Closed: supplementary unit test for XGrammarChecker

[Echo-Nie]

Objective

1. guided_json: Verify whether the JSON schema is compatible with XGrammar.
2. guided_grammar: Verify the validity of custom grammar strings.
3. guided_json_object: Verify that the simplified JSON object pattern { "type": "object" } can be correctly generated.
4. guided_choice: Verify whether a list of choices can be correctly transformed into grammar rules.
5. structural_tag: Verify that structured tags can be correctly parsed.
6. regex: Verify that regular expressions are passed through without validation errors.

Test Results

/opt/conda/envs/python35-paddle120-env/lib/python3.10/site-packages/paddle/utils/cpp_extension/extension_utils.py:717: UserWarning: No ccache found. Please be aware that recompiling all source files may be required. You can download and install ccache from: https://github.com/ccache/ccache/blob/master/doc/INSTALL.md
  warnings.warn(warning_message)
/opt/conda/envs/python35-paddle120-env/lib/python3.10/site-packages/_distutils_hack/__init__.py:30: UserWarning: Setuptools is replacing distutils. Support for replacing an already imported distutils is deprecated. In the future, this condition will fail. Register concerns at https://github.com/pypa/setuptools/issues/new?template=distutils-deprecation.yml
  warnings.warn(
test_guided_choice_invalid (__main__.TestXGrammarChecker) ... ok
test_guided_choice_valid (__main__.TestXGrammarChecker) ... ok
test_guided_grammar_invalid (__main__.TestXGrammarChecker) ... ok
test_guided_grammar_valid (__main__.TestXGrammarChecker) ... ok
test_guided_json_invalid (__main__.TestXGrammarChecker) ... ok
test_guided_json_object (__main__.TestXGrammarChecker) ... ok
test_guided_json_valid (__main__.TestXGrammarChecker) ... ok
test_regex_passthrough (__main__.TestXGrammarChecker) ... ok
test_structural_tag_invalid (__main__.TestXGrammarChecker) ... ok
test_structural_tag_valid (__main__.TestXGrammarChecker) ... ok

----------------------------------------------------------------------
Ran 10 tests in 0.002s

OK

Analysis

guided_json: Valid schemas pass correctly. Invalid schemas correctly return an error.

guided_grammar: Valid grammar strings generate the corresponding Grammar object. Invalid grammar strings return errors as expected.

guided_json_object: Simplified JSON object patterns are generated correctly.

guided_choice: Valid choice lists are correctly converted into grammar rules. Invalid choices (e.g., containing None) trigger exceptions.

structural_tag: Correct structured tags are parsed successfully. Missing trigger fields trigger a KeyError.

regex: Regular expressions are passed through directly without validation errors.

[paddle-bot]

Thanks for your contribution!

[luotao1]

@kevincheng2

[luotao1]

___
ERROR
collecting
model_executor/guided_decoding/test_xgrammar_checker.py
model_executor/guided_decoding/test_xgrammar_checker.py
Error: Process completed with exit code 8.

[Echo-Nie]

___
ERROR
collecting
model_executor/guided_decoding/test_xgrammar_checker.py

刚刚检查了一下log，原因如下：

model_executor/guided_decoding/test_xgrammar_checker.py:20: in <module>
    from fastdeploy.model_executor.guided_decoding.xgrammar_backend import XGrammarChecker
/usr/local/lib/python3.10/site-packages/fastdeploy/model_executor/guided_decoding/xgrammar_backend.py:23: in <module>
    import torch
E   ModuleNotFoundError: No module named 'torch'

原因发生在xgrammar_backend.py源码里面的 import torch，与单测无关。
https://github.com/PaddlePaddle/FastDeploy/blob/develop/fastdeploy/model_executor/guided_decoding/xgrammar_backend.py#L23

好像是CI在安装的时候没有torch？

在AIStudio上，安装好xgrammer 和 torch 之后能够通过

@luotao1 @kevincheng2

[kevincheng2]

___
ERROR
collecting
model_executor/guided_decoding/test_xgrammar_checker.py

刚刚检查了一下log，原因如下：

model_executor/guided_decoding/test_xgrammar_checker.py:20: in <module>
    from fastdeploy.model_executor.guided_decoding.xgrammar_backend import XGrammarChecker
/usr/local/lib/python3.10/site-packages/fastdeploy/model_executor/guided_decoding/xgrammar_backend.py:23: in <module>
    import torch
E   ModuleNotFoundError: No module named 'torch'

原因发生在xgrammar_backend.py源码里面的 import torch，与单测无关。 https://github.com/PaddlePaddle/FastDeploy/blob/develop/fastdeploy/model_executor/guided_decoding/xgrammar_backend.py#L23

好像是CI在安装的时候没有torch？

在AIStudio上，安装好xgrammer 和 torch 之后能够通过

@luotao1 @kevincheng2

可以在 .github/workflows/_unit_test_coverage.yml 文件中增加 python -m pip install xgrammar==0.1.19 torch==2.6.0，用于在单测镜像环境中安装 torch 和 xgrammar

[Echo-Nie]

可以在 .github/workflows/_unit_test_coverage.yml 文件中增加 python -m pip install xgrammar==0.1.19 torch==2.6.0，用于在单测镜像环境中安装 torch 和 xgrammar

Thx. Done

[luotao1]

单侧失败

[kevincheng2]

看起来tokenizers 的版本又冲突了，麻烦在`.github/workflows/_unit_test_coverage.yml` 所有pip包安装完成后指定 `python -m pip install tokenizers==0.22.0`吧

[Echo-Nie]

看起来tokenizers 的版本又冲突了，麻烦在.github/workflows/_unit_test_coverage.yml 所有pip包安装完成后指定 python -m pip install tokenizers==0.22.0吧

Thx, Done

[kevincheng2]

看起来 qwen3 启动时挂掉了，另外 fp8 的精度也受了影响，等晚点我找个环境复现一下找一下问题哈，晚点再合入

[Echo-Nie]

看起来 qwen3 启动时挂掉了，另外 fp8 的精度也受了影响，等晚点我找个环境复现一下找一下问题哈，晚点再合入

Get ! Thx

[kevincheng2]

可以取消 .github/workflows/_unit_test_coverage.yml 文件中的修改，在 scripts/unittest_requirement.txt 中增加 xgrammar==0.1.19，这种情况下可以通过ci，有时间辛苦修改一下

[Echo-Nie]

可以取消 .github/workflows/_unit_test_coverage.yml 文件中的修改，在 scripts/unittest_requirement.txt 中增加 xgrammar==0.1.19，这种情况下可以通过ci，有时间辛苦修改一下

好的老师，因为这个PR目前有点乱了，所以我重新提交一个，辛苦老师！