你所不知道的Spring的@Autowired实现细节分析

前言

相信很多Java开发都遇到过一个面试题：Resource和Autowired的区别是什么？这个问题的答案相信基本都清楚，但是这两者在Spring中是如何实现的呢？这就要分析Spring源码才能知道了。友情提示：本篇主要是讲解Autowired的实现原理，不会分析Spring初始化的过程，不熟悉的读者可以先阅读笔者之前的一篇文章《这一次搞懂Spring的Bean实例化原理》。

正文

在Spring Bean的整个创建过程中会调用很多BeanPostProcessor接口的的实现类：

上图是我整理的Spring Bean的创建过程及整个过程中各个BeanPostProcessor和回调的调用，右边相同颜色的连线代表是同一组调用，主要看到AutowiredAnnotationBeanPostProcessor和CommonAnnotationBeanPostProcessor，前者就是支持 @Autowired和@Value注解，后者则是支持@PostConstruct、@PreDestroy、@Resource注解。先了解这两个Processor的作用，下面从头分析。

从图中可以看到，在createBeanInstance方法中会调用SmartInstantiationAwareBeanPostProcessor类型的determineCandidateConstructors，这个方法是做什么的呢？看代码：

protected BeanWrapper createBeanInstance(String beanName, RootBeanDefinition mbd, @Nullable Object[] args) {
 ......
 // Candidate constructors for autowiring?
 //寻找当前正在实例化的bean中有@Autowired注解的构造函数
 Constructor<?>[] ctors = determineConstructorsFromBeanPostProcessors(beanClass, beanName);
 if (ctors != null || mbd.getResolvedAutowireMode() == AUTOWIRE_CONSTRUCTOR ||
  mbd.hasConstructorArgumentValues() || !ObjectUtils.isEmpty(args)) {
  //如果ctors不为空，就说明构造函数上有@Autowired注解
  return autowireConstructor(beanName, mbd, ctors, args);
 }
 ......

 // No special handling: simply use no-arg constructor.
 return instantiateBean(beanName, mbd);
 }

 protected Constructor<?>[] determineConstructorsFromBeanPostProcessors(@Nullable Class<?> beanClass, String beanName)
  throws BeansException {

 if (beanClass != null && hasInstantiationAwareBeanPostProcessors()) {
  for (BeanPostProcessor bp : getBeanPostProcessors()) {
  if (bp instanceof SmartInstantiationAwareBeanPostProcessor) {
   SmartInstantiationAwareBeanPostProcessor ibp = (SmartInstantiationAwareBeanPostProcessor) bp;
   Constructor<?>[] ctors = ibp.determineCandidateConstructors(beanClass, beanName);
   if (ctors != null) {
   return ctors;
   }
  }
  }
 }
 return null;
 }

createBeanInstance方法是去实例化Bean，而调用AutowiredAnnotationBeanPostProcessor.determineCandidateConstructors的目的就是先去找到带有@Autowired注解的构造方法（自动注入有三种模式：属性、构造方法、普通方法），也就是通过构造方法注入，如果没有找到则通过反射调用无参构造实例化。平时我们基本上都是使用的属性注入，所以一般都不会进入determineCandidateConstructors方法，所以这里也不详细阐述，感兴趣的读者可自行看看。

接着回到doCreateBean方法中，就可以看到调用了applyMergedBeanDefinitionPostProcessors方法：

 protected void applyMergedBeanDefinitionPostProcessors(RootBeanDefinition mbd, Class<?> beanType, String beanName) {
 for (BeanPostProcessor bp : getBeanPostProcessors()) {
  if (bp instanceof MergedBeanDefinitionPostProcessor) {
  /**
   * CommonAnnotationBeanPostProcessor 支持了@PostConstruct，@PreDestroy,@Resource注解
   * AutowiredAnnotationBeanPostProcessor 支持 @Autowired,@Value注解
   */
  MergedBeanDefinitionPostProcessor bdp = (MergedBeanDefinitionPostProcessor) bp;
  bdp.postProcessMergedBeanDefinition(mbd, beanType, beanName);
  }
 }
 }

