修复单元测试问题

Reviewed-on: #54

.drone.yml

@@ -13,7 +13,7 @@ steps:
   - name: submodules
     image: alpine/git
     commands:
-    - git submodule update --init --recursive
+      - git submodule update --init --recursive
   - name: 构建 Docker 镜像
     image: plugins/docker:latest
     privileged: true
@@ -30,7 +30,7 @@ steps:
     volumes:
       - name: docker-socket
         path: /var/run/docker.sock
-  - name: 在容器中测试插件加载
+  - name: 在容器中进行若干测试
     image: docker:dind
     privileged: true
     volumes:
@@ -38,6 +38,8 @@ steps:
         path: /var/run/docker.sock
     commands:
       - docker run --rm gitea.service.jazzwhom.top/mttu-developers/konabot:nightly-${DRONE_COMMIT_SHA} python scripts/test_plugin_load.py
+      - docker run --rm gitea.service.jazzwhom.top/mttu-developers/konabot:nightly-${DRONE_COMMIT_SHA} python scripts/test_playwright.py
+      - docker run --rm gitea.service.jazzwhom.top/mttu-developers/konabot:nightly-${DRONE_COMMIT_SHA} python -m pytest --cov-report term-missing:skip-covered
   - name: 发送构建结果到 ntfy
     image: parrazam/drone-ntfy
     when:
@@ -68,7 +70,7 @@ steps:
   - name: submodules
     image: alpine/git
     commands:
-    - git submodule update --init --recursive
+      - git submodule update --init --recursive
   - name: 构建并推送 Release Docker 镜像
     image: plugins/docker:latest
     privileged: true

.env.example

@@ -1,4 +1,4 @@
 ENVIRONMENT=dev
 PORT=21333
-
+DATABASE_PATH="./data/database.db"
 ENABLE_CONSOLE=true

.gitignore

@@ -1,4 +1,16 @@
+# 基本的数据文件，以及环境用文件
 /.env
 /data
+/pyrightconfig.json
+/pyrightconfig.toml
 
-__pycache__
+# 缓存文件
+__pycache__
+
+# 可能会偶然生成的 diff 文件
+/*.diff
+
+# 代码覆盖报告
+/.coverage
+/.coverage.db
+/htmlcov

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "assets/lexicon/THUOCL"]
 	path = assets/lexicon/THUOCL
 	url = https://github.com/thunlp/THUOCL.git
+[submodule "assets/oracle"]
+	path = assets/oracle
+	url = https://gitea.service.jazzwhom.top/mttu-developers/oracle-source.git

.sqls.yml

@@ -0,0 +1,6 @@
+lowercaseKeywords: false
+connections:
+  - driver: sqlite
+    dataSourceName: "./data/database.db"
+  - driver: sqlite
+    dataSourceName: "./data/perm.sqlite3"

AGENTS.md

@@ -0,0 +1,188 @@
+# AGENTS.md
+
+本文件面向两类协作者：
+
+- 手写代码的人类朋友
+- 会在此仓库中协助开发的 AI Agents
+
+这个项目以手写为主，欢迎协作，但请先理解这里的约束和结构，再开始改动。
+
+## 项目定位
+
+- 这是一个娱乐性质的、私域使用的 QQ Bot 项目。
+- 虽然主要用于熟人环境，但依然要按“不信任输入”的标准写代码。
+- 不要因为使用场景偏内部，就默认消息内容、安全边界、调用参数一定可靠。
+
+## 基本原则
+
+### 1. 默认不信任用户输入
+
+所有来自聊天消息、命令参数、平台事件等的输入，都应视为不可信。
+
+开发时至少注意以下几点：
+
+- 不假设输入类型正确，先校验再使用。
+- 不假设输入长度合理，注意超长文本、大量参数、异常嵌套结构。
+- 不假设输入内容安全，避免直接拼接到文件路径、SQL、shell 参数、HTML 或模板中。
+- 不假设用户一定按预期使用命令，错误输入要能优雅失败。
+- 对任何外部请求、文件读写、渲染、执行型逻辑，都要先考虑滥用风险。
+
+### 2. 优先保持现有风格
+
+- 这是一个以人工维护为主的项目，改动应尽量贴近现有写法。
+- 除非有明确收益，不要为了“看起来更现代”而大规模重构。
+- 新增能力时，优先复用已有通用模块，而不是重复造轮子。
+
+### 3. 小步修改，影响清晰
+
+- 尽量做局部、明确、可解释的改动。
+- 修改插件时，避免顺手改动无关插件。
+- 如果要调整公共模块，先确认是否会影响大量插件行为。
+
+## 仓库结构
+
+### `konabot/`
+
+核心代码目录。
+
+#### `konabot/common/`
+
+通用模块目录。
+
+- 放置可复用的基础能力、工具模块、公共逻辑。
+- 如果某段逻辑可能被多个插件共享，应优先考虑放到这里。
+- 修改这里的代码时，要额外关注兼容性，因为它可能被很多插件依赖。
+
+#### `konabot/docs/`
+
+Bot 内部文档系统使用的文档目录。
+
+- 这是给用户看的文档来源。
+- 文档会通过 `man` 指令被触发和展示。
+- 虽然文档文件通常使用 `.txt` 后缀，但内容可以按 markdown 风格书写。
+- `.md` 后缀文件会被忽略，因此 `.md` 更适合只留给仓库维护者阅读的附加说明。
+- 文档文件名就是用户查询时使用的指令名，应保持简洁、稳定、易理解。
+
+补充说明：
+
+- `konabot/docs/user/` 是直接面向用户检索的文档。
+- `konabot/docs/lib/` 更偏向维护者参考。
+- `konabot/docs/concepts/` 用于记录概念。
+- `konabot/docs/sys/` 用于特定范围可见的系统文档。
+
+#### `konabot/plugins/`
+
+插件目录。
+
+- 插件数量很多，是本项目最主要的功能承载位置。
+- 插件可以是单文件，也可以是文件夹形式。
+- 新增插件或修改插件时，请先观察相邻插件的组织方式，再决定采用单文件还是目录结构。
+- 如果逻辑已经明显超出单文件可维护范围，应拆成目录插件，不要把一个文件堆得过大。
+
+## 根目录文档
+
+### `docs/`
+
+仓库根目录下的 `docs/` 主要用于记录一些可以通用的模块说明和开发文档。
+
+- 这里的内容主要面向开发和维护。
+- 适合放公共模块说明、集成说明、配置说明、开发笔记。
+- 不要把面向 `man` 指令直接展示给用户的文档放到这里；那类内容应放在 `konabot/docs/` 下。
+
+## 对 AI Agents 的具体要求
+
+如果你是 AI Agent，请遵守以下约定：
+
+### 修改前
+
+- 先阅读将要修改的文件以及相关上下文，不要只凭文件名猜用途。
+- 先判断目标逻辑属于公共模块、用户文档，还是某个具体插件。
+- 如果需求可以在局部完成，就不要扩大改动范围。
+
+### 修改时
+
+- 优先延续现有命名、目录结构和编码风格。
+- 不要因为“顺手”而批量格式化整个项目。
+- 不要擅自重命名大量文件、移动目录、替换现有架构。
+- 涉及用户输入、路径、网络、数据库、渲染时，主动补上必要的校验与防御。
+- 如果要新增 `konabot/common/` 或其他会被多处依赖的模块，优先考虑 NoneBot2 框架下的依赖注入方式，而不是把全局状态或硬编码依赖散落到调用方。
+- 写文档时，区分清楚是给 `man` 系统看的，还是给仓库维护者看的。
+
+### 修改后
+
+- 检查改动是否误伤其他插件或公共模块。
+- 如果新增了用户可见功能，考虑是否需要补充 `konabot/docs/` 下对应文档。
+- 如果新增或调整了通用能力，考虑是否需要补充根目录 `docs/` 下的说明。
+
+## 插件开发建议
+
+- 单个插件内部优先保持自洽，不要把特定业务逻辑过早抽成公共模块。
+- 当多个插件开始重复同类逻辑时，再考虑上移到 `konabot/common/`。
+- 插件应尽量对异常输入有稳定反馈，而不是直接抛出难理解的错误。
+- 如果插件会访问外部服务，要考虑超时、失败降级和返回内容校验。
+
+### 最基本的用户交互书写建议
+
+- 先用清晰、可收敛的规则匹配消息，再进入处理逻辑，不要一上来就在 handler 里兜底解析所有输入。
+- 在 handler 里尽早提取纯文本、拆分命令和参数，并对缺失参数、非法参数、异常格式给出稳定反馈。
+- 如果用户输入只允许有限枚举值，先定义允许集合，再进行归一化和校验。
+- 输出优先保持简单直接；能一句话说明问题时，不要返回难懂的异常堆栈或过度技术化提示。
+- 涉及渲染、网络请求、图片生成等较重操作时，先确认输入合理，再执行昂贵逻辑。
+- 如果插件只是做单一交互，优先保持 handler 简短，把渲染、请求、转换等逻辑拆成独立函数。
+- 倾向于使用 `UniMessage` / `UniMsg` 这一套消息抽象来组织收发消息，而不是把平台细节和文本拼接散落在各处。
+- 倾向于显式构造返回消息并发送，而不是大量依赖 NoneBot2 原生的 `.finish()` 作为主要输出路径，除非该场景确实更简单清晰。
+
+### 关于公共能力的依赖方式
+
+- 新建通用能力时，优先设计成可注入、可替换、可测试的接口。
+- 如果一个模块未来可能被多个插件依赖，优先考虑 NoneBot2 的依赖注入，而不是让调用方手动维护重复的初始化流程。
+- 除非确有必要，不要让插件直接依赖隐藏的全局副作用。
+- 如果使用单例、缓存或全局管理器，要明确其生命周期、并发行为以及关闭时机。
+
+## 运行环境与部署限制
+
+这个项目默认会跑在 Docker 环境里，修改功能时请先意识到运行环境不是一台“什么都有”的开发机。
+
+### 容器环境
+
+- 运行时基础镜像是 `python:3.13-slim`，不是完整桌面 Linux；很多系统库默认不存在。
+- 项目运行依赖 Playwright Chromium、字体库、图形相关库，以及部分额外二进制工具。
+- 构建阶段和运行阶段是分离的；不要假设在 builder 里装过的系统包，runtime 里也一定可用。
+- 额外制品目前通过多阶段构建放进镜像，例如 `typst`。
+
+### Docker 相关要求
+
+- 如果你新增的 Python 依赖背后还需要 Linux 动态库、字体、图形库、编译工具或其他系统包，必须同步检查并在 `Dockerfile` 中补齐。
+- 不要只让本地虚拟环境能跑；要默认以容器可运行作为完成标准之一。
+- 如果新功能依赖系统命令、共享库、浏览器能力或字体，请在提交说明里明确写出原因。
+- `.dockerignore` 当前会排除 `/.env`、`/.git`、`/data` 等内容；不要依赖这些文件被复制进镜像。
+- 关于额外制品的管理，优先先阅读根目录文档 `docs/artifact.md`；适合统一管理的二进制或外部资源，倾向于复用 `konabot/common/artifact.py`，而不是在各插件里各自处理下载和校验。
+
+### 本地运行
+
+- 本地开发可参考 `justfile`，当前主要入口是 `just watch`。
+- 如果你的改动影响启动方式、依赖准备方式或运行命令，记得同步更新对应文档或脚本。
+
+## 分支与协作流程
+
+- 本项目托管在个人 Gitea 实例：`https://gitea.service.jazzwhom.top/mttu-developers/konabot`。
+- 如果需要创建 Pull Request，优先倾向使用 `tea` CLI：`https://gitea.com/gitea/tea`。
+- Pull Request 创建后，当前主要会有自动机器人做初步评审，项目维护者会手动查看；不要催促立即合并，也不要默认会马上进主分支。
+- 如果当前是在仓库本体上直接开发、而不是在 fork 上工作，尽量提醒用户不要直接在主分支持续改动，优先使用功能分支。
+- 除非用户明确要求，否则不要擅自把改动直接合并到主分支。
+
+## 文档编写建议
+
+### 面向 `man` 的文档
+
+- 放在 `konabot/docs/` 对应子目录。
+- 文件名直接对应用户查询名称。
+- 建议内容简洁，优先说明“做什么、怎么用、示例、注意事项”。
+- 使用 `.txt` 后缀；内容可以写成接近 markdown 的可读格式。
+
+### 面向开发者的文档
+
+- 放在仓库根目录 `docs/`。
+- 主要描述公共模块、配置方法、设计说明、维护经验。
+- 可以使用 `.md`。
+

Dockerfile

@@ -1,8 +1,21 @@
+FROM alpine:latest AS artifacts
+
+RUN apk add --no-cache curl xz
+WORKDIR /tmp
+
+RUN mkdir -p /artifacts
+RUN curl -L -o typst.tar.xz "https://github.com/typst/typst/releases/download/v0.14.2/typst-x86_64-unknown-linux-musl.tar.xz" \
+    && tar -xJf typst.tar.xz \
+    && mv typst-x86_64-unknown-linux-musl/typst /artifacts
+
+RUN chmod -R +x /artifacts/
+
+
 FROM python:3.13-slim AS base
 
 ENV VIRTUAL_ENV=/app/.venv \
     PATH="/app/.venv/bin:$PATH" \
-    PLAYWRIGHT_BROWSERS_PATH=0
+    PLAYWRIGHT_BROWSERS_PATH=/usr/lib/pw-browsers
 
 # 安装所有都需要的底层依赖
 RUN apt-get update && \
@@ -19,7 +32,6 @@ RUN apt-get update && \
     && rm -rf /var/lib/apt/lists/*
 
 
-
 FROM base AS builder
 
 # 安装构建依赖
@@ -27,23 +39,19 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     build-essential cmake git \
     && rm -rf /var/lib/apt/lists/*
 
-ENV POETRY_NO_INTERACTION=1 \
-    POETRY_VIRTUALENVS_IN_PROJECT=1 \
-    POETRY_VIRTUALENVS_CREATE=1 \
-    POETRY_CACHE_DIR=/tmp/poetry_cache
+RUN pip install --no-cache-dir uv
 
 WORKDIR /app
 
-RUN pip install --no-cache-dir poetry
-
 COPY pyproject.toml poetry.lock ./
-RUN python -m poetry install --no-root && rm -rf $POETRY_CACHE_DIR
+RUN uv sync --no-install-project
 
 
 
 FROM base AS runtime
 
 COPY --from=builder ${VIRTUAL_ENV} ${VIRTUAL_ENV}
+COPY --from=artifacts /artifacts/ /usr/local/bin/
 
 WORKDIR /app
 

README.md

@@ -71,12 +71,16 @@ code .
 
 详见[konabot-web 配置文档](/docs/konabot-web.md)
 
+#### 数据库配置
+
+本项目使用SQLite作为数据库，默认数据库文件位于`./data/database.db`。可以通过设置`DATABASE_PATH`环境变量来指定其他位置。
+
 ### 运行
 
 使用命令行手动启动 Bot：
 
 ```bash
-poetry run watchfiles bot.main . --filter scripts.watch_filter.filter
+poetry run just watch
 ```
 
 如果你不希望自动重载，只是想运行 Bot，可以直接运行：
@@ -91,3 +95,22 @@ poetry run python bot.py
     - [事件响应器](https://nonebot.dev/docs/tutorial/matcher)
     - [事件处理](https://nonebot.dev/docs/tutorial/handler)
     - [Alconna 插件](https://nonebot.dev/docs/best-practice/alconna/)
+
+## 代码测试
+
+本项目使用 pytest 进行自动化测试，你可以把你的测试代码放在 `./tests` 目录下。
+
+使用命令行执行测试：
+
+```bash
+poetry run just test
+```
+
+使用命令行，在浏览器查看测试覆盖率报告：
+
+```bash
+poetry run just coverage
+# 此时会打开一个 :8000 端口的 Web 服务器
+# 你可以在 http://localhost:8000 查看覆盖率报告
+# 在控制台使用 Ctrl+C 关闭这个 Web 服务器
+```

bot.py

@@ -10,6 +10,8 @@ from nonebot.adapters.onebot.v11 import Adapter as OnebotAdapter
 from konabot.common.log import init_logger
 from konabot.common.nb.exc import BotExceptionMessage
 from konabot.common.path import LOG_PATH
+from konabot.common.database import get_global_db_manager
+
 
 dotenv.load_dotenv()
 env = os.environ.get("ENVIRONMENT", "prod")
@@ -20,19 +22,25 @@ env_enable_minecraft = os.environ.get("ENABLE_MINECRAFT", "none")
 
 
 def main():
-    if env.upper() == 'DEBUG' or env.upper() == 'DEV':
-        console_log_level = 'DEBUG'
+    if env.upper() == "DEBUG" or env.upper() == "DEV":
+        console_log_level = "DEBUG"
     else:
-        console_log_level = 'INFO'
-    init_logger(LOG_PATH, [
-        BotExceptionMessage,
-    ], console_log_level=console_log_level)
+        console_log_level = "INFO"
+    init_logger(
+        LOG_PATH,
+        [
+            BotExceptionMessage,
+        ],
+        console_log_level=console_log_level,
+    )
 
     nonebot.init()
 
     driver = nonebot.get_driver()
 
-    if (env != "prod" and env != "test" and env_enable_console.upper() != "FALSE") or (env_enable_console.upper() == "TRUE"):
+    if (env != "prod" and env != "test" and env_enable_console.upper() != "FALSE") or (
+        env_enable_console.upper() == "TRUE"
+    ):
         driver.register_adapter(ConsoleAdapter)
 
     if env_enable_qq.upper() == "TRUE":
@@ -48,7 +56,19 @@ def main():
     nonebot.load_plugins("konabot/plugins")
     nonebot.load_plugin("nonebot_plugin_analysis_bilibili")
 
+    from konabot.common import permsys
+
+    permsys.create_startup()
+
+    # 注册关闭钩子
+    @driver.on_shutdown
+    async def _():
+        # 关闭全局数据库管理器
+        db_manager = get_global_db_manager()
+        await db_manager.close_all_connections()
+
     nonebot.run()
 
+
 if __name__ == "__main__":
     main()

docs/artifact.md

@@ -0,0 +1,26 @@
+# artifact 模块说明
+
+`konabot/common/artifact.py` 用于管理项目运行过程中依赖的额外制品，尤其是二进制文件、外部工具和按平台区分的运行时资源。
+
+## 适用场景
+
+- 某个插件或公共模块依赖额外下载的可执行文件或二进制资源。
+- 依赖需要按操作系统或架构区分。
+- 希望在启动时统一检测、按需下载并校验哈希。
+
+如果额外制品适合在镜像构建阶段直接打包进 Docker 镜像，也可以在 `Dockerfile` 中通过多阶段构建处理；但对于需要在运行环境按平台管理、懒下载或统一校验的资源，优先考虑复用 `artifact.py`。
+
+## 推荐做法
+
+- 新增额外制品时，先判断它更适合放进镜像构建阶段，还是更适合交给 `artifact.py` 管理。
+- 如果该资源会被多个插件或环境复用，倾向于统一通过 `ArtifactDepends` 和 `register_artifacts(...)` 管理。
+- 为下载资源提供稳定来源，并填写 `sha256` 校验值，不要只校验“能不能下载下来”。
+- 使用 `required_os` 和 `required_arch` 限制平台，避免无意义下载。
+- 需要代理时，确认其行为与当前 NoneBot2 配置兼容。
+
+## 注意事项
+
+- 不要把是否存在额外制品的判断散落到多个插件里各自实现。
+- 不要跳过哈希校验，除非该资源确实无法提供稳定校验值，并且有明确理由。
+- 如果一个新能力除了额外制品，还依赖 Linux 动态库、字体、浏览器或系统命令，仍然需要同步检查并更新 `Dockerfile`。
+- 如果镜像构建和运行阶段都依赖该制品，要分别确认 builder 和 runtime 的可用性。

docs/database.md

@@ -0,0 +1,223 @@
+# 数据库系统使用文档
+
+本文档详细介绍了本项目中使用的异步数据库系统，包括其架构设计、使用方法和最佳实践。
+
+## 系统概述
+
+本项目的数据库系统基于 `aiosqlite` 库构建，提供了异步的 SQLite 数据库访问接口。系统主要特性包括：
+
+1. **异步操作**：完全支持异步/await模式，适配NoneBot2框架
+2. **连接池**：内置连接池机制，提高数据库访问性能
+3. **参数化查询**：支持安全的参数化查询，防止SQL注入
+4. **SQL文件支持**：可以直接执行SQL文件中的脚本
+5. **类型支持**：支持 `pathlib.Path` 和 `str` 类型的路径参数
+
+## 核心类和方法
+
+### DatabaseManager 类
+
+`DatabaseManager` 是数据库操作的核心类，提供了以下主要方法：
+
+#### 初始化
+```python
+from konabot.common.database import DatabaseManager
+from pathlib import Path
+
+# 使用默认数据库路径
+db = DatabaseManager()
+
+# 指定了义数据库路径
+db = DatabaseManager("./data/myapp.db")
+db = DatabaseManager(Path("./data/myapp.db"))
+```
+
+#### 查询操作
+```python
+# 执行查询语句并返回结果
+results = await db.query("SELECT * FROM users WHERE age > ?", (18,))
+
+# 从SQL文件执行查询
+results = await db.query_by_sql_file("./sql/get_users.sql", (18,))
+```
+
+#### 执行操作
+```python
+# 执行非查询语句
+await db.execute("INSERT INTO users (name, email) VALUES (?, ?)", ("张三", "zhangsan@example.com"))
+
+# 执行SQL脚本（不带参数）
+await db.execute_script("""
+    CREATE TABLE IF NOT EXISTS users (
+        id INTEGER PRIMARY KEY,
+        name TEXT NOT NULL,
+        email TEXT UNIQUE
+    );
+    INSERT INTO users (name, email) VALUES ('测试用户', 'test@example.com');
+""")
+
+# 从SQL文件执行非查询语句
+await db.execute_by_sql_file("./sql/create_tables.sql")
+
+# 带参数执行SQL文件
+await db.execute_by_sql_file("./sql/insert_user.sql", ("张三", "zhangsan@example.com"))
+
+# 执行多条语句（每条语句使用相同参数）
+await db.execute_many("INSERT INTO users (name, email) VALUES (?, ?)", [
+    ("张三", "zhangsan@example.com"),
+    ("李四", "lisi@example.com"),
+    ("王五", "wangwu@example.com")
+])
+
+# 从SQL文件执行多条语句（每条语句使用相同参数）
+await db.execute_many_values_by_sql_file("./sql/batch_insert.sql", [
+    ("张三", "zhangsan@example.com"),
+    ("李四", "lisi@example.com")
+])
+```
+
+## SQL文件处理机制
+
+### 单语句SQL文件
+```sql
+-- insert_user.sql
+INSERT INTO users (name, email) VALUES (?, ?);
+```
+
+```python
+# 使用方式
+await db.execute_by_sql_file("./sql/insert_user.sql", ("张三", "zhangsan@example.com"))
+```
+
+### 多语句SQL文件
+```sql
+-- setup.sql
+CREATE TABLE IF NOT EXISTS users (
+    id INTEGER PRIMARY KEY,
+    name TEXT NOT NULL,
+    email TEXT UNIQUE
+);
+
+CREATE TABLE IF NOT EXISTS profiles (
+    user_id INTEGER,
+    age INTEGER,
+    FOREIGN KEY (user_id) REFERENCES users(id)
+);
+```
+
+```python
+# 使用方式
+await db.execute_by_sql_file("./sql/setup.sql")
+```
+
+### 多语句带不同参数的SQL文件
+```sql
+-- batch_operations.sql
+INSERT INTO users (name, email) VALUES (?, ?);
+INSERT INTO profiles (user_id, age) VALUES (?, ?);
+```
+
+```python
+# 使用方式
+await db.execute_by_sql_file("./sql/batch_operations.sql", [
+    ("张三", "zhangsan@example.com"),  # 第一条语句的参数
+    (1, 25)  # 第二条语句的参数
+])
+```
+
+## 最佳实践
+
+### 1. 数据库表设计
+```sql
+-- 推荐的表设计实践
+CREATE TABLE IF NOT EXISTS example_table (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    name TEXT NOT NULL,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+    updated_at DATETIME DEFAULT CURRENT_TIMESTAMP
+);
+```
+
+### 2. SQL文件组织
+建议按照功能模块组织SQL文件：
+```
+plugin/
+├── sql/
+│   ├── create_tables.sql
+│   ├── insert_data.sql
+│   ├── update_data.sql
+│   └── query_data.sql
+└── __init__.py
+```
+
+### 3. 错误处理
+```python
+try:
+    results = await db.query("SELECT * FROM users WHERE id = ?", (user_id,))
+except Exception as e:
+    logger.error(f"数据库查询失败: {e}")
+    # 处理错误情况
+```
+
+### 4. 连接管理
+```python
+# 在应用启动时初始化
+db_manager = DatabaseManager()
+
+# 在应用关闭时清理连接
+async def shutdown():
+    await db_manager.close_all_connections()
+```
+
+## 高级特性
+
+### 连接池配置
+```python
+class DatabaseManager:
+    def __init__(self, db_path: Optional[Union[str, Path]] = None):
+        # 连接池大小配置
+        self._pool_size = 5  # 可根据需要调整
+```
+
+### 事务支持
+```python
+# 通过execute方法的自动提交机制支持事务
+await db.execute("BEGIN TRANSACTION")
+try:
+    await db.execute("INSERT INTO users (name) VALUES (?)", ("张三",))
+    await db.execute("INSERT INTO profiles (user_id, age) VALUES (?, ?)", (1, 25))
+    await db.execute("COMMIT")
+except Exception:
+    await db.execute("ROLLBACK")
+    raise
+```
+
+## 注意事项
+
+1. **异步环境**：所有数据库操作都必须在异步环境中执行
+2. **参数安全**：始终使用参数化查询，避免SQL注入
+3. **资源管理**：确保在应用关闭时调用 `close_all_connections()`
+4. **SQL解析**：使用 `sqlparse` 库准确解析SQL语句，正确处理包含分号的字符串和注释
+5. **错误处理**：适当处理数据库操作可能抛出的异常
+
+## 常见问题
+
+### Q: 如何处理数据库约束错误？
+A: 确保SQL语句中的字段名正确引用，特别是保留字需要使用双引号包围：
+```sql
+CREATE TABLE air_conditioner (
+    id VARCHAR(128) PRIMARY KEY, 
+    "on" BOOLEAN NOT NULL,  -- 使用双引号包围保留字
+    temperature REAL NOT NULL
+);
+```
+
+### Q: 如何处理多个语句和参数的匹配？
+A: 当SQL文件包含多个语句时，参数应该是参数列表，每个语句对应一个参数元组：
+```python
+await db.execute_by_sql_file("./sql/batch.sql", [
+    ("参数1", "参数2"),  # 第一个语句的参数
+    ("参数3", "参数4")   # 第二个语句的参数
+])
+```
+
+通过遵循这些指南和最佳实践，您可以充分利用本项目的异步数据库系统，构建高性能、安全的数据库应用。

docs/permsys.md

@@ -0,0 +1,235 @@
+# 权限系统 `konabot.common.permsys`
+
+本文档面向维护者，说明 `konabot/common/permsys` 模块的职责、数据模型、权限解析规则，以及在插件中接入的推荐方式。
+
+## 模块目标
+
+`permsys` 提供了一套简单的、可继承的权限系统，用于回答两个问题：
+
+1. 某个事件对应的主体是谁。
+2. 该主体是否拥有某项权限。
+
+它适合处理 bot 内部的功能开关、管理权限、平台级授权等场景。
+
+当前模块由以下几部分组成：
+
+- `konabot/common/permsys/__init__.py`
+  - 暴露 `PermManager`、`DepPermManager`、`require_permission`
+  - 负责数据库初始化、启动迁移、超级管理员默认授权
+- `konabot/common/permsys/entity.py`
+  - 定义 `PermEntity`
+  - 将事件转换为可查询的实体链
+- `konabot/common/permsys/repo.py`
+  - 封装 SQLite 读写
+- `konabot/common/permsys/migrates/`
+  - 存放迁移 SQL
+- `konabot/common/permsys/sql/`
+  - 存放查询与更新 SQL
+
+## 核心概念
+
+### 1. `PermEntity`
+
+`PermEntity` 是权限系统中的最小主体标识：
+
+```python
+PermEntity(platform: str, entity_type: str, external_id: str)
+```
+
+示例：
+
+- `PermEntity("sys", "global", "global")`
+- `PermEntity("ob11", "group", "123456")`
+- `PermEntity("ob11", "user", "987654")`
+
+其中：
+
+- `platform` 表示来源平台，如 `sys`、`ob11`、`discord`
+- `entity_type` 表示主体类型，如 `global`、`group`、`user`
+- `external_id` 表示平台侧的外部标识
+
+### 2. 实体链
+
+权限判断不是只看单个实体，而是看一条“实体链”。
+
+以 `get_entity_chain_of_entity()` 为例，传入一个具体实体时，返回的链为：
+
+```python
+[
+    PermEntity(platform, entity_type, external_id),
+    PermEntity(platform, "global", "global"),
+    PermEntity("sys", "global", "global"),
+]
+```
+
+这意味着权限会优先读取更具体的主体，再回退到平台全局，最后回退到系统全局。
+
+`get_entity_chain(event)` 则会根据事件类型自动构造链。例如：
+
+- OneBot V11 群消息：用户 -> 群 -> 平台全局 -> 系统全局
+- OneBot V11 私聊：用户 -> 平台全局 -> 系统全局
+- Discord 频道消息：用户/频道/服务器 -> 平台全局 -> 系统全局
+- Console：控制台用户/频道 -> 平台全局 -> 系统全局
+
+注意：当前 `entity.py` 中的具体链顺序与字段命名应以实现为准；修改这里时要评估现有权限继承是否会被破坏。
+
+### 3. 权限键
+
+权限键使用点分结构，例如：
+
+- `admin`
+- `plugin.weather`
+- `plugin.weather.use`
+
+检查时会自动做前缀回退。以 `plugin.weather.use` 为例，查询顺序是：
+
+1. `plugin.weather.use`
+2. `plugin.weather`
+3. `plugin`
+4. `*`
+
+因此，`*` 可以看作兜底总权限。
+
+## 权限解析规则
+
+`PermManager.check_has_permission_info()` 的逻辑可以概括为：
+
+1. 先把输入转换成实体链。
+2. 对权限键做逐级回退，同时追加 `*`。
+3. 在数据库中批量查出链上所有实体、所有候选键的显式记录。
+4. 按“实体越具体越优先、权限键越具体越优先”的顺序，返回第一条命中的记录。
+
+若没有任何显式记录：
+
+- `check_has_permission_info()` 返回 `None`
+- `check_has_permission()` 返回 `False`
+
+这表示本系统默认是“未授权即拒绝”。
+
+## 数据存储
+
+模块使用 SQLite，默认数据库文件位于：
+
+- `data/perm.sqlite3`
+
+启动时会执行迁移：
+
+- `create_startup()` 在 NoneBot 启动事件中调用 `execute_migration()`
+
+权限值支持三态：
+
+- `True`：显式允许
+- `False`：显式拒绝
+- `None`：删除/清空该层的显式设置，让判断重新回退到继承链
+
+`repo.py` 中的 `update_perm_info()` 会将这个三态直接写入数据库。
+
+## 超级管理员注入
+
+在启动阶段，`create_startup()` 会读取 `konabot.common.nb.is_admin.cfg.admin_qq_account`，并为这些 QQ 账号写入：
+
+```python
+PermEntity("ob11", "user", str(account)), "*", True
+```
+
+也就是说，配置中的超级管理员会直接拥有全部权限。
+
+这属于启动时自动灌入的保底策略，不依赖手工授权命令。
+
+## 在插件中使用
+
+### 1. 直接做权限检查
+
+```python
+from konabot.common.permsys import DepPermManager
+
+
+async def handler(pm: DepPermManager, event):
+    ok = await pm.check_has_permission(event, "plugin.example.use")
+    if not ok:
+        return
+```
+
+适合需要在处理流程中动态决定权限键的场景。
+
+### 2. 挂到 Rule 上做准入控制
+
+```python
+from nonebot_plugin_alconna import Alconna, on_alconna
+from konabot.common.permsys import require_permission
+
+
+cmd = on_alconna(
+    Alconna("example"),
+    rule=require_permission("plugin.example.use"),
+)
+```
+
+适合命令入口明确、未通过时直接拦截的场景。
+
+### 3. 更新权限
+
+```python
+from konabot.common.permsys import DepPermManager
+from konabot.common.permsys.entity import PermEntity
+
+
+await pm.update_permission(
+    PermEntity("ob11", "group", "123456"),
+    "plugin.example.use",
+    True,
+)
+```
+
+建议只在专门的管理插件中开放写权限，避免普通功能插件到处分散改表。
+
+## `perm_manage` 插件与本模块的关系
+
+`konabot/plugins/perm_manage/__init__.py` 是本模块当前的管理入口，提供：
+
+- `konaperm list`：列出实体链上已有的显式权限记录
+- `konaperm get`：查看某个权限最终命中的记录
+- `konaperm set`：写入 allow/deny/null
+
+这个插件本身使用 `require_permission("admin")` 保护，因此只有拥有 `admin` 权限的主体才能管理权限。
+
+## 接入建议
+
+### 权限键命名
+
+建议使用稳定、可扩展的分层键名：
+
+- 推荐：`plugin.xxx`、`plugin.xxx.action`
+- 不推荐：含糊的单词或临时字符串
+
+这样才能利用前缀回退机制做批量授权。
+
+### 输入安全
+
+虽然这个项目偏内部使用，但权限键、实体类型、外部 ID 仍然应视为不可信输入：
+
+- 不要把聊天输入直接拼到 SQL 中
+- 不要让任意用户可随意构造高权限写入
+- 对可写命令至少做权限保护和必要校验
+
+### 改动兼容性
+
+以下改动都可能影响全局权限行为，修改前应充分评估：
+
+- 更改实体链顺序
+- 更改默认兜底键 `*` 的语义
+- 更改 `None` 的处理方式
+- 更改启动时超级管理员注入逻辑
+
+## 调试建议
+
+- 先用 `konaperm get ...` 确认某个权限最终命中了哪一层
+- 再用 `konaperm list ...` 查看该实体链上有哪些显式记录
+- 若表现异常，检查是否是更上层实体或更宽泛权限键提前命中
+
+## 相关文件
+
+- `konabot/common/permsys/__init__.py`
+- `konabot/common/permsys/entity.py`
+- `konabot/common/permsys/repo.py`
+- `konabot/plugins/perm_manage/__init__.py`

justfile

@@ -0,0 +1,9 @@
+watch:
+    poetry run watchfiles bot.main . --filter scripts.watch_filter.filter
+
+test:
+    poetry run pytest --cov-report term-missing:skip-covered
+
+coverage:
+    poetry run pytest --cov-report html
+    python -m http.server -d htmlcov

konabot/common/apis/ali_content_safety.py

@@ -0,0 +1,92 @@
+import asyncio
+import json
+
+from alibabacloud_green20220302.client import Client as AlibabaGreenClient
+from alibabacloud_green20220302.models import TextModerationPlusRequest
+from alibabacloud_tea_openapi.models import Config as AlibabaTeaConfig
+from loguru import logger
+from pydantic import BaseModel
+
+import nonebot
+
+
+class AlibabaGreenPluginConfig(BaseModel):
+    module_aligreen_enable: bool = False
+    module_aligreen_access_key_id: str = ""
+    module_aligreen_access_key_secret: str = ""
+    module_aligreen_region_id: str = "cn-shenzhen"
+    module_aligreen_endpoint: str = "green-cip.cn-shenzhen.aliyuncs.com"
+    module_aligreen_service: str = "llm_query_moderation"
+
+
+class AlibabaGreen:
+    _client: AlibabaGreenClient | None = None
+    _config: AlibabaGreenPluginConfig | None = None
+
+    @staticmethod
+    def get_client() -> AlibabaGreenClient:
+        assert AlibabaGreen._client is not None
+        return AlibabaGreen._client
+
+    @staticmethod
+    def get_config() -> AlibabaGreenPluginConfig:
+        assert AlibabaGreen._config is not None
+        return AlibabaGreen._config
+
+    @staticmethod
+    def init():
+        config = nonebot.get_plugin_config(AlibabaGreenPluginConfig)
+        AlibabaGreen._config = config
+        if not config.module_aligreen_enable:
+            logger.info("该环境未启用阿里内容审查，跳过初始化")
+            return
+        AlibabaGreen._client = AlibabaGreenClient(AlibabaTeaConfig(
+            access_key_id=config.module_aligreen_access_key_id,
+            access_key_secret=config.module_aligreen_access_key_secret,
+            connect_timeout=10000,
+            read_timeout=3000,
+            region_id=config.module_aligreen_region_id,
+            endpoint=config.module_aligreen_endpoint,
+        ))
+
+    @staticmethod
+    def _detect_sync(content: str) -> bool:
+        if len(content) == 0:
+            return True
+        if not AlibabaGreen.get_config().module_aligreen_enable:
+            logger.debug("该环境未启用阿里内容审查，直接跳过")
+            return True
+
+        client = AlibabaGreen.get_client()
+        try:
+            response = client.text_moderation_plus(TextModerationPlusRequest(
+                service=AlibabaGreen.get_config().module_aligreen_service,
+                service_parameters=json.dumps({
+                    "content": content,
+                }),
+            ))
+            if response.status_code == 200:
+                result = response.body
+                logger.info(f"检测违规内容 API 调用成功：{result}")
+                risk_level: str = result.data.risk_level or "none"
+                if risk_level == "high":
+                    return False
+                return True
+            logger.error(f"检测违规内容 API 调用失败：{response}")
+            return True
+        except Exception as e:
+            logger.error("检测违规内容 API 调用失败")
+            logger.exception(e)
+            return True
+
+    @staticmethod
+    async def detect(content: str) -> bool:
+        return await asyncio.to_thread(AlibabaGreen._detect_sync, content)
+
+
+driver = nonebot.get_driver()
+
+@driver.on_startup
+async def _():
+    AlibabaGreen.init()
+

konabot/common/artifact.py

@@ -0,0 +1,112 @@
+import asyncio
+import aiohttp
+import hashlib
+import platform
+
+from dataclasses import dataclass
+from pathlib import Path
+
+import nonebot
+from loguru import logger
+from nonebot.adapters.discord.config import Config as DiscordConfig
+from pydantic import BaseModel
+
+
+@dataclass
+class ArtifactDepends:
+    url: str
+    sha256: str
+    target: Path
+
+    required_os: str | None = None
+    "示例值：Windows, Linux, Darwin"
+
+    required_arch: str | None = None
+    "示例值：AMD64, x86_64, arm64"
+
+    use_proxy: bool = True
+    "网络问题，赫赫；使用的是 Discord 模块配置的 proxy"
+
+    def is_corresponding_platform(self) -> bool:
+        if self.required_os is not None:
+            if self.required_os.lower() != platform.system().lower():
+                return False
+        if self.required_arch is not None:
+            if self.required_arch.lower() != platform.machine().lower():
+                return False
+        return True
+
+
+class Config(BaseModel):
+    prefetch_artifact: bool = False
+    "是否提前下载好二进制依赖"
+
+
+artifact_list = []
+
+
+driver = nonebot.get_driver()
+config = nonebot.get_plugin_config(Config)
+
+@driver.on_startup
+async def _():
+    if config.prefetch_artifact:
+        logger.info("启动检测中：正在检测需求的二进制是否下载")
+        semaphore = asyncio.Semaphore(10)
+        async def _task(artifact: ArtifactDepends):
+            async with semaphore:
+                await ensure_artifact(artifact)
+
+        tasks: set[asyncio.Task] = set()
+        for a in artifact_list:
+            tasks.add(asyncio.Task(_task(a)))
+        await asyncio.gather(*tasks, return_exceptions=False)
+        logger.info("检测好了")
+
+
+async def download_artifact(artifact: ArtifactDepends):
+    proxy = None
+    if artifact.use_proxy:
+        discord_config = nonebot.get_plugin_config(DiscordConfig)
+        proxy = discord_config.discord_proxy
+
+    if proxy is not None:
+        logger.info(f"正在使用 Proxy 下载 TARGET={artifact.target} PROXY={proxy}")
+    else:
+        logger.info(f"正在下载 TARGET={artifact.target}")
+
+    async with aiohttp.ClientSession(proxy=proxy) as client:
+        result = await client.get(artifact.url)
+        if result.status != 200:
+            logger.warning(f"已经下载了二进制，但是注意服务器没有返回 200！ URL={artifact.url} TARGET={artifact.target} CODE={result.status}")
+        data = await result.read()
+    artifact.target.write_bytes(data)
+    if not platform.system().lower() == 'windows':
+        artifact.target.chmod(0o755)
+
+    logger.info(f"下载好了 TARGET={artifact.target} URL={artifact.url}")
+    m = hashlib.sha256(artifact.target.read_bytes())
+    if m.hexdigest().lower() != artifact.sha256.lower():
+        logger.warning(f"下载到的二进制的 sha256 与需求不同 TARGET={artifact.target} REQUESTED={artifact.sha256} ACTUAL={m.hexdigest()}")
+
+
+async def ensure_artifact(artifact: ArtifactDepends):
+    if not artifact.is_corresponding_platform():
+        return
+
+    if not artifact.target.exists():
+        logger.info(f"二进制依赖 {artifact.target} 不存在")
+        if not artifact.target.parent.exists():
+            artifact.target.parent.mkdir(parents=True, exist_ok=True)
+        await download_artifact(artifact)
+    else:
+        m = hashlib.sha256(artifact.target.read_bytes())
+        if m.hexdigest().lower() != artifact.sha256.lower():
+            logger.info(f"二进制依赖 {artifact.target} 的哈希无法对应需求的哈希，准备重新下载")
+            artifact.target.unlink()
+            await download_artifact(artifact)
+
+
+def register_artifacts(*artifacts: ArtifactDepends):
+    artifact_list.extend(artifacts)
+

konabot/common/database/__init__.py

@@ -0,0 +1,231 @@
+from contextlib import asynccontextmanager
+import os
+import asyncio
+import sqlparse
+from pathlib import Path
+from typing import List, Dict, Any, Optional, Union, TYPE_CHECKING
+
+import aiosqlite
+
+if TYPE_CHECKING:
+    from . import DatabaseManager
+
+# 全局数据库管理器实例
+_global_db_manager: Optional["DatabaseManager"] = None
+
+
+def get_global_db_manager() -> "DatabaseManager":
+    """获取全局数据库管理器实例"""
+    global _global_db_manager
+    if _global_db_manager is None:
+        from . import DatabaseManager
+
+        _global_db_manager = DatabaseManager()
+    return _global_db_manager
+
+
+def close_global_db_manager() -> None:
+    """关闭全局数据库管理器实例"""
+    global _global_db_manager
+    if _global_db_manager is not None:
+        # 注意：这个函数应该在async环境中调用close_all_connections
+        _global_db_manager = None
+
+
+class DatabaseManager:
+    """异步数据库管理器"""
+
+    def __init__(self, db_path: Optional[Union[str, Path]] = None, pool_size: int = 5):
+        """
+        初始化数据库管理器
+
+        Args:
+            db_path: 数据库文件路径，支持str和Path类型
+            pool_size: 连接池大小
+        """
+        if db_path is None:
+            self.db_path = os.environ.get("DATABASE_PATH", "./data/database.db")
+        else:
+            self.db_path = str(db_path) if isinstance(db_path, Path) else db_path
+
+        # 连接池
+        self._connection_pool = []
+        self._pool_size = pool_size
+        self._lock = asyncio.Lock()
+        self._in_use = set()  # 跟踪正在使用的连接
+
+    async def _get_connection(self) -> aiosqlite.Connection:
+        """从连接池获取连接"""
+        async with self._lock:
+            # 尝试从池中获取现有连接
+            while self._connection_pool:
+                conn = self._connection_pool.pop()
+                # 检查连接是否仍然有效
+                try:
+                    await conn.execute("SELECT 1")
+                    self._in_use.add(conn)
+                    return conn
+                except:
+                    # 连接已失效，关闭它
+                    try:
+                        await conn.close()
+                    except:
+                        pass
+
+            # 如果连接池为空，创建新连接
+            conn = await aiosqlite.connect(self.db_path)
+            await conn.execute("PRAGMA foreign_keys = ON")
+            self._in_use.add(conn)
+            return conn
+
+    async def _return_connection(self, conn: aiosqlite.Connection) -> None:
+        """将连接返回到连接池"""
+        async with self._lock:
+            self._in_use.discard(conn)
+            if len(self._connection_pool) < self._pool_size:
+                self._connection_pool.append(conn)
+            else:
+                # 池已满，直接关闭连接
+                try:
+                    await conn.close()
+                except:
+                    pass
+
+    @asynccontextmanager
+    async def get_conn(self):
+        conn = await self._get_connection()
+        yield conn
+        await self._return_connection(conn)
+
+    async def query(
+        self, query: str, params: Optional[tuple] = None
+    ) -> List[Dict[str, Any]]:
+        """执行查询语句并返回结果"""
+        conn = await self._get_connection()
+        try:
+            cursor = await conn.execute(query, params or ())
+            columns = [description[0] for description in cursor.description]
+            rows = await cursor.fetchall()
+            results = [dict(zip(columns, row)) for row in rows]
+            await cursor.close()
+            return results
+        except Exception as e:
+            # 记录错误但重新抛出，让调用者处理
+            raise Exception(f"数据库查询失败: {str(e)}") from e
+        finally:
+            await self._return_connection(conn)
+
+    async def query_by_sql_file(
+        self, file_path: Union[str, Path], params: Optional[tuple] = None
+    ) -> List[Dict[str, Any]]:
+        """从 SQL 文件中读取查询语句并执行"""
+        path = str(file_path) if isinstance(file_path, Path) else file_path
+        with open(path, "r", encoding="utf-8") as f:
+            query = f.read()
+        return await self.query(query, params)
+
+    async def execute(self, command: str, params: Optional[tuple] = None) -> None:
+        """执行非查询语句"""
+        conn = await self._get_connection()
+        try:
+            await conn.execute(command, params or ())
+            await conn.commit()
+        except Exception as e:
+            # 记录错误但重新抛出，让调用者处理
+            raise Exception(f"数据库执行失败: {str(e)}") from e
+        finally:
+            await self._return_connection(conn)
+
+    async def execute_script(self, script: str) -> None:
+        """执行SQL脚本"""
+        conn = await self._get_connection()
+        try:
+            await conn.executescript(script)
+            await conn.commit()
+        except Exception as e:
+            # 记录错误但重新抛出，让调用者处理
+            raise Exception(f"数据库脚本执行失败: {str(e)}") from e
+        finally:
+            await self._return_connection(conn)
+
+    def _parse_sql_statements(self, script: str) -> List[str]:
+        """解析SQL脚本，分割成独立的语句"""
+        # 使用sqlparse库更准确地分割SQL语句
+        parsed = sqlparse.split(script)
+        statements = []
+
+        for statement in parsed:
+            statement = statement.strip()
+            if statement:
+                statements.append(statement)
+
+        return statements
+
+    async def execute_by_sql_file(
+        self,
+        file_path: Union[str, Path],
+        params: Optional[Union[tuple, List[tuple]]] = None,
+    ) -> None:
+        """从 SQL 文件中读取非查询语句并执行"""
+        path = str(file_path) if isinstance(file_path, Path) else file_path
+        with open(path, "r", encoding="utf-8") as f:
+            script = f.read()
+
+        # 如果有参数且是元组，使用execute执行整个脚本
+        if params is not None and isinstance(params, tuple):
+            await self.execute(script, params)
+        # 如果有参数且是列表，分别执行每个语句
+        elif params is not None and isinstance(params, list):
+            # 使用sqlparse准确分割SQL语句
+            statements = self._parse_sql_statements(script)
+            if len(statements) != len(params):
+                raise ValueError(
+                    f"语句数量({len(statements)})与参数组数量({len(params)})不匹配"
+                )
+
+            for statement, stmt_params in zip(statements, params):
+                if statement:
+                    await self.execute(statement, stmt_params)
+        # 如果无参数，使用executescript
+        else:
+            await self.execute_script(script)
+
+    async def execute_many(self, command: str, seq_of_params: List[tuple]) -> None:
+        """执行多条非查询语句"""
+        conn = await self._get_connection()
+        try:
+            await conn.executemany(command, seq_of_params)
+            await conn.commit()
+        except Exception as e:
+            # 记录错误但重新抛出，让调用者处理
+            raise Exception(f"数据库批量执行失败: {str(e)}") from e
+        finally:
+            await self._return_connection(conn)
+
+    async def execute_many_values_by_sql_file(
+        self, file_path: Union[str, Path], seq_of_params: List[tuple]
+    ) -> None:
+        """从 SQL 文件中读取一条语句，但是被不同值同时执行"""
+        path = str(file_path) if isinstance(file_path, Path) else file_path
+        with open(path, "r", encoding="utf-8") as f:
+            command = f.read()
+        await self.execute_many(command, seq_of_params)
+
+    async def close_all_connections(self) -> None:
+        """关闭所有连接"""
+        async with self._lock:
+            # 关闭池中的连接
+            for conn in self._connection_pool:
+                try:
+                    await conn.close()
+                except:
+                    pass
+            self._connection_pool.clear()
+
+            # 关闭正在使用的连接
+            for conn in self._in_use.copy():
+                try:
+                    await conn.close()
+                except:
+                    pass
+            self._in_use.clear()

konabot/common/llm/__init__.py

@@ -1,4 +1,4 @@
-from typing import Any
+from typing import Any, cast
 import openai
 
 from loguru import logger
@@ -26,14 +26,14 @@ class LLMInfo(BaseModel):
 
     async def chat(
         self,
-        messages: list[ChatCompletionMessageParam],
+        messages: list[ChatCompletionMessageParam] | list[dict[str, Any]],
         timeout: float | None = 30.0,
         max_tokens: int | None = None,
         **kwargs: Any,
     ) -> ChatCompletionMessage:
         logger.info(f"调用 LLM: BASE_URL={self.base_url} MODEL_NAME={self.model_name}")
         completion: ChatCompletion = await self.get_openai_client().chat.completions.create(
-            messages=messages,
+            messages=cast(Any, messages),
             model=self.model_name,
             max_tokens=max_tokens,
             timeout=timeout,
@@ -59,6 +59,9 @@ def get_llm(llm_model: str | None = None):
     if llm_model is None:
         llm_model = llm_config.default_llm
     if llm_model not in llm_config.llms:
-        raise NotImplementedError("LLM 未配置，该功能无法使用")
+        if llm_config.default_llm in llm_config.llms:
+            logger.warning(f"[LLM] 需求的 LLM 不存在，回退到默认模型 REQUIRED={llm_model}")
+            return llm_config.llms[llm_config.default_llm]
+        raise NotImplementedError("[LLM] LLM 未配置，该功能无法使用")
     return llm_config.llms[llm_model]
 

konabot/common/nb/extract_image.py

@@ -1,4 +1,5 @@
 from io import BytesIO
+from pathlib import Path
 from typing import Annotated
 
 import httpx
@@ -19,15 +20,21 @@ from PIL import UnidentifiedImageError
 from pydantic import BaseModel
 from returns.result import Failure, Result, Success
 
-from konabot.common.path import ASSETS_PATH
-
 
 discordConfig = nonebot.get_plugin_config(DiscordConfig)
 
 
 class ExtractImageConfig(BaseModel):
     module_extract_image_no_download: bool = False
-    "要不要算了，不下载了，直接爆炸算了，适用于一些比较奇怪的网络环境，无法从协议端下载文件"
+    """
+    要不要算了，不下载了，直接爆炸算了，
+    适用于一些比较奇怪的网络环境，无法从协议端下载文件
+    """
+
+    module_extract_image_target: str = './assets/img/other/boom.jpg'
+    """
+    使用哪个图片呢
+    """
 
 
 module_config = nonebot.get_plugin_config(ExtractImageConfig)
@@ -37,7 +44,7 @@ async def download_image_bytes(url: str, proxy: str | None = None) -> Result[byt
     # if "/matcha/cache/" in url:
     #     url = url.replace('127.0.0.1', '10.126.126.101')
     if module_config.module_extract_image_no_download:
-        return Success((ASSETS_PATH / "img" / "other" / "boom.jpg").read_bytes())
+        return Success(Path(module_config.module_extract_image_target).read_bytes())
     logger.debug(f"开始从 {url} 下载图片")
     async with httpx.AsyncClient(proxy=proxy) as c:
         try:
@@ -70,15 +77,22 @@ def bytes_to_pil(raw_data: bytes | BytesIO) -> Result[PIL.Image.Image, str]:
         return Failure("图像无法读取，可能是网络存在问题orz")
 
 
-async def unimsg_img_to_pil(image: Image) -> Result[PIL.Image.Image, str]:
+async def unimsg_img_to_bytes(image: Image) -> Result[bytes, str]:
     if image.url is not None:
         raw_result = await download_image_bytes(image.url)
     elif image.raw is not None:
-        raw_result = Success(image.raw)
+        if isinstance(image.raw, bytes):
+            raw_result = Success(image.raw)
+        else:
+            raw_result = Success(image.raw.getvalue())
     else:
         return Failure("由于一些内部问题，下载图片失败了orz")
 
-    return raw_result.bind(bytes_to_pil)
+    return raw_result
+
+
+async def unimsg_img_to_pil(image: Image) -> Result[PIL.Image.Image, str]:
+    return (await unimsg_img_to_bytes(image)).bind(bytes_to_pil)
 
 
 async def extract_image_from_qq_message(
@@ -86,7 +100,7 @@ async def extract_image_from_qq_message(
     evt: OnebotV11MessageEvent,
     bot: OnebotV11Bot,
     allow_reply: bool = True,
-) -> Result[PIL.Image.Image, str]:
+) -> Result[bytes, str]:
     if allow_reply and (reply := evt.reply) is not None:
         return await extract_image_from_qq_message(
             reply.message,
@@ -118,18 +132,17 @@ async def extract_image_from_qq_message(
             url = seg.data.get("url")
             if url is None:
                 return Failure("无法下载图片，可能有一些网络问题")
-            data = await download_image_bytes(url)
-            return data.bind(bytes_to_pil)
+            return await download_image_bytes(url)
 
     return Failure("请在消息中包含图片，或者引用一个含有图片的消息")
 
 
-async def extract_image_from_message(
+async def extract_image_data_from_message(
     msg: Message,
     evt: Event,
     bot: Bot,
     allow_reply: bool = True,
-) -> Result[PIL.Image.Image, str]:
+) -> Result[bytes, str]:
     if (
         isinstance(bot, OnebotV11Bot) 
         and isinstance(msg, OnebotV11Message) 
@@ -145,18 +158,18 @@ async def extract_image_from_message(
             if "image/" not in a.content_type:
                 continue
             url = a.proxy_url
-            return (await download_image_bytes(url, discordConfig.discord_proxy)).bind(bytes_to_pil)
+            return await download_image_bytes(url, discordConfig.discord_proxy)
 
     for seg in UniMessage.of(msg, bot):
         logger.info(seg)
         if isinstance(seg, Image):
-            return await unimsg_img_to_pil(seg)
+            return await unimsg_img_to_bytes(seg)
         elif isinstance(seg, Reply) and allow_reply:
             msg2 = seg.msg
             logger.debug(f"深入搜索引用的消息：{msg2}")
             if msg2 is None or isinstance(msg2, str):
                 continue
-            return await extract_image_from_message(msg2, evt, bot, False)
+            return await extract_image_data_from_message(msg2, evt, bot, False)
         elif isinstance(seg, RefNode) and allow_reply:
             if isinstance(bot, DiscordBot):
                 return Failure("暂时不支持在 Discord 中通过引用的方式获取图片")
@@ -165,12 +178,12 @@ async def extract_image_from_message(
     return Failure("请在消息中包含图片，或者引用一个含有图片的消息")
 
 
-async def _ext_img(
+async def _ext_img_data(
     evt: Event,
     bot: Bot,
     matcher: Matcher,
-) -> PIL.Image.Image | None:
-    match await extract_image_from_message(evt.get_message(), evt, bot):
+) -> bytes | None:
+    match await extract_image_data_from_message(evt.get_message(), evt, bot):
         case Success(img):
             return img
         case Failure(err):
@@ -180,4 +193,35 @@ async def _ext_img(
     assert False
 
 
-PIL_Image = Annotated[PIL.Image.Image, nonebot.params.Depends(_ext_img)]
+async def _ext_img(
+    evt: Event,
+    bot: Bot,
+    matcher: Matcher,
+) -> PIL.Image.Image | None:
+    r = await _ext_img_data(evt, bot, matcher)
+    if r:
+        match bytes_to_pil(r):
+            case Success(img):
+                return img
+            case Failure(msg):
+                await matcher.send(await UniMessage.text(msg).export())
+    return None
+
+async def _try_ext_img(
+    evt: Event,
+    bot: Bot,
+    matcher: Matcher,
+) -> bytes | None:
+    match await extract_image_data_from_message(evt.get_message(), evt, bot):
+        case Success(img):
+            return img
+        case Failure(err):
+            # raise BotExceptionMessage(err)
+            # await matcher.send(await UniMessage().text(err).export())
+            return None
+    assert False
+
+DepImageBytes = Annotated[bytes, nonebot.params.Depends(_ext_img_data)]
+DepPILImage = Annotated[PIL.Image.Image, nonebot.params.Depends(_ext_img)]
+
+DepImageBytesOrNone = Annotated[bytes | None, nonebot.params.Depends(_try_ext_img)]

konabot/common/pager.py

@@ -0,0 +1,76 @@
+from dataclasses import dataclass
+from math import ceil
+from typing import Any, Callable
+
+from nonebot_plugin_alconna import UniMessage
+
+
+@dataclass
+class PagerQuery:
+    page_index: int
+    page_size: int
+
+    def apply[T](self, ls: list[T]) -> "PagerResult[T]":
+        if self.page_size <= 0:
+            return PagerResult(
+                success=False,
+                message="每页元素数量应该大于 0",
+                data=[],
+                page_count=-1,
+                query=self,
+            )
+        page_count = ceil(len(ls) / self.page_size)
+        if self.page_index <= 0 or self.page_size <= 0:
+            return PagerResult(
+                success=False,
+                message="页数必须大于 0",
+                data=[],
+                page_count=page_count,
+                query=self,
+            )
+        data = ls[(self.page_index - 1) * self.page_size: self.page_index * self.page_size]
+        if len(data) > 0:
+            return PagerResult(
+                success=True,
+                message="",
+                data=data,
+                page_count=page_count,
+                query=self,
+            )
+        return PagerResult(
+            success=False,
+            message="指定的页数超过最大页数",
+            data=data,
+            page_count=page_count,
+            query=self,
+        )
+
+
+@dataclass
+class PagerResult[T]:
+    data: list[T]
+    success: bool
+    message: str
+    page_count: int
+    query: PagerQuery
+
+    def to_unimessage(
+        self,
+        formatter: Callable[[T], str | UniMessage[Any]] = str,
+        title: str = '查询结果',
+        list_indicator: str = '- ',
+    ) -> UniMessage[Any]:
+        msg = UniMessage.text(f'===== {title} =====\n\n')
+
+        if not self.success:
+            msg = msg.text(f'⚠️ {self.message}\n')
+        else:
+            for obj in self.data:
+                msg = msg.text(list_indicator)
+                msg += formatter(obj)
+                msg += '\n'
+
+        msg = msg.text(f'\n===== 第 {self.query.page_index} 页，共 {self.page_count} 页 =====')
+        return msg
+
+

konabot/common/path.py

@@ -5,8 +5,10 @@ FONTS_PATH = ASSETS_PATH / "fonts"
 
 SRC_PATH = Path(__file__).resolve().parent.parent
 DATA_PATH = SRC_PATH.parent / "data"
+TMP_PATH = DATA_PATH / "tmp"
 LOG_PATH = DATA_PATH / "logs"
 CONFIG_PATH = DATA_PATH / "config"
+BINARY_PATH = DATA_PATH / "bin"
 
 DOCS_PATH = SRC_PATH / "docs"
 DOCS_PATH_MAN1 = DOCS_PATH / "user"
@@ -21,4 +23,6 @@ if not LOG_PATH.exists():
     LOG_PATH.mkdir()
 
 CONFIG_PATH.mkdir(exist_ok=True)
+TMP_PATH.mkdir(exist_ok=True)
+BINARY_PATH.mkdir(exist_ok=True)
 

konabot/common/permsys/__init__.py

@@ -0,0 +1,108 @@
+from typing import Annotated
+import nonebot
+from nonebot.adapters import Event
+from nonebot.params import Depends
+from nonebot.rule import Rule
+
+from konabot.common.database import DatabaseManager
+from konabot.common.pager import PagerQuery
+from konabot.common.path import DATA_PATH
+from konabot.common.permsys.entity import PermEntity, get_entity_chain
+from konabot.common.permsys.migrates import execute_migration
+from konabot.common.permsys.repo import PermRepo
+
+
+db = DatabaseManager(DATA_PATH / "perm.sqlite3")
+
+
+_EntityLike = Event | PermEntity | list[PermEntity]
+
+
+async def _to_entity_chain(el: _EntityLike):
+    if isinstance(el, Event):
+        return await get_entity_chain(el)  # pragma: no cover
+    if isinstance(el, PermEntity):
+        return [el]
+    return el
+
+
+class PermManager:
+    def __init__(self, db: DatabaseManager) -> None:
+        self.db = db
+
+    async def check_has_permission_info(self, entities: _EntityLike, key: str):
+        entities = await _to_entity_chain(entities)
+        key = key.removesuffix("*").removesuffix(".")
+        key_split = key.split(".")
+        key_split = [s for s in key_split if len(s) > 0]
+        keys = [".".join(key_split[: i + 1]) for i in range(len(key_split))][::-1] + [
+            "*"
+        ]
+
+        async with self.db.get_conn() as conn:
+            repo = PermRepo(conn)
+            data = await repo.get_perm_info_batch(entities, keys)
+            for entity in entities:
+                for k in keys:
+                    p = data.get((entity, k))
+                    if p is not None:
+                        return (entity, k, p)
+        return None
+
+    async def check_has_permission(self, entities: _EntityLike, key: str) -> bool:
+        res = await self.check_has_permission_info(entities, key)
+        if res is None:
+            return False
+        return res[2]
+
+    async def update_permission(self, entity: PermEntity, key: str, perm: bool | None):
+        async with self.db.get_conn() as conn:
+            repo = PermRepo(conn)
+            await repo.update_perm_info(entity, key, perm)
+
+    async def list_permission(self, entities: _EntityLike, query: PagerQuery):
+        entities = await _to_entity_chain(entities)
+        async with self.db.get_conn() as conn:
+            repo = PermRepo(conn)
+            return await repo.list_perm_info_batch(entities, query)
+
+
+def perm_manager(_db: DatabaseManager | None = None) -> PermManager:  # pragma: no cover
+    if _db is None:
+        _db = db
+    return PermManager(_db)
+
+
+def create_startup():  # pragma: no cover
+    from konabot.common.nb.is_admin import cfg
+
+    driver = nonebot.get_driver()
+
+    @driver.on_startup
+    async def _():
+        async with db.get_conn() as conn:
+            await execute_migration(conn)
+            pm = perm_manager(db)
+            for account in cfg.admin_qq_account:
+                # ^ 这里的是超级管理员！！用环境变量定义的。
+                #   咕嘿嘿嘿！！！夺取全部权限！！！
+                await pm.update_permission(
+                    PermEntity("ob11", "user", str(account)), "*", True
+                )
+
+    @driver.on_shutdown
+    async def _():
+        try:
+            await db.close_all_connections()
+        except Exception:
+            pass
+
+
+DepPermManager = Annotated[PermManager, Depends(perm_manager)]
+
+
+def require_permission(perm: str) -> Rule:  # pragma: no cover
+    async def check_permission(event: Event, pm: DepPermManager) -> bool:
+        return await pm.check_has_permission(event, perm)
+
+    return Rule(check_permission)

konabot/common/permsys/entity.py

@@ -0,0 +1,69 @@
+from dataclasses import dataclass
+from nonebot.internal.adapter import Event
+
+from nonebot.adapters.onebot.v11 import Event as OB11Event
+from nonebot.adapters.onebot.v11.event import GroupMessageEvent as OB11GroupEvent
+from nonebot.adapters.onebot.v11.event import PrivateMessageEvent as OB11PrivateEvent
+
+from nonebot.adapters.discord.event import Event as DiscordEvent
+from nonebot.adapters.discord.event import GuildMessageCreateEvent as DiscordGMEvent
+from nonebot.adapters.discord.event import DirectMessageCreateEvent as DiscordDMEvent
+
+from nonebot.adapters.minecraft.event import MessageEvent as MinecraftMessageEvent
+
+from nonebot.adapters.console.event import MessageEvent as ConsoleEvent
+
+
+@dataclass(frozen=True)
+class PermEntity:
+    platform: str
+    entity_type: str
+    external_id: str
+
+
+def get_entity_chain_of_entity(entity: PermEntity) -> list[PermEntity]:
+    return [
+        PermEntity("sys", "global", "global"),
+        PermEntity(entity.platform, "global", "global"),
+        entity,
+    ][::-1]
+
+
+async def get_entity_chain(event: Event) -> list[PermEntity]:  # pragma: no cover
+    entities = [PermEntity("sys", "global", "global")]
+
+    if isinstance(event, OB11Event):
+        entities.append(PermEntity("ob11", "global", "global"))
+
+    if isinstance(event, OB11GroupEvent):
+        entities.append(PermEntity("ob11", "group", str(event.group_id)))
+        entities.append(PermEntity("ob11", "user", str(event.user_id)))
+
+    if isinstance(event, OB11PrivateEvent):
+        entities.append(PermEntity("ob11", "user", str(event.user_id)))
+
+    if isinstance(event, DiscordEvent):
+        entities.append(PermEntity("discord", "global", "global"))
+
+    if isinstance(event, DiscordGMEvent):
+        entities.append(PermEntity("discord", "guild", str(event.guild_id)))
+        entities.append(PermEntity("discord", "channel", str(event.channel_id)))
+        entities.append(PermEntity("discord", "user", str(event.user_id)))
+
+    if isinstance(event, DiscordDMEvent):
+        entities.append(PermEntity("discord", "channel", str(event.channel_id)))
+        entities.append(PermEntity("discord", "user", str(event.user_id)))
+
+    if isinstance(event, MinecraftMessageEvent):
+        entities.append(PermEntity("minecraft", "global", "global"))
+        entities.append(PermEntity("minecraft", "server", event.server_name))
+        player_uuid = event.player.uuid
+        if player_uuid is not None:
+            entities.append(PermEntity("minecraft", "player", player_uuid.hex))
+
+    if isinstance(event, ConsoleEvent):
+        entities.append(PermEntity("console", "global", "global"))
+        entities.append(PermEntity("console", "channel", event.channel.id))
+        entities.append(PermEntity("console", "user", event.user.id))
+
+    return entities[::-1]

konabot/common/permsys/migrates/__init__.py

@@ -0,0 +1,81 @@
+from dataclasses import dataclass
+from pathlib import Path
+
+import aiosqlite
+from loguru import logger
+
+from konabot.common.database import DatabaseManager
+from konabot.common.path import DATA_PATH
+
+
+PATH_THISFOLDER = Path(__file__).parent
+
+SQL_CHECK_EXISTS = (PATH_THISFOLDER / "./check_migrate_version_exists.sql").read_text()
+SQL_CREATE_TABLE = (PATH_THISFOLDER / "./create_migrate_version_table.sql").read_text()
+SQL_GET_MIGRATE_VERSION = (PATH_THISFOLDER / "get_migrate_version.sql").read_text()
+SQL_UPDATE_VERSION = (PATH_THISFOLDER / "./update_migrate_version.sql").read_text()
+
+db = DatabaseManager(DATA_PATH / "perm.sqlite3")
+
+
+@dataclass
+class Migration:
+    upgrade: str | Path
+    downgrade: str | Path
+
+    def get_upgrade_script(self) -> str:
+        if isinstance(self.upgrade, Path):
+            return self.upgrade.read_text()
+        return self.upgrade
+
+    def get_downgrade_script(self) -> str:
+        if isinstance(self.downgrade, Path):
+            return self.downgrade.read_text()
+        return self.downgrade
+
+
+migrations = [
+    Migration(
+        PATH_THISFOLDER / "./mu1_create_permsys_table.sql",
+        PATH_THISFOLDER / "./md1_remove_permsys_table.sql",
+    ),
+]
+
+
+TARGET_VERSION = len(migrations)
+
+
+async def get_current_version(conn: aiosqlite.Connection) -> int:
+    cursor = await conn.execute(SQL_CHECK_EXISTS)
+    count = await cursor.fetchone()
+    assert count is not None
+    if count[0] < 1:
+        logger.info("权限系统数据表不存在，现在创建表")
+        await conn.executescript(SQL_CREATE_TABLE)
+        await conn.commit()
+        return 0
+    cursor = await conn.execute(SQL_GET_MIGRATE_VERSION)
+    row = await cursor.fetchone()
+    if row is None:
+        return 0
+    return row[0]
+
+
+async def execute_migration(
+    conn: aiosqlite.Connection,
+    version: int = TARGET_VERSION,
+    migrations: list[Migration] = migrations,
+):
+    now_version = await get_current_version(conn)
+    while now_version < version:
+        migration = migrations[now_version]
+        await conn.executescript(migration.get_upgrade_script())
+        now_version += 1
+        await conn.execute(SQL_UPDATE_VERSION, (now_version,))
+        await conn.commit()
+    while now_version > version:
+        migration = migrations[now_version - 1]
+        await conn.executescript(migration.get_downgrade_script())
+        now_version -= 1
+        await conn.execute(SQL_UPDATE_VERSION, (now_version,))
+        await conn.commit()

konabot/common/permsys/migrates/check_migrate_version_exists.sql

@@ -0,0 +1,7 @@
+SELECT
+    COUNT(*)
+FROM
+    sqlite_master
+WHERE
+    type = 'table'
+    AND name = 'migrate_version'

konabot/common/permsys/migrates/create_migrate_version_table.sql

@@ -0,0 +1,3 @@
+CREATE TABLE migrate_version(version INT PRIMARY KEY);
+INSERT INTO migrate_version(version)
+VALUES(0);

konabot/common/permsys/migrates/get_migrate_version.sql

@@ -0,0 +1,4 @@
+SELECT
+    version
+FROM
+    migrate_version;

konabot/common/permsys/migrates/md1_remove_permsys_table.sql

@@ -0,0 +1,2 @@
+DROP TABLE IF EXISTS perm_entity;
+DROP TABLE IF EXISTS perm_info;

konabot/common/permsys/migrates/mu1_create_permsys_table.sql

@@ -0,0 +1,30 @@
+CREATE TABLE perm_entity(
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    platform TEXT NOT NULL,
+    entity_type TEXT NOT NULL,
+    external_id TEXT NOT NULL,
+    created_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+    updated_at DATETIME DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE UNIQUE INDEX idx_perm_entity_lookup
+ON perm_entity(platform, entity_type, external_id);
+
+CREATE TABLE perm_info(
+    entity_id INTEGER NOT NULL,
+    config_key TEXT NOT NULL,
+    value BOOLEAN,
+    updated_at DATETIME DEFAULT CURRENT_TIMESTAMP,
+    -- 联合主键
+    PRIMARY KEY (entity_id, config_key)
+);
+
+CREATE TRIGGER perm_entity_update AFTER UPDATE
+ON perm_entity BEGIN
+    UPDATE perm_entity SET updated_at=CURRENT_TIMESTAMP WHERE id=old.id;
+END;
+CREATE TRIGGER perm_info_update AFTER UPDATE
+ON perm_info BEGIN
+    UPDATE perm_info SET updated_at=CURRENT_TIMESTAMP WHERE entity_id=old.entity_id;
+END;
+

konabot/common/permsys/migrates/update_migrate_version.sql

@@ -0,0 +1,2 @@
+UPDATE migrate_version
+SET version = ?;

konabot/common/permsys/repo.py

@@ -0,0 +1,242 @@
+from dataclasses import dataclass
+import math
+from pathlib import Path
+
+import aiosqlite
+
+from konabot.common.pager import PagerQuery, PagerResult
+
+from .entity import PermEntity
+
+
+def s(p: str):
+    """读取 SQL 文件内容。
+
+    Args:
+        p: SQL 文件名（相对于当前文件所在目录的 sql/ 子目录）。
+
+    Returns:
+        SQL 文件的内容字符串。
+    """
+    return (Path(__file__).parent / "./sql/" / p).read_text()
+
+
+@dataclass
+class PermRepo:
+    """权限实体存储库，负责与数据库交互管理权限实体。
+
+    Attributes:
+        conn: aiosqlite 数据库连接对象。
+    """
+
+    conn: aiosqlite.Connection
+
+    async def create_entity(self, entity: PermEntity) -> int:
+        """创建新的权限实体并返回其 ID。
+
+        Args:
+            entity: 要创建的权限实体对象。
+
+        Returns:
+            新创建实体的数据库 ID。
+
+        Raises:
+            AssertionError: 如果创建后无法获取实体 ID。
+        """
+        await self.conn.execute(
+            s("create_entity.sql"),
+            (entity.platform, entity.entity_type, entity.external_id),
+        )
+        await self.conn.commit()
+        eid = await self._get_entity_id_or_none(entity)
+        assert eid is not None
+        return eid
+
+    async def _get_entity_id_or_none(self, entity: PermEntity) -> int | None:
+        """查询实体 ID，如果不存在则返回 None。
+
+        Args:
+            entity: 要查询的权限实体对象。
+
+        Returns:
+            实体 ID，如果不存在则返回 None。
+        """
+        res = await self.conn.execute(
+            s("get_entity_id.sql"),
+            (entity.platform, entity.entity_type, entity.external_id),
+        )
+        row = await res.fetchone()
+        if row is None:
+            return None
+        return row[0]
+
+    async def get_entity_id(self, entity: PermEntity) -> int:
+        """获取实体 ID，如果不存在则自动创建。
+
+        Args:
+            entity: 权限实体对象。
+
+        Returns:
+            实体的数据库 ID。
+        """
+        eid = await self._get_entity_id_or_none(entity)
+        if eid is None:
+            return await self.create_entity(entity)
+        return eid
+
+    async def get_perm_info(self, entity: PermEntity, config_key: str) -> bool | None:
+        """获取实体的权限配置信息。
+
+        Args:
+            entity: 权限实体对象。
+            config_key: 配置项的键名。
+
+        Returns:
+            配置值（True/False），如果不存在则返回 None。
+        """
+        eid = await self.get_entity_id(entity)
+        res = await self.conn.execute(
+            s("get_perm_info.sql"),
+            (eid, config_key),
+        )
+        row = await res.fetchone()
+        if row is None:
+            return None
+        return bool(row[0])
+
+    async def update_perm_info(
+        self, entity: PermEntity, config_key: str, value: bool | None
+    ):
+        """更新实体的权限配置信息。
+
+        Args:
+            entity: 权限实体对象。
+            config_key: 配置项的键名。
+            value: 要设置的配置值（True/False/None）。
+        """
+        eid = await self.get_entity_id(entity)
+        await self.conn.execute(s("update_perm_info.sql"), (eid, config_key, value))
+        await self.conn.commit()
+
+    async def get_entity_id_batch(
+        self, entities: list[PermEntity]
+    ) -> dict[PermEntity, int]:
+        """批量获取 Entity 的 entity_id
+
+        Args:
+            entities: PermEntity 列表
+
+        Returns:
+            字典，键为 PermEntity，值为对应的 ID
+        """
+
+        # for entity in entities:
+        #     await self.conn.execute(
+        #         s("create_entity.sql"),
+        #         (entity.platform, entity.entity_type, entity.external_id),
+        #     )
+        await self.conn.executemany(
+            s("create_entity.sql"),
+            [(e.platform, e.entity_type, e.external_id) for e in entities],
+        )
+        await self.conn.commit()
+        val_placeholders = ", ".join(["(?, ?, ?)"] * len(entities))
+        params = []
+        for e in entities:
+            params.extend([e.platform, e.entity_type, e.external_id])
+        cursor = await self.conn.execute(
+            f"""
+                SELECT id, platform, entity_type, external_id
+                FROM perm_entity
+                WHERE (platform, entity_type, external_id) IN (VALUES {val_placeholders});
+            """,
+            params,
+        )
+        rows = await cursor.fetchall()
+        return {PermEntity(row[1], row[2], row[3]): row[0] for row in rows}
+
+    async def get_perm_info_batch(
+        self, entities: list[PermEntity], config_keys: list[str]
+    ) -> dict[tuple[PermEntity, str], bool]:
+        """批量获取权限信息
+
+        Args:
+            entities: PermEntity 列表
+            config_keys: 查询的键列表
+
+        Returns:
+            字典，键是 PermEntity 和 config_key 的元组，值是布尔，过滤掉所有空值
+        """
+        entity_ids = {
+            v: k for k, v in (await self.get_entity_id_batch(entities)).items()
+        }
+        placeholders1 = ", ".join("?" * len(entity_ids))
+        placeholders2 = ", ".join("?" * len(config_keys))
+        sql = f"""
+            SELECT entity_id, config_key, value
+            FROM perm_info
+            WHERE entity_id IN ({placeholders1})
+              AND config_key IN ({placeholders2})
+              AND value IS NOT NULL;
+        """
+
+        params = tuple(entity_ids.keys()) + tuple(config_keys)
+        cursor = await self.conn.execute(sql, params)
+        rows = await cursor.fetchall()
+
+        return {(entity_ids[row[0]], row[1]): bool(row[2]) for row in rows}
+
+    async def list_perm_info_batch(
+        self, entities: list[PermEntity], pager: PagerQuery
+    ) -> PagerResult[tuple[PermEntity, str, bool]]:
+        """批量获取某个实体的权限信息
+
+        Args:
+            entities: PermEntity 列表
+            pager: PagerQuery 对象，即分页要求
+
+        Returns:
+            字典，键是 PermEntity，值是权限条目和布尔的元组，过滤掉所有空值
+        """
+        entity_to_id = await self.get_entity_id_batch(entities)
+        id_to_entity = {v: k for k, v in entity_to_id.items()}
+        ordered_ids = [entity_to_id[e] for e in entities if e in entity_to_id]
+
+        placeholders = ", ".join("?" * len(ordered_ids))
+        order_by_cases = " ".join([f"WHEN ? THEN {i}" for i in range(len(ordered_ids))])
+
+        pagecount_sql = f"SELECT COUNT(*) FROM perm_info WHERE entity_id IN ({placeholders}) AND value IS NOT NULL;"
+        count_cursor = await self.conn.execute(pagecount_sql, tuple(ordered_ids))
+        total_count = (await count_cursor.fetchone() or (0,))[0]
+
+        sql = f"""
+            SELECT entity_id, config_key, value
+            FROM perm_info
+            WHERE entity_id IN ({placeholders})
+              AND value IS NOT NULL
+            ORDER BY
+                (CASE entity_id {order_by_cases} END) ASC,
+                config_key ASC
+            LIMIT ?
+            OFFSET ?;
+        """
+
+        params = (
+            tuple(ordered_ids)
+            + tuple(ordered_ids)
+            + (
+                pager.page_size,
+                (pager.page_index - 1) * pager.page_size,
+            )
+        )
+        cursor = await self.conn.execute(sql, params)
+        rows = await cursor.fetchall()
+
+        # return {entity_ids[row[0]]: (row[1], bool(row[2])) for row in rows}
+        return PagerResult(
+            data=[(id_to_entity[row[0]], row[1], row[2]) for row in rows],
+            success=True,
+            message="",
+            page_count=math.ceil(total_count / pager.page_size),
+            query=pager,
+        )

konabot/common/permsys/sql/create_entity.sql

@@ -0,0 +1,11 @@
+INSERT
+OR IGNORE INTO perm_entity(
+    platform,
+    entity_type,
+    external_id
+)
+VALUES(
+    ?,
+    ?,
+    ?
+);

konabot/common/permsys/sql/get_entity_id.sql

@@ -0,0 +1,8 @@
+SELECT
+    id
+FROM
+    perm_entity
+WHERE
+    perm_entity.platform = ?
+    AND perm_entity.entity_type = ?
+    AND perm_entity.external_id = ?;

konabot/common/permsys/sql/get_perm_info.sql

@@ -0,0 +1,7 @@
+SELECT
+    VALUE
+FROM
+    perm_info
+WHERE
+    entity_id = ?
+    AND config_key = ?;

konabot/common/permsys/sql/update_perm_info.sql

@@ -0,0 +1,4 @@
+INSERT INTO perm_info (entity_id, config_key, value)
+VALUES (?, ?, ?)
+ON CONFLICT(entity_id, config_key)
+DO UPDATE SET value=excluded.value;

konabot/common/ptimeparse/README.md

@@ -0,0 +1,3 @@
+# 已废弃
+
+坏枪用简单的 LLM + 提示词工程，完成了这 200 块的 `qwen3-coder-plus` 都搞不定的 nb 功能

konabot/common/ptimeparse/__init__.py

@@ -0,0 +1,58 @@
+"""
+Professional time parsing module for Chinese and English time expressions.
+
+This module provides a robust parser for natural language time expressions,
+supporting both Chinese and English formats with proper whitespace handling.
+"""
+
+import datetime
+from typing import Optional
+
+from .expression import TimeExpression
+
+
+def parse(text: str, now: Optional[datetime.datetime] = None) -> datetime.datetime:
+    """
+    Parse a time expression and return a datetime object.
+    
+    Args:
+        text: The time expression to parse
+        now: The reference time (defaults to current time)
+        
+    Returns:
+        A datetime object representing the parsed time
+        
+    Raises:
+        TokenUnhandledException: If the input cannot be parsed
+    """
+    return TimeExpression.parse(text, now)
+
+
+class Parser:
+    """
+    Parser for time expressions with backward compatibility.
+    
+    Maintains the original interface:
+    >>> parser = Parser()
+    >>> result = parser.parse("10分钟后")
+    """
+    
+    def __init__(self, now: Optional[datetime.datetime] = None):
+        self.now = now or datetime.datetime.now()
+    
+    def parse(self, text: str) -> datetime.datetime:
+        """
+        Parse a time expression and return a datetime object.
+        This maintains backward compatibility with the original interface.
+        
+        Args:
+            text: The time expression to parse
+            
+        Returns:
+            A datetime object representing the parsed time
+            
+        Raises:
+            TokenUnhandledException: If the input cannot be parsed
+        """
+        return TimeExpression.parse(text, self.now)
+

konabot/common/ptimeparse/chinese_number.py

@@ -0,0 +1,133 @@
+"""
+Chinese number parser for the time expression parser.
+"""
+
+import re
+from typing import Tuple
+
+
+class ChineseNumberParser:
+    """Parser for Chinese numbers."""
+    
+    def __init__(self):
+        self.digits = {"零": 0, "一": 1, "二": 2, "三": 3, "四": 4, 
+                      "五": 5, "六": 6, "七": 7, "八": 8, "九": 9}
+        self.units = {"十": 10, "百": 100, "千": 1000, "万": 10000, "亿": 100000000}
+    
+    def digest(self, text: str) -> Tuple[str, int]:
+        """
+        Parse a Chinese number from the beginning of text and return the rest and the parsed number.
+        
+        Args:
+            text: Text that may start with a Chinese number
+            
+        Returns:
+            Tuple of (remaining_text, parsed_number)
+        """
+        if not text:
+            return text, 0
+            
+        # Handle "两" at start
+        if text.startswith("两"):
+            # Check if "两" is followed by a time unit
+            # Look ahead to see if we have a valid pattern like "两小时", "两分钟", etc.
+            if len(text) >= 2:
+                # Check for time units that start with the second character
+                time_units = ["小时", "分钟", "秒"]
+                for unit in time_units:
+                    if text[1:].startswith(unit):
+                        # Return the text starting from the time unit, not after it
+                        # The parser will handle the time unit in the next step
+                        return text[1:], 2
+                # Check for single character time units
+                next_char = text[1]
+                if next_char in "时分秒":
+                    return text[1:], 2
+                # Check for Chinese number units
+                if next_char in "十百千万亿":
+                    # This will be handled by the normal parsing below
+                    pass
+                # If "两" is at the end of string, treat it as standalone
+                elif len(text) == 1:
+                    return "", 2
+                # Also accept "两" followed by whitespace and then time units
+                elif next_char.isspace():
+                    # Check if after whitespace we have time units
+                    rest_after_space = text[2:].lstrip()
+                    for unit in time_units:
+                        if rest_after_space.startswith(unit):
+                            # Return the text starting from the time unit
+                            space_len = len(text[2:]) - len(rest_after_space)
+                            return text[2+space_len:], 2
+                    # Check single character time units after whitespace
+                    if rest_after_space and rest_after_space[0] in "时分秒":
+                        return text[2:], 2
+            else:
+                # Just "两" by itself
+                return "", 2
+                
+        s = "零一二三四五六七八九"
+        i = 0
+        while i < len(text) and text[i] in s + "十百千万亿":
+            i += 1
+        if i == 0:
+            return text, 0
+        num_str = text[:i]
+        rest = text[i:]
+        
+        return rest, self.parse(num_str)
+    
+    def parse(self, text: str) -> int:
+        """
+        Parse a Chinese number string and return its integer value.
+
+        Args:
+            text: Chinese number string
+
+        Returns:
+            Integer value of the Chinese number
+        """
+        if not text:
+            return 0
+        if text == "零":
+            return 0
+        if text == "两":
+            return 2
+
+        # Handle special case for "十"
+        if text == "十":
+            return 10
+
+        # Handle numbers with "亿"
+        if "亿" in text:
+            parts = text.split("亿", 1)
+            a, b = parts[0], parts[1]
+            return self.parse(a) * 100000000 + self.parse(b)
+
+        # Handle numbers with "万"
+        if "万" in text:
+            parts = text.split("万", 1)
+            a, b = parts[0], parts[1]
+            return self.parse(a) * 10000 + self.parse(b)
+
+        # Handle remaining numbers
+        result = 0
+        temp = 0
+
+        for char in text:
+            if char == "零":
+                continue
+            elif char == "两":
+                temp = 2
+            elif char in self.digits:
+                temp = self.digits[char]
+            elif char in self.units:
+                unit = self.units[char]
+                if unit == 10 and temp == 0:
+                    # Special case for numbers like "十三"
+                    temp = 1
+                result += temp * unit
+                temp = 0
+
+        result += temp
+        return result

konabot/common/ptimeparse/err.py

@@ -0,0 +1,11 @@
+class PTimeParseException(Exception):
+    ...
+
+class TokenUnhandledException(PTimeParseException):
+    ...
+
+class MultipleSpecificationException(PTimeParseException):
+    ...
+
+class OutOfRangeSpecificationException(PTimeParseException):
+    ...

konabot/common/ptimeparse/expression.py

@@ -0,0 +1,63 @@
+"""
+Main time expression parser class that integrates all components.
+"""
+
+import datetime
+from typing import Optional
+
+from .lexer import Lexer
+from .parser import Parser
+from .semantic import SemanticAnalyzer
+from .ptime_ast import TimeExpressionNode
+from .err import TokenUnhandledException
+
+
+class TimeExpression:
+    """Main class for parsing time expressions."""
+    
+    def __init__(self, text: str, now: Optional[datetime.datetime] = None):
+        self.text = text.strip()
+        self.now = now or datetime.datetime.now()
+        
+        if not self.text:
+            raise TokenUnhandledException("Empty input")
+        
+        # Initialize components
+        self.lexer = Lexer(self.text, self.now)
+        self.parser = Parser(self.text, self.now)
+        self.semantic_analyzer = SemanticAnalyzer(self.now)
+        
+        # Parse the expression
+        self.ast = self._parse()
+    
+    def _parse(self) -> TimeExpressionNode:
+        """Parse the time expression and return the AST."""
+        try:
+            return self.parser.parse()
+        except Exception as e:
+            raise TokenUnhandledException(f"Failed to parse '{self.text}': {str(e)}")
+    
+    def evaluate(self) -> datetime.datetime:
+        """Evaluate the time expression and return the datetime."""
+        try:
+            return self.semantic_analyzer.evaluate(self.ast)
+        except Exception as e:
+            raise TokenUnhandledException(f"Failed to evaluate '{self.text}': {str(e)}")
+    
+    @classmethod
+    def parse(cls, text: str, now: Optional[datetime.datetime] = None) -> datetime.datetime:
+        """
+        Parse a time expression and return a datetime object.
+        
+        Args:
+            text: The time expression to parse
+            now: The reference time (defaults to current time)
+            
+        Returns:
+            A datetime object representing the parsed time
+            
+        Raises:
+            TokenUnhandledException: If the input cannot be parsed
+        """
+        expression = cls(text, now)
+        return expression.evaluate()

konabot/common/ptimeparse/lexer.py

@@ -0,0 +1,225 @@
+"""
+Lexical analyzer for time expressions.
+"""
+
+import re
+from typing import Iterator, Optional
+import datetime
+
+from .ptime_token import Token, TokenType
+from .chinese_number import ChineseNumberParser
+
+
+class Lexer:
+    """Lexical analyzer for time expressions."""
+
+    def __init__(self, text: str, now: Optional[datetime.datetime] = None):
+        self.text = text
+        self.pos = 0
+        self.current_char = self.text[self.pos] if self.text else None
+        self.now = now or datetime.datetime.now()
+        self.chinese_parser = ChineseNumberParser()
+
+        # Define token patterns
+        self.token_patterns = [
+            # Whitespace
+            (r'^\s+', TokenType.WHITESPACE),
+
+            # Time separators
+            (r'^:', TokenType.TIME_SEPARATOR),
+            (r'^点', TokenType.TIME_SEPARATOR),
+            (r'^时', TokenType.TIME_SEPARATOR),
+            (r'^分', TokenType.TIME_SEPARATOR),
+            (r'^秒', TokenType.TIME_SEPARATOR),
+
+            # Special time markers
+            (r'^半', TokenType.HALF),
+            (r'^一刻', TokenType.QUARTER),
+            (r'^整', TokenType.ZHENG),
+            (r'^钟', TokenType.ZHONG),
+
+            # Period indicators (must come before relative time patterns to avoid conflicts)
+            (r'^(上午|早晨|早上|清晨|早(?!\d))', TokenType.PERIOD_AM),
+            (r'^(中午|下午|晚上|晚(?!\d)|凌晨|午夜)', TokenType.PERIOD_PM),
+
+            # Week scope (more specific patterns first)
+            (r'^本周', TokenType.WEEK_SCOPE_CURRENT),
+            (r'^上周', TokenType.WEEK_SCOPE_LAST),
+            (r'^下周', TokenType.WEEK_SCOPE_NEXT),
+
+            # Relative directions
+            (r'^(后|以后|之后)', TokenType.RELATIVE_DIRECTION_FORWARD),
+            (r'^(前|以前|之前)', TokenType.RELATIVE_DIRECTION_BACKWARD),
+
+            # Extended relative time
+            (r'^明年', TokenType.RELATIVE_NEXT),
+            (r'^去年', TokenType.RELATIVE_LAST),
+            (r'^今年', TokenType.RELATIVE_THIS),
+            (r'^下(?![午年月周])', TokenType.RELATIVE_NEXT),
+            (r'^(上|去)(?![午年月周])', TokenType.RELATIVE_LAST),
+            (r'^这', TokenType.RELATIVE_THIS),
+            (r'^本(?![周月年])', TokenType.RELATIVE_THIS),  # Match "本" but not "本周", "本月", "本年"
+
+            # Week scope (fallback for standalone terms)
+            (r'^本', TokenType.WEEK_SCOPE_CURRENT),
+            (r'^上', TokenType.WEEK_SCOPE_LAST),
+            (r'^下(?![午年月周])', TokenType.WEEK_SCOPE_NEXT),
+
+            # Week days (order matters - longer patterns first)
+            (r'^周一', TokenType.WEEKDAY_MONDAY),
+            (r'^周二', TokenType.WEEKDAY_TUESDAY),
+            (r'^周三', TokenType.WEEKDAY_WEDNESDAY),
+            (r'^周四', TokenType.WEEKDAY_THURSDAY),
+            (r'^周五', TokenType.WEEKDAY_FRIDAY),
+            (r'^周六', TokenType.WEEKDAY_SATURDAY),
+            (r'^周日', TokenType.WEEKDAY_SUNDAY),
+            # Single character weekdays should be matched after numbers
+            # (r'^一', TokenType.WEEKDAY_MONDAY),
+            # (r'^二', TokenType.WEEKDAY_TUESDAY),
+            # (r'^三', TokenType.WEEKDAY_WEDNESDAY),
+            # (r'^四', TokenType.WEEKDAY_THURSDAY),
+            # (r'^五', TokenType.WEEKDAY_FRIDAY),
+            # (r'^六', TokenType.WEEKDAY_SATURDAY),
+            # (r'^日', TokenType.WEEKDAY_SUNDAY),
+
+            # Student-friendly time expressions
+            (r'^早(?=\d)', TokenType.EARLY_MORNING),
+            (r'^晚(?=\d)', TokenType.LATE_NIGHT),
+
+            # Relative today variants
+            (r'^今晚上', TokenType.RELATIVE_TODAY),
+            (r'^今晚', TokenType.RELATIVE_TODAY),
+            (r'^今早', TokenType.RELATIVE_TODAY),
+            (r'^今天早上', TokenType.RELATIVE_TODAY),
+            (r'^今天早晨', TokenType.RELATIVE_TODAY),
+            (r'^今天上午', TokenType.RELATIVE_TODAY),
+            (r'^今天下午', TokenType.RELATIVE_TODAY),
+            (r'^今天晚上', TokenType.RELATIVE_TODAY),
+            (r'^今天', TokenType.RELATIVE_TODAY),
+
+            # Relative days
+            (r'^明天', TokenType.RELATIVE_TOMORROW),
+            (r'^后天', TokenType.RELATIVE_DAY_AFTER_TOMORROW),
+            (r'^大后天', TokenType.RELATIVE_THREE_DAYS_AFTER_TOMORROW),
+            (r'^昨天', TokenType.RELATIVE_YESTERDAY),
+            (r'^前天', TokenType.RELATIVE_DAY_BEFORE_YESTERDAY),
+            (r'^大前天', TokenType.RELATIVE_THREE_DAYS_BEFORE_YESTERDAY),
+
+            # Digits
+            (r'^\d+', TokenType.INTEGER),
+
+            # Time units (must come after date separators to avoid conflicts)
+            (r'^年(?![月日号])', TokenType.YEAR),
+            (r'^月(?![日号])', TokenType.MONTH),
+            (r'^[日号](?![月年])', TokenType.DAY),
+            (r'^天', TokenType.DAY),
+            (r'^周', TokenType.WEEK),
+            (r'^小时', TokenType.HOUR),
+            (r'^分钟', TokenType.MINUTE),
+            (r'^秒', TokenType.SECOND),
+
+            # Date separators (fallback patterns)
+            (r'^年', TokenType.DATE_SEPARATOR),
+            (r'^月', TokenType.DATE_SEPARATOR),
+            (r'^[日号]', TokenType.DATE_SEPARATOR),
+            (r'^[-/]', TokenType.DATE_SEPARATOR),
+        ]
+
+    def advance(self):
+        """Advance the position pointer and set the current character."""
+        self.pos += 1
+        if self.pos >= len(self.text):
+            self.current_char = None
+        else:
+            self.current_char = self.text[self.pos]
+
+    def skip_whitespace(self):
+        """Skip whitespace characters."""
+        while self.current_char is not None and self.current_char.isspace():
+            self.advance()
+
+    def integer(self) -> int:
+        """Parse an integer from the input."""
+        result = ''
+        while self.current_char is not None and self.current_char.isdigit():
+            result += self.current_char
+            self.advance()
+        return int(result)
+
+    def chinese_number(self) -> int:
+        """Parse a Chinese number from the input."""
+        # Find the longest prefix that can be parsed as a Chinese number
+        for i in range(len(self.text) - self.pos, 0, -1):
+            prefix = self.text[self.pos:self.pos + i]
+            try:
+                # Use digest to get both the remaining text and the parsed value
+                remaining, value = self.chinese_parser.digest(prefix)
+                # Check if we actually consumed part of the prefix
+                consumed_length = len(prefix) - len(remaining)
+                if consumed_length > 0:
+                    # Advance position by the length of the consumed text
+                    for _ in range(consumed_length):
+                        self.advance()
+                    return value
+            except ValueError:
+                continue
+        # If no Chinese number found, just return 0
+        return 0
+
+    def get_next_token(self) -> Token:
+        """Lexical analyzer that breaks the sentence into tokens."""
+        while self.current_char is not None:
+            # Skip whitespace
+            if self.current_char.isspace():
+                self.skip_whitespace()
+                continue
+
+            # Try to match each pattern
+            text_remaining = self.text[self.pos:]
+            for pattern, token_type in self.token_patterns:
+                match = re.match(pattern, text_remaining)
+                if match:
+                    value = match.group(0)
+                    position = self.pos
+
+                    # Advance position
+                    for _ in range(len(value)):
+                        self.advance()
+
+                    # Special handling for some tokens
+                    if token_type == TokenType.INTEGER:
+                        value = int(value)
+                    elif token_type == TokenType.RELATIVE_TODAY and value in [
+                        "今早上", "今天早上", "今天早晨", "今天上午"
+                    ]:
+                        token_type = TokenType.PERIOD_AM
+                    elif token_type == TokenType.RELATIVE_TODAY and value in [
+                        "今晚上", "今天下午", "今天晚上"
+                    ]:
+                        token_type = TokenType.PERIOD_PM
+
+                    return Token(token_type, value, position)
+
+            # Try to parse Chinese numbers
+            chinese_start_pos = self.pos
+            try:
+                chinese_value = self.chinese_number()
+                if chinese_value > 0:
+                    # We successfully parsed a Chinese number
+                    return Token(TokenType.CHINESE_NUMBER, chinese_value, chinese_start_pos)
+            except ValueError:
+                pass
+
+            # If no pattern matches, skip the character and continue
+            self.advance()
+
+        # End of file
+        return Token(TokenType.EOF, None, self.pos)
+
+    def tokenize(self) -> Iterator[Token]:
+        """Generate all tokens from the input."""
+        while True:
+            token = self.get_next_token()
+            yield token
+            if token.type == TokenType.EOF:
+                break

