Guide Argument Clinic


Larry Hastings

Source code: Tools/clinic/


Argument Clinic is a preprocessor for CPython C files. It was introduced in Python 3.4 with PEP 436, in order to provide introspection signatures, and to generate performant and tailor-made boilerplate code for argument parsing in CPython builtins, module level functions, and class methods. This document is divided in four major sections:

  • Background talks about the basic concepts and goals of Argument Clinic.

  • Reference describes the command-line interface and Argument Clinic terminology.

  • Tutorial guides you through all the steps required to adapt an existing C function to Argument Clinic.

  • How-to guides details how to handle specific tasks.


Argument Clinic is considered internal-only for CPython. Its use is not supported for files outside CPython, and no guarantees are made regarding backwards compatibility for future versions. In other words: if you maintain an external C extension for CPython, you're welcome to experiment with Argument Clinic in your own code. But the version of Argument Clinic that ships with the next version of CPython could be totally incompatible and break all your code.


Basic concepts

When Argument Clinic is run on a file, either via the Command-line interface or via make clinic, it will scan over the input files looking for start lines:

/*[clinic input]

When it finds one, it reads everything up to the end line:

[clinic start generated code]*/

Everything in between these two lines is Argument Clinic input. When Argument Clinic parses input, it generates output. The output is rewritten into the C file immediately after the input, followed by a checksum line. All of these lines, including the start line and checksum line, are collectively called an Argument Clinic block:

/*[clinic input]
... clinic input goes here ...
[clinic start generated code]*/
... clinic output goes here ...
/*[clinic end generated code: ...]*/

If you run Argument Clinic on the same file a second time, Argument Clinic will discard the old output and write out the new output with a fresh checksum line. If the input hasn't changed, the output won't change either.


You should never modify the output of an Argument Clinic block, as any change will be lost in future Argument Clinic runs; Argument Clinic will detect an output checksum mismatch and regenerate the correct output. If you are not happy with the generated output, you should instead change the input until it produces the output you want.



start line

The line /*[clinic input]. This line marks the beginning of Argument Clinic input. Note that the start line opens a C block comment.

end line

The line [clinic start generated code]*/. The end line marks the _end_ of Argument Clinic input, but at the same time marks the _start_ of Argument Clinic output, thus the text "clinic start start generated code" Note that the end line closes the C block comment opened by the start line.


A hash to distinguish unique inputs and outputs.

checksum line

A line that looks like /*[clinic end generated code: ...]*/. The three dots will be replaced by a checksum generated from the input, and a checksum generated from the output. The checksum line marks the end of Argument Clinic generated code, and is used by Argument Clinic to determine if it needs to regenerate output.


The text between the start line and the end line. Note that the start and end lines open and close a C block comment; the input is thus a part of that same C block comment.


The text between the end line and the checksum line.


All text from the start line to the checksum line inclusively.

Command-line interface

The Argument Clinic CLI is typically used to process a single source file, like this:

$ python3 ./Tools/clinic/ foo.c

The CLI supports the following options:

-h, --help

Print CLI usage.

-f, --force

Force output regeneration.

-o, --output OUTPUT

Redirect file output to OUTPUT

-v, --verbose

Enable verbose mode.


Print a list of all supported converters and return converters.


Walk --srcdir to run over all relevant files.

--srcdir SRCDIR

The directory tree to walk in --make mode.

FILE ...

The list of files to process.

Classes for extending Argument Clinic

class clinic.CConverter

The base class for all converters. See How to write a custom converter for how to subclass this class.


The C type to use for this variable. type should be a Python string specifying the type, e.g. 'int'. If this is a pointer type, the type string should end with ' *'.


The Python default value for this parameter, as a Python value. Or the magic value unspecified if there is no default.


default as it should appear in Python code, as a string. Or None if there is no default.


default as it should appear in C code, as a string. Or None if there is no default.


The default value used to initialize the C variable when there is no default, but not specifying a default may result in an "uninitialized variable" warning. This can easily happen when using option groups—although properly written code will never actually use this value, the variable does get passed in to the impl, and the C compiler will complain about the "use" of the uninitialized value. This value should always be a non-empty string.


The name of the C converter function, as a string.


A boolean value. If true, Argument Clinic will add a & in front of the name of the variable when passing it into the impl function.


A boolean value. If true, Argument Clinic will add a & in front of the name of the variable when passing it into PyArg_ParseTuple().


The best way to get a sense of how Argument Clinic works is to convert a function to work with it. Here, then, are the bare minimum steps you'd need to follow to convert a function to work with Argument Clinic. Note that for code you plan to check in to CPython, you really should take the conversion farther, using some of the advanced concepts you'll see later on in the document, like How to use return converters and How to use the "self converter". But we'll keep it simple for this walkthrough so you can learn.

First, make sure you're working with a freshly updated checkout of the CPython trunk.

Next, find a Python builtin that calls either PyArg_ParseTuple() or PyArg_ParseTupleAndKeywords(), and hasn't been converted to work with Argument Clinic yet. For this tutorial, we'll be using _pickle.Pickler.dump.

If the call to the PyArg_Parse*() function uses any of the following format units...:


... or if it has multiple calls to PyArg_ParseTuple(), you should choose a different function. (See How to use advanced converters for those scenarios.)

Also, if the function has multiple calls to PyArg_ParseTuple() or PyArg_ParseTupleAndKeywords() where it supports different types for the same argument, or if the function uses something besides PyArg_Parse*() functions to parse its arguments, it probably isn't suitable for conversion to Argument Clinic. Argument Clinic doesn't support generic functions or polymorphic parameters.

Next, add the following boilerplate above the function, creating our input block:

/*[clinic input]
[clinic start generated code]*/

Cut the docstring and paste it in between the [clinic] lines, removing all the junk that makes it a properly quoted C string. When you're done you should have just the text, based at the left margin, with no line wider than 80 characters. Argument Clinic will preserve indents inside the docstring.

If the old docstring had a first line that looked like a function signature, throw that line away; The docstring doesn't need it anymore --- when you use help() on your builtin in the future, the first line will be built automatically based on the function's signature.

Example docstring summary line:

/*[clinic input]
Write a pickled representation of obj to the open file.
[clinic start generated code]*/

If your docstring doesn't have a "summary" line, Argument Clinic will complain, so let's make sure it has one. The "summary" line should be a paragraph consisting of a single 80-column line at the beginning of the docstring. (See PEP 257 regarding docstring conventions.)

Our example docstring consists solely of a summary line, so the sample code doesn't have to change for this step.

Now, above the docstring, enter the name of the function, followed by a blank line. This should be the Python name of the function, and should be the full dotted path to the function --- it should start with the name of the module, include any sub-modules, and if the function is a method on a class it should include the class name too.

In our example, _pickle is the module, Pickler is the class, and dump() is the method, so the name becomes _pickle.Pickler.dump():

/*[clinic input]

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

If this is the first time that module or class has been used with Argument Clinic in this C file, you must declare the module and/or class. Proper Argument Clinic hygiene prefers declaring these in a separate block somewhere near the top of the C file, in the same way that include files and statics go at the top. In our sample code we'll just show the two blocks next to each other.

Le nom de la classe et du module doivent être les mêmes que ceux vus par Python. Selon le cas, référez-vous à PyModuleDef ou PyTypeObject

When you declare a class, you must also specify two aspects of its type in C: the type declaration you'd use for a pointer to an instance of this class, and a pointer to the PyTypeObject for this class:

/*[clinic input]
module _pickle
class _pickle.Pickler "PicklerObject *" "&Pickler_Type"
[clinic start generated code]*/

/*[clinic input]

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

Declare each of the parameters to the function. Each parameter should get its own line. All the parameter lines should be indented from the function name and the docstring. The general form of these parameter lines is as follows:

name_of_parameter: converter

Si le paramètre a une valeur par défaut, ajoutez ceci après le convertisseur :

name_of_parameter: converter = default_value

Argument Clinic's support for "default values" is quite sophisticated; see How to assign default values to parameter for more information.

Next, add a blank line below the parameters.

What's a "converter"? It establishes both the type of the variable used in C, and the method to convert the Python value into a C value at runtime. For now you're going to use what's called a "legacy converter" --- a convenience syntax intended to make porting old code into Argument Clinic easier.

For each parameter, copy the "format unit" for that parameter from the PyArg_Parse() format argument and specify that as its converter, as a quoted string. The "format unit" is the formal name for the one-to-three character substring of the format parameter that tells the argument parsing function what the type of the variable is and how to convert it. For more on format units please see Analyse des arguments et construction des valeurs.

Pour des spécifications de format de plusieurs caractères, comme z#, utilisez l'ensemble des 2 ou 3 caractères de la chaîne.

Échantillon :

/*[clinic input]
module _pickle
class _pickle.Pickler "PicklerObject *" "&Pickler_Type"
[clinic start generated code]*/

/*[clinic input]

    obj: 'O'

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

If your function has | in the format string, meaning some parameters have default values, you can ignore it. Argument Clinic infers which parameters are optional based on whether or not they have default values.

Si votre fonction a le caractère $ dans son format, parce qu'elle n'accepte que des arguments nommés, spécifiez * sur une ligne à part, avant le premier argument nommé, avec la même indentation que les lignes de paramètres.

_pickle.Pickler.dump() has neither, so our sample is unchanged.

Next, if the existing C function calls PyArg_ParseTuple() (as opposed to PyArg_ParseTupleAndKeywords()), then all its arguments are positional-only.

To mark parameters as positional-only in Argument Clinic, add a / on a line by itself after the last positional-only parameter, indented the same as the parameter lines.

Échantillon :

/*[clinic input]
module _pickle
class _pickle.Pickler "PicklerObject *" "&Pickler_Type"
[clinic start generated code]*/

/*[clinic input]

    obj: 'O'

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

It can be helpful to write a per-parameter docstring for each parameter. Since per-parameter docstrings are optional, you can skip this step if you prefer.

Nevertheless, here's how to add a per-parameter docstring. The first line of the per-parameter docstring must be indented further than the parameter definition. The left margin of this first line establishes the left margin for the whole per-parameter docstring; all the text you write will be outdented by this amount. You can write as much text as you like, across multiple lines if you wish.

Échantillon :

/*[clinic input]
module _pickle
class _pickle.Pickler "PicklerObject *" "&Pickler_Type"
[clinic start generated code]*/

/*[clinic input]

    obj: 'O'
        The object to be pickled.

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

Save and close the file, then run Tools/clinic/ on it. With luck everything worked---your block now has output, and a .c.h file has been generated! Reload the file in your text editor to see the generated code:

/*[clinic input]

    obj: 'O'
        The object to be pickled.

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

static PyObject *
_pickle_Pickler_dump(PicklerObject *self, PyObject *obj)
/*[clinic end generated code: output=87ecad1261e02ac7 input=552eb1c0f52260d9]*/

Obviously, if Argument Clinic didn't produce any output, it's because it found an error in your input. Keep fixing your errors and retrying until Argument Clinic processes your file without complaint.

For readability, most of the glue code has been generated to a .c.h file. You'll need to include that in your original .c file, typically right after the clinic module block:

#include "clinic/_pickle.c.h"

Vérifiez bien que le code d'analyse d'arguments généré par Argument Clinic ressemble bien au code existant.

Assurez vous premièrement que les deux codes utilisent la même fonction pour analyser les arguments. Le code existant doit appeler soit PyArg_ParseTuple() soit PyArg_ParseTupleAndKeywords() ; assurez vous que le code généré par Argument Clinic appelle exactement la même fonction.

Second, the format string passed in to PyArg_ParseTuple() or PyArg_ParseTupleAndKeywords() should be exactly the same as the hand-written one in the existing function, up to the colon or semi-colon.

Argument Clinic always generates its format strings with a : followed by the name of the function. If the existing code's format string ends with ;, to provide usage help, this change is harmless --- don't worry about it.

Third, for parameters whose format units require two arguments, like a length variable, an encoding string, or a pointer to a conversion function, ensure that the second argument is exactly the same between the two invocations.

Fourth, inside the output portion of the block, you'll find a preprocessor macro defining the appropriate static PyMethodDef structure for this builtin:

{"dump", (PyCFunction)__pickle_Pickler_dump, METH_O, __pickle_Pickler_dump__doc__},

This static structure should be exactly the same as the existing static PyMethodDef structure for this builtin.

Si l'un de ces éléments diffère de quelque façon que se soit, ajustez la spécification de fonction d'Argument Clinic et exécutez de nouveau Tools/clinic/ jusqu'à ce qu'elles soient identiques.

Notez que la dernière ligne de cette sortie est la déclaration de votre fonction impl. C'est là que se trouve l'implémentation de la fonction native. Supprimez le prototype de la fonction que vous modifiez, mais laissez l'accolade ouverte. Maintenant, supprimez tout le code d'analyse d'arguments et les déclarations de toutes les variables auxquelles il assigne les arguments. Vous voyez que désormais les arguments Python sont ceux de cette fonction impl ; si l'implémentation utilise des noms différents pour ces variables, corrigez-les.

Let's reiterate, just because it's kind of weird. Your code should now look like this:

static return_type
/*[clinic end generated code: input=..., output=...]*/

Argument Clinic generated the checksum line and the function prototype just above it. You should write the opening and closing curly braces for the function, and the implementation inside.

Échantillon :

/*[clinic input]
module _pickle
class _pickle.Pickler "PicklerObject *" "&Pickler_Type"
[clinic start generated code]*/
/*[clinic end generated code: checksum=da39a3ee5e6b4b0d3255bfef95601890afd80709]*/

/*[clinic input]

    obj: 'O'
        The object to be pickled.

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

"Write a pickled representation of obj to the open file.\n"
static PyObject *
_pickle_Pickler_dump_impl(PicklerObject *self, PyObject *obj)
/*[clinic end generated code: checksum=3bd30745bf206a48f8b576a1da3d90f55a0a4187]*/
    /* Check whether the Pickler was initialized correctly (issue3664).
       Developers often forget to call __init__() in their subclasses, which
       would trigger a segfault without this check. */
    if (self->write == NULL) {
                     "Pickler.__init__() was not called by %s.__init__()",
        return NULL;

    if (_Pickler_ClearBuffer(self) < 0) {
        return NULL;


Remember the macro with the PyMethodDef structure for this function? Find the existing PyMethodDef structure for this function and replace it with a reference to the macro. If the builtin is at module scope, this will probably be very near the end of the file; if the builtin is a class method, this will probably be below but relatively near to the implementation.

Note that the body of the macro contains a trailing comma; when you replace the existing static PyMethodDef structure with the macro, don't add a comma to the end.

Échantillon :

static struct PyMethodDef Pickler_methods[] = {
    {NULL, NULL}                /* sentinel */

Finally, compile, then run the relevant portions of the regression-test suite. This change should not introduce any new compile-time warnings or errors, and there should be no externally visible change to Python's behavior, except for one difference: inspect.signature() run on your function should now provide a valid signature!

Félicitations, vous avez adapté votre première fonction pour qu'elle utilise Argument Clinic !

How-to guides

How to rename C functions and variables generated by Argument Clinic

Argument Clinic nomme automatiquement les fonctions qu'il génère. Parfois, cela peut poser des problèmes, si le nom généré entre en collision avec le nom d'une fonction C existante. Il y a une solution simple : surcharger les noms utilisés par les fonctions C. Ajoutez simplement le mot clef "as" sur la ligne de la déclaration de la fonction, suivi par le nom de la fonction que vous souhaitez utiliser. Argument Clinic utilisera ce nom de fonction pour la fonction de base (celle générée), et ajoutera "_impl" à la fin et utilisera ce nom pour la fonction impl.

For example, if we wanted to rename the C function names generated for pickle.Pickler.dump(), it'd look like this:

/*[clinic input]
pickle.Pickler.dump as pickler_dumper


The base function would now be named pickler_dumper(), and the impl function would now be named pickler_dumper_impl().

De même, vous pouvez avoir un problème quand vous souhaiterez donner à un paramètre un nom spécifique à Python, mais ce nom peut être gênant en C. Argument Clinic vous permet de donner à un paramètre des noms différents en Python et en C :

/*[clinic input]

    obj: object
    file as file_obj: object
    protocol: object = NULL
    fix_imports: bool = True

Here, the name used in Python (in the signature and the keywords array) would be file, but the C variable would be named file_obj.

You can use this to rename the self parameter too!

How to convert functions using PyArg_UnpackTuple

To convert a function parsing its arguments with PyArg_UnpackTuple(), simply write out all the arguments, specifying each as an object. You may specify the type argument to cast the type as appropriate. All arguments should be marked positional-only (add a / on a line by itself after the last argument).

Actuellement, le code généré utilise PyArg_ParseTuple(), mais cela va bientôt changer.

How to use optional groups

Certaines fonctions de base ont une approche particulière pour analyser leurs arguments : elles comptent le nombre d'arguments positionnels, puis elles utilisent une condition switch basée sur le nombre d'arguments présents pour appeler différentes PyArg_ParseTuple() disponibles (ces fonctions ne peuvent pas avoir des arguments passés uniquement en tant qu'arguments nommés). Cette approche était utilisée pour simuler des arguments optionnels avant que PyArg_ParseTupleAndKeywords() ne soit créée.

While functions using this approach can often be converted to use PyArg_ParseTupleAndKeywords(), optional arguments, and default values, it's not always possible. Some of these legacy functions have behaviors PyArg_ParseTupleAndKeywords() doesn't directly support. The most obvious example is the builtin function range(), which has an optional argument on the left side of its required argument! Another example is curses.window.addch(), which has a group of two arguments that must always be specified together. (The arguments are called x and y; if you call the function passing in x, you must also pass in y — and if you don't pass in x you may not pass in y either.)

Dans tous les cas, le but d'Argument Clinic est de prendre en charge l'analyse des arguments pour toutes les fonctions natives de CPython sans avoir besoin de les modifier. C'est pourquoi Argument Clinic propose cette autre approche pour l'analyse, en utilisant ce qu'on appelle les groupes optionnels. Les groupes optionnels sont des groupes d'arguments qui doivent tous être transmis ensemble. Ils peuvent être situés à droite ou à gauche des arguments requis. Ils ne peuvent être utilisés seulement qu'en tant que paramètres positionnels.


Les groupes optionnels sont uniquement prévus pour convertir les fonctions faisant des appels multiples à PyArg_ParseTuple() ! Les fonctions qui utilisent au moins une des autres approches ne doivent presque jamais être converties à Argument Clinic en utilisant les groupes optionnels. Les fonctions utilisant ces groupes n'ont pas actuellement de signature précise en Python, parce que celui-ci ne peut simplement pas comprendre ce concept. Tâchez d'éviter au maximum d'utiliser ces groupes optionnels si possible.

To specify an optional group, add a [ on a line by itself before the parameters you wish to group together, and a ] on a line by itself after these parameters. As an example, here's how curses.window.addch() uses optional groups to make the first two parameters and the last parameter optional:

/*[clinic input]


    x: int
    y: int

    ch: object
      Character to add.

    attr: long
      Attributes for the character.


Notes :

  • Pour chaque groupe optionnel, un paramètre additionnel sera passé à la fonction impl représentant le groupe. Ce paramètre sera un entier nommé group_{direction}_{number}, où {direction} peut être soit right ou left suivant que le groupe est situé avant ou après les paramètres requis, et {number} sera un entier incrémenté (débutant à 1) indiquant la distance entre le groupe et les paramètres requis. Quand la fonction impl est appelée, ce paramètre est positionné à zéro si le groupe n'a pas été utilisé, et positionné à un nombre entier positif sinon (par inutilisé, on entend que les paramètres n'ont pas reçu de valeur lors de cet appel).

  • S'il n'y a pas d'arguments requis, les groupes optionnels se comportent comme s'ils étaient à droite des arguments requis.

  • En cas d'ambiguïté, le code d'analyse des arguments favorise ceux situés à gauche (avant les paramètres obligatoires).

  • Les groupes optionnels ne peuvent contenir que des arguments positionnels.

  • Les groupes optionnels sont seulement destinés au code hérité. Ne les utilisez pas dans du nouveau code.

How to use real Argument Clinic converters, instead of "legacy converters"

Afin de gagner du temps, et pour minimiser la courbe d'apprentissage pour pouvoir utiliser Argument Clinic, le guide ci-dessus préconise les « adaptateurs de base ». Ceux-ci sont un moyen simple conçu pour porter facilement du code existant sous Argument Clinic. Et pour être clair, leur utilisation est tout à fait acceptable pour porter du code Python 3.4.

Cependant, sur le long terme, il est certainement préférable que tous vos blocs utilisent la syntaxe réelle des adaptateurs d'Argument Clinic. Pourquoi ? Voici quelques raisons :

  • Les adaptateurs sont plus simples et plus clairs.

  • Il existe des formats qui ne sont pas gérés par les « adaptateurs de base », parce qu'ils nécessitent des arguments, et la syntaxe de ces adaptateurs ne supporte pas cela.

  • Dans le futur, on pourrait avoir une nouvelle bibliothèque d'analyse des arguments qui ne serait pas limitée à ce que PyArg_ParseTuple() accepte ; cette flexibilité ne serait pas accessible aux paramètres utilisant des adaptateurs de base.

Ainsi, si vous n'êtes pas contre un petit effort supplémentaire, vous devriez utiliser les adaptateurs normaux plutôt que ceux de base.

En bref, la syntaxe des adaptateurs d'Argument Clinic ressemble à un appel de fonction Python. Mais, s'il n'y a pas d'argument explicite à la fonction (celle-ci utilisant ses valeurs par défaut), vous pouvez omettre les parenthèses. Ainsi bool et bool() représentent le même adaptateur.

Tous les arguments passés aux adaptateurs d'Argument Clinic sont nommés. Tous les adaptateurs d'Argument Clinic acceptent les arguments suivants :


La valeur par défaut de cet argument lorsqu'il est défini en C. Typiquement, il servira à initialiser la variable déclarée dans la « fonction d'analyse ». Voir la section relative aux valeurs par défaut pour apprendre à l'utiliser. Spécifié en tant que chaîne de caractères.


La valeur annotée pour ce paramètre. Actuellement non géré, car la PEP 8 exige que les bibliothèques Python n'utilisent pas d'annotations.

De plus, certains adaptateurs acceptent des arguments additionnels. Voici la liste de ces arguments, avec leur explication :


Un ensemble de types Python (et potentiellement des pseudo-types) ; cela restreint l'argument Python autorisé aux valeurs de ces types (ce n'est pas destiné à une utilisation généralisée ; en fait, il gère seulement les types listés dans la table des adaptateurs de base).

Pour accepter None, ajouter NoneType à cet ensemble.


Autorisé seulement pour les entiers non signés. La valeur native de cet argument Python sera transcrite dans le paramètre sans aucune vérification de plage, même pour des valeurs négatives.


Autorisé seulement pour l'adaptateur object. Spécifie le nom d'une « fonction de conversion » depuis C à utiliser pour convertir cet objet en type natif.


Autorisé seulement pour les chaînes de caractères. Spécifie l'encodage à utiliser lors de la conversion de cette chaîne depuis une valeur de type Python str (Unicode) en valeur C char *.


Autorisé seulement pour l'adaptateur object. Nécessite que la valeur Python soit une sous-classe d'un type Python, telle qu'exprimée en C.


Autorisé seulement pour les adaptateurs object et self. Spécifie le type C qui sera utilisé pour déclarer la variable. La valeur par défaut est "PyObject *".


Autorisé seulement pour les chaînes de caractères. Si vrai, les octets NUL ('\\0') sont permis au sein de la valeur. La taille de la chaîne sera passée à la fonction impl, juste après le paramètre chaîne, en tant que paramètre nommé <parameter_name>_length.

Please note, not every possible combination of arguments will work. Usually these arguments are implemented by specific PyArg_ParseTuple() format units, with specific behavior. For example, currently you cannot call unsigned_short without also specifying bitwise=True. Although it's perfectly reasonable to think this would work, these semantics don't map to any existing format unit. So Argument Clinic doesn't support it. (Or, at least, not yet.)

Vous pouvez voir, ci-dessous, une table présentant la correspondance entre les adaptateurs de base et ceux d'Argument Clinic. À gauche, sont listés les adaptateurs de base et, à droite, le texte qui les remplace.
















str(encoding='name_of_encoding', zeroes=True)


str(encoding='name_of_encoding', accept={bytes, bytearray, str})


str(encoding='name_of_encoding', accept={bytes, bytearray, str}, zeroes=True)


















long long


















Py_buffer(accept={buffer, str})














str(accept={robuffer}, zeroes=True)




Py_UNICODE(accept={str, NoneType})


Py_UNICODE(accept={str, NoneType}, zeroes=True)


str(accept={str, NoneType})


str(accept={str, NoneType}, zeroes=True)


Py_buffer(accept={buffer, str, NoneType})

Par exemple, voici notre code pickle.Pickler.dump via l'adaptateur approprié :

/*[clinic input]

    obj: object
        The object to be pickled.

Write a pickled representation of obj to the open file.
[clinic start generated code]*/

Un avantage des adaptateurs réels est qu'ils sont plus flexibles que les adaptateurs de base. Par exemple, l'adaptateur unsigned_int (ainsi que tous les adaptateurs unsigned_) peut être utilisé sans bitwise=True. Leur comportement par défaut contrôle la valeur, et n'acceptera pas de nombres négatifs. On ne peut pas faire ça avec les adaptateurs de base !

Argument Clinic sait lister tous les adaptateurs disponibles. Pour chaque adaptateur, il vous liste également l'ensemble des paramètres qu'ils acceptent, ainsi que les valeurs par défaut de chacun. Utilisez simplement la commande Tools/clinic/ --converters pour afficher la liste.

How to use the Py_buffer converter

Lorsque vous utilisez l'adaptateur Py_buffer (ou bien les adaptateurs de base 's*', 'w*', '*y', ou 'z*'), vous ne devez pas appeler PyBuffer_Release() sur le tampon fourni. Argument Clinic génère du code qui le fait pour vous (dans la fonction d'analyse).

How to use advanced converters

Vous vous souvenez de ces spécifications de format que vous avez laissées de côté la première fois parce qu'il s'agissait de notions avancées ? Voici comment les utiliser.

The trick is, all those format units take arguments—either conversion functions, or types, or strings specifying an encoding. (But "legacy converters" don't support arguments. That's why we skipped them for your first function.) The argument you specified to the format unit is now an argument to the converter; this argument is either converter (for O&), subclass_of (for O!), or encoding (for all the format units that start with e).

When using subclass_of, you may also want to use the other custom argument for object(): type, which lets you set the type actually used for the parameter. For example, if you want to ensure that the object is a subclass of PyUnicode_Type, you probably want to use the converter object(type='PyUnicodeObject *', subclass_of='&PyUnicode_Type').

One possible problem with using Argument Clinic: it takes away some possible flexibility for the format units starting with e. When writing a PyArg_Parse*() call by hand, you could theoretically decide at runtime what encoding string to pass to that call. But now this string must be hard-coded at Argument-Clinic-preprocessing-time. This limitation is deliberate; it made supporting this format unit much easier, and may allow for future optimizations. This restriction doesn't seem unreasonable; CPython itself always passes in static hard-coded encoding strings for parameters whose format units start with e.

How to assign default values to parameter

Les valeurs par défaut des paramètres peuvent être n'importe quelle valeur. Au plus simple, ce sont des chaînes, des entiers ou des nombres flottants :

foo: str = "abc"
bar: int = 123
bat: float = 45.6

Vous pouvez également utiliser n'importe quelle constante native de Python :

yep:  bool = True
nope: bool = False
nada: object = None

La valeur NULL est également acceptée, ainsi que des expressions simples, comme expliqué dans les sections suivantes.

La valeur par défaut NULL

Pour les paramètres chaînes et objets, vous pouvez les positionner à None pour indiquer qu'il n'y a pas de valeur par défaut. Pour autant, cela signifie que la variable C sera initialisée à Py_None. Par commodité, il existe une valeur spécifique appelée NULL juste pour cette raison : du point de vue de Python, cette valeur se comporte comme la valeur par défaut None, mais la variable C est initialisée à NULL.

Valeurs par défaut

La valeur par défaut que vous fournissez pour un paramètre ne peut pas être n'importe quelle expression. Actuellement, ce qui est géré :

  • Constantes numériques (entier ou nombre flottant)

  • Chaînes constantes

  • True, False et None

  • Simple symbolic constants like sys.maxsize, which must start with the name of the module

(Dans le futur, il est possible que l'on ait besoin de l'améliorer, pour autoriser les expressions complètes comme CONSTANT - 1.)

Expressions as default values

La valeur par défaut d'un paramètre peut être plus qu'une simple valeur littérale. Il peut s'agir d'une expression, utilisant des opérateurs mathématiques et des attributs d'objets. Cependant, cette possibilité n'est pas aussi simple, notamment à cause de sémantiques peu évidentes.

Examinons l'exemple suivant :

foo: Py_ssize_t = sys.maxsize - 1

sys.maxsize can have different values on different platforms. Therefore Argument Clinic can't simply evaluate that expression locally and hard-code it in C. So it stores the default in such a way that it will get evaluated at runtime, when the user asks for the function's signature.

What namespace is available when the expression is evaluated? It's evaluated in the context of the module the builtin came from. So, if your module has an attribute called max_widgets, you may simply use it:

foo: Py_ssize_t = max_widgets

If the symbol isn't found in the current module, it fails over to looking in sys.modules. That's how it can find sys.maxsize for example. (Since you don't know in advance what modules the user will load into their interpreter, it's best to restrict yourself to modules that are preloaded by Python itself.)

Evaluating default values only at runtime means Argument Clinic can't compute the correct equivalent C default value. So you need to tell it explicitly. When you use an expression, you must also specify the equivalent expression in C, using the c_default parameter to the converter:

foo: Py_ssize_t(c_default="PY_SSIZE_T_MAX - 1") = sys.maxsize - 1

Another complication: Argument Clinic can't know in advance whether or not the expression you supply is valid. It parses it to make sure it looks legal, but it can't actually know. You must be very careful when using expressions to specify values that are guaranteed to be valid at runtime!

Finally, because expressions must be representable as static C values, there are many restrictions on legal expressions. Here's a list of Python features you're not permitted to use:

  • des appels de fonction.

  • des instructions if en ligne (3 if foo else 5) ;

  • Automatic sequence unpacking (*[1, 2, 3]).

  • List/set/dict comprehensions and generator expressions.

  • Tuple/list/set/dict literals.

How to use return converters

By default, the impl function Argument Clinic generates for you returns PyObject *. But your C function often computes some C type, then converts it into the PyObject * at the last moment. Argument Clinic handles converting your inputs from Python types into native C types—why not have it convert your return value from a native C type into a Python type too?

That's what a "return converter" does. It changes your impl function to return some C type, then adds code to the generated (non-impl) function to handle converting that value into the appropriate PyObject *.

The syntax for return converters is similar to that of parameter converters. You specify the return converter like it was a return annotation on the function itself, using -> notation.

For example:

/*[clinic input]
add -> int

    a: int
    b: int

[clinic start generated code]*/

Return converters behave much the same as parameter converters; they take arguments, the arguments are all keyword-only, and if you're not changing any of the default arguments you can omit the parentheses.

(If you use both "as" and a return converter for your function, the "as" should come before the return converter.)

There's one additional complication when using return converters: how do you indicate an error has occurred? Normally, a function returns a valid (non-NULL) pointer for success, and NULL for failure. But if you use an integer return converter, all integers are valid. How can Argument Clinic detect an error? Its solution: each return converter implicitly looks for a special value that indicates an error. If you return that value, and an error has been set (c:func:PyErr_Occurred returns a true value), then the generated code will propagate the error. Otherwise it will encode the value you return like normal.

Currently Argument Clinic supports only a few return converters:

unsigned int
unsigned long

None of these take parameters. For all of these, return -1 to indicate error.

(There's also an experimental NoneType converter, which lets you return Py_None on success or NULL on failure, without having to increment the reference count on Py_None. I'm not sure it adds enough clarity to be worth using.)

To see all the return converters Argument Clinic supports, along with their parameters (if any), just run Tools/clinic/ --converters for the full list.

How to clone existing functions

If you have a number of functions that look similar, you may be able to use Clinic's "clone" feature. When you clone an existing function, you reuse:

  • its parameters, including

    • their names,

    • their converters, with all parameters,

    • their default values,

    • their per-parameter docstrings,

    • their kind (whether they're positional only, positional or keyword, or keyword only), and

  • its return converter.

The only thing not copied from the original function is its docstring; the syntax allows you to specify a new docstring.

Here's the syntax for cloning a function:

/*[clinic input]
module.class.new_function [as c_basename] = module.class.existing_function

Docstring for new_function goes here.
[clinic start generated code]*/

(The functions can be in different modules or classes. I wrote module.class in the sample just to illustrate that you must use the full path to both functions.)

Sorry, there's no syntax for partially cloning a function, or cloning a function then modifying it. Cloning is an all-or nothing proposition.

Also, the function you are cloning from must have been previously defined in the current file.

How to call Python code

The rest of the advanced topics require you to write Python code which lives inside your C file and modifies Argument Clinic's runtime state. This is simple: you simply define a Python block.

A Python block uses different delimiter lines than an Argument Clinic function block. It looks like this:

/*[python input]
# python code goes here
[python start generated code]*/

All the code inside the Python block is executed at the time it's parsed. All text written to stdout inside the block is redirected into the "output" after the block.

As an example, here's a Python block that adds a static integer variable to the C code:

/*[python input]
print('static int __ignored_unused_variable__ = 0;')
[python start generated code]*/
static int __ignored_unused_variable__ = 0;
/*[python checksum:...]*/

How to use the "self converter"

Argument Clinic automatically adds a "self" parameter for you using a default converter. It automatically sets the type of this parameter to the "pointer to an instance" you specified when you declared the type. However, you can override Argument Clinic's converter and specify one yourself. Just add your own self parameter as the first parameter in a block, and ensure that its converter is an instance of self_converter or a subclass thereof.

What's the point? This lets you override the type of self, or give it a different default name.

How do you specify the custom type you want to cast self to? If you only have one or two functions with the same type for self, you can directly use Argument Clinic's existing self converter, passing in the type you want to use as the type parameter:

/*[clinic input]


  self: self(type="PicklerObject *")
  obj: object

Write a pickled representation of the given object to the open file.
[clinic start generated code]*/

On the other hand, if you have a lot of functions that will use the same type for self, it's best to create your own converter, subclassing self_converter but overwriting the type member:

/*[python input]
class PicklerObject_converter(self_converter):
    type = "PicklerObject *"
[python start generated code]*/

/*[clinic input]


  self: PicklerObject
  obj: object

Write a pickled representation of the given object to the open file.
[clinic start generated code]*/

How to use the "defining class" converter

Argument Clinic facilitates gaining access to the defining class of a method. This is useful for heap type methods that need to fetch module level state. Use PyType_FromModuleAndSpec() to associate a new heap type with a module. You can now use PyType_GetModuleState() on the defining class to fetch the module state, for example from a module method.

Example from Modules/zlibmodule.c. First, defining_class is added to the clinic input:

/*[clinic input]

  cls: defining_class
  data: Py_buffer
    Binary data to be compressed.

After running the Argument Clinic tool, the following function signature is generated:

/*[clinic start generated code]*/
static PyObject *
zlib_Compress_compress_impl(compobject *self, PyTypeObject *cls,
                            Py_buffer *data)
/*[clinic end generated code: output=6731b3f0ff357ca6 input=04d00f65ab01d260]*/

The following code can now use PyType_GetModuleState(cls) to fetch the module state:

zlibstate *state = PyType_GetModuleState(cls);

Each method may only have one argument using this converter, and it must appear after self, or, if self is not used, as the first argument. The argument will be of type PyTypeObject *. The argument will not appear in the __text_signature__.

The defining_class converter is not compatible with __init__() and __new__() methods, which cannot use the METH_METHOD convention.

It is not possible to use defining_class with slot methods. In order to fetch the module state from such methods, use PyType_GetModuleByDef() to look up the module and then PyModule_GetState() to fetch the module state. Example from the setattro slot method in Modules/_threadmodule.c:

static int
local_setattro(localobject *self, PyObject *name, PyObject *v)
    PyObject *module = PyType_GetModuleByDef(Py_TYPE(self), &thread_module);
    thread_module_state *state = get_thread_state(module);

See also PEP 573.

How to write a custom converter

A converter is a Python class that inherits from CConverter. The main purpose of a custom converter, is for parameters parsed with the O& format unit --- parsing such a parameter means calling a PyArg_ParseTuple() "converter function".

Your converter class should be named ConverterName_converter. By following this convention, your converter class will be automatically registered with Argument Clinic, with its converter name being the name of your converter class with the _converter suffix stripped off.

Instead of subclassing CConverter.__init__(), write a converter_init() method. converter_init() always accepts a self parameter. After self, all additional parameters must be keyword-only. Any arguments passed to the converter in Argument Clinic will be passed along to your converter_init() method. See CConverter for a list of members you may wish to specify in your subclass.

Here's the simplest example of a custom converter, from Modules/zlibmodule.c:

/*[python input]

class ssize_t_converter(CConverter):
    type = 'Py_ssize_t'
    converter = 'ssize_t_converter'

[python start generated code]*/
/*[python end generated code: output=da39a3ee5e6b4b0d input=35521e4e733823c7]*/

This block adds a converter named ssize_t to Argument Clinic. Parameters declared as ssize_t will be declared with type Py_ssize_t, and will be parsed by the 'O&' format unit, which will call the ssize_t_converter() converter C function. ssize_t variables automatically support default values.

More sophisticated custom converters can insert custom C code to handle initialization and cleanup. You can see more examples of custom converters in the CPython source tree; grep the C files for the string CConverter.

How to write a custom return converter

Writing a custom return converter is much like writing a custom converter. Except it's somewhat simpler, because return converters are themselves much simpler.

Return converters must subclass CReturnConverter. There are no examples yet of custom return converters, because they are not widely used yet. If you wish to write your own return converter, please read Tools/clinic/, specifically the implementation of CReturnConverter and all its subclasses.

How to convert METH_O and METH_NOARGS functions

To convert a function using METH_O, make sure the function's single argument is using the object converter, and mark the arguments as positional-only:

/*[clinic input]

     argument: object
[clinic start generated code]*/

To convert a function using METH_NOARGS, just don't specify any arguments.

You can still use a self converter, a return converter, and specify a type argument to the object converter for METH_O.

How to convert tp_new and tp_init functions

You can convert tp_new and tp_init functions. Just name them __new__ or __init__ as appropriate. Notes:

  • The function name generated for __new__ doesn't end in __new__ like it would by default. It's just the name of the class, converted into a valid C identifier.

  • No PyMethodDef #define is generated for these functions.

  • __init__ functions return int, not PyObject *.

  • Use the docstring as the class docstring.

  • Although __new__ and __init__ functions must always accept both the args and kwargs objects, when converting you may specify any signature for these functions that you like. (If your function doesn't support keywords, the parsing function generated will throw an exception if it receives any.)

How to change and redirect Clinic's output

It can be inconvenient to have Clinic's output interspersed with your conventional hand-edited C code. Luckily, Clinic is configurable: you can buffer up its output for printing later (or earlier!), or write its output to a separate file. You can also add a prefix or suffix to every line of Clinic's generated output.

While changing Clinic's output in this manner can be a boon to readability, it may result in Clinic code using types before they are defined, or your code attempting to use Clinic-generated code before it is defined. These problems can be easily solved by rearranging the declarations in your file, or moving where Clinic's generated code goes. (This is why the default behavior of Clinic is to output everything into the current block; while many people consider this hampers readability, it will never require rearranging your code to fix definition-before-use problems.)

Let's start with defining some terminology:


A field, in this context, is a subsection of Clinic's output. For example, the #define for the PyMethodDef structure is a field, called methoddef_define. Clinic has seven different fields it can output per function definition:


All the names are of the form "<a>_<b>", where "<a>" is the semantic object represented (the parsing function, the impl function, the docstring, or the methoddef structure) and "<b>" represents what kind of statement the field is. Field names that end in "_prototype" represent forward declarations of that thing, without the actual body/data of the thing; field names that end in "_definition" represent the actual definition of the thing, with the body/data of the thing. ("methoddef" is special, it's the only one that ends with "_define", representing that it's a preprocessor #define.)


A destination is a place Clinic can write output to. There are five built-in destinations:


The default destination: printed in the output section of the current Clinic block.


A text buffer where you can save text for later. Text sent here is appended to the end of any existing text. It's an error to have any text left in the buffer when Clinic finishes processing a file.


A separate "clinic file" that will be created automatically by Clinic. The filename chosen for the file is {basename}.clinic{extension}, where basename and extension were assigned the output from os.path.splitext() run on the current file. (Example: the file destination for _pickle.c would be written to

Important: When using a file destination, you must check in the generated file!


A buffer like buffer. However, a two-pass buffer can only be dumped once, and it prints out all text sent to it during all processing, even from Clinic blocks after the dumping point.


The text is suppressed—thrown away.

Clinic defines five new directives that let you reconfigure its output.

The first new directive is dump:

dump <destination>

This dumps the current contents of the named destination into the output of the current block, and empties it. This only works with buffer and two-pass destinations.

The second new directive is output. The most basic form of output is like this:

output <field> <destination>

This tells Clinic to output field to destination. output also supports a special meta-destination, called everything, which tells Clinic to output all fields to that destination.

output has a number of other functions:

output push
output pop
output preset <preset>

output push and output pop allow you to push and pop configurations on an internal configuration stack, so that you can temporarily modify the output configuration, then easily restore the previous configuration. Simply push before your change to save the current configuration, then pop when you wish to restore the previous configuration.

output preset sets Clinic's output to one of several built-in preset configurations, as follows:


Clinic's original starting configuration. Writes everything immediately after the input block.

Suppress the parser_prototype and docstring_prototype, write everything else to block.


Designed to write everything to the "clinic file" that it can. You then #include this file near the top of your file. You may need to rearrange your file to make this work, though usually this just means creating forward declarations for various typedef and PyTypeObject definitions.

Suppress the parser_prototype and docstring_prototype, write the impl_definition to block, and write everything else to file.

The default filename is "{dirname}/clinic/{basename}.h".


Save up most of the output from Clinic, to be written into your file near the end. For Python files implementing modules or builtin types, it's recommended that you dump the buffer just above the static structures for your module or builtin type; these are normally very near the end. Using buffer may require even more editing than file, if your file has static PyMethodDef arrays defined in the middle of the file.

Suppress the parser_prototype, impl_prototype, and docstring_prototype, write the impl_definition to block, and write everything else to file.


Similar to the buffer preset, but writes forward declarations to the two-pass buffer, and definitions to the buffer. This is similar to the buffer preset, but may require less editing than buffer. Dump the two-pass buffer near the top of your file, and dump the buffer near the end just like you would when using the buffer preset.

Suppresses the impl_prototype, write the impl_definition to block, write docstring_prototype, methoddef_define, and parser_prototype to two-pass, write everything else to buffer.


Similar to the buffer preset, but writes more things to block, only writing the really big chunks of generated code to buffer. This avoids the definition-before-use problem of buffer completely, at the small cost of having slightly more stuff in the block's output. Dump the buffer near the end, just like you would when using the buffer preset.

Suppresses the impl_prototype, write the docstring_definition and parser_definition to buffer, write everything else to block.

The third new directive is destination:

destination <name> <command> [...]

This performs an operation on the destination named name.

There are two defined subcommands: new and clear.

The new subcommand works like this:

destination <name> new <type>

This creates a new destination with name <name> and type <type>.

There are five destination types:


Throws the text away.


Writes the text to the current block. This is what Clinic originally did.


A simple text buffer, like the "buffer" builtin destination above.


A text file. The file destination takes an extra argument, a template to use for building the filename, like so:

destination <name> new <type> <file_template>

The template can use three strings internally that will be replaced by bits of the filename:


The full path to the file, including directory and full filename.


The name of the directory the file is in.


Just the name of the file, not including the directory.


Basename with the extension clipped off (everything up to but not including the last '.').


The last '.' and everything after it. If the basename does not contain a period, this will be the empty string.

If there are no periods in the filename, {basename} and {filename} are the same, and {extension} is empty. "{basename}{extension}" is always exactly the same as "{filename}"."


A two-pass buffer, like the "two-pass" builtin destination above.

The clear subcommand works like this:

destination <name> clear

It removes all the accumulated text up to this point in the destination. (I don't know what you'd need this for, but I thought maybe it'd be useful while someone's experimenting.)

The fourth new directive is set:

set line_prefix "string"
set line_suffix "string"

set lets you set two internal variables in Clinic. line_prefix is a string that will be prepended to every line of Clinic's output; line_suffix is a string that will be appended to every line of Clinic's output.

Both of these support two format strings:

{block comment start}

Turns into the string /*, the start-comment text sequence for C files.

{block comment end}

Turns into the string */, the end-comment text sequence for C files.

The final new directive is one you shouldn't need to use directly, called preserve:


This tells Clinic that the current contents of the output should be kept, unmodified. This is used internally by Clinic when dumping output into file files; wrapping it in a Clinic block lets Clinic use its existing checksum functionality to ensure the file was not modified by hand before it gets overwritten.

How to use the #ifdef trick

If you're converting a function that isn't available on all platforms, there's a trick you can use to make life a little easier. The existing code probably looks like this:

static module_functionname(...)

And then in the PyMethodDef structure at the bottom the existing code will have:

{'functionname', ... },

In this scenario, you should enclose the body of your impl function inside the #ifdef, like so:

/*[clinic input]
[clinic start generated code]*/
static module_functionname(...)

Then, remove those three lines from the PyMethodDef structure, replacing them with the macro Argument Clinic generated:


(You can find the real name for this macro inside the generated code. Or you can calculate it yourself: it's the name of your function as defined on the first line of your block, but with periods changed to underscores, uppercased, and "_METHODDEF" added to the end.)

Perhaps you're wondering: what if HAVE_FUNCTIONNAME isn't defined? The MODULE_FUNCTIONNAME_METHODDEF macro won't be defined either!

Here's where Argument Clinic gets very clever. It actually detects that the Argument Clinic block might be deactivated by the #ifdef. When that happens, it generates a little extra code that looks like this:


That means the macro always works. If the function is defined, this turns into the correct structure, including the trailing comma. If the function is undefined, this turns into nothing.

However, this causes one ticklish problem: where should Argument Clinic put this extra code when using the "block" output preset? It can't go in the output block, because that could be deactivated by the #ifdef. (That's the whole point!)

In this situation, Argument Clinic writes the extra code to the "buffer" destination. This may mean that you get a complaint from Argument Clinic:

Warning in file "Modules/posixmodule.c" on line 12357:
Destination buffer 'buffer' not empty at end of file, emptying.

When this happens, just open your file, find the dump buffer block that Argument Clinic added to your file (it'll be at the very bottom), then move it above the PyMethodDef structure where that macro is used.

How to use Argument Clinic in Python files

It's actually possible to use Argument Clinic to preprocess Python files. There's no point to using Argument Clinic blocks, of course, as the output wouldn't make any sense to the Python interpreter. But using Argument Clinic to run Python blocks lets you use Python as a Python preprocessor!

Since Python comments are different from C comments, Argument Clinic blocks embedded in Python files look slightly different. They look like this:

#/*[python input]
#print("def foo(): pass")
#[python start generated code]*/
def foo(): pass
#/*[python checksum:...]*/

How to override the generated signature

You can use the @text_signature directive to override the default generated signature in the docstring. This can be useful for complex signatures that Argument Clinic cannot handle. The @text_signature directive takes one argument: the custom signature as a string. The provided signature is copied verbatim to the generated docstring.

Example from Objects/codeobject.c:

/*[clinic input]
@text_signature "($self, /, **changes)"
    co_argcount: int(c_default="self->co_argcount") = unchanged
    co_posonlyargcount: int(c_default="self->co_posonlyargcount") = unchanged
    # etc ...

    Return a copy of the code object with new values for the specified fields.
[clinic start generated output]*/

The generated docstring ends up looking like this:

replace($self, /, **changes)

Return a copy of the code object with new values for the specified fields.