fix(langchain): handle early LangGraph stream closure

[dk3yyyy]

• I have added tests that cover my changes.
• If adding a new instrumentation or changing an existing one, I've added screenshots from some observability platform showing the change.
• PR name follows conventional commits format: feat(instrumentation): ... or fix(instrumentation): ....
• (If applicable) I have updated the documentation accordingly. (Not applicable: this bug fix does not change documented user-facing behavior.)

Summary

• treat GeneratorExit from early LangGraph.astream() termination as normal control flow
• finish the workflow and graph spans without error status or exception events
• preserve error recording and propagation for genuine application exceptions
• safely detach context tokens finalized from a copied async context without emitting misleading traceback logs

Tests

• npx nx run opentelemetry-instrumentation-langchain:lint
• npx nx run opentelemetry-instrumentation-langchain:test
• focused early-return and genuine-error regressions on Python 3.11 and Python 3.12

Closes #4385

Summary by CodeRabbit

• Bug Fixes
  • Improved tracing for asynchronous streaming workflows that end early, preventing normal termination from being reported as an error.
  • Ensured genuine errors during streaming are correctly marked as failures and include exception details.
  • Improved context cleanup to avoid misleading detach-related errors during asynchronous operations.
• Tests
  • Added coverage for early stream termination and error recording in asynchronous workflows.

[CLAassistant]

All committers have signed the CLA.

[coderabbitai]

📝 Walkthrough

Walkthrough

LangGraph async streaming instrumentation now treats GeneratorExit as successful termination, safely handles context detachment, and continues recording ordinary stream exceptions. Regression tests cover early-return and failing-node astream executions.

Changes

LangGraph async stream handling

Layer / File(s)	Summary
Async stream control flow packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/patch.py, packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/callback_handler.py	GeneratorExit ends spans successfully without error recording; ordinary exceptions retain error status and exception recording; graph and flow contexts use safe detachment.
Async stream regression coverage packages/opentelemetry-instrumentation-langchain/tests/test_langgraph.py	Tests verify early-return streams remain unset without exception events and failing streams mark workflow and graph spans as errors.

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

Suggested reviewers: dvirski

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 35.71% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check	✅ Passed	The changes address #4385 by treating GeneratorExit as non-error while preserving genuine exception recording.
Out of Scope Changes check	✅ Passed	The added context-safe detach and regression tests are directly related to the LangGraph stream-closure fix.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the main change: treating early LangGraph stream closure as normal control flow.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[karthik2oo4-git]

Could you please take a look at this PR? It looks correct to me and seems to address the issue I reported. It’s been open for a few days, so it would be great if someone could review it. Thanks!

[karthik2oo4-git]

any update on this PR?