vlambda博客
学习文章列表

从源码角度剖析 Spring 如何管理 mybatis 事务的

Ext1:本文源码解析基于 mybatis-spring-boot-starter 2.1.1,即 mybatis 3.5.3 版本。
Ext2:本文主要是对源码的讲解,着重点会是在源码上。
Ext3:阅读本文前,最好对 mapperProxy、 sqlSession 有一定的了解

一、 XMLMapperBuilder、mapperProxy 与 mapperMethod

上篇文章 讲了 mapper 文件是怎么解析的,在文章开头提到了 SqlSessionFactory 这个重要的对象,是的就是我们经常需要配置的:

  @Bean
@ConditionalOnMissingBean
public SqlSessionFactory sqlSessionFactory(DataSource dataSource) throws Exception {
// 略
}

这里面做了很多自动化的配置,当然我们可以通过重写它来自定义我们自己的 sqlSessionFactory,借用一下上篇文章的图片: 

spring 借助 SqlSessionFactoryBean 来创建 sqlSessionFactory,这可以视作是一个典型的建造者模式,来创建 SqlSessionFactory

上篇文章说到,spring 拿到我们配置的 mapper 路径去扫描我们 mapper.xml 然后进行一个循环进行解析(上篇文章第二章节:二、SqlSessionFactory 的初始化与 XMLMapperBuilder):

-- 代码位于 org.mybatis.spring.SqlSessionFactoryBean#buildSqlSessionFactory --

if (this.mapperLocations != null) {
if (this.mapperLocations.length == 0) {
LOGGER.warn(() -> "Property 'mapperLocations' was specified but matching resources are not found.");
} else {
for (Resource mapperLocation : this.mapperLocations) {
if (mapperLocation == null) {
continue;
}
try {
XMLMapperBuilder xmlMapperBuilder = new XMLMapperBuilder(mapperLocation.getInputStream(),
targetConfiguration, mapperLocation.toString(), targetConfiguration.getSqlFragments());
xmlMapperBuilder.parse();
} catch (Exception e) {
throw new NestedIOException("Failed to parse mapping resource: '" + mapperLocation + "'", e);
} finally {
ErrorContext.instance().reset();
}
LOGGER.debug(() -> "Parsed mapper file: '" + mapperLocation + "'");
}
}
} else {
LOGGER.debug(() -> "Property 'mapperLocations' was not specified.");
}

-- 代码位于 org.apache.ibatis.builder.xml.XMLMapperBuilder#parse --

public void parse() {
if (!configuration.isResourceLoaded(resource)) {
configurationElement(parser.evalNode("/mapper")); // 上篇文章主要说的
configuration.addLoadedResource(resource);
bindMapperForNamespace();// 创建mapperProxy的工厂对象
}

parsePendingResultMaps();
parsePendingCacheRefs();
parsePendingStatements();
}

1.1 从 xml 到 mapperStatement

上篇文章实际上就是在讲解 configurationElement(parser.evalNode("/mapper")); 里面发生的故事,实际上还有后续的步骤,如果对 mybatis 有所了解的,应该知道,mybatis 会为我们的接口创建一个叫做 mapperProxy 的代理对象(划重点),其实就是在这后续的步骤 bindMapperForNamespace(); 做的(不尽然,实际上是创建并绑定了 mapperProxyFactory)。

从源码角度剖析 Spring 如何管理 mybatis 事务的

不贴太多代码,bindMapperForNamespace() 方法里核心做的主要就是调用 configuration.addMapper() 方法

      if (boundType != null) {
if (!configuration.hasMapper(boundType)) {
// Spring may not know the real resource name so we set a flag
// to prevent loading again this resource from the mapper interface
// look at MapperAnnotationBuilder#loadXmlResource
configuration.addLoadedResource("namespace:" + namespace);
configuration.addMapper(boundType);
}
}

这个 boundType 就是我们在 mapper 文件里面指定的 namespace,比如:

<mapper namespace="com.anur.mybatisdemo.test.TrackerConfigMapper">
XXXXXXXXXXXXXXXXXX 里面写的sql语句,resultMap 等等,略
</mapper>

在 configuration.addMapper() 中调用了 mapperRegistry.addMapper(),看到 knowMappers ,这个就是存储我们生产 MapperProxy 的工厂映射 map,我们稍微再讲,先继续往下看。

  public <T> void addMapper(Class<T> type) {
if (type.isInterface()) {
if (hasMapper(type)) {
throw new BindingException("Type " + type + " is already known to the MapperRegistry.");
}
boolean loadCompleted = false;
try {
knownMappers.put(type, new MapperProxyFactory<>(type));
// It's important that the type is added before the parser is run
// otherwise the binding may automatically be attempted by the
// mapper parser. If the type is already known, it won't try.
MapperAnnotationBuilder parser = new MapperAnnotationBuilder(config, type);
parser.parse();
loadCompleted = true;
} finally {
if (!loadCompleted) {
knownMappers.remove(type);
}
}
}
}

1.2 从注解到 mapperStatement

看到 MapperAnnotationBuilder#parse()parse() 中主要是对这个接口里面定义的方法做了 parseStatement 这件事

      for (Method method : methods) {
try {
// issue #237
if (!method.isBridge()) {
parseStatement(method);
}
} catch (IncompleteElementException e) {
configuration.addIncompleteMethod(new MethodResolver(this, method));
}
}

parseStatement() 就是解析注解语句的地方, 如果说我们没有写 xml,将语句以注解的形式写在方法上,则会在这里进行语句解析。它和我们上篇文章讲到的解析xml很像,就是拿到一大堆属性,比如 resultMapkeyGenerator 等等,生成一个 MappedStatement 对象,这里就不赘述了。

void parseStatement(Method method) {
Class<?> parameterTypeClass = getParameterType(method);
LanguageDriver languageDriver = getLanguageDriver(method);
SqlSource sqlSource = getSqlSourceFromAnnotations(method, parameterTypeClass, languageDriver);
if (sqlSource != null) {
// 解析注解式的 sql 语句,略
}
}

1.3 如果写了 xml,也写了注解会怎么样(调皮)

我们知道承载 mapperStatement 的是一个 map 映射,通过我们在上篇文章中反复强调的 id 来作为 key,那么重复添加会出现什么呢?

答案在这里,mybatis 的这个 map 被重写了,同时写这两者的话,会抛出 ...already contains value for... 的异常

-- 代码位置 org.apache.ibatis.session.Configuration.StrictMap#put --
@Override
@SuppressWarnings("unchecked")
public V put(String key, V value) {
if (containsKey(key)) {
throw new IllegalArgumentException(name + " already contains value for " + key
+ (conflictMessageProducer == null ? "" : conflictMessageProducer.apply(super.get(key), value)));
}
if (key.contains(".")) {
final String shortKey = getShortName(key);
if (super.get(shortKey) == null) {
super.put(shortKey, value);
} else {
super.put(shortKey, (V) new Ambiguity(shortKey));
}
}
return super.put(key, value);
}

1.4 回到 MapperProxy

1.4.1 MapperProxy 的创建

刚才在1.1中我们提到了,mapperProxy,也就是刚才 org.apache.ibatis.binding.MapperRegistry#addMapper 的代码:knownMappers.put(type, new MapperProxyFactory<>(type));

看到 MapperProxyFactory 的内部:

-- 有删减 --
public class MapperProxyFactory<T> {

private final Class<T> mapperInterface;
private final Map<Method, MapperMethod> methodCache = new ConcurrentHashMap<>();

public MapperProxyFactory(Class<T> mapperInterface) {
this.mapperInterface = mapperInterface;
}

@SuppressWarnings("unchecked")
protected T newInstance(MapperProxy<T> mapperProxy) {
return (T) Proxy.newProxyInstance(mapperInterface.getClassLoader(), new Class[] { mapperInterface }, mapperProxy);
}

public T newInstance(SqlSession sqlSession) {
final MapperProxy<T> mapperProxy = new MapperProxy<>(sqlSession, mapperInterface, methodCache);
return newInstance(mapperProxy);
}
}

了解JDK动态代理的小伙伴应该很清楚了, newProxyInstance(ClassLoader loader, Class<?>[] interfaces, InvocationHandler h)意为,为接口创建一个实现了 InvocationHandler 的代理对象。我们在调用接口方法的时候,实际上要看代理类是如何实现的。

那么看看 mapperProxy 的内部的 invoke 是如何实现的,这里有三类方法,

  • 一种是一些 Object 对象带来的方法,这里不进行代理,直接 invoke

  • 一种是default方法,一种比较蛋疼的写法,把接口当抽象类写,里面可以放一个default方法写实现,这种代理了也没太大意义

  • 最后一种也就是我们准备代理的方法, 它会为每个非上面两者的方法,懒加载一个 MapperMethod 对象,并调用 MapperMethod#execute 来执行真正的 mybatis 逻辑。

1.4.2 MapperMethod 的创建

-- 有删减 --
public class MapperProxy<T> implements InvocationHandler, Serializable {

public MapperProxy(SqlSession sqlSession, Class<T> mapperInterface, Map<Method, MapperMethod> methodCache) {
this.sqlSession = sqlSession;
this.mapperInterface = mapperInterface;
this.methodCache = methodCache;
}

@Override
public Object invoke(Object proxy, Method method, Object[] args) throws Throwable {
try {
if (Object.class.equals(method.getDeclaringClass())) {// 来自 Object 的方法,比如 toString()
return method.invoke(this, args);
} else if (method.isDefault()) {// 静态方法,我们可以直接忽略
if (privateLookupInMethod == null) {
return invokeDefaultMethodJava8(proxy, method, args);
} else {
return invokeDefaultMethodJava9(proxy, method, args);
}
}
} catch (Throwable t) {
throw ExceptionUtil.unwrapThrowable(t);
}
final MapperMethod mapperMethod = cachedMapperMethod(method);
return mapperMethod.execute(sqlSession, args);
}

private MapperMethod cachedMapperMethod(Method method) {
return methodCache.computeIfAbsent(method,
k -> new MapperMethod(mapperInterface, method, sqlSession.getConfiguration()));
}
}

MapperMethod 的逻辑是怎么样的,也很好猜到,它的构造函数中创建了两个对象,

public class MapperMethod {

private final SqlCommand command;
private final MethodSignature method;

public MapperMethod(Class<?> mapperInterface, Method method, Configuration config) {
this.command = new SqlCommand(config, mapperInterface, method);
this.method = new MethodSignature(config, mapperInterface, method);
}
  • sqlCommand

sqlCommand 实际上就是从 configuration 里面把它对应的 MappedStatement 取出来,持有它的唯一 id 和执行类型。

  public static class SqlCommand {

private final String name;
private final SqlCommandType type;

public SqlCommand(Configuration configuration, Class<?> mapperInterface, Method method) {
final String methodName = method.getName();
final Class<?> declaringClass = method.getDeclaringClass();
MappedStatement ms = resolveMappedStatement(mapperInterface, methodName, declaringClass,
configuration);
if (ms == null) {
if (method.getAnnotation(Flush.class) != null) {
name = null;
type = SqlCommandType.FLUSH;
} else {
throw new BindingException("Invalid bound statement (not found): "
+ mapperInterface.getName() + "." + methodName);
}
} else {
name = ms.getId();
type = ms.getSqlCommandType();
if (type == SqlCommandType.UNKNOWN) {
throw new BindingException("Unknown execution method for: " + name);
}
}
}
  • MethodSignature MethodSignature 是针对接口返回值、参数等值的解析,比如我们的 @Param 注解,就是在 new ParamNameResolver(configuration, method); 里面解析的,比较简单,在之前的文章 简单概括的mybatis sqlSession 源码解析里也提到过,这里就不多说了。

    public MethodSignature(Configuration configuration, Class<?> mapperInterface, Method method) {
Type resolvedReturnType = TypeParameterResolver.resolveReturnType(method, mapperInterface);
if (resolvedReturnType instanceof Class<?>) {
this.returnType = (Class<?>) resolvedReturnType;
} else if (resolvedReturnType instanceof ParameterizedType) {
this.returnType = (Class<?>) ((ParameterizedType) resolvedReturnType).getRawType();
} else {
this.returnType = method.getReturnType();
}
this.returnsVoid = void.class.equals(this.returnType);
this.returnsMany = configuration.getObjectFactory().isCollection(this.returnType) || this.returnType.isArray();
this.returnsCursor = Cursor.class.equals(this.returnType);
this.returnsOptional = Optional.class.equals(this.returnType);
this.mapKey = getMapKey(method);
this.returnsMap = this.mapKey != null;
this.rowBoundsIndex = getUniqueParamIndex(method, RowBounds.class);
this.resultHandlerIndex = getUniqueParamIndex(method, ResultHandler.class);
this.paramNameResolver = new ParamNameResolver(configuration, method);
}

