Merge branch 'master' into mjq/http-route-server-spans

Author: mjq

.github/workflows/flaky-test-detector.yml

@@ -108,7 +108,7 @@ jobs:
       # and the repo, and writes the issue body to flaky-issue-body.md.
       - name: Analyze logs and summarize flaky tests
         if: steps.collect.outputs.collected != '0'
-        uses: anthropics/claude-code-action@1f291e1cfe0f5fc21db2aef19af844591600ade7 # v1.0.206
+        uses: anthropics/claude-code-action@70fec183852c4f82f3f1969faed7dd60c5149ca7 # v1.0.207
         with:
           anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
           github_token: ${{ github.token }}

sentry_sdk/_types.py

@@ -188,7 +188,7 @@ class DataCollectionUserOptions(TypedDict, total=False):
         gen_ai: "GenAICollectionUserOptions"
         database_query_data: bool
         queues: bool
-        stack_frame_variables: bool
+        stack_frame_variables: "Union[bool, KeyValueCollectionBehaviour]"
         frame_context_lines: int
 
     class DataCollection(TypedDict):
@@ -202,7 +202,7 @@ class DataCollection(TypedDict):
         gen_ai: "GenAICollectionBehaviour"
         database_query_data: bool
         queues: bool
-        stack_frame_variables: bool
+        stack_frame_variables: "Union[bool, KeyValueCollectionBehaviour]"
         frame_context_lines: int
 
     # "critical" is an alias of "fatal" recognized by Relay

sentry_sdk/data_collection.py

@@ -23,14 +23,12 @@
 """
 
 import warnings
-from typing import TYPE_CHECKING, List, Mapping, Optional, Union, cast
+from typing import TYPE_CHECKING, Any, Dict, List, Mapping, Optional, Union, cast
 from urllib.parse import parse_qs, urlencode
 
 from sentry_sdk._types import SENSITIVE_DATA_SUBSTITUTE
 
 if TYPE_CHECKING:
-    from typing import Any, Dict
-
     from sentry_sdk._types import (
         DataCollection,
         GenAICollectionBehaviour,
@@ -190,29 +188,33 @@ def _map_from_send_default_pii(
 
 def _resolve_explicit(
     d: "dict[str, Any]",
-    include_local_variables: bool,
-    include_source_context: bool,
 ) -> "DataCollection":
     """
     Build a fully-resolved ``DataCollection`` from a user-supplied
     ``data_collection`` dict, filling in spec defaults for any omitted or
-    partially-specified field. Frame fields fall back to the legacy
-    ``include_local_variables`` / ``include_source_context`` options when unset.
+    partially-specified field.
     """
     # frame_context_lines accepts an integer or a boolean fallback (spec: True
     # -> platform default of 5, False -> 0). bool is a subclass of int, so
     # coerce explicitly before treating it as a line count.
     frame_context_lines = d.get("frame_context_lines")
     if frame_context_lines is None:
-        frame_context_lines = (
-            _DEFAULT_FRAME_CONTEXT_LINES if include_source_context else 0
-        )
+        frame_context_lines = _DEFAULT_FRAME_CONTEXT_LINES
     elif isinstance(frame_context_lines, bool):
         frame_context_lines = _DEFAULT_FRAME_CONTEXT_LINES if frame_context_lines else 0
+    else:
+        if not isinstance(frame_context_lines, int) or frame_context_lines < 0:
+            raise ValueError(
+                "Invalid `frame_context_lines` value: Must be 0 or greater."
+            )
+
+    raw_stack_frame_variables = d.get("stack_frame_variables", True)
+    stack_frame_variables: "Union[bool, KeyValueCollectionBehaviour]"
 
-    stack_frame_variables = d.get("stack_frame_variables")
-    if stack_frame_variables is None:
-        stack_frame_variables = include_local_variables
+    if isinstance(raw_stack_frame_variables, dict):
+        stack_frame_variables = _kvcb_from_value(raw_stack_frame_variables)
+    else:
+        stack_frame_variables = bool(raw_stack_frame_variables)
 
     # http_bodies: omitted means "all valid types"; [] is the explicit opt-out.
     http_bodies = d.get("http_bodies")
@@ -323,8 +325,6 @@ def _resolve_data_collection(options: "Dict[str, Any]") -> "DataCollection":
             )
         return _resolve_explicit(
             user_dc,
-            include_local_variables,
-            include_source_context,
         )
 
     return _map_from_send_default_pii(

sentry_sdk/integrations/aiohttp.py

@@ -561,6 +561,7 @@ def aiohttp_processor(
         if request is None:
             return event
 
+        client_options = sentry_sdk.get_client().options
         with capture_internal_exceptions():
             request_info = event.setdefault("request", {})
 
