entari-plugin-hyw 0.3.5__py3-none-any.whl → 4.0.0rc14__py3-none-any.whl

entari_plugin_hyw/tools.py

@@ -1,236 +0,0 @@
-import asyncio
-import httpx
-import json
-from typing import List, Optional
-from langchain_core.tools import tool
-from langchain_openai import ChatOpenAI
-from loguru import logger
-
-from .compressor import WebpageCompressor
-
-
-# 全局搜索引擎配置
-_global_search_engines = ["auto"]
-
-# 全局压缩器配置
-_global_compressor: Optional[WebpageCompressor] = None
-
-
-def set_global_search_engines(engines: List[str]):
-    """设置全局搜索引擎列表"""
-    global _global_search_engines
-    _global_search_engines = engines
-
-
-def set_global_compressor_llm(llm: ChatOpenAI):
-    """设置全局压缩器LLM"""
-    global _global_compressor
-    _global_compressor = WebpageCompressor(llm)
-    logger.info("[工具] 压缩器已初始化")
-    
-    
-    
-# 智能搜索工具
-@tool
-async def web_search(query: str) -> str:
-    """
-    - web_search 每个查询只能包含一个的关键词建议在 1-2 个词之间 不超过 3 个词
-    - 多引擎并发搜索工具，使用配置文件指定的搜索引擎进行搜索
-    - 支持同时搜索多个搜索引擎并合并结果
-    - 搜索引擎配置在 entari.yml 中的 search_engines 字段，如 ["duckduckgo::exact", "duckduckgo"]，::exact 表示精确搜索
-    - 自动去重和合并来自不同搜索引擎的结果，提供更全面的信息
-    智能搜索工具, 返回JSON格式文本结果
-    query: 要搜索的关键词
-    """
-
-    async def ddsg_search(keyword: str, backend: str, is_exact: bool = False) -> dict:
-        """单个搜索引擎搜索"""
-        from ddgs import DDGS
-        
-        def _sync_search():
-            # 简化的搜索逻辑
-            search_keyword = f'"{keyword}"' if is_exact else keyword
-            logger.info(f"DDGS 搜索: 关键词='{search_keyword}', 引擎='{backend}'")
-            return DDGS().text(search_keyword, safesearch="off", timelimit="y", max_results=7, backend=backend)
-        
-        try:
-            results = await asyncio.to_thread(_sync_search)
-            logger.info(f"搜索完成: {backend} -> {len(results)} 个结果")
-            return {"engine": backend, "results": results, "success": True}
-        except Exception as e:
-            logger.error(f"搜索失败: {backend} -> {e}")
-            return {"engine": backend, "results": [], "success": False, "error": str(e)}
-
-    def parse_engines(engines_list: List[str]) -> List[tuple]:
-        """解析搜索引擎列表，提取引擎名称和是否精确搜索"""
-        parsed = []
-        for engine in engines_list:
-            if "::exact" in engine:
-                base_engine = engine.replace("::exact", "")
-                parsed.append((base_engine, True))
-            else:
-                parsed.append((engine, False))
-        return parsed
-
-    try:
-        # 使用全局配置的搜索引擎列表
-        if len(_global_search_engines) == 1:
-            logger.info(f"使用单引擎搜索模式: {_global_search_engines[0]}")
-            parsed_engines = parse_engines(_global_search_engines)
-            engine_name, is_exact = parsed_engines[0]
-            single_result = await ddsg_search(query, engine_name, is_exact)
-            # 对精确模式进行客户端二次过滤，确保只保留包含完整查询词的结果
-            filtered_results = single_result.get("results", [])
-            if is_exact:
-                q = (query or "").lower()
-                def _hit(res: dict) -> bool:
-                    title = (res.get("title") or "").lower()
-                    body = (res.get("body") or "").lower()
-                    href = (res.get("href") or "").lower()
-                    return (q in title) or (q in body) or (q in href)
-                filtered_results = [r for r in filtered_results if _hit(r)]
-            
-            result_data = {
-                "query": query,
-                "results": filtered_results,
-                "search_type": "single_engine",
-                "engine": _global_search_engines[0],
-                "search_mode": "exact" if is_exact else "normal",
-                "success": single_result.get("success", False),
-                "exact_matched": (len(filtered_results) > 0) if is_exact else None,
-                "note": ("没有精确匹配结果" if (is_exact and len(filtered_results) == 0) else None),
-                "error": single_result.get("error")
-            }
-            
-            return json.dumps(result_data, ensure_ascii=False, indent=2)
-        
-        # 使用全局配置的搜索引擎列表进行多引擎搜索
-        search_engines = _global_search_engines
-        parsed_engines = parse_engines(search_engines)
-        logger.info(f"开始多引擎并发搜索: 查询='{query}', 引擎={search_engines}")
-        
-        # 并发执行所有搜索引擎的搜索
-        search_tasks = [ddsg_search(query, engine_name, is_exact) for engine_name, is_exact in parsed_engines]
-        search_results = await asyncio.gather(*search_tasks, return_exceptions=True)
-        
-        # 简化的结果处理
-        merged_results = {}
-        successful_engines = []
-        failed_engines = []
-        
-        for i, result in enumerate(search_results):
-            if isinstance(result, Exception) or not isinstance(result, dict):
