wjones127 commented on code in PR #35568: URL: https://github.com/apache/arrow/pull/35568#discussion_r1240040981
########## docs/source/python/integration/dataset.rst: ########## @@ -0,0 +1,134 @@ +.. Licensed to the Apache Software Foundation (ASF) under one +.. or more contributor license agreements. See the NOTICE file +.. distributed with this work for additional information +.. regarding copyright ownership. The ASF licenses this file +.. to you under the Apache License, Version 2.0 (the +.. "License"); you may not use this file except in compliance +.. with the License. You may obtain a copy of the License at + +.. http://www.apache.org/licenses/LICENSE-2.0 + +.. Unless required by applicable law or agreed to in writing, +.. software distributed under the License is distributed on an +.. "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +.. KIND, either express or implied. See the License for the +.. specific language governing permissions and limitations +.. under the License. + +Extending PyArrow Datasets +========================== + +PyArrow provides a core protocol for datasets, so third-party libraries can both +produce and consume classes that conform to useful subset of the PyArrow dataset +API. This subset provides enough functionality to provide predicate and filter +pushdown. The subset of the API is contained in ``pyarrow.dataset.protocol``. + +.. image:: pyarrow_dataset_protocol.svg + :alt: A diagram showing the workflow for using the PyArrow Dataset protocol. + There are two flows shown, one for stream and one for tasks. The stream + case shows a linear flow from a producer class, to a dataset, to a + scanner, and finally to a RecordBatchReader. The tasks case shows a + similar diagram, except the dataset is split into fragments, which are + then distributed to tasks, which each create their own scanner and + RecordBatchReader. + +Producers are responsible for outputting a class that conforms to the protocol. + +Consumers are responsible for calling methods on the protocol to get the data +out of the dataset. The protocol supports getting data as a single stream or +as a series of tasks which may be distributed. + +From the perspective of a user, the code looks like: + +.. code-block:: python + + dataset = producer_library.get_dataset(...) + df = consumer_library.read_dataset(dataset) + df.filter("x > 0").select("y") Review Comment: Perhaps I should explain that it's imaginary? Or I could show a concrete usage with DuckDB and then Polars? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
