added : timed_cache sub-module

libp2p/pubsub/gossipsub.py

@@ -580,7 +580,7 @@ class GossipSub(IPubsubRouter, Service):
         # Get list of all seen (seqnos, from) from the (seqno, from) tuples in
         # seen_messages cache
         seen_seqnos_and_peers = [
-            seqno_and_from for seqno_and_from in self.pubsub.seen_messages.keys()
+            seqno_and_from for seqno_and_from in self.pubsub.seen_messages.cache.keys()
         ]
 
         # Add all unknown message ids (ids that appear in ihave_msg but not in

libp2p/pubsub/pubsub.py

@@ -18,9 +18,6 @@ from typing import (
 )
 
 import base58
-from lru import (
-    LRU,
-)
 import trio
 
 from libp2p.abc import (
@@ -56,6 +53,9 @@ from libp2p.network.stream.exceptions import (
 from libp2p.peer.id import (
     ID,
 )
+from libp2p.timed_cache.last_seen_cache import (
+    LastSeenCache,
+)
 from libp2p.tools.async_service import (
     Service,
 )
@@ -112,7 +112,7 @@ class Pubsub(Service, IPubsub):
     peer_receive_channel: trio.MemoryReceiveChannel[ID]
     dead_peer_receive_channel: trio.MemoryReceiveChannel[ID]
 
-    seen_messages: LRU[bytes, bool]
+    seen_messages: LastSeenCache
 
     subscribed_topics_send: dict[str, trio.MemorySendChannel[rpc_pb2.Message]]
     subscribed_topics_receive: dict[str, TrioSubscriptionAPI]
@@ -136,6 +136,7 @@ class Pubsub(Service, IPubsub):
         host: IHost,
         router: IPubsubRouter,
         cache_size: int = None,
+        seen_ttl: int = 120,
         strict_signing: bool = True,
         msg_id_constructor: Callable[
             [rpc_pb2.Message], bytes
@@ -187,7 +188,7 @@ class Pubsub(Service, IPubsub):
         else:
             self.sign_key = None
 
-        self.seen_messages = LRU(self.cache_size)
+        self.seen_messages = LastSeenCache(seen_ttl)
 
         # Map of topics we are subscribed to blocking queues
         # for when the given topic receives a message
@@ -662,11 +663,11 @@ class Pubsub(Service, IPubsub):
 
     def _is_msg_seen(self, msg: rpc_pb2.Message) -> bool:
         msg_id = self._msg_id_constructor(msg)
-        return msg_id in self.seen_messages
+        return self.seen_messages.has(msg_id)
 
     def _mark_msg_seen(self, msg: rpc_pb2.Message) -> None:
         msg_id = self._msg_id_constructor(msg)
-        self.seen_messages[msg_id] = True
+        self.seen_messages.add(msg_id)
 
     def _is_subscribed_to_msg(self, msg: rpc_pb2.Message) -> bool:
         return any(topic in self.topic_ids for topic in msg.topicIDs)

libp2p/timed_cache/basic_time_cache.py

@@ -0,0 +1,51 @@
+import threading
+import time
+
+
+class TimedCache:
+    """Base class for Timed Cache with cleanup mechanism."""
+
+    cache: dict[bytes, int]
+
+    SWEEP_INTERVAL = 60  # 1-minute interval between each sweep
+
+    def __init__(self, ttl: int) -> None:
+        """
+        Initialize a new TimedCache with a time-to-live for cache entries
+
+        :param ttl: no of seconds as time-to-live for each cache entry
+        """
+        self.ttl = ttl
+        self.lock = threading.Lock()
+        self.cache = {}
+        self._stop_event = threading.Event()
+        self._thread = threading.Thread(target=self._background_cleanup, daemon=True)
+        self._thread.start()
+
+    def _background_cleanup(self) -> None:
+        while not self._stop_event.wait(self.SWEEP_INTERVAL):
+            self._sweep()
+
+    def _sweep(self) -> None:
+        """Removes expired entries from the cache."""
+        now = time.time()
+        with self.lock:
+            keys_to_remove = [key for key, expiry in self.cache.items() if expiry < now]
+            for key in keys_to_remove:
+                del self.cache[key]
+
+    def stop(self) -> None:
+        """Stops the background cleanup thread."""
+        self._stop_event.set()
+        self._thread.join()
+
+    def length(self) -> int:
+        return len(self.cache)
+
+    def add(self, key: bytes) -> bool:
+        """To be implemented in subclasses."""
+        raise NotImplementedError
+
+    def has(self, key: bytes) -> bool:
+        """To be implemented in subclasses."""
+        raise NotImplementedError

libp2p/timed_cache/first_seen_cache.py

@@ -0,0 +1,20 @@
+import time
+
+from .basic_time_cache import (
+    TimedCache,
+)
+
+
+class FirstSeenCache(TimedCache):
+    """Cache where expiry is set only when first added."""
+
+    def add(self, key: bytes) -> bool:
+        with self.lock:
+            if key in self.cache:
+                return False
+            self.cache[key] = int(time.time()) + self.ttl
+            return True
+
+    def has(self, key: bytes) -> bool:
+        with self.lock:
+            return key in self.cache

libp2p/timed_cache/last_seen_cache.py

@@ -0,0 +1,22 @@
+import time
+
+from .basic_time_cache import (
+    TimedCache,
+)
+
+
+class LastSeenCache(TimedCache):
+    """Cache where expiry is updated on every access."""
+
+    def add(self, key: bytes) -> bool:
+        with self.lock:
+            is_new = key not in self.cache
+            self.cache[key] = int(time.time()) + self.ttl
+            return is_new
+
+    def has(self, key: bytes) -> bool:
+        with self.lock:
+            if key in self.cache:
+                self.cache[key] = int(time.time()) + self.ttl
+                return True
+            return False

libp2p/tools/factories.py

@@ -447,6 +447,7 @@ class PubsubFactory(factory.Factory):
         host: IHost,
         router: IPubsubRouter,
         cache_size: int,
+        seen_ttl: int,
         strict_signing: bool,
         msg_id_constructor: Callable[[rpc_pb2.Message], bytes] = None,
     ) -> AsyncIterator[Pubsub]:
@@ -454,6 +455,7 @@ class PubsubFactory(factory.Factory):
             host=host,
             router=router,
             cache_size=cache_size,
+            seen_ttl=seen_ttl,
             strict_signing=strict_signing,
             msg_id_constructor=msg_id_constructor,
         )
@@ -468,6 +470,7 @@ class PubsubFactory(factory.Factory):
         number: int,
         routers: Sequence[IPubsubRouter],
         cache_size: int = None,
+        seen_ttl: int = None,
         strict_signing: bool = False,
         security_protocol: TProtocol = None,
         muxer_opt: TMuxerOptions = None,
@@ -481,7 +484,12 @@ class PubsubFactory(factory.Factory):
                 pubsubs = [
                     await stack.enter_async_context(
                         cls.create_and_start(
-                            host, router, cache_size, strict_signing, msg_id_constructor
+                            host,
+                            router,
+                            cache_size,
+                            seen_ttl,
+                            strict_signing,
+                            msg_id_constructor,
                         )
                     )
                     for host, router in zip(hosts, routers)
@@ -494,6 +502,7 @@ class PubsubFactory(factory.Factory):
         cls,
         number: int,
         cache_size: int = None,
+        seen_ttl: int = 120,
         strict_signing: bool = False,
         protocols: Sequence[TProtocol] = None,
         security_protocol: TProtocol = None,
@@ -510,6 +519,7 @@ class PubsubFactory(factory.Factory):
             number,
             floodsubs,
             cache_size,
+            seen_ttl,
             strict_signing,
             security_protocol=security_protocol,
             muxer_opt=muxer_opt,