cpython/Doc/c-api/intro.rst

738 lines
31 KiB
ReStructuredText
Raw Normal View History

.. highlight:: c
2007-08-15 11:28:22 -03:00
.. _api-intro:
************
Introduction
************
The Application Programmer's Interface to Python gives C and C++ programmers
access to the Python interpreter at a variety of levels. The API is equally
usable from C++, but for brevity it is generally referred to as the Python/C
API. There are two fundamentally different reasons for using the Python/C API.
The first reason is to write *extension modules* for specific purposes; these
are C modules that extend the Python interpreter. This is probably the most
common use. The second reason is to use Python as a component in a larger
application; this technique is generally referred to as :dfn:`embedding` Python
in an application.
Writing an extension module is a relatively well-understood process, where a
"cookbook" approach works well. There are several tools that automate the
process to some extent. While people have embedded Python in other
applications since its early existence, the process of embedding Python is
less straightforward than writing an extension.
2007-08-15 11:28:22 -03:00
Many API functions are useful independent of whether you're embedding or
extending Python; moreover, most applications that embed Python will need to
provide a custom extension as well, so it's probably a good idea to become
familiar with writing an extension before attempting to embed Python in a real
application.
Coding standards
================
If you're writing C code for inclusion in CPython, you **must** follow the
guidelines and standards defined in :PEP:`7`. These guidelines apply
regardless of the version of Python you are contributing to. Following these
conventions is not necessary for your own third party extension modules,
unless you eventually expect to contribute them to Python.
2007-08-15 11:28:22 -03:00
.. _api-includes:
Include Files
=============
All function, type and macro definitions needed to use the Python/C API are
included in your code by the following line::
#define PY_SSIZE_T_CLEAN
#include <Python.h>
2007-08-15 11:28:22 -03:00
This implies inclusion of the following standard headers: ``<stdio.h>``,
``<string.h>``, ``<errno.h>``, ``<limits.h>``, ``<assert.h>`` and ``<stdlib.h>``
(if available).
2007-08-15 11:28:22 -03:00
.. note::
2007-08-15 11:28:22 -03:00
Since Python may define some pre-processor definitions which affect the standard
headers on some systems, you *must* include :file:`Python.h` before any standard
headers are included.
It is recommended to always define ``PY_SSIZE_T_CLEAN`` before including
``Python.h``. See :ref:`arg-parsing` for a description of this macro.
2007-08-15 11:28:22 -03:00
All user visible names defined by Python.h (except those defined by the included
standard headers) have one of the prefixes ``Py`` or ``_Py``. Names beginning
with ``_Py`` are for internal use by the Python implementation and should not be
used by extension writers. Structure member names do not have a reserved prefix.
**Important:** user code should never define names that begin with ``Py`` or
``_Py``. This confuses the reader, and jeopardizes the portability of the user
code to future Python versions, which may define additional names beginning with
one of these prefixes.
The header files are typically installed with Python. On Unix, these are
located in the directories :file:`{prefix}/include/pythonversion/` and
:file:`{exec_prefix}/include/pythonversion/`, where :envvar:`prefix` and
:envvar:`exec_prefix` are defined by the corresponding parameters to Python's
:program:`configure` script and *version* is
``'%d.%d' % sys.version_info[:2]``. On Windows, the headers are installed
in :file:`{prefix}/include`, where :envvar:`prefix` is the installation
directory specified to the installer.
2007-08-15 11:28:22 -03:00
To include the headers, place both directories (if different) on your compiler's
search path for includes. Do *not* place the parent directories on the search
path and then use ``#include <pythonX.Y/Python.h>``; this will break on
multi-platform builds since the platform independent headers under
:envvar:`prefix` include the platform specific headers from
:envvar:`exec_prefix`.
C++ users should note that though the API is defined entirely using C, the
header files do properly declare the entry points to be ``extern "C"``, so there
is no need to do anything special to use the API from C++.
Useful macros
=============
Several useful macros are defined in the Python header files. Many are
defined closer to where they are useful (e.g. :c:macro:`Py_RETURN_NONE`).
Others of a more general utility are defined here. This is not necessarily a
complete listing.
.. c:macro:: Py_UNREACHABLE()
Use this when you have a code path that you do not expect to be reached.
For example, in the ``default:`` clause in a ``switch`` statement for which
all possible values are covered in ``case`` statements. Use this in places
where you might be tempted to put an ``assert(0)`` or ``abort()`` call.
.. versionadded:: 3.7
.. c:macro:: Py_ABS(x)
Return the absolute value of ``x``.
.. versionadded:: 3.3
.. c:macro:: Py_MIN(x, y)
Return the minimum value between ``x`` and ``y``.
.. versionadded:: 3.3
.. c:macro:: Py_MAX(x, y)
Return the maximum value between ``x`` and ``y``.
.. versionadded:: 3.3
.. c:macro:: Py_STRINGIFY(x)
Convert ``x`` to a C string. E.g. ``Py_STRINGIFY(123)`` returns
``"123"``.
.. versionadded:: 3.4
.. c:macro:: Py_MEMBER_SIZE(type, member)
Return the size of a structure (``type``) ``member`` in bytes.
.. versionadded:: 3.6
.. c:macro:: Py_CHARMASK(c)
Argument must be a character or an integer in the range [-128, 127] or [0,
255]. This macro returns ``c`` cast to an ``unsigned char``.
2017-10-06 10:53:48 -03:00
.. c:macro:: Py_GETENV(s)
Like ``getenv(s)``, but returns *NULL* if :option:`-E` was passed on the
command line (i.e. if ``Py_IgnoreEnvironmentFlag`` is set).
.. c:macro:: Py_UNUSED(arg)
Use this for unused arguments in a function definition to silence compiler
warnings. Example: ``int func(int a, int Py_UNUSED(b)) { return a; }``.
.. versionadded:: 3.4
.. c:macro:: Py_DEPRECATED(version)
Use this for deprecated declarations. The macro must be placed before the
symbol name.
Example::
Py_DEPRECATED(3.8) PyAPI_FUNC(int) Py_OldFunction(void);
.. versionchanged:: 3.8
MSVC support was added.
2007-08-15 11:28:22 -03:00
.. _api-objects:
Objects, Types and Reference Counts
===================================
.. index:: object: type
Most Python/C API functions have one or more arguments as well as a return value
of type :c:type:`PyObject\*`. This type is a pointer to an opaque data type
2007-08-15 11:28:22 -03:00
representing an arbitrary Python object. Since all Python object types are
treated the same way by the Python language in most situations (e.g.,
assignments, scope rules, and argument passing), it is only fitting that they
should be represented by a single C type. Almost all Python objects live on the
heap: you never declare an automatic or static variable of type
:c:type:`PyObject`, only pointer variables of type :c:type:`PyObject\*` can be
2007-08-15 11:28:22 -03:00
declared. The sole exception are the type objects; since these must never be
deallocated, they are typically static :c:type:`PyTypeObject` objects.
2007-08-15 11:28:22 -03:00
All Python objects (even Python integers) have a :dfn:`type` and a
:dfn:`reference count`. An object's type determines what kind of object it is
(e.g., an integer, a list, or a user-defined function; there are many more as
explained in :ref:`types`). For each of the well-known types there is a macro
to check whether an object is of that type; for instance, ``PyList_Check(a)`` is
true if (and only if) the object pointed to by *a* is a Python list.
.. _api-refcounts:
Reference Counts
----------------
The reference count is important because today's computers have a finite (and
often severely limited) memory size; it counts how many different places there
are that have a reference to an object. Such a place could be another object,
or a global (or static) C variable, or a local variable in some C function.
When an object's reference count becomes zero, the object is deallocated. If
it contains references to other objects, their reference count is decremented.
Those other objects may be deallocated in turn, if this decrement makes their
reference count become zero, and so on. (There's an obvious problem with
objects that reference each other here; for now, the solution is "don't do
that.")
.. index::
single: Py_INCREF()
single: Py_DECREF()
Reference counts are always manipulated explicitly. The normal way is to use
the macro :c:func:`Py_INCREF` to increment an object's reference count by one,
and :c:func:`Py_DECREF` to decrement it by one. The :c:func:`Py_DECREF` macro
2007-08-15 11:28:22 -03:00
is considerably more complex than the incref one, since it must check whether
the reference count becomes zero and then cause the object's deallocator to be
called. The deallocator is a function pointer contained in the object's type
structure. The type-specific deallocator takes care of decrementing the
reference counts for other objects contained in the object if this is a compound
object type, such as a list, as well as performing any additional finalization
that's needed. There's no chance that the reference count can overflow; at
least as many bits are used to hold the reference count as there are distinct
Merged revisions 61239-61249,61252-61257,61260-61264,61269-61275,61278-61279,61285-61286,61288-61290,61298,61303-61305,61312-61314,61317,61329,61332,61344,61350-61351,61363-61376,61378-61379,61382-61383,61387-61388,61392,61395-61396,61402-61403 via svnmerge from svn+ssh://pythondev@svn.python.org/python/trunk ........ r61239 | andrew.kuchling | 2008-03-05 01:44:41 +0100 (Wed, 05 Mar 2008) | 1 line Add more items; add fragmentary notes ........ r61240 | amaury.forgeotdarc | 2008-03-05 02:50:33 +0100 (Wed, 05 Mar 2008) | 13 lines Issue#2238: some syntax errors from *args or **kwargs expressions would give bogus error messages, because of untested exceptions:: >>> f(**g(1=2)) XXX undetected error Traceback (most recent call last): File "<stdin>", line 1, in <module> TypeError: 'int' object is not iterable instead of the expected SyntaxError: keyword can't be an expression Will backport. ........ r61241 | neal.norwitz | 2008-03-05 06:10:48 +0100 (Wed, 05 Mar 2008) | 3 lines Remove the files/dirs after closing the DB so the tests work on Windows. Patch from Trent Nelson. Also simplified removing a file by using test_support. ........ r61242 | neal.norwitz | 2008-03-05 06:14:18 +0100 (Wed, 05 Mar 2008) | 3 lines Get this test to pass even when there is no sound card in the system. Patch from Trent Nelson. (I can't test this.) ........ r61243 | neal.norwitz | 2008-03-05 06:20:44 +0100 (Wed, 05 Mar 2008) | 3 lines Catch OSError when trying to remove a file in case removal fails. This should prevent a failure in tearDown masking any real test failure. ........ r61244 | neal.norwitz | 2008-03-05 06:38:06 +0100 (Wed, 05 Mar 2008) | 5 lines Make the timeout longer to give slow machines a chance to pass the test before timing out. This doesn't change the duration of the test under normal circumstances. This is targetted at fixing the spurious failures on the FreeBSD buildbot primarily. ........ r61245 | neal.norwitz | 2008-03-05 06:49:03 +0100 (Wed, 05 Mar 2008) | 1 line Tabs -> spaces ........ r61246 | neal.norwitz | 2008-03-05 06:50:20 +0100 (Wed, 05 Mar 2008) | 1 line Use -u urlfetch to run more tests ........ r61247 | neal.norwitz | 2008-03-05 06:51:20 +0100 (Wed, 05 Mar 2008) | 1 line test_smtplib sometimes reports leaks too, suppress it ........ r61248 | jeffrey.yasskin | 2008-03-05 07:19:56 +0100 (Wed, 05 Mar 2008) | 5 lines Fix test_socketserver on Windows after r61099 added several signal.alarm() calls (which don't exist on non-Unix platforms). Thanks to Trent Nelson for the report and patch. ........ r61249 | georg.brandl | 2008-03-05 08:10:35 +0100 (Wed, 05 Mar 2008) | 2 lines Fix some rst. ........ r61252 | thomas.heller | 2008-03-05 15:53:39 +0100 (Wed, 05 Mar 2008) | 2 lines News entry for yesterdays commit. ........ r61253 | thomas.heller | 2008-03-05 16:34:29 +0100 (Wed, 05 Mar 2008) | 3 lines Issue 1872: Changed the struct module typecode from 't' to '?', for compatibility with PEP3118. ........ r61254 | skip.montanaro | 2008-03-05 17:41:09 +0100 (Wed, 05 Mar 2008) | 4 lines Elaborate on the role of the altinstall target when installing multiple versions. ........ r61255 | georg.brandl | 2008-03-05 20:31:44 +0100 (Wed, 05 Mar 2008) | 2 lines #2239: PYTHONPATH delimiter is os.pathsep. ........ r61256 | raymond.hettinger | 2008-03-05 21:59:58 +0100 (Wed, 05 Mar 2008) | 1 line C implementation of itertools.permutations(). ........ r61257 | raymond.hettinger | 2008-03-05 22:04:32 +0100 (Wed, 05 Mar 2008) | 1 line Small code cleanup. ........ r61260 | martin.v.loewis | 2008-03-05 23:24:31 +0100 (Wed, 05 Mar 2008) | 2 lines cd PCbuild only after deleting all pyc files. ........ r61261 | raymond.hettinger | 2008-03-06 02:15:52 +0100 (Thu, 06 Mar 2008) | 1 line Add examples. ........ r61262 | andrew.kuchling | 2008-03-06 02:36:27 +0100 (Thu, 06 Mar 2008) | 1 line Add two items ........ r61263 | georg.brandl | 2008-03-06 07:47:18 +0100 (Thu, 06 Mar 2008) | 2 lines #1725737: ignore other VC directories other than CVS and SVN's too. ........ r61264 | martin.v.loewis | 2008-03-06 07:55:22 +0100 (Thu, 06 Mar 2008) | 4 lines Patch #2232: os.tmpfile might fail on Windows if the user has no permission to create files in the root directory. Will backport to 2.5. ........ r61269 | georg.brandl | 2008-03-06 08:19:15 +0100 (Thu, 06 Mar 2008) | 2 lines Expand on re.split behavior with captured expressions. ........ r61270 | georg.brandl | 2008-03-06 08:22:09 +0100 (Thu, 06 Mar 2008) | 2 lines Little clarification of assignments. ........ r61271 | georg.brandl | 2008-03-06 08:31:34 +0100 (Thu, 06 Mar 2008) | 2 lines Add isinstance/issubclass to tutorial. ........ r61272 | georg.brandl | 2008-03-06 08:34:52 +0100 (Thu, 06 Mar 2008) | 2 lines Add missing NEWS entry for r61263. ........ r61273 | georg.brandl | 2008-03-06 08:41:16 +0100 (Thu, 06 Mar 2008) | 2 lines #2225: return nonzero status code from py_compile if not all files could be compiled. ........ r61274 | georg.brandl | 2008-03-06 08:43:02 +0100 (Thu, 06 Mar 2008) | 2 lines #2220: handle matching failure more gracefully. ........ r61275 | georg.brandl | 2008-03-06 08:45:52 +0100 (Thu, 06 Mar 2008) | 2 lines Bug #2220: handle rlcompleter attribute match failure more gracefully. ........ r61278 | martin.v.loewis | 2008-03-06 14:49:47 +0100 (Thu, 06 Mar 2008) | 1 line Rely on x64 platform configuration when building _bsddb on AMD64. ........ r61279 | martin.v.loewis | 2008-03-06 14:50:28 +0100 (Thu, 06 Mar 2008) | 1 line Update db-4.4.20 build procedure. ........ r61285 | raymond.hettinger | 2008-03-06 21:52:01 +0100 (Thu, 06 Mar 2008) | 1 line More tests. ........ r61286 | raymond.hettinger | 2008-03-06 23:51:36 +0100 (Thu, 06 Mar 2008) | 1 line Issue 2246: itertools grouper object did not participate in GC (should be backported). ........ r61288 | raymond.hettinger | 2008-03-07 02:33:20 +0100 (Fri, 07 Mar 2008) | 1 line Tweak recipes and tests ........ r61289 | jeffrey.yasskin | 2008-03-07 07:22:15 +0100 (Fri, 07 Mar 2008) | 5 lines Progress on issue #1193577 by adding a polling .shutdown() method to SocketServers. The core of the patch was written by Pedro Werneck, but any bugs are mine. I've also rearranged the code for timeouts in order to avoid interfering with the shutdown poll. ........ r61290 | nick.coghlan | 2008-03-07 15:13:28 +0100 (Fri, 07 Mar 2008) | 1 line Speed up with statements by storing the __exit__ method on the stack instead of in a temp variable (bumps the magic number for pyc files) ........ r61298 | andrew.kuchling | 2008-03-07 22:09:23 +0100 (Fri, 07 Mar 2008) | 1 line Grammar fix ........ r61303 | georg.brandl | 2008-03-08 10:54:06 +0100 (Sat, 08 Mar 2008) | 2 lines #2253: fix continue vs. finally docs. ........ r61304 | marc-andre.lemburg | 2008-03-08 11:01:43 +0100 (Sat, 08 Mar 2008) | 3 lines Add new name for Mandrake: Mandriva. ........ r61305 | georg.brandl | 2008-03-08 11:05:24 +0100 (Sat, 08 Mar 2008) | 2 lines #1533486: fix types in refcount intro. ........ r61312 | facundo.batista | 2008-03-08 17:50:27 +0100 (Sat, 08 Mar 2008) | 5 lines Issue 1106316. post_mortem()'s parameter, traceback, is now optional: it defaults to the traceback of the exception that is currently being handled. ........ r61313 | jeffrey.yasskin | 2008-03-08 19:26:54 +0100 (Sat, 08 Mar 2008) | 2 lines Add tests for with and finally performance to pybench. ........ r61314 | jeffrey.yasskin | 2008-03-08 21:08:21 +0100 (Sat, 08 Mar 2008) | 2 lines Fix pybench for pythons < 2.6, tested back to 2.3. ........ r61317 | jeffrey.yasskin | 2008-03-08 22:35:15 +0100 (Sat, 08 Mar 2008) | 3 lines Well that was dumb. platform.python_implementation returns a function, not a string. ........ r61329 | georg.brandl | 2008-03-09 16:11:39 +0100 (Sun, 09 Mar 2008) | 2 lines #2249: document assertTrue and assertFalse. ........ r61332 | neal.norwitz | 2008-03-09 20:03:42 +0100 (Sun, 09 Mar 2008) | 4 lines Introduce a lock to fix a race condition which caused an exception in the test. Some buildbots were consistently failing (e.g., amd64). Also remove a couple of semi-colons. ........ r61344 | raymond.hettinger | 2008-03-11 01:19:07 +0100 (Tue, 11 Mar 2008) | 1 line Add recipe to docs. ........ r61350 | guido.van.rossum | 2008-03-11 22:18:06 +0100 (Tue, 11 Mar 2008) | 3 lines Fix the overflows in expandtabs(). "This time for sure!" (Exploit at request.) ........ r61351 | raymond.hettinger | 2008-03-11 22:37:46 +0100 (Tue, 11 Mar 2008) | 1 line Improve docs for itemgetter(). Show that it works with slices. ........ r61363 | georg.brandl | 2008-03-13 08:15:56 +0100 (Thu, 13 Mar 2008) | 2 lines #2265: fix example. ........ r61364 | georg.brandl | 2008-03-13 08:17:14 +0100 (Thu, 13 Mar 2008) | 2 lines #2270: fix typo. ........ r61365 | georg.brandl | 2008-03-13 08:21:41 +0100 (Thu, 13 Mar 2008) | 2 lines #1720705: add docs about import/threading interaction, wording by Nick. ........ r61366 | andrew.kuchling | 2008-03-13 12:07:35 +0100 (Thu, 13 Mar 2008) | 1 line Add class decorators ........ r61367 | raymond.hettinger | 2008-03-13 17:43:17 +0100 (Thu, 13 Mar 2008) | 1 line Add 2-to-3 support for the itertools moved to builtins or renamed. ........ r61368 | raymond.hettinger | 2008-03-13 17:43:59 +0100 (Thu, 13 Mar 2008) | 1 line Consistent tense. ........ r61369 | raymond.hettinger | 2008-03-13 20:03:51 +0100 (Thu, 13 Mar 2008) | 1 line Issue 2274: Add heapq.heappushpop(). ........ r61370 | raymond.hettinger | 2008-03-13 20:33:34 +0100 (Thu, 13 Mar 2008) | 1 line Simplify the nlargest() code using heappushpop(). ........ r61371 | brett.cannon | 2008-03-13 21:27:00 +0100 (Thu, 13 Mar 2008) | 4 lines Move test_thread over to unittest. Commits GHOP 237. Thanks Benjamin Peterson for the patch. ........ r61372 | brett.cannon | 2008-03-13 21:33:10 +0100 (Thu, 13 Mar 2008) | 4 lines Move test_tokenize to doctest. Done as GHOP 238 by Josip Dzolonga. ........ r61373 | brett.cannon | 2008-03-13 21:47:41 +0100 (Thu, 13 Mar 2008) | 4 lines Convert test_contains, test_crypt, and test_select to unittest. Patch from GHOP 294 by David Marek. ........ r61374 | brett.cannon | 2008-03-13 22:02:16 +0100 (Thu, 13 Mar 2008) | 4 lines Move test_gdbm to use unittest. Closes issue #1960. Thanks Giampaolo Rodola. ........ r61375 | brett.cannon | 2008-03-13 22:09:28 +0100 (Thu, 13 Mar 2008) | 4 lines Convert test_fcntl to unittest. Closes issue #2055. Thanks Giampaolo Rodola. ........ r61376 | raymond.hettinger | 2008-03-14 06:03:44 +0100 (Fri, 14 Mar 2008) | 1 line Leave heapreplace() unchanged. ........ r61378 | martin.v.loewis | 2008-03-14 14:56:09 +0100 (Fri, 14 Mar 2008) | 2 lines Patch #2284: add -x64 option to rt.bat. ........ r61379 | martin.v.loewis | 2008-03-14 14:57:59 +0100 (Fri, 14 Mar 2008) | 2 lines Use -x64 flag. ........ r61382 | brett.cannon | 2008-03-14 15:03:10 +0100 (Fri, 14 Mar 2008) | 2 lines Remove a bad test. ........ r61383 | mark.dickinson | 2008-03-14 15:23:37 +0100 (Fri, 14 Mar 2008) | 9 lines Issue 705836: Fix struct.pack(">f", 1e40) to behave consistently across platforms: it should now raise OverflowError on all platforms. (Previously it raised OverflowError only on non IEEE 754 platforms.) Also fix the (already existing) test for this behaviour so that it actually raises TestFailed instead of just referencing it. ........ r61387 | thomas.heller | 2008-03-14 22:06:21 +0100 (Fri, 14 Mar 2008) | 1 line Remove unneeded initializer. ........ r61388 | martin.v.loewis | 2008-03-14 22:19:28 +0100 (Fri, 14 Mar 2008) | 2 lines Run debug version, cd to PCbuild. ........ r61392 | georg.brandl | 2008-03-15 00:10:34 +0100 (Sat, 15 Mar 2008) | 2 lines Remove obsolete paragraph. #2288. ........ r61395 | georg.brandl | 2008-03-15 01:20:19 +0100 (Sat, 15 Mar 2008) | 2 lines Fix lots of broken links in the docs, found by Sphinx' external link checker. ........ r61396 | skip.montanaro | 2008-03-15 03:32:49 +0100 (Sat, 15 Mar 2008) | 1 line note that fork and forkpty raise OSError on failure ........ r61402 | skip.montanaro | 2008-03-15 17:04:45 +0100 (Sat, 15 Mar 2008) | 1 line add %f format to datetime - issue 1158 ........ r61403 | skip.montanaro | 2008-03-15 17:07:11 +0100 (Sat, 15 Mar 2008) | 2 lines . ........
