hadoop - 从 Hive 中的 collect_list 结果构造映射

Question

一系列 UNION ALL 生成我想用来构建 MAP 的键值对列表。

所需的功能是这样的:

select id1, id2, map(collect_list(col)) as measurements
from
(
    select id1, id2, "height" as col
    union all
    select id1, id2, count(*) as col from table1
    union all
    select id1, id2, "weight" as col
    union all
    select id1, id2, count(*) as col from table2
)

什么是正确的实现方式？

我希望得到的结果是:

id1  id2  measurements
1    10   {"height": 10, "weight": 20}
2    20   {"height": 10, "weight": 20}

Answer 1

你的要求不是很清楚，但看起来你需要类似的东西

select id1, id2, named_struct("height", height, "weight", weight) from 
(
select t1.id1,t1,id2,height,weight from 
(select id1, id2, count(*) as height from table1 group by id1,id2) t1
join 
(select id1, id2, count(*) as height from table2 group by id1,id2) t2
on t1.d1=t2.d1 and t1.d2=t2.d2
) t;

我没有运行它，但它应该可以运行。