Merge branch 'major/3.0' into ivana/major/remove-data-collection-from-experiments

Author: sentrivana

MIGRATION_GUIDE.md

@@ -164,6 +164,8 @@ Looking to upgrade from Sentry SDK 2.x to 3.x? Here's a comprehensive list of wh
 - `add_attachment` no longer accepts an `add_to_transactions` argument.
 - The `stream_gen_ai_spans` option was removed. All spans are streamed now.
 - The `data_collection` option is no longer accessible under `_experiments`. Use it as a top-level option instead.
+- The experimental `record_sql_params` option was removed. Use the `database_query_data` setting of `data_collection` instead.
+- The `record_params` option of `AsyncPGIntegration` and `AioMySQLIntegration` was removed. Use the `database_query_data` setting of `data_collection` instead.
 
 
 ## Deprecated

sentry_sdk/client.py

@@ -24,7 +24,6 @@
 )
 from sentry_sdk.envelope import Envelope, Item
 from sentry_sdk.integrations import setup_integrations
-from sentry_sdk.integrations.dedupe import DedupeIntegration
 from sentry_sdk.monitor import Monitor
 from sentry_sdk.profiler.continuous_profiler import setup_continuous_profiler
 from sentry_sdk.serializer import serialize
@@ -594,13 +593,6 @@ def _prepare_event(
                     )
                     self.transport.record_lost_event(reason, data_category="error")
 
-                # If this is an exception, reset the DedupeIntegration. It still
-                # remembers the dropped exception as the last exception, meaning
-                # that if the same exception happens again and is not dropped
-                # in before_send, it'd get dropped by DedupeIntegration.
-                if event.get("exception"):
-                    DedupeIntegration.reset_last_seen()
-
             event = new_event
 
         return event

sentry_sdk/consts.py

@@ -65,7 +65,6 @@ class CompressionAlgo(Enum):
         "Experiments",
         {
             "max_flags": Optional[int],
-            "record_sql_params": Optional[bool],
             "continuous_profiling_auto_start": Optional[bool],
             "transport_zlib_compression_level": Optional[int],
             "transport_compression_level": Optional[int],
@@ -575,6 +574,12 @@ class SPANDATA:
     Example: "qa-pipeline"
     """
 
+    GEN_AI_PROMPT_NAME = "gen_ai.prompt.name"
+    """
+    The name of the prompt that uniquely identifies it.
+    Example: "summarize_text"
+    """
+
     GEN_AI_RESPONSE_FINISH_REASONS = "gen_ai.response.finish_reasons"
     """
     The reason why the model stopped generating.
@@ -1086,12 +1091,18 @@ class SPANDATA:
 
     MCP_TOOL_NAME = "mcp.tool.name"
     """
+    .. deprecated::
+        This attribute is deprecated. Use GEN_AI_TOOL_NAME instead.
+
     The name of the MCP tool being called.
     Example: "get_weather"
     """
 
     MCP_PROMPT_NAME = "mcp.prompt.name"
     """
+    .. deprecated::
+        This attribute is deprecated. Use GEN_AI_PROMPT_NAME instead.
+
     The name of the MCP prompt being retrieved.
     Example: "code_review"
     """
@@ -1116,6 +1127,9 @@ class SPANDATA:
 
     MCP_TOOL_RESULT_CONTENT = "mcp.tool.result.content"
     """
+    .. deprecated::
+        This attribute is deprecated. Use GEN_AI_TOOL_CALL_RESULT instead.
+
     The result/output content from an MCP tool execution.
     Example: "The weather is sunny"
     """

sentry_sdk/integrations/aiomysql.py

@@ -13,7 +13,6 @@
 )
 from sentry_sdk.utils import (
     capture_internal_exceptions,
-    has_data_collection_enabled,
     parse_version,
 )
 
@@ -28,10 +27,6 @@
 class AioMySQLIntegration(Integration):
     identifier = "aiomysql"
     origin = f"auto.db.{identifier}"
-    _record_params = False
-
-    def __init__(self, *, record_params: bool = False):
-        AioMySQLIntegration._record_params = record_params
 
