[IMPALA-10920] UNNEST function for arrays in the select list

• Type: Sub-task
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: Impala 4.1.0
• Component/s: Backend, Frontend
• Labels: complextype
• Epic Color: ghx-label-3
• Assignee: Gabor Kaszab
• Reporter: Gabor Kaszab
• Created: 17/Sep/21 08:39
• Updated: 23/Oct/24 18:20
• Resolved: 23/Nov/21 07:38

Description

There is a need for implementing an UNNEST(array) function that could be given in the select list so that we can replicate Postgres functionality that has the very same.

Considering the following table:

CREATE TABLE tbl (
  id int,
  arr1 array<int>,
  arr2 array<int>)
STORED AS PARQUET;

with this data:

1, {1,2}, {11,22,33}

A sample query would look like this:

SELECT id, UNNEST(arr1), UNNEST(arr2) FROM tbl;

id	unnest(arr1)	unnest(arr2)
1	1	11
1	2	22
1	null	33

Not that the expected is to 'zip' the arrays instead of joining their values to have the same behaviour as Postgres 10+.

Issue Links

• is depended upon by: IMPALA-10977 Make zipping unnest work with absolute paths
• is related to: IMPALA-9509 array_contains() for array complex type
• is related to: IMPALA-11038 Zipping unnest should work on arrays from views
• relates to: IMPALA-11059 Speed up zipping unnest by reading collection elements in batches