The processing flow of insert method, update method and delete method in java (Part 2)

Configuration’s newStatementHandler analysis

SimpleExecutor’s doUpdate method has been analyzed above:

 1 public int doUpdate(MappedStatement ms, Object parameter) throws SQLException { 2     Statement stmt = null; 3     try { 4       Configuration configuration = ms.getConfiguration(); 5       StatementHandler handler = configuration.newStatementHandler(this, ms, parameter, RowBounds.DEFAULT, null, null); 6       stmt = prepareStatement(handler, ms.getStatementLog()); 7       return handler.update(stmt); 8     } finally { 9       closeStatement(stmt);10     }11 }

When I re-read the newStatementHandler method in line 5 in the past two days, I found that the method analyzed above in this method is too simple. Let’s go through the newStatementHandler method of Configuration. The implementation of the method is:

1 public StatementHandler newStatementHandler(Executor executor, MappedStatement mappedStatement, Object parameterObject, RowBounds rowBounds, ResultHandler resultHandler, BoundSql boundSql) {2     StatementHandler statementHandler = new RoutingStatementHandler(executor, mappedStatement, parameterObject, rowBounds, resultHandler, boundSql);3     statementHandler = (StatementHandler) interceptorChain.pluginAll(statementHandler);4     return statementHandler;5 }

The 3rd line of code is for adding a plug-in, which is nothing interesting. Take a look at the 2nd line of code. What is actually instantiated by the StatementHandler interface is RoutingStatementHandler, and the construction method The implementation is:

 1 public RoutingStatementHandler(Executor executor, MappedStatement ms, Object parameter, RowBounds rowBounds, ResultHandler resultHandler, BoundSql boundSql) { 2  3     switch (ms.getStatementType()) { 4       case STATEMENT: 5         delegate = new SimpleStatementHandler(executor, ms, parameter, rowBounds, resultHandler, boundSql); 6         break; 7       case PREPARED: 8         delegate = new PreparedStatementHandler(executor, ms, parameter, rowBounds, resultHandler, boundSql); 9         break;10       case CALLABLE:11         delegate = new CallableStatementHandler(executor, ms, parameter, rowBounds, resultHandler, boundSql);12         break;13       default:14         throw new ExecutorException("Unknown statement type: " + ms.getStatementType());15     }16 17 }

RoutingStatementHandler is also an implementation of the decorator pattern, which implements the StatementHandler interface and holds the StatementHandler interface reference delegate. The StatementType here is PREPARED, so the judgment on line 7 is executed to instantiate PreparedStatementHandler. The instantiation process is:

 1 protected BaseStatementHandler(Executor executor, MappedStatement mappedStatement, Object parameterObject, RowBounds rowBounds, ResultHandler resultHandler, BoundSql boundSql) { 2     this.configuration = mappedStatement.getConfiguration(); 3     this.executor = executor; 4     this.mappedStatement = mappedStatement; 5     this.rowBounds = rowBounds; 6  7     this.typeHandlerRegistry = configuration.getTypeHandlerRegistry(); 8     this.objectFactory = configuration.getObjectFactory(); 9 10     if (boundSql == null) { // issue #435, get the key before calculating the statement11       generateKeys(parameterObject);12       boundSql = mappedStatement.getBoundSql(parameterObject);13     }14 15     this.boundSql = boundSql;16 17     this.parameterHandler = configuration.newParameterHandler(mappedStatement, parameterObject, boundSql);18     this.resultSetHandler = configuration.newResultSetHandler(executor, mappedStatement, rowBounds, parameterHandler, resultHandler, boundSql);19 }

The focus here is BoundSql , which can be obtained through MappedStatement. Several important contents are stored in BoundSql:

1. The parameter object itself

2. Parameter list

3. SQL statement to be executed

Some secondary development frameworks based on MyBatis usually get the SQL statements in BoundSql, modify them and reset them into BoundSql.

Generate Statement

##The process of generating Connection has been written above. This article continues Look, first let’s post the prepareStatement method of SimpleExecutor again:

1 private Statement prepareStatement(StatementHandler handler, Log statementLog) throws SQLException {2     Statement stmt;3     Connection connection = getConnection(statementLog);4     stmt = handler.prepare(connection, transaction.getTimeout());5     handler.parameterize(stmt);6     return stmt;7 }

Then there is the code in line 4, which generates Statement. The code in line 4 is implemented as:

  Statement prepare(Connection connection, Integer transactionTimeout)

delegate The above is the decorated role in the decorator mode. Its interface type is StatementHandler and its real type is PreparedStatementHandler. This has been analyzed at the beginning. Take a look at the prepare method implementation:

