feat(explorer): add event timeseries dict to issue details tool (#102119)

Author: aliu39

src/sentry/seer/explorer/tools.py

@@ -1,5 +1,5 @@
 import logging
-from datetime import datetime, timedelta, timezone
+from datetime import UTC, datetime, timedelta, timezone
 from typing import Any, Literal
 
 from sentry import eventstore
@@ -23,6 +23,7 @@
 from sentry.snuba.referrer import Referrer
 from sentry.snuba.spans_rpc import Spans
 from sentry.snuba.trace import query_trace_data
+from sentry.utils.dates import parse_stats_period
 
 logger = logging.getLogger(__name__)
 
@@ -322,6 +323,69 @@ def get_repository_definition(*, organization_id: int, repo_full_name: str) -> d
     }
 
 
+# Tuples of (total period, interval) (both in sentry stats period format).
+EVENT_TIMESERIES_RESOLUTIONS = (
+    ("6h", "15m"),  # 24 buckets
+    ("24h", "1h"),  # 24 buckets
+    ("3d", "3h"),  # 24 buckets
+    ("7d", "6h"),  # 28 buckets
+    ("14d", "12h"),  # 28 buckets
+    ("30d", "24h"),  # 30 buckets
+    ("90d", "3d"),  # 30 buckets
+)
+
+
+def _get_issue_event_timeseries(
+    *,
+    organization: Organization,
+    project_id: int,
+    issue_short_id: str,
+    first_seen_delta: timedelta,
+) -> tuple[dict[str, Any], str, str] | None:
+    """
+    Get event counts over time for an issue by calling the events-stats endpoint.
+    """
+
+    stats_period, interval = None, None
+    for p, i in EVENT_TIMESERIES_RESOLUTIONS:
+        delta = parse_stats_period(p)
+        if delta and first_seen_delta <= delta:
+            stats_period, interval = p, i
+            break
+    stats_period = stats_period or "90d"
+    interval = interval or "3d"
+
+    params: dict[str, Any] = {
+        "dataset": "issuePlatform",
+        "query": f"issue:{issue_short_id}",
+        "yAxis": "count()",
+        "partial": "1",
+        "statsPeriod": stats_period,
+        "interval": interval,
+        "project": project_id,
+        "referrer": Referrer.SEER_RPC,
+    }
+
+    resp = client.get(
+        auth=ApiKey(organization_id=organization.id, scope_list=["org:read", "project:read"]),
+        user=None,
+        path=f"/organizations/{organization.slug}/events-stats/",
+        params=params,
+    )
+    if resp.status_code != 200 or not (resp.data or {}).get("data"):
+        logger.warning(
+            "Failed to get event counts for issue",
+            extra={
+                "organization_slug": organization.slug,
+                "project_id": project_id,
+                "issue_id": issue_short_id,
+            },
+        )
+        return None
+
+    return {"count()": {"data": resp.data["data"]}}, stats_period, interval
+
+
 def get_issue_details(
     *,
     issue_id: str,
@@ -354,12 +418,12 @@ def get_issue_details(
         )
         return None
 
+    org_project_ids = Project.objects.filter(
+        organization=organization, status=ObjectStatus.ACTIVE
+    ).values_list("id", flat=True)
+
     try:
         if issue_id.isdigit():
-            org_project_ids = Project.objects.filter(
-                organization=organization, status=ObjectStatus.ACTIVE
-            ).values_list("id", flat=True)
-
             group = Group.objects.get(project_id__in=org_project_ids, id=int(issue_id))
         else:
             group = Group.objects.by_qualified_short_id(organization_id, issue_id)
@@ -415,8 +479,24 @@ def get_issue_details(
         )
         tags_overview = None
 
+    ts_result = _get_issue_event_timeseries(
+        organization=organization,
+        project_id=group.project_id,
+        issue_short_id=group.qualified_short_id,
+        first_seen_delta=datetime.now(UTC) - group.first_seen,
+    )
+    if ts_result:
+        timeseries, stats_period, interval = ts_result
+    else:
+        timeseries = None
+        stats_period = None
+        interval = None
+
     return {
         "issue": serialized_group,
+        "event_timeseries": timeseries,
+        "timeseries_stats_period": stats_period,
+        "timeseries_interval": interval,
         "tags_overview": tags_overview,
         "event": serialized_event,
         "event_id": event.event_id,

tests/sentry/seer/explorer/test_tools.py

@@ -1,17 +1,19 @@
 import uuid
-from datetime import datetime, timedelta
+from datetime import UTC, datetime, timedelta
 from typing import Literal
 from unittest.mock import patch
 
 import pytest
 from pydantic import BaseModel
 
+from sentry.api import client
 from sentry.constants import ObjectStatus
 from sentry.models.group import Group
 from sentry.models.groupassignee import GroupAssignee
 from sentry.models.repository import Repository
 from sentry.seer.endpoints.seer_rpc import get_organization_project_ids
 from sentry.seer.explorer.tools import (
+    EVENT_TIMESERIES_RESOLUTIONS,
     execute_trace_query_chart,
     execute_trace_query_table,
     get_issue_details,
@@ -21,6 +23,7 @@
 from sentry.seer.sentry_data_models import EAPTrace
 from sentry.testutils.cases import APITransactionTestCase, SnubaTestCase, SpanTestCase
 from sentry.testutils.helpers.datetime import before_now
+from sentry.utils.dates import parse_stats_period
 from sentry.utils.samples import load_data
 from tests.sentry.issues.test_utils import OccurrenceTestMixin
 
