fix(integrations): Gate REMOTE_ADDR on data_collection.user_info

sentry_sdk/integrations/aiohttp.py

@@ -566,7 +566,14 @@ def aiohttp_processor(
 
             request_info["query_string"] = request.query_string
             request_info["method"] = request.method
-            request_info["env"] = {"REMOTE_ADDR": request.remote}
+
+            # REMOTE_ADDR was unconditionally set pre-data collection, so it
+            # continues to be set when data collection is not enabled.
+            if (
+                not has_data_collection_enabled(client_options)
+                or client_options["data_collection"]["user_info"]
+            ):
+                request_info["env"] = {"REMOTE_ADDR": request.remote}
             request_info["headers"] = _filter_headers(dict(request.headers))
 
             # Just attach raw data here if it is within bounds, if available.

sentry_sdk/integrations/sanic.py

@@ -463,7 +463,14 @@ def sanic_processor(event: "Event", hint: "Optional[Hint]") -> "Optional[Event]"
                 request_info["query_string"] = urlparts.query
 
             request_info["method"] = request.method
-            request_info["env"] = {"REMOTE_ADDR": request.remote_addr}
+
+            # REMOTE_ADDR was unconditionally set pre-data collection, so it
+            # continues to be set when data collection is not enabled.
+            if (
+                not has_data_collection_enabled(client_options)
+                or client_options["data_collection"]["user_info"]
+            ):
+                request_info["env"] = {"REMOTE_ADDR": request.remote_addr}
             request_info["headers"] = _filter_headers(dict(request.headers))
 
         return event

sentry_sdk/integrations/tornado.py

@@ -330,7 +330,14 @@ def tornado_processor(event: "Event", hint: "dict[str, Any]") -> "Event":
                 request_info["query_string"] = request.query
 
             request_info["method"] = request.method
-            request_info["env"] = {"REMOTE_ADDR": request.remote_ip}
+
+            # REMOTE_ADDR was unconditionally set pre-data collection, so it
+            # continues to be set when data collection is not enabled.
+            if (
+                not has_data_collection_enabled(client_options)
+                or client_options["data_collection"]["user_info"]
+            ):
+                request_info["env"] = {"REMOTE_ADDR": request.remote_ip}
             request_info["headers"] = _filter_headers(dict(request.headers))
 
         if has_data_collection_enabled(client_options):

tests/integrations/aiohttp/test_aiohttp.py

@@ -27,7 +27,10 @@
 )
 from sentry_sdk.utils import SENSITIVE_DATA_SUBSTITUTE
 from tests.conftest import ApproxDict
-from tests.integrations.utils import DATA_COLLECTION_USER_INFO_CASES
+from tests.integrations.utils import (
+    DATA_COLLECTION_REMOTE_ADDR_CASES,
+    DATA_COLLECTION_USER_INFO_CASES,
+)
 
 
 @pytest.mark.asyncio
@@ -2268,3 +2271,32 @@ async def hello(request):
         assert "url.query" not in inner_client_span["attributes"]
     else:
         assert inner_client_span["attributes"]["url.query"] == expected_query
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize(
+    "init_kwargs, expect_remote_addr", DATA_COLLECTION_REMOTE_ADDR_CASES
+)
+async def test_remote_addr_data_collection(
+    sentry_init, aiohttp_client, capture_events, init_kwargs, expect_remote_addr
+):
+    sentry_init(integrations=[AioHttpIntegration()], **init_kwargs)
+
+    async def hello(request):
+        capture_message("hi")
+        return web.Response(text="hello")
+
+    app = web.Application()
+    app.router.add_get("/", hello)
+
+    events = capture_events()
+
+    client = await aiohttp_client(app)
+    resp = await client.get("/")
+    assert resp.status == 200
+
+    (event,) = events
+    if expect_remote_addr:
+        assert event["request"]["env"] == {"REMOTE_ADDR": "127.0.0.1"}
+    else:
+        assert "env" not in event["request"]

tests/integrations/sanic/test_sanic.py

@@ -16,7 +16,10 @@
 from sentry_sdk.integrations.sanic import SanicIntegration
 from sentry_sdk.tracing import TransactionSource
 from tests.conftest import get_free_port
-from tests.integrations.utils import DATA_COLLECTION_USER_INFO_CASES
+from tests.integrations.utils import (
+    DATA_COLLECTION_REMOTE_ADDR_CASES,
+    DATA_COLLECTION_USER_INFO_CASES,
+)
 
 try:
     from sanic_testing import TestManager
@@ -874,3 +877,24 @@ def oversized_handler(request):
 
     assert "data" not in event["request"]
     assert "data" not in event.get("_meta", {}).get("request", {})
+
+
+@pytest.mark.parametrize(
+    "init_kwargs, expect_remote_addr", DATA_COLLECTION_REMOTE_ADDR_CASES
+)
+def test_remote_addr_data_collection(
+    sentry_init, app, capture_events, init_kwargs, expect_remote_addr
+):
+    sentry_init(integrations=[SanicIntegration()], **init_kwargs)
+    events = capture_events()
+
+    c = get_client(app)
+    with c as client:
+        _, response = client.get("/message")
+        assert response.status == 200
+
+    (event,) = events
+    if expect_remote_addr:
+        assert event["request"]["env"] == {"REMOTE_ADDR": ""}
+    else:
+        assert "env" not in event["request"]

tests/integrations/tornado/test_tornado.py

@@ -8,7 +8,10 @@
 from sentry_sdk import capture_message, start_transaction
 from sentry_sdk._types import SENSITIVE_DATA_SUBSTITUTE
 from sentry_sdk.integrations.tornado import TornadoIntegration
-from tests.integrations.utils import DATA_COLLECTION_USER_INFO_CASES
+from tests.integrations.utils import (
+    DATA_COLLECTION_REMOTE_ADDR_CASES,
+    DATA_COLLECTION_USER_INFO_CASES,
+)
 
 
 @pytest.fixture
@@ -1133,3 +1136,23 @@ def test_client_address_span_attribute_data_collection(
         assert server_span["attributes"]["client.address"] == "127.0.0.1"
     else:
         assert "client.address" not in server_span["attributes"]
+
+
+@pytest.mark.parametrize(
+    "init_kwargs, expect_remote_addr", DATA_COLLECTION_REMOTE_ADDR_CASES
+)
+def test_remote_addr_data_collection(
+    tornado_testcase, sentry_init, capture_events, init_kwargs, expect_remote_addr
+):
+    sentry_init(integrations=[TornadoIntegration()], **init_kwargs)
+    events = capture_events()
+    client = tornado_testcase(Application([(r"/hi", CrashingHandler)]))
+
+    response = client.fetch("/hi")
+    assert response.code == 500
+
+    (event,) = events
+    if expect_remote_addr:
+        assert event["request"]["env"] == {"REMOTE_ADDR": "127.0.0.1"}
+    else:
+        assert "env" not in event["request"]

tests/integrations/utils.py

@@ -39,6 +39,38 @@
     ),
 ]
 
+# Shared parametrization test matrix for ``REMOTE_ADDR`` on events in
+# integrations that set it unconditionally pre-data collection (tornado, sanic,
+# aiohttp). Each case is ``(init_kwargs, expect_remote_addr)``: the address is
+# only gated once ``data_collection`` is enabled, so the legacy
+# ``send_default_pii`` cases still expect it to be collected.
+DATA_COLLECTION_REMOTE_ADDR_CASES = [
+    pytest.param({}, True, id="defaults"),
+    pytest.param({"send_default_pii": True}, True, id="send_default_pii_true"),
+    pytest.param({"send_default_pii": False}, True, id="send_default_pii_false"),
+    pytest.param(
+        {"_experiments": {"data_collection": {}}}, True, id="data_collection_default"
+    ),
+    pytest.param(
+        {"_experiments": {"data_collection": {"user_info": True}}},
+        True,
+        id="data_collection_user_info_true",
+    ),
+    pytest.param(
+        {"_experiments": {"data_collection": {"user_info": False}}},
+        False,
+        id="data_collection_user_info_false",
+    ),
+    pytest.param(
+        {
+            "send_default_pii": True,
+            "_experiments": {"data_collection": {"user_info": False}},
+        },
+        False,
+        id="data_collection_wins_over_send_default_pii",
+    ),
+]
+
 # Shared parametrization test matrix exercising the interaction between the
 # ``data_collection.queues`` experiment and the legacy ``send_default_pii`` boolean
 # for job/task args and kwargs collected by queue integrations (rq, arq, huey).