dis – Désassembleur pour le code intermédiaire de Python

Code source : Lib/dis.py


La bibliothèque dis supporte l'analyse du bytecode CPython en le désassemblant. Le code intermédiaire CPython, que cette bibliothèque prend en paramètre, est défini dans le fichier Include/opcode.h et est utilisé par le compilateur et l'interpréteur.

CPython implementation detail: Le code intermédiaire est un détail d'implémentation de l'interpréteur CPython. Il n'y a pas de garantie que le code intermédiaire sera ajouté, retiré, ou modifié dans les différentes versions de Python. L'utilisation de cette bibliothèque ne fonctionne pas nécessairement sur les machines virtuelles Python ni les différentes versions de Python.

Modifié dans la version 3.6: Utilisez 2 bits pour chaque instruction. Avant, le nombre de bits variait par instruction.

Exemple : Etant donné la fonction myfunc() :

def myfunc(alist):
    return len(alist)

la commande suivante peut-être utilisé pour afficher le désassemblage de myfunc() :

>>> dis.dis(myfunc)
  2           0 LOAD_GLOBAL              0 (len)
              2 LOAD_FAST                0 (alist)
              4 CALL_FUNCTION            1
              6 RETURN_VALUE

(Le "2" est un numéro de ligne).

Analyse du code intermédiaire

Nouveau dans la version 3.4.

L'analyse de l'API code intermédiaire permet de rassembler des blocs de code en Python dans une classe Bytecode, qui permet un accès facile aux détails du code compilé.

class dis.Bytecode(x, *, first_line=None, current_offset=None)

Analyse le code intermédiaire correspondant à une fonction, un générateur, un générateur asynchrone, une coroutine, une méthode, une chaîne de caractères du code source, ou bien une classe (comme retourne la fonction compile()).

Ceci est wrapper sur plusieurs fonctions de la liste ci-dessous, notamment get_instructions(), étant donné qu'une itération sur une instance de la classe Bytecode rend les opérations du code intermédiaire des instances de Instruction.

Si first_line ne vaut pas None, elle indique le nombre de la ligne qui doit être considérée comme première ligne source dans le code désassemblé. Autrement, les informations sur la ligne source sont prises directement à partir de la classe du code désassemblé.

Si la valeur de current_offset est différente de None, c'est une référence à un offset d'une instruction dans le code désassemblé. Cela veut dire que dis() va générer un marqueur de " l'instruction en cours" contre le code d'opération donné.

classmethod from_traceback(tb)

Construisez une instance Bytecode à partir de la trace d'appel, en mettant current_offet à l'instruction responsable de l'exception.

codeobj

Le code compilé objet.

first_line

La première ligne source du code objet (si disponible)

dis()

Retourne une vue formatée des opérations du code intermédiaire (la même que celle envoyée par dis.dis(), mais comme une chaîne de caractères de plusieurs lignes ).

info()

Retourne une chaîne de caractères de plusieurs lignes formatée avec des informations détaillées sur l'objet code comme code_info().

Modifié dans la version 3.7: Cette version supporte la coroutine et les objets générateurs asynchrones.

Exemple :

>>> bytecode = dis.Bytecode(myfunc)
>>> for instr in bytecode:
...     print(instr.opname)
...
LOAD_GLOBAL
LOAD_FAST
CALL_FUNCTION
RETURN_VALUE

Analyse de fonctions

La bibliothèque dis comprend également l'analyse des fonctions suivantes, qui envoient l'entrée directement à la sortie souhaitée. Elles peuvent être utiles si il n'y a qu'une seule opération à effectuer, la représentation intermédiaire objet n'étant donc pas utile dans ce cas:

dis.code_info(x)

Retourne une chaîne de caractères de plusieurs lignes formatée avec des informations détaillées sur l'objet code pour les fonctions données, les générateurs asynchrone, coroutine, la méthode, la chaine de caractères du code source ou objet.

Il est à noter que le contenu exact des chaînes de caractères figurant dans les informations du code dépendent fortement sur l'implémentation, et peuvent changer arbitrairement sous machines virtuelles Python ou les versions de Python.

Nouveau dans la version 3.2.

Modifié dans la version 3.7: Cette version supporte la coroutine et les objets générateurs asynchrones.

dis.show_code(x, *, file=None)

Affiche des informations détaillées sur le code de la fonction fournie, la méthode, la chaîne de caractère du code source ou du code objet à file (ou bien sys.stdout si file n'est pas spécifié).

Ceci est un raccourci convenable de print(code_info(x), file=file), principalement fait pour l'exploration interactive sur l'invite de l'interpréteur.

Nouveau dans la version 3.2.

Modifié dans la version 3.4: Ajout du paramètre file.

dis.dis(x=None, *, file=None, depth=None)

Désassemble l'objet x. x peut être une bibliothèque , une classe, une méthode, une fonction, un générateur, un générateur asynchrone, une coroutine, un code objet, une chaine de caractères du coude source ou une séquence de bits du code intermédiaire brut. Pour une bibliothèque , elle désassemble toutes les fonctions. Pour une classe, elle désassemble toutes les méthodes (y compris les classes et méthodes statiques). Pour un code objet ou une séquence de code intermédiaire brut, elle affiche une ligne par instruction code intermédiaire. Aussi, elle désassemble les codes objets internes récursivement (le code en compréhension, les expressions des générateurs et les fonctions imbriquées, et le code utilisé pour la construction des classes internes). Les chaînes de caractères sont d'abord compilées pour coder des objets avec les fonctions intégrées de compile() avant qu'elles ne soient désassemblées. Si aucun objet n'est fourni, cette fonction désassemble les dernières traces d'appel.

Le désassemblage est envoyé sous forme de texte à l'argument du fichier file si il est fourni, et à sys.stdout sinon.

La profondeur maximale de récursion est limitée par depth sauf si elle correspond à None. depth=0 indique qu'il n'y a pas de récursion.

Modifié dans la version 3.4: Ajout du paramètre file.

Modifié dans la version 3.7: Le désassemblage récursif a été implémenté, et le paramètre depth a été ajouté.

Modifié dans la version 3.7: Cette version supporte la coroutine et les objets générateurs asynchrones.

dis.distb(tb=None, *, file=None)

Désassemble la fonction du haut de la pile des traces d'appels, en utilisant la dernière trace d'appels si rien n'a été envoyé. L'instruction à l'origine de l'exception est indiquée.

Le désassemblage est envoyé sous forme de texte à l'argument du fichier file si il est fourni, et à sys.stdout sinon.

Modifié dans la version 3.4: Ajout du paramètre file.

dis.disassemble(code, lasti=-1, *, file=None)
dis.disco(code, lasti=-1, *, file=None)

Désassemble un code objet, en indiquant la dernière instruction si lasti est fournie. La sortie est répartie sur les colonnes suivantes :

  1. le numéro de ligne, pour la première instruction de chaque ligne

  2. l'instruction en cours, indiquée par -->,

  3. une instruction libellée, indiquée par > >,

  4. l'adresse de l'instruction,

  5. le nom de le code d'opération,

  6. paramètres de l'opération, et

  7. interprétation des paramètres entre parenthèses.

L'interprétation du paramètre reconnaît les noms des variables locales et globales, des valeurs constantes, des branchements cibles, et des opérateurs de comparaison.

Le désassemblage est envoyé sous forme de texte à l'argument du fichier file si il est fourni, et à sys.stdout sinon.

Modifié dans la version 3.4: Ajout du paramètre file.

dis.get_instructions(x, *, first_line=None)

Retourne un itérateur sur les instructions dans la fonction fournie, la méthode, les chaînes de caractères du code source ou objet.

Cet itérateur génère une série de n-uplets de Instruction qui donnent les détails de chacune des opérations dans le code fourni.

Si first_line ne vaut pas None, elle indique le nombre de la ligne qui doit être considérée comme première ligne source dans le code désassemblé. Autrement, les informations sur la ligne source sont prises directement à partir de la classe du code désassemblé.

Nouveau dans la version 3.4.

dis.findlinestarts(code)

This generator function uses the co_firstlineno and co_lnotab attributes of the code object code to find the offsets which are starts of lines in the source code. They are generated as (offset, lineno) pairs. See Objects/lnotab_notes.txt for the co_lnotab format and how to decode it.

Modifié dans la version 3.6: Les numéros de lignes peuvent être décroissants. Avant, ils étaient toujours croissants.

dis.findlabels(code)

Detect all offsets in the raw compiled bytecode string code which are jump targets, and return a list of these offsets.

dis.stack_effect(opcode, oparg=None, *, jump=None)

Compute the stack effect of opcode with argument oparg.

If the code has a jump target and jump is True, stack_effect() will return the stack effect of jumping. If jump is False, it will return the stack effect of not jumping. And if jump is None (default), it will return the maximal stack effect of both cases.

Nouveau dans la version 3.4.

Modifié dans la version 3.8: Added jump parameter.

Les instructions du code intermédiaire en Python

La fonction get_instructions() et la méthode Bytecode fournit des détails sur le code intermédiaire des instructions comme Instruction instances:

class dis.Instruction

Détails sur le code intermédiaire de l'opération

opcode

code numérique pour l'opération, correspondant aux valeurs de l'opcode ci-dessous et les valeurs du code intermédiaire dans la Opcode collections.

opname

nom lisible/compréhensible de l'opération

arg

le cas échéant, argument numérique de l'opération sinon None

argval

resolved arg value (if known), otherwise same as arg

argrepr

human readable description of operation argument

offset

start index of operation within bytecode sequence

starts_line

line started by this opcode (if any), otherwise None

is_jump_target

True if other code jumps to here, otherwise False

Nouveau dans la version 3.4.

The Python compiler currently generates the following bytecode instructions.

General instructions

NOP

Do nothing code. Used as a placeholder by the bytecode optimizer.

POP_TOP

Removes the top-of-stack (TOS) item.

ROT_TWO

Swaps the two top-most stack items.

ROT_THREE

Lifts second and third stack item one position up, moves top down to position three.

ROT_FOUR

Lifts second, third and fourth stack items one position up, moves top down to position four.

Nouveau dans la version 3.8.

DUP_TOP

Duplicates the reference on top of the stack.

Nouveau dans la version 3.2.

DUP_TOP_TWO

Duplicates the two references on top of the stack, leaving them in the same order.

Nouveau dans la version 3.2.

Unary operations

Unary operations take the top of the stack, apply the operation, and push the result back on the stack.

UNARY_POSITIVE

Implements TOS = +TOS.

UNARY_NEGATIVE

Implements TOS = -TOS.

UNARY_NOT

Implements TOS = not TOS.

UNARY_INVERT

Implements TOS = ~TOS.

GET_ITER

Implements TOS = iter(TOS).

GET_YIELD_FROM_ITER

If TOS is a generator iterator or coroutine object it is left as is. Otherwise, implements TOS = iter(TOS).

Nouveau dans la version 3.5.

Binary operations

Binary operations remove the top of the stack (TOS) and the second top-most stack item (TOS1) from the stack. They perform the operation, and put the result back on the stack.

BINARY_POWER

Implements TOS = TOS1 ** TOS.

BINARY_MULTIPLY

Implements TOS = TOS1 * TOS.

BINARY_MATRIX_MULTIPLY

Implements TOS = TOS1 @ TOS.

Nouveau dans la version 3.5.

BINARY_FLOOR_DIVIDE

Implements TOS = TOS1 // TOS.

BINARY_TRUE_DIVIDE

Implements TOS = TOS1 / TOS.

BINARY_MODULO

Implements TOS = TOS1 % TOS.

BINARY_ADD

Implements TOS = TOS1 + TOS.

BINARY_SUBTRACT

Implements TOS = TOS1 - TOS.

BINARY_SUBSCR

Implements TOS = TOS1[TOS].

BINARY_LSHIFT

Implements TOS = TOS1 << TOS.

BINARY_RSHIFT

Implements TOS = TOS1 >> TOS.

BINARY_AND

Implements TOS = TOS1 & TOS.

BINARY_XOR

Implements TOS = TOS1 ^ TOS.

BINARY_OR

Implements TOS = TOS1 | TOS.

In-place operations

In-place operations are like binary operations, in that they remove TOS and TOS1, and push the result back on the stack, but the operation is done in-place when TOS1 supports it, and the resulting TOS may be (but does not have to be) the original TOS1.

INPLACE_POWER

Implements in-place TOS = TOS1 ** TOS.

INPLACE_MULTIPLY

Implements in-place TOS = TOS1 * TOS.

INPLACE_MATRIX_MULTIPLY

Implements in-place TOS = TOS1 @ TOS.

Nouveau dans la version 3.5.

INPLACE_FLOOR_DIVIDE

Implements in-place TOS = TOS1 // TOS.

INPLACE_TRUE_DIVIDE

Implements in-place TOS = TOS1 / TOS.

INPLACE_MODULO

Implements in-place TOS = TOS1 % TOS.

INPLACE_ADD

Implements in-place TOS = TOS1 + TOS.

INPLACE_SUBTRACT

Implements in-place TOS = TOS1 - TOS.

INPLACE_LSHIFT

Implements in-place TOS = TOS1 << TOS.

INPLACE_RSHIFT

Implements in-place TOS = TOS1 >> TOS.

INPLACE_AND

Implements in-place TOS = TOS1 & TOS.

INPLACE_XOR

Implements in-place TOS = TOS1 ^ TOS.

INPLACE_OR

Implements in-place TOS = TOS1 | TOS.

STORE_SUBSCR

Implements TOS1[TOS] = TOS2.

DELETE_SUBSCR

Implements del TOS1[TOS].

Coroutine opcodes

GET_AWAITABLE

Implements TOS = get_awaitable(TOS), where get_awaitable(o) returns o if o is a coroutine object or a generator object with the CO_ITERABLE_COROUTINE flag, or resolves o.__await__.

Nouveau dans la version 3.5.

GET_AITER

Implements TOS = TOS.__aiter__().

Nouveau dans la version 3.5.

Modifié dans la version 3.7: Returning awaitable objects from __aiter__ is no longer supported.

GET_ANEXT

Implements PUSH(get_awaitable(TOS.__anext__())). See GET_AWAITABLE for details about get_awaitable

Nouveau dans la version 3.5.

END_ASYNC_FOR

Terminates an async for loop. Handles an exception raised when awaiting a next item. If TOS is StopAsyncIteration pop 7 values from the stack and restore the exception state using the second three of them. Otherwise re-raise the exception using the three values from the stack. An exception handler block is removed from the block stack.

Nouveau dans la version 3.8.

BEFORE_ASYNC_WITH

Resolves __aenter__ and __aexit__ from the object on top of the stack. Pushes __aexit__ and result of __aenter__() to the stack.

Nouveau dans la version 3.5.

SETUP_ASYNC_WITH

Creates a new frame object.

Nouveau dans la version 3.5.

Miscellaneous opcodes

PRINT_EXPR

Implements the expression statement for the interactive mode. TOS is removed from the stack and printed. In non-interactive mode, an expression statement is terminated with POP_TOP.

SET_ADD(i)

Calls set.add(TOS1[-i], TOS). Used to implement set comprehensions.

LIST_APPEND(i)

Calls list.append(TOS1[-i], TOS). Used to implement list comprehensions.

MAP_ADD(i)

Calls dict.__setitem__(TOS1[-i], TOS1, TOS). Used to implement dict comprehensions.

Nouveau dans la version 3.1.

Modifié dans la version 3.8: Map value is TOS and map key is TOS1. Before, those were reversed.

For all of the SET_ADD, LIST_APPEND and MAP_ADD instructions, while the added value or key/value pair is popped off, the container object remains on the stack so that it is available for further iterations of the loop.

RETURN_VALUE

Returns with TOS to the caller of the function.

YIELD_VALUE

Pops TOS and yields it from a generator.

YIELD_FROM

Pops TOS and delegates to it as a subiterator from a generator.

Nouveau dans la version 3.3.

SETUP_ANNOTATIONS

Checks whether __annotations__ is defined in locals(), if not it is set up to an empty dict. This opcode is only emitted if a class or module body contains variable annotations statically.

Nouveau dans la version 3.6.

IMPORT_STAR

Loads all symbols not starting with '_' directly from the module TOS to the local namespace. The module is popped after loading all names. This opcode implements from module import *.

POP_BLOCK

Removes one block from the block stack. Per frame, there is a stack of blocks, denoting try statements, and such.

POP_EXCEPT

Removes one block from the block stack. The popped block must be an exception handler block, as implicitly created when entering an except handler. In addition to popping extraneous values from the frame stack, the last three popped values are used to restore the exception state.

POP_FINALLY(preserve_tos)

Cleans up the value stack and the block stack. If preserve_tos is not 0 TOS first is popped from the stack and pushed on the stack after performing other stack operations:

  • If TOS is NULL or an integer (pushed by BEGIN_FINALLY or CALL_FINALLY) it is popped from the stack.

  • If TOS is an exception type (pushed when an exception has been raised) 6 values are popped from the stack, the last three popped values are used to restore the exception state. An exception handler block is removed from the block stack.

It is similar to END_FINALLY, but doesn't change the bytecode counter nor raise an exception. Used for implementing break, continue and return in the finally block.

Nouveau dans la version 3.8.

BEGIN_FINALLY

Pushes NULL onto the stack for using it in END_FINALLY, POP_FINALLY, WITH_CLEANUP_START and WITH_CLEANUP_FINISH. Starts the finally block.

Nouveau dans la version 3.8.

END_FINALLY

Terminates a finally clause. The interpreter recalls whether the exception has to be re-raised or execution has to be continued depending on the value of TOS.

  • If TOS is NULL (pushed by BEGIN_FINALLY) continue from the next instruction. TOS is popped.

  • If TOS is an integer (pushed by CALL_FINALLY), sets the bytecode counter to TOS. TOS is popped.

  • If TOS is an exception type (pushed when an exception has been raised) 6 values are popped from the stack, the first three popped values are used to re-raise the exception and the last three popped values are used to restore the exception state. An exception handler block is removed from the block stack.

LOAD_BUILD_CLASS

Pushes builtins.__build_class__() onto the stack. It is later called by CALL_FUNCTION to construct a class.

SETUP_WITH(delta)

This opcode performs several operations before a with block starts. First, it loads __exit__() from the context manager and pushes it onto the stack for later use by WITH_CLEANUP_START. Then, __enter__() is called, and a finally block pointing to delta is pushed. Finally, the result of calling the __enter__() method is pushed onto the stack. The next opcode will either ignore it (POP_TOP), or store it in (a) variable(s) (STORE_FAST, STORE_NAME, or UNPACK_SEQUENCE).

Nouveau dans la version 3.2.

WITH_CLEANUP_START

Starts cleaning up the stack when a with statement block exits.

At the top of the stack are either NULL (pushed by BEGIN_FINALLY) or 6 values pushed if an exception has been raised in the with block. Below is the context manager's __exit__() or __aexit__() bound method.

If TOS is NULL, calls SECOND(None, None, None), removes the function from the stack, leaving TOS, and pushes None to the stack. Otherwise calls SEVENTH(TOP, SECOND, THIRD), shifts the bottom 3 values of the stack down, replaces the empty spot with NULL and pushes TOS. Finally pushes the result of the call.

WITH_CLEANUP_FINISH

Finishes cleaning up the stack when a with statement block exits.

TOS is result of __exit__() or __aexit__() function call pushed by WITH_CLEANUP_START. SECOND is None or an exception type (pushed when an exception has been raised).

Pops two values from the stack. If SECOND is not None and TOS is true unwinds the EXCEPT_HANDLER block which was created when the exception was caught and pushes NULL to the stack.

All of the following opcodes use their arguments.

STORE_NAME(namei)

Implements name = TOS. namei is the index of name in the attribute co_names of the code object. The compiler tries to use STORE_FAST or STORE_GLOBAL if possible.

DELETE_NAME(namei)

Implements del name, where namei is the index into co_names attribute of the code object.

UNPACK_SEQUENCE(count)

Unpacks TOS into count individual values, which are put onto the stack right-to-left.

UNPACK_EX(counts)

Implements assignment with a starred target: Unpacks an iterable in TOS into individual values, where the total number of values can be smaller than the number of items in the iterable: one of the new values will be a list of all leftover items.

The low byte of counts is the number of values before the list value, the high byte of counts the number of values after it. The resulting values are put onto the stack right-to-left.

STORE_ATTR(namei)

Implements TOS.name = TOS1, where namei is the index of name in co_names.

DELETE_ATTR(namei)

Implements del TOS.name, using namei as index into co_names.

STORE_GLOBAL(namei)

Works as STORE_NAME, but stores the name as a global.

DELETE_GLOBAL(namei)

Works as DELETE_NAME, but deletes a global name.

LOAD_CONST(consti)

Pushes co_consts[consti] onto the stack.

LOAD_NAME(namei)

Pushes the value associated with co_names[namei] onto the stack.

BUILD_TUPLE(count)

Creates a tuple consuming count items from the stack, and pushes the resulting tuple onto the stack.

BUILD_LIST(count)

Works as BUILD_TUPLE, but creates a list.

BUILD_SET(count)

Works as BUILD_TUPLE, but creates a set.

BUILD_MAP(count)

Pushes a new dictionary object onto the stack. Pops 2 * count items so that the dictionary holds count entries: {..., TOS3: TOS2, TOS1: TOS}.

Modifié dans la version 3.5: The dictionary is created from stack items instead of creating an empty dictionary pre-sized to hold count items.

BUILD_CONST_KEY_MAP(count)

The version of BUILD_MAP specialized for constant keys. Pops the top element on the stack which contains a tuple of keys, then starting from TOS1, pops count values to form values in the built dictionary.

Nouveau dans la version 3.6.

BUILD_STRING(count)

Concatenates count strings from the stack and pushes the resulting string onto the stack.

Nouveau dans la version 3.6.

BUILD_TUPLE_UNPACK(count)

Pops count iterables from the stack, joins them in a single tuple, and pushes the result. Implements iterable unpacking in tuple displays (*x, *y, *z).

Nouveau dans la version 3.5.

BUILD_TUPLE_UNPACK_WITH_CALL(count)

This is similar to BUILD_TUPLE_UNPACK, but is used for f(*x, *y, *z) call syntax. The stack item at position count + 1 should be the corresponding callable f.

Nouveau dans la version 3.6.

BUILD_LIST_UNPACK(count)

This is similar to BUILD_TUPLE_UNPACK, but pushes a list instead of tuple. Implements iterable unpacking in list displays [*x, *y, *z].

Nouveau dans la version 3.5.

BUILD_SET_UNPACK(count)

This is similar to BUILD_TUPLE_UNPACK, but pushes a set instead of tuple. Implements iterable unpacking in set displays {*x, *y, *z}.

Nouveau dans la version 3.5.

BUILD_MAP_UNPACK(count)

Pops count mappings from the stack, merges them into a single dictionary, and pushes the result. Implements dictionary unpacking in dictionary displays {**x, **y, **z}.

Nouveau dans la version 3.5.

BUILD_MAP_UNPACK_WITH_CALL(count)

This is similar to BUILD_MAP_UNPACK, but is used for f(**x, **y, **z) call syntax. The stack item at position count + 2 should be the corresponding callable f.

Nouveau dans la version 3.5.

Modifié dans la version 3.6: The position of the callable is determined by adding 2 to the opcode argument instead of encoding it in the second byte of the argument.

LOAD_ATTR(namei)

Replaces TOS with getattr(TOS, co_names[namei]).

COMPARE_OP(opname)

Performs a Boolean operation. The operation name can be found in cmp_op[opname].

IMPORT_NAME(namei)

Imports the module co_names[namei]. TOS and TOS1 are popped and provide the fromlist and level arguments of __import__(). The module object is pushed onto the stack. The current namespace is not affected: for a proper import statement, a subsequent STORE_FAST instruction modifies the namespace.

IMPORT_FROM(namei)

Loads the attribute co_names[namei] from the module found in TOS. The resulting object is pushed onto the stack, to be subsequently stored by a STORE_FAST instruction.

JUMP_FORWARD(delta)

Increments bytecode counter by delta.

POP_JUMP_IF_TRUE(target)

If TOS is true, sets the bytecode counter to target. TOS is popped.

Nouveau dans la version 3.1.

POP_JUMP_IF_FALSE(target)

If TOS is false, sets the bytecode counter to target. TOS is popped.

Nouveau dans la version 3.1.

JUMP_IF_TRUE_OR_POP(target)

If TOS is true, sets the bytecode counter to target and leaves TOS on the stack. Otherwise (TOS is false), TOS is popped.

Nouveau dans la version 3.1.

JUMP_IF_FALSE_OR_POP(target)

If TOS is false, sets the bytecode counter to target and leaves TOS on the stack. Otherwise (TOS is true), TOS is popped.

Nouveau dans la version 3.1.

JUMP_ABSOLUTE(target)

Set bytecode counter to target.

FOR_ITER(delta)

TOS is an iterator. Call its __next__() method. If this yields a new value, push it on the stack (leaving the iterator below it). If the iterator indicates it is exhausted TOS is popped, and the byte code counter is incremented by delta.

LOAD_GLOBAL(namei)

Loads the global named co_names[namei] onto the stack.

SETUP_FINALLY(delta)

Pushes a try block from a try-finally or try-except clause onto the block stack. delta points to the finally block or the first except block.

CALL_FINALLY(delta)

Pushes the address of the next instruction onto the stack and increments bytecode counter by delta. Used for calling the finally block as a "subroutine".

Nouveau dans la version 3.8.

LOAD_FAST(var_num)

Pushes a reference to the local co_varnames[var_num] onto the stack.

STORE_FAST(var_num)

Stores TOS into the local co_varnames[var_num].

DELETE_FAST(var_num)

Deletes local co_varnames[var_num].

LOAD_CLOSURE(i)

Pushes a reference to the cell contained in slot i of the cell and free variable storage. The name of the variable is co_cellvars[i] if i is less than the length of co_cellvars. Otherwise it is co_freevars[i - len(co_cellvars)].

LOAD_DEREF(i)

Loads the cell contained in slot i of the cell and free variable storage. Pushes a reference to the object the cell contains on the stack.

LOAD_CLASSDEREF(i)

Much like LOAD_DEREF but first checks the locals dictionary before consulting the cell. This is used for loading free variables in class bodies.

Nouveau dans la version 3.4.

STORE_DEREF(i)

Stores TOS into the cell contained in slot i of the cell and free variable storage.

DELETE_DEREF(i)

Empties the cell contained in slot i of the cell and free variable storage. Used by the del statement.

Nouveau dans la version 3.2.

RAISE_VARARGS(argc)

Raises an exception using one of the 3 forms of the raise statement, depending on the value of argc:

  • 0: raise (re-raise previous exception)

  • 1: raise TOS (raise exception instance or type at TOS)

  • 2: raise TOS1 from TOS (raise exception instance or type at TOS1 with __cause__ set to TOS)

CALL_FUNCTION(argc)

Calls a callable object with positional arguments. argc indicates the number of positional arguments. The top of the stack contains positional arguments, with the right-most argument on top. Below the arguments is a callable object to call. CALL_FUNCTION pops all arguments and the callable object off the stack, calls the callable object with those arguments, and pushes the return value returned by the callable object.

Modifié dans la version 3.6: This opcode is used only for calls with positional arguments.

CALL_FUNCTION_KW(argc)

Calls a callable object with positional (if any) and keyword arguments. argc indicates the total number of positional and keyword arguments. The top element on the stack contains a tuple of keyword argument names. Below that are keyword arguments in the order corresponding to the tuple. Below that are positional arguments, with the right-most parameter on top. Below the arguments is a callable object to call. CALL_FUNCTION_KW pops all arguments and the callable object off the stack, calls the callable object with those arguments, and pushes the return value returned by the callable object.

Modifié dans la version 3.6: Keyword arguments are packed in a tuple instead of a dictionary, argc indicates the total number of arguments.

CALL_FUNCTION_EX(flags)

Calls a callable object with variable set of positional and keyword arguments. If the lowest bit of flags is set, the top of the stack contains a mapping object containing additional keyword arguments. Below that is an iterable object containing positional arguments and a callable object to call. BUILD_MAP_UNPACK_WITH_CALL and BUILD_TUPLE_UNPACK_WITH_CALL can be used for merging multiple mapping objects and iterables containing arguments. Before the callable is called, the mapping object and iterable object are each "unpacked" and their contents passed in as keyword and positional arguments respectively. CALL_FUNCTION_EX pops all arguments and the callable object off the stack, calls the callable object with those arguments, and pushes the return value returned by the callable object.

Nouveau dans la version 3.6.

LOAD_METHOD(namei)

Loads a method named co_names[namei] from the TOS object. TOS is popped. This bytecode distinguishes two cases: if TOS has a method with the correct name, the bytecode pushes the unbound method and TOS. TOS will be used as the first argument (self) by CALL_METHOD when calling the unbound method. Otherwise, NULL and the object return by the attribute lookup are pushed.

Nouveau dans la version 3.7.

CALL_METHOD(argc)

Calls a method. argc is the number of positional arguments. Keyword arguments are not supported. This opcode is designed to be used with LOAD_METHOD. Positional arguments are on top of the stack. Below them, the two items described in LOAD_METHOD are on the stack (either self and an unbound method object or NULL and an arbitrary callable). All of them are popped and the return value is pushed.

Nouveau dans la version 3.7.

MAKE_FUNCTION(flags)

Pushes a new function object on the stack. From bottom to top, the consumed stack must consist of values if the argument carries a specified flag value

  • 0x01 a tuple of default values for positional-only and positional-or-keyword parameters in positional order

  • 0x02 a dictionary of keyword-only parameters' default values

  • 0x04 an annotation dictionary

  • 0x08 a tuple containing cells for free variables, making a closure

  • the code associated with the function (at TOS1)

  • the qualified name of the function (at TOS)

BUILD_SLICE(argc)

Pushes a slice object on the stack. argc must be 2 or 3. If it is 2, slice(TOS1, TOS) is pushed; if it is 3, slice(TOS2, TOS1, TOS) is pushed. See the slice() built-in function for more information.

EXTENDED_ARG(ext)

Prefixes any opcode which has an argument too big to fit into the default one byte. ext holds an additional byte which act as higher bits in the argument. For each opcode, at most three prefixal EXTENDED_ARG are allowed, forming an argument from two-byte to four-byte.

FORMAT_VALUE(flags)

Used for implementing formatted literal strings (f-strings). Pops an optional fmt_spec from the stack, then a required value. flags is interpreted as follows:

  • (flags & 0x03) == 0x00: value is formatted as-is.

  • (flags & 0x03) == 0x01: call str() on value before formatting it.

  • (flags & 0x03) == 0x02: call repr() on value before formatting it.

  • (flags & 0x03) == 0x03: call ascii() on value before formatting it.

  • (flags & 0x04) == 0x04: pop fmt_spec from the stack and use it, else use an empty fmt_spec.

Formatting is performed using PyObject_Format(). The result is pushed on the stack.

Nouveau dans la version 3.6.

HAVE_ARGUMENT

This is not really an opcode. It identifies the dividing line between opcodes which don't use their argument and those that do (< HAVE_ARGUMENT and >= HAVE_ARGUMENT, respectively).

Modifié dans la version 3.6: Now every instruction has an argument, but opcodes < HAVE_ARGUMENT ignore it. Before, only opcodes >= HAVE_ARGUMENT had an argument.

Opcode collections

These collections are provided for automatic introspection of bytecode instructions:

dis.opname

Sequence of operation names, indexable using the bytecode.

dis.opmap

Dictionary mapping operation names to bytecodes.

dis.cmp_op

Sequence of all compare operation names.

dis.hasconst

Sequence of bytecodes that access a constant.

dis.hasfree

Sequence of bytecodes that access a free variable (note that 'free' in this context refers to names in the current scope that are referenced by inner scopes or names in outer scopes that are referenced from this scope. It does not include references to global or builtin scopes).

dis.hasname

Sequence of bytecodes that access an attribute by name.

dis.hasjrel

Sequence of bytecodes that have a relative jump target.

dis.hasjabs

Sequence of bytecodes that have an absolute jump target.

dis.haslocal

Sequence of bytecodes that access a local variable.

dis.hascompare

Sequence of bytecodes of Boolean operations.