How to sort an array and return the index in hive?

user_12746574 0
Viewed 396

In hive, I wish to sort an array from largest to smallest, and get the index array.

For example, the table is like this:

id  |  value_array
 1  |  {30, 40, 10, 20}
 2  |  {10, 30, 40, 20}

I with to get this:

id  |  value_array
 1  |  {1, 0, 3, 2}
 2  |  {2, 1, 3, 0}

The arries in result are the index of the initial elements. How can I achieve this?

hiveql hive arrays sql
1 Answers

Explode array using posexplode to get index and value, sort by value, collect array of index:

select id, collect_list(pos) as result_array
from
(
select s.id, a.pos, a.v 
  from your_table s
       lateral view posexplode(s.value_array) a as pos, v
distribute by s.id sort by a.v DESC --sort by value
)s
group by id
;

Tested, result:

id  result_array
1   [1,0,3,2]
2   [2,1,3,0]
user_27003440
Related