From ec662bbe413bd976af97f099ea4f11dafaf98b3e Mon Sep 17 00:00:00 2001
From: Neil Johnson <neil@matrix.org>
Date: Wed, 23 Aug 2023 14:00:34 +0100
Subject: [PATCH] Filter out unwanted user_agents from udv. (#16124)

---
 changelog.d/16124.bugfix                     |  1 +
 synapse/storage/databases/main/client_ips.py |  5 ++
 tests/storage/test_client_ips.py             | 65 ++++++++++++++++++++
 3 files changed, 71 insertions(+)
 create mode 100644 changelog.d/16124.bugfix

diff --git a/changelog.d/16124.bugfix b/changelog.d/16124.bugfix
new file mode 100644
index 0000000000..fb1d501a2f
--- /dev/null
+++ b/changelog.d/16124.bugfix
@@ -0,0 +1 @@
+Filter out user agent references to the sliding sync proxy and rust-sdk from the user_daily_visits table to ensure that Element X can be represented fully.
diff --git a/synapse/storage/databases/main/client_ips.py b/synapse/storage/databases/main/client_ips.py
index 0df160d2b0..d8d333e11d 100644
--- a/synapse/storage/databases/main/client_ips.py
+++ b/synapse/storage/databases/main/client_ips.py
@@ -579,6 +579,11 @@ class ClientIpWorkerStore(ClientIpBackgroundUpdateStore, MonthlyActiveUsersWorke
         device_id: Optional[str],
         now: Optional[int] = None,
     ) -> None:
+        # The sync proxy continuously triggers /sync even if the user is not
+        # present so should be excluded from user_ips entries.
+        if user_agent == "sync-v3-proxy-":
+            return
+
         if not now:
             now = int(self._clock.time_msec())
         key = (user_id, access_token, ip)
diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py
index cd0079871c..209d68b40b 100644
--- a/tests/storage/test_client_ips.py
+++ b/tests/storage/test_client_ips.py
@@ -654,6 +654,71 @@ class ClientIpStoreTestCase(unittest.HomeserverTestCase):
             r,
         )
 
+    def test_invalid_user_agents_are_ignored(self) -> None:
+        # First make sure we have completed all updates.
+        self.wait_for_background_updates()
+
+        user_id1 = "@user1:id"
+        user_id2 = "@user2:id"
+        device_id1 = "MY_DEVICE1"
+        device_id2 = "MY_DEVICE2"
+        access_token1 = "access_token1"
+        access_token2 = "access_token2"
+
+        # Insert a user IP 1
+        self.get_success(
+            self.store.store_device(
+                user_id1,
+                device_id1,
+                "display name1",
+            )
+        )
+        # Insert a user IP 2
+        self.get_success(
+            self.store.store_device(
+                user_id2,
+                device_id2,
+                "display name2",
+            )
+        )
+
+        self.get_success(
+            self.store.insert_client_ip(
+                user_id1, access_token1, "ip", "sync-v3-proxy-", device_id1
+            )
+        )
+        self.get_success(
+            self.store.insert_client_ip(
+                user_id2, access_token2, "ip", "user_agent", device_id2
+            )
+        )
+        # Force persisting to disk
+        self.reactor.advance(200)
+
+        # We should see that in the DB
+        result = self.get_success(
+            self.store.db_pool.simple_select_list(
+                table="user_ips",
+                keyvalues={},
+                retcols=["access_token", "ip", "user_agent", "device_id", "last_seen"],
+                desc="get_user_ip_and_agents",
+            )
+        )
+
+        # ensure user1 is filtered out
+        self.assertEqual(
+            result,
+            [
+                {
+                    "access_token": access_token2,
+                    "ip": "ip",
+                    "user_agent": "user_agent",
+                    "device_id": device_id2,
+                    "last_seen": 0,
+                }
+            ],
+        )
+
 
 class ClientIpAuthTestCase(unittest.HomeserverTestCase):
     servlets = [