Skip to content

Add support to Vertex AI transformation for anyOf union type with null fields #9625

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 13 commits into from
Mar 28, 2025

Conversation

NickGrab
Copy link
Contributor

Title

Vertex AI supports anyOf union fields which are otherwise unsupported by Gemini. Support is added by modifying the convert_to_nullable helper function (renamed convert_anyof_null_to_nullable to better reflect its new functionality) to handle null types within anyOf fields (which are still not supported), while leaving anyOf unions otherwise intact.

Null fields are handled according to vertex documentation here. An error is thrown if all fields within the anyOf are null, as a standalone null type within a response schema is considered invalid by Vertex

Relevant issues

Addresses #8864

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

  • I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details
  • I have added a screenshot of my new test passing locally
Screenshot 2025-03-25 at 10 41 47 PM
  • My PR passes all unit tests on (make test-unit)[https://docs.litellm.ai/docs/extras/contributing_code]
  • My PR's scope is as isolated as possible, it only solves 1 specific problem

Type

🆕 New Feature
🐛 Bug Fix

Changes

NickGrab and others added 13 commits March 25, 2025 22:37
…yOf-support

Add support to Vertex AI transformation for anyOf union type with null fields
#9617)

* fix(proxy_server.py): get master key from environment, if not set in general settings or general settings not set at all

* test: mark flaky test

* test(test_proxy_server.py): mock prisma client

* ci: add new github workflow for testing just the mock tests

* fix: fix linting error

* ci(conftest.py): add conftest.py to isolate proxy tests

* build(pyproject.toml): add respx to dev dependencies

* build(pyproject.toml): add prisma to dev dependencies

* test: fix mock prompt management tests to use a mock anthropic key

* ci(test-litellm.yml): parallelize mock testing

make it run faster

* build(pyproject.toml): add hypercorn as dev dep

* build(pyproject.toml): separate proxy vs. core dev dependencies

make it easier for non-proxy contributors to run tests locally - e.g. no need to install hypercorn

* ci(test-litellm.yml): pin python version

* test(test_rerank.py): move test - cannot be mocked, requires aws credentials for e2e testing

* ci: add thank you message to ci

* test: add mock env var to test

* test: add autouse to tests

* test: test mock env vars for e2e tests
)

* fix(logging): add json formatting for uncaught exceptions (#9615)

* fix(_logging.py): cleanup logging to catch unhandled exceptions

* fix(_logging.py): avoid using 'print' '

---------

Co-authored-by: Henrique Cavarsan <hcavarsan@yahoo.com.br>
Copy link

vercel bot commented Mar 28, 2025

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

Name Status Preview Comments Updated (UTC)
litellm ✅ Ready (Inspect) Visit Preview 💬 Add feedback Mar 28, 2025 10:56pm

@NickGrab NickGrab merged commit 220d4c0 into main Mar 28, 2025
37 of 40 checks passed
@NickGrab NickGrab deleted the litellm_mar_28_vertex_fix branch March 28, 2025 23:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants