Update Snowflake JDBC driver to 4.3.4

[vishwasvaidya-cloudsufi]

Summary

Bumps snowflake-jdbc.version from 4.0.2 to 4.3.4.

Driver 4.0.2 fails to create the S3 client when the stage endpoint has no URI scheme. Snowflake Government accounts return their endpoint this way, so every read (downloadStream) fails on those accounts with:

java.lang.NullPointerException: The URI scheme of endpointOverride must not be null.

Snowflake fixed this upstream in snowflakedb/snowflake-jdbc#2572 (SNOW-3344317), first released in 4.2.0. In 4.3.4, an endpoint without a scheme gets https:// added before the client is built.

The plugin code needs no changes: the driver APIs it uses (SnowflakeConnection, UploadStreamConfig, DownloadStreamConfig, SnowflakeBasicDataSource) are the same in 4.3.4. The driver is still built for Java 8, and the CDAP parent range is unchanged, so CDAP 6.10 and 6.11 stay supported.

Changes

• pom.xml: snowflake-jdbc.version 4.0.2 → 4.3.4
• New SnowflakeS3StageEndpointTest: builds the S3 client the same way downloadStream does (StageInfo → StorageClientFactory.createClient) and checks the endpoint it ends up with:
  • endpoint without a scheme → https://<endpoint>
  • https:// endpoint → unchanged
  • no endpoint → no override

Testing

Driver	SnowflakeS3StageEndpointTest
4.0.2	scheme-less case fails with the NPE above; other 2 pass
4.3.4	3/3 pass

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[gemini-code-assist]

Code Review

This pull request introduces the ability to retrieve schemas using a named table in the Snowflake Batch Source, alongside updating the Snowflake JDBC driver to version 4.3.4 and adapting the codebase to its API changes. The review feedback highlights several critical issues: a constructor initialization bug in SnowflakeBatchSourceConfig that defaults the query type and ignores the user's selection, a case-mismatch between the UI and backend enum values, a potential IllegalStateException when calling getFields() on a non-RECORD schema, and a swallowed SQLException that hinders debugging.

[gemini-code-assist]

In the constructor, this.importQueryType is assigned the result of getImportQueryType(). However, at this point, the field this.importQueryType has not been initialized yet (it is null), so getImportQueryType() will always return the default value (IMPORT_QUERY), completely ignoring the constructor parameter importQueryType. This will break the 'Named Table' feature as the configuration will always default to IMPORT_QUERY. Please assign the constructor parameter directly to the field.

Suggested change

	this.importQueryType = getImportQueryType();
	this.importQueryType = importQueryType;

[gemini-code-assist]

The method getImportQueryType() returns ImportQueryType.IMPORT_QUERY.name() (which is "IMPORT_QUERY") when importQueryType is null. However, the widget configuration in Snowflake-batchsource.json defines the option ID as "importQuery" (lowercase). Returning "IMPORT_QUERY" will cause a mismatch in the UI. It should return ImportQueryType.IMPORT_QUERY.getValue() instead.

Suggested change

	return importQueryType == null ? ImportQueryType.IMPORT_QUERY.name() : importQueryType;
	return importQueryType == null ? ImportQueryType.IMPORT_QUERY.getValue() : importQueryType;

[gemini-code-assist]

Using direct string comparison between ImportQueryType.IMPORT_QUERY.getValue() (which is "importQuery") and the raw importQueryType field is error-prone due to potential case mismatches (e.g., "IMPORT_QUERY" vs "importQuery"). Instead, you should use the ImportQueryType.fromString(...) helper method to parse the query type and perform enum comparisons.

Suggested change

	boolean isImportQuerySelected = ImportQueryType.IMPORT_QUERY.getValue().equals(importQueryType);
	boolean isTableNameSelected = ImportQueryType.TABLE_NAME.getValue().equals(importQueryType);
	ImportQueryType queryType = ImportQueryType.fromString(importQueryType);
	boolean isImportQuerySelected = queryType == ImportQueryType.IMPORT_QUERY;
	boolean isTableNameSelected = queryType == ImportQueryType.TABLE_NAME;

[gemini-code-assist]

The caught SQLException is swallowed, and a generic IOException is thrown without a message or the root cause. This makes troubleshooting database connection or query issues extremely difficult. Please pass the original exception as the cause to the IOException.

Suggested change

	throw new IOException();
	throw new IOException("Failed to retrieve field descriptors for table: " + tableName, e);

[gemini-code-assist]

If the configured schema is not of type RECORD, calling schema.getFields() will throw an IllegalStateException. To prevent this, you should verify that the schema type is RECORD before accessing its fields.

Suggested change

	if (schema != null && schema.getFields() != null && !schema.getFields().isEmpty()) {
	if (schema != null && schema.getType() == Schema.Type.RECORD && schema.getFields() != null && !schema.getFields().isEmpty()) {

[vishwasvaidya-cloudsufi]

needs some changes