fix(bedrock): keep dotted model names in cross-region profile IDs (us.openai.gpt-5.6-sol)

[kimnamu]

Thanks for maintaining the Bedrock instrumentation and for the ARN / cross-region parsing added in #2785, which this touches. The prefix-less branch of _cross_region_check already splits with split(".", 1); this limits the geo-prefixed branch the same way, so a dot inside the model name survives.

Fixes #4501

Change (__init__.py, 1 line): parts = value.split(".") becomes value.split(".", 2), which yields exactly prefix, vendor and model.

Before / After (real converse calls in us-east-1; the ARN, no-prefix and eu. rows come from the unit tests)

Model ID	Before (main d13e721)	After
us.openai.gpt-5.6-sol	❌ gpt-5, span chat gpt-5	✅ gpt-5.6-sol, span chat gpt-5.6-sol
us.openai.gpt-5.6-luna	❌ gpt-5, span chat gpt-5	✅ gpt-5.6-luna, span chat gpt-5.6-luna
us.xai.grok-4.6	❌ grok-4, span chat grok-4	✅ grok-4.6, span chat grok-4.6
arn:...:inference-profile/us.openai.gpt-5.6-sol	❌ gpt-5	✅ gpt-5.6-sol
openai.gpt-5.6-sol (no prefix)	✅ gpt-5.6-sol	✅ unchanged
us.openai.gpt-6-sol / -luna / -astra	✅ gpt-6-sol / gpt-6-luna / gpt-6-astra	✅ unchanged
eu.anthropic.claude-3-7-sonnet-..., other IDs	✅	✅ unchanged, 377 existing tests pass

This is two lines below the prefix list my open #4235 extends with global. On its own #4235 would record global.openai.gpt-5.6-sol as gpt-5 too. The branches merge without conflict; merged, the suite passes (386) and global.openai.gpt-5.6-sol resolves to gpt-5.6-sol.

Tests: new tests/test_model_id_parsing.py, 4 cases (profile ID, profile ARN, prefix-less ID, undotted IDs).

RED / GREEN, full suite, lint

Source reverted, new tests kept:

E       AssertionError: assert ('aws.bedrock...nai', 'gpt-5') == ('aws.bedrock...'gpt-5.6-sol')
E         At index 2 diff: 'gpt-5' != 'gpt-5.6-sol'
FAILED tests/test_model_id_parsing.py::TestDottedModelName::test_regional_profile_keeps_full_model_name
FAILED tests/test_model_id_parsing.py::TestDottedModelName::test_regional_profile_arn_keeps_full_model_name
2 failed, 2 passed, 1 warning in 0.08s

Restored:

$ uv run pytest tests/test_model_id_parsing.py
4 passed, 1 warning in 0.01s
$ uv run pytest tests/ --record-mode=none
381 passed, 139 warnings in 6.34s
$ uv run ruff check .
All checks passed!

• 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.

Summary by CodeRabbit

• Bug Fixes
  • Improved parsing of Bedrock model identifiers when model names contain dots, including cross-region and regional inference-profile identifiers. Identifiers without dotted model names continue to be handled as before.

[coderabbitai]

Navigate logical layers of code changes, visualize relationships, and explore their blast radius.

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 4e6b71ac-4c38-4d5f-8c60-bc4c181fe324

📥 Commits

Reviewing files that changed from the base of the PR and between d13e721 and 743508e.

📒 Files selected for processing (2)

• packages/opentelemetry-instrumentation-bedrock/opentelemetry/instrumentation/bedrock/__init__.py
• packages/opentelemetry-instrumentation-bedrock/tests/test_model_id_parsing.py

Included review availability: Your plan provides up to 8 included reviews per hour; 6 remain after this review.

---

📝 Walkthrough

Walkthrough

The Bedrock cross-region model ID parser now preserves dots in model names. New tests cover regional IDs, inference-profile ARNs, prefix-less IDs, and undotted model names.

Changes

Bedrock model ID parsing

Layer / File(s)	Summary
Preserve dotted model names packages/opentelemetry-instrumentation-bedrock/opentelemetry/instrumentation/bedrock/__init__.py, packages/opentelemetry-instrumentation-bedrock/tests/test_model_id_parsing.py	_cross_region_check limits splitting to the first two dots, retaining the remaining model name. Tests check dotted names in regional IDs and ARNs, prefix-less IDs, and undotted regional IDs.

Priority: ➖ Normal

Estimated code review effort: 2 (Simple) | ~8 minutes

Change: Bug fix · Severity of issue fixed: Medium

Merge Risk: ⚪ Minimal · up to 74350

The change addresses truncated dotted model names while preserving the inspected existing ID forms. No actionable merge risk is evident.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	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 5 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	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: preserving dotted model names in Bedrock cross-region profile IDs.
Linked Issues check	✅ Passed	The change satisfies issue #4501. _cross_region_check now uses value.split(".", 2), so geo-prefixed IDs preserve dots in the model name. The added tests cover us.openai.gpt-5.6-sol, its inferenc…
Out of Scope Changes check	✅ Passed	The reported changes are limited to the Bedrock model ID parsing fix and focused tests for issue #4501. No unrelated behavior or files are identified in the pull request summary.

• Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create a new PR

---

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