fix(test): mock headers in test_completion_fine_tuned_model

[joereyna]

Summary

• mock_response = AsyncMock() left headers as an AsyncMock attribute
• raw_response.headers.get("x-gemini-service-tier") returned a truthy AsyncMock, so the walrus operator assigned it to service_tier
• service_tier.lower() then failed with AttributeError: 'coroutine' object has no attribute 'lower'
• Fix: set mock_response.headers = {} so the header lookup returns None and the service tier branch is skipped

Test plan

• test_completion_fine_tuned_model passes in CI (requires real GCP credentials)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Ready	Preview, Comment	Apr 9, 2026 10:30pm

[greptile-apps]

Greptile Summary

This PR fixes a spurious AttributeError in test_completion_fine_tuned_model caused by AsyncMock auto-creating a truthy AsyncMock object for headers, which then flowed into a .lower() call on what the code expected to be a string. The fix sets mock_response.headers = {} and hoists mock_response.status_code = 200 earlier, making the mock faithfully represent a real HTTP response. The remaining whitespace-only removals are cosmetic cleanup.

Confidence Score: 5/5

Safe to merge — fixes a broken test mock with no production code changes.

The only functional change is setting headers = {} on an AsyncMock to prevent a spurious AttributeError; all test assertions remain unchanged and test coverage is not weakened. The rest of the diff is cosmetic whitespace removal.

No files require special attention.

Important Files Changed

Filename	Overview
tests/local_testing/test_amazing_vertex_completion.py	Fixes broken AsyncMock by explicitly setting headers = {} and status_code = 200 upfront; remaining changes are cosmetic blank-line removals.

Sequence Diagram

sequenceDiagram
    participant Test as test_completion_fine_tuned_model
    participant LiteLLM as litellm.acompletion
    participant Handler as AsyncHTTPHandler.post (patched)
    participant Mock as mock_response

    Test->>Mock: headers = {} (NEW)
    Test->>Mock: status_code = 200
    Test->>Mock: json = return_val
    Test->>LiteLLM: acompletion(model, messages)
    LiteLLM->>Handler: POST request (mocked)
    Handler-->>LiteLLM: mock_response
    LiteLLM->>Mock: headers.get("x-gemini-service-tier")
    Mock-->>LiteLLM: None (empty dict, no branch taken)
    LiteLLM->>Mock: json() → response dict
    LiteLLM-->>Test: ModelResponse
    Test->>Test: assert URL, payload, response fields

Loading

_{Reviews (2): Last reviewed commit: "format vertex test file" | Re-trigger Greptile}

[codspeed-hq]

Merging this PR will not alter performance

✅ 16 untouched benchmarks

---

_{Comparing joereyna:litellm_fix_vertex_fine_tuned_model_test (afd46e7) with main (3a6db70)}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[Bakul2006]

Hey, @joereyna Are you facing the same issue of formatting the 330 files in lint check

[joereyna]

Hey, @joereyna Are you facing the same issue of formatting the 330 files in lint check

Yes, seeing the same thing.

[yuneng-berri]

@Bakul2006 this is a known issue, we are actively making the dev process more seamless around this.