Mysql组由两列组成，其中条件和最大日期为第三列

Question

我有一个表有user_id，item_id和last_date作为列。最后一列类型是在datetime中。

我需要通过将列user_id和item_id分组来检索last_date，其中type_id就像foo一样非常导入我需要在结果中包含所有列，因为在mytable中有更多列。

这是mytable结构：

user_id  item_id      last_date         type_id
129678      2     2019-02-17 11:00:09     foo
129678      1     2019-02-17 11:00:15     foo
129678      2     2019-02-17 11:00:10     bar
129678      2     2019-02-17 11:00:10     bar
129678      2     2019-02-17 11:00:11     foo
129678      2     2019-02-17 11:00:15     foo
129678      1     2019-02-17 11:00:09     foo
129678      2     2019-02-17 11:00:14     bar
129678      2     2019-02-17 11:00:08     bar
129678      2     2019-02-17 11:00:11     foo
129678      2     2019-02-17 11:00:14     bar
129678      2     2019-02-17 11:00:10     foo
12a0d8      3     2019-02-17 11:00:08     foo
12a0d8      2     2019-02-17 11:00:12     foo
12a0d8      3     2019-02-17 11:00:08     bar
12a0d8      3     2019-02-17 11:00:12     bar
12a0d8      1     2019-02-17 11:00:10     foo
12a0d8      1     2019-02-17 11:00:11     bar
12a0d8      3     2019-02-17 11:00:14     foo
12a0d8      3     2019-02-17 11:00:12     foo
18ae98      2     2019-02-17 11:00:12     foo
18ae98      3     2019-02-17 11:00:07     bar
18ae98      1     2019-02-17 11:00:13     bar
18ae98      1     2019-02-17 11:00:14     foo
18ae98      2     2019-02-17 11:00:09     foo
18ae98      2     2019-02-17 11:00:13     foo
18ae98      3     2019-02-17 11:00:08     foo
18ae98      1     2019-02-17 11:00:12     foo
18ae98      3     2019-02-17 11:00:10     foo
18ae98      1     2019-02-17 11:00:12     foo

这是我为此目的尝试的查询：

SELECT * FROM mytable
  WHERE last_date IN (
    SELECT MAX(last_date)
    FROM mytable
    WHERE type_id LIKE 'foo'
    GROUP BY user_id, item_id
  )

期望的结果：

user_id  item_id      last_date         type_id
129678      1     2019-02-17 11:00:15     foo
129678      2     2019-02-17 11:00:15     foo
12a0d8      1     2019-02-17 11:00:10     foo
12a0d8      2     2019-02-17 11:00:12     foo
12a0d8      3     2019-02-17 11:00:14     foo
18ae98      1     2019-02-17 11:00:14     foo
18ae98      2     2019-02-17 11:00:13     foo
18ae98      3     2019-02-17 11:00:10     foo

但结果很奇怪!!例如，返回user_id多个item_id具有相同的值...

编辑：

如果我使用此查询接缝工作正常但我必须指定间隔日期。

SELECT * FROM mytable
WHERE type_id LIKE 'foo'
AND last_date > date_sub(curdate(), interval 2 day)
GROUP BY user_id, item_id

Answer 1

以下是一种方式 -

    SELECT 
from       MyTable t1 
INNER JOIN 
           ( 
                    SELECT   user_id, 
                             item_id, 
                             max(last_date) AS lastdate 
                    FROM     MyTable 
                    WHERE    type_id='foo' 
                    GROUP BY user_id, 
                             item_id) t2 
ON         t1.user_id=t2.user_id 
AND        t1.item_id=t2.item_id 
AND        t1.last_date=t2.lastdate

Answer 2

使用相关子查询

   SELECT t1.* FROM mytable t1
   WHERE last_date = (
    SELECT MAX(last_date)
    FROM mytable t2
    WHERE t1.user_id=t2.user_id and t1.itemid=t2.itemid    
    and type_id LIKE 'foo'

  )

Answer 3

您可以简单地对您关注的2列进行分组，然后获取max（last_date）。您之前使用的子查询不会限制max（last_date），因为多个user_id和item_id组合可以应用于同一日期。

下面是使用您的示例数据进行的编写，该数据生成预期结果，而无需编写额外的嵌套查询。

declare @t table (
user_id nvarchar(15),
item_id int,
last_date datetime2,
type_id nvarchar(3)
);

insert into  @t (user_id, item_id, last_date, type_id)
values
('129678',      2,     '2019-02-17 11:00:09',     'foo'),
('129678',      1,     '2019-02-17 11:00:15',     'foo'),
('129678',      2,     '2019-02-17 11:00:10',     'bar'),
('129678',      2,     '2019-02-17 11:00:10',     'bar'),
('129678',      2,     '2019-02-17 11:00:11',     'foo'),
('129678',      2,     '2019-02-17 11:00:15',     'foo'),
('129678',      1,     '2019-02-17 11:00:09',     'foo'),
('129678',      2,     '2019-02-17 11:00:14',     'bar'),
('129678',      2,     '2019-02-17 11:00:08',     'bar'),
('129678',      2,     '2019-02-17 11:00:11',     'foo'),
('129678',      2,     '2019-02-17 11:00:14',     'bar'),
('129678',      2,     '2019-02-17 11:00:10',     'foo'),
('12a0d8',      3,     '2019-02-17 11:00:08',     'foo'),
('12a0d8',      2,     '2019-02-17 11:00:12',     'foo'),
('12a0d8',      3,     '2019-02-17 11:00:08',     'bar'),
('12a0d8',      3,     '2019-02-17 11:00:12',     'bar'),
('12a0d8',      1,     '2019-02-17 11:00:10',     'foo'),
('12a0d8',      1,     '2019-02-17 11:00:11',     'bar'),
('12a0d8',      3,     '2019-02-17 11:00:14',     'foo'),
('12a0d8',      3,     '2019-02-17 11:00:12',     'foo'),
('18ae98',      2,     '2019-02-17 11:00:12',     'foo'),
('18ae98',      3,     '2019-02-17 11:00:07',     'bar'),
('18ae98',      1,     '2019-02-17 11:00:13',     'bar'),
('18ae98',      1,     '2019-02-17 11:00:14',     'foo'),
('18ae98',      2,     '2019-02-17 11:00:09',     'foo'),
('18ae98',      2,     '2019-02-17 11:00:13',     'foo'),
('18ae98',      3,     '2019-02-17 11:00:08',     'foo'),
('18ae98',      1,     '2019-02-17 11:00:12',     'foo'),
('18ae98',      3,     '2019-02-17 11:00:10',     'foo'),
('18ae98',      1,     '2019-02-17 11:00:12',     'foo');

select * from (
select USER_ID, item_id, MAX(last_date) as last_date, type_id 
from @t where type_id = 'foo' group by user_id, item_id, type_id
) x
order by user_id, item_id;