fix(bigquery-jdbc): handle EXPORT DATA, EXPORT MODEL, and LOAD DATA statements

[keshavdandeva]

b/510498449

This PR resolves an issue where executing EXPORT DATA, EXPORT MODEL, or LOAD DATA statements triggered a BigQueryJdbcException: Unexpected value: OTHER. These statement types were mapped to OTHER in BigQuerySqlTypeConverter, which was unhandled in BigQueryStatement.java.

1. EXPORT_DATA

• Mapped to a new SqlType.EXPORT and added a specific handler.
• BigQuery exposes the actual number of rows exported in QueryStatistics.ExportDataStats.getRowCount(). We now fetch this job statistic and return it as the update count, providing full functionality for this statement.

2. EXPORT_MODEL

• Mapped to SqlType.DDL
• Machine learning models are not row-based, and BigQuery does not report a row count for this operation in QueryStatistics. Treating it as DDL allows it to execute successfully and return 0 as the update count, which is appropriate.

3. LOAD_DATA

• Mapped to SqlType.DDL
• When run as a SQL statement, LOAD DATA executes as a Query Job. The resulting QueryStatistics do not expose load statistics directly (unlike ExportDataStats). While we could potentially find the count by inspecting child jobs, mapping it to DDL allows it to succeed immediately and return 0, resolving the blocking error with minimal overhead.

[gemini-code-assist]

Code Review

This pull request introduces support for the EXPORT SQL type in the BigQuery JDBC driver. It maps the EXPORT_DATA statement to a new EXPORT SqlType, and handles the query results for this type in BigQueryStatement by updating the affected row count based on the export statistics. The review feedback suggests improving robustness by adding null checks, validating the type before casting JobStatistics, and restoring the thread's interrupted status when catching InterruptedException.

[keshavdandeva]

/gemini review

[gemini-code-assist]

Code Review

This pull request introduces support for handling EXPORT_DATA statements in the BigQuery JDBC driver. It adds a new EXPORT SQL type, maps EXPORT_DATA to this type, and implements logic in BigQueryStatement to wait for the export job to complete and extract the affected row count from the job statistics. I have no feedback to provide as there are no review comments.