在實際應用中,很可能會碰到一些需要刪除某些欄位的重複記錄,我現在把我能想到的寫下來,望高手們補充。
1、
具體實現如下:

Table         Create Table                                           
————  ——————————————————–
users_groups  CREATE TABLE `users_groups` (                          
                `id` int(10) unsigned NOT NULL AUTO_INCREMENT,       
                `uid` int(11) NOT NULL,                              
                `gid` int(11) NOT NULL,                              
                PRIMARY KEY (`id`)                                   
              ) ENGINE=InnoDB AUTO_INCREMENT=15 DEFAULT CHARSET=utf8 

users_groups.txt內容:
1,11,502
2,107,502
3,100,503
4,110,501
5,112,501
6,104,502
7,100,502
8,100,501
9,102,501
10,104,502
11,100,502
12,100,501
13,102,501
14,110,501

mysql> load data infile `c:\users_groups.txt` into table users_groups fields
terminated by `,` lines terminated by `
`;

Query OK, 14 rows affected (0.05 sec)
Records: 14  Deleted: 0  Skipped: 0  Warnings: 0

mysql> select * from users_groups;

query result(14 records)

id uid gid
1 11 502
2 107 502
3 100 503
4 110 501
5 112 501
6 104 502
7 100 502
8 100 501
9 102 501
10 104 502
11 100 502
12 100 501
13 102 501
14 110 501

14 rows in set (0.00 sec)
根據一位兄弟的建議修改。
mysql> create temporary table tmp_wrap select * from users_groups group by uid having count(1) >= 1;
Query OK, 7 rows affected (0.11 sec)
Records: 7  Duplicates: 0  Warnings: 0

mysql> truncate table users_groups;
Query OK, 14 rows affected (0.03 sec)

mysql> insert into users_groups select * from tmp_wrap;
Query OK, 7 rows affected (0.03 sec)
Records: 7  Duplicates: 0  Warnings: 0

mysql> select * from users_groups;

query result(7 records)

id uid gid
1 11 502
2 107 502
3 100 503
4 110 501
5 112 501
6 104 502
9 102 501

mysql> drop table tmp_wrap;
Query OK, 0 rows affected (0.05 sec)

2、還有一個很精簡的辦法。
查詢重複的,並且除掉最小的那個。
delete users_groups as a from users_groups as a,
(
select *,min(id) from users_groups group by uid having count(1) > 1
) as b
 where a.uid = b.uid and a.id > b.id;
(7 row(s)affected)
(0 ms taken)
 

query result(7 records)

id uid gid
1 11 502
2 107 502
3 100 503
4 110 501
5 112 501
6 104 502
9 102 501
3、現在來看一下這兩個辦法的效率。
執行一下以下SQL 語句
create index f_uid on users_groups(uid);
explain select * from users_groups group by uid having count(1) > 1 union all
select * from users_groups group by uid having count(1) = 1;
explain select * from  users_groups as a,
(
select *,min(id) from users_groups group by uid having count(1) > 1
) as b
 where a.uid = b.uid and a.id > b.id;

query result(3 records)

id select_type table type possible_keys key key_len ref rows Extra
1 PRIMARY users_groups index (NULL) f_uid 4 (NULL) 14  
2 UNION users_groups index (NULL) f_uid 4 (NULL) 14  
(NULL) UNION RESULT <union1,2> ALL (NULL) (NULL) (NULL) (NULL) (NULL)  
 

query result(3 records)

id select_type table type possible_keys key key_len ref rows Extra
1 PRIMARY <derived2> ALL (NULL) (NULL) (NULL) (NULL) 4  
1 PRIMARY a ref PRIMARY,f_uid f_uid 4 b.uid 1 Using where
2 DERIVED users_groups index (NULL) f_uid 4 (NULL) 14  
 
 
很明顯的第二個比第一個掃描的函式要少。