Spring系列之不同数据库异常如何抽象的？

前言

使用Spring-Jdbc的情况下，在有些场景中，我们需要根据数据库报的异常类型的不同，来编写我们的业务代码。比如说，我们有这样一段逻辑，如果我们新插入的记录，存在唯一约束冲突，就会返回给客户端描述：记录已存在，请勿重复操作代码一般是这么写的:

@Resourceprivate JdbcTemplate jdbcTemplate;public String testAdd(){try {jdbcTemplate.execute("INSERT INTO user_info (user_id, user_name, email, nick_name, status, address) VALUES (80002, \'张三丰\', \'xxx@126.com\', \'张真人\', 1, \'武当山\');");return "OK";}catch (DuplicateKeyException e){return "记录已存在，请勿重复操作";}}

测试一下：
如上图提示，并且无论什么更换什么数据库（Spring-Jdbc支持的），代码都不用改动

那么Spring-Jdbc是在使用不同数据库时，Spring如何帮我们实现对异常的抽象的呢?

代码实现

我们来正向看下代码:首先入口JdbcTemplate.execute方法：

public void execute(final String sql) throws DataAccessException {if (this.logger.isDebugEnabled()) {this.logger.debug("Executing SQL statement [" + sql + "]");}...//实际执行入口,调用内部方法this.execute(new ExecuteStatementCallback(), true);}

内部方法execute

@Nullableprivate <T> T execute(StatementCallback<T> action, boolean closeResources) throws DataAccessException {Assert.notNull(action, "Callback object must not be null");Connection con = DataSourceUtils.getConnection(this.obtainDataSource());Statement stmt = null;Object var12;try {...} catch (SQLException var10) {....//SQL出现异常后，所有的异常在这里进行异常转换throw this.translateException("StatementCallback", sql, var10);} finally {if (closeResources) {JdbcUtils.closeStatement(stmt);DataSourceUtils.releaseConnection(con, this.getDataSource());}}return var12;}

异常转换方法translateException

protected DataAccessException translateException(String task, @Nullable String sql, SQLException ex) {//获取异常转换器，然后根据数据库返回码相关信息执行转换操作//转换不成功，也有兜底异常UncategorizedSQLExceptionDataAccessException dae = this.getExceptionTranslator().translate(task, sql, ex);return (DataAccessException)(dae != null ? dae : new UncategorizedSQLException(task, sql, ex));}

获取转换器方法getExceptionTranslator

public SQLExceptionTranslator getExceptionTranslator() {//获取转换器属性，如果为空，则生成一个SQLExceptionTranslator exceptionTranslator = this.exceptionTranslator;if (exceptionTranslator != null) {return exceptionTranslator;} else {synchronized(this) {SQLExceptionTranslator exceptionTranslator = this.exceptionTranslator;if (exceptionTranslator == null) {DataSource dataSource = this.getDataSource();//shouldIgnoreXml是一个标记，就是不通过xml加载bean，默认falseif (shouldIgnoreXml) {exceptionTranslator = new SQLExceptionSubclassTranslator();} else if (dataSource != null) {//如果DataSource不为空，则生成转换器SQLErrorCodeSQLExceptionTranslator，一般情况下首先获取到该转换器exceptionTranslator = new SQLErrorCodeSQLExceptionTranslator(dataSource);} else {// 其他情况，生成SQLStateSQLExceptionTranslator转换器exceptionTranslator = new SQLStateSQLExceptionTranslator();}this.exceptionTranslator = (SQLExceptionTranslator)exceptionTranslator;}return (SQLExceptionTranslator)exceptionTranslator;}}}

转换方法：因为默认的转换器是SQLErrorCodeSQLExceptionTranslator，所以这里调用SQLErrorCodeSQLExceptionTranslator的doTranslate方法类图调用关系如上，实际先调用的是AbstractFallbackSQLExceptionTranslator.translate的方法

@Nullablepublic DataAccessException translate(String task, @Nullable String sql, SQLException ex) {Assert.notNull(ex, "Cannot translate a null SQLException");//这里才真正调用SQLErrorCodeSQLExceptionTranslator.doTranslate方法DataAccessException dae = this.doTranslate(task, sql, ex);if (dae != null) {return dae;} else {//如果没有找到响应的异常，则调用其他转换器，输入递归调用，这里后面说SQLExceptionTranslator fallback = this.getFallbackTranslator();return fallback != null ? fallback.translate(task, sql, ex) : null;}}

实际转换类SQLErrorCodeSQLExceptionTranslator的方法：

//这里省略了一些无关代码，只保留了核心代码//先获取SQLErrorCodes集合，在根据返回的SQLException中获取的ErrorCode进行匹配，根据匹配结果进行返回响应的异常protected DataAccessException doTranslate(String task, @Nullable String sql, SQLException ex) {....SQLErrorCodes sqlErrorCodes = this.getSqlErrorCodes();String errorCode = Integer.toString(ex.getErrorCode());...//这里用1062唯一性约束冲突，所以走到这里的逻辑，从而返回DuplicateKeyExceptionif (Arrays.binarySearch(sqlErrorCodes.getDuplicateKeyCodes(), errorCode) >= 0) {this.logTranslation(task, sql, sqlEx, false);return new DuplicateKeyException(this.buildMessage(task, sql, sqlEx), sqlEx);}...return null;}

上面的SQLErrorCodes是一个错误码集合，但是不是全部数据库的所有错误码集合，而是只取了相应数据库的错误码集合，怎么保证获取的是当前使用的数据库的错误码，而不是其他数据库的错误码呢？当然Spring为我们实现了，在SQLErrorCodeSQLExceptionTranslator中：

public class SQLErrorCodeSQLExceptionTranslator extends AbstractFallbackSQLExceptionTranslator {private SingletonSupplier<SQLErrorCodes> sqlErrorCodes;//默认构造方法，设置了如果转换失败，下一个转换器是SQLExceptionSubclassTranslatorpublic SQLErrorCodeSQLExceptionTranslator() {this.setFallbackTranslator(new SQLExceptionSubclassTranslator());}//前面生成转换器的时候，exceptionTranslator = new SQLErrorCodeSQLExceptionTranslator(dataSource);//使用的是本构造方法，传入了DataSource，其中有数据库厂商信息，本文中是MYSQLpublic SQLErrorCodeSQLExceptionTranslator(DataSource dataSource) {this();this.setDataSource(dataSource);}//从错误码工厂SQLErrorCodesFactory里，获取和数据源对应的厂商的所有错误码public void setDataSource(DataSource dataSource) {this.sqlErrorCodes = SingletonSupplier.of(() -> {return SQLErrorCodesFactory.getInstance().resolveErrorCodes(dataSource);});this.sqlErrorCodes.get();}}

错误码工厂SQLErrorCodesFactory的resolveErrorCodes方法：

//既然是工厂，里面肯定有各种数据库的错误码，本文中使用的是MYSQL，我们看一下实现逻辑@Nullablepublic SQLErrorCodes resolveErrorCodes(DataSource dataSource) {Assert.notNull(dataSource, "DataSource must not be null");if (logger.isDebugEnabled()) {logger.debug("Looking up default SQLErrorCodes for DataSource [" + this.identify(dataSource) + "]");}//从缓存中拿MYSQL对应的SQLErrorCodesSQLErrorCodes sec = (SQLErrorCodes)this.dataSourceCache.get(dataSource);if (sec == null) {synchronized(this.dataSourceCache) {sec = (SQLErrorCodes)this.dataSourceCache.get(dataSource);if (sec == null) {try {String name = (String)JdbcUtils.extractDatabaseMetaData(dataSource, DatabaseMetaData::getDatabaseProductName);if (StringUtils.hasLength(name)) {SQLErrorCodes var10000 = this.registerDatabase(dataSource, name);return var10000;}} catch (MetaDataAccessException var6) {logger.warn("Error while extracting database name", var6);}return null;}}}if (logger.isDebugEnabled()) {logger.debug("SQLErrorCodes found in cache for DataSource [" + this.identify(dataSource) + "]");}return sec;}

