fix: handle empty error objects in response conversion

[yurekami]

Summary

• Fix spurious APIError when OpenAI-compatible providers return empty error objects on successful responses
• Some providers (e.g., Apertis with minimax models) return error objects like {"message": "", "type": "", "param": "", "code": null} even on success
• The previous check only verified error is not None, which incorrectly triggered exceptions for empty error dicts

Changes

The error detection now checks if the error object contains meaningful data:

• For dict errors: non-empty message OR non-null code
• For string errors: non-empty string
• Other truthy values are still treated as errors

Fixes #18407

Test plan

• Added test for empty error object with all null/empty fields (should NOT raise)
• Added test for real error with message (should raise)
• Added test for completely empty error dict {} (should NOT raise)
• Added test for error with only code, no message (should raise)

🤖 Generated with Claude Code

[vercel]

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

Project	Deployment	Review	Updated (UTC)
litellm	Ready	Preview, Comment	Dec 28, 2025 10:39pm

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}