Question

ID   Date1   T1   Date2   T2   Date3   T3   Date4   T4
1   DT11     Y    ----------NULL-----------------------  
2   DT11     N    DT22    Y    ----------NULL----------  
3   DT31     N    DT32    Y    DT33    Y    --NULL-----  
4   ----NULL---   DT42    Y    ---NULL---   DT44    Y
5   DT51     Y    ---NULL--    DT53    Y    --NULL-----

我有一个Table1表格，上面有结构。

要求：

检查（T1-T5）中哪一个有“Y”
如果有多个'Y'，请比较日期以获取最近的日期
使用最新日期的日期字段名称或部分名称在新字段中设置值。

所以，o / p应该是

ID  Most-recent-Date-Flag
---------------------------------------------------    
1  "Date1 or T1"  -> Since all others are null
2  "Date2 or T2"  -> T2 has 'Y'
3  "Date3 or T3"  -> one 'N' and multiple 'Y', DT33 is most recent
4  "Date4 or T4"  -> multiple 'Y', DT44 is most recent
5  "Date5 or T5"  -> multiple 'Y', DT53 is most recent

我已设法创建查询以比较日期。它如下：

    SELECT [ID]
    , MAX(UNPIVOT_DATE) AS MOST_RECENT_DATE
    FROM 
    (
    SELECT
    [ID]
    ,[Date1]
    ,[Date2]
    ,[Date3]
    ,[Date4]
    FROM TABLE1
    )SRC_TABLE
    UNPIVOT
    (
    UNPIVOT_DATE
    FOR DATES IN
    (
    [Date1]
    ,[Date2]
    ,[Date3]
    ,[Date4]
    )
    )UNPIVOT_TABLE
    GROUP BY [ID]

我被困在我检查（T1-T4）字段中的'Y'并获得比较后最近日期的列名称的位置。

希望这个解释有所帮助！

2013年8月4日
非常感谢@MarcinJuraszek。有用！！无论如何我还可以传递字段的列名（T1-T4）吗？我尝试了以下操作，但没有用。

SELECT t.[ID]
,MAX(t.Date) as MOST_RECENT_DATE
--,t.Column_Name 
FROM
(
    SELECT [ID]
    ,[Date1] as Date
    ,[T1] as T
--  ,COL_NAME(OBJECT_ID('Table1'), 2) AS Column_Name 
    FROM Table1
UNION ALL
    SELECT [ID]
    ,[Date2] as Date 
    ,[T2] as T 
--    ,COL_NAME(OBJECT_ID('Table1'), 4) AS Column_Name
    FROM Table1
  UNION ALL
    SELECT [ID] 
    ,[Date3] as Date
    ,[T3] as T 
--    ,COL_NAME(OBJECT_ID('Table1'), 6) AS Column_Name
    FROM Table1
  UNION ALL 
    SELECT [ID]
    ,[Date4] as Date
    ,[T4] as T 
--    ,COL_NAME(OBJECT_ID('Table1'), 8) AS Column_Name 
    FROM Table1
) t
WHERE t.T = 'Y'
GROUP BY t.[ID]
--, t.Column_Name
ORDER by [ID]

Answer 1

怎么样：

SELECT t.ID, MAX(t.Date) FROM
  (SELECT ID, Date1 as Date, T1 as T FROM Test
  UNION ALL
    SELECT ID, Date2 as Date, T2 as T FROM Test
  UNION ALL
    SELECT ID, Date3 as Date, T3 as T FROM Test
  UNION ALL 
    SELECT ID, Date4 as Date, T4 as T FROM Test) t
WHERE t.T = 'Y'
GROUP BY t.ID

<强> Simple Demo

Answer 2

除了GROUP BY + MAX()之外，您还可以使用CROSS APPLY (SELECT TOP 1)。这样您就可以返回其他数据以及最大日期：

SELECT
  t.*,
  x.Date,
  x.FlagName
FROM TABLE1 AS t
CROSS APPLY (
  SELECT TOP 1
    Date,
    FlagName
  FROM (
    SELECT Date1, T1, 'T1' UNION ALL
    SELECT Date2, T2, 'T2' UNION ALL
    SELECT Date3, T3, 'T3' UNION ALL
    SELECT Date4, T4, 'T4' UNION ALL
    SELECT Date5, T5, 'T5'
  ) AS s (Date, T, FlagName)
  WHERE T = 'Y'
  ORDER BY Date
) AS x
;

Answer 3

我最终得到了以下查询（偶然）并且它起作用.....

我创建了一个视图，如下所示

创建视图View_Test为选择 [ID]，当[T1] ='Y'然后[D1]的情况下，空的结尾为[D1]，当[T2] ='Y'然后[D2]的情况下，null结束为[D2]，当[T3] ='Y'然后[D3]的情况下，null结束为[D3]， [T4] ='Y'然后[D4]的情况，否则为[D4] 来自测试

然后我在一个univot查询中使用了视图，如下所示

select [ID], max(DATE_FIELD) as most_recent_date, max(Flag) as most_recent_flag
from View_Test
unpivot (
DATE_FIELD for Flag in 
(
[D1]
, [D2]
, [D3]
,[D4]
)
) as DATE_Unpivoted
group by [ID]
ORDER BY [ID]

当我执行上述查询时，“max（flag）”为我提供了具有最新日期的日期字段的名称。我仍然不知道它是如何工作的。关于“旗帜”的那一点对我来说仍然是不可理解的。但我偶然把它留在了查询中...从我尝试过但不起作用的其他查询中复制并粘贴：）

有人可以开导吗？：）

构造针对以下要求的T-SQL查询

3 个答案: