tool-call(dsv3): Fixed a parse problem when there are multiple function definitions in tool_calls

[Missmiaom]

Motivation

When there are multiple function definitions in tool_calls, an error will be reported TypeError: can only concatenate str (not "dict") to str

Modifications

Add what is missing |tojson

Accuracy Tests

request body:

{
    "model": "deepseek-chat",
    "stream": false,
    "messages": [
        {
            "role": "system",
            "content": "You are a helpful assistant."
        },
        {
            "role": "user",
            "content": "武汉、珠海和北京的天气如何？"
        },
        {
            "role": "assistant",
            "content": "",
            "tool_calls": [
                {
                    "id": "111",
                    "type": "function",
                    "function": {
                        "name": "get_weather",
                        "arguments": "{\"location\": \"武汉\"}"
                    }
                }
                , {
                    "id": "222",
                    "type": "function",
                    "function": {
                        "name": "get_weather",
                        "arguments": "{\"location\": \"珠海\"}"
                    }
                }
                ,{
                    "id": "333",
                    "type": "function",
                    "function": {
                        "name": "get_weather",
                        "arguments": "{\"location\": \"北京\"}"
                    }
                }
            ]
        },
        {
            "role": "tool",
            "content": "{\"location\": \"武汉\", \"temperature\": 32, \"condition\": \"Sunny\", \"humidity\": 45, \"wind_speed\": 5}",
            "tool_call_id": "111"
        },
        {
            "role": "tool",
            "content": "{\"location\": \"珠海\", \"temperature\": 35, \"condition\": \"rainning\", \"humidity\": 80, \"wind_speed\": 10}",
            "tool_call_id": "222"
        },
        {
            "role": "tool",
            "content": "{\"location\": \"北京\", \"temperature\": 22, \"condition\": \"Sunny\", \"humidity\": 15, \"wind_speed\": 2}",
            "tool_call_id": "333"
        },
        {
            "role": "user",
            "content": "哪里天气最舒服？"
        }
    ],
    "tools": [
        {
            "type": "function",
            "function": {
                "name": "get_weather",
                "description": "Get weather of an location, the user shoud supply a location first",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "location": {
                            "type": "string",
                            "description": "The city and state, e.g. San Francisco, CA"
                        }
                    },
                    "required": [
                        "location"
                    ]
                }
            }
        }
    ]
}

[gemini-code-assist]

Summary of Changes

Hello @Missmiaom, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request addresses a critical parsing issue within the tool call mechanism, specifically when handling scenarios with multiple function definitions. The fix ensures that the arguments field within tool calls is correctly serialized to JSON, preventing a TypeError and improving the robustness of the tool calling feature.

Highlights

• Tool Call Parsing Fix: Resolved a TypeError that occurred when multiple function definitions were present in tool_calls by ensuring proper JSON serialization of function arguments.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request correctly fixes a TypeError that occurred with multiple tool calls in the deepseekv3 chat template. The issue was caused by a missing |tojson filter for function arguments on subsequent tool calls. While the fix is accurate, I've pointed out that the bug likely stemmed from code duplication within the template. I've suggested a refactoring to consolidate the duplicated logic, which would improve maintainability and prevent similar errors in the future.

[gemini-code-assist]

While adding |tojson is the correct fix, this line is now almost identical to lines 52 and 54. This code duplication likely caused the original bug, as the fix was not applied consistently across all instances. To improve maintainability and prevent such issues, I recommend refactoring the common logic into a Jinja variable.

For example, you could define a variable inside the for loop:

{%- set tool_call_str = '<｜tool call begin｜>' + tool['type'] + '<｜tool sep｜>' + tool['function']['name'] + '\n' + '```json' + '\n' + (tool['function']['arguments']|tojson) + '\n' + '```' + '<｜tool call end｜>' -%}

Then, you can simplify this line and others to use the variable, e.g., {{- '\n' + tool_call_str }}. This will make the template cleaner and less prone to inconsistencies.

[JustinTong0323]

make sense to me, similar with #9895