TSQL：如果一行的某一列为NULL，则根据另一行将每隔一列设置为NULL

Question

this is my first question on stackoverflow so sorry if its not that understandable. I don't know if this is even possible.

This is a example of the TSQL query:

+--------+---------------------+---------------+-------+--------------------+
| Month  | Like_for_like_sales | Filial_number | Sales | PreviousYear_Sales |
+--------+---------------------+---------------+-------+--------------------+
| 201712 | 1                   |         12345 |   132 | 164                |
| 201801 | 1                   |         12345 |   143 | 167                |
| 201802 | 1                   |         12345 |   145 | 134                |
| 201803 | 1                   |         12345 |   165 | 176                |
| 201804 | 1                   |         12345 |   164 | 145                |
| 201805 | 1                   |         12345 |   186 | 145                |
| 201806 | 1                   |         12345 |   176 | 165                |
| 201807 | 1                   |         12345 |   187 | 121                |
| 201808 | 1                   |         12345 |   134 | 134                |
| 201809 | 1                   |         12345 |   123 | 165                |
| 201810 | 1                   |         12345 |   111 | 134                |
| 201811 | 1                   |         12345 |   143 | 156                |
| 201712 | NULL                |         54321 |   132 | NULL               |
| 201801 | NULL                |         54321 |   143 | NULL               |
| 201802 | NULL                |         54321 |   145 | NULL               |
| 201803 | NULL                |         54321 |   165 | NULL               |
| 201804 | NULL                |         54321 |   164 | NULL               |
| 201805 | NULL                |         54321 |   186 | NULL               |
| 201806 | NULL                |         54321 |   176 | 165                |
| 201807 | 1                   |         54321 |   187 | 121                |
| 201808 | 1                   |         54321 |   134 | 134                |
| 201809 | 1                   |         54321 |   123 | 165                |
| 201810 | 1                   |         54321 |   111 | 134                |
| 201811 | 1                   |         54321 |   143 | 156                |
+--------+---------------------+---------------+-------+--------------------+

Now i want that if one column of the Like_for_like_sales is NULL , then every other column for the specific Filial_number should also be NULL

In this example, some columns of Like_for_like_sales with the Filial_number 54321 are NULL, so every other column of Like_for_like_sales with the Filial_number 54321 should also be set to NULL.

In the end, it should look that way:

+--------+---------------------+---------------+-------+--------------------+
| Month  | Like_for_like_sales | Filial_number | Sales | PreviousYear_Sales |
+--------+---------------------+---------------+-------+--------------------+
| 201712 | 1                   |         12345 |   132 | 164                |
| 201801 | 1                   |         12345 |   143 | 167                |
| 201802 | 1                   |         12345 |   145 | 134                |
| 201803 | 1                   |         12345 |   165 | 176                |
| 201804 | 1                   |         12345 |   164 | 145                |
| 201805 | 1                   |         12345 |   186 | 145                |
| 201806 | 1                   |         12345 |   176 | 165                |
| 201807 | 1                   |         12345 |   187 | 121                |
| 201808 | 1                   |         12345 |   134 | 134                |
| 201809 | 1                   |         12345 |   123 | 165                |
| 201810 | 1                   |         12345 |   111 | 134                |
| 201811 | 1                   |         12345 |   143 | 156                |
| 201712 | NULL                |         54321 |   132 | NULL               |
| 201801 | NULL                |         54321 |   143 | NULL               |
| 201802 | NULL                |         54321 |   145 | NULL               |
| 201803 | NULL                |         54321 |   165 | NULL               |
| 201804 | NULL                |         54321 |   164 | NULL               |
| 201805 | NULL                |         54321 |   186 | NULL               |
| 201806 | NULL                |         54321 |   176 | 165                |
| 201807 | NULL                |         54321 |   187 | 121                |
| 201808 | NULL                |         54321 |   134 | 134                |
| 201809 | NULL                |         54321 |   123 | 165                |
| 201810 | NULL                |         54321 |   111 | 134                |
| 201811 | NULL                |         54321 |   143 | 156                |
+--------+---------------------+---------------+-------+--------------------+

Here is the example query i made for this question, so you can try it by yourself:

WITH test (MONTH, Like_for_like_sales, Filial_number, Sales, PreviousYear_Sales) AS
 (
   SELECT  201712, 1, 12345, 132, 164 
   UNION 
   SELECT  201801, 1, 12345, 143, 167 
   UNION 
   SELECT  201803, 1, 12345, 165, 176 
   UNION 
   SELECT  201804, 1, 12345, 164, 145 
   UNION 
   SELECT  201805, 1, 12345, 186, 145 
   UNION 
   SELECT  201806, 1, 12345, 176, 165 
   UNION 
   SELECT  201807, 1, 12345, 187, 121 
   UNION 
   SELECT  201808, 1, 12345, 134, 134 
   UNION 
   SELECT  201809, 1, 12345, 123, 165 
   UNION 
   SELECT  201810, 1, 12345, 111, 134 
   UNION 
   SELECT  201811, 1, 12345, 143, 156 
   UNION 
   SELECT  201712, NULL, 54321, 132, NULL 
   UNION 
   SELECT  201801, NULL, 54321, 143, NULL 
   UNION 
   SELECT  201802, NULL, 54321, 145, NULL 
   UNION 
   SELECT  201803, NULL, 54321, 165, NULL 
   UNION 
   SELECT  201804, NULL, 54321, 164, NULL 
   UNION 
   SELECT  201805, NULL, 54321, 186, NULL 
   UNION 
   SELECT  201806, NULL, 54321, 176, 165 
   UNION 
   SELECT  201807, 1, 54321, 187, 121 
   UNION 
   SELECT  201808, 1, 54321, 134, 134 
   UNION 
   SELECT  201809, 1, 54321, 123, 165 
   UNION 
   SELECT  201810, 1, 54321, 111, 134 
   UNION 
   SELECT  201811, 1, 54321, 143, 156
 ) 
SELECT Month, Like_for_like_sales, Filial_number, Sales, PreviousYear_Sales 
FROM test 
ORDER BY Filial_number, Month

---

Thanks for your help!

Answer 1

You can apply a Windowed Aggregate:

SELECT  MONTH,
        case when -- any of the rows for a filial for a NULL
                  min(case when Like_for_like_sales is null then 0 end)
                  over (partition by Filial_number) = 0 
             then null
             else Like_for_like_sales
        end,
        Filial_number,
        Sales,
        PreviousYear_Sales 
FROM    test 
ORDER BY Filial_number,
        MONTH

See db<>fiddle

Answer 2

You may use CASE

SELECT Month, CASE WHEN EXISTS (
      SELECT 1
      FROM test t2
      WHERE t1.Filial_number = t2.Filial_number and 
            t2.Like_for_like_sales is null and
            t2.Month < t1.Month
    ) THEN NULL ELSE Like_for_like_sales END, 
       Filial_number, Sales, PreviousYear_Sales 
FROM test t1
ORDER BY Filial_number, Month