MyBatis核心源码深度剖析SQL语句执行过程

1 SQL语句的执行过程介绍

MyBatis核心执行组件：

2 SQL执行的入口分析

2.1 为Mapper接口创建代理对象

// 方式1：
User user = session.selectOne("com.oldlu.dao.UserMapper.findUserById", 101);
// 方式2：
UserMapper mapper = session.getMapper(UserMapper.class);
List<User> userList = mapper.findAll();

2.2 执行代理逻辑

方式1入口分析：
session是DefaultSqlSession类型的，因为sqlSessionFactory默认生成的SqlSession是
DefaultSqlSession类型。
selectOne()会调用selectList()。

// DefaultSqlSession类
public <E> List<E> selectList(String statement, Object parameter, RowBounds
rowBounds) {
 try {
   MappedStatement ms = configuration.getMappedStatement(statement);
   // CURD操作是交给Excetor去处理的
   return executor.query(ms, wrapCollection(parameter), rowBounds,
Executor.NO_RESULT_HANDLER);
} catch (Exception e) {
   throw ExceptionFactory.wrapException("Error querying database. Cause: "
+ e, e);
} finally {
   ErrorContext.instance().reset();
}
}

方式2入口分析：
获取代理对象：

//DefaultSqlSession类 ====================>
@Override
public <T> T getMapper(Class<T> type) {
 return configuration.getMapper(type, this);
 }
// Configuration类 ====================>
public <T> T getMapper(Class<T> type, SqlSession sqlSession) {
 return mapperRegistry.getMapper(type, sqlSession);
}
//MapperRegistry ----> apperProxyFactory.newInstance ====================>
public <T> T getMapper(Class<T> type, SqlSession sqlSession) {
 //从缓存中获取该Mapper接口的代理工厂对象
 final MapperProxyFactory<T> mapperProxyFactory = (MapperProxyFactory<T>)
knownMappers.get(type);
 //如果该Mapper接口没有注册过，则抛异常
 if (mapperProxyFactory == null) {
   throw new BindingException("Type " + type + " is not known to the
MapperRegistry.");
}
 try {
   //【使用代理工厂创建Mapper接口的代理对象】
   return mapperProxyFactory.newInstance(sqlSession);
} catch (Exception e) {
   throw new BindingException("Error getting mapper instance. Cause: " + e,
e);
}
}
//MapperProxyFactory  --->此时生成代理对象 ====================>
protected T newInstance(MapperProxy<T> mapperProxy) {
 //Mybatis底层是调用JDK的Proxy类来创建代理实例
 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);
}

代理对象执行逻辑：

//MapperProxy   ====================>
/**代理对象执行的方法，代理以后，所有Mapper的方法调用时，都会调用这个invoke方法*/
public Object invoke(Object proxy, Method method, Object[] args) throws
Throwable {
try {
 if (Object.class.equals(method.getDeclaringClass())) {
  //如果是Object方法，则调用方法本身
  return method.invoke(this, args);
} else {
  //调用接口方法：根据被调用接口的Method对象，从缓存中获取MapperMethodInvoker对象
  //apper接口中的每一个方法都对应一个MapperMethodInvoker对象，而MapperMethodInvoker
对象里面的MapperMethod保存着对应的SQL信息和返回类型以完成SQL调用 ...
  return cachedInvoker(method).invoke(proxy, method, args, sqlSession);
}
} catch (Throwable t) {
 throw ExceptionUtil.unwrapThrowable(t);
}
}
/**
获取缓存中MapperMethodInvoker，如果没有则创建一个，而MapperMethodInvoker内部封装这一
个MethodHandler
*/
private MapperMethodInvoker cachedInvoker(Method method) throws Throwable {
 try {
   return methodCache.computeIfAbsent(method, m -> {
     if (m.isDefault()) {
       //如果调用接口的是默认方法（default方法）
       try {
         if (privateLookupInMethod == null) {
           return new
DefaultMethodInvoker(getMethodHandleJava8(method));
        } else {
           return new
DefaultMethodInvoker(getMethodHandleJava9(method));
        }
      } catch (IllegalAccessException | InstantiationException |
InvocationTargetException
           | NoSuchMethodException e) {
         throw new RuntimeException(e);
      }
    } else {
       //如果调用的普通方法（非default方法），则创建一个PlainMethodInvoker并放
入缓存，其中MapperMethod保存对应接口方法的SQL以及入参和出参的数据类型等信息
       return new PlainMethodInvoker(new MapperMethod(mapperInterface,
method, sqlSession.getConfiguration()));
    }
  });
} catch (RuntimeException re) {
   Throwable cause = re.getCause();
throw cause == null ? re : cause;
}
}
// MapperProxy内部类： PainMethodInvoker  ====================>
// 当cacheInvoker返回了PalinMethodInvoker实例之后，紧接着调用了这个实例的
PlainMethodInvoker:invoke方法
@Override
public Object invoke(Object proxy, Method method, Object[] args, SqlSession
sqlSession) throws Throwable {
//Mybatis实现接口方法的核心: MapperMethod::execute方法：
return mapperMethod.execute(sqlSession, args);
}
// MapperMethod  ====================>
public Object execute(SqlSession sqlSession, Object[] args) {
 Object result;
 switch (command.getType()) {
   case INSERT: {
     // 将args进行解析，如果是多个参数则，则根据@Param注解指定名称将参数转换为Map，
如果是封装实体则不转换
     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:
     //查询操作
     if (method.returnsVoid() && method.hasResultHandler()) {
       executeWithResultHandler(sqlSession, args);
       result = null;
    } else if (method.returnsMany()) {
       result = executeForMany(sqlSession, args);
    } else if (method.returnsMap()) {
result = executeForMap(sqlSession, args);
    } else if (method.returnsCursor()) {
       result = executeForCursor(sqlSession, args);
    } else {
       //解析参数，因为SqlSession::selectOne方法参数只能传入一个，但是我们
Mapper中可能传入多个参数，
       //有可能是通过@Param注解指定参数名，所以这里需要将Mapper接口方法中的多个参
数转化为一个ParamMap,
       //也就是说如果是传入的单个封装实体，那么直接返回出来；如果传入的是多个参数，
实际上都转换成了Map
       Object param = method.convertArgsToSqlCommandParam(args);
       //可以看到 * 最后还是使用SqlSession操作数据库的
       result = sqlSession.selectOne(command.getName(), param);
       if (method.returnsOptional()
         && (result == null ||
!method.getReturnType().equals(result.getClass()))) {
         result = Optional.ofNullable(result);
      }
    }
     break;
   case FLUSH:
     result = sqlSession.flushStatements();
     break;
   default:
     throw new BindingException("Unknown execution method for: " +
command.getName());
}
 if (result == null && method.getReturnType().isPrimitive() &&
!method.returnsVoid()) {
   throw new BindingException("Mapper method '" + command.getName()
                + " attempted to return null from a method
with a primitive return type (" + method.getReturnType() + ").");
}
 return result;
}
// 此时我们发现： 回到了sqlsession中
private <E> Object executeForMany(SqlSession sqlSession, Object[] args) {
 List<E> result;
 Object param = method.convertArgsToSqlCommandParam(args);
 if (method.hasRowBounds()) {
  RowBounds rowBounds = method.extractRowBounds(args);
  result = sqlSession.selectList(command.getName(), param, rowBounds);
} else {
  result = sqlSession.selectList(command.getName(), param);
}
// ...
 return result;
}

3 查询语句的执行过程分析

3.1 selectOne方法分析

// DefaultSqlSession类  ===============>
// selectOne
@Override
public <T> T selectOne(String statement, Object parameter) {
 // //selectOne()会调用selectList()。
 List<T> list = this.selectList(statement, parameter);
 if (list.size() == 1) {
   return list.get(0);
} else if (list.size() > 1) {
   throw new TooManyResultsException("Expected one result (or null) to be
returned by selectOne(), but found: " + list.size());
} else {
   return null;
}
}
// selectList
public <E> List<E> selectList(String statement, Object parameter, RowBounds
rowBounds) {
 try {
   MappedStatement ms = configuration.getMappedStatement(statement);
   // CURD操作是交给Excetor去处理的
   return executor.query(ms, wrapCollection(parameter), rowBounds,
Executor.NO_RESULT_HANDLER);
} catch (Exception e) {
   throw ExceptionFactory.wrapException("Error querying database. Cause: "
+ e, e);
} finally {
   ErrorContext.instance().reset();
}
}

