使用CTE拆分表中的行

Question

我在SQL Server 2008 R2中的表中有以下行

+-------------------------------------------+
|  ID          EntryType dt         price   |
+-------------------------------------------+
| 14          4         2012-11-07 0.025000 |
| 16          5         2012-11-07 0.026000 |
| 18          6         2012-11-07 0.026000 |
| 20          7         2012-11-07 0.026000 |
+-------------------------------------------+

我想做的是基于EntryType分散行（EntryType不变）

For EntryType = 4 (1 row)
For EntryType = 5 (2 row)
For EntryType = 6 (3 row)
For EntryType = 7 (9 row)

并且dt字段将增加（以一个月为间隔），因此输出如下所示：

+-----------+-----------+-------+
| EntryType |    dt     | Price |
+-----------+-----------+-------+
|         4 | 11/7/2012 | 0.024 |
|         5 | 12/7/2012 | 0.025 |
|         5 | 1/7/2013  | 0.025 |
|         6 | 2/7/2013  | 0.026 |
|         6 | 3/7/2013  | 0.026 |
|         6 | 4/7/2013  | 0.026 |
|         7 | 5/7/2013  | 0.027 |
|         7 | 6/7/2013  | 0.027 |
|         7 | 7/7/2013  | 0.027 |
|         7 | 8/7/2013  | 0.027 |
|         7 | 9/7/2013  | 0.027 |
|         7 | 10/7/2013 | 0.027 |
|         7 | 11/7/2013 | 0.027 |
|         7 | 12/7/2013 | 0.027 |
|         7 | 1/7/2014  | 0.027 |
+-----------+-----------+-------+

使用CTE和SQL可以做到吗？

Answer 1

这是在递归CTE中执行此操作的方法：

;with RecordCounts as (
    -- Establish row counts for each EntryType
    select 4 as EntryType, 1 as RecordCount
    union all select 5, 2
    union all select 6, 3
    union all select 7, 9
), PricesCte as (
    -- Get initial set of records
    select ID, p.EntryType, (select min(dt) from MyTable) as dt, price, 1 as RecordNum
    from MyTable p
        join RecordCounts c on p.EntryType = c.EntryType -- Only get rows where we've established a RecordCount
    -- Add records recursively according to RecordCount
    union all
    select ID, p.EntryType, dt, price, RecordNum + 1
    from PricesCte p
        join RecordCounts c on p.EntryType = c.EntryType
    where RecordNum + 1 <= c.RecordCount
)
select EntryType,
    dateadd(mm, row_number() over (order by EntryType, ID) - 1, dt) as dt,
    price
from PricesCTE
order by EntryType
option (maxrecursion 0) -- Infinite recursion, default limit is 100

这是展示这项工作的SqlFiddle 。

有两件事：

• 我认为随着记录数量的增加，使用Tally表而不是递归地乘以记录可能会更好。 您将与Tally表交叉连接，并让where子句根据RecordCount限制记录
• 我没有看到定价应该如何从输入更改为输出。
• 我不知道您在哪里为每个EntryType建立RecordCount，所以我已将其添加到另一个CTE中。

Answer 2

;WITH e (ID, EntryType, row, dt, Price, [Len])
AS
 (
  SELECT ID, EntryType, CASE EntryType WHEN 4 THEN 1
                                       WHEN 5 THEN 2
                                       WHEN 6 THEN 3
                                       WHEN 7 THEN 9 END AS row,
         dt, Price, 0 AS [Len] 
  FROM dbo.your_table
), x (ID, EntryType, row, dt, Price, [Len]) AS
 (
  SELECT ID, EntryType, row, dt, Price, 1
  FROM e
  UNION ALL
  SELECT e.ID, e.EntryType, e.row, e.dt, e.Price, x.[Len] + 1
  FROM e , x
  WHERE e.ID = x.ID AND e.row > x.[Len]
  )
SELECT EntryType, DATEADD(mm, ROW_NUMBER() OVER(ORDER BY EntryType)-1, dt) AS dt, Price
FROM x
ORDER BY EntryType

我唯一不明白以什么标准获得价格？

SQL小提琴演示

Answer 3

不知道这是否可以由CTE完成，因为我无法告知EntryType和＃行之间的任何逻辑。

对于这个问题，我认为用所需的行数来构建临时表会更容易

DECLARE @Entry TABLE(EntryType INT, seq INT) 
INSERT INTO @Entry (EntryType , seq) VALUES
    (4,1)
    ,(5,1)
    ,(5,2)
    ,(6,1)
    ...
    ,(7,1)
    ..
    ,(7,9)

之后，简单的查询将获得相同的结果

SELECT  t.EntryType , DATEADD(MONTH,e.seq,  t.dt) as dt,t.Price 
FROM    YourTable t
INNER JOIN @Entry e 
    ON  t.EntryType = e.EntryType