chore: refactor get_configuration to use a dataclass (#1940)

* Refactor get_configuration to use a dataclass

I've introduced a new `BenchmarkConfig` frozen dataclass to provide a more structured way of handling benchmark configurations.

I updated the `get_configuration` function in `tests/benchmark/utils.py` to return an instance of this dataclass.

I have partially updated the call sites for `get_configuration` to use the new dataclass object. The following directories have been updated:
- tests/benchmark/db_benchmark/groupby/
- tests/benchmark/db_benchmark/join/
- tests/benchmark/db_benchmark/sort/
- tests/benchmark/read_gbq_colab/ (except for sort_output.py)

The remaining call sites in `tests/benchmark/read_gbq_colab/sort_output.py` and `tests/benchmark/tpch/` still need to be updated.

* I will refactor `get_configuration` to use a dataclass.

This change introduces a new `BenchmarkConfig` frozen dataclass to provide a more structured way of handling benchmark configurations.

The `get_configuration` function in `tests/benchmark/utils.py` will be updated to return an instance of this dataclass.

All call sites for `get_configuration` will be updated to use the new dataclass object.

* I will refactor `get_configuration` to use a dataclass.

This change introduces a new `BenchmarkConfig` frozen dataclass to provide a more structured way of handling benchmark configurations.

The `get_configuration` function in `tests/benchmark/utils.py` will be updated to return an instance of this dataclass.

All call sites for `get_configuration` will be updated to use the new dataclass object.

---------

Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>

Author: tswast

tests/benchmark/db_benchmark/groupby/q1.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q1,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q10.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q10,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q2.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q2,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q3.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q3,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q4.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q4,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q5.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q5,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q6.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q6,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q7.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q7,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/groupby/q8.py

@@ -18,21 +18,15 @@
 import bigframes_vendored.db_benchmark.groupby_queries as vendored_dbbenchmark_groupby_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_groupby_queries.q8,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )

tests/benchmark/db_benchmark/join/q1.py

@@ -18,22 +18,16 @@
 import bigframes_vendored.db_benchmark.join_queries as vendored_dbbenchmark_join_queries
 
 if __name__ == "__main__":
-    (
-        project_id,
-        dataset_id,
-        table_id,
-        session,
-        suffix,
-    ) = utils.get_configuration(include_table_id=True)
+    config = utils.get_configuration(include_table_id=True)
 
     current_path = pathlib.Path(__file__).absolute()
 
     utils.get_execution_time(
         vendored_dbbenchmark_join_queries.q1,
         current_path,
-        suffix,
-        project_id,
-        dataset_id,
-        table_id,
-        session,
+        config.benchmark_suffix,
+        config.project_id,
+        config.dataset_id,
+        config.table_id,
+        config.session,
     )