Rextester

Question

我有OppTable

   ID           actualclosedate    status
--------------------------------------
    1            10-March    Won  
    2            11-June     Lost
    3            13-July     Open

我有阶段变化

--------------------------------------
   ID         stage       createdon   
--------------------------------------
    1         Propsect    1-Jan        
    1         Qualify     15-Jan       
    1         Develop     25-Jan       
    2         Qualify     9-Feb       
    2         Develop     7-March     
    3         Prospect    9-April 
    3         Prospect    9-April

如何填充基于多个条件生成的新列中的NULL值。例如：如果Opp是WON / LOST，则关闭日期如果Opp是Open，请将今日日期

预期结果：

--------------------------------------
   OppID      Stage       createdon   newcolumn 
--------------------------------------
    1         Propsect    1-Jan        15-Jan
    1         Qualify     15-Jan       25-Jan 
    1         Develop     25-Jan       10-March ---(closedate for Wonn Opp1)
    2         Qualify     9-Feb        7-March
    2         Develop     7-March      11-june  ---(closedate for lost Opp2)
    3         Prospect    9-April      10-April
    3         Qualify     10-April     4/2/2017----(todaydate for Open Opp3)

使用此查询，将获得我打开的实际关闭日期，但我想要今天日期

Select *
      ,NewColumn = COALESCE(Lead(A.createdon) over (Partition By A.ID Order By A.CreatedOn),B.actualclosedate)

 From  stagechange A
 Join  OppTable B on A.ID=B.ID

Answer 1

您可以尝试此查询：

WITH cte AS (
    SELECT a.ID, a.Stage, a.CreatedOn, b.CloseDate, b.Status,
           LEAD(a.CreatedOn) OVER (PARTITION BY a.ID ORDER BY a.CreatedOn) ld,
           ROW_NUMBER() OVER (PARTITION BY a.ID ORDER BY a.CreatedOn DESC) rn,
           DENSE_RANK() OVER (PARTITION BY a.ID ORDER BY a.CreatedOn DESC) dr
    FROM stagechange a
    INNER JOIN OppTable b
        ON a.ID = b.ID
)
SELECT t.ID, t.Stage, t.CreatedOn,
       CASE WHEN (t.rn <> 1 OR t.rn = t.dr) AND t.ld IS NOT NULL THEN t.ld
            WHEN t.Status <> 'Open' THEN t.CloseDate
            ELSE GETDATE()
        END AS newcolumn
FROM cte t
ORDER BY t.ID, t.CreatedOn;

我发现同时使用ROW_NUMBER()和DENSE_RANK()来处理给定ID可以有多个等效结束日期的边缘情况是必要的。在这种情况下，我们需要一些方法来确定哪一个是最老的。

<强>输出：

在这里演示：

如何使用多个条件为Lead窗口填充Null

1 个答案:

Rextester