Merge branch 'master' into mz/add-error-context-llm

michellewzhang · web-flow · commit e39dae3cb2e7 · 2025-06-16T23:29:47.000-04:00
diff --git a/devservices/config.yml b/devservices/config.yml
@@ -132,6 +132,20 @@ x-sentry-service-config:
     symbolicator: [postgres, snuba, symbolicator, spotlight]
     memcached: [postgres, snuba, memcached, spotlight]
     profiling: [postgres, snuba, vroom, spotlight]
+    ingest:
+      [
+        snuba,
+        postgres,
+        relay,
+        spotlight,
+        worker,
+        ingest-events,
+        ingest-transactions,
+        ingest-attachments,
+        post-process-forwarder-errors,
+        post-process-forwarder-transactions,
+        post-process-forwarder-issue-platform,
+      ]
     minimal: [postgres, snuba]
     ingest-all:
       [
diff --git a/fixtures/events/performance_problems/sql-injection/sql-injection-test-regex-event.json b/fixtures/events/performance_problems/sql-injection/sql-injection-test-regex-event.json
@@ -0,0 +1,158 @@
+{
+  "event_id": "5d6401994d7949d2ac3474f472564370",
+  "platform": "node",
+  "message": "",
+  "datetime": "2025-05-12T22:42:38.642986+00:00",
+  "breakdowns": {
+    "span_ops": {
+      "ops.db": {
+        "value": 65.715075,
+        "unit": "millisecond"
+      },
+      "total.time": {
+        "value": 67.105293,
+        "unit": "millisecond"
+      }
+    }
+  },
+  "request": {
+    "url": "http://localhost:3001/vulnerable-login",
+    "method": "GET",
+    "query_string": [["type", "account"]]
+  },
+  "spans": [
+    {
+      "timestamp": 1747089758.567536,
+      "start_timestamp": 1747089758.567,
+      "exclusive_time": 0.536203,
+      "op": "middleware.express",
+      "span_id": "4a06692f4abc8dbe",
+      "parent_span_id": "91fa92ff0205967d",
+      "trace_id": "375a86eca09a4a4e91903838dd771f50",
+      "status": "ok",
+      "description": "corsMiddleware",
+      "origin": "auto.http.otel.express",
+      "data": {
+        "express.name": "corsMiddleware",
+        "express.type": "middleware",
+        "sentry.op": "middleware.express",
+        "sentry.origin": "auto.http.otel.express"
+      },
+      "sentry_tags": {
+        "user": "ip:::1",
+        "user.ip": "::1",
+        "environment": "production",
+        "transaction": "GET /vulnerable-login",
+        "transaction.method": "GET",
+        "transaction.op": "http.server",
+        "browser.name": "Chrome",
+        "sdk.name": "sentry.javascript.node",
+        "sdk.version": "9.17.0",
+        "platform": "node",
+        "os.name": "macOS",
+        "category": "middleware",
+        "op": "middleware.express",
+        "status": "ok",
+        "trace.status": "ok"
+      },
+      "hash": "e6088cf8b370ed60"
+    },
+    {
+      "timestamp": 1747089758.568761,
+      "start_timestamp": 1747089758.568,
+      "exclusive_time": 0.761032,
+      "op": "middleware.express",
+      "span_id": "92553d2584d250b8",
+      "parent_span_id": "91fa92ff0205967d",
+      "trace_id": "375a86eca09a4a4e91903838dd771f50",
+      "status": "ok",
+      "description": "jsonParser",
+      "origin": "auto.http.otel.express",
+      "data": {
+        "express.name": "jsonParser",
+        "express.type": "middleware",
+        "sentry.op": "middleware.express",
+        "sentry.origin": "auto.http.otel.express"
+      },
+      "sentry_tags": {
+        "user": "ip:::1",
+        "user.ip": "::1",
+        "environment": "production",
+        "transaction": "GET /vulnerable-login",
+        "transaction.method": "GET",
+        "transaction.op": "http.server",
+        "browser.name": "Chrome",
+        "sdk.name": "sentry.javascript.node",
+        "sdk.version": "9.17.0",
+        "platform": "node",
+        "os.name": "macOS",
+        "category": "middleware",
+        "op": "middleware.express",
+        "status": "ok",
+        "trace.status": "ok"
+      },
+      "hash": "c81e963dad9ebc6c"
+    },
+    {
+      "timestamp": 1747089758.569093,
+      "start_timestamp": 1747089758.569,
+      "exclusive_time": 0.092983,
+      "op": "request_handler.express",
+      "span_id": "435146ab0909419d",
+      "parent_span_id": "91fa92ff0205967d",
+      "trace_id": "375a86eca09a4a4e91903838dd771f50",
+      "status": "ok",
+      "description": "/vulnerable-login",
+      "origin": "auto.http.otel.express",
+      "data": {
+        "express.name": "/vulnerable-login",
+        "express.type": "request_handler",
+        "http.route": "/vulnerable-login",
+        "sentry.op": "request_handler.express",
+        "sentry.origin": "auto.http.otel.express"
+      },
+      "sentry_tags": {
+        "user": "ip:::1",
+        "user.ip": "::1",
+        "environment": "production",
+        "transaction": "GET /vulnerable-login",
+        "transaction.method": "GET",
+        "transaction.op": "http.server",
+        "browser.name": "Chrome",
+        "sdk.name": "sentry.javascript.node",
+        "sdk.version": "9.17.0",
+        "platform": "node",
+        "os.name": "macOS",
+        "op": "request_handler.express",
+        "status": "ok",
+        "trace.status": "ok"
+      },
+      "hash": "872b0c84a6f1c590"
+    },
+    {
+      "timestamp": 1747089758.637715,
+      "start_timestamp": 1747089758.572,
+      "exclusive_time": 65.715075,
+      "op": "db",
+      "span_id": "4703181ac343f71a",
+      "parent_span_id": "91fa92ff0205967d",
+      "trace_id": "375a86eca09a4a4e91903838dd771f50",
+      "status": "ok",
+      "description": "SELECT account.id, \"account\".name, account_type FROM data WHERE type = ?",
+      "origin": "auto.db.otel.mysql2",
+      "data": {
+        "db.system": "mysql",
+        "db.connection_string": "jdbc:mysql://localhost:3306/injection_test",
+        "db.name": "injection_test",
+        "db.statement": "SELECT account.id, \"account\".name, account_type FROM data WHERE type = ?",
+        "db.user": "root",
+        "net.peer.name": "localhost",
+        "net.peer.port": 3306,
+        "otel.kind": "CLIENT",
+        "sentry.op": "db",
+        "sentry.origin": "auto.db.otel.mysql2"
+      },
+      "hash": "45330ba0cafa5997"
+    }
+  ]
+}
diff --git a/src/sentry/api/endpoints/organization_feedback_summary.py b/src/sentry/api/endpoints/organization_feedback_summary.py
@@ -77,7 +77,7 @@ def get(self, request: Request, organization: Organization) -> Response:
 
