11 Objects, values and types
12 =========================
18 :dfn:`Objects` are Python's abstraction for data. All data in a Python program
19 is represented by objects or by relations between objects. (In a sense, and in
20 conformance to Von Neumann's model of a "stored program computer," code is also
21 represented by objects.)
26 single: identity of an object
27 single: value of an object
28 single: type of an object
29 single: mutable object
30 single: immutable object
32 Every object has an identity, a type and a value. An object's *identity* never
33 changes once it has been created; you may think of it as the object's address in
34 memory. The ':keyword:`is`' operator compares the identity of two objects; the
35 :func:`id` function returns an integer representing its identity (currently
36 implemented as its address). An object's :dfn:`type` is also unchangeable. [#]_
37 An object's type determines the operations that the object supports (e.g., "does
38 it have a length?") and also defines the possible values for objects of that
39 type. The :func:`type` function returns an object's type (which is an object
40 itself). The *value* of some objects can change. Objects whose value can
41 change are said to be *mutable*; objects whose value is unchangeable once they
42 are created are called *immutable*. (The value of an immutable container object
43 that contains a reference to a mutable object can change when the latter's value
44 is changed; however the container is still considered immutable, because the
45 collection of objects it contains cannot be changed. So, immutability is not
46 strictly the same as having an unchangeable value, it is more subtle.) An
47 object's mutability is determined by its type; for instance, numbers, strings
48 and tuples are immutable, while dictionaries and lists are mutable.
51 single: garbage collection
52 single: reference counting
53 single: unreachable object
55 Objects are never explicitly destroyed; however, when they become unreachable
56 they may be garbage-collected. An implementation is allowed to postpone garbage
57 collection or omit it altogether --- it is a matter of implementation quality
58 how garbage collection is implemented, as long as no objects are collected that
59 are still reachable. (Implementation note: the current implementation uses a
60 reference-counting scheme with (optional) delayed detection of cyclically linked
61 garbage, which collects most objects as soon as they become unreachable, but is
62 not guaranteed to collect garbage containing circular references. See the
63 documentation of the :mod:`gc` module for information on controlling the
64 collection of cyclic garbage.)
66 Note that the use of the implementation's tracing or debugging facilities may
67 keep objects alive that would normally be collectable. Also note that catching
68 an exception with a ':keyword:`try`...\ :keyword:`except`' statement may keep
71 Some objects contain references to "external" resources such as open files or
72 windows. It is understood that these resources are freed when the object is
73 garbage-collected, but since garbage collection is not guaranteed to happen,
74 such objects also provide an explicit way to release the external resource,
75 usually a :meth:`close` method. Programs are strongly recommended to explicitly
76 close such objects. The ':keyword:`try`...\ :keyword:`finally`' statement
77 provides a convenient way to do this.
79 .. index:: single: container
81 Some objects contain references to other objects; these are called *containers*.
82 Examples of containers are tuples, lists and dictionaries. The references are
83 part of a container's value. In most cases, when we talk about the value of a
84 container, we imply the values, not the identities of the contained objects;
85 however, when we talk about the mutability of a container, only the identities
86 of the immediately contained objects are implied. So, if an immutable container
87 (like a tuple) contains a reference to a mutable object, its value changes if
88 that mutable object is changed.
90 Types affect almost all aspects of object behavior. Even the importance of
91 object identity is affected in some sense: for immutable types, operations that
92 compute new values may actually return a reference to any existing object with
93 the same type and value, while for mutable objects this is not allowed. E.g.,
94 after ``a = 1; b = 1``, ``a`` and ``b`` may or may not refer to the same object
95 with the value one, depending on the implementation, but after ``c = []; d =
96 []``, ``c`` and ``d`` are guaranteed to refer to two different, unique, newly
97 created empty lists. (Note that ``c = d = []`` assigns the same object to both
103 The standard type hierarchy
104 ===========================
109 pair: type; hierarchy
110 pair: extension; module
113 Below is a list of the types that are built into Python. Extension modules
114 (written in C, Java, or other languages, depending on the implementation) can
115 define additional types. Future versions of Python may add types to the type
116 hierarchy (e.g., rational numbers, efficiently stored arrays of integers, etc.).
120 pair: special; attribute
121 triple: generic; special; attribute
123 Some of the type descriptions below contain a paragraph listing 'special
124 attributes.' These are attributes that provide access to the implementation and
125 are not intended for general use. Their definition may change in the future.
128 .. index:: object: None
130 This type has a single value. There is a single object with this value. This
131 object is accessed through the built-in name ``None``. It is used to signify the
132 absence of a value in many situations, e.g., it is returned from functions that
133 don't explicitly return anything. Its truth value is false.
136 .. index:: object: NotImplemented
138 This type has a single value. There is a single object with this value. This
139 object is accessed through the built-in name ``NotImplemented``. Numeric methods
140 and rich comparison methods may return this value if they do not implement the
141 operation for the operands provided. (The interpreter will then try the
142 reflected operation, or some other fallback, depending on the operator.) Its
146 .. index:: object: Ellipsis
148 This type has a single value. There is a single object with this value. This
149 object is accessed through the built-in name ``Ellipsis``. It is used to
150 indicate the presence of the ``...`` syntax in a slice. Its truth value is
153 :class:`numbers.Number`
154 .. index:: object: numeric
156 These are created by numeric literals and returned as results by arithmetic
157 operators and arithmetic built-in functions. Numeric objects are immutable;
158 once created their value never changes. Python numbers are of course strongly
159 related to mathematical numbers, but subject to the limitations of numerical
160 representation in computers.
162 Python distinguishes between integers, floating point numbers, and complex
165 :class:`numbers.Integral`
166 .. index:: object: integer
168 These represent elements from the mathematical set of integers (positive and
171 There are three types of integers:
175 object: plain integer
176 single: OverflowError (built-in exception)
178 These represent numbers in the range -2147483648 through 2147483647. (The range
179 may be larger on machines with a larger natural word size, but not smaller.)
180 When the result of an operation would fall outside this range, the result is
181 normally returned as a long integer (in some cases, the exception
182 :exc:`OverflowError` is raised instead). For the purpose of shift and mask
183 operations, integers are assumed to have a binary, 2's complement notation using
184 32 or more bits, and hiding no bits from the user (i.e., all 4294967296
185 different bit patterns correspond to different values).
188 .. index:: object: long integer
190 These represent numbers in an unlimited range, subject to available (virtual)
191 memory only. For the purpose of shift and mask operations, a binary
192 representation is assumed, and negative numbers are represented in a variant of
193 2's complement which gives the illusion of an infinite string of sign bits
194 extending to the left.
202 These represent the truth values False and True. The two objects representing
203 the values False and True are the only Boolean objects. The Boolean type is a
204 subtype of plain integers, and Boolean values behave like the values 0 and 1,
205 respectively, in almost all contexts, the exception being that when converted to
206 a string, the strings ``"False"`` or ``"True"`` are returned, respectively.
208 .. index:: pair: integer; representation
210 The rules for integer representation are intended to give the most meaningful
211 interpretation of shift and mask operations involving negative integers and the
212 least surprises when switching between the plain and long integer domains. Any
213 operation except left shift, if it yields a result in the plain integer domain
214 without causing overflow, will yield the same result in the long integer domain
215 or when using mixed operands.
217 :class:`numbers.Real` (:class:`float`)
219 object: floating point
220 pair: floating point; number
224 These represent machine-level double precision floating point numbers. You are
225 at the mercy of the underlying machine architecture (and C or Java
226 implementation) for the accepted range and handling of overflow. Python does not
227 support single-precision floating point numbers; the savings in processor and
228 memory usage that are usually the reason for using these is dwarfed by the
229 overhead of using objects in Python, so there is no reason to complicate the
230 language with two kinds of floating point numbers.
232 :class:`numbers.Complex`
235 pair: complex; number
237 These represent complex numbers as a pair of machine-level double precision
238 floating point numbers. The same caveats apply as for floating point numbers.
239 The real and imaginary parts of a complex number ``z`` can be retrieved through
240 the read-only attributes ``z.real`` and ``z.imag``.
246 single: index operation
247 single: item selection
250 These represent finite ordered sets indexed by non-negative numbers. The
251 built-in function :func:`len` returns the number of items of a sequence. When
252 the length of a sequence is *n*, the index set contains the numbers 0, 1,
253 ..., *n*-1. Item *i* of sequence *a* is selected by ``a[i]``.
255 .. index:: single: slicing
257 Sequences also support slicing: ``a[i:j]`` selects all items with index *k* such
258 that *i* ``<=`` *k* ``<`` *j*. When used as an expression, a slice is a
259 sequence of the same type. This implies that the index set is renumbered so
262 .. index:: single: extended slicing
264 Some sequences also support "extended slicing" with a third "step" parameter:
265 ``a[i:j:k]`` selects all items of *a* with index *x* where ``x = i + n*k``, *n*
266 ``>=`` ``0`` and *i* ``<=`` *x* ``<`` *j*.
268 Sequences are distinguished according to their mutability:
272 object: immutable sequence
275 An object of an immutable sequence type cannot change once it is created. (If
276 the object contains references to other objects, these other objects may be
277 mutable and may be changed; however, the collection of objects directly
278 referenced by an immutable object cannot change.)
280 The following types are immutable sequences:
291 The items of a string are characters. There is no separate character type; a
292 character is represented by a string of one item. Characters represent (at
293 least) 8-bit bytes. The built-in functions :func:`chr` and :func:`ord` convert
294 between characters and nonnegative integers representing the byte values. Bytes
295 with the values 0-127 usually represent the corresponding ASCII values, but the
296 interpretation of values is up to the program. The string data type is also
297 used to represent arrays of bytes, e.g., to hold data read from a file.
302 single: character set
303 pair: string; comparison
307 (On systems whose native character set is not ASCII, strings may use EBCDIC in
308 their internal representation, provided the functions :func:`chr` and
309 :func:`ord` implement a mapping between ASCII and EBCDIC, and string comparison
310 preserves the ASCII order. Or perhaps someone can propose a better rule?)
322 The items of a Unicode object are Unicode code units. A Unicode code unit is
323 represented by a Unicode object of one item and can hold either a 16-bit or
324 32-bit value representing a Unicode ordinal (the maximum value for the ordinal
325 is given in ``sys.maxunicode``, and depends on how Python is configured at
326 compile time). Surrogate pairs may be present in the Unicode object, and will
327 be reported as two separate items. The built-in functions :func:`unichr` and
328 :func:`ord` convert between code units and nonnegative integers representing the
329 Unicode ordinals as defined in the Unicode Standard 3.0. Conversion from and to
330 other encodings are possible through the Unicode method :meth:`encode` and the
331 built-in function :func:`unicode`.
336 pair: singleton; tuple
339 The items of a tuple are arbitrary Python objects. Tuples of two or more items
340 are formed by comma-separated lists of expressions. A tuple of one item (a
341 'singleton') can be formed by affixing a comma to an expression (an expression
342 by itself does not create a tuple, since parentheses must be usable for grouping
343 of expressions). An empty tuple can be formed by an empty pair of parentheses.
347 object: mutable sequence
349 pair: assignment; statement
355 Mutable sequences can be changed after they are created. The subscription and
356 slicing notations can be used as the target of assignment and :keyword:`del`
359 There is currently a single intrinsic mutable sequence type:
362 .. index:: object: list
364 The items of a list are arbitrary Python objects. Lists are formed by placing a
365 comma-separated list of expressions in square brackets. (Note that there are no
366 special cases needed to form lists of length 0 or 1.)
368 .. index:: module: array
370 The extension module :mod:`array` provides an additional example of a mutable
378 These represent unordered, finite sets of unique, immutable objects. As such,
379 they cannot be indexed by any subscript. However, they can be iterated over, and
380 the built-in function :func:`len` returns the number of items in a set. Common
381 uses for sets are fast membership testing, removing duplicates from a sequence,
382 and computing mathematical operations such as intersection, union, difference,
383 and symmetric difference.
385 For set elements, the same immutability rules apply as for dictionary keys. Note
386 that numeric types obey the normal rules for numeric comparison: if two numbers
387 compare equal (e.g., ``1`` and ``1.0``), only one of them can be contained in a
390 There are currently two intrinsic set types:
393 .. index:: object: set
395 These represent a mutable set. They are created by the built-in :func:`set`
396 constructor and can be modified afterwards by several methods, such as
400 .. index:: object: frozenset
402 These represent an immutable set. They are created by the built-in
403 :func:`frozenset` constructor. As a frozenset is immutable and
404 :term:`hashable`, it can be used again as an element of another set, or as
413 These represent finite sets of objects indexed by arbitrary index sets. The
414 subscript notation ``a[k]`` selects the item indexed by ``k`` from the mapping
415 ``a``; this can be used in expressions and as the target of assignments or
416 :keyword:`del` statements. The built-in function :func:`len` returns the number
417 of items in a mapping.
419 There is currently a single intrinsic mapping type:
422 .. index:: object: dictionary
424 These represent finite sets of objects indexed by nearly arbitrary values. The
425 only types of values not acceptable as keys are values containing lists or
426 dictionaries or other mutable types that are compared by value rather than by
427 object identity, the reason being that the efficient implementation of
428 dictionaries requires a key's hash value to remain constant. Numeric types used
429 for keys obey the normal rules for numeric comparison: if two numbers compare
430 equal (e.g., ``1`` and ``1.0``) then they can be used interchangeably to index
431 the same dictionary entry.
433 Dictionaries are mutable; they can be created by the ``{...}`` notation (see
434 section :ref:`dict`).
441 The extension modules :mod:`dbm`, :mod:`gdbm`, and :mod:`bsddb` provide
442 additional examples of mapping types.
449 pair: function; argument
451 These are the types to which the function call operation (see section
452 :ref:`calls`) can be applied:
454 User-defined functions
456 pair: user-defined; function
458 object: user-defined function
460 A user-defined function object is created by a function definition (see
461 section :ref:`function`). It should be called with an argument list
462 containing the same number of items as the function's formal parameter
467 +-----------------------+-------------------------------+-----------+
468 | Attribute | Meaning | |
469 +=======================+===============================+===========+
470 | :attr:`func_doc` | The function's documentation | Writable |
471 | | string, or ``None`` if | |
473 +-----------------------+-------------------------------+-----------+
474 | :attr:`__doc__` | Another way of spelling | Writable |
475 | | :attr:`func_doc` | |
476 +-----------------------+-------------------------------+-----------+
477 | :attr:`func_name` | The function's name | Writable |
478 +-----------------------+-------------------------------+-----------+
479 | :attr:`__name__` | Another way of spelling | Writable |
480 | | :attr:`func_name` | |
481 +-----------------------+-------------------------------+-----------+
482 | :attr:`__module__` | The name of the module the | Writable |
483 | | function was defined in, or | |
484 | | ``None`` if unavailable. | |
485 +-----------------------+-------------------------------+-----------+
486 | :attr:`func_defaults` | A tuple containing default | Writable |
487 | | argument values for those | |
488 | | arguments that have defaults, | |
489 | | or ``None`` if no arguments | |
490 | | have a default value | |
491 +-----------------------+-------------------------------+-----------+
492 | :attr:`func_code` | The code object representing | Writable |
493 | | the compiled function body. | |
494 +-----------------------+-------------------------------+-----------+
495 | :attr:`func_globals` | A reference to the dictionary | Read-only |
496 | | that holds the function's | |
497 | | global variables --- the | |
498 | | global namespace of the | |
499 | | module in which the function | |
501 +-----------------------+-------------------------------+-----------+
502 | :attr:`func_dict` | The namespace supporting | Writable |
503 | | arbitrary function | |
505 +-----------------------+-------------------------------+-----------+
506 | :attr:`func_closure` | ``None`` or a tuple of cells | Read-only |
507 | | that contain bindings for the | |
508 | | function's free variables. | |
509 +-----------------------+-------------------------------+-----------+
511 Most of the attributes labelled "Writable" check the type of the assigned value.
513 .. versionchanged:: 2.4
514 ``func_name`` is now writable.
516 Function objects also support getting and setting arbitrary attributes, which
517 can be used, for example, to attach metadata to functions. Regular attribute
518 dot-notation is used to get and set such attributes. *Note that the current
519 implementation only supports function attributes on user-defined functions.
520 Function attributes on built-in functions may be supported in the future.*
522 Additional information about a function's definition can be retrieved from its
523 code object; see the description of internal types below.
526 single: func_doc (function attribute)
527 single: __doc__ (function attribute)
528 single: __name__ (function attribute)
529 single: __module__ (function attribute)
530 single: __dict__ (function attribute)
531 single: func_defaults (function attribute)
532 single: func_closure (function attribute)
533 single: func_code (function attribute)
534 single: func_globals (function attribute)
535 single: func_dict (function attribute)
536 pair: global; namespace
541 object: user-defined method
542 pair: user-defined; method
544 A user-defined method object combines a class, a class instance (or ``None``)
545 and any callable object (normally a user-defined function).
547 Special read-only attributes: :attr:`im_self` is the class instance object,
548 :attr:`im_func` is the function object; :attr:`im_class` is the class of
549 :attr:`im_self` for bound methods or the class that asked for the method for
550 unbound methods; :attr:`__doc__` is the method's documentation (same as
551 ``im_func.__doc__``); :attr:`__name__` is the method name (same as
552 ``im_func.__name__``); :attr:`__module__` is the name of the module the method
553 was defined in, or ``None`` if unavailable.
555 .. versionchanged:: 2.2
556 :attr:`im_self` used to refer to the class that defined the method.
558 .. versionchanged:: 2.6
559 For 3.0 forward-compatibility, :attr:`im_func` is also available as
560 :attr:`__func__`, and :attr:`im_self` as :attr:`__self__`.
563 single: __doc__ (method attribute)
564 single: __name__ (method attribute)
565 single: __module__ (method attribute)
566 single: im_func (method attribute)
567 single: im_self (method attribute)
569 Methods also support accessing (but not setting) the arbitrary function
570 attributes on the underlying function object.
572 User-defined method objects may be created when getting an attribute of a class
573 (perhaps via an instance of that class), if that attribute is a user-defined
574 function object, an unbound user-defined method object, or a class method
575 object. When the attribute is a user-defined method object, a new method object
576 is only created if the class from which it is being retrieved is the same as, or
577 a derived class of, the class stored in the original method object; otherwise,
578 the original method object is used as it is.
581 single: im_class (method attribute)
582 single: im_func (method attribute)
583 single: im_self (method attribute)
585 When a user-defined method object is created by retrieving a user-defined
586 function object from a class, its :attr:`im_self` attribute is ``None``
587 and the method object is said to be unbound. When one is created by
588 retrieving a user-defined function object from a class via one of its
589 instances, its :attr:`im_self` attribute is the instance, and the method
590 object is said to be bound. In either case, the new method's
591 :attr:`im_class` attribute is the class from which the retrieval takes
592 place, and its :attr:`im_func` attribute is the original function object.
594 .. index:: single: im_func (method attribute)
596 When a user-defined method object is created by retrieving another method object
597 from a class or instance, the behaviour is the same as for a function object,
598 except that the :attr:`im_func` attribute of the new instance is not the
599 original method object but its :attr:`im_func` attribute.
602 single: im_class (method attribute)
603 single: im_func (method attribute)
604 single: im_self (method attribute)
606 When a user-defined method object is created by retrieving a class method object
607 from a class or instance, its :attr:`im_self` attribute is the class itself (the
608 same as the :attr:`im_class` attribute), and its :attr:`im_func` attribute is
609 the function object underlying the class method.
611 When an unbound user-defined method object is called, the underlying function
612 (:attr:`im_func`) is called, with the restriction that the first argument must
613 be an instance of the proper class (:attr:`im_class`) or of a derived class
616 When a bound user-defined method object is called, the underlying function
617 (:attr:`im_func`) is called, inserting the class instance (:attr:`im_self`) in
618 front of the argument list. For instance, when :class:`C` is a class which
619 contains a definition for a function :meth:`f`, and ``x`` is an instance of
620 :class:`C`, calling ``x.f(1)`` is equivalent to calling ``C.f(x, 1)``.
622 When a user-defined method object is derived from a class method object, the
623 "class instance" stored in :attr:`im_self` will actually be the class itself, so
624 that calling either ``x.f(1)`` or ``C.f(1)`` is equivalent to calling ``f(C,1)``
625 where ``f`` is the underlying function.
627 Note that the transformation from function object to (unbound or bound) method
628 object happens each time the attribute is retrieved from the class or instance.
629 In some cases, a fruitful optimization is to assign the attribute to a local
630 variable and call that local variable. Also notice that this transformation only
631 happens for user-defined functions; other callable objects (and all non-callable
632 objects) are retrieved without transformation. It is also important to note
633 that user-defined functions which are attributes of a class instance are not
634 converted to bound methods; this *only* happens when the function is an
635 attribute of the class.
639 single: generator; function
640 single: generator; iterator
642 A function or method which uses the :keyword:`yield` statement (see section
643 :ref:`yield`) is called a :dfn:`generator
644 function`. Such a function, when called, always returns an iterator object
645 which can be used to execute the body of the function: calling the iterator's
646 :meth:`next` method will cause the function to execute until it provides a value
647 using the :keyword:`yield` statement. When the function executes a
648 :keyword:`return` statement or falls off the end, a :exc:`StopIteration`
649 exception is raised and the iterator will have reached the end of the set of
650 values to be returned.
654 object: built-in function
658 A built-in function object is a wrapper around a C function. Examples of
659 built-in functions are :func:`len` and :func:`math.sin` (:mod:`math` is a
660 standard built-in module). The number and type of the arguments are
661 determined by the C function. Special read-only attributes:
662 :attr:`__doc__` is the function's documentation string, or ``None`` if
663 unavailable; :attr:`__name__` is the function's name; :attr:`__self__` is
664 set to ``None`` (but see the next item); :attr:`__module__` is the name of
665 the module the function was defined in or ``None`` if unavailable.
669 object: built-in method
671 pair: built-in; method
673 This is really a different disguise of a built-in function, this time containing
674 an object passed to the C function as an implicit extra argument. An example of
675 a built-in method is ``alist.append()``, assuming *alist* is a list object. In
676 this case, the special read-only attribute :attr:`__self__` is set to the object
680 Class types, or "new-style classes," are callable. These objects normally act
681 as factories for new instances of themselves, but variations are possible for
682 class types that override :meth:`__new__`. The arguments of the call are passed
683 to :meth:`__new__` and, in the typical case, to :meth:`__init__` to initialize
688 single: __init__() (object method)
690 object: class instance
692 pair: class object; call
694 Class objects are described below. When a class object is called, a new class
695 instance (also described below) is created and returned. This implies a call to
696 the class's :meth:`__init__` method if it has one. Any arguments are passed on
697 to the :meth:`__init__` method. If there is no :meth:`__init__` method, the
698 class must be called without arguments.
701 Class instances are described below. Class instances are callable only when the
702 class has a :meth:`__call__` method; ``x(arguments)`` is a shorthand for
703 ``x.__call__(arguments)``.
710 Modules are imported by the :keyword:`import` statement (see section
711 :ref:`import`). A module object has a
712 namespace implemented by a dictionary object (this is the dictionary referenced
713 by the func_globals attribute of functions defined in the module). Attribute
714 references are translated to lookups in this dictionary, e.g., ``m.x`` is
715 equivalent to ``m.__dict__["x"]``. A module object does not contain the code
716 object used to initialize the module (since it isn't needed once the
717 initialization is done).
719 Attribute assignment updates the module's namespace dictionary, e.g., ``m.x =
720 1`` is equivalent to ``m.__dict__["x"] = 1``.
722 .. index:: single: __dict__ (module attribute)
724 Special read-only attribute: :attr:`__dict__` is the module's namespace as a
728 single: __name__ (module attribute)
729 single: __doc__ (module attribute)
730 single: __file__ (module attribute)
731 pair: module; namespace
733 Predefined (writable) attributes: :attr:`__name__` is the module's name;
734 :attr:`__doc__` is the module's documentation string, or ``None`` if
735 unavailable; :attr:`__file__` is the pathname of the file from which the module
736 was loaded, if it was loaded from a file. The :attr:`__file__` attribute is not
737 present for C modules that are statically linked into the interpreter; for
738 extension modules loaded dynamically from a shared library, it is the pathname
739 of the shared library file.
742 Class objects are created by class definitions (see section :ref:`class`). A
743 class has a namespace implemented by a dictionary object. Class attribute
744 references are translated to lookups in this dictionary, e.g., ``C.x`` is
745 translated to ``C.__dict__["x"]``. When the attribute name is not found
746 there, the attribute search continues in the base classes. The search is
747 depth-first, left-to-right in the order of occurrence in the base class list.
751 object: class instance
753 pair: class object; call
756 pair: class; attribute
758 When a class attribute reference (for class :class:`C`, say) would yield a
759 user-defined function object or an unbound user-defined method object whose
760 associated class is either :class:`C` or one of its base classes, it is
761 transformed into an unbound user-defined method object whose :attr:`im_class`
762 attribute is :class:`C`. When it would yield a class method object, it is
763 transformed into a bound user-defined method object whose :attr:`im_class`
764 and :attr:`im_self` attributes are both :class:`C`. When it would yield a
765 static method object, it is transformed into the object wrapped by the static
766 method object. See section :ref:`descriptors` for another way in which
767 attributes retrieved from a class may differ from those actually contained in
768 its :attr:`__dict__`.
770 .. index:: triple: class; attribute; assignment
772 Class attribute assignments update the class's dictionary, never the dictionary
775 .. index:: pair: class object; call
777 A class object can be called (see above) to yield a class instance (see below).
780 single: __name__ (class attribute)
781 single: __module__ (class attribute)
782 single: __dict__ (class attribute)
783 single: __bases__ (class attribute)
784 single: __doc__ (class attribute)
786 Special attributes: :attr:`__name__` is the class name; :attr:`__module__` is
787 the module name in which the class was defined; :attr:`__dict__` is the
788 dictionary containing the class's namespace; :attr:`__bases__` is a tuple
789 (possibly empty or a singleton) containing the base classes, in the order of
790 their occurrence in the base class list; :attr:`__doc__` is the class's
791 documentation string, or None if undefined.
795 object: class instance
797 pair: class; instance
798 pair: class instance; attribute
800 A class instance is created by calling a class object (see above). A class
801 instance has a namespace implemented as a dictionary which is the first place in
802 which attribute references are searched. When an attribute is not found there,
803 and the instance's class has an attribute by that name, the search continues
804 with the class attributes. If a class attribute is found that is a user-defined
805 function object or an unbound user-defined method object whose associated class
806 is the class (call it :class:`C`) of the instance for which the attribute
807 reference was initiated or one of its bases, it is transformed into a bound
808 user-defined method object whose :attr:`im_class` attribute is :class:`C` and
809 whose :attr:`im_self` attribute is the instance. Static method and class method
810 objects are also transformed, as if they had been retrieved from class
811 :class:`C`; see above under "Classes". See section :ref:`descriptors` for
812 another way in which attributes of a class retrieved via its instances may
813 differ from the objects actually stored in the class's :attr:`__dict__`. If no
814 class attribute is found, and the object's class has a :meth:`__getattr__`
815 method, that is called to satisfy the lookup.
817 .. index:: triple: class instance; attribute; assignment
819 Attribute assignments and deletions update the instance's dictionary, never a
820 class's dictionary. If the class has a :meth:`__setattr__` or
821 :meth:`__delattr__` method, this is called instead of updating the instance
829 Class instances can pretend to be numbers, sequences, or mappings if they have
830 methods with certain special names. See section :ref:`specialnames`.
833 single: __dict__ (instance attribute)
834 single: __class__ (instance attribute)
836 Special attributes: :attr:`__dict__` is the attribute dictionary;
837 :attr:`__class__` is the instance's class.
843 single: popen() (in module os)
844 single: makefile() (socket method)
849 single: stdin (in module sys)
850 single: stdout (in module sys)
851 single: stderr (in module sys)
853 A file object represents an open file. File objects are created by the
854 :func:`open` built-in function, and also by :func:`os.popen`,
855 :func:`os.fdopen`, and the :meth:`makefile` method of socket objects (and
856 perhaps by other functions or methods provided by extension modules). The
857 objects ``sys.stdin``, ``sys.stdout`` and ``sys.stderr`` are initialized to
858 file objects corresponding to the interpreter's standard input, output and
859 error streams. See :ref:`bltin-file-objects` for complete documentation of
864 single: internal type
865 single: types, internal
867 A few types used internally by the interpreter are exposed to the user. Their
868 definitions may change with future versions of the interpreter, but they are
869 mentioned here for completeness.
876 Code objects represent *byte-compiled* executable Python code, or :term:`bytecode`.
877 The difference between a code object and a function object is that the function
878 object contains an explicit reference to the function's globals (the module in
879 which it was defined), while a code object contains no context; also the default
880 argument values are stored in the function object, not in the code object
881 (because they represent values calculated at run-time). Unlike function
882 objects, code objects are immutable and contain no references (directly or
883 indirectly) to mutable objects.
885 Special read-only attributes: :attr:`co_name` gives the function name;
886 :attr:`co_argcount` is the number of positional arguments (including arguments
887 with default values); :attr:`co_nlocals` is the number of local variables used
888 by the function (including arguments); :attr:`co_varnames` is a tuple containing
889 the names of the local variables (starting with the argument names);
890 :attr:`co_cellvars` is a tuple containing the names of local variables that are
891 referenced by nested functions; :attr:`co_freevars` is a tuple containing the
892 names of free variables; :attr:`co_code` is a string representing the sequence
893 of bytecode instructions; :attr:`co_consts` is a tuple containing the literals
894 used by the bytecode; :attr:`co_names` is a tuple containing the names used by
895 the bytecode; :attr:`co_filename` is the filename from which the code was
896 compiled; :attr:`co_firstlineno` is the first line number of the function;
897 :attr:`co_lnotab` is a string encoding the mapping from bytecode offsets to
898 line numbers (for details see the source code of the interpreter);
899 :attr:`co_stacksize` is the required stack size (including local variables);
900 :attr:`co_flags` is an integer encoding a number of flags for the interpreter.
903 single: co_argcount (code object attribute)
904 single: co_code (code object attribute)
905 single: co_consts (code object attribute)
906 single: co_filename (code object attribute)
907 single: co_firstlineno (code object attribute)
908 single: co_flags (code object attribute)
909 single: co_lnotab (code object attribute)
910 single: co_name (code object attribute)
911 single: co_names (code object attribute)
912 single: co_nlocals (code object attribute)
913 single: co_stacksize (code object attribute)
914 single: co_varnames (code object attribute)
915 single: co_cellvars (code object attribute)
916 single: co_freevars (code object attribute)
918 .. index:: object: generator
920 The following flag bits are defined for :attr:`co_flags`: bit ``0x04`` is set if
921 the function uses the ``*arguments`` syntax to accept an arbitrary number of
922 positional arguments; bit ``0x08`` is set if the function uses the
923 ``**keywords`` syntax to accept arbitrary keyword arguments; bit ``0x20`` is set
924 if the function is a generator.
926 Future feature declarations (``from __future__ import division``) also use bits
927 in :attr:`co_flags` to indicate whether a code object was compiled with a
928 particular feature enabled: bit ``0x2000`` is set if the function was compiled
929 with future division enabled; bits ``0x10`` and ``0x1000`` were used in earlier
932 Other bits in :attr:`co_flags` are reserved for internal use.
934 .. index:: single: documentation string
936 If a code object represents a function, the first item in :attr:`co_consts` is
937 the documentation string of the function, or ``None`` if undefined.
940 .. index:: object: frame
942 Frame objects represent execution frames. They may occur in traceback objects
946 single: f_back (frame attribute)
947 single: f_code (frame attribute)
948 single: f_globals (frame attribute)
949 single: f_locals (frame attribute)
950 single: f_lasti (frame attribute)
951 single: f_builtins (frame attribute)
952 single: f_restricted (frame attribute)
954 Special read-only attributes: :attr:`f_back` is to the previous stack frame
955 (towards the caller), or ``None`` if this is the bottom stack frame;
956 :attr:`f_code` is the code object being executed in this frame; :attr:`f_locals`
957 is the dictionary used to look up local variables; :attr:`f_globals` is used for
958 global variables; :attr:`f_builtins` is used for built-in (intrinsic) names;
959 :attr:`f_restricted` is a flag indicating whether the function is executing in
960 restricted execution mode; :attr:`f_lasti` gives the precise instruction (this
961 is an index into the bytecode string of the code object).
964 single: f_trace (frame attribute)
965 single: f_exc_type (frame attribute)
966 single: f_exc_value (frame attribute)
967 single: f_exc_traceback (frame attribute)
968 single: f_lineno (frame attribute)
970 Special writable attributes: :attr:`f_trace`, if not ``None``, is a function
971 called at the start of each source code line (this is used by the debugger);
972 :attr:`f_exc_type`, :attr:`f_exc_value`, :attr:`f_exc_traceback` represent the
973 last exception raised in the parent frame provided another exception was ever
974 raised in the current frame (in all other cases they are None); :attr:`f_lineno`
975 is the current line number of the frame --- writing to this from within a trace
976 function jumps to the given line (only for the bottom-most frame). A debugger
977 can implement a Jump command (aka Set Next Statement) by writing to f_lineno.
983 pair: exception; handler
984 pair: execution; stack
985 single: exc_info (in module sys)
986 single: exc_traceback (in module sys)
987 single: last_traceback (in module sys)
989 single: sys.exc_traceback
990 single: sys.last_traceback
992 Traceback objects represent a stack trace of an exception. A traceback object
993 is created when an exception occurs. When the search for an exception handler
994 unwinds the execution stack, at each unwound level a traceback object is
995 inserted in front of the current traceback. When an exception handler is
996 entered, the stack trace is made available to the program. (See section
997 :ref:`try`.) It is accessible as ``sys.exc_traceback``,
998 and also as the third item of the tuple returned by ``sys.exc_info()``. The
999 latter is the preferred interface, since it works correctly when the program is
1000 using multiple threads. When the program contains no suitable handler, the stack
1001 trace is written (nicely formatted) to the standard error stream; if the
1002 interpreter is interactive, it is also made available to the user as
1003 ``sys.last_traceback``.
1006 single: tb_next (traceback attribute)
1007 single: tb_frame (traceback attribute)
1008 single: tb_lineno (traceback attribute)
1009 single: tb_lasti (traceback attribute)
1012 Special read-only attributes: :attr:`tb_next` is the next level in the stack
1013 trace (towards the frame where the exception occurred), or ``None`` if there is
1014 no next level; :attr:`tb_frame` points to the execution frame of the current
1015 level; :attr:`tb_lineno` gives the line number where the exception occurred;
1016 :attr:`tb_lasti` indicates the precise instruction. The line number and last
1017 instruction in the traceback may differ from the line number of its frame object
1018 if the exception occurred in a :keyword:`try` statement with no matching except
1019 clause or with a finally clause.
1022 .. index:: builtin: slice
1024 Slice objects are used to represent slices when *extended slice syntax* is used.
1025 This is a slice using two colons, or multiple slices or ellipses separated by
1026 commas, e.g., ``a[i:j:step]``, ``a[i:j, k:l]``, or ``a[..., i:j]``. They are
1027 also created by the built-in :func:`slice` function.
1030 single: start (slice object attribute)
1031 single: stop (slice object attribute)
1032 single: step (slice object attribute)
1034 Special read-only attributes: :attr:`start` is the lower bound; :attr:`stop` is
1035 the upper bound; :attr:`step` is the step value; each is ``None`` if omitted.
1036 These attributes can have any type.
1038 Slice objects support one method:
1041 .. method:: slice.indices(self, length)
1043 This method takes a single integer argument *length* and computes information
1044 about the extended slice that the slice object would describe if applied to a
1045 sequence of *length* items. It returns a tuple of three integers; respectively
1046 these are the *start* and *stop* indices and the *step* or stride length of the
1047 slice. Missing or out-of-bounds indices are handled in a manner consistent with
1050 .. versionadded:: 2.3
1052 Static method objects
1053 Static method objects provide a way of defeating the transformation of function
1054 objects to method objects described above. A static method object is a wrapper
1055 around any other object, usually a user-defined method object. When a static
1056 method object is retrieved from a class or a class instance, the object actually
1057 returned is the wrapped object, which is not subject to any further
1058 transformation. Static method objects are not themselves callable, although the
1059 objects they wrap usually are. Static method objects are created by the built-in
1060 :func:`staticmethod` constructor.
1062 Class method objects
1063 A class method object, like a static method object, is a wrapper around another
1064 object that alters the way in which that object is retrieved from classes and
1065 class instances. The behaviour of class method objects upon such retrieval is
1066 described above, under "User-defined methods". Class method objects are created
1067 by the built-in :func:`classmethod` constructor.
1072 New-style and classic classes
1073 =============================
1075 Classes and instances come in two flavors: old-style or classic, and new-style.
1077 Up to Python 2.1, old-style classes were the only flavour available to the user.
1078 The concept of (old-style) class is unrelated to the concept of type: if *x* is
1079 an instance of an old-style class, then ``x.__class__`` designates the class of
1080 *x*, but ``type(x)`` is always ``<type 'instance'>``. This reflects the fact
1081 that all old-style instances, independently of their class, are implemented with
1082 a single built-in type, called ``instance``.
1084 New-style classes were introduced in Python 2.2 to unify classes and types. A
1085 new-style class is neither more nor less than a user-defined type. If *x* is an
1086 instance of a new-style class, then ``type(x)`` is the same as ``x.__class__``.
1088 The major motivation for introducing new-style classes is to provide a unified
1089 object model with a full meta-model. It also has a number of immediate
1090 benefits, like the ability to subclass most built-in types, or the introduction
1091 of "descriptors", which enable computed properties.
1093 For compatibility reasons, classes are still old-style by default. New-style
1094 classes are created by specifying another new-style class (i.e. a type) as a
1095 parent class, or the "top-level type" :class:`object` if no other parent is
1096 needed. The behaviour of new-style classes differs from that of old-style
1097 classes in a number of important details in addition to what :func:`type`
1098 returns. Some of these changes are fundamental to the new object model, like
1099 the way special methods are invoked. Others are "fixes" that could not be
1100 implemented before for compatibility concerns, like the method resolution order
1101 in case of multiple inheritance.
1103 This manual is not up-to-date with respect to new-style classes. For now,
1104 please see http://www.python.org/doc/newstyle/ for more information.
1107 single: class; new-style
1108 single: class; classic
1109 single: class; old-style
1111 The plan is to eventually drop old-style classes, leaving only the semantics of
1112 new-style classes. This change will probably only be feasible in Python 3.0.
1117 Special method names
1118 ====================
1121 pair: operator; overloading
1122 single: __getitem__() (mapping object method)
1124 A class can implement certain operations that are invoked by special syntax
1125 (such as arithmetic operations or subscripting and slicing) by defining methods
1126 with special names. This is Python's approach to :dfn:`operator overloading`,
1127 allowing classes to define their own behavior with respect to language
1128 operators. For instance, if a class defines a method named :meth:`__getitem__`,
1129 and ``x`` is an instance of this class, then ``x[i]`` is equivalent [#]_ to
1130 ``x.__getitem__(i)``. Except where mentioned, attempts to execute an operation
1131 raise an exception when no appropriate method is defined.
1133 For new-style classes, special methods are only guaranteed to work if defined in
1134 an object's class, not in the object's instance dictionary. That explains why
1141 >>> c.__len__ = lambda: 5
1143 Traceback (most recent call last):
1144 File "<stdin>", line 1, in <module>
1145 TypeError: object of type 'C' has no len()
1148 When implementing a class that emulates any built-in type, it is important that
1149 the emulation only be implemented to the degree that it makes sense for the
1150 object being modelled. For example, some sequences may work well with retrieval
1151 of individual elements, but extracting a slice may not make sense. (One example
1152 of this is the :class:`NodeList` interface in the W3C's Document Object Model.)
1161 .. method:: object.__new__(cls[, ...])
1163 Called to create a new instance of class *cls*. :meth:`__new__` is a static
1164 method (special-cased so you need not declare it as such) that takes the class
1165 of which an instance was requested as its first argument. The remaining
1166 arguments are those passed to the object constructor expression (the call to the
1167 class). The return value of :meth:`__new__` should be the new object instance
1168 (usually an instance of *cls*).
1170 Typical implementations create a new instance of the class by invoking the
1171 superclass's :meth:`__new__` method using ``super(currentclass,
1172 cls).__new__(cls[, ...])`` with appropriate arguments and then modifying the
1173 newly-created instance as necessary before returning it.
1175 If :meth:`__new__` returns an instance of *cls*, then the new instance's
1176 :meth:`__init__` method will be invoked like ``__init__(self[, ...])``, where
1177 *self* is the new instance and the remaining arguments are the same as were
1178 passed to :meth:`__new__`.
1180 If :meth:`__new__` does not return an instance of *cls*, then the new instance's
1181 :meth:`__init__` method will not be invoked.
1183 :meth:`__new__` is intended mainly to allow subclasses of immutable types (like
1184 int, str, or tuple) to customize instance creation. It is also commonly
1185 overridden in custom metaclasses in order to customize class creation.
1188 .. method:: object.__init__(self[, ...])
1190 .. index:: pair: class; constructor
1192 Called when the instance is created. The arguments are those passed to the
1193 class constructor expression. If a base class has an :meth:`__init__` method,
1194 the derived class's :meth:`__init__` method, if any, must explicitly call it to
1195 ensure proper initialization of the base class part of the instance; for
1196 example: ``BaseClass.__init__(self, [args...])``. As a special constraint on
1197 constructors, no value may be returned; doing so will cause a :exc:`TypeError`
1198 to be raised at runtime.
1201 .. method:: object.__del__(self)
1207 Called when the instance is about to be destroyed. This is also called a
1208 destructor. If a base class has a :meth:`__del__` method, the derived class's
1209 :meth:`__del__` method, if any, must explicitly call it to ensure proper
1210 deletion of the base class part of the instance. Note that it is possible
1211 (though not recommended!) for the :meth:`__del__` method to postpone destruction
1212 of the instance by creating a new reference to it. It may then be called at a
1213 later time when this new reference is deleted. It is not guaranteed that
1214 :meth:`__del__` methods are called for objects that still exist when the
1219 ``del x`` doesn't directly call ``x.__del__()`` --- the former decrements
1220 the reference count for ``x`` by one, and the latter is only called when
1221 ``x``'s reference count reaches zero. Some common situations that may
1222 prevent the reference count of an object from going to zero include:
1223 circular references between objects (e.g., a doubly-linked list or a tree
1224 data structure with parent and child pointers); a reference to the object
1225 on the stack frame of a function that caught an exception (the traceback
1226 stored in ``sys.exc_traceback`` keeps the stack frame alive); or a
1227 reference to the object on the stack frame that raised an unhandled
1228 exception in interactive mode (the traceback stored in
1229 ``sys.last_traceback`` keeps the stack frame alive). The first situation
1230 can only be remedied by explicitly breaking the cycles; the latter two
1231 situations can be resolved by storing ``None`` in ``sys.exc_traceback`` or
1232 ``sys.last_traceback``. Circular references which are garbage are
1233 detected when the option cycle detector is enabled (it's on by default),
1234 but can only be cleaned up if there are no Python-level :meth:`__del__`
1235 methods involved. Refer to the documentation for the :mod:`gc` module for
1236 more information about how :meth:`__del__` methods are handled by the
1237 cycle detector, particularly the description of the ``garbage`` value.
1241 Due to the precarious circumstances under which :meth:`__del__` methods are
1242 invoked, exceptions that occur during their execution are ignored, and a warning
1243 is printed to ``sys.stderr`` instead. Also, when :meth:`__del__` is invoked in
1244 response to a module being deleted (e.g., when execution of the program is
1245 done), other globals referenced by the :meth:`__del__` method may already have
1246 been deleted. For this reason, :meth:`__del__` methods should do the absolute
1247 minimum needed to maintain external invariants. Starting with version 1.5,
1248 Python guarantees that globals whose name begins with a single underscore are
1249 deleted from their module before other globals are deleted; if no other
1250 references to such globals exist, this may help in assuring that imported
1251 modules are still available at the time when the :meth:`__del__` method is
1255 .. method:: object.__repr__(self)
1257 .. index:: builtin: repr
1259 Called by the :func:`repr` built-in function and by string conversions (reverse
1260 quotes) to compute the "official" string representation of an object. If at all
1261 possible, this should look like a valid Python expression that could be used to
1262 recreate an object with the same value (given an appropriate environment). If
1263 this is not possible, a string of the form ``<...some useful description...>``
1264 should be returned. The return value must be a string object. If a class
1265 defines :meth:`__repr__` but not :meth:`__str__`, then :meth:`__repr__` is also
1266 used when an "informal" string representation of instances of that class is
1270 pair: string; conversion
1271 pair: reverse; quotes
1272 pair: backward; quotes
1275 This is typically used for debugging, so it is important that the representation
1276 is information-rich and unambiguous.
1279 .. method:: object.__str__(self)
1285 Called by the :func:`str` built-in function and by the :keyword:`print`
1286 statement to compute the "informal" string representation of an object. This
1287 differs from :meth:`__repr__` in that it does not have to be a valid Python
1288 expression: a more convenient or concise representation may be used instead.
1289 The return value must be a string object.
1292 .. method:: object.__lt__(self, other)
1293 object.__le__(self, other)
1294 object.__eq__(self, other)
1295 object.__ne__(self, other)
1296 object.__gt__(self, other)
1297 object.__ge__(self, other)
1299 .. versionadded:: 2.1
1304 These are the so-called "rich comparison" methods, and are called for comparison
1305 operators in preference to :meth:`__cmp__` below. The correspondence between
1306 operator symbols and method names is as follows: ``x<y`` calls ``x.__lt__(y)``,
1307 ``x<=y`` calls ``x.__le__(y)``, ``x==y`` calls ``x.__eq__(y)``, ``x!=y`` and
1308 ``x<>y`` call ``x.__ne__(y)``, ``x>y`` calls ``x.__gt__(y)``, and ``x>=y`` calls
1311 A rich comparison method may return the singleton ``NotImplemented`` if it does
1312 not implement the operation for a given pair of arguments. By convention,
1313 ``False`` and ``True`` are returned for a successful comparison. However, these
1314 methods can return any value, so if the comparison operator is used in a Boolean
1315 context (e.g., in the condition of an ``if`` statement), Python will call
1316 :func:`bool` on the value to determine if the result is true or false.
1318 There are no implied relationships among the comparison operators. The truth
1319 of ``x==y`` does not imply that ``x!=y`` is false. Accordingly, when
1320 defining :meth:`__eq__`, one should also define :meth:`__ne__` so that the
1321 operators will behave as expected. See the paragraph on :meth:`__hash__` for
1322 some important notes on creating :term:`hashable` objects which support
1323 custom comparison operations and are usable as dictionary keys.
1325 There are no swapped-argument versions of these methods (to be used when the
1326 left argument does not support the operation but the right argument does);
1327 rather, :meth:`__lt__` and :meth:`__gt__` are each other's reflection,
1328 :meth:`__le__` and :meth:`__ge__` are each other's reflection, and
1329 :meth:`__eq__` and :meth:`__ne__` are their own reflection.
1331 Arguments to rich comparison methods are never coerced.
1334 .. method:: object.__cmp__(self, other)
1340 Called by comparison operations if rich comparison (see above) is not
1341 defined. Should return a negative integer if ``self < other``, zero if
1342 ``self == other``, a positive integer if ``self > other``. If no
1343 :meth:`__cmp__`, :meth:`__eq__` or :meth:`__ne__` operation is defined, class
1344 instances are compared by object identity ("address"). See also the
1345 description of :meth:`__hash__` for some important notes on creating
1346 :term:`hashable` objects which support custom comparison operations and are
1347 usable as dictionary keys. (Note: the restriction that exceptions are not
1348 propagated by :meth:`__cmp__` has been removed since Python 1.5.)
1351 .. method:: object.__rcmp__(self, other)
1353 .. versionchanged:: 2.1
1354 No longer supported.
1357 .. method:: object.__hash__(self)
1363 Called for the key object for dictionary operations, and by the built-in
1364 function :func:`hash`. Should return an integer usable as a hash value
1365 for dictionary operations. The only required property is that objects which
1366 compare equal have the same hash value; it is advised to somehow mix together
1367 (e.g., using exclusive or) the hash values for the components of the object that
1368 also play a part in comparison of objects.
1370 If a class does not define a :meth:`__cmp__` or :meth:`__eq__` method it
1371 should not define a :meth:`__hash__` operation either; if it defines
1372 :meth:`__cmp__` or :meth:`__eq__` but not :meth:`__hash__`, its instances
1373 will not be usable as dictionary keys. If a class defines mutable objects
1374 and implements a :meth:`__cmp__` or :meth:`__eq__` method, it should not
1375 implement :meth:`__hash__`, since the dictionary implementation requires that
1376 a key's hash value is immutable (if the object's hash value changes, it will
1377 be in the wrong hash bucket).
1379 User-defined classes have :meth:`__cmp__` and :meth:`__hash__` methods
1380 by default; with them, all objects compare unequal and ``x.__hash__()``
1383 .. versionchanged:: 2.5
1384 :meth:`__hash__` may now also return a long integer object; the 32-bit
1385 integer is then derived from the hash of that object.
1388 .. method:: object.__nonzero__(self)
1390 .. index:: single: __len__() (mapping object method)
1392 Called to implement truth value testing, and the built-in operation ``bool()``;
1393 should return ``False`` or ``True``, or their integer equivalents ``0`` or
1394 ``1``. When this method is not defined, :meth:`__len__` is called, if it is
1395 defined (see below). If a class defines neither :meth:`__len__` nor
1396 :meth:`__nonzero__`, all its instances are considered true.
1399 .. method:: object.__unicode__(self)
1401 .. index:: builtin: unicode
1403 Called to implement :func:`unicode` builtin; should return a Unicode object.
1404 When this method is not defined, string conversion is attempted, and the result
1405 of string conversion is converted to Unicode using the system default encoding.
1408 .. _attribute-access:
1410 Customizing attribute access
1411 ----------------------------
1413 The following methods can be defined to customize the meaning of attribute
1414 access (use of, assignment to, or deletion of ``x.name``) for class instances.
1417 .. method:: object.__getattr__(self, name)
1419 Called when an attribute lookup has not found the attribute in the usual places
1420 (i.e. it is not an instance attribute nor is it found in the class tree for
1421 ``self``). ``name`` is the attribute name. This method should return the
1422 (computed) attribute value or raise an :exc:`AttributeError` exception.
1424 .. index:: single: __setattr__() (object method)
1426 Note that if the attribute is found through the normal mechanism,
1427 :meth:`__getattr__` is not called. (This is an intentional asymmetry between
1428 :meth:`__getattr__` and :meth:`__setattr__`.) This is done both for efficiency
1429 reasons and because otherwise :meth:`__setattr__` would have no way to access
1430 other attributes of the instance. Note that at least for instance variables,
1431 you can fake total control by not inserting any values in the instance attribute
1432 dictionary (but instead inserting them in another object). See the
1433 :meth:`__getattribute__` method below for a way to actually get total control in
1437 .. method:: object.__setattr__(self, name, value)
1439 Called when an attribute assignment is attempted. This is called instead of the
1440 normal mechanism (i.e. store the value in the instance dictionary). *name* is
1441 the attribute name, *value* is the value to be assigned to it.
1443 .. index:: single: __dict__ (instance attribute)
1445 If :meth:`__setattr__` wants to assign to an instance attribute, it should not
1446 simply execute ``self.name = value`` --- this would cause a recursive call to
1447 itself. Instead, it should insert the value in the dictionary of instance
1448 attributes, e.g., ``self.__dict__[name] = value``. For new-style classes,
1449 rather than accessing the instance dictionary, it should call the base class
1450 method with the same name, for example, ``object.__setattr__(self, name,
1454 .. method:: object.__delattr__(self, name)
1456 Like :meth:`__setattr__` but for attribute deletion instead of assignment. This
1457 should only be implemented if ``del obj.name`` is meaningful for the object.
1460 .. _new-style-attribute-access:
1462 More attribute access for new-style classes
1463 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1465 The following methods only apply to new-style classes.
1468 .. method:: object.__getattribute__(self, name)
1470 Called unconditionally to implement attribute accesses for instances of the
1471 class. If the class also defines :meth:`__getattr__`, the latter will not be
1472 called unless :meth:`__getattribute__` either calls it explicitly or raises an
1473 :exc:`AttributeError`. This method should return the (computed) attribute value
1474 or raise an :exc:`AttributeError` exception. In order to avoid infinite
1475 recursion in this method, its implementation should always call the base class
1476 method with the same name to access any attributes it needs, for example,
1477 ``object.__getattribute__(self, name)``.
1482 Implementing Descriptors
1483 ^^^^^^^^^^^^^^^^^^^^^^^^
1485 The following methods only apply when an instance of the class containing the
1486 method (a so-called *descriptor* class) appears in the class dictionary of
1487 another new-style class, known as the *owner* class. In the examples below, "the
1488 attribute" refers to the attribute whose name is the key of the property in the
1489 owner class' ``__dict__``. Descriptors can only be implemented as new-style
1493 .. method:: object.__get__(self, instance, owner)
1495 Called to get the attribute of the owner class (class attribute access) or of an
1496 instance of that class (instance attribute access). *owner* is always the owner
1497 class, while *instance* is the instance that the attribute was accessed through,
1498 or ``None`` when the attribute is accessed through the *owner*. This method
1499 should return the (computed) attribute value or raise an :exc:`AttributeError`
1503 .. method:: object.__set__(self, instance, value)
1505 Called to set the attribute on an instance *instance* of the owner class to a
1509 .. method:: object.__delete__(self, instance)
1511 Called to delete the attribute on an instance *instance* of the owner class.
1514 .. _descriptor-invocation:
1516 Invoking Descriptors
1517 ^^^^^^^^^^^^^^^^^^^^
1519 In general, a descriptor is an object attribute with "binding behavior", one
1520 whose attribute access has been overridden by methods in the descriptor
1521 protocol: :meth:`__get__`, :meth:`__set__`, and :meth:`__delete__`. If any of
1522 those methods are defined for an object, it is said to be a descriptor.
1524 The default behavior for attribute access is to get, set, or delete the
1525 attribute from an object's dictionary. For instance, ``a.x`` has a lookup chain
1526 starting with ``a.__dict__['x']``, then ``type(a).__dict__['x']``, and
1527 continuing through the base classes of ``type(a)`` excluding metaclasses.
1529 However, if the looked-up value is an object defining one of the descriptor
1530 methods, then Python may override the default behavior and invoke the descriptor
1531 method instead. Where this occurs in the precedence chain depends on which
1532 descriptor methods were defined and how they were called. Note that descriptors
1533 are only invoked for new style objects or classes (ones that subclass
1534 :class:`object()` or :class:`type()`).
1536 The starting point for descriptor invocation is a binding, ``a.x``. How the
1537 arguments are assembled depends on ``a``:
1540 The simplest and least common call is when user code directly invokes a
1541 descriptor method: ``x.__get__(a)``.
1544 If binding to a new-style object instance, ``a.x`` is transformed into the call:
1545 ``type(a).__dict__['x'].__get__(a, type(a))``.
1548 If binding to a new-style class, ``A.x`` is transformed into the call:
1549 ``A.__dict__['x'].__get__(None, A)``.
1552 If ``a`` is an instance of :class:`super`, then the binding ``super(B,
1553 obj).m()`` searches ``obj.__class__.__mro__`` for the base class ``A``
1554 immediately preceding ``B`` and then invokes the descriptor with the call:
1555 ``A.__dict__['m'].__get__(obj, A)``.
1557 For instance bindings, the precedence of descriptor invocation depends on the
1558 which descriptor methods are defined. Normally, data descriptors define both
1559 :meth:`__get__` and :meth:`__set__`, while non-data descriptors have just the
1560 :meth:`__get__` method. Data descriptors always override a redefinition in an
1561 instance dictionary. In contrast, non-data descriptors can be overridden by
1564 Python methods (including :func:`staticmethod` and :func:`classmethod`) are
1565 implemented as non-data descriptors. Accordingly, instances can redefine and
1566 override methods. This allows individual instances to acquire behaviors that
1567 differ from other instances of the same class.
1569 The :func:`property` function is implemented as a data descriptor. Accordingly,
1570 instances cannot override the behavior of a property.
1578 By default, instances of both old and new-style classes have a dictionary for
1579 attribute storage. This wastes space for objects having very few instance
1580 variables. The space consumption can become acute when creating large numbers
1583 The default can be overridden by defining *__slots__* in a new-style class
1584 definition. The *__slots__* declaration takes a sequence of instance variables
1585 and reserves just enough space in each instance to hold a value for each
1586 variable. Space is saved because *__dict__* is not created for each instance.
1591 This class variable can be assigned a string, iterable, or sequence of strings
1592 with variable names used by instances. If defined in a new-style class,
1593 *__slots__* reserves space for the declared variables and prevents the automatic
1594 creation of *__dict__* and *__weakref__* for each instance.
1596 .. versionadded:: 2.2
1598 Notes on using *__slots__*
1600 * Without a *__dict__* variable, instances cannot be assigned new variables not
1601 listed in the *__slots__* definition. Attempts to assign to an unlisted
1602 variable name raises :exc:`AttributeError`. If dynamic assignment of new
1603 variables is desired, then add ``'__dict__'`` to the sequence of strings in the
1604 *__slots__* declaration.
1606 .. versionchanged:: 2.3
1607 Previously, adding ``'__dict__'`` to the *__slots__* declaration would not
1608 enable the assignment of new attributes not specifically listed in the sequence
1609 of instance variable names.
1611 * Without a *__weakref__* variable for each instance, classes defining
1612 *__slots__* do not support weak references to its instances. If weak reference
1613 support is needed, then add ``'__weakref__'`` to the sequence of strings in the
1614 *__slots__* declaration.
1616 .. versionchanged:: 2.3
1617 Previously, adding ``'__weakref__'`` to the *__slots__* declaration would not
1618 enable support for weak references.
1620 * *__slots__* are implemented at the class level by creating descriptors
1621 (:ref:`descriptors`) for each variable name. As a result, class attributes
1622 cannot be used to set default values for instance variables defined by
1623 *__slots__*; otherwise, the class attribute would overwrite the descriptor
1626 * If a class defines a slot also defined in a base class, the instance variable
1627 defined by the base class slot is inaccessible (except by retrieving its
1628 descriptor directly from the base class). This renders the meaning of the
1629 program undefined. In the future, a check may be added to prevent this.
1631 * The action of a *__slots__* declaration is limited to the class where it is
1632 defined. As a result, subclasses will have a *__dict__* unless they also define
1635 * *__slots__* do not work for classes derived from "variable-length" built-in
1636 types such as :class:`long`, :class:`str` and :class:`tuple`.
1638 * Any non-string iterable may be assigned to *__slots__*. Mappings may also be
1639 used; however, in the future, special meaning may be assigned to the values
1640 corresponding to each key.
1642 * *__class__* assignment works only if both classes have the same *__slots__*.
1644 .. versionchanged:: 2.6
1645 Previously, *__class__* assignment raised an error if either new or old class
1651 Customizing class creation
1652 --------------------------
1654 By default, new-style classes are constructed using :func:`type`. A class
1655 definition is read into a separate namespace and the value of class name is
1656 bound to the result of ``type(name, bases, dict)``.
1658 When the class definition is read, if *__metaclass__* is defined then the
1659 callable assigned to it will be called instead of :func:`type`. This allows
1660 classes or functions to be written which monitor or alter the class creation
1663 * Modifying the class dictionary prior to the class being created.
1665 * Returning an instance of another class -- essentially performing the role of a
1668 These steps will have to be performed in the metaclass's :meth:`__new__` method
1669 -- :meth:`type.__new__` can then be called from this method to create a class
1670 with different properties. This example adds a new element to the class
1671 dictionary before creating the class::
1673 class metacls(type):
1674 def __new__(mcs, name, bases, dict):
1675 dict['foo'] = 'metacls was here'
1676 return type.__new__(mcs, name, bases, dict)
1678 You can of course also override other class methods (or add new methods); for
1679 example defining a custom :meth:`__call__` method in the metaclass allows custom
1680 behavior when the class is called, e.g. not always creating a new instance.
1683 .. data:: __metaclass__
1685 This variable can be any callable accepting arguments for ``name``, ``bases``,
1686 and ``dict``. Upon class creation, the callable is used instead of the built-in
1689 .. versionadded:: 2.2
1691 The appropriate metaclass is determined by the following precedence rules:
1693 * If ``dict['__metaclass__']`` exists, it is used.
1695 * Otherwise, if there is at least one base class, its metaclass is used (this
1696 looks for a *__class__* attribute first and if not found, uses its type).
1698 * Otherwise, if a global variable named __metaclass__ exists, it is used.
1700 * Otherwise, the old-style, classic metaclass (types.ClassType) is used.
1702 The potential uses for metaclasses are boundless. Some ideas that have been
1703 explored including logging, interface checking, automatic delegation, automatic
1704 property creation, proxies, frameworks, and automatic resource
1705 locking/synchronization.
1710 Emulating callable objects
1711 --------------------------
1714 .. method:: object.__call__(self[, args...])
1716 .. index:: pair: call; instance
1718 Called when the instance is "called" as a function; if this method is defined,
1719 ``x(arg1, arg2, ...)`` is a shorthand for ``x.__call__(arg1, arg2, ...)``.
1724 Emulating container types
1725 -------------------------
1727 The following methods can be defined to implement container objects. Containers
1728 usually are sequences (such as lists or tuples) or mappings (like dictionaries),
1729 but can represent other containers as well. The first set of methods is used
1730 either to emulate a sequence or to emulate a mapping; the difference is that for
1731 a sequence, the allowable keys should be the integers *k* for which ``0 <= k <
1732 N`` where *N* is the length of the sequence, or slice objects, which define a
1733 range of items. (For backwards compatibility, the method :meth:`__getslice__`
1734 (see below) can also be defined to handle simple, but not extended slices.) It
1735 is also recommended that mappings provide the methods :meth:`keys`,
1736 :meth:`values`, :meth:`items`, :meth:`has_key`, :meth:`get`, :meth:`clear`,
1737 :meth:`setdefault`, :meth:`iterkeys`, :meth:`itervalues`, :meth:`iteritems`,
1738 :meth:`pop`, :meth:`popitem`, :meth:`copy`, and :meth:`update` behaving similar
1739 to those for Python's standard dictionary objects. The :mod:`UserDict` module
1740 provides a :class:`DictMixin` class to help create those methods from a base set
1741 of :meth:`__getitem__`, :meth:`__setitem__`, :meth:`__delitem__`, and
1742 :meth:`keys`. Mutable sequences should provide methods :meth:`append`,
1743 :meth:`count`, :meth:`index`, :meth:`extend`, :meth:`insert`, :meth:`pop`,
1744 :meth:`remove`, :meth:`reverse` and :meth:`sort`, like Python standard list
1745 objects. Finally, sequence types should implement addition (meaning
1746 concatenation) and multiplication (meaning repetition) by defining the methods
1747 :meth:`__add__`, :meth:`__radd__`, :meth:`__iadd__`, :meth:`__mul__`,
1748 :meth:`__rmul__` and :meth:`__imul__` described below; they should not define
1749 :meth:`__coerce__` or other numerical operators. It is recommended that both
1750 mappings and sequences implement the :meth:`__contains__` method to allow
1751 efficient use of the ``in`` operator; for mappings, ``in`` should be equivalent
1752 of :meth:`has_key`; for sequences, it should search through the values. It is
1753 further recommended that both mappings and sequences implement the
1754 :meth:`__iter__` method to allow efficient iteration through the container; for
1755 mappings, :meth:`__iter__` should be the same as :meth:`iterkeys`; for
1756 sequences, it should iterate through the values.
