Question

表名= t1

ID | FileID | Status
1  |  3     | NO
3  |  5     | YES
4  |  3     | NO

表名= T2

ID | FileID | From stage| Date       | Remark
1  |  3     | 14        | 20/03/2015 | file is submitted
2  |  3     | 14        | 21/03/2015 | file is not completed
3  |  5     | 14        | 20/03/2015 | file is submitted

仅当文件ID 没有“状态

时，我才想要最新日期条目的T2表格详细信息

使用以下查询

select *
from (select distinct(t1.FileID)
      from t1
      where t1.FileID not in (select FileID from t1 where Status='YES')
     ) c1 ,t2 
where t2.FileID=c1.FileID and t2.FromStageID='14
order by c1.FileID

获得结果

FIleID | ID | FileID | From stage| Date       | Remark
3      | 1  |  3     | 14        | 20/03/2015 | file is submitted
3      | 2  |  3     | 14        | 21/03/2015 | file is not completed

我想要什么

FIleID | ID | FileID | From stage| Date       | Remark
3      | 2  |  3     | 14        | 21/03/2015 | file is not completed

Answer 1

传统上是SQL-Server的cross apply版本：

SELECT 
    tt.*
FROM
  T1
cross apply (
SELECT TOP 1 *
from t2
WHERE t2.fileId = t1.FileId
ORDER BY t2.date DESC
) TT
WHERE T1.status = 'YES'

这将产生最佳性能，同时更难理解并需要学习新的操作员（如果还没有）

编辑：虽然下面的查询产生了更糟糕的执行计划（仍然不是非常糟糕），但更容易阅读。对于具有＆lt; 100k记录的表：

，仍然可以很好地执行

SELECT 
    t2.FileId, T2.FromStage, T2.Date, T2.remark
FROM
  T1 JOIN T2 ON T2.FileId = T1.FileId
WHERE T1.status = 'YES' AND 
    T2.Date = (SELECT MAX(T2.Date) FROM T2 WHERE T1.FileId = T2.FileId)

Answer 2

如果你想要最新的东西，请考虑row_number()，然后根据条件获取第一个。

以下使用left join代替not in来查找fileid中没有相应匹配项的t1。这样更安全，因为not in对NULL值的行为非直观。

此外，永远不要在from子句中使用逗号。这是一个简单的规则。始终使用明确的join语法。

select t.*
from (select t2.*,
             row_number() over (partition by fileid order by date desc) as seqnum
      from t2 left join
           t1
           on t2.fileid = t1.fileid and t1.status = 'YES'
      where t1.fileid is null
     ) t
where seqnum = 1;

Answer 3

另一个原因，您可以使用CommonTableExpression和ROW_NUMBER()来实现此结果：

试试这个：Fiddle Demo Here

create table t1(
ID int,
  FieldId int,
  [Status] varchar(50)
)
insert into t1 values(1,3,'NO'),(3,5,'YES'),(4,3,'NO')

create table T2(
Id int,
  FieldId int,
  From_Stage int,
 date1 date,
  Remarks varchar(50)
)
insert into T2 values(1,3,14,'03/20/2015','file is submitted'),
(2,3,14,'03/21/2015','file is not completed'),
(3,5,14,'03/21/2015','file is submitted')

查询：

;with cte as
(
select t1.FieldId,t2.ID,t2.From_Stage,t2.date1,t2.REmarks,
ROW_NUMBER() Over(PARTITION  by t2.FieldId order by t2.date1 desc) as rk
from t1,T2  t2 where t1.FieldId=t2.FieldId and t1.Status!='YES'
)

select * from cte where rk=1

仅加入最大日期条目

3 个答案: