[v0.25.1rc][BugFix][Frontend] Align DeepSeek V4 system tool rendering

[QwertyJack]

What this PR does / why we need it?

This is a follow-up to #13519. The v0.25.1 release monkey patch inherited an
upstream Python renderer mismatch reported in
vllm-project/vllm#51829.

For DeepSeek V4 requests containing both an existing system message and
top-level tools, the Python wrapper always inserted a synthetic system message
and rendered the tools before the caller's system content. This differs from
both vLLM's Rust renderer and the DeepSeek-V4-Flash-0731 checkpoint reference.

This patch aligns the release Python path with those references:

• attach request-level tools to a shallow copy of the first existing system
  message;
• insert a synthetic system message only when no system message exists; and
• leave caller-owned message dictionaries unchanged.

The reasoning-effort normalization introduced by #13519 is unchanged.

Does this PR introduce any user-facing change?

Yes. DeepSeek V4 requests with both a system message and top-level tools now
render the system content and tool schemas in the same order as the Rust and
checkpoint reference renderers. Requests without a system message keep the
existing synthetic-system behavior.

How was this patch tested?

• Tested with vLLM v0.25.1 at commit
  752a3a504485790a2e8491cacbb35c137339ad34.
• VLLM_VERSION=0.25.1 python -m pytest -q tests/ut/patch/platform/test_deepseek_v4_thinking.py
  • Result: 26 passed.
  • Coverage includes existing-system tools, missing-system tools, caller input
    immutability, and default/high/low effort behavior.
• Tokenizer integration comparison against the checkpoint encoder for
  system/no-system x low/high:
  • all four rendered prompts matched exactly;
  • all four input-ID sequences matched exactly; and
  • caller messages remained unchanged in all four cases.
• ruff check and ruff format --check on both changed files passed.
• git diff --check origin/releases/v0.25.1rc...HEAD passed.

vllm-project/vllm@752a3a5

• vLLM version: v0.25.1
• vLLM main: vllm-project/vllm@fe784ff

[gemini-code-assist]

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request addresses a rendering mismatch in the DeepSeek V4 Python wrapper for vLLM v0.25.1. By aligning the tool attachment logic with the Rust renderer and official checkpoint references, the patch ensures consistent prompt formatting when system messages and tools are both provided. The changes improve reliability for tool-calling workflows while maintaining the integrity of user-provided message structures.

Highlights

• Tool Rendering Alignment: Updated the DeepSeek V4 Python renderer to correctly attach tools to existing system messages instead of always inserting a new synthetic system message.
• Synthetic System Message Logic: Ensured that a synthetic system message is only created when no existing system message is present in the request.
• Immutability Preservation: Implemented shallow copying of system messages before modification to ensure caller-owned message dictionaries remain unchanged.
• Test Coverage: Added comprehensive unit tests to verify tool attachment behavior, system message handling, and reasoning effort parameter preservation.

New Features

🧠 You can now enable Memory (public preview) to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

---

Tip

💡 Consider Linking a Related Issue or RFC

Your PR title contains the [BugFix] tag, indicating a bug fix or new feature.

Linking a related issue or RFC in the PR description is strongly encouraged — it gives reviewers helpful context and speeds up the review. You can use any of these keywords:

• Fixes #<issue_number>
• Closes #<issue_number>
• Resolves #<issue_number>
• Refs #<rfc_or_issue_number> (for RFCs)

🙏 Thanks for helping us keep the project well-organized!

[gemini-code-assist]

Code Review

Suggested PR Title:

[Ops][BugFix] Fix DeepSeek V4 tokenizer tool attachment to existing system message

Suggested PR Summary:

### What this PR does / why we need it?
This PR fixes an issue in the DeepSeek V4 tokenizer patch where tools were always inserted as a new system message at the beginning of the message list, even if a system message already existed. The updated logic now searches for an existing system message and attaches the tools to it if found; otherwise, it inserts a new system message at the beginning.

### Does this PR introduce _any_ user-facing change?
Yes, it ensures that tools are correctly attached to the existing system message instead of creating a duplicate system message, which improves compatibility with chat templates that expect a single system message.

### How was this patch tested?
Added unit tests in `tests/ut/patch/platform/test_deepseek_v4_thinking.py` to verify both cases (when a system message is already present and when it is missing).

I have no further feedback to provide as the implementation is correct and well-tested.