fix(portkey): do not drain generator content and capture image parts by SAMurai-16 · Pull Request #843 · open-telemetry/opentelemetry-python-genai

SAMurai-16 · 2026-10-03T13:07:44Z

Description

In _prepare_input_messages (portkey/utils.py), iterating any Iterable message content consumed generators and one-shot iterators before the wrapped Portkey SDK call executed, leaving the downstream provider request with empty content (same class of issue as [genai-openai] With content capture enabled, generator/iterator messages and tools are consumed and the request is sent empty #825).
In addition, _prepare_input_messages previously only extracted type: "text" parts; type: "image_url" items were dropped rather than being recorded as UriPart or BlobPart.

To resolve this:

Only walk materialized Sequences (excluding str, bytes, and bytearray) in _prepare_input_messages so generator content is left unconsumed for the SDK.
Map image_url parts to UriPart (for remote URLs) and BlobPart (for inline base64 data: URLs) via image_from_url.
Support both dictionary parts and object instances with property access (SimpleNamespace/SDK models).
Add unit and integration tests covering generator content and image parts.

Fixes #829

Type of change

Bug fix (non-breaking change which fixes an issue)

How has this been tested?

Added unit and integration tests in instrumentation/opentelemetry-instrumentation-genai-portkey/tests/test_chat_completions.py:

test_chat_completions_generator_content_not_consumed: verifies the caller's generator remains unconsumed while span attributes record parts: [].
test_chat_completions_image_parts: verifies remote URLs map to UriPart and inline data: URLs decode to BlobPart.
test_prepare_input_messages_*: unit tests covering generators, iterators, byte sequence handling, image URLs, and object models.

Executed test suite:

pytest instrumentation/opentelemetry-instrumentation-genai-portkey/tests -v (210 passed, 3 skipped)
Pre-commit lint and format checks (ruff, ruff-format passed)

Checklist

Followed the style guidelines of this project
Changelog updated if the change requires an entry (will add once PR number is assigned)
Unit tests added

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

opentelemetry-pr-dashboard · 2026-10-03T13:09:29Z

Pull request dashboard status

Waiting on the author · refreshed 2026-11-01 08:52 UTC

Respond to 1 review item (e.g. link a commit, explain why not, ask a follow-up):

Inline threads: 1

Status above doesn't look right?

Just replied or pushed? Anything around or after the refresh time above may not be picked up yet — give it a few minutes.
Should this be with reviewers? Comment /dashboard route:reviewers to route it to them.
Anything wrong — including the routing? Report it with what you expected; it helps us improve the dashboard.

- Only walk materialized Sequences in _prepare_input_messages to leave generator content unconsumed for the SDK - Map image_url parts to UriPart and BlobPart via image_from_url - Add unit and integration tests covering generator content and image parts

…right

…ai dep

SAMurai-16 · 2026-10-21T11:09:16Z

I bumped the opentelemetry-util-genai dependency floor to 1.3b0.dev and updated the test mocks so they properly handle the legacy HTTP _post interface in portkey-ai==1.0.0 alongside the modern openai_client wrapper, making sure both the oldest and latest CI test environments pass cleanly.

…or-content-and-image-parts # Conflicts: # README.md

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

sangkyoonnam

I checked 3348b084 and a local merge of it with main at a408e14e (which has #890), using an HTTP transport stub and portkey-ai 2.3.4. With content capture on, main sends "messages": [] when messages is a generator or a map(), including the async, stream=True and HTTP 400 cases. This PR sends both input messages in each case and records them in gen_ai.input.messages. The request now also carries every item from a nested content generator and from an assistant tool_calls generator. With capture off, main and this PR send the same request and record no message content.

Merging main at a408e14e conflicts in README.md and tests/requirements.oldest.txt. Main's util-genai is now 1.4b0.dev, so deps-check treats the 1.3b0.dev floor as an earlier release cycle and rejects the -e install of util-genai in the oldest requirements. On a408e14e, adding that line makes scripts/check_deps.py exit 1 with that error. The suggestion below removes it, and with it applied the file merges cleanly and keeps main's typing-extensions==4.7.1 pin. Apply it together with the main merge; the unmerged PR branch still needs the line. I haven't run the oldest env after this change.

sangkyoonnam · 2026-10-28T05:33:38Z

+
+-e util/opentelemetry-util-genai


Remove this when merging main; deps-check on main rejects it.

Suggested change

-e util/opentelemetry-util-genai

Thanks for testing this thoroughly!
I have merged upstream/main into this branch, removed -e util/opentelemetry-util-genai from tests/requirements.oldest.txt, and kept the typing-extensions==4.7.1 pin.

…or-content-and-image-parts # Conflicts: # README.md # instrumentation/opentelemetry-instrumentation-genai-portkey/tests/requirements.oldest.txt

SAMurai-16 · 2026-10-28T20:29:12Z

@lmolkova i believe this PR is ready for review.

lmolkova · 2026-10-29T04:21:58Z

@@ -0,0 +1 @@
+Do not consume generator message ``content`` before the SDK call, and capture ``image_url`` parts as image parts.


The PR now also stops draining messages/tools/tool_calls generators and records audio, file and refusal parts - could the entry mention those?

Suggested change

Do not consume generator message ``content`` before the SDK call, and capture ``image_url`` parts as image parts.

Do not consume generator ``messages``, ``tools``, ``tool_calls`` or message ``content`` before the SDK call, and capture ``image_url``, ``input_audio``, ``file`` and ``refusal`` content parts.

The PR description could use the same update - it still describes walking only Sequences and lists test_prepare_input_messages_* tests that are no longer there.

lmolkova · 2026-10-29T04:21:58Z

+        if hasattr(p.chat.completions, "openai_client"):
+            call_args = p.chat.completions.openai_client.with_raw_response.chat.completions.create.call_args
+            received_messages = call_args.kwargs["messages"]
+        else:
+            call_args = p.chat.completions._post.call_args
+            body = (
+                call_args.kwargs.get("body")
+                if isinstance(call_args.kwargs.get("body"), dict)
+                else call_args.kwargs
+            )
+            received_messages = body["messages"]


nit: this "what did the SDK get" block is repeated in several new tests. A small helper would shorten them:

def _sent_messages(client): completions = client.chat.completions if hasattr(completions, "openai_client"): create = completions.openai_client.with_raw_response.chat.completions.create return create.call_args.kwargs["messages"] call_args = completions._post.call_args body = call_args.kwargs.get("body") return (body if isinstance(body, dict) else call_args.kwargs)["messages"]

lmolkova · 2026-10-29T04:45:26Z

+            received_messages = body["messages"]
+        received_content = received_messages[0]["content"]
+        assert not isinstance(received_content, list)
+        assert isinstance(received_content, Iterator)


A used-up generator is still an Iterator, so this check (and the same one in generator_messages_and_tools_captured and nested_generators_captured) passes on main too, where the content is drained before the SDK call. Could _consume_chat_request keep what it read, so the tests check the SDK got the full content?

_sent: list[Any] = [] def _consume_chat_request(*args: Any, **kwargs: Any) -> None: ... _sent.append([{**m, "content": list(m["content"])} for m in messages])

assert _sent[-1][0]["content"] == [ {"type": "text", "text": "item 0"}, {"type": "text", "text": "item 1"}, ]

…angelog

Copilot AI balanced review requested due to automatic review settings October 3, 2026 13:07

SAMurai-16 requested a review from a team as a code owner October 3, 2026 13:07

github-project-automation Bot added this to GenAI Semantic Conventions and Instrumentation libraries Oct 3, 2026

Copilot AI reviewed Oct 3, 2026

opentelemetry-pr-dashboard Bot mentioned this pull request Oct 3, 2026

Pull Request Dashboard #171

Open

SAMurai-16 force-pushed the fix/portkey-generator-content-and-image-parts branch from 201b09b to a68881c Compare October 3, 2026 17:36

chore(changelog): add changelog fragment for PR open-telemetry#843

4242760

lmolkova reviewed Oct 6, 2026

View reviewed changes

Comment thread ...metry-instrumentation-genai-portkey/src/opentelemetry/instrumentation/genai/portkey/utils.py Resolved

Comment thread ...metry-instrumentation-genai-portkey/src/opentelemetry/instrumentation/genai/portkey/utils.py Resolved

This comment has been minimized.

