代码之家 › 专栏 › 技术社区 › Mar1009

如何根据需要从SQL Server中检索数据?

ssms-2014 tsql sql-server-2008 sql-server sql

Mar1009 · 技术社区 · 6 年前

我有一张这样的桌子:

  CustName    Country    RecordedTime
 ---------------------------------------------
  Alex        Australia  2018-Jun-01 08:00 AM
  Alex        China      2018-Jun-01 10:00 AM
  Alex        Japan      2018-Jun-01 11:00 AM
  John        Australia  2018-Jun-01 08:00 AM
  John        China      2018-Jun-02 08:00 AM
  Bob         Australia  2018-Jun-02 09:00 AM
  Bob         Brazil     2018-Jun-03 09:50 AM

如果该记录在系统中是全新的,则它应在给定日期的“审核”和“历史记录”字段(结果集中的两个附加字段)中显示“添加”和“新建”。

如果当天编辑了两次记录,则该记录应在审核字段中分别显示两个“添加”和“更改”条目,在给定日期的历史状态字段中分别显示“在”和“当前”条目。

例如,我的结果应该是这样出现的;

当我通过输入日期为2018-Jun-01时,输出应如下:

   CustName    Country    RecordedTime           Audit    History
  ----------------------------------------------------------------
   Alex        Australia  2018-Jun-01 08:00 AM   ADD      NEW
   Alex        China      2018-Jun-01 10:00 AM   CHANGE   BEFORE
   Alex        Japan      2018-Jun-01 11:00 AM   CHANGE   CURRENT
   John        Australia  2018-Jun-01 08:00 AM   ADD      NEW

当我通过输入日期为2018-Jun-02时,输出应如下:

   CustName    Country    RecordedTime           Audit    History
  -----------------------------------------------------------------
   John        China      2018-Jun-02 08:00 AM   CHANGE   CURRENT
   Bob         Australia  2018-Jun-02 09:00 AM   ADD      NEW

当我通过输入日期为2018-Jun-02时,输出应如下:

   CustName    Country    RecordedTime           Audit    History
  ----------------------------------------------------------------
   Bob         Brazil     2018-Jun-03 09:50 AM   CHANGE   CURRENT

我尝试了很多方法,但我仍然错过了一些实现这一点的场景。有人能解释一下吗?

3 回复 | 直到 6 年前

DhruvJoshi 6 年前

一种方法是通过下面这样的CTE,我们有row_number()函数来双向跟踪序列。

参见 live demo

; with cte as 
(
select *, rn= row_number() over(partition by CustName order by RecordedTime),
rn2=row_number() over(partition by CustName order by RecordedTime desc)
from records
    )
, cte2 as
(
select *, audit='New', History='Change' from cte where rn=1
    union 
select *, audit='Change', History='Current' from cte where rn2=1 and rn<>1
    union
select *, audit='Change', History='before' from cte where rn>1 and rn2<>1
)

select 
    CustName,
    Country,
    RecordedTime,
    audit,
    History
from cte2
order by  CustName,RecordedTime

D-Shih 6 年前

我只会用 case 表达。

select t.*,
       (case when seqnum = 1 then 'ADD' else 'CHANGE' end) as audit,
       (case when seqnum = 1 then 'NEW'
             when seqnum_day = 1 then 'CURRENT'
             else 'BEFORE'
        end) as history
from (select t.*,
             row_number() over (partition by custname order by recordedtime) as seqnum,
             row_number() over (partition by custname, cast(recordedtime as date) order by recordedtime desc) as seqnum_day
      from t
     ) t;

sqlfiddle: http://sqlfiddle.com/#!18/43c08/27

D-Shih 6 年前

你可以试试。

CASE WHEN 和 RANK 具有 Windows函数

;WITH CTE (CustName,Country,RecordedTime,rn) AS(
  SELECT *,RANK() OVER(PARTITION BY CustName ORDER BY RecordedTime) rn
  FROM T
)
SELECT t.*,
       (CASE WHEN rn = 1 then 'ADD' ELSE 'CHANGE' END) 'Audit',
       (CASE 
          WHEN rn = 1  then 'NEW'
          WHEN t2.mRn = rn then 'CURRENT' 
       ELSE 'BEFORE' END) 'History'
FROM CTE t LEFT JOIN  (
  SELECT MAX(rn) mRn,CustName  FROM CTE GROUP BY CustName
) t2 on t2.mRn = t.rn and t2.CustName = t.CustName
WHERE CONVERT(char(10), RecordedTime,126) = '2018-06-02'

sqlfiddle: http://sqlfiddle.com/#!18/43c08/26