Support results_queue_size parameter in make_batch_reader api (#783)

* Support results_queue_size parameter in make_batch_reader api

* Test results_queue_size is propagated properly to the workers pool

* Update release notes

* Reformat unit test

Author: s-udhaya

docs/release-notes.rst

@@ -8,6 +8,7 @@ Release notes
 Release 0.12.1 (unreleased)
 ===========================
 - `PR 777 <https://github.com/uber/petastorm/pull/777>`_: Remove ``LocalDiskArrowTableCache`` class as it was using deprecated pyarrow serialization API. Speed up ``LocalDiskCache`` by using the highest pickle protocol in cache serialization.
+- `PR 783 <https://github.com/uber/petastorm/pull/783>`_: Support results_queue_size parameter in make_batch_reader API
 
 Release 0.12.0
 ===========================

petastorm/reader.py

@@ -205,6 +205,7 @@ def make_reader(dataset_url,
 def make_batch_reader(dataset_url_or_urls,
                       schema_fields=None,
                       reader_pool_type='thread', workers_count=10,
+                      results_queue_size=50,
                       seed=None, shuffle_rows=False,
                       shuffle_row_groups=True, shuffle_row_drop_partitions=1,
                       predicate=None,
@@ -243,6 +244,8 @@ def make_batch_reader(dataset_url_or_urls,
         denoting a thread pool, process pool, or running everything in the master thread. Defaults to 'thread'
     :param workers_count: An int for the number of workers to use in the reader pool. This only is used for the
         thread or process pool. Defaults to 10
+    :param results_queue_size: Size of the results queue to store prefetched row-groups. Currently only applicable to
+        thread reader pool type.
     :param seed: Random seed specified for shuffle and sharding with reproducible outputs. Defaults to None
     :param shuffle_rows: Whether to shuffle inside a single row group. Defaults to False.
     :param shuffle_row_groups: Whether to shuffle row groups (the order in which full row groups are read)
@@ -312,7 +315,7 @@ def make_batch_reader(dataset_url_or_urls,
         raise ValueError('Unknown cache_type: {}'.format(cache_type))
 
     if reader_pool_type == 'thread':
-        reader_pool = ThreadPool(workers_count)
+        reader_pool = ThreadPool(workers_count, results_queue_size)
     elif reader_pool_type == 'process':
         serializer = ArrowTableSerializer()
         reader_pool = ProcessPool(workers_count, serializer, zmq_copy_buffers=zmq_copy_buffers)

petastorm/tests/test_parquet_reader.py

@@ -258,3 +258,11 @@ def test_random_seed(scalar_dataset):
             results.append(actual_row_ids)
         # Shuffled results are expected to be same
         np.testing.assert_equal(results[0], results[1])
+
+
+def test_results_queue_size_propagation_in_make_batch_reader(scalar_dataset):
+    expected_results_queue_size = 42
+    with make_batch_reader(scalar_dataset.url, reader_pool_type='thread',
+                           results_queue_size=expected_results_queue_size) as batch_reader:
+        actual_results_queue_size = batch_reader._workers_pool._results_queue_size
+    assert actual_results_queue_size == expected_results_queue_size