缓存dataSourceCache如何生成的？

public SQLErrorCodes registerDatabase(DataSource dataSource, String databaseName) {//根据数据库类型名称（这里是MySQL），获取错误码列表SQLErrorCodes sec = this.getErrorCodes(databaseName);if (logger.isDebugEnabled()) {logger.debug("Caching SQL error codes for DataSource [" + this.identify(dataSource) + "]: database product name is \'" + databaseName + "\'");}this.dataSourceCache.put(dataSource, sec);return sec;}public SQLErrorCodes getErrorCodes(String databaseName) {Assert.notNull(databaseName, "Database product name must not be null");//从errorCodesMap根据key=MYSQL获取SQLErrorCodesSQLErrorCodes sec = (SQLErrorCodes)this.errorCodesMap.get(databaseName);if (sec == null) {Iterator var3 = this.errorCodesMap.values().iterator();while(var3.hasNext()) {SQLErrorCodes candidate = (SQLErrorCodes)var3.next();if (PatternMatchUtils.simpleMatch(candidate.getDatabaseProductNames(), databaseName)) {sec = candidate;break;}}}if (sec != null) {this.checkCustomTranslatorRegistry(databaseName, sec);if (logger.isDebugEnabled()) {logger.debug("SQL error codes for \'" + databaseName + "\' found");}return sec;} else {if (logger.isDebugEnabled()) {logger.debug("SQL error codes for \'" + databaseName + "\' not found");}return new SQLErrorCodes();}}//SQLErrorCodesFactory构造方法中，生成的errorCodesMap，map的内容来自org/springframework/jdbc/support/sql-error-codes.xml文件protected SQLErrorCodesFactory() {Map errorCodes;try {DefaultListableBeanFactory lbf = new DefaultListableBeanFactory();lbf.setBeanClassLoader(this.getClass().getClassLoader());XmlBeanDefinitionReader bdr = new XmlBeanDefinitionReader(lbf);Resource resource = this.loadResource("org/springframework/jdbc/support/sql-error-codes.xml");if (resource != null && resource.exists()) {bdr.loadBeanDefinitions(resource);} else {logger.info("Default sql-error-codes.xml not found (should be included in spring-jdbc jar)");}resource = this.loadResource("sql-error-codes.xml");if (resource != null && resource.exists()) {bdr.loadBeanDefinitions(resource);logger.debug("Found custom sql-error-codes.xml file at the root of the classpath");}errorCodes = lbf.getBeansOfType(SQLErrorCodes.class, true, false);if (logger.isTraceEnabled()) {logger.trace("SQLErrorCodes loaded: " + errorCodes.keySet());}} catch (BeansException var5) {logger.warn("Error loading SQL error codes from config file", var5);errorCodes = Collections.emptyMap();}this.errorCodesMap = errorCodes;}

sql-error-codes.xml文件中配置了各个数据库的主要的错误码这里列举了MYSQL部分，当然还有其他部分，我们可以看到唯一性约束错误码是1062，就可以翻译成DuplicateKeyException异常了

<bean id="MySQL" class="org.springframework.jdbc.support.SQLErrorCodes"><property name="databaseProductNames"><list><value>MySQL</value><value>MariaDB</value></list></property><property name="badSqlGrammarCodes"><value>1054,1064,1146</value></property><property name="duplicateKeyCodes"><value>1062</value></property><property name="dataIntegrityViolationCodes"><value>630,839,840,893,1169,1215,1216,1217,1364,1451,1452,1557</value></property><property name="dataAccessResourceFailureCodes"><value>1</value></property><property name="cannotAcquireLockCodes"><value>1205,3572</value></property><property name="deadlockLoserCodes"><value>1213</value></property></bean>

你已经看到，比如上面的错误码值列举了一部分，如果出现了一个不在其中的错误码肯定是匹配不到，Spring当然能想到这种情况了

/***@公-众-号：程序员阿牛*在AbstractFallbackSQLExceptionTranslator中，看到如果查找失败会获取下一个后续转换器*/@Nullablepublic DataAccessException translate(String task, @Nullable String sql, SQLException ex) {Assert.notNull(ex, "Cannot translate a null SQLException");DataAccessException dae = this.doTranslate(task, sql, ex);if (dae != null) {return dae;} else {SQLExceptionTranslator fallback = this.getFallbackTranslator();return fallback != null ? fallback.translate(task, sql, ex) : null;}}

SQLErrorCodeSQLExceptionTranslator的后置转换器是什么？

//构造方法中已经指定，SQLExceptionSubclassTranslatorpublic SQLErrorCodeSQLExceptionTranslator() {this.setFallbackTranslator(new SQLExceptionSubclassTranslator());}