konabot/common/ptimeparse/parser.py

@@ -0,0 +1,846 @@
+"""
+Parser for time expressions that builds an Abstract Syntax Tree (AST).
+"""
+
+from typing import Iterator, Optional, List
+import datetime
+
+from .ptime_token import Token, TokenType
+from .ptime_ast import (
+    ASTNode, NumberNode, DateNode, TimeNode,
+    RelativeDateNode, RelativeTimeNode, WeekdayNode, TimeExpressionNode
+)
+from .lexer import Lexer
+
+
+class ParserError(Exception):
+    """Exception raised for parser errors."""
+    pass
+
+
+class Parser:
+    """Parser for time expressions that builds an AST."""
+    
+    def __init__(self, text: str, now: Optional[datetime.datetime] = None):
+        self.lexer = Lexer(text, now)
+        self.tokens: List[Token] = list(self.lexer.tokenize())
+        self.pos = 0
+        self.now = now or datetime.datetime.now()
+    
+    @property
+    def current_token(self) -> Token:
+        """Get the current token."""
+        if self.pos < len(self.tokens):
+            return self.tokens[self.pos]
+        return Token(TokenType.EOF, None, len(self.tokens))
+    
+    def eat(self, token_type: TokenType) -> Token:
+        """Consume a token of the expected type."""
+        if self.current_token.type == token_type:
+            token = self.current_token
+            self.pos += 1
+            return token
+        else:
+            raise ParserError(
+                f"Expected token {token_type}, got {self.current_token.type} "
+                f"at position {self.current_token.position}"
+            )
+    
+    def peek(self, offset: int = 1) -> Token:
+        """Look ahead at the next token without consuming it."""
+        next_pos = self.pos + offset
+        if next_pos < len(self.tokens):
+            return self.tokens[next_pos]
+        return Token(TokenType.EOF, None, len(self.tokens))
+    
+    def parse_number(self) -> NumberNode:
+        """Parse a number (integer or Chinese number)."""
+        token = self.current_token
+        if token.type == TokenType.INTEGER:
+            self.eat(TokenType.INTEGER)
+            return NumberNode(value=token.value)
+        elif token.type == TokenType.CHINESE_NUMBER:
+            self.eat(TokenType.CHINESE_NUMBER)
+            return NumberNode(value=token.value)
+        else:
+            raise ParserError(
+                f"Expected number, got {token.type} at position {token.position}"
+            )
+    
+    def parse_date(self) -> DateNode:
+        """Parse a date specification."""
+        year_node = None
+        month_node = None
+        day_node = None
+
+        # Try YYYY-MM-DD or YYYY/MM/DD format
+        if (self.current_token.type == TokenType.INTEGER and
+            self.peek().type == TokenType.DATE_SEPARATOR and
+            self.peek().value in ['-', '/'] and
+            self.peek(2).type == TokenType.INTEGER and
+            self.peek(3).type == TokenType.DATE_SEPARATOR and
+            self.peek(3).value in ['-', '/'] and
+            self.peek(4).type == TokenType.INTEGER):
+
+            year_token = self.current_token
+            self.eat(TokenType.INTEGER)
+            separator1 = self.eat(TokenType.DATE_SEPARATOR).value
+
+            month_token = self.current_token
+            self.eat(TokenType.INTEGER)
+
+            separator2 = self.eat(TokenType.DATE_SEPARATOR).value
+
+            day_token = self.current_token
+            self.eat(TokenType.INTEGER)
+
+            year_node = NumberNode(value=year_token.value)
+            month_node = NumberNode(value=month_token.value)
+            day_node = NumberNode(value=day_token.value)
+
+            return DateNode(year=year_node, month=month_node, day=day_node)
+
+        # Try YYYY年MM月DD[日号] format
+        if (self.current_token.type == TokenType.INTEGER and
+            self.peek().type in [TokenType.DATE_SEPARATOR, TokenType.YEAR] and
+            self.peek(2).type == TokenType.INTEGER and
+            self.peek(3).type in [TokenType.DATE_SEPARATOR, TokenType.MONTH] and
+            self.peek(4).type == TokenType.INTEGER):
+
+            year_token = self.current_token
+            self.eat(TokenType.INTEGER)
+            self.eat(self.current_token.type)  # 年 (could be DATE_SEPARATOR or YEAR)
+
+            month_token = self.current_token
+            self.eat(TokenType.INTEGER)
+            self.eat(self.current_token.type)  # 月 (could be DATE_SEPARATOR or MONTH)
+
+            day_token = self.current_token
+            self.eat(TokenType.INTEGER)
+            # Optional 日 or 号
+            if self.current_token.type in [TokenType.DATE_SEPARATOR, TokenType.DAY]:
+                self.eat(self.current_token.type)
+
+            year_node = NumberNode(value=year_token.value)
+            month_node = NumberNode(value=month_token.value)
+            day_node = NumberNode(value=day_token.value)
+
+            return DateNode(year=year_node, month=month_node, day=day_node)
+
+        # Try MM月DD[日号] format (without year)
+        if (self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+            self.peek().type in [TokenType.DATE_SEPARATOR, TokenType.MONTH] and
+            self.peek().value == '月' and
+            self.peek(2).type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]):
+
+            month_token = self.current_token
+            self.eat(month_token.type)
+            self.eat(self.current_token.type)  # 月 (could be DATE_SEPARATOR or MONTH)
+
+            day_token = self.current_token
+            self.eat(day_token.type)
+            # Optional 日 or 号
+            if self.current_token.type in [TokenType.DATE_SEPARATOR, TokenType.DAY]:
+                self.eat(self.current_token.type)
+
+            month_node = NumberNode(value=month_token.value)
+            day_node = NumberNode(value=day_token.value)
+
+            return DateNode(year=None, month=month_node, day=day_node)
+
+        # Try Chinese MM月DD[日号] format
+        if (self.current_token.type == TokenType.CHINESE_NUMBER and
+            self.peek().type == TokenType.DATE_SEPARATOR and
+            self.peek().value == '月' and
+            self.peek(2).type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]):
+
+            month_token = self.current_token
+            self.eat(TokenType.CHINESE_NUMBER)
+            self.eat(TokenType.DATE_SEPARATOR)  # 月
+
+            day_token = self.current_token
+            self.eat(day_token.type)
+            # Optional 日 or 号
+            if self.current_token.type == TokenType.DATE_SEPARATOR:
+                self.eat(TokenType.DATE_SEPARATOR)
+
+            month_node = NumberNode(value=month_token.value)
+            day_node = NumberNode(value=day_token.value)
+
+            return DateNode(year=None, month=month_node, day=day_node)
+
+        raise ParserError(
+            f"Unable to parse date at position {self.current_token.position}"
+        )
+    
+    def parse_time(self) -> TimeNode:
+        """Parse a time specification."""
+        hour_node = None
+        minute_node = None
+        second_node = None
+        is_24hour = False
+        period = None
+        
+        # Try HH:MM format
+        if (self.current_token.type == TokenType.INTEGER and
+            self.peek().type == TokenType.TIME_SEPARATOR and
+            self.peek().value == ':'):
+            
+            hour_token = self.current_token
+            self.eat(TokenType.INTEGER)
+            self.eat(TokenType.TIME_SEPARATOR)  # :
+            
+            minute_token = self.current_token
+            self.eat(TokenType.INTEGER)
+            
+            hour_node = NumberNode(value=hour_token.value)
+            minute_node = NumberNode(value=minute_token.value)
+            is_24hour = True  # HH:MM is always interpreted as 24-hour
+            
+            # Optional :SS
+            if (self.current_token.type == TokenType.TIME_SEPARATOR and
+                self.peek().type == TokenType.INTEGER):
+                
+                self.eat(TokenType.TIME_SEPARATOR)  # :
+                second_token = self.current_token
+                self.eat(TokenType.INTEGER)
+                second_node = NumberNode(value=second_token.value)
+            
+            return TimeNode(
+                hour=hour_node, 
+                minute=minute_node, 
+                second=second_node,
+                is_24hour=is_24hour,
+                period=period
+            )
+        
+        # Try Chinese time format (X点X分)
+        # First check for period indicators
+        period = None
+        if self.current_token.type in [TokenType.PERIOD_AM, TokenType.PERIOD_PM]:
+            if self.current_token.type == TokenType.PERIOD_AM:
+                period = "AM"
+            else:
+                period = "PM"
+            self.eat(self.current_token.type)
+
+        if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER, TokenType.EARLY_MORNING, TokenType.LATE_NIGHT]:
+            if self.current_token.type == TokenType.EARLY_MORNING:
+                self.eat(TokenType.EARLY_MORNING)
+                is_24hour = True
+                period = "AM"
+
+                # Expect a number next
+                if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]:
+                    hour_token = self.current_token
+                    self.eat(hour_token.type)
+                    hour_node = NumberNode(value=hour_token.value)
+
+                    # "早八" should be interpreted as 08:00
+                    # If hour is greater than 12, treat as 24-hour
+                    if hour_node.value > 12:
+                        is_24hour = True
+                        period = None
+                else:
+                    raise ParserError(
+                        f"Expected number after '早', got {self.current_token.type} "
+                        f"at position {self.current_token.position}"
+                    )
+            elif self.current_token.type == TokenType.LATE_NIGHT:
+                self.eat(TokenType.LATE_NIGHT)
+                is_24hour = True
+                period = "PM"
+
+                # Expect a number next
+                if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]:
+                    hour_token = self.current_token
+                    self.eat(hour_token.type)
+                    hour_node = NumberNode(value=hour_token.value)
+
+                    # "晚十" should be interpreted as 22:00
+                    # Adjust hour to 24-hour format
+                    if hour_node.value <= 12:
+                        hour_node.value += 12
+                    is_24hour = True
+                    period = None
+                else:
+                    raise ParserError(
+                        f"Expected number after '晚', got {self.current_token.type} "
+                        f"at position {self.current_token.position}"
+                    )
+            else:
+                # Regular time parsing
+                hour_token = self.current_token
+                self.eat(hour_token.type)
+
+                # Check for 点 or 时
+                if self.current_token.type == TokenType.TIME_SEPARATOR:
+                    separator = self.current_token.value
+                    self.eat(TokenType.TIME_SEPARATOR)
+
+                    if separator == '点':
+                        is_24hour = False
+                    elif separator == '时':
+                        is_24hour = True
+
+                    hour_node = NumberNode(value=hour_token.value)
+
+                    # Optional minutes
+                    if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]:
+                        minute_token = self.current_token
+                        self.eat(minute_token.type)
+
+                        # Optional 分
+                        if self.current_token.type == TokenType.TIME_SEPARATOR and \
+                           self.current_token.value == '分':
+                            self.eat(TokenType.TIME_SEPARATOR)
+
+                        minute_node = NumberNode(value=minute_token.value)
+
+                    # Handle special markers
+                    if self.current_token.type == TokenType.HALF:
+                        self.eat(TokenType.HALF)
+                        minute_node = NumberNode(value=30)
+                    elif self.current_token.type == TokenType.QUARTER:
+                        self.eat(TokenType.QUARTER)
+                        minute_node = NumberNode(value=15)
+                    elif self.current_token.type == TokenType.ZHENG:
+                        self.eat(TokenType.ZHENG)
+                        if minute_node is None:
+                            minute_node = NumberNode(value=0)
+
+                    # Optional 钟
+                    if self.current_token.type == TokenType.ZHONG:
+                        self.eat(TokenType.ZHONG)
+                else:
+                    # If no separator, treat as hour-only time (like "三点")
+                    hour_node = NumberNode(value=hour_token.value)
+                    is_24hour = False
+
+            return TimeNode(
+                hour=hour_node,
+                minute=minute_node,
+                second=second_node,
+                is_24hour=is_24hour,
+                period=period
+            )
+        
+        raise ParserError(
+            f"Unable to parse time at position {self.current_token.position}"
+        )
+    
+    def parse_relative_date(self) -> RelativeDateNode:
+        """Parse a relative date specification."""
+        years = 0
+        months = 0
+        weeks = 0
+        days = 0
+
+        # Handle today variants
+        if self.current_token.type == TokenType.RELATIVE_TODAY:
+            self.eat(TokenType.RELATIVE_TODAY)
+            days = 0
+        elif self.current_token.type == TokenType.RELATIVE_TOMORROW:
+            self.eat(TokenType.RELATIVE_TOMORROW)
+            days = 1
+        elif self.current_token.type == TokenType.RELATIVE_DAY_AFTER_TOMORROW:
+            self.eat(TokenType.RELATIVE_DAY_AFTER_TOMORROW)
+            days = 2
+        elif self.current_token.type == TokenType.RELATIVE_THREE_DAYS_AFTER_TOMORROW:
+            self.eat(TokenType.RELATIVE_THREE_DAYS_AFTER_TOMORROW)
+            days = 3
+        elif self.current_token.type == TokenType.RELATIVE_YESTERDAY:
+            self.eat(TokenType.RELATIVE_YESTERDAY)
+            days = -1
+        elif self.current_token.type == TokenType.RELATIVE_DAY_BEFORE_YESTERDAY:
+            self.eat(TokenType.RELATIVE_DAY_BEFORE_YESTERDAY)
+            days = -2
+        elif self.current_token.type == TokenType.RELATIVE_THREE_DAYS_BEFORE_YESTERDAY:
+            self.eat(TokenType.RELATIVE_THREE_DAYS_BEFORE_YESTERDAY)
+            days = -3
+        else:
+            # Check if this looks like an absolute date pattern before processing
+            # Look ahead to see if this matches absolute date patterns
+            is_likely_absolute_date = False
+
+            # Check for MM月DD[日号] patterns (like "6月20日")
+            if (self.pos + 2 < len(self.tokens) and
+                self.tokens[self.pos].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                self.tokens[self.pos + 1].type in [TokenType.DATE_SEPARATOR, TokenType.MONTH] and
+                self.tokens[self.pos + 1].value == '月' and
+                self.tokens[self.pos + 2].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]):
+                is_likely_absolute_date = True
+
+            if is_likely_absolute_date:
+                # This looks like an absolute date, skip relative date parsing
+                raise ParserError("Looks like absolute date format")
+
+            # Try to parse extended relative time expressions
+            # Handle patterns like "明年", "去年", "下个月", "上个月", etc.
+            original_pos = self.pos
+            try:
+                # Check for "今年", "明年", "去年"
+                if self.current_token.type == TokenType.RELATIVE_THIS and self.peek().type == TokenType.YEAR:
+                    self.eat(TokenType.RELATIVE_THIS)
+                    self.eat(TokenType.YEAR)
+                    years = 0  # Current year
+                elif self.current_token.type == TokenType.RELATIVE_NEXT and self.peek().type == TokenType.YEAR:
+                    self.eat(TokenType.RELATIVE_NEXT)
+                    self.eat(TokenType.YEAR)
+                    years = 1  # Next year
+                elif self.current_token.type == TokenType.RELATIVE_LAST and self.peek().type == TokenType.YEAR:
+                    self.eat(TokenType.RELATIVE_LAST)
+                    self.eat(TokenType.YEAR)
+                    years = -1  # Last year
+                elif self.current_token.type == TokenType.RELATIVE_NEXT and self.current_token.value == "明年":
+                    self.eat(TokenType.RELATIVE_NEXT)
+                    years = 1  # Next year
+                    # Check if there's a month after "明年"
+                    if (self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                        self.peek().type == TokenType.MONTH):
+                        # Parse the month
+                        month_node = self.parse_number()
+                        self.eat(TokenType.MONTH)  # Eat the "月" token
+                        # Store the month in the months field as a special marker
+                        # We'll handle this in semantic analysis
+                        months = month_node.value - 100  # Use negative offset to indicate absolute month
+                elif self.current_token.type == TokenType.RELATIVE_LAST and self.current_token.value == "去年":
+                    self.eat(TokenType.RELATIVE_LAST)
+                    years = -1  # Last year
+                elif self.current_token.type == TokenType.RELATIVE_THIS and self.current_token.value == "今年":
+                    self.eat(TokenType.RELATIVE_THIS)
+                    years = 0  # Current year
+
+                # Check for "这个月", "下个月", "上个月"
+                elif self.current_token.type == TokenType.RELATIVE_THIS and self.peek().type == TokenType.MONTH:
+                    self.eat(TokenType.RELATIVE_THIS)
+                    self.eat(TokenType.MONTH)
+                    months = 0  # Current month
+                elif self.current_token.type == TokenType.RELATIVE_NEXT and self.peek().type == TokenType.MONTH:
+                    self.eat(TokenType.RELATIVE_NEXT)
+                    self.eat(TokenType.MONTH)
+                    months = 1  # Next month
+
+                    # Handle patterns like "下个月五号"
+                    if (self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                        self.peek().type == TokenType.DAY):
+                        # Parse the day
+                        day_node = self.parse_number()
+                        self.eat(TokenType.DAY)  # Eat the "号" token
+                        # Instead of adding days to the current date, we should set a specific day in the target month
+                        # We'll handle this in semantic analysis by setting a flag or special value
+                        days = 0  # Reset days - we'll handle the day differently
+                        # Use a special marker to indicate we want a specific day in the target month
+                        # For now, we'll just store the target day in the weeks field as a temporary solution
+                        weeks = day_node.value  # This is a hack - we'll fix this in semantic analysis
+                elif self.current_token.type == TokenType.RELATIVE_LAST and self.peek().type == TokenType.MONTH:
+                    self.eat(TokenType.RELATIVE_LAST)
+                    self.eat(TokenType.MONTH)
+                    months = -1  # Last month
+
+                # Check for "下周", "上周"
+                elif self.current_token.type == TokenType.RELATIVE_NEXT and self.peek().type == TokenType.WEEK:
+                    self.eat(TokenType.RELATIVE_NEXT)
+                    self.eat(TokenType.WEEK)
+                    weeks = 1  # Next week
+                elif self.current_token.type == TokenType.RELATIVE_LAST and self.peek().type == TokenType.WEEK:
+                    self.eat(TokenType.RELATIVE_LAST)
+                    self.eat(TokenType.WEEK)
+                    weeks = -1  # Last week
+
+                # Handle more complex patterns like "X年后", "X个月后", etc.
+                elif self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]:
+                    # Check if this is likely an absolute date format (e.g., "2025年11月21日")
+                    # If the next token after the number is a date separator or date unit,
+                    # and the number looks like a year (4 digits) or the pattern continues,
+                    # it might be an absolute date. In that case, skip relative date parsing.
+
+                    # Look ahead to see if this matches absolute date patterns
+                    lookahead_pos = self.pos
+                    is_likely_absolute_date = False
+
+                    # Check for YYYY-MM-DD or YYYY/MM/DD patterns
+                    if (lookahead_pos + 4 < len(self.tokens) and
+                        self.tokens[lookahead_pos].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                        self.tokens[lookahead_pos + 1].type in [TokenType.DATE_SEPARATOR, TokenType.YEAR] and
+                        self.tokens[lookahead_pos + 1].value in ['-', '/', '年'] and
+                        self.tokens[lookahead_pos + 2].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                        self.tokens[lookahead_pos + 3].type in [TokenType.DATE_SEPARATOR, TokenType.MONTH] and
+                        self.tokens[lookahead_pos + 3].value in ['-', '/', '月']):
+                        is_likely_absolute_date = True
+
+                    # Check for YYYY年MM月DD patterns
+                    if (lookahead_pos + 4 < len(self.tokens) and
+                        self.tokens[lookahead_pos].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                        self.tokens[lookahead_pos + 1].type in [TokenType.DATE_SEPARATOR, TokenType.YEAR] and
+                        self.tokens[lookahead_pos + 1].value == '年' and
+                        self.tokens[lookahead_pos + 2].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                        self.tokens[lookahead_pos + 3].type in [TokenType.DATE_SEPARATOR, TokenType.MONTH] and
+                        self.tokens[lookahead_pos + 3].value == '月'):
+                        is_likely_absolute_date = True
+
+                    # Check for MM月DD[日号] patterns (like "6月20日")
+                    if (self.pos + 2 < len(self.tokens) and
+                        self.tokens[self.pos].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER] and
+                        self.tokens[self.pos + 1].type in [TokenType.DATE_SEPARATOR, TokenType.MONTH] and
+                        self.tokens[self.pos + 1].value == '月' and
+                        self.tokens[self.pos + 2].type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]):
+                        is_likely_absolute_date = True
+
+                    if is_likely_absolute_date:
+                        # This looks like an absolute date, skip relative date parsing
+                        raise ParserError("Looks like absolute date format")
+
+                    print(f"DEBUG: Parsing complex relative date pattern")
+                    # Parse the number
+                    number_node = self.parse_number()
+                    number_value = number_node.value
+                    print(f"DEBUG: Parsed number: {number_value}")
+
+                    # Check the unit
+                    if self.current_token.type == TokenType.YEAR:
+                        self.eat(TokenType.YEAR)
+                        years = number_value
+                        print(f"DEBUG: Set years to {years}")
+                    elif self.current_token.type == TokenType.MONTH:
+                        self.eat(TokenType.MONTH)
+                        months = number_value
+                        print(f"DEBUG: Set months to {months}")
+                    elif self.current_token.type == TokenType.WEEK:
+                        self.eat(TokenType.WEEK)
+                        weeks = number_value
+                        print(f"DEBUG: Set weeks to {weeks}")
+                    elif self.current_token.type == TokenType.DAY:
+                        self.eat(TokenType.DAY)
+                        days = number_value
+                        print(f"DEBUG: Set days to {days}")
+                    else:
+                        print(f"DEBUG: Unexpected token type: {self.current_token.type}")
+                        raise ParserError(
+                            f"Expected time unit, got {self.current_token.type} "
+                            f"at position {self.current_token.position}"
+                        )
+
+                    # Check direction (前/后)
+                    if self.current_token.type == TokenType.RELATIVE_DIRECTION_FORWARD:
+                        self.eat(TokenType.RELATIVE_DIRECTION_FORWARD)
+                        print(f"DEBUG: Forward direction, values are already positive")
+                        # Values are already positive
+                    elif self.current_token.type == TokenType.RELATIVE_DIRECTION_BACKWARD:
+                        self.eat(TokenType.RELATIVE_DIRECTION_BACKWARD)
+                        print(f"DEBUG: Backward direction, negating values")
+                        years = -years
+                        months = -months
+                        weeks = -weeks
+                        days = -days
+
+            except ParserError:
+                # Reset position if parsing failed
+                self.pos = original_pos
+                raise ParserError(
+                    f"Expected relative date, got {self.current_token.type} "
+                    f"at position {self.current_token.position}"
+                )
+
+        return RelativeDateNode(years=years, months=months, weeks=weeks, days=days)
+    
+    def parse_weekday(self) -> WeekdayNode:
+        """Parse a weekday specification."""
+        # Parse week scope (本, 上, 下)
+        scope = "current"
+        if self.current_token.type == TokenType.WEEK_SCOPE_CURRENT:
+            self.eat(TokenType.WEEK_SCOPE_CURRENT)
+            scope = "current"
+        elif self.current_token.type == TokenType.WEEK_SCOPE_LAST:
+            self.eat(TokenType.WEEK_SCOPE_LAST)
+            scope = "last"
+        elif self.current_token.type == TokenType.WEEK_SCOPE_NEXT:
+            self.eat(TokenType.WEEK_SCOPE_NEXT)
+            scope = "next"
+
+        # Parse weekday
+        weekday_map = {
+            TokenType.WEEKDAY_MONDAY: 0,
+            TokenType.WEEKDAY_TUESDAY: 1,
+            TokenType.WEEKDAY_WEDNESDAY: 2,
+            TokenType.WEEKDAY_THURSDAY: 3,
+            TokenType.WEEKDAY_FRIDAY: 4,
+            TokenType.WEEKDAY_SATURDAY: 5,
+            TokenType.WEEKDAY_SUNDAY: 6,
+            # Handle Chinese numbers (1=Monday, 2=Tuesday, etc.)
+            TokenType.CHINESE_NUMBER: lambda x: x - 1 if 1 <= x <= 7 else None,
+        }
+
+        if self.current_token.type in weekday_map:
+            if self.current_token.type == TokenType.CHINESE_NUMBER:
+                # Handle numeric weekday (1=Monday, 2=Tuesday, etc.)
+                weekday_num = self.current_token.value
+                if 1 <= weekday_num <= 7:
+                    weekday = weekday_num - 1  # Convert to 0-based index
+                    self.eat(TokenType.CHINESE_NUMBER)
+                    return WeekdayNode(weekday=weekday, scope=scope)
+                else:
+                    raise ParserError(
+                        f"Invalid weekday number: {weekday_num} "
+                        f"at position {self.current_token.position}"
+                    )
+            else:
+                weekday = weekday_map[self.current_token.type]
+                self.eat(self.current_token.type)
+                return WeekdayNode(weekday=weekday, scope=scope)
+
+        raise ParserError(
+            f"Expected weekday, got {self.current_token.type} "
+            f"at position {self.current_token.position}"
+        )
+    
+    def parse_relative_time(self) -> RelativeTimeNode:
+        """Parse a relative time specification."""
+        hours = 0.0
+        minutes = 0.0
+        seconds = 0.0
+        
+    def parse_relative_time(self) -> RelativeTimeNode:
+        """Parse a relative time specification."""
+        hours = 0.0
+        minutes = 0.0
+        seconds = 0.0
+
+        # Parse sequences of relative time expressions
+        while self.current_token.type in [
+            TokenType.INTEGER, TokenType.CHINESE_NUMBER,
+            TokenType.HALF, TokenType.QUARTER
+        ] or (self.current_token.type == TokenType.RELATIVE_DIRECTION_FORWARD or
+              self.current_token.type == TokenType.RELATIVE_DIRECTION_BACKWARD):
+
+            # Handle 半小时
+            if (self.current_token.type == TokenType.HALF):
+                self.eat(TokenType.HALF)
+                # Optional 个
+                if (self.current_token.type == TokenType.INTEGER and
+                    self.current_token.value == "个"):
+                    self.eat(TokenType.INTEGER)
+                # Optional 小时
+                if self.current_token.type == TokenType.HOUR:
+                    self.eat(TokenType.HOUR)
+                hours += 0.5
+                # Check for direction
+                if self.current_token.type == TokenType.RELATIVE_DIRECTION_FORWARD:
+                    self.eat(TokenType.RELATIVE_DIRECTION_FORWARD)
+                elif self.current_token.type == TokenType.RELATIVE_DIRECTION_BACKWARD:
+                    self.eat(TokenType.RELATIVE_DIRECTION_BACKWARD)
+                    hours = -hours
+                continue
+
+            # Handle 一刻钟 (15 minutes)
+            if self.current_token.type == TokenType.QUARTER:
+                self.eat(TokenType.QUARTER)
+                # Optional 钟
+                if self.current_token.type == TokenType.ZHONG:
+                    self.eat(TokenType.ZHONG)
+                minutes += 15
+                # Check for direction
+                if self.current_token.type == TokenType.RELATIVE_DIRECTION_FORWARD:
+                    self.eat(TokenType.RELATIVE_DIRECTION_FORWARD)
+                elif self.current_token.type == TokenType.RELATIVE_DIRECTION_BACKWARD:
+                    self.eat(TokenType.RELATIVE_DIRECTION_BACKWARD)
+                    minutes = -minutes
+                continue
+
+            # Parse number if we have one
+            if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]:
+                number_node = self.parse_number()
+                number_value = number_node.value
+
+                # Determine unit and direction
+                unit = None
+                direction = 1  # Forward by default
+
+                # Check for unit
+                if self.current_token.type == TokenType.HOUR:
+                    self.eat(TokenType.HOUR)
+                    # Optional 个
+                    if (self.current_token.type == TokenType.INTEGER and
+                        self.current_token.value == "个"):
+                        self.eat(TokenType.INTEGER)
+                    unit = "hour"
+                elif self.current_token.type == TokenType.MINUTE:
+                    self.eat(TokenType.MINUTE)
+                    unit = "minute"
+                elif self.current_token.type == TokenType.SECOND:
+                    self.eat(TokenType.SECOND)
+                    unit = "second"
+                elif self.current_token.type == TokenType.TIME_SEPARATOR:
+                    # Handle "X点", "X分", "X秒" format
+                    sep_value = self.current_token.value
+                    self.eat(TokenType.TIME_SEPARATOR)
+                    if sep_value == "点":
+                        unit = "hour"
+                        # Optional 钟
+                        if self.current_token.type == TokenType.ZHONG:
+                            self.eat(TokenType.ZHONG)
+                        # If we have "X点" without a direction, this is likely an absolute time
+                        # Check if there's a direction after
+                        if not (self.current_token.type == TokenType.RELATIVE_DIRECTION_FORWARD or
+                                self.current_token.type == TokenType.RELATIVE_DIRECTION_BACKWARD):
+                            # This is probably an absolute time, not relative time
+                            # Push back the number and break
+                            break
+                    elif sep_value == "分":
+                        unit = "minute"
+                        # Optional 钟
+                        if self.current_token.type == TokenType.ZHONG:
+                            self.eat(TokenType.ZHONG)
+                    elif sep_value == "秒":
+                        unit = "second"
+                else:
+                    # If no unit specified, but we have a number followed by a direction,
+                    # assume it's hours
+                    if (self.current_token.type == TokenType.RELATIVE_DIRECTION_FORWARD or
+                        self.current_token.type == TokenType.RELATIVE_DIRECTION_BACKWARD):
+                        unit = "hour"
+                    else:
+                        # If no unit and no direction, this might not be a relative time expression
+                        # Push the number back and break
+                        # We can't easily push back, so let's break
+                        break
+
+                # Check for direction (后/前)
+                if self.current_token.type == TokenType.RELATIVE_DIRECTION_FORWARD:
+                    self.eat(TokenType.RELATIVE_DIRECTION_FORWARD)
+                    direction = 1
+                elif self.current_token.type == TokenType.RELATIVE_DIRECTION_BACKWARD:
+                    self.eat(TokenType.RELATIVE_DIRECTION_BACKWARD)
+                    direction = -1
+
+                # Apply the value based on unit
+                if unit == "hour":
+                    hours += number_value * direction
+                elif unit == "minute":
+                    minutes += number_value * direction
+                elif unit == "second":
+                    seconds += number_value * direction
+                continue
+
+            # If we still haven't handled the current token, break
+            break
+
+        return RelativeTimeNode(hours=hours, minutes=minutes, seconds=seconds)
+
+    def parse_time_expression(self) -> TimeExpressionNode:
+        """Parse a complete time expression."""
+        date_node = None
+        time_node = None
+        relative_date_node = None
+        relative_time_node = None
+        weekday_node = None
+
+        # Parse different parts of the expression
+        while self.current_token.type != TokenType.EOF:
+            # Try to parse date first (absolute dates should take precedence)
+            if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER]:
+                if date_node is None:
+                    original_pos = self.pos
+                    try:
+                        date_node = self.parse_date()
+                        continue
+                    except ParserError:
+                        # Reset position if parsing failed
+                        self.pos = original_pos
+                        pass
+
+            # Try to parse relative date
+            if self.current_token.type in [
+                TokenType.RELATIVE_TODAY, TokenType.RELATIVE_TOMORROW,
+                TokenType.RELATIVE_DAY_AFTER_TOMORROW, TokenType.RELATIVE_THREE_DAYS_AFTER_TOMORROW,
+                TokenType.RELATIVE_YESTERDAY, TokenType.RELATIVE_DAY_BEFORE_YESTERDAY,
+                TokenType.RELATIVE_THREE_DAYS_BEFORE_YESTERDAY,
+                TokenType.INTEGER, TokenType.CHINESE_NUMBER,  # For patterns like "X年后", "X个月后", etc.
+                TokenType.RELATIVE_NEXT, TokenType.RELATIVE_LAST, TokenType.RELATIVE_THIS
+            ]:
+                if relative_date_node is None:
+                    original_pos = self.pos
+                    try:
+                        relative_date_node = self.parse_relative_date()
+                        continue
+                    except ParserError:
+                        # Reset position if parsing failed
+                        self.pos = original_pos
+                        pass
+
+            # Try to parse relative time first (since it can have numbers)
+            if self.current_token.type in [
+                TokenType.INTEGER, TokenType.CHINESE_NUMBER,
+                TokenType.HALF, TokenType.QUARTER,
+                TokenType.RELATIVE_DIRECTION_FORWARD, TokenType.RELATIVE_DIRECTION_BACKWARD
+            ]:
+                if relative_time_node is None:
+                    original_pos = self.pos
+                    try:
+                        relative_time_node = self.parse_relative_time()
+                        # Only continue if we actually parsed some relative time
+                        if relative_time_node.hours != 0 or relative_time_node.minutes != 0 or relative_time_node.seconds != 0:
+                            continue
+                        else:
+                            # If we didn't parse any relative time, reset position
+                            self.pos = original_pos
+                    except ParserError:
+                        # Reset position if parsing failed
+                        self.pos = original_pos
+                        pass
+
+            # Try to parse time
+            if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER, TokenType.TIME_SEPARATOR, TokenType.PERIOD_AM, TokenType.PERIOD_PM]:
+                if time_node is None:
+                    original_pos = self.pos
+                    try:
+                        time_node = self.parse_time()
+                        continue
+                    except ParserError:
+                        # Reset position if parsing failed
+                        self.pos = original_pos
+                        pass
+
+            # Try to parse time
+            if self.current_token.type in [TokenType.INTEGER, TokenType.CHINESE_NUMBER, TokenType.TIME_SEPARATOR, TokenType.PERIOD_AM, TokenType.PERIOD_PM]:
+                if time_node is None:
+                    original_pos = self.pos
+                    try:
+                        time_node = self.parse_time()
+                        continue
+                    except ParserError:
+                        # Reset position if parsing failed
+                        self.pos = original_pos
+                        pass
+
+            # Try to parse weekday
+            if self.current_token.type in [
+                TokenType.WEEK_SCOPE_CURRENT, TokenType.WEEK_SCOPE_LAST, TokenType.WEEK_SCOPE_NEXT,
+                TokenType.WEEKDAY_MONDAY, TokenType.WEEKDAY_TUESDAY, TokenType.WEEKDAY_WEDNESDAY,
+                TokenType.WEEKDAY_THURSDAY, TokenType.WEEKDAY_FRIDAY, TokenType.WEEKDAY_SATURDAY,
+                TokenType.WEEKDAY_SUNDAY
+            ]:
+                if weekday_node is None:
+                    original_pos = self.pos
+                    try:
+                        weekday_node = self.parse_weekday()
+                        continue
+                    except ParserError:
+                        # Reset position if parsing failed
+                        self.pos = original_pos
+                        pass
+
+            # If we get here and couldn't parse anything, skip the token
+            self.pos += 1
+
+        return TimeExpressionNode(
+            date=date_node,
+            time=time_node,
+            relative_date=relative_date_node,
+            relative_time=relative_time_node,
+            weekday=weekday_node
+        )
+    
+    def parse(self) -> TimeExpressionNode:
+        """Parse the complete time expression and return the AST."""
+        return self.parse_time_expression()

konabot/common/ptimeparse/ptime_ast.py

@@ -0,0 +1,71 @@
+"""
+Abstract Syntax Tree (AST) nodes for the time expression parser.
+"""
+
+from abc import ABC
+from typing import Optional
+from dataclasses import dataclass
+
+
+@dataclass
+class ASTNode(ABC):
+    """Base class for all AST nodes."""
+    pass
+
+
+@dataclass
+class NumberNode(ASTNode):
+    """Represents a numeric value."""
+    value: int
+
+
+@dataclass
+class DateNode(ASTNode):
+    """Represents a date specification."""
+    year: Optional[ASTNode]
+    month: Optional[ASTNode]
+    day: Optional[ASTNode]
+
+
+@dataclass
+class TimeNode(ASTNode):
+    """Represents a time specification."""
+    hour: Optional[ASTNode]
+    minute: Optional[ASTNode]
+    second: Optional[ASTNode]
+    is_24hour: bool = False
+    period: Optional[str] = None  # AM or PM
+
+
+@dataclass
+class RelativeDateNode(ASTNode):
+    """Represents a relative date specification."""
+    years: int = 0
+    months: int = 0
+    weeks: int = 0
+    days: int = 0
+
+
+@dataclass
+class RelativeTimeNode(ASTNode):
+    """Represents a relative time specification."""
+    hours: float = 0.0
+    minutes: float = 0.0
+    seconds: float = 0.0
+
+
+@dataclass
+class WeekdayNode(ASTNode):
+    """Represents a weekday specification."""
+    weekday: int  # 0=Monday, 6=Sunday
+    scope: str  # current, last, next
+
+
+@dataclass
+class TimeExpressionNode(ASTNode):
+    """Represents a complete time expression."""
+    date: Optional[DateNode] = None
+    time: Optional[TimeNode] = None
+    relative_date: Optional[RelativeDateNode] = None
+    relative_time: Optional[RelativeTimeNode] = None
+    weekday: Optional[WeekdayNode] = None

konabot/common/ptimeparse/ptime_token.py

@@ -0,0 +1,95 @@
+"""
+Token definitions for the time parser.
+"""
+
+from enum import Enum
+from typing import Union
+from dataclasses import dataclass
+
+
+class TokenType(Enum):
+    """Types of tokens recognized by the lexer."""
+    
+    # Numbers
+    INTEGER = "INTEGER"
+    CHINESE_NUMBER = "CHINESE_NUMBER"
+    
+    # Time units
+    YEAR = "YEAR"
+    MONTH = "MONTH"
+    DAY = "DAY"
+    WEEK = "WEEK"
+    HOUR = "HOUR"
+    MINUTE = "MINUTE"
+    SECOND = "SECOND"
+    
+    # Date separators
+    DATE_SEPARATOR = "DATE_SEPARATOR"  # -, /, 年, 月, 日, 号
+    
+    # Time separators
+    TIME_SEPARATOR = "TIME_SEPARATOR"  # :, 点, 时, 分, 秒
+    
+    # Period indicators
+    PERIOD_AM = "PERIOD_AM"  # 上午, 早上, 早晨, etc.
+    PERIOD_PM = "PERIOD_PM"  # 下午, 晚上, 中午, etc.
+    
+    # Relative time
+    RELATIVE_TODAY = "RELATIVE_TODAY"  # 今天, 今晚, 今早, etc.
+    RELATIVE_TOMORROW = "RELATIVE_TOMORROW"  # 明天
+    RELATIVE_DAY_AFTER_TOMORROW = "RELATIVE_DAY_AFTER_TOMORROW"  # 后天
+    RELATIVE_THREE_DAYS_AFTER_TOMORROW = "RELATIVE_THREE_DAYS_AFTER_TOMORROW"  # 大后天
+    RELATIVE_YESTERDAY = "RELATIVE_YESTERDAY"  # 昨天
+    RELATIVE_DAY_BEFORE_YESTERDAY = "RELATIVE_DAY_BEFORE_YESTERDAY"  # 前天
+    RELATIVE_THREE_DAYS_BEFORE_YESTERDAY = "RELATIVE_THREE_DAYS_BEFORE_YESTERDAY"  # 大前天
+    RELATIVE_DIRECTION_FORWARD = "RELATIVE_DIRECTION_FORWARD"  # 后, 以后, 之后
+    RELATIVE_DIRECTION_BACKWARD = "RELATIVE_DIRECTION_BACKWARD"  # 前, 以前, 之前
+
+    # Extended relative time
+    RELATIVE_NEXT = "RELATIVE_NEXT"  # 下
+    RELATIVE_LAST = "RELATIVE_LAST"  # 上, 去
+    RELATIVE_THIS = "RELATIVE_THIS"  # 这, 本
+    
+    # Week days
+    WEEKDAY_MONDAY = "WEEKDAY_MONDAY"
+    WEEKDAY_TUESDAY = "WEEKDAY_TUESDAY"
+    WEEKDAY_WEDNESDAY = "WEEKDAY_WEDNESDAY"
+    WEEKDAY_THURSDAY = "WEEKDAY_THURSDAY"
+    WEEKDAY_FRIDAY = "WEEKDAY_FRIDAY"
+    WEEKDAY_SATURDAY = "WEEKDAY_SATURDAY"
+    WEEKDAY_SUNDAY = "WEEKDAY_SUNDAY"
+    
+    # Week scope
+    WEEK_SCOPE_CURRENT = "WEEK_SCOPE_CURRENT"  # 本
+    WEEK_SCOPE_LAST = "WEEK_SCOPE_LAST"  # 上
+    WEEK_SCOPE_NEXT = "WEEK_SCOPE_NEXT"  # 下
+    
+    # Special time markers
+    HALF = "HALF"  # 半
+    QUARTER = "QUARTER"  # 一刻
+    ZHENG = "ZHENG"  # 整
+    ZHONG = "ZHONG"  # 钟
+
+    # Student-friendly time expressions
+    EARLY_MORNING = "EARLY_MORNING"  # 早X
+    LATE_NIGHT = "LATE_NIGHT"  # 晚X
+    
+    # Whitespace
+    WHITESPACE = "WHITESPACE"
+    
+    # End of input
+    EOF = "EOF"
+
+
+@dataclass
+class Token:
+    """Represents a single token from the lexer."""
+    
+    type: TokenType
+    value: Union[str, int]
+    position: int
+    
+    def __str__(self):
+        return f"Token({self.type.value}, {repr(self.value)}, {self.position})"
+    
+    def __repr__(self):
+        return self.__str__()

konabot/common/ptimeparse/semantic.py

@@ -0,0 +1,369 @@
+"""
+Semantic analyzer for time expressions that evaluates the AST and produces datetime objects.
+"""
+
+import datetime
+import calendar
+from typing import Optional
+
+from .ptime_ast import (
+    TimeExpressionNode, DateNode, TimeNode, 
+    RelativeDateNode, RelativeTimeNode, WeekdayNode, NumberNode
+)
+from .err import TokenUnhandledException
+
+
+class SemanticAnalyzer:
+    """Semantic analyzer that evaluates time expression ASTs."""
+
+    def __init__(self, now: Optional[datetime.datetime] = None):
+        self.now = now or datetime.datetime.now()
+
+    def evaluate_number(self, node: NumberNode) -> int:
+        """Evaluate a number node."""
+        return node.value
+
+    def evaluate_date(self, node: DateNode) -> datetime.date:
+        """Evaluate a date node."""
+        year = self.now.year
+        month = 1
+        day = 1
+
+        if node.year is not None:
+            year = self.evaluate_number(node.year)
+        if node.month is not None:
+            month = self.evaluate_number(node.month)
+        if node.day is not None:
+            day = self.evaluate_number(node.day)
+
+        return datetime.date(year, month, day)
+
+    def evaluate_time(self, node: TimeNode) -> datetime.time:
+        """Evaluate a time node."""
+        hour = 0
+        minute = 0
+        second = 0
+
+        if node.hour is not None:
+            hour = self.evaluate_number(node.hour)
+        if node.minute is not None:
+            minute = self.evaluate_number(node.minute)
+        if node.second is not None:
+            second = self.evaluate_number(node.second)
+
+        # Handle 24-hour vs 12-hour format
+        if not node.is_24hour and node.period is not None:
+            if node.period == "AM":
+                if hour == 12:
+                    hour = 0
+            elif node.period == "PM":
+                if hour != 12 and hour <= 12:
+                    hour += 12
+
+        # Validate time values
+        if not (0 <= hour <= 23):
+            raise TokenUnhandledException(f"Invalid hour: {hour}")
+        if not (0 <= minute <= 59):
+            raise TokenUnhandledException(f"Invalid minute: {minute}")
+        if not (0 <= second <= 59):
+            raise TokenUnhandledException(f"Invalid second: {second}")
+
+        return datetime.time(hour, minute, second)
+
+    def evaluate_relative_date(self, node: RelativeDateNode) -> datetime.timedelta:
+        """Evaluate a relative date node."""
+        # Start with current time
+        result = self.now
+
+        # Special case: If weeks contains a target day (hacky way to pass target day info)
+        # This is for patterns like "下个月五号"
+        if node.weeks > 0 and node.weeks <= 31:  # Valid day range
+            target_day = node.weeks
+
+            # Calculate the target month
+            if node.months != 0:
+                # Handle month arithmetic carefully
+                total_months = result.month + node.months - 1
+                new_year = result.year + total_months // 12
+                new_month = total_months % 12 + 1
+
+                # Handle day overflow (e.g., Jan 31 + 1 month = Feb 28/29)
+                max_day_in_target_month = calendar.monthrange(new_year, new_month)[1]
+                target_day = min(target_day, max_day_in_target_month)
+
+                try:
+                    result = result.replace(year=new_year, month=new_month, day=target_day)
+                except ValueError:
+                    # Handle edge cases
+                    result = result.replace(year=new_year, month=new_month, day=max_day_in_target_month)
+
+            # Return the difference between the new date and the original date
+            return result - self.now
+
+        # Apply years
+        if node.years != 0:
+            # Handle year arithmetic carefully due to leap years
+            new_year = result.year + node.years
+            try:
+                result = result.replace(year=new_year)
+            except ValueError:
+                # Handle leap year edge case (Feb 29 -> Feb 28)
+                result = result.replace(year=new_year, month=2, day=28)
+
+        # Apply months
+        if node.months != 0:
+            # Check if this is a special marker for absolute month (negative offset)
+            if node.months < 0:
+                # This is an absolute month specification (e.g., from "明年五月")
+                absolute_month = node.months + 100
+                if 1 <= absolute_month <= 12:
+                    result = result.replace(year=result.year, month=absolute_month, day=result.day)
+            else:
+                # Handle month arithmetic carefully
+                total_months = result.month + node.months - 1
+                new_year = result.year + total_months // 12
+                new_month = total_months % 12 + 1
+
+                # Handle day overflow (e.g., Jan 31 + 1 month = Feb 28/29)
+                new_day = min(result.day, calendar.monthrange(new_year, new_month)[1])
+
+                result = result.replace(year=new_year, month=new_month, day=new_day)
+
+        # Apply weeks and days
+        if node.weeks != 0 or node.days != 0:
+            delta_days = node.weeks * 7 + node.days
+            result = result + datetime.timedelta(days=delta_days)
+
+        return result - self.now
+
+    def evaluate_relative_time(self, node: RelativeTimeNode) -> datetime.timedelta:
+        """Evaluate a relative time node."""
+        # Convert all values to seconds for precise calculation
+        total_seconds = (
+            node.hours * 3600 +
+            node.minutes * 60 +
+            node.seconds
+        )
+
+        return datetime.timedelta(seconds=total_seconds)
+
+    def evaluate_weekday(self, node: WeekdayNode) -> datetime.timedelta:
+        """Evaluate a weekday node."""
+        current_weekday = self.now.weekday()  # 0=Monday, 6=Sunday
+        target_weekday = node.weekday
+
+        if node.scope == "current":
+            delta = target_weekday - current_weekday
+        elif node.scope == "last":
+            delta = target_weekday - current_weekday - 7
+        elif node.scope == "next":
+            delta = target_weekday - current_weekday + 7
+        else:
+            delta = target_weekday - current_weekday
+
+        return datetime.timedelta(days=delta)
+
+    def infer_smart_time(self, hour: int, minute: int = 0, second: int = 0, base_time: Optional[datetime.datetime] = None) -> datetime.datetime:
+        """
+        Smart time inference based on current time.
+
+        For example:
+        - If now is 14:30 and user says "3点", interpret as 15:00
+        - If now is 14:30 and user says "1点", interpret as next day 01:00
+        - If now is 8:00 and user says "3点", interpret as 15:00
+        - If now is 8:00 and user says "9点", interpret as 09:00
+        """
+        # Use base_time if provided, otherwise use self.now
+        now = base_time if base_time is not None else self.now
+
+        # Handle 24-hour format directly (13-23)
+        if 13 <= hour <= 23:
+            candidate = now.replace(hour=hour, minute=minute, second=second, microsecond=0)
+            if candidate <= now:
+                candidate += datetime.timedelta(days=1)
+            return candidate
+
+        # Handle 12 (noon/midnight)
+        if hour == 12:
+            # For 12 specifically, we need to be more careful
+            # Try noon first
+            noon_candidate = now.replace(hour=12, minute=minute, second=second, microsecond=0)
+            midnight_candidate = now.replace(hour=0, minute=minute, second=second, microsecond=0)
+
+            # Special case: If it's afternoon or evening, "十二点" likely means next day midnight
+            if now.hour >= 12:
+                result = midnight_candidate + datetime.timedelta(days=1)
+                return result
+
+            # If noon is in the future and closer than midnight, use it
+            if noon_candidate > now and (midnight_candidate <= now or noon_candidate < midnight_candidate):
+                return noon_candidate
+            # If midnight is in the future, use it
+            elif midnight_candidate > now:
+                return midnight_candidate
+            # Both are in the past, use the closer one
+            elif noon_candidate > midnight_candidate:
+                return noon_candidate
+            # Otherwise use midnight next day
+            else:
+                result = midnight_candidate + datetime.timedelta(days=1)
+                return result
+
+        # Handle 1-11 (12-hour format)
+        if 1 <= hour <= 11:
+            # Calculate 12-hour format candidates
+            pm_hour = hour + 12
+            pm_candidate = now.replace(hour=pm_hour, minute=minute, second=second, microsecond=0)
+            am_candidate = now.replace(hour=hour, minute=minute, second=second, microsecond=0)
+
+            # Special case: If it's afternoon (12:00-18:00) and the hour is 1-6,
+            # user might mean either PM today or AM tomorrow.
+            # But if PM is in the future, that's more likely what they mean.
+            if 12 <= now.hour <= 18 and 1 <= hour <= 6:
+                if pm_candidate > now:
+                    return pm_candidate
+                else:
+                    # PM is in the past, so use AM tomorrow
+                    result = am_candidate + datetime.timedelta(days=1)
+                    return result
+
+            # Special case: If it's late evening (after 22:00) and user specifies early morning hours (1-5),
+            # user likely means next day early morning
+            if now.hour >= 22 and 1 <= hour <= 5:
+                result = am_candidate + datetime.timedelta(days=1)
+                return result
+
+            # Special case: In the morning (0-12:00)
+            if now.hour < 12:
+                # In the morning, for hours 1-11, generally prefer AM interpretation
+                # unless it's a very early hour that's much earlier than current time
+                # Only push to next day for very early hours (1-2) that are significantly earlier
+                if hour <= 2 and hour < now.hour and now.hour - hour >= 6:
+                    # Very early morning hour that's significantly earlier, use next day
+                    result = am_candidate + datetime.timedelta(days=1)
+                    return result
+                else:
+                    # For morning, generally prefer AM if it's in the future
+                    if am_candidate > now:
+                        return am_candidate
+                    # If PM is in the future, use it
+                    elif pm_candidate > now:
+                        return pm_candidate
+                    # Both are in the past, prefer AM if it's closer
+                    elif am_candidate > pm_candidate:
+                        return am_candidate
+                    # Otherwise use PM next day
+                    else:
+                        result = pm_candidate + datetime.timedelta(days=1)
+                        return result
+            else:
+                # General case: choose the one that's in the future and closer
+                if pm_candidate > now and (am_candidate <= now or pm_candidate < am_candidate):
+                    return pm_candidate
+                elif am_candidate > now:
+                    return am_candidate
+                # Both are in the past, use the closer one
+                elif pm_candidate > am_candidate:
+                    return pm_candidate
+                # Otherwise use AM next day
+                else:
+                    result = am_candidate + datetime.timedelta(days=1)
+                    return result
+
+        # Handle 0 (midnight)
+        if hour == 0:
+            candidate = now.replace(hour=0, minute=minute, second=second, microsecond=0)
+            if candidate <= now:
+                candidate += datetime.timedelta(days=1)
+            return candidate
+
+        # Default case (should not happen with valid input)
+        candidate = now.replace(hour=hour, minute=minute, second=second, microsecond=0)
+        if candidate <= now:
+            candidate += datetime.timedelta(days=1)
+        return candidate
+
+    def evaluate(self, node: TimeExpressionNode) -> datetime.datetime:
+        """Evaluate a complete time expression node."""
+        result = self.now
+
+        # Apply relative date (should set time to 00:00:00 for dates)
+        if node.relative_date is not None:
+            delta = self.evaluate_relative_date(node.relative_date)
+            result = result + delta
+            # For relative dates like "今天", "明天", set time to 00:00:00
+            # But only for cases where we're dealing with days, not years/months
+            if (node.date is None and node.time is None and node.weekday is None and
+                node.relative_date.years == 0 and node.relative_date.months == 0):
+                result = result.replace(hour=0, minute=0, second=0, microsecond=0)
+
+        # Apply weekday
+        if node.weekday is not None:
+            delta = self.evaluate_weekday(node.weekday)
+            result = result + delta
+            # For weekdays, set time to 00:00:00
+            if node.date is None and node.time is None:
+                result = result.replace(hour=0, minute=0, second=0, microsecond=0)
+
+        # Apply relative time
+        if node.relative_time is not None:
+            delta = self.evaluate_relative_time(node.relative_time)
+            result = result + delta
+
+        # Apply absolute date
+        if node.date is not None:
+            date = self.evaluate_date(node.date)
+            result = result.replace(year=date.year, month=date.month, day=date.day)
+            # For absolute dates without time, set time to 00:00:00
+            if node.time is None:
+                result = result.replace(hour=0, minute=0, second=0, microsecond=0)
+
+        # Apply time
+        if node.time is not None:
+            time = self.evaluate_time(node.time)
+
+            # Handle explicit period or student-friendly expressions
+            if node.time.is_24hour or node.time.period is not None:
+                # Handle explicit period
+                if not node.time.is_24hour and node.time.period is not None:
+                    hour = time.hour
+                    minute = time.minute
+                    second = time.second
+
+                    if node.time.period == "AM":
+                        if hour == 12:
+                            hour = 0
+                    elif node.time.period == "PM":
+                        # Special case: "晚上十二点" should be interpreted as next day 00:00
+                        if hour == 12 and minute == 0 and second == 0:
+                            # Move to next day at 00:00:00
+                            result = result.replace(hour=0, minute=0, second=0, microsecond=0) + datetime.timedelta(days=1)
+                            # Skip the general replacement since we've already handled it
+                            skip_general_replacement = True
+                        else:
+                            # For other PM times, convert to 24-hour format
+                            if hour != 12 and hour <= 12:
+                                hour += 12
+
+                            # Validate hour
+                            if not (0 <= hour <= 23):
+                                raise TokenUnhandledException(f"Invalid hour: {hour}")
+
+                    # Only do general replacement if we haven't handled it specially
+                    if not locals().get('skip_general_replacement', False):
+                        result = result.replace(hour=hour, minute=minute, second=second, microsecond=0)
+                else:
+                    # Already in 24-hour format
+                    result = result.replace(hour=time.hour, minute=time.minute, second=time.second, microsecond=0)
+            else:
+                # Use smart time inference for regular times
+                # But if we have an explicit date, treat the time as 24-hour format
+                if node.date is not None or node.relative_date is not None:
+                    # For explicit dates, treat time as 24-hour format
+                    result = result.replace(hour=time.hour, minute=time.minute or 0, second=time.second or 0, microsecond=0)
+                else:
+                    # Use smart time inference for regular times
+                    smart_time = self.infer_smart_time(time.hour, time.minute, time.second, base_time=result)
+                    result = smart_time
+
+        return result

konabot/common/render_error_message.py

@@ -0,0 +1,34 @@
+from typing import Any
+from loguru import logger
+from nonebot_plugin_alconna import UniMessage
+import playwright.async_api
+from playwright.async_api import Page
+
+from konabot.common.web_render import WebRenderer, konaweb
+
+
+async def render_error_message(message: str) -> UniMessage[Any]:
+    """
+    渲染文本消息为错误信息图片。
+
+    如果无法访达 Web 端则返回纯文本给用户。
+    """
+
+    async def page_function(page: Page):
+        await page.wait_for_function("typeof setContent === 'function'", timeout=3000)
+        await page.evaluate(
+            """(message) => {return setContent(message);}""",
+            message,
+        )
+
+    try:
+        img_data = await WebRenderer.render(
+            url=konaweb("error_report"),
+            target="#main",
+            other_function=page_function,
+        )
+        return UniMessage.image(raw=img_data)
+    except (playwright.async_api.Error, ConnectionError) as e:
+        logger.warning("渲染报错信息图片时出错了，回退到文本 ERR={}", e)
+        return UniMessage.text(message)
+

konabot/common/web_render/config.py

@@ -5,6 +5,7 @@ from pydantic import BaseModel
 class Config(BaseModel):
     module_web_render_weburl: str = "localhost:5173"
     module_web_render_instance: str = ""
+    module_web_render_playwright_ws: str = ""
 
     def get_instance_baseurl(self):
         if self.module_web_render_instance:

konabot/common/web_render/core.py

@@ -1,6 +1,7 @@
+from abc import ABC, abstractmethod
 import asyncio
 import queue
-from typing import Any, Callable, Coroutine
+from typing import Any, Callable, Coroutine, Generic, TypeVar
 from loguru import logger
 from playwright.async_api import (
     Page,
@@ -8,9 +9,14 @@ from playwright.async_api import (
     async_playwright,
     Browser,
     BrowserContext,
+    Error as PlaywrightError,
 )
 
+from .config import web_render_config
+from playwright.async_api import ConsoleMessage, Page
 
+T = TypeVar("T")
+TFunction = Callable[[T], Coroutine[Any, Any, Any]]
 PageFunction = Callable[[Page], Coroutine[Any, Any, Any]]
 
 
@@ -22,23 +28,17 @@ class WebRenderer:
     @classmethod
     async def get_browser_instance(cls) -> "WebRendererInstance":
         if cls.browser_pool.empty():
-            instance = await WebRendererInstance.create()
+            if web_render_config.module_web_render_playwright_ws:
+                instance = await RemotePlaywrightInstance.create(
+                    web_render_config.module_web_render_playwright_ws
+                )
+            else:
+                instance = await LocalPlaywrightInstance.create()
             cls.browser_pool.put(instance)
         instance = cls.browser_pool.get()
         cls.browser_pool.put(instance)
         return instance
 
-    @classmethod
-    async def get_browser_context(cls) -> BrowserContext:
-        instance = await cls.get_browser_instance()
-        if id(instance) not in cls.context_pool:
-            context = await instance.browser.new_context()
-            cls.context_pool[id(instance)] = context
-            logger.debug(
-                f"Created new persistent browser context for WebRendererInstance {id(instance)}"
-            )
-        return cls.context_pool[id(instance)]
-
     @classmethod
     async def render(
         cls,
@@ -67,49 +67,6 @@ class WebRenderer:
             url, target, params=params, other_function=other_function, timeout=timeout
         )
 
-    @classmethod
-    async def render_persistent_page(
-        cls,
-        page_id: str,
-        url: str,
-        target: str,
-        params: dict = {},
-        other_function: PageFunction | None = None,
-        timeout: int = 30,
-    ) -> bytes:
-        """
-        使用长期挂载的页面访问指定URL并返回截图
-
-        :param page_id: 页面唯一标识符
-        :param url: 目标URL
-        :param target: 渲染目标，如 ".box"、"#main" 等CSS选择器
-        :param timeout: 页面加载超时时间，单位秒
-        :param params: URL键值对参数
-        :param other_function: 其他自定义操作函数，接受page参数
-        :return: 截图的字节数据
-
-        """
-        logger.debug(
-            f"Requesting persistent render for page_id {page_id} at {url} targeting {target} with timeout {timeout}"
-        )
-        instance = await cls.get_browser_instance()
-        if page_id not in cls.page_pool:
-            context = await cls.get_browser_context()
-            page = await context.new_page()
-            cls.page_pool[page_id] = page
-            logger.debug(
-                f"Created new persistent page for page_id {page_id} using WebRendererInstance {id(instance)}"
-            )
-        page = cls.page_pool[page_id]
-        return await instance.render_with_page(
-            page,
-            url,
-            target,
-            params=params,
-            other_function=other_function,
-            timeout=timeout,
-        )
-
     @classmethod
     async def render_file(
         cls,
@@ -142,6 +99,75 @@ class WebRenderer:
             timeout=timeout,
         )
 
+    @classmethod
+    async def render_with_persistent_page(
+        cls,
+        page_id: str,
+        url: str,
+        target: str,
+        params: dict = {},
+        other_function: PageFunction | None = None,
+        timeout: int = 30,
+    ) -> bytes:
+        """
+        使用长期挂载的页面进行渲染
+
+        :param page_id: 页面唯一标识符
+        :param target: 渲染目标，如 ".box"、"#main" 等CSS选择器
+        :param timeout: 页面加载超时时间，单位秒
+        :param params: URL键值对参数
+        :param other_function: 其他自定义操作函数，接受page参数
+        :return: 截图的字节数据
+
+        """
+        instance = await cls.get_browser_instance()
+        logger.debug(
+            f"Using WebRendererInstance {id(instance)} to render with persistent page {page_id} targeting {target}"
+        )
+        return await instance.render_with_persistent_page(
+            page_id,
+            url,
+            target,
+            params=params,
+            other_function=other_function,
+            timeout=timeout,
+        )
+
+    @classmethod
+    async def get_persistent_page(cls, page_id: str, url: str) -> Page:
+        """
+        获取长期挂载的页面，如果不存在则创建一个新的页面并存储
+        """
+        if page_id in cls.page_pool:
+            return cls.page_pool[page_id]
+        
+        async def on_console(msg: ConsoleMessage):
+            logger.debug(f"WEB CONSOLE {msg.text}")
+
+        instance = await cls.get_browser_instance()
+        if isinstance(instance, RemotePlaywrightInstance):
+            context = await instance.browser.new_context()
+            page = await context.new_page()
+            await page.goto(url)
+            cls.page_pool[page_id] = page
+            logger.debug(f"Created new persistent page for page_id {page_id}, navigated to {url}")
+
+            page.on('console', on_console)
+
+            return page
+        elif isinstance(instance, LocalPlaywrightInstance):
+            context = await instance.browser.new_context()
+            page = await context.new_page()
+            await page.goto(url)
+            cls.page_pool[page_id] = page
+            logger.debug(f"Created new persistent page for page_id {page_id}, navigated to {url}")
+
+            page.on('console', on_console)
+
+            return page
+        else:
+            raise NotImplementedError("Unsupported WebRendererInstance type")
+
     @classmethod
     async def close_persistent_page(cls, page_id: str) -> None:
         """
@@ -156,38 +182,56 @@ class WebRenderer:
             logger.debug(f"Closed and removed persistent page for page_id {page_id}")
 
 
-class WebRendererInstance:
-    def __init__(self):
-        self._playwright: Playwright | None = None
-        self._browser: Browser | None = None
+class WebRendererInstance(ABC, Generic[T]):
+    @abstractmethod
+    async def render(
+        self,
+        url: str,
+        target: str,
+        index: int = 0,
+        params: dict[str, Any] | None = None,
+        other_function: TFunction | None = None,
+        timeout: int = 30,
+    ) -> bytes: ...
+
+    @abstractmethod
+    async def render_file(
+        self,
+        file_path: str,
+        target: str,
+        index: int = 0,
+        params: dict[str, Any] | None = None,
+        other_function: PageFunction | None = None,
+        timeout: int = 30,
+    ) -> bytes: ...
+
+    @abstractmethod
+    async def render_with_persistent_page(
+        self,
+        page_id: str,
+        url: str,
+        target: str,
+        params: dict = {},
+        other_function: PageFunction | None = None,
+        timeout: int = 30,
+    ) -> bytes: ...
+
+
+class PlaywrightInstance(WebRendererInstance[Page]):
+    def __init__(self) -> None:
+        super().__init__()
         self.lock = asyncio.Lock()
 
     @property
-    def playwright(self) -> Playwright:
-        assert self._playwright is not None
-        return self._playwright
-
-    @property
-    def browser(self) -> Browser:
-        assert self._browser is not None
-        return self._browser
-
-    async def init(self):
-        self._playwright = await async_playwright().start()
-        self._browser = await self.playwright.chromium.launch(headless=True)
-
-    @classmethod
-    async def create(cls) -> "WebRendererInstance":
-        instance = cls()
-        await instance.init()
-        return instance
+    @abstractmethod
+    def browser(self) -> Browser: ...
 
     async def render(
         self,
         url: str,
         target: str,
         index: int = 0,
-        params: dict = {},
+        params: dict[str, Any] | None = None,
         other_function: PageFunction | None = None,
         timeout: int = 30,
     ) -> bytes:
@@ -207,42 +251,41 @@ class WebRendererInstance:
             context = await self.browser.new_context()
             page = await context.new_page()
             screenshot = await self.inner_render(
-                page, url, target, index, params, other_function, timeout
+                page, url, target, index, params or {}, other_function, timeout
             )
             await page.close()
             await context.close()
             return screenshot
 
-    async def render_with_page(
-        self,
-        page: Page,
-        url: str,
-        target: str,
-        index: int = 0,
-        params: dict = {},
-        other_function: PageFunction | None = None,
-        timeout: int = 30,
-    ) -> bytes:
-        async with self.lock:
-            screenshot = await self.inner_render(
-                page, url, target, index, params, other_function, timeout
-            )
-            return screenshot
-
     async def render_file(
         self,
         file_path: str,
         target: str,
         index: int = 0,
-        params: dict = {},
+        params: dict[str, Any] | None = None,
         other_function: PageFunction | None = None,
         timeout: int = 30,
     ) -> bytes:
         file_path = "file:///" + str(file_path).replace("\\", "/")
         return await self.render(
-            file_path, target, index, params, other_function, timeout
+            file_path, target, index, params or {}, other_function, timeout
         )
 
+    async def render_with_persistent_page(
+        self,
+        page_id: str,
+        url: str,
+        target: str,
+        params: dict = {},
+        other_function: PageFunction | None = None,
+        timeout: int = 30,
+    ) -> bytes:
+        page = await WebRenderer.get_persistent_page(page_id, url)
+        screenshot = await self.inner_render(
+            page, url, target, 0, params, other_function, timeout
+        )
+        return screenshot
+
     async def inner_render(
         self,
         page: Page,
@@ -276,6 +319,85 @@ class WebRendererInstance:
         logger.debug("Screenshot taken successfully")
         return screenshot
 
+
+class LocalPlaywrightInstance(PlaywrightInstance):
+    def __init__(self):
+        self._playwright: Playwright | None = None
+        self._browser: Browser | None = None
+        super().__init__()
+
+    @property
+    def playwright(self) -> Playwright:
+        assert self._playwright is not None
+        return self._playwright
+
+    @property
+    def browser(self) -> Browser:
+        assert self._browser is not None
+        return self._browser
+
+    async def init(self):
+        self._playwright = await async_playwright().start()
+        self._browser = await self.playwright.chromium.launch(headless=True)
+
+    @classmethod
+    async def create(cls) -> "WebRendererInstance":
+        instance = cls()
+        await instance.init()
+        return instance
+
     async def close(self):
         await self.browser.close()
         await self.playwright.stop()
+
+
+class RemotePlaywrightInstance(PlaywrightInstance):
+    def __init__(self, ws_endpoint: str) -> None:
+        self._playwright: Playwright | None = None
+        self._browser: Browser | None = None
+        self._ws_endpoint = ws_endpoint
+        super().__init__()
+
+    @property
+    def playwright(self) -> Playwright:
+        assert self._playwright is not None, "Playwright must be initialized by calling init()."
+        return self._playwright
+
+    @property
+    def browser(self) -> Browser:
+        assert self._browser is not None, "Browser must be connected by calling init()."
+        return self._browser
+
+    async def init(self):
+        logger.info(f"尝试连接远程 Playwright 服务器: {self._ws_endpoint}")
+        self._playwright = await async_playwright().start()
+        try:
+            self._browser = await self.playwright.chromium.connect(
+                self._ws_endpoint
+            )
+            logger.info("成功连接到远程 Playwright 服务器。")
+        except PlaywrightError as e:
+            await self.playwright.stop()
+            raise ConnectionError(
+                f"无法连接到远程 Playwright 服务器 ({self._ws_endpoint})：{e}"
+            ) from e
+
+    @classmethod
+    async def create(cls, ws_endpoint: str) -> "RemotePlaywrightInstance":
+        """
+        创建并初始化远程 Playwright 实例的工厂方法。
+        """
+        instance = cls(ws_endpoint)
+        await instance.init()
+        return instance
+
+    async def close(self):
+        """
+        断开与远程浏览器的连接并停止本地 Playwright 实例。
+        """
+        if self._browser:
+            await self.browser.close()
+        if self._playwright:
+            await self.playwright.stop()
+        print("已断开远程连接，本地 Playwright 实例已停止。")
+

konabot/docs/concepts/中间答案.txt

@@ -1,4 +1,4 @@
-关于「中间答案」或者「提示」：
+# 关于「中间答案」或者「提示」
 
 在 KonaPH 中，当有人发送「提交答案 答案」时，会检查答案是否符合你设置的中间答案的 pattern。这个 pattern 可以有两种方式：
 

konabot/docs/concepts/罗文.txt

@@ -1,2 +1,3 @@
-关于罗文和洛温：
-    AdoreLowen 希望和洛温阿特金森区分，所以最好就不要叫他洛温了！此方 BOT 会在一些群提醒叫错了的人。
+# 关于罗文和洛温
+
+AdoreLowen 希望和洛温阿特金森区分，所以最好就不要叫他洛温了！此方 BOT 会在一些群提醒叫错了的人。

konabot/docs/lib/is_admin.txt

@@ -1,45 +1,48 @@
-指令介绍
-    is_admin - 用于判断当前事件是否来自管理员的内部权限校验函数
+# 指令介绍
 
-格式
-    from konabot.common.nb.is_admin import is_admin
-    from nonebot import on
-    from nonebot.adapters import Event
-    from loguru import logger
+`is_admin` - 用于判断当前事件是否来自管理员的内部权限校验函数
 
-    @on().handle()
-    async def _(event: Event):
-        if is_admin(event):
-            logger.info("管理员发送了消息")
+# 格式
 
-说明
-    is_admin 是 Bot 内部用于权限控制的核心函数，根据事件来源（QQ、Discord、控制台）及插件配置，判断触发事件的用户或群组是否具有管理员权限。
+```python
+from konabot.common.nb.is_admin import is_admin
+from nonebot import on
+from nonebot.adapters import Event
+from loguru import logger
+
+@on().handle()
+async def _(event: Event):
+    if is_admin(event):
+        logger.info("管理员发送了消息")
+```
+
+# 说明
+
+is_admin 是 Bot 内部用于权限控制的核心函数，根据事件来源（QQ、Discord、控制台）及插件配置，判断触发事件的用户或群组是否具有管理员权限。
 
 支持的适配器与判定逻辑：
-    • OneBot V11（QQ）
-        - 若用户 ID 在配置项 admin_qq_account 中，则视为管理员
-        - 若为群聊消息，且群 ID 在配置项 admin_qq_group 中，则视为管理员
 
-    • Discord
-        - 若频道 ID 在配置项 admin_discord_channel 中，则视为管理员
-        - 若用户 ID 在配置项 admin_discord_account 中，则视为管理员
+- OneBot V11（QQ）
+  - 若用户 ID 在配置项 admin_qq_account 中，则视为管理员
+  - 若为群聊消息，且群 ID 在配置项 admin_qq_group 中，则视为管理员
+- Discord
+  - 若频道 ID 在配置项 admin_discord_channel 中，则视为管理员
+  - 若用户 ID 在配置项 admin_discord_account 中，则视为管理员
+- Console（控制台）
+  - 所有控制台输入均默认视为管理员操作，自动返回 True
 
-    • Console（控制台）
-        - 所有控制台输入均默认视为管理员操作，自动返回 True
+# 配置项（位于插件配置中）
 
-配置项（位于插件配置中）
-    ADMIN_QQ_GROUP: list[int]
-        允许的管理员 QQ 群 ID 列表
+- `ADMIN_QQ_GROUP`: `list[int]`
+  - 允许的管理员 QQ 群 ID 列表
+- `ADMIN_QQ_ACCOUNT`: `list[int]`
+  - 允许的管理员 QQ 账号 ID 列表
+- `ADMIN_DISCORD_CHANNEL`: `list[int]`
+  - 允许的管理员 Discord 频道 ID 列表
+- `ADMIN_DISCORD_ACCOUNT`: `list[int]`
+  - 允许的管理员 Discord 用户 ID 列表
 
-    ADMIN_QQ_ACCOUNT: list[int]
-        允许的管理员 QQ 账号 ID 列表
+# 注意事项
 
-    ADMIN_DISCORD_CHANNEL: list[int]
-        允许的管理员 Discord 频道 ID 列表
-
-    ADMIN_DISCORD_ACCOUNT: list[int]
-        允许的管理员 Discord 用户 ID 列表
-
-注意事项
-    - 若未在配置文件中设置任何管理员 ID，该函数对所有非控制台事件返回 False
-    - 控制台事件始终拥有管理员权限，便于本地调试与运维
+- 若未在配置文件中设置任何管理员 ID，该函数对所有非控制台事件返回 False
+- 控制台事件始终拥有管理员权限，便于本地调试与运维

konabot/docs/sys/konaperm.txt

@@ -0,0 +1,212 @@
+# 指令介绍
+
+`konaperm` - 用于查看和修改 Bot 内部权限系统记录的管理员指令
+
+## 权限要求
+
+只有拥有 `admin` 权限的主体才能使用本指令。
+
+## 格式
+
+```text
+konaperm list <platform> <entity_type> <external_id> [page]
+konaperm get <platform> <entity_type> <external_id> <perm>
+konaperm set <platform> <entity_type> <external_id> <perm> <val>
+```
+
+## 子命令说明
+
+### `list`
+
+列出指定对象及其继承链上的显式权限记录，按分页输出。
+
+参数：
+
+- `platform` 平台名，如 `ob11`、`discord`、`sys`
+- `entity_type` 对象类型，如 `user`、`group`、`global`
+- `external_id` 平台侧对象 ID；全局对象通常写 `global`
+- `page` 页码，可省略，默认 `1`
+
+### `get`
+
+查询某个对象对指定权限的最终判断结果，并说明它是从哪一层继承来的。
+
+参数：
+
+- `platform`
+- `entity_type`
+- `external_id`
+- `perm` 权限键，如 `admin`、`plugin.xxx.use`
+
+### `set`
+
+为指定对象写入显式权限。
+
+参数：
+
+- `platform`
+- `entity_type`
+- `external_id`
+- `perm` 权限键
+- `val` 设置值
+
+`val` 支持以下写法：
+
+- 允许：`y` `yes` `allow` `true` `t`
+- 拒绝：`n` `no` `deny` `false` `f`
+- 清除：`null` `none`
+
+其中：
+
+- 允许 表示显式授予该权限
+- 拒绝 表示显式禁止该权限
+- 清除 表示删除该层的显式设置，重新回退到继承链判断
+
+## 对象格式
+
+本指令操作的对象由三段组成：
+
+```text
+<platform>.<entity_type>.<external_id>
+```
+
+例如：
+
+- `ob11.user.123456789`
+- `ob11.group.987654321`
+- `sys.global.global`
+
+## 当前支持的 `PermEntity` 值
+
+以下内容按当前实现整理，便于手工查询和设置权限。
+
+### `sys`
+
+- `sys.global.global`
+
+这是系统总兜底对象。
+
+### `ob11`
+
+- `ob11.global.global`
+- `ob11.group.<group_id>`
+- `ob11.user.<user_id>`
+
+常见场景：
+
+- 给整个 OneBot V11 平台统一授权：`ob11.global.global`
+- 给某个 QQ 群授权：`ob11.group.群号`
+- 给某个 QQ 用户授权：`ob11.user.QQ号`
+
+### `discord`
+
+- `discord.global.global`
+- `discord.guild.<guild_id>`
+- `discord.channel.<channel_id>`
+- `discord.user.<user_id>`
+
+常见场景：
+
+- 给整个 Discord 平台统一授权：`discord.global.global`
+- 给某个服务器授权：`discord.guild.服务器ID`
+- 给某个频道授权：`discord.channel.频道ID`
+- 给某个用户授权：`discord.user.用户ID`
+
+### `minecraft`
+
+- `minecraft.global.global`
+- `minecraft.server.<server_name>`
+- `minecraft.player.<player_uuid_hex>`
+
+常见场景：
+
+- 给整个 Minecraft 平台统一授权：`minecraft.global.global`
+- 给某个服务器授权：`minecraft.server.服务器名`
+- 给某个玩家授权：`minecraft.player.玩家UUID的hex`
+
+### `console`
+
+- `console.global.global`
+- `console.channel.<channel_id>`
+- `console.user.<user_id>`
+
+### 快速参考
+
+```text
+sys.global.global
+
+ob11.global.global
+ob11.group.<group_id>
+ob11.user.<user_id>
+
+discord.global.global
+discord.guild.<guild_id>
+discord.channel.<channel_id>
+discord.user.<user_id>
+
+minecraft.global.global
+minecraft.server.<server_name>
+minecraft.player.<player_uuid_hex>
+
+console.global.global
+console.channel.<channel_id>
+console.user.<user_id>
+```
+
+## 权限继承
+
+权限不是只看当前对象，还会按继承链回退。
+
+例如对 `ob11.user.123456` 查询时，通常会从更具体的对象一路回退到：
+
+1. 当前用户
+2. 平台全局对象
+3. 系统全局对象
+
+权限键本身也支持逐级回退。比如查询 `plugin.demo.use` 时，可能依次命中：
+
+1. `plugin.demo.use`
+2. `plugin.demo`
+3. `plugin`
+4. `*`
+
+所以 `get` 返回的结果可能来自更宽泛的权限键，或更上层的继承对象。
+
+## 示例
+
+```text
+konaperm list ob11 user 123456
+```
+
+查看 `ob11.user.123456` 及其继承链上的权限记录第一页。
+
+```text
+konaperm get ob11 user 123456 admin
+```
+
+查看该用户最终是否拥有 `admin` 权限，以及命中来源。
+
+```text
+konaperm set ob11 user 123456 admin allow
+```
+
+显式授予该用户 `admin` 权限。
+
+```text
+konaperm set ob11 user 123456 admin deny
+```
+
+显式拒绝该用户 `admin` 权限。
+
+```text
+konaperm set ob11 user 123456 admin none
+```
+
+删除该用户这一层对 `admin` 的显式设置，恢复继承判断。
+
+## 注意事项
+
+- 这是系统级管理指令，误操作可能直接影响其他插件的权限控制。
+- `list` 只列出显式记录；没有显示出来不代表最终一定无权限，可能是从上层继承。
+- `get` 显示的是最终命中的结果，比 `list` 更适合排查“为什么有/没有某个权限”。
+- 对 `admin` 或 `*` 这类高影响权限做修改前，建议先确认对象是否写对。

konabot/docs/sys/konaph.txt

@@ -1,4 +1,5 @@
-指令介绍
-    konaph - KonaBot 的 PuzzleHunt 管理工具
+# 指令介绍
+
+`konaph` - KonaBot 的 PuzzleHunt 管理工具
 
 详细介绍请直接输入 konaph 获取使用指引（该指令权限仅对部分人开放。如果你有权限的话才有响应。建议在此方 BOT 私聊使用该指令。）

konabot/docs/sys/宾几人.txt

@@ -0,0 +1,4 @@
+# 宾几人
+
+查询 Bingo 有几个人。直接发送给 Bot 即可。
+

konabot/docs/user/celeste.txt

@@ -0,0 +1,38 @@
+# Celeste
+
+爬山小游戏，移植自 Ccleste，是 Celeste Classic（即 PICO-8）版。
+
+使用 `wasdxc` 和数字进行操作。
+
+## 操作说明
+
+`wsad` 是上下左右摇杆方向，或者是方向键。`c` 是跳跃键，`x` 是冲刺键。
+
+使用空格分隔每一个操作，每个操作持续一帧。如果后面跟着数字，则持续那么多帧。
+
+### 例子 1
+
+```
+xc 180
+```
+
+按下 xc 一帧，然后空置 180 帧。
+
+### 例子 2
+
+```
+d10 cd d10 xdw d20
+```
+
+向右走 10 帧，向右跳一帧，再继续按下右 10 帧，按下向右上冲刺一帧，再按下右 20 帧。
+
+## 指令使用说明
+
+直接说 `celeste` 会开启一个新的游戏。但是，你需要在后面跟有操作，才能够渲染 gif 图出来。
+
+一个常见的开始操作是直接发送 `celeste xc 130`，即按下 xc 两个按键触发 PICO 版的开始游戏，然后等待 130 秒动画播放完毕。
+
+对于一个已经存在而且时间不是非常久远的 gif 图，只要是由 bot 自己发送出来的，就可以在它的基础上继续游戏。回复这条消息，可以继续游戏。
+
+一种很常见的技巧是回复一个已经存在的 gif 图 `celeste 1`，此时会空操作一帧并且渲染画面。你可以用这种方法查看一个 gif 图的游戏目前的状态。
+

konabot/docs/user/fx.txt

@@ -0,0 +1,107 @@
+## 指令介绍
+
+`fx` - 用于对图片应用各种滤镜效果的指令
+
+## 格式
+
+```
+fx [滤镜名称] <参数1> <参数2> ...
+```
+
+## 示例
+
+- `fx 模糊`
+- `fx 阈值 150`
+- `fx 缩放 2.0`
+- `fx 色彩 1.8`
+- `fx 色键 rgb(0,255,0) 50`
+
+## 可用滤镜列表
+
+### 基础滤镜
+* ```fx 轮廓```
+* ```fx 锐化```
+* ```fx 边缘增强```
+* ```fx 浮雕```
+* ```fx 查找边缘```
+* ```fx 平滑```
+* ```fx 暗角 <半径=1.5>```
+* ```fx 发光 <强度=0.5> <模糊半径=15>```
+* ```fx 噪点 <数量=0.05>```
+* ```fx 素描```
+* ```fx 阴影 <偏移量X=10> <偏移量Y=10> <模糊量=10> <不透明度=0.5> <阴影颜色=black>```
+
+### 模糊滤镜
+* ```fx 模糊 <半径=10>```
+* ```fx 马赛克 <像素大小=10>```
+* ```fx 径向模糊 <强度=3.0> <采样量=6>```
+* ```fx 旋转模糊 <强度=30.0> <采样量=6>```
+* ```fx 方向模糊 <角度=0.0> <距离=20> <采样量=6>```
+* ```fx 缩放模糊 <强度=0.1> <采样量=6>```
+* ```fx 边缘模糊 <半径=10.0>```
+
+### 色彩处理滤镜
+* ```fx 反色```
+* ```fx 黑白```
+* ```fx 阈值 <阈值=128>```
+* ```fx 对比度 <因子=1.5>```
+* ```fx 亮度 <因子=1.5>```
+* ```fx 色彩 <因子=1.5>```
+* ```fx 色调 <颜色="rgb(255,0,0)">```
+* ```fx RGB分离 <偏移量=5>```
+* ```fx 叠加颜色 <颜色列表=[rgb(255,0,0)|(0,0)+rgb(0,255,0)|(0,100)+rgb(0,0,255)|(50,100)]> <叠加模式=overlay>```
+* ```fx 像素抖动 <最大偏移量=2>```
+* ```fx 半调 <半径=5>```
+* ```fx 描边 <半径=5> <颜色=black>```
+* ```fx 形状描边 <半径=5> <颜色=black> <粗糙度=None>```
+
+### 几何变换滤镜
+* ```fx 平移 <x偏移量=10> <y偏移量=10>```
+* ```fx 缩放 <比例(X)=1.5> <比例Y=None>```
+* ```fx 旋转 <角度=45>```
+* ```fx 透视变换 <变换矩阵>```
+* ```fx 裁剪 <左=0> <上=0> <右=100> <下=100>（百分比）```
+* ```fx 拓展边缘 <拓展量=10>```
+* ```fx 波纹 <振幅=5> <波长=20>```
+* ```fx 光学补偿 <数量=100> <反转=false>```
+* ```fx 球面化 <强度=0.5>```
+* ```fx 镜像 <角度=90>```
+* ```fx 水平翻转```
+* ```fx 垂直翻转```
+* ```fx 复制 <目标位置=(100,100)> <缩放=1.0> <源区域=(0,0,100,100)>（百分比）```
+
+### 特殊效果滤镜
+* ```fx 设置通道 <通道=A>```
+  * 可用 R、G、B、A。
+* ```fx 设置遮罩```
+* ```fx 色键 <目标颜色="rgb(255,0,0)"> <容差=60>```
+* ```fx 晃动 <最大偏移量=5> <运动模糊=False>```
+* ```fx 动图 <帧率=10>```
+
+### 多图像处理器
+* ```fx 存入图像 <目标名称>```
+  * 目标名称是图像的代名词，图像最长可存 12 小时，如果公用容量满了图像也会被删除。
+  * 该项仅可于首项使用。
+* ```fx 读取图像 <目标名称>```
+  * 该项仅可于首项使用。
+* ```fx 暂存图像```
+  * 此项默认插入存储在暂存列表中第一张图像的后面。
+* ```fx 交换图像 <交换项=2> <交换项=1>```
+* ```fx 删除图像 <删除索引=1>```
+* ```fx 选择图像 <目标索引=2>```
+
+### 多图像混合
+* ```fx 混合图像 <模式=normal> <alpha=0.5>```
+* ```fx 覆盖图像```
+
+### 生成类
+* ```fx 覆加颜色 <颜色列表=[rgb(255,0,0)|(0,0)+rgb(0,255,0)|(0,100)+rgb(0,0,255)|(50,100)]>```
+* ```fx 生成图层 <宽度=512> <高度=512>```
+* ```fx 生成文本 <文本内容=请输入文本> <字体大小=32> <文字颜色=black> <字体文件=HarmonyOS_Sans_SC_Regular.ttf>```
+
+## 颜色名称支持
+- **格式**：颜色列表采用 ```[颜色|位置+颜色|位置+颜色|位置]``` 的格式，位置是形如```(x百分比,y百分比)```的元组。
+- **基本颜色**：红、绿、蓝、黄、紫、黑、白、橙、粉、灰、青、靛、棕
+- **修饰词**：浅、深、亮、暗（可组合使用，如`浅红`、`深蓝`）
+- **RGB格式**：`rgb(255,0,0)`、`rgb(0,255,0)`、`(255,0,0)` 等
+- **HEX格式**：`#66ccff`等

konabot/docs/user/giftool.txt

@@ -1,59 +1,83 @@
-指令介绍
-    giftool - 对 GIF 动图进行裁剪、抽帧等处理
+# giftool - 对 GIF 动图进行裁剪、抽帧等处理
 
-格式
-    giftool [图片] [选项]
+## 格式
 
-示例
-    回复一张 GIF 并发送：
-        `giftool --ss 1.5 -t 2.0`
-        从 1.5 秒处开始，截取 2 秒长度的片段。
+```bash
+giftool [图片] [选项]
+```
 
-    `giftool [图片] --ss 0:10 -to 0:15`
-        截取从 10 秒到 15 秒之间的片段（支持 MM:SS 或 HH:MM:SS 格式）。
+## 示例
 
-    `giftool [图片] --frames:v 10`
-        将整张 GIF 均匀抽帧，最终保留 10 帧。
+- **回复一张 GIF 并发送：**
 
-    `giftool [图片] --ss 2 --frames:v 5`
-        从第 2 秒开始截取，并将结果抽帧为 5 帧。
+  ```bash
+  giftool --ss 1.5 -t 2.0
+  ```
 
-参数说明
-    图片（必需）
-        - 必须是 GIF 动图。
-        - 支持直接附带图片，或回复一条含 GIF 的消息后使用指令。
+  从 1.5 秒处开始，截取 2 秒长度的片段。
 
-    --ss <时间戳>（可选）
-        - 指定开始时间（单位：秒），可使用以下格式：
-            • 纯数字（如 `1.5` 表示 1.5 秒）
-            • 分秒格式（如 `1:30` 表示 1 分 30 秒）
-            • 时分秒格式（如 `0:1:30` 表示 1 分 30 秒）
-        - 默认从开头开始（0 秒）。
+- ```bash
+  giftool [图片] --ss 0:10 -to 0:15
+  ```
 
-    -t <持续时间>（可选）
-        - 指定截取的持续时间（单位：秒），格式同 --ss。
-        - 与 --ss 配合使用：截取 [ss, ss + t] 区间。
-        - 不能与 --to 同时使用。
+  截取从 10 秒到 15 秒之间的片段（支持 `MM:SS` 或 `HH:MM:SS` 格式）。
 
-    --to <时间戳>（可选）
-        - 指定结束时间（单位：秒），格式同 --ss。
-        - 与 --ss 配合使用：截取 [ss, to] 区间。
-        - 不能与 -t 同时使用。
+- ```bash
+  giftool [图片] --frames:v 10
+  ```
 
-    --frames:v <帧数>（可选）
-        - 对截取后的片段进行均匀抽帧，保留指定数量的帧。
-        - 帧数必须为正整数（> 0）。
-        - 若原始帧数 ≤ 指定帧数，则保留全部帧。
+  将整张 GIF 均匀抽帧，最终保留 10 帧。
 
-    --speed <速度>（可选）
-        - 调整 gif 图的速度。若为负数，则代表倒放
+- ```bash
+  giftool [图片] --ss 2 --frames:v 5
+  ```
 
-使用方式
-    1. 发送指令前，请确保：
-        - 消息中附带一张 GIF 动图，或
-        - 回复一条包含 GIF 动图的消息后再发送指令。
-    2. 插件会自动：
-        - 解析 GIF 的每一帧及其持续时间（duration）
-        - 根据时间参数转换为帧索引进行裁剪
-        - 如指定抽帧，则对裁剪后的片段均匀采样
-        - 生成新的 GIF 并保持原始循环设置（loop=0）
+  从第 2 秒开始截取，并将结果抽帧为 5 帧。
+
+## 参数说明
+
+### 图片（必需）
+
+- 必须是 GIF 动图。
+- 支持直接附带图片，或回复一条含 GIF 的消息后使用指令。
+
+### `--ss <时间戳>`（可选）
+
+- 指定开始时间（单位：秒），可使用以下格式：
+  - 纯数字（如 `1.5` 表示 1.5 秒）
+  - 分秒格式（如 `1:30` 表示 1 分 30 秒）
+  - 时分秒格式（如 `0:1:30` 表示 1 分 30 秒）
+- 默认从开头开始（0 秒）。
+
+### `-t <持续时间>`（可选）
+
+- 指定截取的持续时间（单位：秒），格式同 `--ss`。
+- 与 `--ss` 配合使用：截取 `[ss, ss + t]` 区间。
+- **不能与 `--to` 同时使用。**
+
+### `--to <时间戳>`（可选）
+
+- 指定结束时间（单位：秒），格式同 `--ss`。
+- 与 `--ss` 配合使用：截取 `[ss, to]` 区间。
+- **不能与 `-t` 同时使用。**
+
+### `--frames:v <帧数>`（可选）
+
+- 对截取后的片段进行均匀抽帧，保留指定数量的帧。
+- 帧数必须为正整数（> 0）。
+- 若原始帧数 ≤ 指定帧数，则保留全部帧。
+
+### `--speed <速度>`（可选）
+
+- 调整 GIF 图的速度。若为负数，则代表倒放。
+
+## 使用方式
+
+1. 发送指令前，请确保：
+   - 消息中附带一张 GIF 动图，**或**
+   - 回复一条包含 GIF 动图的消息后再发送指令。
+2. 插件会自动：
+   - 解析 GIF 的每一帧及其持续时间（duration）
+   - 根据时间参数转换为帧索引进行裁剪
+   - 如指定抽帧，则对裁剪后的片段均匀采样
+   - 生成新的 GIF 并保持原始循环设置（`loop=0`）

konabot/docs/user/k8x12S.txt

@@ -0,0 +1,10 @@
+# 指令介绍
+
+根据文字生成 k8x12S
+
+> 「现在还不知道k8x12S是什么的可以开除界隈籍了」—— Louis, 2025/12/31
+
+## 使用指南
+
+`k8x12S 安心をしてください`
+

konabot/docs/user/man.txt

@@ -1,20 +1,33 @@
-指令介绍
-    man - 用于展示此方 BOT 使用手册的指令
+# 指令介绍
 
-格式
-    man 文档类型
-    man [文档类型] <指令>
+`man` - 用于展示此方 BOT 使用手册的指令
 
-示例
-    `man` 查看所有有文档的指令清单
-    `man 3` 列举所有可读文档的库函数清单
-    `man 喵` 查看指令「喵」的使用说明
-    `man 8 out` 查看管理员指令「out」的使用说明
+## 格式
 
-文档类型
-    文档类型用来区分同一指令在不同场景下的情景。你可以使用数字编号进行筛选。分为这些种类：
+```
+man 文档类型
+man [文档类型] <指令>
+```
 
-    - 1 用户态指令，用于日常使用的指令
-    - 3 库函数指令，用于 Bot 开发用的函数查询
-    - 7 概念指令，用于概念解释
-    - 8 系统指令，仅管理员可用
+## 示例
+
+- ``man``
+  查看所有有文档的指令清单
+
+- ``man 3``
+  列举所有可读文档的库函数清单
+
+- ``man 喵``
+  查看指令「喵」的使用说明
+
+- ``man 8 out``
+  查看管理员指令「out」的使用说明
+
+## 文档类型
+
+文档类型用来区分同一指令在不同场景下的情景。你可以使用数字编号进行筛选。分为以下种类：
+
+- **1** 用户态指令：用于日常使用的指令
+- **3** 库函数指令：用于 Bot 开发用的函数查询
+- **7** 概念指令：用于概念解释
+- **8** 系统指令：仅管理员可用

konabot/docs/user/ntfy.txt

@@ -1,15 +1,16 @@
-指令介绍
-    ntfy - 配置使用 ntfy 来更好地为你通知此方 BOT 代办
+## 指令介绍
+**`ntfy`** - 配置使用 [ntfy](https://ntfy.sh/) 来更好地为你通知此方 BOT 的待办事项。
 
-指令示例
-    `ntfy 创建`
-        创建一个随机的 ntfy 订阅主题来提醒代办，此方 Bot 将会给你使用指引。你可以前往 https://ntfy.sh/ 官网下载 ntfy APP，或者使用网页版 ntfy。
+## 指令示例
 
-    `ntfy 创建 kagami-notice`
-        创建一个名字含有 kagami-notice 的 ntfy 订阅主题
+- **`ntfy 创建`**
+  创建一个随机的 ntfy 订阅主题来提醒待办。此方 Bot 将会给你使用指引。你可以前往 [https://ntfy.sh/](https://ntfy.sh/) 官网下载 ntfy APP，或者使用网页版 ntfy。
 
-    `ntfy 删除`
-        清除并不再使用 ntfy 向你通知
+- **`ntfy 创建 kagami-notice`**
+  创建一个名称包含 `kagami-notice` 的 ntfy 订阅主题。
 
-另见
-    提醒我(1) 查询提醒(1) 删除提醒(1)
+- **`ntfy 删除`**
+  清除配置，不再使用 ntfy 向你发送通知。
+
+## 另见
+[`提醒我(1)`](#) [`查询提醒(1)`](#) [`删除提醒(1)`](#)

konabot/docs/user/openssl.txt

@@ -1,21 +1,39 @@
-指令介绍
-    openssl - 用于生成指定长度的加密安全随机数据
+# 指令介绍
 
-格式
-    openssl rand <模式> <字节数>
+`openssl rand` — 用于生成指定长度的加密安全随机数据。
 
-示例
-    `openssl rand -hex 16` 生成 16 字节的十六进制随机数
-    `openssl rand -base64 32` 生成 32 字节并以 Base64 编码输出的随机数据
+## 格式
 
-说明
-    该指令使用 Python 的 secrets 模块生成加密安全的随机字节，并支持以十六进制（-hex）或 Base64（-base64）格式输出。
+```bash
+openssl rand <模式> <字节数>
+```
 
-参数说明
-    模式（mode）
-        - -hex     ：以十六进制字符串形式输出随机数据
-        - -base64  ：以 Base64 编码字符串形式输出随机数据
+## 示例
 
-    字节数（num）
-        - 必须为正整数
-        - 最大支持 256 字节
+- ```bash
+  openssl rand -hex 16
+  ```
+  生成 16 字节的十六进制随机数。
+
+- ```bash
+  openssl rand -base64 32
+  ```
+  生成 32 字节并以 Base64 编码输出的随机数据。
+
+## 说明
+
+该指令使用 Python 的 `secrets` 模块生成加密安全的随机字节，并支持以以下格式输出：
+- 十六进制（`-hex`）
+- Base64 编码（`-base64`）
+
+## 参数说明
+
+### 模式（mode）
+
+- `-hex`：以十六进制字符串形式输出随机数据
+- `-base64`：以 Base64 编码字符串形式输出随机数据
+
+### 字节数（num）
+
+- 必须为正整数
+- 最大支持 256 字节

konabot/docs/user/shadertool.txt

@@ -1,47 +1,55 @@
-指令介绍  
-    shadertool - 使用 SkSL（Skia Shader Language）代码实时渲染并生成 GIF 动画
+# 指令介绍
+`shadertool` - 使用 SkSL（Skia Shader Language）代码实时渲染并生成 GIF 动画
 
-格式  
-    shadertool [选项] <SkSL 代码>
+## 格式
+```bash
+shadertool [选项] <SkSL 代码>
+```
 
-示例  
-    shadertool """
-    uniform float u_time;
-    uniform float2 u_resolution;
+## 示例
+```bash
+shadertool """
+uniform float u_time;
+uniform float2 u_resolution;
 
-    half4 main(float2 coord) {
-        return half4(
-            1.0,
-            sin((coord.y / u_resolution.y + u_time) * 3.1415926 * 2) * 0.5 + 0.5,
-            coord.x / u_resolution.x,
-            1.0
-        );
-    }
-    """
+half4 main(float2 coord) {
+    return half4(
+        1.0,
+        sin((coord.y / u_resolution.y + u_time) * 3.1415926 * 2) * 0.5 + 0.5,
+        coord.x / u_resolution.x,
+        1.0
+    );
+}
+"""
+```
 
-参数说明  
-    SkSL 代码（必填）  
-        - 类型：字符串（建议用英文双引号包裹）  
-        - 内容：符合 SkSL 语法的片段着色器代码，必须包含 `void main()` 函数，并为 `sk_FragColor` 赋值。  
-        - 注意：插件会自动去除代码首尾的单引号或双引号，便于命令行输入。
+## 参数说明
 
-    --width <整数>（可选）  
-        - 默认值：320  
-        - 作用：输出 GIF 的宽度（像素），必须大于 0。
+### SkSL 代码（必填）
+- **类型**：字符串（建议用英文双引号包裹）
+- **内容**：符合 SkSL 语法的片段着色器代码，必须包含 `main` 函数，并返回 `half4` 类型的颜色值。
+- **注意**：插件会自动去除代码首尾的单引号或双引号，便于命令行输入。
 
-    --height <整数>（可选）  
-        - 默认值：180  
-        - 作用：输出 GIF 的高度（像素），必须大于 0。
+### `--width <整数>`（可选）
+- **默认值**：`320`
+- **作用**：输出 GIF 的宽度（像素），必须大于 0。
 
-    --duration <浮点数>（可选）  
-        - 默认值：1.0  
-        - 作用：动画总时长（秒），必须大于 0。  
-        - 限制：`duration × fps` 必须 ≥ 1 且 ≤ 100（即至少 1 帧，最多 100 帧）。
+### `--height <整数>`（可选）
+- **默认值**：`180`
+- **作用**：输出 GIF 的高度（像素），必须大于 0。
 
-    --fps <浮点数>（可选）  
-        - 默认值：15.0  
-        - 作用：每秒帧数，控制动画流畅度，必须大于 0。  
-        - 常见值：10（低配流畅）、15（默认）、24/30（电影/视频级）。
+### `--duration <浮点数>`（可选）
+- **默认值**：`1.0`
+- **作用**：动画总时长（秒），必须大于 0。
+- **限制**：`duration × fps` 必须 ≥ 1 且 ≤ 100（即至少 1 帧，最多 100 帧）。
 
-使用方式  
-    直接在群聊或私聊中发送 `shadertool` 指令，附上合法的 SkSL 代码即可。
+### `--fps <浮点数>`（可选）
+- **默认值**：`15.0`
+- **作用**：每秒帧数，控制动画流畅度，必须大于 0。
+- **常见值**：
+  - `10`：低配流畅
+  - `15`：默认
+  - `24` / `30`：电影/视频级流畅度
+
+## 使用方式
+直接在群聊或私聊中发送 `shadertool` 指令，附上合法的 SkSL 代码即可。

konabot/docs/user/textfx.txt

@@ -0,0 +1,174 @@
+# 文字处理机器人使用手册（小白友好版）
+
+欢迎使用文字处理机器人！你不需要懂编程，只要会打字，就能用它完成各种文字操作——比如加密、解密、打乱顺序、进制转换、排版整理等。
+
+---
+
+## 一、基础演示
+
+在 QQ 群里这样使用：
+
+1. **直接输入命令**（适合短文本）：
+   ```
+   /textfx reverse 你好世界
+   ```
+   → 机器人回复：`界世好你`
+
+2. **先发一段文字，再用命令处理它**（适合长文本）：
+   - 先发送：`Hello, World!`
+   - 回复这条消息，输入：
+     ```
+     /textfx b64 encode
+     ```
+   → 机器人返回：`SGVsbG8sIFdvcmxkIQ==`
+
+> 命令可写为 `/textfx`、`/处理文字` 或 `/处理文本`。
+> 若不回复消息，命令会处理当前行后面的文本。
+
+---
+
+## 二、流水线语法（超简单）
+
+- 用 `|` 连接多个操作，前一个的输出自动作为后一个的输入。
+- 用 `;` 分隔多条独立指令，它们各自产生输出，最终合并显示。
+- 用 `>` 或 `>>` 把结果保存起来（见下文），被重定向的指令不会产生输出。
+
+**例子**：把"HELLO"先反转，再转成摩斯电码：（转换为摩斯电码功能暂未实现）
+```
+textfx reverse HELLO | morse en
+```
+→ 输出：`--- .-.. .-.. . ....`
+
+**例子**：多条指令各自输出：
+```
+textfx echo 你好; echo 世界
+```
+→ 输出：
+```
+你好
+世界
+```
+
+**例子**：重定向的指令不输出，其余正常输出：
+```
+textfx echo 1; echo 2 > a; echo 3
+```
+→ 输出：
+```
+1
+3
+```
+
+---
+
+## 三、功能清单（含示例）
+
+### reverse（或 rev、反转）
+反转文字。
+示例：`/textfx reverse 爱你一万年` → `年万一你爱`
+
+### b64（或 base64）
+Base64 编码或解码。
+示例：`/textfx b64 encode 你好` → `5L2g5aW9`
+示例：`/textfx b64 decode 5L2g5aW9` → `你好`
+
+### caesar（或 凯撒、rot）
+凯撒密码（仅对英文字母有效）。
+示例：`/textfx caesar 3 ABC` → `DEF`
+示例：`/textfx caesar -3 DEF` → `ABC`
+
+### morse（或 摩斯）
+将摩斯电码解码为文字（支持英文和日文）。字符间用空格，单词间用 `/`。
+示例：`/textfx morse en .... . .-.. .-.. ---` → `HELLO`
+示例：`/textfx morse jp -... --.-- -.. --.. ..- ..` → `ハアホフウイ`
+
+### baseconv（或 进制转换）
+在不同进制之间转换数字。
+示例：`/textfx baseconv 2 10 1101` → `13`
+示例：`/textfx baseconv 10 16 255` → `FF`
+
+### shuffle（或 打乱）
+随机打乱文字顺序。
+示例：`/textfx shuffle abcdef` → `fcbade`（每次结果不同）
+
+### sort（或 排序）
+将文字按字符顺序排列。
+示例：`/textfx sort dbca` → `abcd`
+
+### b64hex
+在 Base64 和十六进制之间互转。
+示例：`/textfx b64hex dec SGVsbG8=` → `48656c6c6f`
+示例：`/textfx b64hex enc 48656c6c6f` → `SGVsbG8=`
+
+### align（或 format、排版）
+按指定格式分组排版文字。
+示例：`/textfx align 2 4 0123456789abcdef` →
+```
+01 23 45 67
+89 ab cd ef
+```
+
+### echo
+输出指定文字。
+示例：`/textfx echo 你好` → `你好`
+
+### cat
+读取并拼接缓存内容，类似 Unix cat 命令。
+- 无参数时直接传递标准输入（管道输入或回复的消息）。
+- 使用 `-` 代表标准输入，可与缓存名混合使用。
+- 支持多个参数，按顺序拼接输出。
+
+示例：
+- 传递输入：`/textfx echo 你好 | cat` → `你好`
+- 读取缓存：`/textfx cat mytext` → 输出 mytext 的内容
+- 拼接多个缓存：`/textfx cat a b c` → 依次拼接缓存 a、b、c
+- 混合标准输入和缓存：`/textfx echo 前缀 | cat - mytext` → 拼接标准输入与缓存 mytext
+
+### 缓存操作（保存中间结果）
+- 保存：`/textfx reverse 你好 > mytext`（不输出，存入 mytext）
+- 读取：`/textfx cat mytext` → `好你`
+- 追加：`/textfx echo world >> mytext`
+- 删除：`/textfx rm mytext`
+
+> 缓存仅在当前对话中有效，重启后清空。
+
+### replace（或 替换、sed）
+替换文字（支持正则表达式）。
+示例（普通）：`/textfx replace 世界 宇宙 你好世界` → `你好宇宙`
+示例（正则）：`/textfx replace \d+ [数字] 我有123个苹果` → `我有[数字]个苹果`
+
+### trim（或 strip、去空格）
+去除文本首尾空白字符。
+示例：`/textfx trim "  你好  "` → `你好`
+示例：`/textfx echo "  hello  " | trim` → `hello`
+
+### ltrim（或 lstrip）
+去除文本左侧空白字符。
+示例：`/textfx ltrim "  你好  "` → `你好  `
+
+### rtrim（或 rstrip）
+去除文本右侧空白字符。
+示例：`/textfx rtrim "  你好  "` → `  你好`
+
+### squeeze（或 压缩空白）
+将连续的空白字符（空格、制表符）压缩为单个空格。
+示例：`/textfx squeeze "你好   世界"` → `你好 世界`
+
+### lines（或 行处理）
+按行处理文本，支持以下子命令：
+- `lines trim` — 去除每行首尾空白
+- `lines empty` — 去除所有空行
+- `lines squeeze` — 将连续空行压缩为一行
+
+示例：`/textfx echo "  hello\n\n\n  world  " | lines trim` → `hello\n\n\n world`
+示例：`/textfx echo "a\n\n\nb" | lines squeeze` → `a\n\nb`
+
+---
+
+## 常见问题
+
+- **没反应？** 可能内容被安全系统拦截，机器人会提示“内容被拦截”。
+- **只支持纯文字**，暂不支持图片或文件。
+- 命令拼错时，机器人会提示“不存在名为 xxx 的函数”，请检查名称。
+
+快去试试吧！用法核心：**`/textfx` + 你的操作**

konabot/docs/user/tqszm.txt

@@ -0,0 +1,24 @@
+# tqszm
+
+引用一条消息，让此方帮你提取首字母。
+
+例子：
+
+```
+John: 11-28 16:50:37
+谁来总结一下今天的工作？
+
+Jack: 11-28 16:50:55
+[引用John的消息] @此方Bot tqszm
+
+此方Bot: 11-28 16:50:56
+slzjyxjtdgz？
+```
+
+或者，你也可以直接以正常指令的方式调用：
+
+```
+@此方Bot 提取首字母 中山大学
+> zsdx
+```
+

konabot/docs/user/typst.txt

@@ -0,0 +1,4 @@
+# Typst 渲染
+
+只需使用 `typst ...` 就可以渲染 Typst 了
+

konabot/docs/user/ytpgif.txt

@@ -1,41 +1,72 @@
-指令介绍
-    ytpgif - 生成来回镜像翻转的仿 YTPMV 动图
+# `ytpgif` 指令说明
 
-格式
-    ytpgif [倍速]
+## 功能简介
+`ytpgif` 用于生成来回镜像翻转的仿 YTPMV（YouTube Poop Music Video）风格动图。
 
-示例
-    `ytpgif`  
-        使用默认倍速（1.0）处理你发送或回复的图片，生成镜像动图。
+---
 
-    `ytpgif 2.5`  
-        以 2.5 倍速处理图片，生成更快节奏的镜像动图。
+## 命令格式
+```bash
+ytpgif [倍速]
+```
 
-    回复一张图片并发送 `ytpgif 0.5`  
-        以慢速（0.5 倍）生成镜像动图。
+---
 
-参数说明
-    倍速（可选）
-        - 类型：浮点数
-        - 默认值：1.0
-        - 有效范围：0.1 ～ 20.0
-        - 作用：
-            • 对于静态图：控制镜像切换的快慢（值越大，切换越快）。
-            • 对于动图：控制截取原始动图正向和反向片段的时长（值越大，截取的片段越长）。
+## 使用示例
 
-使用方式
-    发送指令前，请确保：
-      - 直接在消息中附带一张图片，或
-      - 回复一条包含图片的消息后再发送指令。
+- **默认倍速**
+  ```bash
+  ytpgif
+  ```
+  使用默认倍速（1.0）处理你发送或回复的图片，生成镜像动图。
 
-    插件会自动：
-      - 下载并识别图片（支持静态图和 GIF 动图）
-      - 自动缩放至最大边长不超过 256 像素（保持宽高比）
-      - 静态图 → 生成“原图↔镜像”循环动图
-      - 动图 → 截取开头一段正向播放 + 同一段镜像翻转播放，拼接成新动图
-      - 保留透明通道（如原图含透明），否则转为 RGB 避免颜色异常
+- **指定倍速（较快）**
+  ```bash
+  ytpgif 2.5
+  ```
+  以 2.5 倍速处理图片，生成节奏更快的镜像动图。
 
-注意事项
-    - 图片过大、格式损坏或网络问题可能导致处理失败。
-    - 动图帧数过多或单帧过短可能无法生成有效输出。
-    - 输出 GIF 最大单段帧数限制为 500 帧，以防资源耗尽。
+- **指定倍速（较慢）**
+  回复一张图片并发送：
+  ```bash
+  ytpgif 0.5
+  ```
+  以 0.5 倍速生成慢节奏的镜像动图。
+
+---
+
+## 参数说明
+
+### `倍速`（可选）
+- **类型**：浮点数
+- **默认值**：`1.0`
+- **有效范围**：`0.1 ～ 20.0`
+
+#### 作用：
+- **静态图**：控制“原图 ↔ 镜像”切换的速度（值越大，切换越快）。
+- **GIF 动图**：控制截取原始动图正向与反向片段的时长（值越大，截取的片段越长）。
+
+---
+
+## 使用方式
+
+在发送指令前，请确保满足以下任一条件：
+- 在消息中**直接附带一张图片**，或
+- **回复一条包含图片的消息**后再发送指令。
+
+插件将自动执行以下操作：
+1. 下载并识别图片（支持静态图和 GIF 动图）。
+2. 自动缩放图像，**最大边长不超过 256 像素**（保持宽高比）。
+3. 根据图片类型处理：
+   - **静态图** → 生成“原图 ↔ 镜像”循环动图。
+   - **GIF 动图** → 截取开头一段正向播放 + 同一段镜像翻转播放，拼接成新动图。
+4. **保留透明通道**（若原图含透明），否则转为 RGB 模式以避免颜色异常。
+
+---
+
+## 注意事项
+
+⚠️ 以下情况可能导致处理失败或效果不佳：
+- 图片过大、格式损坏或网络问题；
+- 动图帧数过多或单帧持续时间过短；
+- 输出 GIF 单段帧数超过 **500 帧**（系统将自动限制以防资源耗尽）。

konabot/docs/user/删除提醒.txt

@@ -1,8 +1,9 @@
-指令介绍
-    删除提醒 - 删除在`查询提醒(1)`中查到的提醒
+## 指令介绍
+**删除提醒** - 删除在 [`查询提醒(1)`](查询提醒(1)) 中查到的提醒
 
-指令示例
-    `删除提醒 1` 在查询提醒后，删除编号为 1 的提醒
+## 指令示例
+`删除提醒 1`
+在查询提醒后，删除编号为 1 的提醒
 
-另见
-    提醒我(1) 查询提醒(1) ntfy(1)
+## 另见
+[`提醒我(1)`](提醒我(1)) [`查询提醒(1)`](查询提醒(1)) [`ntfy(1)`](ntfy(1))

konabot/docs/user/卵总展示.txt

@@ -1,20 +1,24 @@
-指令介绍
-    卵总展示 - 让卵总举起你的图片
+# 指令介绍
 
-格式
-    <引用图片> 卵总展示 [选项]
-    卵总展示 [选项] <图片>
+**卵总展示** - 让卵总举起你的图片
 
-选项
-    `--whiteness <number>` 白度
-        将原图进行指数变换，以调整它的白的程度，默认为 0.0
+## 格式
 
-    `--black-level <number>` 黑色等级
-        将原图减淡，数值越大越淡，范围 0.0-1.0，默认 0.2
+```
+<引用图片> 卵总展示 [选项]
+卵总展示 [选项] <图片>
+```
 
-    `--opacity <number>` 不透明度
-        将你的图片叠放在图片上的不透明度，默认为 0.8
+## 选项
 
-    `--saturation <number>` 饱和度
-        调整原图的饱和度，应该要大于 0.0，默认为 0.85
+- `--whiteness <number>` **白度**
+  将原图进行指数变换，以调整它的白的程度，默认为 `0.0`。
 
+- `--black-level <number>` **黑色等级**
+  将原图减淡，数值越大越淡，范围 `0.0–1.0`，默认为 `0.2`。
+
+- `--opacity <number>` **不透明度**
+  将你的图片叠放在图片上的不透明度，默认为 `0.8`。
+
+- `--saturation <number>` **饱和度**
+  调整原图的饱和度，应大于 `0.0`，默认为 `0.85`。

konabot/docs/user/发起投票.txt

@@ -1,11 +1,16 @@
-指令介绍
-    发起投票 - 发起一个投票
+### 指令介绍
+**发起投票** - 发起一个投票
 
-格式
-    发起投票 <投票标题> <选项1> <选项2> ...
+### 格式
+```
+发起投票 <投票标题> <选项1> <选项2> ...
+```
 
-示例
-    `发起投票 这是一个投票 A B C` 发起标题为“这是一个投票”，选项为“A”、“B”、“C”的投票
+### 示例
+`发起投票 这是一个投票 A B C`
+发起标题为“这是一个投票”，选项为“A”、“B”、“C”的投票。
 
-说明
-    投票各个选项之间用空格分隔，选项数量为2-15项。投票的默认有效期为24小时。
+### 说明
+- 投票的各个选项之间用空格分隔。
+- 选项数量必须为 **2 到 15 项**。
+- 投票的默认有效期为 **24 小时**。

konabot/docs/user/喵.txt

@@ -1,2 +1,3 @@
-指令介绍
-    喵 - 你发喵，此方就会回复喵
+# 指令介绍
+
+喵 - 你发喵，此方就会回复喵

konabot/docs/user/投票.txt

@@ -1,12 +1,16 @@
-指令介绍
-    投票 - 参与已发起的投票
+## 指令介绍
+**投票** - 参与已发起的投票
 
-格式
-    投票 <投票ID/标题> <选项文本>
+## 格式
+```
+投票 <投票ID/标题> <选项文本>
+```
 
-示例
-    `投票 1 A` 在ID为1的投票中，投给“A”
-    `投票 这是一个投票 B` 在标题为“这是一个投票”的投票中，投给“B”
+## 示例
+- `投票 1 A`
+  在 ID 为 1 的投票中，投给 “A”
+- `投票 这是一个投票 B`
+  在标题为 “这是一个投票” 的投票中，投给 “B”
 
-说明
-    目前不支持单人多投，每个人只能投一项。
+## 说明
+目前不支持单人多投，每个人只能投一项。

konabot/docs/user/提醒我.txt

@@ -1,15 +1,18 @@
-指令介绍
-    提醒我 - 在指定的时间提醒人事项的工具
+## 指令介绍
 
-使用示例
-    `下午五点提醒我吃饭`
-        创建一个下午五点的提醒，提醒你吃饭
+**提醒我** - 在指定的时间提醒人事项的工具
 
-    `两分钟后提醒我睡觉`
-        创建一个相对于现在推迟 2 分钟的提醒，提醒你睡觉
+## 使用示例
 
-    `2026年4月25日20点整提醒我生日快乐`
-        创建一个指定日期和时间的提醒
+- `下午五点提醒我吃饭`
+  创建一个下午五点的提醒，提醒你吃饭
 
-另见
-    查询提醒(1) 删除提醒(1) ntfy(1)
+- `两分钟后提醒我睡觉`
+  创建一个相对于现在推迟 2 分钟的提醒，提醒你睡觉
+
+- `2026年4月25日20点整提醒我生日快乐`
+  创建一个指定日期和时间的提醒
+
+## 另见
+
+[`查询提醒(1)`](查询提醒) [`删除提醒(1)`](删除提醒) [`ntfy(1)`](ntfy)

konabot/docs/user/摇数字.txt

@@ -1,7 +1,13 @@
-指令介绍
-    摇数字 - 生成一个随机数字并发送
+## 指令介绍
 
-示例
-    `摇数字` 随机生成一个 1-6 的数字
+**摇数字** - 生成一个随机数字并发送
 
-该指令不接受任何参数，直接调用即可。
+### 示例
+
+```
+摇数字
+```
+
+随机生成一个 1-6 的数字。
+
+> 该指令不接受任何参数，直接调用即可。

konabot/docs/user/摇骰子.txt

@@ -1,22 +1,33 @@
-指令介绍
-    摇骰子 - 用于生成随机数并以骰子图像形式展示的指令
+# 指令介绍
 
-格式
-    摇骰子 [最小值] [最大值]
+**摇骰子** - 用于生成随机数并以骰子图像形式展示的指令
 
-示例
-    `摇骰子` 随机生成一个 1-6 的数字，并显示对应的骰子图像
-    `摇骰子 10` 生成 1 到 10 之间的随机整数
-    `摇骰子 0.5` 生成 0 到 0.5 之间的随机小数
-    `摇骰子 -5 5` 生成 -5 到 5 之间的随机数
+## 格式
 
-说明
-    该指令支持以下几种调用方式：
-      - 不带参数：使用默认范围生成随机数
-      - 仅指定一个参数 f1：
-          - 若 f1 > 1，则生成 [1, f1] 范围内的随机数
-          - 若 0 < f1 ≤ 1，则生成 [0, f1] 范围内的随机数
-          - 若 f1 ≤ 0，则生成 [f1, 0] 范围内的随机数
-      - 指定两个参数 f1 和 f2：生成 [f1, f2] 范围内的随机数（顺序无关，内部会自动处理大小）
+```
+摇骰子 [最小值] [最大值]
+```
+
+## 示例
+
+- `摇骰子`
+  随机生成一个 1–6 的数字，并显示对应的骰子图像
+- `摇骰子 10`
+  生成 1 到 10 之间的随机整数
+- `摇骰子 0.5`
+  生成 0 到 0.5 之间的随机小数
+- `摇骰子 -5 5`
+  生成 -5 到 5 之间的随机数
+
+## 说明
+
+该指令支持以下几种调用方式：
+
+- **不带参数**：使用默认范围（1–6）生成随机数
+- **仅指定一个参数 `f1`**：
+  - 若 `f1 > 1`，则生成 `[1, f1]` 范围内的随机数
+  - 若 `0 < f1 ≤ 1`，则生成 `[0, f1]` 范围内的随机数
+  - 若 `f1 ≤ 0`，则生成 `[f1, 0]` 范围内的随机数
+- **指定两个参数 `f1` 和 `f2`**：生成 `[f1, f2]` 范围内的随机数（顺序无关，内部会自动处理大小）
 
 返回结果将以骰子样式的图像形式展示生成的随机数值。

konabot/docs/user/查看投票.txt

@@ -1,12 +1,22 @@
-指令介绍
-    查看投票 - 查看已发起的投票
+# 指令介绍
 
-格式
-    查看投票 <投票ID或标题>
+**查看投票** - 查看已发起的投票
 
-示例
-    `查看投票 1` 查看ID为1的投票
-    `查看投票 这是一个投票` 查看标题为“这是一个投票”的投票
+## 格式
 
-说明
-    投票在进行时，使用此命令可以看到投票的各个选项；投票结束后，则可以看到各项的票数。
+```
+查看投票 <投票ID或标题>
+```
+
+## 示例
+
+- `查看投票 1`
+  查看 ID 为 1 的投票
+
+- `查看投票 这是一个投票`
+  查看标题为“这是一个投票”的投票
+
+## 说明
+
+- 投票进行中时，使用此命令可查看投票的各个选项；
+- 投票结束后，可查看各选项的最终票数。

konabot/docs/user/查询提醒.txt

@@ -1,9 +1,9 @@
-指令介绍
-    查询提醒 - 查询已经创建的提醒
+# 指令介绍
+**查询提醒** - 查询已经创建的提醒
 
-指令格式
-    `查询提醒` 查询提醒
-    `查询提醒 2` 查询第二页提醒
+## 指令格式
+- `查询提醒`：查询提醒
+- `查询提醒 2`：查询第二页提醒
 
-另见
-    提醒我(1) 删除提醒(1) ntfy(1)
+## 另见
+[提醒我(1)]() [删除提醒(1)]() [ntfy(1)]()

konabot/docs/user/生成二维码.txt

@@ -1,8 +1,17 @@
-指令介绍
-    生成二维码 - 将文本内容转换为二维码
+## 指令介绍
 
-格式
-    生成二维码 <文本内容>
+**生成二维码** - 将文本内容转换为二维码
 
-示例
-    `生成二维码 嗨嗨嗨` 生成扫描结果为“嗨嗨嗨”的二维码图片
+### 格式
+
+```
+生成二维码 <文本内容>
+```
+
+### 示例
+
+```
+生成二维码 嗨嗨嗨
+```
+
+生成扫描结果为“嗨嗨嗨”的二维码图片

konabot/docs/user/订阅.txt

@@ -0,0 +1,30 @@
+# 指令介绍
+
+**订阅** - 收听此方 BOT 的自动消息发送。
+
+---
+
+## 格式
+
+- `订阅 <频道名称>`
+- `取消订阅 <频道名称>`
+- `查询订阅 [页码]`
+- `可用订阅 [页码]`
+
+---
+
+## 示例
+
+- **`订阅 此方谜题`**
+  在当前的聊天上下文订阅「此方谜题」频道。此后会每天推送此方谜题（由 konaph(8) 管理的）。
+  - 如果你是私聊，则能够每天发送此方谜题到你的私聊；
+  - 如果在群聊中使用该指令，则会每天发送题目到这个群里面。
+
+- **`取消订阅 此方谜题`**
+  取消订阅「此方谜题」频道。
+
+- **`查询订阅`**
+  查询当前聊天上下文订阅的所有频道。
+
+- **`可用订阅 2`**
+  查询所有可用的订阅的第二页。

konabot/docs/user/雷达回波.txt

@@ -1,13 +1,20 @@
-指令介绍
-    雷达回波 - 用于获取指定地区的天气雷达回波图像
+# 指令介绍
 
-格式
-    雷达回波 <地区>
+**雷达回波** - 用于获取指定地区的天气雷达回波图像。
 
-示例
-    `雷达回波 华南` 获取华南地区的天气雷达回波图
-    `雷达回波 全国` 获取全国的天气雷达回波图
+## 格式
 
-说明
-    该指令通过查询中国气象局 https://www.nmc.cn/publish/radar/chinaall.html ，获取指定地区的实时天气雷达回波图像。
-    支持的地区有：全国 华北 东北 华东 华中 华南 西南 西北。
+```
+雷达回波 <地区>
+```
+
+## 示例
+
+- `雷达回波 华南`：获取华南地区的天气雷达回波图
+- `雷达回波 全国`：获取全国的天气雷达回波图
+
+## 说明
+
+该指令通过查询中国气象局 [https://www.nmc.cn/publish/radar/chinaall.html](https://www.nmc.cn/publish/radar/chinaall.html)，获取指定地区的实时天气雷达回波图像。
+
+支持的地区有：**全国**、**华北**、**东北**、**华东**、**华中**、**华南**、**西南**、**西北**。

konabot/docs/user/黑白.txt

@@ -1,5 +0,0 @@
-指令介绍
-    黑白 - 将图片经过一个黑白滤镜的处理
-
-示例
-    引用一个带有图片的消息，或者消息本身携带图片，然后发送「黑白」即可

konabot/plugins/ai_extract_text/__init__.py

@@ -0,0 +1,52 @@
+from io import BytesIO
+import base64
+import re
+from loguru import logger
+from nonebot import on_message
+from nonebot.rule import Rule
+
+from konabot.common.apis.ali_content_safety import AlibabaGreen
+from konabot.common.llm import get_llm
+from konabot.common.longtask import DepLongTaskTarget
+from konabot.common.nb.extract_image import DepPILImage
+from konabot.common.nb.match_keyword import match_keyword
+
+
+cmd = on_message(rule=Rule(match_keyword(re.compile(r"^千问识图\s*$"))))
+
+
+@cmd.handle()
+async def _(img: DepPILImage, target: DepLongTaskTarget):
+    if 1:
+        return  #TODO:这里还没写完，还有 Bug 要修
+    jpeg_data = BytesIO()
+    if img.width > 2160:
+        img = img.resize((2160, img.height * 2160 // img.width))
+    if img.height > 2160:
+        img = img.resize((img.width * 2160 // img.height, 2160))
+    img = img.convert("RGB")
+    img.save(jpeg_data, format="jpeg", optimize=True, quality=85)
+    data_url = "data:image/jpeg;base64,"
+    data_url += base64.b64encode(jpeg_data.getvalue()).decode('ascii')
+
+    llm = get_llm("qwen3-vl-plus")
+    res = await llm.chat([
+        { "role": "user", "content": [
+            { "type": "image_url", "image_url": {
+                "url": data_url
+            } },
+            { "type": "text", "text": "请你提取这张图片中的所有文字，并尽量按照原图的排版输出，不需要其他内容" },
+        ] }
+    ])
+    result = res.content
+    logger.info(res)
+    if result is None:
+        await target.send_message("提取失败：可能存在网络异常")
+        return
+
+    if not await AlibabaGreen.detect(result):
+        await target.send_message("提取失败：图片中可能存在一些不合适的内容")
+        return
+
+    await target.send_message(result, at=False)
+

konabot/plugins/air_conditioner/__init__.py

@@ -1,22 +1,29 @@
 from io import BytesIO
 from typing import Optional, Union
 import cv2
+import nonebot
 from nonebot.adapters import Event as BaseEvent
 from nonebot.adapters.console.event import MessageEvent as ConsoleMessageEvent
 from nonebot.adapters.discord.event import MessageEvent as DiscordMessageEvent
 from nonebot_plugin_alconna import Alconna, AlconnaMatcher, Args, UniMessage, on_alconna
 from PIL import Image
 import numpy as np
+from konabot.common.database import DatabaseManager
 from konabot.common.longtask import DepLongTaskTarget
 from konabot.common.path import ASSETS_PATH
 from konabot.common.web_render import WebRenderer
 from konabot.plugins.air_conditioner.ac import AirConditioner, CrashType, generate_ac_image, wiggle_transform
-
+from pathlib import Path
 import random
 import math
 
-def get_ac(id: str) -> AirConditioner:
-    ac = AirConditioner.air_conditioners.get(id)
+ROOT_PATH = Path(__file__).resolve().parent
+
+# 创建全局数据库管理器实例
+db_manager = DatabaseManager()
+
+async def get_ac(id: str) -> AirConditioner:
+    ac = await AirConditioner.get_ac(id)
     if ac is None:
         ac = AirConditioner(id)
     return ac
@@ -43,14 +50,32 @@ async def send_ac_image(event: type[AlconnaMatcher], ac: AirConditioner):
     ac_image = await generate_ac_image(ac)
     await event.send(await UniMessage().image(raw=ac_image).export())
 
+
+driver = nonebot.get_driver()
+
+
+@driver.on_startup
+async def register_startup_hook():
+    """注册启动时需要执行的函数"""
+    # 初始化数据库表
+    await db_manager.execute_by_sql_file(
+        Path(__file__).resolve().parent / "sql" / "create_table.sql"
+    )
+
+@driver.on_shutdown
+async def register_shutdown_hook():
+    """注册关闭时需要执行的函数"""
+    # 关闭所有数据库连接
+    await db_manager.close_all_connections()
+
 evt = on_alconna(Alconna(
     "群空调"
 ), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
 
 @evt.handle()
-async def _(event: BaseEvent, target: DepLongTaskTarget):
+async def _(target: DepLongTaskTarget):
     id = target.channel_id
-    ac = get_ac(id)
+    ac = await get_ac(id)
     await send_ac_image(evt, ac)
 
 evt = on_alconna(Alconna(
@@ -58,10 +83,10 @@ evt = on_alconna(Alconna(
 ), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
 
 @evt.handle()
-async def _(event: BaseEvent, target: DepLongTaskTarget):
+async def _(target: DepLongTaskTarget):
     id = target.channel_id
-    ac = get_ac(id)
-    ac.on = True
+    ac = await get_ac(id)
+    await ac.update_ac(state=True)
     await send_ac_image(evt, ac)
 
 evt = on_alconna(Alconna(
@@ -69,10 +94,10 @@ evt = on_alconna(Alconna(
 ), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
 
 @evt.handle()
-async def _(event: BaseEvent, target: DepLongTaskTarget):
+async def _(target: DepLongTaskTarget):
     id = target.channel_id
-    ac = get_ac(id)
-    ac.on = False
+    ac = await get_ac(id)
+    await ac.update_ac(state=False)
     await send_ac_image(evt, ac)
 
 evt = on_alconna(Alconna(
@@ -81,31 +106,29 @@ evt = on_alconna(Alconna(
 ), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
 
 @evt.handle()
-async def _(event: BaseEvent, target: DepLongTaskTarget, temp: Optional[Union[int, float]] = 1):
+async def _(target: DepLongTaskTarget, temp: Optional[Union[int, float]] = 1):
+    if temp is None:
+        temp = 1
     if temp <= 0:
         return
     id = target.channel_id
-    ac = get_ac(id)
+    ac = await get_ac(id)
     if not ac.on or ac.burnt == True or ac.frozen == True:
         await send_ac_image(evt, ac)
         return
-    ac.temperature += temp
-    if ac.temperature > 40:
-        # 根据温度随机出是否爆炸，40度开始，呈指数增长
-        possibility = -math.e ** ((40-ac.temperature) / 50) + 1
-        if random.random() < possibility:
-            # 打开爆炸图片
-            with open(ASSETS_PATH / "img" / "other" / "boom.jpg", "rb") as f:
-                output = BytesIO()
-                # 爆炸抖动
-                frames = wiggle_transform(np.array(Image.open(f)), intensity=5)
-                pil_frames = [Image.fromarray(frame) for frame in frames]
-                pil_frames[0].save(output, format="GIF", save_all=True, append_images=pil_frames[1:], loop=0, duration=35, disposal=2)
-                output.seek(0)
-                await evt.send(await UniMessage().image(raw=output).export())
-            ac.broke_ac(CrashType.BURNT)
-            await evt.send("太热啦，空调炸了！")
-            return
+    await ac.update_ac(temperature_delta=temp)
+    if ac.burnt:
+        # 打开爆炸图片
+        with open(ASSETS_PATH / "img" / "other" / "boom.jpg", "rb") as f:
+            output = BytesIO()
+            # 爆炸抖动
+            frames = wiggle_transform(np.array(Image.open(f)), intensity=5)
+            pil_frames = [Image.fromarray(frame) for frame in frames]
+            pil_frames[0].save(output, format="GIF", save_all=True, append_images=pil_frames[1:], loop=0, duration=35, disposal=2)
+            output.seek(0)
+            await evt.send(await UniMessage().image(raw=output).export())
+        await evt.send("太热啦，空调炸了！")
+        return
     await send_ac_image(evt, ac)
 
 evt = on_alconna(Alconna(
@@ -114,20 +137,17 @@ evt = on_alconna(Alconna(
 ), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
 
 @evt.handle()
-async def _(event: BaseEvent, target: DepLongTaskTarget, temp: Optional[Union[int, float]] = 1):
+async def _(target: DepLongTaskTarget, temp: Optional[Union[int, float]] = 1):
+    if temp is None:
+        temp = 1
     if temp <= 0:
         return
     id = target.channel_id
-    ac = get_ac(id)
+    ac = await get_ac(id)
     if not ac.on or ac.burnt == True or ac.frozen == True:
         await send_ac_image(evt, ac)
         return
-    ac.temperature -= temp
-    if ac.temperature < 0:
-        # 根据温度随机出是否冻结，0度开始，呈指数增长
-        possibility = -math.e ** (ac.temperature / 50) + 1
-        if random.random() < possibility:
-            ac.broke_ac(CrashType.FROZEN)
+    await ac.update_ac(temperature_delta=-temp)
     await send_ac_image(evt, ac)
 
 evt = on_alconna(Alconna(
@@ -135,21 +155,34 @@ evt = on_alconna(Alconna(
 ), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
 
 @evt.handle()
-async def _(event: BaseEvent, target: DepLongTaskTarget):
+async def _(target: DepLongTaskTarget):
     id = target.channel_id
-    ac = get_ac(id)
-    ac.change_ac()
+    ac = await get_ac(id)
+    await ac.change_ac()
     await send_ac_image(evt, ac)
 
+async def query_number_ranking(id: str) -> tuple[int, int]:
+    result = await db_manager.query_by_sql_file(
+        ROOT_PATH / "sql" / "query_crash_and_rank.sql",
+        (id,id)
+    )
+    if len(result) == 0:
+        return 0, 0
+    else:
+        # 将字典转换为值的元组
+        values = list(result[0].values())
+        return values[0], values[1]
+
 evt = on_alconna(Alconna(
     "空调炸炸排行榜",
 ), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
 
 @evt.handle()
-async def _(event: BaseEvent, target: DepLongTaskTarget):
+async def _(target: DepLongTaskTarget):
     id = target.channel_id
-    ac = get_ac(id)
-    number, ranking = ac.get_crashes_and_ranking()
+    # ac = get_ac(id)
+    # number, ranking = ac.get_crashes_and_ranking()
+    number, ranking = await query_number_ranking(id)
     params = {
         "number": number,
         "ranking": ranking
@@ -159,4 +192,37 @@ async def _(event: BaseEvent, target: DepLongTaskTarget):
         target=".box",
         params=params
     )
-    await evt.send(await UniMessage().image(raw=image).export())
+    await evt.send(await UniMessage().image(raw=image).export())
+
+evt = on_alconna(Alconna(
+    "空调最高峰",
+), use_cmd_start=True, use_cmd_sep=False, skip_for_unmatch=True)
+
+@evt.handle()
+async def _(target: DepLongTaskTarget):
+    result = await db_manager.query_by_sql_file(
+        ROOT_PATH / "sql" / "query_peak.sql"
+    )
+    if len(result) == 0:
+        await evt.send("没有空调记录！")
+        return
+    max_temp = result[0].get("max")
+    min_temp = result[0].get("min")
+    his_max = result[0].get("his_max")
+    his_min = result[0].get("his_min")
+    # 再从内存里的空调池中获取最高温度和最低温度
+    for ac in AirConditioner.InstancesPool.values():
+        if ac.on and not ac.burnt and not ac.frozen:
+            if max_temp is None or min_temp is None:
+                max_temp = ac.temperature
+                min_temp = ac.temperature
+            max_temp = max(max_temp, ac.temperature)
+            min_temp = min(min_temp, ac.temperature)
+    if max_temp is None or min_temp is None:
+        await evt.send(f"目前全部空调都被炸掉了！")
+    else:
+        await evt.send(f"全球在线空调最高温度为 {'%.1f' % max_temp}°C，最低温度为 {'%.1f' % min_temp}°C！")
+    if his_max is None or his_min is None:
+        pass
+    else:
+        await evt.send(f"历史最高温度为 {'%.1f' % his_max}°C，最低温度为 {'%.1f' % his_min}°C！\n（要进入历史记录，温度需至少保持 5 分钟）")

konabot/plugins/air_conditioner/ac.py

@@ -1,20 +1,193 @@
+import asyncio
 from enum import Enum
 from io import BytesIO
+import math
+from pathlib import Path
+import random
+import signal
+import time
 
 import cv2
 import numpy as np
 from PIL import Image, ImageDraw, ImageFont
 
+from nonebot import logger
+
+from konabot.common.database import DatabaseManager
 from konabot.common.path import ASSETS_PATH, FONTS_PATH
 from konabot.common.path import DATA_PATH
+import nonebot
 import json
 
+ROOT_PATH = Path(__file__).resolve().parent
+
+# 创建全局数据库管理器实例
+db_manager = DatabaseManager()
+
 class CrashType(Enum):
     BURNT = 0
     FROZEN = 1
 
+driver = nonebot.get_driver()
+
+@driver.on_startup
+async def register_startup_hook():
+    await ac_manager.start_auto_save()
+
+@driver.on_shutdown
+async def register_shutdown_hook():
+    """注册关闭时需要执行的函数"""
+    # 停止自动保存任务
+    if ac_manager:
+        await ac_manager.stop_auto_save()
+
+class AirConditionerManager:
+    def __init__(self, save_interval: int = 300):  # 默认5分钟保存一次
+        self.save_interval = save_interval
+        self._save_task = None
+        self._running = False
+        
+    async def start_auto_save(self):
+        """启动自动保存任务"""
+        self._running = True
+        self._save_task = asyncio.create_task(self._auto_save_loop())
+            
+        logger.info(f"自动保存任务已启动，间隔: {self.save_interval}秒")
+    
+    async def stop_auto_save(self):
+        """停止自动保存任务"""
+        if self._save_task:
+            self._running = False
+            self._save_task.cancel()
+            try:
+                await self._save_task
+            except asyncio.CancelledError:
+                pass
+            logger.info("自动保存任务已停止")
+        else:
+            logger.warning("没有正在运行的自动保存任务")
+        
+    async def _auto_save_loop(self):
+        """自动保存循环"""
+        while self._running:
+            try:
+                await asyncio.sleep(self.save_interval)
+                await self.save_all_instances()
+            except asyncio.CancelledError:
+                break
+            except Exception as e:
+                logger.error(f"定时保存失败: {e}")
+                
+    async def save_all_instances(self):
+        save_time = time.time()
+        to_remove = []
+        """保存所有实例到数据库"""
+        for ac_id, ac_instance in AirConditioner.InstancesPool.items():
+            try:
+                await db_manager.execute_by_sql_file(
+                    ROOT_PATH / "sql" / "update_ac.sql",
+                    [(ac_instance.on, ac_instance.temperature, 
+                     ac_instance.burnt, ac_instance.frozen, ac_id),(ac_id,)]
+                )
+                if(save_time - ac_instance.instance_get_time >= 300): # 5 分钟
+                    to_remove.append(ac_id)
+            except Exception as e:
+                logger.error(f"保存空调 {ac_id} 失败: {e}")
+
+        logger.info(f"定时保存完成，共保存 {len(AirConditioner.InstancesPool)} 个空调实例")
+
+        # 删除时间过长实例
+        for ac_id in to_remove:
+            del AirConditioner.InstancesPool[ac_id]
+
+        logger.info(f"清理长期不活跃的空调实例完成，目前池内共有 {len(AirConditioner.InstancesPool)} 个实例")
+
+ac_manager = AirConditionerManager(save_interval=300) # 5分钟
+
 class AirConditioner:
-    air_conditioners: dict[str, "AirConditioner"] = {}
+    InstancesPool: dict[str, 'AirConditioner'] = {}
+
+    @classmethod
+    async def refresh_ac(cls, id: str):
+        cls.InstancesPool[id].instance_get_time = time.time()
+
+    @classmethod
+    async def storage_ac(cls, id: str, ac: 'AirConditioner'):
+        cls.InstancesPool[id] = ac
+
+    @classmethod
+    async def get_ac(cls, id: str) -> 'AirConditioner':
+        if(id in cls.InstancesPool):
+            await cls.refresh_ac(id)
+            return cls.InstancesPool[id]
+        # 如果没有，那么从数据库重新实例化一个 AC 出来
+        result = await db_manager.query_by_sql_file(ROOT_PATH / "sql" / "query_ac.sql", (id,))
+        if len(result) == 0:
+            ac = await cls.create_ac(id)
+            return ac
+        ac_data = result[0]
+        ac = AirConditioner(id)
+        ac.on = bool(ac_data["on"])
+        ac.temperature = float(ac_data["temperature"])
+        ac.burnt = bool(ac_data["burnt"])
+        ac.frozen = bool(ac_data["frozen"])
+        await cls.storage_ac(id, ac)
+        return ac
+
+    @classmethod
+    async def create_ac(cls, id: str) -> 'AirConditioner':
+        ac = AirConditioner(id)
+        await db_manager.execute_by_sql_file(
+            ROOT_PATH / "sql" / "insert_ac.sql",
+            [(id, ac.on, ac.temperature, ac.burnt, ac.frozen),(id,)]
+        )
+        await cls.storage_ac(id, ac)
+        return ac
+    
+    async def change_ac_temp(self, temperature_delta: float) -> None:
+        '''
+        改变空调的温度
+        :param temperature_delta: float 温度变化量
+        '''
+        changed_temp = self.temperature + temperature_delta
+        random_poss = random.random()
+        if temperature_delta < 0 and changed_temp < 0:
+            # 根据温度随机出是否冻结，0度开始，呈指数增长
+            possibility = -math.e ** (changed_temp / 50) + 1
+            if random_poss < possibility:
+                await self.broke_ac(CrashType.FROZEN)
+        elif temperature_delta > 0 and changed_temp > 40:
+            # 根据温度随机出是否烧坏，40度开始，呈指数增长
+            possibility = -math.e ** ((40-changed_temp) / 50) + 1
+            if random_poss < possibility:
+                await self.broke_ac(CrashType.BURNT)
+        self.temperature = changed_temp
+
+    async def update_ac(self, state: bool = None, temperature_delta: float = None, burnt: bool = None, frozen: bool = None) -> 'AirConditioner':
+        if state is not None:
+            self.on = state
+        if temperature_delta is not None:
+            await self.change_ac_temp(temperature_delta)
+        if burnt is not None:
+            self.burnt = burnt
+        if frozen is not None:
+            self.frozen = frozen
+        # await db_manager.execute_by_sql_file(
+        #     ROOT_PATH / "sql" / "update_ac.sql",
+        #     (self.on, self.temperature, self.burnt, self.frozen, self.id)
+        # )
+        return self
+
+    async def change_ac(self) -> 'AirConditioner':
+        self.on = False
+        self.temperature = 24
+        self.burnt = False
+        self.frozen = False
+        # await db_manager.execute_by_sql_file(
+        #     ROOT_PATH / "sql" / "update_ac.sql",
+        #     (self.on, self.temperature, self.burnt, self.frozen, self.id)
+        # )
+        return self
 
     def __init__(self, id: str) -> None:
         self.id = id
@@ -22,45 +195,42 @@ class AirConditioner:
         self.temperature = 24  # 默认温度
         self.burnt = False
         self.frozen = False
-        AirConditioner.air_conditioners[id] = self
 
-    def change_ac(self):
-        self.burnt = False
-        self.frozen = False
-        self.on = False
-        self.temperature = 24  # 重置为默认温度
+        self.instance_get_time = time.time()
 
-    def broke_ac(self, crash_type: CrashType):
+    async def broke_ac(self, crash_type: CrashType):
         '''
-        让空调坏掉，并保存数据
-
+        让空调坏掉
         :param crash_type: CrashType 枚举，表示空调坏掉的类型
         '''
         match crash_type:
             case CrashType.BURNT:
-                self.burnt = True
+                await self.update_ac(burnt=True)
             case CrashType.FROZEN:
-                self.frozen = True
-        self.save_crash_data(crash_type)
+                await self.update_ac(frozen=True)
+        await db_manager.execute_by_sql_file(
+            ROOT_PATH / "sql" / "insert_crash.sql",
+            (self.id, crash_type.value)
+        )
 
-    def save_crash_data(self, crash_type: CrashType):
-        '''
-        如果空调爆炸了，就往本地的 ac_crash_data.json 里该 id 的记录加一
-        '''
-        data_file = DATA_PATH / "ac_crash_data.json"
-        crash_data = {}
-        if data_file.exists():
-            with open(data_file, "r", encoding="utf-8") as f:
-                crash_data = json.load(f)
-        if self.id not in crash_data:
-            crash_data[self.id] = {"burnt": 0, "frozen": 0}
-        match crash_type:
-            case CrashType.BURNT:
-                crash_data[self.id]["burnt"] += 1
-            case CrashType.FROZEN:
-                crash_data[self.id]["frozen"] += 1
-        with open(data_file, "w", encoding="utf-8") as f:
-            json.dump(crash_data, f, ensure_ascii=False, indent=4)
+    # def save_crash_data(self, crash_type: CrashType):
+    #     '''
+    #     如果空调爆炸了，就往本地的 ac_crash_data.json 里该 id 的记录加一
+    #     '''
+    #     data_file = DATA_PATH / "ac_crash_data.json"
+    #     crash_data = {}
+    #     if data_file.exists():
+    #         with open(data_file, "r", encoding="utf-8") as f:
+    #             crash_data = json.load(f)
+    #     if self.id not in crash_data:
+    #         crash_data[self.id] = {"burnt": 0, "frozen": 0}
+    #     match crash_type:
+    #         case CrashType.BURNT:
+    #             crash_data[self.id]["burnt"] += 1
+    #         case CrashType.FROZEN:
+    #             crash_data[self.id]["frozen"] += 1
+    #     with open(data_file, "w", encoding="utf-8") as f:
+    #         json.dump(crash_data, f, ensure_ascii=False, indent=4)
 
     def get_crashes_and_ranking(self) -> tuple[int, int]:
         '''

konabot/plugins/air_conditioner/sql/create_table.sql

@@ -0,0 +1,26 @@
+-- 创建所有表
+CREATE TABLE IF NOT EXISTS air_conditioner (
+    id VARCHAR(128) PRIMARY KEY, 
+    "on" BOOLEAN NOT NULL, 
+    temperature REAL NOT NULL, 
+    burnt BOOLEAN NOT NULL, 
+    frozen BOOLEAN NOT NULL
+);
+
+CREATE TABLE IF NOT EXISTS air_conditioner_log (
+    log_id INTEGER PRIMARY KEY AUTOINCREMENT,
+    log_time DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    id VARCHAR(128), 
+    "on" BOOLEAN NOT NULL, 
+    temperature REAL NOT NULL, 
+    burnt BOOLEAN NOT NULL, 
+    frozen BOOLEAN NOT NULL,
+    FOREIGN KEY (id) REFERENCES air_conditioner(id)
+);
+
+CREATE TABLE IF NOT EXISTS air_conditioner_crash_log (
+    id VARCHAR(128) NOT NULL,
+    crash_type INT NOT NULL,
+    timestamp DATETIME NOT NULL,
+    FOREIGN KEY (id) REFERENCES air_conditioner(id)
+);

konabot/plugins/air_conditioner/sql/insert_ac.sql

@@ -0,0 +1,8 @@
+-- 插入一台新空调
+INSERT INTO air_conditioner (id, "on", temperature, burnt, frozen)
+VALUES (?, ?, ?, ?, ?);
+-- 使用返回的数据插入日志
+INSERT INTO air_conditioner_log (id, "on", temperature, burnt, frozen)
+SELECT id, "on", temperature, burnt, frozen
+FROM air_conditioner
+WHERE id = ?;

konabot/plugins/air_conditioner/sql/insert_crash.sql

@@ -0,0 +1,3 @@
+-- 插入一条空调爆炸记录
+INSERT INTO air_conditioner_crash_log (id, crash_type, timestamp)
+VALUES (?, ?, CURRENT_TIMESTAMP);

konabot/plugins/air_conditioner/sql/query_ac.sql

@@ -0,0 +1,4 @@
+-- 查询空调状态
+SELECT *
+FROM air_conditioner
+WHERE id = ?;

konabot/plugins/air_conditioner/sql/query_crash_and_rank.sql

@@ -0,0 +1,23 @@
+-- 从 air_conditioner_crash_log 表中获取指定 id 损坏的次数以及损坏次数的排名
+SELECT crash_count, crash_rank
+FROM (
+    SELECT id,
+           COUNT(*) AS crash_count,
+           RANK() OVER (ORDER BY COUNT(*) DESC) AS crash_rank
+    FROM air_conditioner_crash_log
+    GROUP BY id
+) AS ranked_data
+WHERE id = ?
+-- 如果该 id 没有损坏记录，则返回 0 次损坏和对应的最后一名
+UNION
+SELECT 0 AS crash_count,
+    (SELECT COUNT(DISTINCT id) + 1 FROM air_conditioner_crash_log) AS crash_rank
+FROM (
+    SELECT DISTINCT id
+    FROM air_conditioner_crash_log
+) AS ranked_data
+WHERE NOT EXISTS (
+    SELECT 1
+    FROM air_conditioner_crash_log
+    WHERE id = ?
+);

konabot/plugins/air_conditioner/sql/query_peak.sql

@@ -0,0 +1,13 @@
+-- 查询目前所有空调中的最高温度与最低温度与历史最高低温
+SELECT
+    (SELECT MAX(temperature) FROM air_conditioner 
+     WHERE "on" = TRUE AND NOT frozen AND NOT burnt) AS max,
+    
+    (SELECT MIN(temperature) FROM air_conditioner 
+     WHERE "on" = TRUE AND NOT frozen AND NOT burnt) AS min,
+    
+    (SELECT MAX(temperature) FROM air_conditioner_log 
+     WHERE "on" = TRUE AND NOT frozen AND NOT burnt) AS his_max,
+    
+    (SELECT MIN(temperature) FROM air_conditioner_log 
+     WHERE "on" = TRUE AND NOT frozen AND NOT burnt) AS his_min;

konabot/plugins/air_conditioner/sql/update_ac.sql

@@ -0,0 +1,10 @@
+-- 更新空调状态
+UPDATE air_conditioner
+SET "on" = ?, temperature = ?, burnt = ?, frozen = ?
+WHERE id = ?;
+
+-- 插入日志记录（从更新后的数据获取）
+INSERT INTO air_conditioner_log (id, "on", temperature, burnt, frozen)
+SELECT id, "on", temperature, burnt, frozen
+FROM air_conditioner
+WHERE id = ?;

konabot/plugins/bilibili_fetch/__init__.py

@@ -1,39 +1,41 @@
 import re
 
-from nonebot import on_message
+from nonebot import get_plugin_config, on_message
 from nonebot_plugin_alconna import Reference, Reply, UniMsg
 
 from nonebot.adapters import Event
+from nonebot.adapters.onebot.v11.event import GroupMessageEvent as OB11GroupEvent
+from pydantic import BaseModel
 
 
-matcher_fix = on_message()
+class Config(BaseModel):
+    bilifetch_enabled_groups: list[int] = []
 
+
+config = get_plugin_config(Config)
 pattern = (
     r"^(?:(?:av|cv)\d+|BV[a-zA-Z0-9]{10})|"
     r"(?:b23\.tv|bili(?:22|23|33|2233)\.cn|\.bilibili\.com|QQ小程序(?:]|]|\])哔哩哔哩).{0,500}"
 )
 
 
-@matcher_fix.handle()
-async def _(msg: UniMsg, event: Event):
+def _rule(msg: UniMsg, evt: Event) -> bool:
+    if isinstance(evt, OB11GroupEvent):
+        if evt.group_id not in config.bilifetch_enabled_groups:
+            return False
+
     to_search = msg.exclude(Reply, Reference).dump(json=True)
     to_search2 = msg.exclude(Reply, Reference).extract_plain_text()
     if not re.search(pattern, to_search) and not re.search(pattern, to_search2):
-        return
+        return False
+    return True
 
+
+matcher_fix = on_message(rule=_rule)
+
+@matcher_fix.handle()
+async def _(event: Event):
     from nonebot_plugin_analysis_bilibili import handle_analysis
 
     await handle_analysis(event)
 
-    # b_url: str
-    # b_page: str | None
-    # b_time: str | None
-    #
-    # from nonebot_plugin_analysis_bilibili.analysis_bilibili import extract as bilibili_extract
-    #
-    # b_url, b_page, b_time = bilibili_extract(to_search)
-    # if b_url is None:
-    #     return
-    #
-    # await matcher_fix.send(await UniMessage().text(b_url).export())
-

konabot/plugins/celeste_classic/__init__.py

@@ -0,0 +1,154 @@
+from pathlib import Path
+import subprocess
+import tempfile
+from typing import Any
+from loguru import logger
+from nonebot import on_message
+from pydantic import BaseModel
+
+from nonebot.adapters import Event, Bot
+from nonebot_plugin_alconna import UniMessage, UniMsg
+from nonebot.adapters.onebot.v11.event import MessageEvent as OB11MessageEvent
+
+from konabot.common.artifact import ArtifactDepends, ensure_artifact, register_artifacts
+from konabot.common.data_man import DataManager
+from konabot.common.path import BINARY_PATH, DATA_PATH
+
+
+arti_ccleste_wrap_linux = ArtifactDepends(
+    url="https://github.com/Passthem-desu/pt-ccleste-wrap/releases/download/v0.1.5/ccleste-wrap",
+    sha256="ba4118c6465d1ca1547cdd1bd11c6b9e6a6a98ea8967b55485aeb6b77bb7e921",
+    target=BINARY_PATH / "ccleste-wrap",
+    required_os="Linux",
+    required_arch="x86_64",
+)
+arti_ccleste_wrap_windows = ArtifactDepends(
+    url="https://github.com/Passthem-desu/pt-ccleste-wrap/releases/download/v0.1.5/ccleste-wrap.exe",
+    sha256="7df382486a452485cdcf2115eabd7f772339ece470ab344074dc163fc7981feb",
+    target=BINARY_PATH / "ccleste-wrap.exe",
+    required_os="Windows",
+    required_arch="AMD64",
+)
+
+
+register_artifacts(arti_ccleste_wrap_linux)
+register_artifacts(arti_ccleste_wrap_windows)
+
+
+class CelesteStatus(BaseModel):
+    records: dict[str, str] = {}
+
+
+celeste_status = DataManager(CelesteStatus, DATA_PATH / "celeste-status.json")
+
+
+# ↓ 这里的 Type Hinting 是为了能 fit 进去 set[str | tuple[str, ...]]
+aliases: set[Any] = {"celeste", "蔚蓝", "爬山", "鳌太线"}
+ALLOW_CHARS = "wasdxc0123456789 \t\n\r"
+
+
+async def get_prev(evt: Event, bot: Bot) -> str | None:
+    prev = None
+    if isinstance(evt, OB11MessageEvent):
+        if evt.reply is not None:
+            prev = f"QQ:{bot.self_id}:" + str(evt.reply.message_id)
+        else:
+            for seg in evt.get_message():
+                if seg.type == 'reply':
+                    msgid = seg.get('id')
+                    prev = f"QQ:{bot.self_id}:" + str(msgid)
+    if prev is not None:
+        async with celeste_status.get_data() as data:
+            prev = data.records.get(prev)
+    return prev
+
+
+async def match_celeste(evt: Event, bot: Bot, msg: UniMsg) -> bool:
+    prev = await get_prev(evt, bot)
+    text = msg.extract_plain_text().strip()
+    if any(text.startswith(a) for a in aliases):
+        return True
+    if prev is not None:
+        return True
+    return False
+
+
+# cmd = on_command(cmd="celeste", aliases=aliases)
+cmd = on_message(rule=match_celeste)
+
+
+@cmd.handle()
+async def _(msg: UniMsg, evt: Event, bot: Bot):
+    prev = await get_prev(evt, bot)
+    actions = msg.extract_plain_text().strip()
+    for alias in aliases:
+        actions = actions.removeprefix(alias)
+    actions = actions.strip()
+    if len(actions) == 0:
+        return
+    if any((c not in ALLOW_CHARS) for c in actions):
+        return
+
+    await ensure_artifact(arti_ccleste_wrap_linux)
+    await ensure_artifact(arti_ccleste_wrap_windows)
+
+    bin: Path | None = None
+    for arti in (
+        arti_ccleste_wrap_linux,
+        arti_ccleste_wrap_windows,
+    ):
+        if not arti.is_corresponding_platform():
+            continue
+        bin = arti.target
+        if not bin.exists():
+            continue
+        break
+
+    if bin is None:
+        logger.warning("Celeste 模块没有找到该系统需要的二进制文件")
+        return
+
+    if prev is not None:
+        prev_append = ["-p", prev]
+    else:
+        prev_append = []
+
+    try:
+        with tempfile.TemporaryDirectory() as _tempdir:
+            tempdir = Path(_tempdir)
+            gif_path = tempdir / "render.gif"
+            cmd_celeste = [
+                bin,
+                "-a",
+                actions,
+                "-o",
+                gif_path,
+            ] + prev_append
+            logger.info(f"执行指令调用 celeste: CMD={cmd_celeste}")
+            res = subprocess.run(cmd_celeste, timeout=5, capture_output=True)
+            if res.returncode != 0:
+                logger.warning(f"渲染 Celeste 时的输出不是 0 CODE={res.returncode} STDOUT={res.stdout} STDERR={res.stderr}")
+                await UniMessage.text(f"渲染 Celeste 时出错啦！下面是输出：\n\n{res.stdout.decode()}{res.stderr.decode()}").send(evt, bot, at_sender=True)
+                return
+            if not gif_path.exists():
+                logger.warning("没有找到 Celeste 渲染的文件")
+                await UniMessage.text("渲染 Celeste 时出错啦！").send(evt, bot, at_sender=True)
+                return
+            gif_data = gif_path.read_bytes()
+    except TimeoutError:
+        logger.warning("在渲染 Celeste 时超时了")
+        await UniMessage("渲染 Celeste 时超时了！请检查你的操作清单，不能太长").send(evt, bot, at_sender=True)
+        return
+
+    receipt = await UniMessage.image(raw=gif_data).send(evt, bot)
+    async with celeste_status.get_data() as data:
+        if prev:
+            actions = prev + "\n" + actions
+        if isinstance(evt, OB11MessageEvent):
+            for _msgid in receipt.msg_ids:
+                msgid = _msgid["message_id"]
+                data.records[f"QQ:{bot.self_id}:{msgid}"] = actions
+        else:
+            for msgid in receipt.msg_ids:
+                data.records[f"DISCORD:{bot.self_id}:{msgid}"] = actions
+

konabot/plugins/fx_process/__init__.py

@@ -0,0 +1,277 @@
+import asyncio as asynkio
+from io import BytesIO
+
+from inspect import signature
+import random
+
+from konabot.common.longtask import DepLongTaskTarget
+from konabot.common.nb.exc import BotExceptionMessage
+from konabot.common.nb.extract_image import DepImageBytesOrNone
+from nonebot.adapters import Event as BaseEvent
+from nonebot import on_message, logger
+
+from nonebot_plugin_alconna import (
+    UniMessage,
+    UniMsg
+)
+
+from konabot.plugins.fx_process.fx_handle import ImageFilterStorage
+from konabot.plugins.fx_process.fx_manager import ImageFilterManager
+
+from PIL import Image, ImageSequence
+
+from konabot.plugins.fx_process.types import FilterItem, ImageRequireSignal, ImagesListRequireSignal, SenderInfo, StoredInfo
+
+def try_convert_type(param_type, input_param, sender_info: SenderInfo = None) -> tuple[bool, any]:
+    converted_value = None
+    try:
+        if param_type is float:
+            converted_value = float(input_param)
+        elif param_type is int:
+            converted_value = int(input_param)
+        elif param_type is bool:
+            converted_value = input_param.lower() in ['true', '1', 'yes', '是', '开']
+        elif param_type is Image.Image:
+            converted_value = ImageRequireSignal()
+            return False, converted_value
+        elif param_type is SenderInfo:
+            converted_value = sender_info
+            return False, converted_value
+        elif param_type == list[Image.Image]:
+            converted_value = ImagesListRequireSignal()
+            return False, converted_value
+        elif param_type is str:
+            if input_param is None:
+                return False, None
+            converted_value = str(input_param)
+        else:
+            return False, None
+    except Exception:
+        return False, None
+    return True, converted_value
+
+def prase_input_args(input_str: str, sender_info: SenderInfo = None) -> list[FilterItem]:
+    # 按分号或换行符分割参数
+    args = []
+    for part in input_str.replace('\n', ';').split(';'):
+        part = part.strip()
+        if not part:
+            continue
+        split_part = part.split()
+        filter_name = split_part[0]
+        if not ImageFilterManager.has_filter(filter_name):
+            continue
+        filter_func = ImageFilterManager.get_filter(filter_name)
+        input_filter_args = split_part[1:]
+        # 获取函数最大参数数量
+        sig = signature(filter_func)
+        max_params = len(sig.parameters) - 1  # 减去第一个参数 image
+        # 从 args 提取参数，并转换为适当类型
+        func_args = []
+        for i in range(0, max_params):
+            # 尝试将参数转换为函数签名中对应的类型
+            param = list(sig.parameters.values())[i + 1]
+            param_type = param.annotation
+            # 根据函数所需要的参数，从输入参数中提取，如果不匹配就使用默认值，将当前参数递交给下一个循环
+            input_param = input_filter_args[0] if len(input_filter_args) > 0 else None
+            state, converted_param = try_convert_type(param_type, input_param, sender_info)
+            if state:
+                input_filter_args.pop(0)
+            if converted_param is None and param.default != param.empty:
+                converted_param = param.default
+            func_args.append(converted_param)
+        args.append(FilterItem(name=filter_name,filter=filter_func, args=func_args))
+    return args
+
+def handle_filters_to_image(images: list[Image.Image], filters: list[FilterItem]) -> Image.Image:
+    for filter_item in filters:
+        logger.debug(f"{filter_item}")
+        filter_func = filter_item.filter
+        func_args = filter_item.args
+        # 检测参数中是否有 ImageRequireSignal，如果有则传入对应数量的图像列表
+        if any(isinstance(arg, ImageRequireSignal) for arg in func_args):
+            # 替换 ImageRequireSignal 为 images 对应索引的图像
+            actual_args = []
+            img_signal_count = 1 # 从 images[1] 开始取图像
+            for arg in func_args:
+                if isinstance(arg, ImageRequireSignal):
+                    if img_signal_count >= len(images):
+                        raise BotExceptionMessage("图像数量不足，无法满足滤镜需求！")
+                    actual_args.append(images[img_signal_count])
+                    img_signal_count += 1
+                else:
+                    actual_args.append(arg)
+            func_args = actual_args
+        # 检测参数中是否有 ImagesListRequireSignal，如果有则传入整个图像列表
+        if any(isinstance(arg, ImagesListRequireSignal) for arg in func_args):
+            actual_args = []
+            for arg in func_args:
+                if isinstance(arg, ImagesListRequireSignal):
+                    actual_args.append(images)
+                else:
+                    actual_args.append(arg)
+            func_args = actual_args
+        
+        logger.debug(f"Applying filter: {filter_item.name} with args: {func_args}")
+
+        images[0] = filter_func(images[0], *func_args)        
+    return images[0]
+
+def copy_images_by_index(images: list[Image.Image], index: int) -> list[Image.Image]:
+    # 将导入图像列表复制为新的图像列表，如果是动图，那么就找到对应索引下的帧
+    new_images = []
+    for img in images:
+        if getattr(img, "is_animated", False):
+            frames = img.n_frames
+            frame_idx = index % frames
+            img.seek(frame_idx)
+            new_images.append(img.copy())
+        else:
+            new_images.append(img.copy())
+    
+    return new_images
+
+def generate_image(images: list[Image.Image], filters: list[FilterItem]) -> Image.Image:
+    # 处理位于最前面的生成类滤镜
+    while filters and filters[0].name.strip() in ImageFilterManager.generate_filter_map:
+        gen_filter = filters.pop(0)
+        gen_func = gen_filter.filter
+        func_args = gen_filter.args[1:]  # 去掉第一个 list 参数
+        gen_func(None, images, *func_args)
+
+def save_or_load_image(images: list[Image.Image], filters: list[FilterItem], sender_info: SenderInfo) -> StoredInfo | None:
+    stored_info = None
+    # 处理位于最前面的“读取图像”、“存入图像”
+    if not filters:
+        return
+    while filters and filters[0].name.strip() in ["读取图像", "存入图像"]:
+        if filters[0].name.strip() == "读取图像":
+            load_filter = filters.pop(0)
+            path = load_filter.args[0] if load_filter.args else ""
+            ImageFilterStorage.load_image(None, path, images, sender_info)
+        elif filters[0].name.strip() == "存入图像":
+            store_filter = filters.pop(0)
+            name = store_filter.args[0] if store_filter.args[0] else str(random.randint(10000,99999))
+            stored_info = ImageFilterStorage.store_image(images[0], name, sender_info)
+    # 将剩下的“读取图像”或“存入图像”参数全部删除，避免后续非法操作
+    filters[:] = [f for f in filters if f.name.strip() not in ["读取图像", "存入图像"]]
+    return stored_info
+
+async def apply_filters_to_images(images: list[Image.Image], filters: list[FilterItem], sender_info: SenderInfo) -> BytesIO | StoredInfo:
+    # 先处理存取图像、生成图像的操作
+    stored_info = save_or_load_image(images, filters, sender_info)
+    generate_image(images, filters)
+
+    if stored_info and len(filters) <= 0: 
+        return stored_info
+
+    if len(images) <= 0:
+        raise BotExceptionMessage("没有可处理的图像！")
+
+    # 检测是否需要将静态图视作动图处理
+    frozen_to_move = any(
+        filter_item.name == "动图"
+        for filter_item in filters
+    )
+    static_fps = 10
+    # 找到动图参数 fps
+    if frozen_to_move:
+        for filter_item in filters:
+            if filter_item.name == "动图" and filter_item.args:
+                try:
+                    static_fps = int(filter_item.args[0])
+                except Exception:
+                    static_fps = 10
+                break
+    # 如果 image 是动图，则逐帧处理
+    img = images[0]
+    logger.debug("开始图像处理")
+    output = BytesIO()
+    if getattr(img, "is_animated", False) or frozen_to_move:
+        frames = []
+        append_images = []
+        if getattr(img, "is_animated", False):
+            logger.debug("处理动图帧")
+        else:
+            # 将静态图视作单帧动图处理，拷贝 10 帧
+            logger.debug("处理静态图为多帧动图")
+            append_images = [img.copy() for _ in range(10)]
+            img.info['duration'] = int(1000 / static_fps)
+
+        async def process_single_frame(frame_images: list[Image.Image], frame_idx: int) -> Image.Image:
+            """处理单帧的异步函数"""
+            logger.debug(f"开始处理帧 {frame_idx}")
+            result = await asynkio.to_thread(handle_filters_to_image, frame_images, filters)
+            logger.debug(f"完成处理帧 {frame_idx}")
+            return result
+
+        # 并发处理所有帧
+        tasks = []
+        all_frames = []
+        for i, frame in enumerate(list(ImageSequence.Iterator(img)) + append_images):
+            all_frames.append(frame.copy())
+            images_copy = copy_images_by_index(images, i)
+            task = process_single_frame(images_copy, i)
+            tasks.append(task)
+
+        frames = await asynkio.gather(*tasks, return_exceptions=False)
+
+        # 检查是否有处理失败的帧
+        for i, result in enumerate(frames):
+            if isinstance(result, Exception):
+                logger.error(f"帧 {i} 处理失败: {result}")
+                # 使用原始帧作为回退
+                frames[i] = all_frames[i]
+
+        logger.debug("保存动图")
+        frames[0].save(
+            output,
+            format="GIF",
+            save_all=True,
+            append_images=frames[1:],
+            loop=img.info.get("loop", 0),
+            disposal=img.info.get("disposal", 2),
+            duration=img.info.get("duration", 100),
+        )
+        logger.debug("Animated image saved")
+    else:
+        img = handle_filters_to_image(images=images, filters=filters)
+        img.save(output, format="PNG")
+    logger.debug("Image processing completed")
+    output.seek(0)
+    return output
+    
+def is_fx_mentioned(evt: BaseEvent, msg: UniMsg) -> bool:
+    txt = msg.extract_plain_text()
+    if "fx" not in txt[:3].lower():
+        return False
+    return True
+
+fx_on = on_message(rule=is_fx_mentioned)
+
+@fx_on.handle()
+async def _(msg: UniMsg, event: BaseEvent, target: DepLongTaskTarget, image_data: DepImageBytesOrNone):
+    preload_imgs = []
+    # 提取图像
+    try:
+        preload_imgs.append(Image.open(BytesIO(image_data)))
+    except Exception:
+        logger.info("No image found in message for FX processing.")
+    args = msg.extract_plain_text().split()
+    if len(args) < 2:
+        return
+
+    sender_info = SenderInfo(
+        group_id=target.channel_id,
+        qq_id=target.target_id
+    )
+
+    filters = prase_input_args(msg.extract_plain_text()[2:], sender_info=sender_info)
+    # if not filters:
+    #     return
+    output = await apply_filters_to_images(preload_imgs, filters, sender_info)
+    if isinstance(output,StoredInfo):
+        await fx_on.send(await UniMessage().text(f"图像已存为「{output.name}」！").export())
+    elif isinstance(output,BytesIO):
+        await fx_on.send(await UniMessage().image(raw=output).export())
+    

konabot/plugins/fx_process/color_handle.py

@@ -0,0 +1,50 @@
+from typing import Optional
+from PIL import ImageColor
+
+class ColorHandle:
+    color_name_map = {
+        "红": (255, 0, 0),
+        "绿": (0, 255, 0),
+        "蓝": (0, 0, 255),
+        "黄": (255, 255, 0),
+        "紫": (128, 0, 128),
+        "黑": (0, 0, 0),
+        "白": (255, 255, 255),
+        "橙": (255, 165, 0),
+        "粉": (255, 192, 203),
+        "灰": (128, 128, 128),
+        "青": (0, 255, 255),
+        "靛": (75, 0, 130),
+        "棕": (165, 42, 42),
+        "浅": (200, 200, 200),
+        "深": (50, 50, 50),
+        "亮": (255, 255, 224),
+        "暗": (47, 79, 79),
+    }
+
+    @staticmethod
+    def set_or_blend_color(ori_color: Optional[tuple], target_color: tuple) -> tuple:
+        # 如果没有指定初始颜色，返回目标颜色
+        if ori_color is None:
+            return target_color
+        # 混合颜色，取平均值
+        blended_color = tuple((o + t) // 2 for o, t in zip(ori_color, target_color))
+        return blended_color
+
+    @staticmethod
+    def parse_color(color_str: str) -> tuple:
+        # 如果是纯括号，则加上前缀 rgb
+        if color_str.startswith('(') and color_str.endswith(')'):
+            color_str = 'rgb' + color_str
+        try:
+            return ImageColor.getrgb(color_str)
+        except ValueError:
+            pass
+        base_color = None
+        color_str = color_str.replace('色', '')
+        for name, rgb in ColorHandle.color_name_map.items():
+            if name in color_str:
+                base_color = ColorHandle.set_or_blend_color(base_color, rgb)
+        if base_color is not None:
+            return base_color
+        return (255, 255, 255)  # 默认白色

konabot/plugins/fx_process/fx_manager.py

@@ -0,0 +1,87 @@
+from typing import Optional
+from konabot.plugins.fx_process.fx_handle import ImageFilterEmpty, ImageFilterImplement, ImageFilterStorage
+
+class ImageFilterManager:
+    filter_map = {
+        "模糊": ImageFilterImplement.apply_blur,
+        "马赛克": ImageFilterImplement.apply_mosaic,
+        "轮廓": ImageFilterImplement.apply_contour,
+        "锐化": ImageFilterImplement.apply_sharpen,
+        "边缘增强": ImageFilterImplement.apply_edge_enhance,
+        "浮雕": ImageFilterImplement.apply_emboss,
+        "查找边缘": ImageFilterImplement.apply_find_edges,
+        "平滑": ImageFilterImplement.apply_smooth,
+        "反色": ImageFilterImplement.apply_invert,
+        "黑白": ImageFilterImplement.apply_black_white,
+        "阈值": ImageFilterImplement.apply_threshold,
+        "对比度": ImageFilterImplement.apply_contrast,
+        "亮度": ImageFilterImplement.apply_brightness,
+        "色彩": ImageFilterImplement.apply_color,
+        "色调": ImageFilterImplement.apply_to_color,
+        "缩放": ImageFilterImplement.apply_resize,
+        "波纹": ImageFilterImplement.apply_wave,
+        "色键": ImageFilterImplement.apply_color_key,
+        "暗角": ImageFilterImplement.apply_vignette,
+        "发光": ImageFilterImplement.apply_glow,
+        "RGB分离": ImageFilterImplement.apply_rgb_split,
+        "光学补偿": ImageFilterImplement.apply_optical_compensation,
+        "球面化": ImageFilterImplement.apply_spherize,
+        "旋转": ImageFilterImplement.apply_rotate,
+        "透视变换": ImageFilterImplement.apply_perspective_transform,
+        "裁剪": ImageFilterImplement.apply_crop,
+        "噪点": ImageFilterImplement.apply_noise,
+        "平移": ImageFilterImplement.apply_translate,
+        "拓展边缘": ImageFilterImplement.apply_expand_edges,
+        "素描": ImageFilterImplement.apply_sketch,
+        "叠加颜色": ImageFilterImplement.apply_gradient_overlay,
+        "阴影": ImageFilterImplement.apply_shadow,
+        "径向模糊": ImageFilterImplement.apply_radial_blur,
+        "旋转模糊": ImageFilterImplement.apply_spin_blur,
+        "方向模糊": ImageFilterImplement.apply_directional_blur,
+        "边缘模糊": ImageFilterImplement.apply_focus_blur,
+        "缩放模糊": ImageFilterImplement.apply_zoom_blur,
+        "镜像": ImageFilterImplement.apply_mirror_half,
+        "水平翻转": ImageFilterImplement.apply_flip_horizontal,
+        "垂直翻转": ImageFilterImplement.apply_flip_vertical,
+        "复制": ImageFilterImplement.copy_area,
+        "晃动": ImageFilterImplement.apply_random_wiggle,
+        "动图": ImageFilterEmpty.empty_filter_param,
+        "像素抖动": ImageFilterImplement.apply_pixel_jitter,
+        "描边": ImageFilterImplement.apply_stroke,
+        "形状描边": ImageFilterImplement.apply_shape_stroke,
+        "半调": ImageFilterImplement.apply_halftone,
+        "设置通道": ImageFilterImplement.apply_set_channel,
+        "设置遮罩": ImageFilterImplement.apply_set_mask,
+        # 图像处理
+        "存入图像": ImageFilterStorage.store_image,
+        "读取图像": ImageFilterStorage.load_image,
+        "暂存图像": ImageFilterStorage.temp_store_image,
+        "交换图像": ImageFilterStorage.swap_image_index,
+        "删除图像": ImageFilterStorage.delete_image_by_index,
+        "选择图像": ImageFilterStorage.select_image_by_index,
+        # 多图像处理
+        "混合图像": ImageFilterImplement.apply_blend,
+        "覆盖图像": ImageFilterImplement.apply_overlay,
+        # 生成式
+        "覆加颜色": ImageFilterImplement.generate_solid,
+    }
+
+    generate_filter_map = {
+        "生成图层": ImageFilterImplement.generate_empty,
+        "生成文本": ImageFilterImplement.generate_text
+    }
+
+    @classmethod
+    def get_filter(cls, name: str) -> Optional[callable]:
+        if name in cls.filter_map:
+            return cls.filter_map[name]
+        elif name in cls.generate_filter_map:
+            return cls.generate_filter_map[name]
+        else:
+            return None
+    
+    @classmethod
+    def has_filter(cls, name: str) -> bool:
+        return name in cls.filter_map or name in cls.generate_filter_map
+    
+    

konabot/plugins/fx_process/gradient.py

@@ -0,0 +1,344 @@
+import re
+from konabot.plugins.fx_process.color_handle import ColorHandle
+import numpy as np
+from PIL import Image, ImageDraw
+from typing import List, Tuple, Dict, Optional
+
+class GradientGenerator:
+    """渐变生成器类"""
+    
+    def __init__(self):
+        self.has_numpy = hasattr(np, '__version__')
+    
+    def parse_color_list(self, color_list_str: str) -> List[Dict]:
+        """解析渐变颜色列表字符串
+        
+        Args:
+            color_list_str: 格式如 '[rgb(255,0,0)|(0,0)+rgb(0,255,0)|(0,100)+rgb(0,0,255)|(50,100)]'
+        
+        Returns:
+            list: 包含颜色和位置信息的字典列表
+        """
+        color_nodes = []
+        color_list_str = color_list_str.strip('[]').strip()
+        matches = color_list_str.split('+')
+        
+        for single_str in matches:
+            color_str = single_str.split('|')[0]
+            pos_str = single_str.split('|')[1] if '|' in single_str else '0,0'
+
+            color = ColorHandle.parse_color(color_str.strip())
+
+            try:
+                pos_str = pos_str.replace('(', '').replace(')', '')
+                x_str, y_str = pos_str.split(',')
+                x_percent = float(x_str.strip().replace('%', ''))
+                y_percent = float(y_str.strip().replace('%', ''))
+                x_percent = max(0, min(100, x_percent))
+                y_percent = max(0, min(100, y_percent))
+            except:
+                x_percent = 0
+                y_percent = 0
+            
+            color_nodes.append({
+                'color': color,
+                'position': (x_percent / 100.0, y_percent / 100.0)
+            })
+        
+        if not color_nodes:
+            color_nodes = [
+                {'color': (255, 0, 0), 'position': (0, 0)},
+                {'color': (0, 0, 255), 'position': (1, 1)}
+            ]
+        
+        return color_nodes
+    
+    def create_gradient(self, width: int, height: int, color_nodes: List[Dict]) -> Image.Image:
+        """创建渐变图像
+        
+        Args:
+            width: 图像宽度
+            height: 图像高度
+            color_nodes: 颜色节点列表
+        
+        Returns:
+            Image.Image: 渐变图像
+        """
+        if len(color_nodes) == 1:
+            return Image.new('RGB', (width, height), color_nodes[0]['color'])
+        elif len(color_nodes) == 2:
+            return self._create_linear_gradient(width, height, color_nodes)
+        else:
+            return self._create_radial_gradient(width, height, color_nodes)
+    
+    def _create_linear_gradient(self, width: int, height: int, color_nodes: List[Dict]) -> Image.Image:
+        """创建线性渐变"""
+        color1 = color_nodes[0]['color']
+        color2 = color_nodes[1]['color']
+        pos1 = color_nodes[0]['position']
+        pos2 = color_nodes[1]['position']
+        
+        if self.has_numpy:
+            return self._create_linear_gradient_numpy(width, height, color1, color2, pos1, pos2)
+        else:
+            return self._create_linear_gradient_pil(width, height, color1, color2, pos1, pos2)
+    
+    def _create_linear_gradient_numpy(self, width: int, height: int, 
+                                    color1: Tuple, color2: Tuple, 
+                                    pos1: Tuple, pos2: Tuple) -> Image.Image:
+        """使用numpy创建线性渐变"""
+        # 创建坐标网格
+        x = np.linspace(0, 1, width)
+        y = np.linspace(0, 1, height)
+        xx, yy = np.meshgrid(x, y)
+        
+        # 计算渐变方向
+        dx = pos2[0] - pos1[0]
+        dy = pos2[1] - pos1[1]
+        length_sq = dx * dx + dy * dy
+        
+        if length_sq > 0:
+            # 计算投影参数
+            t = ((xx - pos1[0]) * dx + (yy - pos1[1]) * dy) / length_sq
+            t = np.clip(t, 0, 1)
+        else:
+            t = np.zeros_like(xx)
+        
+        # 插值颜色
+        r = color1[0] + (color2[0] - color1[0]) * t
+        g = color1[1] + (color2[1] - color1[1]) * t
+        b = color1[2] + (color2[2] - color1[2]) * t
+        
+        # 创建图像
+        gradient_array = np.stack([r, g, b], axis=-1).astype(np.uint8)
+        return Image.fromarray(gradient_array)
+    
+    def _create_linear_gradient_pil(self, width: int, height: int, 
+                                   color1: Tuple, color2: Tuple, 
+                                   pos1: Tuple, pos2: Tuple) -> Image.Image:
+        """使用PIL创建线性渐变（没有numpy时使用）"""
+        gradient = Image.new('RGB', (width, height))
+        draw = ImageDraw.Draw(gradient)
+        
+        # 判断渐变方向
+        if abs(pos1[0] - pos2[0]) < 0.01:  # 垂直渐变
+            y1 = int(pos1[1] * (height - 1))
+            y2 = int(pos2[1] * (height - 1))
+            
+            if y2 < y1:
+                y1, y2 = y2, y1
+                color1, color2 = color2, color1
+            
+            if y2 > y1:
+                for y in range(height):
+                    if y <= y1:
+                        fill_color = color1
+                    elif y >= y2:
+                        fill_color = color2
+                    else:
+                        ratio = (y - y1) / (y2 - y1)
+                        r = int(color1[0] + (color2[0] - color1[0]) * ratio)
+                        g = int(color1[1] + (color2[1] - color1[1]) * ratio)
+                        b = int(color1[2] + (color2[2] - color1[2]) * ratio)
+                        fill_color = (r, g, b)
+                    
+                    draw.line([(0, y), (width, y)], fill=fill_color)
+            else:
+                draw.rectangle([0, 0, width, height], fill=color1)
+                
+        elif abs(pos1[1] - pos2[1]) < 0.01:  # 水平渐变
+            x1 = int(pos1[0] * (width - 1))
+            x2 = int(pos2[0] * (width - 1))
+            
+            if x2 < x1:
+                x1, x2 = x2, x1
+                color1, color2 = color2, color1
+            
+            if x2 > x1:
+                for x in range(width):
+                    if x <= x1:
+                        fill_color = color1
+                    elif x >= x2:
+                        fill_color = color2
+                    else:
+                        ratio = (x - x1) / (x2 - x1)
+                        r = int(color1[0] + (color2[0] - color1[0]) * ratio)
+                        g = int(color1[1] + (color2[1] - color1[1]) * ratio)
+                        b = int(color1[2] + (color2[2] - color1[2]) * ratio)
+                        fill_color = (r, g, b)
+                    
+                    draw.line([(x, 0), (x, height)], fill=fill_color)
+            else:
+                draw.rectangle([0, 0, width, height], fill=color1)
+                
+        else:  # 对角渐变（简化处理为左上到右下）
+            for y in range(height):
+                for x in range(width):
+                    distance = (x/width + y/height) / 2
+                    r = int(color1[0] + (color2[0] - color1[0]) * distance)
+                    g = int(color1[1] + (color2[1] - color1[1]) * distance)
+                    b = int(color1[2] + (color2[2] - color1[2]) * distance)
+                    draw.point((x, y), fill=(r, g, b))
+        
+        return gradient
+    
+    def _create_radial_gradient(self, width: int, height: int, color_nodes: List[Dict]) -> Image.Image:
+        """创建径向渐变"""
+        if self.has_numpy and len(color_nodes) > 2:
+            return self._create_radial_gradient_numpy(width, height, color_nodes)
+        else:
+            return self._create_simple_gradient(width, height, color_nodes)
+    
+    def _create_radial_gradient_numpy(self, width: int, height: int, color_nodes: List[Dict]) -> Image.Image:
+        """使用numpy创建径向渐变（多色）"""
+        # 创建坐标网格
+        x = np.linspace(0, 1, width)
+        y = np.linspace(0, 1, height)
+        xx, yy = np.meshgrid(x, y)
+        
+        # 提取颜色和位置
+        positions = np.array([node['position'] for node in color_nodes])
+        colors = np.array([node['color'] for node in color_nodes])
+        
+        # 计算每个点到所有节点的距离
+        distances = np.sqrt((xx[:, :, np.newaxis] - positions[np.newaxis, np.newaxis, :, 0]) ** 2 +
+                           (yy[:, :, np.newaxis] - positions[np.newaxis, np.newaxis, :, 1]) ** 2)
+        
+        # 找到最近的两个节点
+        sorted_indices = np.argsort(distances, axis=2)
+        nearest_idx = sorted_indices[:, :, 0]
+        second_idx = sorted_indices[:, :, 1]
+        
+        # 获取对应的颜色
+        nearest_colors = colors[nearest_idx]
+        second_colors = colors[second_idx]
+        
+        # 获取距离并计算权重
+        nearest_dist = np.take_along_axis(distances, np.expand_dims(nearest_idx, axis=2), axis=2)[:, :, 0]
+        second_dist = np.take_along_axis(distances, np.expand_dims(second_idx, axis=2), axis=2)[:, :, 0]
+        
+        total_dist = nearest_dist + second_dist
+        mask = total_dist > 0
+        weight1 = np.zeros_like(nearest_dist)
+        weight1[mask] = second_dist[mask] / total_dist[mask]
+        weight2 = 1 - weight1
+        
+        # 插值颜色
+        r = nearest_colors[:, :, 0] * weight1 + second_colors[:, :, 0] * weight2
+        g = nearest_colors[:, :, 1] * weight1 + second_colors[:, :, 1] * weight2
+        b = nearest_colors[:, :, 2] * weight1 + second_colors[:, :, 2] * weight2
+        
+        gradient_array = np.stack([r, g, b], axis=-1).astype(np.uint8)
+        return Image.fromarray(gradient_array)
+    
+    def _create_simple_gradient(self, width: int, height: int, color_nodes: List[Dict]) -> Image.Image:
+        """创建简化渐变（没有numpy或多色时使用）"""
+        gradient = Image.new('RGB', (width, height))
+        draw = ImageDraw.Draw(gradient)
+        
+        if len(color_nodes) >= 2:
+            # 使用第一个和最后一个颜色创建简单渐变
+            color1 = color_nodes[0]['color']
+            color2 = color_nodes[-1]['color']
+            
+            # 判断节点分布
+            x_positions = [node['position'][0] for node in color_nodes]
+            y_positions = [node['position'][1] for node in color_nodes]
+            
+            if all(abs(x - x_positions[0]) < 0.01 for x in x_positions):
+                # 垂直渐变
+                for y in range(height):
+                    ratio = y / (height - 1) if height > 1 else 0
+                    r = int(color1[0] + (color2[0] - color1[0]) * ratio)
+                    g = int(color1[1] + (color2[1] - color1[1]) * ratio)
+                    b = int(color1[2] + (color2[2] - color1[2]) * ratio)
+                    draw.line([(0, y), (width, y)], fill=(r, g, b))
+            else:
+                # 水平渐变
+                for x in range(width):
+                    ratio = x / (width - 1) if width > 1 else 0
+                    r = int(color1[0] + (color2[0] - color1[0]) * ratio)
+                    g = int(color1[1] + (color2[1] - color1[1]) * ratio)
+                    b = int(color1[2] + (color2[2] - color1[2]) * ratio)
+                    draw.line([(x, 0), (x, height)], fill=(r, g, b))
+        else:
+            # 单色
+            draw.rectangle([0, 0, width, height], fill=color_nodes[0]['color'])
+        
+        return gradient
+    
+    def create_simple_gradient(self, width: int, height: int, 
+                              start_color: Tuple, end_color: Tuple, 
+                              direction: str = 'vertical') -> Image.Image:
+        """创建简单双色渐变
+        
+        Args:
+            width: 图像宽度
+            height: 图像高度
+            start_color: 起始颜色
+            end_color: 结束颜色
+            direction: 渐变方向 'vertical', 'horizontal', 'diagonal'
+        
+        Returns:
+            Image.Image: 渐变图像
+        """
+        if direction == 'vertical':
+            return self._create_vertical_gradient(width, height, start_color, end_color)
+        elif direction == 'horizontal':
+            return self._create_horizontal_gradient(width, height, start_color, end_color)
+        else:  # diagonal
+            return self._create_diagonal_gradient(width, height, start_color, end_color)
+    
+    def _create_vertical_gradient(self, width: int, height: int, 
+                                 color1: Tuple, color2: Tuple) -> Image.Image:
+        """创建垂直渐变"""
+        gradient = Image.new('RGB', (width, height))
+        draw = ImageDraw.Draw(gradient)
+        
+        for y in range(height):
+            ratio = y / (height - 1) if height > 1 else 0
+            r = int(color1[0] + (color2[0] - color1[0]) * ratio)
+            g = int(color1[1] + (color2[1] - color1[1]) * ratio)
+            b = int(color1[2] + (color2[2] - color1[2]) * ratio)
+            draw.line([(0, y), (width, y)], fill=(r, g, b))
+        
+        return gradient
+    
+    def _create_horizontal_gradient(self, width: int, height: int, 
+                                   color1: Tuple, color2: Tuple) -> Image.Image:
+        """创建水平渐变"""
+        gradient = Image.new('RGB', (width, height))
+        draw = ImageDraw.Draw(gradient)
+        
+        for x in range(width):
+            ratio = x / (width - 1) if width > 1 else 0
+            r = int(color1[0] + (color2[0] - color1[0]) * ratio)
+            g = int(color1[1] + (color2[1] - color1[1]) * ratio)
+            b = int(color1[2] + (color2[2] - color1[2]) * ratio)
+            draw.line([(x, 0), (x, height)], fill=(r, g, b))
+        
+        return gradient
+    
+    def _create_diagonal_gradient(self, width: int, height: int, 
+                                 color1: Tuple, color2: Tuple) -> Image.Image:
+        """创建对角渐变"""
+        if self.has_numpy:
+            return self._create_diagonal_gradient_numpy(width, height, color1, color2)
+        else:
+            return self._create_horizontal_gradient(width, height, color1, color2)  # 降级为水平渐变
+    
+    def _create_diagonal_gradient_numpy(self, width: int, height: int, 
+                                       color1: Tuple, color2: Tuple) -> Image.Image:
+        """使用numpy创建对角渐变"""
+        x = np.linspace(0, 1, width)
+        y = np.linspace(0, 1, height)
+        xx, yy = np.meshgrid(x, y)
+        
+        distance = (xx + yy) / 2.0
+        
+        r = color1[0] + (color2[0] - color1[0]) * distance
+        g = color1[1] + (color2[1] - color1[1]) * distance
+        b = color1[2] + (color2[2] - color1[2]) * distance
+        
+        gradient_array = np.stack([r, g, b], axis=-1).astype(np.uint8)
+        return Image.fromarray(gradient_array)

konabot/plugins/fx_process/image_storage.py

@@ -0,0 +1,182 @@
+import asyncio
+from dataclasses import dataclass
+from hashlib import md5
+import time
+
+from nonebot import logger
+from nonebot_plugin_apscheduler import driver
+from konabot.common.path import DATA_PATH
+import os
+from PIL import Image
+from io import BytesIO
+
+IMAGE_PATH = DATA_PATH / "temp" / "images"
+
+@dataclass
+class ImageResource:
+    filename: str
+    expire: int
+
+@dataclass
+class StorageImage:
+    name: str
+    resources: dict[str,
+                    dict[str,ImageResource]] # {群号: {QQ号: ImageResource}}
+
+class ImageStorager:
+    images_pool: dict[str,StorageImage] = {}
+
+    max_storage: int = 10 * 1024 * 1024  # 最大存储10MB
+    max_image_count: int = 200  # 最大存储图片数量
+
+    @staticmethod
+    def init():
+        if not IMAGE_PATH.exists():
+            IMAGE_PATH.mkdir(parents=True, exist_ok=True)
+
+    @staticmethod
+    def delete_path_image(name: str):
+        resource_path = IMAGE_PATH / name
+        if resource_path.exists():
+            os.remove(resource_path)
+
+    @staticmethod
+    async def clear_all_image():
+        # 清理 temp 目录下的所有图片资源
+        for file in os.listdir(IMAGE_PATH):
+            file_path = IMAGE_PATH / file
+            if file_path.is_file():
+                os.remove(file_path)
+
+    @classmethod
+    async def clear_expire_image(cls):
+        # 清理过期的图片资源，将未被删除的放入列表中，如果超过最大数量则删除最早过期的
+        remaining_images = []
+        current_time = time.time()
+        for name, storage_image in list(ImageStorager.images_pool.items()):
+            for group_id, resources in list(storage_image.resources.items()):
+                for qq_id, resource in list(resources.items()):
+                    if resource.expire < current_time:
+                        del storage_image.resources[group_id][qq_id]
+                        cls.delete_path_image(name)
+                    else:
+                        remaining_images.append((name, group_id, qq_id, resource.expire))
+            if not storage_image.resources:
+                del ImageStorager.images_pool[name]
+        # 如果剩余图片超过最大数量，按过期时间排序并删除最早过期的
+        if len(remaining_images) > ImageStorager.max_image_count:
+            remaining_images.sort(key=lambda x: x[3])  # 按过期时间排序
+            to_delete = len(remaining_images) - ImageStorager.max_image_count
+            for i in range(to_delete):
+                name, group_id, qq_id, _ = remaining_images[i]
+                resource = ImageStorager.images_pool[name].resources[group_id][qq_id]
+                del ImageStorager.images_pool[name].resources[group_id][qq_id]
+                cls.delete_path_image(name)
+        logger.info("过期图片清理完成")
+
+    @classmethod
+    def _add_to_pool(cls, filename: str, name: str, group_id: str, qq_id: str, expire: int = 36000):
+        expire_time = time.time() + expire
+        if name not in cls.images_pool:
+            cls.images_pool[name] = StorageImage(name=name,resources={})
+        if group_id not in cls.images_pool[name].resources:
+            cls.images_pool[name].resources[group_id] = {}
+        cls.images_pool[name].resources[group_id][qq_id] = ImageResource(filename=filename, expire=expire_time)
+        logger.debug(f"{cls.images_pool}")
+
+    @classmethod
+    def save_image(cls, image: bytes, name: str, group_id: str, qq_id: str) -> None:
+        """
+        以哈希值命名保存图片，并返回图片资源信息
+        """
+        # 检测图像大小，不得超过 10 MB
+        if len(image) > cls.max_storage:
+            raise ValueError("图片大小超过 10 MB 限制")
+        hash_name = md5(image).hexdigest()
+        ext = os.path.splitext(name)[1]
+        file_name = f"{hash_name}{ext}"
+        full_path = IMAGE_PATH / file_name
+        with open(full_path, "wb") as f:
+            f.write(image)
+        # 将文件写入 images_pool
+        logger.debug(f"Image saved: {file_name} for group {group_id}, qq {qq_id}")
+        cls._add_to_pool(file_name, name, group_id, qq_id)
+
+    @classmethod
+    def save_image_by_pil(cls, image: Image.Image, name: str, group_id: str, qq_id: str) -> None:
+        """
+        以哈希值命名保存图片，并返回图片资源信息
+        """
+        img_byte_arr = BytesIO()
+        # 如果图片是动图，保存为 GIF 格式
+        if getattr(image, "is_animated", False):
+            image.save(img_byte_arr, format="GIF", save_all=True, loop=0)
+        else:
+            image.save(img_byte_arr, format=image.format or "PNG")
+        img_bytes = img_byte_arr.getvalue()
+        cls.save_image(img_bytes, name, group_id, qq_id)
+
+    @classmethod
+    def load_image(cls, name: str, group_id: str, qq_id: str) -> Image:
+        logger.debug(f"Loading image: {name} for group {group_id}, qq {qq_id}")
+        if name not in cls.images_pool:
+            logger.debug(f"Image {name} not found in pool")
+            return None
+        if group_id not in cls.images_pool[name].resources:
+            logger.debug(f"No resources for group {group_id} in image {name}")
+            return None
+        # 寻找对应 QQ 号 的资源，如果没有就返回相同群下的第一个资源
+        if qq_id not in cls.images_pool[name].resources[group_id]:
+            first_qq_id = next(iter(cls.images_pool[name].resources[group_id]))
+            qq_id = first_qq_id
+        resource = cls.images_pool[name].resources[group_id][qq_id]
+        resource_path = IMAGE_PATH / resource.filename
+        logger.debug(f"Image path: {resource_path}")
+        return Image.open(resource_path)
+
+class ImageStoragerManager:
+    def __init__(self, interval: int = 300):  # 默认 5 分钟执行一次
+        self.interval = interval
+        self._clear_task = None
+        self._running = False
+        
+    async def start_auto_clear(self):
+        """启动自动任务"""
+        # 先清理一次
+        await ImageStorager.clear_all_image()
+        self._running = True
+        self._clear_task = asyncio.create_task(self._auto_clear_loop())
+            
+        logger.info(f"自动清理任务已启动，间隔: {self.interval}秒")
+    
+    async def stop_auto_clear(self):
+        """停止自动清理任务"""
+        if self._clear_task:
+            self._running = False
+            self._clear_task.cancel()
+            try:
+                await self._clear_task
+            except asyncio.CancelledError:
+                pass
+            logger.info("自动清理任务已停止")
+        else:
+            logger.warning("没有正在运行的自动清理任务")
+        
+    async def _auto_clear_loop(self):
+        """自动清理循环"""
+        while self._running:
+            try:
+                await asyncio.sleep(self.interval)
+                await ImageStorager.clear_expire_image()
+            except asyncio.CancelledError:
+                break
+            except Exception as e:
+                logger.error(f"定时清理失败: {e}")
+
+image_manager = ImageStoragerManager(interval=300)  # 每5分钟清理一次
+
+@driver.on_startup
+async def init_image_storage():
+    ImageStorager.init()
+    # 启用定时任务清理过期图片
+    await image_manager.start_auto_clear()

konabot/plugins/fx_process/math_helper.py

@@ -0,0 +1,125 @@
+import cv2
+from nonebot import logger
+import numpy as np
+
+from shapely.geometry import Polygon
+from shapely.ops import unary_union
+
+def fix_with_shapely(contours: list) -> np.ndarray:
+    """
+    使用Shapely库处理复杂自相交
+    """
+    fixed_polygons = []
+    for contour in contours:
+        # 转换输入为正确的格式
+        contour_array = contour.reshape(-1, 2)
+        # 转换为Shapely多边形
+        polygon = Polygon(contour_array)
+        if not polygon.is_valid:
+            polygon = polygon.buffer(0)
+        fixed_polygons.append(polygon)
+    # 接下来把所有轮廓合并为一个
+    if len(fixed_polygons) >= 1:
+        merged_polygon = unary_union(fixed_polygons)
+        if merged_polygon.geom_type == 'Polygon':
+            merged_points = np.array(merged_polygon.exterior.coords, dtype=np.int32)
+        elif merged_polygon.geom_type == 'MultiPolygon':
+            largest = max(merged_polygon.geoms, key=lambda p: p.area)
+            merged_points = np.array(largest.exterior.coords, dtype=np.int32)
+        return [merged_points.reshape(-1, 1, 2)]
+    else:
+        logger.warning("No valid contours found after fixing with Shapely.")
+        return [np.array([], dtype=np.int32).reshape(0, 1, 2)]
+
+def expand_contours(contours, stroke_width):
+    """
+    将轮廓向外扩展指定宽度
+    
+    参数:
+        contours: OpenCV轮廓列表
+        stroke_width: 扩展宽度（像素）
+        
+    返回:
+        扩展后的轮廓列表
+    """
+    expanded_contours = []
+    
+    for cnt in contours:
+        # 将轮廓转换为点列表
+        points = cnt.reshape(-1, 2).astype(np.float32)
+        n = len(points)
+        
+        if n < 3:
+            continue  # 至少需要3个点才能形成多边形
+            
+        expanded_points = []
+        
+        for i in range(n):
+            # 获取当前点、前一个点和后一个点
+            p_curr = points[i]
+            p_prev = points[(i - 1) % n]
+            p_next = points[(i + 1) % n]
+            
+            # 计算两条边的向量
+            v1 = p_curr - p_prev  # 前一条边（从prev到curr）
+            v2 = p_next - p_curr  # 后一条边（从curr到next）
+            
+            # 归一化
+            norm1 = np.linalg.norm(v1)
+            norm2 = np.linalg.norm(v2)
+            
+            if norm1 == 0 or norm2 == 0:
+                # 如果有零向量，直接沿着法线方向扩展
+                edge_dir = np.array([0, 0])
+                if norm1 > 0:
+                    edge_dir = v1 / norm1
+                elif norm2 > 0:
+                    edge_dir = v2 / norm2
+                normal = np.array([-edge_dir[1], edge_dir[0]])
+                expanded_point = p_curr + normal * stroke_width
+            else:
+                # 归一化向量
+                v1_norm = v1 / norm1
+                v2_norm = v2 / norm2
+                
+                # 计算两条边的单位法向量（指向多边形外部）
+                n1 = np.array([-v1_norm[1], v1_norm[0]])
+                n2 = np.array([-v2_norm[1], v2_norm[0]])
+                
+                # 计算角平分线方向（两个法向量的和）
+                bisector = n1 + n2
+                
+                # 计算平分线的长度
+                bisector_norm = np.linalg.norm(bisector)
+                
+                if bisector_norm == 0:
+                    # 如果两条边平行（同向或反向），取任一法线方向
+                    expanded_point = p_curr + n1 * stroke_width
+                else:
+                    # 归一化平分线
+                    bisector_normalized = bisector / bisector_norm
+                    
+                    # 计算偏移距离（考虑夹角）
+                    # 使用余弦定理计算正确的偏移距离
+                    cos_angle = np.dot(v1_norm, v2_norm)
+                    angle = np.arccos(np.clip(cos_angle, -1.0, 1.0))
+                    
+                    if abs(np.pi - angle) < 1e-6:  # 近似平角
+                        # 接近直线的情况
+                        offset_distance = stroke_width
+                    else:
+                        # 计算正确的偏移距离
+                        offset_distance = stroke_width / np.sin(angle / 2)
+                    
+                    # 计算扩展点
+                    expanded_point = p_curr + bisector_normalized * offset_distance
+            
+            expanded_points.append(expanded_point)
+        
+        # 将扩展后的点转换为整数坐标
+        expanded_cnt = np.array(expanded_points, dtype=np.float32).reshape(-1, 1, 2)
+        expanded_contours.append(expanded_cnt.astype(np.int32))
+
+    expanded_contours = fix_with_shapely(expanded_contours)
+
+    return expanded_contours