根据记录数在 SQL 服务器中将单列拆分为多列

Question

我在一个列中有大约 28 条记录（这是动态的，有时是奇数条记录，有时是偶数条记录）。

想根据变量将它们分成多列。每列应有 5 或 6 条记录。

注意：这5或6条记录是一个变量，应该改变。

示例输入 table:

所需的输出 table 将是：

            Col1    col2    col3    col4    col5
            -------------------------------------
            A        F       K       P       U
            B        G       L       Q       V
            C        H       M       R       W
            D        I       N       S       X
            E        J       O       T       Y
                                             Z

这是我尝试过的，这两个不同的查询提供了不同的结果：

            select Col1
                from Table1 where  Col1 is not null
                order by [col1] asc

--This provides all records

            WITH CTE AS
            (
            SELECT [Col1],
                   (ROW_NUMBER() OVER (ORDER BY Col1) -1)%5 AS Col,
                   (ROW_NUMBER() OVER (ORDER BY Col1) -1)/5 AS Row
            FROM Table1  where  Col1 is not null     
            )
            SELECT [0], [1], [2], [3], [4]
            FROM CTE 
            PIVOT (MAX([COL1]) FOR Col IN ([0], [1], [2], [3], [4])) AS Pvt
            ORDER BY Row

-- 在这种情况下记录丢失（当记录数很高时，因为记录数可以动态增加）

            WITH CTE AS
            (
            SELECT [Col1],
                   (ROW_NUMBER() OVER (ORDER BY Col1) -1)%5 AS Col,
                   (ROW_NUMBER() OVER (ORDER BY Col1) -1)/5 AS Row
            FROM Table1  where  Col1 is not null    
            )
            SELECT [0], [1]
            FROM CTE 
            PIVOT (MAX([COL1]) FOR Col IN ([0], [1])) AS Pvt
            ORDER BY Row

我做错了什么？我怎样才能达到预期的输出？谢谢。

Answer 1

您做错的是在 Col、

中获取五个值

(ROW_NUMBER() OVER (ORDER BY Col1) -1)%5 AS Col,

但只有其中 2 个 PIVOTing。

 PIVOT (MAX([COL1]) FOR Col IN ([0], [1])) AS Pvt

因此您忽略了所有值为 2、3 和 4 的 Col。

Answer 2

根据经验，我能够提出以下数据透视查询：

WITH cte AS (
    SELECT Col1, ROW_NUMBER() OVER (ORDER BY Col1) rn
    FROM yourTable
)

SELECT
    MAX(CASE WHEN (rn-1) / 5 = 0 THEN Col1 END) AS col1,
    MAX(CASE WHEN (rn-1) / 5 = 1 THEN Col1 END) AS col2,
    MAX(CASE WHEN (rn-1) / 5 = 2 THEN Col1 END) AS col3,
    MAX(CASE WHEN (rn-1) / 5 = 3 THEN Col1 END) AS col4,
    MAX(CASE WHEN (rn-1) / 5 = 4 THEN Col1 END) AS co15
FROM cte
GROUP BY
    (rn-1) % 5;

Demo

这里的想法是，每个组（一行）由当前行号以 5 为模的那一点确定。列或数据透视表由当前行号的 5 的倍数决定。

Answer 3

您需要先为每条记录分配一个列，这可以通过NTILE():

来完成

WITH Table1 AS
(
    SELECT  TOP 26 Col1 = CHAR(64 + ROW_NUMBER() OVER(ORDER BY object_id))
    FROM    sys.all_objects
)
SELECT [Col1],
        5 - NTILE(5) OVER(ORDER BY Col1 DESC) AS Col
FROM Table1 ;

给予：

Col1    Col
---------
Z       4
Y       4
X       4
W       4
V       4
U       4
T       3
....
Y       4
Z       4

注意，要从最后一列开始填充，您必须以相反的顺序分配列 (ORDER BY Col1 DESC)，然后从列总数中扣除。

然后您可以通过在您的列中排序来计算您的行：

WITH Table1 AS
(
    SELECT  TOP 26 Col1 = CHAR(64 + ROW_NUMBER() OVER(ORDER BY object_id))
    FROM    sys.all_objects
)
SELECT  Col1, Col, ROW_NUMBER() OVER(PARTITION BY Col ORDER BY Col1) AS Row
FROM    (   SELECT [Col1],
                    5 - NTILE(5) OVER(ORDER BY Col1 DESC) AS Col
            FROM Table1  
        ) c;

给予：

Col1    Col     Row
--------------------
A       0       1
B       0       2
C       0       3
D       0       4
E       0       5
F       1       1
G       1       2

然后你可以应用枢轴：

WITH Table1 AS
(
    SELECT  TOP 26 Col1 = CHAR(64 + ROW_NUMBER() OVER(ORDER BY object_id))
    FROM    sys.all_objects
), CTE AS
(
    SELECT  Col1, COL, ROW_NUMBER() OVER(PARTITION BY Col ORDER BY Col1) AS Row
    FROM    (   SELECT [Col1],
                        5 - NTILE(5) OVER(ORDER BY Col1 DESC) AS Col
                FROM Table1  
            ) c
)
SELECT [0], [1], [2], [3], [4]
FROM CTE 
PIVOT (MAX([COL1]) FOR Col IN ([0], [1], [2], [3], [4])) AS Pvt
ORDER BY Row;

给予：

0       1       2       3       4
-------------------------------------
A       F       K       P       U
B       G       L       Q       V
C       H       M       R       W
D       I       N       S       X
E       J       O       T       Y
NULL    NULL    NULL    NULL    Z

附录

我已经设置了一个示例 table，并使用动态 SQL 的过程来简化 re-use 来演示此解决方案，它似乎按我的预期工作。

设置

-- SET UP TABLE AND INSERT RANDOM VALUES
IF OBJECT_ID(N'tempdb..#Table1', 'U') IS NOT NULL
    DROP TABLE #Table1;

CREATE TABLE #Table1 (Col1 CHAR(2));
INSERT #Table1 (Col1)
SELECT  CONCAT(Letter, Number)
FROM    (SELECT  TOP 26 Letter = CHAR(64 + ROW_NUMBER() OVER(ORDER BY object_id))
        FROM  sys.all_objects) l
        CROSS JOIN (VALUES (1),(2),(3),(4),(5),(6),(7),(8),(9)) n (Number);

GO
IF OBJECT_ID(N'tempdb..#GenerateMatrix', 'P') IS NOT NULL
    DROP PROCEDURE #GenerateMatrix;

GO
CREATE PROCEDURE #GenerateMatrix @Records INT, @Columns INT
AS
BEGIN
    -- GENERATE COLUMNS FOR PIVOT AND SELECT
    DECLARE @ColSQL NVARCHAR(MAX) = 
            STUFF((SELECT   TOP (@Columns) 
                            CONCAT(',', QUOTENAME(ROW_NUMBER() OVER(ORDER BY Col1) - 1))
                    FROM #Table1
                    FOR XML PATH(''), TYPE
                ).value('.', 'NVARCHAR(MAX)'), 1, 1, '');


    -- FOR @Cols = 5 Generates "[0],[1],[2],[3],[4]"

    DECLARE @SQL NVARCHAR(MAX) = 
        CONCAT('SELECT ', @ColSQL, '
                FROM (SELECT  Col1, Col, Row = ROW_NUMBER() OVER(PARTITION BY Col ORDER BY Col1)
                    FROM    (   SELECT [Col1], Col = @Columns - NTILE(@Columns) OVER(ORDER BY Col1 DESC)
                                FROM (SELECT TOP (@Records) Col1 FROM #Table1 Col1) t
                            ) c) c 
                PIVOT (MAX([COL1]) FOR Col IN (', @ColSQL, ')) AS Pvt
                ORDER BY Row;');

    EXECUTE sp_executesql @SQL, N'@Columns INT, @Records INT', @Columns = @Columns, @Records = @Records;

END
GO

测试 1

EXECUTE #GenerateMatrix @Records = 26, @Columns = 5;

0       1       2       3       4
----------------------------------
A1      A6      B2      B7      C3
A2      A7      B3      B8      C4
A3      A8      B4      B9      C5
A4      A9      B5      C1      C6
A5      B1      B6      C2      C7
NULL    NULL    NULL    NULL    C8

测试 2

EXECUTE #GenerateMatrix @Records = 8, @Columns = 4;

0       1       2       3
----------------------------
A1      A3      A5      A7
A2      A4      A6      A8

测试 3

EXECUTE #GenerateMatrix @Records = 40, @Columns = 8;

0       1       2       3       4       5       6       7
-----------------------------------------------------------
A1      A6      B2      B7      C3      C8      D4      D9
A2      A7      B3      B8      C4      C9      D5      E1
A3      A8      B4      B9      C5      D1      D6      E2
A4      A9      B5      C1      C6      D2      D7      E3
A5      B1      B6      C2      C7      D3      D8      E4

测试 4

EXECUTE #GenerateMatrix @Records = 50, @Columns = 6;

0       1       2       3       4       5
    ---------------------------------------
A1      A9      B8      C7      D6      E6
A2      B1      B9      C8      D7      E7
A3      B2      C1      C9      D8      E8
A4      B3      C2      D1      D9      E9
A5      B4      C3      D2      E1      F1
A6      B5      C4      D3      E2      F2
A7      B6      C5      D4      E3      F3
A8      B7      C6      D5      E4      F4
NULL    NULL    NULL    NULL    E5      F5

在任何测试中都没有丢失任何记录，也没有只有空值的行。

Example on DB Fiddle

根据记录数在 SQL 服务器中将单列拆分为多列

Split single column into multiple columns in SQL server based on number of records

python

tsql

sql-server

sql-server-2014

sql-server-2017

Demo