-                continue
-                
-            engine_config = parsed_engines[i] if i < len(parsed_engines) else ("unknown", False)
-            engine_name, is_exact = engine_config
-            display_name = f"{engine_name}(精确)" if is_exact else f"{engine_name}(普通)"
-            
-            # 简单格式化结果 + 精确模式客户端二次过滤
-            raw_results = result.get("results", [])
-            if is_exact:
-                q = (query or "").lower()
-                def _hit(res: dict) -> bool:
-                    title = (res.get("title") or "").lower()
-                    body = (res.get("body") or "").lower()
-                    href = (res.get("href") or "").lower()
-                    return (q in title) or (q in body) or (q in href)
-                raw_results = [r for r in raw_results if _hit(r)]
-                if len(raw_results) == 0:
-                    logger.info(f"精确模式无匹配: 查询='{query}', 引擎='{engine_name}'")
-            formatted_results = []
-            for j, res in enumerate(raw_results):
-                formatted_results.append({
-                    "title": res.get("title", ""),
-                    "href": res.get("href", ""),
-                    "body": res.get("body", ""),
-                    "source": display_name
-                })
-            
-            # 使用不同的键名来区分同一引擎的不同搜索模式
-            result_key = f"{engine_name}_exact" if is_exact else engine_name
-            merged_results[result_key] = {
-                "results": formatted_results,
-                "success": result.get("success", False),
-                "display_name": display_name,
-                "engine_name": engine_name,
-                "is_exact": is_exact,
-                "exact_matched": (len(formatted_results) > 0) if is_exact else None,
-                "note": ("没有精确匹配结果" if (is_exact and len(formatted_results) == 0) else None)
-            }
-            
-            if result.get("success", False):
-                successful_engines.append(result_key)
-            else:
-                failed_engines.append(result_key)
-        
-        # 不去重，保留所有结果，让LLM能够看到不同搜索模式的差异
-        unique_results = []
-        for result in merged_results.values():
-            unique_results.extend(result["results"])
-        
-        # 构建结果
-        result_data = {
-            "query": query,
-            "merged_results": unique_results,
-            "engine_results": merged_results,
-            "search_type": "multi_engine",
-            "stats": {
-                "total_results": len(unique_results),
-                "successful_engines_count": len(successful_engines),
-                "failed_engines_count": len(failed_engines)
-            }
-        }
-        
-        logger.info(f"多引擎搜索完成: 成功引擎 {len(successful_engines)}/{len(search_engines)}, 总结果: {unique_results}")
-        return json.dumps(result_data, ensure_ascii=False, indent=2)
-        
-    except Exception as e:
-        error_result = {
-            "query": query,
-            "search_engines": {
-                "successful": [],
-                "failed": _global_search_engines,
-                "total_requested": len(_global_search_engines)
-            },
-            "merged_results": [],
-            "engine_results": {},
-            "search_type": "multi_engine",
-            "stats": {
-                "total_results": 0,
-                "successful_engines_count": 0,
-                "failed_engines_count": len(_global_search_engines)
-            },
-            "error": str(e)
-        }
-        logger.error(f"多引擎搜索失败: {e}")
-        return json.dumps(error_result, ensure_ascii=False, indent=2)
-
-
-@tool
-async def jina_fetch_webpage(url: str) -> str:
-    """
-    输入网址, 获取网页内容
-    """
-    try:
-        async with httpx.AsyncClient(timeout=30.0) as client:
-            logger.info(f"Jina 获取页面: URL='{url}'")
-            resp = await client.get(f"https://r.jina.ai/{url}", headers={"X-Engine": "direct"})
-            if resp.status_code == 200:
-                content = resp.text
-                logger.info(f"网页获取成功，原始长度: {len(content)} 字符")
-                
-                # 如果压缩器已初始化，则使用压缩器压缩内容
-                if _global_compressor is not None:
-                    try:
-                        content = await _global_compressor.compress_webpage(content, url)
-                        logger.info(f"网页内容压缩完成，最终长度: {len(content)} 字符")
-                    except Exception as e:
-                        logger.error(f"网页压缩失败: {e}，返回原始内容")
-                
-                return content
-            else:
-                return f"获取网页失败，状态码: {resp.status_code}"
-    except Exception as e:
-        return f"获取网页失败: {str(e)}"
-