         if groups.count() < MIN_FEEDBACKS_TO_SUMMARIZE:
             logger.error("Too few feedbacks to summarize")
-            return Response({"summary": None, "success": False, "num_feedbacks_used": 0})
+            return Response({"summary": None, "success": False, "numFeedbacksUsed": 0})
 
         # Also cap the number of characters that we send to the LLM
         group_feedbacks = []
@@ -100,5 +100,5 @@ def get(self, request: Request, organization: Organization) -> Response:
             return Response({"detail": "Error generating summary"}, status=500)
 
         return Response(
-            {"summary": summary, "success": True, "num_feedbacks_used": len(group_feedbacks)}
+            {"summary": summary, "success": True, "numFeedbacksUsed": len(group_feedbacks)}
         )
diff --git a/src/sentry/eventstream/kafka/dispatch.py b/src/sentry/eventstream/kafka/dispatch.py
@@ -100,4 +100,5 @@ class EventPostProcessForwarderStrategyFactory(PostProcessForwarderStrategyFacto
     def _dispatch_function(
         message: Message[KafkaPayload], eventstream_type: str | None = None
     ) -> None:
-        return _get_task_kwargs_and_dispatch(message, eventstream_type)
+        with _sampled_eventstream_timer(instance="_get_task_kwargs_and_dispatch"):
+            return _get_task_kwargs_and_dispatch(message, eventstream_type)
diff --git a/src/sentry/performance_issues/detectors/sql_injection_detector.py b/src/sentry/performance_issues/detectors/sql_injection_detector.py
@@ -122,8 +122,8 @@ def visit_span(self, span: Span) -> None:
         for parameter in self.request_parameters:
             value = parameter[1]
             key = parameter[0]
-            if re.search(f"\\b{re.escape(key)}\\b", description) and re.search(
-                f"\\b{re.escape(value)}\\b", description
+            if re.search(rf'(?<![\w.])"?{re.escape(key)}"?(?![\w."])', description) and re.search(
+                rf'(?<![\w.])"?{re.escape(value)}"?(?![\w."])', description
             ):
                 description = description.replace(value, "?")
                 vulnerable_parameters.append(key)
diff --git a/src/sentry/workflow_engine/processors/delayed_workflow.py b/src/sentry/workflow_engine/processors/delayed_workflow.py
@@ -1,5 +1,6 @@
 import logging
 from collections import defaultdict
+from collections.abc import Mapping, Sequence
 from dataclasses import dataclass
 from datetime import timedelta
 from typing import Any
@@ -35,6 +36,7 @@
 from sentry.utils.retries import ConditionalRetryPolicy, exponential_delay
 from sentry.workflow_engine.handlers.condition.event_frequency_query_handlers import (
     BaseEventFrequencyQueryHandler,
+    QueryFilter,
     QueryResult,
     slow_condition_query_handler_registry,
 )
@@ -83,7 +85,26 @@ class UniqueConditionQuery:
     interval: str
     environment_id: int | None
     comparison_interval: str | None = None
-    filters: list[dict[str, Any]] | None = None
+    # Hashable representation of the filters
+    frozen_filters: Sequence[frozenset[tuple[str, Any]]] | None = None
+
+    @staticmethod
+    def freeze_filters(
+        filters: Sequence[Mapping[str, Any]] | None,
+    ) -> Sequence[frozenset[tuple[str, Any]]] | None:
+        """
+        Convert the sorted representation of filters into a frozen one that can
+        be safely hashed.
+        """
+        if filters is None:
+            return None
+        return tuple(frozenset(sorted(filter.items())) for filter in filters)
+
+    @property
+    def filters(self) -> list[QueryFilter] | None:
+        if self.frozen_filters is None:
+            return None
+        return [dict(filter) for filter in self.frozen_filters]
 
     def __repr__(self):
         return f"UniqueConditionQuery(handler={self.handler.__name__}, interval={self.interval}, environment_id={self.environment_id}, comparison_interval={self.comparison_interval}, filters={self.filters})"
@@ -203,7 +224,7 @@ def generate_unique_queries(
             handler=handler,
             interval=condition.comparison["interval"],
             environment_id=environment_id,
-            filters=condition.comparison.get("filters"),
+            frozen_filters=UniqueConditionQuery.freeze_filters(condition.comparison.get("filters")),
         )
     ]
     if condition_type in PERCENT_CONDITIONS:
@@ -213,7 +234,9 @@ def generate_unique_queries(
                 interval=condition.comparison["interval"],
                 environment_id=environment_id,
                 comparison_interval=condition.comparison.get("comparison_interval"),
-                filters=condition.comparison.get("filters"),
+                frozen_filters=UniqueConditionQuery.freeze_filters(
+                    condition.comparison.get("filters")
+                ),
             )
         )
     return unique_queries
diff --git a/tests/sentry/api/endpoints/test_organization_feedback_summary.py b/tests/sentry/api/endpoints/test_organization_feedback_summary.py
@@ -91,7 +91,7 @@ def test_get_feedback_summary_basic(self, mock_generate_summary):
 
         assert response.data["success"] is True
         assert response.data["summary"] == "Test summary of feedback"
-        assert response.data["num_feedbacks_used"] == 15
+        assert response.data["numFeedbacksUsed"] == 15
 
     @django_db_all
     @patch(
@@ -122,7 +122,7 @@ def test_get_feedback_summary_with_date_filter(self, mock_generate_summary):
 
         assert response.data["success"] is True
         assert response.data["summary"] == "Test summary of feedback"
-        assert response.data["num_feedbacks_used"] == 12
+        assert response.data["numFeedbacksUsed"] == 12
 
