Revert(databricks-jdbc-driver): Switch to the latest OSS Databricks JDBC driver

This reverts commit 29fdd61.

Author: KSDaemon

packages/cubejs-backend-shared/src/env.ts

@@ -952,15 +952,9 @@ const variables: Record<string, (...args: any) => any> = {
    * Accept Databricks policy flag. This environment variable doesn't
    * need to be split by the data source.
    */
-  databrickAcceptPolicy: () => {
-    const val = get('CUBEJS_DB_DATABRICKS_ACCEPT_POLICY').asBoolStrict();
-
-    if (val !== undefined) {
-      console.warn(
-        'The CUBEJS_DB_DATABRICKS_ACCEPT_POLICY is not needed anymore. Please, remove it'
-      );
-    }
-  },
+  databrickAcceptPolicy: () => (
+    get('CUBEJS_DB_DATABRICKS_ACCEPT_POLICY').asBoolStrict()
+  ),
 
   /**
    * Databricks jdbc-connection url.

packages/cubejs-backend-shared/src/http-utils.ts

@@ -66,11 +66,9 @@ export async function streamWithProgress(
 type DownloadAndExtractFile = {
   showProgress: boolean;
   cwd: string;
-  noExtract?: boolean;
-  dstFileName?: string;
 };
 
-export async function downloadAndExtractFile(url: string, { cwd, noExtract, dstFileName }: DownloadAndExtractFile) {
+export async function downloadAndExtractFile(url: string, { cwd }: DownloadAndExtractFile) {
   const request = new Request(url, {
     headers: new Headers({
       'Content-Type': 'application/octet-stream',
@@ -101,15 +99,7 @@ export async function downloadAndExtractFile(url: string, { cwd, noExtract, dstF
     });
   });
 
-  if (noExtract) {
-    if (dstFileName) {
-      fs.copyFileSync(savedFilePath, path.resolve(path.join(cwd, dstFileName)));
-    } else {
-      fs.copyFileSync(savedFilePath, cwd);
-    }
-  } else {
-    await decompress(savedFilePath, cwd);
-  }
+  await decompress(savedFilePath, cwd);
 
   try {
     fs.unlinkSync(savedFilePath);

packages/cubejs-backend-shared/test/db_env_multi.test.ts

@@ -1192,6 +1192,34 @@ describe('Multiple datasources', () => {
     );
   });
 
+  test('getEnv("databrickAcceptPolicy")', () => {
+    process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY = 'true';
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toEqual(true);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toEqual(true);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toEqual(true);
+
+    process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY = 'false';
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toEqual(false);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toEqual(false);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toEqual(false);
+
+    process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY = 'wrong';
+    expect(() => getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toThrow(
+      'env-var: "CUBEJS_DB_DATABRICKS_ACCEPT_POLICY" should be either "true", "false", "TRUE", or "FALSE"'
+    );
+    expect(() => getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toThrow(
+      'env-var: "CUBEJS_DB_DATABRICKS_ACCEPT_POLICY" should be either "true", "false", "TRUE", or "FALSE"'
+    );
+    expect(() => getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toThrow(
+      'env-var: "CUBEJS_DB_DATABRICKS_ACCEPT_POLICY" should be either "true", "false", "TRUE", or "FALSE"'
+    );
+
+    delete process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY;
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toBeUndefined();
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toBeUndefined();
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toBeUndefined();
+  });
+
   test('getEnv("athenaAwsKey")', () => {
     process.env.CUBEJS_AWS_KEY = 'default1';
     process.env.CUBEJS_DS_POSTGRES_AWS_KEY = 'postgres1';

packages/cubejs-backend-shared/test/db_env_single.test.ts

@@ -760,6 +760,34 @@ describe('Single datasources', () => {
     expect(getEnv('databricksCatalog', { dataSource: 'wrong' })).toBeUndefined();
   });
 
