--第一种
SELECT * FROM
( SELECT DISTINCT column1 FROM table1 )A
CROSS APPLY
( SELECT column2 =
( SELECT [*] = column2 FROM table1
WHERE column1 = A.column1 FOR XML PATH(''), TYPE ).value('/', 'nvarchar(20)')
)B
--第二种
SELECT *
FROM
(SELECT DISTINCT column1
FROM table1)A
OUTER APPLY
(SELECT [values]= STUFF(REPLACE(REPLACE((
SELECT column2 FROM table1 N
WHERE column1 = A.column1 FOR XML AUTO), '', ''), 1, 1, '')
)N select * from table1 select * from B
--第三种的方法
CREATE FUNCTION dbo.f_str(@id int)
RETURNS varchar(8000)
AS
BEGIN
DECLARE @r varchar(8000)
SET @r = ''
SELECT @r = @r + ',' + value
FROM tb
WHERE id=@id
RETURN STUFF(@r, 1, 1, '')
END
GO
-- 调用函数
SELECt id, values=dbo.f_str(id)
FROM tb
GROUP BY id
-- 2. 新的解决方法
-- 示例数据
DECLARE @t TABLE(id int, value varchar(10))
INSERT @t SELECT 1, 'aa'
UNION ALL SELECT 1, 'bb'
UNION ALL SELECT 2, 'aaa'
UNION ALL SELECT 2, 'bbb'
UNION ALL SELECT 2, 'ccc'
-- 查询处理
SELECT *
FROM(
SELECT DISTINCT
id
FROM @t
)A
OUTER APPLY(
SELECT
[values]= STUFF(REPLACE(REPLACE(
(
SELECT value FROM @t N
WHERE id = A.id
FOR XML AUTO
), '', ''), 1, 1, '')
)N
/*--结果
id values
----------- ----------------
1 aa,bb
2 aaa,bbb,ccc
(2 行受影响)
--*/
--各种字符串分函数
--3.3.1 使用游标法进行字符串合并处理的示例。
--处理的数据
CREATE TABLE tb(col1 varchar(10),col2 int)
INSERT tb SELECT 'a',1
UNION ALL SELECT 'a',2
UNION ALL SELECT 'b',1
UNION ALL SELECT 'b',2
UNION ALL SELECT 'b',3
--合并处理
--定义结果集表变量
DECLARE @t TABLE(col1 varchar(10),col2 varchar(100))
--定义游标并进行合并处理
DECLARE tb CURSOR LOCAL
FOR
SELECT col1,col2 FROM tb ORDER BY col1,col2
DECLARE @col1_old varchar(10),@col1 varchar(10),@col2 int,@s varchar(100)
OPEN tb
FETCH tb INTO @col1,@col2
SELECT @col1_old=@col1,@s=''
WHILE @@FETCH_STATUS=0
BEGIN
IF @col1=@col1_old
SELECT @s=@s+','+CAST(@col2 as varchar)
ELSE
BEGIN
INSERT @t VALUES(@col1_old,STUFF(@s,1,1,''))
SELECT @s=','+CAST(@col2 as varchar),@col1_old=@col1
END
FETCH tb INTO @col1,@col2
END
INSERT @t VALUES(@col1_old,STUFF(@s,1,1,''))
CLOSE tb
DEALLOCATE tb
--显示结果并删除测试数据
SELECT * FROM @t
DROP TABLE tb
/*--结果
col1 col2
---------- -----------
a 1,2
b 1,2,3
--*/
GO
/*==============================================*/
--3.3.2 使用用户定义函数,配合SELECT处理完成字符串合并处理的示例
--处理的数据
CREATE TABLE tb(col1 varchar(10),col2 int)
INSERT tb SELECT 'a',1
UNION ALL SELECT 'a',2
UNION ALL SELECT 'b',1
UNION ALL SELECT 'b',2
UNION ALL SELECT 'b',3
GO
--合并处理函数
CREATE FUNCTION dbo.f_str(@col1 varchar(10))
RETURNS varchar(100)
AS
BEGIN
DECLARE @re varchar(100)
SET @re=''
SELECT @re=@re+','+CAST(col2 as varchar)
FROM tb
WHERE col1=@col1
RETURN(STUFF(@re,1,1,''))
END
GO
--调用函数
SELECT col1,col2=dbo.f_str(col1) FROM tb GROUP BY col1
