如何明智地将列增加一组

Question

我有一张表叫productLocation，它的数据结构如下，SQLFiddle

+--------------+-------------+-----------+
| FkLocationId | FkProductId | SortValue |
+--------------+-------------+-----------+
| 1            | 100         | 0         |
+--------------+-------------+-----------+
| 1            | 101         | 0         |
+--------------+-------------+-----------+
| 1            | 102         | 0         |
+--------------+-------------+-----------+
| 1            | 103         | 2         |
+--------------+-------------+-----------+
| 1            | 104         | 1         |
+--------------+-------------+-----------+
| 1            | 105         | 3         |
+--------------+-------------+-----------+
| 2            | 100         | 0         |
+--------------+-------------+-----------+
| 2            | 101         | 0         |
+--------------+-------------+-----------+
| 2            | 102         | 0         |
+--------------+-------------+-----------+
| 2            | 103         | 1         |
+--------------+-------------+-----------+
| 2            | 104         | 3         |
+--------------+-------------+-----------+
| 2            | 105         | 2         |
+--------------+-------------+-----------+
| 3            | 100         | 0         |
+--------------+-------------+-----------+
| 3            | 101         | 0         |
+--------------+-------------+-----------+
| 3            | 102         | 0         |
+--------------+-------------+-----------+
| 3            | 103         | 1         |
+--------------+-------------+-----------+
| 3            | 104         | 2         |
+--------------+-------------+-----------+

每个位置都有自己的产品排序顺序。但有些产品有 0 作为SortValue 现在我需要编写一个查询来将SortValue 更新为，

如果我考虑一个位置，FkLocationId = 1

+--------------+-------------+-----------+
| FkLocationId | FkProductId | SortValue |
+--------------+-------------+-----------+
| 1            | 100         | 0         |
+--------------+-------------+-----------+
| 1            | 101         | 0         |
+--------------+-------------+-----------+
| 1            | 102         | 0         |
+--------------+-------------+-----------+
| 1            | 103         | 2         |
+--------------+-------------+-----------+
| 1            | 104         | 1         |
+--------------+-------------+-----------+
| 1            | 105         | 3         |
+--------------+-------------+-----------+

在上面的数据表中，可以看到FkProductId = 100,101,102有0作为SortValue。我需要按FkProductId 降序更新其sortValue。我需要更新它的样子

+--------------+-------------+-----------+
| FkLocationId | FkProductId | SortValue |
+--------------+-------------+-----------+
| 1            | 100         | 3         |
+--------------+-------------+-----------+
| 1            | 101         | 2         |
+--------------+-------------+-----------+
| 1            | 102         | 1         |
+--------------+-------------+-----------+

并且还将之前的SortValue一一更新。

那么完整的输出应该是，

+--------------+-------------+-----------+
| FkLocationId | FkProductId | SortValue |
+--------------+-------------+-----------+
| 1            | 100         | 3         |
+--------------+-------------+-----------+
| 1            | 101         | 2         |
+--------------+-------------+-----------+
| 1            | 102         | 1         |
+--------------+-------------+-----------+
| 1            | 103         | 5         |
+--------------+-------------+-----------+
| 1            | 104         | 4         |
+--------------+-------------+-----------+
| 1            | 105         | 6         |
+--------------+-------------+-----------+

这可能吗？我真的很困惑，请帮我解决这个问题。谢谢

更新：

假设我有另一个表作为产品。其数据结构如下，

+-----------+-----------+
| ProdcutId | SortValue |
+-----------+-----------+
| 100       | 0         |
+-----------+-----------+
| 101       | 0         |
+-----------+-----------+
| 107       | 0         |
+-----------+-----------+
| 108       | 1         |
+-----------+-----------+
| 109       | 2         |
+-----------+-----------+
| 110       | 6         |
+-----------+-----------+
| 111       | 5         |
+-----------+-----------+
| 112       | 4         |
+-----------+-----------+
| 113       | 3         |
+-----------+-----------+

我也需要为这张桌子做同样的事情，我该怎么做

预期输出：

+-----------+-----------+
| ProdcutId | SortValue |
+-----------+-----------+
| 100       | 3         |
+-----------+-----------+
| 101       | 2         |
+-----------+-----------+
| 107       | 1         |
+-----------+-----------+
| 108       | 4         |
+-----------+-----------+
| 109       | 5         |
+-----------+-----------+
| 110       | 9         |
+-----------+-----------+
| 111       | 8         |
+-----------+-----------+
| 112       | 7         |
+-----------+-----------+
| 113       | 6         |
+-----------+-----------+

Answer 1

以下是一个示例查询，您可以使用 CTE 检查并使用相同的逻辑进行更新过程-

DEMO HERE

重要提示：更新是一个有风险的过程，您应该先尝试使用您的测试数据。

WITH CTE
AS
(
    SELECT *,
    (
        SELECT COUNT(*) 
        FROM #Temp B 
        WHERE B.FkLocationId = A.FkLocationId
        AND SortValue = 0
    ) Zero_Count,
    -- The above Zero count is a simple count of rows with 0
    -- for a specific FkLocationId. This is for adding to other 
    -- rows where there are already value in column SortValue. The logic
    -- is simple, the number of row with 0 in column SortValue, 
    -- should be add to existing value where there is value not equal 0.
    ROW_NUMBER() OVER (PARTITION BY FkLocationId ORDER BY SortValue ASC, FkProductId DESC) RN
    -- ROW_NUMBER is simply creating the Number you should replace 0 by.
    -- As you are looking for 1 to start for the MAX FkProductId with 0,
    -- I applied DESC order on that column
    FROM #Temp A
) 

UPDATE A
SET A.SortValue = 
CASE 
    WHEN A.SortValue = 0 THEN B.RN 
    -- RN is created in such way, so that you can directly 
    -- Replace your value in column SortValue by RN if SortValue = 0 
    ELSE A.SortValue + B.Zero_Count
    -- IF SortValue contains already value > 0, 
    -- You need to just increment the value with Number 
    -- of rows is there with value 0. I have calculated that value
    -- by a sub query in the CTE you can check.
END
FROM  #Temp A
INNER JOIN CTE B ON A.FkLocationId = B.FkLocationId
AND A.FkProductId = B.FkProductId


SELECT * FROM #Temp

只需在我使用过 #temp 的地方使用您的“table_name”

Answer 2

with cte as
(select FkLocationId, FkProductId, SortValue,
ROW_NUMBER() over(partition by FkLocationId order by sortvalue,FkProductId desc) new_sort_value
from productLocation where FkLocationId =1  --and SortValue =0
)

update pl SET
pl.SortValue = c.new_sort_value
from productLocation pl inner join cte c
on pl.FkLocationId = c.FkLocationId and pl.FkProductId = c.FkProductId

Select * from productLocation where FkLocationId =1 order by FkProductId

我采用的方法的核心是基于row_number的窗口函数，它将fkproductid上的数据按sortvalue和fkproductid降序排序。

涉及到 2 个查询，虽然可以做单个查询，但为简单起见，我使用了公用表表达式 (cte)

从查询一派生的 CTE 在第二个更新查询中用作连接表。

我已经为自我探索留下了很多东西，请记住必须根据集合进行思考。