A plugin for Pentaho Data Integration (Kettle) that adds support for DuckDB table input/output.

The following should already be installed and configured on your local machine:

• Maven, version 3+
• Java JDK 11 (or OpenJDK)
• This settings.xml in your /.m2 directory

Begin by cloning this repository locally.

git clone https://github.com/forgineer/duckdb-kettle-plugin.git

After cloning, review the pom.xml file for the project <version> tag and ensure it is the same version as the Pentaho Data Integration (Kettle) version you are using. This version can be verified under the Maven repository.

...
<groupId>com.forgineer</groupId>
<artifactId>duckdb-kettle-plugin</artifactId>
<!-- The version should reflect the PDI (Kettle version) -->
<!-- Ex: 9.4.0.0-343, 9.3.0.0-428, etc. -->
<version>9.4.0.0-343</version>
...

Also, review the DuckDB version and update it to the desired version.

...
<dependency>
    <groupId>org.duckdb</groupId>
    <artifactId>duckdb_jdbc</artifactId>
    <!-- The version should reflect JDBC driver used and vice versa -->
    <version>0.9.2</version>
</dependency>
...

Lastly, update the JDBC driver jar file name and version to match the pom.xml file. This file can be verified by downloading from Maven or directly from DuckDB.

@Override
public String[] getUsedLibraries() {
    // The version should match POM
    return new String[] {"duckdb_jdbc-0.9.2.jar"};
}

After reviewing and/or updating the above, proceed with packaging the jar file.

mvn package 

This will create a jar and zip file inside of the target directory:

• duckdb-kettle-plugin-x.x.x.x-xxx.jar
• duckdb-kettle-plugin-x.x.x.x-xxx.zip

Unpack the zip file into the plugins directory (\data-integration\plugins) of your local Kettle install. The zip file should include the plugin and necessary JDBC driver for DuckDB.

data-integration\
    plugins\
        duckdb-kettle-plugin-x.x.x.x-xxx\
            lib\
                duckdb_jdbc-x.x.x.jar
            duckdb-kettle-plugin-x.x.x.x-xxx.jar

Restart Kettle (Spoon).

DuckDB should now be available as a connection type from a table input or output step. Only the database name is required to create the DuckDB file.

✋ Use caution when using Kettle properties within your database name parameter. In recent testing, this has caused errors with the interpreted file path that Kettle generates.