[google-genai] Restore caller context between stream reads

[JacksonWeber]

Description

Move invocation suspension into the shared SyncStreamWrapper and AsyncStreamWrapper constructors, removing duplicate Google GenAI and tool-wrapper calls. Keep spans current only during stream reads and cleanup, with shared sync/async regressions and 24 Google GenAI API cases covering generation, Interactions model/agent calls, and fetch streams.

Part of #872. This shared fix also applies to other invocation-backed stream wrappers. The separate stream-timing metrics fix is not included.

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How has this been tested?

Dependencies were restored with pip after uv TLS failures. The Python 3.10 environment was manually pinned to the declared dependency floors, including Google GenAI 1.32.0. Disabling suspension makes all eight shared constructor regression cases fail.

• Package tests, focused lower-bound regressions, utility typechecking, API compatibility, and pre-commit checks passed

# Passed: 1261 tests, 240 skipped, 27 subtests
.\.venv\Scripts\python.exe -m pytest util\opentelemetry-util-genai\tests instrumentation\opentelemetry-instrumentation-google-genai\tests --ignore=instrumentation\opentelemetry-instrumentation-google-genai\tests\test_conformance.py --import-mode=importlib --vcr-record=none -o addopts= -o log_cli=false -q --tb=short

# Passed: 134 tests, 18 skipped because Interactions is unavailable
.\.tox\py310-test-instrumentation-google-genai-oldest\Scripts\python.exe -m pytest instrumentation\opentelemetry-instrumentation-google-genai\tests\test_stream_context.py util\opentelemetry-util-genai\tests\test_stream.py --import-mode=importlib --vcr-record=none -o addopts= -o log_cli=false -q --tb=short

# Passed: 0 errors
python -m pyright --pythonpath .venv\Scripts\python.exe util\opentelemetry-util-genai\src

# Passed
python -m tox -e check-util-api --runner virtualenv

# Passed
python -m tox -e precommit --runner virtualenv

Full-repository typechecking and conformance were not rerun for this follow-up.

Checklist

See CONTRIBUTING.md
for the style guide, changelog guidance, and more.

• Followed the style guidelines of this project
• Changelog updated if the change requires an entry
• Unit tests added
• Documentation updated

[opentelemetry-pr-dashboard]

Pull request dashboard status

Waiting on maintainers · refreshed 2026-10-17 01:45 UTC

Merge when ready.

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.
• Anything look wrong? Report it with what you expected; it helps us improve the dashboard.

[Copilot]

🟢 Approval recommended

The focused changes follow the shared stream lifecycle contract and include comprehensive regression coverage, with no identified blockers.

0 open findings

What changed in this PR

Restores caller context between Google GenAI stream reads while preserving invocation context during iteration and cleanup.

Changes:

• Suspends invocations in all six sync/async stream wrappers.
• Adds 72 lifecycle regression cases and includes them in the SDK v2 test environment.
• Documents the fix in a changelog fragment.

File	Description
tox.ini	Runs context regression tests against SDK v2.
instrumentation/​opentelemetry-instrumentation-google-genai/​tests/​test_stream_context.py	Covers context restoration across stream lifecycle paths.
instrumentation/​opentelemetry-instrumentation-google-genai/​src/​opentelemetry/​instrumentation/​google_genai/​interactions.py	Suspends Interactions and fetch stream invocations.
instrumentation/​opentelemetry-instrumentation-google-genai/​src/​opentelemetry/​instrumentation/​google_genai/​generate_content.py	Suspends sync/async generation stream invocations.
instrumentation/​opentelemetry-instrumentation-google-genai/​.changelog/​899.fixed	Records the context-restoration fix.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[DylanRussell]

why does simply calling invocation.suspend in the initializer of these stream wrappers require such a massive test ? Can we have a simpler test or can you explain why this is necessary ?

[JacksonWeber]

I’ve trimmed it to context restoration for sync and async calls. Tests should be simpler now, let me know if they look good now

[DylanRussell]

its better but i dont love all the branching logic.. maybe split it in 2 1 for sync 1 with async ? I think that'd get rid of a lot of it

[JacksonWeber]

Split it into separate sync and async tests, should remove the branching logic