fix: include_tracing_fields arg to control unvetted data in rust_tracing integration

Author: antonpirker

sentry_sdk/integrations/rust_tracing.py

@@ -151,12 +151,21 @@ def __init__(
             [Dict[str, Any]], EventTypeMapping
         ] = default_event_type_mapping,
         span_filter: Callable[[Dict[str, Any]], bool] = default_span_filter,
-        send_sensitive_data: Optional[bool] = None,
+        include_tracing_fields: Optional[bool] = None,
     ):
         self.origin = origin
         self.event_type_mapping = event_type_mapping
         self.span_filter = span_filter
-        self.send_sensitive_data = send_sensitive_data
+        self.include_tracing_fields = include_tracing_fields
+
+    def _include_tracing_fields(self) -> bool:
+        """
+        By default, the values of tracing fields are not included in case they
+        contain PII. A user may override that by passing `True` for the
+        `include_tracing_fields` keyword argument of this integration or by
+        setting `send_default_pii` to `True` in their Sentry client options.
+        """
+        return should_send_default_pii() if self.include_tracing_fields is None else self.include_tracing_fields
 
     def on_event(self, event: str, _span_state: TraceState) -> None:
         deserialized_event = json.loads(event)
@@ -207,7 +216,10 @@ def on_new_span(self, attrs: str, span_id: str) -> TraceState:
 
         fields = metadata.get("fields", [])
         for field in fields:
-            sentry_span.set_data(field, attrs.get(field))
+            if self._include_tracing_fields():
+                sentry_span.set_data(field, attrs.get(field))
+            else:
+                sentry_span.set_data(field, SENSITIVE_DATA_SUBSTITUTE)
 
         scope.span = sentry_span
         return (parent_sentry_span, sentry_span)
@@ -225,15 +237,9 @@ def on_record(self, span_id: str, values: str, span_state: TraceState) -> None:
             return
         _parent_sentry_span, sentry_span = span_state
 
-        send_sensitive_data = (
-            should_send_default_pii()
-            if self.send_sensitive_data is None
-            else self.send_sensitive_data
-        )
-
         deserialized_values = json.loads(values)
         for key, value in deserialized_values.items():
-            if send_sensitive_data:
+            if self._include_tracing_fields():
                 sentry_span.set_data(key, value)
             else:
                 sentry_span.set_data(key, SENSITIVE_DATA_SUBSTITUTE)
@@ -259,12 +265,12 @@ def __init__(
             [Dict[str, Any]], EventTypeMapping
         ] = default_event_type_mapping,
         span_filter: Callable[[Dict[str, Any]], bool] = default_span_filter,
-        send_sensitive_data: Optional[bool] = None,
+        include_tracing_fields: Optional[bool] = None,
     ):
         self.identifier = identifier
         origin = f"auto.function.rust_tracing.{identifier}"
         self.tracing_layer = RustTracingLayer(
-            origin, event_type_mapping, span_filter, send_sensitive_data
+            origin, event_type_mapping, span_filter, include_tracing_fields
         )
 
         initializer(self.tracing_layer)

tests/integrations/rust_tracing/test_rust_tracing.py

@@ -1,3 +1,4 @@
+from unittest import mock
 import pytest
 
 from string import Template
@@ -66,7 +67,9 @@ def record(self, span_id: int):
 def test_on_new_span_on_close(sentry_init, capture_events):
     rust_tracing = FakeRustTracing()
     integration = RustTracingIntegration(
-        "test_on_new_span_on_close", rust_tracing.set_layer_impl
+        "test_on_new_span_on_close",
+        initializer=rust_tracing.set_layer_impl,
+        include_tracing_fields=True,
     )
     sentry_init(integrations=[integration], traces_sample_rate=1.0)
 
@@ -105,7 +108,9 @@ def test_on_new_span_on_close(sentry_init, capture_events):
 def test_nested_on_new_span_on_close(sentry_init, capture_events):
     rust_tracing = FakeRustTracing()
     integration = RustTracingIntegration(
-        "test_nested_on_new_span_on_close", rust_tracing.set_layer_impl
+        "test_nested_on_new_span_on_close",
+        initializer=rust_tracing.set_layer_impl,
+        include_tracing_fields=True,
     )
     sentry_init(integrations=[integration], traces_sample_rate=1.0)
 
@@ -331,7 +336,10 @@ def span_filter(metadata: Dict[str, object]) -> bool:
 
     rust_tracing = FakeRustTracing()
     integration = RustTracingIntegration(
-        "test_span_filter", rust_tracing.set_layer_impl, span_filter=span_filter
+        "test_span_filter",
+        initializer=rust_tracing.set_layer_impl,
+        span_filter=span_filter,
+        include_tracing_fields=True,
     )
     sentry_init(integrations=[integration], traces_sample_rate=1.0)
 
@@ -365,7 +373,7 @@ def test_record(sentry_init):
     integration = RustTracingIntegration(
         "test_record",
         initializer=rust_tracing.set_layer_impl,
-        send_sensitive_data=True,
+        include_tracing_fields=True,
     )
     sentry_init(integrations=[integration], traces_sample_rate=1.0)
 
@@ -391,6 +399,7 @@ def span_filter(metadata: Dict[str, object]) -> bool:
         "test_record_in_ignored_span",
         rust_tracing.set_layer_impl,
         span_filter=span_filter,
+        include_tracing_fields=True,
     )
     sentry_init(integrations=[integration], traces_sample_rate=1.0)
 
@@ -409,7 +418,7 @@ def span_filter(metadata: Dict[str, object]) -> bool:
 
 
 @pytest.mark.parametrize(
-    "send_default_pii, send_sensitive_data, sensitive_data_expected",
+    "send_default_pii, include_tracing_fields, tracing_fields_expected",
     [
         (True, True, True),
         (True, False, False),
@@ -419,14 +428,14 @@ def span_filter(metadata: Dict[str, object]) -> bool:
         (False, None, False),
     ],
 )
-def test_sensitive_data(
-    sentry_init, send_default_pii, send_sensitive_data, sensitive_data_expected
+def test_include_tracing_fields(
+    sentry_init, send_default_pii, include_tracing_fields, tracing_fields_expected
 ):
     rust_tracing = FakeRustTracing()
     integration = RustTracingIntegration(
         "test_record",
         initializer=rust_tracing.set_layer_impl,
-        send_sensitive_data=send_sensitive_data,
+        include_tracing_fields=include_tracing_fields,
     )
 
     sentry_init(
@@ -438,13 +447,29 @@ def test_sensitive_data(
         rust_tracing.new_span(RustTracingLevel.Info, 3)
 
         span_before_record = sentry_sdk.get_current_span().to_json()
-        assert span_before_record["data"]["version"] is None
+        if tracing_fields_expected:
+            assert span_before_record["data"]["version"] is None
+        else:
+            assert span_before_record["data"]["version"] == "[Filtered]"
 
         rust_tracing.record(3)
 
         span_after_record = sentry_sdk.get_current_span().to_json()
 
-        if sensitive_data_expected:
-            assert span_after_record["data"]["version"] == "memoized"
+        if tracing_fields_expected:
+            assert span_after_record["data"] == {
+                "thread.id": mock.ANY,
+                "thread.name": mock.ANY,
+                "use_memoized": True,
+                "version": "memoized",
+                "index": 10,
+            }
+
         else:
-            assert span_after_record["data"]["version"] == "[Filtered]"
+            assert span_after_record["data"] == {
+                "thread.id": mock.ANY,
+                "thread.name": mock.ANY,
+                "use_memoized": "[Filtered]",
+                "version": "[Filtered]",
+                "index": "[Filtered]",
+            }