SQL删除重复数据，只保留一行

概述

在sql的使用中，咱们老是碰到须要删除重复数据的状况，可是又不能所有删除完，必需要保留至少一个重复的数据。重复的记录根据两个字段a2,a3判断（实际使用中能够拓展为多个）

实现过程

1. 建立表A，数据以下

a1	a2	a3
1	1	1
2	1	2
3	2	2
4	2	2
5	3	3
6	2	2

2. 选择重复的行：

select *,count(*) 
 from A group by a2,a3 
 having count(*)>1;

a1	a2	a3	count(*)
3	2	2	3

使用in来找到咱们想要的ID

SELECT *
FROM A 
WHERE (a2,a3) IN 
(SELECT A.`a2`,A.`a3` 
FROM A 
GROUP BY A.`a2`,A.`a3` 
HAVING COUNT(*)>1)

a1	a2	a3
3	2	2
4	2	2
6	2	2

那么后面就很好办了：

3. 选出要删除的值：

SELECT * 
FROM A 
WHERE (a2, a3) IN 
  (SELECT `a2`,`a3` 
  FROM A 
  GROUP BY A.`a2`,A.`a3` 
  HAVING COUNT(*) > 1) 
  AND a1 NOT IN 
  (SELECT MIN(a1) 
  FROM A 
  GROUP BY A.`a2`,A.`a3` 
  HAVING COUNT(*) > 1) ;

-- 结果是保留a1最小的值，其余选项所有选出，
-- 请注意此时并非将Select 改成delete就能够了，若是你直接这样改的话，会报以下错误：
-- You can't specify target table 'A' for update in FROM clause
-- 该错误提示你，不能先select出同一表中的某些值，再update这个表(在同一语句中）。因此要稍微修改一下。

4. 删除值

sql语句以下：

//建立中间表
CREATE TABLE F(a1 INTEGER,a2 INTEGER,a3 INTEGER);
//将要删除的数据插入中间表
INSERT INTO F (
SELECT *  
FROM A 
WHERE (a2, a3) IN (SELECT `a2`,`a3` 
  FROM A GROUP BY A.`a2`,A.`a3` 
  HAVING COUNT(*) > 1) 
  AND a1 NOT IN 
  (SELECT MIN(a1) FROM A 
  GROUP BY A.`a2`,A.`a3` 
  HAVING COUNT(*) > 1)) ;
//删除中间表
DELETE FROM A WHERE a1 IN (SELECT a1 FROM F);
SELECT *FROM A;

a1	a2	a3
1	1	1
2	1	2
3	2	2
5	3	3

备注

若是说不用保留一行数据的话那么就简单多了，只须要一个很简单的sql语句：

DELETE FROM A WHERE (a2,a3) IN (SELECT a2,a3 FROM A GROUP BY a2,a3 HAVING COUNT(*)>1)