[Bugfix] Add validation for tool requests when tool_parser is unavailable

[majiayu000]

Summary

When tool_choice is set to "required" or a named tool (e.g., {"type": "function", "function": {"name": "get_weather"}}), but the server was not started with --enable-auto-tool-choice and --tool-call-parser, vLLM would silently degrade instead of returning a clear error.

This PR adds validation to return an appropriate error message indicating that these flags are required for tool parsing.

Changes

• Added validation in serving_chat.py to check if tool_parser is available when tool_choice requires tool parsing
• Added test case covering both tool_choice="required" and named tool scenarios

Test Plan

• Added unit test test_tool_choice_validation_without_parser
• Test verifies error response contains guidance about --enable-auto-tool-choice

Fixes #29432

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors.

You ask your reviewers to trigger select CI tests on top of fastcheck CI.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

If you have any questions, please reach out to us on Slack at https://slack.vllm.ai.

🚀

[chatgpt-codex-connector]

Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits.

[gemini-code-assist]

Code Review

This pull request correctly addresses a bug where vLLM would fail silently when tool choice was required but the tool parser was not enabled. The added validation ensures a clear error message is returned to the user, improving the developer experience. The new logic is covered by a unit test. I have one suggestion to refactor the validation logic to improve maintainability by reducing code duplication.

[gemini-code-assist]

This validation logic duplicates the condition from the tool_choice='auto' check on lines 207-212. This could lead to maintenance issues if the conditions for tool parsing availability change in the future.

Consider refactoring to centralize the check. You could define a tool_parsing_unavailable variable and then use a single if block to handle all tool_choice values that require a parser ('auto', 'required', and named tools).

Example refactoring for lines 207-232:

tool_parsing_unavailable = (tool_parser is None and
                            not isinstance(tokenizer, MistralTokenizer) and
                            not self.use_harmony)

if tool_parsing_unavailable and request.tool_choice not in (None, "none"):
    if request.tool_choice == "auto":
        return self.create_error_response(
            '"auto" tool choice requires '
            "--enable-auto-tool-choice and --tool-call-parser to be set")
    else:
        return self.create_error_response(
            f'tool_choice="{request.tool_choice}" requires '
            "--enable-auto-tool-choice and --tool-call-parser to be set")

Since this refactoring touches lines outside the current diff, I'm providing it as a code block in the comment body.

[DarkLight1337]

Thanks for improving the UX!

[majiayu000]

Done, thanks for the suggestion!

[DarkLight1337]

PTAL at the failing test

[mergify]

Hi @majiayu000, the pre-commit checks have failed. Please run:

uv pip install pre-commit
pre-commit install
pre-commit run --all-files

Then, commit the changes and push to your branch.

For future commits, pre-commit will run automatically on changed files before each commit.

Tip

Is mypy or markdownlint failing?

mypy and markdownlint are run differently in CI. If the failure is related to either of these checks, please use the following commands to run them locally:

# For mypy (substitute "3.10" with the failing version if needed)
pre-commit run --hook-stage manual mypy-3.10
# For markdownlint
pre-commit run --hook-stage manual markdownlint

[mergify]

Documentation preview: https://vllm--30613.org.readthedocs.build/en/30613/