entari_plugin_hyw/vision.py

@@ -1,35 +0,0 @@
-import base64
-from typing import Optional
-from langchain_openai import ChatOpenAI
-from langchain_core.messages import HumanMessage, SystemMessage
-from loguru import logger
-
-
-async def vision_expert_analysis(vision_llm: ChatOpenAI, image_data: bytes, query: str = "") -> str:
-    """视觉专家分析工具"""
-    logger.info("调用视觉专家分析")
-    try:
-        img_data = base64.b64encode(image_data).decode()
-        message_content = [
-            {"type": "text", "text": f"请分析这张图片内容。用户问题：{query}" if query else "请详细分析这张图片的内容"},
-            {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{img_data}"}}
-        ]
-        
-        vision_prompt = """你是视觉分析专家，请分析图片内容：
-
-- 一大段话详尽的描述主要内容, 
-- 如果出现文字, 请给出所有文字内容
-
-输出格式：
-第一张图描述了... 此外...
-...
-"""
-        
-        result = await vision_llm.ainvoke([
-            SystemMessage(content=vision_prompt),
-            HumanMessage(content=message_content)
-        ])
-        logger.info(f"视觉专家分析完成: {result}")
-        return str(result.content) if hasattr(result, 'content') else str(result)
-    except Exception as e:
-        return f"视觉分析失败: {str(e)}"

