔會進入到這裏
								// 建立獲取有@Aspect註解類的實例工廠，負責獲取有@Aspect註解類的實例
								MetadataAwareAspectInstanceFactory factory =
										new BeanFactoryAspectInstanceFactory(this.beanFactory, beanName);

								//建立切面advisor對象
								List<Advisor> classAdvisors = this.advisorFactory.getAdvisors(factory);
								if (this.beanFactory.isSingleton(beanName)) {
									this.advisorsCache.put(beanName, classAdvisors);
								}
								else {
									this.aspectFactoryCache.put(beanName, factory);
								}
								advisors.addAll(classAdvisors);
							}
							else {
								MetadataAwareAspectInstanceFactory factory =
										new PrototypeAspectInstanceFactory(this.beanFactory, beanName);
								this.aspectFactoryCache.put(beanName, factory);
								advisors.addAll(this.advisorFactory.getAdvisors(factory));
							}
						}
					}
					this.aspectBeanNames = aspectNames;
					return advisors;
				}
			}
		}
		return advisors;
	}

這個方法裏面首先從IOC中拿到全部Bean的名稱，並循環判斷該類上是否帶有@Aspect註解，若是有則將BeanName和Bean的Class類型封裝到BeanFactoryAspectInstanceFactory中，並調用ReflectiveAspectJAdvisorFactory.getAdvisors建立切面對象：ide

public List<Advisor> getAdvisors(MetadataAwareAspectInstanceFactory aspectInstanceFactory) {
		//從工廠中獲取有@Aspect註解的類Class
		Class<?> aspectClass = aspectInstanceFactory.getAspectMetadata().getAspectClass();
		//從工廠中獲取有@Aspect註解的類的名稱
		String aspectName = aspectInstanceFactory.getAspectMetadata().getAspectName();
		validate(aspectClass);

		// 建立工廠的裝飾類，獲取實例只會獲取一次
		MetadataAwareAspectInstanceFactory lazySingletonAspectInstanceFactory =
				new LazySingletonAspectInstanceFactoryDecorator(aspectInstanceFactory);

		List<Advisor> advisors = new ArrayList<>();

		//這裏循環沒有@Pointcut註解的方法
		for (Method method : getAdvisorMethods(aspectClass)) {

			//很是重要重點看看
			Advisor advisor = getAdvisor(method, lazySingletonAspectInstanceFactory, advisors.size(), aspectName);
			if (advisor != null) {
				advisors.add(advisor);
			}
		}

		if (!advisors.isEmpty() && lazySingletonAspectInstanceFactory.getAspectMetadata().isLazilyInstantiated()) {
			Advisor instantiationAdvisor = new SyntheticInstantiationAdvisor(lazySingletonAspectInstanceFactory);
			advisors.add(0, instantiationAdvisor);
		}

		//判斷屬性上是否有引介註解，這裏能夠不看
		for (Field field : aspectClass.getDeclaredFields()) {
			//判斷屬性上是否有DeclareParents註解，若是有返回切面
			Advisor advisor = getDeclareParentsAdvisor(field);
			if (advisor != null) {
				advisors.add(advisor);
			}
		}

		return advisors;
	}

	private List<Method> getAdvisorMethods(Class<?> aspectClass) {
		final List<Method> methods = new ArrayList<>();
		ReflectionUtils.doWithMethods(aspectClass, method -> {
			// Exclude pointcuts
			if (AnnotationUtils.getAnnotation(method, Pointcut.class) == null) {
				methods.add(method);
			}
		});
		methods.sort(METHOD_COMPARATOR);
		return methods;
	}

根據Aspect的Class拿到全部不帶@Pointcut註解的方法對象（爲何是不帶@Pointcut註解的方法？仔細想一想不難理解），另外要注意這裏對method進行了排序，看看這個METHOD_COMPARATOR比較器：函數

private static final Comparator<Method> METHOD_COMPARATOR;

	static {
		Comparator<Method> adviceKindComparator = new ConvertingComparator<>(
				new InstanceComparator<>(
						Around.class, Before.class, After.class, AfterReturning.class, AfterThrowing.class),
				(Converter<Method, Annotation>) method -> {
					AspectJAnnotation<?> annotation =
						AbstractAspectJAdvisorFactory.findAspectJAnnotationOnMethod(method);
					return (annotation != null ? annotation.getAnnotation() : null);
				});
		Comparator<Method> methodNameComparator = new ConvertingComparator<>(Method::getName);
		METHOD_COMPARATOR = adviceKindComparator.thenComparing(methodNameComparator);
	}

關注InstanceComparator構造函數參數，記住它們的順序，這就是AOP鏈式調用中同一個@Aspect類中Advice的執行順序。接着往下看，在getAdvisors方法中循環獲取到的methods，分別調用getAdvisor方法，也就是根據方法逐個去建立切面：post

public Advisor getAdvisor(Method candidateAdviceMethod, MetadataAwareAspectInstanceFactory aspectInstanceFactory,
			int declarationOrderInAspect, String aspectName) {

		validate(aspectInstanceFactory.getAspectMetadata().getAspectClass());

		//獲取pointCut對象，最重要的是從註解中獲取表達式
		AspectJExpressionPointcut expressionPointcut = getPointcut(
				candidateAdviceMethod, aspectInstanceFactory.getAspectMetadata().getAspectClass());
		if (expressionPointcut == null) {
			return null;
		}

		//建立Advisor切面類，這纔是真正的切面類，一個切面類裏面確定要有一、pointCut 二、advice
		//這裏pointCut是expressionPointcut， advice 加強方法是 candidateAdviceMethod
		return new InstantiationModelAwarePointcutAdvisorImpl(expressionPointcut, candidateAdviceMethod,
				this, aspectInstanceFactory, declarationOrderInAspect, aspectName);
	}

	private static final Class<?>[] ASPECTJ_ANNOTATION_CLASSES = new Class<?>[] {
			Pointcut.class, Around.class, Before.class, After.class, AfterReturning.class, AfterThrowing.class};
			
	private AspectJExpressionPointcut getPointcut(Method candidateAdviceMethod, Class<?> candidateAspectClass) {
		//從候選的加強方法裏面 candidateAdviceMethod  找有有註解
		//Pointcut.class, Around.class, Before.class, After.class, AfterReturning.class, AfterThrowing.class
		//並把註解信息封裝成AspectJAnnotation對象
		AspectJAnnotation<?> aspectJAnnotation =
				AbstractAspectJAdvisorFactory.findAspectJAnnotationOnMethod(candidateAdviceMethod);
		if (aspectJAnnotation == null) {
			return null;
		}

