[ZhuLinsen]

Overview

Ernie45ToolParser.extract_tool_calls reports tools_called=True with an empty tool_calls list when the output contains <tool_call> but no complete <tool_call>...</tool_call> block, and drops everything after the start marker. This PR returns the full output as content with tools_called=False in that case.

Claims

• Truncated tool-call output (e.g. generation stopped by max_tokens in the middle of a call) is no longer silently lost: the request now gets the full text back as content.
• tools_called is consistent with tool_calls, as tests/tool_parsers/utils.py already asserts for other parsers (extracted.tools_called == bool(extracted.tool_calls)).
• Outputs that contain at least one complete tool call are unchanged.

Validation

New regression test test_extract_tool_calls_truncated_keeps_content in tests/tool_parsers/test_ernie45_moe_tool_parser.py (real ERNIE 4.5 tokenizer, real parser).

Before the fix (parser from main, new test):

$ pytest tests/tool_parsers/test_ernie45_moe_tool_parser.py -k truncated
E       AssertionError: assert not True
E        +  where True = ExtractedToolCallInformation(tools_called=True, tool_calls=[], content='Sure.').tools_called
FAILED tests/tool_parsers/test_ernie45_moe_tool_parser.py::test_extract_tool_calls_truncated_keeps_content
1 failed, 7 deselected

After the fix:

$ pytest tests/tool_parsers/test_ernie45_moe_tool_parser.py
8 passed

pre-commit run --files <changed files> passes (ruff check/format, mypy, typos, check-test-tethering).

Environment: CPU only (ARM Neoverse-N1, 4 cores), Python 3.12, torch 2.14.0, transformers 5.20.2. I did not run an end-to-end RemoteOpenAIServer test with real ERNIE 4.5 weights because I have no GPU; the change is confined to the non-streaming parser and is covered at that level by the test above.

Details

Root cause: tool_call_regex only matches complete <tool_call>...</tool_call> blocks, but the code returned tools_called=True whenever <tool_call> appeared in the output, even if the regex found nothing. DelegatingParser._extract_tool_calls (vllm/parser/abstract_parser.py) then replaces the content with the text before the marker, so the rest of the output is dropped.

The fix adds an early return when no call was extracted, matching what other parsers already do (e.g. step3_tool_parser.py, functiongemma_tool_parser.py). The streaming path is not touched.

Duplicate check: gh pr list --state open --search "ernie45 tool" returns #57677 (regex timeouts), #52841 (Rust frontend parser), #48295 (streaming path) and #48386 (reasoning parser); none changes the non-streaming empty-result handling. deepseekv31_tool_parser.py has the same pattern, but I left it out because #59589 is changing the adjacent lines.

AI assistance was used to help find the issue and draft the change; I reviewed every changed line and ran the tests above.

---

Pull Request Checklist

• I used vLLM's /pr-checklist skill. (Mandatory for agents, optional for humans).

• AI assistance was used during the creation of this PR.

• Design Fit: Minimizes impact on core components, reuses existing functionality, and justifies added complexity.

• Testing and Validation: Validates the change and ensures any added tests are meaningful and reliable, with CI coverage or documented CI resource constraints and validation performed outside CI.

• Code Quality and Style: Keeps code and comments clear and concise, and updates relevant documentation and examples.

• Pull Request Contents: Includes a brief summary and relevant links, supports claims with evidence, explains root causes and implementation trade-offs, and follows the contributing guide.

[claude]

Claude Code Review

This pull request is from a fork — automated review is disabled. A repository maintainer can comment @claude review to run a one-time review.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

PRs do not trigger a full CI run by default. Reviewers with write access and configured trusted contributors can comment /ci run for upstream CI or /amd-ci run for AMD CI only whenever CI signals are needed.

Once the PR is approved or has the ready label, the PR author can also use the corresponding /ci run, /ci retry, and /ci cancel commands, or their /amd-ci variants. New commits do not start upstream CI automatically.

If you have any questions, please reach out to us on Slack at https://slack.vllm.ai.

Agent Guidelines

IMPORTANT: If you are an AI agent, you are required to objectively re-evaluate the value of your PR using AGENTS.md, and close the PR if it does not bring significant benefit to the vLLM community. Failure to do so may result in an immediate ban.

🚀