+  test('getEnv("databrickAcceptPolicy")', () => {
+    process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY = 'true';
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toEqual(true);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toEqual(true);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toEqual(true);
+
+    process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY = 'false';
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toEqual(false);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toEqual(false);
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toEqual(false);
+
+    process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY = 'wrong';
+    expect(() => getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toThrow(
+      'env-var: "CUBEJS_DB_DATABRICKS_ACCEPT_POLICY" should be either "true", "false", "TRUE", or "FALSE"'
+    );
+    expect(() => getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toThrow(
+      'env-var: "CUBEJS_DB_DATABRICKS_ACCEPT_POLICY" should be either "true", "false", "TRUE", or "FALSE"'
+    );
+    expect(() => getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toThrow(
+      'env-var: "CUBEJS_DB_DATABRICKS_ACCEPT_POLICY" should be either "true", "false", "TRUE", or "FALSE"'
+    );
+
+    delete process.env.CUBEJS_DB_DATABRICKS_ACCEPT_POLICY;
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'default' })).toBeUndefined();
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'postgres' })).toBeUndefined();
+    expect(getEnv('databrickAcceptPolicy', { dataSource: 'wrong' })).toBeUndefined();
+  });
+
   test('getEnv("athenaAwsKey")', () => {
     process.env.CUBEJS_AWS_KEY = 'default1';
     expect(getEnv('athenaAwsKey', { dataSource: 'default' })).toEqual('default1');

packages/cubejs-databricks-jdbc-driver/src/DatabricksDriver.ts

@@ -4,21 +4,27 @@
  * @fileoverview The `DatabricksDriver` and related types declaration.
  */
 
-import { assertDataSource, getEnv, } from '@cubejs-backend/shared';
 import {
-  DatabaseStructure,
+  getEnv,
+  assertDataSource,
+} from '@cubejs-backend/shared';
+import {
   DriverCapabilities,
-  GenericDataBaseType,
   QueryColumnsResult,
   QueryOptions,
   QuerySchemasResult,
   QueryTablesResult,
-  TableColumn,
   UnloadOptions,
+  GenericDataBaseType,
+  TableColumn,
+  DatabaseStructure,
 } from '@cubejs-backend/base-driver';
-import { JDBCDriver, JDBCDriverConfiguration, } from '@cubejs-backend/jdbc-driver';
+import {
+  JDBCDriver,
+  JDBCDriverConfiguration,
+} from '@cubejs-backend/jdbc-driver';
 import { DatabricksQuery } from './DatabricksQuery';
-import { extractUidFromJdbcUrl, resolveJDBCDriver } from './helpers';
+import { resolveJDBCDriver, extractUidFromJdbcUrl } from './helpers';
 
 export type DatabricksDriverConfiguration = JDBCDriverConfiguration &
   {
@@ -126,7 +132,7 @@ export class DatabricksDriver extends JDBCDriver {
   /**
    * Show warning message flag.
    */
-  private readonly showSparkProtocolWarn: boolean;
+  private showSparkProtocolWarn: boolean;
 
   /**
    * Driver Configuration.
@@ -423,7 +429,8 @@ export class DatabricksDriver extends JDBCDriver {
         metadata[database] = {};
       }
 
-      metadata[database][tableName] = await this.tableColumnTypes(`${database}.${tableName}`);
+      const columns = await this.tableColumnTypes(`${database}.${tableName}`);
+      metadata[database][tableName] = columns;
     }));
 
     return metadata;

packages/cubejs-databricks-jdbc-driver/src/helpers.ts

@@ -1,7 +1,7 @@
 import fs from 'fs';
 import path from 'path';
 
-import { downloadJDBCDriver, OSS_DRIVER_VERSION } from './installer';
+import { downloadJDBCDriver } from './installer';
 
 async function fileExistsOr(
   fsPath: string,
@@ -15,16 +15,16 @@ async function fileExistsOr(
 
 export async function resolveJDBCDriver(): Promise<string> {
   return fileExistsOr(
-    path.join(process.cwd(), `databricks-jdbc-${OSS_DRIVER_VERSION}-oss.jar`),
+    path.join(process.cwd(), 'DatabricksJDBC42.jar'),
     async () => fileExistsOr(
-      path.join(__dirname, '..', 'download', `databricks-jdbc-${OSS_DRIVER_VERSION}-oss.jar`),
+      path.join(__dirname, '..', 'download', 'DatabricksJDBC42.jar'),
       async () => {
         const pathOrNull = await downloadJDBCDriver();
         if (pathOrNull) {
           return pathOrNull;
         }
         throw new Error(
-          `Please download and place databricks-jdbc-${OSS_DRIVER_VERSION}-oss.jar inside your ` +
+          'Please download and place DatabricksJDBC42.jar inside your ' +
           'project directory'
         );
       }

packages/cubejs-databricks-jdbc-driver/src/installer.ts

@@ -1,32 +1,38 @@
 import path from 'path';
 import { downloadAndExtractFile, getEnv } from '@cubejs-backend/shared';
 
-export const OSS_DRIVER_VERSION = '1.0.2';
-
-/**
- * In the beginning of 2025 Databricks released their open-source version of JDBC driver and encourage
- * all users to migrate to it as company plans to focus on improving and evolving it over legacy simba driver.
- * More info about OSS Driver could be found at https://docs.databricks.com/aws/en/integrations/jdbc/oss
- * As of March 2025 To use the Databricks JDBC Driver (OSS), the following requirements must be met:
- * Java Runtime Environment (JRE) 11.0 or above. CI testing is supported on JRE 11, 17, and 21.
- */
+function acceptedByEnv() {
+  const acceptStatus = getEnv('databrickAcceptPolicy');
+  if (acceptStatus) {
+    console.log('You accepted Terms & Conditions for JDBC driver from DataBricks by CUBEJS_DB_DATABRICKS_ACCEPT_POLICY');
+  }
+
+  if (acceptStatus === false) {
+    console.log('You declined Terms & Conditions for JDBC driver from DataBricks by CUBEJS_DB_DATABRICKS_ACCEPT_POLICY');
+    console.log('Installation will be skipped');
+  }
+
+  return acceptStatus;
+}
+
 export async function downloadJDBCDriver(): Promise<string | null> {
-  // TODO: Just to throw a console warning that this ENV is obsolete and could be safely removed
-  getEnv('databrickAcceptPolicy');
+  const driverAccepted = acceptedByEnv();
+
+  if (driverAccepted) {
+    console.log('Downloading DatabricksJDBC42-2.6.29.1051');
 
-  console.log(`Downloading databricks-jdbc-${OSS_DRIVER_VERSION}-oss.jar`);
+    await downloadAndExtractFile(
+      'https://databricks-bi-artifacts.s3.us-east-2.amazonaws.com/simbaspark-drivers/jdbc/2.6.29/DatabricksJDBC42-2.6.29.1051.zip',
+      {
+        showProgress: true,
+        cwd: path.resolve(path.join(__dirname, '..', 'download')),
+      }
+    );
 
-  await downloadAndExtractFile(
-    `https://repo1.maven.org/maven2/com/databricks/databricks-jdbc/${OSS_DRIVER_VERSION}-oss/databricks-jdbc-${OSS_DRIVER_VERSION}-oss.jar`,
-    {
-      showProgress: true,
-      cwd: path.resolve(path.join(__dirname, '..', 'download')),
-      noExtract: true,
-      dstFileName: `databricks-jdbc-${OSS_DRIVER_VERSION}-oss.jar`,
-    }
-  );
+    console.log('Release notes: https://databricks-bi-artifacts.s3.us-east-2.amazonaws.com/simbaspark-drivers/jdbc/2.6.29/docs/release-notes.txt');
 
-  console.log(`Release notes: https://mvnrepository.com/artifact/com.databricks/databricks-jdbc/${OSS_DRIVER_VERSION}-oss`);
+    return path.resolve(path.join(__dirname, '..', 'download', 'DatabricksJDBC42.jar'));
+  }
 
-  return path.resolve(path.join(__dirname, '..', 'download', `databricks-jdbc-${OSS_DRIVER_VERSION}-oss.jar`));
+  return null;
 }

packages/cubejs-databricks-jdbc-driver/src/post-install.ts

@@ -1,11 +1,16 @@
 import 'source-map-support/register';
 
 import { displayCLIError } from '@cubejs-backend/shared';
-import { resolveJDBCDriver } from './helpers';
+
+import fs from 'fs';
+import path from 'path';
+import { downloadJDBCDriver } from './installer';
 
 (async () => {
   try {
-    await resolveJDBCDriver();
+    if (!fs.existsSync(path.join(__dirname, '..', 'download', 'SparkJDBC42.jar'))) {
+      await downloadJDBCDriver();
+    }
   } catch (e: any) {
     await displayCLIError(e, 'Cube.js Databricks JDBC Installer');
   }