Pandas-like data science and ETL in pure Java, executed through JShell with Maven dependencies resolved automatically at run time.

Provided free and as is, without warranty of any kind — including merchantability, fitness for a particular purpose, and the accuracy or completeness of any result. See the licence. You are responsible for checking what this produces before relying on it.

DataFrame Tool

Pandas-like data science and ETL in pure Java using JShell with automatic Maven dependency resolution.

Tool ID

dataframe

Credential Required

Operations

1. `execute` — Run Java code in JShell

Property	Type	Default	Description
`code`	string	—	Required. Java code to execute in JShell
`libraries`	string	—	XML-formatted Maven dependency declarations
`workingDir`	string	`INTOUCH_HOME/tasks/dataframe`	Working directory for execution
`jshellCommand`	string	`jshell`	Path to jshell executable

Maven Dependencies

Specify dependencies using XML format in the libraries property:

<dependency>
    <groupId>tech.tablesaw</groupId>
    <artifactId>tablesaw-core</artifactId>
    <version>0.43.1</version>
</dependency>
<dependency>
    <groupId>com.google.code.gson</groupId>
    <artifactId>gson</artifactId>
    <version>2.10.1</version>
</dependency>

Dependencies are: - Downloaded from Maven Central on first use - Cached in {workingDir}/lib/ for subsequent runs - Automatically added to the JShell classpath

Code Environment

Full Java language support via JShell
returnValue variable (int) controls exit code — set it to non-zero for warning/failure
##intouch[set name=X value=Y] property markers work via System.out.println()
All JShell prompts and noise are filtered from output

Example Code

// Import Tablesaw
import tech.tablesaw.api.*;
import tech.tablesaw.io.csv.*;

// Read a CSV
Table sales = Table.read().csv("/data/sales.csv");

// Filter and aggregate
Table result = sales
    .where(sales.stringColumn("region").isEqualTo("East"))
    .summarize("amount", AggregateFunctions.sum).by("product");

// Write results
result.write().csv("/data/east_sales_summary.csv");

// Publish a property
System.out.println("##intouch[set name=rowCount value=" + result.rowCount() + "]");

Published Outputs: - exitCode — JShell exit code - output — filtered stdout - Any properties set via ##intouch[set name=X value=Y] markers

Killable

Yes — destroys the JShell process forcibly.

Technology

JShell — Java REPL (JDK 9+)
Maven Aether — automatic dependency resolution from Maven Central
Tablesaw — Java DataFrame library (up to 2 billion rows)

Chaining Patterns

SQL export → DataFrame — export data to CSV, process with Tablesaw
DataFrame → SQL import — process data, import results into database
DataFrame → Email — generate analytics, email results
DataFrame → PDF — analyze data, generate PDF report
FTP download → DataFrame — download data files, process them

Use Cases

Data transformation and ETL
Statistical analysis and aggregation
CSV/data file manipulation
Report generation from structured data
Any Java library available on Maven Central

Limitations

Requires JShell (JDK 11+)
Dependencies must be in Maven Central
No persistent state between executions
Local repository cleaned up after each run
No graphical output (server-side execution)

Publishes

The keys a later task may reference as {{taskName.key}}. Referencing anything not listed here resolves to nothing at run time.

exitCode
output

DataFrame