Change LRUCache to be tree-based so we can delete subtrees.

synapse/push/push_rule_evaluator.py

@@ -309,14 +309,14 @@ def _flatten_dict(d, prefix=[], result={}):
     return result
 
 
-regex_cache = LruCache(5000)
+regex_cache = LruCache(5000, 1)
 
 
 def _compile_regex(regex_str):
-    r = regex_cache.get(regex_str, None)
+    r = regex_cache.get((regex_str,), None)
     if r:
         return r
 
     r = re.compile(regex_str, flags=re.IGNORECASE)
-    regex_cache[regex_str] = r
+    regex_cache[(regex_str,)] = r
     return r

synapse/util/caches/descriptors.py

@@ -38,7 +38,7 @@ class Cache(object):
 
     def __init__(self, name, max_entries=1000, keylen=1, lru=True):
         if lru:
-            self.cache = LruCache(max_size=max_entries)
+            self.cache = LruCache(max_size=max_entries, keylen=keylen)
             self.max_entries = None
         else:
             self.cache = OrderedDict()
@@ -99,6 +99,15 @@ class Cache(object):
         self.sequence += 1
         self.cache.pop(key, None)
 
+    def invalidate_many(self, key):
+        self.check_thread()
+        if not isinstance(key, tuple):
+            raise TypeError(
+                "The cache key must be a tuple not %r" % (type(key),)
+            )
+        self.sequence += 1
+        self.cache.del_multi(key)
+
     def invalidate_all(self):
         self.check_thread()
         self.sequence += 1

synapse/util/caches/dictionary_cache.py

@@ -32,7 +32,7 @@ class DictionaryCache(object):
     """
 
     def __init__(self, name, max_entries=1000):
-        self.cache = LruCache(max_size=max_entries)
+        self.cache = LruCache(max_size=max_entries, keylen=1)
 
         self.name = name
         self.sequence = 0
@@ -56,7 +56,7 @@ class DictionaryCache(object):
                 )
 
     def get(self, key, dict_keys=None):
-        entry = self.cache.get(key, self.sentinel)
+        entry = self.cache.get((key,), self.sentinel)
         if entry is not self.sentinel:
             cache_counter.inc_hits(self.name)
 
@@ -78,7 +78,7 @@ class DictionaryCache(object):
         # Increment the sequence number so that any SELECT statements that
         # raced with the INSERT don't update the cache (SYN-369)
         self.sequence += 1
-        self.cache.pop(key, None)
+        self.cache.pop((key,), None)
 
     def invalidate_all(self):
         self.check_thread()
@@ -96,8 +96,8 @@ class DictionaryCache(object):
                 self._update_or_insert(key, value)
 
     def _update_or_insert(self, key, value):
-        entry = self.cache.setdefault(key, DictionaryEntry(False, {}))
+        entry = self.cache.setdefault((key,), DictionaryEntry(False, {}))
         entry.value.update(value)
 
     def _insert(self, key, value):
-        self.cache[key] = DictionaryEntry(True, value)
+        self.cache[(key,)] = DictionaryEntry(True, value)

synapse/util/caches/lrucache.py

@@ -17,11 +17,23 @@
 from functools import wraps
 import threading
 
+from synapse.util.caches.treecache import TreeCache
+
+
+def enumerate_leaves(node, depth):
+    if depth == 0:
+        yield node
+    else:
+        for n in node.values():
+            for m in enumerate_leaves(n, depth - 1):
+                yield m
+
 
 class LruCache(object):
     """Least-recently-used cache."""
-    def __init__(self, max_size):
-        cache = {}
+    def __init__(self, max_size, keylen):
+        cache = TreeCache()
+        self.size = 0
         list_root = []
         list_root[:] = [list_root, list_root, None, None]
 
@@ -44,6 +56,7 @@ class LruCache(object):
             prev_node[NEXT] = node
             next_node[PREV] = node
             cache[key] = node
+            self.size += 1
 
         def move_node_to_front(node):
             prev_node = node[PREV]
@@ -62,7 +75,7 @@ class LruCache(object):
             next_node = node[NEXT]
             prev_node[NEXT] = next_node
             next_node[PREV] = prev_node
-            cache.pop(node[KEY], None)
+            self.size -= 1
 
         @synchronized
         def cache_get(key, default=None):
@@ -81,8 +94,10 @@ class LruCache(object):
                 node[VALUE] = value
             else:
                 add_node(key, value)
-                if len(cache) > max_size:
-                    delete_node(list_root[PREV])
+                if self.size > max_size:
+                    todelete = list_root[PREV]
+                    delete_node(todelete)
+                    cache.pop(todelete[KEY], None)
 
         @synchronized
         def cache_set_default(key, value):
@@ -91,8 +106,10 @@ class LruCache(object):
                 return node[VALUE]
             else:
                 add_node(key, value)
-                if len(cache) > max_size:
-                    delete_node(list_root[PREV])
+                if self.size > max_size:
+                    todelete = list_root[PREV]
+                    delete_node(todelete)
+                    cache.pop(todelete[KEY], None)
                 return value
 
         @synchronized
@@ -100,10 +117,19 @@ class LruCache(object):
             node = cache.get(key, None)
             if node:
                 delete_node(node)
+                cache.pop(node[KEY], None)
                 return node[VALUE]
             else:
                 return default
 
+        @synchronized
+        def cache_del_multi(key):
+            popped = cache.pop(key)
+            if popped is None:
+                return
+            for leaf in enumerate_leaves(popped, keylen - len(key)):
+                delete_node(leaf)
+
         @synchronized
         def cache_clear():
             list_root[NEXT] = list_root
@@ -112,7 +138,7 @@ class LruCache(object):
 
         @synchronized
         def cache_len():
-            return len(cache)
+            return self.size
 
         @synchronized
         def cache_contains(key):
@@ -123,6 +149,7 @@ class LruCache(object):
         self.set = cache_set
         self.setdefault = cache_set_default
         self.pop = cache_pop
+        self.del_multi = cache_del_multi
         self.len = cache_len
         self.contains = cache_contains
         self.clear = cache_clear

synapse/util/caches/treecache.py

@@ -0,0 +1,52 @@
+SENTINEL = object()
+
+
+class TreeCache(object):
+    def __init__(self):
+        self.root = {}
+
+    def __setitem__(self, key, value):
+        return self.set(key, value)
+
+    def set(self, key, value):
+        node = self.root
+        for k in key[:-1]:
+            node = node.setdefault(k, {})
+        node[key[-1]] = value
+
+    def get(self, key, default=None):
+        node = self.root
+        for k in key[:-1]:
+            node = node.get(k, None)
+            if node is None:
+                return default
+        return node.get(key[-1], default)
+
+    def clear(self):
+        self.root = {}
+
+    def pop(self, key, default=None):
+        nodes = []
+
+        node = self.root
+        for k in key[:-1]:
+            node = node.get(k, None)
+            nodes.append(node)  # don't add the root node
+            if node is None:
+                return default
+        popped = node.pop(key[-1], SENTINEL)
+        if popped is SENTINEL:
+            return default
+
+        node_and_keys = zip(nodes, key)
+        node_and_keys.reverse()
+        node_and_keys.append((self.root, None))
+
+        for i in range(len(node_and_keys) - 1):
+            n,k = node_and_keys[i]
+
+            if n:
+                break
+            node_and_keys[i+1][0].pop(k)
+
+        return popped

tests/storage/test__base.py

@@ -56,42 +56,42 @@ class CacheTestCase(unittest.TestCase):
     def test_eviction(self):
         cache = Cache("test", max_entries=2)
 
-        cache.prefill(1, "one")
-        cache.prefill(2, "two")
-        cache.prefill(3, "three")  # 1 will be evicted
+        cache.prefill((1,), "one")
+        cache.prefill((2,), "two")
+        cache.prefill((3,), "three")  # 1 will be evicted
 
         failed = False
         try:
-            cache.get(1)
+            cache.get((1,))
         except KeyError:
             failed = True
 
         self.assertTrue(failed)
 
-        cache.get(2)
-        cache.get(3)
+        cache.get((2,))
+        cache.get((3,))
 
     def test_eviction_lru(self):
         cache = Cache("test", max_entries=2, lru=True)
 
-        cache.prefill(1, "one")
-        cache.prefill(2, "two")
+        cache.prefill((1,), "one")
+        cache.prefill((2,), "two")
 
         # Now access 1 again, thus causing 2 to be least-recently used
-        cache.get(1)
+        cache.get((1,))
 
-        cache.prefill(3, "three")
+        cache.prefill((3,), "three")
 
         failed = False
         try:
-            cache.get(2)
+            cache.get((2,))
         except KeyError:
             failed = True
 
         self.assertTrue(failed)
 
-        cache.get(1)
-        cache.get(3)
+        cache.get((1,))
+        cache.get((3,))
 
 
 class CacheDecoratorTestCase(unittest.TestCase):

tests/util/test_lrucache.py

@@ -21,34 +21,34 @@ from synapse.util.caches.lrucache import LruCache
 class LruCacheTestCase(unittest.TestCase):
 
     def test_get_set(self):
-        cache = LruCache(1)
-        cache["key"] = "value"
-        self.assertEquals(cache.get("key"), "value")
-        self.assertEquals(cache["key"], "value")
+        cache = LruCache(1, 1)
+        cache[("key",)] = "value"
+        self.assertEquals(cache.get(("key",)), "value")
+        self.assertEquals(cache[("key",)], "value")
 
     def test_eviction(self):
-        cache = LruCache(2)
-        cache[1] = 1
-        cache[2] = 2
+        cache = LruCache(2, 1)
+        cache[(1,)] = 1
+        cache[(2,)] = 2
 
-        self.assertEquals(cache.get(1), 1)
-        self.assertEquals(cache.get(2), 2)
+        self.assertEquals(cache.get((1,)), 1)
+        self.assertEquals(cache.get((2,)), 2)
 
-        cache[3] = 3
+        cache[(3,)] = 3
 
-        self.assertEquals(cache.get(1), None)
-        self.assertEquals(cache.get(2), 2)
-        self.assertEquals(cache.get(3), 3)
+        self.assertEquals(cache.get((1,)), None)
+        self.assertEquals(cache.get((2,)), 2)
+        self.assertEquals(cache.get((3,)), 3)
 
     def test_setdefault(self):
-        cache = LruCache(1)
-        self.assertEquals(cache.setdefault("key", 1), 1)
-        self.assertEquals(cache.get("key"), 1)
-        self.assertEquals(cache.setdefault("key", 2), 1)
-        self.assertEquals(cache.get("key"), 1)
+        cache = LruCache(1, 1)
+        self.assertEquals(cache.setdefault(("key",), 1), 1)
+        self.assertEquals(cache.get(("key",)), 1)
+        self.assertEquals(cache.setdefault(("key",), 2), 1)
+        self.assertEquals(cache.get(("key",)), 1)
 
     def test_pop(self):
-        cache = LruCache(1)
-        cache["key"] = 1
-        self.assertEquals(cache.pop("key"), 1)
-        self.assertEquals(cache.pop("key"), None)
+        cache = LruCache(1, 1)
+        cache[("key",)] = 1
+        self.assertEquals(cache.pop(("key",)), 1)
+        self.assertEquals(cache.pop(("key",)), None)