path	# lines of code
readthedocs.yml	9
setup.py	91
setup.cfg	2
petastorm/benchmark/dummy_reader.py	56
petastorm/benchmark/throughput.py	124
petastorm/benchmark/__init__.py	1
petastorm/benchmark/cli.py	63
petastorm/arrow_reader_worker.py	188
petastorm/ngram.py	105
petastorm/predicates.py	93
petastorm/tools/__init__.py	1
petastorm/tools/copy_dataset.py	77
petastorm/tools/spark_session_cli.py	28
petastorm/local_disk_cache.py	28
petastorm/weighted_sampling_reader.py	48
petastorm/etl/legacy.py	36
petastorm/etl/petastorm_generate_metadata.py	98
petastorm/etl/__init__.py	21
petastorm/etl/dataset_metadata.py	201
petastorm/etl/rowgroup_indexing.py	71
petastorm/etl/metadata_util.py	45
petastorm/etl/rowgroup_indexers.py	73
petastorm/codecs.py	154
petastorm/cache.py	9
petastorm/__init__.py	4
petastorm/py_dict_reader_worker.py	155
petastorm/reader_impl/pickle_serializer.py	6
petastorm/reader_impl/__init__.py	1
petastorm/reader_impl/pytorch_shuffling_buffer.py	123
petastorm/reader_impl/shuffling_buffer.py	75
petastorm/reader_impl/arrow_table_serializer.py	12
petastorm/selectors.py	46
petastorm/pyarrow_helpers/__init__.py	1
petastorm/pyarrow_helpers/batching_table_queue.py	30
petastorm/reader.py	367
petastorm/generator.py	15
petastorm/pytorch.py	248
petastorm/utils.py	61
petastorm/fs_utils.py	106
petastorm/transform.py	32
petastorm/namedtuple_gt_255_fields.py	65
petastorm/workers_pool/worker_base.py	11
petastorm/workers_pool/dummy_pool.py	38
petastorm/workers_pool/thread_pool.py	126
petastorm/workers_pool/ventilator.py	83
petastorm/workers_pool/__init__.py	3
petastorm/workers_pool/exec_in_new_process.py	18
petastorm/workers_pool/process_pool.py	214
petastorm/workers_pool/exec_in_new_process_entrypoint.py	18
petastorm/hdfs/__init__.py	1
petastorm/hdfs/namenode.py	159
petastorm/unischema.py	266
petastorm/errors.py	1
petastorm/spark_utils.py	18
petastorm/tf_utils.py	180
petastorm/spark/spark_dataset_converter.py	403
petastorm/spark/__init__.py	2