feat(breadcrumbs): add _meta information for truncation of breadcrumbs (#4007)

- Implements annotations for breadcrumbs
- Adds an `int` field to `Scope` to track the number of truncated
breadcrumbs
- When scopes are merged, the number of breadcrumbs that were removed
are added
- If breadcrumbs were truncated, add the original number of breadcrumbs
to `_meta`

- Closes getsentry/projects#593

---------

Co-authored-by: Anton Pirker <[email protected]>

Author: shellmayr

sentry_sdk/_types.py

@@ -30,6 +30,17 @@ def __eq__(self, other):
 
         return self.value == other.value and self.metadata == other.metadata
 
+    def __str__(self):
+        # type: (AnnotatedValue) -> str
+        return str({"value": str(self.value), "metadata": str(self.metadata)})
+
+    def __len__(self):
+        # type: (AnnotatedValue) -> int
+        if self.value is not None:
+            return len(self.value)
+        else:
+            return 0
+
     @classmethod
     def removed_because_raw_data(cls):
         # type: () -> AnnotatedValue
@@ -152,8 +163,8 @@ class SDKInfo(TypedDict):
     Event = TypedDict(
         "Event",
         {
-            "breadcrumbs": dict[
-                Literal["values"], list[dict[str, Any]]
+            "breadcrumbs": Annotated[
+                dict[Literal["values"], list[dict[str, Any]]]
             ],  # TODO: We can expand on this type
             "check_in_id": str,
             "contexts": dict[str, dict[str, object]],

sentry_sdk/client.py

@@ -498,6 +498,7 @@ def _prepare_event(
         # type: (...) -> Optional[Event]
 
         previous_total_spans = None  # type: Optional[int]
+        previous_total_breadcrumbs = None  # type: Optional[int]
 
         if event.get("timestamp") is None:
             event["timestamp"] = datetime.now(timezone.utc)
@@ -534,6 +535,16 @@ def _prepare_event(
             dropped_spans = event.pop("_dropped_spans", 0) + spans_delta  # type: int
             if dropped_spans > 0:
                 previous_total_spans = spans_before + dropped_spans
+            if scope._n_breadcrumbs_truncated > 0:
+                breadcrumbs = event.get("breadcrumbs", {})
+                values = (
+                    breadcrumbs.get("values", [])
+                    if not isinstance(breadcrumbs, AnnotatedValue)
+                    else []
+                )
+                previous_total_breadcrumbs = (
+                    len(values) + scope._n_breadcrumbs_truncated
+                )
 
         if (
             self.options["attach_stacktrace"]
@@ -586,7 +597,10 @@ def _prepare_event(
             event["spans"] = AnnotatedValue(
                 event.get("spans", []), {"len": previous_total_spans}
             )
-
+        if previous_total_breadcrumbs is not None:
+            event["breadcrumbs"] = AnnotatedValue(
+                event.get("breadcrumbs", []), {"len": previous_total_breadcrumbs}
+            )
         # Postprocess the event here so that annotated types do
         # generally not surface in before_send
         if event is not None:

sentry_sdk/scope.py

@@ -9,6 +9,7 @@
 from functools import wraps
 from itertools import chain
 
+from sentry_sdk._types import AnnotatedValue
 from sentry_sdk.attachments import Attachment
 from sentry_sdk.consts import DEFAULT_MAX_BREADCRUMBS, FALSE_VALUES, INSTRUMENTER
 from sentry_sdk.feature_flags import FlagBuffer, DEFAULT_FLAG_CAPACITY
@@ -186,6 +187,7 @@ class Scope:
         "_contexts",
         "_extras",
         "_breadcrumbs",
+        "_n_breadcrumbs_truncated",
         "_event_processors",
         "_error_processors",
         "_should_capture",
@@ -210,6 +212,7 @@ def __init__(self, ty=None, client=None):
 
         self._name = None  # type: Optional[str]
         self._propagation_context = None  # type: Optional[PropagationContext]
+        self._n_breadcrumbs_truncated = 0  # type: int
 
         self.client = NonRecordingClient()  # type: sentry_sdk.client.BaseClient
 
@@ -243,6 +246,7 @@ def __copy__(self):
         rv._extras = dict(self._extras)
 
         rv._breadcrumbs = copy(self._breadcrumbs)
+        rv._n_breadcrumbs_truncated = copy(self._n_breadcrumbs_truncated)
         rv._event_processors = list(self._event_processors)
         rv._error_processors = list(self._error_processors)
         rv._propagation_context = self._propagation_context
@@ -916,6 +920,7 @@ def clear_breadcrumbs(self):
         # type: () -> None
         """Clears breadcrumb buffer."""
         self._breadcrumbs = deque()  # type: Deque[Breadcrumb]
+        self._n_breadcrumbs_truncated = 0
 
     def add_attachment(
         self,
@@ -983,6 +988,7 @@ def add_breadcrumb(self, crumb=None, hint=None, **kwargs):
 
         while len(self._breadcrumbs) > max_breadcrumbs:
             self._breadcrumbs.popleft()
+            self._n_breadcrumbs_truncated += 1
 
     def start_transaction(
         self,
@@ -1366,17 +1372,23 @@ def _apply_level_to_event(self, event, hint, options):
 
     def _apply_breadcrumbs_to_event(self, event, hint, options):
         # type: (Event, Hint, Optional[Dict[str, Any]]) -> None
-        event.setdefault("breadcrumbs", {}).setdefault("values", []).extend(
-            self._breadcrumbs
-        )
+        event.setdefault("breadcrumbs", {})
+
+        # This check is just for mypy -
+        if not isinstance(event["breadcrumbs"], AnnotatedValue):
+            event["breadcrumbs"].setdefault("values", [])
+            event["breadcrumbs"]["values"].extend(self._breadcrumbs)
 
         # Attempt to sort timestamps
         try:
-            for crumb in event["breadcrumbs"]["values"]:
-                if isinstance(crumb["timestamp"], str):
-                    crumb["timestamp"] = datetime_from_isoformat(crumb["timestamp"])
+            if not isinstance(event["breadcrumbs"], AnnotatedValue):
+                for crumb in event["breadcrumbs"]["values"]:
+                    if isinstance(crumb["timestamp"], str):
+                        crumb["timestamp"] = datetime_from_isoformat(crumb["timestamp"])
 
-            event["breadcrumbs"]["values"].sort(key=lambda crumb: crumb["timestamp"])
+                event["breadcrumbs"]["values"].sort(
+                    key=lambda crumb: crumb["timestamp"]
+                )
         except Exception as err:
             logger.debug("Error when sorting breadcrumbs", exc_info=err)
             pass
@@ -1564,6 +1576,10 @@ def update_from_scope(self, scope):
             self._extras.update(scope._extras)
         if scope._breadcrumbs:
             self._breadcrumbs.extend(scope._breadcrumbs)
+        if scope._n_breadcrumbs_truncated:
+            self._n_breadcrumbs_truncated = (
+                self._n_breadcrumbs_truncated + scope._n_breadcrumbs_truncated
+            )
         if scope._span:
             self._span = scope._span
         if scope._attachments:

sentry_sdk/scrubber.py

@@ -144,7 +144,10 @@ def scrub_breadcrumbs(self, event):
         # type: (Event) -> None
         with capture_internal_exceptions():
             if "breadcrumbs" in event:
-                if "values" in event["breadcrumbs"]:
+                if (
+                    not isinstance(event["breadcrumbs"], AnnotatedValue)
+                    and "values" in event["breadcrumbs"]
+                ):
                     for value in event["breadcrumbs"]["values"]:
                         if "data" in value:
                             self.scrub_dict(value["data"])

tests/test_scrubber.py

@@ -119,25 +119,33 @@ def test_stack_var_scrubbing(sentry_init, capture_events):
 
 
 def test_breadcrumb_extra_scrubbing(sentry_init, capture_events):
-    sentry_init()
+    sentry_init(max_breadcrumbs=2)
     events = capture_events()
-
-    logger.info("bread", extra=dict(foo=42, password="secret"))
+    logger.info("breadcrumb 1", extra=dict(foo=1, password="secret"))
+    logger.info("breadcrumb 2", extra=dict(bar=2, auth="secret"))
+    logger.info("breadcrumb 3", extra=dict(foobar=3, password="secret"))
     logger.critical("whoops", extra=dict(bar=69, auth="secret"))
 
     (event,) = events
 
     assert event["extra"]["bar"] == 69
     assert event["extra"]["auth"] == "[Filtered]"
-
     assert event["breadcrumbs"]["values"][0]["data"] == {
-        "foo": 42,
+        "bar": 2,
+        "auth": "[Filtered]",
+    }
+    assert event["breadcrumbs"]["values"][1]["data"] == {
+        "foobar": 3,
         "password": "[Filtered]",
     }
 
     assert event["_meta"]["extra"]["auth"] == {"": {"rem": [["!config", "s"]]}}
     assert event["_meta"]["breadcrumbs"] == {
-        "values": {"0": {"data": {"password": {"": {"rem": [["!config", "s"]]}}}}}
+        "": {"len": 3},
+        "values": {
+            "0": {"data": {"auth": {"": {"rem": [["!config", "s"]]}}}},
+            "1": {"data": {"password": {"": {"rem": [["!config", "s"]]}}}},
+        },
     }