1759 .. method:: object.__len__(self)
1763 single: __nonzero__() (object method)
1765 Called to implement the built-in function :func:`len`. Should return the length
1766 of the object, an integer ``>=`` 0. Also, an object that doesn't define a
1767 :meth:`__nonzero__` method and whose :meth:`__len__` method returns zero is
1768 considered to be false in a Boolean context.
1771 .. method:: object.__getitem__(self, key)
1773 .. index:: object: slice
1775 Called to implement evaluation of ``self[key]``. For sequence types, the
1776 accepted keys should be integers and slice objects. Note that the special
1777 interpretation of negative indexes (if the class wishes to emulate a sequence
1778 type) is up to the :meth:`__getitem__` method. If *key* is of an inappropriate
1779 type, :exc:`TypeError` may be raised; if of a value outside the set of indexes
1780 for the sequence (after any special interpretation of negative values),
1781 :exc:`IndexError` should be raised. For mapping types, if *key* is missing (not
1782 in the container), :exc:`KeyError` should be raised.
1786 :keyword:`for` loops expect that an :exc:`IndexError` will be raised for illegal
1787 indexes to allow proper detection of the end of the sequence.
1790 .. method:: object.__setitem__(self, key, value)
1792 Called to implement assignment to ``self[key]``. Same note as for
1793 :meth:`__getitem__`. This should only be implemented for mappings if the
1794 objects support changes to the values for keys, or if new keys can be added, or
1795 for sequences if elements can be replaced. The same exceptions should be raised
1796 for improper *key* values as for the :meth:`__getitem__` method.
1799 .. method:: object.__delitem__(self, key)
1801 Called to implement deletion of ``self[key]``. Same note as for
1802 :meth:`__getitem__`. This should only be implemented for mappings if the
1803 objects support removal of keys, or for sequences if elements can be removed
1804 from the sequence. The same exceptions should be raised for improper *key*
1805 values as for the :meth:`__getitem__` method.
1808 .. method:: object.__iter__(self)
1810 This method is called when an iterator is required for a container. This method
1811 should return a new iterator object that can iterate over all the objects in the
1812 container. For mappings, it should iterate over the keys of the container, and
1813 should also be made available as the method :meth:`iterkeys`.
1815 Iterator objects also need to implement this method; they are required to return
1816 themselves. For more information on iterator objects, see :ref:`typeiter`.
1819 .. method:: object.__reversed__(self)
1821 Called (if present) by the :func:`reversed` builtin to implement
1822 reverse iteration. It should return a new iterator object that iterates
1823 over all the objects in the container in reverse order.
1825 If the :meth:`__reversed__` method is not provided, the
1826 :func:`reversed` builtin will fall back to using the sequence protocol
1827 (:meth:`__len__` and :meth:`__getitem__`). Objects should normally
1828 only provide :meth:`__reversed__` if they do not support the sequence
1829 protocol and an efficient implementation of reverse iteration is possible.
1831 .. versionadded:: 2.6
1834 The membership test operators (:keyword:`in` and :keyword:`not in`) are normally
1835 implemented as an iteration through a sequence. However, container objects can
1836 supply the following special method with a more efficient implementation, which
1837 also does not require the object be a sequence.
1840 .. method:: object.__contains__(self, item)
1842 Called to implement membership test operators. Should return true if *item* is
1843 in *self*, false otherwise. For mapping objects, this should consider the keys
1844 of the mapping rather than the values or the key-item pairs.
1847 .. _sequence-methods:
1849 Additional methods for emulation of sequence types
1850 --------------------------------------------------
1852 The following optional methods can be defined to further emulate sequence
1853 objects. Immutable sequences methods should at most only define
1854 :meth:`__getslice__`; mutable sequences might define all three methods.
1857 .. method:: object.__getslice__(self, i, j)
1860 Support slice objects as parameters to the :meth:`__getitem__` method.
1861 (However, built-in types in CPython currently still implement
1862 :meth:`__getslice__`. Therefore, you have to override it in derived
1863 classes when implementing slicing.)
1865 Called to implement evaluation of ``self[i:j]``. The returned object should be
1866 of the same type as *self*. Note that missing *i* or *j* in the slice
1867 expression are replaced by zero or ``sys.maxint``, respectively. If negative
1868 indexes are used in the slice, the length of the sequence is added to that
1869 index. If the instance does not implement the :meth:`__len__` method, an
1870 :exc:`AttributeError` is raised. No guarantee is made that indexes adjusted this
1871 way are not still negative. Indexes which are greater than the length of the
1872 sequence are not modified. If no :meth:`__getslice__` is found, a slice object
1873 is created instead, and passed to :meth:`__getitem__` instead.
1876 .. method:: object.__setslice__(self, i, j, sequence)
1878 Called to implement assignment to ``self[i:j]``. Same notes for *i* and *j* as
1879 for :meth:`__getslice__`.
1881 This method is deprecated. If no :meth:`__setslice__` is found, or for extended
1882 slicing of the form ``self[i:j:k]``, a slice object is created, and passed to
1883 :meth:`__setitem__`, instead of :meth:`__setslice__` being called.
1886 .. method:: object.__delslice__(self, i, j)
1888 Called to implement deletion of ``self[i:j]``. Same notes for *i* and *j* as for
1889 :meth:`__getslice__`. This method is deprecated. If no :meth:`__delslice__` is
1890 found, or for extended slicing of the form ``self[i:j:k]``, a slice object is
1891 created, and passed to :meth:`__delitem__`, instead of :meth:`__delslice__`
1894 Notice that these methods are only invoked when a single slice with a single
1895 colon is used, and the slice method is available. For slice operations
1896 involving extended slice notation, or in absence of the slice methods,
1897 :meth:`__getitem__`, :meth:`__setitem__` or :meth:`__delitem__` is called with a
1898 slice object as argument.
1900 The following example demonstrate how to make your program or module compatible
1901 with earlier versions of Python (assuming that methods :meth:`__getitem__`,
1902 :meth:`__setitem__` and :meth:`__delitem__` support slice objects as
1907 def __getitem__(self, index):
1909 def __setitem__(self, index, value):
1911 def __delitem__(self, index):
1914 if sys.version_info < (2, 0):
1915 # They won't be defined if version is at least 2.0 final
1917 def __getslice__(self, i, j):
1918 return self[max(0, i):max(0, j):]
1919 def __setslice__(self, i, j, seq):
1920 self[max(0, i):max(0, j):] = seq
1921 def __delslice__(self, i, j):
1922 del self[max(0, i):max(0, j):]
1925 Note the calls to :func:`max`; these are necessary because of the handling of
1926 negative indices before the :meth:`__\*slice__` methods are called. When
1927 negative indexes are used, the :meth:`__\*item__` methods receive them as
1928 provided, but the :meth:`__\*slice__` methods get a "cooked" form of the index
1929 values. For each negative index value, the length of the sequence is added to
1930 the index before calling the method (which may still result in a negative
1931 index); this is the customary handling of negative indexes by the built-in
1932 sequence types, and the :meth:`__\*item__` methods are expected to do this as
1933 well. However, since they should already be doing that, negative indexes cannot
1934 be passed in; they must be constrained to the bounds of the sequence before
1935 being passed to the :meth:`__\*item__` methods. Calling ``max(0, i)``
1936 conveniently returns the proper value.
1941 Emulating numeric types
1942 -----------------------
1944 The following methods can be defined to emulate numeric objects. Methods
1945 corresponding to operations that are not supported by the particular kind of
1946 number implemented (e.g., bitwise operations for non-integral numbers) should be
1950 .. method:: object.__add__(self, other)
1951 object.__sub__(self, other)
1952 object.__mul__(self, other)
1953 object.__floordiv__(self, other)
1954 object.__mod__(self, other)
1955 object.__divmod__(self, other)
1956 object.__pow__(self, other[, modulo])
1957 object.__lshift__(self, other)
1958 object.__rshift__(self, other)
1959 object.__and__(self, other)
1960 object.__xor__(self, other)
1961 object.__or__(self, other)
1968 These methods are called to implement the binary arithmetic operations (``+``,
1969 ``-``, ``*``, ``//``, ``%``, :func:`divmod`, :func:`pow`, ``**``, ``<<``,
1970 ``>>``, ``&``, ``^``, ``|``). For instance, to evaluate the expression
1971 *x*``+``*y*, where *x* is an instance of a class that has an :meth:`__add__`
1972 method, ``x.__add__(y)`` is called. The :meth:`__divmod__` method should be the
1973 equivalent to using :meth:`__floordiv__` and :meth:`__mod__`; it should not be
1974 related to :meth:`__truediv__` (described below). Note that :meth:`__pow__`
1975 should be defined to accept an optional third argument if the ternary version of
1976 the built-in :func:`pow` function is to be supported.
1978 If one of those methods does not support the operation with the supplied
1979 arguments, it should return ``NotImplemented``.
1982 .. method:: object.__div__(self, other)
1983 object.__truediv__(self, other)
1985 The division operator (``/``) is implemented by these methods. The
1986 :meth:`__truediv__` method is used when ``__future__.division`` is in effect,
1987 otherwise :meth:`__div__` is used. If only one of these two methods is defined,
1988 the object will not support division in the alternate context; :exc:`TypeError`
1989 will be raised instead.
1992 .. method:: object.__radd__(self, other)
1993 object.__rsub__(self, other)
1994 object.__rmul__(self, other)
1995 object.__rdiv__(self, other)
1996 object.__rtruediv__(self, other)
1997 object.__rfloordiv__(self, other)
1998 object.__rmod__(self, other)
1999 object.__rdivmod__(self, other)
2000 object.__rpow__(self, other)
2001 object.__rlshift__(self, other)
2002 object.__rrshift__(self, other)
2003 object.__rand__(self, other)
2004 object.__rxor__(self, other)
2005 object.__ror__(self, other)
2011 These methods are called to implement the binary arithmetic operations (``+``,
2012 ``-``, ``*``, ``/``, ``%``, :func:`divmod`, :func:`pow`, ``**``, ``<<``, ``>>``,
2013 ``&``, ``^``, ``|``) with reflected (swapped) operands. These functions are
2014 only called if the left operand does not support the corresponding operation and
2015 the operands are of different types. [#]_ For instance, to evaluate the
2016 expression *x*``-``*y*, where *y* is an instance of a class that has an
2017 :meth:`__rsub__` method, ``y.__rsub__(x)`` is called if ``x.__sub__(y)`` returns
2020 .. index:: builtin: pow
2022 Note that ternary :func:`pow` will not try calling :meth:`__rpow__` (the
2023 coercion rules would become too complicated).
2027 If the right operand's type is a subclass of the left operand's type and that
2028 subclass provides the reflected method for the operation, this method will be
2029 called before the left operand's non-reflected method. This behavior allows
2030 subclasses to override their ancestors' operations.
2033 .. method:: object.__iadd__(self, other)
2034 object.__isub__(self, other)
2035 object.__imul__(self, other)
2036 object.__idiv__(self, other)
2037 object.__itruediv__(self, other)
2038 object.__ifloordiv__(self, other)
2039 object.__imod__(self, other)
2040 object.__ipow__(self, other[, modulo])
2041 object.__ilshift__(self, other)
2042 object.__irshift__(self, other)
2043 object.__iand__(self, other)
2044 object.__ixor__(self, other)
2045 object.__ior__(self, other)
2047 These methods are called to implement the augmented arithmetic operations
2048 (``+=``, ``-=``, ``*=``, ``/=``, ``//=``, ``%=``, ``**=``, ``<<=``, ``>>=``,
2049 ``&=``, ``^=``, ``|=``). These methods should attempt to do the operation
2050 in-place (modifying *self*) and return the result (which could be, but does
2051 not have to be, *self*). If a specific method is not defined, the augmented
2052 operation falls back to the normal methods. For instance, to evaluate the
2053 expression *x*``+=``*y*, where *x* is an instance of a class that has an
2054 :meth:`__iadd__` method, ``x.__iadd__(y)`` is called. If *x* is an instance
2055 of a class that does not define a :meth:`__iadd__` method, ``x.__add__(y)``
2056 and ``y.__radd__(x)`` are considered, as with the evaluation of *x*``+``*y*.
2059 .. method:: object.__neg__(self)
2060 object.__pos__(self)
2061 object.__abs__(self)
2062 object.__invert__(self)
2064 .. index:: builtin: abs
2066 Called to implement the unary arithmetic operations (``-``, ``+``, :func:`abs`
2070 .. method:: object.__complex__(self)
2071 object.__int__(self)
2072 object.__long__(self)
2073 object.__float__(self)
2081 Called to implement the built-in functions :func:`complex`, :func:`int`,
2082 :func:`long`, and :func:`float`. Should return a value of the appropriate type.
2085 .. method:: object.__oct__(self)
2086 object.__hex__(self)
2092 Called to implement the built-in functions :func:`oct` and :func:`hex`. Should
2093 return a string value.
2096 .. method:: object.__index__(self)
2098 Called to implement :func:`operator.index`. Also called whenever Python needs
2099 an integer object (such as in slicing). Must return an integer (int or long).
2101 .. versionadded:: 2.5
2104 .. method:: object.__coerce__(self, other)
2106 Called to implement "mixed-mode" numeric arithmetic. Should either return a
2107 2-tuple containing *self* and *other* converted to a common numeric type, or
2108 ``None`` if conversion is impossible. When the common type would be the type of
2109 ``other``, it is sufficient to return ``None``, since the interpreter will also
2110 ask the other object to attempt a coercion (but sometimes, if the implementation
2111 of the other type cannot be changed, it is useful to do the conversion to the
2112 other type here). A return value of ``NotImplemented`` is equivalent to
2121 This section used to document the rules for coercion. As the language has
2122 evolved, the coercion rules have become hard to document precisely; documenting
2123 what one version of one particular implementation does is undesirable. Instead,
2124 here are some informal guidelines regarding coercion. In Python 3.0, coercion
2125 will not be supported.
2129 If the left operand of a % operator is a string or Unicode object, no coercion
2130 takes place and the string formatting operation is invoked instead.
2134 It is no longer recommended to define a coercion operation. Mixed-mode
2135 operations on types that don't define coercion pass the original arguments to
2140 New-style classes (those derived from :class:`object`) never invoke the
2141 :meth:`__coerce__` method in response to a binary operator; the only time
2142 :meth:`__coerce__` is invoked is when the built-in function :func:`coerce` is
2147 For most intents and purposes, an operator that returns ``NotImplemented`` is
2148 treated the same as one that is not implemented at all.
2152 Below, :meth:`__op__` and :meth:`__rop__` are used to signify the generic method
2153 names corresponding to an operator; :meth:`__iop__` is used for the
2154 corresponding in-place operator. For example, for the operator '``+``',
2155 :meth:`__add__` and :meth:`__radd__` are used for the left and right variant of
2156 the binary operator, and :meth:`__iadd__` for the in-place variant.
2160 For objects *x* and *y*, first ``x.__op__(y)`` is tried. If this is not
2161 implemented or returns ``NotImplemented``, ``y.__rop__(x)`` is tried. If this
2162 is also not implemented or returns ``NotImplemented``, a :exc:`TypeError`
2163 exception is raised. But see the following exception:
2167 Exception to the previous item: if the left operand is an instance of a built-in
2168 type or a new-style class, and the right operand is an instance of a proper
2169 subclass of that type or class and overrides the base's :meth:`__rop__` method,
2170 the right operand's :meth:`__rop__` method is tried *before* the left operand's
2171 :meth:`__op__` method.
2173 This is done so that a subclass can completely override binary operators.
2174 Otherwise, the left operand's :meth:`__op__` method would always accept the
2175 right operand: when an instance of a given class is expected, an instance of a
2176 subclass of that class is always acceptable.
2180 When either operand type defines a coercion, this coercion is called before that
2181 type's :meth:`__op__` or :meth:`__rop__` method is called, but no sooner. If
2182 the coercion returns an object of a different type for the operand whose
2183 coercion is invoked, part of the process is redone using the new object.
2187 When an in-place operator (like '``+=``') is used, if the left operand
2188 implements :meth:`__iop__`, it is invoked without any coercion. When the
2189 operation falls back to :meth:`__op__` and/or :meth:`__rop__`, the normal
2190 coercion rules apply.
2194 In *x*``+``*y*, if *x* is a sequence that implements sequence concatenation,
2195 sequence concatenation is invoked.
2199 In *x*``*``*y*, if one operator is a sequence that implements sequence
2200 repetition, and the other is an integer (:class:`int` or :class:`long`),
2201 sequence repetition is invoked.
2205 Rich comparisons (implemented by methods :meth:`__eq__` and so on) never use
2206 coercion. Three-way comparison (implemented by :meth:`__cmp__`) does use
2207 coercion under the same conditions as other binary operations use it.
2211 In the current implementation, the built-in numeric types :class:`int`,
2212 :class:`long` and :class:`float` do not use coercion; the type :class:`complex`
2213 however does use it. The difference can become apparent when subclassing these
2214 types. Over time, the type :class:`complex` may be fixed to avoid coercion.
2215 All these types implement a :meth:`__coerce__` method, for use by the built-in
2216 :func:`coerce` function.
2219 .. _context-managers:
2221 With Statement Context Managers
2222 -------------------------------
2224 .. versionadded:: 2.5
2226 A :dfn:`context manager` is an object that defines the runtime context to be
2227 established when executing a :keyword:`with` statement. The context manager
2228 handles the entry into, and the exit from, the desired runtime context for the
2229 execution of the block of code. Context managers are normally invoked using the
2230 :keyword:`with` statement (described in section :ref:`with`), but can also be
2231 used by directly invoking their methods.
2235 single: context manager
2237 Typical uses of context managers include saving and restoring various kinds of
2238 global state, locking and unlocking resources, closing opened files, etc.
2240 For more information on context managers, see :ref:`typecontextmanager`.
2243 .. method:: object.__enter__(self)
2245 Enter the runtime context related to this object. The :keyword:`with` statement
2246 will bind this method's return value to the target(s) specified in the
2247 :keyword:`as` clause of the statement, if any.
2250 .. method:: object.__exit__(self, exc_type, exc_value, traceback)
2252 Exit the runtime context related to this object. The parameters describe the
2253 exception that caused the context to be exited. If the context was exited
2254 without an exception, all three arguments will be :const:`None`.
2256 If an exception is supplied, and the method wishes to suppress the exception
2257 (i.e., prevent it from being propagated), it should return a true value.
2258 Otherwise, the exception will be processed normally upon exit from this method.
2260 Note that :meth:`__exit__` methods should not reraise the passed-in exception;
2261 this is the caller's responsibility.
2266 :pep:`0343` - The "with" statement
2267 The specification, background, and examples for the Python :keyword:`with`
2270 .. rubric:: Footnotes
2272 .. [#] Since Python 2.2, a gradual merging of types and classes has been started that
2273 makes this and a few other assertions made in this manual not 100% accurate and
2274 complete: for example, it *is* now possible in some cases to change an object's
2275 type, under certain controlled conditions. Until this manual undergoes
2276 extensive revision, it must now be taken as authoritative only regarding
2277 "classic classes", that are still the default, for compatibility purposes, in
2278 Python 2.2 and 2.3. For more information, see
2279 http://www.python.org/doc/newstyle/.
2281 .. [#] This, and other statements, are only roughly true for instances of new-style
2284 .. [#] A descriptor can define any combination of :meth:`__get__`,
2285 :meth:`__set__` and :meth:`__delete__`. If it does not define :meth:`__get__`,
2286 then accessing the attribute even on an instance will return the descriptor
2287 object itself. If the descriptor defines :meth:`__set__` and/or
2288 :meth:`__delete__`, it is a data descriptor; if it defines neither, it is a
2289 non-data descriptor.
2291 .. [#] For operands of the same type, it is assumed that if the non-reflected method
2292 (such as :meth:`__add__`) fails the operation is not supported, which is why the
2293 reflected method is not called.