这个方法本质上就是调用MergedBeanDefinitionPostProcessor类型的postProcessMergedBeanDefinition方法，通过这个方法去收集@Autowired、@Resource等注解，这里主要分析AutowiredAnnotationBeanPostProcessor的实现：

 public void postProcessMergedBeanDefinition(RootBeanDefinition beanDefinition, Class<?> beanType, String beanName) {
 InjectionMetadata metadata = findAutowiringMetadata(beanName, beanType, null);
 metadata.checkConfigMembers(beanDefinition);
 }

 private InjectionMetadata findAutowiringMetadata(String beanName, Class<?> clazz, @Nullable PropertyValues pvs) {
 // Fall back to class name as cache key, for backwards compatibility with custom callers.
 String cacheKey = (StringUtils.hasLength(beanName) ? beanName : clazz.getName());
 // Quick check on the concurrent map first, with minimal locking.
 InjectionMetadata metadata = this.injectionMetadataCache.get(cacheKey);
 if (InjectionMetadata.needsRefresh(metadata, clazz)) {
  synchronized (this.injectionMetadataCache) {
  metadata = this.injectionMetadataCache.get(cacheKey);
  if (InjectionMetadata.needsRefresh(metadata, clazz)) {
   if (metadata != null) {
   metadata.clear(pvs);
   }
   //主要看这个方法
   metadata = buildAutowiringMetadata(clazz);
   this.injectionMetadataCache.put(cacheKey, metadata);
  }
  }
 }
 return metadata;
 }

 private InjectionMetadata buildAutowiringMetadata(final Class<?> clazz) {
 List<InjectionMetadata.InjectedElement> elements = new ArrayList<>();
 Class<?> targetClass = clazz;

 do {
  final List<InjectionMetadata.InjectedElement> currElements = new ArrayList<>();

  // 找到带有@Autowired注解的属性并封装为AnnotationAttributes
  ReflectionUtils.doWithLocalFields(targetClass, field -> {
  AnnotationAttributes ann = findAutowiredAnnotation(field);
  if (ann != null) {
   if (Modifier.isStatic(field.getModifiers())) {
   if (logger.isInfoEnabled()) {
    logger.info("Autowired annotation is not supported on static fields: " + field);
   }
   return;
   }
   boolean required = determineRequiredStatus(ann);
   currElements.add(new AutowiredFieldElement(field, required));
  }
  });

  // 找到带有@Autowired注解的方法并封装为AnnotationAttributes
  ReflectionUtils.doWithLocalMethods(targetClass, method -> {
  Method bridgedMethod = BridgeMethodResolver.findBridgedMethod(method);
  if (!BridgeMethodResolver.isVisibilityBridgeMethodPair(method, bridgedMethod)) {
   return;
  }
  AnnotationAttributes ann = findAutowiredAnnotation(bridgedMethod);
  if (ann != null && method.equals(ClassUtils.getMostSpecificMethod(method, clazz))) {
   if (Modifier.isStatic(method.getModifiers())) {
   if (logger.isInfoEnabled()) {
    logger.info("Autowired annotation is not supported on static methods: " + method);
   }
   return;
   }
   if (method.getParameterCount() == 0) {
   if (logger.isInfoEnabled()) {
    logger.info("Autowired annotation should only be used on methods with parameters: " +
     method);
   }
   }
   boolean required = determineRequiredStatus(ann);
   PropertyDescriptor pd = BeanUtils.findPropertyForMethod(bridgedMethod, clazz);
   currElements.add(new AutowiredMethodElement(method, required, pd));
  }
  });

  elements.addAll(0, currElements);
  targetClass = targetClass.getSuperclass();
 }
 while (targetClass != null && targetClass != Object.class);

 return new InjectionMetadata(clazz, elements);
 }

收集的逻辑主要在findAutowiringMetadata方法中，层层调用后可以看到是通过findAutowiredAnnotation这个方法去找到带有@Autowired和@Value注解的属性和方法：

 private final Set<Class<? extends Annotation>> autowiredAnnotationTypes = new LinkedHashSet<>(4);

 public AutowiredAnnotationBeanPostProcessor() {
 this.autowiredAnnotationTypes.add(Autowired.class);
 this.autowiredAnnotationTypes.add(Value.class);
 try {
  this.autowiredAnnotationTypes.add((Class<? extends Annotation>)
   ClassUtils.forName("javax.inject.Inject", AutowiredAnnotationBeanPostProcessor.class.getClassLoader()));
  logger.trace("JSR-330 'javax.inject.Inject' annotation found and supported for autowiring");
 }
 catch (ClassNotFoundException ex) {
  // JSR-330 API not available - simply skip.
 }
 }

 private AnnotationAttributes findAutowiredAnnotation(AccessibleObject ao) {
 if (ao.getAnnotations().length > 0) { // autowiring annotations have to be local
  for (Class<? extends Annotation> type : this.autowiredAnnotationTypes) {
  AnnotationAttributes attributes = AnnotatedElementUtils.getMergedAnnotationAttributes(ao, type);
  if (attributes != null) {
   return attributes;
  }
  }
 }
 return null;
 }

最后将其封装为AutowiredFieldElement和AutowiredMethodElement对象的list并连同Class一起封装成InjectionMetadata返回，这就完成了相关注解的收集。

收集完成后在哪里使用呢？对Bean生命周期熟悉的读者都知道，之后就会进行依赖注入，自然相关的调用就在populateBean这个方法里：

protected void populateBean(String beanName, RootBeanDefinition mbd, @Nullable BeanWrapper bw) {

 ......

 PropertyValues pvs = (mbd.hasPropertyValues() ? mbd.getPropertyValues() : null);

 if (mbd.getResolvedAutowireMode() == AUTOWIRE_BY_NAME || mbd.getResolvedAutowireMode() == AUTOWIRE_BY_TYPE) {
  MutablePropertyValues newPvs = new MutablePropertyValues(pvs);
  // Add property values based on autowire by name if applicable.
  if (mbd.getResolvedAutowireMode() == AUTOWIRE_BY_NAME) {
  autowireByName(beanName, mbd, bw, newPvs);
  }
  // Add property values based on autowire by type if applicable.
  if (mbd.getResolvedAutowireMode() == AUTOWIRE_BY_TYPE) {
  autowireByType(beanName, mbd, bw, newPvs);
  }
  pvs = newPvs;
 }

 boolean hasInstAwareBpps = hasInstantiationAwareBeanPostProcessors();
 boolean needsDepCheck = (mbd.getDependencyCheck() != AbstractBeanDefinition.DEPENDENCY_CHECK_NONE);

 PropertyDescriptor[] filteredPds = null;

 //重点看这个if代码块，重要程度 5
 if (hasInstAwareBpps) {
  if (pvs == null) {
  pvs = mbd.getPropertyValues();
  }
  for (BeanPostProcessor bp : getBeanPostProcessors()) {
  if (bp instanceof InstantiationAwareBeanPostProcessor) {
   InstantiationAwareBeanPostProcessor ibp = (InstantiationAwareBeanPostProcessor) bp;
   // 依赖注入过程，@Autowired的支持
   PropertyValues pvsToUse = ibp.postProcessProperties(pvs, bw.getWrappedInstance(), beanName);
   if (pvsToUse == null) {
   if (filteredPds == null) {
    filteredPds = filterPropertyDescriptorsForDependencyCheck(bw, mbd.allowCaching);
   }

   // 老版本用这个完成依赖注入过程，@Autowired的支持
   pvsToUse = ibp.postProcessPropertyValues(pvs, filteredPds, bw.getWrappedInstance(), beanName);
   if (pvsToUse == null) {
    return;
   }
   }
   pvs = pvsToUse;
  }
  }
 }
 if (needsDepCheck) {
  if (filteredPds == null) {
  filteredPds = filterPropertyDescriptorsForDependencyCheck(bw, mbd.allowCaching);
  }
  checkDependencies(beanName, mbd, filteredPds, pvs);
 }

 // xml中<property>标签的依赖注入
 if (pvs != null) {
  applyPropertyValues(beanName, mbd, bw, pvs);
 }
 }

mbd.getResolvedAutowireMode() == AUTOWIRE_BY_NAME || mbd.getResolvedAutowireMode() == AUTOWIRE_BY_TYPE