--删除测试
DROP TABLE tb
DROP FUNCTION f_str
/*--结果
col1 col2
---------- -----------
a 1,2
b 1,2,3
--*/
GO
------------------------------------------
--3.3.3 使用临时表实现字符串合并处理的示例
--处理的数据
CREATE TABLE tb(col1 varchar(10),col2 int)
INSERT tb SELECT 'a',1
UNION ALL SELECT 'a',2
UNION ALL SELECT 'b',1
UNION ALL SELECT 'b',2
UNION ALL SELECT 'b',3
--合并处理
SELECT col1,col2=CAST(col2 as varchar(100))
INTO #t FROM tb
ORDER BY col1,col2
DECLARE @col1 varchar(10),@col2 varchar(100)
UPDATE #t SET
@col2=CASE WHEN @col1=col1 THEN @col2+','+col2 ELSE col2 END,
@col1=col1,
col2=@col2
SELECT * FROM #t
/*--更新处理后的临时表
col1 col2
---------- -------------
a 1
a 1,2
b 1
b 1,2
b 1,2,3
--*/
--得到最终结果
SELECT col1,col2=MAX(col2) FROM #t GROUP BY col1
/*--结果
col1 col2
---------- -----------
a 1,2
b 1,2,3
--*/
--删除测试
DROP TABLE tb,#t
GO
/*==============================================*/
--3.3.4.1 每组 <=2 条记录的合并
--处理的数据
CREATE TABLE tb(col1 varchar(10),col2 int)
INSERT tb SELECT 'a',1
UNION ALL SELECT 'a',2
UNION ALL SELECT 'b',1
UNION ALL SELECT 'b',2
UNION ALL SELECT 'c',3
--合并处理
SELECT col1,
col2=CAST(MIN(col2) as varchar)
+CASE
WHEN COUNT(*)=1 THEN ''
ELSE ','+CAST(MAX(col2) as varchar)
END
FROM tb
GROUP BY col1
DROP TABLE tb
/*--结果
col1 col2
---------- ----------
a 1,2
b 1,2
c 3
--*/
--3.3.4.2 每组 <=3 条记录的合并
--处理的数据
CREATE TABLE tb(col1 varchar(10),col2 int)
INSERT tb SELECT 'a',1
UNION ALL SELECT 'a',2
UNION ALL SELECT 'b',1
UNION ALL SELECT 'b',2
UNION ALL SELECT 'b',3
UNION ALL SELECT 'c',3
--合并处理
SELECT col1,
col2=CAST(MIN(col2) as varchar)
+CASE
WHEN COUNT(*)=3 THEN ','
+CAST((SELECT col2 FROM tb WHERE col1=a.col1 AND col2 NOT IN(MAX(a.col2),MIN(a.col2))) as varchar)
ELSE ''
END
+CASE
WHEN COUNT(*)>=2 THEN ','+CAST(MAX(col2) as varchar)
ELSE ''
END
FROM tb a
GROUP BY col1
DROP TABLE tb
/*--结果
col1 col2
---------- ------------
a 1,2
b 1,2,3
c 3
--*/
GO
if not object_id('A') is null
drop table A
Go
Create table A([id] int,[cname] nvarchar(2))
Insert A
select 1,N'张三' union all
select 2,N'李四' union all
select 3,N'王五' union all
select 4,N'蔡六'
Go
--> -->
if not object_id('B') is null
drop table B
Go
Create table B([id] int,[cname] nvarchar(5))
Insert B
select 1,N'1,2,3' union all
select 2,N'3,4'
Go
create function F_str(@cname nvarchar(100))
returns nvarchar(100)
as
begin
select @cname=replace(@cname,ID,[cname]) from A where patindex('%,'+rtrim(ID)+',%',','+@cname+',')>0
return @cname
end
go
select [id],dbo.F_str([cname])[cname] from B
id cname
----------- ----------------------------------------------------------------------------------------------------
1 张三,李四,王五
2 王五,蔡六
(2 個資料列受到影響)