     @django_db_all
     @patch(
@@ -151,7 +151,7 @@ def test_get_feedback_summary_with_project_filter(self, mock_generate_summary):
 
         assert response.data["success"] is True
         assert response.data["summary"] == "Test summary of feedback"
-        assert response.data["num_feedbacks_used"] == 10
+        assert response.data["numFeedbacksUsed"] == 10
 
     @django_db_all
     @patch(
@@ -180,7 +180,7 @@ def test_get_feedback_summary_with_many_project_filter_as_list(self, mock_genera
 
         assert response.data["success"] is True
         assert response.data["summary"] == "Test summary of feedback"
-        assert response.data["num_feedbacks_used"] == 22
+        assert response.data["numFeedbacksUsed"] == 22
 
     @django_db_all
     @patch(
@@ -208,7 +208,7 @@ def test_get_feedback_summary_with_many_project_filter_separate(self, mock_gener
         assert response.status_code == 200
         assert response.data["success"] is True
         assert response.data["summary"] == "Test summary of feedback"
-        assert response.data["num_feedbacks_used"] == 22
+        assert response.data["numFeedbacksUsed"] == 22
 
     @django_db_all
     @patch(
@@ -259,4 +259,4 @@ def test_get_feedback_summary_character_limit(self, mock_generate_summary):
 
         assert response.data["success"] is True
         assert response.data["summary"] == "Test summary of feedback"
-        assert response.data["num_feedbacks_used"] == 12
+        assert response.data["numFeedbacksUsed"] == 12
diff --git a/tests/sentry/performance_issues/test_sql_injection_detector.py b/tests/sentry/performance_issues/test_sql_injection_detector.py
@@ -54,6 +54,10 @@ def test_sql_injection_detection_in_body(self):
         assert problem.evidence_data["vulnerable_parameters"] == ["username"]
         assert problem.evidence_data["request_url"] == "http://localhost:3001/vulnerable-login"
 
+    def test_sql_injection_regex(self):
+        injection_event = get_event("sql-injection/sql-injection-test-regex-event")
+        assert len(self.find_problems(injection_event)) == 0
+
     def test_sql_injection_on_non_vulnerable_query(self):
         injection_event = get_event("sql-injection/sql-injection-event-non-vulnerable")
         assert len(self.find_problems(injection_event)) == 0
diff --git a/tests/sentry/workflow_engine/processors/test_delayed_workflow.py b/tests/sentry/workflow_engine/processors/test_delayed_workflow.py
@@ -13,6 +13,7 @@
 from sentry.models.project import Project
 from sentry.notifications.models.notificationaction import ActionTarget
 from sentry.rules.conditions.event_frequency import ComparisonType
+from sentry.rules.match import MatchType
 from sentry.rules.processing.buffer_processing import process_in_batches
 from sentry.rules.processing.delayed_processing import fetch_project
 from sentry.testutils.helpers import override_options, with_feature
@@ -353,6 +354,36 @@ def test_generate_unique_queries(self):
         expected_comparison_query = UniqueConditionQuery(**comparison_query_dict)
         assert percent_queries[1] == expected_comparison_query
 
+    def test_generate_unique_queries__filters_hashable(self):
+        dc = self.create_data_condition(
+            condition_group=self.create_data_condition_group(
+                logic_type=DataConditionGroup.Type.ALL
+            ),
+            type=Condition.EVENT_FREQUENCY_COUNT,
+            comparison={
+                "interval": "1h",
+                "value": 100,
+                "filters": [
+                    {
+                        "key": "http.method",
+                        "match": MatchType.IS_IN,
+                        "value": "GET,POST",
+                    }
+                ],
+            },
+            condition_result=True,
+        )
+        queries = generate_unique_queries(dc, None)
+        [hash(query) for query in queries]  # shouldn't raise
+        assert len(queries) == 1
+        assert queries[0].filters == [
+            {
+                "key": "http.method",
+                "match": MatchType.IS_IN,
+                "value": "GET,POST",
+            }
+        ]
+
     def test_generate_unique_queries__invalid(self):
         dc = self.create_data_condition(
             condition_group=self.workflow_triggers,
