entari-plugin-hyw 4.0.0rc7__py3-none-any.whl → 4.0.0rc8__py3-none-any.whl

hyw_core/core.py

@@ -0,0 +1,322 @@
+"""
+hyw_core.core - Main HywCore Class
+
+Provides the unified LLM query interface and search capabilities.
+"""
+
+import asyncio
+import time
+from dataclasses import dataclass, field
+from typing import Dict, List, Any, Optional, Callable, Awaitable
+
+from loguru import logger
+from openai import AsyncOpenAI
+
+from .config import HywCoreConfig, ModelConfig
+from .pipeline import ModularPipeline
+from .search import SearchService
+from .stages.base import StageContext
+
+
+@dataclass
+class QueryRequest:
+    """Request for the query interface."""
+    user_input: str
+    images: List[str] = field(default_factory=list)  # base64 encoded images
+    conversation_history: List[Dict] = field(default_factory=list)
+    model_name: Optional[str] = None  # Override model
+    
+    # Optional callbacks
+    send_notification: Optional[Callable[[str], Awaitable[None]]] = None
+
+
+@dataclass
+class QueryResponse:
+    """Response from the query interface."""
+    success: bool
+    content: str  # Markdown response
+    image_path: Optional[str] = None  # Path to rendered image
+    
+    # Statistics
+    usage: Dict[str, int] = field(default_factory=dict)
+    cost: float = 0.0
+    total_time: float = 0.0
+    
+    # References
+    references: List[Dict[str, Any]] = field(default_factory=list)
+    page_references: List[Dict[str, Any]] = field(default_factory=list)
+    image_references: List[Dict[str, Any]] = field(default_factory=list)
+    
+    # Trace information
+    stages_trace: Dict[str, Any] = field(default_factory=dict)
+    
+    # Error handling
+    error: Optional[str] = None
+    should_refuse: bool = False
+    refuse_reason: str = ""
+    
+    # Debug/Save
+    web_results: List[Dict[str, Any]] = field(default_factory=list)
+    stages_used: List[Dict[str, Any]] = field(default_factory=list)
+
+
+
+class HywCore:
+    """
+    HYW Core Service.
+    
+    Provides the unified LLM query interface (/q command) and search capabilities.
+    
+    Usage:
+        from hyw_core import HywCore, HywCoreConfig, QueryRequest
+        
+        config = HywCoreConfig.from_yaml("config.yaml")
+        core = HywCore(config)
+        
+        response = await core.query(QueryRequest(
+            user_input="What is Python?",
+            images=[],
+            conversation_history=[]
+        ))
+    """
+    
+    def __init__(
+        self, 
+        config: HywCoreConfig,
+        send_func: Optional[Callable[[str], Awaitable[None]]] = None
+    ):
+        """
+        Initialize HywCore.
+        
+        Args:
+            config: HywCoreConfig instance
+            send_func: Optional callback for sending notifications
+        """
+        self.config = config
+        self._send_func = send_func
+        
+        # Create OpenAI client
+        self._client = AsyncOpenAI(
+            api_key=config.api_key,
+            base_url=config.base_url if config.base_url else None
+        )
+        
+        # Create search service
+        self._search_service = SearchService(config)
+        
+        # Create pipeline
+        self._pipeline = ModularPipeline(
+            config=config,
+            search_service=self._search_service,
+            send_func=send_func
+        )
+        
+        # Create renderer (lazy init)
+        self._renderer = None
+        
+        logger.info("HywCore initialized")
+    
+    async def _ensure_renderer(self):
+        """Lazy initialize renderer."""
+        if self._renderer is None:
+            from .browser_control import ContentRenderer
+            self._renderer = ContentRenderer(headless=self.config.headless)
+            await self._renderer.start()
+    
+    async def query(
+        self, 
+        request: QueryRequest,
+        output_path: Optional[str] = None
+    ) -> QueryResponse:
+        """
+        Unified query interface.
+        
+        This is the main entry point for /q commands.
