feat: Improve ollama_chat structured-output resilience for mixed text+JSON responses

[ansh-info]

Fixes #17807 (Ollama /api/chat returns non-JSON, breaking structured output/tool calls)

Pre-Submission checklist

• I have Added testing in the tests/litellm/ (https://github.com/BerriAI/litellm/tree/main/tests/litellm) directory, Adding at least 1 test is a hard requirement - see details (https://docs.litellm.ai/docs/extras/contributing_code)

• I have added a screenshot of my new test passing locally

• 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

  🐛 Bug Fix

  Changes

  • Add a best-effort JSON coercion in ollama_chat when format/structured output is requested but /api/chat returns mixed natural text + JSON; applies to sync/async paths.
  • Add a regression test covering mixed text + JSON to ensure coercion extracts a valid JSON object.
  • Goal: prevent structured-output/tool-call parsing from failing when the model includes a JSON fragment in otherwise natural-language replies.

[vercel]

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

Project	Deployment	Preview	Comments	Updated (UTC)
litellm	Error			Dec 11, 2025 9:57am

[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.}

[krrishdholakia]

@ansh-info can you share a screenshot of your code being run on an ollama_chat call?

i think this is a dead/unused file