feat(replays): support querying segment_names

src/sentry/replays/post_process.py

@@ -69,6 +69,7 @@ class ReplayDetailsResponse(TypedDict, total=False):
     ota_updates: OTAUpdatesResponseType
     is_archived: bool | None
     urls: list[str] | None
+    segment_names: list[str] | None
     clicks: list[dict[str, Any]]
     count_dead_clicks: int | None
     count_rage_clicks: int | None
@@ -179,6 +180,7 @@ def generate_normalized_output(response: list[dict[str, Any]]) -> Generator[Repl
 
         item.pop("agg_urls", None)
         ret_item["urls"] = item.pop("urls_sorted", None)
+        ret_item["segment_names"] = item.pop("segment_names", None)
 
         ret_item["is_archived"] = bool(item.pop("isArchived", 0))
 
@@ -260,6 +262,7 @@ def _archived_row(replay_id: str, project_id: int) -> ReplayDetailsResponse:
         "device": {"name": None, "brand": None, "model": None, "family": None},
         "ota_updates": {"channel": None, "runtime_version": None, "update_id": None},
         "urls": None,
+        "segment_names": None,
         "activity": None,
         "count_dead_clicks": None,
         "count_rage_clicks": None,

src/sentry/replays/query.py

@@ -635,6 +635,7 @@ def _empty_uuids_lambda():
     "duration": ["duration", "started_at", "finished_at"],
     "urls": ["urls_sorted", "agg_urls"],
     "url": ["urls_sorted", "agg_urls"],
+    "segment_names": ["segment_names"],
     "count_errors": ["count_errors"],
     "count_urls": ["count_urls"],
     "count_segments": ["count_segments"],
@@ -770,6 +771,11 @@ def _empty_uuids_lambda():
         parameters=[Function("tuple", parameters=[Column("segment_id"), Column("urls")])],
         alias="agg_urls",
     ),
+    "segment_names": Function(
+        "groupUniqArrayArray",
+        parameters=[Column("segment_names")],
+        alias="segment_names",
+    ),
     "count_segments": Function("count", parameters=[Column("segment_id")], alias="count_segments"),
     "count_urls": Function(
         "sum",

src/sentry/replays/testutils.py

@@ -130,6 +130,7 @@ def mock_expected_response(
                 "subdivision": kwargs.pop("user_geo_subdivision", "California"),
             },
         },
+        "segment_names": kwargs.pop("segment_names", []),
         "tags": kwargs.pop("tags", {}),
         "activity": kwargs.pop("activity", 0),
         "is_archived": kwargs.pop("is_archived", False),
@@ -166,6 +167,7 @@ def mock_replay(
             "segment_id": kwargs.pop("segment_id", 0),
             "tags": tags,
             "urls": kwargs.pop("urls", []),
+            "segment_names": kwargs.pop("segment_names", []),
             "is_archived": kwargs.pop("is_archived", None),
             "error_ids": kwargs.pop("error_ids", ["a3a62ef6-ac86-415b-83c2-416fc2f76db1"]),
             "trace_ids": kwargs.pop("trace_ids", ["44916572-43ba-4dbe-bd2f-6bd62b733080"]),
@@ -524,6 +526,7 @@ def mock_replay_event(replay_id="b58a67446c914f44a4e329763420047b", **kwargs):
         "segment_id": kwargs.pop("segment_id", 0),
         "tags": tags,
         "urls": kwargs.pop("urls", []),
+        "segment_names": kwargs.pop("segment_names", []),
         "is_archived": kwargs.pop("is_archived", None),
         "error_ids": kwargs.pop("error_ids", ["a3a62ef6-ac86-415b-83c2-416fc2f76db1"]),
         "trace_ids": kwargs.pop("trace_ids", ["44916572-43ba-4dbe-bd2f-6bd62b733080"]),

src/sentry/replays/usecases/query/configs/aggregate.py

@@ -133,6 +133,7 @@ def array_string_field(column_name: str) -> StringColumnField:
     "tap.view_class": string_field("tap_view_class"),
     "tap.view_id": string_field("tap_view_id"),
     "trace_ids": UUIDColumnField("trace_ids", parse_uuid, SumOfUUIDArray),
+    "segment_names": array_string_field("segment_names"),
     "urls": array_string_field("urls"),
     "user.email": string_field("user_email"),
     "user.id": string_field("user_id"),

src/sentry/replays/usecases/query/configs/scalar.py

@@ -70,6 +70,7 @@ def string_field(column_name: str) -> StringColumnField:
 varying_search_config: dict[str, FieldProtocol] = {
     "error_ids": ComputedField(parse_uuid, ErrorIdScalar),
     "trace_ids": UUIDColumnField("trace_ids", parse_uuid, UUIDArray),
+    "segment_names": StringColumnField("segment_names", parse_str, StringArray),
     "urls": StringColumnField("urls", parse_str, StringArray),
 }
 

src/sentry/replays/validators.py

@@ -26,6 +26,7 @@
     "started_at",
     "tags",
     "trace_ids",
+    "segment_names",
     "urls",
     "user",
     "clicks",

tests/sentry/replays/endpoints/test_organization_replay_index.py

@@ -623,6 +623,7 @@ def test_get_replays_user_filters(self) -> None:
                 device_model="10",
                 tags={"a": "m", "b": "q", "c": "test"},
                 urls=["example.com"],
+                segment_names=["/api/checkout"],
                 segment_id=0,
             )
         )
@@ -783,6 +784,7 @@ def test_get_replays_user_filters(self) -> None:
                 "url:example.com",
                 "screens:example.com",
                 "screen:example.com",
+                "segment_names:/api/checkout",
                 "activity:8",
                 "activity:>2",
                 "count_warnings:1",
@@ -853,6 +855,8 @@ def test_get_replays_user_filters(self) -> None:
                 "release:[a,b]",
                 "c:*zz",
                 "!c:*st",
+                "segment_names:/api/orders",
+                "!segment_names:/api/checkout",
                 "!activity:8",
                 "activity:<2",
                 f"viewed_by_id:{self.user.id + 1}",
@@ -1808,6 +1812,38 @@ def test_query_branches_array_of_string_conditions(self) -> None:
                 response_data = response.json()
                 assert len(response_data["data"]) == 1, query
 
+    def test_query_branches_segment_names_conditions(self) -> None:
+        project = self.create_project(teams=[self.team])
+
+        replay1_id = uuid.uuid4().hex
+        seq1_timestamp = datetime.datetime.now() - datetime.timedelta(seconds=22)
+        seq2_timestamp = datetime.datetime.now() - datetime.timedelta(seconds=5)
+
+        self.store_replays(
+            mock_replay(
+                seq1_timestamp,
+                project.id,
+                replay1_id,
+                segment_names=["/api/checkout"],
+            )
+        )
+        self.store_replays(mock_replay(seq2_timestamp, project.id, replay1_id, segment_names=[]))
+
+        with self.feature(self.features):
+            queries = [
+                "segment_names:/api/checkout",
+                "!segment_names:/api/orders",
+                "segment_names:[/api/checkout,/api/orders]",
+                "!segment_names:[/api/orders,/api/users]",
+                "segment_names:/api/*",
+                "!segment_names:/web/*",
+            ]
+            for query in queries:
+                response = self.client.get(self.url + f"?field=id&query={query}")
+                assert response.status_code == 200
+                response_data = response.json()
+                assert len(response_data["data"]) == 1, query
+
     def test_query_branches_integer_conditions(self) -> None:
         project = self.create_project(teams=[self.team])