@@ -578,7 +579,16 @@ def aiohttp_processor(
             # Just attach raw data here if it is within bounds, if available.
             # Unfortunately there's no way to get structured data from aiohttp
             # without awaiting on some coroutine.
-            request_info["data"] = get_aiohttp_request_data(request)
+            if has_data_collection_enabled(client_options):
+                if (
+                    "incoming_request"
+                    in client_options["data_collection"]["http_bodies"]
+                ):
+                    request_info["data"] = get_aiohttp_request_data(request)
+            else:
+                # We never gated this prior to data collection, so it should be attached
+                # when data collection is not enabled.
+                request_info["data"] = get_aiohttp_request_data(request)
 
         return event
 

sentry_sdk/integrations/aws_lambda.py

@@ -453,6 +453,11 @@ def event_processor(
                 ip = identity.get("sourceIp")
                 if ip is not None:
                     user_info.setdefault("ip_address", ip)
+
+            if "incoming_request" in client_options["data_collection"]["http_bodies"]:
+                if "body" in aws_event:
+                    request["data"] = aws_event.get("body", "")
+
         elif should_send_default_pii():
             user_info = sentry_event.setdefault("user", {})
 

sentry_sdk/integrations/dramatiq.py

@@ -17,6 +17,7 @@
     AnnotatedValue,
     capture_internal_exceptions,
     event_from_exception,
+    has_data_collection_enabled,
 )
 
 R = TypeVar("R")
@@ -241,10 +242,17 @@ def extract_into_event(self, event: "Event") -> None:
         request_info = contexts.setdefault("dramatiq", {})
         request_info["type"] = "dramatiq"
 
-        data: "Optional[Union[AnnotatedValue, Dict[str, Any]]]" = None
-        if not request_body_within_bounds(client, self.content_length()):
-            data = AnnotatedValue.removed_because_over_size_limit()
-        else:
-            data = self.message_data
+        attach_request_body = True
+        if has_data_collection_enabled(client.options):
+            attach_request_body = (
+                "incoming_request" in client.options["data_collection"]["http_bodies"]
+            )
+
+        if attach_request_body:
+            data: "Optional[Union[AnnotatedValue, Dict[str, Any]]]" = None
+            if not request_body_within_bounds(client, self.content_length()):
+                data = AnnotatedValue.removed_because_over_size_limit()
+            else:
+                data = self.message_data
 
-        request_info["data"] = data
+            request_info["data"] = data

sentry_sdk/integrations/fastapi.py

@@ -9,7 +9,7 @@
 from sentry_sdk.traces import StreamedSpan, get_current_span
 from sentry_sdk.tracing import SOURCE_FOR_STYLE, TransactionSource
 from sentry_sdk.tracing_utils import has_span_streaming_enabled
-from sentry_sdk.utils import transaction_from_function
+from sentry_sdk.utils import has_data_collection_enabled, transaction_from_function
 
 if TYPE_CHECKING:
     from typing import Any, Awaitable, Callable, Dict
@@ -122,7 +122,15 @@ def event_processor(event: "Event", hint: "Dict[str, Any]") -> "Event":
                 if "cookies" in info:
                     request_info["cookies"] = info["cookies"]
                 if "data" in info:
-                    request_info["data"] = info["data"]
+                    attach_request_data = True
+                    if has_data_collection_enabled(client.options):
+                        attach_request_data = (
+                            "incoming_request"
+                            in client.options["data_collection"]["http_bodies"]
+                        )
+
+                    if attach_request_data:
+                        request_info["data"] = info["data"]
             event["request"] = deepcopy(request_info)
 
             return event
@@ -140,14 +148,22 @@ def event_processor(event: "Event", hint: "Dict[str, Any]") -> "Event":
         current_span = get_current_span()
 
         if type(current_span) is StreamedSpan:
-            request_body = _get_cached_request_body_attribute(
-                client=client, request=request
-            )
-            if request_body:
-                current_span._segment.set_attribute(
-                    SPANDATA.HTTP_REQUEST_BODY_DATA,
-                    request_body,
+            attach_request_data = True
+            if has_data_collection_enabled(client.options):
+                attach_request_data = (
+                    "incoming_request"
+                    in client.options["data_collection"]["http_bodies"]
+                )
+
+            if attach_request_data:
+                request_body = _get_cached_request_body_attribute(
+                    client=client, request=request
                 )
+                if request_body:
+                    current_span._segment.set_attribute(
+                        SPANDATA.HTTP_REQUEST_BODY_DATA,
+                        request_body,
+                    )
 
 
 def patch_get_request_handler() -> None:

sentry_sdk/integrations/gcp.py

@@ -246,9 +246,10 @@ def event_processor(event: "Event", hint: "Hint") -> "Optional[Event]":
         if hasattr(gcp_event, "method"):
             request["method"] = gcp_event.method
 
+        client_options = sentry_sdk.get_client().options
+
         if hasattr(gcp_event, "query_string"):
             query_string = gcp_event.query_string.decode("utf-8", errors="replace")
-            client_options = sentry_sdk.get_client().options
             if has_data_collection_enabled(client_options):
                 if query_string:
                     filtered_qs = _apply_data_collection_filtering_to_query_string(
@@ -263,11 +264,16 @@ def event_processor(event: "Event", hint: "Hint") -> "Optional[Event]":
         if hasattr(gcp_event, "headers"):
             request["headers"] = _filter_headers(gcp_event.headers)
 
-        if should_send_default_pii():
-            if hasattr(gcp_event, "data"):
+        if hasattr(gcp_event, "data"):
+            if has_data_collection_enabled(client_options):
+                if (
+                    "incoming_request"
+                    in client_options["data_collection"]["http_bodies"]
+                ):
+                    request["data"] = gcp_event.data
+            elif should_send_default_pii():
                 request["data"] = gcp_event.data
-        else:
-            if hasattr(gcp_event, "data"):
+            else:
                 # Unfortunately couldn't find a way to get structured body from GCP
                 # event. Meaning every body is unstructured to us.
                 request["data"] = AnnotatedValue.removed_because_raw_data()

sentry_sdk/integrations/litestar.py

@@ -321,16 +321,24 @@ async def handle_wrapper(
         def event_processor(event: "Event", _: "Hint") -> "Event":
             request_info = event.get("request", {})
             request_info["content_length"] = len(scope.get("_body", b""))
+            should_attach_request_body = True
+
             if has_data_collection_enabled(client.options):
                 cookies = _apply_key_value_collection_filtering(
                     items=extracted_request_data["cookies"],
                     behaviour=client.options["data_collection"]["cookies"],
                 )
                 if cookies:
                     request_info["cookies"] = cookies
+
+                should_attach_request_body = (
+                    "incoming_request"
+                    in client.options["data_collection"]["http_bodies"]
+                )
             elif should_send_default_pii():
                 request_info["cookies"] = extracted_request_data["cookies"]
-            if request_data is not None:
+
+            if request_data is not None and should_attach_request_body:
                 request_info["data"] = request_data
 
             event["request"] = deepcopy(request_info)

sentry_sdk/integrations/starlette.py

@@ -562,7 +562,15 @@ def event_processor(event: "Event", hint: "Dict[str, Any]") -> "Event":
                 if "cookies" in info:
                     request_info["cookies"] = info["cookies"]
                 if "data" in info:
-                    request_info["data"] = info["data"]
+                    attach_request_data = True
+                    if has_data_collection_enabled(client.options):
+                        attach_request_data = (
+                            "incoming_request"
+                            in client.options["data_collection"]["http_bodies"]
+                        )
+
+                    if attach_request_data:
+                        request_info["data"] = info["data"]
             event["request"] = deepcopy(request_info)
 
             return event
@@ -580,15 +588,23 @@ def event_processor(event: "Event", hint: "Dict[str, Any]") -> "Event":
         current_span = get_current_span()
 
         if type(current_span) is StreamedSpan:
-            request_body = _get_cached_request_body_attribute(
-                client=client, request=request
-            )
-            if request_body:
-                current_span._segment.set_attribute(
-                    SPANDATA.HTTP_REQUEST_BODY_DATA,
-                    request_body,
+            attach_request_data = True
+            if has_data_collection_enabled(client.options):
+                attach_request_data = (
+                    "incoming_request"
+                    in client.options["data_collection"]["http_bodies"]
                 )
 
+            if attach_request_data:
+                request_body = _get_cached_request_body_attribute(
+                    client=client, request=request
+                )
+                if request_body:
+                    current_span._segment.set_attribute(
+                        SPANDATA.HTTP_REQUEST_BODY_DATA,
+                        request_body,
+                    )
+
 
 def patch_request_response() -> None:
     old_request_response = starlette.routing.request_response
@@ -827,21 +843,24 @@ async def json(self: "StarletteRequestExtractor") -> "Optional[Dict[str, Any]]":
             return None
 
 
-def _transaction_name_from_router(scope: "StarletteScope") -> "Optional[str]":
+def _transaction_name_and_source_from_router(
+    scope: "StarletteScope",
+) -> "Tuple[Optional[str], TransactionSource]":
     router = scope.get("router")
     if not router:
-        return None
+        return None, TransactionSource.ROUTE
 
     for route in router.routes:
         match = route.matches(scope)
         if match[0] == Match.FULL:
             try:
-                return route.path
+                return route.path, TransactionSource.ROUTE
             except AttributeError:
-                # routes added via app.host() won't have a path attribute
-                return scope.get("path")
+                # Host routes have no path template, so fall back to the
+                # concrete request path and classify it as a URL.
+                return scope.get("path"), TransactionSource.URL
 
-    return None
+    return None, TransactionSource.ROUTE
 
 
 def _set_transaction_name_and_source(
@@ -856,7 +875,7 @@ def _set_transaction_name_and_source(
             name = transaction_from_function(endpoint) or None
 
     elif transaction_style == "url":
-        name = _transaction_name_from_router(request.scope)
+        name, source = _transaction_name_and_source_from_router(request.scope)
 
     if name is None:
         name = _DEFAULT_TRANSACTION_NAME
@@ -877,7 +896,6 @@ def _get_transaction_from_middleware(
         name = transaction_from_function(app.__class__)
         source = TransactionSource.COMPONENT
     elif integration.transaction_style == "url":
-        name = _transaction_name_from_router(asgi_scope)
-        source = TransactionSource.ROUTE
+        name, source = _transaction_name_and_source_from_router(asgi_scope)
 
     return name, source