从同一个表中的逗号分隔列更新行列

Question

我有以下表定义

CREATE TABLE _Table 
(
    [Pat] NVARCHAR(8), 
    [Codes] NVARCHAR(50), 
    [C1] NVARCHAR(6), 
    [C2] NVARCHAR(6), 
    [C3] NVARCHAR(6), 
    [C4] NVARCHAR(6), 
    [C5] NVARCHAR(6)
);
GO

INSERT INTO _Table ([Pat], [Codes], [C1], [C2], [C3], [C4], [C5])
VALUES
    ('Pat1', 'U212,Y973,Y982', null, null, null, null, null),
    ('Pat2', 'M653', null, null, null, null, null), 
    ('Pat3', 'U212,Y973,Y983,Z924,Z926', null, null, null, null, null);
GO  

SQL小提琴在这里.

现在,我想分割每行的代码并填充Cn列,以便最终得到

Pat     Codes                       C1      C2      C3      C4      C5
Pat1    'U212,Y973,Y982'            U212    Y973    Y982    NULL    NULL
Pat2    'M653'                      M653    NULL    NULL    NULL    NULL
Pat3    'U212,Y973,Y983,Z924,Z926'  U212    Y973    Y983    Z924    Z926

我正在研究动态SQL,但有更好的方法......

我已经开始了CTE路线,但我在这里很弱.我基本上是循环,删除第一个逗号分隔的代码,并使用左来获取该代码并选择它C1.

;WITH tmp([Pat], [Codes], [C1], [C2], [C3], [C4], [C5]) AS
(
    SELECT
        Pat,
        STUFF(Codes, 1, CHARINDEX(',', Codes + ','), ''), 
        LEFT(Codes, CHARINDEX(',', Codes + ',') - 1), 
        [C2], 
        [C3], 
        [C4], 
        [C5]
    FROM _Table 
    UNION all
    SELECT
        Pat,
        STUFF(Codes, 1, CHARINDEX(',', Codes + ','), ''), 
        LEFT(Codes, CHARINDEX(',', Codes + ',') - 1), 
        [C2], 
        [C3], 
        [C4], 
        [C5]
    FROM _Table 
    WHERE
        Codes > ''
)
SELECT Pat, Codes,  [C1], [C2], [C3], [C4], [C5] 
FROM tmp 
ORDER BY Pat

这适用于一个代码,但我如何做所有5？注意,在实践中,这可以增加到N个代码.

Answer 1

如果我正确理解要求,这非常简单.不需要拆分或其他类型的功能,动态SQL,递归CTE,PIVOTING或任何其他技术.

要执行"拆分",您可以这样使用CROSS APPLY:

SELECT 
  Pat,
  Codes,
  C1 = SUBSTRING(Codes,1,ISNULL(d1.d-1,8000)),
  C2 = SUBSTRING(Codes,d1.d+1, d2.d-d1.d-1),
  C3 = SUBSTRING(Codes,d2.d+1, d3.d-d2.d-1),
  C4 = SUBSTRING(Codes,d3.d+1, d4.d-d3.d-1),
  C5 = SUBSTRING(Codes,d4.d+1, 8000)
FROM _Table
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes),0)))        d1(d)
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes,d1.d+1),0))) d2(d)
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes,d2.d+1),0))) d3(d)
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes,d3.d+1),0))) d4(d);

返回

Pat      Codes                         C1    C2     C3    C4    C5    
-------- ----------------------------- ----- ------ ----- ----- ------
Pat1     U212,Y973,Y982                U212  Y973   NULL  NULL  NULL
Pat2     M653                          M653  NULL   NULL  NULL  NULL
Pat3     U212,Y973,Y983,Z924,Z926      U212  Y973   Y983  Z924  Z926

请注意超级简单且超高效的执行计划:

如果代码总是四个字符,你可以进一步简化这个:

SELECT
  Pat,
  Codes,
  C1 = NULLIF(SUBSTRING(Codes,1,4),''),
  C2 = NULLIF(SUBSTRING(Codes,6,4),''),
  C3 = NULLIF(SUBSTRING(Codes,11,4),''),
  C4 = NULLIF(SUBSTRING(Codes,16,4),''),
  C2 = NULLIF(SUBSTRING(Codes,21,4),'')
FROM _Table;

要执行更新,您将为第一个解决方案执行此操作:

UPDATE _Table
SET 
  C1 = SUBSTRING(Codes,1,ISNULL(d1.d-1,8000)),
  C2 = SUBSTRING(Codes,d1.d+1, d2.d-d1.d-1),
  C3 = SUBSTRING(Codes,d2.d+1, d3.d-d2.d-1),
  C4 = SUBSTRING(Codes,d3.d+1, d4.d-d3.d-1),
  C5 = SUBSTRING(Codes,d4.d+1, 8000)
FROM _Table
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes),0)))        d1(d)
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes,d1.d+1),0))) d2(d)
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes,d2.d+1),0))) d3(d)
CROSS APPLY (VALUES (NULLIF(CHARINDEX(',',Codes,d3.d+1),0))) d4(d);

如果代码再次只有四个字符,则更新非常简单,感觉就像作弊:

UPDATE _Table
SET C1 = NULLIF(SUBSTRING(Codes,1,4),''),
    C2 = NULLIF(SUBSTRING(Codes,6,4),''),
    C3 = NULLIF(SUBSTRING(Codes,11,4),''),
    C4 = NULLIF(SUBSTRING(Codes,16,4),''),
    C5 = NULLIF(SUBSTRING(Codes,21,4),'');