现在小弟有如此一数据表
结构如下:
select 1 as tname,'01:0102;02:0102;03:0102;04:0102,0101;05:0102'as tstr into tb1
insert into tb1 select 2,'01:0101,0102'
insert into tb1 select 3,'01:0102;02:0102;03:0102;04:0102;05:0102'
insert into tb1 select 4,'01:0102,0103'
insert into tb1 select 5,'0104'
insert into tb1 select 6,'01:0102;02:0102;03:0102;04:0102;05:0102'
insert into tb1 select 7,'01:0102;02:0102;03:0102;04:0102,0101;05:0102'
上表数据代表的含义
tname 为序号,不用管
tstr 为编组内容
01:0101,0102 意思为01组,组员为编号0101和0102两人组成。
01:0102;02:0102 意思为编号0102的组员,即在01组,又在02组。
可见有很多重复。
现在查询想得到如下结果:即找出这几些编组到底有哪些组员
0101,0102,0103,0104
求存储过程,不能创建自定义函数,谢谢
注意数据库版本为sqlserver 2k
------解决方案--------------------
是这样吗:
select --*,
distinct
case when vv like '%:%' then SUBSTRING(vv,charindex(':',vv)+1,len(vv))
else vv
end '编组'
from
(
select tname,
tstr,
v,
SUBSTRING(t.v, number ,CHARINDEX(',',t.v+',',number)-number) vv
from
(
select tname,
tstr,
SUBSTRING(t.tstr, number ,CHARINDEX(';',t.tstr+';',number)-number) v
from tbl t,master..spt_values s
where s.number >=1
and s.type = 'P'
and SUBSTRING(';'+t.tstr,s.number,1) = ';'
)t,master..spt_values s
where s.number >=1
and s.type = 'P'
and SUBSTRING(','+t.v,s.number,1) = ','
)t
/*
编组
0101
0102
0103
0104
*/
------解决方案--------------------
SELECT * INTO #tb
FROM
(
Select
a.tname,tstr=substring(substring(a.tstr,b.number,charindex(';',a.tstr+';',b.number)-b.number),4,LEN(substring(a.tstr,b.number,charindex(';',a.tstr+';',b.number)-b.number)))
from
Tb1 a join master..spt_values b
ON B.type='p' AND B.number BETWEEN 1 AND LEN(A.tstr)
where
substring(';'+a.tstr,b.number,1)=';')t
SELECT * FROM
(
SELECT PARSENAME(REPLACE(tstr,',','.'),2) AS col FROM #tb
UNION
SELECT PARSENAME(REPLACE(tstr,',','.'),1) FROM #tb)t WHERE col IS NOT NULL
------解决方案--------------------
完整的:
建表:
--drop table tbl
create table tbl(tname int,tstr varchar(100))
insert into tbl
select 1 ,'01:0102;02:0102;03:0102;04:0102,0101;05:0102'
insert into tbl
select 2,'01:0101,0102'
insert into tbl
select 3,'01:0102;02:0102;03:0102;04:0102;05:0102'
insert into tbl
select 4,'01:0102,0103'
insert into tbl
select 5,'0104'
insert into tbl
select 6,'01:0102;02:0102;03:0102;04:0102;05:0102'
insert into tbl
select 7,'01:0102;02:0102;03:0102;04:0102,0101;05:0102'
go
存储过程:
create proc dbo.proc_group
as
select --*,
distinct
case when vv like '%:%' then SUBSTRING(vv,charindex(':',vv)+1,len(vv))
else vv
end '编组'
from
(
select tname,
tstr,
v,
SUBSTRING(t.v, number ,CHARINDEX(',',t.v+',',number)-number) vv
from
(
select tname,
tstr,
SUBSTRING(t.tstr, number ,CHARINDEX(';',t.tstr+';',number)-number) v
from tbl t,master..spt_values s
where s.number >=1
and s.type = 'P'
and SUBSTRING(';'+t.tstr,s.number,1) = ';'
)t,master..spt_values s
where s.number >=1
and s.type = 'P'
and SUBSTRING(','+t.v,s.number,1) = ','
)t
go
exec dbo.proc_group
/*
编组
0101
0102
0103
0104
*/
------解决方案--------------------
这个在2000中,需要借助变量来实现:
declare @str varchar(1000)