当组中的任何值为0时，将组内的所有值设置为0

Question

如果组中的任何值为0，我想扫描表并将组内的值重新赋值为0。通过搜索Group By,Partition和Any的各种组合来找出解决方案。

我开始看起来像

CREATE TABLE #QP 
(
    [Method] VARCHAR(1), 
    [Station] VARCHAR(1), 
    [Instrument] VARCHAR(20), 
    [LastAnalysis] DATE, 
    [DaysPassed] INT 
) 

INSERT INTO #QP 
(Method, Station, Instrument, LastAnalysis, DaysPassed) 
VALUES 
('A', 1, 'Polaris', '2016-07-19', 21), 
('B', 1, 'Polaris', '2016-08-04', 5), 
('C', 1, 'Polaris', '2016-07-31', 9), 
('A', 2, 'Polaris', '2016-07-31', 9), 
('B', 2, 'Polaris', '2016-08-09', 0), 
('C', 2, 'Polaris', '2016-07-23', 17), 
('A', 3, 'Polaris', '2016-08-09', 0), 
('B', 3, 'Polaris', '2016-07-27', 13), 
('C', 3, 'Polaris', '2016-07-19', 21) 

而且我想结果显示为数据（包括为便于解释换行符）

Method Station Instrument LastAnalysis DaysPassed Weight 
A    1 Polaris 2016-07-19   21  21 
B    1 Polaris 2016-08-04   5  5 
C    1 Polaris 2016-07-31   9  6 

A    2 Polaris 2016-07-31   9  0 
B    2 Polaris 2016-08-09   0  0 
C    2 Polaris 2016-07-23   17  0 

A    3 Polaris 2016-08-09   0  0 
B    3 Polaris 2016-07-27   13  0 
C    3 Polaris 2016-07-19   21  0 

我已经得到的最接近的到目前为止使用的是

SELECT *, 
    CASE WHEN 0 = ANY(SELECT DaysPassed FROM #QP) THEN 0 ELSE DaysPassed END AS [Weight] 
FROM #QP 
WHERE Instrument = 'Polaris' 
ORDER BY Station, Method 

但是这设定了每个值在Weight co lumn为0，Station组中的值应保持原样。

如果这有一个答案，我很想知道正确的搜索条件来找到它。

Answer 1

我认为使用min()窗口函数将工作。试试看：

SELECT *, 
     case when min(DaysPassed) over (partition by station) = 0 then 0 else DaysPassed end as [Weight] 
FROM #QP 
WHERE Instrument = 'Polaris' 
ORDER BY Station, Method 

Answer 2

SELECT * 
     ,Weight= Sign(min(DaysPassed) over (Partition By Station)) * DaysPassed 
FROM #QP 
WHERE Instrument = 'Polaris' 
ORDER BY Station, Method 

返回

Method Station Instrument LastAnalysis DaysPassed Weight 
A  1  Polaris  2016-07-19  21   21 
B  1  Polaris  2016-08-04  5   5 
C  1  Polaris  2016-07-31  9   9 
A  2  Polaris  2016-07-31  9   0 
B  2  Polaris  2016-08-09  0   0 
C  2  Polaris  2016-07-23  17   0 
A  3  Polaris  2016-08-09  0   0 
B  3  Polaris  2016-07-27  13   0 
C  3  Polaris  2016-07-19  21   0