entari_plugin_hyw-0.3.5.dist-info/METADATA

@@ -1,112 +0,0 @@
-Metadata-Version: 2.4
-Name: entari_plugin_hyw
-Version: 0.3.5
-Summary: Use large language models to interpret chat messages
-Author-email: kumoSleeping <zjr2992@outlook.com>
-License: MIT
-Project-URL: Homepage, https://github.com/kumoSleeping/entari-plugin-hyw
-Project-URL: Repository, https://github.com/kumoSleeping/entari-plugin-hyw
-Project-URL: Issues, https://github.com/kumoSleeping/entari-plugin-hyw/issues
-Keywords: entari,llm,ai,bot,chat
-Classifier: Development Status :: 3 - Alpha
-Classifier: Intended Audience :: Developers
-Classifier: License :: OSI Approved :: MIT License
-Classifier: Programming Language :: Python :: 3
-Classifier: Programming Language :: Python :: 3.8
-Classifier: Programming Language :: Python :: 3.9
-Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3.12
-Requires-Python: >=3.8
-Description-Content-Type: text/markdown
-Requires-Dist: arclet-entari[full]>=0.16.5
-Requires-Dist: langchain-openai
-Requires-Dist: httpx
-Requires-Dist: bs4>=0.0.2
-Requires-Dist: langchain>=1.0.1
-Requires-Dist: langgraph>=1.0.0
-Requires-Dist: entari-plugin-server>=0.5.0
-Requires-Dist: satori-python-adapter-onebot11>=0.2.5
-Requires-Dist: nekobox>=0.1.6
-Requires-Dist: ddgs>=9.6.1
-
-# entari-plugin-hyw
-
-
-![License](https://img.shields.io/badge/License-MIT-green.svg) ![PyPI](https://img.shields.io/badge/PyPI-Available-brightgreen.svg)
-
-**使用大语言模型在聊天环境解释大家的hyw**
-
-### 目前的局限
-
-> 目前仅支持 satori-python-adapter-onebot11 使用此插件, 更多适配请等一会...
-
-> 目前没有考虑 video , 小程序 等复杂消息类型的处理
-
-
-## 🚀 快速开始
-
-### 安装
-
-```bash
-pip install entari-plugin-hyw
-```
-
-### 配置
-
-在你的 `entari.yml` 配置文件中根据您的情况添加以下配置：
-
-```yaml
-plugins:
-  entari_plugin_hyw:
-    hyw_command_name: ["/hyw", "hyw"]
-    
-    # 文本模型配置
-    text_llm_model_name: "qwen3-max"
-    text_llm_api_key: "your-api-key"
-    text_llm_model_base_url: "https://xxx/v1"
-    text_llm_enable_search: false
-    
-    # 视觉模型配置
-    vision_llm_model_name: "qwen3-vl-plus"
-    vision_llm_api_key: "your-api-key"
-    vision_llm_model_base_url: "https://xxx/v1"
-    vision_llm_enable_search: false
-```
-
-## 📖 使用方法
-
-```
-hyw 什么是人工智能？
-hyw [图片]
-```
-
-
-```
-[引用消息[图片, 文字]] hyw
-[引用消息[图片, 文字]] [At] hyw 什么是人工智能？ [图片]
-```
-
-> 自动屏蔽 At 元素
-
-## ⚙️ 配置参数
-
-| 参数 | 类型 | 默认值 | 说明 |
-|------|------|--------|------|
-| `hyw_command_name` | `str \| List[str]` | `"hyw"` | 触发命令名称 |
-| `text_llm_model_name` | `str` | - | 文本模型名称 |
-| `text_llm_api_key` | `str` | - | 文本模型 API 密钥 |
-| `text_llm_model_base_url` | `str` | - | 文本模型 API 地址 |
-| `text_llm_temperature` | `float` | `0.4` | 文本模型温度参数 |
-| `text_llm_enable_search` | `bool` | `false` | 是否启用搜索功能 |
-| `vision_llm_model_name` | `str` | - | 视觉模型名称 |
-| `vision_llm_api_key` | `str` | - | 视觉模型 API 密钥 |
-| `vision_llm_model_base_url` | `str` | - | 视觉模型 API 地址 |
-| `vision_llm_temperature` | `float` | `0.4` | 视觉模型温度参数 |
-| `vision_llm_enable_search` | `bool` | `false` | 是否启用视觉搜索 |
-| `hyw_prompt` | `str` | 默认提示词 | 自定义系统提示词 |
-
-
-
-
-

entari_plugin_hyw-0.3.5.dist-info/RECORD

@@ -1,9 +0,0 @@
-entari_plugin_hyw/__init__.py,sha256=17FyK84OlwkJTD6mtfC_bQBAY2T-B3g_zhmCo2M7EfU,5040
-entari_plugin_hyw/agent.py,sha256=rSQHZmJdUKEHFciSh-JP3mg4hBn5zrRVQsnrcfAKWJU,20984
-entari_plugin_hyw/compressor.py,sha256=3TYXXYRFi0iktMDsZUnK2v1gaBCTLDCvX89aWU4RZnU,2252
-entari_plugin_hyw/tools.py,sha256=xuhlrhB8DaMjUzrGAEwouATFycbDPPKgugeSlLG17MY,10249
-entari_plugin_hyw/vision.py,sha256=oqPECLS4d6doUn9qSHe4d5075qpitDd9tR2IxgM3ISg,1322
-entari_plugin_hyw-0.3.5.dist-info/METADATA,sha256=9xy1UfLypvkpWKXwaEWwQK4eu4lYgMUetmf5BqoOhEM,3464
-entari_plugin_hyw-0.3.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-entari_plugin_hyw-0.3.5.dist-info/top_level.txt,sha256=TIDsn6XPs6KA5e3ezsE65JoXsy03ejDdrB41I4SPjmo,18
-entari_plugin_hyw-0.3.5.dist-info/RECORD,,