Nested CASE in SQL

问题:

TABLE1:

ARTIKEL      SUPPLIERID   SALE_SUM_PIECES   
TV            SONY            7            

TABLE2:

 ROW_ID     ARTIKEL      SUPPLIERID     PIECES    
    1           TV            SONY       6       
    2           TV            SONY       10      
    3           TV            SONY       6      
    4           TV            SONY       14       
    5           TV            SONY       18      
    6           TV            SONY       4  

I need to subtract value X=23 on TABLE2."PIECES", only when the value TABLE1."SALE_SUM_PIECES" is less than the SUM of "PIECES" in TABLE2. For example: the value of TABLE1."SALE_SUM_PIECES" is 7. NOw I need to check at which row the value 7 goes less than the SUM of TABLE2."PIECES".In the below example the first row in TABLE2 is not valid because 7 is greater than 6. But the second row in TABLE2 is valid since the SUM OF "PIECES" from row1 and row2 in TABLE2 i.e 6+10=16 is greater than 7. So, I need to subtract value of X=23 from the second row to the following rows in TABLE2. The query I have is as follows:

SELECT "SUPPLIERID", "ARTIKEL",
    (case when ( cumulativesum - (select "SALE_SUM_PIECES" from  T1 where T1."SUPPLIERID"=T2."SUPPLIERID" and T1."ARTIKEL" = T2."ARTIKEL" )) <= 0
        then NULL
        when ( cumulativesum - (select "SALE_SUM_PIECES" from  TABLE1 T1 where T1."SUPPLIERID"=T2."SUPPLIERID" and T1."ARTIKEL" = T2."ARTIKEL" )) > 0 
        then
            (case when @x - cumulativesum <= 0 and @x - (cumulativesum -PIECES) > 0
                 then 0
                 when @x - "cumulativesum" <= 0
                 then NULL
                 else @x - "cumulativesum"
            end) as "VALUE_DRILL_DOWN"
    from (SELECT T1."ARTIKEL", T1."SUPPLIERID", T1.PIECES
             (select sum("PIECES")
              from EXAMPLE_TABLE T2
              where T2."ROW_ID" <= T1."ROW_ID" and T2."SUPPLIERID" = T1."SUPPLIERID" and T2."ARTIKEL"=T1."ARTIKEL"
             ) as "cumulativesum"
     from EXAMPLE_TABLE T1
     ) 

When I execute the above query I get the result as follows:

 ROW_ID     ARTIKEL      SUPPLIERID    PIECES     VALUE_DRILL_DOWN
    1           TV            SONY      6            NULL
    2           TV            SONY      10           7
    3           TV            SONY      6            1
    4           TV            SONY      14           0
    5           TV            SONY      18           Null
    6           TV            SONY       4           Null 

But I expect a result to be as follows:

   ROW_ID       ARTIKEL      SUPPLIERID    PIECES    VALUE_DRILL_DOWN
    1           TV            SONY           6           NULL
    2           TV            SONY          10           13
    3           TV            SONY          6            7
    4           TV            SONY          14           0
    5           TV            SONY          18           Null
    6           TV            SONY          4            Null

I want the subtraction of ´X=23´ to start from the row in TABLE2 where the condition TABLE1."SALE_SUM_PIECES" < TABLE2."PIECES" i.e from row2. Any suggestions?

Thanks in advance.

回答1:

The below solution gives the desired results. See SqlFiddle

1   TV  SONY    922 6   110 2.50    NULL
2   TV  SONY    922 10  80  1.00    13
3   TV  SONY    922 6   65  1.50    7
4   TV  SONY    922 14  95  1.50    0
5   TV  SONY    922 18  95  1.50    NULL
6   TV  SONY    922 4   95  1.50    NULL



DECLARE @x INT = 23

; WITH cte AS
(
    SELECT t2.*, t1.SALE_SUM_PIECES,
            CASE 
                WHEN SUM(t2.PIECES) OVER (PARTITION BY t2.ARTIKEL, t2.SUPPLIERID ORDER BY ROW_ID) < t1.SALE_SUM_PIECES THEN 'None'
                ELSE t2.ARTIKEL + t2.SUPPLIERID
            END AS GroupId
    FROM @Table2 t2
    JOIN @Table1 t1 ON t2.ARTIKEL = t1.ARTIKEL AND t2.SUPPLIERID = t1.SUPPLIERID
),
cumulative AS
(
    SELECT *, SUM(PIECES) OVER (PARTITION BY GroupId ORDER BY ROW_ID) AS CumulativeSum
    FROM cte
)
SELECT ROW_ID, ARTIKEL, SUPPLIERID, ORGID, PIECES, COSTPRICE, DISCOUNT, 
        CASE
            WHEN CumulativeSum < SALE_SUM_PIECES THEN NULL
            WHEN @x - CumulativeSum <= 0 AND @x - (CumulativeSum - PIECES) > 0 THEN 0
            WHEN @x - CumulativeSum <= 0 THEN NULL
            ELSE @x - CumulativeSum
        END AS VALUE_DRILL_DOWN
FROM cumulative

回答2:

I'm not 100% sure what the expected calulation for VALUE_DRILL_DOWN is supposed to be. You've said that for a valid row, it should be (sum of all rows at or above row) - 23

In this case the following code should work, but it gives different numbers from your expected results.