+        
+        Args:
+            request: QueryRequest with user input, images, history
+            output_path: Optional path to save rendered image
+            
+        Returns:
+            QueryResponse with content, rendered image path, and metadata
+        """
+        start_time = time.time()
+        
+        try:
+            # Override model if specified
+            model_name = request.model_name or self.config.model_name
+            
+            # Use notification callback from request if provided
+            send_func = request.send_notification or self._send_func
+            if send_func and self._pipeline._send_func != send_func:
+                self._pipeline._send_func = send_func
+            
+            # Execute pipeline
+            result = await self._pipeline.execute(
+                user_input=request.user_input,
+                conversation_history=request.conversation_history,
+                model_name=model_name,
+                images=request.images if request.images else None
+            )
+            
+            total_time = time.time() - start_time
+            
+            # Check for refusal
+            if result.get("should_refuse"):
+                return QueryResponse(
+                    success=True,
+                    content="",
+                    should_refuse=True,
+                    refuse_reason=result.get("refuse_reason", ""),
+                    total_time=total_time
+                )
+            
+            # Extract response data
+            content = result.get("llm_response", "")
+            structured = result.get("structured_response", {})
+            billing = result.get("billing_info", {})
+            
+            usage = {
+                "input_tokens": billing.get("input_tokens", 0),
+                "output_tokens": billing.get("output_tokens", 0)
+            }
+            
+            # Calculate cost
+            model_cfg = self.config.get_model_config("main")
+            cost = (
+                usage["input_tokens"] * (model_cfg.input_price or 0) / 1_000_000 +
+                usage["output_tokens"] * (model_cfg.output_price or 0) / 1_000_000
+            )
+            
+            # Build response
+            response = QueryResponse(
+                success=True,
+                content=content,
+                usage=usage,
+                cost=cost,
+                total_time=total_time,
+                references=structured.get("references", []),
+                page_references=structured.get("page_references", []),
+                image_references=structured.get("image_references", []),
+                stages_trace=result.get("trace", {}),
+                web_results=result.get("web_results", []),
+                stages_used=result.get("stages_used", [])
+            )
+            
+            # Render image if output path provided
+            if output_path and content:
+                await self._ensure_renderer()
+                
+                render_success = await self._renderer.render(
+                    markdown_content=content,
+                    output_path=output_path,
+                    stats=result.get("stats", {}),
+                    references=result.get("references", []),
+                    page_references=result.get("page_references", []),
+                    image_references=result.get("image_references", []),
+                    stages_used=result.get("stages_used", []),
+                    theme_color=self.config.theme_color
+                )
+                
+                if render_success:
+                    response.image_path = output_path
+            
+            return response
+            
+        except Exception as e:
+            logger.error(f"HywCore query failed: {e}")
+            return QueryResponse(
+                success=False,
+                content="",
+                error=str(e),
+                total_time=time.time() - start_time
+            )
+    
+    async def search(
+        self,
+        queries: List[str],
+        engine: Optional[str] = None,
+        limit: int = 10
+    ) -> List[List[Dict[str, Any]]]:
+        """
+        Independent search interface.
+        
+        For future step-by-step search functionality.
+        
+        Args:
+            queries: List of search queries
+            engine: Optional search engine override
+            limit: Results per query
+            
+        Returns:
+            List of search results for each query
+        """
+        # TODO: Support engine override per-call
+        return await self._search_service.search_batch(queries)
+
+    async def screenshot(self, url: str) -> Optional[str]:
+        """
+        Capture full page screenshot of a URL.
+        Returns: base64 string or None
+        """
+        # Default to full_page=True as requested for /w command
+        return await self._search_service.screenshot_url(url, full_page=True)
+    
+    async def screenshot_batch(self, urls: List[str]) -> List[Optional[str]]:
+        """
+        Capture full page screenshots of multiple URLs concurrently.
+        Returns: list of base64 strings (None for failed ones)
+        """
+        return await self._search_service.screenshot_urls_batch(urls, full_page=True)
+    
+    async def fetch_pages(
+        self,
+        urls: List[str],
+        include_screenshot: bool = False
+    ) -> List[Dict[str, Any]]:
+        """
+        Fetch multiple pages.