 1 public Statement prepare(Connection connection, Integer transactionTimeout) throws SQLException { 2     ErrorContext.instance().sql(boundSql.getSql()); 3     Statement statement = null; 4     try { 5       statement = instantiateStatement(connection); 6       setStatementTimeout(statement, transactionTimeout); 7       setFetchSize(statement); 8       return statement; 9     } catch (SQLException e) {10       closeStatement(statement);11       throw e;12     } catch (Exception e) {13       closeStatement(statement);14       throw new ExecutorException("Error preparing statement.  Cause: " + e, e);15     }16 }

The code in line 6 sets the query timeout, and the code in line 7 sets the received data size. I won’t follow up and look at it. Let’s look at the implementation of the instantiateStatement method in line 6:

 1 protected Statement instantiateStatement(Connection connection) throws SQLException { 2     String sql = boundSql.getSql(); 3     if (mappedStatement.getKeyGenerator() instanceof Jdbc3KeyGenerator) { 4       String[] keyColumnNames = mappedStatement.getKeyColumns(); 5       if (keyColumnNames == null) { 6         return connection.prepareStatement(sql, PreparedStatement.RETURN_GENERATED_KEYS); 7       } else { 8         return connection.prepareStatement(sql, keyColumnNames); 9       }10     } else if (mappedStatement.getResultSetType() != null) {11       return connection.prepareStatement(sql, mappedStatement.getResultSetType().getValue(), ResultSet.CONCUR_READ_ONLY);12     } else {13       return connection.prepareStatement(sql);14     }15 }

Line 2, get the real SQL statement from boundSql, The first part has already been analyzed. After getting the SQL statement, execute the judgment on lines 3 and 5. Here is the code we are familiar with to get the Statement through Connection. PreparedStatement is obtained through the prepareStatement method. Its true type is com.mysql.jdbc.JDBC4PreparedStatement, which is Subclass of PreparedStatement.

Statement parameter setting

After obtaining the Statement, the next step is to set the parameters. , look at the code for setting parameters, or go back to the prepareStatement method of SimpleExecutor:

1 private Statement prepareStatement(StatementHandler handler, Log statementLog) throws SQLException {2     Statement stmt;3     Connection connection = getConnection(statementLog);4     stmt = handler.prepare(connection, transaction.getTimeout());5     handler.parameterize(stmt);6     return stmt;7 }

and the code on line 5:

 1 public void parameterize(Statement statement) throws SQLException { 2     parameterHandler.setParameters((PreparedStatement) statement); 3 }

Continue with the code in line 2:

 1 public void setParameters(PreparedStatement ps) { 2     ErrorContext.instance().activity("setting parameters").object(mappedStatement.getParameterMap().getId()); 3     List<parametermapping> parameterMappings = boundSql.getParameterMappings(); 4     if (parameterMappings != null) { 5       for (int i = 0; i <div class="cnblogs_code">
<p><span style="font-size: 13px; font-family: 宋体">最终执行的是第26行的代码，从26行的代码我们可以知道，<span style="color: #ff0000"><strong>参数设置到最后都是通过参数的TypeHandler来执行的</strong></span>，JDBC为我们预定义了很多TypeHandler，比如int值的TypeHandler就是IntegerTypeHandler，当然我们也可以定义自己的TypeHandler，通常来说继承BaseTypeHandler就可以了。</span></p>
<p><span style="font-size: 13px; font-family: 宋体">但是在此之前，会获取到Statement（setParameters方法形参</span><span style="font-size: 13px; font-family: 宋体">）、占位符位置号（for循环的遍历参数i）、参数值（通过属性名获取）与jdbcType（配置在配置文件中，默认为null），最终执行TypeHandler的setParameters方法，这是BaseTypeHandler中的一个方法：</span></p>
<div class="cnblogs_code"><pre class="brush:php;toolbar:false"> 1 public void setParameter(PreparedStatement ps, int i, T parameter, JdbcType jdbcType) throws SQLException { 2     if (parameter == null) { 3       if (jdbcType == null) { 4         throw new TypeException("JDBC requires that the JdbcType must be specified for all nullable parameters."); 5       } 6       try { 7         ps.setNull(i, jdbcType.TYPE_CODE); 8       } catch (SQLException e) { 9         throw new TypeException("Error setting null for parameter #" + i + " with JdbcType " + jdbcType + " . " +10                 "Try setting a different JdbcType for this parameter or a different jdbcTypeForNull configuration property. " +11                 "Cause: " + e, e);12       }13     } else {14       try {15         setNonNullParameter(ps, i, parameter, jdbcType);16       } catch (Exception e) {17         throw new TypeException("Error setting non null for parameter #" + i + " with JdbcType " + jdbcType + " . " +18                 "Try setting a different JdbcType for this parameter or a different configuration property. " +19                 "Cause: " + e, e);20       }21     }22 }