[FLINK-32262] Add MAP_ENTRIES support in SQL & Table API

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.18.0
• Fix Version/s: 1.19.0
• Component/s: Table SQL / Planner
• Labels: pull-request-available stale-assigned
• Assignee: Hanyu Zheng
• Reporter: Bonnie Varghese
• Created: 05/Jun/23 18:25
• Updated: 20/Feb/24 17:46
• Resolved: 20/Feb/24 17:44

Description

Implement the map_entries method to transform a map into an array of key-value structs.

Description: The current implementation of the map_entries method in the Flink library does not provide a way to transform a map into an array of key-value structs. This enhancement aims to add this functionality, allowing users to convert a map into a more structured format for further processing.

Syntax:

 

 

map_entries[map: Map[K, V]] -> Array[Struct<K, V>] 

Arguments:

• map: The input map to be transformed.

Returns: An array of key-value structs obtained from the input map. Each struct consists of two fields: key of type K and value of type V.

Examples:

1. Transforming a map into key-value structs:

 
 

input_map = [1: 'apple', 2: 'banana', 3: 'cherry'] 
 map_entries[input_map] 
 Output: [{'key': 1, 'value': 'apple'}, {'key': 2, 'value': 'banana'}, {'key': 3, 'value': 'cherry'}]

1. Handling an empty map:

empty_map = {} 
map_entries[empty_map]
Output: []

See also:

spark:https://spark.apache.org/docs/latest/api/sql/index.html#map_entries

Issue Links

• links to: GitHub Pull Request #23098