     @staticmethod
     def setup_once() -> None:
@@ -84,12 +79,7 @@ async def _inner(*args: Any, **kwargs: Any) -> T:
         integration = sentry_sdk.get_client().get_integration(AioMySQLIntegration)
 
         client = sentry_sdk.get_client()
-        should_record_params = False
-        if has_data_collection_enabled(client.options):
-            if client.options["data_collection"]["database_query_data"]:
-                should_record_params = True
-        else:
-            should_record_params = integration._record_params if integration else False
+        should_record_params = client.options["data_collection"]["database_query_data"]
 
         params_list = params if integration and should_record_params else None
         param_style = "pyformat" if params_list else None
@@ -134,12 +124,7 @@ async def _inner(*args: Any, **kwargs: Any) -> T:
 
         integration = sentry_sdk.get_client().get_integration(AioMySQLIntegration)
         client = sentry_sdk.get_client()
-        should_record_params = False
-        if has_data_collection_enabled(client.options):
-            if client.options["data_collection"]["database_query_data"]:
-                should_record_params = True
-        else:
-            should_record_params = integration._record_params if integration else False
+        should_record_params = client.options["data_collection"]["database_query_data"]
 
         params_list = seq_of_params if integration and should_record_params else None
         param_style = "pyformat" if params_list else None

sentry_sdk/integrations/asyncpg.py

@@ -32,10 +32,6 @@
 class AsyncPGIntegration(Integration):
     identifier = "asyncpg"
     origin = f"auto.db.{identifier}"
-    _record_params = False
-
-    def __init__(self, *, record_params: bool = False):
-        AsyncPGIntegration._record_params = record_params
 
     @staticmethod
     def setup_once() -> None:
@@ -114,7 +110,10 @@ def _record(
 ) -> "Iterator[Span]":
     client = sentry_sdk.get_client()
     integration = client.get_integration(AsyncPGIntegration)
-    if integration is not None and not integration._record_params:
+    if (
+        integration is not None
+        and not client.options["data_collection"]["database_query_data"]
+    ):
         params_list = None
 
     param_style = "pyformat" if params_list else None

sentry_sdk/integrations/dedupe.py

@@ -1,24 +1,19 @@
-import weakref
-from contextvars import ContextVar
 from typing import TYPE_CHECKING
 
 import sentry_sdk
 from sentry_sdk.integrations import Integration
 from sentry_sdk.scope import add_global_event_processor
-from sentry_sdk.utils import logger
+from sentry_sdk.utils import capture_internal_exceptions, logger
 
 if TYPE_CHECKING:
-    from typing import Any, Optional
+    from typing import Optional
 
     from sentry_sdk._types import Event, Hint
 
 
 class DedupeIntegration(Integration):
     identifier = "dedupe"
 
-    def __init__(self) -> None:
-        self._last_seen: "ContextVar[Any]" = ContextVar("last-seen")
-
     @staticmethod
     def setup_once() -> None:
         @add_global_event_processor
@@ -34,30 +29,12 @@ def processor(event: "Event", hint: "Optional[Hint]") -> "Optional[Event]":
             if exc_info is None:
                 return event
 
-            last_seen = integration._last_seen.get(None)
-            if last_seen is not None:
-                # last_seen is either a weakref or the original instance
-                last_seen = (
-                    last_seen() if isinstance(last_seen, weakref.ref) else last_seen
-                )
-
             exc = exc_info[1]
-            if last_seen is exc:
+
+            if getattr(exc, "_handled_by_sentry", False):
                 logger.info("DedupeIntegration dropped duplicated error event %s", exc)
                 return None
-
-            # we can only weakref non builtin types
-            try:
-                integration._last_seen.set(weakref.ref(exc))
-            except TypeError:
-                integration._last_seen.set(exc)
-
-            return event
-
-    @staticmethod
-    def reset_last_seen() -> None:
-        integration = sentry_sdk.get_client().get_integration(DedupeIntegration)
-        if integration is None:
-            return
-
-        integration._last_seen.set(None)
+            else:
+                with capture_internal_exceptions():
+                    exc._handled_by_sentry = True
+                return event

sentry_sdk/integrations/gcp.py

@@ -11,14 +11,13 @@
 from sentry_sdk.integrations import Integration
 from sentry_sdk.integrations._wsgi_common import _filter_headers
 from sentry_sdk.integrations.cloud_resource_context import CLOUD_PROVIDER
-from sentry_sdk.scope import Scope, should_send_default_pii
+from sentry_sdk.scope import Scope
 from sentry_sdk.traces import SegmentNameSource
 from sentry_sdk.utils import (
     AnnotatedValue,
     TimeoutThread,
     capture_internal_exceptions,
     event_from_exception,
-    has_data_collection_enabled,
     logger,
     reraise,
 )
@@ -98,17 +97,12 @@ def sentry_func(
             if hasattr(gcp_event, "query_string"):
                 query_string = gcp_event.query_string.decode("utf-8", errors="replace")
                 if query_string:
-                    if has_data_collection_enabled(client.options):
-                        filtered_qs = _apply_data_collection_filtering_to_query_string(
-                            query_string=query_string,
-                            behaviour=client.options["data_collection"][
-                                "url_query_params"
-                            ],
-                        )
-                        if filtered_qs:
-                            additional_attributes["url.query"] = filtered_qs
-                    elif should_send_default_pii():
-                        additional_attributes["url.query"] = query_string
+                    filtered_qs = _apply_data_collection_filtering_to_query_string(
+                        query_string=query_string,
+                        behaviour=client.options["data_collection"]["url_query_params"],
+                    )
+                    if filtered_qs:
+                        additional_attributes["url.query"] = filtered_qs
 
             sampling_context = {
                 "gcp_env": {
@@ -229,28 +223,19 @@ def event_processor(event: "Event", hint: "Hint") -> "Optional[Event]":
 
         if hasattr(gcp_event, "query_string"):
             query_string = gcp_event.query_string.decode("utf-8", errors="replace")
-            if has_data_collection_enabled(client_options):
-                if query_string:
-                    filtered_qs = _apply_data_collection_filtering_to_query_string(
-                        query_string=query_string,
-                        behaviour=client_options["data_collection"]["url_query_params"],
-                    )
-                    if filtered_qs:
-                        request["query_string"] = filtered_qs
-            else:
-                request["query_string"] = query_string
+            if query_string:
+                filtered_qs = _apply_data_collection_filtering_to_query_string(
+                    query_string=query_string,
+                    behaviour=client_options["data_collection"]["url_query_params"],
+                )
+                if filtered_qs:
+                    request["query_string"] = filtered_qs
 
         if hasattr(gcp_event, "headers"):
             request["headers"] = _filter_headers(gcp_event.headers)
 
         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():
+            if "incoming_request" in client_options["data_collection"]["http_bodies"]:
                 request["data"] = gcp_event.data
             else:
                 # Unfortunately couldn't find a way to get structured body from GCP

sentry_sdk/integrations/google_genai/utils.py

@@ -710,7 +710,7 @@ async def async_wrapped(*args: "Any", **kwargs: "Any") -> "Any":
                 tool_input = _capture_tool_input(args, kwargs, tool)
                 with capture_internal_exceptions():
                     span.set_attribute(
-                        SPANDATA.GEN_AI_TOOL_INPUT, safe_serialize(tool_input)
+                        SPANDATA.GEN_AI_TOOL_CALL_ARGUMENTS, safe_serialize(tool_input)
                     )
 
                 try:
@@ -719,7 +719,7 @@ async def async_wrapped(*args: "Any", **kwargs: "Any") -> "Any":
                     # Capture tool output
                     with capture_internal_exceptions():
                         span.set_attribute(
-                            SPANDATA.GEN_AI_TOOL_OUTPUT, safe_serialize(result)
+                            SPANDATA.GEN_AI_TOOL_CALL_RESULT, safe_serialize(result)
                         )
 
                     return result
@@ -737,7 +737,7 @@ def sync_wrapped(*args: "Any", **kwargs: "Any") -> "Any":
                 tool_input = _capture_tool_input(args, kwargs, tool)
                 with capture_internal_exceptions():
                     span.set_attribute(
-                        SPANDATA.GEN_AI_TOOL_INPUT, safe_serialize(tool_input)
+                        SPANDATA.GEN_AI_TOOL_CALL_ARGUMENTS, safe_serialize(tool_input)
                     )
 
                 try:
@@ -746,7 +746,7 @@ def sync_wrapped(*args: "Any", **kwargs: "Any") -> "Any":
                     # Capture tool output
                     with capture_internal_exceptions():
                         span.set_attribute(
-                            SPANDATA.GEN_AI_TOOL_OUTPUT, safe_serialize(result)
+                            SPANDATA.GEN_AI_TOOL_CALL_RESULT, safe_serialize(result)
                         )
 
                     return result

sentry_sdk/integrations/gql.py

@@ -1,10 +1,8 @@
 import sentry_sdk
 from sentry_sdk.integrations import DidNotEnable, Integration, _check_minimum_version
-from sentry_sdk.scope import should_send_default_pii
 from sentry_sdk.utils import (
     ensure_integration_enabled,
     event_from_exception,
-    has_data_collection_enabled,
     parse_version,
 )
 
@@ -61,10 +59,7 @@ def _data_from_document(document: "DocumentNode") -> "EventDataType":
         data: "EventDataType" = {"query": print_ast(document)}
 
         if operation_ast is not None:
-            if has_data_collection_enabled(client_options):
-                if client_options["data_collection"]["graphql"]["variables"]:
-                    data["variables"] = operation_ast.variable_definitions
-            elif should_send_default_pii():
+            if client_options["data_collection"]["graphql"]["variables"]:
                 data["variables"] = operation_ast.variable_definitions
 
             if operation_ast.name is not None:
@@ -155,29 +150,7 @@ def processor(event: "Event", hint: "dict[str, Any]") -> "Event":
         )
 
         client_options = sentry_sdk.get_client().options
-        if has_data_collection_enabled(client_options):
-            if client_options["data_collection"]["graphql"]["document"]:
-                if GraphQLRequest is not None and isinstance(
-                    document_or_request, GraphQLRequest
-                ):
-                    # In v4.0.0, gql moved to using GraphQLRequest instead of
-                    # DocumentNode in execute
-                    # https://github.com/graphql-python/gql/pull/556
-                    document = document_or_request.document
-                else:
-                    document = document_or_request
-
-                request["data"] = _data_from_document(document)
-                contexts = event.setdefault("contexts", {})
-                response = contexts.setdefault("response", {})
-                response.update(
-                    {
-                        "data": {"errors": errors},
-                        "type": response,
-                    }
-                )
-
-        elif should_send_default_pii():
+        if client_options["data_collection"]["graphql"]["document"]:
             if GraphQLRequest is not None and isinstance(
                 document_or_request, GraphQLRequest
             ):

sentry_sdk/integrations/langchain.py

@@ -735,7 +735,7 @@ def on_tool_start(
 
             set_data_normalized(
                 span,
-                SPANDATA.GEN_AI_TOOL_INPUT,
+                SPANDATA.GEN_AI_TOOL_CALL_ARGUMENTS,
                 kwargs.get("inputs", [input_str]),
             )
 
@@ -755,7 +755,7 @@ def on_tool_end(
             )
 
             if sentry_sdk.get_client().options["data_collection"]["gen_ai"]["outputs"]:
-                set_data_normalized(span, SPANDATA.GEN_AI_TOOL_OUTPUT, output)
+                set_data_normalized(span, SPANDATA.GEN_AI_TOOL_CALL_RESULT, output)
 
             self._exit_span(span, run_id)