MySQL：获取用户在给定日期的最少记录

Question

我有一张活动表，每个活动都有负责人。每天可能有多个此类事件，但我需要在给定日期为每个用户的 first 查询记录。

例如，如果我有以下事件表：

+----------+-------------+---------------------+
| event_id | director_id | event_start         |
+----------+-------------+---------------------+
|        1 |         111 | 2015-04-27 10:00:00 |
+----------+-------------+---------------------+
|        2 |         222 | 2015-04-27 11:00:00 |
+----------+-------------+---------------------+
|        3 |         333 | 2015-04-27 12:00:00 |
+----------+-------------+---------------------+
|        4 |         111 | 2015-04-27 13:00:00 |
+----------+-------------+---------------------+
|        5 |         222 | 2015-04-27 09:00:00 |
+----------+-------------+---------------------+

我希望返回以下内容：

+----------+-------------+---------------------+
| event_id | director_id | event_start         |
+----------+-------------+---------------------+
|        1 |         111 | 2015-04-27 10:00:00 |
+----------+-------------+---------------------+
|        5 |         222 | 2015-04-27 09:00:00 |
+----------+-------------+---------------------+
|        3 |         333 | 2015-04-27 12:00:00 |
+----------+-------------+---------------------+

我认为像下面这样的查询会起作用，但事实证明 MySQL 不支持 WHERE 子句中的 MIN (simple SQL query giving Invalid use of group function)：

SELECT
        event_id, director_id, MIN(event_start) AS event_start 
    FROM events 
    WHERE MIN(event_start) >= '2015-04-27 00:00:00' 
        AND MIN(event_start) < '2015-04-28 00:00:00'
    GROUP BY director_id;

我怎样才能以最有效的方式做到这一点？我的events 表可能很容易有 10,000-100,000 条记录。

Answer 1

您可以通过与您类似的查询获得每天的最短活动时间：

SELECT director_id, date(event_start) as dte, MIN(event_start) AS event_start 
FROM events e
GROUP BY director_id, date(event_start);

然后您可以将其用作子查询以从行中获取所有其他信息：

select e.*
from events e join
     (SELECT e.director_id, date(e.event_start) as dte, MIN(e.event_start) AS event_start 
      FROM events e
      GROUP BY e.director_id, date(e.event_start)
     ) ee
     on e.event_start = ee.event_start -- note, this has both the date and time;

如果要将结果限制为一天，可以将where 子句放在子查询中。

Answer 2

您不能在查询的 where 子句中使用 group by / 聚合函数。做你想做的一种方法是像这样使用左连接：

select e1.*
  from events e1
    left join events e2
      on e1.director_id = e2.director_id
        and e1.event_start > e2.event_start
        and date(e1.event_start) = date(e2.event_start)
  where e2.director_id is null

fiddle here

如果您有跨 (director_id, event_start) 的索引，则性能可能会提高

您还可以通过更改and date(e1.event_start) = date(e2.event_start) 来进一步限制结果大小以检查特定日期。

Answer 3

你可以试试这个：

SELECT
  e1.* 
FROM events AS e1
  INNER JOIN ( SELECT director_id, MIN(event_start) AS `eventStart` 
               FROM `events` GROUP BY director_id ) AS e2
  ON e1.director_id = e2.director_id
  AND e1.event_start = e2.eventStart
WHERE e2.eventStart >= '2015-04-27 00:00:00' 
  AND e2.eventStart < '2015-04-28 00:00:00';

这里是sqlfiddle。