Skip to content

JDBC driver (datapress-jdbc)

datapress-jdbc is a read-only, Type‑4 (pure Java, no native components) JDBC driver for a running DataPress server. It wraps the HTTP + Arrow IPC API behind the standard java.sql interfaces, so BI tools, SQL consoles, and JVM applications can talk to DataPress without knowing anything about its REST endpoints.

  • Read-only and analytics-oriented — SELECT, WITH … SELECT, and DESCRIBE only.
  • Arrow-native — results stream as Apache Arrow IPC and are decoded lazily, so large result sets use roughly constant memory.
  • Tool compatible — DBeaver, DataGrip, and other clients see a synthesized DatabaseMetaData (catalogs, schemas, tables, columns) rather than any SQL rewriting.

Requires the raw-SQL endpoint

The driver issues SQL against the server's POST /api/v1/sql endpoint, which is disabled by default. Enable it server-side with [sql].enabled = true (or sql_enabled=True when launching from Python) before connecting.

Install

The driver is published on Maven Central as org.datap-rs:datapress-jdbc:0.1.0.

<dependency>
  <groupId>org.datap-rs</groupId>
  <artifactId>datapress-jdbc</artifactId>
  <version>0.1.0</version>
</dependency>

implementation("org.datap-rs:datapress-jdbc:0.1.0")

implementation 'org.datap-rs:datapress-jdbc:0.1.0'

For GUI tools, download the self-contained (shaded) jar directly — it bundles a relocated Arrow, Jackson, and FlatBuffers, so no extra jars are needed:

https://repo1.maven.org/maven2/org/datap-rs/datapress-jdbc/0.1.0/datapress-jdbc-0.1.0.jar

JDK 16+ JVM flag

Arrow's off-heap memory needs the java.nio module opened. Add this to the JVM running the driver (your application, DBeaver, DataGrip, …):

--add-opens=java.base/java.nio=ALL-UNNAMED

Connection URL

jdbc:datapress://host[:port][/][?token=<bearer>&tls=false&...]
Part Meaning
host DataPress server host.
port Optional server port.
token Bearer token for a server with auth enabled. May also be supplied as the JDBC password / a token connection property.
tls Set tls=false to talk plain HTTP to a non-TLS server.

The driver class is org.datapress.jdbc.DataPressDriver. On a modern JDBC runtime it registers itself via the service loader — no Class.forName(...) needed.

Java example

import java.sql.*;

public class Demo {
    public static void main(String[] args) throws Exception {
        String url = "jdbc:datapress://127.0.0.1:8000/?tls=false";

        try (Connection conn = DriverManager.getConnection(url);
             Statement stmt = conn.createStatement();
             ResultSet rs = stmt.executeQuery(
                 "SELECT State, COUNT(*) AS n FROM accidents GROUP BY State ORDER BY n DESC")) {

            while (rs.next()) {
                System.out.printf("%s: %d%n", rs.getString("State"), rs.getLong("n"));
            }
        }
    }
}

With auth enabled, pass the token via properties instead of the URL:

java.util.Properties props = new java.util.Properties();
props.setProperty("token", System.getenv("DATAPRESS_TOKEN"));
Connection conn = DriverManager.getConnection("jdbc:datapress://my-host:8000/", props);

Using it with DBeaver

The driver exposes tables and columns through DatabaseMetaData, so DBeaver's database navigator, data preview, and SQL editor all work.

1. Register the driver

Database → Driver Manager → New, then set:

Field Value
Driver Name DataPress
Class Name org.datapress.jdbc.DataPressDriver
URL Template jdbc:datapress://{host}:{port}/
Default Port 8000

Under Libraries, either Add Artifact with the Maven coordinates org.datap-rs:datapress-jdbc:0.1.0 (DBeaver downloads it from Maven Central), or Add File and select the shaded jar you downloaded above. "Find Class" should then resolve org.datapress.jdbc.DataPressDriver. Click OK.

2. Create a connection

  • JDBC URL: jdbc:datapress://127.0.0.1:8000/ (append ?tls=false for a plain-HTTP server).
  • Leave user/password empty for a server without auth. If your server requires a bearer token, add it as a connection property named token (or as the password).
  • Test Connection should succeed — the driver performs a GET /version handshake.

3. What you should see

Expand the connection in the navigator:

  • Catalog datapress → Schema main.
  • Your datasets appear as tables; expanding a table's Columns shows the mapped SQL types (e.g. BIGINT, VARCHAR, BOOLEAN, DOUBLE, TIMESTAMP WITH TIME ZONE).
  • Double-click a table → Data tab to preview rows, or open a SQL Editor and run a SELECT.

Limitations (by design)

  • Read-only. Only SELECT / WITH … SELECT / DESCRIBE are accepted.
  • One dataset per SQL statement — matches the server's raw-SQL endpoint (no cross-dataset joins yet).
  • No server-side parameter binding. PreparedStatement substitutes parameters client-side.
  • The server's raw-SQL endpoint must be enabled ([sql].enabled = true).

See also