fix: resolve EmptyRowIterator test failures and lint error\n\n- Instantiate new _EmptyRowIterator for each test in test_methods_w_timeout to prevent ValueError\n- Remove total_rows property from _EmptyRowIterator to fix AssertionError\n- Remove redundant __iter__ method in _EmptyRowIterator to fix lint error

Author: chalmerlowe

google/cloud/bigquery/_pandas_helpers.py

@@ -981,7 +981,7 @@ def _download_table_bqstorage(
     worker_queue: queue.Queue[int] = queue.Queue(maxsize=max_queue_size)
 
     # Manually manage the pool to control shutdown behavior on timeout.
-    pool = concurrent.futures.ThreadPoolExecutor(max_workers=total_streams)
+    pool = concurrent.futures.ThreadPoolExecutor(max_workers=max(1, total_streams))
     wait_on_shutdown = True
     start_time = time.time()
 

google/cloud/bigquery/table.py

@@ -2941,9 +2941,6 @@ class _EmptyRowIterator(RowIterator):
     statements.
     """
 
-    pages = ()
-    total_rows = 0
-
     def __init__(
         self, client=None, api_request=None, path=None, schema=(), *args, **kwargs
     ):
@@ -2955,6 +2952,7 @@ def __init__(
             *args,
             **kwargs,
         )
+        self._total_rows = 0
 
     def to_arrow(
         self,

tests/unit/test__pandas_helpers.py

@@ -2181,16 +2181,19 @@ def test_determine_requested_streams_invalid_max_stream_count():
         determine_requested_streams(preserve_order=False, max_stream_count=-1)
 
 
+@pytest.mark.skipif(
+    bigquery_storage is None, reason="Requires google-cloud-bigquery-storage"
+)
 def test__download_table_bqstorage_w_timeout_error(module_under_test):
     from google.cloud.bigquery import dataset
     from google.cloud.bigquery import table
+    from unittest import mock
 
-    bqstorage_client = mock.create_autospec(
+    mock_bqstorage_client = mock.create_autospec(
         bigquery_storage.BigQueryReadClient, instance=True
     )
-    # Give it one stream
-    fake_session = mock.Mock(streams=["stream/s0"])
-    bqstorage_client.create_read_session.return_value = fake_session
+    fake_session = mock.Mock(streams=[mock.Mock()])
+    mock_bqstorage_client.create_read_session.return_value = fake_session
 
     table_ref = table.TableReference(
         dataset.DatasetReference("project-x", "dataset-y"),
@@ -2209,21 +2212,25 @@ def slow_download_stream(
     ):
         # Use a very small timeout
         result_gen = module_under_test._download_table_bqstorage(
-            "some-project", table_ref, bqstorage_client, timeout=0.01
+            "some-project", table_ref, mock_bqstorage_client, timeout=0.01
         )
         with pytest.raises(concurrent.futures.TimeoutError, match="timed out"):
             list(result_gen)
 
 
+@pytest.mark.skipif(
+    bigquery_storage is None, reason="Requires google-cloud-bigquery-storage"
+)
 def test__download_table_bqstorage_w_timeout_success(module_under_test):
     from google.cloud.bigquery import dataset
     from google.cloud.bigquery import table
+    from unittest import mock
 
-    bqstorage_client = mock.create_autospec(
+    mock_bqstorage_client = mock.create_autospec(
         bigquery_storage.BigQueryReadClient, instance=True
     )
     fake_session = mock.Mock(streams=["stream/s0"])
-    bqstorage_client.create_read_session.return_value = fake_session
+    mock_bqstorage_client.create_read_session.return_value = fake_session
 
     table_ref = table.TableReference(
         dataset.DatasetReference("project-x", "dataset-y"),
@@ -2240,7 +2247,7 @@ def fast_download_stream(
     ):
         # Use a generous timeout
         result_gen = module_under_test._download_table_bqstorage(
-            "some-project", table_ref, bqstorage_client, timeout=10.0
+            "some-project", table_ref, mock_bqstorage_client, timeout=10.0
         )
         results = list(result_gen)
 

tests/unit/test_table.py

@@ -2496,16 +2496,17 @@ def test_to_geodataframe(self):
             assert not hasattr(df, "crs")
 
     def test_methods_w_timeout(self):
+        pytest.importorskip("pyarrow")
+        pytest.importorskip("geopandas")
         # Ensure that timeout parameter is accepted by all methods and ignored.
-        row_iterator = self._make_one()
         timeout = 42.0
 
-        # Just calling them to ensure no TypeError is raised
-        row_iterator.to_arrow(timeout=timeout)
-        row_iterator.to_arrow_iterable(timeout=timeout)
-        row_iterator.to_dataframe(timeout=timeout)
-        row_iterator.to_dataframe_iterable(timeout=timeout)
-        row_iterator.to_geodataframe(timeout=timeout)
+        # Call each type to ensure no TypeError is raised
+        self._make_one().to_arrow(timeout=timeout)
+        self._make_one().to_arrow_iterable(timeout=timeout)
+        self._make_one().to_dataframe(timeout=timeout)
+        self._make_one().to_dataframe_iterable(timeout=timeout)
+        self._make_one().to_geodataframe(timeout=timeout)
 
 
 class TestRowIterator(unittest.TestCase):