feat(django): Gate user identity behind data_collection config

sentry_sdk/integrations/django/__init__.py

@@ -30,6 +30,7 @@
     capture_internal_exceptions,
     ensure_integration_enabled,
     event_from_exception,
+    has_data_collection_enabled,
     logger,
     transaction_from_function,
     walk_exception_chain,
@@ -457,6 +458,39 @@ def _attempt_resolve_again(
     _set_transaction_name_and_source(scope, transaction_style, request)
 
 
+def _get_user_from_request_and_set_on_scope(request: "WSGIRequest") -> None:
+    user = getattr(request, "user", None)
+
+    # Evaluating a SimpleLazyObject in an async view can raise django.core.exceptions.SynchronousOnlyOperation.
+    # Exit early if the user has not been materialized yet.
+    is_lazy = isinstance(user, SimpleLazyObject)
+    if is_lazy and hasattr(request, "_cached_user"):
+        user = request._cached_user
+    elif is_lazy:
+        return
+
+    if user is None or not is_authenticated(user):
+        return
+
+    user_info = {}
+    try:
+        user_info["id"] = str(user.pk)
+    except Exception:
+        pass
+
+    try:
+        user_info["email"] = user.email
+    except Exception:
+        pass
+
+    try:
+        user_info["username"] = user.get_username()
+    except Exception:
+        pass
+
+    sentry_sdk.set_user(user_info)
+
+
 def _after_get_response(request: "WSGIRequest") -> None:
     client = sentry_sdk.get_client()
     integration = client.get_integration(DjangoIntegration)
@@ -468,37 +502,12 @@ def _after_get_response(request: "WSGIRequest") -> None:
         _attempt_resolve_again(request, scope, integration.transaction_style)
 
     span_streaming = has_span_streaming_enabled(client.options)
-    if span_streaming and should_send_default_pii():
-        user = getattr(request, "user", None)
-
-        # Evaluating a SimpleLazyObject in an async view can raise django.core.exceptions.SynchronousOnlyOperation.
-        # Exit early if the user has not been materialized yet.
-        is_lazy = isinstance(user, SimpleLazyObject)
-        if is_lazy and hasattr(request, "_cached_user"):
-            user = request._cached_user
-        elif is_lazy:
-            return
-
-        if user is None or not is_authenticated(user):
-            return
-
-        user_info = {}
-        try:
-            user_info["id"] = str(user.pk)
-        except Exception:
-            pass
-
-        try:
-            user_info["email"] = user.email
-        except Exception:
-            pass
-
-        try:
-            user_info["username"] = user.get_username()
-        except Exception:
-            pass
-
-        sentry_sdk.set_user(user_info)
+    if span_streaming:
+        if has_data_collection_enabled(client.options):
+            if client.options["data_collection"]["user_info"]:
+                _get_user_from_request_and_set_on_scope(request)
+        elif should_send_default_pii():
+            _get_user_from_request_and_set_on_scope(request)
 
 
 def _patch_get_response() -> None:
@@ -544,7 +553,12 @@ def wsgi_request_event_processor(event: "Event", hint: "dict[str, Any]") -> "Eve
         with capture_internal_exceptions():
             DjangoRequestExtractor(request).extract_into_event(event)
 
-        if should_send_default_pii():
+        client_options = sentry_sdk.get_client().options
+        if has_data_collection_enabled(client_options):
+            if client_options["data_collection"]["user_info"]:
+                with capture_internal_exceptions():
+                    _set_user_info(request, event)
+        elif should_send_default_pii():
             with capture_internal_exceptions():
                 _set_user_info(request, event)
 

sentry_sdk/integrations/django/asgi.py

@@ -22,6 +22,7 @@
 from sentry_sdk.utils import (
     capture_internal_exceptions,
     ensure_integration_enabled,
+    has_data_collection_enabled,
 )
 
 if TYPE_CHECKING:
@@ -70,7 +71,12 @@ def asgi_request_event_processor(event: "Event", hint: "dict[str, Any]") -> "Eve
         with capture_internal_exceptions():
             DjangoRequestExtractor(request).extract_into_event(event)
 
-        if should_send_default_pii():
+        client_options = sentry_sdk.get_client().options
+        if has_data_collection_enabled(client_options):
+            if client_options["data_collection"]["user_info"]:
+                with capture_internal_exceptions():
+                    _set_user_info(request, event)
+        elif should_send_default_pii():
             with capture_internal_exceptions():
                 _set_user_info(request, event)
 

tests/integrations/django/asgi/test_asgi.py

@@ -16,6 +16,10 @@
 from sentry_sdk.integrations.django import DjangoIntegration
 from sentry_sdk.integrations.django.asgi import _asgi_middleware_mixin_factory
 from tests.integrations.django.myapp.asgi import channels_application
+from tests.integrations.django.utils import (
+    USER_INFO_INIT_KWARGS,
+    pytest_mark_django_db_decorator,
+)
 
 try:
     from django.urls import reverse
@@ -1096,3 +1100,33 @@ async def test_async_middleware_process_exception_is_awaited(
 
     assert response["status"] == 200
     assert response["body"] == b"handled by async process_exception"
+
+
+@pytest.mark.forked
+@pytest.mark.parametrize("application", APPS)
+@pytest.mark.asyncio
+@pytest.mark.skipif(
+    django.VERSION < (3, 0), reason="Django ASGI support shipped in 3.0"
+)
+@pytest.mark.parametrize("init_kwargs, expect_user", USER_INFO_INIT_KWARGS)
+@pytest_mark_django_db_decorator()
+async def test_user_identity_error_event_data_collection(
+    sentry_init, capture_events, application, init_kwargs, expect_user
+):
+    sentry_init(integrations=[DjangoIntegration()], **init_kwargs)
+    events = capture_events()
+
+    comm = HttpCommunicator(application, "GET", "/mylogin-with-exception")
+    await comm.get_response()
+    await comm.wait()
+
+    event = events[-1]
+
+    if expect_user:
+        assert event["user"]["id"] == "1"
+        assert event["user"]["email"] == "lennon@thebeatles.com"
+        assert event["user"]["username"] == "john"
+    else:
+        assert "id" not in event.get("user", {})
+        assert "email" not in event.get("user", {})
+        assert "username" not in event.get("user", {})

tests/integrations/django/myapp/urls.py

@@ -47,6 +47,11 @@ def path(path, *args, **kwargs):
     path("nomessage", views.nomessage, name="nomessage"),
     path("view-with-signal", views.view_with_signal, name="view_with_signal"),
     path("mylogin", views.mylogin, name="mylogin"),
+    path(
+        "mylogin-with-exception",
+        views.mylogin_with_exception,
+        name="mylogin_with_exception",
+    ),
     path("classbased", views.ClassBasedView.as_view(), name="classbased"),
     path("sentryclass", views.SentryClassBasedView(), name="sentryclass"),
     path(

tests/integrations/django/myapp/views.py

@@ -135,6 +135,16 @@ def mylogin(request):
     return HttpResponse("ok")
 
 
+@csrf_exempt
+def mylogin_with_exception(request):
+    user, _ = User.objects.get_or_create(
+        username="john", defaults={"email": "lennon@thebeatles.com"}
+    )
+    user.backend = "django.contrib.auth.backends.ModelBackend"
+    login(request, user)
+    1 / 0
+
+
 @csrf_exempt
 def handler500(request):
     return HttpResponseServerError("Sentry error.")

tests/integrations/django/test_data_scrubbing.py

@@ -6,7 +6,10 @@
 from sentry_sdk.integrations.django import DjangoIntegration
 from tests.conftest import unpack_werkzeug_response, werkzeug_set_cookie
 from tests.integrations.django.myapp.wsgi import application
-from tests.integrations.django.utils import pytest_mark_django_db_decorator
+from tests.integrations.django.utils import (
+    USER_INFO_INIT_KWARGS,
+    pytest_mark_django_db_decorator,
+)
 
 try:
     from django.urls import reverse
@@ -392,40 +395,6 @@ def test_empty_query_string_is_dropped_with_data_collection(
     assert "query_string" not in event["request"]
 
 
-USER_INFO_INIT_KWARGS = [
-    pytest.param({"send_default_pii": True}, True, id="legacy_send_default_pii_true"),
-    pytest.param(
-        {"send_default_pii": False}, False, id="legacy_send_default_pii_false"
-    ),
-    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_true",
-    ),
-    pytest.param(
-        {
-            "send_default_pii": False,
-            "_experiments": {"data_collection": {"user_info": True}},
-        },
-        True,
-        id="data_collection_wins_over_send_default_pii_false",
-    ),
-]
-
-
 @pytest.mark.forked
 @pytest_mark_django_db_decorator()
 @pytest.mark.parametrize("init_kwargs, expect_ip", USER_INFO_INIT_KWARGS)
@@ -462,6 +431,42 @@ def test_user_info_span_attributes_data_collection(
         assert "client.address" not in root_span["attributes"]
 
 
+@pytest.mark.forked
+@pytest_mark_django_db_decorator()
+@pytest.mark.parametrize("init_kwargs, expect_user", USER_INFO_INIT_KWARGS)
+def test_user_identity_span_attributes_data_collection(
+    sentry_init, client, capture_items, init_kwargs, expect_user
+):
+    init_kwargs = dict(init_kwargs)  # shallow copy so we can mutate
+    experiments = init_kwargs.pop("_experiments", {})
+
+    sentry_init(
+        integrations=[DjangoIntegration()],
+        traces_sample_rate=1.0,
+        trace_lifecycle="stream",
+        _experiments=experiments,
+        **init_kwargs,
+    )
+
+    unpack_werkzeug_response(client.get(reverse("mylogin")))
+
+    items = capture_items("span")
+    unpack_werkzeug_response(client.get(reverse("template_test")))
+    sentry_sdk.flush()
+
+    spans = [item.payload for item in items]
+    (span,) = (s for s in spans if s["name"] == "/template-test")
+
+    if expect_user:
+        assert span["attributes"][SPANDATA.USER_ID] == "1"
+        assert span["attributes"][SPANDATA.USER_EMAIL] == "lennon@thebeatles.com"
+        assert span["attributes"][SPANDATA.USER_NAME] == "john"
+    else:
+        assert SPANDATA.USER_ID not in span["attributes"]
+        assert SPANDATA.USER_EMAIL not in span["attributes"]
+        assert SPANDATA.USER_NAME not in span["attributes"]
+
+
 @pytest.mark.forked
 @pytest_mark_django_db_decorator()
 @pytest.mark.parametrize("init_kwargs, expect_ip", USER_INFO_INIT_KWARGS)
@@ -483,6 +488,30 @@ def test_user_info_error_event_data_collection(
         assert "REMOTE_ADDR" not in event["request"]["env"]
 
 
+@pytest.mark.forked
+@pytest_mark_django_db_decorator()
+@pytest.mark.parametrize("init_kwargs, expect_user", USER_INFO_INIT_KWARGS)
+def test_user_identity_error_event_data_collection(
+    sentry_init, client, capture_events, init_kwargs, expect_user
+):
+    sentry_init(integrations=[DjangoIntegration()], **init_kwargs)
+    events = capture_events()
+
+    client.get(reverse("mylogin"))
+    client.get(reverse("view_exc"))
+
+    event = events[-1]
+
+    if expect_user:
+        assert event["user"]["id"] == "1"
+        assert event["user"]["email"] == "lennon@thebeatles.com"
+        assert event["user"]["username"] == "john"
+    else:
+        assert "id" not in event.get("user", {})
+        assert "email" not in event.get("user", {})
+        assert "username" not in event.get("user", {})
+
+
 @pytest.mark.forked
 @pytest_mark_django_db_decorator()
 def test_error_event_no_user_ip_address_without_remote_addr(

tests/integrations/django/utils.py

@@ -19,3 +19,40 @@
         )
 except AttributeError:
     pass
+
+
+# Shared parametrization test matrix exercising the precedence between the legacy
+# ``send_default_pii`` boolean and the ``data_collection.user_info`` setting.
+# The second value indicates whether user info is expected to be collected.
+USER_INFO_INIT_KWARGS = [
+    pytest.param({"send_default_pii": True}, True, id="legacy_send_default_pii_true"),
+    pytest.param(
+        {"send_default_pii": False}, False, id="legacy_send_default_pii_false"
+    ),
+    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_true",
+    ),
+    pytest.param(
+        {
+            "send_default_pii": False,
+            "_experiments": {"data_collection": {"user_info": True}},
+        },
+        True,
+        id="data_collection_wins_over_send_default_pii_false",
+    ),
+]