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: CPython currently 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. Other implementations act differently and CPython
67 Note that the use of the implementation's tracing or debugging facilities may
68 keep objects alive that would normally be collectable. Also note that catching
69 an exception with a ':keyword:`try`...\ :keyword:`except`' statement may keep
72 Some objects contain references to "external" resources such as open files or
73 windows. It is understood that these resources are freed when the object is
74 garbage-collected, but since garbage collection is not guaranteed to happen,
75 such objects also provide an explicit way to release the external resource,
76 usually a :meth:`close` method. Programs are strongly recommended to explicitly
77 close such objects. The ':keyword:`try`...\ :keyword:`finally`' statement
78 provides a convenient way to do this.
80 .. index:: single: container
82 Some objects contain references to other objects; these are called *containers*.
83 Examples of containers are tuples, lists and dictionaries. The references are
84 part of a container's value. In most cases, when we talk about the value of a
85 container, we imply the values, not the identities of the contained objects;
86 however, when we talk about the mutability of a container, only the identities
87 of the immediately contained objects are implied. So, if an immutable container
88 (like a tuple) contains a reference to a mutable object, its value changes if
89 that mutable object is changed.
91 Types affect almost all aspects of object behavior. Even the importance of
92 object identity is affected in some sense: for immutable types, operations that
93 compute new values may actually return a reference to any existing object with
94 the same type and value, while for mutable objects this is not allowed. E.g.,
95 after ``a = 1; b = 1``, ``a`` and ``b`` may or may not refer to the same object
96 with the value one, depending on the implementation, but after ``c = []; d =
97 []``, ``c`` and ``d`` are guaranteed to refer to two different, unique, newly
98 created empty lists. (Note that ``c = d = []`` assigns the same object to both
104 The standard type hierarchy
105 ===========================
110 pair: type; hierarchy
111 pair: extension; module
114 Below is a list of the types that are built into Python. Extension modules
115 (written in C, Java, or other languages, depending on the implementation) can
116 define additional types. Future versions of Python may add types to the type
117 hierarchy (e.g., rational numbers, efficiently stored arrays of integers, etc.).
121 pair: special; attribute
122 triple: generic; special; attribute
124 Some of the type descriptions below contain a paragraph listing 'special
125 attributes.' These are attributes that provide access to the implementation and
126 are not intended for general use. Their definition may change in the future.
129 .. index:: object: None
131 This type has a single value. There is a single object with this value. This
132 object is accessed through the built-in name ``None``. It is used to signify the
133 absence of a value in many situations, e.g., it is returned from functions that
134 don't explicitly return anything. Its truth value is false.
137 .. index:: object: NotImplemented
139 This type has a single value. There is a single object with this value. This
140 object is accessed through the built-in name ``NotImplemented``. Numeric methods
141 and rich comparison methods may return this value if they do not implement the
142 operation for the operands provided. (The interpreter will then try the
143 reflected operation, or some other fallback, depending on the operator.) Its
147 .. index:: object: Ellipsis
149 This type has a single value. There is a single object with this value. This
150 object is accessed through the built-in name ``Ellipsis``. It is used to
151 indicate the presence of the ``...`` syntax in a slice. Its truth value is
154 :class:`numbers.Number`
155 .. index:: object: numeric
157 These are created by numeric literals and returned as results by arithmetic
158 operators and arithmetic built-in functions. Numeric objects are immutable;
159 once created their value never changes. Python numbers are of course strongly
160 related to mathematical numbers, but subject to the limitations of numerical
161 representation in computers.
163 Python distinguishes between integers, floating point numbers, and complex
166 :class:`numbers.Integral`
167 .. index:: object: integer
169 These represent elements from the mathematical set of integers (positive and
172 There are three types of integers:
176 object: plain integer
177 single: OverflowError (built-in exception)
179 These represent numbers in the range -2147483648 through 2147483647.
180 (The range may be larger on machines with a larger natural word size,
181 but not smaller.) When the result of an operation would fall outside
182 this range, the result is normally returned as a long integer (in some
183 cases, the exception :exc:`OverflowError` is raised instead). For the
184 purpose of shift and mask operations, integers are assumed to have a
185 binary, 2's complement notation using 32 or more bits, and hiding no
186 bits from the user (i.e., all 4294967296 different bit patterns
187 correspond to different values).
190 .. index:: object: long integer
192 These represent numbers in an unlimited range, subject to available
193 (virtual) memory only. For the purpose of shift and mask operations, a
194 binary representation is assumed, and negative numbers are represented
195 in a variant of 2's complement which gives the illusion of an infinite
196 string of sign bits extending to the left.
204 These represent the truth values False and True. The two objects
205 representing the values False and True are the only Boolean objects.
206 The Boolean type is a subtype of plain integers, and Boolean values
207 behave like the values 0 and 1, respectively, in almost all contexts,
208 the exception being that when converted to a string, the strings
209 ``"False"`` or ``"True"`` are returned, respectively.
211 .. index:: pair: integer; representation
213 The rules for integer representation are intended to give the most
214 meaningful interpretation of shift and mask operations involving negative
215 integers and the least surprises when switching between the plain and long
216 integer domains. Any operation, if it yields a result in the plain
217 integer domain, will yield the same result in the long integer domain or
218 when using mixed operands. The switch between domains is transparent to
221 :class:`numbers.Real` (:class:`float`)
223 object: floating point
224 pair: floating point; number
228 These represent machine-level double precision floating point numbers. You are
229 at the mercy of the underlying machine architecture (and C or Java
230 implementation) for the accepted range and handling of overflow. Python does not
231 support single-precision floating point numbers; the savings in processor and
232 memory usage that are usually the reason for using these is dwarfed by the
233 overhead of using objects in Python, so there is no reason to complicate the
234 language with two kinds of floating point numbers.
236 :class:`numbers.Complex`
239 pair: complex; number
241 These represent complex numbers as a pair of machine-level double precision
242 floating point numbers. The same caveats apply as for floating point numbers.
243 The real and imaginary parts of a complex number ``z`` can be retrieved through
244 the read-only attributes ``z.real`` and ``z.imag``.
250 single: index operation
251 single: item selection
254 These represent finite ordered sets indexed by non-negative numbers. The
255 built-in function :func:`len` returns the number of items of a sequence. When
256 the length of a sequence is *n*, the index set contains the numbers 0, 1,
257 ..., *n*-1. Item *i* of sequence *a* is selected by ``a[i]``.
259 .. index:: single: slicing
261 Sequences also support slicing: ``a[i:j]`` selects all items with index *k* such
262 that *i* ``<=`` *k* ``<`` *j*. When used as an expression, a slice is a
263 sequence of the same type. This implies that the index set is renumbered so
266 .. index:: single: extended slicing
268 Some sequences also support "extended slicing" with a third "step" parameter:
269 ``a[i:j:k]`` selects all items of *a* with index *x* where ``x = i + n*k``, *n*
270 ``>=`` ``0`` and *i* ``<=`` *x* ``<`` *j*.
272 Sequences are distinguished according to their mutability:
276 object: immutable sequence
279 An object of an immutable sequence type cannot change once it is created. (If
280 the object contains references to other objects, these other objects may be
281 mutable and may be changed; however, the collection of objects directly
282 referenced by an immutable object cannot change.)
284 The following types are immutable sequences:
295 The items of a string are characters. There is no separate character type; a
296 character is represented by a string of one item. Characters represent (at
297 least) 8-bit bytes. The built-in functions :func:`chr` and :func:`ord` convert
298 between characters and nonnegative integers representing the byte values. Bytes
299 with the values 0-127 usually represent the corresponding ASCII values, but the
300 interpretation of values is up to the program. The string data type is also
301 used to represent arrays of bytes, e.g., to hold data read from a file.
306 single: character set
307 pair: string; comparison
311 (On systems whose native character set is not ASCII, strings may use EBCDIC in
312 their internal representation, provided the functions :func:`chr` and
313 :func:`ord` implement a mapping between ASCII and EBCDIC, and string comparison
314 preserves the ASCII order. Or perhaps someone can propose a better rule?)
326 The items of a Unicode object are Unicode code units. A Unicode code unit is
327 represented by a Unicode object of one item and can hold either a 16-bit or
328 32-bit value representing a Unicode ordinal (the maximum value for the ordinal
329 is given in ``sys.maxunicode``, and depends on how Python is configured at
330 compile time). Surrogate pairs may be present in the Unicode object, and will
331 be reported as two separate items. The built-in functions :func:`unichr` and
332 :func:`ord` convert between code units and nonnegative integers representing the
333 Unicode ordinals as defined in the Unicode Standard 3.0. Conversion from and to
334 other encodings are possible through the Unicode method :meth:`encode` and the
335 built-in function :func:`unicode`.
340 pair: singleton; tuple
343 The items of a tuple are arbitrary Python objects. Tuples of two or more items
344 are formed by comma-separated lists of expressions. A tuple of one item (a
345 'singleton') can be formed by affixing a comma to an expression (an expression
346 by itself does not create a tuple, since parentheses must be usable for grouping
347 of expressions). An empty tuple can be formed by an empty pair of parentheses.
351 object: mutable sequence
353 pair: assignment; statement
359 Mutable sequences can be changed after they are created. The subscription and
360 slicing notations can be used as the target of assignment and :keyword:`del`
363 There are currently two intrinsic mutable sequence types:
366 .. index:: object: list
368 The items of a list are arbitrary Python objects. Lists are formed by placing a
369 comma-separated list of expressions in square brackets. (Note that there are no
370 special cases needed to form lists of length 0 or 1.)
375 A bytearray object is a mutable array. They are created by the built-in
376 :func:`bytearray` constructor. Aside from being mutable (and hence
377 unhashable), byte arrays otherwise provide the same interface and
378 functionality as immutable bytes objects.
380 .. index:: module: array
382 The extension module :mod:`array` provides an additional example of a mutable
390 These represent unordered, finite sets of unique, immutable objects. As such,
391 they cannot be indexed by any subscript. However, they can be iterated over, and
392 the built-in function :func:`len` returns the number of items in a set. Common
393 uses for sets are fast membership testing, removing duplicates from a sequence,
394 and computing mathematical operations such as intersection, union, difference,
395 and symmetric difference.
397 For set elements, the same immutability rules apply as for dictionary keys. Note
398 that numeric types obey the normal rules for numeric comparison: if two numbers
399 compare equal (e.g., ``1`` and ``1.0``), only one of them can be contained in a
402 There are currently two intrinsic set types:
405 .. index:: object: set
407 These represent a mutable set. They are created by the built-in :func:`set`
408 constructor and can be modified afterwards by several methods, such as
412 .. index:: object: frozenset
414 These represent an immutable set. They are created by the built-in
415 :func:`frozenset` constructor. As a frozenset is immutable and
416 :term:`hashable`, it can be used again as an element of another set, or as
425 These represent finite sets of objects indexed by arbitrary index sets. The
426 subscript notation ``a[k]`` selects the item indexed by ``k`` from the mapping
427 ``a``; this can be used in expressions and as the target of assignments or
428 :keyword:`del` statements. The built-in function :func:`len` returns the number
429 of items in a mapping.
431 There is currently a single intrinsic mapping type:
434 .. index:: object: dictionary
436 These represent finite sets of objects indexed by nearly arbitrary values. The
437 only types of values not acceptable as keys are values containing lists or
438 dictionaries or other mutable types that are compared by value rather than by
439 object identity, the reason being that the efficient implementation of
440 dictionaries requires a key's hash value to remain constant. Numeric types used
441 for keys obey the normal rules for numeric comparison: if two numbers compare
442 equal (e.g., ``1`` and ``1.0``) then they can be used interchangeably to index
443 the same dictionary entry.
445 Dictionaries are mutable; they can be created by the ``{...}`` notation (see
446 section :ref:`dict`).
453 The extension modules :mod:`dbm`, :mod:`gdbm`, and :mod:`bsddb` provide
454 additional examples of mapping types.
461 pair: function; argument
463 These are the types to which the function call operation (see section
464 :ref:`calls`) can be applied:
466 User-defined functions
468 pair: user-defined; function
470 object: user-defined function
472 A user-defined function object is created by a function definition (see
473 section :ref:`function`). It should be called with an argument list
474 containing the same number of items as the function's formal parameter
479 +-----------------------+-------------------------------+-----------+
480 | Attribute | Meaning | |
481 +=======================+===============================+===========+
482 | :attr:`func_doc` | The function's documentation | Writable |
483 | | string, or ``None`` if | |
485 +-----------------------+-------------------------------+-----------+
486 | :attr:`__doc__` | Another way of spelling | Writable |
487 | | :attr:`func_doc` | |
488 +-----------------------+-------------------------------+-----------+
489 | :attr:`func_name` | The function's name | Writable |
490 +-----------------------+-------------------------------+-----------+
491 | :attr:`__name__` | Another way of spelling | Writable |
492 | | :attr:`func_name` | |
493 +-----------------------+-------------------------------+-----------+
494 | :attr:`__module__` | The name of the module the | Writable |
495 | | function was defined in, or | |
496 | | ``None`` if unavailable. | |
497 +-----------------------+-------------------------------+-----------+
498 | :attr:`func_defaults` | A tuple containing default | Writable |
499 | | argument values for those | |
500 | | arguments that have defaults, | |
501 | | or ``None`` if no arguments | |
502 | | have a default value | |
503 +-----------------------+-------------------------------+-----------+
504 | :attr:`func_code` | The code object representing | Writable |
505 | | the compiled function body. | |
506 +-----------------------+-------------------------------+-----------+
507 | :attr:`func_globals` | A reference to the dictionary | Read-only |
508 | | that holds the function's | |
509 | | global variables --- the | |
510 | | global namespace of the | |
511 | | module in which the function | |
513 +-----------------------+-------------------------------+-----------+
514 | :attr:`func_dict` | The namespace supporting | Writable |
515 | | arbitrary function | |
517 +-----------------------+-------------------------------+-----------+
518 | :attr:`func_closure` | ``None`` or a tuple of cells | Read-only |
519 | | that contain bindings for the | |
520 | | function's free variables. | |
521 +-----------------------+-------------------------------+-----------+
523 Most of the attributes labelled "Writable" check the type of the assigned value.
525 .. versionchanged:: 2.4
526 ``func_name`` is now writable.
528 Function objects also support getting and setting arbitrary attributes, which
529 can be used, for example, to attach metadata to functions. Regular attribute
530 dot-notation is used to get and set such attributes. *Note that the current
531 implementation only supports function attributes on user-defined functions.
532 Function attributes on built-in functions may be supported in the future.*
534 Additional information about a function's definition can be retrieved from its
535 code object; see the description of internal types below.
538 single: func_doc (function attribute)
539 single: __doc__ (function attribute)
540 single: __name__ (function attribute)
541 single: __module__ (function attribute)
542 single: __dict__ (function attribute)
543 single: func_defaults (function attribute)
544 single: func_closure (function attribute)
545 single: func_code (function attribute)
546 single: func_globals (function attribute)
547 single: func_dict (function attribute)
548 pair: global; namespace
553 object: user-defined method
554 pair: user-defined; method
556 A user-defined method object combines a class, a class instance (or ``None``)
557 and any callable object (normally a user-defined function).
559 Special read-only attributes: :attr:`im_self` is the class instance object,
560 :attr:`im_func` is the function object; :attr:`im_class` is the class of
561 :attr:`im_self` for bound methods or the class that asked for the method for
562 unbound methods; :attr:`__doc__` is the method's documentation (same as
563 ``im_func.__doc__``); :attr:`__name__` is the method name (same as
564 ``im_func.__name__``); :attr:`__module__` is the name of the module the method
565 was defined in, or ``None`` if unavailable.
567 .. versionchanged:: 2.2
568 :attr:`im_self` used to refer to the class that defined the method.
570 .. versionchanged:: 2.6
571 For 3.0 forward-compatibility, :attr:`im_func` is also available as
572 :attr:`__func__`, and :attr:`im_self` as :attr:`__self__`.
575 single: __doc__ (method attribute)
576 single: __name__ (method attribute)
577 single: __module__ (method attribute)
578 single: im_func (method attribute)
579 single: im_self (method attribute)
581 Methods also support accessing (but not setting) the arbitrary function
582 attributes on the underlying function object.
584 User-defined method objects may be created when getting an attribute of a class
585 (perhaps via an instance of that class), if that attribute is a user-defined
586 function object, an unbound user-defined method object, or a class method
587 object. When the attribute is a user-defined method object, a new method object
588 is only created if the class from which it is being retrieved is the same as, or
589 a derived class of, the class stored in the original method object; otherwise,
590 the original method object is used as it is.
593 single: im_class (method attribute)
594 single: im_func (method attribute)
595 single: im_self (method attribute)
597 When a user-defined method object is created by retrieving a user-defined
598 function object from a class, its :attr:`im_self` attribute is ``None``
599 and the method object is said to be unbound. When one is created by
600 retrieving a user-defined function object from a class via one of its
601 instances, its :attr:`im_self` attribute is the instance, and the method
602 object is said to be bound. In either case, the new method's
603 :attr:`im_class` attribute is the class from which the retrieval takes
604 place, and its :attr:`im_func` attribute is the original function object.
606 .. index:: single: im_func (method attribute)
608 When a user-defined method object is created by retrieving another method object
609 from a class or instance, the behaviour is the same as for a function object,
610 except that the :attr:`im_func` attribute of the new instance is not the
611 original method object but its :attr:`im_func` attribute.
614 single: im_class (method attribute)
615 single: im_func (method attribute)
616 single: im_self (method attribute)
618 When a user-defined method object is created by retrieving a class method object
619 from a class or instance, its :attr:`im_self` attribute is the class itself (the
620 same as the :attr:`im_class` attribute), and its :attr:`im_func` attribute is
621 the function object underlying the class method.
623 When an unbound user-defined method object is called, the underlying function
624 (:attr:`im_func`) is called, with the restriction that the first argument must
625 be an instance of the proper class (:attr:`im_class`) or of a derived class
628 When a bound user-defined method object is called, the underlying function
629 (:attr:`im_func`) is called, inserting the class instance (:attr:`im_self`) in
630 front of the argument list. For instance, when :class:`C` is a class which
631 contains a definition for a function :meth:`f`, and ``x`` is an instance of
632 :class:`C`, calling ``x.f(1)`` is equivalent to calling ``C.f(x, 1)``.
634 When a user-defined method object is derived from a class method object, the
635 "class instance" stored in :attr:`im_self` will actually be the class itself, so
636 that calling either ``x.f(1)`` or ``C.f(1)`` is equivalent to calling ``f(C,1)``
637 where ``f`` is the underlying function.
639 Note that the transformation from function object to (unbound or bound) method
640 object happens each time the attribute is retrieved from the class or instance.
641 In some cases, a fruitful optimization is to assign the attribute to a local
642 variable and call that local variable. Also notice that this transformation only
643 happens for user-defined functions; other callable objects (and all non-callable
644 objects) are retrieved without transformation. It is also important to note
645 that user-defined functions which are attributes of a class instance are not
646 converted to bound methods; this *only* happens when the function is an
647 attribute of the class.
651 single: generator; function
652 single: generator; iterator
654 A function or method which uses the :keyword:`yield` statement (see section
655 :ref:`yield`) is called a :dfn:`generator
656 function`. Such a function, when called, always returns an iterator object
657 which can be used to execute the body of the function: calling the iterator's
658 :meth:`next` method will cause the function to execute until it provides a value
659 using the :keyword:`yield` statement. When the function executes a
660 :keyword:`return` statement or falls off the end, a :exc:`StopIteration`
661 exception is raised and the iterator will have reached the end of the set of
662 values to be returned.
666 object: built-in function
670 A built-in function object is a wrapper around a C function. Examples of
671 built-in functions are :func:`len` and :func:`math.sin` (:mod:`math` is a
672 standard built-in module). The number and type of the arguments are
673 determined by the C function. Special read-only attributes:
674 :attr:`__doc__` is the function's documentation string, or ``None`` if
675 unavailable; :attr:`__name__` is the function's name; :attr:`__self__` is
676 set to ``None`` (but see the next item); :attr:`__module__` is the name of
677 the module the function was defined in or ``None`` if unavailable.
681 object: built-in method
683 pair: built-in; method
685 This is really a different disguise of a built-in function, this time containing
686 an object passed to the C function as an implicit extra argument. An example of
687 a built-in method is ``alist.append()``, assuming *alist* is a list object. In
688 this case, the special read-only attribute :attr:`__self__` is set to the object
692 Class types, or "new-style classes," are callable. These objects normally act
693 as factories for new instances of themselves, but variations are possible for
694 class types that override :meth:`__new__`. The arguments of the call are passed
695 to :meth:`__new__` and, in the typical case, to :meth:`__init__` to initialize
700 single: __init__() (object method)
702 object: class instance
704 pair: class object; call
706 Class objects are described below. When a class object is called, a new class
707 instance (also described below) is created and returned. This implies a call to
708 the class's :meth:`__init__` method if it has one. Any arguments are passed on
709 to the :meth:`__init__` method. If there is no :meth:`__init__` method, the
710 class must be called without arguments.
713 Class instances are described below. Class instances are callable only when the
714 class has a :meth:`__call__` method; ``x(arguments)`` is a shorthand for
715 ``x.__call__(arguments)``.
722 Modules are imported by the :keyword:`import` statement (see section
723 :ref:`import`). A module object has a
724 namespace implemented by a dictionary object (this is the dictionary referenced
725 by the func_globals attribute of functions defined in the module). Attribute
726 references are translated to lookups in this dictionary, e.g., ``m.x`` is
727 equivalent to ``m.__dict__["x"]``. A module object does not contain the code
728 object used to initialize the module (since it isn't needed once the
729 initialization is done).
731 Attribute assignment updates the module's namespace dictionary, e.g., ``m.x =
732 1`` is equivalent to ``m.__dict__["x"] = 1``.
734 .. index:: single: __dict__ (module attribute)
736 Special read-only attribute: :attr:`__dict__` is the module's namespace as a
740 single: __name__ (module attribute)
741 single: __doc__ (module attribute)
742 single: __file__ (module attribute)
743 pair: module; namespace
745 Predefined (writable) attributes: :attr:`__name__` is the module's name;
746 :attr:`__doc__` is the module's documentation string, or ``None`` if
747 unavailable; :attr:`__file__` is the pathname of the file from which the module
748 was loaded, if it was loaded from a file. The :attr:`__file__` attribute is not
749 present for C modules that are statically linked into the interpreter; for
750 extension modules loaded dynamically from a shared library, it is the pathname
751 of the shared library file.
754 Both class types (new-style classes) and class objects (old-style/classic
755 classes) are typically created by class definitions (see section
756 :ref:`class`). A class has a namespace implemented by a dictionary object.
757 Class attribute references are translated to lookups in this dictionary, e.g.,
758 ``C.x`` is translated to ``C.__dict__["x"]`` (although for new-style classes
759 in particular there are a number of hooks which allow for other means of
760 locating attributes). When the attribute name is not found there, the
761 attribute search continues in the base classes. For old-style classes, the
762 search is depth-first, left-to-right in the order of occurrence in the base
763 class list. New-style classes use the more complex C3 method resolution
764 order which behaves correctly even in the presence of 'diamond'
765 inheritance structures where there are multiple inheritance paths
766 leading back to a common ancestor. Additional details on the C3 MRO used by
767 new-style classes can be found in the documentation accompanying the
768 2.3 release at http://www.python.org/download/releases/2.3/mro/.
770 .. XXX: Could we add that MRO doc as an appendix to the language ref?
774 object: class instance
776 pair: class object; call
779 pair: class; attribute
781 When a class attribute reference (for class :class:`C`, say) would yield a
782 user-defined function object or an unbound user-defined method object whose
783 associated class is either :class:`C` or one of its base classes, it is
784 transformed into an unbound user-defined method object whose :attr:`im_class`
785 attribute is :class:`C`. When it would yield a class method object, it is
786 transformed into a bound user-defined method object whose :attr:`im_class`
787 and :attr:`im_self` attributes are both :class:`C`. When it would yield a
788 static method object, it is transformed into the object wrapped by the static
789 method object. See section :ref:`descriptors` for another way in which
790 attributes retrieved from a class may differ from those actually contained in
791 its :attr:`__dict__` (note that only new-style classes support descriptors).
793 .. index:: triple: class; attribute; assignment
795 Class attribute assignments update the class's dictionary, never the dictionary
798 .. index:: pair: class object; call
800 A class object can be called (see above) to yield a class instance (see below).
803 single: __name__ (class attribute)
804 single: __module__ (class attribute)
805 single: __dict__ (class attribute)
806 single: __bases__ (class attribute)
807 single: __doc__ (class attribute)
809 Special attributes: :attr:`__name__` is the class name; :attr:`__module__` is
810 the module name in which the class was defined; :attr:`__dict__` is the
811 dictionary containing the class's namespace; :attr:`__bases__` is a tuple
812 (possibly empty or a singleton) containing the base classes, in the order of
813 their occurrence in the base class list; :attr:`__doc__` is the class's
814 documentation string, or None if undefined.
818 object: class instance
820 pair: class; instance
821 pair: class instance; attribute
823 A class instance is created by calling a class object (see above). A class
824 instance has a namespace implemented as a dictionary which is the first place in
825 which attribute references are searched. When an attribute is not found there,
826 and the instance's class has an attribute by that name, the search continues
827 with the class attributes. If a class attribute is found that is a user-defined
828 function object or an unbound user-defined method object whose associated class
829 is the class (call it :class:`C`) of the instance for which the attribute
830 reference was initiated or one of its bases, it is transformed into a bound
831 user-defined method object whose :attr:`im_class` attribute is :class:`C` and
832 whose :attr:`im_self` attribute is the instance. Static method and class method
833 objects are also transformed, as if they had been retrieved from class
834 :class:`C`; see above under "Classes". See section :ref:`descriptors` for
835 another way in which attributes of a class retrieved via its instances may
836 differ from the objects actually stored in the class's :attr:`__dict__`. If no
837 class attribute is found, and the object's class has a :meth:`__getattr__`
838 method, that is called to satisfy the lookup.
840 .. index:: triple: class instance; attribute; assignment
842 Attribute assignments and deletions update the instance's dictionary, never a
843 class's dictionary. If the class has a :meth:`__setattr__` or
844 :meth:`__delattr__` method, this is called instead of updating the instance
852 Class instances can pretend to be numbers, sequences, or mappings if they have
853 methods with certain special names. See section :ref:`specialnames`.
856 single: __dict__ (instance attribute)
857 single: __class__ (instance attribute)
859 Special attributes: :attr:`__dict__` is the attribute dictionary;
860 :attr:`__class__` is the instance's class.
866 single: popen() (in module os)
867 single: makefile() (socket method)
872 single: stdin (in module sys)
873 single: stdout (in module sys)
874 single: stderr (in module sys)
876 A file object represents an open file. File objects are created by the
877 :func:`open` built-in function, and also by :func:`os.popen`,
878 :func:`os.fdopen`, and the :meth:`makefile` method of socket objects (and
879 perhaps by other functions or methods provided by extension modules). The
880 objects ``sys.stdin``, ``sys.stdout`` and ``sys.stderr`` are initialized to
881 file objects corresponding to the interpreter's standard input, output and
882 error streams. See :ref:`bltin-file-objects` for complete documentation of
887 single: internal type
888 single: types, internal
890 A few types used internally by the interpreter are exposed to the user. Their
891 definitions may change with future versions of the interpreter, but they are
892 mentioned here for completeness.
899 Code objects represent *byte-compiled* executable Python code, or :term:`bytecode`.
900 The difference between a code object and a function object is that the function
901 object contains an explicit reference to the function's globals (the module in
902 which it was defined), while a code object contains no context; also the default
903 argument values are stored in the function object, not in the code object
904 (because they represent values calculated at run-time). Unlike function
905 objects, code objects are immutable and contain no references (directly or
906 indirectly) to mutable objects.
908 Special read-only attributes: :attr:`co_name` gives the function name;
909 :attr:`co_argcount` is the number of positional arguments (including arguments
910 with default values); :attr:`co_nlocals` is the number of local variables used
911 by the function (including arguments); :attr:`co_varnames` is a tuple containing
912 the names of the local variables (starting with the argument names);
913 :attr:`co_cellvars` is a tuple containing the names of local variables that are
914 referenced by nested functions; :attr:`co_freevars` is a tuple containing the
915 names of free variables; :attr:`co_code` is a string representing the sequence
916 of bytecode instructions; :attr:`co_consts` is a tuple containing the literals
917 used by the bytecode; :attr:`co_names` is a tuple containing the names used by
918 the bytecode; :attr:`co_filename` is the filename from which the code was
919 compiled; :attr:`co_firstlineno` is the first line number of the function;
920 :attr:`co_lnotab` is a string encoding the mapping from bytecode offsets to
921 line numbers (for details see the source code of the interpreter);
922 :attr:`co_stacksize` is the required stack size (including local variables);
923 :attr:`co_flags` is an integer encoding a number of flags for the interpreter.
926 single: co_argcount (code object attribute)
927 single: co_code (code object attribute)
928 single: co_consts (code object attribute)
929 single: co_filename (code object attribute)
930 single: co_firstlineno (code object attribute)
931 single: co_flags (code object attribute)
932 single: co_lnotab (code object attribute)
933 single: co_name (code object attribute)
934 single: co_names (code object attribute)
935 single: co_nlocals (code object attribute)
936 single: co_stacksize (code object attribute)
937 single: co_varnames (code object attribute)
938 single: co_cellvars (code object attribute)
939 single: co_freevars (code object attribute)
941 .. index:: object: generator
943 The following flag bits are defined for :attr:`co_flags`: bit ``0x04`` is set if
944 the function uses the ``*arguments`` syntax to accept an arbitrary number of
945 positional arguments; bit ``0x08`` is set if the function uses the
946 ``**keywords`` syntax to accept arbitrary keyword arguments; bit ``0x20`` is set
947 if the function is a generator.
949 Future feature declarations (``from __future__ import division``) also use bits
950 in :attr:`co_flags` to indicate whether a code object was compiled with a
951 particular feature enabled: bit ``0x2000`` is set if the function was compiled
952 with future division enabled; bits ``0x10`` and ``0x1000`` were used in earlier
955 Other bits in :attr:`co_flags` are reserved for internal use.
957 .. index:: single: documentation string
959 If a code object represents a function, the first item in :attr:`co_consts` is
960 the documentation string of the function, or ``None`` if undefined.
963 .. index:: object: frame
965 Frame objects represent execution frames. They may occur in traceback objects
969 single: f_back (frame attribute)
970 single: f_code (frame attribute)
971 single: f_globals (frame attribute)
972 single: f_locals (frame attribute)
973 single: f_lasti (frame attribute)
974 single: f_builtins (frame attribute)
975 single: f_restricted (frame attribute)
977 Special read-only attributes: :attr:`f_back` is to the previous stack frame
978 (towards the caller), or ``None`` if this is the bottom stack frame;
979 :attr:`f_code` is the code object being executed in this frame; :attr:`f_locals`
980 is the dictionary used to look up local variables; :attr:`f_globals` is used for
981 global variables; :attr:`f_builtins` is used for built-in (intrinsic) names;
982 :attr:`f_restricted` is a flag indicating whether the function is executing in
983 restricted execution mode; :attr:`f_lasti` gives the precise instruction (this
984 is an index into the bytecode string of the code object).
987 single: f_trace (frame attribute)
988 single: f_exc_type (frame attribute)
989 single: f_exc_value (frame attribute)
990 single: f_exc_traceback (frame attribute)
991 single: f_lineno (frame attribute)
993 Special writable attributes: :attr:`f_trace`, if not ``None``, is a function
994 called at the start of each source code line (this is used by the debugger);
995 :attr:`f_exc_type`, :attr:`f_exc_value`, :attr:`f_exc_traceback` represent the
996 last exception raised in the parent frame provided another exception was ever
997 raised in the current frame (in all other cases they are None); :attr:`f_lineno`
998 is the current line number of the frame --- writing to this from within a trace
999 function jumps to the given line (only for the bottom-most frame). A debugger
1000 can implement a Jump command (aka Set Next Statement) by writing to f_lineno.
1006 pair: exception; handler
1007 pair: execution; stack
1008 single: exc_info (in module sys)
1009 single: exc_traceback (in module sys)
1010 single: last_traceback (in module sys)
1011 single: sys.exc_info
1012 single: sys.exc_traceback
1013 single: sys.last_traceback
1015 Traceback objects represent a stack trace of an exception. A traceback object
1016 is created when an exception occurs. When the search for an exception handler
1017 unwinds the execution stack, at each unwound level a traceback object is
1018 inserted in front of the current traceback. When an exception handler is
1019 entered, the stack trace is made available to the program. (See section
1020 :ref:`try`.) It is accessible as ``sys.exc_traceback``,
1021 and also as the third item of the tuple returned by ``sys.exc_info()``. The
1022 latter is the preferred interface, since it works correctly when the program is
1023 using multiple threads. When the program contains no suitable handler, the stack
1024 trace is written (nicely formatted) to the standard error stream; if the
1025 interpreter is interactive, it is also made available to the user as
1026 ``sys.last_traceback``.
1029 single: tb_next (traceback attribute)
1030 single: tb_frame (traceback attribute)
1031 single: tb_lineno (traceback attribute)
1032 single: tb_lasti (traceback attribute)
1035 Special read-only attributes: :attr:`tb_next` is the next level in the stack
1036 trace (towards the frame where the exception occurred), or ``None`` if there is
1037 no next level; :attr:`tb_frame` points to the execution frame of the current
1038 level; :attr:`tb_lineno` gives the line number where the exception occurred;
1039 :attr:`tb_lasti` indicates the precise instruction. The line number and last
1040 instruction in the traceback may differ from the line number of its frame object
1041 if the exception occurred in a :keyword:`try` statement with no matching except
1042 clause or with a finally clause.
1045 .. index:: builtin: slice
1047 Slice objects are used to represent slices when *extended slice syntax* is used.
1048 This is a slice using two colons, or multiple slices or ellipses separated by
1049 commas, e.g., ``a[i:j:step]``, ``a[i:j, k:l]``, or ``a[..., i:j]``. They are
1050 also created by the built-in :func:`slice` function.
1053 single: start (slice object attribute)
1054 single: stop (slice object attribute)
1055 single: step (slice object attribute)
1057 Special read-only attributes: :attr:`start` is the lower bound; :attr:`stop` is
1058 the upper bound; :attr:`step` is the step value; each is ``None`` if omitted.
1059 These attributes can have any type.
1061 Slice objects support one method:
1064 .. method:: slice.indices(self, length)
1066 This method takes a single integer argument *length* and computes information
1067 about the extended slice that the slice object would describe if applied to a
1068 sequence of *length* items. It returns a tuple of three integers; respectively
1069 these are the *start* and *stop* indices and the *step* or stride length of the
1070 slice. Missing or out-of-bounds indices are handled in a manner consistent with
1073 .. versionadded:: 2.3
1075 Static method objects
1076 Static method objects provide a way of defeating the transformation of function
1077 objects to method objects described above. A static method object is a wrapper
1078 around any other object, usually a user-defined method object. When a static
1079 method object is retrieved from a class or a class instance, the object actually
1080 returned is the wrapped object, which is not subject to any further
1081 transformation. Static method objects are not themselves callable, although the
1082 objects they wrap usually are. Static method objects are created by the built-in
1083 :func:`staticmethod` constructor.
1085 Class method objects
1086 A class method object, like a static method object, is a wrapper around another
1087 object that alters the way in which that object is retrieved from classes and
1088 class instances. The behaviour of class method objects upon such retrieval is
1089 described above, under "User-defined methods". Class method objects are created
1090 by the built-in :func:`classmethod` constructor.
1095 New-style and classic classes
1096 =============================
1098 Classes and instances come in two flavors: old-style (or classic) and new-style.
1100 Up to Python 2.1, old-style classes were the only flavour available to the user.
1101 The concept of (old-style) class is unrelated to the concept of type: if *x* is
1102 an instance of an old-style class, then ``x.__class__`` designates the class of
1103 *x*, but ``type(x)`` is always ``<type 'instance'>``. This reflects the fact
1104 that all old-style instances, independently of their class, are implemented with
1105 a single built-in type, called ``instance``.
1107 New-style classes were introduced in Python 2.2 to unify classes and types. A
1108 new-style class is neither more nor less than a user-defined type. If *x* is an
1109 instance of a new-style class, then ``type(x)`` is typically the same as
1110 ``x.__class__`` (although this is not guaranteed - a new-style class instance is
1111 permitted to override the value returned for ``x.__class__``).
1113 The major motivation for introducing new-style classes is to provide a unified
1114 object model with a full meta-model. It also has a number of practical
1115 benefits, like the ability to subclass most built-in types, or the introduction
1116 of "descriptors", which enable computed properties.
1118 For compatibility reasons, classes are still old-style by default. New-style
1119 classes are created by specifying another new-style class (i.e. a type) as a
1120 parent class, or the "top-level type" :class:`object` if no other parent is
1121 needed. The behaviour of new-style classes differs from that of old-style
1122 classes in a number of important details in addition to what :func:`type`
1123 returns. Some of these changes are fundamental to the new object model, like
1124 the way special methods are invoked. Others are "fixes" that could not be
1125 implemented before for compatibility concerns, like the method resolution order
1126 in case of multiple inheritance.
1128 While this manual aims to provide comprehensive coverage of Python's class
1129 mechanics, it may still be lacking in some areas when it comes to its coverage
1130 of new-style classes. Please see http://www.python.org/doc/newstyle/ for
1131 sources of additional information.
1134 single: class; new-style
1135 single: class; classic
1136 single: class; old-style
1138 Old-style classes are removed in Python 3.0, leaving only the semantics of
1144 Special method names
1145 ====================
1148 pair: operator; overloading
1149 single: __getitem__() (mapping object method)
1151 A class can implement certain operations that are invoked by special syntax
1152 (such as arithmetic operations or subscripting and slicing) by defining methods
1153 with special names. This is Python's approach to :dfn:`operator overloading`,
1154 allowing classes to define their own behavior with respect to language
1155 operators. For instance, if a class defines a method named :meth:`__getitem__`,
1156 and ``x`` is an instance of this class, then ``x[i]`` is roughly equivalent
1157 to ``x.__getitem__(i)`` for old-style classes and ``type(x).__getitem__(x, i)``
1158 for new-style classes. Except where mentioned, attempts to execute an
1159 operation raise an exception when no appropriate method is defined (typically
1160 :exc:`AttributeError` or :exc:`TypeError`).
1162 When implementing a class that emulates any built-in type, it is important that
1163 the emulation only be implemented to the degree that it makes sense for the
1164 object being modelled. For example, some sequences may work well with retrieval
1165 of individual elements, but extracting a slice may not make sense. (One example
1166 of this is the :class:`NodeList` interface in the W3C's Document Object Model.)
1174 .. method:: object.__new__(cls[, ...])
1176 .. index:: pair: subclassing; immutable types
1178 Called to create a new instance of class *cls*. :meth:`__new__` is a static
1179 method (special-cased so you need not declare it as such) that takes the class
1180 of which an instance was requested as its first argument. The remaining
1181 arguments are those passed to the object constructor expression (the call to the
1182 class). The return value of :meth:`__new__` should be the new object instance
1183 (usually an instance of *cls*).
1185 Typical implementations create a new instance of the class by invoking the
1186 superclass's :meth:`__new__` method using ``super(currentclass,
1187 cls).__new__(cls[, ...])`` with appropriate arguments and then modifying the
1188 newly-created instance as necessary before returning it.
1190 If :meth:`__new__` returns an instance of *cls*, then the new instance's
1191 :meth:`__init__` method will be invoked like ``__init__(self[, ...])``, where
1192 *self* is the new instance and the remaining arguments are the same as were
1193 passed to :meth:`__new__`.
1195 If :meth:`__new__` does not return an instance of *cls*, then the new instance's
1196 :meth:`__init__` method will not be invoked.
1198 :meth:`__new__` is intended mainly to allow subclasses of immutable types (like
1199 int, str, or tuple) to customize instance creation. It is also commonly
1200 overridden in custom metaclasses in order to customize class creation.
1203 .. method:: object.__init__(self[, ...])
1205 .. index:: pair: class; constructor
1207 Called when the instance is created. The arguments are those passed to the
1208 class constructor expression. If a base class has an :meth:`__init__` method,
1209 the derived class's :meth:`__init__` method, if any, must explicitly call it to
1210 ensure proper initialization of the base class part of the instance; for
1211 example: ``BaseClass.__init__(self, [args...])``. As a special constraint on
1212 constructors, no value may be returned; doing so will cause a :exc:`TypeError`
1213 to be raised at runtime.
1216 .. method:: object.__del__(self)
1222 Called when the instance is about to be destroyed. This is also called a
1223 destructor. If a base class has a :meth:`__del__` method, the derived class's
1224 :meth:`__del__` method, if any, must explicitly call it to ensure proper
1225 deletion of the base class part of the instance. Note that it is possible
1226 (though not recommended!) for the :meth:`__del__` method to postpone destruction
1227 of the instance by creating a new reference to it. It may then be called at a
1228 later time when this new reference is deleted. It is not guaranteed that
1229 :meth:`__del__` methods are called for objects that still exist when the
1234 ``del x`` doesn't directly call ``x.__del__()`` --- the former decrements
1235 the reference count for ``x`` by one, and the latter is only called when
1236 ``x``'s reference count reaches zero. Some common situations that may
1237 prevent the reference count of an object from going to zero include:
1238 circular references between objects (e.g., a doubly-linked list or a tree
1239 data structure with parent and child pointers); a reference to the object
1240 on the stack frame of a function that caught an exception (the traceback
1241 stored in ``sys.exc_traceback`` keeps the stack frame alive); or a
1242 reference to the object on the stack frame that raised an unhandled
1243 exception in interactive mode (the traceback stored in
1244 ``sys.last_traceback`` keeps the stack frame alive). The first situation
1245 can only be remedied by explicitly breaking the cycles; the latter two
1246 situations can be resolved by storing ``None`` in ``sys.exc_traceback`` or
1247 ``sys.last_traceback``. Circular references which are garbage are
1248 detected when the option cycle detector is enabled (it's on by default),
1249 but can only be cleaned up if there are no Python-level :meth:`__del__`
1250 methods involved. Refer to the documentation for the :mod:`gc` module for
1251 more information about how :meth:`__del__` methods are handled by the
1252 cycle detector, particularly the description of the ``garbage`` value.
1256 Due to the precarious circumstances under which :meth:`__del__` methods are
1257 invoked, exceptions that occur during their execution are ignored, and a warning
1258 is printed to ``sys.stderr`` instead. Also, when :meth:`__del__` is invoked in
1259 response to a module being deleted (e.g., when execution of the program is
1260 done), other globals referenced by the :meth:`__del__` method may already have
1261 been deleted or in the process of being torn down (e.g. the import
1262 machinery shutting down). For this reason, :meth:`__del__` methods
1263 should do the absolute
1264 minimum needed to maintain external invariants. Starting with version 1.5,
1265 Python guarantees that globals whose name begins with a single underscore are
1266 deleted from their module before other globals are deleted; if no other
1267 references to such globals exist, this may help in assuring that imported
1268 modules are still available at the time when the :meth:`__del__` method is
1272 .. method:: object.__repr__(self)
1274 .. index:: builtin: repr
1276 Called by the :func:`repr` built-in function and by string conversions (reverse
1277 quotes) to compute the "official" string representation of an object. If at all
1278 possible, this should look like a valid Python expression that could be used to
1279 recreate an object with the same value (given an appropriate environment). If
1280 this is not possible, a string of the form ``<...some useful description...>``
1281 should be returned. The return value must be a string object. If a class
1282 defines :meth:`__repr__` but not :meth:`__str__`, then :meth:`__repr__` is also
1283 used when an "informal" string representation of instances of that class is
1287 pair: string; conversion
1288 pair: reverse; quotes
1289 pair: backward; quotes
1292 This is typically used for debugging, so it is important that the representation
1293 is information-rich and unambiguous.
1296 .. method:: object.__str__(self)
1302 Called by the :func:`str` built-in function and by the :keyword:`print`
1303 statement to compute the "informal" string representation of an object. This
1304 differs from :meth:`__repr__` in that it does not have to be a valid Python
1305 expression: a more convenient or concise representation may be used instead.
1306 The return value must be a string object.
1309 .. method:: object.__lt__(self, other)
1310 object.__le__(self, other)
1311 object.__eq__(self, other)
1312 object.__ne__(self, other)
1313 object.__gt__(self, other)
1314 object.__ge__(self, other)
1316 .. versionadded:: 2.1
1321 These are the so-called "rich comparison" methods, and are called for comparison
1322 operators in preference to :meth:`__cmp__` below. The correspondence between
1323 operator symbols and method names is as follows: ``x<y`` calls ``x.__lt__(y)``,
1324 ``x<=y`` calls ``x.__le__(y)``, ``x==y`` calls ``x.__eq__(y)``, ``x!=y`` and
1325 ``x<>y`` call ``x.__ne__(y)``, ``x>y`` calls ``x.__gt__(y)``, and ``x>=y`` calls
1328 A rich comparison method may return the singleton ``NotImplemented`` if it does
1329 not implement the operation for a given pair of arguments. By convention,
1330 ``False`` and ``True`` are returned for a successful comparison. However, these
1331 methods can return any value, so if the comparison operator is used in a Boolean
1332 context (e.g., in the condition of an ``if`` statement), Python will call
1333 :func:`bool` on the value to determine if the result is true or false.
1335 There are no implied relationships among the comparison operators. The truth
1336 of ``x==y`` does not imply that ``x!=y`` is false. Accordingly, when
1337 defining :meth:`__eq__`, one should also define :meth:`__ne__` so that the
1338 operators will behave as expected. See the paragraph on :meth:`__hash__` for
1339 some important notes on creating :term:`hashable` objects which support
1340 custom comparison operations and are usable as dictionary keys.
1342 There are no swapped-argument versions of these methods (to be used when the
1343 left argument does not support the operation but the right argument does);
1344 rather, :meth:`__lt__` and :meth:`__gt__` are each other's reflection,
1345 :meth:`__le__` and :meth:`__ge__` are each other's reflection, and
1346 :meth:`__eq__` and :meth:`__ne__` are their own reflection.
1348 Arguments to rich comparison methods are never coerced.
1350 To automatically generate ordering operations from a single root operation,
1351 see the `Total Ordering recipe in the ASPN cookbook
1352 <http://code.activestate.com/recipes/576529/>`_\.
1354 .. method:: object.__cmp__(self, other)
1360 Called by comparison operations if rich comparison (see above) is not
1361 defined. Should return a negative integer if ``self < other``, zero if
1362 ``self == other``, a positive integer if ``self > other``. If no
1363 :meth:`__cmp__`, :meth:`__eq__` or :meth:`__ne__` operation is defined, class
1364 instances are compared by object identity ("address"). See also the
1365 description of :meth:`__hash__` for some important notes on creating
1366 :term:`hashable` objects which support custom comparison operations and are
1367 usable as dictionary keys. (Note: the restriction that exceptions are not
1368 propagated by :meth:`__cmp__` has been removed since Python 1.5.)
1371 .. method:: object.__rcmp__(self, other)
1373 .. versionchanged:: 2.1
1374 No longer supported.
1377 .. method:: object.__hash__(self)
1383 Called by built-in function :func:`hash` and for operations on members of
1384 hashed collections including :class:`set`, :class:`frozenset`, and
1385 :class:`dict`. :meth:`__hash__` should return an integer. The only required
1386 property is that objects which compare equal have the same hash value; it is
1387 advised to somehow mix together (e.g. using exclusive or) the hash values for
1388 the components of the object that also play a part in comparison of objects.
1390 If a class does not define a :meth:`__cmp__` or :meth:`__eq__` method it
1391 should not define a :meth:`__hash__` operation either; if it defines
1392 :meth:`__cmp__` or :meth:`__eq__` but not :meth:`__hash__`, its instances
1393 will not be usable in hashed collections. If a class defines mutable objects
1394 and implements a :meth:`__cmp__` or :meth:`__eq__` method, it should not
1395 implement :meth:`__hash__`, since hashable collection implementations require
1396 that a object's hash value is immutable (if the object's hash value changes,
1397 it will be in the wrong hash bucket).
1399 User-defined classes have :meth:`__cmp__` and :meth:`__hash__` methods
1400 by default; with them, all objects compare unequal (except with themselves)
1401 and ``x.__hash__()`` returns ``id(x)``.
1403 Classes which inherit a :meth:`__hash__` method from a parent class but
1404 change the meaning of :meth:`__cmp__` or :meth:`__eq__` such that the hash
1405 value returned is no longer appropriate (e.g. by switching to a value-based
1406 concept of equality instead of the default identity based equality) can
1407 explicitly flag themselves as being unhashable by setting ``__hash__ = None``
1408 in the class definition. Doing so means that not only will instances of the
1409 class raise an appropriate :exc:`TypeError` when a program attempts to
1410 retrieve their hash value, but they will also be correctly identified as
1411 unhashable when checking ``isinstance(obj, collections.Hashable)`` (unlike
1412 classes which define their own :meth:`__hash__` to explicitly raise
1415 .. versionchanged:: 2.5
1416 :meth:`__hash__` may now also return a long integer object; the 32-bit
1417 integer is then derived from the hash of that object.
1419 .. versionchanged:: 2.6
1420 :attr:`__hash__` may now be set to :const:`None` to explicitly flag
1421 instances of a class as unhashable.
1424 .. method:: object.__nonzero__(self)
1426 .. index:: single: __len__() (mapping object method)
1428 Called to implement truth value testing and the built-in operation ``bool()``;
1429 should return ``False`` or ``True``, or their integer equivalents ``0`` or
1430 ``1``. When this method is not defined, :meth:`__len__` is called, if it is
1431 defined, and the object is considered true if its result is nonzero.
1432 If a class defines neither :meth:`__len__` nor :meth:`__nonzero__`, all its
1433 instances are considered true.
1436 .. method:: object.__unicode__(self)
1438 .. index:: builtin: unicode
1440 Called to implement :func:`unicode` builtin; should return a Unicode object.
1441 When this method is not defined, string conversion is attempted, and the result
1442 of string conversion is converted to Unicode using the system default encoding.
1445 .. _attribute-access:
1447 Customizing attribute access
1448 ----------------------------
1450 The following methods can be defined to customize the meaning of attribute
1451 access (use of, assignment to, or deletion of ``x.name``) for class instances.
1454 .. method:: object.__getattr__(self, name)
1456 Called when an attribute lookup has not found the attribute in the usual places
1457 (i.e. it is not an instance attribute nor is it found in the class tree for
1458 ``self``). ``name`` is the attribute name. This method should return the
1459 (computed) attribute value or raise an :exc:`AttributeError` exception.
1461 .. index:: single: __setattr__() (object method)
1463 Note that if the attribute is found through the normal mechanism,
1464 :meth:`__getattr__` is not called. (This is an intentional asymmetry between
1465 :meth:`__getattr__` and :meth:`__setattr__`.) This is done both for efficiency
1466 reasons and because otherwise :meth:`__getattr__` would have no way to access
1467 other attributes of the instance. Note that at least for instance variables,
1468 you can fake total control by not inserting any values in the instance attribute
1469 dictionary (but instead inserting them in another object). See the
1470 :meth:`__getattribute__` method below for a way to actually get total control in
1474 .. method:: object.__setattr__(self, name, value)
1476 Called when an attribute assignment is attempted. This is called instead of the
1477 normal mechanism (i.e. store the value in the instance dictionary). *name* is
1478 the attribute name, *value* is the value to be assigned to it.
1480 .. index:: single: __dict__ (instance attribute)
1482 If :meth:`__setattr__` wants to assign to an instance attribute, it should not
1483 simply execute ``self.name = value`` --- this would cause a recursive call to
1484 itself. Instead, it should insert the value in the dictionary of instance
1485 attributes, e.g., ``self.__dict__[name] = value``. For new-style classes,
1486 rather than accessing the instance dictionary, it should call the base class
1487 method with the same name, for example, ``object.__setattr__(self, name,
1491 .. method:: object.__delattr__(self, name)
1493 Like :meth:`__setattr__` but for attribute deletion instead of assignment. This
1494 should only be implemented if ``del obj.name`` is meaningful for the object.
1497 .. _new-style-attribute-access:
1499 More attribute access for new-style classes
1500 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1502 The following methods only apply to new-style classes.
1505 .. method:: object.__getattribute__(self, name)
1507 Called unconditionally to implement attribute accesses for instances of the
1508 class. If the class also defines :meth:`__getattr__`, the latter will not be
1509 called unless :meth:`__getattribute__` either calls it explicitly or raises an
1510 :exc:`AttributeError`. This method should return the (computed) attribute value
1511 or raise an :exc:`AttributeError` exception. In order to avoid infinite
1512 recursion in this method, its implementation should always call the base class
1513 method with the same name to access any attributes it needs, for example,
1514 ``object.__getattribute__(self, name)``.
1518 This method may still be bypassed when looking up special methods as the
1519 result of implicit invocation via language syntax or builtin functions.
1520 See :ref:`new-style-special-lookup`.
1525 Implementing Descriptors
1526 ^^^^^^^^^^^^^^^^^^^^^^^^
1528 The following methods only apply when an instance of the class containing the
1529 method (a so-called *descriptor* class) appears in the class dictionary of
1530 another new-style class, known as the *owner* class. In the examples below, "the
1531 attribute" refers to the attribute whose name is the key of the property in the
1532 owner class' ``__dict__``. Descriptors can only be implemented as new-style
1536 .. method:: object.__get__(self, instance, owner)
1538 Called to get the attribute of the owner class (class attribute access) or of an
1539 instance of that class (instance attribute access). *owner* is always the owner
1540 class, while *instance* is the instance that the attribute was accessed through,
1541 or ``None`` when the attribute is accessed through the *owner*. This method
1542 should return the (computed) attribute value or raise an :exc:`AttributeError`
1546 .. method:: object.__set__(self, instance, value)
1548 Called to set the attribute on an instance *instance* of the owner class to a
1552 .. method:: object.__delete__(self, instance)
1554 Called to delete the attribute on an instance *instance* of the owner class.
1557 .. _descriptor-invocation:
1559 Invoking Descriptors
1560 ^^^^^^^^^^^^^^^^^^^^
1562 In general, a descriptor is an object attribute with "binding behavior", one
1563 whose attribute access has been overridden by methods in the descriptor
1564 protocol: :meth:`__get__`, :meth:`__set__`, and :meth:`__delete__`. If any of
1565 those methods are defined for an object, it is said to be a descriptor.
1567 The default behavior for attribute access is to get, set, or delete the
1568 attribute from an object's dictionary. For instance, ``a.x`` has a lookup chain
1569 starting with ``a.__dict__['x']``, then ``type(a).__dict__['x']``, and
1570 continuing through the base classes of ``type(a)`` excluding metaclasses.
1572 However, if the looked-up value is an object defining one of the descriptor
1573 methods, then Python may override the default behavior and invoke the descriptor
1574 method instead. Where this occurs in the precedence chain depends on which
1575 descriptor methods were defined and how they were called. Note that descriptors
1576 are only invoked for new style objects or classes (ones that subclass
1577 :class:`object()` or :class:`type()`).
1579 The starting point for descriptor invocation is a binding, ``a.x``. How the
1580 arguments are assembled depends on ``a``:
1583 The simplest and least common call is when user code directly invokes a
1584 descriptor method: ``x.__get__(a)``.
1587 If binding to a new-style object instance, ``a.x`` is transformed into the call:
1588 ``type(a).__dict__['x'].__get__(a, type(a))``.
1591 If binding to a new-style class, ``A.x`` is transformed into the call:
1592 ``A.__dict__['x'].__get__(None, A)``.
1595 If ``a`` is an instance of :class:`super`, then the binding ``super(B,
1596 obj).m()`` searches ``obj.__class__.__mro__`` for the base class ``A``
1597 immediately preceding ``B`` and then invokes the descriptor with the call:
1598 ``A.__dict__['m'].__get__(obj, A)``.
1600 For instance bindings, the precedence of descriptor invocation depends on the
1601 which descriptor methods are defined. Normally, data descriptors define both
1602 :meth:`__get__` and :meth:`__set__`, while non-data descriptors have just the
1603 :meth:`__get__` method. Data descriptors always override a redefinition in an
1604 instance dictionary. In contrast, non-data descriptors can be overridden by
1607 Python methods (including :func:`staticmethod` and :func:`classmethod`) are
1608 implemented as non-data descriptors. Accordingly, instances can redefine and
1609 override methods. This allows individual instances to acquire behaviors that
1610 differ from other instances of the same class.
1612 The :func:`property` function is implemented as a data descriptor. Accordingly,
1613 instances cannot override the behavior of a property.
1621 By default, instances of both old and new-style classes have a dictionary for
1622 attribute storage. This wastes space for objects having very few instance
1623 variables. The space consumption can become acute when creating large numbers
1626 The default can be overridden by defining *__slots__* in a new-style class
1627 definition. The *__slots__* declaration takes a sequence of instance variables
1628 and reserves just enough space in each instance to hold a value for each
1629 variable. Space is saved because *__dict__* is not created for each instance.
1634 This class variable can be assigned a string, iterable, or sequence of strings
1635 with variable names used by instances. If defined in a new-style class,
1636 *__slots__* reserves space for the declared variables and prevents the automatic
1637 creation of *__dict__* and *__weakref__* for each instance.
1639 .. versionadded:: 2.2
1641 Notes on using *__slots__*
1643 * When inheriting from a class without *__slots__*, the *__dict__* attribute of
1644 that class will always be accessible, so a *__slots__* definition in the
1645 subclass is meaningless.
1647 * Without a *__dict__* variable, instances cannot be assigned new variables not
1648 listed in the *__slots__* definition. Attempts to assign to an unlisted
1649 variable name raises :exc:`AttributeError`. If dynamic assignment of new
1650 variables is desired, then add ``'__dict__'`` to the sequence of strings in the
1651 *__slots__* declaration.
1653 .. versionchanged:: 2.3
1654 Previously, adding ``'__dict__'`` to the *__slots__* declaration would not
1655 enable the assignment of new attributes not specifically listed in the sequence
1656 of instance variable names.
1658 * Without a *__weakref__* variable for each instance, classes defining
1659 *__slots__* do not support weak references to its instances. If weak reference
1660 support is needed, then add ``'__weakref__'`` to the sequence of strings in the
1661 *__slots__* declaration.
1663 .. versionchanged:: 2.3
1664 Previously, adding ``'__weakref__'`` to the *__slots__* declaration would not
1665 enable support for weak references.
1667 * *__slots__* are implemented at the class level by creating descriptors
1668 (:ref:`descriptors`) for each variable name. As a result, class attributes
1669 cannot be used to set default values for instance variables defined by
1670 *__slots__*; otherwise, the class attribute would overwrite the descriptor
1673 * If a class defines a slot also defined in a base class, the instance variable
1674 defined by the base class slot is inaccessible (except by retrieving its
1675 descriptor directly from the base class). This renders the meaning of the
1676 program undefined. In the future, a check may be added to prevent this.
1678 * The action of a *__slots__* declaration is limited to the class where it is
1679 defined. As a result, subclasses will have a *__dict__* unless they also define
1682 * Nonempty *__slots__* does not work for classes derived from "variable-length"
1683 built-in types such as :class:`long`, :class:`str` and :class:`tuple`.
1685 * Any non-string iterable may be assigned to *__slots__*. Mappings may also be
1686 used; however, in the future, special meaning may be assigned to the values
1687 corresponding to each key.
1689 * *__class__* assignment works only if both classes have the same *__slots__*.
1691 .. versionchanged:: 2.6
1692 Previously, *__class__* assignment raised an error if either new or old class
1698 Customizing class creation
1699 --------------------------
1701 By default, new-style classes are constructed using :func:`type`. A class
1702 definition is read into a separate namespace and the value of class name is
1703 bound to the result of ``type(name, bases, dict)``.
1705 When the class definition is read, if *__metaclass__* is defined then the
1706 callable assigned to it will be called instead of :func:`type`. This allows
1707 classes or functions to be written which monitor or alter the class creation
1710 * Modifying the class dictionary prior to the class being created.
1712 * Returning an instance of another class -- essentially performing the role of a
1715 These steps will have to be performed in the metaclass's :meth:`__new__` method
1716 -- :meth:`type.__new__` can then be called from this method to create a class
1717 with different properties. This example adds a new element to the class
1718 dictionary before creating the class::
1720 class metacls(type):
1721 def __new__(mcs, name, bases, dict):
1722 dict['foo'] = 'metacls was here'
1723 return type.__new__(mcs, name, bases, dict)
1725 You can of course also override other class methods (or add new methods); for
1726 example defining a custom :meth:`__call__` method in the metaclass allows custom
1727 behavior when the class is called, e.g. not always creating a new instance.
1730 .. data:: __metaclass__
1732 This variable can be any callable accepting arguments for ``name``, ``bases``,
1733 and ``dict``. Upon class creation, the callable is used instead of the built-in
1736 .. versionadded:: 2.2
1738 The appropriate metaclass is determined by the following precedence rules:
1740 * If ``dict['__metaclass__']`` exists, it is used.
1742 * Otherwise, if there is at least one base class, its metaclass is used (this
1743 looks for a *__class__* attribute first and if not found, uses its type).
1745 * Otherwise, if a global variable named __metaclass__ exists, it is used.
1747 * Otherwise, the old-style, classic metaclass (types.ClassType) is used.
1749 The potential uses for metaclasses are boundless. Some ideas that have been
1750 explored including logging, interface checking, automatic delegation, automatic
1751 property creation, proxies, frameworks, and automatic resource
1752 locking/synchronization.
1757 Emulating callable objects
1758 --------------------------
1761 .. method:: object.__call__(self[, args...])
1763 .. index:: pair: call; instance
1765 Called when the instance is "called" as a function; if this method is defined,
1766 ``x(arg1, arg2, ...)`` is a shorthand for ``x.__call__(arg1, arg2, ...)``.
1771 Emulating container types
1772 -------------------------
1774 The following methods can be defined to implement container objects. Containers
1775 usually are sequences (such as lists or tuples) or mappings (like dictionaries),
1776 but can represent other containers as well. The first set of methods is used
1777 either to emulate a sequence or to emulate a mapping; the difference is that for
1778 a sequence, the allowable keys should be the integers *k* for which ``0 <= k <
1779 N`` where *N* is the length of the sequence, or slice objects, which define a
1780 range of items. (For backwards compatibility, the method :meth:`__getslice__`
1781 (see below) can also be defined to handle simple, but not extended slices.) It
1782 is also recommended that mappings provide the methods :meth:`keys`,
1783 :meth:`values`, :meth:`items`, :meth:`has_key`, :meth:`get`, :meth:`clear`,
1784 :meth:`setdefault`, :meth:`iterkeys`, :meth:`itervalues`, :meth:`iteritems`,
1785 :meth:`pop`, :meth:`popitem`, :meth:`copy`, and :meth:`update` behaving similar
1786 to those for Python's standard dictionary objects. The :mod:`UserDict` module
1787 provides a :class:`DictMixin` class to help create those methods from a base set
1788 of :meth:`__getitem__`, :meth:`__setitem__`, :meth:`__delitem__`, and
1789 :meth:`keys`. Mutable sequences should provide methods :meth:`append`,
1790 :meth:`count`, :meth:`index`, :meth:`extend`, :meth:`insert`, :meth:`pop`,
1791 :meth:`remove`, :meth:`reverse` and :meth:`sort`, like Python standard list
1792 objects. Finally, sequence types should implement addition (meaning
1793 concatenation) and multiplication (meaning repetition) by defining the methods
1794 :meth:`__add__`, :meth:`__radd__`, :meth:`__iadd__`, :meth:`__mul__`,
1795 :meth:`__rmul__` and :meth:`__imul__` described below; they should not define
1796 :meth:`__coerce__` or other numerical operators. It is recommended that both
1797 mappings and sequences implement the :meth:`__contains__` method to allow
1798 efficient use of the ``in`` operator; for mappings, ``in`` should be equivalent
1799 of :meth:`has_key`; for sequences, it should search through the values. It is
1800 further recommended that both mappings and sequences implement the
1801 :meth:`__iter__` method to allow efficient iteration through the container; for
1802 mappings, :meth:`__iter__` should be the same as :meth:`iterkeys`; for
1803 sequences, it should iterate through the values.
1806 .. method:: object.__len__(self)
1810 single: __nonzero__() (object method)
1812 Called to implement the built-in function :func:`len`. Should return the length
1813 of the object, an integer ``>=`` 0. Also, an object that doesn't define a
1814 :meth:`__nonzero__` method and whose :meth:`__len__` method returns zero is
1815 considered to be false in a Boolean context.
1818 .. method:: object.__getitem__(self, key)
1820 .. index:: object: slice
1822 Called to implement evaluation of ``self[key]``. For sequence types, the
1823 accepted keys should be integers and slice objects. Note that the special
1824 interpretation of negative indexes (if the class wishes to emulate a sequence
1825 type) is up to the :meth:`__getitem__` method. If *key* is of an inappropriate
1826 type, :exc:`TypeError` may be raised; if of a value outside the set of indexes
1827 for the sequence (after any special interpretation of negative values),
1828 :exc:`IndexError` should be raised. For mapping types, if *key* is missing (not
1829 in the container), :exc:`KeyError` should be raised.
1833 :keyword:`for` loops expect that an :exc:`IndexError` will be raised for illegal
1834 indexes to allow proper detection of the end of the sequence.
1837 .. method:: object.__setitem__(self, key, value)
1839 Called to implement assignment to ``self[key]``. Same note as for
1840 :meth:`__getitem__`. This should only be implemented for mappings if the
1841 objects support changes to the values for keys, or if new keys can be added, or
1842 for sequences if elements can be replaced. The same exceptions should be raised
1843 for improper *key* values as for the :meth:`__getitem__` method.
1846 .. method:: object.__delitem__(self, key)
1848 Called to implement deletion of ``self[key]``. Same note as for
1849 :meth:`__getitem__`. This should only be implemented for mappings if the
1850 objects support removal of keys, or for sequences if elements can be removed
1851 from the sequence. The same exceptions should be raised for improper *key*
1852 values as for the :meth:`__getitem__` method.
1855 .. method:: object.__iter__(self)
1857 This method is called when an iterator is required for a container. This method
1858 should return a new iterator object that can iterate over all the objects in the
1859 container. For mappings, it should iterate over the keys of the container, and
1860 should also be made available as the method :meth:`iterkeys`.
1862 Iterator objects also need to implement this method; they are required to return
1863 themselves. For more information on iterator objects, see :ref:`typeiter`.
1866 .. method:: object.__reversed__(self)
1868 Called (if present) by the :func:`reversed` builtin to implement
1869 reverse iteration. It should return a new iterator object that iterates
1870 over all the objects in the container in reverse order.
1872 If the :meth:`__reversed__` method is not provided, the
1873 :func:`reversed` builtin will fall back to using the sequence protocol
1874 (:meth:`__len__` and :meth:`__getitem__`). Objects should normally
1875 only provide :meth:`__reversed__` if they do not support the sequence
1876 protocol and an efficient implementation of reverse iteration is possible.
1878 .. versionadded:: 2.6
1881 The membership test operators (:keyword:`in` and :keyword:`not in`) are normally
1882 implemented as an iteration through a sequence. However, container objects can
1883 supply the following special method with a more efficient implementation, which
1884 also does not require the object be a sequence.
1887 .. method:: object.__contains__(self, item)
1889 Called to implement membership test operators. Should return true if *item* is
1890 in *self*, false otherwise. For mapping objects, this should consider the keys
1891 of the mapping rather than the values or the key-item pairs.
1894 .. _sequence-methods:
1896 Additional methods for emulation of sequence types
1897 --------------------------------------------------
1899 The following optional methods can be defined to further emulate sequence
1900 objects. Immutable sequences methods should at most only define
1901 :meth:`__getslice__`; mutable sequences might define all three methods.
1904 .. method:: object.__getslice__(self, i, j)
1907 Support slice objects as parameters to the :meth:`__getitem__` method.
1908 (However, built-in types in CPython currently still implement
1909 :meth:`__getslice__`. Therefore, you have to override it in derived
1910 classes when implementing slicing.)
1912 Called to implement evaluation of ``self[i:j]``. The returned object should be
1913 of the same type as *self*. Note that missing *i* or *j* in the slice
1914 expression are replaced by zero or ``sys.maxint``, respectively. If negative
1915 indexes are used in the slice, the length of the sequence is added to that
1916 index. If the instance does not implement the :meth:`__len__` method, an
1917 :exc:`AttributeError` is raised. No guarantee is made that indexes adjusted this
1918 way are not still negative. Indexes which are greater than the length of the
1919 sequence are not modified. If no :meth:`__getslice__` is found, a slice object
1920 is created instead, and passed to :meth:`__getitem__` instead.
1923 .. method:: object.__setslice__(self, i, j, sequence)
1925 Called to implement assignment to ``self[i:j]``. Same notes for *i* and *j* as
1926 for :meth:`__getslice__`.
1928 This method is deprecated. If no :meth:`__setslice__` is found, or for extended
1929 slicing of the form ``self[i:j:k]``, a slice object is created, and passed to
1930 :meth:`__setitem__`, instead of :meth:`__setslice__` being called.
1933 .. method:: object.__delslice__(self, i, j)
1935 Called to implement deletion of ``self[i:j]``. Same notes for *i* and *j* as for
1936 :meth:`__getslice__`. This method is deprecated. If no :meth:`__delslice__` is
1937 found, or for extended slicing of the form ``self[i:j:k]``, a slice object is
1938 created, and passed to :meth:`__delitem__`, instead of :meth:`__delslice__`
1941 Notice that these methods are only invoked when a single slice with a single
1942 colon is used, and the slice method is available. For slice operations
1943 involving extended slice notation, or in absence of the slice methods,
1944 :meth:`__getitem__`, :meth:`__setitem__` or :meth:`__delitem__` is called with a
1945 slice object as argument.
1947 The following example demonstrate how to make your program or module compatible
1948 with earlier versions of Python (assuming that methods :meth:`__getitem__`,
1949 :meth:`__setitem__` and :meth:`__delitem__` support slice objects as
1954 def __getitem__(self, index):
1956 def __setitem__(self, index, value):
1958 def __delitem__(self, index):
1961 if sys.version_info < (2, 0):
1962 # They won't be defined if version is at least 2.0 final
1964 def __getslice__(self, i, j):
1965 return self[max(0, i):max(0, j):]
1966 def __setslice__(self, i, j, seq):
1967 self[max(0, i):max(0, j):] = seq
1968 def __delslice__(self, i, j):
1969 del self[max(0, i):max(0, j):]
1972 Note the calls to :func:`max`; these are necessary because of the handling of
1973 negative indices before the :meth:`__\*slice__` methods are called. When
1974 negative indexes are used, the :meth:`__\*item__` methods receive them as
1975 provided, but the :meth:`__\*slice__` methods get a "cooked" form of the index
1976 values. For each negative index value, the length of the sequence is added to
1977 the index before calling the method (which may still result in a negative
1978 index); this is the customary handling of negative indexes by the built-in
1979 sequence types, and the :meth:`__\*item__` methods are expected to do this as
1980 well. However, since they should already be doing that, negative indexes cannot
1981 be passed in; they must be constrained to the bounds of the sequence before
1982 being passed to the :meth:`__\*item__` methods. Calling ``max(0, i)``
1983 conveniently returns the proper value.
1988 Emulating numeric types
1989 -----------------------
1991 The following methods can be defined to emulate numeric objects. Methods
1992 corresponding to operations that are not supported by the particular kind of
1993 number implemented (e.g., bitwise operations for non-integral numbers) should be
1997 .. method:: object.__add__(self, other)
1998 object.__sub__(self, other)
1999 object.__mul__(self, other)
2000 object.__floordiv__(self, other)
2001 object.__mod__(self, other)
2002 object.__divmod__(self, other)
2003 object.__pow__(self, other[, modulo])
2004 object.__lshift__(self, other)
2005 object.__rshift__(self, other)
2006 object.__and__(self, other)
2007 object.__xor__(self, other)
2008 object.__or__(self, other)
2015 These methods are called to implement the binary arithmetic operations (``+``,
2016 ``-``, ``*``, ``//``, ``%``, :func:`divmod`, :func:`pow`, ``**``, ``<<``,
2017 ``>>``, ``&``, ``^``, ``|``). For instance, to evaluate the expression
2018 ``x + y``, where *x* is an instance of a class that has an :meth:`__add__`
2019 method, ``x.__add__(y)`` is called. The :meth:`__divmod__` method should be the
2020 equivalent to using :meth:`__floordiv__` and :meth:`__mod__`; it should not be
2021 related to :meth:`__truediv__` (described below). Note that :meth:`__pow__`
2022 should be defined to accept an optional third argument if the ternary version of
2023 the built-in :func:`pow` function is to be supported.
2025 If one of those methods does not support the operation with the supplied
2026 arguments, it should return ``NotImplemented``.
2029 .. method:: object.__div__(self, other)
2030 object.__truediv__(self, other)
2032 The division operator (``/``) is implemented by these methods. The
2033 :meth:`__truediv__` method is used when ``__future__.division`` is in effect,
2034 otherwise :meth:`__div__` is used. If only one of these two methods is defined,
2035 the object will not support division in the alternate context; :exc:`TypeError`
2036 will be raised instead.
2039 .. method:: object.__radd__(self, other)
2040 object.__rsub__(self, other)
2041 object.__rmul__(self, other)
2042 object.__rdiv__(self, other)
2043 object.__rtruediv__(self, other)
2044 object.__rfloordiv__(self, other)
2045 object.__rmod__(self, other)
2046 object.__rdivmod__(self, other)
2047 object.__rpow__(self, other)
2048 object.__rlshift__(self, other)
2049 object.__rrshift__(self, other)
2050 object.__rand__(self, other)
2051 object.__rxor__(self, other)
2052 object.__ror__(self, other)
2058 These methods are called to implement the binary arithmetic operations (``+``,
2059 ``-``, ``*``, ``/``, ``%``, :func:`divmod`, :func:`pow`, ``**``, ``<<``, ``>>``,
2060 ``&``, ``^``, ``|``) with reflected (swapped) operands. These functions are
2061 only called if the left operand does not support the corresponding operation and
2062 the operands are of different types. [#]_ For instance, to evaluate the
2063 expression ``x - y``, where *y* is an instance of a class that has an
2064 :meth:`__rsub__` method, ``y.__rsub__(x)`` is called if ``x.__sub__(y)`` returns
2067 .. index:: builtin: pow
2069 Note that ternary :func:`pow` will not try calling :meth:`__rpow__` (the
2070 coercion rules would become too complicated).
2074 If the right operand's type is a subclass of the left operand's type and that
2075 subclass provides the reflected method for the operation, this method will be
2076 called before the left operand's non-reflected method. This behavior allows
2077 subclasses to override their ancestors' operations.
2080 .. method:: object.__iadd__(self, other)
2081 object.__isub__(self, other)
2082 object.__imul__(self, other)
2083 object.__idiv__(self, other)
2084 object.__itruediv__(self, other)
2085 object.__ifloordiv__(self, other)
2086 object.__imod__(self, other)
2087 object.__ipow__(self, other[, modulo])
2088 object.__ilshift__(self, other)
2089 object.__irshift__(self, other)
2090 object.__iand__(self, other)
2091 object.__ixor__(self, other)
2092 object.__ior__(self, other)
2094 These methods are called to implement the augmented arithmetic assignments
2095 (``+=``, ``-=``, ``*=``, ``/=``, ``//=``, ``%=``, ``**=``, ``<<=``, ``>>=``,
2096 ``&=``, ``^=``, ``|=``). These methods should attempt to do the operation
2097 in-place (modifying *self*) and return the result (which could be, but does
2098 not have to be, *self*). If a specific method is not defined, the augmented
2099 assignment falls back to the normal methods. For instance, to execute the
2100 statement ``x += y``, where *x* is an instance of a class that has an
2101 :meth:`__iadd__` method, ``x.__iadd__(y)`` is called. If *x* is an instance
2102 of a class that does not define a :meth:`__iadd__` method, ``x.__add__(y)``
2103 and ``y.__radd__(x)`` are considered, as with the evaluation of ``x + y``.
2106 .. method:: object.__neg__(self)
2107 object.__pos__(self)
2108 object.__abs__(self)
2109 object.__invert__(self)
2111 .. index:: builtin: abs
2113 Called to implement the unary arithmetic operations (``-``, ``+``, :func:`abs`
2117 .. method:: object.__complex__(self)
2118 object.__int__(self)
2119 object.__long__(self)
2120 object.__float__(self)
2128 Called to implement the built-in functions :func:`complex`, :func:`int`,
2129 :func:`long`, and :func:`float`. Should return a value of the appropriate type.
2132 .. method:: object.__oct__(self)
2133 object.__hex__(self)
2139 Called to implement the built-in functions :func:`oct` and :func:`hex`. Should
2140 return a string value.
2143 .. method:: object.__index__(self)
2145 Called to implement :func:`operator.index`. Also called whenever Python needs
2146 an integer object (such as in slicing). Must return an integer (int or long).
2148 .. versionadded:: 2.5
2151 .. method:: object.__coerce__(self, other)
2153 Called to implement "mixed-mode" numeric arithmetic. Should either return a
2154 2-tuple containing *self* and *other* converted to a common numeric type, or
2155 ``None`` if conversion is impossible. When the common type would be the type of
2156 ``other``, it is sufficient to return ``None``, since the interpreter will also
2157 ask the other object to attempt a coercion (but sometimes, if the implementation
2158 of the other type cannot be changed, it is useful to do the conversion to the
2159 other type here). A return value of ``NotImplemented`` is equivalent to
2168 This section used to document the rules for coercion. As the language has
2169 evolved, the coercion rules have become hard to document precisely; documenting
2170 what one version of one particular implementation does is undesirable. Instead,
2171 here are some informal guidelines regarding coercion. In Python 3.0, coercion
2172 will not be supported.
2176 If the left operand of a % operator is a string or Unicode object, no coercion
2177 takes place and the string formatting operation is invoked instead.
2181 It is no longer recommended to define a coercion operation. Mixed-mode
2182 operations on types that don't define coercion pass the original arguments to
2187 New-style classes (those derived from :class:`object`) never invoke the
2188 :meth:`__coerce__` method in response to a binary operator; the only time
2189 :meth:`__coerce__` is invoked is when the built-in function :func:`coerce` is
2194 For most intents and purposes, an operator that returns ``NotImplemented`` is
2195 treated the same as one that is not implemented at all.
2199 Below, :meth:`__op__` and :meth:`__rop__` are used to signify the generic method
2200 names corresponding to an operator; :meth:`__iop__` is used for the
2201 corresponding in-place operator. For example, for the operator '``+``',
2202 :meth:`__add__` and :meth:`__radd__` are used for the left and right variant of
2203 the binary operator, and :meth:`__iadd__` for the in-place variant.
2207 For objects *x* and *y*, first ``x.__op__(y)`` is tried. If this is not
2208 implemented or returns ``NotImplemented``, ``y.__rop__(x)`` is tried. If this
2209 is also not implemented or returns ``NotImplemented``, a :exc:`TypeError`
2210 exception is raised. But see the following exception:
2214 Exception to the previous item: if the left operand is an instance of a built-in
2215 type or a new-style class, and the right operand is an instance of a proper
2216 subclass of that type or class and overrides the base's :meth:`__rop__` method,
2217 the right operand's :meth:`__rop__` method is tried *before* the left operand's
2218 :meth:`__op__` method.
2220 This is done so that a subclass can completely override binary operators.
2221 Otherwise, the left operand's :meth:`__op__` method would always accept the
2222 right operand: when an instance of a given class is expected, an instance of a
2223 subclass of that class is always acceptable.
2227 When either operand type defines a coercion, this coercion is called before that
2228 type's :meth:`__op__` or :meth:`__rop__` method is called, but no sooner. If
2229 the coercion returns an object of a different type for the operand whose
2230 coercion is invoked, part of the process is redone using the new object.
2234 When an in-place operator (like '``+=``') is used, if the left operand
2235 implements :meth:`__iop__`, it is invoked without any coercion. When the
2236 operation falls back to :meth:`__op__` and/or :meth:`__rop__`, the normal
2237 coercion rules apply.
2241 In ``x + y``, if *x* is a sequence that implements sequence concatenation,
2242 sequence concatenation is invoked.
2246 In ``x * y``, if one operator is a sequence that implements sequence
2247 repetition, and the other is an integer (:class:`int` or :class:`long`),
2248 sequence repetition is invoked.
2252 Rich comparisons (implemented by methods :meth:`__eq__` and so on) never use
2253 coercion. Three-way comparison (implemented by :meth:`__cmp__`) does use
2254 coercion under the same conditions as other binary operations use it.
2258 In the current implementation, the built-in numeric types :class:`int`,
2259 :class:`long` and :class:`float` do not use coercion; the type :class:`complex`
2260 however does use coercion for binary operators and rich comparisons, despite
2261 the above rules. The difference can become apparent when subclassing these
2262 types. Over time, the type :class:`complex` may be fixed to avoid coercion.
2263 All these types implement a :meth:`__coerce__` method, for use by the built-in
2264 :func:`coerce` function.
2267 .. _context-managers:
2269 With Statement Context Managers
2270 -------------------------------
2272 .. versionadded:: 2.5
2274 A :dfn:`context manager` is an object that defines the runtime context to be
2275 established when executing a :keyword:`with` statement. The context manager
2276 handles the entry into, and the exit from, the desired runtime context for the
2277 execution of the block of code. Context managers are normally invoked using the
2278 :keyword:`with` statement (described in section :ref:`with`), but can also be
2279 used by directly invoking their methods.
2283 single: context manager
2285 Typical uses of context managers include saving and restoring various kinds of
2286 global state, locking and unlocking resources, closing opened files, etc.
2288 For more information on context managers, see :ref:`typecontextmanager`.
2291 .. method:: object.__enter__(self)
2293 Enter the runtime context related to this object. The :keyword:`with` statement
2294 will bind this method's return value to the target(s) specified in the
2295 :keyword:`as` clause of the statement, if any.
2298 .. method:: object.__exit__(self, exc_type, exc_value, traceback)
2300 Exit the runtime context related to this object. The parameters describe the
2301 exception that caused the context to be exited. If the context was exited
2302 without an exception, all three arguments will be :const:`None`.
2304 If an exception is supplied, and the method wishes to suppress the exception
2305 (i.e., prevent it from being propagated), it should return a true value.
2306 Otherwise, the exception will be processed normally upon exit from this method.
2308 Note that :meth:`__exit__` methods should not reraise the passed-in exception;
2309 this is the caller's responsibility.
2314 :pep:`0343` - The "with" statement
2315 The specification, background, and examples for the Python :keyword:`with`
2319 .. _old-style-special-lookup:
2321 Special method lookup for old-style classes
2322 -------------------------------------------
2324 For old-style classes, special methods are always looked up in exactly the
2325 same way as any other method or attribute. This is the case regardless of
2326 whether the method is being looked up explicitly as in ``x.__getitem__(i)``
2327 or implicitly as in ``x[i]``.
2329 This behaviour means that special methods may exhibit different behaviour
2330 for different instances of a single old-style class if the appropriate
2331 special attributes are set differently::
2338 >>> c1.__len__ = lambda: 5
2339 >>> c2.__len__ = lambda: 9
2346 .. _new-style-special-lookup:
2348 Special method lookup for new-style classes
2349 -------------------------------------------
2351 For new-style classes, implicit invocations of special methods are only guaranteed
2352 to work correctly if defined on an object's type, not in the object's instance
2353 dictionary. That behaviour is the reason why the following code raises an
2354 exception (unlike the equivalent example with old-style classes)::
2356 >>> class C(object):
2360 >>> c.__len__ = lambda: 5
2362 Traceback (most recent call last):
2363 File "<stdin>", line 1, in <module>
2364 TypeError: object of type 'C' has no len()
2366 The rationale behind this behaviour lies with a number of special methods such
2367 as :meth:`__hash__` and :meth:`__repr__` that are implemented by all objects,
2368 including type objects. If the implicit lookup of these methods used the
2369 conventional lookup process, they would fail when invoked on the type object
2372 >>> 1 .__hash__() == hash(1)
2374 >>> int.__hash__() == hash(int)
2375 Traceback (most recent call last):
2376 File "<stdin>", line 1, in <module>
2377 TypeError: descriptor '__hash__' of 'int' object needs an argument
2379 Incorrectly attempting to invoke an unbound method of a class in this way is
2380 sometimes referred to as 'metaclass confusion', and is avoided by bypassing
2381 the instance when looking up special methods::
2383 >>> type(1).__hash__(1) == hash(1)
2385 >>> type(int).__hash__(int) == hash(int)
2388 In addition to bypassing any instance attributes in the interest of
2389 correctness, implicit special method lookup generally also bypasses the
2390 :meth:`__getattribute__` method even of the object's metaclass::
2392 >>> class Meta(type):
2393 ... def __getattribute__(*args):
2394 ... print "Metaclass getattribute invoked"
2395 ... return type.__getattribute__(*args)
2397 >>> class C(object):
2398 ... __metaclass__ = Meta
2399 ... def __len__(self):
2401 ... def __getattribute__(*args):
2402 ... print "Class getattribute invoked"
2403 ... return object.__getattribute__(*args)
2406 >>> c.__len__() # Explicit lookup via instance
2407 Class getattribute invoked
2409 >>> type(c).__len__(c) # Explicit lookup via type
2410 Metaclass getattribute invoked
2412 >>> len(c) # Implicit lookup
2415 Bypassing the :meth:`__getattribute__` machinery in this fashion
2416 provides significant scope for speed optimisations within the
2417 interpreter, at the cost of some flexibility in the handling of
2418 special methods (the special method *must* be set on the class
2419 object itself in order to be consistently invoked by the interpreter).
2422 .. rubric:: Footnotes
2424 .. [#] It *is* possible in some cases to change an object's type, under certain
2425 controlled conditions. It generally isn't a good idea though, since it can
2426 lead to some very strange behaviour if it is handled incorrectly.
2428 .. [#] A descriptor can define any combination of :meth:`__get__`,
2429 :meth:`__set__` and :meth:`__delete__`. If it does not define :meth:`__get__`,
2430 then accessing the attribute even on an instance will return the descriptor
2431 object itself. If the descriptor defines :meth:`__set__` and/or
2432 :meth:`__delete__`, it is a data descriptor; if it defines neither, it is a
2433 non-data descriptor.
2435 .. [#] For operands of the same type, it is assumed that if the non-reflected method
2436 (such as :meth:`__add__`) fails the operation is not supported, which is why the
2437 reflected method is not called.