Question

我正在使用Oracle数据库，我有一个查询，我必须执行7个不同表的连接。现在，我的问题是，我需要获得满足连接要求的行（显然），但即使它们与上一次连接的条件不匹配，我需要从前6个表中获取行。我不能做左外连接，所以我有什么替代方案？

代码看起来像这样：

    with
tmp as (select col1, col2, col3, col4, row_number() over (partition by col1 order by col2 desc) rn
            from
            (select /*+ MATERIALIZE */
              col1, col2, col3, col4
             from
              table1
             where 
              col3 in ('A','R','F') and
              somedate >= sysdate-720 and
              col5 is null
              and col1<> '0000000000'))
select /*+ use_hash(a,b,c,d,e,f,g,h) */
b.col5,
a.col1, 
 d.col6, 
 e.col7, 
 c.col8 , 
 (CASE when f.col9= 'B' then 'Foo' else 'Bar' END) as "col9", 
 a.col2, 
 a.col3,
 h.col10
from tmp a
     join table2 b on
      a.col1= b.col1 and
      a.col4=b.col4 and
      b.col11='P' and
      (b.otherDate>= sysdate OR b.otherDate is null)  and
      b.col5 is null 
     join table3 c on
      b.col12 = c.col12 and
      (c.otherDate is null or b.otherDate >= sysdate) and
      c.col5 is null 
     join table4 d on
      a.col1= d.col1 and
      d.col13 in ('R','A','F') and
      d.col5 is null 
     join table5 e on
      e.col1=b.col1 and
      e.col14=d.col14 and
      d.col6=e.col6 and
      d.col15 = e.col15 and
      e.col5 is null       
    join table6 f on
      f.col4= a.col4 and 
      f.col5 is null
    join table7 g on 
      g.col16=  case when f.col15 is null then null else f.col15 end
      and g.col5is null
      and (g.otherDate is null or g.otherDate >= sysdate)
    join table8 h on 
      h.col17= g.col17
      and (h.otherDate >= sysdate or h.otherDate is null)
      and h.col5 is null
      and a.rn=1;

Answer 1

我不打算尝试使用您的实际查询，但原则上您可以更改：

select tab1.col1, tab2.col2, tab3.col3
from tab1
join tab2 on tab2.fk = tab1.pk
join tab3 on tab3.fk = tab2.pk

成：

select tab1.col1, tab2.col2, tab3.col3
from tab1
join tab2 on tab2.fk = tab1.pk
left join tab3 on tab3.fk = tab2.pk

你可以用以下内容替换（在你的外联接 - 不允许的世界中）：

with tmp as (
    select tab1.col1, tab2.col2, tab3.pk
    from tab1
    join tab2 on tab2.fk = tab1.pk
)
select tmp.col1, tmp.col2, tab3.col3
from tmp
join tab3 on tab3.fk = tmp.pk
union all
select tmp.col1, tmp.col2, null as col3
from tmp
where not exists (
    select null from tab3
    where tab3.fk = tmp.pk
)

哪个非常难看 - 我用CTE最小化了重复次数，但即便如此也不好 - 并且很可能表现得不如外连接那么好。

当然，在不知道为什么你不能使用外连接的情况下，我不知道是否还有其他限制会使这种方法不可接受......

从第一个表中获取行，即使它们在连接的最后一个表中不存在而没有左连接

1 个答案: