StarRocks/starrocks

Support converting JSON to STRUCT

Open

Aperta il 25 mar 2024

Vedi su GitHub
 (12 commenti) (2 reazioni) (1 assegnatario)Java (5717 star) (1246 fork)batch import
good first issuetype/feature-requestversion:3.2.9version:3.3.0version:3.4

Descrizione

Feature request

Is your feature request related to a problem? Please describe.

STRUCT type is more performant that JSON in terms of query, but which use the fixed schema and annoying to construct.

If it's possible to convert JSON to STRUCT directly, like to_struct(JSON), it can be much easier to use.

Describe the solution you'd like

  1. Use a function to convert JSON to STRUCT:
    • Example: select to_struct(parse_json("{"a": 1}"))
    • Difficulty: cannot inference the schema of struct before parsing the JSON
  2. Specify the STRUCT schema and convert:
    • Example: select to_struct<a int, b int>(parse_json("a": 1, "b": 2))
    • Difficulty: not easy to use

Describe alternatives you've considered

Additional context

Guida contributor

Tech stack
javasql
Dominio
backenddatabase
Tipo issue
feature
DifficoltàQuanto dovrebbe essere impegnativa per un nuovo contributor.
4
Tempo stimatoTempo stimato per completare e verificare uno scope piccolo.
over 1 week
Stato attivitàQuanto è probabile che la issue sia ancora attiva e reviewable.
blocked
ChiarezzaQuanto chiaramente la issue descrive problema, scope e risultato atteso.
mostly clear
Prerequisiti
Familiarity with StarRocks codebaseKnowledge of JSON and STRUCT typesUnderstanding of SQL function registration
Adatta ai principiantiQuanto la issue è adatta a contributor alla prima esperienza.
15
Direzione di ricerca
Examine the three linked PRs (#47233, #46448, #45406) to understand previous work on type conversions in StarRocks. Review the issue comments for any maintainer suggestions on the preferred approach. Investigate the SQL parser and type system in the StarRocks codebase to determine how to register a new function like `to struct`. Consider the schema inference challenge mentioned in the issue.