1.4.3 MapperMethod 的执行

mapperMethod 就是 sqlSession 与 mappedStatement 的一个整合。它的执行是一个策略模式:

  public Object execute(SqlSession sqlSession, Object[] args) {
Object result;
switch (command.getType()) {
case INSERT: {
Object param = method.convertArgsToSqlCommandParam(args);
result = rowCountResult(sqlSession.insert(command.getName(), param));
break;
}
case UPDATE: {
Object param = method.convertArgsToSqlCommandParam(args);
result = rowCountResult(sqlSession.update(command.getName(), param));
break;
}
case DELETE: {
Object param = method.convertArgsToSqlCommandParam(args);
result = rowCountResult(sqlSession.delete(command.getName(), param));
break;
}
case SELECT:
// 略..
}

具体是怎么执行的在文章 简单概括的mybatis sqlSession 源码解析 提到过,这里也不过多赘述。

这里对 MapperProxy 在初始化与调用过程中的关系做一下罗列:

从源码角度剖析 Spring 如何管理 mybatis 事务的

二、 下文序言

上面的 MapperProxy 讲解的比较粗略,因为真的很简单(复杂一点的在 MepperMethod 的策略模式,也就是调用 sqlSession去执行语句的时候,但是那个本文不会详细说明,后续的文章会解析这部分代码)

本文要讲的是几个在很多文章或者书里都没有提到,或者只是简单提了一下的点:本文将会把 sqlSession、MapperProxy、Spring事务管理几个关联密切的功能点进行总结,比如如下这样的疑问:

  • 1、我们知道一个 sqlSession 对应一个数据库连接,在创建 MapperProxy 的时候,又注入了 sqlSession ,难道我们用的一直是同一个 sqlSession?或者难道每次使用不同的数据库连接,会创建不同的 MapperProxy 代理?

  • 2、事务传播等级是怎么实现的,和 sqlSession 有关系吗?

  • 3、代理对象 MapperProxy 是如何和 spring 产生关联的?

三、 SqlSession 的初始化及其运作总览

为了避免有小伙伴对 sqlSession 完全没有概念,这里将接口代码贴出,可以看出 sqlSession 是执行语句的一个入口,同时也提供了事务的一些操作,实际上就是如此:

public interface SqlSession extends Closeable {
<T> T selectOne(String statement);
<T> T selectOne(String statement, Object parameter);
<E> List<E> selectList(String statement);
<E> List<E> selectList(String statement, Object parameter);
<E> List<E> selectList(String statement, Object parameter, RowBounds rowBounds);
<K, V> Map<K, V> selectMap(String statement, String mapKey);
<K, V> Map<K, V> selectMap(String statement, Object parameter, String mapKey);
<K, V> Map<K, V> selectMap(String statement, Object parameter, String mapKey, RowBounds rowBounds);
<T> Cursor<T> selectCursor(String statement);
<T> Cursor<T> selectCursor(String statement, Object parameter);
<T> Cursor<T> selectCursor(String statement, Object parameter, RowBounds rowBounds);
void select(String statement, Object parameter, ResultHandler handler);
void select(String statement, ResultHandler handler);
void select(String statement, Object parameter, RowBounds rowBounds, ResultHandler handler);
int insert(String statement);
int insert(String statement, Object parameter);
int update(String statement);
int update(String statement, Object parameter);
int delete(String statement);
int delete(String statement, Object parameter);
void commit();
void commit(boolean force);
void rollback();
void rollback(boolean force);
List<BatchResult> flushStatements();
void close();
void clearCache();
Configuration getConfiguration();
<T> T getMapper(Class<T> type);
Connection getConnection();
}

3.1 sqlSession 的创建

3.1.1 Environment 与 Transaction

首先忘掉 spring 为我们提供的便利,看一下基础的,脱离了 spring 托管的 mybatis 是怎么进行 sql 操作的:

        SqlSession sqlSession = sqlSessionFactory.openSession();
TrackerConfigMapper mapper = sqlSession.getMapper(TrackerConfigMapper.class);
TrackerConfigDO one = mapper.getOne(1);

SqlSessionFactory 有两个子类实现:DefaultSqlSessionFactory 和 SqlSessionManagerSqlSessionManager 使用动态代理 + 静态代理对 DefaultSqlSessionFactory 进行了代理,不过不用太在意这个 SqlSessionManager,后面会说明原因。

上面不管怎么代理,实际逻辑的执行者都是 DefaultSqlSessionFactory,我们看看它的创建方法,也就是 openSession() 实际执行的方法:

  private SqlSession openSessionFromDataSource(ExecutorType execType, TransactionIsolationLevel level, boolean autoCommit) {
Transaction tx = null;
try {
final Environment environment = configuration.getEnvironment();
final TransactionFactory transactionFactory = getTransactionFactoryFromEnvironment(environment);
tx = transactionFactory.newTransaction(environment.getDataSource(), level, autoCommit);
final Executor executor = configuration.newExecutor(tx, execType);
return new DefaultSqlSession(configuration, executor, autoCommit);
} catch (Exception e) {
closeTransaction(tx); // may have fetched a connection so lets call close()
throw ExceptionFactory.wrapException("Error opening session. Cause: " + e, e);
} finally {
ErrorContext.instance().reset();
}
}

environment 可用于数据源切换,那么提到数据源切换,就很容易想到了,连接的相关信息是这货维持的。所以看到我们的代码: tx = transactionFactory.newTransaction(environment.getDataSource(), level, autoCommit);, TransactionFactory 有三个实现,它们分别是 JdbcTransactionFactoryManagedTransactionFactory 和 SpringManagedTransactionFactory

JdbcTransactionFactory 和 ManagedTransactionFactory 最大的区别就在于 ManagedTransactionFactory 实现了空的 commit 与 rollback,源码中这样说道:付与容器来管理 transaction 的生命周期,这个博主不是特别熟悉,因为没这么用过,tomcat、jetty 等容器实现了对 jdbc 的代理。要注意,不管如何都是使用的 jdbc 这套接口规范进行数据库操作的。

/**
* {@link Transaction} that lets the container manage the full lifecycle of the transaction.
* Delays connection retrieval until getConnection() is called.
* Ignores all commit or rollback requests.
* By default, it closes the connection but can be configured not to do it.
*
* @author Clinton Begin
*
* @see ManagedTransactionFactory
*/

Transaction 是 mybatis 创建的一个对象,它实际上是对 jdbc connection 对象的一个封装:

-- 代码位于 org.apache.ibatis.transaction.jdbc.JdbcTransaction --

@Override
public Connection getConnection() throws SQLException {
if (connection == null) {
openConnection();
}
return connection;
}

@Override
public void commit() throws SQLException {
if (connection != null && !connection.getAutoCommit()) {
if (log.isDebugEnabled()) {
log.debug("Committing JDBC Connection [" + connection + "]");
}
connection.commit();
}
}