		//建立一個PointCut類，而且把前面從註解裏面解析的表達式設置進去
		AspectJExpressionPointcut ajexp =
				new AspectJExpressionPointcut(candidateAspectClass, new String[0], new Class<?>[0]);
		ajexp.setExpression(aspectJAnnotation.getPointcutExpression());
		if (this.beanFactory != null) {
			ajexp.setBeanFactory(this.beanFactory);
		}
		return ajexp;
	}

以前就說過切面的定義，是切點和加強的組合，因此這裏首先經過getPointcut獲取到註解對象，而後new了一個Pointcut對象，並將表達式設置進去。而後在getAdvisor方法中最後new了一個InstantiationModelAwarePointcutAdvisorImpl對象：

public InstantiationModelAwarePointcutAdvisorImpl(AspectJExpressionPointcut declaredPointcut,
			Method aspectJAdviceMethod, AspectJAdvisorFactory aspectJAdvisorFactory,
			MetadataAwareAspectInstanceFactory aspectInstanceFactory, int declarationOrder, String aspectName) {

		this.declaredPointcut = declaredPointcut;
		this.declaringClass = aspectJAdviceMethod.getDeclaringClass();
		this.methodName = aspectJAdviceMethod.getName();
		this.parameterTypes = aspectJAdviceMethod.getParameterTypes();
		this.aspectJAdviceMethod = aspectJAdviceMethod;
		this.aspectJAdvisorFactory = aspectJAdvisorFactory;
		this.aspectInstanceFactory = aspectInstanceFactory;
		this.declarationOrder = declarationOrder;
		this.aspectName = aspectName;

		if (aspectInstanceFactory.getAspectMetadata().isLazilyInstantiated()) {
			// Static part of the pointcut is a lazy type.
			Pointcut preInstantiationPointcut = Pointcuts.union(
					aspectInstanceFactory.getAspectMetadata().getPerClausePointcut(), this.declaredPointcut);

			// Make it dynamic: must mutate from pre-instantiation to post-instantiation state.
			// If it's not a dynamic pointcut, it may be optimized out
			// by the Spring AOP infrastructure after the first evaluation.
			this.pointcut = new PerTargetInstantiationModelPointcut(
					this.declaredPointcut, preInstantiationPointcut, aspectInstanceFactory);
			this.lazy = true;
		}
		else {
			// A singleton aspect.
			this.pointcut = this.declaredPointcut;
			this.lazy = false;
			//這個方法重點看看，建立advice對象
			this.instantiatedAdvice = instantiateAdvice(this.declaredPointcut);
		}
	}

這個就是咱們的切面類，在其構造方法的最後經過instantiateAdvice建立了Advice對象。注意這裏傳進來的declarationOrder參數，它就是循環method時的序號，其做用就是賦值給這裏的declarationOrder屬性以及Advice的declarationOrder屬性，在後面排序時就會經過這個序號來比較，所以Advice的執行順序是固定的，至於爲何要固定，後面分析完AOP鏈式調用過程天然就明白了。

public Advice getAdvice(Method candidateAdviceMethod, AspectJExpressionPointcut expressionPointcut,
			MetadataAwareAspectInstanceFactory aspectInstanceFactory, int declarationOrder, String aspectName) {

		//獲取有@Aspect註解的類
		Class<?> candidateAspectClass = aspectInstanceFactory.getAspectMetadata().getAspectClass();
		validate(candidateAspectClass);

		//找到candidateAdviceMethod方法上面的註解，而且包裝成AspectJAnnotation對象，這個對象中就有註解類型
		AspectJAnnotation<?> aspectJAnnotation =
				AbstractAspectJAdvisorFactory.findAspectJAnnotationOnMethod(candidateAdviceMethod);
		if (aspectJAnnotation == null) {
			return null;
		}
		
		AbstractAspectJAdvice springAdvice;

		//根據不一樣的註解類型建立不一樣的advice類實例
		switch (aspectJAnnotation.getAnnotationType()) {
			case AtPointcut:
				if (logger.isDebugEnabled()) {
					logger.debug("Processing pointcut '" + candidateAdviceMethod.getName() + "'");
				}
				return null;
			case AtAround:
				//實現了MethodInterceptor接口
				springAdvice = new AspectJAroundAdvice(
						candidateAdviceMethod, expressionPointcut, aspectInstanceFactory);
				break;
			case AtBefore:
				//實現了MethodBeforeAdvice接口，沒有實現MethodInterceptor接口
				springAdvice = new AspectJMethodBeforeAdvice(
						candidateAdviceMethod, expressionPointcut, aspectInstanceFactory);
				break;
			case AtAfter:
				//實現了MethodInterceptor接口
				springAdvice = new AspectJAfterAdvice(
						candidateAdviceMethod, expressionPointcut, aspectInstanceFactory);
				break;
			case AtAfterReturning:
				//實現了AfterReturningAdvice接口，沒有實現MethodInterceptor接口
				springAdvice = new AspectJAfterReturningAdvice(
						candidateAdviceMethod, expressionPointcut, aspectInstanceFactory);
				AfterReturning afterReturningAnnotation = (AfterReturning) aspectJAnnotation.getAnnotation();
				if (StringUtils.hasText(afterReturningAnnotation.returning())) {
					springAdvice.setReturningName(afterReturningAnnotation.returning());
				}
				break;
			case AtAfterThrowing:
				//實現了MethodInterceptor接口
				springAdvice = new AspectJAfterThrowingAdvice(
						candidateAdviceMethod, expressionPointcut, aspectInstanceFactory);
				AfterThrowing afterThrowingAnnotation = (AfterThrowing) aspectJAnnotation.getAnnotation();
				if (StringUtils.hasText(afterThrowingAnnotation.throwing())) {
					springAdvice.setThrowingName(afterThrowingAnnotation.throwing());
				}
				break;
			default:
				throw new UnsupportedOperationException(
						"Unsupported advice type on method: " + candidateAdviceMethod);
		}

		// Now to configure the advice...
		springAdvice.setAspectName(aspectName);
		springAdvice.setDeclarationOrder(declarationOrder);
		String[] argNames = this.parameterNameDiscoverer.getParameterNames(candidateAdviceMethod);
		if (argNames != null) {
			springAdvice.setArgumentNamesFromStringArray(argNames);
		}

		//計算argNames和類型的對應關係
		springAdvice.calculateArgumentBindings();

		return springAdvice;
	}

這裏邏輯很清晰，就是拿到方法上的註解類型，根據類型建立不一樣的加強Advice對象：AspectJAroundAdvice、AspectJMethodBeforeAdvice、AspectJAfterAdvice、AspectJAfterReturningAdvice、AspectJAfterThrowingAdvice。完成以後經過calculateArgumentBindings方法進行參數綁定，感興趣的可自行研究。這裏主要看看幾個Advice的繼承體系：

能夠看到有兩個Advice是沒有實現MethodInterceptor接口的：AspectJMethodBeforeAdvice和AspectJAfterReturningAdvice。而MethodInterceptor有一個invoke方法，這個方法就是鏈式調用的核心方法，但那兩個沒有實現該方法的Advice怎麼處理呢？稍後會分析。
到這裏切面對象就建立完成了，接下來就是判斷當前建立的Bean實例是否和這些切面匹配以及對切面排序。匹配過程比較複雜，對理解主流程也沒什麼幫助，因此這裏就不展開分析，感興趣的自行分析（AbstractAdvisorAutoProxyCreator.findAdvisorsThatCanApply()）。下面看看排序的過程，回到AbstractAdvisorAutoProxyCreator.findEligibleAdvisors方法：

protected List<Advisor> findEligibleAdvisors(Class<?> beanClass, String beanName) {
		//找到候選的切面,其實就是一個尋找有@Aspectj註解的過程，把工程中全部有這個註解的類封裝成Advisor返回
		List<Advisor> candidateAdvisors = findCandidateAdvisors();

		//判斷候選的切面是否做用在當前beanClass上面，就是一個匹配過程。。如今就是一個匹配
		List<Advisor> eligibleAdvisors = findAdvisorsThatCanApply(candidateAdvisors, beanClass, beanName);
		extendAdvisors(eligibleAdvisors);
		if (!eligibleAdvisors.isEmpty()) {
			//對有@Order@Priority進行排序
			eligibleAdvisors = sortAdvisors(eligibleAdvisors);
		}
		return eligibleAdvisors;
	}

sortAdvisors方法就是排序，但這個方法有兩個實現：當前類AbstractAdvisorAutoProxyCreator和子類AspectJAwareAdvisorAutoProxyCreator，應該走哪一個呢？

經過類圖咱們能夠確定是進入的AspectJAwareAdvisorAutoProxyCreator類，由於AnnotationAwareAspectJAutoProxyCreator的父類是它。

protected List<Advisor> sortAdvisors(List<Advisor> advisors) {
		List<PartiallyComparableAdvisorHolder> partiallyComparableAdvisors = new ArrayList<>(advisors.size());
		for (Advisor element : advisors) {
			partiallyComparableAdvisors.add(
					new PartiallyComparableAdvisorHolder(element, DEFAULT_PRECEDENCE_COMPARATOR));
		}
		List<PartiallyComparableAdvisorHolder> sorted = PartialOrder.sort(partiallyComparableAdvisors);
		if (sorted != null) {
			List<Advisor> result = new ArrayList<>(advisors.size());
			for (PartiallyComparableAdvisorHolder pcAdvisor : sorted) {
				result.add(pcAdvisor.getAdvisor());
			}
			return result;
		}
		else {
			return super.sortAdvisors(advisors);
		}
	}

這裏排序主要是委託給PartialOrder進行的，而在此以前將全部的切面都封裝成了PartiallyComparableAdvisorHolder對象，注意傳入的DEFAULT_PRECEDENCE_COMPARATOR參數，這個就是比較器對象：

private static final Comparator<Advisor> DEFAULT_PRECEDENCE_COMPARATOR = new AspectJPrecedenceComparator();

因此咱們直接看這個比較器的compare方法：

public int compare(Advisor o1, Advisor o2) {
		int advisorPrecedence = this.advisorComparator.compare(o1, o2);
		if (advisorPrecedence == SAME_PRECEDENCE && declaredInSameAspect(o1, o2)) {
			advisorPrecedence = comparePrecedenceWithinAspect(o1, o2);
		}
		return advisorPrecedence;
	}

	private final Comparator<? super Advisor> advisorComparator;
	public AspectJPrecedenceComparator() {
		this.advisorComparator = AnnotationAwareOrderComparator.INSTANCE;
	}

第一步先經過AnnotationAwareOrderComparator去比較，點進去看能夠發現是對實現了PriorityOrdered和Ordered接口以及標記了Priority和Order註解的非同一個@Aspect類中的切面進行排序。這個和以前分析BeanFacotryPostProcessor類是同樣的原理。而對同一個@Aspect類中的切面排序主要是comparePrecedenceWithinAspect方法：

private int comparePrecedenceWithinAspect(Advisor advisor1, Advisor advisor2) {
		boolean oneOrOtherIsAfterAdvice =
				(AspectJAopUtils.isAfterAdvice(advisor1) || AspectJAopUtils.isAfterAdvice(advisor2));
		int adviceDeclarationOrderDelta = getAspectDeclarationOrder(advisor1) - getAspectDeclarationOrder(advisor2);

		if (oneOrOtherIsAfterAdvice) {
			// the advice declared last has higher precedence
			if (adviceDeclarationOrderDelta < 0) {
				// advice1 was declared before advice2
				// so advice1 has lower precedence
				return LOWER_PRECEDENCE;
			}
			else if (adviceDeclarationOrderDelta == 0) {
				return SAME_PRECEDENCE;
			}
			else {
				return HIGHER_PRECEDENCE;
			}
		}
		else {
			// the advice declared first has higher precedence
			if (adviceDeclarationOrderDelta < 0) {
				// advice1 was declared before advice2
				// so advice1 has higher precedence
				return HIGHER_PRECEDENCE;
			}
			else if (adviceDeclarationOrderDelta == 0) {
				return SAME_PRECEDENCE;
			}
			else {
				return LOWER_PRECEDENCE;
			}
		}
	}

	private int getAspectDeclarationOrder(Advisor anAdvisor) {
		AspectJPrecedenceInformation precedenceInfo =
			AspectJAopUtils.getAspectJPrecedenceInformationFor(anAdvisor);
		if (precedenceInfo != null) {
			return precedenceInfo.getDeclarationOrder();
		}
		else {
			return 0;
		}
	}

這裏就是經過precedenceInfo.getDeclarationOrder拿到在建立InstantiationModelAwarePointcutAdvisorImpl對象時設置的declarationOrder屬性，這就驗證了以前的說法（實際上這裏排序過程很是複雜，不是簡單的按照這個屬性進行排序）。
當上面的一切都進行完成後，就該建立代理對象了，回到AbstractAutoProxyCreator.wrapIfNecessary，看關鍵部分代碼：

//若是有切面，則生成該bean的代理
	if (specificInterceptors != DO_NOT_PROXY) {
		this.advisedBeans.put(cacheKey, Boolean.TRUE);
		//把被代理對象bean實例封裝到SingletonTargetSource對象中
		Object proxy = createProxy(
				bean.getClass(), beanName, specificInterceptors, new SingletonTargetSource(bean));
		this.proxyTypes.put(cacheKey, proxy.getClass());
		return proxy;
	}

注意這裏將被代理對象封裝成了一個SingletonTargetSource對象，它是TargetSource的實現類。

protected Object createProxy(Class<?> beanClass, @Nullable String beanName,
			@Nullable Object[] specificInterceptors, TargetSource targetSource) {

		if (this.beanFactory instanceof ConfigurableListableBeanFactory) {
			AutoProxyUtils.exposeTargetClass((ConfigurableListableBeanFactory) this.beanFactory, beanName, beanClass);
		}

		//建立代理工廠
		ProxyFactory proxyFactory = new ProxyFactory();
		proxyFactory.copyFrom(this);

		if (!proxyFactory.isProxyTargetClass()) {
			if (shouldProxyTargetClass(beanClass, beanName)) {
				//proxyTargetClass 是否對類進行代理，而不是對接口進行代理，設置爲true時，使用CGLib代理。
				proxyFactory.setProxyTargetClass(true);
			}
			else {
				evaluateProxyInterfaces(beanClass, proxyFactory);
			}
		}

		//把advice類型的加強包裝成advisor切面
		Advisor[] advisors = buildAdvisors(beanName, specificInterceptors);
		proxyFactory.addAdvisors(advisors);
		proxyFactory.setTargetSource(targetSource);
		customizeProxyFactory(proxyFactory);

		////用來控制代理工廠被配置後，是否還容許修改代理的配置,默認爲false
		proxyFactory.setFrozen(this.freezeProxy);
		if (advisorsPreFiltered()) {
			proxyFactory.setPreFiltered(true);
		}

		//獲取代理實例
		return proxyFactory.getProxy(getProxyClassLoader());
	}

這裏經過ProxyFactory對象去建立代理實例，這是工廠模式的體現，但在建立代理對象以前還有幾個準備動做：須要判斷是JDK代理仍是CGLIB代理以及經過buildAdvisors方法將擴展的Advice封裝成Advisor切面。準備完成則經過getProxy建立代理對象：

public Object getProxy(@Nullable ClassLoader classLoader) {
		//根據目標對象是否有接口來判斷採用什麼代理方式，cglib代理仍是jdk動態代理
		return createAopProxy().getProxy(classLoader);
	}

	protected final synchronized AopProxy createAopProxy() {
		if (!this.active) {
			activate();
		}
		return getAopProxyFactory().createAopProxy(this);
	}

	public AopProxy createAopProxy(AdvisedSupport config) throws AopConfigException {
		if (config.isOptimize() || config.isProxyTargetClass() || hasNoUserSuppliedProxyInterfaces(config)) {
			Class<?> targetClass = config.getTargetClass();
			if (targetClass == null) {
				throw new AopConfigException("TargetSource cannot determine target class: " +
						"Either an interface or a target is required for proxy creation.");
			}
			if (targetClass.isInterface() || Proxy.isProxyClass(targetClass)) {
				return new JdkDynamicAopProxy(config);
			}
			return new ObjenesisCglibAopProxy(config);
		}
		else {
			return new JdkDynamicAopProxy(config);
		}
	}

首先經過配置拿到對應的代理類：ObjenesisCglibAopProxy和JdkDynamicAopProxy，而後再經過getProxy建立Bean的代理，這裏以JdkDynamicAopProxy爲例：

public Object getProxy(@Nullable ClassLoader classLoader) {
		//advised是代理工廠對象
		Class<?>[] proxiedInterfaces = AopProxyUtils.completeProxiedInterfaces(this.advised, true);
		findDefinedEqualsAndHashCodeMethods(proxiedInterfaces);
		return Proxy.newProxyInstance(classLoader, proxiedInterfaces, this);
	}

這裏的代碼你應該不陌生了，就是JDK的原生API，newProxyInstance方法傳入的InvocationHandler對象是this，所以，最終AOP代理的調用就是從該類中的invoke方法開始。至此，代理對象的建立就完成了，下面來看下整個過程的時序圖：

小結

代理對象的建立過程總體來講並不複雜，首先找到全部帶有@Aspect註解的類，並獲取其中沒有@Pointcut註解的方法，循環建立切面，而建立切面須要切點和加強兩個元素，其中切點可簡單理解爲咱們寫的表達式，加強則是根據@Before、@Around、@After等註解建立的對應的Advice類。切面建立好後則須要循環判斷哪些切面能對當前的Bean實例的方法進行加強並排序，最後經過ProxyFactory建立代理對象。

AOP鏈式調用

熟悉JDK動態代理的都知道經過代理對象調用方法時，會進入到InvocationHandler對象的invoke方法，因此咱們直接從JdkDynamicAopProxy的這個方法開始：

public Object invoke(Object proxy, Method method, Object[] args) throws Throwable {
		MethodInvocation invocation;
		Object oldProxy = null;
		boolean setProxyContext = false;

		//從代理工廠中拿到TargetSource對象，該對象包裝了被代理實例bean
		TargetSource targetSource = this.advised.targetSource;
		Object target = null;

		try {
			//被代理對象的equals方法和hashCode方法是不能被代理的，不會走切面
			.......
			
			Object retVal;

			// 能夠從當前線程中拿到代理對象
			if (this.advised.exposeProxy) {
				// Make invocation available if necessary.
				oldProxy = AopContext.setCurrentProxy(proxy);
				setProxyContext = true;
			}

			//這個target就是被代理實例
			target = targetSource.getTarget();
			Class<?> targetClass = (target != null ? target.getClass() : null);
			
			//從代理工廠中拿過濾器鏈 Object是一個MethodInterceptor類型的對象，其實就是一個advice對象
			List<Object> chain = this.advised.getInterceptorsAndDynamicInterceptionAdvice(method, targetClass);

			//若是該方法沒有執行鏈，則說明這個方法不須要被攔截，則直接反射調用
			if (chain.isEmpty()) {
				Object[] argsToUse = AopProxyUtils.adaptArgumentsIfNecessary(method, args);
				retVal = AopUtils.invokeJoinpointUsingReflection(target, method, argsToUse);
			}
			else {
				invocation = new ReflectiveMethodInvocation(proxy, target, method, args, targetClass, chain);
				retVal = invocation.proceed();
			}

			// Massage return value if necessary.
			Class<?> returnType = method.getReturnType();
			if (retVal != null && retVal == target &&
					returnType != Object.class && returnType.isInstance(proxy) &&
					!RawTargetAccess.class.isAssignableFrom(method.getDeclaringClass())) {
				retVal = proxy;
			}
			return retVal;
		}
		finally {
			if (target != null && !targetSource.isStatic()) {
				// Must have come from TargetSource.
				targetSource.releaseTarget(target);
			}
			if (setProxyContext) {
				// Restore old proxy.
				AopContext.setCurrentProxy(oldProxy);
			}
		}
	}

這段代碼比較長，我刪掉了不關鍵的地方。首先來看this.advised.exposeProxy這個屬性，這在@EnableAspectJAutoProxy註解中能夠配置，當爲true時，會將該代理對象設置到當前線程的ThreadLocal對象中，這樣就能夠經過AopContext.currentProxy拿到代理對象。這個有什麼用呢？我相信有經驗的Java開發都遇到過這樣一個BUG，在Service實現類中調用本類中的另外一個方法時，事務不會生效，這是由於直接經過this調用就不會調用到代理對象的方法，而是原對象的，因此事務切面就沒有生效。所以這種狀況下就能夠從當前線程的ThreadLocal對象拿到代理對象，不過實際上直接使用@Autowired注入本身自己也能夠拿到代理對象。
接下來就是經過getInterceptorsAndDynamicInterceptionAdvice拿到執行鏈，看看具體作了哪些事情：

public List<Object> getInterceptorsAndDynamicInterceptionAdvice(
			Advised config, Method method, @Nullable Class<?> targetClass) {

		AdvisorAdapterRegistry registry = GlobalAdvisorAdapterRegistry.getInstance();
		//從代理工廠中得到該被代理類的全部切面advisor，config就是代理工廠對象
		Advisor[] advisors = config.getAdvisors();
		List<Object> interceptorList = new ArrayList<>(advisors.length);
		Class<?> actualClass = (targetClass != null ? targetClass : method.getDeclaringClass());
		Boolean hasIntroductions = null;

		for (Advisor advisor : advisors) {
			//大部分走這裏
			if (advisor instanceof PointcutAdvisor) {
				// Add it conditionally.
				PointcutAdvisor pointcutAdvisor = (PointcutAdvisor) advisor;
				//若是切面的pointCut和被代理對象是匹配的，說明是切面要攔截的對象
				if (config.isPreFiltered() || pointcutAdvisor.getPointcut().getClassFilter().matches(actualClass)) {
					MethodMatcher mm = pointcutAdvisor.getPointcut().getMethodMatcher();
					boolean match;
					if (mm instanceof IntroductionAwareMethodMatcher) {
						if (hasIntroductions == null) {
							hasIntroductions = hasMatchingIntroductions(advisors, actualClass);
						}
						match = ((IntroductionAwareMethodMatcher) mm).matches(method, actualClass, hasIntroductions);
					}
					else {
						//接下來判斷方法是不是切面pointcut須要攔截的方法
						match = mm.matches(method, actualClass);
					}
					//若是類和方法都匹配
					if (match) {

						//獲取到切面advisor中的advice，而且包裝成MethodInterceptor類型的對象
						MethodInterceptor[] interceptors = registry.getInterceptors(advisor);
						if (mm.isRuntime()) {
							for (MethodInterceptor interceptor : interceptors) {
								interceptorList.add(new InterceptorAndDynamicMethodMatcher(interceptor, mm));
							}
						}
						else {
							interceptorList.addAll(Arrays.asList(interceptors));
						}
					}
				}
			}
			//若是是引介切面
			else if (advisor instanceof IntroductionAdvisor) {
				IntroductionAdvisor ia = (IntroductionAdvisor) advisor;
				if (config.isPreFiltered() || ia.getClassFilter().matches(actualClass)) {
					Interceptor[] interceptors = registry.getInterceptors(advisor);
					interceptorList.addAll(Arrays.asList(interceptors));
				}
			}
			else {
				Interceptor[] interceptors = registry.getInterceptors(advisor);
				interceptorList.addAll(Arrays.asList(interceptors));
			}
		}

		return interceptorList;
	}

這也是個長方法，看關鍵的部分，由於以前咱們建立的基本上都是InstantiationModelAwarePointcutAdvisorImpl對象，該類是PointcutAdvisor的實現類，因此會進入第一個if判斷裏，這裏首先進行匹配，看切點和當前對象以及該對象的哪些方法匹配，若是能匹配上，則調用getInterceptors獲取執行鏈：

private final List<AdvisorAdapter> adapters = new ArrayList<>(3);
	public DefaultAdvisorAdapterRegistry() {
		registerAdvisorAdapter(new MethodBeforeAdviceAdapter());
		registerAdvisorAdapter(new AfterReturningAdviceAdapter());
		registerAdvisorAdapter(new ThrowsAdviceAdapter());
	}

	public MethodInterceptor[] getInterceptors(Advisor advisor) throws UnknownAdviceTypeException {
		List<MethodInterceptor> interceptors = new ArrayList<>(3);
		Advice advice = advisor.getAdvice();
		//若是是MethodInterceptor類型的，如：AspectJAroundAdvice
		//AspectJAfterAdvice
		//AspectJAfterThrowingAdvice
		if (advice instanceof MethodInterceptor) {
			interceptors.add((MethodInterceptor) advice);
		}

