Skip to content

C
cpython
Kaydet (Commit) 9d0ced36 authored tarafından Benjamin Peterson's avatar Benjamin Peterson
Dosyalara gözat
Seçenekler

add documentation for the new buffer interface based on PEP 3118; I hope it's at least 60% right...

üst 74423693
Hide whitespace changes
Inline Side-by-side
Showing with 280 additions and 145 deletions
Doc/c-api/buffer.rst
Dosyayı görüntüle @ 9d0ced36
......@@ -6,19 +6,20 @@ Buffer Objects
--------------
.. sectionauthor:: Greg Stein <gstein@lyra.org>
.. sectionauthor:: Benjamin Peterson
.. index::
object: buffer
single: buffer interface
Python objects implemented in C can export a group of functions called the
"buffer interface." These functions can be used by an object to expose its data
in a raw, byte-oriented format. Clients of the object can use the buffer
interface to access the object data directly, without needing to copy it first.
Python objects implemented in C can export a "buffer interface." These
functions can be used by an object to expose its data in a raw, byte-oriented
format. Clients of the object can use the buffer interface to access the object
data directly, without needing to copy it first.
Two examples of objects that support the buffer interface are strings and
arrays. The string object exposes the character contents in the buffer
Two examples of objects that support the buffer interface are bytes and
arrays. The bytes object exposes the character contents in the buffer
interface's byte-oriented form. An array can also expose its contents, but it
should be noted that array elements may be multi-byte values.
......@@ -33,87 +34,275 @@ returning data from the target object.
More information on the buffer interface is provided in the section
:ref:`buffer-structs`, under the description for :ctype:`PyBufferProcs`.
A "buffer object" is defined in the :file:`bufferobject.h` header (included by
:file:`Python.h`). These objects look very similar to string objects at the
Python programming level: they support slicing, indexing, concatenation, and
some other standard string operations. However, their data can come from one of
two sources: from a block of memory, or from another object which exports the
buffer interface.
Buffer objects are useful as a way to expose the data from another object's
buffer interface to the Python programmer. They can also be used as a zero-copy
slicing mechanism. Using their ability to reference a block of memory, it is
possible to expose any data to the Python programmer quite easily. The memory
buffer interface to the Python programmer. They can also be used as a zero-copy
slicing mechanism. Using their ability to reference a block of memory, it is
possible to expose any data to the Python programmer quite easily. The memory
could be a large, constant array in a C extension, it could be a raw block of
memory for manipulation before passing to an operating system library, or it
could be used to pass around structured data in its native, in-memory format.
.. ctype:: PyBufferObject
This subtype of :ctype:`PyObject` represents a buffer object.
.. cvar:: PyTypeObject PyBuffer_Type
.. index:: single: BufferType (in module types)
The instance of :ctype:`PyTypeObject` which represents the Python buffer type;
it is the same object as ``buffer`` and ``types.BufferType`` in the Python
layer. .
.. cvar:: int Py_END_OF_BUFFER
This constant may be passed as the *size* parameter to
:cfunc:`PyBuffer_FromObject` or :cfunc:`PyBuffer_FromReadWriteObject`. It
indicates that the new :ctype:`PyBufferObject` should refer to *base* object
from the specified *offset* to the end of its exported buffer. Using this
enables the caller to avoid querying the *base* object for its length.
.. ctype:: Py_buffer
.. cfunction:: int PyBuffer_Check(PyObject *p)
.. cmember:: void *buf
Return true if the argument has type :cdata:`PyBuffer_Type`.
A pointer to the start of the memory for the object.
.. cmember:: Py_ssize_t len
.. cfunction:: PyObject* PyBuffer_FromObject(PyObject *base, Py_ssize_t offset, Py_ssize_t size)
The total length of the memory in bytes.
Return a new read-only buffer object. This raises :exc:`TypeError` if *base*
doesn't support the read-only buffer protocol or doesn't provide exactly one
buffer segment, or it raises :exc:`ValueError` if *offset* is less than zero.
The buffer will hold a reference to the *base* object, and the buffer's contents
will refer to the *base* object's buffer interface, starting as position
*offset* and extending for *size* bytes. If *size* is :const:`Py_END_OF_BUFFER`,
then the new buffer's contents extend to the length of the *base* object's
exported buffer data.
.. cmember:: int readonly
An indicator of whether the buffer is read only.
.. cfunction:: PyObject* PyBuffer_FromReadWriteObject(PyObject *base, Py_ssize_t offset, Py_ssize_t size)
.. cmember:: const char *format
Return a new writable buffer object. Parameters and exceptions are similar to
those for :cfunc:`PyBuffer_FromObject`. If the *base* object does not export
the writable buffer protocol, then :exc:`TypeError` is raised.
A *NULL* terminated string in :mod:`struct` module style syntax giving the
contents of the elements available through the buffer. If this is *NULL*,
``"B"`` (unsigned bytes) is assumed.
.. cmember:: int ndim
.. cfunction:: PyObject* PyBuffer_FromMemory(void *ptr, Py_ssize_t size)
The number of dimensions the memory represents as a multi-dimensional
array. If it is 0, :cdata:`strides` and :cdata:`suboffsets` must be
*NULL*.
Return a new read-only buffer object that reads from a specified location in
memory, with a specified size. The caller is responsible for ensuring that the
memory buffer, passed in as *ptr*, is not deallocated while the returned buffer
object exists. Raises :exc:`ValueError` if *size* is less than zero. Note that
:const:`Py_END_OF_BUFFER` may *not* be passed for the *size* parameter;
:exc:`ValueError` will be raised in that case.
.. cmember:: Py_ssize_t *shape
An array of :ctype:`Py_ssize_t`\s the length of :cdata:`ndim` giving the
shape of the memory as a multi-dimensional array. Note that
``((*shape)[0] * ... * (*shape)[ndims-1])*itemsize`` should be equal to
:cdata:`len`.
.. cfunction:: PyObject* PyBuffer_FromReadWriteMemory(void *ptr, Py_ssize_t size)
.. cmember:: Py_ssize_t *strides
Similar to :cfunc:`PyBuffer_FromMemory`, but the returned buffer is writable.
An array of :ctype:`Py_ssize_t`\s the length of :cdata:`ndim` giving the
number of bytes to skip to get to a new element in each dimension.
.. cmember:: Py_ssize_t *suboffsets
.. cfunction:: PyObject* PyBuffer_New(Py_ssize_t size)
An array of :ctype:`Py_ssize_t`\s the length of :cdata:`ndim`. If these
suboffset numbers are greater than or equal to 0, then the value stored
along the indicated dimension is a pointer and the suboffset value
dictates how many bytes to add to the pointer after de-referencing. A
suboffset value that it negative indicates that no de-referencing should
occur (striding in a contiguous memory block).
Return a new writable buffer object that maintains its own memory buffer of
*size* bytes. :exc:`ValueError` is returned if *size* is not zero or positive.
Note that the memory buffer (as returned by :cfunc:`PyObject_AsWriteBuffer`) is
not specifically aligned.
Here is a function that returns a pointer to the element in an N-D array
pointed to by an N-dimesional index when there are both non-NULL strides
and suboffsets::
void *get_item_pointer(int ndim, void *buf, Py_ssize_t *strides,
Py_ssize_t *suboffsets, Py_ssize_t *indices) {
char *pointer = (char*)buf;
int i;
for (i = 0; i < ndim; i++) {
pointer += strides[i] * indices[i];
if (suboffsets[i] >=0 ) {
pointer = *((char**)pointer) + suboffsets[i];
}
}
return (void*)pointer;
}
.. cmember:: Py_ssize_t itemsize
This is a storage for the itemsize (in bytes) of each element of the
shared memory. It is technically un-necessary as it can be obtained using
:cfunc:`PyBuffer_SizeFromFormat`, however an exporter may know this
information without parsing the format string and it is necessary to know
the itemsize for proper interpretation of striding. Therefore, storing it
is more convenient and faster.
.. cmember:: void *internal
This is for use internally by the exporting object. For example, this
might be re-cast as an integer by the exporter and used to store flags
about whether or not the shape, strides, and suboffsets arrays must be
freed when the buffer is released. The consumer should never alter this
value.
Buffer related functions
========================
.. cfunction:: int PyObject_CheckBuffer(PyObject *obj)
Return 1 if *obj* supports the buffer interface otherwise 0.
.. cfunction:: int PyObject_GetBuffer(PyObject *obj, PyObject *view, int flags)
Export *obj* into a :ctype:`Py_buffer`, *view*. These arguments must
never be *NULL*. The *flags* argument is a bit field indicating what kind
of buffer the caller is prepared to deal with and therefore what kind of
buffer the exporter is allowed to return. The buffer interface allows for
complicated memory sharing possibilities, but some caller may not be able
to handle all the complexibity but may want to see if the exporter will
let them take a simpler view to its memory.
