fix(trends): fix active users persons date range

posthog/queries/trends/test/test_person.py

@@ -1,8 +1,11 @@
+import json
+from datetime import datetime
 from uuid import UUID
 
 from dateutil.relativedelta import relativedelta
 from django.utils import timezone
 from freezegun.api import freeze_time
+from unittest.case import skip
 
 from posthog.models.entity import Entity
 from posthog.models.filters import Filter
@@ -15,12 +18,12 @@
     ClickhouseTestMixin,
     _create_event,
     _create_person,
+    flush_persons_and_events,
     snapshot_clickhouse_queries,
 )
 
 
 class TestPerson(ClickhouseTestMixin, APIBaseTest):
-
     # Note: not using `@snapshot_clickhouse_queries` here because the ordering of the session_ids in the recording
     # query is not guaranteed, so adding it would lead to a flaky test.
     @freeze_time("2021-01-21T20:00:00.000Z")
@@ -155,3 +158,145 @@ def test_group_query_includes_recording_events(self):
                 }
             ],
         )
+
+
+class TestPersonIntegration(ClickhouseTestMixin, APIBaseTest):
+    def test_weekly_active_users(self):
+        for d in range(10, 18):  # create a person and event for each day 10. Sep - 17. Sep
+            _create_person(team_id=self.team.pk, distinct_ids=[f"u_{d}"])
+            _create_event(
+                event="pageview",
+                distinct_id=f"u_{d}",
+                team=self.team,
+                timestamp=datetime(2023, 9, d, 00, 42),
+            )
+        flush_persons_and_events()
+
+        # request weekly active users in the following week
+        filter = {
+            "insight": "TRENDS",
+            "date_from": "2023-09-17T13:37:00",
+            "date_to": "2023-09-24T13:37:00",
+            "events": json.dumps([{"id": "pageview", "math": "weekly_active"}]),
+        }
+        insight_response = self.client.get(f"/api/projects/{self.team.pk}/insights/trend", data=filter)
+        insight_response = (insight_response.json()).get("result")
+
+        self.assertEqual(insight_response[0].get("labels")[5], "22-Sep-2023")
+        self.assertEqual(insight_response[0].get("data")[5], 2)
+
+        persons_url = insight_response[0].get("persons_urls")[5].get("url")
+        response = self.client.get("/" + persons_url)
+
+        data = response.json()
+        self.assertEqual(data.get("results")[0].get("count"), 2)
+        self.assertEqual([item["name"] for item in data.get("results")[0].get("people")], ["u_17", "u_16"])
+
+    def test_weekly_active_users_grouped_by_week(self):
+        for d in range(10, 18):  # create a person and event for each day 10. Sep - 17. Sep
+            _create_person(team_id=self.team.pk, distinct_ids=[f"u_{d}"])
+            _create_event(
+                event="pageview",
+                distinct_id=f"u_{d}",
+                team=self.team,
+                timestamp=datetime(2023, 9, d, 00, 42),
+            )
+        flush_persons_and_events()
+
+        # request weekly active users in the following week
+        filter = {
+            "insight": "TRENDS",
+            "date_from": "2023-09-17T13:37:00",
+            "date_to": "2023-09-24T13:37:00",
+            "interval": "week",
+            "events": json.dumps([{"id": "pageview", "math": "weekly_active"}]),
+        }
+        insight_response = self.client.get(f"/api/projects/{self.team.pk}/insights/trend", data=filter)
+        insight_response = (insight_response.json()).get("result")
+
+        self.assertEqual(insight_response[0].get("labels")[0], "17-Sep-2023")
+        self.assertEqual(insight_response[0].get("data")[0], 7)
+
+        persons_url = insight_response[0].get("persons_urls")[0].get("url")
+        response = self.client.get("/" + persons_url)
+
+        data = response.json()
+        self.assertEqual(data.get("results")[0].get("count"), 7)
+        self.assertEqual(
+            [item["name"] for item in data.get("results")[0].get("people")],
+            ["u_17", "u_16", "u_15", "u_14", "u_13", "u_12", "u_11"],
+        )
+
+    def test_weekly_active_users_cumulative(self):
+        for d in range(10, 18):  # create a person and event for each day 10. Sep - 17. Sep
+            _create_person(team_id=self.team.pk, distinct_ids=[f"u_{d}"])
+            _create_event(
+                event="pageview",
+                distinct_id=f"u_{d}",
+                team=self.team,
+                timestamp=datetime(2023, 9, d, 00, 42),
+            )
+        flush_persons_and_events()
+
+        # request weekly active users in the following week
+        filter = {
+            "insight": "TRENDS",
+            "date_from": "2023-09-10T13:37:00",
+            "date_to": "2023-09-24T13:37:00",
+            "events": json.dumps([{"id": "pageview", "math": "weekly_active"}]),
+            "display": "ActionsLineGraphCumulative",
+        }
+        insight_response = self.client.get(f"/api/projects/{self.team.pk}/insights/trend", data=filter)
+        insight_response = (insight_response.json()).get("result")
+
+        self.assertEqual(insight_response[0].get("labels")[1], "11-Sep-2023")
+        self.assertEqual(insight_response[0].get("data")[1], 3)
+
+        persons_url = insight_response[0].get("persons_urls")[1].get("url")
+        response = self.client.get("/" + persons_url)
+
+        data = response.json()
+        self.assertEqual(data.get("results")[0].get("count"), 2)
+        self.assertEqual([item["name"] for item in data.get("results")[0].get("people")], ["u_11", "u_10"])
+
+    @skip("see PR 17356")
+    def test_weekly_active_users_breakdown(self):
+        for d in range(10, 18):  # create a person and event for each day 10. Sep - 17. Sep
+            _create_person(team_id=self.team.pk, distinct_ids=[f"a_{d}"])
+            _create_person(team_id=self.team.pk, distinct_ids=[f"b_{d}"])
+            _create_event(
+                event="pageview",
+                distinct_id=f"a_{d}",
+                properties={"some_prop": "a"},
+                team=self.team,
+                timestamp=datetime(2023, 9, d, 00, 42),
+            )
+            _create_event(
+                event="pageview",
+                distinct_id=f"b_{d}",
+                properties={"some_prop": "b"},
+                team=self.team,
+                timestamp=datetime(2023, 9, d, 00, 42),
+            )
+        flush_persons_and_events()
+
+        # request weekly active users in the following week
+        filter = {
+            "insight": "TRENDS",
+            "date_from": "2023-09-17T13:37:00",
+            "date_to": "2023-09-24T13:37:00",
+            "events": json.dumps([{"id": "pageview", "math": "weekly_active"}]),
+            "breakdown": "some_prop",
+        }
+        insight_response = self.client.get(f"/api/projects/{self.team.pk}/insights/trend", data=filter)
+        insight_response = (insight_response.json()).get("result")
+
+        self.assertEqual(insight_response[0].get("labels")[5], "22-Sep-2023")
+        # self.assertEqual(insight_response[0].get("data")[5], 2)
+
+        persons_url = insight_response[0].get("persons_urls")[5].get("url")
+        response = self.client.get("/" + persons_url)
+
+        data = response.json()
+        # self.assertEqual(data.get("results")[0].get("count"), 2)
+        self.assertEqual([item["name"] for item in data.get("results")[0].get("people")], ["a_17", "a_16"])

posthog/queries/trends/total_volume.py

@@ -1,5 +1,5 @@
 import urllib.parse
-from datetime import date, datetime
+from datetime import date, datetime, timedelta
 from typing import Any, Callable, Dict, List, Tuple, Union
 
 from posthog.clickhouse.query_tagging import tag_queries
@@ -256,6 +256,21 @@ def _parse(result: List) -> List:
 
         return _parse
 
+    def _offset_date_from(self, point_datetime: datetime, filter: Filter, entity: Entity) -> datetime | None:
+        if filter.display == TRENDS_CUMULATIVE:
+            return filter.date_from
+        elif entity.math in [WEEKLY_ACTIVE, MONTHLY_ACTIVE]:
+            # :TRICKY: We have to offset the date by one, as the final query already subtracts 7 days
+            return point_datetime + timedelta(days=1)
+        else:
+            return point_datetime
+
+    def _offset_date_to(self, point_datetime: datetime, filter: Filter, entity: Entity, team: Team) -> datetime:
+        if entity.math in [WEEKLY_ACTIVE, MONTHLY_ACTIVE]:
+            return point_datetime
+        else:
+            return offset_time_series_date_by_interval(point_datetime, filter=filter, team=team)
+
     def _get_persons_url(
         self, filter: Filter, entity: Entity, team: Team, point_datetimes: List[datetime]
     ) -> List[Dict[str, Any]]:
@@ -267,8 +282,8 @@ def _get_persons_url(
                 "entity_id": entity.id,
                 "entity_type": entity.type,
                 "entity_math": entity.math,
-                "date_from": filter.date_from if filter.display == TRENDS_CUMULATIVE else point_datetime,
-                "date_to": offset_time_series_date_by_interval(point_datetime, filter=filter, team=team),
+                "date_from": self._offset_date_from(point_datetime, filter=filter, entity=entity),
+                "date_to": self._offset_date_to(point_datetime, filter=filter, entity=entity, team=team),
                 "entity_order": entity.order,
             }