[
https://issues.apache.org/jira/browse/FLINK-31207?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18096173#comment-18096173
]
Dominik Bünzli edited comment on FLINK-31207 at 7/14/26 12:53 PM:
------------------------------------------------------------------
Hi [~twalthr], Hi @[~martijnvisser]
I took some time to do a PoC implementation for the higher order functions
"ARRAY_FILTER" and "TRANSFORM", which allow filtering elements from arrays
based on a predicate and applying lambda-based transformations to arrays or
maps.
They can be used like:
====================================================================
SELECT id, ARRAY_FILTER(vals, x -> x > 2)
FROM (VALUES (1, ARRAY[1, 2, 3, 4]), (2, ARRAY[0, 1]), (3, CAST(NULL AS
ARRAY<INT>)))
AS t(id, vals)
====================================================================
SELECT TRANSFORM(MAP['a', 1, 'b', 2], (k, v) -> v * 100)
====================================================================
SELECT TRANSFORM(ARRAY['a', 'bb', 'ccc'], s -> CHAR_LENGTH(s))
====================================================================
I would love to contribute my code, do I need to create a FLIP? If so, I read
that I need to open a Google Doc. Anything else needed or just sending the
Google Doc to the mailing list?
Kind Regards
Dominik
was (Author: JIRAUSER309386):
Hi [~twalthr], Hi @[~martijnvisser]
I took some time to do a PoC implementation for the higher order functions
"ARRAY_FILTER" and "TRANSFORM", which allow filtering elements from arrays
based on a predicate and applying lambda-based transformations to arrays or
maps.
They can be used like:
SELECT id, ARRAY_FILTER(vals, x -> x > 2)
FROM (VALUES (1, ARRAY[1, 2, 3, 4]), (2, ARRAY[0, 1]), (3, CAST(NULL AS
ARRAY<INT>)))
AS t(id, vals)
SELECT TRANSFORM(MAP['a', 1, 'b', 2], (k, v) -> v * 100)
SELECT TRANSFORM(ARRAY['a', 'bb', 'ccc'], s -> CHAR_LENGTH(s))
I would love to contribute my code, do I need to create a FLIP? If so, I read
that I need to open a Google Doc. Anything else needed or just sending the
Google Doc to the mailing list?
Kind Regards
Dominik
> Supports high order function like other engine
> ----------------------------------------------
>
> Key: FLINK-31207
> URL: https://issues.apache.org/jira/browse/FLINK-31207
> Project: Flink
> Issue Type: New Feature
> Reporter: Jacky Lau
> Priority: Major
>
> spark [https://spark.apache.org/docs/latest/api/sql/index.html#transform]
> transform/transform_keys/transform_values
> after calcite https://issues.apache.org/jira/browse/CALCITE-3679s upports
> high order functions, we should supports many high order funcsions like
> spark/presto
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
