Add regex cache. Only caculate push actions for users that have sent read receipts, and are on that server

7 files changed, 50 insertions, 14 deletions

diff --git a/synapse/handlers/_base.py b/synapse/handlers/_base.py
index 2d1167296a..5c7617de44 100644
--- a/synapse/handlers/_base.py
+++ b/synapse/handlers/_base.py
@@ -266,7 +266,7 @@ class BaseHandler(object):
             event, context=context
         )
 
-        action_generator = ActionGenerator(self.store)
+        action_generator = ActionGenerator(self.hs)
         yield action_generator.handle_push_actions_for_event(
             event, self
         )
diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py
index 4b94940e99..6c19d6ae8c 100644
--- a/synapse/handlers/federation.py
+++ b/synapse/handlers/federation.py
@@ -245,7 +245,7 @@ class FederationHandler(BaseHandler):
                     yield user_joined_room(self.distributor, user, event.room_id)
 
         if not backfilled and not event.internal_metadata.is_outlier():
-            action_generator = ActionGenerator(self.store)
+            action_generator = ActionGenerator(self.hs)
             yield action_generator.handle_push_actions_for_event(
                 event, self
             )
diff --git a/synapse/push/action_generator.py b/synapse/push/action_generator.py
index 4cf94f6c61..1d2e558f9a 100644
--- a/synapse/push/action_generator.py
+++ b/synapse/push/action_generator.py
@@ -25,8 +25,9 @@ logger = logging.getLogger(__name__)
 
 
 class ActionGenerator:
-    def __init__(self, store):
-        self.store = store
+    def __init__(self, hs):
+        self.hs = hs
+        self.store = hs.get_datastore()
         # really we want to get all user ids and all profile tags too,
         # since we want the actions for each profile tag for every user and
         # also actions for a client with no profile tag for each user.
@@ -42,7 +43,7 @@ class ActionGenerator:
             )
 
         bulk_evaluator = yield bulk_push_rule_evaluator.evaluator_for_room_id(
-            event.room_id, self.store
+            event.room_id, self.hs, self.store
         )
 
         actions_by_user = yield bulk_evaluator.action_for_event_by_user(event, handler)
diff --git a/synapse/push/bulk_push_rule_evaluator.py b/synapse/push/bulk_push_rule_evaluator.py
index efd686fa6e..1000ae6301 100644
--- a/synapse/push/bulk_push_rule_evaluator.py
+++ b/synapse/push/bulk_push_rule_evaluator.py
@@ -36,6 +36,7 @@ def decode_rule_json(rule):
 @defer.inlineCallbacks
 def _get_rules(room_id, user_ids, store):
     rules_by_user = yield store.bulk_get_push_rules(user_ids)
+
     rules_by_user = {
         uid: baserules.list_with_base_rules([
             decode_rule_json(rule_list)
@@ -47,12 +48,16 @@ def _get_rules(room_id, user_ids, store):
 
 
 @defer.inlineCallbacks
-def evaluator_for_room_id(room_id, store):
-    users = yield store.get_users_in_room(room_id)
-    rules_by_user = yield _get_rules(room_id, users, store)
+def evaluator_for_room_id(room_id, hs, store):
+    results = yield store.get_receipts_for_room(room_id, "m.read")
+    user_ids = [
+        row["user_id"] for row in results
+        if hs.is_mine_id(row["user_id"])
+    ]
+    rules_by_user = yield _get_rules(room_id, user_ids, store)
 
     defer.returnValue(BulkPushRuleEvaluator(
-        room_id, rules_by_user, users, store
+        room_id, rules_by_user, user_ids, store
     ))
 
 
@@ -129,7 +134,7 @@ def _condition_checker(evaluator, conditions, uid, display_name, cache):
 
         res = evaluator.matches(cond, uid, display_name, None)
         if _id:
-            cache[_id] = res
+            cache[_id] = bool(res)
 
         if not res:
             return False
diff --git a/synapse/push/push_rule_evaluator.py b/synapse/push/push_rule_evaluator.py
index 4654994d2d..753b6469e2 100644
--- a/synapse/push/push_rule_evaluator.py
+++ b/synapse/push/push_rule_evaluator.py
@@ -22,6 +22,7 @@ import simplejson as json
 import re
 
 from synapse.types import UserID
+from synapse.util.caches.lrucache import LruCache
 
 logger = logging.getLogger(__name__)
 
@@ -277,18 +278,18 @@ def _glob_matches(glob, value, word_boundary=False):
             )
             if word_boundary:
                 r = r"\b%s\b" % (r,)
-                r = re.compile(r, flags=re.IGNORECASE)
+                r = _compile_regex(r)
 
                 return r.search(value)
             else:
                 r = r + "$"
-                r = re.compile(r, flags=re.IGNORECASE)
+                r = _compile_regex(r)
 
                 return r.match(value)
         elif word_boundary:
             r = re.escape(glob)
             r = r"\b%s\b" % (r,)
-            r = re.compile(r, flags=re.IGNORECASE)
+            r = _compile_regex(r)
 
             return r.search(value)
         else:
@@ -306,3 +307,16 @@ def _flatten_dict(d, prefix=[], result={}):
             _flatten_dict(value, prefix=(prefix+[key]), result=result)
 
     return result
+
+
+regex_cache = LruCache(100000)
+
+
+def _compile_regex(regex_str):
+    r = regex_cache.get(regex_str, None)
+    if r:
+        return r
+
+    r = re.compile(regex_str, flags=re.IGNORECASE)
+    regex_cache[regex_str] = r
+    return r
diff --git a/synapse/server.py b/synapse/server.py
index ffd4f936d0..63f9059837 100644
--- a/synapse/server.py
+++ b/synapse/server.py
@@ -96,6 +96,7 @@ class BaseHomeServer(object):
             hostname : The hostname for the server.
         """
         self.hostname = hostname
+        self.hostname_with_colon = ":" + hostname
         self._building = {}
 
         # Other kwargs are explicit dependencies
@@ -139,6 +140,9 @@ class BaseHomeServer(object):
     def is_mine(self, domain_specific_string):
         return domain_specific_string.domain == self.hostname
 
+    def is_mine_id(self, string):
+        return string.endswith(self.hostname_with_colon)
+
 # Build magic accessors for every dependency
 for depname in BaseHomeServer.DEPENDENCIES:
     BaseHomeServer._make_dependency_method(depname)
diff --git a/synapse/storage/receipts.py b/synapse/storage/receipts.py
index 21cf88b3da..c80e576620 100644
--- a/synapse/storage/receipts.py
+++ b/synapse/storage/receipts.py
@@ -14,7 +14,7 @@
 # limitations under the License.
 
 from ._base import SQLBaseStore
-from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList
+from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList, cached
 from synapse.util.caches import cache_counter, caches_by_name
 
 from twisted.internet import defer
@@ -33,6 +33,18 @@ class ReceiptsStore(SQLBaseStore):
 
         self._receipts_stream_cache = _RoomStreamChangeCache()
 
+    @cached(num_args=2)
+    def get_receipts_for_room(self, room_id, receipt_type):
+        return self._simple_select_list(
+            table="receipts_linearized",
+            keyvalues={
+                "room_id": room_id,
+                "receipt_type": receipt_type,
+            },
+            retcols=("user_id", "event_id"),
+            desc="get_receipts_for_room",
+        )
+
     @defer.inlineCallbacks
     def get_linearized_receipts_for_rooms(self, room_ids, to_key, from_key=None):
         """Get receipts for multiple rooms for sending to clients.