[None][fix] use tool parser for named tool choice responses

[youngzheng-oss]

Description

• remove named-tool-choice special handling in chat postprocessors
• always run tool output through apply_tool_parser for parity with auto tool choice

Test coverage

• tests/unittest/llmapi/apps/_test_openai_tool_call.py

Summary by CodeRabbit

• Bug Fixes
  • Tool call handling is now unified and consistent between streaming and non-streaming response processing modes.
  • Tool calls are reliably parsed and integrated into messages regardless of the processing path.
  • Improved stability and consistency when handling tool calls across all post-processing scenarios.

[coderabbitai]

📝 Walkthrough

Walkthrough

The pull request consolidates tool-call handling across streaming and non-streaming post-processing paths in a single file. Tool parsing is now applied uniformly regardless of tool_choice type, and tool_call_id generation is restricted to first chunks only. Duplicate branching logic is eliminated in favor of a single, consistent flow for both functions.

Changes

Cohort / File(s)	Summary
Tool-call handling unification tensorrt_llm/serve/postprocess_handlers.py	Refactored yield_first_chat and chat_response_post_processor to consistently apply tool parsing and construct tool call entries. Removed conditional branching based on tool_choice type; tool_call_id now generated only for first chunks. DeltaMessage and ChatMessage construction consolidated to single path for both functions.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.
Description check	❓ Inconclusive	The PR description provides a concise explanation of the changes but is missing required sections from the template such as PR title format, detailed problem/solution explanation, and the PR checklist.	Add a proper PR title following the template format [ticket/issue][type] Summary, expand the description section with more context about why this change is needed, and ensure the PR checklist section is included with appropriate items reviewed.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: applying the tool parser to named tool choice responses for consistency with auto tool choice handling.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

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

Tip

Try Coding Plans. Let us write the prompt for your AI agent so you can ship faster (with fewer bugs).
Share your feedback on Discord.

---

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

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@tensorrt_llm/serve/postprocess_handlers.py`:
- Around line 219-250: apply_tool_parser currently lacks access to
args.tool_choice so when a forced named tool
(ChatCompletionNamedToolChoiceParam) is used and the model emits only argument
deltas, resulting calls lose their function name (call.name is None) and
streaming/non-streaming behavior breaks; fix by updating apply_tool_parser to
accept the forced tool name (extract from args.tool_choice when it is a
ChatCompletionNamedToolChoiceParam) and propagate it into returned call objects,
or if you prefer minimal changes, after apply_tool_parser returns, inspect calls
in the postprocess flow (the loop building DeltaToolCall/DeltaFunctionCall) and
if call_item.name is missing but args.tool_choice is a named tool, restore
function_name from args.tool_choice before creating the DeltaFunctionCall so
DeltaToolCall.id/name are preserved.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 0a5bb47d-6e2d-4266-9f34-0b2844115b36

📥 Commits

Reviewing files that changed from the base of the PR and between 35ccedd and 2dcda86.

📒 Files selected for processing (1)

• tensorrt_llm/serve/postprocess_handlers.py

[pengbowang-nv]

Hi @LinPoly , could you review this PR as it is related to tool parser code? Thanks!