学习sql有一段时间了，发现在我建了一个用来测试的表（没有建索引）中出现了许多的重复记录。后来总结了一些删除重复记录的方法，在Oracle中，可以通过唯一rowid实现删除重复记录；还可以建临时表来实现...这个只提到其中的几种简单实用的方法，希望可以和大家分享（以表employee为例）。

SQL> desc employee

 Name                                      Null?    Type
 ----------------------------------------- -------- ------------------

emp_id                                                NUMBER(10)
emp_name                                           VARCHAR2(20)

salary                                                  NUMBER(10,2)

可以通过下面的语句查询重复的记录：

SQL> select * from employee;

    EMP_ID EMP_NAME                                  SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         1 sunshine                                      10000

         2 semon                                         20000

         2 semon                                         20000

         3 xyz                                           30000

         2 semon                                         20000

SQL> select distinct * from employee;

    EMP_ID EMP_NAME                                     SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         2 semon                                         20000

         3 xyz                                             30000

SQL>  select * from employee group by emp_id,emp_name,salary having count (*)>1

    EMP_ID EMP_NAME                                     SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         2 semon                                          20000

SQL> select * from employee e1

where rowid in (select max(rowid) from employe e2
 where e1.emp_id=e2.emp_id and

  e1.emp_name=e2.emp_name and e1.salary=e2.salary);

    EMP_ID EMP_NAME                                     SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         3 xyz                                             30000

         2 semon                                         20000

2. 删除的几种方法：

（1）通过建立临时表来实现

SQL>create table temp_emp as (select distinct * from employee) 

SQL> truncate table employee; (清空employee表的数据）

SQL> insert into employee select * from temp_emp;  (再将临时表里的内容插回来）

( 2）通过唯一rowid实现删除重复记录.在Oracle中，每一条记录都有一个rowid，rowid在整个数据库中是唯一的，rowid确定了每条记录是在Oracle中的哪一个数据文件、块、行上。在重复的记录中，可能所有列的内容都相同，但rowid不会相同，所以只要确定出重复记录中那些具有最大或最小rowid的就可以了，其余全部删除。

SQL>delete from employee e2 where rowid not in (
        select max(e1.rowid) from employee e1 where

        e1.emp_id=e2.emp_id and e1.emp_name=e2.emp_name and e1.salary=e2.salary);--这里用min(rowid)也可以。

SQL>delete from employee e2 where rowid <(
        select max(e1.rowid) from employee e1 where
        e1.emp_id=e2.emp_id and e1.emp_name=e2.emp_name and

                  e1.salary=e2.salary);

（3）也是通过rowid，但效率更高。

SQL>delete from employee where rowid not in (
        select max(t1.rowid) from employee t1 group by

         t1.emp_id,t1.emp_name,t1.salary);--这里用min(rowid)也可以。

    EMP_ID EMP_NAME                                     SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         3 xyz                                             30000

         2 semon                                         20000

SQL> desc employee

 Name                                      Null?    Type
 ----------------------------------------- -------- ------------------

emp_id                                                NUMBER(10)
emp_name                                           VARCHAR2(20)

salary               ,                                    NUMBER(10,2)

可以通过下面的语句查询重复的记录：

SQL> select * from employee;

    EMP_ID EMP_NAME                                  SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         1 sunshine                                      10000

         2 semon                                         20000

         2 semon                                         20000

         3 xyz                                           30000

         2 semon                                         20000

SQL> select distinct * from employee;

    EMP_ID EMP_NAME                                     SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         2 semon                                         20000

         3 xyz                                             30000

SQL>  select * from employee group by emp_id,emp_name,salary having count (*)>1

    EMP_ID EMP_NAME                                     SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         2 semon                                          20000

SQL> select * from employee e1

where rowid in (select max(rowid) from employe e2
 where e1.emp_id=e2.emp_id and

  e1.emp_name=e2.emp_name and e1.salary=e2.salary);

    EMP_ID EMP_NAME                                     SALARY

---------- ---------------------------------------- ----------

         1 sunshine                                      10000

         3 xyz                                             30000

         2 semon                                         20000

 20000