LEFT JOIN具有特定条件
问题描述:
I have two tables
table A
---------
uid c1(distinct)
---------
1 a
1 b
2 c
3 d
3 e
3 f
4 g
and
table B
---------
uid c2(non unique)
---------
1 x
2 y
3 z
3 z
3 v
I need a resultant table which maps table A to table B on uid like this
Result Table
-------------
uid c1 c2
-------------
1 a x
1 b -
2 c y
3 d z
3 e z
3 f v
4 g -
Basically once a row from table B is used in the resultant table, it must be exhausted. None of the JOINS work in this case.
Both these tables contain large number of records, so I can't write individual queries for each uid.
我有两个表 p>
表A
-- -------
uid c1(distinct)
---------
1 a
1 b
2 c
3 d
3 e
3 f
4 g
code> pre>
和 p>
表B
---------
uid c2 (非唯一)
---------
1 x
2 y
3 z
3 z
3 v
code> pre>
我需要一个结果表,它将表A映射到uid上的表B,如此 p>
结果表
------------- \ nuid c1 c2
-------------
1 ax
1 b -
2 cy
3 dz
3 ez
3 fv
4 g -
code> pre>
基本上,一旦在结果表中使用了表B中的一行,就必须将其耗尽。
在这种情况下,没有JOINS工作。 p>
这两个表都包含大量记录,因此我无法为每个uid编写单独的查询。 p>
div>
答
Let's count the rows with the same uid and join equal uids and that numbers
select t.uid, c1, ifnull(c2, '-') c2
from
(select *, @n:=if(@i=uid, @n+1, if(@i:=uid, 1, 1)) n
from tableA
cross join
(select @n:=0, @i:='') param1
order by uid
) t
left join
(select *, @n1:=if(@i1=uid, @n1+1, if(@i1:=uid, 1, 1)) n1
from tableB
cross join
(select @n1:=0, @i1:='') param2
order by uid
) tt
on t.uid=tt.uid and t.n=tt.n