3.2 sql获取

// CachingExecutor ===============>
public <E> List<E> query(MappedStatement ms, Object parameterObject, RowBounds
rowBounds, ResultHandler resultHandler) throws SQLException {
 // 获取绑定的sql命令，比如"SELECT * FROM xxx"
 BoundSql boundSql = ms.getBoundSql(parameterObject);
 CacheKey key = createCacheKey(ms, parameterObject, rowBounds, boundSql);
 return query(ms, parameterObject, rowBounds, resultHandler, key, boundSql);
}
@Override
public <E> List<E> query(MappedStatement ms, Object parameterObject, RowBounds
rowBounds, ResultHandler resultHandler, CacheKey key, BoundSql boundSql)
 throws SQLException {
 Cache cache = ms.getCache();
 if (cache != null) {
   flushCacheIfRequired(ms);
   if (ms.isUseCache() && resultHandler == null) {
     ensureNoOutParams(ms, boundSql);
     @SuppressWarnings("unchecked")
     List<E> list = (List<E>) tcm.getObject(cache, key);
     if (list == null) {
       list = delegate.query(ms, parameterObject, rowBounds,
resultHandler, key, boundSql);
       tcm.putObject(cache, key, list); // issue #578 and #116
   }
     return list;
  }
}
 return delegate.query(ms, parameterObject, rowBounds, resultHandler, key,
boundSql);
}
//真正执行query操作的是SimplyExecutor代理来完成的，SimplyExecutor的父类BaseExecutor的
query方法中:
// BaseExecutor类:SimplyExecutor的父类 =================>
@Override
public <E> List<E> query(MappedStatement ms, Object parameter, RowBounds
rowBounds, ResultHandler resultHandler, CacheKey key, BoundSql boundSql) throws
SQLException {
 ErrorContext.instance().resource(ms.getResource()).activity("executing a
query").object(ms.getId());
 if (closed) {
   throw new ExecutorException("Executor was closed.");
}
 if (queryStack == 0 && ms.isFlushCacheRequired()) {
   clearLocalCache();
}
 List<E> list;
 try {
   queryStack++;
   //localCache是一级缓存，如果找不到就调用queryFromDatabase从数据库中查找
   list = resultHandler == null ? (List<E>) localCache.getObject(key) :
null;
   if (list != null) {
     handleLocallyCachedOutputParameters(ms, key, parameter, boundSql);
  } else {
     list = queryFromDatabase(ms, parameter, rowBounds, resultHandler,
key, boundSql);
  }
} finally {
   queryStack--;
}
 if (queryStack == 0) {
   for (DeferredLoad deferredLoad : deferredLoads) {
     deferredLoad.load();
  }
   deferredLoads.clear();
   if (configuration.getLocalCacheScope() == LocalCacheScope.STATEMENT) {
     clearLocalCache();
  }
}
 return list;
}
//第一次，没有缓存，所以会调用queryFromDatabase方法来执行查询。
private <E> List<E> queryFromDatabase(...) throws SQLException {
 List<E> list;
 localCache.putObject(key, EXECUTION_PLACEHOLDER);
 try {
   // 查询
   list = doQuery(ms, parameter, rowBounds, resultHandler, boundSql);
} finally {
   localCache.removeObject(key);
}
 localCache.putObject(key, list);
 if (ms.getStatementType() == StatementType.CALLABLE) {
   localOutputParameterCache.putObject(key, parameter);
}
 return list;
}
// SimpleExecutor类 ============================>
public <E> List<E> doQuery(...) throws SQLException {
 Statement stmt = null;
 try {
   Configuration configuration = ms.getConfiguration();
   StatementHandler handler = configuration.newStatementHandler(....);
   // 1：SQL查询参数的设置
   stmt = prepareStatement(handler, ms.getStatementLog());
   // StatementHandler封装了Statement
   // 2：SQL查询操作和结果集的封装
   return handler.<E>query(stmt);
} finally {
   closeStatement(stmt);
}
}

3.3 参数设置

// SimplyExecutor类 ============================>
// 【1】 参数设置： prepareStatement
private Statement prepareStatement(StatementHandler handler, Log statementLog)
throws SQLException {
 Statement stmt;
 // 通过getConnection方法来获取一个Connection，
 Connection connection = getConnection(statementLog);
 // 调用prepare方法来获取一个Statement
 stmt = handler.prepare(connection, transaction.getTimeout());

