在日常的工作中经常会遇到客户反馈,针对一个等值查询,为什么查询出来的结果是错误的呢?而此刻我的内存独白是:一定是sql语句写的有问题呗,找我干啥?当然了,这也就是开玩笑,客户是上帝啊,客户虐我千万遍,我待客户如初恋!接下来肯定就是收集相关的信息,比如建表语句,SQL语句,查询结果等;
下面针对客户所反馈的情况,我们去动手实验一下;
MySQL中隐式转换详细查看官方文档相关的说明:
https://dev.mysql.com/doc/refman/5.7/en/type-conversion.html
环境说明:
MYSQL 5.7
测试表:
root@localhost [wjq]>show create table emp\G; *************************** 1. row *************************** Table: emp Create Table: CREATE TABLE `emp` ( `EMPNO` int(11) NOT NULL, `ENAME` varchar(15) NOT NULL, `JOB` varchar(15) NOT NULL, `MGR` int(11) DEFAULT '0', `HIREDATE` timestamp NULL DEFAULT NULL, `SAL` int(20) DEFAULT '0', `COMM` int(11) DEFAULT '0', `DEPTNO` int(11) NOT NULL, PRIMARY KEY (`EMPNO`), KEY `idx_deptno` (`DEPTNO`), KEY `idx_sal` (`SAL`), KEY `idx_comm` (`COMM`), KEY `idx_ename` (`ENAME`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 1 row in set (0.00 sec)
1、过滤字段为数值类型(int)
在如上测试表emp中empno是主键,类型为int,那么:
select * from emp where empno=’7788′;
会产生隐式转换吗?
下面通过实验证明:
root@localhost [wjq]>select * from emp where empno=7788; +-------+-------+---------+------+---------------------+------+------+--------+ | EMPNO | ENAME | JOB | MGR | HIREDATE | SAL | COMM | DEPTNO | +-------+-------+---------+------+---------------------+------+------+--------+ | 7788 | SCOTT | ANALYST | 7566 | 1987-04-19 00:00:00 | 3000 | NULL | 20 | +-------+-------+---------+------+---------------------+------+------+--------+ 1 row in set (0.00 sec) root@localhost [wjq]>explain select * from emp where empno=7788; +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ | 1 | SIMPLE | emp | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL | +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ 1 row in set, 1 warning (0.00 sec) root@localhost [wjq]>select * from emp where empno='7788'; +-------+-------+---------+------+---------------------+------+------+--------+ | EMPNO | ENAME | JOB | MGR | HIREDATE | SAL | COMM | DEPTNO | +-------+-------+---------+------+---------------------+------+------+--------+ | 7788 | SCOTT | ANALYST | 7566 | 1987-04-19 00:00:00 | 3000 | NULL | 20 | +-------+-------+---------+------+---------------------+------+------+--------+ 1 row in set (0.00 sec) root@localhost [wjq]>explain select * from emp where empno='7788'; +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ | 1 | SIMPLE | emp | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL | +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ 1 row in set, 1 warning (0.00 sec)
通过上述的测试结果可以发现,针对数据类型字段,即使类型不一致,并不影响是否使用索引,执行计划是一样的,不会产生隐式转换。但仍然建议在开发程序和生产库中尽量避免出现这样的SQL。
注意:
在过滤字段为数值类型的时候,数值类型有一种隐式转换,如果以数字开头的,包含有字符,后面的字符将被截断,只取前面的数字值,如果不以数字开关的将被置为0。
测试如下:
root@localhost [wjq]>select * from emp where empno='7788wjq123'; +-------+-------+---------+------+---------------------+------+------+--------+ | EMPNO | ENAME | JOB | MGR | HIREDATE | SAL | COMM | DEPTNO | +-------+-------+---------+------+---------------------+------+------+--------+ | 7788 | SCOTT | ANALYST | 7566 | 1987-04-19 00:00:00 | 3000 | NULL | 20 | +-------+-------+---------+------+---------------------+------+------+--------+ 1 row in set, 1 warning (0.00 sec) root@localhost [wjq]>show warnings; +---------+------+------------------------------------------------+ | Level | Code | Message | +---------+------+------------------------------------------------+ | Warning | 1292 | Truncated incorrect DOUBLE value: '7788wjq123' | +---------+------+------------------------------------------------+ 1 row in set (0.00 sec)
## 这个就相当于empno=7788,后面的wjq123将被截断,但是并且不影响索引的使用,如下是执行计划:
root@localhost [wjq]>explain select * from emp where empno='7788wjq123'; +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ | 1 | SIMPLE | emp | NULL | const | PRIMARY | PRIMARY | 4 | const | 1 | 100.00 | NULL | +----+-------------+-------+------------+-------+---------------+---------+---------+-------+------+----------+-------+ 1 row in set, 2 warnings (0.00 sec) root@localhost [wjq]> SELECT '7788wjq123'=7788; +-------------------+ | '7788wjq123'=7788 | +-------------------+ | 1 | +-------------------+ 1 row in set, 1 warning (0.00 sec) root@localhost [wjq]>SELECT CAST('7788wjq123' AS SIGNED)=7788; +-----------------------------------+ | CAST('7788wjq123' AS SIGNED)=7788 | +-----------------------------------+ | 1 | +-----------------------------------+ 1 row in set, 1 warning (0.00 sec) 下面的这个就相当于empno=0 root@localhost [wjq]>select * from emp where empno='wjq7788'; Empty set (0.00 sec) root@localhost [wjq]>select 'wjq7788'=7788; +----------------+ | 'wjq7788'=7788 | +----------------+ | 0 | +----------------+ 1 row in set, 1 warning (0.00 sec) root@localhost [wjq]>SELECT CAST('wjq7788' AS SIGNED)=7788; +--------------------------------+ | CAST('wjq7788' AS SIGNED)=7788 | +--------------------------------+ | 0 | +--------------------------------+ 1 row in set, 1 warning (0.00 sec)
2、过滤字段为字符类型(varchar)
针对测试表emp中的ename字段(varchar类型),上面有一辅助索引idx_ename,并且ename中有两个值是全数字的,若有这样的查询:
select * from emp where ename=123456;
上面的SQL会不会出现隐式转换呢?
下面实验证明:
root@localhost [wjq]>select * from emp where ename=123456; +-------+--------+-----+------+---------------------+------+------+--------+ | EMPNO | ENAME | JOB | MGR | HIREDATE | SAL | COMM | DEPTNO | +-------+--------+-----+------+---------------------+------+------+--------+ | 7936 | 123456 | DBA | 7788 | 2019-03-08 16:13:56 | 1230 | 0 | 10 | +-------+--------+-----+------+---------------------+------+------+--------+ 1 row in set, 16 warnings (0.00 sec) root@localhost [wjq]>show warnings; +---------+------+--------------------------------------------+ | Level | Code | Message | +---------+------+--------------------------------------------+ | Warning | 1292 | Truncated incorrect DOUBLE value: 'SMITH' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'ALLEN' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'WARD' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'JONES' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'MARTIN' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'BLAKE' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'CLARK' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'SCOTT' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'KING' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'TURNER' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'ADAMS' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'JAMES' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'FORD' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'MILLER' | | Warning | 1292 | Truncated incorrect DOUBLE value: 'wjq345' | | Warning | 1292 | Truncated incorrect DOUBLE value: '12wjq5' | +---------+------+--------------------------------------------+ 16 rows in set (0.00 sec) root@localhost [wjq]>explain select * from emp where ename=123456; +----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+ | 1 | SIMPLE | emp | NULL | ALL | idx_ename | NULL | NULL | NULL | 18 | 10.00 | Using where | +----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+ 1 row in set, 3 warnings (0.00 sec)
从上面的测试结果可以看出,当过滤的字段是字符类型的时候,没有使用到索引,走的全表扫描;
下面接着
root@localhost [wjq]>select * from emp where ename=12; +-------+--------+-----+------+---------------------+------+------+--------+ | EMPNO | ENAME | JOB | MGR | HIREDATE | SAL | COMM | DEPTNO | +-------+--------+-----+------+---------------------+------+------+--------+ | 7939 | 12wjq5 | DBA | 7788 | 2019-03-08 16:13:56 | 2567 | 0 | 30 | +-------+--------+-----+------+---------------------+------+------+--------+ 1 row in set, 16 warnings (0.00 sec) root@localhost [wjq]> root@localhost [wjq]>explain select * from emp where ename=12; +----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+ | id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra | +----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+ | 1 | SIMPLE | emp | NULL | ALL | idx_ename | NULL | NULL | NULL | 18 | 10.00 | Using where | +----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+-------------+
当varchar类型不对时,仍然是可以查出结果,后面的wjq5被截断,无法使用索引,查询的结果也是错误的
我们用数值型的12和ename进行比较的时候,不仅无法利用索引,同时查询出来的结果也是错误的,
root@localhost [wjq]>show warnings\G; *************************** 1. row *************************** Level: Warning Code: 1739 Message: Cannot use ref access on index 'idx_ename' due to type or collation conversion on field 'ENAME' *************************** 2. row *************************** Level: Warning Code: 1739 Message: Cannot use range access on index 'idx_ename' due to type or collation conversion on field 'ENAME' *************************** 3. row *************************** Level: Note Code: 1003 Message: /* select#1 */ select `wjq`.`emp`.`EMPNO` AS `EMPNO`,`wjq`.`emp`.`ENAME` AS `ENAME`,`wjq`.`emp`.`JOB` AS `JOB`,`wjq`.`emp`.`MGR` AS `MGR`,`wjq`.`emp`.`HIREDATE` AS `HIREDATE`,`wjq`.`emp`.`SAL` AS `SAL`,`wjq`.`emp`.`COMM` AS `COMM`,`wjq`.`emp`.`DEPTNO` AS `DEPTNO` from `wjq`.`emp` where (`wjq`.`emp`.`ENAME` = 12) 3 rows in set (0.00 sec) root@localhost [wjq]> SELECT CAST('12' AS SIGNED)=12; +-------------------------+ | CAST('12' AS SIGNED)=12 | +-------------------------+ | 1 | +-------------------------+ 1 row in set (0.00 sec) root@localhost [wjq]> SELECT CAST('12wjq5' AS SIGNED)=12; +-----------------------------+ | CAST('12wjq5' AS SIGNED)=12 | +-----------------------------+ | 1 | +-----------------------------+ 1 row in set, 1 warning (0.00 sec)
之所以上述查看有结果,是因为MySQL针对12wjq5的值进行了转化,变成了12;
通过上述的测试,如果是字符类型,当出现类型不一致时,是会影响索引的使用的,会产生隐式转换的,并且查询出来的结果很有可能是错误的。
如下是官方文档说明:
For comparisons of a string column with a number, MySQL cannot use an index on the column to look up the value quickly.
3、过滤字段为浮点类型(float或double)
Comparisons that use floating-point numbers (or values that are converted to floating-point numbers) are approximate because such numbers are inexact. This might lead to results that appear inconsistent:
如果查询过滤中使用了浮点型,那么比较会是近似的,将导致结果看起来不一致,也就是可能导致查询结果错误。
root@localhost [wjq]>SELECT '180153763202434582' = 180153763202434582; +-------------------------------------------+ | '180153763202434582' = 180153763202434582 | +-------------------------------------------+ | 1 | +-------------------------------------------+ 1 row in set (0.00 sec) root@localhost [wjq]>SELECT '180153763202434582' = 180153763202434585; +-------------------------------------------+ | '180153763202434582' = 180153763202434585 | +-------------------------------------------+ | 1 | +-------------------------------------------+ 1 row in set (0.00 sec)
通过上面的结果可以发现,字符串的’180153763202434582′ 和 数值的180153763202434585比较结果竟然是相等的。
我们再看下字符串’180153763202434582′ 和字符串’180153763202434585′ 转化为浮点型的结果
root@localhost [wjq]>select '180153763202434582'+0.0; +--------------------------+ | '180153763202434582'+0.0 | +--------------------------+ | 1.801537632024346e17 | +--------------------------+ 1 row in set (0.00 sec) root@localhost [wjq]>select '180153763202434585'+0.0; +--------------------------+ | '180153763202434585'+0.0 | +--------------------------+ | 1.801537632024346e17 | +--------------------------+ 1 row in set (0.00 sec)
我们发现,将两个不同的字符串转化为浮点数后,结果是一样的,
所以只要是转化为浮点数之后的值是相等的,那么,经过隐式转化后的比较也会相等,我们继续进行测试其他转化为浮点型相等的字符串的结果
root@localhost [wjq]>SELECT '180153763202434589'+0.0; +--------------------------+ | '180153763202434589'+0.0 | +--------------------------+ | 1.801537632024346e17 | +--------------------------+ 1 row in set (0.00 sec) root@localhost [wjq]>SELECT '180153763202434594'+0.0; +--------------------------+ | '180153763202434594'+0.0 | +--------------------------+ | 1.801537632024346e17 | +--------------------------+ 1 row in set (0.00 sec)
字符串’180153763202434589’和’180153763202434594’转化为浮点型结果一样,我们看下他们和数值的比较结果
root@localhost [wjq]>SELECT '180153763202434589' = 180153763202434585; +-------------------------------------------+ | '180153763202434589' = 180153763202434585 | +-------------------------------------------+ | 1 | +-------------------------------------------+ 1 row in set (0.00 sec) root@localhost [wjq]>SELECT '180153763202434594' = 180153763202434585; +-------------------------------------------+ | '180153763202434594' = 180153763202434585 | +-------------------------------------------+ | 1 | +-------------------------------------------+ 1 row in set (0.00 sec)
结果也是符合预期的。因此,当MySQL遇到字段类型不匹配的时候,会进行各种隐式转化,一定要小心,有可能导致精度丢失。
总结
不管是Oracle还是MySQL,在数据库中进行查询的时候,在查询过滤的时候,过滤条件一定要注意字段类型,杜绝隐式转化,这样不仅会导致查询缓慢,还会导致结果错误,这是生产业务所不能接受的;