不涉及Spring完整的启动流程，仅仅从Mybatis的视角去分析几个关键的方法，找到Mybatis是如何通过这几个扩展点植入进去的，反过来看Spring是如何设计，埋下这些伏笔，实现其可扩展性。

springContext-mybatis.xml的配置：

<!-- simplest possible SqlSessionFactory configuration -->
<bean id="sqlSessionFactory" class="org.mybatis.spring.SqlSessionFactoryBean">
    <property name="dataSource" ref="dataSource" />
    <property name="configLocation" value="classpath:/mybatis-config.xml"></property>
</bean>

<!-- Scan for mappers and let them be autowired; notice there is no UserDaoImplementation 
    needed. The required SqlSessionFactory will be autowired. -->
<bean class="org.mybatis.spring.mapper.MapperScannerConfigurer">
    <property name="basePackage" value="com.fcs.**.dao,fcs.common.**.dao" />
</bean>

一般情况下，要植入进去，必须通过接口实现，这也是Spring的扩展点。SqlSessionFactoryBean和MapperScannerConfigurer分别实现了这些接口：

public class SqlSessionFactoryBean implements FactoryBean<SqlSessionFactory>,
                            InitializingBean,
                            ApplicationListener<ApplicationEvent> {
}


public class MapperScannerConfigurer implements BeanDefinitionRegistryPostProcessor, 
                            InitializingBean, 
                            ApplicationContextAware, 
                            BeanNameAware {
}

public interface BeanDefinitionRegistryPostProcessor extends BeanFactoryPostProcessor {

    void postProcessBeanDefinitionRegistry(BeanDefinitionRegistry registry) throws BeansException;

}

从AbstractApplicationContext的invokeBeanFactoryPostProcessors方法开始分析：

Map<String, BeanDefinitionRegistryPostProcessor> beanMap = beanFactory.getBeansOfType(BeanDefinitionRegistryPostProcessor.class, true, false);
List<BeanDefinitionRegistryPostProcessor> registryPostProcessorBeans = new ArrayList<BeanDefinitionRegistryPostProcessor>(beanMap.values());

OrderComparator.sort(registryPostProcessorBeans);
for (BeanDefinitionRegistryPostProcessor postProcessor : registryPostProcessorBeans) {
    postProcessor.postProcessBeanDefinitionRegistry(registry);
}

第一步：获取指定类型（BeanDefinitionRegistryPostProcessor）的beanMap

DefaultListableBeanFactory#getBeansOfType：

public <T> Map<String, T> getBeansOfType(Class<T> type, boolean includeNonSingletons, boolean allowEagerInit)
        throws BeansException {

    String[] beanNames = getBeanNamesForType(type, includeNonSingletons, allowEagerInit);
    Map<String, T> result = new LinkedHashMap<String, T>(beanNames.length);
    for (String beanName : beanNames) {
        try {
            result.put(beanName, getBean(beanName, type));
        }
        catch (BeanCreationException ex) {
            Throwable rootCause = ex.getMostSpecificCause();
            if (rootCause instanceof BeanCurrentlyInCreationException) {
                BeanCreationException bce = (BeanCreationException) rootCause;
                if (isCurrentlyInCreation(bce.getBeanName())) {
                    if (this.logger.isDebugEnabled()) {
                        this.logger.debug("Ignoring match to currently created bean '" + beanName + "': " +
                                ex.getMessage());
                    }
                    onSuppressedException(ex);
                    // Ignore: indicates a circular reference when autowiring constructors.
                    // We want to find matches other than the currently created bean itself.
                    continue;
                }
            }
            throw ex;
        }
    }
    return result;
}

• String[] beanNames = getBeanNamesForType(type, includeNonSingletons, allowEagerInit);

获取匹配类型的beanNames

• result.put(beanName, getBean(beanName, type));

根据名称和类型获取bean，放到result集合中

接下来会走bean的创建流程，在AbstractAutowireCapableBeanFactory的doCreateBean方法中：

// Initialize the bean instance.
Object exposedObject = bean;
try {
    populateBean(beanName, mbd, instanceWrapper);
    if (exposedObject != null) {
        exposedObject = initializeBean(beanName, exposedObject, mbd);
    }
}

这里会调用MapperScannerConfigurer的afterPropertiesSet方法：

public void afterPropertiesSet() throws Exception {
    notNull(this.basePackage, "Property 'basePackage' is required");
}

第二步：执行postProcessor的postProcessBeanDefinitionRegistry方法

postProcessor.postProcessBeanDefinitionRegistry(registry);

这样就会调用MapperScannerConfigurer的postProcessBeanDefinitionRegistry方法：

public void postProcessBeanDefinitionRegistry(BeanDefinitionRegistry registry) throws BeansException {
    if (this.processPropertyPlaceHolders) {
      processPropertyPlaceHolders();
    }
    
    ClassPathMapperScanner scanner = new ClassPathMapperScanner(registry);
    scanner.setAddToConfig(this.addToConfig);
    scanner.setAnnotationClass(this.annotationClass);
    scanner.setMarkerInterface(this.markerInterface);
    scanner.setSqlSessionFactory(this.sqlSessionFactory);
    scanner.setSqlSessionTemplate(this.sqlSessionTemplate);
    scanner.setSqlSessionFactoryBeanName(this.sqlSessionFactoryBeanName);
    scanner.setSqlSessionTemplateBeanName(this.sqlSessionTemplateBeanName);
    scanner.setResourceLoader(this.applicationContext);
    scanner.setBeanNameGenerator(this.nameGenerator);
    scanner.registerFilters();
    scanner.scan(StringUtils.tokenizeToStringArray(this.basePackage, ConfigurableApplicationContext.CONFIG_LOCATION_DELIMITERS));
}

这里会开启扫描mapper.xml，ClassPathMapperScanner继承自Spring的ClassPathBeanDefinitionScanner，重写了doScan方法：

public Set<BeanDefinitionHolder> doScan(String... basePackages) {
    Set<BeanDefinitionHolder> beanDefinitions = super.doScan(basePackages);

    if (beanDefinitions.isEmpty()) {
      logger.warn("No MyBatis mapper was found in '" + Arrays.toString(basePackages) + "' package. Please check your configuration.");
    } else {
      for (BeanDefinitionHolder holder : beanDefinitions) {
        GenericBeanDefinition definition = (GenericBeanDefinition) holder.getBeanDefinition();

        if (logger.isDebugEnabled()) {
          logger.debug("Creating MapperFactoryBean with name '" + holder.getBeanName() 
              + "' and '" + definition.getBeanClassName() + "' mapperInterface");
        }

        // the mapper interface is the original class of the bean
        // but, the actual class of the bean is MapperFactoryBean
        definition.getPropertyValues().add("mapperInterface", definition.getBeanClassName());
        definition.setBeanClass(MapperFactoryBean.class);

        definition.getPropertyValues().add("addToConfig", this.addToConfig);

        // ......
      }
    }

    return beanDefinitions;
 }

• Set<BeanDefinitionHolder> beanDefinitions = super.doScan(basePackages);

调用父类的doScan方法构造每个Mapper的BeanDefinitionHolder，同service一样。

definition.getPropertyValues().add("mapperInterface", definition.getBeanClassName());
definition.setBeanClass(MapperFactoryBean.class);

definition.getPropertyValues().add("addToConfig", this.addToConfig);

往definition里设置相关属性，方便后面构造mapper代理类。

第三步：获取指定类型的postProcessorNames（BeanFactoryPostProcessor）并处理

String[] postProcessorNames =
        beanFactory.getBeanNamesForType(BeanFactoryPostProcessor.class, true, false);

// Separate between BeanFactoryPostProcessors that implement PriorityOrdered,
// Ordered, and the rest.
List<BeanFactoryPostProcessor> priorityOrderedPostProcessors = new ArrayList<BeanFactoryPostProcessor>();
List<String> orderedPostProcessorNames = new ArrayList<String>();
List<String> nonOrderedPostProcessorNames = new ArrayList<String>();
for (String ppName : postProcessorNames) {
    if (processedBeans.contains(ppName)) {
        // skip - already processed in first phase above
    }
    else if (isTypeMatch(ppName, PriorityOrdered.class)) {
        priorityOrderedPostProcessors.add(beanFactory.getBean(ppName, BeanFactoryPostProcessor.class));
    }
    else if (isTypeMatch(ppName, Ordered.class)) {
        orderedPostProcessorNames.add(ppName);
    }
    else {
        nonOrderedPostProcessorNames.add(ppName);
    }
}

// First, invoke the BeanFactoryPostProcessors that implement PriorityOrdered.
OrderComparator.sort(priorityOrderedPostProcessors);
invokeBeanFactoryPostProcessors(priorityOrderedPostProcessors, beanFactory);

processedBeans集合包含之前两个beanName：

0 = "org.mybatis.spring.mapper.MapperScannerConfigurer#0"
1 = "org.springframework.context.annotation.internalConfigurationAnnotationProcessor"

postProcessorNames数组包含三个beanName

0 = "propertyConfigurer"
1 = "org.springframework.context.annotation.internalConfigurationAnnotationProcessor"
2 = "org.mybatis.spring.mapper.MapperScannerConfigurer#0"

之前的两个已经创建过了，现在处理剩下的"propertyConfigurer" 一个了。

根据类关系图，自定义的propertyConfigurer刚好也实现了PriorityOrdered接口：

image

所以将会调用其postProcessBeanFactory方法：

private void invokeBeanFactoryPostProcessors(
        Collection<? extends BeanFactoryPostProcessor> postProcessors, ConfigurableListableBeanFactory beanFactory) {

    for (BeanFactoryPostProcessor postProcessor : postProcessors) {
        postProcessor.postProcessBeanFactory(beanFactory);
    }
}

propertyConfigurer可以是我们自定义的XxPropertyPlaceholderConfigurer，继承自PropertyPlaceholderConfigurer就行了。这里也是一个切入点（关于dataSource的配置等）。

AbstractApplicationContext的registerBeanPostProcessors方法：

第四步：处理BeanPostProcessor相关

String[] postProcessorNames = beanFactory.getBeanNamesForType(BeanPostProcessor.class, true, false);

这里找到7个类：

org.springframework.context.annotation.internalAutowiredAnnotationProcessor

org.springframework.context.annotation.internalRequiredAnnotationProcessor

org.springframework.context.annotation.internalCommonAnnotationProcessor

org.springframework.aop.config.internalAutoProxyCreator

&shiroFilter

lifecycleBeanPostProcessor

org.springframework.context.annotation.ConfigurationClassPostProcessor.importAwareProcessor

同处理BeanFactoryPostProcessor一样，也是分为PriorityOrdered、Ordered和其他类型的顺序分别获取bean，然后调用其registerBeanPostProcessors方法注册：

private void registerBeanPostProcessors(
        ConfigurableListableBeanFactory beanFactory, List<BeanPostProcessor> postProcessors) {

    for (BeanPostProcessor postProcessor : postProcessors) {
        beanFactory.addBeanPostProcessor(postProcessor);
    }
}

lifecycleBeanPostProcessor在shiroFilter之前初始化。

这里重点看postProcessorNames中name为&shiroFilter。

在AbstractAutowireCapableBeanFactory的applyPropertyValues方法中找到了securityMannger属性，然后又在安全管理器中带出了realm属性，接着又找到adminUserService属性。

第五步：由一个Mapper的初始化引起所有Mapper的创建

这个的重点在于sqlSessionFactory这个属性，中间是根据这个特殊的type去匹配所有的beandefine，adminUserService中有个@autowire注解的adminUserMapper：

if (mbd.getResolvedAutowireMode() == RootBeanDefinition.AUTOWIRE_BY_NAME ||
        mbd.getResolvedAutowireMode() == RootBeanDefinition.AUTOWIRE_BY_TYPE) {
    MutablePropertyValues newPvs = new MutablePropertyValues(pvs);

    // Add property values based on autowire by name if applicable.
    if (mbd.getResolvedAutowireMode() == RootBeanDefinition.AUTOWIRE_BY_NAME) {
        autowireByName(beanName, mbd, bw, newPvs);
    }

    // Add property values based on autowire by type if applicable.
    if (mbd.getResolvedAutowireMode() == RootBeanDefinition.AUTOWIRE_BY_TYPE) {
        autowireByType(beanName, mbd, bw, newPvs);
    }

    pvs = newPvs;
}

看了下AbstractAutowireCapableBeanFactory中的autowireByType方法：

protected void autowireByType(
        String beanName, AbstractBeanDefinition mbd, BeanWrapper bw, MutablePropertyValues pvs) {

    TypeConverter converter = getCustomTypeConverter();
    if (converter == null) {
        converter = bw;
    }

    Set<String> autowiredBeanNames = new LinkedHashSet<String>(4);
    String[] propertyNames = unsatisfiedNonSimpleProperties(mbd, bw);
    for (String propertyName : propertyNames) {
        try {
            PropertyDescriptor pd = bw.getPropertyDescriptor(propertyName);
            // Don't try autowiring by type for type Object: never makes sense,
            // even if it technically is a unsatisfied, non-simple property.
            if (!Object.class.equals(pd.getPropertyType())) {
                MethodParameter methodParam = BeanUtils.getWriteMethodParameter(pd);
                // Do not allow eager init for type matching in case of a prioritized post-processor.
                boolean eager = !PriorityOrdered.class.isAssignableFrom(bw.getWrappedClass());
                DependencyDescriptor desc = new AutowireByTypeDependencyDescriptor(methodParam, eager);
                Object autowiredArgument = resolveDependency(desc, beanName, autowiredBeanNames, converter);
                if (autowiredArgument != null) {
                    pvs.add(propertyName, autowiredArgument);
                }
                for (String autowiredBeanName : autowiredBeanNames) {
                    registerDependentBean(autowiredBeanName, beanName);
                    if (logger.isDebugEnabled()) {
                        logger.debug("Autowiring by type from bean name '" + beanName + "' via property '" +
                                propertyName + "' to bean named '" + autowiredBeanName + "'");
                    }
                }
                autowiredBeanNames.clear();
            }
        }
        catch (BeansException ex) {
            throw new UnsatisfiedDependencyException(mbd.getResourceDescription(), beanName, propertyName, ex);
        }
    }
}

• String[] propertyNames = unsatisfiedNonSimpleProperties(mbd, bw);

此方法获取了两个属性：

0 = "sqlSessionFactory"
1 = "sqlSessionTemplate"

MapperFactoryBean继承了SqlSessionDaoSupport类：

public abstract class SqlSessionDaoSupport extends DaoSupport {

  private SqlSession sqlSession;

  private boolean externalSqlSession;

  public void setSqlSessionFactory(SqlSessionFactory sqlSessionFactory) {
    if (!this.externalSqlSession) {
      this.sqlSession = new SqlSessionTemplate(sqlSessionFactory);
    }
  }

  public void setSqlSessionTemplate(SqlSessionTemplate sqlSessionTemplate) {
    this.sqlSession = sqlSessionTemplate;
    this.externalSqlSession = true;
  }
  
  // ......

}

遍历这两个属性名称，对于sqlSessionFactory：

MethodParameter methodParam = BeanUtils.getWriteMethodParameter(pd);
// Do not allow eager init for type matching in case of a prioritized post-processor.
boolean eager = !PriorityOrdered.class.isAssignableFrom(bw.getWrappedClass());

MethodParameter跟SqlSessionDaoSupport中的set方法有关，eager判断也为true，接着：

Object autowiredArgument = resolveDependency(desc, beanName, autowiredBeanNames, converter);

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

    descriptor.initParameterNameDiscovery(getParameterNameDiscoverer());
    if (descriptor.getDependencyType().equals(ObjectFactory.class)) {
        return new DependencyObjectFactory(descriptor, beanName);
    }
    else if (descriptor.getDependencyType().equals(javaxInjectProviderClass)) {
        return new DependencyProviderFactory().createDependencyProvider(descriptor, beanName);
    }
    else {
        return doResolveDependency(descriptor, descriptor.getDependencyType(), beanName, autowiredBeanNames, typeConverter);
    }
}

在doResolveDependency方法中：

Map<String, Object> matchingBeans = findAutowireCandidates(beanName, type, descriptor);

DefaultListableBeanFactory的findAutowireCandidates方法：

String[] candidateNames = BeanFactoryUtils.beanNamesForTypeIncludingAncestors(
this, requiredType, true, descriptor.isEager());

在这个方法中传入的还是adminUserMapper，要求的类型是SqlSessionFactory，还有个参数是DependencyDescriptor。

BeanFactoryUtils的beanNamesForTypeIncludingAncestors方法中：

String[] result = lbf.getBeanNamesForType(type, includeNonSingletons, allowEagerInit);

然后在DefaultListableBeanFactory中getBeanNamesForType：

public String[] getBeanNamesForType(Class<?> type, boolean includeNonSingletons, boolean allowEagerInit) {
    if (!isConfigurationFrozen() || type == null || !allowEagerInit) {
        return doGetBeanNamesForType(type, includeNonSingletons, allowEagerInit);
    }
    
    //......
}

doGetBeanNamesForType方法中遍历所有beanDefinitionName并判断是否匹配：

boolean matchFound = (allowEagerInit || !isFactoryBean || containsSingleton(beanName)) &&
    (includeNonSingletons || isSingleton(beanName)) && isTypeMatch(beanName, type);

isTypeMatch方法中就会针对不同的beanName，与type为SqlSessionFactory做匹配：

if (FactoryBean.class.isAssignableFrom(beanType)) {
    if (!BeanFactoryUtils.isFactoryDereference(name)) {
        // If it's a FactoryBean, we want to look at what it creates, not the factory class.
        beanType = getTypeForFactoryBean(beanName, mbd);
        if (beanType == null) {
            return false;
        }
    }
}

beanType由于是MapperFactoryBean，就会获取相关bean。

getTypeForFactoryBean方法到这里也该终结了：

 FactoryBean<?> factoryBean = doGetBean(FACTORY_BEAN_PREFIX + beanName, FactoryBean.class, null, true);

这样就开始创建一个个mapper了。

试想一下，就是有没有这样一种可能，当Spring在创建bean的过程中依赖了其他bean，它就会去处理其依赖（创建依赖对象），然后它根据某种情况判断，将所有用到该属性的bean一起都创建了，这样就引起了所有Mapper对象的创建，合乎逻辑。

留下几个问题：

• 其他的属性都是通过配置找到的 mapper是通过注解 如何处理这些差异
• 如何牵一发而动全身 引起200多个mapper的初始化
• 如果没有使用shiro，初始化过程又会是怎么样的 怎么触发 有无规律可循