5 Parsing arguments and building values
6 =====================================
8 These functions are useful when creating your own extensions functions and
9 methods. Additional information and examples are available in
10 :ref:`extending-index`.
12 The first three of these functions described, :cfunc:`PyArg_ParseTuple`,
13 :cfunc:`PyArg_ParseTupleAndKeywords`, and :cfunc:`PyArg_Parse`, all use
14 *format strings* which are used to tell the function about the expected
15 arguments. The format strings use the same syntax for each of these
18 A format string consists of zero or more "format units." A format unit
19 describes one Python object; it is usually a single character or a
20 parenthesized sequence of format units. With a few exceptions, a format unit
21 that is not a parenthesized sequence normally corresponds to a single address
22 argument to these functions. In the following description, the quoted form is
23 the format unit; the entry in (round) parentheses is the Python object type
24 that matches the format unit; and the entry in [square] brackets is the type
25 of the C variable(s) whose address should be passed.
27 ``s`` (string or Unicode object) [const char \*]
28 Convert a Python string or Unicode object to a C pointer to a character
29 string. You must not provide storage for the string itself; a pointer to
30 an existing string is stored into the character pointer variable whose
31 address you pass. The C string is NUL-terminated. The Python string must
32 not contain embedded NUL bytes; if it does, a :exc:`TypeError` exception is
33 raised. Unicode objects are converted to C strings using the default
34 encoding. If this conversion fails, a :exc:`UnicodeError` is raised.
36 ``s#`` (string, Unicode or any read buffer compatible object) [const char \*, int (or :ctype:`Py_ssize_t`, see below)]
37 This variant on ``s`` stores into two C variables, the first one a pointer
38 to a character string, the second one its length. In this case the Python
39 string may contain embedded null bytes. Unicode objects pass back a
40 pointer to the default encoded string version of the object if such a
41 conversion is possible. All other read-buffer compatible objects pass back
42 a reference to the raw internal data representation.
44 Starting with Python 2.5 the type of the length argument can be controlled
45 by defining the macro :cmacro:`PY_SSIZE_T_CLEAN` before including
46 :file:`Python.h`. If the macro is defined, length is a :ctype:`Py_ssize_t`
49 ``s*`` (string, Unicode, or any buffer compatible object) [Py_buffer \*]
50 Similar to ``s#``, this code fills a Py_buffer structure provided by the
51 caller. The buffer gets locked, so that the caller can subsequently use
52 the buffer even inside a ``Py_BEGIN_ALLOW_THREADS`` block; the caller is
53 responsible for calling ``PyBuffer_Release`` with the structure after it
54 has processed the data.
58 ``z`` (string or ``None``) [const char \*]
59 Like ``s``, but the Python object may also be ``None``, in which case the C
60 pointer is set to *NULL*.
62 ``z#`` (string or ``None`` or any read buffer compatible object) [const char \*, int]
63 This is to ``s#`` as ``z`` is to ``s``.
65 ``z*`` (string or ``None`` or any buffer compatible object) [Py_buffer*]
66 This is to ``s*`` as ``z`` is to ``s``.
70 ``u`` (Unicode object) [Py_UNICODE \*]
71 Convert a Python Unicode object to a C pointer to a NUL-terminated buffer
72 of 16-bit Unicode (UTF-16) data. As with ``s``, there is no need to
73 provide storage for the Unicode data buffer; a pointer to the existing
74 Unicode data is stored into the :ctype:`Py_UNICODE` pointer variable whose
77 ``u#`` (Unicode object) [Py_UNICODE \*, int]
78 This variant on ``u`` stores into two C variables, the first one a pointer
79 to a Unicode data buffer, the second one its length. Non-Unicode objects
80 are handled by interpreting their read-buffer pointer as pointer to a
81 :ctype:`Py_UNICODE` array.
