John6797/sc 7670/move test result description prompts to backend

[johnwalz97]

Internal Notes for Reviewers

Moving test result description generation to the backend.

External Release Notes

[juanmleng]

Would this change still support passing external context via VALIDMIND_LLM_DESCRIPTIONS_CONTEXT?

[cachafla]

Would this change still support passing external context via VALIDMIND_LLM_DESCRIPTIONS_CONTEXT?

The context is still being passed to the request, except that now the request is sent to the API and unpacked here: https://github.com/validmind/backend/pull/1140/files#diff-1f286300bd4ef710c7d4c921c3702e65de730379fc39f4b68c686cc4d92c1345R412

If you check the the generate_description() function you'll see that context is still being retrieved in the same way:

context = _get_llm_global_context()

[cachafla]

Awesome 👍

[AnilSorathiya]

left one comment. Otherwise it is LGTM 👍

[github-actions]

PR Summary

This pull request introduces significant refactoring to the validmind/ai/test_descriptions.py module, primarily focusing on simplifying the process of generating test result descriptions. The changes include:

• Removal of the _load_prompt and prompt_to_message functions, which were responsible for loading and formatting prompt templates for generating descriptions.
• Elimination of the context.py, config.yaml, image_processing.py, system.jinja, and user.jinja files, which were previously used for handling context and image processing related to test result descriptions.
• Introduction of a new function generate_test_result_description in validmind/api_client.py, which centralizes the description generation process by making an API call to generate the test result description.
• The generate_description function in test_descriptions.py now directly calls generate_test_result_description from api_client.py, simplifying the code and removing the need for local prompt handling.

These changes aim to streamline the codebase by removing redundant code and improving the maintainability of the test description generation process.

Test Suggestions

• Test the generate_description function to ensure it correctly calls the API and returns the expected description.
• Verify that the removal of the _load_prompt and prompt_to_message functions does not affect other parts of the codebase.
• Check the API endpoint used in generate_test_result_description to ensure it handles various input scenarios correctly.
• Ensure that the refactored code handles cases where no tables, figures, or metrics are provided, raising appropriate errors.