基于 Python + LangChain + MCP（新闻查询）

创建 news_server.py MCP 服务器

"""新闻搜索 MCP Server - 使用百度新闻搜索"""
import re
import urllib.parse
import requests
from mcp.server.fastmcp import FastMCP

mcp = FastMCP("News")

HEADERS = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 "
                  "(KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
    "Accept-Language": "zh-CN,zh;q=0.9",
}


def _search_baidu_news(keyword: str, count: int = 5) -> list[dict]:
    """从百度新闻搜索结果中提取新闻列表"""
    url = "https://www.baidu.com/s"
    params = {
        "tn": "news",
        "word": keyword,
        "rn": str(count),
    }
    resp = requests.get(url, params=params, headers=HEADERS, timeout=15)
    resp.encoding = "utf-8"
    html = resp.text

    results = []
    # 百度新闻搜索结果的每个条目在 <div class="result-op"> 或 <div class="result"> 中
    # 提取标题、链接、来源、时间、摘要
    blocks = re.findall(
        r'<div\s+class="result[^"]*"[^>]*>(.*?)</div>\s*</div>',
        html, re.DOTALL,
    )
    if not blocks:
        # 备用正则：匹配更宽泛的结构
        blocks = re.findall(
            r'<h3[^>]*class="c-title[^"]*"[^>]*>(.*?)</div>\s*(?:<div[^>]*>)?\s*</div>',
            html, re.DOTALL,
        )

    for block in blocks[:count]:
        news = {}

        # 标题
        title_match = re.search(r'<h3[^>]*>(.*?)</h3>', block, re.DOTALL)
        if title_match:
            news["title"] = re.sub(r"<[^>]+>", "", title_match.group(1)).strip()
        else:
            continue

        # 链接
        link_match = re.search(r'href=["\']?(https?://[^"\'>\s]+)', block)
        news["url"] = link_match.group(1) if link_match else ""

        # 来源和时间 - 通常在 <p class="c-author"> 或 <span class="c-color-gray"> 中
        author_match = re.search(
            r'(?:c-author|c-color-gray|news-meta)[^>]*>(.*?)<',
            block, re.DOTALL,
        )
        if author_match:
            meta = re.sub(r"<[^>]+>", "", author_match.group(1)).strip()
            news["meta"] = meta
        else:
            news["meta"] = ""

        # 摘要
        summary_match = re.search(
            r'<span\s+class="c-font-normal[^"]*"[^>]*>(.*?)</span>', block, re.DOTALL
        )
        if not summary_match:
            summary_match = re.search(
                r'class="c-span-last[^"]*"[^>]*>(.*?)</div>', block, re.DOTALL
            )
        if summary_match:
            news["summary"] = re.sub(r"<[^>]+>", "", summary_match.group(1)).strip()
        else:
            news["summary"] = ""

        results.append(news)

    return results


@mcp.tool()
def search_news(keyword: str, count: int = 5) -> str:
    """根据关键词搜索百度新闻。参数 keyword 为搜索关键词，count 为返回条数（默认5条，最多10条）。"""
    count = max(1, min(10, count))
    try:
        results = _search_baidu_news(keyword, count)
        if not results:
            return f"未找到与「{keyword}」相关的新闻。"

        lines = [f"🔍 关键词「{keyword}」的搜索结果（共 {len(results)} 条）：\n"]
        for i, news in enumerate(results, 1):
            lines.append(f"{i}. {news['title']}")
            if news["meta"]:
                lines.append(f"   📰 {news['meta']}")
            if news["summary"]:
                lines.append(f"   📝 {news['summary']}")
            if news["url"]:
                lines.append(f"   🔗 {news['url']}")
            lines.append("")
        return "\n".join(lines)
    except requests.exceptions.RequestException as e:
        return f"搜索新闻失败: {e}"


if __name__ == "__main__":
    mcp.run(transport="stdio")

在 mcp_client.py 中注册新闻 MCP

# MCP 服务器配置 - 在此注册所有 MCP 服务器
MCP_SERVERS = {
    "weather": {
        "transport": "stdio",
        "command": sys.executable,
        "args": [os.path.join(os.path.dirname(__file__), "mcp_servers", "weather_server.py")],
    },
    "news": {
        "transport": "stdio",
        "command": sys.executable,
        "args": [os.path.join(os.path.dirname(__file__), "mcp_servers", "news_server.py")],
    },
}