[do not merge] feat: Span streaming & new span API

CHANGELOG.md

@@ -1,5 +1,88 @@
 # Changelog
 
+## 2.52.0a7
+
+### Bug Fixes 🐛
+
+#### Openai
+
+- Avoid consuming iterables passed to the Completions API by @alexander-alderman-webb in [#5489](https://github.com/getsentry/sentry-python/pull/5489)
+- Avoid consuming iterables passed to the Embeddings API by @alexander-alderman-webb in [#5491](https://github.com/getsentry/sentry-python/pull/5491)
+
+#### Other
+
+- (grpc) Read method from handler_call_details for grpcio >= 1.76 compat by @yeung108 in [#5521](https://github.com/getsentry/sentry-python/pull/5521)
+- (httpx) Correctly append baggage in async client by @sentrivana in [#5530](https://github.com/getsentry/sentry-python/pull/5530)
+- (pydantic-ai) Adapt to missing `ToolManager._call_tool` by @sentrivana in [#5522](https://github.com/getsentry/sentry-python/pull/5522)
+- (utils) Use HEROKU_BUILD_COMMIT env var for default release by @ericapisani in [#5499](https://github.com/getsentry/sentry-python/pull/5499)
+
+### Documentation 📚
+
+- Add debugging advice by @alexander-alderman-webb in [#5517](https://github.com/getsentry/sentry-python/pull/5517)
+
+### Internal Changes 🔧
+
+- (agents) Add security-review skill to agent configuration by @ericapisani in [#5498](https://github.com/getsentry/sentry-python/pull/5498)
+- (anthropic) Remove `set_data_normalized` for primitive attributes by @alexander-alderman-webb in [#5504](https://github.com/getsentry/sentry-python/pull/5504)
+- (openai-agents) Remove `set_data_normalized` for primitive attributes by @alexander-alderman-webb in [#5509](https://github.com/getsentry/sentry-python/pull/5509)
+- (pydantic-ai) Remove `set_data_normalized` for the `gen_ai.response.model` attribute by @alexander-alderman-webb in [#5512](https://github.com/getsentry/sentry-python/pull/5512)
+- 🤖 Update test matrix with new releases (02/24) by @github-actions in [#5524](https://github.com/getsentry/sentry-python/pull/5524)
+- 🤖 Update test matrix with new releases (02/23) by @github-actions in [#5503](https://github.com/getsentry/sentry-python/pull/5503)
+
+### Other
+
+- [do not merge] feat: Span streaming & new span API by @sentrivana in [#5317](https://github.com/getsentry/sentry-python/pull/5317)
+
+## 2.52.0a6
+
+### Bug Fixes 🐛
+
+#### Openai Agents
+
+- Patch `execute_final_output()` functions following library refactor by @alexander-alderman-webb in [#5453](https://github.com/getsentry/sentry-python/pull/5453)
+- Patch `execute_handoffs()` functions following library refactor by @alexander-alderman-webb in [#5452](https://github.com/getsentry/sentry-python/pull/5452)
+- Patch `run_single_turn_streamed()` functions following library refactor by @alexander-alderman-webb in [#5451](https://github.com/getsentry/sentry-python/pull/5451)
+- Patch `run_single_turn()` functions following library refactor by @alexander-alderman-webb in [#5450](https://github.com/getsentry/sentry-python/pull/5450)
+- Patch models functions following library refactor by @alexander-alderman-webb in [#5449](https://github.com/getsentry/sentry-python/pull/5449)
+- Patch tool functions following library refactor by @alexander-alderman-webb in [#5445](https://github.com/getsentry/sentry-python/pull/5445)
+
+#### Other
+
+- (anthropic) Fix token accounting by @shellmayr in [#5490](https://github.com/getsentry/sentry-python/pull/5490)
+- (google-genai) Remove agent spans for simple requests by @alexander-alderman-webb in [#5443](https://github.com/getsentry/sentry-python/pull/5443)
+
+### Documentation 📚
+
+- New integration guide by @alexander-alderman-webb in [#5476](https://github.com/getsentry/sentry-python/pull/5476)
+
+### Internal Changes 🔧
+
+#### Openai Agents
+
+- Expect new tool fields by @alexander-alderman-webb in [#5471](https://github.com/getsentry/sentry-python/pull/5471)
+- New tool field and library error log by @alexander-alderman-webb in [#5454](https://github.com/getsentry/sentry-python/pull/5454)
+
+#### Other
+
+- (agents) Add sentry skills to be used by warden in CI reviews by @ericapisani in [#5485](https://github.com/getsentry/sentry-python/pull/5485)
+- (ai) Add configuration for dotagents by @ericapisani in [#5480](https://github.com/getsentry/sentry-python/pull/5480)
+- (github) Add warden configuration by @ericapisani in [#5484](https://github.com/getsentry/sentry-python/pull/5484)
+- (repo) Add .serena to .gitignore by @ericapisani in [#5464](https://github.com/getsentry/sentry-python/pull/5464)
+- 🤖 Update test matrix with new releases (02/19) by @github-actions in [#5483](https://github.com/getsentry/sentry-python/pull/5483)
+- 🤖 Update test matrix with new releases (02/18) by @github-actions in [#5475](https://github.com/getsentry/sentry-python/pull/5475)
+- Use fixed clickhouse action, remove aws-sam-cli dependency by @sentrivana in [#5457](https://github.com/getsentry/sentry-python/pull/5457)
+
+### Other
+
+- [do not merge] feat: Span streaming & new span API by @sentrivana in [#5317](https://github.com/getsentry/sentry-python/pull/5317)
+- Update CHANGELOG.md by @alexander-alderman-webb in [f75a9ac1](https://github.com/getsentry/sentry-python/commit/f75a9ac1059b648644c05189deffd3c7ddc0931a)
+- Update CHANGELOG.md by @alexander-alderman-webb in [b700fa85](https://github.com/getsentry/sentry-python/commit/b700fa85d8d722891ad84cefaec73a9aeeaafa16)
+- release: 2.53.0 by @alexander-alderman-webb in [45379e20](https://github.com/getsentry/sentry-python/commit/45379e206c992d5f5ab782b1e13dc609cedbdd5f)
+
+## 2.52.0a5
+
+This is an alpha release for internal testing of span streaming.
+
 ## 2.53.0
 