// 设置SQL查询中的参数值 ***
 handler.parameterize(stmt);
 return stmt;
}
// RoutingStatementHandler ============================>
// PreparedStatementHandler ============================>
@Override
public void parameterize(Statement statement) throws SQLException {
 parameterHandler.setParameters((PreparedStatement) statement);
}
// DefaultParameterHandler ============================> 此时参数设置成功
@Override
public void setParameters(PreparedStatement ps) {
 ErrorContext.instance().activity("setting
parameters").object(mappedStatement.getParameterMap().getId());
 List<ParameterMapping> parameterMappings = boundSql.getParameterMappings();
 if (parameterMappings != null) {
   for (int i = 0; i < parameterMappings.size(); i++) {
     ParameterMapping parameterMapping = parameterMappings.get(i);
     if (parameterMapping.getMode() != ParameterMode.OUT) {
       Object value;
       String propertyName = parameterMapping.getProperty();
       if (boundSql.hasAdditionalParameter(propertyName)) {
         value = boundSql.getAdditionalParameter(propertyName);
      } else if (parameterObject == null) {
         value = null;
      } else if
(typeHandlerRegistry.hasTypeHandler(parameterObject.getClass())) {
         value = parameterObject;
      } else {
         MetaObject metaObject =
configuration.newMetaObject(parameterObject);
         value = metaObject.getValue(propertyName);
      }
       TypeHandler typeHandler = parameterMapping.getTypeHandler();
       JdbcType jdbcType = parameterMapping.getJdbcType();
       if (value == null && jdbcType == null) {
         jdbcType = configuration.getJdbcTypeForNull();
      }
     try {
         typeHandler.setParameter(ps, i + 1, value, jdbcType);
      } catch (TypeException | SQLException e) {
         throw new TypeException("Could not set parameters for
mapping.....");
      }
    }
  }
}
}

3.4 SQL执行和结果集的封装

// RoutingStatementHandler ============================>
@Override
public <E> List<E> query(Statement statement) throws SQLException {
 return delegate.<E>query(statement);
}
// PreparedStatementHandler ============================>
@Override
public <E> List<E> query(Statement statement, ResultHandler resultHandler)
throws SQLException {
 // 这里就到了熟悉的PreparedStatement了
 PreparedStatement ps = (PreparedStatement) statement;
 // 执行SQL查询操作
 ps.execute();
 // 结果交给ResultHandler来处理
 return resultSetHandler.<E> handleResultSets(ps);
}
// DefaultResultSetHandler类（封装返回值，将查询结果封装成Object对象）
@Override
public List<Object> handleResultSets(Statement stmt) throws SQLException {
 ErrorContext.instance().activity("handling
results").object(mappedStatement.getId());
 final List<Object> multipleResults = new ArrayList<Object>();
 int resultSetCount = 0;
 ResultSetWrapper rsw = getFirstResultSet(stmt);
 List<ResultMap> resultMaps = mappedStatement.getResultMaps();
int resultMapCount = resultMaps.size();
 validateResultMapsCount(rsw, resultMapCount);
 while (rsw != null && resultMapCount > resultSetCount) {
   ResultMap resultMap = resultMaps.get(resultSetCount);
   handleResultSet(rsw, resultMap, multipleResults, null);
   rsw = getNextResultSet(stmt);
   cleanUpAfterHandlingResultSet();
   resultSetCount++;
}
 String[] resultSets = mappedStatement.getResultSets();
 if (resultSets != null) {
   while (rsw != null && resultSetCount < resultSets.length) {
     ResultMapping parentMapping =
nextResultMaps.get(resultSets[resultSetCount]);
     if (parentMapping != null) {
       String nestedResultMapId = parentMapping.getNestedResultMapId();
       ResultMap resultMap =
configuration.getResultMap(nestedResultMapId);
       handleResultSet(rsw, resultMap, null, parentMapping);
    }
     rsw = getNextResultSet(stmt);
     cleanUpAfterHandlingResultSet();
     resultSetCount++;
  }
}
 return collapseSingleResultList(multipleResults);
}

4 更新语句的执行过程分析

• xecutor 的 update 方法分析

• insert、update 和 delete 操作都会清空一二级缓存

• doUpdate 方法

• PreparedStatementHandler 的 update 方法

• 默认是创建PreparedStatementHandler，然后执行prepareStatement方法。

• 执行结果为受影响行数

• 执行更新语句的SQL

4.1 sqlsession增删改方法分析

// DefaultSqlSession ===============>
@Override
public int insert(...) {
 return update(statement, parameter);
}
@Override
public int update(String statement) {
 return update(statement, null);
}
@Override
 public int delete(...) {
 return update(....);
}
// insert 、delete操作是通过调用update语句进行的相关逻辑
@Override
public int update(String statement, Object parameter) {
 try {
  dirty = true;
  MappedStatement ms = configuration.getMappedStatement(statement);
  // 增删改 最终底层都是 update
  return executor.update(ms, wrapCollection(parameter));

} catch (Exception e) {
  throw ExceptionFactory.wrapException("Error updating database. Cause: " +
e, e);
} finally {
  ErrorContext.instance().reset();
}
}

