zoukankan      html  css  js  c++  java
  • 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;

  • 相关阅读:
    BZOJ 3028 食物 ——生成函数
    BZOJ 1933 [Shoi2007]Bookcase 书柜的尺寸 ——动态规划
    论咸鱼的自我修养之网络流
    SPOJ LCS2 Longest Common Substring II ——后缀自动机
    SPOJ NSUBSTR Substrings ——后缀自动机
    BZOJ 1879 [Sdoi2009]Bill的挑战 ——状压DP
    BZOJ 1875 [SDOI2009]HH去散步 ——动态规划 矩阵乘法
    BZOJ 1226 [SDOI2009]学校食堂Dining ——状压DP
    BZOJ 4566 [Haoi2016]找相同字符 ——广义后缀自动机
    BZOJ 3473 字符串 ——广义后缀自动机
  • 原文地址:https://www.cnblogs.com/xinyuyuanm/p/3022897.html
Copyright © 2011-2022 走看看