注,仅适用于MYSQL8以上,不然无法使用窗口函数

示例数据如下:

lot_id	file_name	file_time	type
13629111	13629111_20231101_165326.csv	2023-11-01 16:53:18	1
13629111	13629111_20231110_095855.csv	2023-11-10 09:58:31	2
13629111	13629111_20231110_084255.csv	2023-11-10 08:41:56	2
13629111	13629111_20231110_093731.csv	2023-11-10 08:41:56	2
1234	1234_20231102_134030.csv	2023-11-02 13:40:21	1
1234	1234_20231102_171750.csv	2023-11-02 17:17:31	1
1234	1234_20231106_122026.csv	2023-11-06 12:18:36	2

若想取出数据,按lot_id和type分组,依据file_time最新的那条数据,获取到file_name. 即取出如下数据

lot_id	file_name	file_time	type
1234	1234_20231102_171750.csv	2023-11-02 17:17:00	1
1234	1234_20231106_122026.csv	2023-11-06 12:18:00	2
13629111	13629111_20231101_165326.csv	2023-11-01 16:53:00	1
13629111	13629111_20231110_095855.csv	2023-11-10 09:58:00	2

可使用如下sql完成

-- 首先按照需求排序记得行号
select lot_id,file_name,file_time,`type`,
ROW_NUMBER() over(PARTITION BY lot_id,`type` order by file_time desc) as `rank` FROM test;-- 完整sql,取出行号为1的数据,即最新的数据
select * from 
(
select lot_id,file_name,file_time,`type`,
ROW_NUMBER() over(PARTITION BY lot_id,`type` order by file_time desc) as `rank` FROM test
)a where a.`rank` = 1

其中

• PARTITION BY lot_id,type. 语法类似与group by,指定分组要求.
• order by file_time desc即指定组内按什么排序.
• ROW_NUMBER()为窗口函数,记为组内的行号
  注意:窗口函数不会改变原有的数据行数,可以认为是查看分析数据的一种方式,即在原数据上加上一列

上半句片段执行结果如下,rank表示按照需求的排序情况

lot_id	file_name	file_time	type	rank
13629111	13629111_20231101_165326.csv	2023-11-01 16:53:18	1	1
13629111	13629111_20231110_095855.csv	2023-11-10 09:58:31	2	1
13629111	13629111_20231110_084255.csv	2023-11-10 08:41:56	2	2
13629111	13629111_20231110_093731.csv	2023-11-10 08:41:56	2	3
1234	1234_20231102_134030.csv	2023-11-02 13:40:21	1	2
1234	1234_20231102_171750.csv	2023-11-02 17:17:31	1	1
1234	1234_20231106_122026.csv	2023-11-06 12:18:36	2	1