feat(spanner): add DataBoost and auto_partition_mode support to DBAPI driver

[sakthivelmanii]

Description

Add support for Cloud Spanner DataBoost and auto_partition_mode in the Python DBAPI (PEP 249) driver (google.cloud.spanner_dbapi).

Commit Structure

1. Commit 1 (cff1650f719): feat(spanner): add DataBoost support to DBAPI driver and client-side statements

   • Adds data_boost_enabled: bool = False to connect() and Connection.
   • Adds SET DATA_BOOST_ENABLED and SHOW VARIABLE DATA_BOOST_ENABLED (returning DATA_BOOST_ENABLED column with BOOL type).
   • Forwards data_boost_enabled in Connection.partition_query() and Connection.run_partitioned_query().
   • Confines DataBoost execution to partitioned queries (per Spanner RPC requirement of partition_token).

2. Commit 2 (9c53e3f9a81): feat(spanner): add auto_partition_mode support to DBAPI driver and client-side statements

   • Adds auto_partition_mode: bool = False to connect() and Connection.
   • Adds SET AUTO_PARTITION_MODE and SHOW VARIABLE AUTO_PARTITION_MODE (returning AUTO_PARTITION_MODE column with BOOL type).
   • Automatically partitions and parallelizes standard queries (cursor.execute() / pandas.read_sql()) when auto_partition_mode=True.
   • Enables seamless DataBoost execution for standard queries when both auto_partition_mode=True and data_boost_enabled=True.
   • Introduces reusable _parse_bool helper for client-side boolean parameter extraction.

Testing

• All 234 unit and gRPC mock server tests passing.
• Verified flake8 and ruff formatting/import sorting compliance.

[gemini-code-assist]

Code Review

This pull request adds support for configuring and querying the DATA_BOOST_ENABLED setting on Spanner DB-API connections, including client-side statement parsing and execution for SET DATA_BOOST_ENABLED and SHOW VARIABLE DATA_BOOST_ENABLED. The reviewer feedback suggests improving the robustness of parsing by stripping quotes from the parameter value and initializing column_values directly as a list for consistency and safety.

[olavloite]

Regarding this part in the PR description:

enabling analytical and bulk read workloads (e.g. pandas.read_sql) to execute on serverless compute

Note that this feature in its current implementation can only be used with RUN PARTITIONED QUERY .... Other Spanner drivers, like JDBC, additionally also support an autoPartitionMode=true|false variable to turn all queries into partitioned queries, which makes it easier to use DataBoost for all queries.

[olavloite]

nit: add documentation here for our new input arguments data_boost_enabled and auto_partition_mode

[sakthivelmanii]

Done! Added documentation for both data_boost_enabled and auto_partition_mode arguments to the Connection class docstring.

[olavloite]

nit: This function does not work if the application added a semicolon to the end of the statement. That is:

SET AUTO_PARTITION_MODE = TRUE;

Will produce an error. We could easily fix this by adjusting the regex parsing a bit:

def _parse_bool(raw_val: str, var_name: str) -> bool:
    cleaned = raw_val.strip().rstrip(";").strip().strip("'\"").lower()
    if cleaned not in _BOOL_MAP:
        raise ProgrammingError(
            f"Invalid value for {var_name}: '{raw_val}'. Expected TRUE or FALSE."
        )
    return _BOOL_MAP[cleaned]

[sakthivelmanii]

Done! Updated _parse_bool to strip trailing semicolons, and updated the SHOW VARIABLE parser patterns to accept optional trailing semicolons as well.