ref(wsgi): Drop send_default_pii support

sentry_sdk/integrations/_asgi_common.py

@@ -4,7 +4,7 @@
 
 import sentry_sdk
 from sentry_sdk.data_collection import _apply_data_collection_filtering_to_query_string
-from sentry_sdk.integrations._wsgi_common import _filter_headers
+from sentry_sdk.integrations._wsgi_common import _filter_headers_legacy
 from sentry_sdk.scope import should_send_default_pii
 from sentry_sdk.utils import has_data_collection_enabled
 
@@ -121,7 +121,7 @@ def _get_request_data(
 
         headers = _get_headers(asgi_scope)
 
-        request_data["headers"] = _filter_headers(
+        request_data["headers"] = _filter_headers_legacy(
             headers,
             use_annotated_value=False,
         )
@@ -175,7 +175,7 @@ def _get_request_attributes(
 
         headers = _get_headers(asgi_scope)
 
-        filtered_headers = _filter_headers(headers, use_annotated_value=False)
+        filtered_headers = _filter_headers_legacy(headers, use_annotated_value=False)
         for header, value in filtered_headers.items():
             attributes[f"http.request.header.{header.lower()}"] = value
 

sentry_sdk/integrations/_wsgi_common.py

@@ -23,6 +23,18 @@
     from sentry_sdk._types import Event
 
 
+DEFAULT_HTTP_METHODS_TO_CAPTURE = (
+    "CONNECT",
+    "DELETE",
+    "GET",
+    # "HEAD",  # do not capture HEAD requests by default
+    # "OPTIONS",  # do not capture OPTIONS requests by default
+    "PATCH",
+    "POST",
+    "PUT",
+    "TRACE",
+)
+
 SENSITIVE_ENV_KEYS = (
     "REMOTE_ADDR",
     "HTTP_X_FORWARDED_FOR",
@@ -39,18 +51,6 @@
     x[len("HTTP_") :] for x in SENSITIVE_ENV_KEYS if x.startswith("HTTP_")
 )
 
-DEFAULT_HTTP_METHODS_TO_CAPTURE = (
-    "CONNECT",
-    "DELETE",
-    "GET",
-    # "HEAD",  # do not capture HEAD requests by default
-    # "OPTIONS",  # do not capture OPTIONS requests by default
-    "PATCH",
-    "POST",
-    "PUT",
-    "TRACE",
-)
-
 
 def request_body_within_bounds(
     client: "Optional[sentry_sdk.client.BaseClient]", content_length: int
@@ -89,23 +89,15 @@ def extract_into_event(self, event: "Event") -> None:
         content_length = self.content_length()
         request_info = event.get("request", {})
 
-        # Prior to data collection being implemented we unconditionally attached
-        # the request body, which is why we default to True here.
-        attach_request_body = True
-
-        if has_data_collection_enabled(client.options):
-            cookies = _apply_key_value_collection_filtering(
-                items=dict(self.cookies()),
-                behaviour=client.options["data_collection"]["cookies"],
-            )
-            if cookies:
-                request_info["cookies"] = cookies
+        data_collection = client.options["data_collection"]
+        cookies = _apply_key_value_collection_filtering(
+            items=dict(self.cookies()),
+            behaviour=data_collection["cookies"],
+        )
+        if cookies:
+            request_info["cookies"] = cookies
 
-            attach_request_body = (
-                "incoming_request" in client.options["data_collection"]["http_bodies"]
-            )
-        elif should_send_default_pii():
-            request_info["cookies"] = dict(self.cookies())
+        attach_request_body = "incoming_request" in data_collection["http_bodies"]
 
         if attach_request_body:
             if not request_body_within_bounds(client, content_length):
@@ -210,6 +202,63 @@ def env(self) -> "Dict[str, Any]":
         raise NotImplementedError()
 
 
+class LegacyRequestExtractor(RequestExtractor):
+    def extract_into_event(self, event: "Event") -> None:
+        client = sentry_sdk.get_client()
+        if not client.is_active():
+            return
+
+        data: "Optional[Union[AnnotatedValue, Dict[str, Any]]]" = None
+        content_length = self.content_length()
+        request_info = event.get("request", {})
+
+        # Prior to data collection being implemented we unconditionally attached
+        # the request body, which is why we default to True here.
+        attach_request_body = True
+        if has_data_collection_enabled(client.options):
+            cookies = _apply_key_value_collection_filtering(
+                items=dict(self.cookies()),
+                behaviour=client.options["data_collection"]["cookies"],
+            )
+            if cookies:
+                request_info["cookies"] = cookies
+
+            attach_request_body = (
+                "incoming_request" in client.options["data_collection"]["http_bodies"]
+            )
+        elif should_send_default_pii():
+            request_info["cookies"] = dict(self.cookies())
+
+        if attach_request_body:
+            if not request_body_within_bounds(client, content_length):
+                data = AnnotatedValue.removed_because_over_size_limit()
+            else:
+                # First read the raw body data
+                # It is important to read this first because if it is Django
+                # it will cache the body and then we can read the cached version
+                # again in parsed_body() (or json() or wherever).
+                raw_data = None
+                try:
+                    raw_data = self.raw_data()
+                except _RAW_DATA_EXCEPTIONS:
+                    # If DjangoRestFramework is used it already read the body for us
+                    # so reading it here will fail. We can ignore this.
+                    pass
+
+                parsed_body = self.parsed_body()
+                if parsed_body is not None:
+                    data = parsed_body
+                elif raw_data:
+                    data = AnnotatedValue.removed_because_raw_data()
+                else:
+                    data = None
+
+        if data is not None:
+            request_info["data"] = data
+
+        event["request"] = deepcopy(request_info)
+
+
 def _is_json_content_type(ct: "Optional[str]") -> bool:
     mt = (ct or "").split(";", 1)[0]
     return (
@@ -222,12 +271,29 @@ def _is_json_content_type(ct: "Optional[str]") -> bool:
 def _filter_headers(
     headers: "Mapping[str, str]",
     use_annotated_value: bool = True,
+) -> "Mapping[str, Union[AnnotatedValue, str]]":
+    client = sentry_sdk.get_client()
+
+    filtered = _apply_key_value_collection_filtering(
+        items=headers,
+        behaviour=client.options["data_collection"]["http_headers"]["request"],
+    )
+
+    for key in filtered:
+        if isinstance(key, str) and key.lower() in ("cookie", "set-cookie"):
+            filtered[key] = SENSITIVE_DATA_SUBSTITUTE
+
+    return filtered
+
+
+def _filter_headers_legacy(
+    headers: "Mapping[str, str]",
+    use_annotated_value: bool = True,
 ) -> "Mapping[str, Union[AnnotatedValue, str]]":
     client_options = sentry_sdk.get_client().options
 
     if has_data_collection_enabled(client_options):
         data_collection_configuration = client_options["data_collection"]
-
         filtered = _apply_key_value_collection_filtering(
             items=headers,
             behaviour=data_collection_configuration["http_headers"]["request"],
@@ -238,6 +304,7 @@ def _filter_headers(
                 filtered[key] = SENSITIVE_DATA_SUBSTITUTE
 
         return filtered
+
     else:
         if should_send_default_pii():
             return headers

sentry_sdk/integrations/django/__init__.py

@@ -15,7 +15,7 @@
 )
 from sentry_sdk.integrations._wsgi_common import (
     DEFAULT_HTTP_METHODS_TO_CAPTURE,
-    RequestExtractor,
+    LegacyRequestExtractor,
 )
 from sentry_sdk.integrations.logging import ignore_logger_for_events
 from sentry_sdk.integrations.wsgi import SentryWsgiMiddleware
@@ -703,7 +703,7 @@ def sentry_patched_response_for_exception(
     exception_handler.response_for_exception = sentry_patched_response_for_exception
 
 
-class DjangoRequestExtractor(RequestExtractor):
+class DjangoRequestExtractor(LegacyRequestExtractor):
     def __init__(self, request: "Union[WSGIRequest, ASGIRequest]") -> None:
         try:
             drf_request = request._sentry_drf_request_backref()
@@ -747,7 +747,7 @@ def parsed_body(self) -> "Optional[Dict[str, Any]]":
         try:
             return self.request.data
         except Exception:
-            return RequestExtractor.parsed_body(self)
+            return LegacyRequestExtractor.parsed_body(self)
 
 
 def _set_user_info(request: "WSGIRequest", event: "Event") -> None:

sentry_sdk/integrations/flask.py

@@ -5,7 +5,7 @@
 from sentry_sdk.integrations import DidNotEnable, Integration, _check_minimum_version
 from sentry_sdk.integrations._wsgi_common import (
     DEFAULT_HTTP_METHODS_TO_CAPTURE,
-    RequestExtractor,
+    LegacyRequestExtractor,
 )
 from sentry_sdk.integrations.wsgi import SentryWsgiMiddleware
 from sentry_sdk.scope import should_send_default_pii
@@ -185,7 +185,7 @@ def _request_started(app: "Flask", **kwargs: "Any") -> None:
     scope.add_event_processor(evt_processor)
 
 
-class FlaskRequestExtractor(RequestExtractor):
+class FlaskRequestExtractor(LegacyRequestExtractor):
     def env(self) -> "Dict[str, str]":
         return self.request.environ
 

sentry_sdk/integrations/gcp.py

@@ -9,7 +9,7 @@
 from sentry_sdk.consts import OP
 from sentry_sdk.data_collection import _apply_data_collection_filtering_to_query_string
 from sentry_sdk.integrations import Integration
-from sentry_sdk.integrations._wsgi_common import _filter_headers
+from sentry_sdk.integrations._wsgi_common import _filter_headers_legacy
 from sentry_sdk.integrations.cloud_resource_context import CLOUD_PROVIDER
 from sentry_sdk.scope import Scope, should_send_default_pii
 from sentry_sdk.traces import SegmentNameSource
@@ -86,7 +86,7 @@ def sentry_func(
             header_attributes: "dict[str, Any]" = {}
             if hasattr(gcp_event, "headers"):
                 headers = gcp_event.headers
-                for header, header_value in _filter_headers(
+                for header, header_value in _filter_headers_legacy(
                     headers, use_annotated_value=False
                 ).items():
                     header_attributes[f"http.request.header.{header.lower()}"] = (
@@ -244,7 +244,7 @@ def event_processor(event: "Event", hint: "Hint") -> "Optional[Event]":
                 request["query_string"] = query_string
 
         if hasattr(gcp_event, "headers"):
-            request["headers"] = _filter_headers(gcp_event.headers)
+            request["headers"] = _filter_headers_legacy(gcp_event.headers)
 
         if hasattr(gcp_event, "data"):
             if has_data_collection_enabled(client_options):

sentry_sdk/integrations/quart.py

@@ -8,7 +8,7 @@
 from sentry_sdk.consts import SPANDATA
 from sentry_sdk.data_collection import _apply_data_collection_filtering_to_query_string
 from sentry_sdk.integrations import DidNotEnable, Integration
-from sentry_sdk.integrations._wsgi_common import _filter_headers
+from sentry_sdk.integrations._wsgi_common import _filter_headers_legacy
 from sentry_sdk.integrations.asgi import SentryAsgiMiddleware
 from sentry_sdk.scope import should_send_default_pii
 from sentry_sdk.traces import SOURCE_FOR_STYLE as SEGMENT_SOURCE_FOR_STYLE
@@ -187,7 +187,7 @@ async def _request_websocket_started(app: "Quart", **kwargs: "Any") -> None:
         segment.set_attribute("http.request.method", request_websocket.method)
         header_attributes: "dict[str, Any]" = {}
 
-        for header, header_value in _filter_headers(
+        for header, header_value in _filter_headers_legacy(
             dict(request_websocket.headers), use_annotated_value=False
         ).items():
             header_attributes[f"http.request.header.{header.lower()}"] = header_value
@@ -282,7 +282,7 @@ def inner(event: "Event", hint: "dict[str, Any]") -> "Event":
             request_info["url"] = request.url
             request_info["query_string"] = request.query_string
             request_info["method"] = request.method
-            request_info["headers"] = _filter_headers(dict(request.headers))
+            request_info["headers"] = _filter_headers_legacy(dict(request.headers))
 
             client_options = sentry_sdk.get_client().options
             if has_data_collection_enabled(client_options):

sentry_sdk/integrations/sanic.py

@@ -10,7 +10,10 @@
     _apply_data_collection_filtering_to_query_string,
 )
 from sentry_sdk.integrations import DidNotEnable, Integration, _check_minimum_version
-from sentry_sdk.integrations._wsgi_common import RequestExtractor, _filter_headers
+from sentry_sdk.integrations._wsgi_common import (
+    LegacyRequestExtractor,
+    _filter_headers_legacy,
+)
 from sentry_sdk.scope import should_send_default_pii
 from sentry_sdk.traces import SegmentNameSource
 from sentry_sdk.utils import (
@@ -64,7 +67,7 @@ def setup_once() -> None:
         _setup_sanic()
 
 
-class SanicRequestExtractor(RequestExtractor):
+class SanicRequestExtractor(LegacyRequestExtractor):
     def content_length(self) -> int:
         if self.request.body is None:
             return 0
@@ -251,7 +254,7 @@ def _get_request_attributes(request: "Request") -> "Dict[str, Any]":
     if request.method:
         attributes[SPANDATA.HTTP_REQUEST_METHOD] = request.method.upper()
 
-    headers = _filter_headers(dict(request.headers), use_annotated_value=False)
+    headers = _filter_headers_legacy(dict(request.headers), use_annotated_value=False)
     for header, value in headers.items():
         attributes[f"{SPANDATA.HTTP_REQUEST_HEADER}.{header.lower()}"] = value
 
@@ -340,7 +343,7 @@ def sanic_processor(event: "Event", hint: "Optional[Hint]") -> "Optional[Event]"
                 or client_options["data_collection"]["user_info"]
             ):
                 request_info["env"] = {"REMOTE_ADDR": request.remote_addr}
-            request_info["headers"] = _filter_headers(dict(request.headers))
+            request_info["headers"] = _filter_headers_legacy(dict(request.headers))
 
         return event
 

sentry_sdk/integrations/tornado.py

@@ -6,8 +6,8 @@
 from sentry_sdk.data_collection import _apply_data_collection_filtering_to_query_string
 from sentry_sdk.integrations import DidNotEnable, Integration, _check_minimum_version
 from sentry_sdk.integrations._wsgi_common import (
-    RequestExtractor,
-    _filter_headers,
+    LegacyRequestExtractor,
+    _filter_headers_legacy,
     _is_json_content_type,
     request_body_within_bounds,
 )
@@ -147,7 +147,7 @@ def _get_request_attributes(request: "Any") -> "Dict[str, Any]":
     if request.method:
         attributes[SPANDATA.HTTP_REQUEST_METHOD] = request.method.upper()
 
-    headers = _filter_headers(dict(request.headers), use_annotated_value=False)
+    headers = _filter_headers_legacy(dict(request.headers), use_annotated_value=False)
     for header, value in headers.items():
         attributes[f"{SPANDATA.HTTP_REQUEST_HEADER}.{header.lower()}"] = value
 
@@ -280,7 +280,7 @@ def tornado_processor(event: "Event", hint: "dict[str, Any]") -> "Event":
                 or client_options["data_collection"]["user_info"]
             ):
                 request_info["env"] = {"REMOTE_ADDR": request.remote_ip}
-            request_info["headers"] = _filter_headers(dict(request.headers))
+            request_info["headers"] = _filter_headers_legacy(dict(request.headers))
 
         if has_data_collection_enabled(client_options):
             if client_options["data_collection"]["user_info"]:
@@ -305,7 +305,7 @@ def tornado_processor(event: "Event", hint: "dict[str, Any]") -> "Event":
     return tornado_processor
 
 
-class TornadoRequestExtractor(RequestExtractor):
+class TornadoRequestExtractor(LegacyRequestExtractor):
     def content_length(self) -> int:
         if self.request.body is None:
             return 0