Question

我有以下数据库（小例子）：

ID | username | action  | points
1  | matt     | login   | 3
2  | john     | comment | 6
3  | john     | login   | 6
4  | peter    | login   | 8
5  | peter    | login   | 8

我不确定如何选择和分组具有相同操作（=登录）和高于5的用户名（对于所有操作）。

预期结果：

username | COUNT | points(SUM)
peter    | 2     | 16

我尝试了子查询，但找不到合适的解决方案。你知道怎么做吗？非常感谢你的帮助！

对不起，我没有强调这样一个事实，即所有行动必须是login。因此，结果将没有john，因为有与此用户名相关的操作comment。

Answer 1

SELECT    username, COUNT(*) AS cnt, SUM(points) AS points
FROM      tableX AS t
GROUP BY  username
HAVING    COUNT(*) = COUNT(CASE WHEN action = 'login' THEN action END) 
   AND    SUM(points) > 5 ;

或：

SELECT    username, COUNT(*) AS cnt, SUM(points) AS points
FROM      tableX AS t
WHERE     action = 'login'
  AND     NOT EXISTS
            ( SELECT    *
              FROM      tableX AS tt
              WHERE     tt.username = t.username
                AND     ( tt.action <> 'login' OR tt.action IS NULL )
            )
GROUP BY  username
HAVING    SUM(points) > 5 ;

但我认为，(username, login)上的索引和(username, points)上的索引，这将是效率最高的：

SELECT    username, COUNT(*) AS cnt, SUM(points) AS points
FROM      tableX AS t
GROUP BY  username
HAVING    MIN(action) = 'login'
   AND    MAX(action) = 'login' 
   AND    SUM(points) > 5 ;

Answer 2

您可以使用聚合和聪明的having子句来执行此操作：

select username, sum(points), count(*) 
from tbl
group by username
having sum(case when points <= 5 then 1 else 0 end) = 0 and   -- count number with points < 5
       max(action) = min(action) and  -- all the actions are the same and
       min(action) = 'login'

Answer 3

尝试此查询

select username, action, sum(points), count(*) 
from 
tbl
group by username, action 
having sum(if (points<=5, 1, 0)) =0  and count(*) >=2

的 SQL FIDDLE ：

| USERNAME | ACTION | SUM(POINTS) | COUNT(*) | ---------------------------------------------- | peter | login | 16 | 2 |

MySQL：仅选择一些分组的行

3 个答案:

的 SQL FIDDLE ：