很多读者包括网上很多文章在看到这个判断时，都认为自动注入的逻辑就是这里，但实际上并不是，这里是自动注入没错，但却是针对以前xml配置，如下：

<?xml version="1.0" encoding="UTF-8"?>
<beans xmlns="http://www.springframework.org/schema/beans"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xsi:schemaLocation="
      http://www.springframework.org/schema/beans
      http://www.springframework.org/schema/beans/spring-beans-2.5.xsd"
    default-autowire="byName">
 <bean id="a" class="com.A" autowire="byName"/>
 <bean id="b" class="com.B" autowire="byType"/>
</beans> 

头文件中的default-autowire属性就是开启全局自动注入，而bean标签上的autowire则是特定针对当前bean的，会覆盖全局配置。这样我们配置后，bean标签就无需配置prototype子标签，也能自动注入对应的对象。这些属性包含了以下几个值：

no：默认值。表示不进行自动注入。对应BeanDefinition中autowireMode的值为 0。

byName：根据名称进行自动注入。对应BeanDefinition中autowireMode的值为1。

byType：根据类型进行自动注入，如果容器中找到两个及以上符合该类型的Bean就将抛出异常。对应BeanDefinition中autowireMode的值为2。

constructor：等同于byType，只是当指定autowire=”constructor”时表示将通过构造方法根据类型进行自动注入。对应BeanDefinition中autowireMode的值为3。

这就是xml配置中的自动注入，而我们使用@Autowired注解时，BeanDefinition中autowireMode的值为 0，即表示不进行自动注入。插一句题外话，网上很多人在争论@Autowired是自动注入还是手动注入，我个人认为都算自动注入，不能说它没有进入这段逻辑就不能叫自动注入，只是它以另外一种方式实现了，至少也没有让我们自己手动new并设置属性。

那这另外一种方式是什么呢？就是下面这个代码干的事：

if (hasInstAwareBpps) {
 if (pvs == null) {
 pvs = mbd.getPropertyValues();
 }
 for (BeanPostProcessor bp : getBeanPostProcessors()) {
 if (bp instanceof InstantiationAwareBeanPostProcessor) {
  InstantiationAwareBeanPostProcessor ibp = (InstantiationAwareBeanPostProcessor) bp;
  // 依赖注入过程，@Autowired的支持
  PropertyValues pvsToUse = ibp.postProcessProperties(pvs, bw.getWrappedInstance(), beanName);
  if (pvsToUse == null) {
  if (filteredPds == null) {
   filteredPds = filterPropertyDescriptorsForDependencyCheck(bw, mbd.allowCaching);
  }

  // 老版本用这个完成依赖注入过程，@Autowired的支持
  pvsToUse = ibp.postProcessPropertyValues(pvs, filteredPds, bw.getWrappedInstance(), beanName);
  if (pvsToUse == null) {
   return;
  }
  }
  pvs = pvsToUse;
 }
 }
}

可以看到这里又是调用了InstantiationAwareBeanPostProcessor类型的postProcessProperties和postProcessPropertyValues方法，后者是老版本中的实现，已经废弃，所以直接看postProcessProperties，还是进入到AutowiredAnnotationBeanPostProcessor类中：

public PropertyValues postProcessProperties(PropertyValues pvs, Object bean, String beanName) {
 InjectionMetadata metadata = findAutowiringMetadata(beanName, bean.getClass(), pvs);
 try {
  metadata.inject(bean, beanName, pvs);
 }
 catch (BeanCreationException ex) {
  throw ex;
 }
 catch (Throwable ex) {
  throw new BeanCreationException(beanName, "Injection of autowired dependencies failed", ex);
 }
 return pvs;
 }

findAutowiringMetadata这个方法不陌生的，刚刚已经分析了，是去收集对应注解并封装为InjectionMetadata对象放入到缓存，这里就是从缓存中拿到值，注入则是通过inject实现的：

public void inject(Object target, @Nullable String beanName, @Nullable PropertyValues pvs) throws Throwable {
 Collection<InjectedElement> checkedElements = this.checkedElements;
 Collection<InjectedElement> elementsToIterate =
  (checkedElements != null ? checkedElements : this.injectedElements);
 if (!elementsToIterate.isEmpty()) {
 for (InjectedElement element : elementsToIterate) {
  if (logger.isTraceEnabled()) {
  logger.trace("Processing injected element of bean '" + beanName + "': " + element);
  }
  if(element.isField) {
  Field field = (Field)element.member;
  System.out.println("==IOC/DI===beanName==" + beanName + "==field[" + field.getName() +"]-> getBean(" + field.getName() + ")");
  }
  element.inject(target, beanName, pvs);
 }
 }
}

最后就是调用element.inject实现注入，element我们刚刚也看到了，就是AutowiredFieldElement和AutowiredMethodElement，分别实现属性注入和方法注入，这里我们看最常用的属性注入就行了：

protected void inject(Object bean, @Nullable String beanName, @Nullable PropertyValues pvs) throws Throwable {
 Field field = (Field) this.member;
 Object value;
 if (this.cached) {
 value = resolvedCachedArgument(beanName, this.cachedFieldValue);
 }
 else {
 DependencyDescriptor desc = new DependencyDescriptor(field, this.required);
 desc.setContainingClass(bean.getClass());
 Set<String> autowiredBeanNames = new LinkedHashSet<>(1);
 Assert.state(beanFactory != null, "No BeanFactory available");
 TypeConverter typeConverter = beanFactory.getTypeConverter();
 try {
  // 找到依赖对象
  value = beanFactory.resolveDependency(desc, beanName, autowiredBeanNames, typeConverter);
 }
 catch (BeansException ex) {
  throw new UnsatisfiedDependencyException(null, beanName, new InjectionPoint(field), ex);
 }
 synchronized (this) {
  if (!this.cached) {
  if (value != null || this.required) {
   this.cachedFieldValue = desc;
   registerDependentBeans(beanName, autowiredBeanNames);
   if (autowiredBeanNames.size() == 1) {
   String autowiredBeanName = autowiredBeanNames.iterator().next();
   if (beanFactory.containsBean(autowiredBeanName) &&
    beanFactory.isTypeMatch(autowiredBeanName, field.getType())) {
    this.cachedFieldValue = new ShortcutDependencyDescriptor(
     desc, autowiredBeanName, field.getType());
   }
   }
  }
  else {
   this.cachedFieldValue = null;
  }
  this.cached = true;
  }
 }
 }

 if (value != null) {
 ReflectionUtils.makeAccessible(field);
 field.set(bean, value);
 }
}

这段代码整体逻辑比较清晰，首先根据field创建一个依赖对象的抽象DependencyDescriptor对象，然后通过beanFactory.resolveDependency解析拿到对应的实例，最后通过反射注入即可。

因此我们主要看resolveDependency方法中做了什么：

public Object resolveDependency(DependencyDescriptor descriptor, @Nullable String requestingBeanName,
 @Nullable Set<String> autowiredBeanNames, @Nullable TypeConverter typeConverter) throws BeansException {

 descriptor.initParameterNameDiscovery(getParameterNameDiscoverer());
 if (Optional.class == descriptor.getDependencyType()) {
 return createOptionalDependency(descriptor, requestingBeanName);
 }
 else if (ObjectFactory.class == descriptor.getDependencyType() ||
  ObjectProvider.class == descriptor.getDependencyType()) {
 return new DependencyObjectProvider(descriptor, requestingBeanName);
 }
 else if (javaxInjectProviderClass == descriptor.getDependencyType()) {
 return new Jsr330Factory().createDependencyProvider(descriptor, requestingBeanName);
 }
 else {
 Object result = getAutowireCandidateResolver().getLazyResolutionProxyIfNecessary(
  descriptor, requestingBeanName);
 if (result == null) {
  result = doResolveDependency(descriptor, requestingBeanName, autowiredBeanNames, typeConverter);
 }
 return result;
 }
}

