Skip to content

Add option to disable any_whitespace for xgrammar and llguidance backends.#8919

Merged
hnyls2002 merged 8 commits intosgl-project:mainfrom
lulor:xgrammar-disable-any-whitespace
Oct 11, 2025
Merged

Add option to disable any_whitespace for xgrammar and llguidance backends.#8919
hnyls2002 merged 8 commits intosgl-project:mainfrom
lulor:xgrammar-disable-any-whitespace

Conversation

@lulor
Copy link
Copy Markdown
Contributor

@lulor lulor commented Aug 7, 2025

Motivation

See #8250.

Modifications

This PR simply adds a server arg which is passed through to the xgrammar and llguidance backends, preserving the current defaults.

Accuracy Test

Benchmark & Profiling

Checklist

@lulor lulor requested a review from hnyls2002 as a code owner August 7, 2025 15:10
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Summary of Changes

Hello @lulor, 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 introduces a new server argument that provides granular control over whitespace handling in constrained JSON generation for the xgrammar and llguidance backends. My aim is to allow users to enforce a more compact JSON representation by disabling flexible whitespace, which can be beneficial for specific parsing requirements or reducing output size.

Highlights

  • New Server Argument: A new command-line argument, --constrained-json-disable-any-whitespace, has been added to ServerArgs, allowing users to control whether flexible whitespace is permitted in constrained JSON output.
  • XGrammar Backend Integration: The XGrammarGrammarBackend now accepts an any_whitespace parameter during initialization, which is then passed to the grammar compiler when generating JSON schemas, enabling more control over whitespace in the output.
  • LLGuidance Backend Integration: The GuidanceBackend has been updated to incorporate the any_whitespace parameter in its constructor and utilizes this setting when dispatching JSON schemas, influencing the flexibility of whitespace in the generated JSON.
  • Whitespace Pattern Type Correction: The type hint for the whitespace_pattern parameter in the OutlinesGrammarBackend's constructor has been corrected from bool to str | None, improving type accuracy.
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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

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 issue comments and review comments.
Help /gemini help Displays a list of available commands.

Customization

To customize 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 counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments or fill out our survey to provide feedback.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

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.

Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request introduces an option to disable flexible whitespace in JSON-constrained generation for the xgrammar and llguidance backends. The changes are well-implemented and consistent across the affected files. My main feedback is to improve the naming of the new server argument for better clarity and maintainability. I've provided suggestions to rename constrained_json_disable_any_whitespace to constrained_json_compact.

Comment thread python/sglang/srt/server_args.py
Comment thread python/sglang/srt/server_args.py
Comment thread python/sglang/srt/constrained/base_grammar_backend.py
@CatherineSue
Copy link
Copy Markdown
Collaborator

@lulor Thank you for this change. Can you take a look at the lint error?

@CatherineSue CatherineSue self-assigned this Oct 6, 2025
@lulor
Copy link
Copy Markdown
Contributor Author

lulor commented Oct 7, 2025

@CatherineSue thank you for pushing this forward. I believe the lint check is passing now, is there anything I can do on my side?

@CatherineSue
Copy link
Copy Markdown
Collaborator

@lulor All good right now. I'm waiting for the CI to finish. If it passes, we'll merge.

@Swipe4057
Copy link
Copy Markdown
Contributor

Swipe4057 commented Oct 9, 2025

@CatherineSue

ci was successful

Copy link
Copy Markdown
Collaborator

@hnyls2002 hnyls2002 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@hnyls2002 hnyls2002 merged commit b5dcfd4 into sgl-project:main Oct 11, 2025
91 of 98 checks passed
lpc0220 pushed a commit to lpc0220/sglang that referenced this pull request Oct 29, 2025
…` backends. (sgl-project#8919)

Co-authored-by: Chang Su <chang.s.su@oracle.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants