1
# Copyright (C) 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
"""A simple least-recently-used (LRU) cache."""
19
from collections import deque
23
class LRUCache(object):
24
"""A class which manages a cache of entries, removing unused ones."""
26
def __init__(self, max_cache=100, after_cleanup_size=None):
27
self._max_cache = max_cache
28
if after_cleanup_size is None:
29
self._after_cleanup_size = self._max_cache
31
self._after_cleanup_size = min(after_cleanup_size, self._max_cache)
33
self._compact_queue_length = 4*self._max_cache
37
self._queue = deque() # Track when things are accessed
38
self._refcount = {} # number of entries in self._queue for each key
40
def __contains__(self, key):
41
return key in self._cache
43
def __getitem__(self, key):
44
val = self._cache[key]
45
self._record_access(key)
49
return len(self._cache)
51
def add(self, key, value, cleanup=None):
52
"""Add a new value to the cache.
54
Also, if the entry is ever removed from the queue, call cleanup.
55
Passing it the key and value being removed.
57
:param key: The key to store it under
58
:param value: The object to store
59
:param cleanup: None or a function taking (key, value) to indicate
60
'value' sohuld be cleaned up.
62
if key in self._cache:
64
self._cache[key] = value
65
self._cleanup[key] = cleanup
66
self._record_access(key)
68
if len(self._cache) > self._max_cache:
72
def get(self, key, default=None):
73
if key in self._cache:
78
"""Get the list of keys currently cached.
80
Note that values returned here may not be available by the time you
81
request them later. This is simply meant as a peak into the current
84
:return: An unordered list of keys that are currently cached.
86
return self._cache.keys()
89
"""Clear the cache until it shrinks to the requested size.
91
This does not completely wipe the cache, just makes sure it is under
92
the after_cleanup_size.
94
# Make sure the cache is shrunk to the correct size
95
while len(self._cache) > self._after_cleanup_size:
98
def __setitem__(self, key, value):
99
"""Add a value to the cache, there will be no cleanup function."""
100
self.add(key, value, cleanup=None)
102
def _record_access(self, key):
103
"""Record that key was accessed."""
104
self._queue.append(key)
105
# Can't use setdefault because you can't += 1 the result
106
self._refcount[key] = self._refcount.get(key, 0) + 1
108
# If our access queue is too large, clean it up too
109
if len(self._queue) > self._compact_queue_length:
110
self._compact_queue()
112
def _compact_queue(self):
113
"""Compact the queue, leaving things in sorted last appended order."""
115
for item in self._queue:
116
if self._refcount[item] == 1:
117
new_queue.append(item)
119
self._refcount[item] -= 1
120
self._queue = new_queue
121
# All entries should be of the same size. There should be one entry in
122
# queue for each entry in cache, and all refcounts should == 1
123
if not (len(self._queue) == len(self._cache) ==
124
len(self._refcount) == sum(self._refcount.itervalues())):
125
raise AssertionError()
127
def _remove(self, key):
128
"""Remove an entry, making sure to maintain the invariants."""
129
cleanup = self._cleanup.pop(key)
130
val = self._cache.pop(key)
131
if cleanup is not None:
135
def _remove_lru(self):
136
"""Remove one entry from the lru, and handle consequences.
138
If there are no more references to the lru, then this entry should be
139
removed from the cache.
141
key = self._queue.popleft()
142
self._refcount[key] -= 1
143
if not self._refcount[key]:
144
del self._refcount[key]
148
"""Clear out all of the cache."""
149
# Clean up in LRU order
154
class LRUSizeCache(LRUCache):
155
"""An LRUCache that removes things based on the size of the values.
157
This differs in that it doesn't care how many actual items there are,
158
it just restricts the cache to be cleaned up after so much data is stored.
160
The values that are added must support len(value).
163
def __init__(self, max_size=1024*1024, after_cleanup_size=None,
165
"""Create a new LRUSizeCache.
167
:param max_size: The max number of bytes to store before we start
168
clearing out entries.
169
:param after_cleanup_size: After cleaning up, shrink everything to this
171
:param compute_size: A function to compute the size of the values. We
172
use a function here, so that you can pass 'len' if you are just
173
using simple strings, or a more complex function if you are using
174
something like a list of strings, or even a custom object.
175
The function should take the form "compute_size(value) => integer".
176
If not supplied, it defaults to 'len()'
178
# This approximates that texts are > 0.5k in size. It only really
179
# effects when we clean up the queue, so we don't want it to be too
181
LRUCache.__init__(self, max_cache=int(max_size/512))
182
self._max_size = max_size
183
if after_cleanup_size is None:
184
self._after_cleanup_size = self._max_size
186
self._after_cleanup_size = min(after_cleanup_size, self._max_size)
189
self._compute_size = compute_size
190
if compute_size is None:
191
self._compute_size = len
193
def add(self, key, value, cleanup=None):
194
"""Add a new value to the cache.
196
Also, if the entry is ever removed from the queue, call cleanup.
197
Passing it the key and value being removed.
199
:param key: The key to store it under
200
:param value: The object to store
201
:param cleanup: None or a function taking (key, value) to indicate
202
'value' sohuld be cleaned up.
204
if key in self._cache:
206
value_len = self._compute_size(value)
207
if value_len >= self._after_cleanup_size:
209
self._value_size += value_len
210
self._cache[key] = value
211
self._cleanup[key] = cleanup
212
self._record_access(key)
214
if self._value_size > self._max_size:
219
"""Clear the cache until it shrinks to the requested size.
221
This does not completely wipe the cache, just makes sure it is under
222
the after_cleanup_size.
224
# Make sure the cache is shrunk to the correct size
225
while self._value_size > self._after_cleanup_size:
228
def _remove(self, key):
229
"""Remove an entry, making sure to maintain the invariants."""
230
val = LRUCache._remove(self, key)
231
self._value_size -= self._compute_size(val)
added
removed
bzrlib/lru_cache.py