Some exporters may not be able to share memory in every possible way and
may need to raise errors to signal to some consumers that something is
just not possible. These errors should be a :exc:`BufferError` unless
there is another error that is actually causing the problem. The exporter
can use flags information to simplify how much of the :cdata:`Py_buffer`
structure is filled in with non-default values and/or raise an error if
the object can't support a simpler view of its memory.
0 is returned on success and -1 on error.
The following table gives possible values to the *flags* arguments.
+------------------------------+-----------------------------------------------+
| Flag | Description |
+==============================+===============================================+
| :cmacro:`PyBUF_SIMPLE` |This is the default flag state. The returned |
| |buffer may or may not have writable memory. |
| |The format will be assumed to be unsigned bytes|
| |. This is a "stand-alone" flag constant. It |
| |never needs to be |'d to the others. The |
| |exporter will raise an error if it cannot |
| |provide such a contiguous buffer of bytes. |
| | |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_WRITABLE` |The returned buffer must be writable. If it is |
| |not writable, then raise an error. |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_STRIDES` |This implies :cmacro:`PyBUF_ND`. The returned |
| |buffer must provide strides information |
| |(i.e. the strides cannot be NULL). This would |
| |be used when the consumer can handle strided, |
| |discontiguous arrays. Handling strides |
| |automatically assumes you can handle shape. The|
| |exporter may raise an error if cannot provide a|
| |strided-only representation of the data |
| |(i.e. without the suboffsets). |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_ND` |The returned buffer must provide shape |
| |information. The memory will be assumed C-style|
| |contiguous (last dimension varies the |
| |fastest). The exporter may raise an error if it|
| |cannot provide this kind of contiguous |
| |buffer. If this is not given then shape will be|
| |*NULL*. |
| | |
| | |
+------------------------------+-----------------------------------------------+
|:cmacro:`PyBUF_C_CONTIGUOUS` |These flags indicate that the contiguoity |
|:cmacro:`PyBUF_F_CONTIGUOUS` |returned buffer must be respectively, |
|:cmacro:`PyBUF_ANY_CONTIGUOUS`|C-contiguous (last dimension varies the |
| |fastest), Fortran contiguous (first dimension |
| |varies the fastest) or either one. All of |
| |these flags imply :cmacro:`PyBUF_STRIDES` and |
| |guarantee that the strides buffer info |
| |structure will be filled in correctly. |
| | |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_INDIRECT` |This implies :cmacro:`PyBUF_STRIDES`. The |
| |returned buffer must have suboffsets |
| |information (which can be NULL if no suboffsets|
| |are needed). This would be used when the |
| |consumer can handle indirect array referencing |
| |implied by these suboffsets. |
| | |
| | |
| | |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_FORMAT` |The returned buffer must have true format |
| |information if this flag is provided. This |
| |would be used when the consumer is going to be |
| |checking for what 'kind' of data is actually |
| |stored. An exporter should always be able to |
| |provide this information if requested. If |
| |format is not explicitly requested then the |
| |format must be returned as *NULL* (which means |
| |``'B'``, or unsigned bytes) |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_STRIDED` |This is equivalent to ``(PyBUF_STRIDES | |
| |PyBUF_WRITABLE)``. |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_STRIDED_RO` |This is equivalent to ``(PyBUF_STRIDES)``. |
| | |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_RECORDS` |This is equivalent to ``(PyBUF_STRIDES | |
| |PyBUF_FORMAT | PyBUF_WRITABLE)``. |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_RECORDS_RO` |This is equivalent to ``(PyBUF_STRIDES | |
| |PyBUF_FORMAT)``. |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_FULL` |This is equivalent to ``(PyBUF_INDIRECT | |
| |PyBUF_FORMAT | PyBUF_WRITABLE)``. |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_FULL_RO`` |This is equivalent to ``(PyBUF_INDIRECT | |
| |PyBUF_FORMAT)``. |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_CONTIG` |This is equivalent to ``(PyBUF_ND | |
| |PyBUF_WRITABLE)``. |
+------------------------------+-----------------------------------------------+
| :cmacro:`PyBUF_CONTIG_RO` |This is equivalent to ``(PyBUF_ND)``. |
| | |
+------------------------------+-----------------------------------------------+
.. cfunction:: void PyBuffer_Release(PyObject *obj, Py_buffer *view)
Release the buffer *view* over *obj*. This shouldd be called when the buffer
is no longer being used as it may free memory from it.
