平时工作中可能会遇到当试图对库表中的某一列或几列创建唯一索引时，系统提示 ORA-01452 ：不能创建唯一索引，发现重复记录。
下面总结一下几种查找和删除重复记录的方法（以表CZ为例）：
表CZ的结构如下：
1 SQL> desc cz 2 Name Null? Type 3 ------------------------------------------------------------------- 4 C1 NUMBER(10) 5 C10 NUMBER(5) 6 C20 VARCHAR2(3)
删除重复记录的方法原理：
(1).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。
(2).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rowid的就可以了，其余全部删除。
重复记录判断的标准是：
C1，C10和C20这三列的值都相同才算是重复记录。
经查看表CZ总共有16条记录：
1 SQL>set pagesize 100 2 SQL>select * from cz; 3 C1 C10 C20 4 ---------- ---------- --- 5 1 2 dsf 6 1 2 dsf 7 1 2 dsf 8 1 2 dsf 9 2 3 che 10 1 2 dsf 11 1 2 dsf 12 1 2 dsf 13 1 2 dsf 14 2 3 che 15 2 3 che 16 2 3 che 17 2 3 che 18 3 4 dff 19 3 4 dff 20 3 4 dff 21 4 5 err 22 5 3 dar 23 6 1 wee 24 7 2 zxc 25 20 rows selected.
1.查找重复记录的几种方法：
(1).SQL>select * from cz group by c1,c10,c20 having count(*) >1;
1 SQL>select * from cz group by c1,c10,c20 having count(*) >1; 2 3 C1 C10 C20 4 ----------------------- 5 1 2 dsf 6 2 3 che 7 3 4 dff
(2).SQL>select distinct * from cz;
1 SQL>select distinct * from cz; 2 3 C1 C10 C20 4 ----------------------- 5 1 2 dsf 6 2 3 che 7 3 4 dff
(3).SQL>select * from cz a where rowid=(select max(rowid) from cz where c1=a.c1 and c10=a.c10 and c20=a.c20);
1 SQL>select * from cz a where rowid=(select max(rowid) from cz where c1=a.c1 and c10=a.c10 and c20=a.c20); 2 3 C1 C10 C20 4 ----------------------- 5 1 2 dsf 6 2 3 che 7 3 4 dff
2.删除重复记录的几种方法：
(1).适用于有大量重复记录的情况（在C1，C10和C20列上建有索引的时候，用以下语句效率会很高）：
1 SQL>DELETE FROM cz 2 WHERE (c1, c10, c20) IN ( SELECT c1, c10, c20 3 FROM cz 4 GROUP BY c1, c10, c20 5 HAVING COUNT (*) > 1) 6 AND ROWID NOT IN ( SELECT MIN (ROWID) 7 FROM cz 8 GROUP BY c1, c10, c20 9 HAVING COUNT (*) > 1); 10 11 SQL>DELETE fron cz 12 WHERE ROWID NOT IN ( SELECT MIN (ROWID) 13 FROM cz 14 GROUP BY c1, c10, c20);
(2).适用于有少量重复记录的情况（注意，对于有大量重复记录的情况，用以下语句效率会很低）：
1 SQL>DELETE FROM cz a 2 WHERE a.ROWID != 3 (SELECT MAX (ROWID) 4 FROM cz b 5 WHERE a.c1 = b.c1 AND a.c10 = b.c10 AND a.c20 = b.c20); 6 7 SQL>DELETE FROM cz a 8 WHERE a.ROWID < 9 (SELECT MAX (ROWID) 10 FROM cz b 11 WHERE a.c1 = b.c1 AND a.c10 = b.c10 AND a.c20 = b.c20); 12 13 SQL>DELETE FROM cz a 14 WHERE ROWID < (SELECT MAX (ROWID) 15 FROM cz 16 WHERE c1 = a.c1 AND c10 = a.c10 AND c20 = a.c20);
(3).适用于有少量重复记录的情况（临时表法）：
1 SQL>create table test as select distinct * from cz; (建一个临时表test用来存放重复的记录) 2 3 SQL>truncate table cz; (清空cz表的数据，但保留cz表的结构) 4 5 SQL>insert into cz select * from test; (再将临时表test里的内容反插回来)
(4).适用于有大量重复记录的情况(Exception into 子句法)：
采用alter table 命令中的 Exception into 子句也可以确定出库表中重复的记录。这种方法稍微麻烦一些，为了使用“excepeion into ”子句，必须首先创建 EXCEPTIONS 表。创建该表的 SQL 脚本文件为 utlexcpt.sql 。对于win2000系统和 UNIX 系统， Oracle 存放该文件的位置稍有不同，在win2000系统下，该脚本文件存放在$ORACLE_HOMEOra90 dbmsadmin 目录下;而对于 UNIX 系统，该脚本文件存放在$ORACLE_HOME/rdbms/admin 目录下。
具体步骤如下：
1 SQL>@?/rdbms/admin/utlexcpt.sql 2 Table created. 3 SQL>desc exceptions 4 Name Null? Type 5 ----------------------------------------- -------- -------------- 6 ROW_ID ROWID 7 OWNER VARCHAR2(30) 8 TABLE_NAME VARCHAR2(30) 9 CONSTRAINT VARCHAR2(30) 10 SQL>alter table cz add constraint cz_unique unique(c1,c10,c20) exceptions into exceptions; 11 * 12 ERROR at line 1: 13 ORA-02299: cannot validate (TEST.CZ_UNIQUE) - duplicate keys found 14 SQL>create table dups as select * from cz where rowid in (select row_id from exceptions); 15 Table created. 16 SQL>select * from dups; 17 C1 C10 C20 18 ---------- ---------- --- 19 2 dsf 20 2 dsf 21 2 dsf 22 2 dsf 23 3 che 24 2 dsf 25 2 dsf 26 2 dsf 27 2 dsf 28 3 che 29 3 che 30 3 che 31 3 che 32 4 dff 33 4 dff 34 4 dff 35 rows selected. 36 SQL>select row_id from exceptions; 37 ROW_ID 38 ---------- ---------- --- 39 AAAHD/AAIAAAADSAAA 40 AAAHD/AAIAAAADSAAB 41 AAAHD/AAIAAAADSAAC 42 AAAHD/AAIAAAADSAAF 43 AAAHD/AAIAAAADSAAH 44 AAAHD/AAIAAAADSAAI 45 AAAHD/AAIAAAADSAAG 46 AAAHD/AAIAAAADSAAD 47 AAAHD/AAIAAAADSAAE 48 AAAHD/AAIAAAADSAAJ 49 AAAHD/AAIAAAADSAAK 50 AAAHD/AAIAAAADSAAL 51 AAAHD/AAIAAAADSAAM 52 AAAHD/AAIAAAADSAAN 53 AAAHD/AAIAAAADSAAO 54 AAAHD/AAIAAAADSAAP 55 rows selected. 56 SQL>delete from cz where rowid in ( select row_id from exceptions); 57 rows deleted. 58 SQL>insert into cz select distinct * from dups; 59 rows created. 60 SQL>select *from cz; 61 C1 C10 C20 62 ---------- ---------- --- 63 2 dsf 64 3 che 65 4 dff 66 5 err 67 3 dar 68 1 wee 69 2 zxc 70 rows selected.
从结果里可以看到重复记录已经删除。