feat(boto3): Gate user identifying resource id behind user_info

sentry_sdk/integrations/boto3/_services/sns.py

@@ -1,5 +1,6 @@
 from typing import TYPE_CHECKING
 
+import sentry_sdk
 from sentry_sdk.integrations.boto3._services.base import _ServiceExtension
 from sentry_sdk.integrations.boto3._utils import _get_aws_arn_attributes
 from sentry_sdk.utils import capture_internal_exceptions
@@ -13,13 +14,26 @@ class _SNSExtension(_ServiceExtension):
     def get_request_attributes(self, ctx: "AwsCallContext") -> "Attributes":
         attributes: "Attributes" = {}
         with capture_internal_exceptions():
+            endpoint_arn = ctx.params.get("EndpointArn")
+            subscription_arn = ctx.params.get("SubscriptionArn")
+            target_arn = ctx.params.get("TargetArn")
+
+            # platform endpoint and subscription ARNs identify a single device or
+            # subscriber, so gate them with `user_info`; endpoint can be a device
+            # token, target ARN can be an endpoint, and subscriber email/phone number.
+            if not sentry_sdk.get_client().options["data_collection"]["user_info"]:
+                endpoint_arn = None
+                subscription_arn = None
+                if target_arn and ":endpoint/" in target_arn:
+                    target_arn = None
+
             attributes.update(
                 _get_aws_arn_attributes(
                     (
-                        ctx.params.get("EndpointArn"),
+                        endpoint_arn,
                         ctx.params.get("PlatformApplicationArn"),
-                        ctx.params.get("SubscriptionArn"),
-                        ctx.params.get("TargetArn"),
+                        subscription_arn,
+                        target_arn,
                         ctx.params.get("TopicArn"),
                     ),
                     ("sns",),

tests/integrations/boto3/test_sns.py

@@ -1,8 +1,11 @@
+import pytest
 from botocore.stub import Stubber
 
 from sentry_sdk.consts import OP, SPANDATA
+from sentry_sdk.integrations.boto3 import Boto3Integration
 from tests.integrations.boto3.helpers import capture_spans_by_op
 from tests.integrations.boto3.helpers import client_factory as client_factory
+from tests.integrations.utils import DATA_COLLECTION_USER_INFO_CASES
 
 
 def test_request_attributes(client_factory, capture_items):
@@ -19,3 +22,86 @@ def test_request_attributes(client_factory, capture_items):
     attributes = span["attributes"]
     assert attributes[SPANDATA.CLOUD_ACCOUNT_ID] == "123456789012"
     assert attributes[SPANDATA.CLOUD_RESOURCE_ID] == topic_arn
+
+
+ENDPOINT_ARN = "arn:aws:sns:eu-north-1:123456789012:endpoint/GCM/myapp/5f1c8a3e-0b7d-3c55-9b5e-1a2b3c4d5e6f"
+
+
+@pytest.mark.parametrize(
+    "method,params,response,user_identifying",
+    [
+        # `EndpointARN` can be a device token.
+        pytest.param(
+            "get_endpoint_attributes",
+            {"EndpointArn": ENDPOINT_ARN},
+            {"Attributes": {}},
+            True,
+            id="endpoint_arn",
+        ),
+        # `SubscriptionArn` can be a subscriber email/phone number.
+        pytest.param(
+            "unsubscribe",
+            {
+                "SubscriptionArn": "arn:aws:sns:eu-north-1:123456789012:orders:8e2a6b1c-4d3f-4f6a-9c1e-2b3d4e5f6a7b"
+            },
+            {},
+            True,
+            id="subscription_arn",
+        ),
+        # `TargetArn` can be an endpoint ARN.
+        pytest.param(
+            "publish",
+            {"TargetArn": ENDPOINT_ARN, "Message": "hi"},
+            {"MessageId": "1"},
+            True,
+            id="target_arn_endpoint",
+        ),
+        # `TargetArn` can be a topic ARN, which is not gated.
+        pytest.param(
+            "publish",
+            {
+                "TargetArn": "arn:aws:sns:eu-north-1:123456789012:orders",
+                "Message": "hi",
+            },
+            {"MessageId": "1"},
+            False,
+            id="target_arn_topic",
+        ),
+    ],
+)
+@pytest.mark.parametrize(
+    "data_collection,expect_user_info", DATA_COLLECTION_USER_INFO_CASES
+)
+def test_resource_id_span_attributes_data_collection(
+    sentry_init,
+    client_factory,
+    capture_items,
+    method,
+    params,
+    response,
+    user_identifying,
+    data_collection,
+    expect_user_info,
+):
+    sentry_init(
+        traces_sample_rate=1.0,
+        integrations=[Boto3Integration()],
+        data_collection=data_collection,
+    )
+    client = client_factory("sns")
+    (arn,) = (value for key, value in params.items() if key.endswith("Arn"))
+
+    with Stubber(client) as stubber:
+        stubber.add_response(method, response, params)
+        spans = capture_spans_by_op(
+            lambda: getattr(client, method)(**params), capture_items
+        )
+
+    (span,) = spans[OP.HTTP_CLIENT]
+    attributes = span["attributes"]
+    if expect_user_info or not user_identifying:
+        assert attributes[SPANDATA.CLOUD_RESOURCE_ID] == arn
+        assert attributes[SPANDATA.CLOUD_ACCOUNT_ID] == "123456789012"
+    else:
+        assert SPANDATA.CLOUD_RESOURCE_ID not in attributes
+        assert SPANDATA.CLOUD_ACCOUNT_ID not in attributes