0

我想从三个表中选择值。每个表都有一个buyer_entity_id 列。

我想出了在其中两个表上进行外部联接的语法，但是如何添加第三个表却让我望而却步。

这是两个表连接的语句，它完全按照我希望的方式工作：

select * from (select  b.buyer_entity_id, count(distinct(a.row_id)) as imps 
from imps a, anl_line b
where b.line_item_id=a.buyer_line_id 
and a.entity_id=3
group by b.buyer_entity_id
order by b.buyer_entity_id) tab1
full outer join (select b.buyer_entity_id, count(distinct(a.row_id)) as clicks 
from clicks a, anl_line b 
where a.buyer_line_id=b.line_item_id 
and a.entity_id=3
group by b.buyer_entity_id 
order by b.buyer_entity_id) tab2
on tab1.buyer_entity_id = tab2.buyer_entity_id;

第三个表将具有相同的 select 语句，并且也将连接到 buy_entity_id 值。但是，当我添加第三个选择语句时，我收到“缺少关键字”错误。下面是我的三路全外连接语句：

select * from ((select  b.buyer_entity_id, count(distinct(a.row_id)) as imps 
from imps_table a, line_table b
where b.line_item_id=a.buyer_line_id 
and a.entity_id=3
group by b.buyer_entity_id
order by b.buyer_entity_id) tab1
full outer join (select b.buyer_entity_id, count(distinct(a.row_id)) as clicks 
from clicks_table a, line_table b 
where a.buyer_line_id=b.line_item_id 
and a.entity_id=3
group by b.buyer_entity_id 
order by b.buyer_entity_id) tab2)
outer join (select  b.buyer_entity_id, count(distinct(a.row_id)) as vers 
from vers_table a, line_table b
where b.line_item_id=a.buyer_line_id 
and a.entity_id=3
group by b.buyer_entity_id
order by  b.buyer_entity_id) tab3
on tab1.buyer_entity_id = tab2.buyer_entity_id and tab2.buyer_entity_id=tab3.buyer_entity_id;
4

3 回答 3

1

错误在这里：order by b.buyer_entity_id) tab2)。您需要一个on子句来指定 tab1 和 tab2 之间的连接条件。order by b.buyer_entity) tab1 on <join condition)

一旦解决了。您将需要添加full到下一行，以便outer join (select b.buyer_entity_id, count(distinct(a.row_id)) as vers变为full outer join (select b.buyer_entity_id, count(distinct(a.row_id)) as vers.

这涵盖了语法问题，但不包括语义问题。

连接是成对的，表 1 在某些条件下连接到表 2，然后结果成为下一个连接的左侧。如果 tab1 和 tab3 匹配，但 tab2 不匹配，您希望发生什么？我猜想用 tab1 和 tab3 数据和 tab2 空值生成一行。就像是：

select * 
from (select  b.buyer_entity_id, count(distinct(a.row_id)) as imps 
        from imps_table a, line_table b
        where b.line_item_id=a.buyer_line_id 
        and a.entity_id=3
        group by b.buyer_entity_id) tab1
full outer join (select b.buyer_entity_id, count(distinct(a.row_id)) as clicks 
        from clicks_table a, line_table b 
        where a.buyer_line_id=b.line_item_id 
        and a.entity_id=3
        group by b.buyer_entity_id) tab2
    on tab1.buyer_entity_id = tab2.buyer_entity_id
full outer join (select  b.buyer_entity_id, count(distinct(a.row_id)) as vers 
        from vers_table a, line_table b
        where b.line_item_id=a.buyer_line_id 
        and a.entity_id=3
        group by b.buyer_entity_id) tab3
    on tab3.buyer_entity_id in (tab1.buyer_entity_id, tab2.buyer_entity_id)
order by coalesce(tab1.buyer_entity_id
    , tab2.buyer_entity_id
    , tab3.buyer_entity_id);

另外，请order by在子查询中丢失。他们没有为您做任何事情，因为不能保证订单在连接中存在。

于 2011-11-04T20:21:12.233 回答
1

您可以使用 WITH 子句进行一些简化：

使用 first_part 作为 ( ( select .... ) 外连接 ( select .... )); select * from first_part 外连接（选择 ....）；

不过，您可能想重新审视整体逻辑，看看是否有更好的方法来完成您正在尝试做的事情。多个外连接通常不是最有效的解决方案。

于 2011-11-04T19:49:53.880 回答
1

一种更快的方法将如此处所述 https://forums.oracle.com/thread/2388229

SELECT       COALESCE (a.id, b.id, c.id)     AS common_id
,       NVL2 (a.id, 1, NULL)           AS table_a_flag
,       NVL2 (b.id, 1, NULL)           AS table_b_flag
,       NVL2 (c.id, 1, NULL)           AS table_c_flag
FROM              table_a  a
FULL OUTER JOIN  table_b  b  ON  b.id  =           a.id
FULL OUTER JOIN      table_c  c  ON  c.id  = COALESCE (a.id, b.id)
ORDER BY  common_id;
于 2013-06-17T09:30:23.383 回答