hive-行转列按顺序合并

目录

一、背景

二、实现

1.建表ddl

2.示例数据

3.按顺序合并

4.按顺序合并结果

5.可以看到最后一条最长的才是我们需要的数据

6.结果

---

---

一、背景

想实现行转列按顺序合并，但是impala不支持，故用hive实现

二、实现

1.建表ddl

create table a(

id bigint comment '主键',

type bigint comment '分类',

start_time bigint comment '开始时间，时间戳',

end_time bigint comment '结束时间，时间戳', 

text string comment '内容'

)stored as textfile;

2.示例数据

insert into a values(1,1,1648963299613,1648963319545,'哈哈');

insert into a values(2,1,1648963319545,1648963325635,'嘻嘻');

insert into a values(3,1,1648963325635,1648963331726,'嘿嘿');

insert into a values(4,1,1648963331726,1648963335417,'哦哦');

insert into a values(5,1,1648963335417,1648963346121,'嗯嗯');


insert into a values(6,2,1648964757698,1648964764500,'你');

insert into a values(7,2,1648964764500,1648964778843,'我');

insert into a values(8,2,1648964778843,1648964813889,'他');

insert into a values(9,2,1648964813889,1648964832633,'它');

insert into a values(10,2,1648964832633,1648964840608,'她');

insert into a values(11,2,1648965496307,1648965510042,'123');

insert into a values(12,2,1648965510042,1648965518099,'456');

3.按顺序合并

select 
     id
    ,type
    ,start_time
    ,end_time
    ,concat_ws(',',collect_set(text) over(partition by type order by start_time asc)) as  text_by_type
from a 
order by start_time

4.按顺序合并结果

5.可以看到最后一条最长的才是我们需要的数据

select *
from (
select 
     id
    ,type
    ,start_time
    ,end_time
    ,row_number() over(partition by type order by start_time desc) as rn
    ,concat_ws(',',collect_set(text) over(partition by type order by start_time asc)) as  text_by_type
from a 
order by start_time
) t      
where rn=1

6.结果