在sql的使用中,我們總是碰到需要?jiǎng)h除重復(fù)數(shù)據(jù)的情況,但是又不能全部刪除完,必須要保留至少一個(gè)重復(fù)的數(shù)據(jù)。重復(fù)的記錄根據(jù)兩個(gè)字段a2,a3判斷(實(shí)際使用中可以拓展為多個(gè))
eg:表A
| a1 | a2 | a3 |
|---|---|---|
| 1 | 1 | 1 |
| 2 | 1 | 2 |
| 3 | 2 | 2 |
| 4 | 2 | 2 |
| 5 | 3 | 3 |
| 6 | 2 | 2 |
在上述的表中第三行和第四行重復(fù),我們要選擇一行刪除,流程如下:
- 選擇重復(fù)的行:
select *,count(*)
from A group by a2,a3
having count(*)>1;
結(jié)果如下:
| a1 | a2 | a3 | count(*) |
|---|---|---|---|
| 3 | 2 | 2 | 3 |
- 使用in來找到我們想要的ID
SELECT *
FROM A
WHERE (a2,a3) IN
(SELECT A.`a2`,A.`a3`
FROM A
GROUP BY A.`a2`,A.`a3`
HAVING COUNT(*)>1)
得到的結(jié)果如下:
|a1|a2|a3|
|---|---|
|3|2|2|
|4|2|2|
|6|2|2|
那么后面就很好辦了:
3.選出要?jiǎng)h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) ;
結(jié)果是保留a1最小的值,其他選項(xiàng)全部選出,
請(qǐng)注意此時(shí)并不是將Select 改為delete就可以了,如果你直接這樣子改的話,會(huì)報(bào)如下錯(cuò)誤:
You can't specify target table 'A' for update in FROM clause
該錯(cuò)誤提示你,不能先select出同一表中的某些值,再update這個(gè)表(在同一語句中)。所以要稍微修改一下。
- 刪除值
sql語句如下:
//創(chuàng)建中間表
CREATE TABLE F(a1 INTEGER,a2 INTEGER,a3 INTEGER);
//將要?jiǎng)h除的數(shù)據(jù)插入中間表
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)) ;
//刪除中間表
DELETE FROM A WHERE a1 IN (SELECT a1 FROM F);
SELECT *FROM A;
結(jié)果如下:
|a1|a2|a3|
|---|---|
|1|1|1|
|2|1|2|
|3|2|2|
|5|3|3|
完畢
注:如果說不用保留一行數(shù)據(jù)的話那么就簡(jiǎn)單多了,只需要一個(gè)很簡(jiǎn)單的sql語句:
DELETE FROM A WHERE (a2,a3) IN (SELECT a2,a3 FROM A GROUP BY a2,a3 HAVING COUNT(*)>1)