[JAVA][C++]Support Parquet Read and Write in Java
- Langage dominant
- Java
- Étoiles
- 94
- Forks
- 152
- Merge moyen
- 3 j 16 h
- PR mergées (30 j)
- 11
Description
We added a new java interface to support parquet read and write from hdfs or local file.
The purpose of this implementation is that when we loading and dumping parquet data in Java, we can only use rowBased put and get methods. Since arrow already has C++ implementation to load and dump parquet, so we wrapped those codes as Java APIs.
After test, we noticed in our workload, performance improved more than 2x comparing with rowBased load and dump. So we want to contribute codes to arrow.
since this is a total independent change, there is no codes change to current arrow codes. We added two folders as listed: java/adapter/parquet and cpp/src/jni/parquet
**Reporter**: [Chendi.Xue](https://issues.apache.org/jira/browse/ARROW-6720)
#### Related issues:
- [[Java][Dataset] Implement Datasets Java API ](https://github.com/apache/arrow/issues/17055) (incorporates)
- [[Java][Dataset] Support writing to files within dataset scanner via JNI](https://github.com/apache/arrow/issues/27628) (incorporates)
#### PRs and other links:
- [GitHub Pull Request apache/arrow#5522](https://github.com/apache/arrow/pull/5522)
- [GitHub Pull Request apache/arrow#5717](https://github.com/apache/arrow/pull/5717)
- [GitHub Pull Request apache/arrow#5719](https://github.com/apache/arrow/pull/5719)
**Note**: *This issue was originally created as [ARROW-6720](https://issues.apache.org/jira/browse/ARROW-6720). Please see the [migration documentation](https://github.com/apache/arrow/issues/14542) for further details.*
Guide de contribution
Ouvrir le guide de contribution
Piste de recherche
Commencez par examiner java/adapter/parquet et cpp/src/jni/parquet, puis comparez les pull requests liés #5522, #5717 et #5719 avec les issues associées aux datasets. Vérifiez les API Java pour la lecture et l’écriture de Parquet dans HDFS et dans des fichiers locaux, ainsi que leurs tests ; le travail est considéré comme terminé lorsque la prise en charge requise est présente sans dupliquer le travail sur les datasets déjà intégré.
Rédigé par le modèle d'indexation à partir du texte de l'issue.
Évaluation
- Stack technique
- cpp, java
- Domaine
- backend, data
- Type d'issue
- Fonctionnalité
- Difficulté
- 4/5
- Temps estimé
- 3-5 jours
- Activité
- À l'abandon
- Clarté
- Plutôt claire
- Accessibilité débutants
- 20/100