Raymond Hettinger | e5820c6 | 2011-03-22 09:11:39 -0700 | [diff] [blame] | 1 | |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 2 | :mod:`collections` --- Container datatypes |
| 3 | ========================================== |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 4 | |
| 5 | .. module:: collections |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 6 | :synopsis: Container datatypes |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 7 | .. moduleauthor:: Raymond Hettinger <python@rcn.com> |
| 8 | .. sectionauthor:: Raymond Hettinger <python@rcn.com> |
| 9 | |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 10 | .. testsetup:: * |
| 11 | |
| 12 | from collections import * |
| 13 | import itertools |
| 14 | __name__ = '<doctest>' |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 15 | |
Raymond Hettinger | 158c9c2 | 2011-02-22 00:41:50 +0000 | [diff] [blame] | 16 | **Source code:** :source:`Lib/collections/__init__.py` |
Raymond Hettinger | 1048094 | 2011-01-10 03:26:08 +0000 | [diff] [blame] | 17 | |
Raymond Hettinger | 4f707fd | 2011-01-10 19:54:11 +0000 | [diff] [blame] | 18 | -------------- |
| 19 | |
Raymond Hettinger | a6b76ba | 2010-08-08 00:29:08 +0000 | [diff] [blame] | 20 | This module implements specialized container datatypes providing alternatives to |
| 21 | Python's general purpose built-in containers, :class:`dict`, :class:`list`, |
| 22 | :class:`set`, and :class:`tuple`. |
Christian Heimes | 0bd4e11 | 2008-02-12 22:59:25 +0000 | [diff] [blame] | 23 | |
Raymond Hettinger | a6b76ba | 2010-08-08 00:29:08 +0000 | [diff] [blame] | 24 | ===================== ==================================================================== |
| 25 | :func:`namedtuple` factory function for creating tuple subclasses with named fields |
| 26 | :class:`deque` list-like container with fast appends and pops on either end |
Raymond Hettinger | 9fe1ccf | 2011-02-26 01:02:51 +0000 | [diff] [blame] | 27 | :class:`ChainMap` dict-like class for creating a single view of multiple mappings |
Raymond Hettinger | a6b76ba | 2010-08-08 00:29:08 +0000 | [diff] [blame] | 28 | :class:`Counter` dict subclass for counting hashable objects |
| 29 | :class:`OrderedDict` dict subclass that remembers the order entries were added |
| 30 | :class:`defaultdict` dict subclass that calls a factory function to supply missing values |
| 31 | :class:`UserDict` wrapper around dictionary objects for easier dict subclassing |
| 32 | :class:`UserList` wrapper around list objects for easier list subclassing |
| 33 | :class:`UserString` wrapper around string objects for easier string subclassing |
| 34 | ===================== ==================================================================== |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 35 | |
Raymond Hettinger | 158c9c2 | 2011-02-22 00:41:50 +0000 | [diff] [blame] | 36 | .. versionchanged:: 3.3 |
| 37 | Moved :ref:`abstract-base-classes` to the :mod:`collections.abc` module. |
| 38 | For backwards compatibility, they continue to be visible in this module |
| 39 | as well. |
Mark Summerfield | 08898b4 | 2007-09-05 08:43:04 +0000 | [diff] [blame] | 40 | |
| 41 | |
Raymond Hettinger | 9fe1ccf | 2011-02-26 01:02:51 +0000 | [diff] [blame] | 42 | :class:`ChainMap` objects |
| 43 | ------------------------- |
| 44 | |
| 45 | A :class:`ChainMap` class is provided for quickly linking a number of mappings |
| 46 | so they can be treated as a single unit. It is often much faster than creating |
| 47 | a new dictionary and running multiple :meth:`~dict.update` calls. |
| 48 | |
| 49 | The class can be used to simulate nested scopes and is useful in templating. |
| 50 | |
| 51 | .. class:: ChainMap(*maps) |
| 52 | |
| 53 | A :class:`ChainMap` groups multiple dicts or other mappings together to |
| 54 | create a single, updateable view. If no *maps* are specified, a single empty |
| 55 | dictionary is provided so that a new chain always has at least one mapping. |
| 56 | |
| 57 | The underlying mappings are stored in a list. That list is public and can |
| 58 | accessed or updated using the *maps* attribute. There is no other state. |
| 59 | |
| 60 | Lookups search the underlying mappings successively until a key is found. In |
| 61 | contrast, writes, updates, and deletions only operate on the first mapping. |
| 62 | |
Raymond Hettinger | 4bbde72 | 2011-04-11 17:57:21 -0700 | [diff] [blame] | 63 | A :class:`ChainMap` incorporates the underlying mappings by reference. So, if |
Raymond Hettinger | 9fe1ccf | 2011-02-26 01:02:51 +0000 | [diff] [blame] | 64 | one of the underlying mappings gets updated, those changes will be reflected |
Raymond Hettinger | 4bbde72 | 2011-04-11 17:57:21 -0700 | [diff] [blame] | 65 | in :class:`ChainMap`. |
Raymond Hettinger | 9fe1ccf | 2011-02-26 01:02:51 +0000 | [diff] [blame] | 66 | |
| 67 | All of the usual dictionary methods are supported. In addition, there is a |
| 68 | *maps* attribute, a method for creating new subcontexts, and a property for |
| 69 | accessing all but the first mapping: |
| 70 | |
| 71 | .. attribute:: maps |
| 72 | |
| 73 | A user updateable list of mappings. The list is ordered from |
| 74 | first-searched to last-searched. It is the only stored state and can |
| 75 | modified to change which mappings are searched. The list should |
| 76 | always contain at least one mapping. |
| 77 | |
| 78 | .. method:: new_child() |
| 79 | |
| 80 | Returns a new :class:`ChainMap` containing a new :class:`dict` followed by |
| 81 | all of the maps in the current instance. A call to ``d.new_child()`` is |
| 82 | equivalent to: ``ChainMap({}, *d.maps)``. This method is used for |
| 83 | creating subcontexts that can be updated without altering values in any |
| 84 | of the parent mappings. |
| 85 | |
| 86 | .. attribute:: parents() |
| 87 | |
| 88 | Returns a new :class:`ChainMap` containing all of the maps in the current |
| 89 | instance except the first one. This is useful for skipping the first map |
| 90 | in the search. The use-cases are similar to those for the |
| 91 | :keyword:`nonlocal` keyword used in :term:`nested scopes <nested scope>`. |
| 92 | The use-cases also parallel those for the builtin :func:`super` function. |
| 93 | A reference to ``d.parents`` is equivalent to: ``ChainMap(*d.maps[1:])``. |
| 94 | |
| 95 | .. versionadded:: 3.3 |
| 96 | |
| 97 | Example of simulating Python's internal lookup chain:: |
| 98 | |
Raymond Hettinger | 94ca211 | 2011-02-26 02:48:44 +0000 | [diff] [blame] | 99 | import builtins |
| 100 | pylookup = ChainMap(locals(), globals(), vars(builtins)) |
Raymond Hettinger | 9fe1ccf | 2011-02-26 01:02:51 +0000 | [diff] [blame] | 101 | |
| 102 | Example of letting user specified values take precedence over environment |
| 103 | variables which in turn take precedence over default values:: |
| 104 | |
| 105 | import os, argparse |
| 106 | defaults = {'color': 'red', 'user': guest} |
| 107 | parser = argparse.ArgumentParser() |
| 108 | parser.add_argument('-u', '--user') |
| 109 | parser.add_argument('-c', '--color') |
| 110 | user_specified = vars(parser.parse_args()) |
| 111 | combined = ChainMap(user_specified, os.environ, defaults) |
| 112 | |
| 113 | Example patterns for using the :class:`ChainMap` class to simulate nested |
| 114 | contexts:: |
| 115 | |
| 116 | c = ChainMap() Create root context |
| 117 | d = c.new_child() Create nested child context |
| 118 | e = c.new_child() Child of c, independent from d |
| 119 | e.maps[0] Current context dictionary -- like Python's locals() |
| 120 | e.maps[-1] Root context -- like Python's globals() |
| 121 | e.parents Enclosing context chain -- like Python's nonlocals |
| 122 | |
| 123 | d['x'] Get first key in the chain of contexts |
| 124 | d['x'] = 1 Set value in current context |
| 125 | del['x'] Delete from current context |
| 126 | list(d) All nested values |
| 127 | k in d Check all nested values |
| 128 | len(d) Number of nested values |
| 129 | d.items() All nested items |
| 130 | dict(d) Flatten into a regular dictionary |
| 131 | |
| 132 | .. seealso:: |
| 133 | |
| 134 | * The `MultiContext class |
| 135 | <http://svn.enthought.com/svn/enthought/CodeTools/trunk/enthought/contexts/multi_context.py>`_ |
| 136 | in the Enthought `CodeTools package |
| 137 | <https://github.com/enthought/codetools>`_\ has options to support |
| 138 | writing to any mapping in the chain. |
| 139 | |
| 140 | * Django's `Context class |
| 141 | <http://code.djangoproject.com/browser/django/trunk/django/template/context.py>`_ |
| 142 | for templating is a read-only chain of mappings. It also features |
| 143 | pushing and popping of contexts similar to the |
| 144 | :meth:`~collections.ChainMap.new_child` method and the |
| 145 | :meth:`~collections.ChainMap.parents` property. |
| 146 | |
| 147 | * The `Nested Contexts recipe |
| 148 | <http://code.activestate.com/recipes/577434/>`_ has options to control |
| 149 | whether writes and other mutations apply only to the first mapping or to |
| 150 | any mapping in the chain. |
| 151 | |
| 152 | * A `greatly simplified read-only version of Chainmap |
| 153 | <http://code.activestate.com/recipes/305268/>`_\. |
| 154 | |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 155 | :class:`Counter` objects |
| 156 | ------------------------ |
| 157 | |
| 158 | A counter tool is provided to support convenient and rapid tallies. |
| 159 | For example:: |
| 160 | |
Raymond Hettinger | 1c62dc9 | 2009-02-04 11:41:45 +0000 | [diff] [blame] | 161 | >>> # Tally occurrences of words in a list |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 162 | >>> cnt = Counter() |
Raymond Hettinger | 670eaec | 2009-01-21 23:14:07 +0000 | [diff] [blame] | 163 | >>> for word in ['red', 'blue', 'red', 'green', 'blue', 'blue']: |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 164 | ... cnt[word] += 1 |
| 165 | >>> cnt |
| 166 | Counter({'blue': 3, 'red': 2, 'green': 1}) |
| 167 | |
Raymond Hettinger | 1c62dc9 | 2009-02-04 11:41:45 +0000 | [diff] [blame] | 168 | >>> # Find the ten most common words in Hamlet |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 169 | >>> import re |
| 170 | >>> words = re.findall('\w+', open('hamlet.txt').read().lower()) |
Raymond Hettinger | 0bae662 | 2009-01-20 13:00:59 +0000 | [diff] [blame] | 171 | >>> Counter(words).most_common(10) |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 172 | [('the', 1143), ('and', 966), ('to', 762), ('of', 669), ('i', 631), |
| 173 | ('you', 554), ('a', 546), ('my', 514), ('hamlet', 471), ('in', 451)] |
| 174 | |
| 175 | .. class:: Counter([iterable-or-mapping]) |
| 176 | |
Raymond Hettinger | 670eaec | 2009-01-21 23:14:07 +0000 | [diff] [blame] | 177 | A :class:`Counter` is a :class:`dict` subclass for counting hashable objects. |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 178 | It is an unordered collection where elements are stored as dictionary keys |
| 179 | and their counts are stored as dictionary values. Counts are allowed to be |
| 180 | any integer value including zero or negative counts. The :class:`Counter` |
| 181 | class is similar to bags or multisets in other languages. |
| 182 | |
| 183 | Elements are counted from an *iterable* or initialized from another |
Benjamin Peterson | 25c95f1 | 2009-05-08 20:42:26 +0000 | [diff] [blame] | 184 | *mapping* (or counter): |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 185 | |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 186 | >>> c = Counter() # a new, empty counter |
| 187 | >>> c = Counter('gallahad') # a new counter from an iterable |
| 188 | >>> c = Counter({'red': 4, 'blue': 2}) # a new counter from a mapping |
| 189 | >>> c = Counter(cats=4, dogs=8) # a new counter from keyword args |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 190 | |
Raymond Hettinger | 670eaec | 2009-01-21 23:14:07 +0000 | [diff] [blame] | 191 | Counter objects have a dictionary interface except that they return a zero |
Benjamin Peterson | 25c95f1 | 2009-05-08 20:42:26 +0000 | [diff] [blame] | 192 | count for missing items instead of raising a :exc:`KeyError`: |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 193 | |
Raymond Hettinger | 94adc8e | 2009-01-22 05:27:37 +0000 | [diff] [blame] | 194 | >>> c = Counter(['eggs', 'ham']) |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 195 | >>> c['bacon'] # count of a missing element is zero |
| 196 | 0 |
| 197 | |
Raymond Hettinger | 94adc8e | 2009-01-22 05:27:37 +0000 | [diff] [blame] | 198 | Setting a count to zero does not remove an element from a counter. |
| 199 | Use ``del`` to remove it entirely: |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 200 | |
Raymond Hettinger | 94adc8e | 2009-01-22 05:27:37 +0000 | [diff] [blame] | 201 | >>> c['sausage'] = 0 # counter entry with a zero count |
| 202 | >>> del c['sausage'] # del actually removes the entry |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 203 | |
Benjamin Peterson | d45bf58 | 2009-03-02 21:44:54 +0000 | [diff] [blame] | 204 | .. versionadded:: 3.1 |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 205 | |
| 206 | |
Ezio Melotti | 0be8b1c | 2010-04-04 06:53:44 +0000 | [diff] [blame] | 207 | Counter objects support three methods beyond those available for all |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 208 | dictionaries: |
| 209 | |
| 210 | .. method:: elements() |
| 211 | |
Raymond Hettinger | 670eaec | 2009-01-21 23:14:07 +0000 | [diff] [blame] | 212 | Return an iterator over elements repeating each as many times as its |
| 213 | count. Elements are returned in arbitrary order. If an element's count |
| 214 | is less than one, :meth:`elements` will ignore it. |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 215 | |
Raymond Hettinger | 0bae662 | 2009-01-20 13:00:59 +0000 | [diff] [blame] | 216 | >>> c = Counter(a=4, b=2, c=0, d=-2) |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 217 | >>> list(c.elements()) |
| 218 | ['a', 'a', 'a', 'a', 'b', 'b'] |
| 219 | |
| 220 | .. method:: most_common([n]) |
| 221 | |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 222 | Return a list of the *n* most common elements and their counts from the |
Raymond Hettinger | d04fa31 | 2009-02-04 19:45:13 +0000 | [diff] [blame] | 223 | most common to the least. If *n* is not specified, :func:`most_common` |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 224 | returns *all* elements in the counter. Elements with equal counts are |
Benjamin Peterson | 25c95f1 | 2009-05-08 20:42:26 +0000 | [diff] [blame] | 225 | ordered arbitrarily: |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 226 | |
| 227 | >>> Counter('abracadabra').most_common(3) |
| 228 | [('a', 5), ('r', 2), ('b', 2)] |
| 229 | |
Raymond Hettinger | 9c01e44 | 2010-04-03 10:32:58 +0000 | [diff] [blame] | 230 | .. method:: subtract([iterable-or-mapping]) |
| 231 | |
| 232 | Elements are subtracted from an *iterable* or from another *mapping* |
| 233 | (or counter). Like :meth:`dict.update` but subtracts counts instead |
| 234 | of replacing them. Both inputs and outputs may be zero or negative. |
| 235 | |
| 236 | >>> c = Counter(a=4, b=2, c=0, d=-2) |
| 237 | >>> d = Counter(a=1, b=2, c=3, d=4) |
| 238 | >>> c.subtract(d) |
| 239 | Counter({'a': 3, 'b': 0, 'c': -3, 'd': -6}) |
| 240 | |
Ezio Melotti | 0be8b1c | 2010-04-04 06:53:44 +0000 | [diff] [blame] | 241 | .. versionadded:: 3.2 |
| 242 | |
Raymond Hettinger | 670eaec | 2009-01-21 23:14:07 +0000 | [diff] [blame] | 243 | The usual dictionary methods are available for :class:`Counter` objects |
| 244 | except for two which work differently for counters. |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 245 | |
| 246 | .. method:: fromkeys(iterable) |
| 247 | |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 248 | This class method is not implemented for :class:`Counter` objects. |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 249 | |
| 250 | .. method:: update([iterable-or-mapping]) |
| 251 | |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 252 | Elements are counted from an *iterable* or added-in from another |
| 253 | *mapping* (or counter). Like :meth:`dict.update` but adds counts |
| 254 | instead of replacing them. Also, the *iterable* is expected to be a |
| 255 | sequence of elements, not a sequence of ``(key, value)`` pairs. |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 256 | |
| 257 | Common patterns for working with :class:`Counter` objects:: |
| 258 | |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 259 | sum(c.values()) # total of all counts |
| 260 | c.clear() # reset all counts |
| 261 | list(c) # list unique elements |
| 262 | set(c) # convert to a set |
| 263 | dict(c) # convert to a regular dictionary |
| 264 | c.items() # convert to a list of (elem, cnt) pairs |
| 265 | Counter(dict(list_of_pairs)) # convert from a list of (elem, cnt) pairs |
| 266 | c.most_common()[:-n:-1] # n least common elements |
| 267 | c += Counter() # remove zero and negative counts |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 268 | |
Raymond Hettinger | 72a95cc | 2009-02-25 22:51:40 +0000 | [diff] [blame] | 269 | Several mathematical operations are provided for combining :class:`Counter` |
| 270 | objects to produce multisets (counters that have counts greater than zero). |
| 271 | Addition and subtraction combine counters by adding or subtracting the counts |
| 272 | of corresponding elements. Intersection and union return the minimum and |
| 273 | maximum of corresponding counts. Each operation can accept inputs with signed |
| 274 | counts, but the output will exclude results with counts of zero or less. |
Raymond Hettinger | 4d2073a | 2009-01-20 03:41:22 +0000 | [diff] [blame] | 275 | |
Raymond Hettinger | e0d1b9f | 2009-01-21 20:36:27 +0000 | [diff] [blame] | 276 | >>> c = Counter(a=3, b=1) |
| 277 | >>> d = Counter(a=1, b=2) |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 278 | >>> c + d # add two counters together: c[x] + d[x] |
Raymond Hettinger | 4d2073a | 2009-01-20 03:41:22 +0000 | [diff] [blame] | 279 | Counter({'a': 4, 'b': 3}) |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 280 | >>> c - d # subtract (keeping only positive counts) |
Raymond Hettinger | 4d2073a | 2009-01-20 03:41:22 +0000 | [diff] [blame] | 281 | Counter({'a': 2}) |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 282 | >>> c & d # intersection: min(c[x], d[x]) |
Raymond Hettinger | 4d2073a | 2009-01-20 03:41:22 +0000 | [diff] [blame] | 283 | Counter({'a': 1, 'b': 1}) |
Raymond Hettinger | 73662a5 | 2009-01-27 02:38:22 +0000 | [diff] [blame] | 284 | >>> c | d # union: max(c[x], d[x]) |
Raymond Hettinger | 4d2073a | 2009-01-20 03:41:22 +0000 | [diff] [blame] | 285 | Counter({'a': 3, 'b': 2}) |
| 286 | |
Raymond Hettinger | 22f1885 | 2010-04-12 21:45:14 +0000 | [diff] [blame] | 287 | .. note:: |
| 288 | |
| 289 | Counters were primarily designed to work with positive integers to represent |
| 290 | running counts; however, care was taken to not unnecessarily preclude use |
| 291 | cases needing other types or negative values. To help with those use cases, |
| 292 | this section documents the minimum range and type restrictions. |
| 293 | |
| 294 | * The :class:`Counter` class itself is a dictionary subclass with no |
| 295 | restrictions on its keys and values. The values are intended to be numbers |
| 296 | representing counts, but you *could* store anything in the value field. |
| 297 | |
| 298 | * The :meth:`most_common` method requires only that the values be orderable. |
| 299 | |
| 300 | * For in-place operations such as ``c[key] += 1``, the value type need only |
| 301 | support addition and subtraction. So fractions, floats, and decimals would |
| 302 | work and negative values are supported. The same is also true for |
| 303 | :meth:`update` and :meth:`subtract` which allow negative and zero values |
| 304 | for both inputs and outputs. |
| 305 | |
| 306 | * The multiset methods are designed only for use cases with positive values. |
| 307 | The inputs may be negative or zero, but only outputs with positive values |
| 308 | are created. There are no type restrictions, but the value type needs to |
| 309 | support support addition, subtraction, and comparison. |
| 310 | |
| 311 | * The :meth:`elements` method requires integer counts. It ignores zero and |
| 312 | negative counts. |
| 313 | |
Raymond Hettinger | b14043c | 2009-01-20 23:44:31 +0000 | [diff] [blame] | 314 | .. seealso:: |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 315 | |
Raymond Hettinger | 94adc8e | 2009-01-22 05:27:37 +0000 | [diff] [blame] | 316 | * `Counter class <http://code.activestate.com/recipes/576611/>`_ |
| 317 | adapted for Python 2.5 and an early `Bag recipe |
| 318 | <http://code.activestate.com/recipes/259174/>`_ for Python 2.4. |
| 319 | |
Raymond Hettinger | b14043c | 2009-01-20 23:44:31 +0000 | [diff] [blame] | 320 | * `Bag class <http://www.gnu.org/software/smalltalk/manual-base/html_node/Bag.html>`_ |
| 321 | in Smalltalk. |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 322 | |
Raymond Hettinger | b14043c | 2009-01-20 23:44:31 +0000 | [diff] [blame] | 323 | * Wikipedia entry for `Multisets <http://en.wikipedia.org/wiki/Multiset>`_\. |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 324 | |
Raymond Hettinger | b14043c | 2009-01-20 23:44:31 +0000 | [diff] [blame] | 325 | * `C++ multisets <http://www.demo2s.com/Tutorial/Cpp/0380__set-multiset/Catalog0380__set-multiset.htm>`_ |
Raymond Hettinger | 94adc8e | 2009-01-22 05:27:37 +0000 | [diff] [blame] | 326 | tutorial with examples. |
Raymond Hettinger | b14043c | 2009-01-20 23:44:31 +0000 | [diff] [blame] | 327 | |
Raymond Hettinger | 94adc8e | 2009-01-22 05:27:37 +0000 | [diff] [blame] | 328 | * For mathematical operations on multisets and their use cases, see |
Raymond Hettinger | b14043c | 2009-01-20 23:44:31 +0000 | [diff] [blame] | 329 | *Knuth, Donald. The Art of Computer Programming Volume II, |
| 330 | Section 4.6.3, Exercise 19*\. |
| 331 | |
Raymond Hettinger | 670eaec | 2009-01-21 23:14:07 +0000 | [diff] [blame] | 332 | * To enumerate all distinct multisets of a given size over a given set of |
Raymond Hettinger | d07d939 | 2009-01-27 04:20:44 +0000 | [diff] [blame] | 333 | elements, see :func:`itertools.combinations_with_replacement`. |
Raymond Hettinger | b14043c | 2009-01-20 23:44:31 +0000 | [diff] [blame] | 334 | |
Raymond Hettinger | 94adc8e | 2009-01-22 05:27:37 +0000 | [diff] [blame] | 335 | map(Counter, combinations_with_replacement('ABC', 2)) --> AA AB AC BB BC CC |
Raymond Hettinger | b8baf63 | 2009-01-14 02:20:07 +0000 | [diff] [blame] | 336 | |
| 337 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 338 | :class:`deque` objects |
| 339 | ---------------------- |
| 340 | |
Georg Brandl | c2a4f4f | 2009-04-10 09:03:43 +0000 | [diff] [blame] | 341 | .. class:: deque([iterable, [maxlen]]) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 342 | |
| 343 | Returns a new deque object initialized left-to-right (using :meth:`append`) with |
| 344 | data from *iterable*. If *iterable* is not specified, the new deque is empty. |
| 345 | |
| 346 | Deques are a generalization of stacks and queues (the name is pronounced "deck" |
| 347 | and is short for "double-ended queue"). Deques support thread-safe, memory |
| 348 | efficient appends and pops from either side of the deque with approximately the |
| 349 | same O(1) performance in either direction. |
| 350 | |
| 351 | Though :class:`list` objects support similar operations, they are optimized for |
| 352 | fast fixed-length operations and incur O(n) memory movement costs for |
| 353 | ``pop(0)`` and ``insert(0, v)`` operations which change both the size and |
| 354 | position of the underlying data representation. |
| 355 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 356 | |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 357 | If *maxlen* is not specified or is *None*, deques may grow to an |
| 358 | arbitrary length. Otherwise, the deque is bounded to the specified maximum |
| 359 | length. Once a bounded length deque is full, when new items are added, a |
| 360 | corresponding number of items are discarded from the opposite end. Bounded |
| 361 | length deques provide functionality similar to the ``tail`` filter in |
| 362 | Unix. They are also useful for tracking transactions and other pools of data |
| 363 | where only the most recent activity is of interest. |
| 364 | |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 365 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 366 | Deque objects support the following methods: |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 367 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 368 | .. method:: append(x) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 369 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 370 | Add *x* to the right side of the deque. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 371 | |
| 372 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 373 | .. method:: appendleft(x) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 374 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 375 | Add *x* to the left side of the deque. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 376 | |
| 377 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 378 | .. method:: clear() |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 379 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 380 | Remove all elements from the deque leaving it with length 0. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 381 | |
| 382 | |
Raymond Hettinger | 44459de | 2010-04-03 23:20:46 +0000 | [diff] [blame] | 383 | .. method:: count(x) |
| 384 | |
| 385 | Count the number of deque elements equal to *x*. |
| 386 | |
| 387 | .. versionadded:: 3.2 |
| 388 | |
Georg Brandl | 67b21b7 | 2010-08-17 15:07:14 +0000 | [diff] [blame] | 389 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 390 | .. method:: extend(iterable) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 391 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 392 | Extend the right side of the deque by appending elements from the iterable |
| 393 | argument. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 394 | |
| 395 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 396 | .. method:: extendleft(iterable) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 397 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 398 | Extend the left side of the deque by appending elements from *iterable*. |
| 399 | Note, the series of left appends results in reversing the order of |
| 400 | elements in the iterable argument. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 401 | |
| 402 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 403 | .. method:: pop() |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 404 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 405 | Remove and return an element from the right side of the deque. If no |
| 406 | elements are present, raises an :exc:`IndexError`. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 407 | |
| 408 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 409 | .. method:: popleft() |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 410 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 411 | Remove and return an element from the left side of the deque. If no |
| 412 | elements are present, raises an :exc:`IndexError`. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 413 | |
| 414 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 415 | .. method:: remove(value) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 416 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 417 | Removed the first occurrence of *value*. If not found, raises a |
| 418 | :exc:`ValueError`. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 419 | |
Georg Brandl | 67b21b7 | 2010-08-17 15:07:14 +0000 | [diff] [blame] | 420 | |
Raymond Hettinger | e5fdedb | 2009-12-10 00:47:21 +0000 | [diff] [blame] | 421 | .. method:: reverse() |
| 422 | |
| 423 | Reverse the elements of the deque in-place and then return ``None``. |
| 424 | |
| 425 | .. versionadded:: 3.2 |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 426 | |
Georg Brandl | 67b21b7 | 2010-08-17 15:07:14 +0000 | [diff] [blame] | 427 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 428 | .. method:: rotate(n) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 429 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 430 | Rotate the deque *n* steps to the right. If *n* is negative, rotate to |
| 431 | the left. Rotating one step to the right is equivalent to: |
| 432 | ``d.appendleft(d.pop())``. |
| 433 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 434 | |
Raymond Hettinger | 5bb0f0e | 2009-03-10 12:56:32 +0000 | [diff] [blame] | 435 | Deque objects also provide one read-only attribute: |
| 436 | |
| 437 | .. attribute:: maxlen |
| 438 | |
| 439 | Maximum size of a deque or *None* if unbounded. |
| 440 | |
Raymond Hettinger | 150fb9c | 2009-03-10 22:48:06 +0000 | [diff] [blame] | 441 | .. versionadded:: 3.1 |
Raymond Hettinger | 5bb0f0e | 2009-03-10 12:56:32 +0000 | [diff] [blame] | 442 | |
| 443 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 444 | In addition to the above, deques support iteration, pickling, ``len(d)``, |
| 445 | ``reversed(d)``, ``copy.copy(d)``, ``copy.deepcopy(d)``, membership testing with |
Benjamin Peterson | 206e307 | 2008-10-19 14:07:49 +0000 | [diff] [blame] | 446 | the :keyword:`in` operator, and subscript references such as ``d[-1]``. Indexed |
| 447 | access is O(1) at both ends but slows to O(n) in the middle. For fast random |
| 448 | access, use lists instead. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 449 | |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 450 | Example: |
| 451 | |
| 452 | .. doctest:: |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 453 | |
| 454 | >>> from collections import deque |
| 455 | >>> d = deque('ghi') # make a new deque with three items |
| 456 | >>> for elem in d: # iterate over the deque's elements |
Neal Norwitz | 752abd0 | 2008-05-13 04:55:24 +0000 | [diff] [blame] | 457 | ... print(elem.upper()) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 458 | G |
| 459 | H |
| 460 | I |
| 461 | |
| 462 | >>> d.append('j') # add a new entry to the right side |
| 463 | >>> d.appendleft('f') # add a new entry to the left side |
| 464 | >>> d # show the representation of the deque |
| 465 | deque(['f', 'g', 'h', 'i', 'j']) |
| 466 | |
| 467 | >>> d.pop() # return and remove the rightmost item |
| 468 | 'j' |
| 469 | >>> d.popleft() # return and remove the leftmost item |
| 470 | 'f' |
| 471 | >>> list(d) # list the contents of the deque |
| 472 | ['g', 'h', 'i'] |
| 473 | >>> d[0] # peek at leftmost item |
| 474 | 'g' |
| 475 | >>> d[-1] # peek at rightmost item |
| 476 | 'i' |
| 477 | |
| 478 | >>> list(reversed(d)) # list the contents of a deque in reverse |
| 479 | ['i', 'h', 'g'] |
| 480 | >>> 'h' in d # search the deque |
| 481 | True |
| 482 | >>> d.extend('jkl') # add multiple elements at once |
| 483 | >>> d |
| 484 | deque(['g', 'h', 'i', 'j', 'k', 'l']) |
| 485 | >>> d.rotate(1) # right rotation |
| 486 | >>> d |
| 487 | deque(['l', 'g', 'h', 'i', 'j', 'k']) |
| 488 | >>> d.rotate(-1) # left rotation |
| 489 | >>> d |
| 490 | deque(['g', 'h', 'i', 'j', 'k', 'l']) |
| 491 | |
| 492 | >>> deque(reversed(d)) # make a new deque in reverse order |
| 493 | deque(['l', 'k', 'j', 'i', 'h', 'g']) |
| 494 | >>> d.clear() # empty the deque |
| 495 | >>> d.pop() # cannot pop from an empty deque |
| 496 | Traceback (most recent call last): |
| 497 | File "<pyshell#6>", line 1, in -toplevel- |
| 498 | d.pop() |
| 499 | IndexError: pop from an empty deque |
| 500 | |
| 501 | >>> d.extendleft('abc') # extendleft() reverses the input order |
| 502 | >>> d |
| 503 | deque(['c', 'b', 'a']) |
| 504 | |
| 505 | |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 506 | :class:`deque` Recipes |
| 507 | ^^^^^^^^^^^^^^^^^^^^^^ |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 508 | |
| 509 | This section shows various approaches to working with deques. |
| 510 | |
Raymond Hettinger | d2ee64d | 2009-03-31 22:52:48 +0000 | [diff] [blame] | 511 | Bounded length deques provide functionality similar to the ``tail`` filter |
| 512 | in Unix:: |
| 513 | |
| 514 | def tail(filename, n=10): |
| 515 | 'Return the last n lines of a file' |
Éric Araujo | a3dd56b | 2011-03-11 17:42:48 +0100 | [diff] [blame] | 516 | with open(filename) as f: |
| 517 | return deque(f, n) |
Raymond Hettinger | d2ee64d | 2009-03-31 22:52:48 +0000 | [diff] [blame] | 518 | |
| 519 | Another approach to using deques is to maintain a sequence of recently |
| 520 | added elements by appending to the right and popping to the left:: |
| 521 | |
| 522 | def moving_average(iterable, n=3): |
| 523 | # moving_average([40, 30, 50, 46, 39, 44]) --> 40.0 42.0 45.0 43.0 |
| 524 | # http://en.wikipedia.org/wiki/Moving_average |
| 525 | it = iter(iterable) |
Raymond Hettinger | d40285a | 2009-05-22 01:11:26 +0000 | [diff] [blame] | 526 | d = deque(itertools.islice(it, n-1)) |
| 527 | d.appendleft(0) |
Raymond Hettinger | d2ee64d | 2009-03-31 22:52:48 +0000 | [diff] [blame] | 528 | s = sum(d) |
Raymond Hettinger | d2ee64d | 2009-03-31 22:52:48 +0000 | [diff] [blame] | 529 | for elem in it: |
| 530 | s += elem - d.popleft() |
| 531 | d.append(elem) |
| 532 | yield s / n |
| 533 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 534 | The :meth:`rotate` method provides a way to implement :class:`deque` slicing and |
Ezio Melotti | 0639d5a | 2009-12-19 23:26:38 +0000 | [diff] [blame] | 535 | deletion. For example, a pure Python implementation of ``del d[n]`` relies on |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 536 | the :meth:`rotate` method to position elements to be popped:: |
| 537 | |
| 538 | def delete_nth(d, n): |
| 539 | d.rotate(-n) |
| 540 | d.popleft() |
| 541 | d.rotate(n) |
| 542 | |
| 543 | To implement :class:`deque` slicing, use a similar approach applying |
| 544 | :meth:`rotate` to bring a target element to the left side of the deque. Remove |
| 545 | old entries with :meth:`popleft`, add new entries with :meth:`extend`, and then |
| 546 | reverse the rotation. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 547 | With minor variations on that approach, it is easy to implement Forth style |
| 548 | stack manipulations such as ``dup``, ``drop``, ``swap``, ``over``, ``pick``, |
| 549 | ``rot``, and ``roll``. |
| 550 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 551 | |
| 552 | :class:`defaultdict` objects |
| 553 | ---------------------------- |
| 554 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 555 | .. class:: defaultdict([default_factory[, ...]]) |
| 556 | |
| 557 | Returns a new dictionary-like object. :class:`defaultdict` is a subclass of the |
Georg Brandl | 22b3431 | 2009-07-26 14:54:51 +0000 | [diff] [blame] | 558 | built-in :class:`dict` class. It overrides one method and adds one writable |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 559 | instance variable. The remaining functionality is the same as for the |
| 560 | :class:`dict` class and is not documented here. |
| 561 | |
| 562 | The first argument provides the initial value for the :attr:`default_factory` |
| 563 | attribute; it defaults to ``None``. All remaining arguments are treated the same |
| 564 | as if they were passed to the :class:`dict` constructor, including keyword |
| 565 | arguments. |
| 566 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 567 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 568 | :class:`defaultdict` objects support the following method in addition to the |
| 569 | standard :class:`dict` operations: |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 570 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 571 | .. method:: __missing__(key) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 572 | |
Benjamin Peterson | 5478b47 | 2008-09-17 22:25:09 +0000 | [diff] [blame] | 573 | If the :attr:`default_factory` attribute is ``None``, this raises a |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 574 | :exc:`KeyError` exception with the *key* as argument. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 575 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 576 | If :attr:`default_factory` is not ``None``, it is called without arguments |
| 577 | to provide a default value for the given *key*, this value is inserted in |
| 578 | the dictionary for the *key*, and returned. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 579 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 580 | If calling :attr:`default_factory` raises an exception this exception is |
| 581 | propagated unchanged. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 582 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 583 | This method is called by the :meth:`__getitem__` method of the |
| 584 | :class:`dict` class when the requested key is not found; whatever it |
| 585 | returns or raises is then returned or raised by :meth:`__getitem__`. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 586 | |
| 587 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 588 | :class:`defaultdict` objects support the following instance variable: |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 589 | |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 590 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 591 | .. attribute:: default_factory |
Benjamin Peterson | e41251e | 2008-04-25 01:59:09 +0000 | [diff] [blame] | 592 | |
| 593 | This attribute is used by the :meth:`__missing__` method; it is |
| 594 | initialized from the first argument to the constructor, if present, or to |
| 595 | ``None``, if absent. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 596 | |
| 597 | |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 598 | :class:`defaultdict` Examples |
| 599 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 600 | |
| 601 | Using :class:`list` as the :attr:`default_factory`, it is easy to group a |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 602 | sequence of key-value pairs into a dictionary of lists: |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 603 | |
| 604 | >>> s = [('yellow', 1), ('blue', 2), ('yellow', 3), ('blue', 4), ('red', 1)] |
| 605 | >>> d = defaultdict(list) |
| 606 | >>> for k, v in s: |
| 607 | ... d[k].append(v) |
| 608 | ... |
Ezio Melotti | c53a894 | 2009-09-12 01:52:05 +0000 | [diff] [blame] | 609 | >>> list(d.items()) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 610 | [('blue', [2, 4]), ('red', [1]), ('yellow', [1, 3])] |
| 611 | |
| 612 | When each key is encountered for the first time, it is not already in the |
| 613 | mapping; so an entry is automatically created using the :attr:`default_factory` |
| 614 | function which returns an empty :class:`list`. The :meth:`list.append` |
| 615 | operation then attaches the value to the new list. When keys are encountered |
| 616 | again, the look-up proceeds normally (returning the list for that key) and the |
| 617 | :meth:`list.append` operation adds another value to the list. This technique is |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 618 | simpler and faster than an equivalent technique using :meth:`dict.setdefault`: |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 619 | |
| 620 | >>> d = {} |
| 621 | >>> for k, v in s: |
| 622 | ... d.setdefault(k, []).append(v) |
| 623 | ... |
Ezio Melotti | c53a894 | 2009-09-12 01:52:05 +0000 | [diff] [blame] | 624 | >>> list(d.items()) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 625 | [('blue', [2, 4]), ('red', [1]), ('yellow', [1, 3])] |
| 626 | |
| 627 | Setting the :attr:`default_factory` to :class:`int` makes the |
| 628 | :class:`defaultdict` useful for counting (like a bag or multiset in other |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 629 | languages): |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 630 | |
| 631 | >>> s = 'mississippi' |
| 632 | >>> d = defaultdict(int) |
| 633 | >>> for k in s: |
| 634 | ... d[k] += 1 |
| 635 | ... |
Ezio Melotti | c53a894 | 2009-09-12 01:52:05 +0000 | [diff] [blame] | 636 | >>> list(d.items()) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 637 | [('i', 4), ('p', 2), ('s', 4), ('m', 1)] |
| 638 | |
| 639 | When a letter is first encountered, it is missing from the mapping, so the |
| 640 | :attr:`default_factory` function calls :func:`int` to supply a default count of |
| 641 | zero. The increment operation then builds up the count for each letter. |
| 642 | |
| 643 | The function :func:`int` which always returns zero is just a special case of |
| 644 | constant functions. A faster and more flexible way to create constant functions |
| 645 | is to use a lambda function which can supply any constant value (not just |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 646 | zero): |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 647 | |
| 648 | >>> def constant_factory(value): |
| 649 | ... return lambda: value |
| 650 | >>> d = defaultdict(constant_factory('<missing>')) |
| 651 | >>> d.update(name='John', action='ran') |
| 652 | >>> '%(name)s %(action)s to %(object)s' % d |
| 653 | 'John ran to <missing>' |
| 654 | |
| 655 | Setting the :attr:`default_factory` to :class:`set` makes the |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 656 | :class:`defaultdict` useful for building a dictionary of sets: |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 657 | |
| 658 | >>> s = [('red', 1), ('blue', 2), ('red', 3), ('blue', 4), ('red', 1), ('blue', 4)] |
| 659 | >>> d = defaultdict(set) |
| 660 | >>> for k, v in s: |
| 661 | ... d[k].add(v) |
| 662 | ... |
Ezio Melotti | c53a894 | 2009-09-12 01:52:05 +0000 | [diff] [blame] | 663 | >>> list(d.items()) |
Raymond Hettinger | ba7b560 | 2011-03-22 22:57:49 -0700 | [diff] [blame] | 664 | [('blue', {2, 4}), ('red', {1, 3})] |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 665 | |
| 666 | |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 667 | :func:`namedtuple` Factory Function for Tuples with Named Fields |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 668 | ---------------------------------------------------------------- |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 669 | |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 670 | Named tuples assign meaning to each position in a tuple and allow for more readable, |
| 671 | self-documenting code. They can be used wherever regular tuples are used, and |
| 672 | they add the ability to access fields by name instead of position index. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 673 | |
Georg Brandl | c2a4f4f | 2009-04-10 09:03:43 +0000 | [diff] [blame] | 674 | .. function:: namedtuple(typename, field_names, verbose=False, rename=False) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 675 | |
| 676 | Returns a new tuple subclass named *typename*. The new subclass is used to |
Christian Heimes | c3f30c4 | 2008-02-22 16:37:40 +0000 | [diff] [blame] | 677 | create tuple-like objects that have fields accessible by attribute lookup as |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 678 | well as being indexable and iterable. Instances of the subclass also have a |
Benjamin Peterson | 4469d0c | 2008-11-30 22:46:23 +0000 | [diff] [blame] | 679 | helpful docstring (with typename and field_names) and a helpful :meth:`__repr__` |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 680 | method which lists the tuple contents in a ``name=value`` format. |
| 681 | |
Benjamin Peterson | 4469d0c | 2008-11-30 22:46:23 +0000 | [diff] [blame] | 682 | The *field_names* are a single string with each fieldname separated by whitespace |
| 683 | and/or commas, for example ``'x y'`` or ``'x, y'``. Alternatively, *field_names* |
Christian Heimes | 25bb783 | 2008-01-11 16:17:00 +0000 | [diff] [blame] | 684 | can be a sequence of strings such as ``['x', 'y']``. |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 685 | |
| 686 | Any valid Python identifier may be used for a fieldname except for names |
Christian Heimes | 0449f63 | 2007-12-15 01:27:15 +0000 | [diff] [blame] | 687 | starting with an underscore. Valid identifiers consist of letters, digits, |
| 688 | and underscores but do not start with a digit or underscore and cannot be |
Georg Brandl | f694518 | 2008-02-01 11:56:49 +0000 | [diff] [blame] | 689 | a :mod:`keyword` such as *class*, *for*, *return*, *global*, *pass*, |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 690 | or *raise*. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 691 | |
Benjamin Peterson | a86f2c0 | 2009-02-10 02:41:10 +0000 | [diff] [blame] | 692 | If *rename* is true, invalid fieldnames are automatically replaced |
| 693 | with positional names. For example, ``['abc', 'def', 'ghi', 'abc']`` is |
Raymond Hettinger | 85737b8 | 2009-04-02 22:37:59 +0000 | [diff] [blame] | 694 | converted to ``['abc', '_1', 'ghi', '_3']``, eliminating the keyword |
Benjamin Peterson | a86f2c0 | 2009-02-10 02:41:10 +0000 | [diff] [blame] | 695 | ``def`` and the duplicate fieldname ``abc``. |
| 696 | |
Raymond Hettinger | 2ebea41 | 2011-03-23 12:52:23 -0700 | [diff] [blame] | 697 | If *verbose* is true, the class definition is printed after it is |
| 698 | built. This option is outdated; instead, it is simpler to print the |
| 699 | :attr:`_source` attribute. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 700 | |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 701 | Named tuple instances do not have per-instance dictionaries, so they are |
Thomas Wouters | 8ce81f7 | 2007-09-20 18:22:40 +0000 | [diff] [blame] | 702 | lightweight and require no more memory than regular tuples. |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 703 | |
Raymond Hettinger | b62ad24 | 2009-03-02 22:16:43 +0000 | [diff] [blame] | 704 | .. versionchanged:: 3.1 |
Georg Brandl | 67b21b7 | 2010-08-17 15:07:14 +0000 | [diff] [blame] | 705 | Added support for *rename*. |
Benjamin Peterson | a86f2c0 | 2009-02-10 02:41:10 +0000 | [diff] [blame] | 706 | |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 707 | |
| 708 | .. doctest:: |
| 709 | :options: +NORMALIZE_WHITESPACE |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 710 | |
Raymond Hettinger | 0ef956f | 2010-11-21 23:23:29 +0000 | [diff] [blame] | 711 | >>> # Basic example |
Raymond Hettinger | 15aded8 | 2011-03-15 17:25:51 -0700 | [diff] [blame] | 712 | >>> Point = namedtuple('Point', ['x', 'y']) |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 713 | >>> p = Point(11, y=22) # instantiate with positional or keyword arguments |
Christian Heimes | 99170a5 | 2007-12-19 02:07:34 +0000 | [diff] [blame] | 714 | >>> p[0] + p[1] # indexable like the plain tuple (11, 22) |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 715 | 33 |
| 716 | >>> x, y = p # unpack like a regular tuple |
| 717 | >>> x, y |
| 718 | (11, 22) |
Christian Heimes | c3f30c4 | 2008-02-22 16:37:40 +0000 | [diff] [blame] | 719 | >>> p.x + p.y # fields also accessible by name |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 720 | 33 |
| 721 | >>> p # readable __repr__ with a name=value style |
| 722 | Point(x=11, y=22) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 723 | |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 724 | Named tuples are especially useful for assigning field names to result tuples returned |
| 725 | by the :mod:`csv` or :mod:`sqlite3` modules:: |
| 726 | |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 727 | EmployeeRecord = namedtuple('EmployeeRecord', 'name, age, title, department, paygrade') |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 728 | |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 729 | import csv |
Christian Heimes | faf2f63 | 2008-01-06 16:59:19 +0000 | [diff] [blame] | 730 | for emp in map(EmployeeRecord._make, csv.reader(open("employees.csv", "rb"))): |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 731 | print(emp.name, emp.title) |
| 732 | |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 733 | import sqlite3 |
| 734 | conn = sqlite3.connect('/companydata') |
| 735 | cursor = conn.cursor() |
| 736 | cursor.execute('SELECT name, age, title, department, paygrade FROM employees') |
Christian Heimes | faf2f63 | 2008-01-06 16:59:19 +0000 | [diff] [blame] | 737 | for emp in map(EmployeeRecord._make, cursor.fetchall()): |
Christian Heimes | 0041223 | 2008-01-10 16:02:19 +0000 | [diff] [blame] | 738 | print(emp.name, emp.title) |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 739 | |
Christian Heimes | 99170a5 | 2007-12-19 02:07:34 +0000 | [diff] [blame] | 740 | In addition to the methods inherited from tuples, named tuples support |
Raymond Hettinger | 2ebea41 | 2011-03-23 12:52:23 -0700 | [diff] [blame] | 741 | three additional methods and two attributes. To prevent conflicts with |
Christian Heimes | 2380ac7 | 2008-01-09 00:17:24 +0000 | [diff] [blame] | 742 | field names, the method and attribute names start with an underscore. |
Christian Heimes | 99170a5 | 2007-12-19 02:07:34 +0000 | [diff] [blame] | 743 | |
Benjamin Peterson | 0b9fb80 | 2010-07-18 14:23:36 +0000 | [diff] [blame] | 744 | .. classmethod:: somenamedtuple._make(iterable) |
Christian Heimes | 99170a5 | 2007-12-19 02:07:34 +0000 | [diff] [blame] | 745 | |
Christian Heimes | faf2f63 | 2008-01-06 16:59:19 +0000 | [diff] [blame] | 746 | Class method that makes a new instance from an existing sequence or iterable. |
Christian Heimes | 99170a5 | 2007-12-19 02:07:34 +0000 | [diff] [blame] | 747 | |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 748 | .. doctest:: |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 749 | |
Christian Heimes | faf2f63 | 2008-01-06 16:59:19 +0000 | [diff] [blame] | 750 | >>> t = [11, 22] |
| 751 | >>> Point._make(t) |
| 752 | Point(x=11, y=22) |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 753 | |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 754 | .. method:: somenamedtuple._asdict() |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 755 | |
Raymond Hettinger | a4f52b1 | 2009-03-02 22:28:31 +0000 | [diff] [blame] | 756 | Return a new :class:`OrderedDict` which maps field names to their corresponding |
| 757 | values:: |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 758 | |
Christian Heimes | 0449f63 | 2007-12-15 01:27:15 +0000 | [diff] [blame] | 759 | >>> p._asdict() |
Raymond Hettinger | a4f52b1 | 2009-03-02 22:28:31 +0000 | [diff] [blame] | 760 | OrderedDict([('x', 11), ('y', 22)]) |
| 761 | |
Raymond Hettinger | a88e4da | 2009-03-03 05:12:27 +0000 | [diff] [blame] | 762 | .. versionchanged:: 3.1 |
Raymond Hettinger | a4f52b1 | 2009-03-02 22:28:31 +0000 | [diff] [blame] | 763 | Returns an :class:`OrderedDict` instead of a regular :class:`dict`. |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 764 | |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 765 | .. method:: somenamedtuple._replace(kwargs) |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 766 | |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 767 | Return a new instance of the named tuple replacing specified fields with new |
| 768 | values: |
Thomas Wouters | 8ce81f7 | 2007-09-20 18:22:40 +0000 | [diff] [blame] | 769 | |
| 770 | :: |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 771 | |
| 772 | >>> p = Point(x=11, y=22) |
Christian Heimes | 0449f63 | 2007-12-15 01:27:15 +0000 | [diff] [blame] | 773 | >>> p._replace(x=33) |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 774 | Point(x=33, y=22) |
| 775 | |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 776 | >>> for partnum, record in inventory.items(): |
Christian Heimes | 454f37b | 2008-01-10 00:10:02 +0000 | [diff] [blame] | 777 | ... inventory[partnum] = record._replace(price=newprices[partnum], timestamp=time.now()) |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 778 | |
Raymond Hettinger | 2ebea41 | 2011-03-23 12:52:23 -0700 | [diff] [blame] | 779 | .. attribute:: somenamedtuple._source |
| 780 | |
| 781 | A string with the pure Python source code used to create the named |
| 782 | tuple class. The source makes the named tuple self-documenting. |
| 783 | It can be printed, executed using :func:`exec`, or saved to a file |
| 784 | and imported. |
| 785 | |
| 786 | .. versionadded:: 3.3 |
| 787 | |
Christian Heimes | 790c823 | 2008-01-07 21:14:23 +0000 | [diff] [blame] | 788 | .. attribute:: somenamedtuple._fields |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 789 | |
Christian Heimes | 2380ac7 | 2008-01-09 00:17:24 +0000 | [diff] [blame] | 790 | Tuple of strings listing the field names. Useful for introspection |
Georg Brandl | 9afde1c | 2007-11-01 20:32:30 +0000 | [diff] [blame] | 791 | and for creating new named tuple types from existing named tuples. |
Thomas Wouters | 8ce81f7 | 2007-09-20 18:22:40 +0000 | [diff] [blame] | 792 | |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 793 | .. doctest:: |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 794 | |
Christian Heimes | 0449f63 | 2007-12-15 01:27:15 +0000 | [diff] [blame] | 795 | >>> p._fields # view the field names |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 796 | ('x', 'y') |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 797 | |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 798 | >>> Color = namedtuple('Color', 'red green blue') |
Christian Heimes | 0449f63 | 2007-12-15 01:27:15 +0000 | [diff] [blame] | 799 | >>> Pixel = namedtuple('Pixel', Point._fields + Color._fields) |
Thomas Wouters | 1b7f891 | 2007-09-19 03:06:30 +0000 | [diff] [blame] | 800 | >>> Pixel(11, 22, 128, 255, 0) |
Christian Heimes | 454f37b | 2008-01-10 00:10:02 +0000 | [diff] [blame] | 801 | Pixel(x=11, y=22, red=128, green=255, blue=0) |
Georg Brandl | 116aa62 | 2007-08-15 14:28:22 +0000 | [diff] [blame] | 802 | |
Christian Heimes | 0449f63 | 2007-12-15 01:27:15 +0000 | [diff] [blame] | 803 | To retrieve a field whose name is stored in a string, use the :func:`getattr` |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 804 | function: |
Christian Heimes | 0449f63 | 2007-12-15 01:27:15 +0000 | [diff] [blame] | 805 | |
| 806 | >>> getattr(p, 'x') |
| 807 | 11 |
| 808 | |
Raymond Hettinger | 651453a | 2009-02-11 00:20:02 +0000 | [diff] [blame] | 809 | To convert a dictionary to a named tuple, use the double-star-operator |
| 810 | (as described in :ref:`tut-unpacking-arguments`): |
Christian Heimes | 99170a5 | 2007-12-19 02:07:34 +0000 | [diff] [blame] | 811 | |
| 812 | >>> d = {'x': 11, 'y': 22} |
| 813 | >>> Point(**d) |
| 814 | Point(x=11, y=22) |
| 815 | |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 816 | Since a named tuple is a regular Python class, it is easy to add or change |
Christian Heimes | 043d6f6 | 2008-01-07 17:19:16 +0000 | [diff] [blame] | 817 | functionality with a subclass. Here is how to add a calculated field and |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 818 | a fixed-width print format: |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 819 | |
Christian Heimes | 043d6f6 | 2008-01-07 17:19:16 +0000 | [diff] [blame] | 820 | >>> class Point(namedtuple('Point', 'x y')): |
Raymond Hettinger | 15aded8 | 2011-03-15 17:25:51 -0700 | [diff] [blame] | 821 | __slots__ = () |
| 822 | @property |
| 823 | def hypot(self): |
| 824 | return (self.x ** 2 + self.y ** 2) ** 0.5 |
| 825 | def __str__(self): |
| 826 | return 'Point: x=%6.3f y=%6.3f hypot=%6.3f' % (self.x, self.y, self.hypot) |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 827 | |
Georg Brandl | 0df7979 | 2008-10-04 18:33:26 +0000 | [diff] [blame] | 828 | >>> for p in Point(3, 4), Point(14, 5/7): |
Raymond Hettinger | 15aded8 | 2011-03-15 17:25:51 -0700 | [diff] [blame] | 829 | print(p) |
Christian Heimes | 25bb783 | 2008-01-11 16:17:00 +0000 | [diff] [blame] | 830 | Point: x= 3.000 y= 4.000 hypot= 5.000 |
| 831 | Point: x=14.000 y= 0.714 hypot=14.018 |
Christian Heimes | 043d6f6 | 2008-01-07 17:19:16 +0000 | [diff] [blame] | 832 | |
Georg Brandl | af5c238 | 2009-12-28 08:02:38 +0000 | [diff] [blame] | 833 | The subclass shown above sets ``__slots__`` to an empty tuple. This helps |
Christian Heimes | 679db4a | 2008-01-18 09:56:22 +0000 | [diff] [blame] | 834 | keep memory requirements low by preventing the creation of instance dictionaries. |
| 835 | |
Christian Heimes | 2380ac7 | 2008-01-09 00:17:24 +0000 | [diff] [blame] | 836 | Subclassing is not useful for adding new, stored fields. Instead, simply |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 837 | create a new named tuple type from the :attr:`_fields` attribute: |
Christian Heimes | 2380ac7 | 2008-01-09 00:17:24 +0000 | [diff] [blame] | 838 | |
Christian Heimes | 25bb783 | 2008-01-11 16:17:00 +0000 | [diff] [blame] | 839 | >>> Point3D = namedtuple('Point3D', Point._fields + ('z',)) |
Christian Heimes | 2380ac7 | 2008-01-09 00:17:24 +0000 | [diff] [blame] | 840 | |
| 841 | Default values can be implemented by using :meth:`_replace` to |
Christian Heimes | fe337bf | 2008-03-23 21:54:12 +0000 | [diff] [blame] | 842 | customize a prototype instance: |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 843 | |
| 844 | >>> Account = namedtuple('Account', 'owner balance transaction_count') |
Christian Heimes | 587c2bf | 2008-01-19 16:21:02 +0000 | [diff] [blame] | 845 | >>> default_account = Account('<owner name>', 0.0, 0) |
| 846 | >>> johns_account = default_account._replace(owner='John') |
Raymond Hettinger | b2d0945 | 2011-03-22 22:36:21 -0700 | [diff] [blame] | 847 | >>> janes_account = default_account._replace(owner='Jane') |
Guido van Rossum | 3d392eb | 2007-11-16 00:35:22 +0000 | [diff] [blame] | 848 | |
Christian Heimes | e4ca815 | 2008-05-08 17:18:53 +0000 | [diff] [blame] | 849 | Enumerated constants can be implemented with named tuples, but it is simpler |
| 850 | and more efficient to use a simple class declaration: |
| 851 | |
| 852 | >>> Status = namedtuple('Status', 'open pending closed')._make(range(3)) |
| 853 | >>> Status.open, Status.pending, Status.closed |
| 854 | (0, 1, 2) |
| 855 | >>> class Status: |
Raymond Hettinger | 15aded8 | 2011-03-15 17:25:51 -0700 | [diff] [blame] | 856 | open, pending, closed = range(3) |
Christian Heimes | e4ca815 | 2008-05-08 17:18:53 +0000 | [diff] [blame] | 857 | |
Raymond Hettinger | 651453a | 2009-02-11 00:20:02 +0000 | [diff] [blame] | 858 | .. seealso:: |
Thomas Wouters | 47b49bf | 2007-08-30 22:15:33 +0000 | [diff] [blame] | 859 | |
Raymond Hettinger | 6c94e6f | 2011-03-31 15:46:06 -0700 | [diff] [blame] | 860 | * `Named tuple recipe <http://code.activestate.com/recipes/500261/>`_ |
| 861 | adapted for Python 2.4. |
| 862 | |
| 863 | * `Recipe for named tuple abstract base class with a metaclass mix-in |
| 864 | <http://code.activestate.com/recipes/577629-namedtupleabc-abstract-base-class-mix-in-for-named/>`_ |
| 865 | by Jan Kaliszewski. Besides providing an :term:`abstract base class` for |
| 866 | named tuples, it also supports an alternate :term:`metaclass`-based |
| 867 | constructor that is convenient for use cases where named tuples are being |
| 868 | subclassed. |
Raymond Hettinger | e4c96ad | 2008-02-06 01:23:58 +0000 | [diff] [blame] | 869 | |
| 870 | |
Raymond Hettinger | 2d32f63 | 2009-03-02 21:24:57 +0000 | [diff] [blame] | 871 | :class:`OrderedDict` objects |
| 872 | ---------------------------- |
| 873 | |
| 874 | Ordered dictionaries are just like regular dictionaries but they remember the |
| 875 | order that items were inserted. When iterating over an ordered dictionary, |
| 876 | the items are returned in the order their keys were first added. |
| 877 | |
| 878 | .. class:: OrderedDict([items]) |
| 879 | |
| 880 | Return an instance of a dict subclass, supporting the usual :class:`dict` |
| 881 | methods. An *OrderedDict* is a dict that remembers the order that keys |
| 882 | were first inserted. If a new entry overwrites an existing entry, the |
| 883 | original insertion position is left unchanged. Deleting an entry and |
| 884 | reinserting it will move it to the end. |
| 885 | |
Benjamin Peterson | d45bf58 | 2009-03-02 21:44:54 +0000 | [diff] [blame] | 886 | .. versionadded:: 3.1 |
Raymond Hettinger | 2d32f63 | 2009-03-02 21:24:57 +0000 | [diff] [blame] | 887 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 888 | .. method:: popitem(last=True) |
Raymond Hettinger | dc879f0 | 2009-03-19 20:30:56 +0000 | [diff] [blame] | 889 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 890 | The :meth:`popitem` method for ordered dictionaries returns and removes a |
| 891 | (key, value) pair. The pairs are returned in LIFO order if *last* is true |
| 892 | or FIFO order if false. |
Raymond Hettinger | 2d32f63 | 2009-03-02 21:24:57 +0000 | [diff] [blame] | 893 | |
Raymond Hettinger | f45abc9 | 2010-09-06 21:26:09 +0000 | [diff] [blame] | 894 | .. method:: move_to_end(key, last=True) |
| 895 | |
| 896 | Move an existing *key* to either end of an ordered dictionary. The item |
| 897 | is moved to the right end if *last* is true (the default) or to the |
| 898 | beginning if *last* is false. Raises :exc:`KeyError` if the *key* does |
| 899 | not exist:: |
| 900 | |
| 901 | >>> d = OrderedDict.fromkeys('abcde') |
| 902 | >>> d.move_to_end('b') |
| 903 | >>> ''.join(d.keys) |
| 904 | 'acdeb' |
Éric Araujo | 1cb25aa | 2010-11-06 07:03:07 +0000 | [diff] [blame] | 905 | >>> d.move_to_end('b', last=False) |
Raymond Hettinger | f45abc9 | 2010-09-06 21:26:09 +0000 | [diff] [blame] | 906 | >>> ''.join(d.keys) |
| 907 | 'bacde' |
| 908 | |
| 909 | .. versionadded:: 3.2 |
| 910 | |
Raymond Hettinger | e909150 | 2009-05-19 17:40:07 +0000 | [diff] [blame] | 911 | In addition to the usual mapping methods, ordered dictionaries also support |
| 912 | reverse iteration using :func:`reversed`. |
| 913 | |
Raymond Hettinger | 2d32f63 | 2009-03-02 21:24:57 +0000 | [diff] [blame] | 914 | Equality tests between :class:`OrderedDict` objects are order-sensitive |
| 915 | and are implemented as ``list(od1.items())==list(od2.items())``. |
| 916 | Equality tests between :class:`OrderedDict` objects and other |
| 917 | :class:`Mapping` objects are order-insensitive like regular dictionaries. |
| 918 | This allows :class:`OrderedDict` objects to be substituted anywhere a |
| 919 | regular dictionary is used. |
| 920 | |
Raymond Hettinger | 3618078 | 2009-04-09 22:34:23 +0000 | [diff] [blame] | 921 | The :class:`OrderedDict` constructor and :meth:`update` method both accept |
| 922 | keyword arguments, but their order is lost because Python's function call |
| 923 | semantics pass-in keyword arguments using a regular unordered dictionary. |
| 924 | |
Raymond Hettinger | dc879f0 | 2009-03-19 20:30:56 +0000 | [diff] [blame] | 925 | .. seealso:: |
| 926 | |
| 927 | `Equivalent OrderedDict recipe <http://code.activestate.com/recipes/576693/>`_ |
| 928 | that runs on Python 2.4 or later. |
| 929 | |
Raymond Hettinger | 7bba683 | 2011-04-15 17:43:19 -0700 | [diff] [blame] | 930 | :class:`OrderedDict` Examples and Recipes |
| 931 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 932 | |
Raymond Hettinger | 0e31201 | 2009-11-10 18:35:46 +0000 | [diff] [blame] | 933 | Since an ordered dictionary remembers its insertion order, it can be used |
| 934 | in conjuction with sorting to make a sorted dictionary:: |
| 935 | |
| 936 | >>> # regular unsorted dictionary |
| 937 | >>> d = {'banana': 3, 'apple':4, 'pear': 1, 'orange': 2} |
| 938 | |
| 939 | >>> # dictionary sorted by key |
| 940 | >>> OrderedDict(sorted(d.items(), key=lambda t: t[0])) |
| 941 | OrderedDict([('apple', 4), ('banana', 3), ('orange', 2), ('pear', 1)]) |
| 942 | |
| 943 | >>> # dictionary sorted by value |
| 944 | >>> OrderedDict(sorted(d.items(), key=lambda t: t[1])) |
| 945 | OrderedDict([('pear', 1), ('orange', 2), ('banana', 3), ('apple', 4)]) |
| 946 | |
| 947 | >>> # dictionary sorted by length of the key string |
| 948 | >>> OrderedDict(sorted(d.items(), key=lambda t: len(t[0]))) |
| 949 | OrderedDict([('pear', 1), ('apple', 4), ('orange', 2), ('banana', 3)]) |
| 950 | |
| 951 | The new sorted dictionaries maintain their sort order when entries |
| 952 | are deleted. But when new keys are added, the keys are appended |
| 953 | to the end and the sort is not maintained. |
| 954 | |
Raymond Hettinger | 4821ef8 | 2010-07-31 10:14:41 +0000 | [diff] [blame] | 955 | It is also straight-forward to create an ordered dictionary variant |
| 956 | that the remembers the order the keys were *last* inserted. |
| 957 | If a new entry overwrites an existing entry, the |
| 958 | original insertion position is changed and moved to the end:: |
| 959 | |
| 960 | class LastUpdatedOrderedDict(OrderedDict): |
Georg Brandl | 77570e2 | 2010-12-18 16:21:58 +0000 | [diff] [blame] | 961 | 'Store items in the order the keys were last added' |
Raymond Hettinger | 7bba683 | 2011-04-15 17:43:19 -0700 | [diff] [blame] | 962 | |
Raymond Hettinger | 4821ef8 | 2010-07-31 10:14:41 +0000 | [diff] [blame] | 963 | def __setitem__(self, key, value): |
| 964 | if key in self: |
| 965 | del self[key] |
| 966 | OrderedDict.__setitem__(self, key, value) |
| 967 | |
Raymond Hettinger | 7bba683 | 2011-04-15 17:43:19 -0700 | [diff] [blame] | 968 | An ordered dictionary can combined with the :class:`Counter` class |
| 969 | so that the counter remembers the order elements are first encountered:: |
| 970 | |
| 971 | class OrderedCounter(Counter, OrderedDict): |
| 972 | 'Counter that remembers the order elements are first encountered' |
| 973 | |
| 974 | def __init__(self, iterable=None, **kwds): |
| 975 | OrderedDict.__init__(self) |
| 976 | Counter.__init__(self, iterable, **kwds) |
| 977 | |
| 978 | def __repr__(self): |
| 979 | return '%s(%r)' % (self.__class__.__name__, OrderedDict(self)) |
| 980 | |
| 981 | def __reduce__(self): |
| 982 | return self.__class__, (OrderedDict(self),) |
| 983 | |
Raymond Hettinger | e4c96ad | 2008-02-06 01:23:58 +0000 | [diff] [blame] | 984 | |
| 985 | :class:`UserDict` objects |
Mark Summerfield | 8f2d006 | 2008-02-06 13:30:44 +0000 | [diff] [blame] | 986 | ------------------------- |
Raymond Hettinger | e4c96ad | 2008-02-06 01:23:58 +0000 | [diff] [blame] | 987 | |
Georg Brandl | 48310cd | 2009-01-03 21:18:54 +0000 | [diff] [blame] | 988 | The class, :class:`UserDict` acts as a wrapper around dictionary objects. |
| 989 | The need for this class has been partially supplanted by the ability to |
Raymond Hettinger | e4c96ad | 2008-02-06 01:23:58 +0000 | [diff] [blame] | 990 | subclass directly from :class:`dict`; however, this class can be easier |
| 991 | to work with because the underlying dictionary is accessible as an |
| 992 | attribute. |
| 993 | |
| 994 | .. class:: UserDict([initialdata]) |
| 995 | |
| 996 | Class that simulates a dictionary. The instance's contents are kept in a |
| 997 | regular dictionary, which is accessible via the :attr:`data` attribute of |
| 998 | :class:`UserDict` instances. If *initialdata* is provided, :attr:`data` is |
| 999 | initialized with its contents; note that a reference to *initialdata* will not |
| 1000 | be kept, allowing it be used for other purposes. |
| 1001 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 1002 | In addition to supporting the methods and operations of mappings, |
| 1003 | :class:`UserDict` instances provide the following attribute: |
Raymond Hettinger | e4c96ad | 2008-02-06 01:23:58 +0000 | [diff] [blame] | 1004 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 1005 | .. attribute:: data |
Raymond Hettinger | e4c96ad | 2008-02-06 01:23:58 +0000 | [diff] [blame] | 1006 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 1007 | A real dictionary used to store the contents of the :class:`UserDict` |
| 1008 | class. |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 1009 | |
| 1010 | |
| 1011 | |
| 1012 | :class:`UserList` objects |
| 1013 | ------------------------- |
| 1014 | |
| 1015 | This class acts as a wrapper around list objects. It is a useful base class |
Georg Brandl | 48310cd | 2009-01-03 21:18:54 +0000 | [diff] [blame] | 1016 | for your own list-like classes which can inherit from them and override |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 1017 | existing methods or add new ones. In this way, one can add new behaviors to |
| 1018 | lists. |
| 1019 | |
Georg Brandl | 48310cd | 2009-01-03 21:18:54 +0000 | [diff] [blame] | 1020 | The need for this class has been partially supplanted by the ability to |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 1021 | subclass directly from :class:`list`; however, this class can be easier |
| 1022 | to work with because the underlying list is accessible as an attribute. |
| 1023 | |
| 1024 | .. class:: UserList([list]) |
| 1025 | |
| 1026 | Class that simulates a list. The instance's contents are kept in a regular |
| 1027 | list, which is accessible via the :attr:`data` attribute of :class:`UserList` |
| 1028 | instances. The instance's contents are initially set to a copy of *list*, |
| 1029 | defaulting to the empty list ``[]``. *list* can be any iterable, for |
| 1030 | example a real Python list or a :class:`UserList` object. |
| 1031 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 1032 | In addition to supporting the methods and operations of mutable sequences, |
| 1033 | :class:`UserList` instances provide the following attribute: |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 1034 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 1035 | .. attribute:: data |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 1036 | |
Benjamin Peterson | d319ad5 | 2010-07-18 14:27:02 +0000 | [diff] [blame] | 1037 | A real :class:`list` object used to store the contents of the |
| 1038 | :class:`UserList` class. |
Raymond Hettinger | 53dbe39 | 2008-02-12 20:03:09 +0000 | [diff] [blame] | 1039 | |
| 1040 | **Subclassing requirements:** Subclasses of :class:`UserList` are expect to |
| 1041 | offer a constructor which can be called with either no arguments or one |
| 1042 | argument. List operations which return a new sequence attempt to create an |
| 1043 | instance of the actual implementation class. To do so, it assumes that the |
| 1044 | constructor can be called with a single parameter, which is a sequence object |
| 1045 | used as a data source. |
| 1046 | |
| 1047 | If a derived class does not wish to comply with this requirement, all of the |
| 1048 | special methods supported by this class will need to be overridden; please |
| 1049 | consult the sources for information about the methods which need to be provided |
| 1050 | in that case. |
Raymond Hettinger | b3a65f8 | 2008-02-21 22:11:37 +0000 | [diff] [blame] | 1051 | |
| 1052 | :class:`UserString` objects |
Christian Heimes | c3f30c4 | 2008-02-22 16:37:40 +0000 | [diff] [blame] | 1053 | --------------------------- |
Raymond Hettinger | b3a65f8 | 2008-02-21 22:11:37 +0000 | [diff] [blame] | 1054 | |
Georg Brandl | 48310cd | 2009-01-03 21:18:54 +0000 | [diff] [blame] | 1055 | The class, :class:`UserString` acts as a wrapper around string objects. |
| 1056 | The need for this class has been partially supplanted by the ability to |
Raymond Hettinger | b3a65f8 | 2008-02-21 22:11:37 +0000 | [diff] [blame] | 1057 | subclass directly from :class:`str`; however, this class can be easier |
| 1058 | to work with because the underlying string is accessible as an |
| 1059 | attribute. |
| 1060 | |
| 1061 | .. class:: UserString([sequence]) |
| 1062 | |
| 1063 | Class that simulates a string or a Unicode string object. The instance's |
Georg Brandl | 48310cd | 2009-01-03 21:18:54 +0000 | [diff] [blame] | 1064 | content is kept in a regular string object, which is accessible via the |
| 1065 | :attr:`data` attribute of :class:`UserString` instances. The instance's |
Raymond Hettinger | b3a65f8 | 2008-02-21 22:11:37 +0000 | [diff] [blame] | 1066 | contents are initially set to a copy of *sequence*. The *sequence* can |
| 1067 | be an instance of :class:`bytes`, :class:`str`, :class:`UserString` (or a |
| 1068 | subclass) or an arbitrary sequence which can be converted into a string using |
| 1069 | the built-in :func:`str` function. |