Fix BigQuery empty SELECT response typing

[ianu82]

Summary

• update BigQuery native_query to treat any dataframe result as RESPONSE_TYPE.TABLE, including empty rowsets
• add a regression unit test covering empty SELECT behavior

Why

Empty SELECT queries produce dataframes with schema but no rows. Returning OK for these responses drops the table contract and breaks metadata/column-aware downstream logic.

Validation

• python3 -m py_compile mindsdb/integrations/handlers/bigquery_handler/bigquery_handler.py tests/unit/handlers/test_bigquery.py
• pytest -q tests/unit/handlers/test_bigquery.py -k empty_select_returns_table (fails in this environment due to missing optional dependency google.api_core)

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[entelligence-ai-pr-reviews]

Correctness: ⚠️ This change makes the condition always true because to_dataframe() always returns a DataFrame, so you never return RESPONSE_TYPE.OK (e.g., for DDL/empty results). That’s a behavior regression from the previous not result.empty logic. Restore the empty check (optionally keep the type guard).

🤖 AI Agent Prompt for Cursor/Windsurf

📋 Copy this prompt to your AI coding assistant (Cursor, Windsurf, etc.) to get help fixing this issue

In `mindsdb/integrations/handlers/bigquery_handler/bigquery_handler.py` around the `native_query` result handling, reintroduce the empty-result check so that non-result queries still return `RESPONSE_TYPE.OK`. Update the condition to `isinstance(result, pd.DataFrame) and not result.empty`.

[ianu82]

Addressed. I refined native_query result typing so non-result operations still return RESPONSE_TYPE.OK, while empty SELECT queries with schema still return RESPONSE_TYPE.TABLE.

Implemented condition:

• isinstance(result, pd.DataFrame) and (not result.empty or len(result.columns) > 0)

Also added a regression test for the no-schema empty dataframe case (pd.DataFrame() -> OK) alongside the existing empty-select test (columns present -> TABLE).