[Frontend][2/n] remove empty content from _parse_tool_calls_from_content

[qandrew]

Purpose

I noticed that our tool parser often times may leave whitespace after the tool call, which gets outputted to the user. We should strip it out.

Test Plan

qwen3 example:

vllm serve Qwen/Qwen3-1.7B --reasoning-parser qwen3       --structured-outputs-config.backend xgrammar       --enable-auto-tool-choice --tool-call-parser hermes

client

 curl -X POST "http://localhost:8000/v1/responses"   -H "Content-Type: application/json"   -H "Authorization: Bearer dummy-api-key"   -d '{
        "model": "Qwen/Qwen3-1.7B",
        "input": "Whats the weather like in San Francisco? use celsius.",
        "tools": [{
            "type": "function",
            "name": "get_weather",
            "description": "Get the current weather in a given location",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {"type": "string"},
                    "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]}
                },
                "required": ["location", "unit"]
            }
        }]
      }'

Test Result

{
    "id": "resp_3814bea59f4449a7a29c9a406b50ab4e",
    "created_at": 1762398952,
    "incomplete_details": null,
    "instructions": null,
    "metadata": null,
    "model": "Qwen/Qwen3-1.7B",
    "object": "response",
    "output": [
        {
            "id": "rs_8a39e9f4178d40abac111d595d801126",
            "summary": [],
            "type": "reasoning",
            "content": [
                {
                    "text": "\nOkay, the user is asking for the weather in San Francisco using Celsius. Let me check the tools provided. There's a function called get_weather that requires location and unit. The parameters are location (string) and unit (enum: celsius or fahrenheit). The user specified San Francisco and Celsius, so I need to call get_weather with those arguments. I should make sure the arguments are in the correct format. Let me structure the JSON accordingly.\n",
                    "type": "reasoning_text"
                }
            ],
            "encrypted_content": null,
            "status": null
        },
### BELOW GETS REMOVED AFTER THIS PR ####
        {
            "id": "msg_33be381a036e49e8abc0faead1e05752",
            "content": [
                {
                    "annotations": [],
                    "text": "\n\n",
                    "type": "output_text",
                    "logprobs": null
                }
            ],
            "role": "assistant",
            "status": "completed",
            "type": "message"
        },
### ABOVE GETS REMOVED AFTER THIS PR ####
        {
            "arguments": "{\"location\": \"San Francisco\", \"unit\": \"celsius\"}",
            "call_id": "call_62165488d94f43f1ab949be8f63ab20f",
            "name": "get_weather",
            "type": "function_call",
            "id": "fc_9e77818e2fee464ea9853758d198ee01",
            "status": "completed"
        }
    ],
    "parallel_tool_calls": true,
    "temperature": 0.6,
    "tool_choice": "auto",
    "tools": [
        {
            "name": "get_weather",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {
                        "type": "string"
                    },
                    "unit": {
                        "type": "string",
                        "enum": [
                            "celsius",
                            "fahrenheit"
                        ]
                    }
                },
                "required": [
                    "location",
                    "unit"
                ]
            },
            "strict": null,
            "type": "function",
            "description": "Get the current weather in a given location"
        }
    ],
    "top_p": 0.95,
    "background": false,
    "max_output_tokens": 40766,
    "max_tool_calls": null,
    "previous_response_id": null,
    "prompt": null,
    "reasoning": null,
    "service_tier": "auto",
    "status": "completed",
    "text": null,
    "top_logprobs": null,
    "truncation": "disabled",
    "usage": {
        "input_tokens": 194,
        "input_tokens_details": {
            "cached_tokens": 0,
            "input_tokens_per_turn": [],
            "cached_tokens_per_turn": []
        },
        "output_tokens": 123,
        "output_tokens_details": {
            "reasoning_tokens": 0,
            "tool_output_tokens": 0,
            "output_tokens_per_turn": [],
            "tool_output_tokens_per_turn": []
        },
        "total_tokens": 317
    },
    "user": null,
    "input_messages": null,
    "output_messages": null
}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Guard None before trimming parsed content

When extract_tool_calls() returns content=None (e.g., OpenAIToolParser leaves final_content unset when a model emits only tool calls), the new content.strip() call will raise AttributeError and abort the request. Previously this branch safely returned (function_calls, None); now any tool-only response will crash automatic tool parsing instead of succeeding. A simple if content and content.strip() == "": ... avoids the regression.

Useful? React with 👍 / 👎.

[chaunceyjiang]

Why should we remove whitespace characters?

[qandrew]

@chaunceyjiang if we keep it, we get an extra output item like this, which is pointless to show for the user

        {
            "id": "msg_33be381a036e49e8abc0faead1e05752",
            "content": [
                {
                    "annotations": [],
                    "text": "\n\n",
                    "type": "output_text",
                    "logprobs": null
                }
            ],
            "role": "assistant",
            "status": "completed",
            "type": "message"
        },