		//處理 AspectJMethodBeforeAdvice  AspectJAfterReturningAdvice
		for (AdvisorAdapter adapter : this.adapters) {
			if (adapter.supportsAdvice(advice)) {
				interceptors.add(adapter.getInterceptor(advisor));
			}
		}
		if (interceptors.isEmpty()) {
			throw new UnknownAdviceTypeException(advisor.getAdvice());
		}
		return interceptors.toArray(new MethodInterceptor[0]);
	}

這裏咱們能夠看到若是是MethodInterceptor的實現類，則直接添加到鏈中，若是不是，則須要經過適配器去包裝後添加，恰好這裏有MethodBeforeAdviceAdapter和AfterReturningAdviceAdapter兩個適配器對應上文兩個沒有實現MethodInterceptor接口的類。最後將Interceptors返回。

if (chain.isEmpty()) {
	Object[] argsToUse = AopProxyUtils.adaptArgumentsIfNecessary(method, args);
	retVal = AopUtils.invokeJoinpointUsingReflection(target, method, argsToUse);
}
else {
	// We need to create a method invocation...
	invocation = new ReflectiveMethodInvocation(proxy, target, method, args, targetClass, chain);
	// Proceed to the joinpoint through the interceptor chain.
	retVal = invocation.proceed();
}

返回到invoke方法後，若是執行鏈爲空，說明該方法不須要被加強，因此直接反射調用原對象的方法（注意傳入的是TargetSource封裝的被代理對象）；反之，則經過ReflectiveMethodInvocation類進行鏈式調用，關鍵方法就是proceed：

private int currentInterceptorIndex = -1;
	
	public Object proceed() throws Throwable {
		//若是執行鏈中的advice所有執行完，則直接調用joinPoint方法，就是被代理方法
		if (this.currentInterceptorIndex == this.interceptorsAndDynamicMethodMatchers.size() - 1) {
			return invokeJoinpoint();
		}

		Object interceptorOrInterceptionAdvice =
				this.interceptorsAndDynamicMethodMatchers.get(++this.currentInterceptorIndex);
		if (interceptorOrInterceptionAdvice instanceof InterceptorAndDynamicMethodMatcher) {
			InterceptorAndDynamicMethodMatcher dm =
					(InterceptorAndDynamicMethodMatcher) interceptorOrInterceptionAdvice;
			Class<?> targetClass = (this.targetClass != null ? this.targetClass : this.method.getDeclaringClass());
			if (dm.methodMatcher.matches(this.method, targetClass, this.arguments)) {
				return dm.interceptor.invoke(this);
			}
			else {
				return proceed();
			}
		}
		else {
			//調用MethodInterceptor中的invoke方法
			return ((MethodInterceptor) interceptorOrInterceptionAdvice).invoke(this);
		}
	}

這個方法的核心就在兩個地方：invokeJoinpoint和interceptorOrInterceptionAdvice.invoke(this)。當加強方法調用完後就會經過前者調用到被代理的方法，不然則是依次調用Interceptor的invoke方法。下面就分別看看每一個Interceptor是怎麼實現的。

AspectJAroundAdvice

public Object invoke(MethodInvocation mi) throws Throwable {
		if (!(mi instanceof ProxyMethodInvocation)) {
			throw new IllegalStateException("MethodInvocation is not a Spring ProxyMethodInvocation: " + mi);
		}
		ProxyMethodInvocation pmi = (ProxyMethodInvocation) mi;
		ProceedingJoinPoint pjp = lazyGetProceedingJoinPoint(pmi);
		JoinPointMatch jpm = getJoinPointMatch(pmi);
		return invokeAdviceMethod(pjp, jpm, null, null);
	}

MethodBeforeAdviceInterceptor -> AspectJMethodBeforeAdvice

public Object invoke(MethodInvocation mi) throws Throwable {
		this.advice.before(mi.getMethod(), mi.getArguments(), mi.getThis());
		return mi.proceed();
	}

	public void before(Method method, Object[] args, @Nullable Object target) throws Throwable {
		invokeAdviceMethod(getJoinPointMatch(), null, null);
	}

AspectJAfterAdvice

public Object invoke(MethodInvocation mi) throws Throwable {
		try {
			return mi.proceed();
		}
		finally {
			invokeAdviceMethod(getJoinPointMatch(), null, null);
		}
	}

AfterReturningAdviceInterceptor -> AspectJAfterReturningAdvice

public Object invoke(MethodInvocation mi) throws Throwable {
		Object retVal = mi.proceed();
		this.advice.afterReturning(retVal, mi.getMethod(), mi.getArguments(), mi.getThis());
		return retVal;
	}

	public void afterReturning(@Nullable Object returnValue, Method method, Object[] args, @Nullable Object target) throws Throwable {
		if (shouldInvokeOnReturnValueOf(method, returnValue)) {
			invokeAdviceMethod(getJoinPointMatch(), returnValue, null);
		}
	}

AspectJAfterThrowingAdvice

public Object invoke(MethodInvocation mi) throws Throwable {
		try {
			return mi.proceed();
		}
		catch (Throwable ex) {
			if (shouldInvokeOnThrowing(ex)) {
				invokeAdviceMethod(getJoinPointMatch(), null, ex);
			}
			throw ex;
		}
	}

這裏的調用順序是怎樣的呢？其核心就是經過proceed方法控制流程，每執行完一個Advice就會回到proceed方法中調用下一個Advice。能夠思考一下，怎麼才能讓調用結果知足以下圖的執行順序。

以上就是AOP的鏈式調用過程，可是這只是只有一個切面類的狀況，若是有多個@Aspect類呢，這個調用過程又是怎樣的？其核心思想和「棧」同樣，就是「先進後出，後進先出」。

AOP擴展知識

1、自定義全局攔截器Interceptor

在上文建立代理對象的時候有這樣一個方法：

protected Advisor[] buildAdvisors(@Nullable String beanName, @Nullable Object[] specificInterceptors) {
		//自定義MethodInterceptor.拿到setInterceptorNames方法注入的Interceptor對象
		Advisor[] commonInterceptors = resolveInterceptorNames();

		List<Object> allInterceptors = new ArrayList<>();
		if (specificInterceptors != null) {
			allInterceptors.addAll(Arrays.asList(specificInterceptors));
			if (commonInterceptors.length > 0) {
				if (this.applyCommonInterceptorsFirst) {
					allInterceptors.addAll(0, Arrays.asList(commonInterceptors));
				}
				else {
					allInterceptors.addAll(Arrays.asList(commonInterceptors));
				}
			}
		}

		Advisor[] advisors = new Advisor[allInterceptors.size()];
		for (int i = 0; i < allInterceptors.size(); i++) {
			//對自定義的advice要進行包裝，把advice包裝成advisor對象，切面對象
			advisors[i] = this.advisorAdapterRegistry.wrap(allInterceptors.get(i));
		}
		return advisors;
	}

這個方法的做用就在於咱們能夠擴展咱們本身的Interceptor，首先經過resolveInterceptorNames方法獲取到經過setInterceptorNames方法設置的Interceptor，而後調用DefaultAdvisorAdapterRegistry.wrap方法將其包裝爲DefaultPointcutAdvisor對象並返回：

public Advisor wrap(Object adviceObject) throws UnknownAdviceTypeException {
		if (adviceObject instanceof Advisor) {
			return (Advisor) adviceObject;
		}
		if (!(adviceObject instanceof Advice)) {
			throw new UnknownAdviceTypeException(adviceObject);
		}
		Advice advice = (Advice) adviceObject;
		if (advice instanceof MethodInterceptor) {
			return new DefaultPointcutAdvisor(advice);
		}
		for (AdvisorAdapter adapter : this.adapters) {
			if (adapter.supportsAdvice(advice)) {
				return new DefaultPointcutAdvisor(advice);
			}
		}
		throw new UnknownAdviceTypeException(advice);
	}

	public DefaultPointcutAdvisor(Advice advice) {
		this(Pointcut.TRUE, advice);
	}

須要注意DefaultPointcutAdvisor構造器裏面傳入了一個Pointcut.TRUE，表示這種擴展的Interceptor是全局的攔截器。下面來看看如何使用：

public class MyMethodInterceptor implements MethodInterceptor {
    @Override
    public Object invoke(MethodInvocation invocation) throws Throwable {

        System.out.println("自定義攔截器");
        return invocation.proceed();
    }
}

首先寫一個類實現MethodInterceptor 接口，在invoke方法中實現咱們的攔截邏輯，而後經過下面的方式測試，只要UserService 有AOP攔截就會發現自定義的MyMethodInterceptor也生效了。

public void costomInterceptorTest() {
        AnnotationAwareAspectJAutoProxyCreator bean = applicationContext.getBean(AnnotationAwareAspectJAutoProxyCreator.class);
        bean.setInterceptorNames("myMethodInterceptor ");

        UserService userService = applicationContext.getBean(UserService.class);
        userService.queryUser("dark");
    }

可是若是換個順序，像下面這樣：

public void costomInterceptorTest() {

        UserService userService = applicationContext.getBean(UserService.class);

        AnnotationAwareAspectJAutoProxyCreator bean = applicationContext.getBean(AnnotationAwareAspectJAutoProxyCreator.class);
        bean.setInterceptorNames("myMethodInterceptor ");

        userService.queryUser("dark");
    }

這時自定義的全局攔截器就沒有做用了，這是爲何呢？由於當執行getBean的時候，若是有切面匹配就會經過ProxyFactory去建立代理對象，注意Interceptor是存到這個Factory對象中的，而這個對象和代理對象是一一對應的，所以調用getBean時，尚未myMethodInterceptor這個對象，自定義攔截器就沒有效果了，也就是說要想自定義攔截器生效，就必須在代理對象生成以前註冊進去。

2、循環依賴三級緩存存在的必要性

在上一篇文章我分析了Spring是如何經過三級緩存來解決循環依賴的問題的，但你是否考慮過第三級緩存爲何要存在？我直接將bean存到二級不就好了麼，爲何還要存一個ObjectFactory對象到第三級緩存中？這個在學習了AOP以後就很清楚了，由於咱們在@Autowired對象時，想要注入的不必定是Bean自己，而是想要注入一個修改事後的對象，如代理對象。在AbstractAutowireCapableBeanFactory.getEarlyBeanReference方法中循環調用了SmartInstantiationAwareBeanPostProcessor.getEarlyBeanReference方法，AbstractAutoProxyCreator對象就實現了該方法：

public Object getEarlyBeanReference(Object bean, String beanName) {
		Object cacheKey = getCacheKey(bean.getClass(), beanName);
		if (!this.earlyProxyReferences.contains(cacheKey)) {
			this.earlyProxyReferences.add(cacheKey);
		}
		// 建立代理對象
		return wrapIfNecessary(bean, beanName, cacheKey);
	}

所以，當咱們想要對循壞依賴的Bean作出修改時，就能夠像AOP這樣作。

3、如何在Bean建立以前提早建立代理對象

Spring的代理對象基本上都是在Bean實例化完成以後建立的，但在文章開始我就說過，Spring也提供了一個機會在建立Bean對象以前就建立代理對象，在AbstractAutowireCapableBeanFactory.resolveBeforeInstantiation方法中：

protected Object resolveBeforeInstantiation(String beanName, RootBeanDefinition mbd) {
		Object bean = null;
		if (!Boolean.FALSE.equals(mbd.beforeInstantiationResolved)) {
			// Make sure bean class is actually resolved at this point.
			if (!mbd.isSynthetic() && hasInstantiationAwareBeanPostProcessors()) {
				Class<?> targetType = determineTargetType(beanName, mbd);
				if (targetType != null) {
					bean = applyBeanPostProcessorsBeforeInstantiation(targetType, beanName);
					if (bean != null) {
						bean = applyBeanPostProcessorsAfterInitialization(bean, beanName);
					}
				}
			}
			mbd.beforeInstantiationResolved = (bean != null);
		}
		return bean;
	}

	protected Object applyBeanPostProcessorsBeforeInstantiation(Class<?> beanClass, String beanName) {
		for (BeanPostProcessor bp : getBeanPostProcessors()) {
			if (bp instanceof InstantiationAwareBeanPostProcessor) {
				InstantiationAwareBeanPostProcessor ibp = (InstantiationAwareBeanPostProcessor) bp;
				Object result = ibp.postProcessBeforeInstantiation(beanClass, beanName);
				if (result != null) {
					return result;
				}
			}
		}
		return null;
	}

主要是InstantiationAwareBeanPostProcessor.postProcessBeforeInstantiation方法中，這裏又會進入到AbstractAutoProxyCreator類中：

public Object postProcessBeforeInstantiation(Class<?> beanClass, String beanName) {
		TargetSource targetSource = getCustomTargetSource(beanClass, beanName);
		if (targetSource != null) {
			if (StringUtils.hasLength(beanName)) {
				this.targetSourcedBeans.add(beanName);
			}
			Object[] specificInterceptors = getAdvicesAndAdvisorsForBean(beanClass, beanName, targetSource);
			Object proxy = createProxy(beanClass, beanName, specificInterceptors, targetSource);
			this.proxyTypes.put(cacheKey, proxy.getClass());
			return proxy;
		}

		return null;
	}

	protected TargetSource getCustomTargetSource(Class<?> beanClass, String beanName) {
		// We can't create fancy target sources for directly registered singletons.
		if (this.customTargetSourceCreators != null &&
				this.beanFactory != null && this.beanFactory.containsBean(beanName)) {
			for (TargetSourceCreator tsc : this.customTargetSourceCreators) {
				TargetSource ts = tsc.getTargetSource(beanClass, beanName);
				if (ts != null) {
					return ts;
				}
			}
		}

		// No custom TargetSource found.
		return null;
	}

看到這裏大體應該明白了，先是獲取到一個自定義的TargetSource對象，而後建立代理對象，因此咱們首先須要本身實現一個TargetSource類，這裏直接繼承一個抽象類，getTarget方法則返回原始對象：

public class MyTargetSource extends AbstractBeanFactoryBasedTargetSource {
    @Override
    public Object getTarget() throws Exception {
        return getBeanFactory().getBean(getTargetBeanName());
    }
}

但這還不夠，上面首先判斷了customTargetSourceCreators!=null，而這個屬性是個數組，能夠經過下面這個方法設置進來：

public void setCustomTargetSourceCreators(TargetSourceCreator... targetSourceCreators) {
		this.customTargetSourceCreators = targetSourceCreators;
	}

因此咱們還要實現一個TargetSourceCreator類，一樣繼承一個抽象類實現，並只對userServiceImpl對象進行攔截：

public class MyTargetSourceCreator extends AbstractBeanFactoryBasedTargetSourceCreator {
    @Override
    protected AbstractBeanFactoryBasedTargetSource createBeanFactoryBasedTargetSource(Class<?> beanClass, String beanName) {

        if (getBeanFactory() instanceof ConfigurableListableBeanFactory) {
            if(beanName.equalsIgnoreCase("userServiceImpl")) {
                return new MyTargetSource();
            }
        }

        return null;
    }
}

createBeanFactoryBasedTargetSource方法是在AbstractBeanFactoryBasedTargetSourceCreator.getTargetSource中調用的，而getTargetSource就是在上面getCustomTargetSource中調用的。以上工做作完後，還須要將其設置到AnnotationAwareAspectJAutoProxyCreator對象中，所以須要咱們注入這個對象：

@Configuration
public class TargetSourceCreatorBean {

    @Autowired
    private BeanFactory beanFactory;

   @Bean
    public AnnotationAwareAspectJAutoProxyCreator annotationAwareAspectJAutoProxyCreator() {
        AnnotationAwareAspectJAutoProxyCreator creator = new AnnotationAwareAspectJAutoProxyCreator();
        MyTargetSourceCreator myTargetSourceCreator = new MyTargetSourceCreator();
        myTargetSourceCreator.setBeanFactory(beanFactory);
        creator.setCustomTargetSourceCreators(myTargetSourceCreator);
        return creator;
    }
}

這樣，當咱們經過getBean獲取userServiceImpl的對象時，就會優先生成代理對象，而後在調用執行鏈的過程當中再經過TargetSource.getTarget獲取到被代理對象。可是，爲何咱們在getTarget方法中調用getBean就能拿到被代理對象呢？
繼續探究，經過斷點我發現從getTarget進入時，在resolveBeforeInstantiation方法中返回的bean就是null了，而getBeanPostProcessors方法返回的Processors中也沒有了AnnotationAwareAspectJAutoProxyCreator對象，也就是沒有進入到AbstractAutoProxyCreator.postProcessBeforeInstantiation方法中，因此不會再次獲取到代理對象，那AnnotationAwareAspectJAutoProxyCreator對象是在何時移除的呢？
帶着問題，我開始反推，發如今AbstractBeanFactoryBasedTargetSourceCreator類中有這樣一個方法buildInternalBeanFactory：

protected DefaultListableBeanFactory buildInternalBeanFactory(ConfigurableBeanFactory containingFactory) {
		DefaultListableBeanFactory internalBeanFactory = new DefaultListableBeanFactory(containingFactory);

		// Required so that all BeanPostProcessors, Scopes, etc become available.
		internalBeanFactory.copyConfigurationFrom(containingFactory);

		// Filter out BeanPostProcessors that are part of the AOP infrastructure,
		// since those are only meant to apply to beans defined in the original factory.
		internalBeanFactory.getBeanPostProcessors().removeIf(beanPostProcessor ->
				beanPostProcessor instanceof AopInfrastructureBean);

		return internalBeanFactory;
	}

在這裏移除掉了全部AopInfrastructureBean的子類，而AnnotationAwareAspectJAutoProxyCreator就是其子類，那這個方法是在哪裏調用的呢？繼續反推：

protected DefaultListableBeanFactory getInternalBeanFactoryForBean(String beanName) {
		synchronized (this.internalBeanFactories) {
			DefaultListableBeanFactory internalBeanFactory = this.internalBeanFactories.get(beanName);
			if (internalBeanFactory == null) {
				internalBeanFactory = buildInternalBeanFactory(this.beanFactory);
				this.internalBeanFactories.put(beanName, internalBeanFactory);
			}
			return internalBeanFactory;
		}
	}

	public final TargetSource getTargetSource(Class<?> beanClass, String beanName) {
		AbstractBeanFactoryBasedTargetSource targetSource =
				createBeanFactoryBasedTargetSource(beanClass, beanName);
		
		// 建立完targetSource後就移除掉AopInfrastructureBean類型的BeanPostProcessor對象，如AnnotationAwareAspectJAutoProxyCreator
		DefaultListableBeanFactory internalBeanFactory = getInternalBeanFactoryForBean(beanName);

		......
		return targetSource;
	}

至此，關於TargetSource接口擴展的原理就搞明白了。