[Eapp1e]

What

_abuild_from_streaming_response accumulates the text of every streamed chunk into complete_response, but it passes the async generator object itself to handle_streaming_response:

async for item in response:
    complete_response += str(item.text)
    ...
handle_streaming_response(span, event_logger, llm_model, response, token_usage)

set_response_attributes then writes that object into gen_ai.completion.0.content. The span ends up holding a memory address (<async_generator object ... at 0x...>) instead of the model output, and the SDK logs Invalid type <class 'async_generator'> for attribute value on every call. With events enabled (use_legacy_attributes=False) the async path emits no gen_ai.choice event at all, because emit_response_events only handles str and GenerationResponse.

The sync sibling _build_from_streaming_response already does it right — it passes complete_response (lines 159-161).

How

Pass the accumulated text, matching the sync path:

handle_streaming_response(
    span, event_logger, llm_model, complete_response, token_usage
)

Tests

New tests/test_streaming_attributes.py drives _abuild_from_streaming_response with an in-memory AsyncGenerator of chunks and asserts the exported span:

• gen_ai.completion.0.content == "Hello world!" (before the fix this is the generator repr),
• gen_ai.completion.0.role == "assistant".

Config.use_legacy_attributes is pinned with monkeypatch so the assertion is mode-independent. The whole package suite passes locally: 18 passed.

The expectation matches tests/disabled_test_bison.py:134, which asserts gen_ai.completion.0.content == "".join(response) for the same async streaming path.

Prepared with the help of an AI assistant.

Summary by CodeRabbit

• Bug Fixes
  • Streaming responses now record the accumulated completion text rather than an async-generator representation.
  • Completion attributes now include the assistant role, improving the accuracy of streaming response details in telemetry.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[coderabbitai]

No actionable comments were generated in the recent review.

Recent review info

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: defaults
• Review profile: CHILL
• Plan: Advanced
• Run ID: b8ac6195-bcc8-4f39-8f2f-c2b96a578cfb

Commits

Reviewing files that changed from the base of the PR and between 8d9c93f and e44efe9.

Files selected for processing: 2

• packages/opentelemetry-instrumentation-vertexai/opentelemetry/instrumentation/vertexai/__init__.py
• packages/opentelemetry-instrumentation-vertexai/tests/test_streaming_attributes.py

Included review availability: This review used your included allowance. Your plan provides up to 8 included reviews per hour; 7 remain after this review.

Walkthrough

The async Vertex AI streaming path now passes accumulated response text to handle_streaming_response. New tests check that streamed completion content is concatenated and that the recorded completion role is "assistant".

Changes: Vertex AI async streaming attributes

Layer / File(s)	Summary
Pass accumulated text and verify attributes packages/opentelemetry-instrumentation-vertexai/opentelemetry/instrumentation/vertexai/__init__.py, packages/opentelemetry-instrumentation-vertexai/tests/test_streaming_attributes.py	The async streaming path passes accumulated response text to handle_streaming_response. Tests check concatenated completion content and the "assistant" role.

Priority: Normal

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Bug fix

Suggested reviewers: luizdmm

Merge Risk: Minimal · up to e44ef

The async streaming fix and its focused attribute tests are mergeable with no identified actionable risk.

Pre-merge checks: 4 · 1

Failed checks: 1 warning

Check name	Status	Explanation	Resolution
Docstring Coverage		Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 7 functions across 2 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

Passed checks: 4 passed

Check name	Status	Explanation
Description Check		Check skipped - CodeRabbit’s high-level summary is enabled.
Title check		The title clearly and concisely describes the main change: recording accumulated text for asynchronous Vertex AI streaming responses.
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.

Finishing Touches

🧪 Generate unit tests (beta)

• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}

[Eapp1e]

This project requires signing the CLA, and I cannot complete that on behalf of my account, so I am closing this PR rather than leaving it blocked. The analysis and patch stay available in the branch if someone able to sign wants to take them over.