@Override
public void rollback() throws SQLException {
if (connection != null && !connection.getAutoCommit()) {
if (log.isDebugEnabled()) {
log.debug("Rolling back JDBC Connection [" + connection + "]");
}
connection.rollback();
}
}

3.1.2 Executor 与 SqlSession

我们知道 sqlSession 的 四大对象之一,Executor,负责统领全局,从语句获取(从 mappedStatement),到参数拼装(parameterHandler),再到执行语句(statementHandler),最后结果集封装(resultHandler),都是它负责“指挥”的。

我们看到它使用 Transaction 进行初始化,另外的一个参数是它的类型,这里不多说,REUSE 是带语句缓存的,和普通的 SimpleExecutor 没有特别大的区别,BATCH 类型则是通过 jdbc 提供的批量提交来对网络请求进行优化。

public enum ExecutorType {  SIMPLE, REUSE, BATCH}

最后将持有 Transaction 的 Executor 置入 SqlSession ,完成一个 SqlSession 对象的创建。

可以看到,我们的确是一个SqlSession 对应一个连接(Transaction),MapperProxy 这个业务接口的动态代理对象又持有一个 SqlSession 对象,那么总不可能一直用同一个连接吧?

当然有疑问是好的,而且通过对 SqlSession 初始化过程的剖析,我们已经完善了我们对 mybatis 的认知:

从源码角度剖析 Spring 如何管理 mybatis 事务的

接下来就是来打消这个疑问,MapperProxy 持有的 sqlSession 和 SqlSessionFactory 创建的这个有什么关系?

3.2 SqlSessionTemplate 对 sqlSession 的代理

实际上答案就在 SqlSessionTemplateSqlSessionTemplate 是 spring 对 mybatis SqlSessionFactory 的封装,同时,它还是 SqlSession 的代理。

SqlSessionTemplate 和 mybatis 提供的 SqlSessionManagerSqlSessionFactory 的另一个实现类,也是DefaultSqlSessionFactory 的代理类,可以细想一下,业务是否共用同一个 sqlSession 还要在业务里面去传递,去控制是不是很麻烦) 是一样的思路,不过 spring 直接代理了 sqlSession

-- 代码位于 org.mybatis.spring.SqlSessionTemplate --

private final SqlSessionFactory sqlSessionFactory;

private final ExecutorType executorType;

private final SqlSession sqlSessionProxy;

private final PersistenceExceptionTranslator exceptionTranslator;

/**
* Constructs a Spring managed {@code SqlSession} with the given
* {@code SqlSessionFactory} and {@code ExecutorType}.
* A custom {@code SQLExceptionTranslator} can be provided as an
* argument so any {@code PersistenceException} thrown by MyBatis
* can be custom translated to a {@code RuntimeException}
* The {@code SQLExceptionTranslator} can also be null and thus no
* exception translation will be done and MyBatis exceptions will be
* thrown
*
* @param sqlSessionFactory a factory of SqlSession
* @param executorType an executor type on session
* @param exceptionTranslator a translator of exception
*/

public SqlSessionTemplate(SqlSessionFactory sqlSessionFactory, ExecutorType executorType,
PersistenceExceptionTranslator exceptionTranslator)
{

notNull(sqlSessionFactory, "Property 'sqlSessionFactory' is required");
notNull(executorType, "Property 'executorType' is required");

this.sqlSessionFactory = sqlSessionFactory;
this.executorType = executorType;
this.exceptionTranslator = exceptionTranslator;
this.sqlSessionProxy = (SqlSession) newProxyInstance(
SqlSessionFactory.class.getClassLoader(),
new Class[]
{ SqlSession.class },
new SqlSessionInterceptor())
;
}

还是熟悉的配方,就是 jdk 的动态代理,SqlSessionTemplate 在初始化时创建了一个 SqlSession 代理,也内置了 ExecutorTypeSqlSessionFactory 等 defaultSqlSession 初始化的必要组件。

想必看到这里,已经有很多小伙伴知道这里是怎么回事了,是的,我们对 SqlSession 的操作都是经由这个代理来完成,代理的内部,实现了真正 SqlSession 的创建与销毁,回滚与提交等,我们先纵览以下它的代理实现。

3.2.1 sqlSession 常规代理流程赏析

对于这种jdk动态代理,我们看到 SqlSessionInterceptor#invoke 方法就明了了。我们先过一遍常规的流程,也就是没有使用 spring 事务功能支持,执行完 sql 就直接提交事务的常规操作:

  • 1、getSqlSession() 创建 sqlSession

  • 2、执行 MapperProxy,也就是前面讲了一大堆的,MapperProxy 中,通过 MapperMethod 来调用 sqlSession 和我们生成好的 mappedStatement 操作 sql 语句。

  • 3、提交事务

  • 4、关闭事务

注:代码有很大删减

 private class SqlSessionInterceptor implements InvocationHandler {
@Override
public Object invoke(Object proxy, Method method, Object[] args) throws Throwable {
SqlSession sqlSession = getSqlSession(
SqlSessionTemplate.this.sqlSessionFactory,
SqlSessionTemplate.this.executorType,
SqlSessionTemplate.this.exceptionTranslator); // 创建或者获取真正需要的 SqlSession
try {
Object result = method.invoke(sqlSession, args); // 执行原本想对 SqlSession 做的事情
if (!isSqlSessionTransactional(sqlSession, SqlSessionTemplate.this.sqlSessionFactory)) {
// force commit even on non-dirty sessions because some databases require
// a commit/rollback before calling close()
sqlSession.commit(true);// 如非 spring 管理事务,则直接提交
} finally {
if (sqlSession != null) {
closeSqlSession(sqlSession, SqlSessionTemplate.this.sqlSessionFactory);
}
}
}
}

注意:注释掉的代码在此类型的操作中没有什么意义,getSqlSession() 在这里只是简单通过 sessionFactory 创建了一个 sqlSession

  public static SqlSession getSqlSession(SqlSessionFactory sessionFactory, ExecutorType executorType, PersistenceExceptionTranslator exceptionTranslator) {
// SqlSessionHolder holder = (SqlSessionHolder) TransactionSynchronizationManager.getResource(sessionFactory);

// SqlSession session = sessionHolder(executorType, holder);
// if (session != null) {
// return session;
// }

LOGGER.debug(() -> "Creating a new SqlSession");
session = sessionFactory.openSession(executorType);
// registerSessionHolder(sessionFactory, executorType, exceptionTranslator, session);
return session;
}

3.2.2 sqlSession 借助 TransactionSynchronizationManager 代理流程赏析

看完前面的实现,有小伙伴会好奇,我的 @Transactional 注解呢?我的事务传播等级呢?

实际上,除去上述常规流程,更多的是要借助 TransactionSynchronizationManager 这个对象来完成,比如刚才步骤一,getSqlSession() 我暂时注释掉的代码里面,有一个很重要的操作:

我们把刚才 getSqlSession() 中注释掉的代码再拿回来看看:

    SqlSessionHolder holder = (SqlSessionHolder) TransactionSynchronizationManager.getResource(sessionFactory);

SqlSession session = sessionHolder(executorType, holder);
if (session != null) {
return session;
}

session = sessionFactory.openSession(executorType);
registerSessionHolder(sessionFactory, executorType, exceptionTranslator, session);
return session;

我们可以看到首先获取一个叫做 SqlSessionHolder 的东西,如果里面没有 sqlSession 则调用 sessionFactory.openSession(executorType); 创建一个,并把它注册到 TransactionSynchronizationManager

sqlSessionHolder 没什么可说的,它就只是个纯粹的容器,里面主要就是装着一个 SqlSession :

  public SqlSessionHolder(SqlSession sqlSession,
ExecutorType executorType,
PersistenceExceptionTranslator exceptionTranslator)
{

notNull(sqlSession, "SqlSession must not be null");
notNull(executorType, "ExecutorType must not be null");

this.sqlSession = sqlSession;
this.executorType = executorType;
this.exceptionTranslator = exceptionTranslator;
}

所以说我们只需要把目光焦距在 TransactionSynchronizationManager 就可以了,它的内部持有了很多个元素为 Map<Object, Object> 的 ThreadLocal(代码示例中只贴出了 resources 这一个 ThreadLocal ):

public abstract class TransactionSynchronizationManager {

private static final Log logger = LogFactory.getLog(TransactionSynchronizationManager.class);

private static final ThreadLocal<Map<Object, Object>> resources =
new NamedThreadLocal<>("Transactional resources");

@Nullable
public static Object getResource(Object key) {
Object actualKey = TransactionSynchronizationUtils.unwrapResourceIfNecessary(key);
Object value = doGetResource(actualKey);
if (value != null && logger.isTraceEnabled()) {
logger.trace("Retrieved value [" + value + "] for key [" + actualKey + "] bound to thread [" +
Thread.currentThread().getName() + "]");
}
return value;
}

@Nullable
private static Object doGetResource(Object actualKey) {
Map<Object, Object> map = resources.get();
if (map == null) {
return null;
}
Object value = map.get(actualKey);
// Transparently remove ResourceHolder that was marked as void...
if (value instanceof ResourceHolder && ((ResourceHolder) value).isVoid()) {
map.remove(actualKey);
// Remove entire ThreadLocal if empty...
if (map.isEmpty()) {
resources.remove();
}
value = null;
}
return value;
}

也就是说,spring 的事务,是借助 TransactionSynchronizationManager SqlSessionHolder 对 sqlSession 的控制来实现的。

那么这样就很清晰了,如下总结,也如下图:

  • MapperProxy 内置的 sqlSession 是 sqlSessiontemplate

  • sqlSessiontemplate 通过持有 SqlSessionFactory 来创建真正的 SqlSession

  • TransactionSynchronizationManager + SqlSessionHolder 则扮演着 SqlSession 管理的角色

从源码角度剖析 Spring 如何管理 mybatis 事务的

四、spring 如何管理 sqlSession

上一个小节只是讲了是什么,没有讲为什么,到了这里如果有好奇宝宝一定会好奇诸如 spring 的一系列事务控制是怎么实现的,当然本文不会讲太多 spring 事务管理相关的太多东西,以后会有后续文章专门剖析事务管理。

我们可以简单看下 TransactionInterceptor ,这是 @Transactional 注解的代理类。

/**
* AOP Alliance MethodInterceptor for declarative transaction
* management using the common Spring transaction infrastructure
* ({@link org.springframework.transaction.PlatformTransactionManager}/
* {@link org.springframework.transaction.ReactiveTransactionManager}).
*
* <p>Derives from the {@link TransactionAspectSupport} class which
* contains the integration with Spring's underlying transaction API.
* TransactionInterceptor simply calls the relevant superclass methods
* such as {@link #invokeWithinTransaction} in the correct order.
*
* <p>TransactionInterceptors are thread-safe.
*
* @author Rod Johnson
* @author Juergen Hoeller
* @see TransactionProxyFactoryBean
* @see org.springframework.aop.framework.ProxyFactoryBean
* @see org.springframework.aop.framework.ProxyFactory
*/

@SuppressWarnings("serial")
public class TransactionInterceptor extends TransactionAspectSupport implements MethodInterceptor, Serializable {

/**
* Create a new TransactionInterceptor.
* <p>Transaction manager and transaction attributes still need to be set.
* @see #setTransactionManager
* @see #setTransactionAttributes(java.util.Properties)
* @see #setTransactionAttributeSource(TransactionAttributeSource)
*/

public TransactionInterceptor() {
}

@Override
@Nullable
public Object invoke(MethodInvocation invocation) throws Throwable {
// Work out the target class: may be {@code null}.
// The TransactionAttributeSource should be passed the target class
// as well as the method, which may be from an interface.
Class<?> targetClass = (invocation.getThis() != null ? AopUtils.getTargetClass(invocation.getThis()) : null);

// Adapt to TransactionAspectSupport's invokeWithinTransaction...
return invokeWithinTransaction(invocation.getMethod(), targetClass, invocation::proceed);
}

可以看到它的代理方法 invoke() 的执行逻辑在 invokeWithinTransaction() 里:

--代码位于 org.springframework.transaction.interceptor.TransactionAspectSupport#invokeWithinTransaction --
@Nullable
protected Object invokeWithinTransaction(Method method, @Nullable Class<?> targetClass,
final InvocationCallback invocation)
throws Throwable
{

// If the transaction attribute is null, the method is non-transactional.
TransactionAttributeSource tas = getTransactionAttributeSource();
final TransactionAttribute txAttr = (tas != null ? tas.getTransactionAttribute(method, targetClass) : null);
final TransactionManager tm = determineTransactionManager(txAttr);

if (this.reactiveAdapterRegistry != null && tm instanceof ReactiveTransactionManager) {
// 响应式事务相关
}

PlatformTransactionManager ptm = asPlatformTransactionManager(tm);
final String joinpointIdentification = methodIdentification(method, targetClass, txAttr);

if (txAttr == null || !(ptm instanceof CallbackPreferringPlatformTransactionManager)) {
// Standard transaction demarcation with getTransaction and commit/rollback calls.
TransactionInfo txInfo = createTransactionIfNecessary(ptm, txAttr, joinpointIdentification);

Object retVal;
try {
// This is an around advice: Invoke the next interceptor in the chain.
// This will normally result in a target object being invoked.
retVal = invocation.proceedWithInvocation();
}
catch (Throwable ex) {
// target invocation exception
completeTransactionAfterThrowing(txInfo, ex);
throw ex;
}
finally {
cleanupTransactionInfo(txInfo);
}

if (vavrPresent && VavrDelegate.isVavrTry(retVal)) {
// Set rollback-only in case of Vavr failure matching our rollback rules...
TransactionStatus status = txInfo.getTransactionStatus();
if (status != null && txAttr != null) {
retVal = VavrDelegate.evaluateTryFailure(retVal, txAttr, status);
}
}

commitTransactionAfterReturning(txInfo);
return retVal;
}

else {
// CallbackPreferringPlatformTransactionManager 的处理逻辑
}
}

invokeWithinTransaction() 的代码虽然长,我们还是把它分段来看:

4.1 TransactionDefinition 与 TransactionManager 的创建

  • 第一部分,准备阶段

也就是这部分代码:

		// If the transaction attribute is null, the method is non-transactional.
TransactionAttributeSource tas = getTransactionAttributeSource();
final TransactionAttribute txAttr = (tas != null ? tas.getTransactionAttribute(method, targetClass) : null);
final TransactionManager tm = determineTransactionManager(txAttr);
PlatformTransactionManager ptm = asPlatformTransactionManager(tm);
final String joinpointIdentification = methodIdentification(method, targetClass, txAttr);

获取 TransactionAttribute(TransactionDefinition(底层接口),这里面装载了事务传播等级,隔离级别等属性。TransactionAttribute 的创建依据配置,或者我们的事务传播等级注解,对什么异常进行回滚等,后续会继续对它的应用做说明, PlatformTransactionManager 则是进行事务管理的主要操作者。

4.2 获取 TransactionInfo

  • 第二部分,事务开启或者获取与准备,也就是我们执行逻辑的第一行代码 createTransactionIfNecessary()(是不是和前面说到的 SqlSession的创建或者获取很像?)

我们可以看到 createTransactionIfNecessary() 的实现就做了两件事,其一是获取一个叫做 TransactionStatus 的东西,另外则是调用 prepareTransactionInfo(),获取一个 TransactionInfo

	// Standard transaction demarcation with getTransaction and commit/rollback calls.
TransactionInfo txInfo = createTransactionIfNecessary(ptm, txAttr, joinpointIdentification);

--代码位于 org.springframework.transaction.interceptor.TransactionAspectSupport#createTransactionIfNecessary --

protected TransactionInfo createTransactionIfNecessary(@Nullable PlatformTransactionManager tm,
@Nullable TransactionAttribute txAttr, final String joinpointIdentification)
{

TransactionStatus status = tm.getTransaction(txAttr);
return prepareTransactionInfo(tm, txAttr, joinpointIdentification, status);
}

先看看第一件事,也就是获取 TransactionStatus,它保存了事务的 savePoint ,是否新事物等。删减掉一些判断方法,代码如下:

	public final TransactionStatus getTransaction(@Nullable TransactionDefinition definition)
throws TransactionException
{

// Use defaults if no transaction definition given.
TransactionDefinition def = (definition != null ? definition : TransactionDefinition.withDefaults());

Object transaction = doGetTransaction();
boolean debugEnabled = logger.isDebugEnabled();

if (isExistingTransaction(transaction)) {
// Existing transaction found -> check propagation behavior to find out how to behave.
return handleExistingTransaction(def, transaction, debugEnabled);
}
if (def.getPropagationBehavior() == TransactionDefinition.PROPAGATION_REQUIRED ||
def.getPropagationBehavior() == TransactionDefinition.PROPAGATION_REQUIRES_NEW ||
def.getPropagationBehavior() == TransactionDefinition.PROPAGATION_NESTED) {
SuspendedResourcesHolder suspendedResources = suspend(null);
try {
boolean newSynchronization = (getTransactionSynchronization() != SYNCHRONIZATION_NEVER);
DefaultTransactionStatus status = newTransactionStatus(
def, transaction, true, newSynchronization, debugEnabled, suspendedResources);
doBegin(transaction, def);
prepareSynchronization(status, def);
return status;
}
catch (RuntimeException | Error ex) {
resume(null, suspendedResources);
throw ex;
}
}
else {
// Create "empty" transaction: no actual transaction, but potentially synchronization.
if (def.getIsolationLevel() != TransactionDefinition.ISOLATION_DEFAULT && logger.isWarnEnabled()) {
logger.warn("Custom isolation level specified but no actual transaction initiated; " +
"isolation level will effectively be ignored: " + def);
}
boolean newSynchronization = (getTransactionSynchronization() == SYNCHRONIZATION_ALWAYS);
return prepareTransactionStatus(def, null, true, newSynchronization, debugEnabled, null);
}
}

代码很长,但是不急,我们可以简单看出它分为两个部分:

  • 第一部分是获取事务 doGetTransaction()

  • 第二部分则是判断是否新事物,

    • 则 TransactionDefinition.PROPAGATION_REQUIREDTransactionDefinition.PROPAGATION_REQUIRES_NEWTransactionDefinition.PROPAGATION_NESTED 是一种逻辑

    • 其余是另一种逻辑,信息量有点大,但是慢慢来:

    • 如果不是新事物,则执行 handleExistingTransaction

    • 如果是新事物

4.2.1 doGetTransaction

	protected Object doGetTransaction() {
DataSourceTransactionObject txObject = new DataSourceTransactionObject();
txObject.setSavepointAllowed(isNestedTransactionAllowed());
ConnectionHolder conHolder =
(ConnectionHolder) TransactionSynchronizationManager.getResource(obtainDataSource());
txObject.setConnectionHolder(conHolder, false);
return txObject;
}

doGetTransaction 获取我们的事务对象,这里也使用了 TransactionSynchronizationManager(前面说到的 SqlSession 的管理类),事务对象会尝试获取本事务所使用的连接对象,这个和事务传播等级有关,先立个 flag。