I've used a self-joining common table expression to work out the cumulative sums of the rows as you count up .

The (total of all rows at or above) is equal to the (total) - (cumulative sum at row before)

Therefore the following code should be correct...

DECLARE @table1 TABLE(ARTIKEL NVARCHAR(50), SUPPLIERID NVARCHAR(50), ORGID INT, 
STORE INT, SALE_SUM_PIECES INT, [DATE] DATETIME)
DECLARE @table2 TABLE(ROW_ID INT, ARTIKEL NVARCHAR(50), SUPPLIERID NVARCHAR(50), 
ORGID INT, PIECES INT, COSTPRICE MONEY, DISCOUNT DECIMAL(9,5))
DECLARE @x INT = 23
DECLARE @SumOfAll INT
INSERT @table1
        ( ARTIKEL , SUPPLIERID , ORGID ,  STORE ,  SALE_SUM_PIECES ,  [DATE] )
VALUES  ( 'TV', 'SONY',  922,100,7 ,'2014-01-01'  )
INSERT @table2
        (ROW_ID, ARTIKEL , SUPPLIERID ,  ORGID , PIECES ,  COSTPRICE , DISCOUNT  )
VALUES  (1, 'TV', 'SONY',  922, 6, 110, 2.5 ),
    (2, 'TV', 'SONY',  922, 10 ,  80,  1 )  ,  
    (3, 'TV', 'SONY',  922, 6 , 65 ,  1.5  ) ,
    (4, 'TV', 'SONY',  922, 14 , 95  ,  1.5  ),
    (5, 'TV', 'SONY',  922, 18 ,  95 ,   1.5  ),
    (6, 'TV', 'SONY',  922, 4  ,  95 ,  1.5  )

SELECT @SumOfAll = SUM(PIECES) FROM @table2 AS t

;WITH t2C AS (
    SELECT ROW_ID, PIECES AS [cumulativeSum]
    FROM @table2 AS t
    WHERE ROW_ID = 1 -- assume starting at 1
    UNION ALL
    SELECT t.ROW_ID, t.PIECES + cumulativeSum AS [cumlativeSum]
    FROM @table2 AS t
    JOIN t2C ON t2C.ROW_ID + 1 = t.ROW_ID --assume rowIDs are sequential
)
SELECT
    t2.ROW_ID,  
    t1.SUPPLIERID,
    t1.ORGID,
    t2.PIECES,
    t2.COSTPRICE,
    t2.DISCOUNT,
    CASE WHEN t2C.cumulativeSum - t1.SALE_SUM_PIECES <= 0 THEN NULL
        ELSE
            CASE 
                WHEN @x - t2C.cumulativeSum <= 0 
                    AND @x - (t2C.cumulativeSum - t2.PIECES) > 0 THEN 0
                WHEN @x - t2C.cumulativeSum <= 0 THEN NULL
                ELSE  (@SumOfAll - t2cPrev.cumulativeSum) - @x
            END
        END AS [VALUE_DRILL_DOWN]
FROM t2C
LEFT JOIN t2C AS t2cPrev ON t2cPrev.ROW_ID = t2C.ROW_ID - 1
JOIN @table2 AS t2 ON t2.ROW_ID = t2C.ROW_ID
JOIN @table1 AS t1 ON t1.SUPPLIERID = t2.SUPPLIERID AND t1.ARTIKEL = t2.ARTIKEL

However this gives:

ROW_ID  SUPPLIERID  ORGID   PIECES  COSTPRICE   DISCOUNT    VALUE_DRILL_DOWN    cumulativeSum   sumOfRowsOnOrAfter
1       SONY        922     6       110.00      2.50000     NULL                6               NULL
2       SONY        922     10      80.00       1.00000     29                  16              52
3       SONY        922     6       65.00       1.50000     19                  22              42
4       SONY        922     14      95.00       1.50000     0                   36              36
5       SONY        922     18      95.00       1.50000     NULL                54              22
6       SONY        922     4       95.00       1.50000     NULL                58              4

EDIT:

SQL Fiddle by WiiMaxx

回答3:

select
  *,
  case when 23 - sum(c) over (order by row_id) >= 0
       then 23 - sum(c) over (order by row_id)
       when 23 - sum(c) over (order by row_id) + pieces > 0
       then 0
  end value_drill_down
from (
       select *,
         case when sum(pieces) over (order by row_id) >= (select sale_sum_pieces from t1)
              then pieces
         end c
       from t2) x;