Oracle 查找并删除表中的重复记录

案例：一个应用表中的一个字段是主键，向表中插入数据时，先把数据放在临时表中（没有主键）然后再插入应用表。

这时候如果临时表中有重复数据，无论是主键字段businessid有重复，还是一整行有重复都会报出违反唯一主键约束错误。

方法：group by  XX having count(*)>1,rowid,distinct,temporary table,procedure

1、查询表中的重复数据

a.重复一个字段

b.重复多个字段

c.重复一整行

创建测试表：

create table cfa (businessid number,customer varchar2(50),branchcode varchar2(10),data_date varchar2(10));
insert into cfa values (1,'Albert','SCB','2011-11-11');
insert into cfa values (2,'Andy','DB','2011-11-12');
insert into cfa values (3,'Allen','HSBC','2011-11-13');

---------------以下为重复数据----------------------------------------------
insert into cfa values (1,'Alex','ICBC','2011-11-14');
insert into cfa values (1,'Albert','CTBK','2011-11-15');
insert into cfa values (1,'Albert','SCB','2011-11-11');

对于a的情况,只有businessid重复

select * from cfa where businessid in  (select businessid from cfa group by businessid having count(businessid)>1);

如果是b的情况，businessid 和name同时存在重复

select * from cfa where (businessid,customer) in (select businessid,customer from cfa group by businessid,customer having count(*)>1);

对于c的情况，重复一整行

参考b的方法：select * from cfa where (businessid,customer,branchcode,data_date) in (select * from cfa group by businessid,customer,branchcode,data_date having count(*)>1);

2、删除表中的重复数据

a情况，删除表中多余的重复记录，重复记录是根据单个字段（businessid）来判断，只留有rowid最小的记录

也可以只保留rowid不是最小记录，需要把代码中的min改为max这里不再赘述。

delete from cfa
where businessid in (select businessid
                        from cfa
                       group by businessid
                      having count(businessid) > 1)
   and rowid not in (select min(rowid)
                       from cfa
                      group by businessid
                     having count(businessid) > 1);

或者，使用下面更简单高效的语句

DELETE FROM cfa t
WHERE t.ROWID >
       (SELECT MIN(X.ROWID) FROM cfa X WHERE X.businessid = t.businessid);

b情况，删除表中多余的重复记录（多个字段），只留有rowid最小的记录

delete from cfa
where (businessid,customer) in (select businessid,customer
                        from cfa
                       group by businessid,customer
                      having count(*) > 1)
   and rowid not in (select min(rowid)
                       from cfa
                      group by businessid,customer
                     having count(*) > 1);

或者，使用下面更简单高效的语句

DELETE FROM cfa t
WHERE t.ROWID > (SELECT MIN(X.ROWID)
                    FROM cfa X
                   WHERE X.businessid = t.businessid
                     and x.customer = t.customer);

c情况，这种情况就比较简单，使用临时表方法

create table cfabak as select distinct * from cfa;

truncate table cfa;--如果是生产最好对该表backup

Insert into cfa select * from cfabak;

commit;