Sign in to view

SAMurai-16 and others added 5 commits October 16, 2026 17:43

fix(portkey): cover input_audio, file, and refusal content parts

42b7f62

fix(portkey): materialize generator messages and tools before SDK call

592109f

style(portkey): clean up unnecessary comments in utils.py

7c61066

Merge branch 'main' into fix/portkey-generator-content-and-image-parts

eaee4c6

fix(portkey): restore import sys and cast iterable in patch.py for py…

9986728

…right

SAMurai-16 requested a review from lmolkova October 16, 2026 18:55

lmolkova reviewed Oct 20, 2026

View reviewed changes

Comment thread ...metry-instrumentation-genai-portkey/src/opentelemetry/instrumentation/genai/portkey/patch.py Resolved

Comment thread ...metry-instrumentation-genai-portkey/src/opentelemetry/instrumentation/genai/portkey/utils.py Resolved

SAMurai-16 and others added 3 commits October 21, 2026 13:14

Merge branch 'main' into fix/portkey-generator-content-and-image-parts

6317e27

fix(portkey): use IteratorCapture for generator messages and tools

7510ac1

fix(portkey): support legacy client in oldest tests and bump util-gen…

bb2287e

…ai dep

SAMurai-16 requested a review from lmolkova October 21, 2026 11:09

lmolkova reviewed Oct 22, 2026

View reviewed changes

SAMurai-16 added 3 commits October 23, 2026 09:33

fix(portkey): simplify message extraction and add generator error tests

7efed6b

chore: regenerate instrumentation readme

80cfc27

Merge remote-tracking branch 'upstream/main' into fix/portkey-generat…

3348b08

…or-content-and-image-parts # Conflicts: # README.md

SAMurai-16 requested a balanced review from Copilot October 23, 2026 13:05

Copilot AI reviewed Oct 23, 2026

SAMurai-16 requested a review from lmolkova October 23, 2026 13:12

sangkyoonnam reviewed Oct 28, 2026

View reviewed changes

DylanRussell mentioned this pull request Oct 28, 2026

fix(portkey): do not consume generator content and map image parts (#829) #988

Open

2 of 4 tasks

Merge remote-tracking branch 'upstream/main' into fix/portkey-generat…

2359c7c

…or-content-and-image-parts # Conflicts: # README.md # instrumentation/opentelemetry-instrumentation-genai-portkey/tests/requirements.oldest.txt

lmolkova approved these changes Oct 29, 2026

View reviewed changes

test(portkey): verify full generator contents reach SDK and update ch…

ab09814

…angelog

SAMurai-16 requested a review from lmolkova November 1, 2026 09:52

		@@ -0,0 +1 @@
		Do not consume generator message ``content`` before the SDK call, and capture ``image_url`` parts as image parts.

	Do not consume generator message ``content`` before the SDK call, and capture ``image_url`` parts as image parts.
	Do not consume generator ``messages``, ``tools``, ``tool_calls`` or message ``content`` before the SDK call, and capture ``image_url``, ``input_audio``, ``file`` and ``refusal`` content parts.

Conversation

SAMurai-16 commented Oct 3, 2026

Description

Type of change

How has this been tested?

Checklist

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

opentelemetry-pr-dashboard Bot commented Oct 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull request dashboard status

Uh oh!

Uh oh!

Uh oh!

This comment has been minimized.

Uh oh!

Uh oh!

SAMurai-16 commented Oct 21, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

sangkyoonnam left a comment

Choose a reason for hiding this comment

Uh oh!

sangkyoonnam Oct 28, 2026

Choose a reason for hiding this comment

Uh oh!

SAMurai-16 Oct 28, 2026

Choose a reason for hiding this comment

Uh oh!

SAMurai-16 commented Oct 28, 2026

Uh oh!

lmolkova Oct 29, 2026

Choose a reason for hiding this comment

Uh oh!

lmolkova Oct 29, 2026

Choose a reason for hiding this comment

Uh oh!

lmolkova Oct 29, 2026

Choose a reason for hiding this comment

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

opentelemetry-pr-dashboard Bot commented Oct 3, 2026 •

edited

Loading