litellm - 💡(How to fix) Fix refactor: extract LiteLLM Data Access Layer (DAL) for control & inference nodes [1 participants]

TL;DR

Create a single Data Access Layer (DAL) module, litellm_db.py, to centralize raw SQL functions for LiteLLM tables and refactor existing files to delegate SQL operations to this module.

Guidance

• Identify and consolidate duplicated queries across files (auth.py, provisioning.py, invitations.py, budget_sync.py, consumer.py, worldclaw/service.py) into the proposed litellm_db.py module.
• Refactor the listed source files to use the new DAL module for SQL operations, ensuring a clean data access boundary between business logic and LiteLLM table operations.
• Document performance and correctness trade-offs for SQL queries in the DAL module, including considerations for deadlock prevention, spend, and idempotency.
• Implement unit tests for the DAL module to ensure its correctness and reliability.

Example

No specific code example is provided due to the lack of explicit code snippets in the issue body. However, the process would involve creating a litellm_db.py file with functions like _find_user_by_sso and then updating files like provisioning.py and worldclaw/service.py to import and use these centralized functions.

Notes

The proposed solution aims to improve code organization, reduce duplication, and enhance maintainability by introducing a clear data access layer. However, the success of this approach depends on correctly identifying all relevant SQL queries and ensuring that the new DAL module is properly tested and integrated into the existing codebase.

Recommendation

Apply the proposed workaround by creating the litellm_db.py module and refactoring the existing files to use it, as this approach addresses the identified issues of duplicated queries, missing documentation, and lack of a clean data access boundary.