SQL Server 2005:将字符串拆分为数组并获取数组(x)?

时间:2012-11-30 12:33:12

标签: arrays sql-server-2005 split

我在数据库表中有这样的字符串:

Peter/Parker/Spiderman/Marvel
Bruce/Wayne/Batman/DC

SQL中是否有一种方法可以从字符串中提取特定值,例如

Name = MyColumn(0)
SurName = MyColumn(1)
Character = MyColumn(3)
Company = MyColumn(4)

由于

2 个答案:

答案 0 :(得分:13)

我在这里假设总有4个部分。

如果是这样,您可以将/替换为.并使用内置的ParseName功能。你的例子中唯一的问题是它从最后开始计算,所以你需要注意你想要的那个部分:

DECLARE @test VARCHAR(max);
SET @test = 'Peter/Parker/Spiderman/Marvel';
SET @test = Replace(@test, '/', '.');

SELECT Parsename(@test, 4),--returns Peter
       Parsename(@test, 3),--returns Parker
       Parsename(@test, 2),--returns Spiderman
       Parsename(@test, 1) --returns Marvel

如果有可变数量的部件,你需要找到一个字符串拆分功能来为你做这个,内置的不是很好。搜索SO时可以找到很多选项:{{3} }

警告提示 - 如果您尝试使用除{1}以外的PARSENAME的数字,则结果将始终为NULL。

答案 1 :(得分:5)

如果您知道您将有4列,那么您也可以使用此嵌套CTE版本:

;with s1 (name, extra) as
(
  select left(data, charindex('/', data)-1), 
    substring(data, charindex('/', data) +1, len(data))
  from yourtable
),
s2 (name, surname, extra) as
(
  select name, 
    left(extra, charindex('/', extra)-1), 
    substring(extra, charindex('/', extra)+1, len(extra))
  from s1
),
s3 (name, surname, [character], company) as
(
  select name, 
    surname, 
    left(extra, charindex('/', extra)-1), 
    substring(extra, charindex('/', extra)+1, len(extra))
  from s2
)
select *
from s3;

请参阅SQL Fiddle with Demo

结果是:

|  NAME | SURNAME | CHARACTER | COMPANY |
-----------------------------------------
| Peter |  Parker | Spiderman |  Marvel |
| Bruce |   Wayne |    Batman |      DC |

或者您可以实现分割数据的CTE,然后实现PIVOT

;with cte (item, data, colNum, rn) as
(
  select cast(left(data, charindex('/',data+'/')-1) as varchar(50)) item,
    stuff(data, 1, charindex('/',data+'/'), '') data,
    1 colNum,
    row_number() over(order by data) rn
  from yourtable
  union all
  select cast(left(data, charindex('/',data+'/')-1) as varchar(50)) ,
    stuff(data, 1, charindex('/',data+'/'), '') data,
    colNum+1, 
    rn
  from cte
  where data > ''
) 
select [1] as Name, 
  [2] as Surname, 
  [3] as [character], 
  [4] as company
from
(
  select item, colnum, rn
  from cte
) src
pivot
(
  max(item)
  for colnum in ([1], [2], [3], [4])
) piv

请参阅SQL Fiddle with Demo