我正在尝试为每个唯一的用户ID选择最近的行,其中pid = 50且active = 1.我无法弄明白。
这是一个示例表
+-----+----------+-------+-----------------------+---------+
| id | userid | pid | start_date | active |
+-----+----------+-------+-----------------------+---------+
| 1 | 4 | 50 | 2015-05-15 12:00:00 | 1 |
| 2 | 4 | 50 | 2015-05-16 12:00:00 | 1 |
| 3 | 4 | 50 | 2015-05-17 12:00:00 | 0 |
| 4 | 4 | 51 | 2015-06-29 12:00:00 | 1 |
| 5 | 4 | 51 | 2015-06-30 12:00:00 | 1 |
| 6 | 5 | 50 | 2015-07-05 12:00:00 | 1 |
| 7 | 5 | 50 | 2015-07-06 12:00:00 | 1 |
| 8 | 5 | 51 | 2015-07-08 12:00:00 | 1 |
+-----+----------+-------+-----------------------+---------+
期望的结果
+-----+----------+-------+-----------------------+---------+
| id | userid | pid | start_date | active |
+-----+----------+-------+-----------------------+---------+
| 2 | 4 | 50 | 2015-05-16 12:00:00 | 1 |
| 7 | 5 | 50 | 2015-07-06 12:00:00 | 1 |
+-----+----------+-------+-----------------------+---------+
我尝试了很多东西,这是我得到的最接近但不幸的是它并没有放弃。
SELECT *
FROM mytable t1
WHERE
(
SELECT COUNT(*)
FROM mytable t2
WHERE
t1.userid = t2.userid
AND t1.start_date < t2.start_date
) < 1
AND pid = 50
AND active = 1
ORDER BY start_date DESC
答案 0 :(得分:2)
<强>计划强>
- 获取用户ID的最后一条记录分组,其中pid为50且处于活动状态
- 内部联接到mytable以获取与上次
相关联的记录信息 醇>
<强>查询强>
select
my.*
from
(
select userid, pid, active, max(start_date) as lst
from mytable
where pid = 50
and active = 1
group by userid, pid, active
) maxd
inner join mytable my
on maxd.userid = my.userid
and maxd.pid = my.pid
and maxd.active = my.active
and maxd.lst = my.start_date
;
<强>输出强>
+----+--------+-----+------------------------+--------+
| id | userid | pid | start_date | active |
+----+--------+-----+------------------------+--------+
| 2 | 4 | 50 | May, 16 2015 12:00:00 | 1 |
| 7 | 5 | 50 | July, 06 2015 12:00:00 | 1 |
+----+--------+-----+------------------------+--------+
备注强>
正如@Strawberry建议的那样,更新后也加入了pid和active。这将避免记录不活跃或不是pid 50的可能性,但也会提供完全相同的日期。