一般情况下，都是走的else分支并调用doResolveDependency方法：

public Object doResolveDependency(DependencyDescriptor descriptor, @Nullable String beanName,
 @Nullable Set<String> autowiredBeanNames, @Nullable TypeConverter typeConverter) throws BeansException {

 InjectionPoint previousInjectionPoint = ConstructorResolver.setCurrentInjectionPoint(descriptor);
 try {
 Object shortcut = descriptor.resolveShortcut(this);
 if (shortcut != null) {
  return shortcut;
 }

 Class<?> type = descriptor.getDependencyType();
 Object value = getAutowireCandidateResolver().getSuggestedValue(descriptor);
 if (value != null) {
  if (value instanceof String) {
  String strVal = resolveEmbeddedValue((String) value);
  BeanDefinition bd = (beanName != null && containsBean(beanName) ? getMergedBeanDefinition(beanName) : null);
  value = evaluateBeanDefinitionString(strVal, bd);
  }
  TypeConverter converter = (typeConverter != null ? typeConverter : getTypeConverter());
  return (descriptor.getField() != null ?
   converter.convertIfNecessary(value, type, descriptor.getField()) :
   converter.convertIfNecessary(value, type, descriptor.getMethodParameter()));
 }

 // 有多个实现类需要注入，特指注入的是数组、集合或者Map
 Object multipleBeans = resolveMultipleBeans(descriptor, beanName, autowiredBeanNames, typeConverter);
 if (multipleBeans != null) {
  return multipleBeans;
 }

 // 找到依赖对象的所有实现类
 Map<String, Object> matchingBeans = findAutowireCandidates(beanName, type, descriptor);
 if (matchingBeans.isEmpty()) {
  if (isRequired(descriptor)) {
  raiseNoMatchingBeanFound(type, descriptor.getResolvableType(), descriptor);
  }
  return null;
 }

 String autowiredBeanName;
 Object instanceCandidate;

 // 依赖的对象有多个实例
 if (matchingBeans.size() > 1) {
  // 根据@Primary、@Priority和名称依次进行匹配注入
  autowiredBeanName = determineAutowireCandidate(matchingBeans, descriptor);
  if (autowiredBeanName == null) {
  if (isRequired(descriptor) || !indicatesMultipleBeans(type)) {
   return descriptor.resolveNotUnique(descriptor.getResolvableType(), matchingBeans);
  }
  else {
   // In case of an optional Collection/Map, silently ignore a non-unique case:
   // possibly it was meant to be an empty collection of multiple regular beans
   // (before 4.3 in particular when we didn't even look for collection beans).
   return null;
  }
  }
  instanceCandidate = matchingBeans.get(autowiredBeanName);
 }
 else {
  // We have exactly one match.
  Map.Entry<String, Object> entry = matchingBeans.entrySet().iterator().next();
  autowiredBeanName = entry.getKey();
  instanceCandidate = entry.getValue();
 }

 if (autowiredBeanNames != null) {
  autowiredBeanNames.add(autowiredBeanName);
 }
 // 如果拿到的是class对象，通过getBean实例化返回
 if (instanceCandidate instanceof Class) {
  instanceCandidate = descriptor.resolveCandidate(autowiredBeanName, type, this);
 }
 Object result = instanceCandidate;
 if (result instanceof NullBean) {
  if (isRequired(descriptor)) {
  raiseNoMatchingBeanFound(type, descriptor.getResolvableType(), descriptor);
  }
  result = null;
 }
 if (!ClassUtils.isAssignableValue(type, result)) {
  throw new BeanNotOfRequiredTypeException(autowiredBeanName, type, instanceCandidate.getClass());
 }
 return result;
 }
 finally {
 ConstructorResolver.setCurrentInjectionPoint(previousInjectionPoint);
 }
}

这里面首先是通过 getAutowireCandidateResolver().getSuggestedValue(descriptor)拿到@Value注解的值，然后通过TypeConverter进行转换，默认可转换的类型是JDK和Spring内置的一些类型，自然不包含我们自定义的类，所以如果不进行扩展在@Autowired注入我们自定义类对象时同时使用@Value注解是会报错的。

