Magic-API MCP Server

"""Web文档知识库模块 - 用于搜索web-docs目录下的Markdown文档。 这个模块提供搜索web-docs目录下Markdown文档的功能。 """ from __future__ import annotations import os import re from pathlib import Path from typing import Any, Dict, List, Optional from datetime import datetime def read_md_file_content(file_path: str) -> Dict[str, Any]: """读取markdown文件的内容，解析frontmatter和正文。 Args: file_path: markdown文件路径 Returns: 包含文件信息的字典 """ try: with open(file_path, 'r', encoding='utf-8') as f: content = f.read() # 解析frontmatter（如果有） frontmatter_match = re.match(r'^---\n(.*?)\n---\n(.*)', content, re.DOTALL) frontmatter = {} body_content = content if frontmatter_match: frontmatter_str = frontmatter_match.group(1) body_content = frontmatter_match.group(2) # 简单解析frontmatter for line in frontmatter_str.split('\n'): if ':' in line: key, value = line.split(':', 1) key = key.strip() value = value.strip().strip('"\'') frontmatter[key] = value # 尝试从文件路径推断分类 relative_path = os.path.relpath(file_path) path_parts = relative_path.split(os.sep) # 提取目录名作为分类信息 category = "" subcategory = "" if len(path_parts) >= 2: # 移除数字前缀，保留有意义的分类名 category = re.sub(r'^\d+\.', '', path_parts[1]).strip() if len(path_parts) >= 3: subcategory = re.sub(r'^\d+\.', '', path_parts[2]).strip() return { "title": frontmatter.get("title", ""), "date": frontmatter.get("date", ""), "permalink": frontmatter.get("permalink", ""), "content": body_content, "raw_content": content, "file_path": file_path, "category": category, "subcategory": subcategory, "relative_path": relative_path } except Exception as e: print(f"Error reading file {file_path}: {str(e)}") return { "title": "", "date": "", "permalink": "", "content": "", "raw_content": "", "file_path": file_path, "category": "", "subcategory": "", "relative_path": "" } def load_all_web_docs(base_path: Optional[str] = None) -> List[Dict[str, Any]]: """加载web-docs目录下的所有markdown文档。 Args: base_path: web-docs目录的基础路径 Returns: 包含所有文档信息的列表 """ if base_path is None: # First, try the development/relative path approach current_dir = Path(__file__).parent dev_path = current_dir.parent.parent / "web-docs" if os.path.exists(dev_path): base_path = str(dev_path) else: # For installation, we need to handle the package data differently # The MANIFEST.in and pyproject.toml should ensure the files are included # Let's try to find the web-docs relative to the package import magicapi_tools package_dir = Path(magicapi_tools.__file__).parent installed_path = package_dir / "web-docs" if os.path.exists(installed_path): base_path = str(installed_path) else: # If both approaches fail, return empty list return [] if isinstance(base_path, Path): base_path = str(base_path) docs = [] # Walk through the directory to find all markdown files for root, dirs, files in os.walk(base_path): for file in files: if file.endswith('.md'): file_path = os.path.join(root, file) doc_info = read_md_file_content(file_path) docs.append(doc_info) return docs def search_web_docs(keyword: str, base_path: Optional[str] = None) -> List[Dict[str, Any]]: """在web-docs的markdown文档中搜索关键词。 Args: keyword: 搜索关键词 base_path: web-docs目录的基础路径 Returns: 匹配的文档列表 """ docs = load_all_web_docs(base_path) results = [] keyword_lower = keyword.lower() for doc in docs: # 在标题、分类、子分类和内容中搜索关键词 title_match = keyword_lower in doc.get("title", "").lower() category_match = keyword_lower in doc.get("category", "").lower() subcategory_match = keyword_lower in doc.get("subcategory", "").lower() content_match = keyword_lower in doc.get("content", "").lower() if title_match or category_match or subcategory_match or content_match: result = { "category": "web_docs", "topic": doc.get("title", "Untitled"), "title": doc.get("title", ""), "description": doc.get("category", "") + " / " + doc.get("subcategory", ""), "content_preview": doc.get("content", "")[:200] + "..." if len(doc.get("content", "")) > 200 else doc.get("content", ""), "file_path": doc.get("relative_path", ""), "type": "文档", "full_content": doc.get("content", ""), "permalink": doc.get("permalink", ""), "date": doc.get("date", "") } results.append(result) return results # 预加载web-docs内容以提高搜索性能 WEB_DOCS_KNOWLEDGE = load_all_web_docs() def get_web_docs_knowledge() -> List[Dict[str, Any]]: """获取所有web-docs知识。 Returns: web-docs知识列表 """ return WEB_DOCS_KNOWLEDGE def search_web_docs_by_keyword(keyword: str) -> List[Dict[str, Any]]: """根据关键词搜索web-docs内容。 Args: keyword: 搜索关键词 Returns: 搜索结果列表 """ return search_web_docs(keyword)