+        
+        Args:
+            urls: List of URLs to fetch
+            include_screenshot: Whether to capture screenshots
+            
+        Returns:
+            List of page data dicts
+        """
+        return await self._search_service.fetch_pages_batch(
+            urls, 
+            include_screenshot=include_screenshot
+        )
+    
+    async def render(
+        self,
+        markdown_content: str,
+        output_path: str,
+        **kwargs
+    ) -> bool:
+        """
+        Render markdown to image.
+        
+        Args:
+            markdown_content: Markdown to render
+            output_path: Path to save image
+            **kwargs: Additional render options
+            
+        Returns:
+            True if successful
+        """
+        await self._ensure_renderer()
+        return await self._renderer.render(
+            markdown_content=markdown_content,
+            output_path=output_path,
+            theme_color=kwargs.pop("theme_color", self.config.theme_color),
+            **kwargs
+        )
+    
+    async def close(self):
+        """Close all resources."""
+        if self._renderer:
+            await self._renderer.close()
+        await self._pipeline.close()
+        logger.info("HywCore closed")

hyw_core/definitions.py

@@ -0,0 +1,83 @@
+"""
+Centralized Definitions
+
+All global prompts and tool definitions for the pipeline stages.
+"""
+
+from typing import Dict, Any
+
+# =============================================================================
+# PROMPTS
+# =============================================================================
+
+
+
+
+
+SUMMARY_REPORT_SP = """# 你是一个信息整合专家 (Summary Agent).
+你需要根据用户问题、搜索结果和网页详细内容，生成最终的回答.
+如果用户发送你好或空内容回应你好即可.
+
+## 内容审查
+以下情况调用 `refuse_answer`:
+- 内容涉及隐喻政治事件、任务、现役国家领导人
+- r18+、r18g(但不包含正常galgame、科普等)
+
+## 过程要求
+- 用户要求的回复语言(包裹在 language 标签内)
+```language
+{language}
+```
+- 字数控制在600字以内, 百科式风格, 语言严谨不啰嗦.
+- 视觉信息: 输入中如果包含自动获取的网页截图，请分析图片中的信息作为参考.
+- 注意分辨搜索内容是否和用户问题有直接关系, 避免盲目相信混为一谈.
+- 正文格式: 
+  - 先给出一个 `# `大标题约 8-10 个字, 不要有多余废话, 不要直接回答用户的提问.
+  - 然后紧接着给出一个 <summary>...</summary>, 除了给出一个约 100 字的纯文本简介, 介绍本次输出的长文的清晰、重点概括.
+  - 随后开始详细二级标题 + markdown 正文, 语言描绘格式丰富多样, 简洁准确可信.
+  - 请不要给出过长的代码、表格列数等, 只讲重点和准确的数据.
+  - 不支持渲染: 链接, 图片链接, mermaid
+  - 支持渲染: 公式, 代码高亮, 只在需要的时候给出.
+  - 图片链接、链接框架会自动渲染出, 你无需显式给出.
+- 引用:
+  > 重要: 所有正文内容必须基于实际信息, 保证百分百真实度
+  - 信息来源已按获取顺序编号为 [1], [2], [3]...
+  - 优先引用优质 fetch 抓取的页面的资源, 但如果抓取到需要登录、需要验证码、需要跳转到其他网站等无法获取的资源, 则不引用此资源
+  - 正文中直接使用 [1] 格式引用, 只引用对回答有帮助的来源, 只使用官方性较强的 wiki、官方网站、资源站等等, 不使用第三方转载新闻网站.
+  - 无需给出参考文献列表, 系统会自动生成
+"""
+
+
+# =============================================================================
+# VISION DESCRIPTION PROMPT
+# =============================================================================
+
+
+
+
+# =============================================================================
+# TOOL DEFINITIONS
+# =============================================================================
+
+def get_refuse_answer_tool() -> Dict[str, Any]:
+    """Tool for refusing to answer inappropriate content."""
+    return {
+        "type": "function",
+        "function": {
+            "name": "refuse_answer",
+            "description": "违规内容拒绝回答",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "reason": {"type": "string", "description": "拒绝回答的原因（展示给用户）"},
+                },
+                "required": ["reason"],
+            },
+        },
+    }
+
+
+
+
+
+