fix LRUCache

shadowsocks/lru_cache.py

@@ -22,11 +22,22 @@
 import logging
 import time
 
+if __name__ == '__main__':
+    import os, sys, inspect
+    file_path = os.path.dirname(os.path.realpath(inspect.getfile(inspect.currentframe())))
+    sys.path.insert(0, os.path.join(file_path, '../'))
+
+try:
+    from collections import OrderedDict
+    print("loaded collections.OrderedDict")
+except:
+    from shadowsocks.ordereddict import OrderedDict
+
 # this LRUCache is optimized for concurrency, not QPS
 # n: concurrency, keys stored in the cache
 # m: visits not timed out, proportional to QPS * timeout
-# get & set is O(log(n)), not O(n). thus we can support very large n
-# sweep is O((n - m)*log(n)) or O(1024*log(n)) at most,
+# get & set is O(1), not O(n). thus we can support very large n
+# sweep is O((n - m)) or O(1024) at most,
 # no metter how large the cache or timeout value is
 
 SWEEP_MAX_ITEMS = 1024
@@ -38,39 +49,30 @@ def __init__(self, timeout=60, close_callback=None, *args, **kwargs):
         self.timeout = timeout
         self.close_callback = close_callback
         self._store = {}
-        self._time_to_keys = collections.OrderedDict()
-        self._keys_to_last_time = {}
-        self._visit_id = 0
+        self._keys_to_last_time = OrderedDict()
         self.update(dict(*args, **kwargs))  # use the free update to set keys
 
     def __getitem__(self, key):
-        # O(log(n))
+        # O(1)
         t = time.time()
-        last_t, vid = self._keys_to_last_time[key]
-        self._keys_to_last_time[key] = (t, vid)
-        if last_t != t:
-            del self._time_to_keys[(last_t, vid)]
-            self._time_to_keys[(t, vid)] = key
+        last_t = self._keys_to_last_time[key]
+        del self._keys_to_last_time[key]
+        self._keys_to_last_time[key] = t
         return self._store[key]
 
     def __setitem__(self, key, value):
-        # O(log(n))
+        # O(1)
         t = time.time()
         if key in self._keys_to_last_time:
-            last_t, vid = self._keys_to_last_time[key]
-            del self._time_to_keys[(last_t, vid)]
-        vid = self._visit_id
-        self._visit_id += 1
-        self._keys_to_last_time[key] = (t, vid)
+            del self._keys_to_last_time[key]
+        self._keys_to_last_time[key] = t
         self._store[key] = value
-        self._time_to_keys[(t, vid)] = key
 
     def __delitem__(self, key):
-        # O(log(n))
-        last_t, vid = self._keys_to_last_time[key]
+        # O(1)
+        last_t = self._keys_to_last_time[key]
         del self._store[key]
         del self._keys_to_last_time[key]
-        del self._time_to_keys[(last_t, vid)]
 
     def __iter__(self):
         return iter(self._store)
@@ -83,18 +85,17 @@ def sweep(self):
         now = time.time()
         c = 0
         while c < SWEEP_MAX_ITEMS:
-            if len(self._time_to_keys) == 0:
+            if len(self._keys_to_last_time) == 0:
                 break
-            last_t, vid = iter(self._time_to_keys).next()
+            key = iter(self._keys_to_last_time).next()
+            last_t = self._keys_to_last_time[key]
             if now - last_t <= self.timeout:
                 break
-            key = self._time_to_keys[(last_t, vid)]
             value = self._store[key]
             if self.close_callback is not None:
                 self.close_callback(value)
             del self._store[key]
             del self._keys_to_last_time[key]
-            del self._time_to_keys[(last_t, vid)]
             c += 1
         if c:
             logging.debug('%d keys swept' % c)

shadowsocks/ordereddict.py

@@ -0,0 +1,214 @@
+import collections
+
+################################################################################
+### OrderedDict
+################################################################################
+
+class OrderedDict(dict):
+    'Dictionary that remembers insertion order'
+    # An inherited dict maps keys to values.
+    # The inherited dict provides __getitem__, __len__, __contains__, and get.
+    # The remaining methods are order-aware.
+    # Big-O running times for all methods are the same as regular dictionaries.
+
+    # The internal self.__map dict maps keys to links in a doubly linked list.
+    # The circular doubly linked list starts and ends with a sentinel element.
+    # The sentinel element never gets deleted (this simplifies the algorithm).
+    # Each link is stored as a list of length three:  [PREV, NEXT, KEY].
+
+    def __init__(*args, **kwds):
+        '''Initialize an ordered dictionary.  The signature is the same as
+        regular dictionaries, but keyword arguments are not recommended because
+        their insertion order is arbitrary.
+
+        '''
+        if not args:
+            raise TypeError("descriptor '__init__' of 'OrderedDict' object "
+                            "needs an argument")
+        self = args[0]
+        args = args[1:]
+        if len(args) > 1:
+            raise TypeError('expected at most 1 arguments, got %d' % len(args))
+        try:
+            self.__root
+        except AttributeError:
+            self.__root = root = []                     # sentinel node
+            root[:] = [root, root, None]
+            self.__map = {}
+        self.__update(*args, **kwds)
+
+    def __setitem__(self, key, value, dict_setitem=dict.__setitem__):
+        'od.__setitem__(i, y) <==> od[i]=y'
+        # Setting a new item creates a new link at the end of the linked list,
+        # and the inherited dictionary is updated with the new key/value pair.
+        if key not in self:
+            root = self.__root
+            last = root[0]
+            last[1] = root[0] = self.__map[key] = [last, root, key]
+        return dict_setitem(self, key, value)
+
+    def __delitem__(self, key, dict_delitem=dict.__delitem__):
+        'od.__delitem__(y) <==> del od[y]'
+        # Deleting an existing item uses self.__map to find the link which gets
+        # removed by updating the links in the predecessor and successor nodes.
+        dict_delitem(self, key)
+        link_prev, link_next, _ = self.__map.pop(key)
+        link_prev[1] = link_next                        # update link_prev[NEXT]
+        link_next[0] = link_prev                        # update link_next[PREV]
+
+    def __iter__(self):
+        'od.__iter__() <==> iter(od)'
+        # Traverse the linked list in order.
+        root = self.__root
+        curr = root[1]                                  # start at the first node
+        while curr is not root:
+            yield curr[2]                               # yield the curr[KEY]
+            curr = curr[1]                              # move to next node
+
+    def __reversed__(self):
+        'od.__reversed__() <==> reversed(od)'
+        # Traverse the linked list in reverse order.
+        root = self.__root
+        curr = root[0]                                  # start at the last node
+        while curr is not root:
+            yield curr[2]                               # yield the curr[KEY]
+            curr = curr[0]                              # move to previous node
+
+    def clear(self):
+        'od.clear() -> None.  Remove all items from od.'
+        root = self.__root
+        root[:] = [root, root, None]
+        self.__map.clear()
+        dict.clear(self)
+
+    # -- the following methods do not depend on the internal structure --
+
+    def keys(self):
+        'od.keys() -> list of keys in od'
+        return list(self)
+
+    def values(self):
+        'od.values() -> list of values in od'
+        return [self[key] for key in self]
+
+    def items(self):
+        'od.items() -> list of (key, value) pairs in od'
+        return [(key, self[key]) for key in self]
+
+    def iterkeys(self):
+        'od.iterkeys() -> an iterator over the keys in od'
+        return iter(self)
+
+    def itervalues(self):
+        'od.itervalues -> an iterator over the values in od'
+        for k in self:
+            yield self[k]
+
+    def iteritems(self):
+        'od.iteritems -> an iterator over the (key, value) pairs in od'
+        for k in self:
+            yield (k, self[k])
+
+    update = collections.MutableMapping.update
+
+    __update = update # let subclasses override update without breaking __init__
+
+    __marker = object()
+
+    def pop(self, key, default=__marker):
+        '''od.pop(k[,d]) -> v, remove specified key and return the corresponding
+        value.  If key is not found, d is returned if given, otherwise KeyError
+        is raised.
+
+        '''
+        if key in self:
+            result = self[key]
+            del self[key]
+            return result
+        if default is self.__marker:
+            raise KeyError(key)
+        return default
+
+    def setdefault(self, key, default=None):
+        'od.setdefault(k[,d]) -> od.get(k,d), also set od[k]=d if k not in od'
+        if key in self:
+            return self[key]
+        self[key] = default
+        return default
+
+    def popitem(self, last=True):
+        '''od.popitem() -> (k, v), return and remove a (key, value) pair.
+        Pairs are returned in LIFO order if last is true or FIFO order if false.
+
+        '''
+        if not self:
+            raise KeyError('dictionary is empty')
+        key = next(reversed(self) if last else iter(self))
+        value = self.pop(key)
+        return key, value
+
+    def __repr__(self, _repr_running={}):
+        'od.__repr__() <==> repr(od)'
+        call_key = id(self), _get_ident()
+        if call_key in _repr_running:
+            return '...'
+        _repr_running[call_key] = 1
+        try:
+            if not self:
+                return '%s()' % (self.__class__.__name__,)
+            return '%s(%r)' % (self.__class__.__name__, self.items())
+        finally:
+            del _repr_running[call_key]
+
+    def __reduce__(self):
+        'Return state information for pickling'
+        items = [[k, self[k]] for k in self]
+        inst_dict = vars(self).copy()
+        for k in vars(OrderedDict()):
+            inst_dict.pop(k, None)
+        if inst_dict:
+            return (self.__class__, (items,), inst_dict)
+        return self.__class__, (items,)
+
+    def copy(self):
+        'od.copy() -> a shallow copy of od'
+        return self.__class__(self)
+
+    @classmethod
+    def fromkeys(cls, iterable, value=None):
+        '''OD.fromkeys(S[, v]) -> New ordered dictionary with keys from S.
+        If not specified, the value defaults to None.
+
+        '''
+        self = cls()
+        for key in iterable:
+            self[key] = value
+        return self
+
+    def __eq__(self, other):
+        '''od.__eq__(y) <==> od==y.  Comparison to another OD is order-sensitive
+        while comparison to a regular mapping is order-insensitive.
+
+        '''
+        if isinstance(other, OrderedDict):
+            return dict.__eq__(self, other) and all(_imap(_eq, self, other))
+        return dict.__eq__(self, other)
+
+    def __ne__(self, other):
+        'od.__ne__(y) <==> od!=y'
+        return not self == other
+
+    # -- the following methods support python 3.x style dictionary views --
+
+    def viewkeys(self):
+        "od.viewkeys() -> a set-like object providing a view on od's keys"
+        return KeysView(self)
+
+    def viewvalues(self):
+        "od.viewvalues() -> an object providing a view on od's values"
+        return ValuesView(self)
+
+    def viewitems(self):
+        "od.viewitems() -> a set-like object providing a view on od's items"
+        return ItemsView(self)
+