接着是调用resolveMultipleBeans方法实现对Map、List、数组等属性的注入，本质上还是调用findAutowireCandidates方法找到所有的实现类的对象装入对应的集合数组中，所以直接看findAutowireCandidates：

protected Map<String, Object> findAutowireCandidates(
 @Nullable String beanName, Class<?> requiredType, DependencyDescriptor descriptor) {

 String[] candidateNames = BeanFactoryUtils.beanNamesForTypeIncludingAncestors(
  this, requiredType, true, descriptor.isEager());
 Map<String, Object> result = new LinkedHashMap<>(candidateNames.length);

 ......

 for (String candidate : candidateNames) {
 if (!isSelfReference(beanName, candidate) && isAutowireCandidate(candidate, descriptor)) {
  addCandidateEntry(result, candidate, descriptor, requiredType);
 }
 }

 ......
 return result;
}

private void addCandidateEntry(Map<String, Object> candidates, String candidateName,
 DependencyDescriptor descriptor, Class<?> requiredType) {

 if (descriptor instanceof MultiElementDescriptor) {
 Object beanInstance = descriptor.resolveCandidate(candidateName, requiredType, this);
 if (!(beanInstance instanceof NullBean)) {
  candidates.put(candidateName, beanInstance);
 }
 }
 else if (containsSingleton(candidateName) || (descriptor instanceof StreamDependencyDescriptor &&
  ((StreamDependencyDescriptor) descriptor).isOrdered())) {
 Object beanInstance = descriptor.resolveCandidate(candidateName, requiredType, this);
 candidates.put(candidateName, (beanInstance instanceof NullBean ? null : beanInstance));
 }
 else {
 candidates.put(candidateName, getType(candidateName));
 }
}

首先通过beanNamesForTypeIncludingAncestors方法拿到依赖类所有的实现类的beanName，然后调用addCandidateEntry将beanName及对应的实例或者Class对象放入到Map中。

接着回到doResolveDependency方法中：

if (matchingBeans.size() > 1) {
 // 根据@Primary、@Priority和名称依次进行匹配注入
 autowiredBeanName = determineAutowireCandidate(matchingBeans, descriptor);
 if (autowiredBeanName == null) {
 if (isRequired(descriptor) || !indicatesMultipleBeans(type)) {
  return descriptor.resolveNotUnique(descriptor.getResolvableType(), matchingBeans);
 }
 else {
  // In case of an optional Collection/Map, silently ignore a non-unique case:
  // possibly it was meant to be an empty collection of multiple regular beans
  // (before 4.3 in particular when we didn't even look for collection beans).
  return null;
 }
 }
 instanceCandidate = matchingBeans.get(autowiredBeanName);
}
else {
 // We have exactly one match.
 Map.Entry<String, Object> entry = matchingBeans.entrySet().iterator().next();
 autowiredBeanName = entry.getKey();
 instanceCandidate = entry.getValue();
}

if (autowiredBeanNames != null) {
 autowiredBeanNames.add(autowiredBeanName);
}
// 如果拿到的是class对象，通过getBean实例化返回
if (instanceCandidate instanceof Class) {
 instanceCandidate = descriptor.resolveCandidate(autowiredBeanName, type, this);
}
Object result = instanceCandidate;

如果只有一个实例则直接返回该实例，如果实现类有多个则调用determineAutowireCandidate进行判断该使用哪一个实例对象，判断规则如下：

首先判断实现类上是否有@Primary注解，找到一个则返回当前实例，找到多个则报错。

若没有标注@Primary注解，则判断是否指定了优先级，且只能是通过@Priority注解指定的，@Order不支持。

上述都没有拿到合适的Bean则通过属性名称获取Bean。

拿到对应的实例后，最后进行反射注入即可。以上就是@Autowired的实现细节。

总结

本篇从源码角度详细分析了@Autowired的实现细节，只有真正阅读了源码才能了解更多的细节，在开发中更加清楚如何注入多个实例以及如何指定注入的优先级，同时在面试中也能更有理有据，而不是统一的大众回答，先根据byType，再根据byName。另外对于方法注入和@Resource注解的处理本篇没有涉及，但是相信看完本文读者自己也能轻松分析这部分源码。