83 ``es`` (string, Unicode object or character buffer compatible object) [const char \*encoding, char \*\*buffer]
84 This variant on ``s`` is used for encoding Unicode and objects convertible
85 to Unicode into a character buffer. It only works for encoded data without
88 This format requires two arguments. The first is only used as input, and
89 must be a :ctype:`const char\*` which points to the name of an encoding as
90 a NUL-terminated string, or *NULL*, in which case the default encoding is
91 used. An exception is raised if the named encoding is not known to Python.
92 The second argument must be a :ctype:`char\*\*`; the value of the pointer
93 it references will be set to a buffer with the contents of the argument
94 text. The text will be encoded in the encoding specified by the first
97 :cfunc:`PyArg_ParseTuple` will allocate a buffer of the needed size, copy
98 the encoded data into this buffer and adjust *\*buffer* to reference the
99 newly allocated storage. The caller is responsible for calling
100 :cfunc:`PyMem_Free` to free the allocated buffer after use.
102 ``et`` (string, Unicode object or character buffer compatible object) [const char \*encoding, char \*\*buffer]
103 Same as ``es`` except that 8-bit string objects are passed through without
104 recoding them. Instead, the implementation assumes that the string object
105 uses the encoding passed in as parameter.
107 ``es#`` (string, Unicode object or character buffer compatible object) [const char \*encoding, char \*\*buffer, int \*buffer_length]
108 This variant on ``s#`` is used for encoding Unicode and objects convertible
109 to Unicode into a character buffer. Unlike the ``es`` format, this variant
110 allows input data which contains NUL characters.
112 It requires three arguments. The first is only used as input, and must be
113 a :ctype:`const char\*` which points to the name of an encoding as a
114 NUL-terminated string, or *NULL*, in which case the default encoding is
115 used. An exception is raised if the named encoding is not known to Python.
116 The second argument must be a :ctype:`char\*\*`; the value of the pointer
117 it references will be set to a buffer with the contents of the argument
118 text. The text will be encoded in the encoding specified by the first
119 argument. The third argument must be a pointer to an integer; the
120 referenced integer will be set to the number of bytes in the output buffer.
122 There are two modes of operation:
124 If *\*buffer* points a *NULL* pointer, the function will allocate a buffer
125 of the needed size, copy the encoded data into this buffer and set
126 *\*buffer* to reference the newly allocated storage. The caller is
127 responsible for calling :cfunc:`PyMem_Free` to free the allocated buffer
130 If *\*buffer* points to a non-*NULL* pointer (an already allocated buffer),
131 :cfunc:`PyArg_ParseTuple` will use this location as the buffer and
132 interpret the initial value of *\*buffer_length* as the buffer size. It
133 will then copy the encoded data into the buffer and NUL-terminate it. If
134 the buffer is not large enough, a :exc:`ValueError` will be set.
136 In both cases, *\*buffer_length* is set to the length of the encoded data
137 without the trailing NUL byte.
139 ``et#`` (string, Unicode object or character buffer compatible object) [const char \*encoding, char \*\*buffer, int \*buffer_length]
140 Same as ``es#`` except that string objects are passed through without
141 recoding them. Instead, the implementation assumes that the string object
142 uses the encoding passed in as parameter.
144 ``b`` (integer) [unsigned char]
145 Convert a nonnegative Python integer to an unsigned tiny int, stored in a C
146 :ctype:`unsigned char`.
148 ``B`` (integer) [unsigned char]
149 Convert a Python integer to a tiny int without overflow checking, stored in
150 a C :ctype:`unsigned char`.
152 .. versionadded:: 2.3
154 ``h`` (integer) [short int]
155 Convert a Python integer to a C :ctype:`short int`.
157 ``H`` (integer) [unsigned short int]
158 Convert a Python integer to a C :ctype:`unsigned short int`, without
161 .. versionadded:: 2.3
163 ``i`` (integer) [int]
164 Convert a Python integer to a plain C :ctype:`int`.
166 ``I`` (integer) [unsigned int]
167 Convert a Python integer to a C :ctype:`unsigned int`, without overflow
170 .. versionadded:: 2.3
172 ``l`` (integer) [long int]
173 Convert a Python integer to a C :ctype:`long int`.
175 ``k`` (integer) [unsigned long]
176 Convert a Python integer or long integer to a C :ctype:`unsigned long`
177 without overflow checking.
179 .. versionadded:: 2.3
181 ``L`` (integer) [PY_LONG_LONG]
182 Convert a Python integer to a C :ctype:`long long`. This format is only
183 available on platforms that support :ctype:`long long` (or :ctype:`_int64`
186 ``K`` (integer) [unsigned PY_LONG_LONG]
187 Convert a Python integer or long integer to a C :ctype:`unsigned long long`
188 without overflow checking. This format is only available on platforms that
189 support :ctype:`unsigned long long` (or :ctype:`unsigned _int64` on
192 .. versionadded:: 2.3
194 ``n`` (integer) [Py_ssize_t]
195 Convert a Python integer or long integer to a C :ctype:`Py_ssize_t`.
197 .. versionadded:: 2.5
199 ``c`` (string of length 1) [char]
200 Convert a Python character, represented as a string of length 1, to a C
203 ``f`` (float) [float]
204 Convert a Python floating point number to a C :ctype:`float`.
206 ``d`` (float) [double]
207 Convert a Python floating point number to a C :ctype:`double`.
209 ``D`` (complex) [Py_complex]
210 Convert a Python complex number to a C :ctype:`Py_complex` structure.
212 ``O`` (object) [PyObject \*]
213 Store a Python object (without any conversion) in a C object pointer. The
214 C program thus receives the actual object that was passed. The object's
215 reference count is not increased. The pointer stored is not *NULL*.
217 ``O!`` (object) [*typeobject*, PyObject \*]
218 Store a Python object in a C object pointer. This is similar to ``O``, but
219 takes two C arguments: the first is the address of a Python type object,
220 the second is the address of the C variable (of type :ctype:`PyObject\*`)
221 into which the object pointer is stored. If the Python object does not
222 have the required type, :exc:`TypeError` is raised.
224 ``O&`` (object) [*converter*, *anything*]
225 Convert a Python object to a C variable through a *converter* function.
226 This takes two arguments: the first is a function, the second is the
227 address of a C variable (of arbitrary type), converted to :ctype:`void \*`.
228 The *converter* function in turn is called as follows::
230 status = converter(object, address);
232 where *object* is the Python object to be converted and *address* is the
233 :ctype:`void\*` argument that was passed to the :cfunc:`PyArg_Parse\*`
234 function. The returned *status* should be ``1`` for a successful
235 conversion and ``0`` if the conversion has failed. When the conversion
236 fails, the *converter* function should raise an exception and leave the
237 content of *address* unmodified.
239 ``S`` (string) [PyStringObject \*]
240 Like ``O`` but requires that the Python object is a string object. Raises
241 :exc:`TypeError` if the object is not a string object. The C variable may
242 also be declared as :ctype:`PyObject\*`.
244 ``U`` (Unicode string) [PyUnicodeObject \*]
245 Like ``O`` but requires that the Python object is a Unicode object. Raises
246 :exc:`TypeError` if the object is not a Unicode object. The C variable may
247 also be declared as :ctype:`PyObject\*`.
249 ``t#`` (read-only character buffer) [char \*, int]
250 Like ``s#``, but accepts any object which implements the read-only buffer
251 interface. The :ctype:`char\*` variable is set to point to the first byte
252 of the buffer, and the :ctype:`int` is set to the length of the buffer.
253 Only single-segment buffer objects are accepted; :exc:`TypeError` is raised
256 ``w`` (read-write character buffer) [char \*]
257 Similar to ``s``, but accepts any object which implements the read-write
258 buffer interface. The caller must determine the length of the buffer by
259 other means, or use ``w#`` instead. Only single-segment buffer objects are
260 accepted; :exc:`TypeError` is raised for all others.
262 ``w#`` (read-write character buffer) [char \*, Py_ssize_t]
263 Like ``s#``, but accepts any object which implements the read-write buffer
264 interface. The :ctype:`char \*` variable is set to point to the first byte
265 of the buffer, and the :ctype:`Py_ssize_t` is set to the length of the
266 buffer. Only single-segment buffer objects are accepted; :exc:`TypeError`
267 is raised for all others.
269 ``w*`` (read-write byte-oriented buffer) [Py_buffer \*]
270 This is to ``w`` what ``s*`` is to ``s``.
272 .. versionadded:: 2.6
274 ``(items)`` (tuple) [*matching-items*]
275 The object must be a Python sequence whose length is the number of format
276 units in *items*. The C arguments must correspond to the individual format
277 units in *items*. Format units for sequences may be nested.
281 Prior to Python version 1.5.2, this format specifier only accepted a
282 tuple containing the individual parameters, not an arbitrary sequence.
283 Code which previously caused :exc:`TypeError` to be raised here may now
284 proceed without an exception. This is not expected to be a problem for
287 It is possible to pass Python long integers where integers are requested;
288 however no proper range checking is done --- the most significant bits are
289 silently truncated when the receiving field is too small to receive the value
290 (actually, the semantics are inherited from downcasts in C --- your mileage
293 A few other characters have a meaning in a format string. These may not occur
294 inside nested parentheses. They are:
297 Indicates that the remaining arguments in the Python argument list are
298 optional. The C variables corresponding to optional arguments should be
299 initialized to their default value --- when an optional argument is not
300 specified, :cfunc:`PyArg_ParseTuple` does not touch the contents of the
301 corresponding C variable(s).
304 The list of format units ends here; the string after the colon is used as
305 the function name in error messages (the "associated value" of the
306 exception that :cfunc:`PyArg_ParseTuple` raises).
309 The list of format units ends here; the string after the semicolon is used
310 as the error message *instead* of the default error message. ``:`` and
311 ``;`` mutually exclude each other.
313 Note that any Python object references which are provided to the caller are
314 *borrowed* references; do not decrement their reference count!
316 Additional arguments passed to these functions must be addresses of variables
317 whose type is determined by the format string; these are used to store values
318 from the input tuple. There are a few cases, as described in the list of
319 format units above, where these parameters are used as input values; they
320 should match what is specified for the corresponding format unit in that case.
322 For the conversion to succeed, the *arg* object must match the format and the
323 format must be exhausted. On success, the :cfunc:`PyArg_Parse\*` functions
324 return true, otherwise they return false and raise an appropriate exception.
325 When the :cfunc:`PyArg_Parse\*` functions fail due to conversion failure in
326 one of the format units, the variables at the addresses corresponding to that
327 and the following format units are left untouched.
330 .. cfunction:: int PyArg_ParseTuple(PyObject *args, const char *format, ...)
332 Parse the parameters of a function that takes only positional parameters
333 into local variables. Returns true on success; on failure, it returns
334 false and raises the appropriate exception.
337 .. cfunction:: int PyArg_VaParse(PyObject *args, const char *format, va_list vargs)
339 Identical to :cfunc:`PyArg_ParseTuple`, except that it accepts a va_list
340 rather than a variable number of arguments.
343 .. cfunction:: int PyArg_ParseTupleAndKeywords(PyObject *args, PyObject *kw, const char *format, char *keywords[], ...)
345 Parse the parameters of a function that takes both positional and keyword
346 parameters into local variables. Returns true on success; on failure, it
347 returns false and raises the appropriate exception.
350 .. cfunction:: int PyArg_VaParseTupleAndKeywords(PyObject *args, PyObject *kw, const char *format, char *keywords[], va_list vargs)
352 Identical to :cfunc:`PyArg_ParseTupleAndKeywords`, except that it accepts a
353 va_list rather than a variable number of arguments.
356 .. cfunction:: int PyArg_Parse(PyObject *args, const char *format, ...)
358 Function used to deconstruct the argument lists of "old-style" functions
359 --- these are functions which use the :const:`METH_OLDARGS` parameter
360 parsing method. This is not recommended for use in parameter parsing in
361 new code, and most code in the standard interpreter has been modified to no
362 longer use this for that purpose. It does remain a convenient way to
363 decompose other tuples, however, and may continue to be used for that
367 .. cfunction:: int PyArg_UnpackTuple(PyObject *args, const char *name, Py_ssize_t min, Py_ssize_t max, ...)
369 A simpler form of parameter retrieval which does not use a format string to
370 specify the types of the arguments. Functions which use this method to
371 retrieve their parameters should be declared as :const:`METH_VARARGS` in
372 function or method tables. The tuple containing the actual parameters
373 should be passed as *args*; it must actually be a tuple. The length of the
374 tuple must be at least *min* and no more than *max*; *min* and *max* may be
375 equal. Additional arguments must be passed to the function, each of which
376 should be a pointer to a :ctype:`PyObject\*` variable; these will be filled
377 in with the values from *args*; they will contain borrowed references. The
378 variables which correspond to optional parameters not given by *args* will
379 not be filled in; these should be initialized by the caller. This function
380 returns true on success and false if *args* is not a tuple or contains the
381 wrong number of elements; an exception will be set if there was a failure.
383 This is an example of the use of this function, taken from the sources for
384 the :mod:`_weakref` helper module for weak references::
387 weakref_ref(PyObject *self, PyObject *args)
390 PyObject *callback = NULL;
391 PyObject *result = NULL;
393 if (PyArg_UnpackTuple(args, "ref", 1, 2, &object, &callback)) {
394 result = PyWeakref_NewRef(object, callback);
399 The call to :cfunc:`PyArg_UnpackTuple` in this example is entirely
400 equivalent to this call to :cfunc:`PyArg_ParseTuple`::
402 PyArg_ParseTuple(args, "O|O:ref", &object, &callback)
404 .. versionadded:: 2.2
406 .. versionchanged:: 2.5
407 This function used an :ctype:`int` type for *min* and *max*. This might
408 require changes in your code for properly supporting 64-bit systems.
411 .. cfunction:: PyObject* Py_BuildValue(const char *format, ...)
413 Create a new value based on a format string similar to those accepted by
414 the :cfunc:`PyArg_Parse\*` family of functions and a sequence of values.
415 Returns the value or *NULL* in the case of an error; an exception will be
416 raised if *NULL* is returned.
418 :cfunc:`Py_BuildValue` does not always build a tuple. It builds a tuple
419 only if its format string contains two or more format units. If the format
420 string is empty, it returns ``None``; if it contains exactly one format
421 unit, it returns whatever object is described by that format unit. To
422 force it to return a tuple of size 0 or one, parenthesize the format
425 When memory buffers are passed as parameters to supply data to build
426 objects, as for the ``s`` and ``s#`` formats, the required data is copied.
427 Buffers provided by the caller are never referenced by the objects created
428 by :cfunc:`Py_BuildValue`. In other words, if your code invokes
429 :cfunc:`malloc` and passes the allocated memory to :cfunc:`Py_BuildValue`,
430 your code is responsible for calling :cfunc:`free` for that memory once
431 :cfunc:`Py_BuildValue` returns.
433 In the following description, the quoted form is the format unit; the entry
434 in (round) parentheses is the Python object type that the format unit will
435 return; and the entry in [square] brackets is the type of the C value(s) to
438 The characters space, tab, colon and comma are ignored in format strings
439 (but not within format units such as ``s#``). This can be used to make
440 long format strings a tad more readable.
442 ``s`` (string) [char \*]
443 Convert a null-terminated C string to a Python object. If the C string
444 pointer is *NULL*, ``None`` is used.
446 ``s#`` (string) [char \*, int]
447 Convert a C string and its length to a Python object. If the C string
448 pointer is *NULL*, the length is ignored and ``None`` is returned.
450 ``z`` (string or ``None``) [char \*]
453 ``z#`` (string or ``None``) [char \*, int]
456 ``u`` (Unicode string) [Py_UNICODE \*]
457 Convert a null-terminated buffer of Unicode (UCS-2 or UCS-4) data to a
458 Python Unicode object. If the Unicode buffer pointer is *NULL*,
459 ``None`` is returned.
461 ``u#`` (Unicode string) [Py_UNICODE \*, int]
462 Convert a Unicode (UCS-2 or UCS-4) data buffer and its length to a
463 Python Unicode object. If the Unicode buffer pointer is *NULL*, the
464 length is ignored and ``None`` is returned.
466 ``i`` (integer) [int]
467 Convert a plain C :ctype:`int` to a Python integer object.
469 ``b`` (integer) [char]
470 Convert a plain C :ctype:`char` to a Python integer object.
472 ``h`` (integer) [short int]
473 Convert a plain C :ctype:`short int` to a Python integer object.
475 ``l`` (integer) [long int]
476 Convert a C :ctype:`long int` to a Python integer object.
478 ``B`` (integer) [unsigned char]
479 Convert a C :ctype:`unsigned char` to a Python integer object.
481 ``H`` (integer) [unsigned short int]
482 Convert a C :ctype:`unsigned short int` to a Python integer object.
484 ``I`` (integer/long) [unsigned int]
485 Convert a C :ctype:`unsigned int` to a Python integer object or a Python
486 long integer object, if it is larger than ``sys.maxint``.
488 ``k`` (integer/long) [unsigned long]
489 Convert a C :ctype:`unsigned long` to a Python integer object or a
490 Python long integer object, if it is larger than ``sys.maxint``.
492 ``L`` (long) [PY_LONG_LONG]
493 Convert a C :ctype:`long long` to a Python long integer object. Only
494 available on platforms that support :ctype:`long long`.
496 ``K`` (long) [unsigned PY_LONG_LONG]
497 Convert a C :ctype:`unsigned long long` to a Python long integer object.
498 Only available on platforms that support :ctype:`unsigned long long`.
500 ``n`` (int) [Py_ssize_t]
501 Convert a C :ctype:`Py_ssize_t` to a Python integer or long integer.
503 .. versionadded:: 2.5
505 ``c`` (string of length 1) [char]
506 Convert a C :ctype:`int` representing a character to a Python string of
509 ``d`` (float) [double]
510 Convert a C :ctype:`double` to a Python floating point number.
512 ``f`` (float) [float]
515 ``D`` (complex) [Py_complex \*]
516 Convert a C :ctype:`Py_complex` structure to a Python complex number.
518 ``O`` (object) [PyObject \*]
519 Pass a Python object untouched (except for its reference count, which is
520 incremented by one). If the object passed in is a *NULL* pointer, it is
521 assumed that this was caused because the call producing the argument
522 found an error and set an exception. Therefore, :cfunc:`Py_BuildValue`
523 will return *NULL* but won't raise an exception. If no exception has
524 been raised yet, :exc:`SystemError` is set.
526 ``S`` (object) [PyObject \*]
529 ``N`` (object) [PyObject \*]
530 Same as ``O``, except it doesn't increment the reference count on the
531 object. Useful when the object is created by a call to an object
532 constructor in the argument list.
534 ``O&`` (object) [*converter*, *anything*]
535 Convert *anything* to a Python object through a *converter* function.
536 The function is called with *anything* (which should be compatible with
537 :ctype:`void \*`) as its argument and should return a "new" Python
538 object, or *NULL* if an error occurred.
540 ``(items)`` (tuple) [*matching-items*]
541 Convert a sequence of C values to a Python tuple with the same number of
544 ``[items]`` (list) [*matching-items*]
545 Convert a sequence of C values to a Python list with the same number of
548 ``{items}`` (dictionary) [*matching-items*]
549 Convert a sequence of C values to a Python dictionary. Each pair of
550 consecutive C values adds one item to the dictionary, serving as key and
553 If there is an error in the format string, the :exc:`SystemError` exception
554 is set and *NULL* returned.
556 .. cfunction:: PyObject* Py_VaBuildValue(const char *format, va_list vargs)
558 Identical to :cfunc:`Py_BuildValue`, except that it accepts a va_list
559 rather than a variable number of arguments.