SQLExceptionSubclassTranslator的转换方法逻辑如下：

/***@公-众-号：程序员阿牛*可以看出实际按照子类类型来判断，返回相应的错误类，如果匹配不到，则找到下一个处理器，这里的处理其我们可以根据构造方法青松找到*SQLStateSQLExceptionTranslator*/@Nullableprotected DataAccessException doTranslate(String task, @Nullable String sql, SQLException ex) {if (ex instanceof SQLTransientException) {if (ex instanceof SQLTransientConnectionException) {return new TransientDataAccessResourceException(this.buildMessage(task, sql, ex), ex);}if (ex instanceof SQLTransactionRollbackException) {return new ConcurrencyFailureException(this.buildMessage(task, sql, ex), ex);}if (ex instanceof SQLTimeoutException) {return new QueryTimeoutException(this.buildMessage(task, sql, ex), ex);}} else if (ex instanceof SQLNonTransientException) {if (ex instanceof SQLNonTransientConnectionException) {return new DataAccessResourceFailureException(this.buildMessage(task, sql, ex), ex);}if (ex instanceof SQLDataException) {return new DataIntegrityViolationException(this.buildMessage(task, sql, ex), ex);}if (ex instanceof SQLIntegrityConstraintViolationException) {return new DataIntegrityViolationException(this.buildMessage(task, sql, ex), ex);}if (ex instanceof SQLInvalidAuthorizationSpecException) {return new PermissionDeniedDataAccessException(this.buildMessage(task, sql, ex), ex);}if (ex instanceof SQLSyntaxErrorException) {return new BadSqlGrammarException(task, sql != null ? sql : "", ex);}if (ex instanceof SQLFeatureNotSupportedException) {return new InvalidDataAccessApiUsageException(this.buildMessage(task, sql, ex), ex);}} else if (ex instanceof SQLRecoverableException) {return new RecoverableDataAccessException(this.buildMessage(task, sql, ex), ex);}return null;}

SQLStateSQLExceptionTranslator的转换方法：

/***@公-众-号：程序员阿牛*可以看出根据SQLState的前两位来判断异常，根据匹配结果返回相应的异常信息*/@Nullableprotected DataAccessException doTranslate(String task, @Nullable String sql, SQLException ex) {String sqlState = this.getSqlState(ex);if (sqlState != null && sqlState.length() >= 2) {String classCode = sqlState.substring(0, 2);if (this.logger.isDebugEnabled()) {this.logger.debug("Extracted SQL state class \'" + classCode + "\' from value \'" + sqlState + "\'");}if (BAD_SQL_GRAMMAR_CODES.contains(classCode)) {return new BadSqlGrammarException(task, sql != null ? sql : "", ex);}if (DATA_INTEGRITY_VIOLATION_CODES.contains(classCode)) {return new DataIntegrityViolationException(this.buildMessage(task, sql, ex), ex);}if (DATA_ACCESS_RESOURCE_FAILURE_CODES.contains(classCode)) {return new DataAccessResourceFailureException(this.buildMessage(task, sql, ex), ex);}if (TRANSIENT_DATA_ACCESS_RESOURCE_CODES.contains(classCode)) {return new TransientDataAccessResourceException(this.buildMessage(task, sql, ex), ex);}if (CONCURRENCY_FAILURE_CODES.contains(classCode)) {return new ConcurrencyFailureException(this.buildMessage(task, sql, ex), ex);}}return ex.getClass().getName().contains("Timeout") ? new QueryTimeoutException(this.buildMessage(task, sql, ex), ex) : null;}

为什么SQLState可以得出错误类型？

因为数据库是根据 X/Open 和 SQL Access Group SQL CAE 规范 (1992) 所进行的定义，SQLERROR 返回 SQLSTATE 值。SQLSTATE 值是包含五个字符的字符串。五个字符包含数值或者大写字母，代表各种错误或者警告条件的代码。SQLSTATE 有个层次化的模式：头两个字符标识条件的通常表示错误条件的类别，后三个字符表示在该通用类中的子类。成功的状态是由 00000 标识的。SQLSTATE 代码在大多数地方都是定义在 SQL 标准里