feat: 集成r1_filter至框架

[Raven95676]

Motivation

用户需要

Modifications

集成r1_filter至框架，并更名为thinking_filter

Check

• 😊 我的 Commit Message 符合良好的规范
• 👀 我的更改经过良好的测试
• 🤓 我确保没有引入新依赖库，或者引入了新依赖库的同时将其添加到了 requirements.txt 和 pyproject.toml 文件相应位置。
• 😮 我的更改没有引入恶意代码

好的，这是翻译成中文的 pull request 总结：

Sourcery 总结

集成 r1_filter 并将其重命名为核心框架中的 thinking_filter，并添加一个配置驱动的开关来显示或隐藏模型的内部推理。

新特性：

• 集成一个新的 thinking_filter 插件来处理 LLM 推理内容。
• 引入 display_reasoning_text 配置选项来切换是否在回复中显示模型的推理。

增强功能：

• 从 ‘reasoning’ 和 ‘reasoning_content’ 字段中提取推理，以支持不同的提供商。
• 当推理显示被禁用时，移除 <think>…</think> 标签。

Original summary in English

Summary by Sourcery

Integrate and rename the r1_filter into the core framework as thinking_filter, adding a config-driven switch to show or hide the model's internal reasoning.

New Features:

• Integrate a new thinking_filter plugin to handle LLM reasoning content.
• Introduce a display_reasoning_text configuration option to toggle showing the model's reasoning in replies.

Enhancements:

• Extract reasoning from both ‘reasoning’ and ‘reasoning_content’ fields for different providers.
• Strip out … tags when reasoning display is disabled.

[sourcery-ai]

## 审查者指南

此 PR 通过添加新的配置标志并注册一个 Star 插件，将 R1 推理过滤器作为“thinking_filter”集成到框架中，该插件有条件地包含或删除响应中的模型推理内容。

#### provider_settings 配置更新的实体关系图

```mermaid
erDiagram
    PROVIDER_SETTINGS {
        bool display_reasoning_text
    }
    PROVIDER_SETTINGS ||--o{ THINKING_FILTER : uses

新的 thinking_filter 插件的类图

classDiagram
    class R1Filter {
        - display_reasoning_text: bool
        + __init__(context: Context)
        + resp(event: AstrMessageEvent, response: LLMResponse)
    }
    R1Filter --|> Star
    class Star {
    }
    class Context {
    }
    class AstrMessageEvent {
    }
    class LLMResponse {
        + completion_text
        + raw_completion
    }
    class ChatCompletion {
        + choices
    }
    R1Filter o-- Context
    R1Filter ..> AstrMessageEvent
    R1Filter ..> LLMResponse
    LLMResponse o-- ChatCompletion

Loading

文件级别变更

变更	详情	文件
引入 display_reasoning_text 配置标志	在核心设置下添加默认布尔标志 在 provider_settings 下定义模式（描述、类型、提示）	astrbot/core/config/default.py
添加 thinking_filter Star 插件以进行推理控制	创建带有 @register 元数据的 Star 子类 从 init 中的上下文中读取 display_reasoning_text 实现 on_llm_response 以在启用时包含推理 使用正则表达式在禁用时删除 <think> 标签	packages/thinking_filter/main.py

---

提示和命令

与 Sourcery 交互

• 触发新的审查： 在 pull request 上评论 @sourcery-ai review。
• 继续讨论： 直接回复 Sourcery 的审查评论。
• 从审查评论生成 GitHub issue： 通过回复审查评论，要求 Sourcery 从审查评论创建一个 issue。 您还可以回复审查评论并使用 @sourcery-ai issue 从中创建一个 issue。
• 生成 pull request 标题： 在 pull request 标题中的任何位置写入 @sourcery-ai 以随时生成标题。 您也可以在 pull request 上评论 @sourcery-ai title 以随时（重新）生成标题。
• 生成 pull request 摘要： 在 pull request 正文中的任何位置写入 @sourcery-ai summary 以随时在你想要的任何位置生成 PR 摘要。 您也可以在 pull request 上评论 @sourcery-ai summary 以随时（重新）生成摘要。
• 生成审查者指南： 在 pull request 上评论 @sourcery-ai guide 以随时（重新）生成审查者指南。
• 解决所有 Sourcery 评论： 在 pull request 上评论 @sourcery-ai resolve 以解决所有 Sourcery 评论。 如果您已经解决了所有评论并且不想再看到它们，这将非常有用。
• 驳回所有 Sourcery 审查： 在 pull request 上评论 @sourcery-ai dismiss 以驳回所有现有的 Sourcery 审查。 如果你想用一个新的审查重新开始，这将特别有用 - 不要忘记评论 @sourcery-ai review 以触发一个新的审查！

自定义您的体验

访问您的 仪表板 以：

• 启用或禁用审查功能，例如 Sourcery 生成的 pull request 摘要、审查者指南等。
• 更改审查语言。
• 添加、删除或编辑自定义审查说明。
• 调整其他审查设置。

获得帮助

• 联系我们的支持团队 提出问题或反馈。
• 访问我们的 文档 以获取详细的指南和信息。
• 通过在 X/Twitter、LinkedIn 或 GitHub 上关注我们，与 Sourcery 团队保持联系。

```

Original review guide in English

Reviewer's Guide

This PR integrates the R1 reasoning filter into the framework as “thinking_filter” by adding a new configuration flag and registering a Star plugin that conditionally includes or strips model reasoning content in responses.

Entity relationship diagram for provider_settings config update

erDiagram
    PROVIDER_SETTINGS {
        bool display_reasoning_text
    }
    PROVIDER_SETTINGS ||--o{ THINKING_FILTER : uses

Loading

Class diagram for the new thinking_filter plugin

classDiagram
    class R1Filter {
        - display_reasoning_text: bool
        + __init__(context: Context)
        + resp(event: AstrMessageEvent, response: LLMResponse)
    }
    R1Filter --|> Star
    class Star {
    }
    class Context {
    }
    class AstrMessageEvent {
    }
    class LLMResponse {
        + completion_text
        + raw_completion
    }
    class ChatCompletion {
        + choices
    }
    R1Filter o-- Context
    R1Filter ..> AstrMessageEvent
    R1Filter ..> LLMResponse
    LLMResponse o-- ChatCompletion

Loading

File-Level Changes

Change	Details	Files
Introduce display_reasoning_text config flag	Add default boolean flag under core settings Define schema (description, type, hint) under provider_settings	astrbot/core/config/default.py
Add thinking_filter Star plugin for reasoning control	Create Star subclass with @register metadata Read display_reasoning_text from context in init Implement on_llm_response to include reasoning when enabled Use regex to strip tags when disabled	packages/thinking_filter/main.py

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[sourcery-ai]

嘿 @Raven95676 - 我已经查看了你的更改 - 这里有一些反馈：

• 澄清或更正else分支中的注释——它对“reason_content”的引用与实际的<think>标签过滤逻辑相冲突，可能会使未来的维护者感到困惑。
• 将推理提取和标签剥离块提取到单独的辅助函数中，以保持resp方法简洁并提高可读性。

Prompt for AI Agents

Please address the comments from this code review:
## Overall Comments
- Clarify or correct the comment in the `else` branch—its reference to “reason_content” conflicts with the actual `<think>` tag filtering logic and may confuse future maintainers.
- Extract the reasoning‐extraction and tag‐stripping blocks into separate helper functions to keep the `resp` method concise and improve readability.

## Individual Comments

### Comment 1
<location> `packages/thinking_filter/main.py:25` </location>
<code_context>
+        )
+
+    @filter.on_llm_response()
+    async def resp(self, event: AstrMessageEvent, response: LLMResponse):
+        if self.display_reasoning_text:
+            if (
</code_context>

<issue_to_address>
Consider refactoring repeated logic into helper functions and simplifying the main method for clarity.

```python
# helper functions at top of file
def extract_reasoning(message) -> str:
    for attr in ("reasoning", "reasoning_content"):
        val = getattr(message, attr, None)
        if val:
            return val
    return ""

def strip_think_tags(text: str) -> str:
    text = re.sub(r"<think>.*?</think>", "", text, flags=re.DOTALL)
    return text.replace("<think>", "").replace("</think>", "").strip()

# simplified resp()
@filter.on_llm_response()
async def resp(self, event: AstrMessageEvent, response: LLMResponse):
    if not self.display_reasoning_text:
        # fast‐exit when not displaying reasoning
        response.completion_text = strip_think_tags(response.completion_text or "")
        return

    raw = response.raw_completion
    if not (isinstance(raw, ChatCompletion) and raw.choices):
        return

    msg = raw.choices[0].message
    reasoning = extract_reasoning(msg)
    if reasoning:
        response.completion_text = f"🤔思考：{reasoning}\n\n{msg.content}"
    else:
        response.completion_text = msg.content
```
</issue_to_address>

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

<sub>Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.</sub>

Original comment in English

Hey @Raven95676 - I've reviewed your changes - here's some feedback:

• Clarify or correct the comment in the else branch—its reference to “reason_content” conflicts with the actual <think> tag filtering logic and may confuse future maintainers.
• Extract the reasoning‐extraction and tag‐stripping blocks into separate helper functions to keep the resp method concise and improve readability.

Prompt for AI Agents

Please address the comments from this code review:
## Overall Comments
- Clarify or correct the comment in the `else` branch—its reference to “reason_content” conflicts with the actual `<think>` tag filtering logic and may confuse future maintainers.
- Extract the reasoning‐extraction and tag‐stripping blocks into separate helper functions to keep the `resp` method concise and improve readability.

## Individual Comments

### Comment 1
<location> `packages/thinking_filter/main.py:25` </location>
<code_context>
+        )
+
+    @filter.on_llm_response()
+    async def resp(self, event: AstrMessageEvent, response: LLMResponse):
+        if self.display_reasoning_text:
+            if (
</code_context>

<issue_to_address>
Consider refactoring repeated logic into helper functions and simplifying the main method for clarity.

```python
# helper functions at top of file
def extract_reasoning(message) -> str:
    for attr in ("reasoning", "reasoning_content"):
        val = getattr(message, attr, None)
        if val:
            return val
    return ""

def strip_think_tags(text: str) -> str:
    text = re.sub(r"<think>.*?</think>", "", text, flags=re.DOTALL)
    return text.replace("<think>", "").replace("</think>", "").strip()

# simplified resp()
@filter.on_llm_response()
async def resp(self, event: AstrMessageEvent, response: LLMResponse):
    if not self.display_reasoning_text:
        # fast‐exit when not displaying reasoning
        response.completion_text = strip_think_tags(response.completion_text or "")
        return

    raw = response.raw_completion
    if not (isinstance(raw, ChatCompletion) and raw.choices):
        return

    msg = raw.choices[0].message
    reasoning = extract_reasoning(msg)
    if reasoning:
        response.completion_text = f"🤔思考：{reasoning}\n\n{msg.content}"
    else:
        response.completion_text = msg.content
```
</issue_to_address>

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

<sub>Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.</sub>