♻️ fix test

Author: michellewzhang

src/sentry/replays/usecases/summarize.py

@@ -247,31 +247,40 @@ def get_summary_logs(
     error_events: list[EventDict],
     project_id: int,
     is_mobile_replay: bool = False,
+    replay_start: str | None = None,
 ) -> list[str]:
     # Sort error events by timestamp. This list includes all feedback events still.
     error_events.sort(key=lambda x: x["timestamp"])
-    return list(generate_summary_logs(segment_data, error_events, project_id, is_mobile_replay))
+    return list(
+        generate_summary_logs(
+            segment_data, error_events, project_id, is_mobile_replay, replay_start
+        )
+    )
 
 
 def generate_summary_logs(
     segment_data: Iterator[tuple[int, memoryview]],
     error_events: list[EventDict],
     project_id,
     is_mobile_replay: bool = False,
+    replay_start: str | None = None,
 ) -> Generator[str]:
     """
     Generate log messages from events and errors in chronological order.
     Avoid processing duplicate feedback events.
     """
     error_idx = 0
     seen_feedback_ids = {error["id"] for error in error_events if error["category"] == "feedback"}
+    replay_start_ms = _parse_iso_timestamp_to_ms(replay_start) if replay_start else 0.0
 
     # Process segments
     for _, segment in segment_data:
         events = json.loads(segment.tobytes().decode("utf-8"))
         for event in events:
             event_type = which(event)
             timestamp = get_replay_event_timestamp_ms(event, event_type)
+            if timestamp < replay_start_ms:
+                continue
 
             # Check if we need to yield any error messages that occurred before this event
             while (
@@ -582,5 +591,5 @@ def rpc_get_replay_summary_logs(
     segment_data = iter_segment_data(segment_md)
 
     # Combine replay and error data and parse into logs.
-    logs = get_summary_logs(segment_data, error_events, project.id, is_mobile_replay)
+    logs = get_summary_logs(segment_data, error_events, project.id, is_mobile_replay, replay_start)
     return {"logs": logs}

tests/sentry/replays/usecases/test_summarize.py

@@ -1057,18 +1057,21 @@ def save_recording_segment(
         FilestoreBlob().set(metadata, zlib.compress(data) if compressed else data)
 
     def test_rpc_simple(self) -> None:
+        now = datetime.now(UTC)
+        replay_start = now - timedelta(minutes=1)
+
         data = [
             {
                 "type": 5,
-                "timestamp": 0.0,
+                "timestamp": replay_start.timestamp() * 1000,
                 "data": {
                     "tag": "breadcrumb",
                     "payload": {"category": "console", "message": "hello"},
                 },
             },
             {
                 "type": 5,
-                "timestamp": 0.0,
+                "timestamp": replay_start.timestamp() * 1000,
                 "data": {
                     "tag": "breadcrumb",
                     "payload": {"category": "console", "message": "world"},
@@ -1077,15 +1080,18 @@ def test_rpc_simple(self) -> None:
         ]
         self.save_recording_segment(0, json.dumps(data).encode())
         self.save_recording_segment(1, json.dumps([]).encode())
-        self.store_replay()
+        self.store_replay(dt=replay_start)
 
         response = rpc_get_replay_summary_logs(
             self.project.id,
             self.replay_id,
             2,
         )
 
-        assert response == {"logs": ["Logged: 'hello' at 0.0", "Logged: 'world' at 0.0"]}
+        timestamp_ms = replay_start.timestamp() * 1000
+        assert response == {
+            "logs": [f"Logged: 'hello' at {timestamp_ms}", f"Logged: 'world' at {timestamp_ms}"]
+        }
 
     def test_rpc_with_both_direct_and_trace_connected_errors(self) -> None:
         """Test handling of breadcrumbs with both direct and trace connected errors. Error logs should not be duplicated."""
@@ -1212,7 +1218,7 @@ def test_rpc_with_feedback_breadcrumb(self) -> None:
         data = [
             {
                 "type": 5,
-                "timestamp": dt.timestamp(),
+                "timestamp": dt.timestamp() * 1000,
                 "data": {
                     "tag": "breadcrumb",
                     "payload": {
@@ -1421,7 +1427,8 @@ def test_rpc_with_trace_errors_duplicate_feedback(
             id=feedback_event_id,
             title="User Feedback",
             message=feedback_data["contexts"]["feedback"]["message"],
-            timestamp=float(feedback_data["timestamp"]),
+            timestamp=float(feedback_data["timestamp"])
+            * 1000,  # EventDict timestamps are in milliseconds
             category="feedback",
         )
 
@@ -1516,3 +1523,46 @@ def test_rpc_web_replay_navigation(self) -> None:
         logs = response["logs"]
         # Web replays should not include navigation events, so logs should be empty
         assert len(logs) == 0
+
+    def test_rpc_filters_out_events_before_replay_start(self) -> None:
+        """Test that events before the replay start are not logged."""
+        now = datetime.now(UTC)
+        replay_start = now - timedelta(minutes=1)
+
+        self.store_replay(dt=replay_start)
+
+        data = [
+            {
+                "type": 5,
+                "timestamp": float((replay_start - timedelta(minutes=2)).timestamp() * 1000),
+                "data": {
+                    "tag": "breadcrumb",
+                    "payload": {
+                        "category": "console",
+                        "message": "hello",
+                    },
+                },
+            },
+            {
+                "type": 5,
+                "timestamp": float((replay_start + timedelta(minutes=3)).timestamp() * 1000),
+                "data": {
+                    "tag": "breadcrumb",
+                    "payload": {
+                        "category": "console",
+                        "message": "world",
+                    },
+                },
+            },
+        ]
+        self.save_recording_segment(0, json.dumps(data).encode())
+
+        response = rpc_get_replay_summary_logs(
+            self.project.id,
+            self.replay_id,
+            1,
+        )
+
+        logs = response["logs"]
+        assert len(logs) == 1
+        assert "world" in logs[0]