@@ -631,6 +634,20 @@ class _SentryEventData(BaseModel):
 
 class TestGetIssueDetails(APITransactionTestCase, SnubaTestCase, OccurrenceTestMixin):
 
+    def _validate_event_timeseries(self, timeseries: dict):
+        assert isinstance(timeseries, dict)
+        assert "count()" in timeseries
+        assert "data" in timeseries["count()"]
+        assert isinstance(timeseries["count()"]["data"], list)
+        for item in timeseries["count()"]["data"]:
+            assert len(item) == 2
+            assert isinstance(item[0], int)
+            assert isinstance(item[1], list)
+            assert len(item[1]) == 1
+            assert isinstance(item[1][0], dict)
+            assert "count" in item[1][0]
+            assert isinstance(item[1][0]["count"], int)
+
     @patch("sentry.models.group.get_recommended_event")
     @patch("sentry.seer.explorer.tools.get_all_tags_overview")
     def _test_get_issue_details_success(
@@ -717,6 +734,9 @@ def _test_get_issue_details_success(
             else:
                 assert result["event_trace_id"] is None
 
+            # Validate timeseries dict structure.
+            self._validate_event_timeseries(result["event_timeseries"])
+
     def test_get_issue_details_success_int_id(self):
         self._test_get_issue_details_success(use_short_id=False)
 
@@ -861,6 +881,64 @@ def test_get_issue_details_with_assigned_team(self, mock_get_tags, mock_get_reco
         assert md.assignedTo.name == self.team.slug
         assert md.assignedTo.email is None
 
+    @patch("sentry.seer.explorer.tools.client")
+    @patch("sentry.models.group.get_recommended_event")
+    @patch("sentry.seer.explorer.tools.get_all_tags_overview")
+    def test_get_issue_details_timeseries_resolution(
+        self,
+        mock_get_tags,
+        mock_get_recommended_event,
+        mock_api_client,
+    ):
+        """Test groups with different first_seen dates"""
+        mock_get_tags.return_value = {"tags_overview": [{"key": "test_tag", "top_values": []}]}
+        # Passthrough to real client - allows testing call args
+        mock_api_client.get.side_effect = client.get
+
+        for stats_period, interval in EVENT_TIMESERIES_RESOLUTIONS:
+            delta = parse_stats_period(stats_period)
+            assert delta is not None
+            if delta > timedelta(days=30):
+                # Skip the 90d test as the retention for testutils is 30d.
+                continue
+
+            # Set a first_seen date slightly newer than the stats period we're testing.
+            first_seen = datetime.now(UTC) - delta + timedelta(minutes=6, seconds=7)
+            data = load_data("python", timestamp=first_seen)
+            data["exception"] = {"values": [{"type": "Exception", "value": "Test exception"}]}
+            event = self.store_event(data=data, project_id=self.project.id)
+            mock_get_recommended_event.return_value = event
+
+            # Second newer event
+            data = load_data("python", timestamp=first_seen + timedelta(minutes=6, seconds=7))
+            data["exception"] = {"values": [{"type": "Exception", "value": "Test exception"}]}
+            self.store_event(data=data, project_id=self.project.id)
+
+            group = event.group
+            assert isinstance(group, Group)
+            assert group.first_seen == first_seen
+
+            result = get_issue_details(
+                issue_id=str(group.id),
+                organization_id=self.organization.id,
+                selected_event="recommended",
+            )
+
+            # Assert expected stats params were passed to the API.
+            _, kwargs = mock_api_client.get.call_args
+            assert kwargs["path"] == f"/organizations/{self.organization.slug}/events-stats/"
+            assert kwargs["params"]["statsPeriod"] == stats_period
+            assert kwargs["params"]["interval"] == interval
+
+            # Validate final results.
+            assert result is not None
+            self._validate_event_timeseries(result["event_timeseries"])
+            assert result["timeseries_stats_period"] == stats_period
+            assert result["timeseries_interval"] == interval
+
+            # Ensure next iteration makes a fresh group.
+            group.delete()
+
 
 @pytest.mark.django_db(databases=["default", "control"])
 class TestGetRepositoryDefinition(APITransactionTestCase):