[TRTLLM-9677][feat] Support DeepSeek-V3.2 tool parser #10126

lfr-0531 · 2025-12-18T13:11:18Z

Summary by CodeRabbit

New Features
- Added support for DeepSeek V3.2 model with tool-call parsing capabilities for both streaming and non-streaming inference modes.
Tests
- Added comprehensive test coverage for DeepSeek V3.2 tool parser, validating tool-call parsing, parameter extraction, and error handling.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

Description

Features:

Port tool parser for DeepSeek-V3.2 from sgl.
Enable both streaming and one-time parsing modes.
Add tests for DeepSeekV32ToolParser.

Dependency: #9814

Test Coverage

PR Checklist

Please review the following before submitting your PR:

PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
Update tava architecture diagram if there is a significant design change in PR.
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...

Provide a user friendly way for developers to interact with a Jenkins server.

Run /bot [-h|--help] to print this help message.

See details below for each supported subcommand.

Details

run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]

Launch build/test pipelines. All previously running jobs will be killed.

--reuse-test (optional)pipeline-id (OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.

--disable-reuse-test (OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.

--disable-fail-fast (OPTIONAL) : Disable fail fast on build/tests/infra failures.

--skip-test (OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.

--stage-list "A10-PyTorch-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.

--gpu-type "A30, H100_PCIe" (OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.

--test-backend "pytorch, cpp" (OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.

--only-multi-gpu-test (OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.

--disable-multi-gpu-test (OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.

--add-multi-gpu-test (OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.

--post-merge (OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.

--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" (OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".

--detailed-log (OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.

--debug (OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in the stage-list parameter to access the appropriate container environment. Note: Does NOT update GitHub check status.

For guidance on mapping tests to stage names, see docs/source/reference/ci-overview.md
and the scripts/test_to_stage_mapping.py helper.

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip testing for latest commit on pull request. --comment "Reason for skipping build/test" is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

reuse-pipeline

reuse-pipeline

Reuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

Signed-off-by: Fanrong Li <23290157+lfr-0531@users.noreply.github.com>

lfr-0531 · 2025-12-20T08:07:46Z

/bot run

coderabbitai · 2025-12-20T08:10:15Z

📝 Walkthrough

Walkthrough

Adds support for DeepSeek V3.2 tool-call format parsing by introducing a new DeepSeekV32Parser class, registering it in the tool parser factory, configuring the OpenAI server to recognize the model type, and providing comprehensive test coverage. The parser supports both streaming and non-streaming parsing modes using DSML-formatted tool-call sequences.

Changes

Cohort / File(s)	Change Summary
DeepSeekV32Parser Implementation `tensorrt_llm/serve/tool_parser/deepseekv32_parser.py`	New parser class extending BaseToolParser with detect_and_parse and parse_streaming_increment methods for DSML tool-call format, including XML/JSON parameter extraction and streaming state management.
Factory and Server Configuration `tensorrt_llm/serve/tool_parser/tool_parser_factory.py`, `tensorrt_llm/serve/openai_server.py`	Registers DeepSeekV32Parser in factory mapping; adds "deepseek_v32" model type branch in OpenAI server to set corresponding tool_call_id_type.
Test Coverage `tests/unittest/llmapi/apps/test_tool_parsers.py`	Adds TestDeepSeekV32Parser class with initialization, streaming parse, structure_info, and format/compliance validation tests; includes duplicate test blocks.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

deepseekv32_parser.py: Review streaming buffer management, state tracking (current_tool_id, prev_tool_call_arr, streamed_args_for_tool), and the dual-format parameter parsing logic in _parse_parameters_from_xml for correctness and edge cases.
test_tool_parsers.py: Identify and address the duplicate TestDeepSeekV32Parser test blocks to prevent redundant test execution and maintenance debt.
tool_parser_factory.py & openai_server.py: Verify factory registration and server configuration correctly map to the new parser implementation.

