arrow/docs/source/python/memory.rst

.. 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.

.. currentmodule:: pyarrow
.. highlight:: python

.. _io:

========================
Memory and IO Interfaces
========================

This section will introduce you to the major concepts in PyArrow's memory
management and IO systems:

* Buffers
* Memory pools
* File-like and stream-like objects

Referencing and Allocating Memory
=================================

pyarrow.Buffer
--------------

The :class:`Buffer` object wraps the C++ :cpp:class:`arrow::Buffer` type
which is the primary tool for memory management in Apache Arrow in C++. It permits
higher-level array classes to safely interact with memory which they may or may
not own. ``arrow::Buffer`` can be zero-copy sliced to permit Buffers to cheaply
reference other Buffers, while preserving memory lifetime and clean
parent-child relationships.

There are many implementations of ``arrow::Buffer``, but they all provide a
standard interface: a data pointer and length. This is similar to Python's
built-in ``buffer protocol`` and ``memoryview`` objects.

A :class:`Buffer` can be created from any Python object implementing
the buffer protocol by calling the :func:`py_buffer` function. Let's consider
a bytes object:

.. code-block:: python

   >>> import pyarrow as pa
   >>> data = b'abcdefghijklmnopqrstuvwxyz'
   >>> buf = pa.py_buffer(data)
   >>> buf
   <pyarrow.Buffer address=... size=26 is_cpu=True is_mutable=False>
   >>> buf.size
   26

Creating a Buffer in this way does not allocate any memory; it is a zero-copy
view on the memory exported from the ``data`` bytes object.

External memory, under the form of a raw pointer and size, can also be
referenced using the :func:`foreign_buffer` function.

Buffers can be used in circumstances where a Python buffer or memoryview is
required, and such conversions are zero-copy:

.. code-block:: python

   >>> memoryview(buf)
   <memory at ...>

The Buffer's :meth:`~Buffer.to_pybytes` method converts the Buffer's data to a
Python bytestring (thus making a copy of the data):

.. code-block:: python

   >>> buf.to_pybytes()
   b'abcdefghijklmnopqrstuvwxyz'

Memory Pools
------------

All memory allocations and deallocations (like ``malloc`` and ``free`` in C)
are tracked in an instance of :class:`MemoryPool`. This means that we can
then precisely track amount of memory that has been allocated:

.. code-block:: python

   >>> pa.total_allocated_bytes()
   0

Let's allocate a resizable :class:`Buffer` from the default pool:

.. code-block:: python

   >>> buf = pa.allocate_buffer(1024, resizable=True)
   >>> pa.total_allocated_bytes()
   1024
   >>> buf.resize(2048)
   >>> pa.total_allocated_bytes()
   2048

The default allocator requests memory in a minimum increment of 64 bytes. If
the buffer is garbage-collected, all of the memory is freed:

.. code-block:: python

   >>> buf = None
   >>> pa.total_allocated_bytes()
   0

Besides the default built-in memory pool, there may be additional memory pools
to choose from (such as `jemalloc <http://jemalloc.net/>`_)
depending on how Arrow was built.  One can get the backend name for a memory
pool::

   >>> pa.default_memory_pool().backend_name
   'mimalloc'

.. seealso::
   :ref:`API documentation for memory pools <api.memory_pool>`.

.. seealso::
   On-GPU buffers using Arrow's optional :doc:`CUDA integration <integration/cuda>`.


Input and Output
================

.. _io.native_file:

The Arrow C++ libraries have several abstract interfaces for different kinds of
IO objects:

* Read-only streams
* Read-only files supporting random access
* Write-only streams
* Write-only files supporting random access
* File supporting reads, writes, and random access

In the interest of making these objects behave more like Python's built-in
``file`` objects, we have defined a :class:`~pyarrow.NativeFile` base class
which implements the same API as regular Python file objects.

:class:`~pyarrow.NativeFile` has some important features which make it
preferable to using Python files with PyArrow where possible:

* Other Arrow classes can access the internal C++ IO objects natively, and do
  not need to acquire the Python GIL
* Native C++ IO may be able to do zero-copy IO, such as with memory maps

There are several kinds of :class:`~pyarrow.NativeFile` options available:

* :class:`~pyarrow.OSFile`, a native file that uses your operating system's
  file descriptors
* :class:`~pyarrow.MemoryMappedFile`, for reading (zero-copy) and writing with
  memory maps
* :class:`~pyarrow.BufferReader`, for reading :class:`~pyarrow.Buffer` objects
  as a file
* :class:`~pyarrow.BufferOutputStream`, for writing data in-memory, producing a
  Buffer at the end
* :class:`~pyarrow.FixedSizeBufferWriter`, for writing data into an already
  allocated Buffer
* :class:`~pyarrow.PythonFile`, for interfacing with Python file objects in C++
* :class:`~pyarrow.CompressedInputStream` and
  :class:`~pyarrow.CompressedOutputStream`, for on-the-fly compression or
  decompression to/from another stream

There are also high-level APIs to make instantiating common kinds of streams
easier.

High-Level API
--------------

Input Streams
~~~~~~~~~~~~~

The :func:`~pyarrow.input_stream` function allows creating a readable
:class:`~pyarrow.NativeFile` from various kinds of sources.

* If passed a :class:`~pyarrow.Buffer` or a ``memoryview`` object, a
  :class:`~pyarrow.BufferReader` will be returned:

   .. code-block:: python

      >>> buf = memoryview(b"some data")
      >>> stream = pa.input_stream(buf)
      >>> stream.read(4)
      b'some'

* If passed a string or file path, it will open the given file on disk
  for reading, creating a :class:`~pyarrow.OSFile`.  Optionally, the file
  can be compressed: if its filename ends with a recognized extension
  such as ``.gz``, its contents will automatically be decompressed on
  reading.

  .. code-block:: python

     >>> import gzip
     >>> with gzip.open('example.gz', 'wb') as f:
     ...     f.write(b'some data\n' * 3)
     30
     >>> stream = pa.input_stream('example.gz')
     >>> stream.read()
     b'some data\nsome data\nsome data\n'

* If passed a Python file object, it will wrapped in a :class:`PythonFile`
  such that the Arrow C++ libraries can read data from it (at the expense
  of a slight overhead).

Output Streams
~~~~~~~~~~~~~~

:func:`~pyarrow.output_stream` is the equivalent function for output streams
and allows creating a writable :class:`~pyarrow.NativeFile`.  It has the same
features as explained above for :func:`~pyarrow.input_stream`, such as being
able to write to buffers or do on-the-fly compression.

.. code-block:: python

   >>> with pa.output_stream('example1.dat') as stream:
   ...     stream.write(b'some data')
   9
   >>> with open('example1.dat', 'rb') as f:
   ...     f.read()
   b'some data'


On-Disk and Memory Mapped Files
-------------------------------

PyArrow includes two ways to interact with data on disk: standard operating
system-level file APIs, and memory-mapped files. In regular Python we can
write:

.. code-block:: python

   >>> with open('example2.dat', 'wb') as f:
   ...     f.write(b'some example data')
   17

Using pyarrow's :class:`~pyarrow.OSFile` class, you can write:

.. code-block:: python

   >>> with pa.OSFile('example3.dat', 'wb') as f:
   ...     f.write(b'some example data')
   17

For reading files, you can use :class:`~pyarrow.OSFile` or
:class:`~pyarrow.MemoryMappedFile`. The difference between these is that
:class:`~pyarrow.OSFile` allocates new memory on each read, like Python file
objects. In reads from memory maps, the library constructs a buffer referencing
the mapped memory without any memory allocation or copying:

.. code-block:: python

   >>> file_obj = pa.OSFile('example2.dat')
   >>> mmap = pa.memory_map('example3.dat')
   >>> file_obj.read(4)
   b'some'
   >>> mmap.read(4)
   b'some'

The ``read`` method implements the standard Python file ``read`` API. To read
into Arrow Buffer objects, use ``read_buffer``:

.. code-block:: python

   >>> mmap.seek(0)
   0
   >>> buf = mmap.read_buffer(4)
   >>> buf
   <pyarrow.Buffer address=... size=4 is_cpu=True is_mutable=False>
   >>> buf.to_pybytes()
   b'some'

Many tools in PyArrow, particular the Apache Parquet interface and the file and
stream messaging tools, are more efficient when used with these ``NativeFile``
types than with normal Python file objects.

In-Memory Reading and Writing
-----------------------------

To assist with serialization and deserialization of in-memory data, we have
file interfaces that can read and write to Arrow Buffers.

.. code-block:: python

   >>> writer = pa.BufferOutputStream()
   >>> writer.write(b'hello, friends')
   14
   >>> buf = writer.getvalue()
   >>> buf
   <pyarrow.Buffer address=... size=14 is_cpu=True is_mutable=True>
   >>> buf.size
   14
   >>> reader = pa.BufferReader(buf)
   >>> reader.seek(7)
   7
   >>> reader.read(7)
   b'friends'

These have similar semantics to Python's built-in ``io.BytesIO``.