.. cfunction:: Py_ssize_t PyBuffer_SizeFromFormat(const char *)
Return the implied :cdata:`~Py_buffer.itemsize` from the struct-stype
:cdata:`~Py_buffer.format`.
.. cfunction:: int PyObject_CopyToObject(PyObject *obj, void *buf, Py_ssize_t len, char fortran)
Copy *len* bytes of data pointed to by the contiguous chunk of memory pointed
to by *buf* into the buffer exported by obj. The buffer must of course be
writable. Return 0 on success and return -1 and raise an error on failure.
If the object does not have a writable buffer, then an error is raised. If
*fortran* is ``'F'``, then if the object is multi-dimensional, then the data
will be copied into the array in Fortran-style (first dimension varies the
fastest). If *fortran* is ``'C'``, then the data will be copied into the
array in C-style (last dimension varies the fastest). If *fortran* is
``'A'``, then it does not matter and the copy will be made in whatever way is
more efficient.
.. cfunction:: int PyBuffer_IsContiguous(Py_buffer *view, char fortran)
Return 1 if the memory defined by the *view* is C-style (*fortran* is
``'C'``) or Fortran-style (*fortran* is ``'F'``) contiguous or either one
(*fortran* is ``'A'``). Return 0 otherwise.
.. cfunction:: void PyBuffer_FillContiguousStrides(int ndim, Py_ssize_t *shape, Py_ssize_t *strides, Py_ssize_t itemsize, char fortran)
Fill the *strides* array with byte-strides of a contiguous (C-style if
*fortran* is ``'C'`` or Fortran-style if *fortran* is ``'F'`` array of the
given shape with the given number of bytes per element.
.. cfunction:: int PyBuffer_FillInfo(Py_buffer *view, void *buf, Py_ssize_t len, int readonly, int infoflags)
Fill in a buffer-info structure, *view*, correctly for an exporter that can
only share a contiguous chunk of memory of "unsigned bytes" of the given
length. Return 0 on success and -1 (with raising an error) on error.
MemoryView objects
==================
A memoryview object is an extended buffer object that could replace the buffer
object (but doesn't have to as that could be kept as a simple 1-d memoryview
object). It, unlike :ctype:`Py_buffer`, is a Python object (exposed as
:class:`memoryview` in :mod:`builtins`), so it can be used with Python code.
.. cfunction:: PyObject* PyMemoryView_FromObject(PyObject *obj)
Return a memoryview object from an object that defines the buffer interface.
Doc/c-api/typeobj.rst
Dosyayı görüntüle @ 9d0ced36
......@@ -1196,101 +1196,47 @@ Buffer Object Structures
========================
.. sectionauthor:: Greg J. Stein <greg@lyra.org>
.. sectionauthor:: Benjamin Peterson
The buffer interface exports a model where an object can expose its internal
data as a set of chunks of data, where each chunk is specified as a
pointer/length pair. These chunks are called :dfn:`segments` and are presumed
to be non-contiguous in memory.
data.
If an object does not export the buffer interface, then its :attr:`tp_as_buffer`
member in the :ctype:`PyTypeObject` structure should be *NULL*. Otherwise, the
:attr:`tp_as_buffer` will point to a :ctype:`PyBufferProcs` structure.
.. note::
It is very important that your :ctype:`PyTypeObject` structure uses
:const:`Py_TPFLAGS_DEFAULT` for the value of the :attr:`tp_flags` member rather
than ``0``. This tells the Python runtime that your :ctype:`PyBufferProcs`
structure contains the :attr:`bf_getcharbuffer` slot. Older versions of Python
did not have this member, so a new Python interpreter using an old extension
needs to be able to test for its presence before using it.
.. XXX out of date!
.. ctype:: PyBufferProcs
Structure used to hold the function pointers which define an implementation of
the buffer protocol.
The first slot is :attr:`bf_getreadbuffer`, of type :ctype:`getreadbufferproc`.
If this slot is *NULL*, then the object does not support reading from the
internal data. This is non-sensical, so implementors should fill this in, but
callers should test that the slot contains a non-*NULL* value.
The next slot is :attr:`bf_getwritebuffer` having type
:ctype:`getwritebufferproc`. This slot may be *NULL* if the object does not
allow writing into its returned buffers.
The third slot is :attr:`bf_getsegcount`, with type :ctype:`getsegcountproc`.
This slot must not be *NULL* and is used to inform the caller how many segments
the object contains. Simple objects such as :ctype:`PyString_Type` and
:ctype:`PyBuffer_Type` objects contain a single segment.
.. index:: single: PyType_HasFeature()
The last slot is :attr:`bf_getcharbuffer`, of type :ctype:`getcharbufferproc`.
This slot will only be present if the :const:`Py_TPFLAGS_HAVE_GETCHARBUFFER`
flag is present in the :attr:`tp_flags` field of the object's
:ctype:`PyTypeObject`. Before using this slot, the caller should test whether it
is present by using the :cfunc:`PyType_HasFeature` function. If the flag is
present, :attr:`bf_getcharbuffer` may be *NULL*, indicating that the object's
contents cannot be used as *8-bit characters*. The slot function may also raise
an error if the object's contents cannot be interpreted as 8-bit characters.
For example, if the object is an array which is configured to hold floating
point values, an exception may be raised if a caller attempts to use
:attr:`bf_getcharbuffer` to fetch a sequence of 8-bit characters. This notion of
exporting the internal buffers as "text" is used to distinguish between objects
that are binary in nature, and those which have character-based content.
.. note::
The current policy seems to state that these characters may be multi-byte
characters. This implies that a buffer size of *N* does not mean there are *N*
characters present.
.. ctype:: Py_ssize_t (*readbufferproc) (PyObject *self, Py_ssize_t segment, void **ptrptr)
Return a pointer to a readable segment of the buffer in ``*ptrptr``. This
function is allowed to raise an exception, in which case it must return ``-1``.
The *segment* which is specified must be zero or positive, and strictly less
than the number of segments returned by the :attr:`bf_getsegcount` slot
function. On success, it returns the length of the segment, and sets
``*ptrptr`` to a pointer to that memory.
.. ctype:: Py_ssize_t (*writebufferproc) (PyObject *self, Py_ssize_t segment, void **ptrptr)
Return a pointer to a writable memory buffer in ``*ptrptr``, and the length of
that segment as the function return value. The memory buffer must correspond to
buffer segment *segment*. Must return ``-1`` and set an exception on error.
:exc:`TypeError` should be raised if the object only supports read-only buffers,
and :exc:`SystemError` should be raised when *segment* specifies a segment that
doesn't exist.
.. Why doesn't it raise ValueError for this one?
GJS: because you shouldn't be calling it with an invalid
segment. That indicates a blatant programming error in the C code.
.. cmember:: getbufferproc bf_getbuffer
This should fill a :ctype:`Py_buffer` with the necessary data for
exporting the type. The signature of :data:`getbufferproc` is ``int
(PyObject *obj, PyObject *view, int flags)``. *obj* is the object to
export, *view* is the :ctype:`Py_buffer` struct to fill, and *flags* gives
the conditions the caller wants the memory under. (See
:cfunc:`PyObject_GetBuffer` for all flags.) :cmember:`bf_getbuffer` is
responsible for filling *view* with the approiate information.
(:cfunc:`PyBuffer_FillView` can be used in simple cases.) See
:ctype:`Py_buffer`\s docs for what needs to be filled in.
.. ctype:: Py_ssize_t (*segcountproc) (PyObject *self, Py_ssize_t *lenp)
Return the number of memory segments which comprise the buffer. If *lenp* is
not *NULL*, the implementation must report the sum of the sizes (in bytes) of
all segments in ``*lenp``. The function cannot fail.
.. cmember:: releasebufferproc bf_releasebuffer
This should release the resources of the buffer. The signature of
:cdata:`releasebufferproc` is ``void (PyObject *obj, Py_buffer *view)``.
If the :cdata:`bf_releasebuffer` function is not provided (i.e. it is
*NULL*), then it does not ever need to be called.
.. ctype:: Py_ssize_t (*charbufferproc) (PyObject *self, Py_ssize_t segment, const char **ptrptr)
The exporter of the buffer interface must make sure that any memory
pointed to in the :ctype:`Py_buffer` structure remains valid until
releasebuffer is called. Exporters will need to define a
:cdata:`bf_releasebuffer` function if they can re-allocate their memory,
strides, shape, suboffsets, or format variables which they might share
through the struct bufferinfo.
Return the size of the segment *segment* that *ptrptr* is set to. ``*ptrptr``
is set to the memory buffer. Returns ``-1`` on error.
See :cfunc:`PyBuffer_Release`.
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment