MySQL数据库行去重复和列去重复

数据库表中数据行去重复

0.起因
我在创建唯一索引时，提示有重复数据，创建失败。于是，我得先去重，然后再创建唯一索引。

1.建表

sqlCREATE TABLE `demo_table` (  `id` int(10) unsigned NOT NULL,  `name` char(255) NOT NULL,  `email` char(255) NOT NULL,  PRIMARY KEY (`id`)) ENGINE=InnoDB DEFAULT CHARSET=utf8;

2.插入数据，并尝试建立唯一索引

sqlINSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('1', 'u1', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('2', 'u2', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('3', 'u3', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('4', 'u4', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('5', 'u5', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('6', 'u6', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('7', 'u7', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('8', 'u8', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('9', 'u9', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('10', 'u10', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('11', 'u11', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('12', 'u12', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('13', 'u13', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('14', 'u14', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('15', 'u15', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('16', 'u16', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('17', 'u17', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('18', 'u18', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('19', 'u19', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('20', 'u20', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('21', 'u21', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('22', 'u22', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('23', 'u23', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('24', 'u24', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('25', 'u25', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('26', 'u26', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('27', 'u27', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('28', 'u28', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('29', 'u29', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('30', 'u30', '[e-mail]');

创建唯一索引

sqlCREATE UNIQUE INDEX idx_email ON demo_table(email);

执行结果：
ERROR 1062 (23000): Duplicate entry '[e-mail]' for key 'idx_email'
提示有重复数据，导致创建唯一索引失败。

3.看看我们插入的数据

sqlselect * from demo_table;

执行结果：

sql+----+------+---------------+| id | name | email         |+----+------+---------------+|  1 | u1   | [e-mail]  ||  2 | u2   | [e-mail]  ||  3 | u3   | [e-mail]  ||  4 | u4   | [e-mail]  ||  5 | u5   | [e-mail]  ||  6 | u6   | [e-mail]  ||  7 | u7   | [e-mail]  ||  8 | u8   | [e-mail]  ||  9 | u9   | [e-mail]  || 10 | u10  | [e-mail] || 11 | u11  | [e-mail] || 12 | u12  | [e-mail] || 13 | u13  | [e-mail] || 14 | u14  | [e-mail] || 15 | u15  | [e-mail] || 16 | u16  | [e-mail] || 17 | u17  | [e-mail] || 18 | u18  | [e-mail] || 19 | u19  | [e-mail] || 20 | u20  | [e-mail] || 21 | u21  | [e-mail]  || 22 | u22  | [e-mail]  || 23 | u23  | [e-mail]  || 24 | u24  | [e-mail]  || 25 | u25  | [e-mail]  || 26 | u26  | [e-mail]  || 27 | u27  | [e-mail]  || 28 | u28  | [e-mail]  || 29 | u29  | [e-mail]  || 30 | u30  | [e-mail]  |+----+------+---------------+30 rows in set (0.00 sec)

可以看到，我们插入的email字段的确有重复数据。当然，我们是故意的。

4.使用group by看看email字段完全不重复的数据。为了让结果看起来比较顺眼，我们使用order by对结果排一下序，这个不是必选项啦。

sqlselect * from demo_table group by email order by id;

执行结果：

sql+----+------+---------------+| id | name | email         |+----+------+---------------+|  1 | u1   | [e-mail]  ||  2 | u2   | [e-mail]  ||  3 | u3   | [e-mail]  ||  4 | u4   | [e-mail]  ||  5 | u5   | [e-mail]  ||  6 | u6   | [e-mail]  ||  7 | u7   | [e-mail]  ||  8 | u8   | [e-mail]  ||  9 | u9   | [e-mail]  || 10 | u10  | [e-mail] || 11 | u11  | [e-mail] || 12 | u12  | [e-mail] || 13 | u13  | [e-mail] || 14 | u14  | [e-mail] || 15 | u15  | [e-mail] || 16 | u16  | [e-mail] || 17 | u17  | [e-mail] || 18 | u18  | [e-mail] || 19 | u19  | [e-mail] || 20 | u20  | [e-mail] |+----+------+---------------+20 rows in set (0.00 sec)

可以发现，表中原本有30条记录，group by得出的结果是只有20条记录，这就说明有10条重复记录。
不相信？好吧，我们再使用having看看吧。group by后，分组记录条数是多少就表示该字段存在多少条。大于1就表示出现了重复。

sqlselect *,count(id) from demo_table group by email having count(id)>1 order by id;

执行结果：

sql+----+------+--------------+-----------+| id | name | email        | count(id) |+----+------+--------------+-----------+|  1 | u1   | [e-mail] |         4 ||  2 | u2   | [e-mail] |         4 ||  3 | u3   | [e-mail] |         3 ||  4 | u4   | [e-mail] |         2 ||  5 | u5   | [e-mail] |         2 |+----+------+--------------+-----------+5 rows in set (0.00 sec)

这下子，我们不仅仅知道重复记录的条数是10条，而且还知道是哪些email出现了重复，并且出现了多次数。
还不相信？好吧，那你自己数吧

5.删除重复数据
我们只保留id最小的数据，其他重复的都给删掉。

sqldelete from demo_table where id not in (select min(id) from demo_table group by email);

执行结果：
ERROR 1093 (HY000): You can't specify target table 'demo_table' for update in FROM clause

貌似是MySQL有保护机制，不可以这样操作。
题外话：
如果不是纯使用SQL方式处理，而是通过编程处理，可以把select min(id) from demo_table group by email查询出来的id结果集保存在变量中，然后再执行delete from demo_table where id not id结果集。有一点需要注意的是，可能id结果集很大很大，保存在变量中会吃内存。

6.把需要保留的数据id存在临时表中，然后删除重复数据。

sqlcreate table tmp_table as select min(id) from demo_table group by email;

执行结果：
Query OK, 20 rows affected (0.43 sec)
Records: 20 Duplicates: 0 Warnings: 0

看看我们创建的临时表：

sqlselect * from tmp_table;

执行结果：

sql+---------+| min(id) |+---------+|      10 ||      11 ||      12 ||      13 ||      14 ||      15 ||      16 ||      17 ||      18 ||      19 ||       1 ||      20 ||       2 ||       3 ||       4 ||       5 ||       6 ||       7 ||       8 ||       9 |+---------+20 rows in set (0.00 sec)

这就是我们要保留的记录的id列表。

删除重复数据：

sqldelete from demo_table where id not in (select * from tmp_table);

执行结果：
Query OK, 10 rows affected (0.05 sec)

让我们再看一眼demo_table表吧：

sqlselect * from demo_table;

执行结果：

sql+----+------+---------------+| id | name | email         |+----+------+---------------+|  1 | u1   | [e-mail]  ||  2 | u2   | [e-mail]  ||  3 | u3   | [e-mail]  ||  4 | u4   | [e-mail]  ||  5 | u5   | [e-mail]  ||  6 | u6   | [e-mail]  ||  7 | u7   | [e-mail]  ||  8 | u8   | [e-mail]  ||  9 | u9   | [e-mail]  || 10 | u10  | [e-mail] || 11 | u11  | [e-mail] || 12 | u12  | [e-mail] || 13 | u13  | [e-mail] || 14 | u14  | [e-mail] || 15 | u15  | [e-mail] || 16 | u16  | [e-mail] || 17 | u17  | [e-mail] || 18 | u18  | [e-mail] || 19 | u19  | [e-mail] || 20 | u20  | [e-mail] |+----+------+---------------+20 rows in set (0.00 sec)

好的，重复的数据已经被我们成功干掉了。大功告成。等等，我们还得把那个已经没有用处的临时表给删除。貌似有点卸磨杀驴的感觉啊。

sqldrop table tmp_table;

执行结果：
Query OK, 0 rows affected (0.12 sec)

7.总结一下吧
其实就三条SQL语句：

sqlcreate table tmp_table as select min(id) from demo_table group by email;delete from demo_table where id not in (select * from tmp_table);drop table tmp_table;

可能有童鞋要问，如果要去重的不是一个字段，而是多个怎么办？比如，某2条记录的email相同并且name也相同，就算为重复。
处理方式：

sqlgroup by email,name

---

再说说数据列去重复吧

1.故意插入一些列重复的数据

sqlINSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('31', '[e-mail]', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('32', '[e-mail]', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('33', '[e-mail]', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('34', '[e-mail]', '[e-mail]');INSERT INTO `test`.`demo_table` (`id`, `name`, `email`) VALUES ('35', '[e-mail]', '[e-mail]');

查找表中name字段和email字段相同的数据：

sqlselect * from demo_table as a, demo_table as b where a.id=b.id and a.name=b.email;

执行结果：

sql+----+--------------+--------------+----+--------------+--------------+| id | name         | email        | id | name         | email        |+----+--------------+--------------+----+--------------+--------------+| 31 | [e-mail] | [e-mail] | 31 | [e-mail] | [e-mail] || 32 | [e-mail] | [e-mail] | 32 | [e-mail] | [e-mail] || 33 | [e-mail] | [e-mail] | 33 | [e-mail] | [e-mail] || 34 | [e-mail] | [e-mail] | 34 | [e-mail] | [e-mail] || 35 | [e-mail] | [e-mail] | 35 | [e-mail] | [e-mail] |+----+--------------+--------------+----+--------------+--------------+5 rows in set (0.00 sec)

重复记录的id都已经找出来了，删除就参考上述方式处理了。