在cte中使用交叉连接与多个变量

Question

我有一些订单号，想检查是否有任何号码被跳过。我将使用左 table 方法，即：

select * from 
    #CreatedCrossTable (which has all possibilities)
    Left Join #MainTableWhichHaveRealSerialNo mt
where
    mt is null

没有结构的订单是：“CodeType.Year.Month.SerialNo”。例如：“DP.21.07.001”、“DP.21.07.002”.. 或不是 DP，而是“FB.21.07.001”、“FB.21.07.002”等

我想创建一个交叉 table 模式来确定跳过的 SerilNo 值 (CreatedCrossTable 以上):

（序列号每月重置）

CodeType | Year | Month | SerialNo
DP         21       1        1
DP         21       1        2
DP         21       1        3
DP         21       1        4
...
(All SerialNos must increase max serial number of the original table's SerialNo (MainTableWhichHaveRealSerialNo) Also codeType,year and month values must match)

DP         21       2        1
DP         21       2        2
...

FB         21       1        1
FB         21       1        2
...
FB         21       1        1
FB         21       1        2
FB         21       1        3
...

每个代码和月份的序列号都有不同的最大数量用于创建 CrossTable。我写了那个代码：

;WITH cteSerialNo AS
        (
            SELECT 1 AS ORDERNO
            UNION ALL
            SELECT (ORDERNO+1) AS ORDERNO FROM cteSerialNo WHERE ORDERNO < MAX_ORDER_NO
        )
,cteMonthYear AS
        (
            SELECT CAST('2021.01.01' AS DATE) AS Dt
 
            UNION ALL

            SELECT DATEADD(MONTH , 1, Dt) AS Dt
            FROM cteMonthYear
            WHERE DATEADD (MONTH, 1, Dt) < GETDATE()
        )
SELECT
*
FROM
        (
            SELECT
                CODES.CODETYPE,
                YEAR(Dts.Dt) AS 'YEAR', 
                MONTH(Dts.Dt) AS 'MONTH'
            FROM
                ##KK_TBL_CODETYPES AS CODES
                CROSS JOIN cteMonthYear AS Dts
        ) AS CROSSTABLE
        CROSS JOIN cteSerialNo AS cSN

如何为这段代码中的每个变量输入 (MAX_ORDER_NO)？

Answer 1

假设最大 SerialNo 值基于 SerialNo 列中的现有值，您可能只想找到该 SerialNo 值之前的所有可能组合，然后删除源数据中匹配的那些：

-- Define test data
declare @t table(CodeType varchar(2),[Year] int,[Month] int,SerialNo int);
insert into @t values
 ('DP',21,1,1)
,('DP',21,1,2)
,('DP',21,1,3)
--,('DP',21,1,4) -- Missing so should be in Output
,('DP',21,1,5)
,('DP',21,2,1)
,('DP',21,2,2)
,('FB',21,1,1)
,('FB',21,1,2)
,('FB',21,2,1)
,('FB',21,2,2)
--,('FB',21,2,3) -- Missing so should be in Output
,('FB',21,2,4)
;

with m as  -- Get Max SerialNo for each grouping
(
    select CodeType
          ,[Year]
          ,[Month]
          ,max(SerialNo) as MaxSerialNo
    from @t
    group by CodeType
            ,[Year]
            ,[Month]
)
,t as  -- Create a table with 10 rows in
(
    select t
    from(values(1),(1),(1),(1),(1),(1),(1),(1),(1),(1)) as t(t)
)
,n as  -- Self join those 10 rows 5 times to generate a possible 10*10*10*10*10 = 100,000 incrementing numbers using row_number
(
    select top(select max(MaxSerialNo) from m) row_number() over (order by (select null)) as n
    from t,t t2,t t3,t t4,t t5
)
-- Join from the numbers table to the source data to generate all possible SerialNo values up to the Max
select m.CodeType
      ,m.[Year]
      ,m.[Month]
      ,n.n as SerialNo
from n
    left join m
        on n.n <= m.MaxSerialNo

except -- Then exclude any that are in the source data

select CodeType
      ,[Year]
      ,[Month]
      ,SerialNo
from @t

order by CodeType
        ,[Year]
        ,[Month]
        ,SerialNo

输出

CodeType	Year	Month	SerialNo
DP	21	1	4
FB	21	2	3

在cte中使用交叉连接与多个变量

Using cross join with multiple variable in cte

sql

sql-server

sql-server-2012

cross-join

输出