Pre-merge checks and finishing touches

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 62.50% which is insufficient. The required threshold is 80.00%.	You can run `@coderabbitai generate docstrings` to improve docstring coverage.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely describes the main change: adding support for the DeepSeek-V3.2 tool parser.
Description check	✅ Passed	The description provides a brief explanation of features (porting parser, streaming/one-time modes, tests) but the Test Coverage section is incomplete, lacking specific test references.

✨ Finishing touches

📝 Generate docstrings

🧪 Generate unit tests (beta)

Create PR with unit tests
Post copyable unit tests in a comment

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

coderabbitai

Actionable comments posted: 1

🧹 Nitpick comments (5)

tensorrt_llm/serve/tool_parser/deepseekv32_parser.py (4)
64-74: Initialize all instance attributes in __init__.

_tool_indices (checked via hasattr at line 184) and current_tool_name_sent (set at line 270) should be initialized in __init__ to ensure all externally visible members are declared upfront. Per coding guidelines, initialize all externally visible members in the constructor.
🔎 Proposed fix
 def __init__(self):
     super().__init__()
     self.bot_token = "<｜DSML｜function_calls>"
     self.eot_token = "</｜DSML｜function_calls>"
     self.invoke_begin_regex = r'<｜DSML｜invoke\s+name="([^"]+)"\s*>'
     self.invoke_end_token = "</｜DSML｜invoke>"
     self.parameter_regex = (
         r'<｜DSML｜parameter\s+name="([^"]+)"\s+string="([^"]+)"\s*>(.*?)</｜DSML｜parameter>'
     )
     self._last_arguments = ""
     self.current_tool_id = -1
+    self.current_tool_name_sent = False
+    self._tool_indices = {}
151-154: Narrow the exception clause.

Per coding guidelines, limit except to the smallest set of errors possible. The broad except Exception could mask unexpected errors. Consider catching specific exceptions like re.error, json.JSONDecodeError, or KeyError.
🔎 Proposed fix
-        except Exception as e:
+        except (re.error, json.JSONDecodeError, KeyError, AttributeError) as e:
             logger.error(f"Error in detect_and_parse: {e}")
             # return the normal text if parsing fails
             return StreamingParseResult(normal_text=text)
184-185: Avoid hasattr check for instance attribute.

If _tool_indices is initialized in __init__, this hasattr check becomes unnecessary. Currently, this lazy initialization pattern makes the code harder to follow.

286-288: Narrow the exception clause.

Same as the detect_and_parse method - consider catching specific exceptions rather than broad Exception.
🔎 Proposed fix
-        except Exception as e:
+        except (re.error, json.JSONDecodeError, KeyError, AttributeError) as e:
             logger.error(f"Error in parse_streaming_increment: {e}")
             return StreamingParseResult(normal_text=current_text)
tests/unittest/llmapi/apps/test_tool_parsers.py (1)
1257-1281: Note: Docstring doesn't match test purpose.

The docstring says "Test streaming parser handles end token correctly" but the test is actually verifying multiple tools streaming behavior.
🔎 Proposed fix
     def test_parse_streaming_increment_multiple_tools_streaming(
             self, sample_tools, parser):
-        """Test streaming parser handles end token correctly."""
+        """Test streaming parser handles multiple tool calls."""
         # First tool

📜 Review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 3f25db9 and 34eff6a.