2008-03-15 21:07:10 -03:00
memory locations in virtual memory (assuming ``sizeof(Py_ssize_t) >= sizeof(void*)``).
2007-08-15 11:28:22 -03:00
Thus, the reference count increment is a simple operation.
It is not necessary to increment an object's reference count for every local
variable that contains a pointer to an object. In theory, the object's
reference count goes up by one when the variable is made to point to it and it
goes down by one when the variable goes out of scope. However, these two
cancel each other out, so at the end the reference count hasn't changed. The
only real reason to use the reference count is to prevent the object from being
deallocated as long as our variable is pointing to it. If we know that there
is at least one other reference to the object that lives at least as long as
our variable, there is no need to increment the reference count temporarily.
An important situation where this arises is in objects that are passed as
arguments to C functions in an extension module that are called from Python;
the call mechanism guarantees to hold a reference to every argument for the
duration of the call.
However, a common pitfall is to extract an object from a list and hold on to it
for a while without incrementing its reference count. Some other operation might
conceivably remove the object from the list, decrementing its reference count
and possibly deallocating it. The real danger is that innocent-looking
2007-08-15 11:28:22 -03:00
operations may invoke arbitrary Python code which could do this; there is a code
path which allows control to flow back to the user from a :c:func:`Py_DECREF`, so
2007-08-15 11:28:22 -03:00
almost any operation is potentially dangerous.
A safe approach is to always use the generic operations (functions whose name
begins with ``PyObject_``, ``PyNumber_``, ``PySequence_`` or ``PyMapping_``).
These operations always increment the reference count of the object they return.
This leaves the caller with the responsibility to call :c:func:`Py_DECREF` when
2007-08-15 11:28:22 -03:00
they are done with the result; this soon becomes second nature.
.. _api-refcountdetails:
Reference Count Details
^^^^^^^^^^^^^^^^^^^^^^^
The reference count behavior of functions in the Python/C API is best explained
in terms of *ownership of references*. Ownership pertains to references, never
to objects (objects are not owned: they are always shared). "Owning a
reference" means being responsible for calling Py_DECREF on it when the
reference is no longer needed. Ownership can also be transferred, meaning that
the code that receives ownership of the reference then becomes responsible for
eventually decref'ing it by calling :c:func:`Py_DECREF` or :c:func:`Py_XDECREF`
2007-08-15 11:28:22 -03:00
when it's no longer needed---or passing on this responsibility (usually to its
caller). When a function passes ownership of a reference on to its caller, the
caller is said to receive a *new* reference. When no ownership is transferred,
the caller is said to *borrow* the reference. Nothing needs to be done for a
borrowed reference.
Conversely, when a calling function passes in a reference to an object, there
2007-08-15 11:28:22 -03:00
are two possibilities: the function *steals* a reference to the object, or it
does not. *Stealing a reference* means that when you pass a reference to a
function, that function assumes that it now owns that reference, and you are not
responsible for it any longer.
.. index::
single: PyList_SetItem()
single: PyTuple_SetItem()
Few functions steal references; the two notable exceptions are
:c:func:`PyList_SetItem` and :c:func:`PyTuple_SetItem`, which steal a reference
2007-08-15 11:28:22 -03:00
to the item (but not to the tuple or list into which the item is put!). These
functions were designed to steal a reference because of a common idiom for
populating a tuple or list with newly created objects; for example, the code to
create the tuple ``(1, 2, "three")`` could look like this (forgetting about
error handling for the moment; a better way to code this is shown below)::
PyObject *t;
t = PyTuple_New(3);
PyTuple_SetItem(t, 0, PyLong_FromLong(1L));
PyTuple_SetItem(t, 1, PyLong_FromLong(2L));
PyTuple_SetItem(t, 2, PyUnicode_FromString("three"));
2007-08-15 11:28:22 -03:00
Here, :c:func:`PyLong_FromLong` returns a new reference which is immediately
stolen by :c:func:`PyTuple_SetItem`. When you want to keep using an object
although the reference to it will be stolen, use :c:func:`Py_INCREF` to grab
2007-08-15 11:28:22 -03:00
another reference before calling the reference-stealing function.
Incidentally, :c:func:`PyTuple_SetItem` is the *only* way to set tuple items;
:c:func:`PySequence_SetItem` and :c:func:`PyObject_SetItem` refuse to do this
2007-08-15 11:28:22 -03:00
since tuples are an immutable data type. You should only use
:c:func:`PyTuple_SetItem` for tuples that you are creating yourself.
2007-08-15 11:28:22 -03:00
Equivalent code for populating a list can be written using :c:func:`PyList_New`
and :c:func:`PyList_SetItem`.
2007-08-15 11:28:22 -03:00
However, in practice, you will rarely use these ways of creating and populating
a tuple or list. There's a generic function, :c:func:`Py_BuildValue`, that can
2007-08-15 11:28:22 -03:00
create most common objects from C values, directed by a :dfn:`format string`.
For example, the above two blocks of code could be replaced by the following
(which also takes care of the error checking)::
PyObject *tuple, *list;
tuple = Py_BuildValue("(iis)", 1, 2, "three");
list = Py_BuildValue("[iis]", 1, 2, "three");
It is much more common to use :c:func:`PyObject_SetItem` and friends with items
2007-08-15 11:28:22 -03:00
whose references you are only borrowing, like arguments that were passed in to
the function you are writing. In that case, their behaviour regarding reference
counts is much saner, since you don't have to increment a reference count so you
can give a reference away ("have it be stolen"). For example, this function
sets all items of a list (actually, any mutable sequence) to a given item::
int
set_all(PyObject *target, PyObject *item)
{
Py_ssize_t i, n;
2007-08-15 11:28:22 -03:00
n = PyObject_Length(target);
if (n < 0)
return -1;
for (i = 0; i < n; i++) {
PyObject *index = PyLong_FromSsize_t(i);
2007-08-15 11:28:22 -03:00
if (!index)
return -1;
if (PyObject_SetItem(target, index, item) < 0) {
Py_DECREF(index);
2007-08-15 11:28:22 -03:00
return -1;
}
2007-08-15 11:28:22 -03:00
Py_DECREF(index);
}
return 0;
}
.. index:: single: set_all()
The situation is slightly different for function return values. While passing
a reference to most functions does not change your ownership responsibilities
for that reference, many functions that return a reference to an object give
you ownership of the reference. The reason is simple: in many cases, the
returned object is created on the fly, and the reference you get is the only
reference to the object. Therefore, the generic functions that return object
references, like :c:func:`PyObject_GetItem` and :c:func:`PySequence_GetItem`,
2007-08-15 11:28:22 -03:00
always return a new reference (the caller becomes the owner of the reference).
It is important to realize that whether you own a reference returned by a
function depends on which function you call only --- *the plumage* (the type of
the object passed as an argument to the function) *doesn't enter into it!*
Thus, if you extract an item from a list using :c:func:`PyList_GetItem`, you
2007-08-15 11:28:22 -03:00
don't own the reference --- but if you obtain the same item from the same list
using :c:func:`PySequence_GetItem` (which happens to take exactly the same
2007-08-15 11:28:22 -03:00
arguments), you do own a reference to the returned object.
.. index::
single: PyList_GetItem()
single: PySequence_GetItem()
Here is an example of how you could write a function that computes the sum of
the items in a list of integers; once using :c:func:`PyList_GetItem`, and once
using :c:func:`PySequence_GetItem`. ::
2007-08-15 11:28:22 -03:00
long
sum_list(PyObject *list)
{
Py_ssize_t i, n;
long total = 0, value;
2007-08-15 11:28:22 -03:00
PyObject *item;
n = PyList_Size(list);
if (n < 0)
return -1; /* Not a list */
for (i = 0; i < n; i++) {
item = PyList_GetItem(list, i); /* Can't fail */
if (!PyLong_Check(item)) continue; /* Skip non-integers */
value = PyLong_AsLong(item);
if (value == -1 && PyErr_Occurred())
/* Integer too big to fit in a C long, bail out */
return -1;
total += value;
2007-08-15 11:28:22 -03:00
}
return total;
}
.. index:: single: sum_list()
::
long
sum_sequence(PyObject *sequence)
{
Py_ssize_t i, n;
long total = 0, value;
2007-08-15 11:28:22 -03:00
PyObject *item;
n = PySequence_Length(sequence);
if (n < 0)
return -1; /* Has no length */
for (i = 0; i < n; i++) {
item = PySequence_GetItem(sequence, i);
if (item == NULL)
return -1; /* Not a sequence, or other failure */
if (PyLong_Check(item)) {
value = PyLong_AsLong(item);
Py_DECREF(item);
if (value == -1 && PyErr_Occurred())
/* Integer too big to fit in a C long, bail out */
return -1;
total += value;
}
else {
Py_DECREF(item); /* Discard reference ownership */
}
2007-08-15 11:28:22 -03:00
}
return total;
}
.. index:: single: sum_sequence()
.. _api-types:
Types
-----
There are few other data types that play a significant role in the Python/C
API; most are simple C types such as :c:type:`int`, :c:type:`long`,
:c:type:`double` and :c:type:`char\*`. A few structure types are used to
2007-08-15 11:28:22 -03:00
describe static tables used to list the functions exported by a module or the
data attributes of a new object type, and another is used to describe the value
of a complex number. These will be discussed together with the functions that
use them.
.. _api-exceptions:
Exceptions
==========
The Python programmer only needs to deal with exceptions if specific error
handling is required; unhandled exceptions are automatically propagated to the
caller, then to the caller's caller, and so on, until they reach the top-level
interpreter, where they are reported to the user accompanied by a stack
traceback.
.. index:: single: PyErr_Occurred()
For C programmers, however, error checking always has to be explicit. All
functions in the Python/C API can raise exceptions, unless an explicit claim is
made otherwise in a function's documentation. In general, when a function
encounters an error, it sets an exception, discards any object references that
it owns, and returns an error indicator. If not documented otherwise, this
indicator is either *NULL* or ``-1``, depending on the function's return type.
A few functions return a Boolean true/false result, with false indicating an
error. Very few functions return no explicit error indicator or have an
ambiguous return value, and require explicit testing for errors with
:c:func:`PyErr_Occurred`. These exceptions are always explicitly documented.
2007-08-15 11:28:22 -03:00
.. index::
single: PyErr_SetString()
single: PyErr_Clear()
Exception state is maintained in per-thread storage (this is equivalent to
using global storage in an unthreaded application). A thread can be in one of
two states: an exception has occurred, or not. The function
:c:func:`PyErr_Occurred` can be used to check for this: it returns a borrowed
2007-08-15 11:28:22 -03:00
reference to the exception type object when an exception has occurred, and
*NULL* otherwise. There are a number of functions to set the exception state:
:c:func:`PyErr_SetString` is the most common (though not the most general)
function to set the exception state, and :c:func:`PyErr_Clear` clears the
2007-08-15 11:28:22 -03:00
exception state.
The full exception state consists of three objects (all of which can be
*NULL*): the exception type, the corresponding exception value, and the
traceback. These have the same meanings as the Python result of
``sys.exc_info()``; however, they are not the same: the Python objects represent
the last exception being handled by a Python :keyword:`try` ...
:keyword:`except` statement, while the C level exception state only exists while
an exception is being passed on between C functions until it reaches the Python
bytecode interpreter's main loop, which takes care of transferring it to
``sys.exc_info()`` and friends.
.. index:: single: exc_info() (in module sys)
Note that starting with Python 1.5, the preferred, thread-safe way to access the
exception state from Python code is to call the function :func:`sys.exc_info`,
which returns the per-thread exception state for Python code. Also, the
semantics of both ways to access the exception state have changed so that a
function which catches an exception will save and restore its thread's exception
state so as to preserve the exception state of its caller. This prevents common
bugs in exception handling code caused by an innocent-looking function
overwriting the exception being handled; it also reduces the often unwanted
lifetime extension for objects that are referenced by the stack frames in the
traceback.
As a general principle, a function that calls another function to perform some
task should check whether the called function raised an exception, and if so,
pass the exception state on to its caller. It should discard any object
references that it owns, and return an error indicator, but it should *not* set
another exception --- that would overwrite the exception that was just raised,
and lose important information about the exact cause of the error.
.. index:: single: sum_sequence()
A simple example of detecting exceptions and passing them on is shown in the
:c:func:`sum_sequence` example above. It so happens that this example doesn't
2007-08-15 11:28:22 -03:00
need to clean up any owned references when it detects an error. The following
example function shows some error cleanup. First, to remind you why you like
Python, we show the equivalent Python code::
def incr_item(dict, key):
try:
item = dict[key]
except KeyError:
item = 0
dict[key] = item + 1
.. index:: single: incr_item()
Here is the corresponding C code, in all its glory::
int
incr_item(PyObject *dict, PyObject *key)
{
/* Objects all initialized to NULL for Py_XDECREF */
PyObject *item = NULL, *const_one = NULL, *incremented_item = NULL;
int rv = -1; /* Return value initialized to -1 (failure) */
item = PyObject_GetItem(dict, key);
if (item == NULL) {
/* Handle KeyError only: */
if (!PyErr_ExceptionMatches(PyExc_KeyError))
goto error;
/* Clear the error and use zero: */
PyErr_Clear();
item = PyLong_FromLong(0L);
2007-08-15 11:28:22 -03:00
if (item == NULL)
goto error;
}
const_one = PyLong_FromLong(1L);
2007-08-15 11:28:22 -03:00
if (const_one == NULL)
goto error;
incremented_item = PyNumber_Add(item, const_one);
if (incremented_item == NULL)
goto error;
if (PyObject_SetItem(dict, key, incremented_item) < 0)
goto error;
rv = 0; /* Success */
/* Continue with cleanup code */
error:
/* Cleanup code, shared by success and failure path */
/* Use Py_XDECREF() to ignore NULL references */
Py_XDECREF(item);
Py_XDECREF(const_one);
Py_XDECREF(incremented_item);
return rv; /* -1 for error, 0 for success */
}
.. index:: single: incr_item()
.. index::
single: PyErr_ExceptionMatches()
single: PyErr_Clear()
single: Py_XDECREF()
This example represents an endorsed use of the ``goto`` statement in C!
It illustrates the use of :c:func:`PyErr_ExceptionMatches` and
:c:func:`PyErr_Clear` to handle specific exceptions, and the use of
:c:func:`Py_XDECREF` to dispose of owned references that may be *NULL* (note the
``'X'`` in the name; :c:func:`Py_DECREF` would crash when confronted with a
2007-08-15 11:28:22 -03:00
*NULL* reference). It is important that the variables used to hold owned
references are initialized to *NULL* for this to work; likewise, the proposed
return value is initialized to ``-1`` (failure) and only set to success after
the final call made is successful.
.. _api-embedding:
Embedding Python
================
The one important task that only embedders (as opposed to extension writers) of
the Python interpreter have to worry about is the initialization, and possibly
the finalization, of the Python interpreter. Most functionality of the
interpreter can only be used after the interpreter has been initialized.
.. index::
single: Py_Initialize()
module: builtins
2007-08-15 11:28:22 -03:00
module: __main__
module: sys
triple: module; search; path
single: path (in module sys)
The basic initialization function is :c:func:`Py_Initialize`. This initializes
2007-08-15 11:28:22 -03:00
the table of loaded modules, and creates the fundamental modules
2011-03-26 03:22:01 -03:00
:mod:`builtins`, :mod:`__main__`, and :mod:`sys`. It also
2007-08-15 11:28:22 -03:00
initializes the module search path (``sys.path``).
.. index:: single: PySys_SetArgvEx()
2007-08-15 11:28:22 -03:00
:c:func:`Py_Initialize` does not set the "script argument list" (``sys.argv``).
If this variable is needed by Python code that will be executed later, it must
be set explicitly with a call to ``PySys_SetArgvEx(argc, argv, updatepath)``
after the call to :c:func:`Py_Initialize`.
2007-08-15 11:28:22 -03:00
On most systems (in particular, on Unix and Windows, although the details are
slightly different), :c:func:`Py_Initialize` calculates the module search path
2007-08-15 11:28:22 -03:00
based upon its best guess for the location of the standard Python interpreter
executable, assuming that the Python library is found in a fixed location
relative to the Python interpreter executable. In particular, it looks for a
directory named :file:`lib/python{X.Y}` relative to the parent directory
where the executable named :file:`python` is found on the shell command search
path (the environment variable :envvar:`PATH`).
For instance, if the Python executable is found in
:file:`/usr/local/bin/python`, it will assume that the libraries are in
:file:`/usr/local/lib/python{X.Y}`. (In fact, this particular path is also
the "fallback" location, used when no executable file named :file:`python` is
found along :envvar:`PATH`.) The user can override this behavior by setting the
environment variable :envvar:`PYTHONHOME`, or insert additional directories in
front of the standard path by setting :envvar:`PYTHONPATH`.
.. index::
single: Py_SetProgramName()
single: Py_GetPath()
single: Py_GetPrefix()
single: Py_GetExecPrefix()
single: Py_GetProgramFullPath()
The embedding application can steer the search by calling
``Py_SetProgramName(file)`` *before* calling :c:func:`Py_Initialize`. Note that
2007-08-15 11:28:22 -03:00
:envvar:`PYTHONHOME` still overrides this and :envvar:`PYTHONPATH` is still
inserted in front of the standard path. An application that requires total
control has to provide its own implementation of :c:func:`Py_GetPath`,
:c:func:`Py_GetPrefix`, :c:func:`Py_GetExecPrefix`, and
:c:func:`Py_GetProgramFullPath` (all defined in :file:`Modules/getpath.c`).
2007-08-15 11:28:22 -03:00
.. index:: single: Py_IsInitialized()
Sometimes, it is desirable to "uninitialize" Python. For instance, the
application may want to start over (make another call to
:c:func:`Py_Initialize`) or the application is simply done with its use of
2007-08-15 11:28:22 -03:00
Python and wants to free memory allocated by Python. This can be accomplished
by calling :c:func:`Py_FinalizeEx`. The function :c:func:`Py_IsInitialized` returns
2007-08-15 11:28:22 -03:00
true if Python is currently in the initialized state. More information about
these functions is given in a later chapter. Notice that :c:func:`Py_FinalizeEx`
2007-08-15 11:28:22 -03:00
does *not* free all memory allocated by the Python interpreter, e.g. memory
allocated by extension modules currently cannot be released.
.. _api-debugging:
Debugging Builds
================
Python can be built with several macros to enable extra checks of the
interpreter and extension modules. These checks tend to add a large amount of
overhead to the runtime so they are not enabled by default.
A full list of the various types of debugging builds is in the file
:file:`Misc/SpecialBuilds.txt` in the Python source distribution. Builds are
available that support tracing of reference counts, debugging the memory
allocator, or low-level profiling of the main interpreter loop. Only the most
frequently-used builds will be described in the remainder of this section.
Compiling the interpreter with the :c:macro:`Py_DEBUG` macro defined produces
what is generally meant by "a debug build" of Python. :c:macro:`Py_DEBUG` is
enabled in the Unix build by adding ``--with-pydebug`` to the
:file:`./configure` command. It is also implied by the presence of the
not-Python-specific :c:macro:`_DEBUG` macro. When :c:macro:`Py_DEBUG` is enabled
2007-08-15 11:28:22 -03:00
in the Unix build, compiler optimization is disabled.
In addition to the reference count debugging described below, the following
extra checks are performed:
* Extra checks are added to the object allocator.
* Extra checks are added to the parser and compiler.
* Downcasts from wide types to narrow types are checked for loss of information.
* A number of assertions are added to the dictionary and set implementations.
In addition, the set object acquires a :meth:`test_c_api` method.
* Sanity checks of the input arguments are added to frame creation.
* The storage for ints is initialized with a known invalid pattern to catch
2007-08-15 11:28:22 -03:00
reference to uninitialized digits.
* Low-level tracing and extra exception checking are added to the runtime
virtual machine.
* Extra checks are added to the memory arena implementation.
* Extra debugging is added to the thread module.
There may be additional checks not mentioned here.
Defining :c:macro:`Py_TRACE_REFS` enables reference tracing. When defined, a
2007-08-15 11:28:22 -03:00
circular doubly linked list of active objects is maintained by adding two extra
fields to every :c:type:`PyObject`. Total allocations are tracked as well. Upon
2007-08-15 11:28:22 -03:00
exit, all existing references are printed. (In interactive mode this happens
after every statement run by the interpreter.) Implied by :c:macro:`Py_DEBUG`.
2007-08-15 11:28:22 -03:00
Please refer to :file:`Misc/SpecialBuilds.txt` in the Python source distribution
for more detailed information.