[Boulea7]

What does this PR do?

When an OpenAI stream yields partial output and then raises, record the generation as ERROR with the exception message. Keep partial output and usage, re-raise the original exception, and finalize once across native and fallback sync/async streams. Normal exhaustion, early closing, and cancellation retain their existing behavior.

Fixes #2039.

Type of change

• Bug fix
• New feature
• Breaking change
• Refactor
• Documentation update
• Tooling, CI, or repo maintenance

Verification

python -m pytest -p pytest_asyncio.plugin -p pytest_timeout tests/unit/test_openai.py tests/unit/test_openai_prompt_extraction.py -q --timeout=30
ruff check .
ruff format --check langfuse/openai.py tests/unit/test_openai.py
mypy langfuse --no-error-summary

68 tests passed; Ruff and mypy passed. The SSE regression tests use the real OpenAI client with httpx.MockTransport and an in-memory exporter. Verified with locked OpenAI 2.29.0; live-provider tests were not run.

Checklist

• I self-reviewed the diff using code_review.md.
• I added or updated tests for behavior changes.
• I updated docs, examples, or .env.template if needed. (Not applicable.)
• I did not hand-edit generated files; if generated files changed, I used the upstream regeneration path.
• I did not commit secrets or credentials.

Confidence Score: 5/5

This PR appears safe to merge; no actionable issues were found.

Summary

Records OpenAI stream iteration failures as ERROR with the exception message.

• OpenAI streams record errors without losing the response chunks they already yielded.

No actionable issues were found. Tests were inspected but not run during this review.

_{Reviews (1) · Last reviewed commit: "fix(openai): record stream iteration err..." · Reviewed by Greptile}

[claude]

Claude Code Review

This pull request is from a fork — automated review is disabled. A repository maintainer can comment @claude review to run a one-time review.