Question

我有一个看起来像这样的表：

station     node        group
13859160    13195576    high
13859160    42502030    low
13859165    42283197    low
13859165    42283198    high
13859166    13190800    low
13859166    13190801    low
13859166    13195587    high

可以使用以下语句创建：

CREATE TABLE nodes
    ("station" int, "node" int, "group" varchar(6))
;

并填写了这句话：

insert into nodes values (13859160, 13195576,   'high');
insert into nodes values (13859160, 42502030,   'low');
insert into nodes values (13859165, 42283197,   'low');
insert into nodes values (13859165, 42283198,   'high');
insert into nodes values (13859166, 13190800,   'low');
insert into nodes values (13859166, 13190801,   'low');
insert into nodes values (13859166, 13195587,   'high');

我想查询此表，以便高和低成为列，因此结果如下所示：

Station     high        low
13859160    13195576    42502030
13859165    42283198    42283197
13859166    13195587    13190801
13859166    13195587    13190800

如果那里没有多个“低”行的电台，那就很容易了，我可以做这样的事情来获得高点和低点：

select 
    "station",
    case when "group" = 'high' then "node" end as high,
    case when "group" = 'low' then "node" end as low
from "NODES";

导致：

然后使用聚合函数获取结果，如：

select 
    "station",
    max(case when "group" = 'high' then "node" end) as high,
    max(case when "group" = 'low' then "node" end) as low
from "NODES"
group by "station";

这导致：

但是，当1个startion有多个“low”enties时，我需要为每个低项创建一行，该行应该在两行中包含该站的相同“高”条目。

知道怎么做到这一点？

Answer 1

您可以使用条件聚合执行此操作。

select * from (
select 
    "station",
    max(case when "group" = 'high' then "node" end) over(partition by "station") as high,
    case when "group" = 'low' then "node" end as low
from nodes
) n
where low is not null

Answer 2

首先，我可能会建议listagg()进行聚合查询：

select "station",
       listagg(case when "group" = 'high' then "node" end) as high,
       listagg(case when "group" = 'low' then "node" end) as low
from "NODES"
group by "station";

要为单个工作站获取多行，您可以使用row_number()：

select "station",
       max(case when "group" = 'high' then "node" end) as high,
       max(case when "group" = 'low' then "node" end) as low
from (select n.*,
             row_number() over (partition by "station", "group" order by "node") as seqnum
      from "NODES" n
     ) n
group by "station", seqnum;

最后，如果您想要重复上一个最大值：

select "station",
       coalesce(max(case when "group" = 'high' then "node" end),
                max(max(case when "group" = 'high' then "node" end)) over (partition by "station)
               ) as high,
       max(case when "group" = 'low' then "node" end) as low
from (select n.*,
             row_number() over (partition by "station", "group" order by "node") as seqnum
      from "NODES"
     ) n
group by "station", seqnum;

Oracle将行转换为列

2 个答案: