Builtins

Built-in Constants
Built-in Exceptions
Built-in functions
Python built-ins reference
Built-in Types
Thread Safety Guarantees
Time complexity of operations on built-in types

Built-in Constants

A small number of constants live in the built-in namespace. They are:

False

The false value of the bool type. Assignments to False are illegal and raise a SyntaxError.

True

The true value of the bool type. Assignments to True are illegal and raise a SyntaxError.

None

An object frequently used to represent the absence of a value, as when default arguments are not passed to a function. Assignments to None are illegal and raise a SyntaxError. None is the sole instance of the ~types.NoneType type.

NotImplemented

A special value which should be returned by the binary special methods (e.g. ~object.__eq__, ~object.__lt__, ~object.__add__, ~object.__rsub__, etc.) to indicate that the operation is not implemented with respect to the other type; may be returned by the in-place binary special methods (e.g. ~object.__imul__, ~object.__iand__, etc.) for the same purpose. It should not be evaluated in a boolean context. !NotImplemented is the sole instance of the types.NotImplementedType type.

[!NOTE] When a binary (or in-place) method returns !NotImplemented the interpreter will try the reflected operation on the other type (or some other fallback, depending on the operator). If all attempts return !NotImplemented, the interpreter will raise an appropriate exception. Incorrectly returning !NotImplemented will result in a misleading error message or the !NotImplemented value being returned to Python code.

See implementing-the-arithmetic-operations for examples.

[!CAUTION] !NotImplemented and !NotImplementedError are not interchangeable. This constant should only be used as described above; see NotImplementedError for details on correct usage of the exception.

3.9 Evaluating !NotImplemented in a boolean context was deprecated.

3.14 Evaluating !NotImplemented in a boolean context now raises a TypeError. It previously evaluated to True and emitted a DeprecationWarning since Python 3.9.

single: ...; ellipsis literal

Ellipsis

The same as the ellipsis literal "...", an object frequently used to indicate that something is omitted. Assignment to Ellipsis is possible, but assignment to ... raises a SyntaxError. Ellipsis is the sole instance of the types.EllipsisType type.

__debug__

This constant is true if Python was not started with an -O option. See also the assert statement.

[!NOTE] The names None, False, True and __debug__ cannot be reassigned (assignments to them, even as an attribute name, raise SyntaxError), so they can be considered "true" constants.

Constants added by the `site` module

The site module (which is imported automatically during startup, except if the -S command-line option is given) adds several constants to the built-in namespace. They are useful for the interactive interpreter shell and should not be used in programs.

quit(code=None) exit(code=None)

Objects that when printed, print a message like "Use quit() or Ctrl-D (i.e. EOF) to exit", and when accessed directly in the interactive interpreter or called as functions, raise SystemExit with the specified exit code.

help

Object that when printed, prints the message "Type help() for interactive help, or help(object) for help about object.", and when accessed directly in the interactive interpreter, invokes the built-in help system (see help).

Objects that when printed or called, print the text of copyright or credits, respectively.

license

Object that when printed, prints the message "Type license() to see the full license text", and when called, displays the full license text in a pager-like fashion (one screen at a time).

Built-in Exceptions

pair: statement; try pair: statement; except

In Python, all exceptions must be instances of a class that derives from BaseException. In a try statement with an except clause that mentions a particular class, that clause also handles any exception classes derived from that class (but not exception classes from which it is derived). Two exception classes that are not related via subclassing are never equivalent, even if they have the same name.

pair: statement; raise

The built-in exceptions listed in this chapter can be generated by the interpreter or built-in functions. Except where mentioned, they have an "associated value" indicating the detailed cause of the error. This may be a string or a tuple of several items of information (e.g., an error code and a string explaining the code). The associated value is usually passed as arguments to the exception class's constructor.

User code can raise built-in exceptions. This can be used to test an exception handler or to report an error condition "just like" the situation in which the interpreter raises the same exception; but beware that there is nothing to prevent user code from raising an inappropriate error.

The built-in exception classes can be subclassed to define new exceptions; programmers are encouraged to derive new exceptions from the Exception class or one of its subclasses, and not from BaseException. More information on defining exceptions is available in the Python Tutorial under tut-userexceptions.

Exception context

pair: exception; chaining __cause__ (exception attribute) __context__ (exception attribute) __suppress_context__ (exception attribute)

Three attributes on exception objects provide information about the context in which the exception was raised:

BaseException.__context__ BaseException.__cause__ BaseException.__suppress_context__

When raising a new exception while another exception is already being handled, the new exception's !__context__ attribute is automatically set to the handled exception. An exception may be handled when an except or finally clause, or a with statement, is used.

This implicit exception context can be supplemented with an explicit cause by using !from with raise:

raise new_exc from original_exc

The expression following from<raise> must be an exception or None. It will be set as !__cause__ on the raised exception. Setting !__cause__ also implicitly sets the !__suppress_context__ attribute to True, so that using raise new_exc from None effectively replaces the old exception with the new one for display purposes (e.g. converting KeyError to AttributeError), while leaving the old exception available in !__context__ for introspection when debugging.

The default traceback display code shows these chained exceptions in addition to the traceback for the exception itself. An explicitly chained exception in !__cause__ is always shown when present. An implicitly chained exception in !__context__ is shown only if !__cause__ is None and !__suppress_context__ is false.

In either case, the exception itself is always shown after any chained exceptions so that the final line of the traceback always shows the last exception that was raised.

Inheriting from built-in exceptions

User code can create subclasses that inherit from an exception type. It's recommended to only subclass one exception type at a time to avoid any possible conflicts between how the bases handle the args attribute, as well as due to possible memory layout incompatibilities.

Most built-in exceptions are implemented in C for efficiency, see: Objects/exceptions.c. Some have custom memory layouts which makes it impossible to create a subclass that inherits from multiple exception types. The memory layout of a type is an implementation detail and might change between Python versions, leading to new conflicts in the future. Therefore, it's recommended to avoid subclassing multiple exception types altogether.

Base classes

The following exceptions are used mostly as base classes for other exceptions.

BaseException

The base class for all built-in exceptions. It is not meant to be directly inherited by user-defined classes (for that, use Exception). If str is called on an instance of this class, the representation of the argument(s) to the instance are returned, or the empty string when there were no arguments.

args

The tuple of arguments given to the exception constructor. Some built-in exceptions (like OSError) expect a certain number of arguments and assign a special meaning to the elements of this tuple, while others are usually called only with a single string giving an error message.

with_traceback(tb)

This method sets tb as the new traceback for the exception and returns the exception object. It was more commonly used before the exception chaining features of 3134 became available. The following example shows how we can convert an instance of SomeException into an instance of OtherException while preserving the traceback. Once raised, the current frame is pushed onto the traceback of the OtherException, as would have happened to the traceback of the original SomeException had we allowed it to propagate to the caller. :

try:
    ...
except SomeException:
    tb = sys.exception().__traceback__
    raise OtherException(...).with_traceback(tb)

__traceback__

A writable field that holds the traceback object <traceback-objects> associated with this exception. See also: raise.

add_note(note)

Add the string note to the exception's notes which appear in the standard traceback after the exception string. A TypeError is raised if note is not a string.

3.11

__notes__

A list of the notes of this exception, which were added with add_note. This attribute is created when add_note is called.

3.11

Exception

All built-in, non-system-exiting exceptions are derived from this class. All user-defined exceptions should also be derived from this class.

ArithmeticError

The base class for those built-in exceptions that are raised for various arithmetic errors: OverflowError, ZeroDivisionError, FloatingPointError.

BufferError

Raised when a buffer <bufferobjects> related operation cannot be performed.

LookupError

The base class for the exceptions that are raised when a key or index used on a mapping or sequence is invalid: IndexError, KeyError. This can be raised directly by codecs.lookup.

Concrete exceptions

The following exceptions are the exceptions that are usually raised.

AssertionError

pair: statement; assert

Raised when an assert statement fails.

AttributeError

Raised when an attribute reference (see attribute-references) or assignment fails. (When an object does not support attribute references or attribute assignments at all, TypeError is raised.)

The optional name and obj keyword-only arguments set the corresponding attributes:

name

The name of the attribute that was attempted to be accessed.

obj

The object that was accessed for the named attribute.

When possible, name and obj are set automatically.

3.10 Added the name and obj attributes.

EOFError

Raised when the input function hits an end-of-file condition (EOF) without reading any data. (Note: the io.TextIOBase.read and io.IOBase.readline methods return an empty string when they hit EOF.)

FloatingPointError

Not currently used.

GeneratorExit

Raised when a generator or coroutine is closed; see generator.close and coroutine.close. It directly inherits from BaseException instead of Exception since it is technically not an error.

ImportError

Raised when the import statement has troubles trying to load a module. Also raised when the "from list" in from ... import has a name that cannot be found.

The optional name and path keyword-only arguments set the corresponding attributes:

name

The name of the module that was attempted to be imported.

path

The path to any file which triggered the exception.

3.3 Added the name and path attributes.

ModuleNotFoundError

A subclass of ImportError which is raised by import when a module could not be located. It is also raised when None is found in sys.modules.

3.6

ImportCycleError

A subclass of ImportError which is raised when a lazy import fails because it (directly or indirectly) tries to import itself.

3.15

IndexError

Raised when a sequence subscript is out of range. (Slice indices are silently truncated to fall in the allowed range; if an index is not an integer, TypeError is raised.)

KeyError

Raised when a mapping (dictionary) key is not found in the set of existing keys.

KeyboardInterrupt

Raised when the user hits the interrupt key (normally Control-C or Delete). During execution, a check for interrupts is made regularly. The exception inherits from BaseException so as to not be accidentally caught by code that catches Exception and thus prevent the interpreter from exiting.

[!NOTE] Catching a KeyboardInterrupt requires special consideration. Because it can be raised at unpredictable points, it may, in some circumstances, leave the running program in an inconsistent state. It is generally best to allow KeyboardInterrupt to end the program as quickly as possible or avoid raising it entirely. (See handlers-and-exceptions.)

MemoryError

Raised when an operation runs out of memory but the situation may still be rescued (by deleting some objects). The associated value is a string indicating what kind of (internal) operation ran out of memory. Note that because of the underlying memory management architecture (C's malloc function), the interpreter may not always be able to completely recover from this situation; it nevertheless raises an exception so that a stack traceback can be printed, in case a run-away program was the cause.

NameError

Raised when a local or global name is not found. This applies only to unqualified names. The associated value is an error message that includes the name that could not be found.

The optional name keyword-only argument sets the attribute:

name

The name of the variable that was attempted to be accessed.

3.10 Added the name attribute.

NotImplementedError

This exception is derived from RuntimeError. In user defined base classes, abstract methods should raise this exception when they require derived classes to override the method, or while the class is being developed to indicate that the real implementation still needs to be added.

[!NOTE] It should not be used to indicate that an operator or method is not meant to be supported at all -- in that case either leave the operator / method undefined or, if a subclass, set it to None.

[!CAUTION] !NotImplementedError and !NotImplemented are not interchangeable. This exception should only be used as described above; see NotImplemented for details on correct usage of the built-in constant.

OSError([arg]) OSError(errno, strerror[, filename[, winerror[, filename2]]])

pair: module; errno

This exception is raised when a system function returns a system-related error, including I/O failures such as "file not found" or "disk full" (not for illegal argument types or other incidental errors).

The second form of the constructor sets the corresponding attributes, described below. The attributes default to None if not specified. For backwards compatibility, if three arguments are passed, the ~BaseException.args attribute contains only a 2-tuple of the first two constructor arguments.

The constructor often actually returns a subclass of OSError, as described in OS exceptions below. The particular subclass depends on the final .errno value. This behaviour only occurs when constructing OSError directly or via an alias, and is not inherited when subclassing.

errno

A numeric error code from the C variable errno.

winerror

Under Windows, this gives you the native Windows error code. The .errno attribute is then an approximate translation, in POSIX terms, of that native error code.

Under Windows, if the winerror constructor argument is an integer, the .errno attribute is determined from the Windows error code, and the errno argument is ignored. On other platforms, the winerror argument is ignored, and the winerror attribute does not exist.

strerror

The corresponding error message, as provided by the operating system. Under POSIX, it is obtained using the C functions !strerror or !strerror_r; under Windows, !FormatMessage is used.

filename filename2

For exceptions that involve a file system path (such as open or os.unlink), filename is the file name passed to the function. For functions that involve two file system paths (such as os.rename), filename2 corresponds to the second file name passed to the function.

3.3 EnvironmentError, IOError, WindowsError, socket.error, select.error and !mmap.error have been merged into OSError, and the constructor may return a subclass.

3.4 The filename attribute is now the original file name passed to the function, instead of the name encoded to or decoded from the filesystem encoding and error handler. Also, the filename2 constructor argument and attribute was added.

OverflowError

Raised when the result of an arithmetic operation is too large to be represented. This cannot occur for integers (which would rather raise MemoryError than give up). However, for historical reasons, OverflowError is sometimes raised for integers that are outside a required range. Because of the lack of standardization of floating-point exception handling in C, most floating-point operations are not checked.

PythonFinalizationError

This exception is derived from RuntimeError. It is raised when an operation is blocked during interpreter shutdown also known as Python finalization <interpreter shutdown>.

Examples of operations which can be blocked with a PythonFinalizationError during the Python finalization:

Creating a new Python thread.
Joining <threading.Thread.join> a running daemon thread.
os.fork,
acquiring a lock such as threading.Lock, when it is known that the operation would otherwise deadlock.

See also the sys.is_finalizing function.

3.13 Previously, a plain RuntimeError was raised.

3.14

threading.Thread.join can now raise this exception.

3.15

This exception may be raised when acquiring threading.Lock or threading.RLock.

RecursionError

This exception is derived from RuntimeError. It is raised when the interpreter detects that the maximum recursion depth (see sys.getrecursionlimit) is exceeded.

3.5 Previously, a plain RuntimeError was raised.

ReferenceError

This exception is raised when a weak reference proxy, created by the weakref.proxy function, is used to access an attribute of the referent after it has been garbage collected. For more information on weak references, see the weakref module.

RuntimeError

Raised when an error is detected that doesn't fall in any of the other categories. The associated value is a string indicating what precisely went wrong.

StopIteration

Raised by built-in function next and an iterator's ~iterator.__next__ method to signal that there are no further items produced by the iterator.

StopIteration.value

The exception object has a single attribute !value, which is given as an argument when constructing the exception, and defaults to None.

When a generator or coroutine function returns, a new StopIteration instance is raised, and the value returned by the function is used as the value parameter to the constructor of the exception.

If a generator code directly or indirectly raises StopIteration, it is converted into a RuntimeError (retaining the StopIteration as the new exception's cause).

3.3 Added value attribute and the ability for generator functions to use it to return a value.

3.5 Introduced the RuntimeError transformation via from __future__ import generator_stop, see 479.

3.7 Enable 479 for all code by default: a StopIteration error raised in a generator is transformed into a RuntimeError.

StopAsyncIteration

Must be raised by ~object.__anext__ method of an asynchronous iterator object to stop the iteration.

3.5

SyntaxError(message, details)

Raised when the parser encounters a syntax error. This may occur in an import statement, in a call to the built-in functions compile, exec, or eval, or when reading the initial script or standard input (also interactively).

The str of the exception instance returns only the error message. Details is a tuple whose members are also available as separate attributes.

filename

The name of the file the syntax error occurred in.

lineno

Which line number in the file the error occurred in. This is 1-indexed: the first line in the file has a lineno of 1.

offset

The column in the line where the error occurred. This is 1-indexed: the first character in the line has an offset of 1.

text

The source code text involved in the error.

end_lineno

Which line number in the file the error occurred ends in. This is 1-indexed: the first line in the file has a lineno of 1.

end_offset

The column in the end line where the error occurred finishes. This is 1-indexed: the first character in the line has an offset of 1.

3.10 Added the end_lineno and end_offset attributes.

IndentationError

Base class for syntax errors related to incorrect indentation. This is a subclass of SyntaxError.

TabError

Raised when indentation contains an inconsistent use of tabs and spaces. This is a subclass of IndentationError.

SystemError

Raised when the interpreter finds an internal error, but the situation does not look so serious to cause it to abandon all hope. The associated value is a string indicating what went wrong (in low-level terms). In CPython, this could be raised by incorrectly using Python's C API, such as returning a NULL value without an exception set.

If you're confident that this exception wasn't your fault, or the fault of a package you're using, you should report this to the author or maintainer of your Python interpreter. Be sure to report the version of the Python interpreter (sys.version; it is also printed at the start of an interactive Python session), the exact error message (the exception's associated value) and if possible the source of the program that triggered the error.

SystemExit

This exception is raised by the sys.exit function. It inherits from BaseException instead of Exception so that it is not accidentally caught by code that catches Exception. This allows the exception to properly propagate up and cause the interpreter to exit. When it is not handled, the Python interpreter exits; no stack traceback is printed. The constructor accepts the same optional argument passed to sys.exit. If the value is an integer, it specifies the system exit status (passed to C's !exit function); if it is None, the exit status is zero; if it has another type (such as a string), the object's value is printed and the exit status is one.

A call to sys.exit is translated into an exception so that clean-up handlers (finally clauses of try statements) can be executed, and so that a debugger can execute a script without running the risk of losing control. The os._exit function can be used if it is absolutely positively necessary to exit immediately (for example, in the child process after a call to os.fork).

code

The exit status or error message that is passed to the constructor. (Defaults to None.)

TypeError

Raised when an operation or function is applied to an object of inappropriate type. The associated value is a string giving details about the type mismatch.

This exception may be raised by user code to indicate that an attempted operation on an object is not supported, and is not meant to be. If an object is meant to support a given operation but has not yet provided an implementation, NotImplementedError is the proper exception to raise.

Passing arguments of the wrong type (e.g. passing a list when an int is expected) should result in a TypeError, but passing arguments with the wrong value (e.g. a number outside expected boundaries) should result in a ValueError.

UnboundLocalError

Raised when a reference is made to a local variable in a function or method, but no value has been bound to that variable. This is a subclass of NameError.

UnicodeError

Raised when a Unicode-related encoding or decoding error occurs. It is a subclass of ValueError.

UnicodeError has attributes that describe the encoding or decoding error. For example, err.object[err.start:err.end] gives the particular invalid input that the codec failed on.

encoding

The name of the encoding that raised the error.

reason

A string describing the specific codec error.

object

The object the codec was attempting to encode or decode.

start

The first index of invalid data in object.

This value should not be negative as it is interpreted as an absolute offset but this constraint is not enforced at runtime.

end

The index after the last invalid data in object.

This value should not be negative as it is interpreted as an absolute offset but this constraint is not enforced at runtime.

UnicodeEncodeError

Raised when a Unicode-related error occurs during encoding. It is a subclass of UnicodeError.

UnicodeDecodeError

Raised when a Unicode-related error occurs during decoding. It is a subclass of UnicodeError.

UnicodeTranslateError

Raised when a Unicode-related error occurs during translating. It is a subclass of UnicodeError.

ValueError

Raised when an operation or function receives an argument that has the right type but an inappropriate value, and the situation is not described by a more precise exception such as IndexError.

ZeroDivisionError

Raised when the second argument of a division or modulo operation is zero. The associated value is a string indicating the type of the operands and the operation.

The following exceptions are kept for compatibility with previous versions; starting from Python 3.3, they are aliases of OSError.

EnvironmentError

IOError

WindowsError

Only available on Windows.

OS exceptions

The following exceptions are subclasses of OSError, they get raised depending on the system error code.

BlockingIOError

Raised when an operation would block on an object (e.g. socket) set for non-blocking operation. Corresponds to errno ~errno.EAGAIN, ~errno.EALREADY, ~errno.EWOULDBLOCK and ~errno.EINPROGRESS.

In addition to those of OSError, BlockingIOError can have one more attribute:

characters_written

An integer containing the number of bytes written to the stream before it blocked. This attribute is available when using the buffered I/O classes from the io module.

ChildProcessError

Raised when an operation on a child process failed. Corresponds to errno ~errno.ECHILD.

ConnectionError

A base class for connection-related issues.

Subclasses are BrokenPipeError, ConnectionAbortedError, ConnectionRefusedError and ConnectionResetError.

BrokenPipeError

A subclass of ConnectionError, raised when trying to write on a pipe while the other end has been closed, or trying to write on a socket which has been shutdown for writing. Corresponds to errno ~errno.EPIPE and ~errno.ESHUTDOWN.

ConnectionAbortedError

A subclass of ConnectionError, raised when a connection attempt is aborted by the peer. Corresponds to errno ~errno.ECONNABORTED.

ConnectionRefusedError

A subclass of ConnectionError, raised when a connection attempt is refused by the peer. Corresponds to errno ~errno.ECONNREFUSED.

ConnectionResetError

A subclass of ConnectionError, raised when a connection is reset by the peer. Corresponds to errno ~errno.ECONNRESET.

FileExistsError

Raised when trying to create a file or directory which already exists. Corresponds to errno ~errno.EEXIST.

FileNotFoundError

Raised when a file or directory is requested but doesn't exist. Corresponds to errno ~errno.ENOENT.

InterruptedError

Raised when a system call is interrupted by an incoming signal. Corresponds to errno ~errno.EINTR.

3.5 Python now retries system calls when a syscall is interrupted by a signal, except if the signal handler raises an exception (see 475 for the rationale), instead of raising InterruptedError.

IsADirectoryError

Raised when a file operation (such as os.remove) is requested on a directory. Corresponds to errno ~errno.EISDIR.

NotADirectoryError

Raised when a directory operation (such as os.listdir) is requested on something which is not a directory. On most POSIX platforms, it may also be raised if an operation attempts to open or traverse a non-directory file as if it were a directory. Corresponds to errno ~errno.ENOTDIR.

PermissionError

Raised when trying to run an operation without the adequate access rights - for example filesystem permissions. Corresponds to errno ~errno.EACCES, ~errno.EPERM, and ~errno.ENOTCAPABLE.

3.11.1 WASI's ~errno.ENOTCAPABLE is now mapped to PermissionError.

ProcessLookupError

Raised when a given process doesn't exist. Corresponds to errno ~errno.ESRCH.

TimeoutError

Raised when a system function timed out at the system level. Corresponds to errno ~errno.ETIMEDOUT.

3.3 All the above OSError subclasses were added.

3151 - Reworking the OS and IO exception hierarchy

Warnings

The following exceptions are used as warning categories; see the warning-categories documentation for more details.

Warning

Base class for warning categories.

UserWarning

Base class for warnings generated by user code.

DeprecationWarning

Base class for warnings about deprecated features when those warnings are intended for other Python developers.

Ignored by the default warning filters, except in the __main__ module (565). Enabling the Python Development Mode <devmode> shows this warning.

The deprecation policy is described in 387.

PendingDeprecationWarning

Base class for warnings about features which are obsolete and expected to be deprecated in the future, but are not deprecated at the moment.

This class is rarely used as emitting a warning about a possible upcoming deprecation is unusual, and DeprecationWarning is preferred for already active deprecations.

Ignored by the default warning filters. Enabling the Python Development Mode <devmode> shows this warning.

The deprecation policy is described in 387.

SyntaxWarning

Base class for warnings about dubious syntax.

This warning is typically emitted when compiling Python source code, and usually won't be reported when running already compiled code.

RuntimeWarning

Base class for warnings about dubious runtime behavior.

FutureWarning

Base class for warnings about deprecated features when those warnings are intended for end users of applications that are written in Python.

ImportWarning

Base class for warnings about probable mistakes in module imports.

Ignored by the default warning filters. Enabling the Python Development Mode <devmode> shows this warning.

UnicodeWarning

Base class for warnings related to Unicode.

EncodingWarning

Base class for warnings related to encodings.

See io-encoding-warning for details.

3.10

BytesWarning

Base class for warnings related to bytes and bytearray.

ResourceWarning

Base class for warnings related to resource usage.

Ignored by the default warning filters. Enabling the Python Development Mode <devmode> shows this warning.

3.2

Exception groups

The following are used when it is necessary to raise multiple unrelated exceptions. They are part of the exception hierarchy so they can be handled with except like all other exceptions. In addition, they are recognised by except*<except_star>, which matches their subgroups based on the types of the contained exceptions.

ExceptionGroup(msg, excs)

BaseExceptionGroup(msg, excs)

Both of these exception types wrap the exceptions in the sequence excs. The msg parameter must be a string. The difference between the two classes is that BaseExceptionGroup extends BaseException and it can wrap any exception, while ExceptionGroup extends Exception and it can only wrap subclasses of Exception. This design is so that except Exception catches an ExceptionGroup but not BaseExceptionGroup.

The BaseExceptionGroup constructor returns an ExceptionGroup rather than a BaseExceptionGroup if all contained exceptions are Exception instances, so it can be used to make the selection automatic. The ExceptionGroup constructor, on the other hand, raises a TypeError if any contained exception is not an Exception subclass.

Exception groups are generic <generics> over the type of their contained exceptions.

The excs parameter may be any sequence, but lists and tuples are specifically processed more efficiently here. For optimal performance, pass a tuple as excs.

message

The msg argument to the constructor. This is a read-only attribute.

exceptions

A tuple of the exceptions in the excs sequence given to the constructor. This is a read-only attribute.

subgroup(condition)

Returns an exception group that contains only the exceptions from the current group that match condition, or None if the result is empty.

The condition can be an exception type or tuple of exception types, in which case each exception is checked for a match using the same check that is used in an except clause. The condition can also be a callable (other than a type object) that accepts an exception as its single argument and returns true for the exceptions that should be in the subgroup.

The nesting structure of the current exception is preserved in the result, as are the values of its message, ~BaseException.__traceback__, ~BaseException.__cause__, ~BaseException.__context__ and ~BaseException.__notes__ fields. Empty nested groups are omitted from the result.

The condition is checked for all exceptions in the nested exception group, including the top-level and any nested exception groups. If the condition is true for such an exception group, it is included in the result in full.

3.13 condition can be any callable which is not a type object.

split(condition)

Like subgroup, but returns the pair (match, rest) where match is subgroup(condition) and rest is the remaining non-matching part.

derive(excs)

Returns an exception group with the same message, but which wraps the exceptions in excs.

This method is used by subgroup and split, which are used in various contexts to break up an exception group. A subclass needs to override it in order to make subgroup and split return instances of the subclass rather than ExceptionGroup.

subgroup and split copy the ~BaseException.__traceback__, ~BaseException.__cause__, ~BaseException.__context__ and ~BaseException.__notes__ fields from the original exception group to the one returned by derive, so these fields do not need to be updated by derive.

>>> class MyGroup(ExceptionGroup): ... def derive(self, excs): ... return MyGroup(self.message, excs) ... >>> e = MyGroup("eg", [ValueError(1), TypeError(2)]) >>> e.add_note("a note") >>> e.__context__ = Exception("context") >>> e.__cause__ = Exception("cause") >>> try: ... raise e ... except Exception as e: ... exc = e ... >>> match, rest = exc.split(ValueError) >>> exc, exc.__context__, exc.__cause__, exc.__notes__ (MyGroup('eg', [ValueError(1), TypeError(2)]), Exception('context'), Exception('cause'), ['a note']) >>> match, match.__context__, match.__cause__, match.__notes__ (MyGroup('eg', [ValueError(1)]), Exception('context'), Exception('cause'), ['a note']) >>> rest, rest.__context__, rest.__cause__, rest.__notes__ (MyGroup('eg', [TypeError(2)]), Exception('context'), Exception('cause'), ['a note']) >>> exc.__traceback__ is match.__traceback__ is rest.__traceback__ True

Note that BaseExceptionGroup defines ~object.__new__, so subclasses that need a different constructor signature need to override that rather than ~object.__init__. For example, the following defines an exception group subclass which accepts an exit_code and constructs the group's message from it. :

class Errors(ExceptionGroup):
   def __new__(cls, errors, exit_code):
      self = super().__new__(Errors, f"exit code: {exit_code}", errors)
      self.exit_code = exit_code
      return self

   def derive(self, excs):
      return Errors(excs, self.exit_code)

Like ExceptionGroup, any subclass of BaseExceptionGroup which is also a subclass of Exception can only wrap instances of Exception.

3.11

Exception hierarchy

The class hierarchy for built-in exceptions is:

../../Lib/test/exception_hierarchy.txt

Built-in functions

The Python interpreter has a number of functions and types built into it that are always available. They are listed here in alphabetical order.

Built-in Functions
A `abs` `aiter` `all` `anext` `any` `ascii` B `bin` `bool` `breakpoint` `bytearray()` `bytes()` C `callable` `chr` `classmethod` `compile` `complex` D `delattr` `dict()` `dir` `divmod`	E `enumerate` `eval` `exec` F `filter` `float` `format` `frozendict()` `frozenset()` G `getattr` `globals` H `hasattr` `hash` `help` `hex` I `id` `input` `int` `isinstance` `issubclass` `iter`	L `len` `list()` `locals` M `map` `max` `memoryview()` `min` N `next` O `object` `oct` `open` `ord` P `pow` `print` `property`	R `range()` `repr` `reversed` `round` S `set()` `setattr` `sentinel` `slice` `sorted` `staticmethod` `str()` `sum` `super` T `tuple()` `type` V `vars` Z `zip` _ `__import__`

abs(number, /)

Return the absolute value of a number. The argument may be an integer, a floating-point number, or an object implementing ~object.__abs__. If the argument is a complex number, its magnitude is returned.

aiter(async_iterable, /) aiter(callable, /, stop_value, , stop_exception=StopAsyncIteration) aiter(callable, /,, stop_exception)

Return an asynchronous iterator object. The first argument is interpreted very differently depending on the presence of the other arguments. Without other arguments, the single argument must be an asynchronous iterable, and the result is equivalent to calling x.__aiter__().

If stop_value or stop_exception is given, then the first argument must be a callable object. The asynchronous iterator created in this case calls callable with no arguments and awaits the result for each call to its ~object.__anext__ method; if the awaited value is equal to stop_value, or if the call raises an exception matching stop_exception, StopAsyncIteration will be raised, otherwise the value will be returned. The callable is only called when the result of ~object.__anext__ is awaited.

stop_exception is an exception class or a tuple of exception classes. If stop_value is not specified, the iteration stops only when the callable raises an exception. If the callable raises StopAsyncIteration which does not match stop_exception, it is replaced with a RuntimeError, as for asynchronous generators (see 525).

For example, reading fixed-size chunks from an asynchronous stream until the end of file is reached:

from functools import partial
async for chunk in aiter(partial(reader.read, 1024), b''):
    process_chunk(chunk)

Or consuming an asyncio.Queue until it is shut down:

from asyncio import QueueShutDown
async for item in aiter(queue.get, stop_exception=QueueShutDown):
    process_item(item)

3.10

3.16 Added the stop_value and stop_exception parameters.

all(iterable, /)

Return True if all elements of the iterable are true (or if the iterable is empty). Equivalent to:

def all(iterable):
    for element in iterable:
        if not element:
            return False
    return True

anext(async_iterator, /) anext(async_iterator, default, /)

When awaited, return the next item from the given asynchronous iterator, or default if given and the iterator is exhausted.

This is the async variant of the next builtin, and behaves similarly.

This calls the ~object.__anext__ method of async_iterator, returning an awaitable. Awaiting this returns the next value of the iterator. If default is given, it is returned if the iterator is exhausted, otherwise StopAsyncIteration is raised.

3.10

any(iterable, /)

Return True if any element of the iterable is true. If the iterable is empty, return False. Equivalent to:

def any(iterable):
    for element in iterable:
        if element:
            return True
    return False

ascii(object, /)

As repr, return a string containing a printable representation of an object, but escape the non-ASCII characters in the string returned by repr using \x, \u, or \U escapes. This generates a string similar to that returned by repr in Python 2.

bin(integer, /)

Convert an integer number to a binary string prefixed with "0b". The result is a valid Python expression. If integer is not a Python int object, it has to define an ~object.__index__ method that returns an integer. Some examples:

>>> bin(3) '0b11' >>> bin(-10) '-0b1010'

If the prefix "0b" is desired or not, you can use either of the following ways.

>>> format(14, '#b'), format(14, 'b') ('0b1110', '1110') >>> f'{14:#b}', f'{14:b}' ('0b1110', '1110')

See also enum.bin to represent negative values as twos-complement.

Character	Meaning
`'r'`	open for reading (default)
`'w'`	open for writing, truncating the file first
`'x'`	open for exclusive creation, failing if the file already exists
`'a'`	open for writing, appending to the end of file if it exists
`'b'`	binary mode
`'t'`	text mode (default)
`'+'`	open for updating (reading and writing)

Python built-ins reference

Python comes with a number of built-in functions and classes.

The built-in classes include data types that would normally be considered part of the "core" of a language, such as numbers and lists. For these types, the Python language core defines the form of literals and places some constraints on their semantics, but does not fully define the semantics.

The built-ins also include functions and exceptions --- objects that can be used by all Python code without the need of an import statement. Some of these are defined by the core language, but many are not essential for the core semantics and are only described here.

In addition to the built-ins, Python provides an extensive importable standard library, see library-index.

stdtypes.rst constants.rst functions.rst exceptions.rst threadsafety.rst time-complexity.rst

Built-in Types

The following sections describe the standard types that are built into the interpreter.

pair: built-in; types

The principal built-in types are numerics, sequences, mappings, classes, instances and exceptions.

Some collection classes are mutable. The methods that add, subtract, or rearrange their members in place, and don't return a specific item, never return the collection instance itself but None.

Some operations are supported by several object types; in particular, practically all objects can be compared for equality, tested for truth value, and converted to a string (with the repr function or the slightly different str function). The latter function is implicitly used when an object is written by the print function.

Truth Value Testing

pair: statement; if pair: statement; while pair: truth; value pair: Boolean; operations single: false

Any object can be tested for truth value, for use in an if or while condition or as operand of the Boolean operations below.

single: true

By default, an object is considered true unless its class defines either a ~object.__bool__ method that returns False or a ~object.__len__ method that returns zero, when called with the object.[^1] If one of the methods raises an exception when called, the exception is propagated and the object does not have a truth value (for example, NotImplemented). Here are most of the built-in objects considered false:

single: None (Built-in object) single: False (Built-in object)

constants defined to be false: None and False
zero of any numeric type: 0, 0.0, 0j, Decimal(0), Fraction(0, 1)
empty sequences and collections: '', (), [], {}, set(), range(0)

pair: operator; or pair: operator; and single: False single: True

Operations and built-in functions that have a Boolean result always return 0 or False for false and 1 or True for true, unless otherwise stated. (Important exception: the Boolean operations or and and always return one of their operands.)

Boolean Operations --- `!and`, `!or`, `!not`

pair: Boolean; operations

These are the Boolean operations, ordered by ascending priority:

Operation	Result	Notes
`x or y`	if x is true, then x, else y	(1)
`x and y`	if x is false, then x, else y	(2)
`not x`	if x is false, then `True`, else `False`	(3)

pair: operator; and pair: operator; or pair: operator; not

Notes:

(1)
This is a short-circuit operator, so it only evaluates the second argument if the first one is false.

(2)
This is a short-circuit operator, so it only evaluates the second argument if the first one is true.

(3)
not has a lower priority than non-Boolean operators, so not a == b is interpreted as not (a == b), and a == not b is a syntax error.

Comparisons

pair: chaining; comparisons pair: operator; comparison pair: operator; == pair: operator; < (less) pair: operator; <= pair: operator; > (greater) pair: operator; >= pair: operator; != pair: operator; is pair: operator; is not

There are eight comparison operations in Python. They all have the same priority (which is higher than that of the Boolean operations). Comparisons can be chained arbitrarily; for example, x < y <= z is equivalent to x < y and y <= z, except that y is evaluated only once (but in both cases z is not evaluated at all when x < y is found to be false).

This table summarizes the comparison operations:

Operation	Meaning
`<`	strictly less than
`<=`	less than or equal
`>`	strictly greater than
`>=`	greater than or equal
`==`	equal
`!=`	not equal
`is`	object identity
`is not`	negated object identity

pair: object; numeric pair: objects; comparing

Unless stated otherwise, objects of different types never compare equal. The == operator is always defined but for some object types (for example, class objects) is equivalent to is. The <, <=, > and >= operators are only defined where they make sense; for example, they raise a TypeError exception when one of the arguments is a complex number.

single: __eq__() (instance method) single: __ne__() (instance method) single: __lt__() (instance method) single: __le__() (instance method) single: __gt__() (instance method) single: __ge__() (instance method)

Non-identical instances of a class normally compare as non-equal unless the class defines the ~object.__eq__ method.

Instances of a class cannot be ordered with respect to other instances of the same class, or other types of object, unless the class defines enough of the methods ~object.__lt__, ~object.__le__, ~object.__gt__, and ~object.__ge__ (in general, ~object.__lt__ and ~object.__eq__ are sufficient, if you want the conventional meanings of the comparison operators).

The behavior of the is and is not operators cannot be customized; also they can be applied to any two objects and never raise an exception.

pair: operator; in pair: operator; not in

Two more operations with the same syntactic priority, in and not in, are supported by types that are iterable or implement the ~object.__contains__ method.

Numeric Types --- `int`, `float`, `complex`

pair: object; numeric pair: object; Boolean pair: object; integer pair: object; floating-point pair: object; complex number pair: C; language

There are three distinct numeric types: integers, floating-point numbers, and complex numbers. In addition, Booleans are a subtype of integers. Integers have unlimited precision. Floating-point numbers are usually implemented using double in C; information about the precision and internal representation of floating-point numbers for the machine on which your program is running is available in sys.float_info. Complex numbers have a real and imaginary part, which are each a floating-point number. To extract these parts from a complex number z, use z.real and z.imag. (The standard library includes the additional numeric types fractions.Fraction, for rationals, and decimal.Decimal, for floating-point numbers with user-definable precision.)

pair: numeric; literals pair: integer; literals pair: floating-point; literals pair: complex number; literals pair: hexadecimal; literals pair: octal; literals pair: binary; literals

Numbers are created by numeric literals or as the result of built-in functions and operators. Unadorned integer literals (including hex, octal and binary numbers) yield integers. Numeric literals containing a decimal point or an exponent sign yield floating-point numbers. Appending 'j' or 'J' to a numeric literal yields an imaginary number (a complex number with a zero real part) which you can add to an integer or float to get a complex number with real and imaginary parts.

The constructors int, float, and complex can be used to produce numbers of a specific type.

single: arithmetic pair: built-in function; int pair: built-in function; float pair: built-in function; complex single: operator; + (plus) single: + (plus); unary operator single: + (plus); binary operator single: operator; - (minus) single: - (minus); unary operator single: - (minus); binary operator pair: operator; * (asterisk) pair: operator; / (slash) pair: operator; // pair: operator; % (percent) pair: operator; **

Python fully supports mixed arithmetic: when a binary arithmetic operator has operands of different built-in numeric types, the operand with the "narrower" type is widened to that of the other:

If both arguments are complex numbers, no conversion is performed;
if either argument is a complex or a floating-point number, the other is converted to a floating-point number;
otherwise, both must be integers and no conversion is necessary.

Arithmetic with complex and real operands is defined by the usual mathematical formula, for example:

x + complex(u, v) = complex(x + u, v)
x * complex(u, v) = complex(x * u, x * v)

A comparison between numbers of different types behaves as though the exact values of those numbers were being compared.[^2]

All numeric types (except complex) support the following operations (for priorities of the operations, see operator-summary):

Operation	Result	Notes	Full documentation
`x + y`	sum of x and y
`x - y`	difference of x and y
`x * y`	product of x and y
`x / y`	quotient of x and y
`x // y`	floored quotient of x and y	(1)(2)
`x % y`	remainder of `x / y`	(2)
`-x`	x negated
`+x`	x unchanged
`abs(x)`	absolute value or magnitude of x		`abs`
`int(x)`	x converted to integer	(3)(6)	`int`
`float(x)`	x converted to floating point	(4)(6)	`float`
`complex(re, im)`	a complex number with real part re, imaginary part im. im defaults to zero.	(6)	`complex`
`c.conjugate()`	conjugate of the complex number c
`divmod(x, y)`	the pair `(x // y, x % y)`	(2)	`divmod`
`pow(x, y)`	x to the power y	(5)	`pow`
`x ** y`	x to the power y	(5)

triple: operations on; numeric; types single: conjugate() (complex number method)

Notes:

(1)
Also referred to as integer division. For operands of type int, the result has type int. For operands of type float, the result has type float. In general, the result is a whole integer, though the result's type is not necessarily int. The result is always rounded towards minus infinity: 1//2 is 0, (-1)//2 is -1, 1//(-2) is -1, and (-1)//(-2) is 0.

(2)
Not for complex numbers. Instead convert to floats using abs if appropriate.

(3)

pair: module; math single: floor() (in module math) single: ceil() (in module math) single: trunc() (in module math) pair: numeric; conversions

Conversion from float to int truncates, discarding the fractional part. See functions math.floor and math.ceil for alternative conversions.

(4)
float also accepts the strings "nan" and "inf" with an optional prefix "+" or "-" for Not a Number (NaN) and positive or negative infinity.

(5)
Python defines pow(0, 0) and 0 ** 0 to be 1, as is common for programming languages.

(6)
The numeric literals accepted include the digits 0 to 9 or any Unicode equivalent (code points with the Nd property).

See the Unicode Standard for a complete list of code points with the Nd property.

All numbers.Real types (int and float) also include the following operations:

Operation	Result
:func:`math.trunc(x) <math.trunc>`	x truncated to `~numbers.Integral`
`round(x[,
n]) `	x rounded to n digits, rounding half to even. If n is omitted, it defaults to 0.
:func:`math.floor(x) <math.floor>`	the greatest `~numbers.Integral` <= x
`math.ceil(x)
<math.ceil>`	the least `~numbers.Integral` >= x

For additional numeric operations see the math and cmath modules.

Bitwise Operations on Integer Types

triple: operations on; integer; types pair: bitwise; operations pair: shifting; operations pair: masking; operations pair: operator; | (vertical bar) pair: operator; ^ (caret) pair: operator; & (ampersand) pair: operator; << pair: operator; >> pair: operator; ~ (tilde)

Bitwise operations only make sense for integers. The result of bitwise operations is calculated as though carried out in two's complement with an infinite number of sign bits.

The priorities of the binary bitwise operations are all lower than the numeric operations and higher than the comparisons; the unary operation ~ has the same priority as the other unary numeric operations (+ and -).

This table lists the bitwise operations sorted in ascending priority:

Operation	Result	Notes
`x	y`	bitwise `or` of x and y
`x ^ y`	bitwise `exclusive or` of x and y	(4)
`x & y`	bitwise `and` of x and y	(4)
`x << n`	x shifted left by n bits	(1)(2)
`x >> n`	x shifted right by n bits	(1)(3)
`~x`	the bits of x inverted

Notes:

(1)
Negative shift counts are illegal and cause a ValueError to be raised.

(2)
A left shift by n bits is equivalent to multiplication by pow(2, n).

(3)
A right shift by n bits is equivalent to floor division by pow(2, n).

(4)
Performing these calculations with at least one extra sign extension bit in a finite two's complement representation (a working bit-width of 1 + max(x.bit_length(), y.bit_length()) or more) is sufficient to get the same result as if there were an infinite number of sign bits.

Additional Methods on Integer Types

The int type implements the numbers.Integral abstract base class. In addition, it provides a few more methods:

int.bit_length()

Return the number of bits necessary to represent an integer in binary, excluding the sign and leading zeros:

>>> n = -37
>>> bin(n)
'-0b100101'
>>> n.bit_length()
6

More precisely, if x is nonzero, then x.bit_length() is the unique positive integer k such that 2**(k-1) <= abs(x) < 2**k. Equivalently, when abs(x) is small enough to have a correctly rounded logarithm, then k = 1 + int(log(abs(x), 2)). If x is zero, then x.bit_length() returns 0.

Equivalent to:

def bit_length(self):
    s = bin(self)       # binary representation:  bin(-37) --> '-0b100101'
    s = s.lstrip('-0b') # remove leading zeros and minus sign
    return len(s)       # len('100101') --> 6

3.1

int.bit_count()

Return the number of ones in the binary representation of the absolute value of the integer. This is also known as the population count. Example:

>>> n = 19
>>> bin(n)
'0b10011'
>>> n.bit_count()
3
>>> (-n).bit_count()
3

Equivalent to:

def bit_count(self):
    return bin(self).count("1")

3.10

int.to_bytes(length=1, byteorder='big', *, signed=False)

Return an array of bytes representing an integer.

>>> (1024).to_bytes(2, byteorder='big') b'x04x00' >>> (1024).to_bytes(10, byteorder='big') b'x00x00x00x00x00x00x00x00x04x00' >>> (-1024).to_bytes(10, byteorder='big', signed=True) b'xffxffxffxffxffxffxffxffxfcx00' >>> x = 1000 >>> x.to_bytes((x.bit_length() + 7) // 8, byteorder='little') b'xe8x03'

The integer is represented using length bytes, and defaults to 1. An OverflowError is raised if the integer is not representable with the given number of bytes.

The byteorder argument determines the byte order used to represent the integer, and defaults to "big". If byteorder is "big", the most significant byte is at the beginning of the byte array. If byteorder is "little", the most significant byte is at the end of the byte array.

The signed argument determines whether two's complement is used to represent the integer. If signed is False and a negative integer is given, an OverflowError is raised. The default value for signed is False.

The default values can be used to conveniently turn an integer into a single byte object:

>>> (65).to_bytes()
b'A'

However, when using the default arguments, don't try to convert a value greater than 255 or you'll get an OverflowError.

Equivalent to:

def to_bytes(n, length=1, byteorder='big', signed=False):
    if byteorder == 'little':
        order = range(length)
    elif byteorder == 'big':
        order = reversed(range(length))
    else:
        raise ValueError("byteorder must be either 'little' or 'big'")

    return bytes((n >> i*8) & 0xff for i in order)

3.2

3.11 Added default argument values for length and byteorder.

int.from_bytes(bytes, byteorder='big', *, signed=False)

Return the integer represented by the given array of bytes.

>>> int.from_bytes(b'x00x10', byteorder='big') 16 >>> int.from_bytes(b'x00x10', byteorder='little') 4096 >>> int.from_bytes(b'xfcx00', byteorder='big', signed=True) -1024 >>> int.from_bytes(b'xfcx00', byteorder='big', signed=False) 64512 >>> int.from_bytes([255, 0, 0], byteorder='big') 16711680

The argument bytes must either be a bytes-like object or an iterable producing bytes.

The signed argument indicates whether two's complement is used to represent the integer.

Equivalent to:

def from_bytes(bytes, byteorder='big', signed=False):
    if byteorder == 'little':
        little_ordered = list(bytes)
    elif byteorder == 'big':
        little_ordered = list(reversed(bytes))
    else:
        raise ValueError("byteorder must be either 'little' or 'big'")

    n = sum(b << i*8 for i, b in enumerate(little_ordered))
    if signed and little_ordered and (little_ordered[-1] & 0x80):
        n -= 1 << 8*len(little_ordered)

    return n

3.2

3.11 Added default argument value for byteorder.

int.as_integer_ratio()

Return a pair of integers whose ratio is equal to the original integer and has a positive denominator. The integer ratio of integers (whole numbers) is always the integer as the numerator and 1 as the denominator.

3.8

int.is_integer()

Returns True. Exists for duck type compatibility with float.is_integer.

3.12

Additional Methods on Float

The float type implements the numbers.Real abstract base class. float also has the following additional methods.

float.from_number(x)

Class method to return a floating-point number constructed from a number x.

If the argument is an integer or a floating-point number, a floating-point number with the same value (within Python's floating-point precision) is returned. If the argument is outside the range of a Python float, an OverflowError will be raised.

For a general Python object x, float.from_number(x) delegates to x.__float__(). If ~object.__float__ is not defined then it falls back to ~object.__index__.

3.14

float.as_integer_ratio()

Return a pair of integers whose ratio is exactly equal to the original float. The ratio is in lowest terms and has a positive denominator. Raises OverflowError on infinities and a ValueError on NaNs.

float.is_integer()

Return True if the float instance is finite with integral value, and False otherwise:

>>> (-2.0).is_integer()
True
>>> (3.2).is_integer()
False

Two methods support conversion to and from hexadecimal strings. Since Python's floats are stored internally as binary numbers, converting a float to or from a decimal string usually involves a small rounding error. In contrast, hexadecimal strings allow exact representation and specification of floating-point numbers. This can be useful when debugging, and in numerical work.

float.hex()

Return a representation of a floating-point number as a hexadecimal string. For finite floating-point numbers, this representation will always include a leading 0x and a trailing p and exponent.

float.fromhex(s)

Class method to return the float represented by a hexadecimal string s. The string s may have leading and trailing whitespace.

Note that float.hex is an instance method, while float.fromhex is a class method.

A hexadecimal string takes the form:

[sign] ['0x'] integer ['.' fraction] ['p' exponent]

where the optional sign may be either + or -, integer and fraction are strings of hexadecimal digits, and exponent is a decimal integer with an optional leading sign. Case is not significant, and there must be at least one hexadecimal digit in either the integer or the fraction. This syntax is similar to the syntax specified in section 6.4.4.2 of the C99 standard, and also to the syntax used in Java 1.5 onwards. In particular, the output of float.hex is usable as a hexadecimal floating-point literal in C or Java code, and hexadecimal strings produced by C's %a format character or Java's Double.toHexString are accepted by float.fromhex.

Note that the exponent is written in decimal rather than hexadecimal, and that it gives the power of 2 by which to multiply the coefficient. For example, the hexadecimal string 0x3.a7p10 represents the floating-point number (3 + 10./16 + 7./16**2) * 2.0**10, or 3740.0:

>>> float.fromhex('0x3.a7p10')
3740.0

Applying the reverse conversion to 3740.0 gives a different hexadecimal string representing the same number:

>>> float.hex(3740.0)
'0x1.d380000000000p+11'

Additional Methods on Complex

The !complex type implements the numbers.Complex abstract base class. !complex also has the following additional methods.

complex.from_number(x)

Class method to convert a number to a complex number.

For a general Python object x, complex.from_number(x) delegates to x.__complex__(). If ~object.__complex__ is not defined then it falls back to ~object.__float__. If !__float__ is not defined then it falls back to ~object.__index__.

3.14

Hashing of numeric types

For numbers x and y, possibly of different types, it's a requirement that hash(x) == hash(y) whenever x == y (see the ~object.__hash__ method documentation for more details). For ease of implementation and efficiency across a variety of numeric types (including int, float, decimal.Decimal and fractions.Fraction) Python's hash for numeric types is based on a single mathematical function that's defined for any rational number, and hence applies to all instances of int and fractions.Fraction, and all finite instances of float and decimal.Decimal. Essentially, this function is given by reduction modulo P for a fixed prime P. The value of P is made available to Python as the ~sys.hash_info.modulus attribute of sys.hash_info.

Currently, the prime used is P = 2**31 - 1 on machines with 32-bit C longs and P = 2**61 - 1 on machines with 64-bit C longs.

Here are the rules in detail:

If x = m / n is a nonnegative rational number and n is not divisible by P, define hash(x) as m * invmod(n, P) % P, where invmod(n, P) gives the inverse of n modulo P.
If x = m / n is a nonnegative rational number and n is divisible by P (but m is not) then n has no inverse modulo P and the rule above doesn't apply; in this case define hash(x) to be the constant value sys.hash_info.inf.
If x = m / n is a negative rational number define hash(x) as -hash(-x). If the resulting hash is -1, replace it with -2.
The particular values sys.hash_info.inf and -sys.hash_info.inf are used as hash values for positive infinity or negative infinity (respectively).
For a complex number z, the hash values of the real and imaginary parts are combined by computing hash(z.real) + sys.hash_info.imag * hash(z.imag), reduced modulo 2**sys.hash_info.width so that it lies in range(-2**(sys.hash_info.width - 1), 2**(sys.hash_info.width - 1)). Again, if the result is -1, it's replaced with -2.

To clarify the above rules, here's some example Python code, equivalent to the built-in hash, for computing the hash of a rational number, float, or complex:

import sys, math

def hash_fraction(m, n):
    """Compute the hash of a rational number m / n.

    Assumes m and n are integers, with n positive.
    Equivalent to hash(fractions.Fraction(m, n)).

    """
    P = sys.hash_info.modulus
    # Remove common factors of P.  (Unnecessary if m and n already coprime.)
    while m % P == n % P == 0:
        m, n = m // P, n // P

    if n % P == 0:
        hash_value = sys.hash_info.inf
    else:
        # Fermat's Little Theorem: pow(n, P-1, P) is 1, so
        # pow(n, P-2, P) gives the inverse of n modulo P.
        hash_value = (abs(m) % P) * pow(n, P - 2, P) % P
    if m < 0:
        hash_value = -hash_value
    if hash_value == -1:
        hash_value = -2
    return hash_value

def hash_float(x):
    """Compute the hash of a float x."""

    if math.isnan(x):
        return object.__hash__(x)
    elif math.isinf(x):
        return sys.hash_info.inf if x > 0 else -sys.hash_info.inf
    else:
        return hash_fraction(*x.as_integer_ratio())

def hash_complex(z):
    """Compute the hash of a complex number z."""

    hash_value = hash_float(z.real) + sys.hash_info.imag * hash_float(z.imag)
    # do a signed reduction modulo 2**sys.hash_info.width
    M = 2**(sys.hash_info.width - 1)
    hash_value = (hash_value & (M - 1)) - (hash_value & M)
    if hash_value == -1:
        hash_value = -2
    return hash_value

Boolean Type - `bool`

Booleans represent truth values. The bool type has exactly two constant instances: True and False.

single: False single: True pair: Boolean; values

The built-in function bool converts any value to a boolean, if the value can be interpreted as a truth value (see section truth above).

For logical operations, use the boolean operators <boolean> and, or and not. When applying the bitwise operators &, |, ^ to two booleans, they return a bool equivalent to the logical operations "and", "or", "xor". However, the logical operators and, or and != should be preferred over &, | and ^.

3.12

The use of the bitwise inversion operator ~ is deprecated and will raise an error in Python 3.16.

bool is a subclass of int (see typesnumeric). In many numeric contexts, False and True behave like the integers 0 and 1, respectively. However, relying on this is discouraged; explicitly convert using int instead.

single: iterator protocol single: protocol; iterator single: sequence; iteration single: container; iteration over

Python supports a concept of iteration over containers. This is implemented using two distinct methods; these are used to allow user-defined classes to support iteration. Sequences, described below in more detail, always support the iteration methods.

Iterables

One method needs to be defined for container objects to provide iterable support:

container.__iter__()

Return an iterator object. The object is required to support the iterator protocol described below. If a container supports different types of iteration, additional methods can be provided to specifically request iterators for those iteration types. (An example of an object supporting multiple forms of iteration would be a tree structure which supports both breadth-first and depth-first traversal.) This method corresponds to the ~PyTypeObject.tp_iter slot of the type structure for Python objects in the Python/C API.

Iterators

The iterator objects themselves are required to support the following two methods, which together form the iterator protocol:

iterator.__iter__()

Return the iterator object itself. This is required to allow both containers and iterators to be used with the for and in statements. This method corresponds to the ~PyTypeObject.tp_iter slot of the type structure for Python objects in the Python/C API.

iterator.__next__()

Return the next item from the iterator. If there are no further items, raise the StopIteration exception. This method corresponds to the ~PyTypeObject.tp_iternext slot of the type structure for Python objects in the Python/C API.

Python defines several iterator objects to support iteration over general and specific sequence types, dictionaries, and other more specialized forms. The specific types are not important beyond their implementation of the iterator protocol.

Once an iterator's ~iterator.__next__ method raises StopIteration, it must continue to do so on subsequent calls. Implementations that do not obey this property are deemed broken.

Generator types

Python's generators <generator> -- or more precisely, generator functions <generator function> and generator iterators <generator iterator> -- provide a convenient way to implement the iterator protocol.

A function that contains one or more yield expressions <yieldexpr> is a generator function. For example:

>>> def count_to_three():
...     yield 0
...     yield 1
...     yield 2
...     yield 3

Generator functions behave as regular user-defined functions <user-defined-funcs> (for example, they have the same attributes), except that calling a generator function returns a generator iterator <generator-methods>:

>>> count_to_three()
<generator object count_to_three at 0x7f33a2305000>

Iterating a generator iterator executes code of the underlying generator function, producing each yielded value in turn:

>>> for number in count_to_three():
...     print(number)
0
1
2
3

>>> list(count_to_three())
[0, 1, 2, 3]

One common use for generator functions is implementing the ~object.__iter__ method of custom iterable objects. For example:

>>> class CardDeck:
...     def __iter__(self):
...         yield 'three of clubs'
...         yield 'ace of hearts'

>>> list(CardDeck())
['three of clubs', 'ace of hearts']

pair: object; generator

Generator iterators

Generator iterators implement the iterator protocol <stdtypes-iterators>. Iterating them drives execution of the underlying generator function.

pair: exception; StopIteration

generator.__next__()

Starts the execution of a generator function or resumes it at the yield expression <yieldexpr> where the function is currently suspended. When a generator function is resumed with a ~generator.__next__ method, the current yield expression always evaluates to None. The execution then continues to the next yield expression, where the generator is suspended again, and the value of the expression after the yield keyword is returned to ~generator.__next__'s caller. If the generator exits without yielding another value, ~generator.__next__ raises a StopIteration exception, signalling that iteration has completed.

This method is normally called implicitly, for example by a for loop, or by the built-in next function.

Generator iterators have a few more methods than generic iterators, which can be used to control the execution of the underlying generator function:

generator.send(value)

"Sends" a value into the generator function: the value argument becomes the result of the current yield expression.

Otherwise, this method behaves like ~generator.__next__: it resumes the underlying function and either returns the next yielded value or raises StopIteration.

When send is called to start the generator, it must be called with None as the argument, because there is no current yield expression that could receive the value.

generator.throw(value) generator.throw(type[, value[, traceback]])

Raises an exception at the point where the generator is currently suspended.

Otherwise, this method behaves like ~generator.__next__: it resumes the underlying function and either returns the next yielded value or raises StopIteration. If the generator function does not catch the passed-in exception, or raises a different exception, then that exception propagates to the caller.

When throw is called to start the generator, the generator immediately exits (that is, subsequent calls to ~generator.__next__ will raise StopIteration) and the thrown exception is propagated to throw's caller.

In typical use, this is called with a single argument, an exception instance, similar to the way the raise keyword is used.

For backwards compatibility, however, the second signature is supported, following a convention from older versions of Python. The type argument should be an exception class, and value should be an exception instance. If the value is not provided, the type constructor is called to get an instance. If traceback is provided, it is set on the exception, otherwise any existing ~BaseException.__traceback__ attribute stored in value may be cleared.

3.12

The second signature (type[, value[, traceback]]) is deprecated and may be removed in a future version of Python.

pair: exception; GeneratorExit

generator.close()

Raises a GeneratorExit exception at the point where the generator function is currently suspended (equivalent to calling throw(GeneratorExit)).

If the generator function has already exited (due to an exception or normal return), or raises GeneratorExit (by not catching the exception), close returns None. If the generator yields a value, a RuntimeError is raised. If the generator raises any other exception, it is propagated to the caller. If a generator returns a value upon being closed, that value is returned by close.

When a generator iterator is garbage collected before it has exited, ~generator.close is called automatically.

3.13

If a generator returns a value upon being closed, the value is returned by close. Previously, it returned None.

Calling any of the generator methods (~generator.__next__, ~generator.send, ~generator.throw, ~generator.close) while one of these methods is already executing raises a ValueError exception.

pair: object; asynchronous-generator

Asynchronous generator iterators

This subsection describes the methods of an asynchronous generator iterator, which are used to control the execution of an asynchronous generator function.

pair: exception; StopAsyncIteration

agen.__anext__()

Returns an awaitable which when run starts to execute the asynchronous generator function or resumes it at the yield expression <yieldexpr> where the function is currently suspended. When an asynchronous generator function is resumed with an ~agen.__anext__ method, the current yield expression always evaluates to None in the returned awaitable, which when run will continue to the next yield expression. The value of the expression after the yield keyword is the value of the StopIteration exception raised by the completing coroutine. If the asynchronous generator exits without yielding another value, the awaitable instead raises a StopAsyncIteration exception, signalling that the asynchronous iteration has completed.

This method is normally called implicitly by an async for loop, or by the built-in anext function.

Asynchronous generator-iterators have a few more methods than generic asynchronous iterators, which can be used to control the execution of the underlying generator function:

agen.asend(value)

Returns an awaitable which, when run, "sends" a value into the underlying asynchronous generator function: the value argument becomes the result of the current yield expression.

Otherwise, this method behaves like ~agen.__anext__: when the returned awaitable runs, it resumes the underlying function and either returns the next yielded value as the value of the raised StopIteration, or raises StopAsyncIteration.

When asend is called to start the asynchronous generator, it must be called with None as the argument, because there is no yield expression that could receive the value.

agen.athrow(value) agen.athrow(type[, value[, traceback]])

Returns an awaitable that, when run, raises an exception at the point where the underlying asynchronous generator function is currently suspended.

Otherwise, this method behaves like ~agen.__anext__: when the returned awaitable runs, it resumes the underlying function (with an exception raised) and either returns the next yielded value as the value of the raised StopIteration, or raises StopAsyncIteration. If the underlying function does not catch the passed-in exception, or raises a different exception, then when the awaitable is run, that exception propagates to the caller of the awaitable.

When ~agen.athrow is called to start the generator, the generator exits when the awaitable runs (that is, subsequent results from ~agen.__anext__ will raise StopAsyncIteration when run) and the thrown exception is propagated to the awaitable's caller.

In typical use, this is called with a single argument, an exception instance, similar to the way the raise keyword is used.

For backwards compatibility, however, the second signature is supported. An exception instance is created from three arguments in the same way as in generator.throw.

3.12

The second signature (type[, value[, traceback]]) is deprecated and may be removed in a future version of Python.

pair: exception; GeneratorExit

agen.aclose()

Returns an awaitable that when run will throw a GeneratorExit into the underlying asynchronous generator function at the point where it is currently suspended (equivalent to calling athrow(GeneratorExit)).

If the asynchronous generator function then exits gracefully, is already closed, or raises GeneratorExit (by not catching the exception), then the returned awaitable will raise a StopIteration exception. Any further awaitables returned by subsequent calls to the asynchronous generator will raise a StopAsyncIteration exception.

If the asynchronous generator yields a value, a RuntimeError is raised by the awaitable. If the asynchronous generator raises any other exception, that exception is propagated to the caller of the awaitable.

If the asynchronous generator has already exited due to an exception or normal exit, then further calls to aclose will return an awaitable that does nothing.

Sequence Types --- `list`, `tuple`, `range`

There are three basic sequence types: lists, tuples, and range objects. Additional sequence types tailored for processing of binary data <binaryseq> and text strings <textseq> are described in dedicated sections.

Common Sequence Operations

pair: object; sequence

The operations in the following table are supported by most sequence types, both mutable and immutable. The collections.abc.Sequence ABC is provided to make it easier to correctly implement these operations on custom sequence types.

This table lists the sequence operations sorted in ascending priority. In the table, s and t are sequences of the same type, n, i, j and k are integers and x is an arbitrary object that meets any type and value restrictions imposed by s.

The in and not in operations have the same priorities as the comparison operations. The + (concatenation) and * (repetition) operations have the same priority as the corresponding numeric operations.[^3]

See time-complexity for the costs of the various sequence operations.

triple: operations on; sequence; types pair: built-in function; len pair: built-in function; min pair: built-in function; max pair: concatenation; operation pair: repetition; operation pair: subscript; operation pair: slice; operation pair: operator; in pair: operator; not in

Operation	Result	Notes
`x in s`	`True` if an item of s is equal to x, else `False`	(1)
`x not in s`	`False` if an item of s is equal to x, else `True`	(1)
`s + t`	the concatenation of s and t	(6)(7)
`s * n` or `n * s`	equivalent to adding s to itself n times	(2)(7)
`s[i]`	ith item of s, origin 0	(3)(8)
`s[i:j]`	slice of s from i to j	(3)(4)
`s[i:j:k]`	slice of s from i to j with step k	(3)(5)
`len(s)`	length of s
`min(s)`	smallest item of s
`max(s)`	largest item of s

Sequences of the same type also support comparisons. In particular, tuples and lists are compared lexicographically by comparing corresponding elements. This means that to compare equal, every element must compare equal and the two sequences must be of the same type and have the same length. (For full details see comparisons in the language reference.)

single: loop; over mutable sequence single: mutable sequence; loop over

Forward and reversed iterators over mutable sequences access values using an index. That index will continue to march forward (or backward) even if the underlying sequence is mutated. The iterator terminates only when an IndexError or a StopIteration is encountered (or when the index drops below zero).

Notes:

(1)
While the in and not in operations are used only for simple containment testing in the general case, some specialised sequences (such as str, bytes and bytearray) also use them for subsequence testing:

>>> "gg" in "eggs"
True

(2)
Values of n less than 0 are treated as 0 (which yields an empty sequence of the same type as s). Note that items in the sequence s are not copied; they are referenced multiple times. This often haunts new Python programmers; consider:

>>> lists = [[]] * 3
>>> lists
[[], [], []]
>>> lists[0].append(3)
>>> lists
[[3], [3], [3]]

What has happened is that [[]] is a one-element list containing an empty list, so all three elements of [[]] * 3 are references to this single empty list. Modifying any of the elements of lists modifies this single list. You can create a list of different lists this way:

>>> lists = [[] for i in range(3)]
>>> lists[0].append(3)
>>> lists[1].append(5)
>>> lists[2].append(7)
>>> lists
[[3], [5], [7]]

Further explanation is available in the FAQ entry faq-multidimensional-list.

(3)
If i or j is negative, the index is relative to the end of sequence s: len(s) + i or len(s) + j is substituted. But note that -0 is still 0.

(4)
The slice of s from i to j is defined as the sequence of items with index k such that i <= k < j.

If i is omitted or None, use 0.
If j is omitted or None, use len(s).
If i or j is less than -len(s), use 0.
If i or j is greater than len(s), use len(s).
If i is greater than or equal to j, the slice is empty.

(5)
The slice of s from i to j with step k is defined as the sequence of items with index x = i + n*k such that 0 <= n < (j-i)/k. In other words, the indices are i, i+k, i+2*k, i+3*k and so on, stopping when j is reached (but never including j). When k is positive, i and j are reduced to len(s) if they are greater. When k is negative, i and j are reduced to len(s) - 1 if they are greater. If i or j are omitted or None, they become "end" values (which end depends on the sign of k). Note, k cannot be zero. If k is None, it is treated like 1.

(6)
Concatenating immutable sequences always results in a new object. This means that building up a sequence by repeated concatenation will have a quadratic runtime cost in the total sequence length. To get a linear runtime cost, you must switch to one of the alternatives below:

if concatenating str objects, you can build a list and use str.join at the end or else write to an io.StringIO instance and retrieve its value when complete
if concatenating bytes objects, you can similarly use bytes.join or io.BytesIO, or you can do in-place concatenation with a bytearray object. bytearray objects are mutable and have an efficient overallocation mechanism
if concatenating tuple objects, extend a list instead
for other types, investigate the relevant class documentation

(7)
Some sequence types (such as range) only support item sequences that follow specific patterns, and hence don't support sequence concatenation or repetition.

(8)
An IndexError is raised if i is outside the sequence range.

Sequence Methods

Sequence types also support the following methods:

list.count(value, /) range.count(value, /) tuple.count(value, /)

sequence.count(value, /)

Return the total number of occurrences of value in sequence.

list.index(value[, start[, stop]]) range.index(value[, start[, stop]]) tuple.index(value[, start[, stop]])

sequence.index(value[, start[, stop]])

Return the index of the first occurrence of value in sequence.

Raises ValueError if value is not found in sequence.

The start or stop arguments allow for efficient searching of subsections of the sequence, beginning at start and ending at stop. This is roughly equivalent to start + sequence[start:stop].index(value), only without copying any data.

[!CAUTION] Not all sequence types support passing the start and stop arguments.

Immutable Sequence Types

triple: immutable; sequence; types pair: object; tuple pair: built-in function; hash

The only operation that immutable sequence types generally implement that is not also implemented by mutable sequence types is support for the hash built-in.

This support allows immutable sequences, such as tuple instances, to be used as dict keys and stored in set and frozenset instances.

Attempting to hash an immutable sequence that contains unhashable values will result in TypeError.

Mutable Sequence Types

triple: mutable; sequence; types pair: object; list pair: object; bytearray

The operations in the following table are defined on mutable sequence types. The collections.abc.MutableSequence ABC is provided to make it easier to correctly implement these operations on custom sequence types.

In the table s is an instance of a mutable sequence type, t is any iterable object and x is an arbitrary object that meets any type and value restrictions imposed by s (for example, bytearray only accepts integers that meet the value restriction 0 <= x <= 255).

triple: operations on; sequence; types triple: operations on; list; type pair: subscript; assignment pair: slice; assignment pair: statement; del

Operation	Result	Notes
`s[i] = x`	item i of s is replaced by x
`del s[i]`	removes item i of s
`s[i:j] = t`	slice of s from i to j is replaced by the contents of the iterable t
`del s[i:j]`	removes the elements of `s[i:j]` from the list (same as `s[i:j] = []`)
`s[i:j:k] = t`	the elements of `s[i:j:k]` are replaced by those of t	(1)
`del s[i:j:k]`	removes the elements of `s[i:j:k]` from the list
`s += t`	extends s with the contents of t (for the most part the same as `s[len(s):len(s)] = t`)
`s *= n`	updates s with its contents repeated n times	(2)

Notes:

(1)
If k is not equal to 1, t must have the same length as the slice it is replacing.

(2)
The value n is an integer, or an object implementing ~object.__index__. Zero and negative values of n clear the sequence. Items in the sequence are not copied; they are referenced multiple times, as explained for s * n under typesseq-common.

Mutable Sequence Methods

Mutable sequence types also support the following methods:

bytearray.append(value, /) list.append(value, /)

sequence.append(value, /)

Append value to the end of the sequence. This is equivalent to writing seq[len(seq):len(seq)] = [value].

bytearray.clear() list.clear()

sequence.clear()

3.3

Remove all items from sequence. This is equivalent to writing del sequence[:].

bytearray.copy() list.copy()

sequence.copy()

3.3

Create a shallow copy of sequence. This is equivalent to writing sequence[:].

Hint

The !copy method is not part of the ~collections.abc.MutableSequence ~abc.ABC, but most concrete mutable sequence types provide it.

bytearray.extend(iterable, /) list.extend(iterable, /)

sequence.extend(iterable, /)

Extend sequence with the contents of iterable. For the most part, this is the same as writing seq[len(seq):len(seq)] = iterable.

bytearray.insert(index, value, /) list.insert(index, value, /)

sequence.insert(index, value, /)

Insert value into sequence at the given index. This is equivalent to writing sequence[index:index] = [value].

bytearray.pop(index=-1, /) list.pop(index=-1, /)

sequence.pop(index=-1, /)

Retrieve the item at index and also remove it from sequence. By default, the last item in sequence is removed and returned.

bytearray.remove(value, /) list.remove(value, /)

sequence.remove(value, /)

Remove the first item from sequence where sequence[i] == value.

Raises ValueError if value is not found in sequence.

bytearray.reverse() list.reverse()

sequence.reverse()

Reverse the items of sequence in place. This method maintains economy of space when reversing a large sequence. To remind users that it operates by side-effect, it returns None.

Lists

pair: object; list

Lists are mutable sequences, typically used to store collections of homogeneous items (where the precise degree of similarity will vary by application).

Lists may be constructed in several ways:

Using a pair of square brackets to denote the empty list: []
Using square brackets, separating items with commas: [a], [a, b, c]
Using a list comprehension: [x for x in iterable]
Using the type constructor: list() or list(iterable)

The constructor builds a list whose items are the same and in the same order as iterable's items. iterable may be either a sequence, a container that supports iteration, or an iterator object. If iterable is already a list, a copy is made and returned, similar to iterable[:]. For example, list('abc') returns ['a', 'b', 'c'] and list( (1, 2, 3) ) returns [1, 2, 3]. If no argument is given, the constructor creates a new empty list, [].

Many other operations also produce lists, including the sorted built-in.

Lists are generic <generics> over the types of their items.

Lists implement all of the common <typesseq-common> and mutable <typesseq-mutable> sequence operations. Lists also provide the following additional method:

list.sort(*, key=None, reverse=False)

This method sorts the list in place, using only < comparisons between items. Exceptions are not suppressed - if any comparison operations fail, the entire sort operation will fail (and the list will likely be left in a partially modified state).

sort accepts two arguments that can only be passed by keyword (keyword-only arguments <keyword-only_parameter>):

key specifies a function of one argument that is used to extract a comparison key from each list element (for example, key=str.lower). The key corresponding to each item in the list is calculated once and then used for the entire sorting process. The default value of None means that list items are sorted directly without calculating a separate key value.

The functools.cmp_to_key utility is available to convert a 2.x style cmp function to a key function.

reverse is a boolean value. If set to True, then the list elements are sorted as if each comparison were reversed.

This method modifies the sequence in place for economy of space when sorting a large sequence. To remind users that it operates by side effect, it does not return the sorted sequence (use sorted to explicitly request a new sorted list instance).

The sort method is guaranteed to be stable. A sort is stable if it guarantees not to change the relative order of elements that compare equal --- this is helpful for sorting in multiple passes (for example, sort by department, then by salary grade).

For sorting examples and a brief sorting tutorial, see sortinghowto.

While a list is being sorted, the effect of attempting to mutate, or even inspect, the list is undefined. The C implementation of Python makes the list appear empty for the duration, and raises ValueError if it can detect that the list has been mutated during a sort.

For detailed information on thread-safety guarantees for list objects, see thread-safety-list.

Tuples

pair: object; tuple

Tuples are immutable sequences, typically used to store collections of heterogeneous data (such as the 2-tuples produced by the enumerate built-in). Tuples are also used for cases where an immutable sequence of homogeneous data is needed (such as allowing storage in a set or dict instance).

Tuples may be constructed in a number of ways:

Using a pair of parentheses to denote the empty tuple: ()
Using a trailing comma for a singleton tuple: a, or (a,)
Separating items with commas: a, b, c or (a, b, c)
Using the tuple built-in: tuple() or tuple(iterable)

The constructor builds a tuple whose items are the same and in the same order as iterable's items. iterable may be either a sequence, a container that supports iteration, or an iterator object. If iterable is already a tuple, it is returned unchanged. For example, tuple('abc') returns ('a', 'b', 'c') and tuple( [1, 2, 3] ) returns (1, 2, 3). If no argument is given, the constructor creates a new empty tuple, ().

Note that it is actually the comma which makes a tuple, not the parentheses. The parentheses are optional, except in the empty tuple case, or when they are needed to avoid syntactic ambiguity. For example, f(a, b, c) is a function call with three arguments, while f((a, b, c)) is a function call with a 3-tuple as the sole argument.

Tuples implement all of the common <typesseq-common> sequence operations.

Tuples are generic <generics> over the types of their contents. For more information, refer to the typing documentation on annotating tuples <annotating-tuples>.

For heterogeneous collections of data where access by name is clearer than access by index, collections.namedtuple may be a more appropriate choice than a simple tuple object.

Ranges

pair: object; range

The range type represents an immutable sequence of numbers and is commonly used for looping a specific number of times in for loops.

The arguments to the range constructor must be integers (either built-in int or any object that implements the ~object.__index__ special method). If the step argument is omitted, it defaults to 1. If the start argument is omitted, it defaults to 0. If step is zero, ValueError is raised.

For a positive step, the contents of a range r are determined by the formula r[i] = start + step*i where i >= 0 and r[i] < stop.

For a negative step, the contents of the range are still determined by the formula r[i] = start + step*i, but the constraints are i >= 0 and r[i] > stop.

A range object will be empty if r[0] does not meet the value constraint. Ranges do support negative indices, but these are interpreted as indexing from the end of the sequence determined by the positive indices.

Ranges containing absolute values larger than sys.maxsize are permitted but some features (such as len) may raise OverflowError.

Range examples:

>>> list(range(10))
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
>>> list(range(1, 11))
[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
>>> list(range(0, 30, 5))
[0, 5, 10, 15, 20, 25]
>>> list(range(0, 10, 3))
[0, 3, 6, 9]
>>> list(range(0, -10, -1))
[0, -1, -2, -3, -4, -5, -6, -7, -8, -9]
>>> list(range(0))
[]
>>> list(range(1, 0))
[]

Ranges implement all of the common <typesseq-common> sequence operations except concatenation and repetition (due to the fact that range objects can only represent sequences that follow a strict pattern and repetition and concatenation will usually violate that pattern).

start

The value of the start parameter (or 0 if the parameter was not supplied)

stop

The value of the stop parameter

step

The value of the step parameter (or 1 if the parameter was not supplied)

The advantage of the range type over a regular list or tuple is that a range object will always take the same (small) amount of memory, no matter the size of the range it represents (as it only stores the start, stop and step values, calculating individual items and subranges as needed).

Range objects implement the collections.abc.Sequence ABC, and provide features such as containment tests, element index lookup, slicing and support for negative indices (see typesseq):

>>> r = range(0, 20, 2) >>> r range(0, 20, 2) >>> 11 in r False >>> 10 in r True >>> r.index(10) 5 >>> r[5] 10 >>> r[:5] range(0, 10, 2) >>> r[-1] 18

Testing range objects for equality with == and != compares them as sequences. That is, two range objects are considered equal if they represent the same sequence of values. (Note that two range objects that compare equal might have different ~range.start, ~range.stop and ~range.step attributes, for example range(0) == range(2, 1, 3) or range(0, 3, 2) == range(0, 4, 2).)

3.2 Implement the Sequence ABC. Support slicing and negative indices. Test int objects for membership in constant time instead of iterating through all items.

3.3 Define '==' and '!=' to compare range objects based on the sequence of values they define (instead of comparing based on object identity).

Added the ~range.start, ~range.stop and ~range.step attributes.

The linspace recipe shows how to implement a lazy version of range suitable for floating-point applications.

single: string; text sequence type single: str (built-in class); (see also string) pair: object; string

Text and Binary Sequence Type Methods Summary

The following table summarizes the text and binary sequence types methods by category.

Category	`str` methods				`bytes` and `bytearray` methods
Formatting	`str.format`
	`str.format_map`
	`f-strings`
	`old-string-formatting`				`bytes-formatting`
Searching and Replacing	`str.find`	`str.rfind`			`bytes.find`	`bytes.rfind`
	`str.index`	`str.rindex`			`bytes.index`	`bytes.rindex`
	`str.startswith`				`bytes.startswith`
	`str.endswith`				`bytes.endswith`
	`str.count`				`bytes.count`
	`str.replace`				`bytes.replace`
Splitting and Joining	`str.split`		`str.rsplit`		`bytes.split`		`bytes.rsplit`
	`str.splitlines`				`bytes.splitlines`
	`str.partition`				`bytes.partition`
	`str.rpartition`				`bytes.rpartition`
	`str.join`				`bytes.join`
String Classification	`str.isalpha`				`bytes.isalpha`
	`str.isdecimal`
	`str.isdigit`				`bytes.isdigit`
	`str.isnumeric`
	`str.isalnum`				`bytes.isalnum`
	`str.isidentifier`
	`str.islower`				`bytes.islower`
	`str.isupper`				`bytes.isupper`
	`str.istitle`				`bytes.istitle`
	`str.isspace`				`bytes.isspace`
	`str.isprintable`
Case Manipulation	`str.lower`				`bytes.lower`
	`str.upper`				`bytes.upper`
	`str.casefold`
	`str.capitalize`				`bytes.capitalize`
	`str.title`				`bytes.title`
	`str.swapcase`				`bytes.swapcase`
Padding and Stripping	`str.ljust`		`str.rjust`		`bytes.ljust`		`bytes.rjust`
	`str.center`				`bytes.center`
	`str.expandtabs`				`bytes.expandtabs`
	`str.strip`				`bytes.strip`
	`str.lstrip`			`str.rstrip`	`bytes.lstrip`			`bytes.rstrip`
	`str.removeprefix`				`bytes.removeprefix`
	`str.removesuffix`				`bytes.removesuffix`
Translation and Encoding	`str.translate`				`bytes.translate`
	`str.maketrans`				`bytes.maketrans`
	`str.encode`
					`bytes.decode`

Text Sequence Type --- `str`

Textual data in Python is handled with str objects, or strings. Strings are immutable sequences <typesseq> of Unicode code points. String literals are written in a variety of ways:

Single quotes: 'allows embedded "double" quotes'
Double quotes: "allows embedded 'single' quotes"
Triple quoted: '''Three single quotes''', """Three double quotes"""

Triple quoted strings may span multiple lines - all associated whitespace will be included in the string literal.

String literals that are part of a single expression and have only whitespace between them will be implicitly converted to a single string literal. That is, ("spam " "eggs") == "spam eggs".

See strings for more about the various forms of string literal, including supported escape sequences <escape-sequences>, and the r ("raw") prefix that disables most escape sequence processing.

Strings may also be created from other objects using the str constructor.

Since there is no separate "character" type, indexing a string produces strings of length 1. That is, for a non-empty string s, s[0] == s[0:1].

pair: object; io.StringIO

There is also no mutable string type, but str.join or io.StringIO can be used to efficiently construct strings from multiple fragments.

3.3 For backwards compatibility with the Python 2 series, the u prefix is once again permitted on string literals. It has no effect on the meaning of string literals and cannot be combined with the r prefix.

single: string; str (built-in class)

Return a string <textseq> version of object. If object is not provided, returns the empty string. Otherwise, the behavior of str() depends on whether encoding or errors is given, as follows.

If neither encoding nor errors is given, str(object) returns type(object).__str__(object) <object.__str__>, which is the "informal" or nicely printable string representation of object. For string objects, this is the string itself. If object does not have a ~object.__str__ method, then str falls back to returning repr(object) <repr>.

single: buffer protocol; str (built-in class) single: bytes; str (built-in class)

If at least one of encoding or errors is given, object should be a bytes-like object (e.g. bytes or bytearray). In this case, if object is a bytes (or bytearray) object, then str(bytes, encoding, errors) is equivalent to bytes.decode(encoding, errors) <bytes.decode>. Otherwise, the bytes object underlying the buffer object is obtained before calling bytes.decode. See binaryseq and bufferobjects for information on buffer objects.

Passing a bytes object to str without the encoding or errors arguments falls under the first case of returning the informal string representation (see also the -b command-line option to Python). For example:

>>> str(b'Zoot!')
"b'Zoot!'"

For more information on the str class and its methods, see textseq and the string-methods section below. To output formatted strings, see the f-strings and formatstrings sections. In addition, see the stringservices section.

pair: string; methods

String Methods

pair: module; re

Strings implement all of the common <typesseq-common> sequence operations, along with the additional methods described below.

Strings also support two styles of string formatting, one providing a large degree of flexibility and customization (see str.format, formatstrings and string-formatting) and the other based on C printf style formatting that handles a narrower range of types and is slightly harder to use correctly, but is often faster for the cases it can handle (old-string-formatting).

The textservices section of the standard library covers a number of other modules that provide various text related utilities (including regular expression support in the re module).

str.capitalize()

Return a copy of the string with its first character capitalized and the rest lowercased.

3.8 The first character is now put into titlecase rather than uppercase. This means that characters like digraphs will only have their first letter capitalized, instead of the full character.

str.casefold()

Return a casefolded copy of the string. Casefolded strings may be used for caseless matching.

Casefolding is similar to lowercasing but more aggressive because it is intended to remove all case distinctions in a string. For example, the German lowercase letter 'ß' is equivalent to "ss". Since it is already lowercase, lower would do nothing to 'ß'; casefold converts it to "ss". For example:

>>> 'straße'.lower() 'straße' >>> 'straße'.casefold() 'strasse'

The casefolding algorithm is described in section 3.13.3 'Default Case Folding' of the Unicode Standard.

3.3

str.center(width, fillchar=' ', /)

Return centered in a string of length width. Padding is done using the specified fillchar (default is an ASCII space). The original string is returned if width is less than or equal to len(s). For example:

>>> 'Python'.center(10)
'  Python  '
>>> 'Python'.center(10, '-')
'--Python--'
>>> 'Python'.center(4)
'Python'

str.count(sub[, start[, end]])

Return the number of non-overlapping occurrences of substring sub in the range [start, end]. Optional arguments start and end are interpreted as in slice notation.

If sub is empty, returns the number of empty strings between characters which is the length of the string plus one. For example:

>>> 'spam, spam, spam'.count('spam')
3
>>> 'spam, spam, spam'.count('spam', 5)
2
>>> 'spam, spam, spam'.count('spam', 5, 10)
1
>>> 'spam, spam, spam'.count('eggs')
0
>>> 'spam, spam, spam'.count('')
17

str.encode(encoding="utf-8", errors="strict")

Return the string encoded to bytes.

encoding defaults to 'utf-8'; see standard-encodings for possible values.

errors controls how encoding errors are handled. If 'strict' (the default), a UnicodeError exception is raised. Other possible values are 'ignore', 'replace', 'xmlcharrefreplace', 'backslashreplace' and any other name registered via codecs.register_error. See error-handlers for details.

For performance reasons, the value of errors is not checked for validity unless an encoding error actually occurs, devmode is enabled or a debug build <debug-build> is used. For example:

>>> encoded_str_to_bytes = 'Python'.encode()
>>> type(encoded_str_to_bytes)
<class 'bytes'>
>>> encoded_str_to_bytes
b'Python'

3.1 Added support for keyword arguments.

3.9 The value of the errors argument is now checked in devmode and in debug mode <debug-build>.

str.endswith(suffix[, start[, end]])

Return True if the string ends with the specified suffix, otherwise return False. suffix can also be a tuple of suffixes to look for. With optional start, test beginning at that position. With optional end, stop comparing at that position. Using start and end is equivalent to str[start:end].endswith(suffix). For example:

>>> 'Python'.endswith('on')
True
>>> 'a tuple of suffixes'.endswith(('at', 'in'))
False
>>> 'a tuple of suffixes'.endswith(('at', 'es'))
True
>>> 'Python is amazing'.endswith('is', 0, 9)
True

See also startswith and removesuffix.

str.expandtabs(tabsize=8)

Return a copy of the string where all tab characters are replaced by one or more spaces, depending on the current column and the given tab size. Tab positions occur every tabsize characters (default is 8, giving tab positions at columns 0, 8, 16 and so on). To expand the string, the current column is set to zero and the string is examined character by character. If the character is a tab (\t), one or more space characters are inserted in the result until the current column is equal to the next tab position. (The tab character itself is not copied.) If the character is a newline (\n) or return (\r), it is copied and the current column is reset to zero. Any other character is copied unchanged and the current column is incremented by one regardless of how the character is represented when printed. For example:

>>> '01\t012\t0123\t01234'.expandtabs()
'01      012     0123    01234'
>>> '01\t012\t0123\t01234'.expandtabs(4)
'01  012 0123    01234'
>>> print('01\t012\n0123\t01234'.expandtabs(4))
01  012
0123    01234

str.find(sub[, start[, end]])

Return the lowest index in the string where substring sub is found within the slice s[start:end]. Optional arguments start and end are interpreted as in slice notation. Return -1 if sub is not found. For example:

>>> 'spam, spam, spam'.find('sp')
0
>>> 'spam, spam, spam'.find('sp', 5)
6

Representation	Description
`\n`	Line Feed
`\r`	Carriage Return
`\r\n`	Carriage Return + Line Feed
`\v` or `\x0b`	Line Tabulation
`\f` or `\x0c`	Form Feed
`\x1c`	File Separator
`\x1d`	Group Separator
`\x1e`	Record Separator
`\x85`	Next Line (C1 Control Code)
`\u2028`	Line Separator
`\u2029`	Paragraph Separator

Formatted String Literals (f-strings)

3.6

3.7 The await and async for can be used in expressions within f-strings.

3.8 Added the debug specifier (=)

3.12 Many restrictions on expressions within f-strings have been removed. Notably, nested strings, comments, and backslashes are now permitted.

An f-string (formally a formatted string literal) is a string literal that is prefixed with f or F. This type of string literal allows embedding the results of arbitrary Python expressions within replacement fields, which are delimited by curly brackets ({}). Each replacement field must contain an expression, optionally followed by:

a debug specifier -- an equal sign (=);
a conversion specifier -- !s, !r or !a; and/or
a format specifier prefixed with a colon (:).

See the Lexical Analysis section on f-strings <f-strings> for details on the syntax of these fields.

Debug specifier

3.8

If a debug specifier -- an equal sign (=) -- appears after the replacement field expression, the resulting f-string will contain the expression's source, the equal sign, and the value of the expression. This is often useful for debugging:

>>> number = 14.3
>>> f'{number=}'
'number=14.3'

Whitespace before, inside and after the expression, as well as whitespace after the equal sign, is significant --- it is retained in the result:

>>> f'{ number  -  4  = }'
' number  -  4  = 10.3'

Conversion specifier

By default, the value of a replacement field expression is converted to a string using str:

>>> from fractions import Fraction
>>> one_third = Fraction(1, 3)
>>> f'{one_third}'
'1/3'

When a debug specifier but no format specifier is used, the default conversion instead uses repr:

>>> f'{one_third = }'
'one_third = Fraction(1, 3)'

The conversion can be specified explicitly using one of these specifiers:

!s for str
!r for repr
!a for ascii

For example:

>>> str(one_third)
'1/3'
>>> repr(one_third)
'Fraction(1, 3)'

>>> f'{one_third!s} is {one_third!r}'
'1/3 is Fraction(1, 3)'

>>> string = "¡kočka 😸!"
>>> ascii(string)
"'\\xa1ko\\u010dka \\U0001f638!'"

>>> f'{string = !a}'
"string = '\\xa1ko\\u010dka \\U0001f638!'"

Format specifier

After the expression has been evaluated, and possibly converted using an explicit conversion specifier, it is formatted using the format function. If the replacement field includes a format specifier introduced by a colon (:), the specifier is passed to !format as the second argument. The result of !format is then used as the final value for the replacement field. For example:

>>> from fractions import Fraction
>>> one_third = Fraction(1, 3)
>>> f'{one_third:.6f}'
'0.333333'
>>> f'{one_third:_^+10}'
'___+1/3___'
>>> f'{one_third!r:_^20}'
'___Fraction(1, 3)___'
>>> f'{one_third = :~>10}~'
'one_third = ~~~~~~~1/3~'

Template String Literals (t-strings)

A t-string (formally a template string literal) is a string literal that is prefixed with t or T.

These strings follow the same syntax and evaluation rules as formatted string literals <stdtypes-fstrings>, with the following differences:

Rather than evaluating to a str object, template string literals evaluate to a string.templatelib.Template object.
The format protocol is not used. Instead, the format specifier and conversions (if any) are passed to a new ~string.templatelib.Interpolation object that is created for each evaluated expression. It is up to code that processes the resulting ~string.templatelib.Template object to decide how to handle format specifiers and conversions.
Format specifiers containing nested replacement fields are evaluated eagerly, prior to being passed to the ~string.templatelib.Interpolation object. For instance, an interpolation of the form {amount:.{precision}f} will evaluate the inner expression {precision} to determine the value of the format_spec attribute. If precision were to be 2, the resulting format specifier would be '.2f'.
When the equals sign '=' is provided in an interpolation expression, the text of the expression is appended to the literal string that precedes the relevant interpolation. This includes the equals sign and any surrounding whitespace. The !Interpolation instance for the expression will be created as normal, except that ~string.templatelib.Interpolation.conversion will be set to 'r' (repr) by default. If an explicit conversion or format specifier is provided, this will override the default behaviour.

`printf`-style String Formatting

single: formatting, string (%) single: interpolation, string (%) single: string; formatting, printf single: string; interpolation, printf single: printf-style formatting single: sprintf-style formatting single: % (percent); printf-style formatting

[!NOTE] The formatting operations described here exhibit a variety of quirks that lead to a number of common errors (such as failing to display tuples and dictionaries correctly).

Using formatted string literals <f-strings>, the str.format interface, or string.Template may help avoid these errors. Each of these alternatives provides their own trade-offs and benefits of simplicity, flexibility, and/or extensibility.

String objects have one unique built-in operation: the % operator (modulo). This is also known as the string formatting or interpolation operator. Given format % values (where format is a string), % conversion specifications in format are replaced with zero or more elements of values. The effect is similar to using the sprintf function in the C language. For example:

>>> print('%s has %d quote types.' % ('Python', 2)) Python has 2 quote types.

If format requires a single argument, values may be a single non-tuple object.[^8] Otherwise, values must be a tuple with exactly the number of items specified by the format string, or a single mapping object (for example, a dictionary).

single: () (parentheses); in printf-style formatting single: * (asterisk); in printf-style formatting single: . (dot); in printf-style formatting

A conversion specifier contains two or more characters and has the following components, which must occur in this order:

The '%' character, which marks the start of the specifier.
Mapping key (optional), consisting of a parenthesised sequence of characters (for example, (somename)).
Conversion flags (optional), which affect the result of some conversion types.
Minimum field width (optional). If specified as an '*' (asterisk), the actual width is read from the next element of the tuple in values, and the object to convert comes after the minimum field width and optional precision.
Precision (optional), given as a '.' (dot) followed by the precision. If specified as '*' (an asterisk), the actual precision is read from the next element of the tuple in values, and the value to convert comes after the precision.
Length modifier (optional).
Conversion type.

When the right argument is a dictionary (or other mapping type), then the formats in the string must include a parenthesised mapping key into that dictionary inserted immediately after the '%' character. The mapping key selects the value to be formatted from the mapping. For example:

>>> print('%(language)s has %(number)03d quote types.' % ... {'language': "Python", "number": 2}) Python has 002 quote types.

In this case no * specifiers may occur in a format (since they require a sequential parameter list).

The conversion flag characters are:

single: # (hash); in printf-style formatting single: - (minus); in printf-style formatting single: + (plus); in printf-style formatting single: space; in printf-style formatting

Flag	Meaning
`'#'`	The value conversion will use the "alternate form" (where defined below).
`'0'`	The conversion will be zero padded for numeric values.
`'-'`	The converted value is left adjusted (overrides the `'0'` conversion if both are given).
`' '`	(a space) A blank should be left before a positive number (or empty string) produced by a signed conversion.
`'+'`	A sign character (`'+'` or `'-'`) will precede the conversion (overrides a "space" flag).

A length modifier (h, l, or L) may be present, but is ignored as it is not necessary for Python -- so e.g. %ld is identical to %d.

The conversion types are:

Conversion	Meaning	Notes
`'d'`	Signed integer decimal.
`'i'`	Signed integer decimal.
`'o'`	Signed octal value.	(1)
`'u'`	Obsolete type -- it is identical to `'d'`.	(6)
`'x'`	Signed hexadecimal (lowercase).	(2)
`'X'`	Signed hexadecimal (uppercase).	(2)
`'e'`	Floating-point exponential format (lowercase).	(3)
`'E'`	Floating-point exponential format (uppercase).	(3)
`'f'`	Floating-point decimal format.	(3)
`'F'`	Floating-point decimal format.	(3)
`'g'`	Floating-point format. Uses lowercase exponential format if exponent is less than -4 or not less than precision, decimal format otherwise.	(4)
`'G'`	Floating-point format. Uses uppercase exponential format if exponent is less than -4 or not less than precision, decimal format otherwise.	(4)
`'c'`	Single character (accepts integer or single character string).
`'r'`	String (converts any Python object using `repr`).	(5)
`'s'`	String (converts any Python object using `str`).	(5)
`'a'`	String (converts any Python object using `ascii`).	(5)
`'%'`	No argument is converted, results in a `'%'` character in the result.

For floating-point formats, the result should be correctly rounded to a given precision p of digits after the decimal point. The rounding mode matches that of the round builtin.

Notes:

(1)
The alternate form causes a leading octal specifier ('0o') to be inserted before the first digit.

(2)
The alternate form causes a leading '0x' or '0X' (depending on whether the 'x' or 'X' format was used) to be inserted before the first digit.

(3)
The alternate form causes the result to always contain a decimal point, even if no digits follow it.

The precision determines the number of digits after the decimal point and defaults to 6.

(4)
The alternate form causes the result to always contain a decimal point, and trailing zeroes are not removed as they would otherwise be.

The precision determines the number of significant digits before and after the decimal point and defaults to 6.

(5)
If precision is N, the output is truncated to N characters.

(6)
See 237.

Since Python strings have an explicit length, %s conversions do not assume that '\0' is the end of the string.

3.1 %f conversions for numbers whose absolute value is over 1e50 are no longer replaced by %g conversions.

single: buffer protocol; binary sequence types

Binary Sequence Types --- `bytes`, `bytearray`, `memoryview`

pair: object; bytes pair: object; bytearray pair: object; memoryview pair: module; array

The core built-in types for manipulating binary data are bytes and bytearray. They are supported by memoryview which uses the buffer protocol <bufferobjects> to access the memory of other binary objects without needing to make a copy.

The array module supports efficient storage of basic data types like 32-bit integers and IEEE754 double-precision floating values.

Bytes Objects

pair: object; bytes

Bytes objects are immutable sequences of single bytes. Since many major binary protocols are based on the ASCII text encoding, bytes objects offer several methods that are only valid when working with ASCII compatible data and are closely related to string objects in a variety of other ways.

Firstly, the syntax for bytes literals is largely the same as that for string literals, except that a b prefix is added:

Single quotes: b'still allows embedded "double" quotes'
Double quotes: b"still allows embedded 'single' quotes"
Triple quoted: b'''3 single quotes''', b"""3 double quotes"""

Only ASCII characters are permitted in bytes literals (regardless of the declared source code encoding). Any binary values over 127 must be entered into bytes literals using the appropriate escape sequence.

As with string literals, bytes literals may also use a r prefix to disable processing of escape sequences. See strings for more about the various forms of bytes literal, including supported escape sequences.

While bytes literals and representations are based on ASCII text, bytes objects actually behave like immutable sequences of integers, with each value in the sequence restricted such that 0 <= x < 256 (attempts to violate this restriction will trigger ValueError). This is done deliberately to emphasise that while many binary formats include ASCII based elements and can be usefully manipulated with some text-oriented algorithms, this is not generally the case for arbitrary binary data (blindly applying text processing algorithms to binary data formats that are not ASCII compatible will usually lead to data corruption).

In addition to the literal forms, bytes objects can be created in a number of other ways:

A zero-filled bytes object of a specified length: bytes(10)
From an iterable of integers: bytes(range(20))
Copying existing binary data via the buffer protocol: bytes(obj)

Also see the bytes <func-bytes> built-in.

Since 2 hexadecimal digits correspond precisely to a single byte, hexadecimal numbers are a commonly used format for describing binary data. Accordingly, the bytes type has an additional class method to read data in that format:

fromhex(string, /)

This bytes class method returns a bytes object, decoding the given string object. The string must contain two hexadecimal digits per byte, with ASCII whitespace being ignored.

>>> bytes.fromhex('2Ef0 F1f2 ') b'.xf0xf1xf2'

3.7 bytes.fromhex now skips all ASCII whitespace in the string, not just spaces.

3.14 bytes.fromhex now accepts ASCII bytes and bytes-like objects <bytes-like object> as input.

A reverse conversion function exists to transform a bytes object into its hexadecimal representation.

hex(*, bytes_per_sep=1) hex(sep, bytes_per_sep=1)

Return a string object containing two hexadecimal digits for each byte in the instance.

>>> b'xf0xf1xf2'.hex() 'f0f1f2'

If you want to make the hex string easier to read, you can specify a single character separator sep parameter to include in the output. By default, this separator will be included between each byte. A second optional bytes_per_sep parameter controls the spacing. Positive values calculate the separator position from the right, negative values from the left.

>>> value = b'xf0xf1xf2' >>> value.hex('-') 'f0-f1-f2' >>> value.hex('_', 2) 'f0_f1f2' >>> b'UUDDLRLRAB'.hex(' ', -4) '55554444 4c524c52 4142'

3.5

3.8 bytes.hex now supports optional sep and bytes_per_sep parameters to insert separators between bytes in the hex output.

Since bytes objects are sequences of integers (akin to a tuple), for a bytes object b, b[0] will be an integer, while b[0:1] will be a bytes object of length 1. (This contrasts with text strings, where both indexing and slicing will produce a string of length 1)

The representation of bytes objects uses the literal format (b'...') since it is often more useful than e.g. bytes([46, 46, 46]). You can always convert a bytes object into a list of integers using list(b).

Bytearray Objects

pair: object; bytearray

bytearray objects are a mutable counterpart to bytes objects.

There is no dedicated literal syntax for bytearray objects, instead they are always created by calling the constructor:

Creating an empty instance: bytearray()
Creating a zero-filled instance with a given length: bytearray(10)
From an iterable of integers: bytearray(range(20))
Copying existing binary data via the buffer protocol: bytearray(b'Hi!')

As bytearray objects are mutable, they support the mutable <typesseq-mutable> sequence operations in addition to the common bytes and bytearray operations described in bytes-methods.

Also see the bytearray <func-bytearray> built-in.

Since 2 hexadecimal digits correspond precisely to a single byte, hexadecimal numbers are a commonly used format for describing binary data. Accordingly, the bytearray type has an additional class method to read data in that format:

fromhex(string, /)

This bytearray class method returns a bytearray object, decoding the given string object. The string must contain two hexadecimal digits per byte, with ASCII whitespace being ignored.

>>> bytearray.fromhex('2Ef0 F1f2 ') bytearray(b'.xf0xf1xf2')

3.7 bytearray.fromhex now skips all ASCII whitespace in the string, not just spaces.

3.14 bytearray.fromhex now accepts ASCII bytes and bytes-like objects <bytes-like object> as input.

A reverse conversion function exists to transform a bytearray object into its hexadecimal representation.

hex(*, bytes_per_sep=1) hex(sep, bytes_per_sep=1)

Return a string object containing two hexadecimal digits for each byte in the instance.

>>> bytearray(b'xf0xf1xf2').hex() 'f0f1f2'

3.5

3.8 Similar to bytes.hex, bytearray.hex now supports optional sep and bytes_per_sep parameters to insert separators between bytes in the hex output.

resize(size, /)

Resize the bytearray to contain size bytes. size must be greater than or equal to 0.

If the bytearray needs to shrink, bytes beyond size are truncated.

If the bytearray needs to grow, all new bytes will be set to null bytes.

This is equivalent to:

>>> def resize(ba, size): ... if len(ba) > size: ... del ba[size:] ... else: ... ba += b'0' * (size - len(ba))

Examples:

>>> shrink = bytearray(b'abc') >>> shrink.resize(1) >>> (shrink, len(shrink)) (bytearray(b'a'), 1) >>> grow = bytearray(b'abc') >>> grow.resize(5) >>> (grow, len(grow)) (bytearray(b'abcx00x00'), 5)

3.14

take_bytes(n=None, /)

Remove the first n bytes from the bytearray and return them as an immutable bytes. By default (if n is None), return all bytes and clear the bytearray.

If n is negative, index from the end and take the first len plus n bytes. If n is out of bounds, raise IndexError.

Taking less than the full length will leave remaining bytes in the bytearray, which requires a copy. If the remaining bytes should be discarded, use ~bytearray.resize or del to truncate then ~bytearray.take_bytes without a size.

Taking all bytes is a zero-copy operation.

3.15

See the What's New <whatsnew315-bytearray-take-bytes> entry for common code patterns which can be optimized with bytearray.take_bytes.

Since bytearray objects are sequences of integers (akin to a list), for a bytearray object b, b[0] will be an integer, while b[0:1] will be a bytearray object of length 1. (This contrasts with text strings, where both indexing and slicing will produce a string of length 1)

The representation of bytearray objects uses the bytes literal format (bytearray(b'...')) since it is often more useful than e.g. bytearray([46, 46, 46]). You can always convert a bytearray object into a list of integers using list(b).

For detailed information on thread-safety guarantees for bytearray objects, see thread-safety-bytearray.

Bytes and Bytearray Operations

pair: bytes; methods pair: bytearray; methods

Both bytes and bytearray objects support the common <typesseq-common> sequence operations. They interoperate not just with operands of the same type, but with any bytes-like object. Due to this flexibility, they can be freely mixed in operations without causing errors. However, the return type of the result may depend on the order of operands.

[!NOTE] The methods on bytes and bytearray objects don't accept strings as their arguments, just as the methods on strings don't accept bytes as their arguments. For example, you have to write:
a = "abc"
b = a.replace("a", "f")
and:
a = b"abc"
b = a.replace(b"a", b"f")

Some bytes and bytearray operations assume the use of ASCII compatible binary formats, and hence should be avoided when working with arbitrary binary data. These restrictions are covered below.

[!NOTE] Using these ASCII based operations to manipulate binary data that is not stored in an ASCII based format may lead to data corruption.

The following methods on bytes and bytearray objects can be used with arbitrary binary data.

bytes.count(sub[, start[, end]]) bytearray.count(sub[, start[, end]])

Return the number of non-overlapping occurrences of subsequence sub in the range [start, end]. Optional arguments start and end are interpreted as in slice notation.

The subsequence to search for may be any bytes-like object or an integer in the range 0 to 255.

If sub is empty, returns the number of empty slices between characters which is the length of the bytes object plus one.

3.3 Also accept an integer in the range 0 to 255 as the subsequence.

bytes.removeprefix(prefix, /) bytearray.removeprefix(prefix, /)

If the binary data starts with the prefix string, return bytes[len(prefix):]. Otherwise, return a copy of the original binary data:

>>> b'TestHook'.removeprefix(b'Test')
b'Hook'
>>> b'BaseTestCase'.removeprefix(b'Test')
b'BaseTestCase'

The prefix may be any bytes-like object.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

3.9

bytes.removesuffix(suffix, /) bytearray.removesuffix(suffix, /)

If the binary data ends with the suffix string and that suffix is not empty, return bytes[:-len(suffix)]. Otherwise, return a copy of the original binary data:

>>> b'MiscTests'.removesuffix(b'Tests')
b'Misc'
>>> b'TmpDirMixin'.removesuffix(b'Tests')
b'TmpDirMixin'

The suffix may be any bytes-like object.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

3.9

bytes.decode(encoding="utf-8", errors="strict") bytearray.decode(encoding="utf-8", errors="strict")

Return the bytes decoded to a str.

encoding defaults to 'utf-8'; see standard-encodings for possible values.

errors controls how decoding errors are handled. If 'strict' (the default), a UnicodeError exception is raised. Other possible values are 'ignore', 'replace', and any other name registered via codecs.register_error. See error-handlers for details.

For performance reasons, the value of errors is not checked for validity unless a decoding error actually occurs, devmode is enabled or a debug build <debug-build> is used.

[!NOTE] Passing the encoding argument to str allows decoding any bytes-like object directly, without needing to make a temporary !bytes or !bytearray object.

3.1 Added support for keyword arguments.

3.9 The value of the errors argument is now checked in devmode and in debug mode <debug-build>.

bytes.endswith(suffix[, start[, end]]) bytearray.endswith(suffix[, start[, end]])

Return True if the binary data ends with the specified suffix, otherwise return False. suffix can also be a tuple of suffixes to look for. With optional start, test beginning at that position. With optional end, stop comparing at that position.

The suffix(es) to search for may be any bytes-like object.

bytes.find(sub[, start[, end]]) bytearray.find(sub[, start[, end]])

Return the lowest index in the data where the subsequence sub is found, such that sub is contained in the slice s[start:end]. Optional arguments start and end are interpreted as in slice notation. Return -1 if sub is not found.

The subsequence to search for may be any bytes-like object or an integer in the range 0 to 255.

[!NOTE] The ~bytes.find method should be used only if you need to know the position of sub. To check if sub is a substring or not, use the in operator:
>>> b'Py' in b'Python'
True

3.3 Also accept an integer in the range 0 to 255 as the subsequence.

bytes.index(sub[, start[, end]]) bytearray.index(sub[, start[, end]])

Like ~bytes.find, but raise ValueError when the subsequence is not found.

The subsequence to search for may be any bytes-like object or an integer in the range 0 to 255.

3.3 Also accept an integer in the range 0 to 255 as the subsequence.

bytes.join(iterable, /) bytearray.join(iterable, /)

Return a bytes or bytearray object which is the concatenation of the binary data sequences in iterable. A TypeError will be raised if there are any values in iterable that are not bytes-like objects <bytes-like object>, including str objects. The separator between elements is the contents of the bytes or bytearray object providing this method.

bytes.maketrans(from, to, /) bytearray.maketrans(from, to, /)

This static method returns a translation table usable for bytes.translate that will map each character in from into the character at the same position in to; from and to must both be bytes-like objects <bytes-like object> and have the same length.

3.1

bytes.partition(sep, /) bytearray.partition(sep, /)

Split the sequence at the first occurrence of sep, and return a 3-tuple containing the part before the separator, the separator itself or its bytearray copy, and the part after the separator. If the separator is not found, return a 3-tuple containing a copy of the original sequence, followed by two empty bytes or bytearray objects.

The separator to search for may be any bytes-like object.

bytes.replace(old, new, /, count=-1) bytearray.replace(old, new, /, count=-1)

Return a copy of the sequence with all occurrences of subsequence old replaced by new. If count is given, only the first count occurrences are replaced. If count is not specified or -1, then all occurrences are replaced.

The subsequence to search for and its replacement may be any bytes-like object.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

3.15 count is now supported as a keyword argument.

bytes.rfind(sub[, start[, end]]) bytearray.rfind(sub[, start[, end]])

Return the highest index in the sequence where the subsequence sub is found, such that sub is contained within s[start:end]. Optional arguments start and end are interpreted as in slice notation. Return -1 on failure.

The subsequence to search for may be any bytes-like object or an integer in the range 0 to 255.

3.3 Also accept an integer in the range 0 to 255 as the subsequence.

bytes.rindex(sub[, start[, end]]) bytearray.rindex(sub[, start[, end]])

Like ~bytes.rfind but raises ValueError when the subsequence sub is not found.

The subsequence to search for may be any bytes-like object or an integer in the range 0 to 255.

3.3 Also accept an integer in the range 0 to 255 as the subsequence.

bytes.rpartition(sep, /) bytearray.rpartition(sep, /)

Split the sequence at the last occurrence of sep, and return a 3-tuple containing the part before the separator, the separator itself or its bytearray copy, and the part after the separator. If the separator is not found, return a 3-tuple containing two empty bytes or bytearray objects, followed by a copy of the original sequence.

The separator to search for may be any bytes-like object.

bytes.startswith(prefix[, start[, end]]) bytearray.startswith(prefix[, start[, end]])

Return True if the binary data starts with the specified prefix, otherwise return False. prefix can also be a tuple of prefixes to look for. With optional start, test beginning at that position. With optional end, stop comparing at that position.

The prefix(es) to search for may be any bytes-like object.

bytes.translate(table, /, delete=b'') bytearray.translate(table, /, delete=b'')

Return a copy of the bytes or bytearray object where all bytes occurring in the optional argument delete are removed, and the remaining bytes have been mapped through the given translation table, which must be a bytes object of length 256.

You can use the bytes.maketrans method to create a translation table.

Set the table argument to None for translations that only delete characters:

>>> b'read this short text'.translate(None, b'aeiou')
b'rd ths shrt txt'

3.6 delete is now supported as a keyword argument.

The following methods on bytes and bytearray objects have default behaviours that assume the use of ASCII compatible binary formats, but can still be used with arbitrary binary data by passing appropriate arguments. Note that all of the bytearray methods in this section do not operate in place, and instead produce new objects.

bytes.center(width, fillbyte=b' ', /) bytearray.center(width, fillbyte=b' ', /)

Return a copy of the object centered in a sequence of length width. Padding is done using the specified fillbyte (default is an ASCII space). For bytes objects, the original sequence is returned if width is less than or equal to len(s).

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.ljust(width, fillbyte=b' ', /) bytearray.ljust(width, fillbyte=b' ', /)

Return a copy of the object left justified in a sequence of length width. Padding is done using the specified fillbyte (default is an ASCII space). For bytes objects, the original sequence is returned if width is less than or equal to len(s).

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.lstrip(bytes=None, /) bytearray.lstrip(bytes=None, /)

Return a copy of the sequence with specified leading bytes removed. The bytes argument is a binary sequence specifying the set of byte values to be removed. If omitted or None, the bytes argument defaults to removing ASCII whitespace <bytes.isspace>. The bytes argument is not a prefix; rather, all combinations of its values are stripped:

>>> b'   spacious   '.lstrip()
b'spacious   '
>>> b'www.example.com'.lstrip(b'cmowz.')
b'example.com'

The binary sequence of byte values to remove may be any bytes-like object. See ~bytes.removeprefix for a method that will remove a single prefix string rather than all of a set of characters. For example:

>>> b'Arthur: three!'.lstrip(b'Arthur: ')
b'ee!'
>>> b'Arthur: three!'.removeprefix(b'Arthur: ')
b'three!'

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.rjust(width, fillbyte=b' ', /) bytearray.rjust(width, fillbyte=b' ', /)

Return a copy of the object right justified in a sequence of length width. Padding is done using the specified fillbyte (default is an ASCII space). For bytes objects, the original sequence is returned if width is less than or equal to len(s).

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.rsplit(sep=None, maxsplit=-1) bytearray.rsplit(sep=None, maxsplit=-1)

Split the binary sequence into subsequences of the same type, using sep as the delimiter string. If maxsplit is given, at most maxsplit splits are done, the rightmost ones. If sep is not specified or None, any subsequence consisting solely of ASCII whitespace <bytes.isspace> is a separator. Except for splitting from the right, rsplit behaves like split which is described in detail below.

bytes.rstrip(bytes=None, /) bytearray.rstrip(bytes=None, /)

Return a copy of the sequence with specified trailing bytes removed. The bytes argument is a binary sequence specifying the set of byte values to be removed. If omitted or None, the bytes argument defaults to removing ASCII whitespace <bytes.isspace>. The bytes argument is not a suffix; rather, all combinations of its values are stripped:

>>> b'   spacious   '.rstrip()
b'   spacious'
>>> b'mississippi'.rstrip(b'ipz')
b'mississ'

The binary sequence of byte values to remove may be any bytes-like object. See ~bytes.removesuffix for a method that will remove a single suffix string rather than all of a set of characters. For example:

>>> b'Monty Python'.rstrip(b' Python')
b'M'
>>> b'Monty Python'.removesuffix(b' Python')
b'Monty'

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.split(sep=None, maxsplit=-1) bytearray.split(sep=None, maxsplit=-1)

Split the binary sequence into subsequences of the same type, using sep as the delimiter string. If maxsplit is given and non-negative, at most maxsplit splits are done (thus, the list will have at most maxsplit+1 elements). If maxsplit is not specified or is -1, then there is no limit on the number of splits (all possible splits are made).

If sep is given, consecutive delimiters are not grouped together and are deemed to delimit empty subsequences (for example, b'1,,2'.split(b',') returns [b'1', b'', b'2']). The sep argument may consist of a multibyte sequence as a single delimiter. Splitting an empty sequence with a specified separator returns [b''] or [bytearray(b'')] depending on the type of object being split. The sep argument may be any bytes-like object.

For example:

>>> b'1,2,3'.split(b',')
[b'1', b'2', b'3']
>>> b'1,2,3'.split(b',', maxsplit=1)
[b'1', b'2,3']
>>> b'1,2,,3,'.split(b',')
[b'1', b'2', b'', b'3', b'']
>>> b'1<>2<>3<4'.split(b'<>')
[b'1', b'2', b'3<4']

If sep is not specified or is None, a different splitting algorithm is applied: runs of consecutive ASCII whitespace <bytes.isspace> are regarded as a single separator, and the result will contain no empty strings at the start or end if the sequence has leading or trailing whitespace. Consequently, splitting an empty sequence or a sequence consisting solely of ASCII whitespace without a specified separator returns [].

For example:

>>> b'1 2 3'.split()
[b'1', b'2', b'3']
>>> b'1 2 3'.split(maxsplit=1)
[b'1', b'2 3']
>>> b'   1   2   3   '.split()
[b'1', b'2', b'3']

bytes.strip(bytes=None, /) bytearray.strip(bytes=None, /)

Return a copy of the sequence with specified leading and trailing bytes removed. The bytes argument is a binary sequence specifying the set of byte values to be removed. If omitted or None, the bytes argument defaults to removing ASCII whitespace <bytes.isspace>. The bytes argument is not a prefix or suffix; rather, all combinations of its values are stripped:

>>> b'   spacious   '.strip()
b'spacious'
>>> b'www.example.com'.strip(b'cmowz.')
b'example'

The binary sequence of byte values to remove may be any bytes-like object.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

The following methods on bytes and bytearray objects assume the use of ASCII compatible binary formats and should not be applied to arbitrary binary data. Note that all of the bytearray methods in this section do not operate in place, and instead produce new objects.

bytes.capitalize() bytearray.capitalize()

Return a copy of the sequence with each byte interpreted as an ASCII character, and the first byte capitalized and the rest lowercased. Non-ASCII byte values are passed through unchanged.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.expandtabs(tabsize=8) bytearray.expandtabs(tabsize=8)

Return a copy of the sequence where all ASCII tab characters are replaced by one or more ASCII spaces, depending on the current column and the given tab size. Tab positions occur every tabsize bytes (default is 8, giving tab positions at columns 0, 8, 16 and so on). To expand the sequence, the current column is set to zero and the sequence is examined byte by byte. If the byte is an ASCII tab character (b'\t'), one or more space characters are inserted in the result until the current column is equal to the next tab position. (The tab character itself is not copied.) If the current byte is an ASCII newline (b'\n') or carriage return (b'\r'), it is copied and the current column is reset to zero. Any other byte value is copied unchanged and the current column is incremented by one regardless of how the byte value is represented when printed:

>>> b'01\t012\t0123\t01234'.expandtabs()
b'01      012     0123    01234'
>>> b'01\t012\t0123\t01234'.expandtabs(4)
b'01  012 0123    01234'

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.isalnum() bytearray.isalnum()

Return True if all bytes in the sequence are alphabetical ASCII characters or ASCII decimal digits and the sequence is not empty, False otherwise. Alphabetic ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'. ASCII decimal digits are those byte values in the sequence b'0123456789'.

For example:

>>> b'ABCabc1'.isalnum()
True
>>> b'ABC abc1'.isalnum()
False

bytes.isalpha() bytearray.isalpha()

Return True if all bytes in the sequence are alphabetic ASCII characters and the sequence is not empty, False otherwise. Alphabetic ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'.

For example:

>>> b'ABCabc'.isalpha()
True
>>> b'ABCabc1'.isalpha()
False

bytes.isascii() bytearray.isascii()

Return True if the sequence is empty or all bytes in the sequence are ASCII, False otherwise. ASCII bytes are in the range 0-0x7F.

3.7

bytes.isdigit() bytearray.isdigit()

Return True if all bytes in the sequence are ASCII decimal digits and the sequence is not empty, False otherwise. ASCII decimal digits are those byte values in the sequence b'0123456789'.

For example:

>>> b'1234'.isdigit()
True
>>> b'1.23'.isdigit()
False

bytes.islower() bytearray.islower()

Return True if there is at least one lowercase ASCII character in the sequence and no uppercase ASCII characters, False otherwise.

For example:

>>> b'hello world'.islower()
True
>>> b'Hello world'.islower()
False

Lowercase ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyz'. Uppercase ASCII characters are those byte values in the sequence b'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.

bytes.isspace() bytearray.isspace()

Return True if all bytes in the sequence are ASCII whitespace and the sequence is not empty, False otherwise. ASCII whitespace characters are those byte values in the sequence b' \t\n\r\x0b\f' (space, tab, newline, carriage return, vertical tab, form feed).

bytes.istitle() bytearray.istitle()

Return True if the sequence is ASCII titlecase and the sequence is not empty, False otherwise. See bytes.title for more details on the definition of "titlecase".

For example:

>>> b'Hello World'.istitle()
True
>>> b'Hello world'.istitle()
False

bytes.isupper() bytearray.isupper()

Return True if there is at least one uppercase alphabetic ASCII character in the sequence and no lowercase ASCII characters, False otherwise.

For example:

>>> b'HELLO WORLD'.isupper()
True
>>> b'Hello world'.isupper()
False

Lowercase ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyz'. Uppercase ASCII characters are those byte values in the sequence b'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.

bytes.lower() bytearray.lower()

Return a copy of the sequence with all the uppercase ASCII characters converted to their corresponding lowercase counterpart.

For example:

>>> b'Hello World'.lower()
b'hello world'

Lowercase ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyz'. Uppercase ASCII characters are those byte values in the sequence b'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

single: universal newlines; bytes.splitlines method single: universal newlines; bytearray.splitlines method

bytes.splitlines(keepends=False) bytearray.splitlines(keepends=False)

Return a list of the lines in the binary sequence, breaking at ASCII line boundaries. This method uses the universal newlines approach to splitting lines. Line breaks are not included in the resulting list unless keepends is given and true.

For example:

>>> b'ab c\n\nde fg\rkl\r\n'.splitlines()
[b'ab c', b'', b'de fg', b'kl']
>>> b'ab c\n\nde fg\rkl\r\n'.splitlines(keepends=True)
[b'ab c\n', b'\n', b'de fg\r', b'kl\r\n']

Unlike ~bytes.split when a delimiter string sep is given, this method returns an empty list for the empty string, and a terminal line break does not result in an extra line:

>>> b"".split(b'\n'), b"Two lines\n".split(b'\n')
([b''], [b'Two lines', b''])
>>> b"".splitlines(), b"One line\n".splitlines()
([], [b'One line'])

bytes.swapcase() bytearray.swapcase()

Return a copy of the sequence with all the lowercase ASCII characters converted to their corresponding uppercase counterpart and vice-versa.

For example:

>>> b'Hello World'.swapcase()
b'hELLO wORLD'

Lowercase ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyz'. Uppercase ASCII characters are those byte values in the sequence b'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.

Unlike str.swapcase, it is always the case that bin.swapcase().swapcase() == bin for the binary versions. Case conversions are symmetrical in ASCII, even though that is not generally true for arbitrary Unicode code points.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.title() bytearray.title()

Return a titlecased version of the binary sequence where words start with an uppercase ASCII character and the remaining characters are lowercase. Uncased byte values are left unmodified.

For example:

>>> b'Hello world'.title()
b'Hello World'

Lowercase ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyz'. Uppercase ASCII characters are those byte values in the sequence b'ABCDEFGHIJKLMNOPQRSTUVWXYZ'. All other byte values are uncased.

>>> b"they're bill's friends from the UK".title()
b"They'Re Bill'S Friends From The Uk"

A workaround for apostrophes can be constructed using regular expressions:

>>> import re
>>> def titlecase(s):
...     return re.sub(rb"[A-Za-z]+('[A-Za-z]+)?",
...                   lambda mo: mo.group(0)[0:1].upper() +
...                              mo.group(0)[1:].lower(),
...                   s)
...
>>> titlecase(b"they're bill's friends.")
b"They're Bill's Friends."

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.upper() bytearray.upper()

Return a copy of the sequence with all the lowercase ASCII characters converted to their corresponding uppercase counterpart.

For example:

>>> b'Hello World'.upper()
b'HELLO WORLD'

Lowercase ASCII characters are those byte values in the sequence b'abcdefghijklmnopqrstuvwxyz'. Uppercase ASCII characters are those byte values in the sequence b'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

bytes.zfill(width, /) bytearray.zfill(width, /)

Return a copy of the sequence left filled with ASCII b'0' digits to make a sequence of length width. A leading sign prefix (b'+'/ b'-') is handled by inserting the padding after the sign character rather than before. For bytes objects, the original sequence is returned if width is less than or equal to len(seq).

For example:

>>> b"42".zfill(5)
b'00042'
>>> b"-42".zfill(5)
b'-0042'

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

`printf`-style Bytes Formatting

single: formatting; bytes (%) single: formatting; bytearray (%) single: interpolation; bytes (%) single: interpolation; bytearray (%) single: bytes; formatting single: bytearray; formatting single: bytes; interpolation single: bytearray; interpolation single: printf-style formatting single: sprintf-style formatting single: % (percent); printf-style formatting

[!NOTE] The formatting operations described here exhibit a variety of quirks that lead to a number of common errors (such as failing to display tuples and dictionaries correctly). If the value being printed may be a tuple or dictionary, wrap it in a tuple.

Bytes objects (bytes/bytearray) have one unique built-in operation: the % operator (modulo). This is also known as the bytes formatting or interpolation operator. Given format % values (where format is a bytes object), % conversion specifications in format are replaced with zero or more elements of values. The effect is similar to using the sprintf function in the C language.

If format requires a single argument, values may be a single non-tuple object.[^9] Otherwise, values must be a tuple with exactly the number of items specified by the format bytes object, or a single mapping object (for example, a dictionary).

single: () (parentheses); in printf-style formatting single: * (asterisk); in printf-style formatting single: . (dot); in printf-style formatting

A conversion specifier contains two or more characters and has the following components, which must occur in this order:

The '%' character, which marks the start of the specifier.
Mapping key (optional), consisting of a parenthesised sequence of characters (for example, (somename)).
Conversion flags (optional), which affect the result of some conversion types.
Minimum field width (optional). If specified as an '*' (asterisk), the actual width is read from the next element of the tuple in values, and the object to convert comes after the minimum field width and optional precision.
Precision (optional), given as a '.' (dot) followed by the precision. If specified as '*' (an asterisk), the actual precision is read from the next element of the tuple in values, and the value to convert comes after the precision.
Length modifier (optional).
Conversion type.

When the right argument is a dictionary (or other mapping type), then the formats in the bytes object must include a parenthesised mapping key into that dictionary inserted immediately after the '%' character. The mapping key selects the value to be formatted from the mapping. For example:

>>> print(b'%(language)s has %(number)03d quote types.' % ... {b'language': b"Python", b"number": 2}) b'Python has 002 quote types.'

In this case no * specifiers may occur in a format (since they require a sequential parameter list).

The conversion flag characters are:

single: # (hash); in printf-style formatting single: - (minus); in printf-style formatting single: + (plus); in printf-style formatting single: space; in printf-style formatting

Flag	Meaning
`'#'`	The value conversion will use the "alternate form" (where defined below).
`'0'`	The conversion will be zero padded for numeric values.
`'-'`	The converted value is left adjusted (overrides the `'0'` conversion if both are given).
`' '`	(a space) A blank should be left before a positive number (or empty string) produced by a signed conversion.
`'+'`	A sign character (`'+'` or `'-'`) will precede the conversion (overrides a "space" flag).

A length modifier (h, l, or L) may be present, but is ignored as it is not necessary for Python -- so e.g. %ld is identical to %d.

The conversion types are:

Conversion	Meaning	Notes
`'d'`	Signed integer decimal.
`'i'`	Signed integer decimal.
`'o'`	Signed octal value.	(1)
`'u'`	Obsolete type -- it is identical to `'d'`.	(8)
`'x'`	Signed hexadecimal (lowercase).	(2)
`'X'`	Signed hexadecimal (uppercase).	(2)
`'e'`	Floating-point exponential format (lowercase).	(3)
`'E'`	Floating-point exponential format (uppercase).	(3)
`'f'`	Floating-point decimal format.	(3)
`'F'`	Floating-point decimal format.	(3)
`'g'`	Floating-point format. Uses lowercase exponential format if exponent is less than -4 or not less than precision, decimal format otherwise.	(4)
`'G'`	Floating-point format. Uses uppercase exponential format if exponent is less than -4 or not less than precision, decimal format otherwise.	(4)
`'c'`	Single byte (accepts integer or single byte objects).
`'b'`	Bytes (any object that follows the `buffer protocol <bufferobjects>` or has `~object.__bytes__`).	(5)
`'s'`	`'s'` is an alias for `'b'` and should only be used for Python2/3 code bases.	(6)
`'a'`	Bytes (converts any Python object using `repr(obj).encode('ascii', 'backslashreplace')`).	(5)
`'r'`	`'r'` is an alias for `'a'` and should only be used for Python2/3 code bases.	(7)
`'%'`	No argument is converted, results in a `'%'` character in the result.

Notes:

(1)
The alternate form causes a leading octal specifier ('0o') to be inserted before the first digit.

(2)
The alternate form causes a leading '0x' or '0X' (depending on whether the 'x' or 'X' format was used) to be inserted before the first digit.

(3)
The alternate form causes the result to always contain a decimal point, even if no digits follow it.

The precision determines the number of digits after the decimal point and defaults to 6.

(4)
The alternate form causes the result to always contain a decimal point, and trailing zeroes are not removed as they would otherwise be.

The precision determines the number of significant digits before and after the decimal point and defaults to 6.

(5)
If precision is N, the output is truncated to N characters.

(6)
b'%s' is deprecated, but will not be removed during the 3.x series.

(7)
b'%r' is deprecated, but will not be removed during the 3.x series.

(8)
See 237.

[!NOTE] The bytearray version of this method does not operate in place - it always produces a new object, even if no changes were made.

461 - Adding % formatting to bytes and bytearray

3.5

Memory Views

memoryview objects allow Python code to access the internal data of an object that supports the buffer protocol <bufferobjects> without copying.

Create a memoryview that references object. object must support the buffer protocol. Built-in objects that support the buffer protocol include bytes and bytearray.

A memoryview has the notion of an element, which is the atomic memory unit handled by the originating object. For many simple types such as bytes and bytearray, an element is a single byte, but other types such as array.array may have bigger elements.

!memoryviews are generic <generics> over the type of their underlying data.

len(view) is equal to the length of ~memoryview.tolist, which is the nested list representation of the view. If view.ndim == 1, this is equal to the number of elements in the view.

3.12 If view.ndim == 0, len(view) now raises TypeError instead of returning 1.

The ~memoryview.itemsize attribute will give you the number of bytes in a single element.

A memoryview supports slicing and indexing to expose its data. One-dimensional slicing will result in a subview:

>>> v = memoryview(b'abcefg')
>>> v[1]
98
>>> v[-1]
103
>>> v[1:4]
<memory at 0x7f3ddc9f4350>
>>> bytes(v[1:4])
b'bce'

If ~memoryview.format is one of the native format specifiers from the struct module, indexing with an integer or a tuple of integers is also supported and returns a single element with the correct type. One-dimensional memoryviews can be indexed with an integer or a one-integer tuple. Multi-dimensional memoryviews can be indexed with tuples of exactly ndim integers where ndim is the number of dimensions. Zero-dimensional memoryviews can be indexed with the empty tuple.

Here is an example with a non-byte format:

>>> import array
>>> a = array.array('l', [-11111111, 22222222, -33333333, 44444444])
>>> m = memoryview(a)
>>> m[0]
-11111111
>>> m[-1]
44444444
>>> m[::2].tolist()
[-11111111, -33333333]

If the underlying object is writable, the memoryview supports one-dimensional slice assignment. Resizing is not allowed:

>>> data = bytearray(b'abcefg')
>>> v = memoryview(data)
>>> v.readonly
False
>>> v[0] = ord(b'z')
>>> data
bytearray(b'zbcefg')
>>> v[1:4] = b'123'
>>> data
bytearray(b'z123fg')
>>> v[2:3] = b'spam'
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: memoryview assignment: lvalue and rvalue have different structures
>>> v[2:6] = b'spam'
>>> data
bytearray(b'z1spam')

One-dimensional memoryviews of hashable (read-only) types with formats 'B', 'b' or 'c' are also hashable. The hash is defined as hash(m) == hash(m.tobytes()):

>>> v = memoryview(b'abcefg')
>>> hash(v) == hash(b'abcefg')
True
>>> hash(v[2:4]) == hash(b'ce')
True
>>> hash(v[::-2]) == hash(b'abcefg'[::-2])
True

3.3 One-dimensional memoryviews can now be sliced. One-dimensional memoryviews with formats 'B', 'b' or 'c' are now hashable.

3.4 memoryview is now registered automatically with collections.abc.Sequence

3.5 memoryviews can now be indexed with a tuple of integers.

3.14 memoryview is now a generic type.

memoryview has several methods:

__eq__(exporter)

A memoryview and a 3118 exporter are equal if their shapes are equivalent and if all corresponding values are equal when the operands' respective format codes are interpreted using struct syntax.

For the subset of struct format strings currently supported by tolist, v and w are equal if v.tolist() == w.tolist():

>>> import array
>>> a = array.array('I', [1, 2, 3, 4, 5])
>>> b = array.array('d', [1.0, 2.0, 3.0, 4.0, 5.0])
>>> c = array.array('b', [5, 3, 1])
>>> x = memoryview(a)
>>> y = memoryview(b)
>>> x == a == y == b
True
>>> x.tolist() == a.tolist() == y.tolist() == b.tolist()
True
>>> z = y[::-2]
>>> z == c
True
>>> z.tolist() == c.tolist()
True

If either format string is not supported by the struct module, then the objects will always compare as unequal (even if the format strings and buffer contents are identical):

>>> from ctypes import BigEndianStructure, c_long
>>> class BEPoint(BigEndianStructure):
...     _fields_ = [("x", c_long), ("y", c_long)]
...
>>> point = BEPoint(100, 200)
>>> a = memoryview(point)
>>> b = memoryview(point)
>>> a == point
False
>>> a == b
False

Note that, as with floating-point numbers, v is w does not imply v == w for memoryview objects.

3.3 Previous versions compared the raw memory disregarding the item format and the logical array structure.

tobytes(order='C')

Return the data in the buffer as a bytestring. This is equivalent to calling the bytes constructor on the memoryview. :

>>> m = memoryview(b"abc")
>>> m.tobytes()
b'abc'
>>> bytes(m)
b'abc'

For non-contiguous arrays the result is equal to the flattened list representation with all elements converted to bytes. tobytes supports all format strings, including those that are not in struct module syntax.

3.8 order can be {'C', 'F', 'A'}. When order is 'C' or 'F', the data of the original array is converted to C or Fortran order. For contiguous views, 'A' returns an exact copy of the physical memory. In particular, in-memory Fortran order is preserved. For non-contiguous views, the data is converted to C first. order=None is the same as order='C'.

hex(*, bytes_per_sep=1) hex(sep, bytes_per_sep=1)

Return a string object containing two hexadecimal digits for each byte in the buffer. :

>>> m = memoryview(b"abc")
>>> m.hex()
'616263'

3.5

3.8 Similar to bytes.hex, memoryview.hex now supports optional sep and bytes_per_sep parameters to insert separators between bytes in the hex output.

tolist()

Return the data in the buffer as a list of elements. :

>>> memoryview(b'abc').tolist()
[97, 98, 99]
>>> import array
>>> a = array.array('d', [1.1, 2.2, 3.3])
>>> m = memoryview(a)
>>> m.tolist()
[1.1, 2.2, 3.3]

3.3 tolist now supports all single character native formats in struct module syntax as well as multi-dimensional representations.

toreadonly()

Return a readonly version of the memoryview object. The original memoryview object is unchanged. :

>>> m = memoryview(bytearray(b'abc'))
>>> mm = m.toreadonly()
>>> mm.tolist()
[97, 98, 99]
>>> mm[0] = 42
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: cannot modify read-only memory
>>> m[0] = 43
>>> mm.tolist()
[43, 98, 99]

3.8

release()

Release the underlying buffer exposed by the memoryview object. Many objects take special actions when a view is held on them (for example, a bytearray would temporarily forbid resizing); therefore, calling release() is handy to remove these restrictions (and free any dangling resources) as soon as possible.

After this method has been called, any further operation on the view raises a ValueError (except release itself which can be called multiple times):

>>> m = memoryview(b'abc')
>>> m.release()
>>> m[0]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: operation forbidden on released memoryview object

The context management protocol can be used for a similar effect, using the with statement:

>>> with memoryview(b'abc') as m:
...     m[0]
...
97
>>> m[0]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: operation forbidden on released memoryview object

3.2

cast(format, /) cast(format, shape, /, *, order='C')

Cast a memoryview to a new format or shape. shape defaults to [byte_length//new_itemsize], which means that the result view will be one-dimensional. The return value is a new memoryview, but the buffer itself is not copied. Supported casts are 1D -> C-contiguous, C-contiguous -> 1D, and F-contiguous -> 1D.

With a multidimensional shape, order selects the memory layout of the result: 'C' for C-contiguous (row-major, the default) or 'F' for Fortran-contiguous (column-major). The buffer is still not copied, so order='F' gives a zero-copy view over a buffer holding column-major data.

The destination format is restricted to a single element native format in struct syntax. One of the formats must be a byte format ('B', 'b' or 'c'). The byte length of the result must be the same as the original length. Note that all byte lengths may depend on the operating system.

Cast 1D/long to 1D/unsigned bytes:

>>> import array
>>> a = array.array('l', [1,2,3])
>>> x = memoryview(a)
>>> x.format
'l'
>>> x.itemsize
8
>>> len(x)
3
>>> x.nbytes
24
>>> y = x.cast('B')
>>> y.format
'B'
>>> y.itemsize
1
>>> len(y)
24
>>> y.nbytes
24

Cast 1D/unsigned bytes to 1D/char:

>>> b = bytearray(b'zyz')
>>> x = memoryview(b)
>>> x[0] = b'a'
Traceback (most recent call last):
  ...
TypeError: memoryview: invalid type for format 'B'
>>> y = x.cast('c')
>>> y[0] = b'a'
>>> b
bytearray(b'ayz')

Cast 1D/bytes to 3D/ints to 1D/signed char:

>>> import struct
>>> buf = struct.pack("i"*12, *list(range(12)))
>>> x = memoryview(buf)
>>> y = x.cast('i', shape=[2,2,3])
>>> y.tolist()
[[[0, 1, 2], [3, 4, 5]], [[6, 7, 8], [9, 10, 11]]]
>>> y.format
'i'
>>> y.itemsize
4
>>> len(y)
2
>>> y.nbytes
48
>>> z = y.cast('b')
>>> z.format
'b'
>>> z.itemsize
1
>>> len(z)
48
>>> z.nbytes
48

Cast 1D/unsigned long to 2D/unsigned long:

>>> buf = struct.pack("L"*6, *list(range(6)))
>>> x = memoryview(buf)
>>> y = x.cast('L', shape=[2,3])
>>> len(y)
2
>>> y.nbytes
48
>>> y.tolist()
[[0, 1, 2], [3, 4, 5]]

3.3

3.5 The source format is no longer restricted when casting to a byte view.

3.16 Casting a multi-dimensional F-contiguous view to a one-dimensional view is now supported.

count(value, /)

Count the number of occurrences of value.

3.14

index(value, start=0, stop=sys.maxsize, /)

Return the index of the first occurrence of value (at or after index start and before index stop).

Raises a ValueError if value cannot be found.

3.14

There are also several readonly attributes available:

obj

The underlying object of the memoryview:

>>> b  = bytearray(b'xyz')
>>> m = memoryview(b)
>>> m.obj is b
True

3.3

nbytes

nbytes == product(shape) * itemsize == len(m.tobytes()). This is the amount of space in bytes that the array would use in a contiguous representation. It is not necessarily equal to len(m):

>>> import array
>>> a = array.array('i', [1,2,3,4,5])
>>> m = memoryview(a)
>>> len(m)
5
>>> m.nbytes
20
>>> y = m[::2]
>>> len(y)
3
>>> y.nbytes
12
>>> len(y.tobytes())
12

Multi-dimensional arrays:

>>> import struct
>>> buf = struct.pack("d"*12, *[1.5*x for x in range(12)])
>>> x = memoryview(buf)
>>> y = x.cast('d', shape=[3,4])
>>> y.tolist()
[[0.0, 1.5, 3.0, 4.5], [6.0, 7.5, 9.0, 10.5], [12.0, 13.5, 15.0, 16.5]]
>>> len(y)
3
>>> y.nbytes
96

Interpret a flat buffer as a Fortran-contiguous (column-major) array:

>>> buf = bytes(range(6))
>>> y = memoryview(buf).cast('B', shape=[3, 2], order='F')
>>> y.f_contiguous
True
>>> y.tolist()
[[0, 3], [1, 4], [2, 5]]

3.3

3.16 Added the order parameter.

readonly

A bool indicating whether the memory is read only.

format

A string containing the format (in struct module style) for each element in the view. A memoryview can be created from exporters with arbitrary format strings, but some methods (e.g. tolist) are restricted to native single element formats.

3.3 format 'B' is now handled according to the struct module syntax. This means that memoryview(b'abc')[0] == b'abc'[0] == 97.

itemsize

The size in bytes of each element of the memoryview:

>>> import array, struct
>>> m = memoryview(array.array('H', [32000, 32001, 32002]))
>>> m.itemsize
2
>>> m[0]
32000
>>> struct.calcsize('H') == m.itemsize
True

ndim

An integer indicating how many dimensions of a multi-dimensional array the memory represents.

shape

A tuple of integers the length of ndim giving the shape of the memory as an N-dimensional array.

3.3 An empty tuple instead of None when ndim = 0.

strides

A tuple of integers the length of ndim giving the size in bytes to access each element for each dimension of the array.

3.3 An empty tuple instead of None when ndim = 0.

suboffsets

Used internally for PIL-style arrays. The value is informational only.

c_contiguous

A bool indicating whether the memory is C-contiguous.

3.3

f_contiguous

A bool indicating whether the memory is Fortran contiguous.

3.3

contiguous

A bool indicating whether the memory is contiguous.

3.3

For information on the thread safety of memoryview objects in the free-threaded build, see thread-safety-memoryview.

Set Types --- `set`, `frozenset`

pair: object; set

A set object is an unordered collection of distinct hashable objects. Common uses include membership testing, removing duplicates from a sequence, and computing mathematical operations such as intersection, union, difference, and symmetric difference. (For other containers see the built-in dict, list, and tuple classes, and the collections module.) See time-complexity for the costs of the various set operations.

Like other collections, sets support x in set, len(set), and for x in set. Being an unordered collection, sets do not record element position or order of insertion. Accordingly, sets do not support indexing, slicing, or other sequence-like behavior.

There are currently two built-in set types, set and frozenset. The set type is mutable --- the contents can be changed using methods like ~set.add and ~set.remove. Since it is mutable, it has no hash value and cannot be used as either a dictionary key or as an element of another set. The frozenset type is immutable and hashable --- its contents cannot be altered after it is created; it can therefore be used as a dictionary key or as an element of another set.

Non-empty sets (not frozensets) can be created by placing a comma-separated list of elements within braces, for example: {'jack', 'sjoerd'}, in addition to the set constructor.

The constructors for both classes work the same:

Return a new set or frozenset object whose elements are taken from iterable. The elements of a set must be hashable. To represent sets of sets, the inner sets must be frozenset objects. If iterable is not specified, a new empty set is returned.

Sets can be created by several means:

Use a comma-separated list of elements within braces: {'jack', 'sjoerd'}
Use a set comprehension: {c for c in 'abracadabra' if c not in 'abc'}
Use the type constructor: set(), set('foobar'), set(['a', 'b', 'foo'])

Instances of set and frozenset provide the following operations:

len(s)

Return the number of elements in set s (cardinality of s).

x in s

Test x for membership in s.

x not in s

Test x for non-membership in s.

frozenset.isdisjoint(other, /) set.isdisjoint(other, /)

Return True if the set has no elements in common with other. Sets are disjoint if and only if their intersection is the empty set.

frozenset.issubset(other, /) set.issubset(other, /)

set <= other

Test whether every element in the set is in other.

set < other

Test whether the set is a proper subset of other, that is, set <= other and set != other.

frozenset.issuperset(other, /) set.issuperset(other, /)

set >= other

Test whether every element in other is in the set.

set > other

Test whether the set is a proper superset of other, that is, set >= other and set != other.

frozenset.union(*others) set.union(*others)

set | other | ...

Return a new set with elements from the set and all others.

frozenset.intersection(*others) set.intersection(*others)

set & other & ...

Return a new set with elements common to the set and all others.

frozenset.difference(*others) set.difference(*others)

set - other - ...

Return a new set with elements in the set that are not in the others.

frozenset.symmetric_difference(other, /) set.symmetric_difference(other, /)

set ^ other

Return a new set with elements in either the set or other but not both.

frozenset.copy() set.copy()

Return a shallow copy of the set.

Note, the non-operator versions of ~frozenset.union, ~frozenset.intersection, ~frozenset.difference, ~frozenset.symmetric_difference, ~frozenset.issubset, and ~frozenset.issuperset methods will accept any iterable as an argument. In contrast, their operator based counterparts require their arguments to be sets. This precludes error-prone constructions like set('abc') & 'cbs' in favor of the more readable set('abc').intersection('cbs').

Both set and frozenset support set to set comparisons. Two sets are equal if and only if every element of each set is contained in the other (each is a subset of the other). A set is less than another set if and only if the first set is a proper subset of the second set (is a subset, but is not equal). A set is greater than another set if and only if the first set is a proper superset of the second set (is a superset, but is not equal).

Instances of set are compared to instances of frozenset based on their members. For example, set('abc') == frozenset('abc') returns True and so does set('abc') in set([frozenset('abc')]).

The subset and equality comparisons do not generalize to a total ordering function. For example, any two nonempty disjoint sets are not equal and are not subsets of each other, so all of the following return False: a<b, a==b, or a>b.

Since sets only define partial ordering (subset relationships), the output of the list.sort method is undefined for lists of sets.

Set elements, like dictionary keys, must be hashable.

Binary operations that mix set instances with frozenset return the type of the first operand. For example: frozenset('ab') | set('bc') returns an instance of frozenset.

The following table lists operations available for set that do not apply to immutable instances of frozenset:

set.update(*others)

set |= other | ...

Update the set, adding elements from all others.

set.intersection_update(*others)

set &= other & ...

Update the set, keeping only elements found in it and all others.

set.difference_update(*others)

set -= other | ...

Update the set, removing elements found in others.

set.symmetric_difference_update(other, /)

set ^= other

Update the set, keeping only elements found in either set, but not in both.

set.add(elem, /)

Add element elem to the set.

set.remove(elem, /)

Remove element elem from the set. Raises KeyError if elem is not contained in the set.

set.discard(elem, /)

Remove element elem from the set if it is present.

set.pop()

Remove and return an arbitrary element from the set. Raises KeyError if the set is empty.

set.clear()

Remove all elements from the set.

Note, the non-operator versions of the ~set.update, ~set.intersection_update, ~set.difference_update, and ~set.symmetric_difference_update methods will accept any iterable as an argument.

Note, the elem argument to the ~object.__contains__, ~set.remove, and ~set.discard methods may be a set. In this case, an equivalent frozenset is searched for.

Sets and frozensets are generic <generics> over the type of their elements.

For detailed information on thread-safety guarantees for set objects, see thread-safety-set.

Mapping types --- `!dict`, `!frozendict`

pair: object; mapping pair: object; dictionary triple: operations on; mapping; types triple: operations on; dictionary; type pair: statement; del pair: built-in function; len

A mapping object maps hashable values to arbitrary objects. There are currently two standard mapping types, the dictionary and frozendict. (For other containers see the built-in list, set, and tuple classes, and the collections module.) See time-complexity for the costs of the various dictionary operations.

A dictionary's keys are almost arbitrary values. Values that are not hashable, that is, values containing lists, dictionaries or other mutable types (that are compared by value rather than by object identity) may not be used as keys. Values that compare equal (such as 1, 1.0, and True) can be used interchangeably to index the same dictionary entry.

Return a new dictionary initialized from an optional positional argument and a possibly empty set of keyword arguments.

Dictionaries can be created by several means:

Use a comma-separated list of key: value pairs within braces: {'jack': 4098, 'sjoerd': 4127} or {4098: 'jack', 4127: 'sjoerd'}
Use a dict comprehension: {x: x ** 2 for x in range(10)}
Use the type constructor: dict(), dict([('foo', 100), ('bar', 200)]), dict(foo=100, bar=200)

If no positional argument is given, an empty dictionary is created. If a positional argument is given and it defines a keys() method, a dictionary is created by calling ~object.__getitem__ on the argument with each returned key from the method. Otherwise, the positional argument must be an iterable object. Each item in the iterable must itself be an iterable with exactly two elements. The first element of each item becomes a key in the new dictionary, and the second element the corresponding value. If a key occurs more than once, the last value for that key becomes the corresponding value in the new dictionary.

If keyword arguments are given, the keyword arguments and their values are added to the dictionary created from the positional argument. If a key being added is already present, the value from the keyword argument replaces the value from the positional argument.

Dictionaries compare equal if and only if they have the same (key, value) pairs (regardless of ordering). Order comparisons ('<', '<=', '>=', '>') raise TypeError. To illustrate dictionary creation and equality, the following examples all return a dictionary equal to {"one": 1, "two": 2, "three": 3}:

>>> a = dict(one=1, two=2, three=3)
>>> b = {'one': 1, 'two': 2, 'three': 3}
>>> c = dict(zip(['one', 'two', 'three'], [1, 2, 3]))
>>> d = dict([('two', 2), ('one', 1), ('three', 3)])
>>> e = dict({'three': 3, 'one': 1, 'two': 2})
>>> f = dict({'one': 1, 'three': 3}, two=2)
>>> a == b == c == d == e == f
True

Providing keyword arguments as in the first example only works for keys that are valid Python identifiers. Otherwise, any valid keys can be used.

Dictionaries preserve insertion order. Note that updating a key does not affect the order. Keys added after deletion are inserted at the end. :

>>> d = {"one": 1, "two": 2, "three": 3, "four": 4}
>>> d
{'one': 1, 'two': 2, 'three': 3, 'four': 4}
>>> list(d)
['one', 'two', 'three', 'four']
>>> list(d.values())
[1, 2, 3, 4]
>>> d["one"] = 42
>>> d
{'one': 42, 'two': 2, 'three': 3, 'four': 4}
>>> del d["two"]
>>> d["two"] = None
>>> d
{'one': 42, 'three': 3, 'four': 4, 'two': None}

3.7 Dictionary order is guaranteed to be insertion order. This behavior was an implementation detail of CPython from 3.6.

Dictionaries are generic <generics> over two types, signifying (respectively) the types of the dictionary's keys and values.

These are the operations that dictionaries support (and therefore, custom mapping types should support too):

list(d)

Return a list of all the keys used in the dictionary d.

len(d)

Return the number of items in the dictionary d.

d[key]

Return the item of d with key key. Raises a KeyError if key is not in the map.

__missing__()

If a subclass of dict defines a method ~object.__missing__ and key is not present, the d[key] operation calls that method with the key key as argument. The d[key] operation then returns or raises whatever is returned or raised by the __missing__(key) call. No other operations or methods invoke ~object.__missing__. If ~object.__missing__ is not defined, KeyError is raised. ~object.__missing__ must be a method; it cannot be an instance variable:

>>> class Counter(dict):
...     def __missing__(self, key):
...         return 0
...
>>> c = Counter()
>>> c['red']
0
>>> c['red'] += 1
>>> c['red']
1

The example above shows part of the implementation of collections.Counter. A different !__missing__ method is used by collections.defaultdict.

d[key] = value

Set d[key] to value.

del d[key]

Remove d[key] from d. Raises a KeyError if key is not in the map.

key in d

Return True if d has a key key, else False.

key not in d

Equivalent to not key in d.

iter(d)

Return an iterator over the keys of the dictionary. This is a shortcut for iter(d.keys()).

clear()

Remove all items from the dictionary.

copy()

Return a shallow copy of the dictionary.

fromkeys(iterable, value=None, /)

Create a new dictionary with keys from iterable and values set to value.

fromkeys is a class method that returns a new dictionary. value defaults to None. All of the values refer to just a single instance, so it generally doesn't make sense for value to be a mutable object such as an empty list. To get distinct values, use a dict comprehension <dict> instead.

get(key, default=None, /)

Return the value for key if key is in the dictionary, else default. If default is not given, it defaults to None, so that this method never raises a KeyError.

items()

Return a new view of the dictionary's items ((key, value) pairs). See the documentation of view objects <dict-views>.

keys()

Return a new view of the dictionary's keys. See the documentation of view objects <dict-views>.

pop(key, /) pop(key, default, /)

If key is in the dictionary, remove it and return its value, else return default. If default is not given and key is not in the dictionary, a KeyError is raised.

popitem()

Remove and return a (key, value) pair from the dictionary. Pairs are returned in LIFO (last-in, first-out) order.

popitem is useful to destructively iterate over a dictionary, as often used in set algorithms. If the dictionary is empty, calling popitem raises a KeyError.

3.7 LIFO order is now guaranteed. In prior versions, popitem would return an arbitrary key/value pair.

reversed(d)

Return a reverse iterator over the keys of the dictionary. This is a shortcut for reversed(d.keys()).

3.8

setdefault(key, default=None, /)

If key is in the dictionary, return its value. If not, insert key with a value of default and return default. default defaults to None.

update(**kwargs) update(mapping, /,**kwargs) update(iterable, /, **kwargs)

Update the dictionary with the key/value pairs from mapping or iterable and kwargs, overwriting existing keys. Return None.

update accepts either another object with a keys() method (in which case ~object.__getitem__ is called with every key returned from the method) or an iterable of key/value pairs (as tuples or other iterables of length two). If keyword arguments are specified, the dictionary is then updated with those key/value pairs: d.update(red=1, blue=2).

values()

Return a new view of the dictionary's values. See the documentation of view objects <dict-views>.

An equality comparison between two distinct dict.values() views will always return False, even for views of the same dictionary:

>>> d = {'a': 1}
>>> d.values() == d.values()
False

d | other

Create a new dictionary with the merged keys and values of d and other, which must both be dictionaries. The values of other take priority when d and other share keys.

3.9

d |= other

Update the dictionary d with keys and values from other, which may be either a mapping or an iterable of key/value pairs. The values of other take priority when d and other share keys.

3.9

Dictionaries and dictionary views are reversible. :

>>> d = {"one": 1, "two": 2, "three": 3, "four": 4}
>>> d
{'one': 1, 'two': 2, 'three': 3, 'four': 4}
>>> list(reversed(d))
['four', 'three', 'two', 'one']
>>> list(reversed(d.values()))
[4, 3, 2, 1]
>>> list(reversed(d.items()))
[('four', 4), ('three', 3), ('two', 2), ('one', 1)]

3.8 Dictionaries are now reversible.

frozendict and types.MappingProxyType can be used to create a read-only view of a dict.

For detailed information on thread-safety guarantees for dict objects, see thread-safety-dict.

Dictionary view objects

The objects returned by dict.keys, dict.values and dict.items are view objects. They provide a dynamic view on the dictionary's entries, which means that when the dictionary changes, the view reflects these changes.

Dictionary views can be iterated over to yield their respective data, and support membership tests:

len(dictview)

Return the number of entries in the dictionary.

iter(dictview)

Return an iterator over the keys, values or items (represented as tuples of (key, value)) in the dictionary.

Keys and values are iterated over in insertion order. This allows the creation of (value, key) pairs using zip: pairs = zip(d.values(), d.keys()). Another way to create the same list is pairs = [(v, k) for (k, v) in d.items()].

Iterating views while adding or deleting entries in the dictionary may raise a RuntimeError or fail to iterate over all entries.

3.7 Dictionary order is guaranteed to be insertion order.

x in dictview

Return True if x is in the underlying dictionary's keys, values or items (in the latter case, x should be a (key, value) tuple).

reversed(dictview)

Return a reverse iterator over the keys, values or items of the dictionary. The view will be iterated in reverse order of the insertion.

3.8 Dictionary views are now reversible.

dictview.mapping

Return a types.MappingProxyType that wraps the original dictionary to which the view refers.

3.10

Keys views are set-like since their entries are unique and hashable. Items views also have set-like operations since the (key, value) pairs are unique and the keys are hashable. If all values in an items view are hashable as well, then the items view can interoperate with other sets. (Values views are not treated as set-like since the entries are generally not unique.) For set-like views, all of the operations defined for the abstract base class collections.abc.Set are available (for example, ==, <, or ^). While using set operators, set-like views accept any iterable as the other operand, unlike sets which only accept sets as the input.

An example of dictionary view usage:

>>> dishes = {'eggs': 2, 'sausage': 1, 'bacon': 1, 'spam': 500}
>>> keys = dishes.keys()
>>> values = dishes.values()

>>> # iteration
>>> n = 0
>>> for val in values:
...     n += val
...
>>> print(n)
504

>>> # keys and values are iterated over in the same order (insertion order)
>>> list(keys)
['eggs', 'sausage', 'bacon', 'spam']
>>> list(values)
[2, 1, 1, 500]

>>> # view objects are dynamic and reflect dict changes
>>> del dishes['eggs']
>>> del dishes['sausage']
>>> list(keys)
['bacon', 'spam']

>>> # set operations
>>> keys & {'eggs', 'bacon', 'salad'}
{'bacon'}
>>> keys ^ {'sausage', 'juice'} == {'juice', 'sausage', 'bacon', 'spam'}
True
>>> keys | ['juice', 'juice', 'juice'] == {'bacon', 'spam', 'juice'}
True

>>> # get back a read-only proxy for the original dictionary
>>> values.mapping
mappingproxy({'bacon': 1, 'spam': 500})
>>> values.mapping['spam']
500

Frozen dictionaries

Return a new frozen dictionary initialized from an optional positional argument and a possibly empty set of keyword arguments.

A !frozendict has a similar API to the dict API, with the following differences:

!dict has more methods than !frozendict:
- !__delitem__
- !__setitem__
- ~dict.clear
- ~dict.pop
- ~dict.popitem
- ~dict.setdefault
- ~dict.update
A !frozendict can be hashed with hash(frozendict) if all keys and values can be hashed.
frozendict |= other does not modify the !frozendict in-place but creates a new frozen dictionary.

!frozendict is not a !dict subclass but inherits directly from object.

Like dictionaries, frozendicts are generic <generics> over two types, signifying (respectively) the types of the frozendict's keys and values.

fromkeys(iterable, value=None, /)

Similar to dict.fromkeys, but call again the type constructor with an initialized frozendict if the type is a frozendict subclass or if the constructor returned a frozendict.

3.15

Context Manager Types

single: context manager single: context management protocol single: protocol; context management

Python's with statement supports the concept of a runtime context defined by a context manager. This is implemented using a pair of methods that allow user-defined classes to define a runtime context that is entered before the statement body is executed and exited when the statement ends:

contextmanager.__enter__()

Enter the runtime context and return either this object or another object related to the runtime context. The value returned by this method is bound to the identifier in the !as clause of with statements using this context manager.

An example of a context manager that returns itself is a file object. File objects return themselves from __enter__() to allow open to be used as the context expression in a with statement.

An example of a context manager that returns a related object is the one returned by decimal.localcontext. These managers set the active decimal context to a copy of the original decimal context and then return the copy. This allows changes to be made to the current decimal context in the body of the with statement without affecting code outside the !with statement.

contextmanager.__exit__(exc_type, exc_val, exc_tb)

Exit the runtime context and return a Boolean flag indicating if any exception that occurred should be suppressed. If an exception occurred while executing the body of the with statement, the arguments contain the exception type, value and traceback information. Otherwise, all three arguments are None.

Returning a true value from this method will cause the with statement to suppress the exception and continue execution with the statement immediately following the !with statement. Otherwise the exception continues propagating after this method has finished executing.

If this method raises an exception while handling an earlier exception from the with block, the new exception is raised, and the original exception is stored in its ~BaseException.__context__ attribute.

The exception passed in should never be reraised explicitly - instead, this method should return a false value to indicate that the method completed successfully and does not want to suppress the raised exception. This allows context management code to easily detect whether or not an ~object.__exit__ method has actually failed.

Python defines several context managers to support easy thread synchronisation, prompt closure of files or other objects, and simpler manipulation of the active decimal arithmetic context. The specific types are not treated specially beyond their implementation of the context management protocol. See the contextlib module for some examples.

Python's generators and the contextlib.contextmanager decorator provide a convenient way to implement these protocols. If a generator function is decorated with the contextlib.contextmanager decorator, it will return a context manager implementing the necessary ~contextmanager.__enter__ and ~contextmanager.__exit__ methods, rather than the iterator produced by an undecorated generator function.

Note that there is no specific slot for any of these methods in the type structure for Python objects in the Python/C API. Extension types wanting to define these methods must provide them as a normal Python accessible method. Compared to the overhead of setting up the runtime context, the overhead of a single class dictionary lookup is negligible.

Type Annotation Types --- `Generic Alias <types-genericalias>`, `Union <types-union>`

single: annotation; type annotation; type hint

The core built-in types for type annotations <annotation> are Generic Alias <types-genericalias> and Union <types-union>.

Generic Alias Type

pair: object; GenericAlias pair: Generic; Alias

GenericAlias objects are generally created by subscripting <subscriptions> a class. They are most often used with container classes <sequence-types>, such as list or dict. For example, list[int] is a GenericAlias object created by subscripting the list class with the argument int. GenericAlias objects are intended primarily for use with type annotations <annotation>.

[!NOTE] It is generally only possible to subscript a class if the class implements the special method ~object.__class_getitem__.

A GenericAlias object acts as a proxy for a generic type, implementing parameterized generics.

For a container class, the argument(s) supplied to a subscription <subscriptions> of the class may indicate the type(s) of the elements an object contains. For example, set[bytes] can be used in type annotations to signify a set in which all the elements are of type bytes.

For a class which defines ~object.__class_getitem__ but is not a container, the argument(s) supplied to a subscription of the class will often indicate the return type(s) of one or more methods defined on an object. For example, regular expressions <re> can be used on both the str data type and the bytes data type:

If x = re.search('foo', 'foo'), x will be a re.Match <match-objects> object where the return values of x.group(0) and x[0] will both be of type str. We can represent this kind of object in type annotations with the GenericAlias re.Match[str].
If y = re.search(b'bar', b'bar'), (note the b for bytes), y will also be an instance of re.Match, but the return values of y.group(0) and y[0] will both be of type bytes. In type annotations, we would represent this variety of re.Match <match-objects> objects with re.Match[bytes].

GenericAlias objects are instances of the class types.GenericAlias, which can also be used to create GenericAlias objects directly. Specializations of user-defined generic classes <generic-classes> may not be instances of types.GenericAlias, but they provide similar functionality.

T[X, Y, ...]

Creates a GenericAlias representing a type T parameterized by types X, Y, and more depending on the T used. For example, a function expecting a list containing float elements:

def average(values: list[float]) -> float:
    return sum(values) / len(values)

Another example for mapping objects, using a dict, which is a generic type expecting two type parameters representing the key type and the value type. In this example, the function expects a dict with keys of type str and values of type int:

def send_post_request(url: str, body: dict[str, int]) -> None:
    ...

The builtin functions isinstance and issubclass do not accept GenericAlias types for their second argument:

>>> isinstance([1, 2], list[str])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: isinstance() argument 2 cannot be a parameterized generic

The Python runtime does not enforce type annotations <annotation>. This extends to generic types and their type parameters. When creating a container object from a GenericAlias, the elements in the container are not checked against their type. For example, the following code is discouraged, but will run without errors:

>>> t = list[str]
>>> t([1, 2, 3])
[1, 2, 3]

Furthermore, parameterized generics erase type parameters during object creation:

>>> t = list[str]
>>> type(t)
<class 'types.GenericAlias'>

>>> l = t()
>>> type(l)
<class 'list'>

Instances of GenericAlias are not classes at runtime, even though they behave like classes (they can be instantiated and subclassed):

>>> import inspect
>>> inspect.isclass(list[int])
False

This is true for user-defined generics <user-defined-generics> also.

Calling repr or str on a generic shows the parameterized type:

>>> repr(list[int])
'list[int]'

>>> str(list[int])
'list[int]'

The ~object.__getitem__ method of generic containers will raise an exception to disallow mistakes like dict[str][str]:

>>> dict[str][str]
Traceback (most recent call last):
  ...
TypeError: dict[str] is not a generic class

However, such expressions are valid when type variables <generics> are used. The index must have as many elements as there are type variable items in the GenericAlias object's ~genericalias.__args__. :

>>> from typing import TypeVar
>>> Y = TypeVar('Y')
>>> dict[str, Y][int]
dict[str, int]

Standard Generic Classes

The following standard library classes support parameterized generics. This list is non-exhaustive.

tuple
list
dict
set
frozendict
frozenset
type
asyncio.Future
asyncio.Task
collections.deque
collections.defaultdict
collections.OrderedDict
collections.Counter
collections.ChainMap
collections.abc.Awaitable
collections.abc.Coroutine
collections.abc.AsyncIterable
collections.abc.AsyncIterator
collections.abc.AsyncGenerator
collections.abc.Iterable
collections.abc.Iterator
collections.abc.Generator
collections.abc.Reversible
collections.abc.Container
collections.abc.Collection
collections.abc.Callable
collections.abc.Set
collections.abc.MutableSet
collections.abc.Mapping
collections.abc.MutableMapping
collections.abc.Sequence
collections.abc.MutableSequence
collections.abc.ByteString
collections.abc.MappingView
collections.abc.KeysView
collections.abc.ItemsView
collections.abc.ValuesView
contextlib.AbstractContextManager
contextlib.AbstractAsyncContextManager
dataclasses.Field
functools.cached_property
functools.partialmethod
os.PathLike
queue.LifoQueue
queue.Queue
queue.PriorityQueue
queue.SimpleQueue
re.Pattern <re-objects>
re.Match <match-objects>
shelve.BsdDbShelf
shelve.DbfilenameShelf
shelve.Shelf
types.MappingProxyType
weakref.WeakKeyDictionary
weakref.WeakMethod
weakref.WeakSet
weakref.WeakValueDictionary

Special Attributes of `GenericAlias` objects

All parameterized generics implement special read-only attributes.

genericalias.__origin__

This attribute points at the non-parameterized generic class:

>>> list[int].__origin__
<class 'list'>

genericalias.__args__

This attribute is a tuple (possibly of length 1) of generic types passed to the original ~object.__class_getitem__ of the generic class:

>>> dict[str, list[int]].__args__
(<class 'str'>, list[int])

genericalias.__parameters__

This attribute is a lazily computed tuple (possibly empty) of unique type variables found in __args__:

>>> from typing import TypeVar

>>> T = TypeVar('T')
>>> list[T].__parameters__
(~T,)

[!NOTE] A GenericAlias object with typing.ParamSpec parameters may not have correct __parameters__ after substitution because typing.ParamSpec is intended primarily for static type checking.

genericalias.__unpacked__

A boolean that is true if the alias has been unpacked using the * operator (see ~typing.TypeVarTuple).

3.11

484 - Type Hints
Introducing Python's framework for type annotations.

585 - Type Hinting Generics In Standard Collections
Introducing the ability to natively parameterize standard-library classes, provided they implement the special class method ~object.__class_getitem__.

Generics, user-defined generics <user-defined-generics> and typing.Generic
Documentation on how to implement generic classes that can be parameterized at runtime and understood by static type-checkers.

3.9

Union Type

pair: object; Union pair: union; type

A union object holds the value of the | (bitwise or) operation on multiple type objects <bltin-type-objects>. These types are intended primarily for type annotations <annotation>. The union type expression enables cleaner type hinting syntax compared to subscripting typing.Union.

X | Y | ...

Defines a union object which holds types X, Y, and so forth. X | Y means either X or Y. It is equivalent to typing.Union[X, Y]. For example, the following function expects an argument of type int or float:

def square(number: int | float) -> int | float:
    return number ** 2

[!NOTE] The | operator cannot be used at runtime to define unions where one or more members is a forward reference. For example, int | "Foo", where "Foo" is a reference to a class not yet defined, will fail at runtime. For unions which include forward references, present the whole expression as a string, e.g. "int | Foo".

union_object == other

Union objects can be tested for equality with other union objects. Details:

Unions of unions are flattened:

(int | str) | float == int | str | float

Redundant types are removed:
```
int | str | int == int | str
```
When comparing unions, the order is ignored:
```
int | str == str | int
```

It creates instances of typing.Union:

int | str == typing.Union[int, str]
type(int | str) is typing.Union

Optional types can be spelled as a union with None:
```
str | None == typing.Optional[str]
```

isinstance(obj, union_object)

issubclass(obj, union_object)

Calls to isinstance and issubclass are also supported with a union object:

>>> isinstance("", int | str)
True

However, parameterized generics <types-genericalias> in union objects cannot be checked:

>>> isinstance(1, int | list[int])  # short-circuit evaluation
True
>>> isinstance([1], int | list[int])
Traceback (most recent call last):
  ...
TypeError: isinstance() argument 2 cannot be a parameterized generic

The user-exposed type for the union object can be accessed from typing.Union and used for isinstance checks:

>>> import typing
>>> isinstance(int | str, typing.Union)
True
>>> typing.Union()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: cannot create 'typing.Union' instances

[!NOTE] The !__or__ method for type objects was added to support the syntax X | Y. If a metaclass implements !__or__, the Union may override it:

>>> class M(type): ... def __or__(self, other): ... return "Hello" ... >>> class C(metaclass=M): ... pass ... >>> C | int 'Hello' >>> int | C int | C

604 -- PEP proposing the X | Y syntax and the Union type.

3.10

3.14

Union objects are now instances of typing.Union. Previously, they were instances of types.UnionType, which remains an alias for typing.Union.

Other Built-in Types

The interpreter supports several other kinds of objects. Most of these support only one or two operations.

Modules

The only special operation on a module is attribute access: m.name, where m is a module and name accesses a name defined in m's symbol table. Module attributes can be assigned to. (Note that the import statement is not, strictly speaking, an operation on a module object; import foo does not require a module object named foo to exist, rather it requires an (external) definition for a module named foo somewhere.)

A special attribute of every module is ~object.__dict__. This is the dictionary containing the module's symbol table. Modifying this dictionary will actually change the module's symbol table, but direct assignment to the ~object.__dict__ attribute is not possible (you can write m.__dict__['a'] = 1, which defines m.a to be 1, but you can't write m.__dict__ = {}). Modifying ~object.__dict__ directly is not recommended.

Modules built into the interpreter are written like this: <module 'sys' (built-in)>. If loaded from a file, they are written as <module 'os' from '/usr/local/lib/pythonX.Y/os.pyc'>.

Classes and Class Instances

See objects and class for these.

Functions

Function objects are created by function definitions. The only operation on a function object is to call it: func(argument-list).

There are really two flavors of function objects: built-in functions and user-defined functions. Both support the same operation (to call the function), but the implementation is different, hence the different object types.

See function for more information.

Methods

pair: object; method

Methods are functions that are called using the attribute notation. There are two flavors: built-in methods <builtin-methods> (such as ~list.append on lists) and class instance methods <instance-methods>. Built-in methods are described with the types that support them.

If you access a method (a function defined in a class namespace) through an instance, you get a special object: a bound method (also called instance method <instance-methods>) object. When called, it will add the self argument to the argument list. Bound methods have two special read-only attributes: m.__self__ <method.__self__> is the object on which the method operates, and m.__func__ <method.__func__> is the function implementing the method. Calling m(arg-1, arg-2, ..., arg-n) is completely equivalent to calling m.__func__(m.__self__, arg-1, arg-2, ..., arg-n).

Like function objects <user-defined-funcs>, bound method objects support getting arbitrary attributes. However, since method attributes are actually stored on the underlying function object (method.__func__), setting method attributes on bound methods is disallowed. Attempting to set an attribute on a method results in an AttributeError being raised. In order to set a method attribute, you need to explicitly set it on the underlying function object:

>>> class C: ... def method(self): ... pass ... >>> c = C() >>> c.method.whoami = 'my name is method' # can't set on the method Traceback (most recent call last): File "<stdin>", line 1, in <module> AttributeError: 'method' object has no attribute 'whoami' >>> c.method.__func__.whoami = 'my name is method' >>> c.method.whoami 'my name is method'

See instance-methods for more information.

object; code, code object

Code Objects

pair: built-in function; compile single: __code__ (function object attribute)

Code objects are used by the implementation to represent "pseudo-compiled" executable Python code such as a function body. They differ from function objects because they don't contain a reference to their global execution environment. Code objects are returned by the built-in compile function and can be extracted from function objects through their ~function.__code__ attribute. See also the code module.

Accessing ~function.__code__ raises an auditing event <auditing> object.__getattr__ with arguments obj and "__code__".

pair: built-in function; exec pair: built-in function; eval

A code object can be executed or evaluated by passing it (instead of a source string) to the exec or eval built-in functions.

See types for more information.

Type Objects

pair: built-in function; type pair: module; types

Type objects represent the various object types. An object's type is accessed by the built-in function type. There are no special operations on types. The standard module types defines names for all standard built-in types.

Types are written like this: <class 'int'>.

The Null Object

This object is returned by functions that don't explicitly return a value. It supports no special operations. There is exactly one null object, named None (a built-in name). type(None)() produces the same singleton.

It is written as None.

single: ...; ellipsis literal

The Ellipsis Object

This object is commonly used to indicate that something is omitted. It supports no special operations. There is exactly one ellipsis object, named Ellipsis (a built-in name). type(Ellipsis)() produces the Ellipsis singleton.

It is written as Ellipsis or ....

In typical use, ... as the Ellipsis object appears in a few different places, for instance:

In type annotations, such as callable arguments <annotating-callables> or tuple elements <annotating-tuples>.
As the body of a function instead of a pass statement <tut-pass>.
In third-party libraries, such as Numpy's slicing and striding.

Python also uses three dots in ways that are not Ellipsis objects, for instance:

Doctest's ELLIPSIS <doctest.ELLIPSIS>, as a pattern for missing content.
The default Python prompt of the interactive shell when partial input is incomplete.

Lastly, the Python documentation often uses three dots in conventional English usage to mean omitted content, even in code examples that also use them as the Ellipsis.

The NotImplemented Object

This object is returned from comparisons and binary operations when they are asked to operate on types they don't support. See comparisons for more information. There is exactly one NotImplemented object. type(NotImplemented)() produces the singleton instance.

It is written as NotImplemented.

Internal Objects

See types for this information. It describes stack frame objects <frame-objects>, traceback objects <traceback-objects>, and slice objects.

Special Attributes

The implementation adds a few special read-only attributes to several object types, where they are relevant. Some of these are not reported by the dir built-in function.

definition.__name__

The name of the class, function, method, descriptor, or generator instance.

definition.__qualname__

The qualified name of the class, function, method, descriptor, or generator instance.

3.3

definition.__module__

The name of the module in which a class or function was defined.

definition.__doc__

The documentation string of a class or function, or None if undefined.

definition.__type_params__

The type parameters <type-params> of generic classes, functions, and type aliases <type-aliases>. For classes and functions that are not generic, this will be an empty tuple.

3.12

Integer string conversion length limitation

CPython has a global limit for converting between int and str to mitigate denial of service attacks. This limit only applies to decimal or other non-power-of-two number bases. Hexadecimal, octal, and binary conversions are unlimited. The limit can be configured.

The int type in CPython is an arbitrary length number stored in binary form (commonly known as a "bignum"). There exists no algorithm that can convert a string to a binary integer or a binary integer to a string in linear time, unless the base is a power of 2. Even the best known algorithms for base 10 have sub-quadratic complexity. Converting a large value such as int('1' * 500_000) can take over a second on a fast CPU.

Limiting conversion size offers a practical way to avoid 2020-10735.

The limit is applied to the number of digit characters in the input or output string when a non-linear conversion algorithm would be involved. Underscores and the sign are not counted towards the limit.

When an operation would exceed the limit, a ValueError is raised:

>>> import sys >>> sys.set_int_max_str_digits(4300) # Illustrative, this is the default. >>> _ = int('2' * 5432) Traceback (most recent call last): ... ValueError: Exceeds the limit (4300 digits) for integer string conversion: value has 5432 digits; use sys.set_int_max_str_digits() to increase the limit >>> i = int('2' * 4300) >>> len(str(i)) 4300 >>> i_squared = i*i >>> len(str(i_squared)) Traceback (most recent call last): ... ValueError: Exceeds the limit (4300 digits) for integer string conversion; use sys.set_int_max_str_digits() to increase the limit >>> len(hex(i_squared)) 7144 >>> assert int(hex(i_squared), base=16) == i*i # Hexadecimal is unlimited.

The default limit is 4300 digits as provided in sys.int_info.default_max_str_digits <sys.int_info>. The lowest limit that can be configured is 640 digits as provided in sys.int_info.str_digits_check_threshold <sys.int_info>.

Verification:

>>> import sys >>> assert sys.int_info.default_max_str_digits == 4300, sys.int_info >>> assert sys.int_info.str_digits_check_threshold == 640, sys.int_info >>> msg = int('578966293710682886880994035146873798396722250538762761564' ... '9252925514383915483333812743580549779436104706260696366600' ... '571186405732').to_bytes(53, 'big') ...

3.11

Affected APIs

The limitation only applies to potentially slow conversions between int and str or bytes:

int(string) with default base 10.
int(string, base) for all bases that are not a power of 2.
str(integer).
repr(integer).
any other string conversion to base 10, for example f"{integer}", "{}".format(integer), or b"%d" % integer.

The limitations do not apply to functions with a linear algorithm:

int(string, base) with base 2, 4, 8, 16, or 32.
int.from_bytes and int.to_bytes.
hex, oct, bin.
formatspec for hex, octal, and binary numbers.
str to float.
str to decimal.Decimal.

Configuring the limit

Before Python starts up you can use an environment variable or an interpreter command line flag to configure the limit:

PYTHONINTMAXSTRDIGITS, e.g. PYTHONINTMAXSTRDIGITS=640 python3 to set the limit to 640 or PYTHONINTMAXSTRDIGITS=0 python3 to disable the limitation.
-X int_max_str_digits <-X>, e.g. python3 -X int_max_str_digits=640
sys.flags.int_max_str_digits contains the value of PYTHONINTMAXSTRDIGITS or -X int_max_str_digits <-X>. If both the env var and the -X option are set, the -X option takes precedence. A value of -1 indicates that both were unset, thus a value of sys.int_info.default_max_str_digits was used during initialization.

From code, you can inspect the current limit and set a new one using these sys APIs:

sys.get_int_max_str_digits and sys.set_int_max_str_digits are a getter and setter for the interpreter-wide limit. Subinterpreters have their own limit.

Information about the default and minimum can be found in sys.int_info:

sys.int_info.default_max_str_digits <sys.int_info> is the compiled-in default limit.
sys.int_info.str_digits_check_threshold <sys.int_info> is the lowest accepted value for the limit (other than 0 which disables it).

3.11

[!CAUTION] Setting a low limit can lead to problems. While rare, code exists that contains integer constants in decimal in their source that exceed the minimum threshold. A consequence of setting the limit is that Python source code containing decimal integer literals longer than the limit will encounter an error during parsing, usually at startup time or import time or even at installation time - anytime an up to date .pyc does not already exist for the code. A workaround for source that contains such large constants is to convert them to 0x hexadecimal form as it has no limit.

Test your application thoroughly if you use a low limit. Ensure your tests run with the limit set early via the environment or flag so that it applies during startup and even during any installation step that may invoke Python to precompile .py sources to .pyc files.

Recommended configuration

The default sys.int_info.default_max_str_digits is expected to be reasonable for most applications. If your application requires a different limit, set it from your main entry point using Python version agnostic code as these APIs were added in security patch releases in versions before 3.12.

Example:

>>> import sys
>>> if hasattr(sys, "set_int_max_str_digits"):
...     upper_bound = 68000
...     lower_bound = 4004
...     current_limit = sys.get_int_max_str_digits()
...     if current_limit == 0 or current_limit > upper_bound:
...         sys.set_int_max_str_digits(upper_bound)
...     elif current_limit < lower_bound:
...         sys.set_int_max_str_digits(lower_bound)

If you need to disable it entirely, set it to 0.

Footnotes

[^1]: Additional information on these special methods may be found in the Python Reference Manual (customization).

[^2]: As a consequence, the list [1, 2] is considered equal to [1.0, 2.0], and similarly for tuples.

[^3]: They must have since the parser can't tell the type of the operands.

[^4]: Cased characters are those with general category property being one of "Lu" (Letter, uppercase), "Ll" (Letter, lowercase), or "Lt" (Letter, titlecase).

[^5]: Cased characters are those with general category property being one of "Lu" (Letter, uppercase), "Ll" (Letter, lowercase), or "Lt" (Letter, titlecase).

[^6]: Cased characters are those with general category property being one of "Lu" (Letter, uppercase), "Ll" (Letter, lowercase), or "Lt" (Letter, titlecase).

[^7]: Cased characters are those with general category property being one of "Lu" (Letter, uppercase), "Ll" (Letter, lowercase), or "Lt" (Letter, titlecase).

[^8]: To format only a tuple you should therefore provide a singleton tuple whose only element is the tuple to be formatted.

[^9]: To format only a tuple you should therefore provide a singleton tuple whose only element is the tuple to be formatted.

Thread Safety Guarantees

This page documents thread-safety guarantees for built-in types in Python's free-threaded build. The guarantees described here apply when using Python with the GIL disabled (free-threaded mode). When the GIL is enabled, most operations are implicitly serialized.

For general guidance on writing thread-safe code in free-threaded Python, see freethreading-python-howto.

Thread safety levels

The C API documentation uses the following levels to describe the thread safety guarantees of each function. The levels are listed from least to most safe.

Incompatible

A function or operation that cannot be made safe for concurrent use even with external synchronization. Incompatible code typically accesses global state in an unsynchronized way and must only be called from a single thread throughout the program's lifetime.

Example: a function that modifies process-wide state such as signal handlers or environment variables, where concurrent calls from any threads, even with external locking, can conflict with the runtime or other libraries.

Compatible

A function or operation that is safe to call from multiple threads provided the caller supplies appropriate external synchronization, for example by holding a lock for the duration of each call. Without such synchronization, concurrent calls may produce race conditions <race condition> or data races <data race>.

Example: a function that reads from or writes to an object whose internal state is not protected by a lock. Callers must ensure that no two threads access the same object at the same time.

Safe on distinct objects

A function or operation that is safe to call from multiple threads without external synchronization, as long as each thread operates on a different object. Two threads may call the function at the same time, but they must not pass the same object (or objects that share underlying state) as arguments.

Example: a function that modifies fields of a struct using non-atomic writes. Two threads can each call the function on their own struct instance safely, but concurrent calls on the same instance require external synchronization.

Safe on shared objects

A function or operation that is safe for concurrent use on the same object. The implementation uses internal synchronization (such as per-object locks <per-object lock> or critical sections <python-critical-section-api>) to protect shared mutable state, so callers do not need to supply their own locking.

Example: PyList_GetItemRef can be called from multiple threads on the same PyListObject - it uses internal synchronization to serialize access.

Atomic

A function or operation that appears atomic <atomic operation> with respect to other threads - it executes instantaneously from the perspective of other threads. This is the strongest form of thread safety.

Example: PyMutex_IsLocked performs an atomic read of the mutex state and can be called from any thread at any time.

Thread safety for list objects

Reading a single element from a list is atomic <atomic operation>:

lst[i]   # list.__getitem__

The following methods traverse the list and use atomic <atomic operation> reads of each item to perform their function. That means that they may return results affected by concurrent modifications:

item in lst
lst.index(item)
lst.count(item)

All of the above operations avoid acquiring per-object locks <per-object lock>. They do not block concurrent modifications. Other operations that hold a lock will not block these from observing intermediate states.

All other operations from here on block using the per-object lock.

Writing a single item via lst[i] = x is safe to call from multiple threads and will not corrupt the list.

The following operations return new objects and appear atomic <atomic operation> to other threads:

lst1 + lst2    # concatenates two lists into a new list
x * lst        # repeats lst x times into a new list
lst.copy()     # returns a shallow copy of the list

The following methods that only operate on a single element with no shifting required are atomic <atomic operation>:

lst.append(x)  # append to the end of the list, no shifting required
lst.pop()      # pop element from the end of the list, no shifting required

The ~list.clear method is also atomic <atomic operation>. Other threads cannot observe elements being removed.

The ~list.sort method is not atomic <atomic operation>. Other threads cannot observe intermediate states during sorting, but the list appears empty for the duration of the sort.

The following operations may allow lock-free operations to observe intermediate states since they modify multiple elements in place:

lst.insert(idx, item)  # shifts elements
lst.pop(idx)           # idx not at the end of the list, shifts elements
lst *= x               # copies elements in place

The ~list.remove method may allow concurrent modifications since element comparison may execute arbitrary Python code (via ~object.__eq__).

~list.extend is safe to call from multiple threads. However, its guarantees depend on the iterable passed to it. If it is a list, a tuple, a set, a frozenset, a dict or a dictionary view object <dict-views> (but not their subclasses), the extend operation is safe from concurrent modifications to the iterable. Otherwise, an iterator is created which can be concurrently modified by another thread. The same applies to inplace concatenation of a list with other iterables when using lst += iterable.

Similarly, assigning to a list slice with lst[i:j] = iterable is safe to call from multiple threads, but iterable is only locked when it is also a list (but not its subclasses).

Operations that involve multiple accesses, as well as iteration, are never atomic. For example:

# NOT atomic: read-modify-write
lst[i] = lst[i] + 1

# NOT atomic: check-then-act
if lst:
    item = lst.pop()

# NOT thread-safe: iteration while modifying
for item in lst:
    process(item)  # another thread may modify lst

Consider external synchronization when sharing list instances across threads.

Thread safety for dict objects

Creating a dictionary with the dict constructor is atomic when the argument to it is a dict or a tuple. When using the dict.fromkeys method, dictionary creation is atomic when the argument is a dict, tuple, set or frozenset.

The following operations and functions are lock-free and atomic <atomic operation>.

d[key]       # dict.__getitem__
d.get(key)   # dict.get
key in d     # dict.__contains__
len(d)       # dict.__len__

All other operations from here on hold the per-object lock.

Writing or removing a single item is safe to call from multiple threads and will not corrupt the dictionary:

d[key] = value        # write
del d[key]            # delete
d.pop(key)            # remove and return
d.popitem()           # remove and return last item
d.setdefault(key, v)  # insert if missing

These operations may compare keys using ~object.__eq__, which can execute arbitrary Python code. During such comparisons, the dictionary may be modified by another thread. For built-in types like str, int, and float, that implement ~object.__eq__ in C, the underlying lock is not released during comparisons and this is not a concern.

The following operations return new objects and hold the per-object lock for the duration of the operation:

d.copy()      # returns a shallow copy of the dictionary
d | other     # merges two dicts into a new dict
d.keys()      # returns a new dict_keys view object
d.values()    # returns a new dict_values view object
d.items()     # returns a new dict_items view object

The ~dict.clear method holds the lock for its duration. Other threads cannot observe elements being removed.

The following operations lock both dictionaries. For ~dict.update and |=, this applies only when the other operand is a dict that uses the standard dict iterator (but not subclasses that override iteration). For equality comparison, this applies to dict and its subclasses:

d.update(other_dict)  # both locked when other_dict is a dict
d |= other_dict       # both locked when other_dict is a dict
d == other_dict       # both locked for dict and subclasses

All comparison operations also compare values using ~object.__eq__, so for non-built-in types the lock may be released during comparison.

~dict.fromkeys locks both the new dictionary and the iterable when the iterable is exactly a dict, set, or frozenset (not subclasses):

dict.fromkeys(a_dict)      # locks both
dict.fromkeys(a_set)       # locks both
dict.fromkeys(a_frozenset) # locks both

When updating from a non-dict iterable, only the target dictionary is locked. The iterable may be concurrently modified by another thread:

d.update(iterable)        # iterable is not a dict: only d locked
d |= iterable             # iterable is not a dict: only d locked
dict.fromkeys(iterable)   # iterable is not a dict/set/frozenset: only result locked

Operations that involve multiple accesses, as well as iteration, are never atomic:

# NOT atomic: read-modify-write
d[key] = d[key] + 1

# NOT atomic: check-then-act (TOCTOU)
if key in d:
    del d[key]

# NOT thread-safe: iteration while modifying
for key, value in d.items():
    process(key)  # another thread may modify d

To avoid time-of-check to time-of-use (TOCTOU) issues, use atomic operations or handle exceptions:

# Use pop() with default instead of check-then-delete
d.pop(key, None)

# Or handle the exception
try:
    del d[key]
except KeyError:
    pass

To safely iterate over a dictionary that may be modified by another thread, iterate over a copy:

# Make a copy to iterate safely
for key, value in d.copy().items():
    process(key)

Consider external synchronization when sharing dict instances across threads.

Thread safety for set objects

The len function is lock-free and atomic <atomic operation>.

The following read operation is lock-free. It does not block concurrent modifications and may observe intermediate states from operations that hold the per-object lock:

elem in s    # set.__contains__

This operation may compare elements using ~object.__eq__, which can execute arbitrary Python code. During such comparisons, the set may be modified by another thread. For built-in types like str, int, and float, !__eq__ does not release the underlying lock during comparisons and this is not a concern.

All other operations from here on hold the per-object lock.

Adding or removing a single element is safe to call from multiple threads and will not corrupt the set:

s.add(elem)      # add element
s.remove(elem)   # remove element, raise if missing
s.discard(elem)  # remove element if present
s.pop()          # remove and return arbitrary element

These operations also compare elements, so the same ~object.__eq__ considerations as above apply.

The ~set.copy method returns a new object and holds the per-object lock for the duration so that it is always atomic.

The ~set.clear method holds the lock for its duration. Other threads cannot observe elements being removed.

The following operations only accept set or frozenset as operands and always lock both objects:

s |= other                   # other must be set/frozenset
s &= other                   # other must be set/frozenset
s -= other                   # other must be set/frozenset
s ^= other                   # other must be set/frozenset
s & other                    # other must be set/frozenset
s | other                    # other must be set/frozenset
s - other                    # other must be set/frozenset
s ^ other                    # other must be set/frozenset

set.update, set.union, set.intersection and set.difference can take multiple iterables as arguments. They all iterate through all the passed iterables and do the following:

set.update and set.union lock both objects only when the other operand is a set, frozenset, or dict.

set.intersection and set.difference always try to lock all objects.

set.symmetric_difference tries to lock both objects.

The update variants of the above methods also have some differences between them:

set.difference_update and set.intersection_update try to lock all objects one-by-one.

set.symmetric_difference_update only locks the arguments if it is of type set, frozenset, or dict.

The following methods always try to lock both objects:

s.isdisjoint(other)          # both locked
s.issubset(other)            # both locked
s.issuperset(other)          # both locked

Operations that involve multiple accesses, as well as iteration, are never atomic:

# NOT atomic: check-then-act
if elem in s:
      s.remove(elem)

# NOT thread-safe: iteration while modifying
for elem in s:
      process(elem)  # another thread may modify s

Consider external synchronization when sharing set instances across threads. See freethreading-python-howto for more information.

Thread safety for bytearray objects

The len function is lock-free and atomic <atomic operation>.

Concatenation and comparisons use the buffer protocol, which prevents resizing but does not hold the per-object lock. These operations may observe intermediate states from concurrent modifications:
ba + other    # may observe concurrent writes
ba == other   # may observe concurrent writes
ba < other    # may observe concurrent writes
All other operations from here on hold the per-object lock.

Reading a single element or slice is safe to call from multiple threads:
ba[i]        # bytearray.__getitem__
ba[i:j]      # slice
The following operations are safe to call from multiple threads and will not corrupt the bytearray:
ba[i] = x         # write single byte
ba[i:j] = values  # write slice
ba.append(x)      # append single byte
ba.extend(other)  # extend with iterable
ba.insert(i, x)   # insert single byte
ba.pop()          # remove and return last byte
ba.pop(i)         # remove and return byte at index
ba.remove(x)      # remove first occurrence
ba.reverse()      # reverse in place
ba.clear()        # remove all bytes
Slice assignment locks both objects when values is a bytearray:
ba[i:j] = other_bytearray  # both locked
The following operations return new objects and hold the per-object lock for the duration:
ba.copy()     # returns a shallow copy
ba * n        # repeat into new bytearray
The membership test holds the lock for its duration:
x in ba       # bytearray.__contains__
All other bytearray methods (such as ~bytearray.find, ~bytearray.replace, ~bytearray.split, ~bytearray.decode, etc.) hold the per-object lock for their duration.

Operations that involve multiple accesses, as well as iteration, are never atomic:
# NOT atomic: check-then-act
if x in ba:
    ba.remove(x)

# NOT thread-safe: iteration while modifying
for byte in ba:
    process(byte)  # another thread may modify ba
To safely iterate over a bytearray that may be modified by another thread, iterate over a copy:
# Make a copy to iterate safely
for byte in ba.copy():
    process(byte)
Consider external synchronization when sharing bytearray instances across threads. See freethreading-python-howto for more information.

Thread safety for memoryview objects

memoryview objects provide access to the internal data of an underlying object without copying. Thread safety depends on both the memoryview itself and the underlying buffer exporter.

The memoryview implementation uses atomic operations to track its own exports in the free-threaded build. Creating and releasing a memoryview are thread-safe. Attribute access (e.g., ~memoryview.shape, ~memoryview.format) reads fields that are immutable for the lifetime of the memoryview, so concurrent reads are safe as long as the memoryview has not been released.

However, the actual data accessed through the memoryview is owned by the underlying object. Concurrent access to this data is only safe if the underlying object supports it:

For immutable objects like bytes, concurrent reads through multiple memoryviews are safe.
For mutable objects like bytearray, reading and writing the same memory region from multiple threads without external synchronization is not safe and may result in data corruption. Note that even read-only memoryviews of mutable objects do not prevent data races if the underlying object is modified from another thread.

# NOT safe: concurrent writes to the same buffer
data = bytearray(1000)
view = memoryview(data)
# Thread 1: view[0:500] = b'x' * 500
# Thread 2: view[0:500] = b'y' * 500

# Safe: use a lock for concurrent access
import threading
lock = threading.Lock()
data = bytearray(1000)
view = memoryview(data)

with lock:
    view[0:500] = b'x' * 500

Resizing or reallocating the underlying object (such as calling bytearray.resize) while a memoryview is exported raises BufferError. This is enforced regardless of threading.

Time complexity of operations on built-in types

This page documents the time complexity of various operations on built-in types in CPython. Other Python implementations may have different performance characteristics. Additionally, the listed costs assume exact built-in types, as instances of subclasses may have different costs.

We use Big O notation to describe how the running time of an operation grows with the size of its inputs. Unless stated otherwise, n denotes the number of elements currently in the container, and k is the value of a numeric parameter, such as an index or a repeat count.

`!list`

Lists are mutable sequences; for more detail on the implementation see how-are-lists-implemented. The largest costs come from growing beyond the current allocation size (because everything must move), or from inserting or deleting somewhere near the beginning (because everything after that must move). If you need to add or remove at both ends, consider using a collections.deque instead.

Operation	Complexity
Copy (`l.copy()`)	O(n)
Append (`l.append(x)`)[^1]	O(1)
Pop (`l.pop(k)`)[^2][^3]	O(n - k)
Insert (`l.insert(k, x)`)[^4][^5]	O(n - k)
Get item (`l[k]`)	O(1)
Set item (`l[k] = x`)	O(1)
Delete item (`del l[k]`)[^6]	O(n - k)
Iteration	O(n)
Get slice (`l[i:j]`)	O(j - i)
Set slice (`l[i:j] = t`)[^7]	O(j - i) if len(t) == j - i, otherwise O(n - i + len(t))
Delete slice (`del l[i:j]`)	O(n - i)
Extend (`l.extend(t)`)[^8][^9]	O(len(t))
Sort (`l.sort()`)[^10]	O(n log n)
Concatenate (`l1 + l2`)	O(len(l1) + len(l2))
Multiply (`l * k`)	O(nk)
`x in l`	O(n)
`min(l)`, `max(l)`	O(n)
Get length (`len(l)`)[^11]	O(1)

`!tuple`

A tuple is an immutable sequence. Because a tuple can never change, there are no insertion or deletion costs, and making a copy simply returns the same object, so is constant time (O(1)).

Operation	Complexity
Copy (`tuple(t)`)	O(1)
Get item (`t[k]`)	O(1)
Get slice (`t[i:j]`)	O(j - i)
Concatenate (`t1 + t2`)	O(len(t1) + len(t2))
Multiply (`t * k`)	O(nk)
Iteration	O(n)
`x in t`	O(n)
`min(t)`, `max(t)`	O(n)
Get length (`len(t)`)[^12]	O(1)

`!dict`, `!frozendict`

The times listed for dict objects are average-case times, as they assume the hash function for the objects is sufficiently robust to make collisions uncommon. They also assume the keys are well-distributed among the set of possible keys. In the worst case, when every key hashes to the same value, each of the O(1) operations below instead takes O(n) time. They also assume that hashing and comparing a key is O(1). For more detail on the implementation, see how-are-dictionaries-implemented.

A frozendict is immutable, so it does not support setting, deleting, or updating items. The other operations below apply to it at the same costs.

Operation	Complexity
`key in d`	O(1)
Copy (`d.copy()`)[^13][^14]	O(n)
Get item (`d[key]`, `d.get(key)`)	O(1)
Set item (`d[key] = value`)[^15]	O(1)
Delete item (`del d[key]`, `d.pop(key)`)	O(1)
Update (`d.update(t)`, `d	= t`)[^16][^17][^18]
Iteration[^19]	O(n)
Get length (`len(d)`)[^20]	O(1)

`!set`, `!frozenset`

See dict as the set and frozenset implementations are similar, and the same caveats apply. In the worst case, O(1) operations instead take O(n) time, and operations that look up every element degrade accordingly.

A frozenset is immutable, so it does not support adding, discarding, or the in-place update operations. The others below apply to it at the same costs.

Operation	Complexity
`x in s`	O(1)
Copy (`s.copy()`)[^21][^22]	O(n)
Add (`s.add(x)`)[^23]	O(1)
Discard (`s.discard(x)`, `s.remove(x)`)	O(1)
Union (`s1	s2`,` s1.union(s2)`)[^24]
Update (`s1	= s2`,` s1.update(s2)`)[^25][^26]
Intersection (`s1 & s2`, `s1.intersection(s2)`)[^27][^28]	O(min(len(s1), len(s2)))
Intersection update (`s1 &= s2`, `s1.intersection_update(s2)`)[^29][^30][^31]	O(min(len(s1), len(s2)))
Difference (`s1 - s2`, `s1.difference(s2)`)[^32][^33]	O(len(s1))
Difference update (`s1 -= s2`, `s1.difference_update(s2)`)[^34][^35][^36]	O(min(len(s1), len(s2)))
Symmetric difference (`s1 ^ s2`, `s1.symmetric_difference(s2)`)[^37]	O(len(s1) + len(s2))
Symmetric difference update (`s1 ^= s2`, `s1.symmetric_difference_update(s2)`)[^38][^39]	O(len(s2))
Get length (`len(s)`)[^40]	O(1)

`!str`, `!bytes`, `!bytearray`

str and bytes objects are immutable sequences of characters and bytes, respectively. As with tuples, copying one returns the original object. A bytearray is mutable, and additionally supports the mutating operations of list (except !sort), at the same costs. However, deleting at the front with del (del b[0], del b[:k]) only advances the start of the buffer instead of moving the remaining bytes, and is amortized O(1).

Operation	Complexity
Get item (`s[k]`)	O(1)
Get slice (`s[i:j]`)	O(j - i)
Concatenate (`s + t`)[^41]	O(len(s) + len(t))
Multiply (`s * k`)	O(nk)
Substring search (`x in s`, `s.find(x)`, `s.index(x)`)[^42]	O(n)
Reverse substring search (`s.rfind(x)`, `s.rindex(x)`)[^43][^44]	O(n × len(x))
Encode or decode[^45]	O(n)
Iteration	O(n)
Get length (`len(s)`)[^46]	O(1)

`!memoryview`

memoryview objects allow Python code to access the internal data of an object that supports the buffer protocol <bufferobjects> without copying. In particular, slicing a memory view returns a new view onto the same buffer.

Operation	Complexity
Create (`memoryview(obj)`)	O(1)
Get item (`v[k]`)	O(1)
Get slice (`v[i:j]`)	O(1)
Index (`v.index(x)`)[^47][^48]	O(n)
Count (`v.count(x)`)[^49]	O(n)
Convert to bytes (`v.tobytes()`, `bytes(v)`)	O(n)
Get length (`len(v)`)[^50]	O(1)

`!range`

A range object computes its items on demand from its start, stop and step values, so most operations do not depend on the length of the range.

Operation	Complexity
Get item (`r[k]`)	O(1)
Get slice (`r[i:j]`)	O(1)
`x in r`[^51]	O(1)
Index and count (`r.index(x)`, `r.count(x)`)[^52]	O(1)
Iteration	O(n)
`min(r)`, `max(r)`	O(n)
Get length (`len(r)`)[^53]	O(1)

Notes

[^1]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^2]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^3]: Popping or deleting the element at index k of a list of size n shifts all elements after k one slot to the left, moving n - k - 1 elements; inserting at index k shifts the elements from k onwards one slot to the right, moving n - k elements. The worst case is index 0, where the whole rest of the list has to be moved; the average case, an index in the middle of the list, takes O(n/2) = O(n) operations; and operating at the end of the list moves nothing and is O(1).

[^4]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^5]: Popping or deleting the element at index k of a list of size n shifts all elements after k one slot to the left, moving n - k - 1 elements; inserting at index k shifts the elements from k onwards one slot to the right, moving n - k elements. The worst case is index 0, where the whole rest of the list has to be moved; the average case, an index in the middle of the list, takes O(n/2) = O(n) operations; and operating at the end of the list moves nothing and is O(1).

[^6]: Popping or deleting the element at index k of a list of size n shifts all elements after k one slot to the left, moving n - k - 1 elements; inserting at index k shifts the elements from k onwards one slot to the right, moving n - k elements. The worst case is index 0, where the whole rest of the list has to be moved; the average case, an index in the middle of the list, takes O(n/2) = O(n) operations; and operating at the end of the list moves nothing and is O(1).

[^7]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^8]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^9]: Plus the cost of iterating over t, which may be expensive for an arbitrary iterable.

[^10]: This is the worst case scenario. Sorting is adaptive and input that is already sorted or reverse-sorted takes only O(n) comparisons. See Objects/listsort.txt for more information.

[^11]: The number of elements is stored in the object, so len() does not need to count them.

[^12]: The number of elements is stored in the object, so len() does not need to count them.

[^13]: Copying a frozendict or a frozenset is O(1) as it returns the original object.

[^14]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^15]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^16]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^17]: Plus the cost of iterating over t, which may be expensive for an arbitrary iterable.

[^18]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^19]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^20]: The number of elements is stored in the object, so len() does not need to count them.

[^21]: Copying a frozendict or a frozenset is O(1) as it returns the original object.

[^22]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^23]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^24]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^25]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^26]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^27]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^28]: O(len(t)) if t is not a set.

[^29]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^30]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^31]: O(len(t)) if t is not a set.

[^32]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^33]: O(len(s) + len(t)) if t is not a set.

[^34]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^35]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^36]: O(len(t)) if t is not a set.

[^37]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^38]: Amortized. An individual operation may occasionally be O(n) when the underlying storage is resized, but this cost is spread over many operations, depending on the history of the container.

[^39]: These operations scan the container's internal hash table, which is not shrunk when elements are removed. After removing most elements, they still take time proportional to the container's former size, until a later insertion triggers a resize.

[^40]: The number of elements is stored in the object, so len() does not need to count them.

[^41]: Each concatenation builds a new object, so building a string by concatenating many pieces in a loop is quadratic in the total length. See the note on concatenating immutable sequences <typesseq-repeated-concatenation> for alternatives.

[^42]: With start and end arguments, n is the length of the region searched rather than of s, and unlike slicing nothing is copied.

[^43]: With start and end arguments, n is the length of the region searched rather than of s, and unlike slicing nothing is copied.

[^44]: This is the worst case. Reverse searches are O(n) on typical input. Forward searches instead use a more elaborate algorithm with a linear worst case, described in Objects/stringlib/stringlib_find_two_way_notes.txt.

[^45]: This assumes a codec that does a constant amount of work per character.

[^46]: The number of elements is stored in the object, so len() does not need to count them.

[^47]: With start and end arguments, n is the length of the region searched rather than of s, and unlike slicing nothing is copied.

[^48]: These unpack and compare each element individually, so they are much slower than the equivalent bytes methods.

[^49]: These unpack and compare each element individually, so they are much slower than the equivalent bytes methods.

[^50]: The number of elements is stored in the object, so len() does not need to count them.

[^51]: Assuming int or bool arguments. For other types, the range is searched like any other sequence in O(n) time.

[^52]: Assuming int or bool arguments. For other types, the range is searched like any other sequence in O(n) time.

[^53]: The number of elements is stored in the object, so len() does not need to count them.

Builtins

Built-in Constants

Constants added by the site module

Built-in Exceptions

Exception context

Inheriting from built-in exceptions

Base classes

Concrete exceptions

OS exceptions

Warnings

Exception groups

Exception hierarchy

Built-in functions

Python built-ins reference

Built-in Types

Truth Value Testing

Boolean Operations --- !and, !or, !not

Comparisons

Numeric Types --- int, float, complex

Bitwise Operations on Integer Types

Additional Methods on Integer Types

Additional Methods on Float

Additional Methods on Complex

Hashing of numeric types

Boolean Type - bool

Iteration-related types

Iterables

Iterators

Generator types

Generator iterators

Asynchronous generator iterators

Sequence Types --- list, tuple, range

Common Sequence Operations

Immutable Sequence Types

Mutable Sequence Types

Lists

Tuples

Ranges

Text and Binary Sequence Type Methods Summary

Text Sequence Type --- str

String Methods

Formatted String Literals (f-strings)

Debug specifier

Conversion specifier

Format specifier

Template String Literals (t-strings)

printf-style String Formatting

Binary Sequence Types --- bytes, bytearray, memoryview

Bytes Objects

Bytearray Objects

Bytes and Bytearray Operations

printf-style Bytes Formatting

Memory Views

Set Types --- set, frozenset

Mapping types --- !dict, !frozendict

Dictionary view objects

Frozen dictionaries

Context Manager Types

Type Annotation Types --- Generic Alias <types-genericalias>, Union <types-union>

Generic Alias Type

Standard Generic Classes

Special Attributes of GenericAlias objects

Union Type

Other Built-in Types

Modules

Classes and Class Instances

Functions

Methods

Code Objects

Type Objects

The Null Object

The Ellipsis Object

The NotImplemented Object

Internal Objects

Special Attributes

Integer string conversion length limitation

Affected APIs

Configuring the limit

Recommended configuration

Thread Safety Guarantees

Constants added by the `site` module

Boolean Operations --- `!and`, `!or`, `!not`

Numeric Types --- `int`, `float`, `complex`

Boolean Type - `bool`

Sequence Types --- `list`, `tuple`, `range`

Text Sequence Type --- `str`

`printf`-style String Formatting

Binary Sequence Types --- `bytes`, `bytearray`, `memoryview`

`printf`-style Bytes Formatting

Set Types --- `set`, `frozenset`

Mapping types --- `!dict`, `!frozendict`

Type Annotation Types --- `Generic Alias <types-genericalias>`, `Union <types-union>`

Special Attributes of `GenericAlias` objects

`!list`

`!tuple`

`!dict`, `!frozendict`

`!set`, `!frozenset`

`!str`, `!bytes`, `!bytearray`

`!memoryview`

`!range`