📒 Files selected for processing (4)

tensorrt_llm/serve/openai_server.py (1 hunks)
tensorrt_llm/serve/tool_parser/deepseekv32_parser.py (1 hunks)
tensorrt_llm/serve/tool_parser/tool_parser_factory.py (2 hunks)
tests/unittest/llmapi/apps/test_tool_parsers.py (2 hunks)

🧰 Additional context used

📓 Path-based instructions (2)

**/*.py

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

**/*.py: Code developed for TensorRT-LLM should conform to Python 3.8+
Indent Python code with 4 spaces. Do not use tabs
Always maintain the namespace when importing in Python, even if only one class or function from a module is used
Python files should use snake_case naming: some_file.py
Python classes should use PascalCase naming: class SomeClass
Python functions and methods should use snake_case naming: def my_awesome_function():
Python local variables should use snake_case naming: my_variable = ...
Python variable names that start with a number should be prefixed with 'k': k_99th_percentile = ...
Python global variables should use upper snake_case with prefix 'G': G_MY_GLOBAL = ...
Python constants should use upper snake_case naming: MY_CONSTANT = ...
Avoid shadowing variables declared in an outer scope in Python
Initialize all externally visible members of a Python class in the constructor
For Python interfaces that may be used outside a file, prefer docstrings over comments
Python comments should be reserved for code within a function, or interfaces that are local to a file
Use Google style docstrings in Python for classes and functions, which can be parsed by Sphinx
Python attributes and variables can be documented inline with type and description
Avoid using reflection in Python when functionality can be easily achieved without reflection
When using try-except blocks in Python, limit the except to the smallest set of errors possible
When using try-except blocks in Python to handle multiple possible variable types (duck-typing), keep the body of the try as small as possible, using the else block for logic

Files:

tests/unittest/llmapi/apps/test_tool_parsers.py
tensorrt_llm/serve/openai_server.py
tensorrt_llm/serve/tool_parser/deepseekv32_parser.py
tensorrt_llm/serve/tool_parser/tool_parser_factory.py

**/*.{cpp,h,cu,cuh,py}

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

All TensorRT-LLM Open Source Software code should contain an NVIDIA copyright header that includes the year of its latest meaningful modification

Files:

tests/unittest/llmapi/apps/test_tool_parsers.py
tensorrt_llm/serve/openai_server.py
tensorrt_llm/serve/tool_parser/deepseekv32_parser.py
tensorrt_llm/serve/tool_parser/tool_parser_factory.py

🧠 Learnings (1)

📚 Learning: 2025-08-14T06:36:40.701Z

Learnt from: timlee0212
Repo: NVIDIA/TensorRT-LLM PR: 6886
File: tensorrt_llm/_torch/models/modeling_deepseekv3.py:0-0
Timestamp: 2025-08-14T06:36:40.701Z
Learning: In DeepSeek V3 model (tensorrt_llm/_torch/models/modeling_deepseekv3.py), the disagreement between AllReduce.__init__ guard and _compute_mlp_tp_size logic for MNNVL usage is expected by design. The AllReduce component and MLP TP-size computation intentionally use different criteria for MNNVL availability decisions.

Applied to files:

tensorrt_llm/serve/openai_server.py

🧬 Code graph analysis (4)

tests/unittest/llmapi/apps/test_tool_parsers.py (1)

tensorrt_llm/serve/tool_parser/deepseekv32_parser.py (3)

DeepSeekV32Parser (13-295)

parse_streaming_increment (156-288)

detect_and_parse (114-154)

tensorrt_llm/serve/openai_server.py (1)

tensorrt_llm/bench/benchmark/__init__.py (1)

model_type (70-71)

tensorrt_llm/serve/tool_parser/deepseekv32_parser.py (2)

tensorrt_llm/serve/openai_protocol.py (1)

ChatCompletionToolsParam (520-522)

tensorrt_llm/serve/tool_parser/core_types.py (2)

StreamingParseResult (16-20)

ToolCallItem (8-13)

tensorrt_llm/serve/tool_parser/tool_parser_factory.py (1)

tensorrt_llm/serve/tool_parser/deepseekv32_parser.py (1)

DeepSeekV32Parser (13-295)

🪛 Ruff (0.14.8)

tests/unittest/llmapi/apps/test_tool_parsers.py

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1189-1189: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)

1193-1193: String contains ambiguous ｜ (FULLWIDTH VERTICAL LINE). Did you mean | (VERTICAL LINE)?

(RUF001)