 ### Bug Fixes 🐛
@@ -35,6 +118,76 @@
 - Remove references to unsupported attribute types by @alexander-alderman-webb in [#5425](https://github.com/getsentry/sentry-python/pull/5425)
 - Pin setuptools for linting and chalice tests by @alexander-alderman-webb in [#5438](https://github.com/getsentry/sentry-python/pull/5438)
 
+## 2.52.0a4
+
+- [do not merge] feat: Span streaming & new span API by @sentrivana in [#5317](https://github.com/getsentry/sentry-python/pull/5317)
+
+## 2.52.0a3
+
+### New Features ✨
+
+- feat(integration): add `gen_ai.conversation.id` if available by @constantinius in [#5307](https://github.com/getsentry/sentry-python/pull/5307)
+
+### Bug Fixes 🐛
+
+#### Mcp
+
+- fix(mcp): Nest MCP spans under HTTP transactions by @alexander-alderman-webb in [#5292](https://github.com/getsentry/sentry-python/pull/5292)
+- fix(mcp): Handle all awaitable return types by @alexander-alderman-webb in [#5415](https://github.com/getsentry/sentry-python/pull/5415)
+
+#### Span Streaming
+
+- fix(span-streaming): Always preserialize attributes by @sentrivana in [#5407](https://github.com/getsentry/sentry-python/pull/5407)
+- fix(span-streaming): Warn about thread usage if any batcher is active by @sentrivana in [#5408](https://github.com/getsentry/sentry-python/pull/5408)
+
+#### Other
+
+- fix(anthropic): Token reporting by @alexander-alderman-webb in [#5403](https://github.com/getsentry/sentry-python/pull/5403)
+- fix(celery): Close the connection we're reading driver_type from by @sentrivana in [#5427](https://github.com/getsentry/sentry-python/pull/5427)
+- fix(google-genai): Token reporting by @alexander-alderman-webb in [#5404](https://github.com/getsentry/sentry-python/pull/5404)
+- fix(openai): Token reporting by @alexander-alderman-webb in [#5406](https://github.com/getsentry/sentry-python/pull/5406)
+- fix(openai-agents): Inject propagation headers for `HostedMCPTool` when streaming by @alexander-alderman-webb in [#5405](https://github.com/getsentry/sentry-python/pull/5405)
+- fix: Fix list attribute type by @sentrivana in [#5417](https://github.com/getsentry/sentry-python/pull/5417)
+- fix: Adapt to new packaging in toxgen by @sentrivana in [#5382](https://github.com/getsentry/sentry-python/pull/5382)
+
+### Internal Changes 🔧
+
+#### Fastmcp
+
+- test(fastmcp): Wrap prompt in `Message` by @alexander-alderman-webb in [#5411](https://github.com/getsentry/sentry-python/pull/5411)
+- test(fastmcp): Remove `test_fastmcp_without_request_context()` by @alexander-alderman-webb in [#5412](https://github.com/getsentry/sentry-python/pull/5412)
+- test(fastmcp): Use `AsyncClient` for SSE by @alexander-alderman-webb in [#5400](https://github.com/getsentry/sentry-python/pull/5400)
+- test(fastmcp): Use `TestClient` for Streamable HTTP by @alexander-alderman-webb in [#5384](https://github.com/getsentry/sentry-python/pull/5384)
+- test(fastmcp): Simulate stdio transport with memory streams by @alexander-alderman-webb in [#5333](https://github.com/getsentry/sentry-python/pull/5333)
+
+#### Mcp
+
+- test(mcp): Use `AsyncClient` for SSE by @alexander-alderman-webb in [#5396](https://github.com/getsentry/sentry-python/pull/5396)
+- test(mcp): Use `TestClient` for Streamable HTTP by @alexander-alderman-webb in [#5383](https://github.com/getsentry/sentry-python/pull/5383)
+- test(mcp): Remove unused stdio helpers by @alexander-alderman-webb in [#5409](https://github.com/getsentry/sentry-python/pull/5409)
+
+#### Other
+
+- ci: Pin setuptools for linting and chalice tests by @alexander-alderman-webb in [#5438](https://github.com/getsentry/sentry-python/pull/5438)
+- ci: Fix lint step by @sentrivana in [#5418](https://github.com/getsentry/sentry-python/pull/5418)
+- ci: 🤖 Update test matrix with new releases (02/02) by @github-actions in [#5413](https://github.com/getsentry/sentry-python/pull/5413)
+- ci: Update tox and pin packaging version for tox by @alexander-alderman-webb in [#5381](https://github.com/getsentry/sentry-python/pull/5381)
+
+### Other
+
+- [do not merge] feat: Span streaming & new span API by @sentrivana in [#5317](https://github.com/getsentry/sentry-python/pull/5317)
+- Update CHANGELOG.md by @alexander-alderman-webb in [bcec5780](https://github.com/getsentry/sentry-python/commit/bcec57806fcb8ef0aa228052aeddf10272df43ba)
+- release: 2.52.0 by @alexander-alderman-webb in [3d2154ce](https://github.com/getsentry/sentry-python/commit/3d2154ce57447e6717f7e38b2ea0d1d27b4ff940)
+- Revert "feat(ai): Add original input length meta attribute (#5375)" by @alexander-alderman-webb in [#5419](https://github.com/getsentry/sentry-python/pull/5419)
+
+## 2.52.0a2
+
+This is an alpha release for internal testing of span streaming.
+
+## 2.51.0a1
+
+This is an alpha release for internal testing of span streaming.
+
 ## 2.52.0
 
 ### New Features ✨

docs/conf.py

@@ -31,7 +31,7 @@
 copyright = "2019-{}, Sentry Team and Contributors".format(datetime.now().year)
 author = "Sentry Team and Contributors"
 
-release = "2.53.0"
+release = "2.52.0a7"
 version = ".".join(release.split(".")[:2])  # The short X.Y version.
 
