"""Classes to represent arbitrary sets (including sets of sets).
This module implements sets using dictionaries whose values are
ignored. The usual operations (union, intersection, deletion, etc.)
are provided as both methods and operators.
Important: sets are not sequences! While they support 'x in s',
'len(s)', and 'for x in s', none of those operations are unique for
sequences; for example, mappings support all three as well. The
characteristic operation for sequences is subscripting with small
integers: s[i], for i in range(len(s)). Sets don't support
subscripting at all. Also, sequences allow multiple occurrences and
their elements have a definite order; sets on the other hand don't
record multiple occurrences and don't remember the order of element
insertion (which is why they don't support s[i]).
The following classes are provided:
BaseSet -- All the operations common to both mutable and immutable
sets. This is an abstract class, not meant to be directly
Set -- Mutable sets, subclass of BaseSet; not hashable.
ImmutableSet -- Immutable sets, subclass of BaseSet; hashable.
An iterable argument is mandatory to create an ImmutableSet.
_TemporarilyImmutableSet -- A wrapper around a Set, hashable,
giving the same hash value as the immutable set equivalent
would have. Do not use this class directly.
Only hashable objects can be added to a Set. In particular, you cannot
really add a Set as an element to another Set; if you try, what is
actually added is an ImmutableSet built from it (it compares equal to
the one you tried adding).
When you ask if `x in y' where x is a Set and y is a Set or
ImmutableSet, x is wrapped into a _TemporarilyImmutableSet z, and
what's tested is actually `z in y'.
# - Greg V. Wilson wrote the first version, using a different approach
# to the mutable/immutable problem, and inheriting from dict.
# - Alex Martelli modified Greg's version to implement the current
# Set/ImmutableSet approach, and make the data an attribute.
# - Guido van Rossum rewrote much of the code, made some API changes,
# and cleaned up the docstrings.
# - Raymond Hettinger added a number of speedups and other
from itertools import ifilter, ifilterfalse
__all__ = ['BaseSet', 'Set', 'ImmutableSet']
warnings.warn("the sets module is deprecated", DeprecationWarning,
"""Common base class for mutable and immutable sets."""
"""This is an abstract class."""
# Don't call this from a concrete subclass!
if self.__class__ is BaseSet:
raise TypeError, ("BaseSet is an abstract class. "
"Use Set or ImmutableSet.")
# Standard protocols: __len__, __repr__, __str__, __iter__
"""Return the number of elements of a set."""
"""Return string representation of a set.
This looks like 'Set([<list of elements>])'.
# __str__ is the same as __repr__
def _repr(self, sorted=False):
elements = self._data.keys()
return '%s(%r)' % (self.__class__.__name__, elements)
"""Return an iterator over the elements or a set.
This is the keys iterator for the underlying dict.
return self._data.iterkeys()
# Three-way comparison is not supported. However, because __eq__ is
# tried before __cmp__, if Set x == Set y, x.__eq__(y) returns True and
# then cmp(x, y) returns 0 (Python doesn't actually call __cmp__ in this
def __cmp__(self, other):
raise TypeError, "can't compare sets using cmp()"
# Equality comparisons using the underlying dicts. Mixed-type comparisons
# are allowed here, where Set == z for non-Set z always returns False,
# and Set != z always True. This allows expressions like "x in y" to
# give the expected result when y is a sequence of mixed types, not
# raising a pointless TypeError just because y contains a Set, or x is
# a Set and y contain's a non-set ("in" invokes only __eq__).
# Subtle: it would be nicer if __eq__ and __ne__ could return
# NotImplemented instead of True or False. Then the other comparand
# would get a chance to determine the result, and if the other comparand
# also returned NotImplemented then it would fall back to object address
# comparison (which would always return False for __eq__ and always
# True for __ne__). However, that doesn't work, because this type
# *also* implements __cmp__: if, e.g., __eq__ returns NotImplemented,
# Python tries __cmp__ next, and the __cmp__ here then raises TypeError.
if isinstance(other, BaseSet):
return self._data == other._data
if isinstance(other, BaseSet):
return self._data != other._data
"""Return a shallow copy of a set."""
result = self.__class__()
result._data.update(self._data)
__copy__ = copy # For the copy module
def __deepcopy__(self, memo):
"""Return a deep copy of a set; used by copy module."""
# This pre-creates the result and inserts it in the memo
# early, in case the deep copy recurses into another reference
# to this same set. A set can't be an element of itself, but
# it can certainly contain an object that has a reference to
from copy import deepcopy
result = self.__class__()
data[deepcopy(elt, memo)] = value
# Standard set operations: union, intersection, both differences.
# Each has an operator version (e.g. __or__, invoked with |) and a
# method version (e.g. union).
# Subtle: Each pair requires distinct code so that the outcome is
# correct when the type of other isn't suitable. For example, if
# we did "union = __or__" instead, then Set().union(3) would return
# NotImplemented instead of raising TypeError (albeit that *why* it
# raises TypeError as-is is also a bit subtle).
"""Return the union of two sets as a new set.
(I.e. all elements that are in either set.)
if not isinstance(other, BaseSet):
"""Return the union of two sets as a new set.
(I.e. all elements that are in either set.)
result = self.__class__(self)
def __and__(self, other):
"""Return the intersection of two sets as a new set.
(I.e. all elements that are in both sets.)
if not isinstance(other, BaseSet):
return self.intersection(other)
def intersection(self, other):
"""Return the intersection of two sets as a new set.
(I.e. all elements that are in both sets.)
if not isinstance(other, BaseSet):
if len(self) <= len(other):
little, big = self, other
little, big = other, self
common = ifilter(big._data.__contains__, little)
return self.__class__(common)
def __xor__(self, other):
"""Return the symmetric difference of two sets as a new set.
(I.e. all elements that are in exactly one of the sets.)
if not isinstance(other, BaseSet):
return self.symmetric_difference(other)
def symmetric_difference(self, other):
"""Return the symmetric difference of two sets as a new set.
(I.e. all elements that are in exactly one of the sets.)
result = self.__class__()
otherdata = Set(other)._data
for elt in ifilterfalse(otherdata.__contains__, selfdata):
for elt in ifilterfalse(selfdata.__contains__, otherdata):
def __sub__(self, other):
"""Return the difference of two sets as a new Set.
(I.e. all elements that are in this set and not in the other.)
if not isinstance(other, BaseSet):
return self.difference(other)
def difference(self, other):
"""Return the difference of two sets as a new Set.
(I.e. all elements that are in this set and not in the other.)
result = self.__class__()
otherdata = Set(other)._data
for elt in ifilterfalse(otherdata.__contains__, self):
def __contains__(self, element):
"""Report whether an element is a member of a set.
(Called in response to the expression `element in self'.)
return element in self._data
transform = getattr(element, "__as_temporarily_immutable__", None)
raise # re-raise the TypeError exception we caught
return transform() in self._data
# Subset and superset test
def issubset(self, other):
"""Report whether another set contains this set."""
self._binary_sanity_check(other)
if len(self) > len(other): # Fast check for obvious cases
for elt in ifilterfalse(other._data.__contains__, self):
def issuperset(self, other):
"""Report whether this set contains another set."""
self._binary_sanity_check(other)
if len(self) < len(other): # Fast check for obvious cases
for elt in ifilterfalse(self._data.__contains__, other):
# Inequality comparisons using the is-subset relation.
self._binary_sanity_check(other)
return len(self) < len(other) and self.issubset(other)
self._binary_sanity_check(other)
return len(self) > len(other) and self.issuperset(other)
# We inherit object.__hash__, so we must deny this explicitly
def _binary_sanity_check(self, other):
# Check that the other argument to a binary operation is also
# a set, raising a TypeError otherwise.
if not isinstance(other, BaseSet):
raise TypeError, "Binary operation only permitted between sets"
# Calculate hash code for a set by xor'ing the hash codes of
# the elements. This ensures that the hash code does not depend
# on the order in which elements are added to the set. This is
# not called __hash__ because a BaseSet should not be hashable;
# only an ImmutableSet is hashable.
def _update(self, iterable):
# The main loop for update() and the subclass __init__() methods.
# Use the fast update() method when a dictionary is available.
if isinstance(iterable, BaseSet):
data.update(iterable._data)
if type(iterable) in (list, tuple, xrange):
# Optimized: we know that __iter__() and next() can't
# raise TypeError, so we can move 'try:' out of the loop.
transform = getattr(element, "__as_immutable__", None)
raise # re-raise the TypeError exception we caught
data[transform()] = value
# Safe: only catch TypeError where intended
transform = getattr(element, "__as_immutable__", None)
raise # re-raise the TypeError exception we caught
data[transform()] = value
class ImmutableSet(BaseSet):
"""Immutable set class."""
__slots__ = ['_hashcode']
def __init__(self, iterable=None):
"""Construct an immutable set from an optional iterable."""
if self._hashcode is None:
self._hashcode = self._compute_hash()
return self._data, self._hashcode
def __setstate__(self, state):
self._data, self._hashcode = state
""" Mutable set class."""
# BaseSet + operations requiring mutability; no hashing
def __init__(self, iterable=None):
"""Construct a set from an optional iterable."""
# getstate's results are ignored if it is not
def __setstate__(self, data):
# In-place union, intersection, differences.
# Subtle: The xyz_update() functions deliberately return None,
# as do all mutating operations on built-in container types.
# The __xyz__ spellings have to return self, though.
def __ior__(self, other):
"""Update a set with the union of itself and another."""
self._binary_sanity_check(other)
self._data.update(other._data)
def union_update(self, other):
"""Update a set with the union of itself and another."""
def __iand__(self, other):
"""Update a set with the intersection of itself and another."""
self._binary_sanity_check(other)
self._data = (self & other)._data
def intersection_update(self, other):
"""Update a set with the intersection of itself and another."""
if isinstance(other, BaseSet):
self._data = (self.intersection(other))._data
def __ixor__(self, other):
"""Update a set with the symmetric difference of itself and another."""
self._binary_sanity_check(other)
self.symmetric_difference_update(other)
def symmetric_difference_update(self, other):
"""Update a set with the symmetric difference of itself and another."""
if not isinstance(other, BaseSet):
def __isub__(self, other):
"""Remove all elements of another set from this set."""
self._binary_sanity_check(other)
self.difference_update(other)
def difference_update(self, other):
"""Remove all elements of another set from this set."""
if not isinstance(other, BaseSet):
for elt in ifilter(data.__contains__, other):
# Python dict-like mass mutations: update, clear
def update(self, iterable):
"""Add all values from an iterable (such as a list or file)."""
"""Remove all elements from this set."""
# Single-element mutations: add, remove, discard
"""Add an element to a set.
It is recommended that you Edit text format, this type of Fix handles quite a lot in one request