我们可以看到这里面主要逻辑就是去获取 ConnectionHolder,实际上很简单,只要能获取到,就是已经存在的事务,获取不到(或者事务已经关闭)就是新事物。

4.2.2 新事物的处理之创建一个真正的事务对象

如果说前面无法从 TransactionSynchronizationManager 获取到 conHolder,或者说,我们的线程中并没有 ConnectionHolder那么将会进入此分支,此分支的支持的三个事务传播等级 TransactionDefinition.PROPAGATION_REQUIREDTransactionDefinition.PROPAGATION_REQUIRES_NEWTransactionDefinition.PROPAGATION_NESTED 都是需要创建新事务的,所以它们在同一个分支里面:

	SuspendedResourcesHolder suspendedResources = suspend(null);
boolean newSynchronization = (getTransactionSynchronization() != SYNCHRONIZATION_NEVER);
DefaultTransactionStatus status = newTransactionStatus(
def, transaction, true, newSynchronization, debugEnabled, suspendedResources);
doBegin(transaction, def);
prepareSynchronization(status, def);
return status;

SuspendedResourcesHolder 与事务的挂起相关,doBegin() 则是对连接对象 connection 的获取和配置,prepareSynchronization() 则是对新事物的一些初始化操作。我们一点点看:

	/**
* This implementation sets the isolation level but ignores the timeout.
*/

@Override
protected void doBegin(Object transaction, TransactionDefinition definition) {
DataSourceTransactionObject txObject = (DataSourceTransactionObject) transaction;
Connection con = null;

if (!txObject.hasConnectionHolder() ||
txObject.getConnectionHolder().isSynchronizedWithTransaction()) {
Connection newCon = obtainDataSource().getConnection();
if (logger.isDebugEnabled()) {
logger.debug("Acquired Connection [" + newCon + "] for JDBC transaction");
}
txObject.setConnectionHolder(new ConnectionHolder(newCon), true);
}

txObject.getConnectionHolder().setSynchronizedWithTransaction(true);
con = txObject.getConnectionHolder().getConnection();

Integer previousIsolationLevel = DataSourceUtils.prepareConnectionForTransaction(con, definition);
txObject.setPreviousIsolationLevel(previousIsolationLevel);
txObject.setReadOnly(definition.isReadOnly());

// Switch to manual commit if necessary. This is very expensive in some JDBC drivers,
// so we don't want to do it unnecessarily (for example if we've explicitly
// configured the connection pool to set it already).
if (con.getAutoCommit()) {
txObject.setMustRestoreAutoCommit(true);
if (logger.isDebugEnabled()) {
logger.debug("Switching JDBC Connection [" + con + "] to manual commit");
}
con.setAutoCommit(false);
}

prepareTransactionalConnection(con, definition);
txObject.getConnectionHolder().setTransactionActive(true);

// Bind the connection holder to the thread.
if (txObject.isNewConnectionHolder()) {
TransactionSynchronizationManager.bindResource(obtainDataSource(), txObject.getConnectionHolder());
}
}
}

可以看到,ConnectionHolder 的创建和连接的打开就是在这里进行的,创建后,设置其隔离级别,取消 connection 的自动提交,将提交操作纳入到 spring 管理,并且将其存到 TransactionSynchronizationManager 使得 4.2.1 提到的 doGetTransaction()可以拿到此 ConnectionHolder


做完连接的获取与配置后,下一步就是对事物的一些初始化:

	/**
* Initialize transaction synchronization as appropriate.
*/

protected void prepareSynchronization(DefaultTransactionStatus status, TransactionDefinition definition) {
if (status.isNewSynchronization()) {
TransactionSynchronizationManager.setActualTransactionActive(status.hasTransaction());
TransactionSynchronizationManager.setCurrentTransactionIsolationLevel(
definition.getIsolationLevel() != TransactionDefinition.ISOLATION_DEFAULT ?
definition.getIsolationLevel() : null);
TransactionSynchronizationManager.setCurrentTransactionReadOnly(definition.isReadOnly());
TransactionSynchronizationManager.setCurrentTransactionName(definition.getName());
TransactionSynchronizationManager.initSynchronization();
}
}

这个代码都是代码字面意义的简单设置,就不赘述了。

4.2.3 新事物的处理之创建一个虚假的事务对象

刚才讲的是 “无法从 TransactionSynchronizationManager 获取到 conHolder”,并且属于一些需要创建新事物的传播等级的情况。

如果说方才没有事务,也不需要创建新的事务,则会进入此分支,创建一个空的 TransactionStatus,内部的事务对象为空,代码很简单就不贴了,有兴趣可以去看看 org.springframework.transaction.support.AbstractPlatformTransactionManager#getTransaction 的最后一个分支。

4.2.4 事务嵌套

刚才说的都是无法获取到 conHolder 的情况,如果获取到了,则又是另一套代码了,handleExistingTransaction 很长,它的第一个部分是对传播等级的控制,有兴趣的小伙伴可以去看看源码,我这里只挑一个简单的传播等级 PROPAGATION_NESTED_NEW做说明(其他的会在专门的事务一期做讲解):

-- 代码位于 org.springframework.transaction.support.AbstractPlatformTransactionManager#handleExistingTransaction --
private TransactionStatus handleExistingTransaction(
TransactionDefinition definition, Object transaction, boolean debugEnabled)

throws TransactionException
{

if (definition.getPropagationBehavior() == TransactionDefinition.PROPAGATION_REQUIRES_NEW) {
if (debugEnabled) {
logger.debug("Suspending current transaction, creating new transaction with name [" +
definition.getName() + "]");
}
SuspendedResourcesHolder suspendedResources = suspend(transaction);
try {
boolean newSynchronization = (getTransactionSynchronization() != SYNCHRONIZATION_NEVER);
DefaultTransactionStatus status = newTransactionStatus(
definition, transaction, true, newSynchronization, debugEnabled, suspendedResources);
doBegin(transaction, definition);
prepareSynchronization(status, definition);
return status;
}
catch (RuntimeException | Error beginEx) {
resumeAfterBeginException(transaction, suspendedResources, beginEx);
throw beginEx;
}
}

... 略
}

我们可以发现和 4.2.2 新事物的处理 代码是一样的,唯一的区别就是此 TransactionStatus 对象会真正内嵌一个事务挂起对象 SuspendedResourcesHolder 

4.3 封装 TransactionInfo

拿到 TransactionStatus 之后, prepareTransactionInfo() 里简单的将刚才那些 PlatformTransactionManager TransactionAttributeTransactionStatus 包装成一个 TransactionInfo 对象,并将其保存在 ThreadLocal 中,这个 bindToThread() 还会将当前已经持有的 TransactionInfo 对象暂存。

protected TransactionInfo prepareTransactionInfo(@Nullable PlatformTransactionManager tm,
@Nullable TransactionAttribute txAttr, String joinpointIdentification,
@Nullable TransactionStatus status)
{