 

requirements-linting.txt

@@ -7,6 +7,8 @@ types-greenlet
 types-redis
 types-setuptools
 types-webob
+types-channels
+types-uWSGI
 opentelemetry-distro[otlp]
 pymongo # There is no separate types module.
 loguru # There is no separate types module.

sentry_sdk/_compat.py

@@ -37,7 +37,7 @@ def check_uwsgi_thread_support() -> bool:
     #    --py-call-uwsgi-fork-hooks to force uWSGI to run regular cpython
     #    after-fork hooks that take care of cleaning up stale thread data.
     try:
-        from uwsgi import opt  # type: ignore
+        from uwsgi import opt
     except ImportError:
         return True
 
@@ -50,11 +50,11 @@ def enabled(option: str) -> bool:
 
         if isinstance(value, bytes):
             try:
-                value = value.decode()
+                value = value.decode()  # type: ignore[assignment]
             except Exception:
                 pass
 
-        return value and str(value).lower() not in FALSE_VALUES
+        return bool(value and str(value).lower() not in FALSE_VALUES)
 
     # When `threads` is passed in as a uwsgi option,
     # `enable-threads` is implied on.

sentry_sdk/_span_batcher.py

@@ -1,24 +1,23 @@
+import json
 import threading
 from collections import defaultdict
 from datetime import datetime, timezone
 from typing import TYPE_CHECKING
 
 from sentry_sdk._batcher import Batcher
-from sentry_sdk.consts import SPANSTATUS
 from sentry_sdk.envelope import Envelope, Item, PayloadRef
-from sentry_sdk.utils import format_timestamp, serialize_attribute, safe_repr
+from sentry_sdk.utils import format_timestamp, serialize_attribute
 
 if TYPE_CHECKING:
     from typing import Any, Callable, Optional
     from sentry_sdk.traces import StreamedSpan
-    from sentry_sdk._types import SerializedAttributeValue
 
 
 class SpanBatcher(Batcher["StreamedSpan"]):
-    # TODO[span-first]: size-based flushes
     # TODO[span-first]: adjust flush/drop defaults
     MAX_BEFORE_FLUSH = 1000
-    MAX_BEFORE_DROP = 5000
+    MAX_BEFORE_DROP = 1000
+    MAX_BYTES_BEFORE_FLUSH = 5 * 1024 * 1024  # 5 MB
     FLUSH_WAIT_TIME = 5.0
 
     TYPE = "span"
@@ -35,6 +34,7 @@
         # envelope.
         # trace_id -> span buffer
         self._span_buffer: dict[str, list["StreamedSpan"]] = defaultdict(list)
+        self._running_size: dict[str, int] = defaultdict(lambda: 0)
         self._capture_func = capture_func
         self._record_lost_func = record_lost_func
         self._running = True
@@ -64,18 +64,39 @@
                 return None
 
             self._span_buffer[span.trace_id].append(span)
+
             if size + 1 >= self.MAX_BEFORE_FLUSH:
                 self._flush_event.set()
+                return
+
+            self._running_size[span.trace_id] += self._estimate_size(span)
+            if self._running_size[span.trace_id] >= self.MAX_BYTES_BEFORE_FLUSH:
+                self._flush_event.set()
+                return
+
+    @staticmethod
+    def _estimate_size(item: "StreamedSpan") -> int:
+        # This is just a quick approximation
+        span_dict = SpanBatcher._to_transport_format(item)
+        return len(str(span_dict))
 
     @staticmethod
     def _to_transport_format(item: "StreamedSpan") -> "Any":
-        # TODO[span-first]
         res: "dict[str, Any]" = {
+            "trace_id": item.trace_id,
             "span_id": item.span_id,
             "name": item._name,
             "status": item._status,
+            "is_segment": item.is_segment(),
+            "start_timestamp": item.start_timestamp.timestamp(),
         }
 
+        if item.timestamp:
+            res["end_timestamp"] = item.timestamp.timestamp()
+
+        if item.parent_span_id:
+            res["parent_span_id"] = item.parent_span_id
+
         if item._attributes:
             res["attributes"] = {
                 k: serialize_attribute(v) for (k, v) in item._attributes.items()
@@ -89,11 +110,9 @@
                 return None
 
             envelopes = []
-            for trace_id, spans in self._span_buffer.items():
+            for spans in self._span_buffer.values():
                 if spans:
-                    # TODO[span-first]
-                    # dsc = spans[0].dynamic_sampling_context()
-                    dsc = None
+                    dsc = spans[0].dynamic_sampling_context()
 
                     envelope = Envelope(
                         headers={
@@ -123,6 +142,7 @@
                     envelopes.append(envelope)
 
             self._span_buffer.clear()
+            self._running_size.clear()
 
         for envelope in envelopes:
             self._capture_func(envelope)

sentry_sdk/_types.py

@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING, TypeVar, Union
+from typing import TYPE_CHECKING, Pattern, TypeVar, Union
 
 
 # Re-exported for compat, since code out there in the wild might use this variable.
@@ -360,3 +360,11 @@ class SDKInfo(TypedDict):
     class TextPart(TypedDict):
         type: Literal["text"]
         content: str
+
+    IgnoreSpansName = Union[str, Pattern[str]]
+    IgnoreSpansContext = TypedDict(
+        "IgnoreSpansContext",
+        {"name": IgnoreSpansName, "attributes": Attributes},
+        total=False,
+    )
+    IgnoreSpansConfig = list[Union[IgnoreSpansName, IgnoreSpansContext]]

sentry_sdk/ai/monitoring.py

@@ -6,6 +6,7 @@
 import sentry_sdk.utils
 from sentry_sdk import start_span
 from sentry_sdk.tracing import Span
+from sentry_sdk.traces import StreamedSpan
 from sentry_sdk.utils import ContextVar, reraise, capture_internal_exceptions
 
 from typing import TYPE_CHECKING
@@ -97,7 +98,7 @@ async def async_wrapped(*args: "Any", **kwargs: "Any") -> "Any":
 
 
 def record_token_usage(
-    span: "Span",
+    span: "Union[Span, StreamedSpan]",
     input_tokens: "Optional[int]" = None,
     input_tokens_cached: "Optional[int]" = None,
     input_tokens_cache_write: "Optional[int]" = None,
@@ -106,30 +107,35 @@ def record_token_usage(
     total_tokens: "Optional[int]" = None,
 ) -> None:
     # TODO: move pipeline name elsewhere
+    if isinstance(span, StreamedSpan):
+        set_on_span = span.set_attribute
+    else:
+        set_on_span = span.set_data
+
     ai_pipeline_name = get_ai_pipeline_name()
     if ai_pipeline_name:
-        span.set_data(SPANDATA.GEN_AI_PIPELINE_NAME, ai_pipeline_name)
+        set_on_span(SPANDATA.GEN_AI_PIPELINE_NAME, ai_pipeline_name)
 
     if input_tokens is not None:
-        span.set_data(SPANDATA.GEN_AI_USAGE_INPUT_TOKENS, input_tokens)
+        set_on_span(SPANDATA.GEN_AI_USAGE_INPUT_TOKENS, input_tokens)
 
     if input_tokens_cached is not None:
-        span.set_data(
+        set_on_span(
             SPANDATA.GEN_AI_USAGE_INPUT_TOKENS_CACHED,
             input_tokens_cached,
         )
 
     if input_tokens_cache_write is not None:
-        span.set_data(
+        set_on_span(
             SPANDATA.GEN_AI_USAGE_INPUT_TOKENS_CACHE_WRITE,
             input_tokens_cache_write,
         )
 
     if output_tokens is not None:
-        span.set_data(SPANDATA.GEN_AI_USAGE_OUTPUT_TOKENS, output_tokens)
+        set_on_span(SPANDATA.GEN_AI_USAGE_OUTPUT_TOKENS, output_tokens)
 
     if output_tokens_reasoning is not None:
-        span.set_data(
+        set_on_span(
             SPANDATA.GEN_AI_USAGE_OUTPUT_TOKENS_REASONING,
             output_tokens_reasoning,
         )
@@ -138,4 +144,4 @@ def record_token_usage(
         total_tokens = input_tokens + output_tokens
 
     if total_tokens is not None:
-        span.set_data(SPANDATA.GEN_AI_USAGE_TOTAL_TOKENS, total_tokens)
+        set_on_span(SPANDATA.GEN_AI_USAGE_TOTAL_TOKENS, total_tokens)