Parquet Hadoop input plugin for Embulk

• Read Parquet files via Hadoop FileSystem.
• Outputs a single record named record (type is json).

Overview

• Plugin type: input
• Resume supported: no
• Cleanup supported: no
• Guess supported: no

Configuration

• config_files: list of paths to Hadoop's configuration files (array of strings, default: [])
• config: overwrites configuration parameters (hash, default: {})
• path: file path on Hdfs. you can use glob pattern (string, required).
• parquet_log_level: set log level of parquet reader module (string, default: "INFO")
  • value is one of java.util.logging.Level (ALL, SEVERE, WARNING, INFO, CONFIG, FINE, FINER, FINEST, OFF)

Hadoop Configuration

Describe parquet reader specific configuration.

• parquet.read.bad.record.threshold: Tolerated percent bad records per file (float, 0.0 to 1.0, default: 0)

Example

in:
  type: parquet_hadoop
  config_files:
    - /etc/hadoop/conf/core-site.xml
    - /etc/hadoop/conf/hdfs-site.xml
  config:
    parquet.read.bad.record.threshold: 0.01
  path: /user/hadoop/example/data/*.parquet
  parquet_log_level: WARNING

Build

$ ./gradlew gem  # -t to watch change of files and rebuild continuously

Install gem

$ embulk gem install <pkg/.gemfile>

Notes

Why implement this as input plugin rather than parser plugin ?

Because to parsing Parquet format needs seekable file stream but parser plugin has only sequential read.

How map Parquet schema to json ?

TBD