TransactionInfo txInfo = new TransactionInfo(tm, txAttr, joinpointIdentification);
if (txAttr != null) {
// The transaction manager will flag an error if an incompatible tx already exists.
txInfo.newTransactionStatus(status);
}

// We always bind the TransactionInfo to the thread, even if we didn't create
// a new transaction here. This guarantees that the TransactionInfo stack
// will be managed correctly even if no transaction was created by this aspect.
txInfo.bindToThread();
return txInfo;
}

到这里思路就很清晰了,代理为我们做的事情就是生成了一个叫做 TransactionInfo 的东西,里面的 TransactionManager 可以使得 spring 去对最底层的 connection 对象做一些回滚,提交操作。TransactionStatus 则保存挂起的事务的信息,以及当前事务的一些状态,如下图:

4.4 纵览流程

让我们回到第四节开头的那段很长的代码,到这里是不是很明了了:

	protected Object invokeWithinTransaction(Method method, @Nullable Class<?> targetClass,
final InvocationCallback invocation)
throws Throwable
{

// If the transaction attribute is null, the method is non-transactional.
TransactionAttributeSource tas = getTransactionAttributeSource();
final TransactionAttribute txAttr = (tas != null ? tas.getTransactionAttribute(method, targetClass) : null);
final TransactionManager tm = determineTransactionManager(txAttr);
PlatformTransactionManager ptm = asPlatformTransactionManager(tm);
final String joinpointIdentification = methodIdentification(method, targetClass, txAttr);

if (txAttr == null || !(ptm instanceof CallbackPreferringPlatformTransactionManager)) {
// Standard transaction demarcation with getTransaction and commit/rollback calls.
TransactionInfo txInfo = createTransactionIfNecessary(ptm, txAttr, joinpointIdentification);

Object retVal;
try {
// This is an around advice: Invoke the next interceptor in the chain.
// This will normally result in a target object being invoked.
retVal = invocation.proceedWithInvocation();
}
catch (Throwable ex) {
// target invocation exception
completeTransactionAfterThrowing(txInfo, ex);
throw ex;
}
finally {
cleanupTransactionInfo(txInfo);
}

if (vavrPresent && VavrDelegate.isVavrTry(retVal)) {
// Set rollback-only in case of Vavr failure matching our rollback rules...
TransactionStatus status = txInfo.getTransactionStatus();
if (status != null && txAttr != null) {
retVal = VavrDelegate.evaluateTryFailure(retVal, txAttr, status);
}
}

commitTransactionAfterReturning(txInfo);
return retVal;
}
}
  • 1、获取 TransactionInfo

  • 2、执行切面

  • 3、将之前挂起的 TransactionInfo 找回:

		private void bindToThread() {
// Expose current TransactionStatus, preserving any existing TransactionStatus
// for restoration after this transaction is complete.
this.oldTransactionInfo = transactionInfoHolder.get();
transactionInfoHolder.set(this);
}

private void restoreThreadLocalStatus() {
// Use stack to restore old transaction TransactionInfo.
// Will be null if none was set.
transactionInfoHolder.set(this.oldTransactionInfo);
}
  • 4、如果需要,则提交当前事务

  • 5、返回切面值

4.5 最后一块拼图,spring 如何与 sqlSession 产生关联:

我们在第三章讲到,mybatis有一个叫做 defualtSqlSessionFactory 的类,负责创建 sqlSession,但是它和 spring 又是怎么产生关联的呢?

答案就在于,spring 实现了自己的 TransactionFactory,以及自己的 Transaction 对象 SpringManagedTransaction 。回顾一下 SqlSession 的创建过程:

  private SqlSession openSessionFromDataSource(ExecutorType execType, TransactionIsolationLevel level, boolean autoCommit) {
Transaction tx = null;
try {
final Environment environment = configuration.getEnvironment();
final TransactionFactory transactionFactory = getTransactionFactoryFromEnvironment(environment);
tx = transactionFactory.newTransaction(environment.getDataSource(), level, autoCommit);
final Executor executor = configuration.newExecutor(tx, execType);
return new DefaultSqlSession(configuration, executor, autoCommit);
} catch (Exception e) {
closeTransaction(tx); // may have fetched a connection so lets call close()
throw ExceptionFactory.wrapException("Error opening session. Cause: " + e, e);
} finally {
ErrorContext.instance().reset();
}
}

看一下 SpringManagedTransaction 是如何管理 connection的:

  private void openConnection() throws SQLException {
this.connection = DataSourceUtils.getConnection(this.dataSource);
this.autoCommit = this.connection.getAutoCommit();
this.isConnectionTransactional = DataSourceUtils.isConnectionTransactional(this.connection, this.dataSource);

LOGGER.debug(() -> "JDBC Connection [" + this.connection + "] will"
+ (this.isConnectionTransactional ? " " : " not ") + "be managed by Spring");
}

DataSourceUtils.getConnection(this.dataSource); 划重点,里面的实现不用我多说了,我们可以看到熟悉的身影,也就是 ConnectionHolder,连接是从这里(优先)拿的:

		ConnectionHolder conHolder = (ConnectionHolder) TransactionSynchronizationManager.getResource(dataSource);
if (conHolder != null && (conHolder.hasConnection() || conHolder.isSynchronizedWithTransaction())) {
conHolder.requested();
if (!conHolder.hasConnection()) {
logger.debug("Fetching resumed JDBC Connection from DataSource");
conHolder.setConnection(fetchConnection(dataSource));
}
return conHolder.getConnection();
}

更新整套体系图:

我们整体简单过一次:

  • mybatis 启动时根据xml、注解创建了 mapperedStatement,用于sql执行,创建了 SqlSessionFactory 用于创建 SqlSession对象。

  • mybatis 启动时创建了 MapperProxyFactory 用于创建接口的代理对象 MapperProxy

  • 在创建 MapperProxy 时,spring 为其注入了一个 sqlSession 用于 sql执行,但是这个 sqlSession 是一个代理对象,叫做 sqlSessionTemplate,它会自动选择我们该使用哪个 sqlSession 去执行

  • 在执行时,spring 切面在执行事务之前,会创建一个叫做 TransactionInfo 的对象,此对象会根据事务传播等级来控制是否创建新连接,是否挂起上一个连接,将信息保存在 TransactionSynchronizationManager

  • 到了真正需要创建或者获取 sqlSession 时,spring 重写的 TransactionFactory 会优先去 TransactionSynchronizationManager 中拿连接对象。


参考资料:
mybatis-spring-boot-starter 2.1.1
mybatis 3.5.3