列转行

SELECT id, 'c1' cn, c1 cv
   FROM t_col_row
UNION ALL
SELECT id, 'c2' cn, c2 cv
   FROM t_col_row
UNION ALL
SELECT id, 'c3' cn, c3 cv FROM t_col_row;


select * from t_col_row;
select j.id ,
  max(decode(j.cn,'c1',j.cv)) as a , 
  max(decode(j.cn,'c2',j.cv)) as b 
 from 
(SELECT id, 'c1' cn, unitname cv
   FROM test
UNION ALL
SELECT id, 'c2' cn, wentiname cv
   FROM test)   j 
   group by j.id

select * from test for update ;
select   m.case_num,                                                       
     max(decode(m.unit_code,'2',m.unit_name)) as a , 

     max(decode(m.unit_code,'3',m.unit_name)) as b , 

     max(decode(m.unit_code,'4',m.unit_name)) as c    
  from   jjkg_ss_sjmj  m                            

  group by m.case_num 


  SELECT id, 'c1' cn, unitname cv
   FROM test
UNION ALL
SELECT id, 'c2' cn, wentiname cv
   FROM test;


   select * from (select m.case_num ,m.police_num from jjkg_ss_sjmj m )

   select id ,unitname||':'||wentiname as b  from test 


Oracle列转行拼接及多行拼接2010-09-28 12:00表结构和数据如下（表名Test）： 
NO VALUE NAME
1       a       测试1
1       b       测试2
1       c       测试3
1       d       测试4
2       e       测试5
4       f        测试6
4       g       测试7

Sql语句：
select No,
       ltrim(max(sys_connect_by_path(Value, ';')), ';') as Value,
       ltrim(max(sys_connect_by_path(Name, ';')), ';') as Name
from (select No,
               Value,
               Name,
               rnFirst,
               lead(rnFirst) over(partition by No order by rnFirst) rnNext
          from (select a.No,
                       a.Value,
                       a.Name,
                       row_number() over(order by a.No, a.Value desc) rnFirst
                  from Test a) tmpTable1) tmpTable2
start with rnNext is null
connect by rnNext = prior rnFirst
group by No;

检索结果如下：

NO VALUE    NAME
1    a;b;c;d   测试1;测试2;测试3;测试4
2    e            测试5
4    f;g          测试6;测试7

简单解释一下那个Sql吧：
1、最内层的Sql（即表tmpTable1），按No和Value排序，并列出行号：
select a.No,
       a.Value,
       a.Name,
       row_number() over(order by a.No, a.Value desc) rnFirst
from Test a
该语句结果如下：
NO VALUE NAME RNFIRST
1     d       测试4     1
1     c       测试3     2
1     b       测试2     3
1     a       测试1     4
2     e       测试5     5
4     g       测试7     6
4     f       测试6     7


2、外层的Sql（即表tmpTable2），根据No分区，取出当前行对应的下一条记录的行号字段：
select No,
       Value,
       Name,
       rnFirst,
       lead(rnFirst) over(partition by No order by rnFirst) rnNext
from (这里是tmpTable1的SQL) tmpTable1

lead(rnFirst)：取得下一行记录的rnFirst字段
over(partition by No order by rnFirst) 按rnFirst排序，并按No分区，分区就是如果下一行的No字段与当前行的No字段不相等时，不取下一行记录显示
该语句结果如下：
NO VALUE NAME RNFIRST RNNEXT
1     d        测试4     1         2
1     c        测试3      2         3
1     b        测试2     3         4
1     a        测试1     4         NULL
2     e        测试5     5         NULL
4     g        测试7     6         7
4     f         测试6     7         NULL


3、最后就是最外层的sys_connect_by_path函数与start递归了
sys_connect_by_path(Value, ';')
start with rnNext is null
connect by rnNext = prior rnFirst
这个大概意思就是从rnNext为null的那条记录开始，递归查找，
如果前一记录的rnFirst字段等于当前记录的rnNext字段，就把2条记录的Value用分号连接起来，
大家可以先试试下面这个没有Max和Group的Sql：
select No,
       sys_connect_by_path(Value, ';') as Value,
       sys_connect_by_path(Name, ';') as Name
from (select No,
               Value,
               Name,
               rnFirst,
               lead(rnFirst) over(partition by No order by rnFirst) rnNext
          from (select a.No,
                       a.Value,
                       a.Name,
                       row_number() over(order by a.No, a.Value desc) rnFirst
                  from Test a) tmpTable1) tmpTable2
start with rnNext is null
connect by rnNext = prior rnFirst

结果是：
NO VALUE       NAME
1     ;a            ;测试1
1     ;a;b         ;测试1;测试2
1     ;a;b;c     ;测试1;测试2;测试3
1     ;a;b;c;d ;测试1;测试2;测试3;测试4
2     ;e            ;测试5
4     ;f             ;测试6
4     ;f;g          ;测试6;测试7

可以看到，每个No的最后一条记录就是我们要的了
所以在sys_connect_by_path外面套一个Max，再加个Group by No，得到的结果就是行转列的结果了
最后再加一个Ltrim，去掉最前面的那个分号，完成。

[color=red][b]奶奶的，最终的答案是：[/b][/color]
select id,
       ltrim(max(sys_connect_by_path(s, ';')), ';') as unitname

from (select id,
               s,

               rnFirst,
               lead(rnFirst) over(partition by id order by rnFirst) rnNext
          from (select a.id,
                       a.s,
                       row_number() over(order by a.id, a.s desc) rnFirst
                  from (select v.id,v.unitname||':'||v.wentiname as s from test v) a) tmpTable1) tmpTable2
start with rnNext is null
connect by rnNext = prior rnFirst
group by id;