4.2 sql获取

// CachingExecutor  ===============>
@Override
public int update(MappedStatement ms, Object parameterObject) throws
SQLException {
 // 执行增删改，清除缓存
 flushCacheIfRequired(ms);
 // 跳转BaseExecutor
 return delegate.update(ms, parameterObject);
}
// BaseExecutor   ===============>
@Override
public int update(MappedStatement ms, Object parameter) throws SQLException {
 ErrorContext.instance().resource(ms.getResource()).activity("executing an
update").object(ms.getId());
 if (closed) {
   throw new ExecutorException("Executor was closed.");
}
 // 清除 LocalCache 一级缓存
 clearLocalCache();
 //执行 doUpdate
 return doUpdate(ms, parameter);
}
// SimpleExecutor  ===============>
// doUpdate
@Override
public int doUpdate(MappedStatement ms, Object parameter) throws SQLException {
 Statement stmt = null;
 try {
   Configuration configuration = ms.getConfiguration();
   StatementHandler handler = configuration.newStatementHandler(...);
   // 【1】.获取statement,并进行参数映射
   stmt = prepareStatement(handler, ms.getStatementLog());
   // 【2】.handler.update()方法执行具体sql指令
   return handler.update(stmt);
} finally {
   closeStatement(stmt);
}
}

4.3 参数设置

// SimplyExecutor类 ============================>
//【1】 prepareStatement
private Statement prepareStatement(StatementHandler handler, Log statementLog)
throws SQLException {
 Statement stmt;
 Connection connection = getConnection(statementLog);
 // 使用connection对象信息创建statement，并将超时时间绑定
 stmt = handler.prepare(connection, transaction.getTimeout());
 // parameterize方法设置sql执行时候需要的参数
 handler.parameterize(stmt);
 return stmt;
}
// RoutingStatementHandler ============================>
// PreparedStatementHandler ============================>
@Override
public void parameterize(Statement statement) throws SQLException {
 parameterHandler.setParameters((PreparedStatement) statement);
}
// DefaultParameterHandler ============================> 此时参数设置成功
@Override
public void setParameters(PreparedStatement ps) {
 ErrorContext.instance().activity("setting
parameters").object(mappedStatement.getParameterMap().getId());
 List<ParameterMapping> parameterMappings = boundSql.getParameterMappings();
 if (parameterMappings != null) {
   for (int i = 0; i < parameterMappings.size(); i++) {
     ParameterMapping parameterMapping = parameterMappings.get(i);
     if (parameterMapping.getMode() != ParameterMode.OUT) {
       Object value;
       String propertyName = parameterMapping.getProperty();
       if (boundSql.hasAdditionalParameter(propertyName)) {
         value = boundSql.getAdditionalParameter(propertyName);
      } else if (parameterObject == null) {
         value = null;
      } else if
(typeHandlerRegistry.hasTypeHandler(parameterObject.getClass())) {
         value = parameterObject;
      } else {
         MetaObject metaObject =
configuration.newMetaObject(parameterObject);
         value = metaObject.getValue(propertyName);
      }
       TypeHandler typeHandler = parameterMapping.getTypeHandler();
       JdbcType jdbcType = parameterMapping.getJdbcType();
       if (value == null && jdbcType == null) {
         jdbcType = configuration.getJdbcTypeForNull();
      }
       try {
         typeHandler.setParameter(ps, i + 1, value, jdbcType);
      } catch (TypeException | SQLException e) {
         throw new TypeException("Could not set parameters for
mapping.....");
      }
    }
  }
}
}

4.4 SQL执行

// RoutingStatementHandler ============================>
@Override
public int update(Statement statement) throws SQLException {
 return delegate.update(statement);
}
// PreparedStatementHandler ============================>
@Override
public int update(Statement statement) throws SQLException {
 // 这里就是底层JDBC的PreparedStatement 操作了
 PreparedStatement ps = (PreparedStatement) statement;
 // 执行SQL增删改操作
 ps.execute();
 // 获取影响的行数
 int rows = ps.getUpdateCount();
 Object parameterObject = boundSql.getParameterObject();
 KeyGenerator keyGenerator = mappedStatement.getKeyGenerator();
 keyGenerator.processAfter(executor, mappedStatement, ps, parameterObject);
 // 返回影响的行数
 return rows;
}

5 小结

mybatis执行SQL的流程都是：
1.根据statement字符串从configuration中获取对应的mappedStatement；
2.根据获取的mappedStatement创建相应的Statement实例；
3.根据传入的参数对statement实例进行参数设置；
4.执行statement并执行后置操作；

来源：https://blog.csdn.net/ZGL_cyy/article/details/124876412