@youhaozhao/cninfo-mcp 1.0.5 → 1.0.7

package/README.md

@@ -9,6 +9,7 @@
 在 Claude Desktop / Claude Code 配置文件中添加：
 
 **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
+
 **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
 
 ```json
@@ -48,3 +49,4 @@
 ## Credits
 
 爬虫逻辑基于 [gaodechen/cninfo_process](https://github.com/gaodechen/cninfo_process)。
+

package/bin/cninfo-mcp.js

@@ -5,22 +5,45 @@
  * 自动检测 Python 并安装依赖，然后启动 Python MCP 服务器。
  */
 
-const { spawn } = require('child_process');
-const path = require('path');
-const fs = require('fs');
+const { spawn } = require("child_process");
+const path = require("path");
+const fs = require("fs");
+const os = require("os");
 
 // 配置路径
-const PYTHON_SCRIPT = path.join(__dirname, '..', 'python', 'mcp_server.py');
-const PYTHON_REQUIREMENTS = path.join(__dirname, '..', 'python', 'requirements.txt');
+const PYTHON_SCRIPT = path.join(__dirname, "..", "python", "mcp_server.py");
+const PYTHON_REQUIREMENTS = path.join(
+  __dirname,
+  "..",
+  "python",
+  "requirements.txt",
+);
+
+// 虚拟环境目录，放在用户目录下保证跨 npx 调用持久化
+const VENV_DIR = path.join(os.homedir(), ".cninfo-mcp", "venv");
+
+// 获取虚拟环境中的 Python 可执行文件路径
+function getVenvPython() {
+  if (process.platform === "win32") {
+    return path.join(VENV_DIR, "Scripts", "python.exe");
+  }
+  return path.join(VENV_DIR, "bin", "python3");
+}
 
-// 查找可用的 Python 可执行文件
+// 查找可用的系统 Python 可执行文件（仅用于创建 venv）
 async function findPython() {
-  const pythonCommands = ['python3', 'python', 'python3.12', 'python3.11', 'python3.10'];
+  const pythonCommands = [
+    "python3",
+    "python",
+    "python3.12",
+    "python3.11",
+    "python3.10",
+  ];
 
   for (const cmd of pythonCommands) {
     try {
-      const result = await spawnAsync(cmd, ['--version']);
-      if (result.stdout && result.stdout.includes('Python')) {
+      const result = await spawnAsync(cmd, ["--version"]);
+      if (result.stdout && result.stdout.includes("Python")) {
         return cmd;
       }
     } catch (error) {
@@ -29,38 +52,57 @@ async function findPython() {
   }
 
   throw new Error(
-    'Python not found. Please install Python 3.10+ from https://python.org\n' +
-    'After installation, restart your terminal and try again.'
+    "Python not found. Please install Python 3.10+ from https://python.org\n" +
+      "After installation, restart your terminal and try again.",
   );
 }
 
-// 检查并安装 Python 依赖
-async function ensureDependencies(pythonCmd) {
+// 创建虚拟环境（如果不存在）
+async function ensureVenv(systemPythonCmd) {
+  const venvPython = getVenvPython();
+  if (fs.existsSync(venvPython)) {
+    return venvPython;
+  }
+
+  console.error("Creating Python virtual environment...");
+  fs.mkdirSync(path.dirname(VENV_DIR), { recursive: true });
+  await spawnAsync(systemPythonCmd, ["-m", "venv", VENV_DIR], {
+    stdio: "inherit",
+  });
+  console.error("Virtual environment created\n");
+  return venvPython;
+}
+
+// 检查并安装 Python 依赖（使用 venv 中的 python）
+async function ensureDependencies(venvPython) {
   const requirementsPath = PYTHON_REQUIREMENTS;
 
   if (!fs.existsSync(requirementsPath)) {
-    console.error('Error: requirements.txt not found at', requirementsPath);
+    console.error("Error: requirements.txt not found at", requirementsPath);
     process.exit(1);
   }
 
   try {
     // 检查 mcp 包是否已安装
-    const checkResult = await spawnAsync(pythonCmd, ['-c', 'import mcp']);
+    await spawnAsync(venvPython, ["-c", "import mcp"]);
   } catch (error) {
     // 未安装，执行安装
-    console.error('Installing Python dependencies...');
-    const installResult = await spawnAsync(pythonCmd, ['-m', 'pip', 'install', '-r', requirementsPath], {
-      stdio: 'inherit'
-    });
-
-    if (installResult.code !== 0) {
-      console.error('\n❌ Failed to install Python dependencies');
-      console.error('Please run manually:');
-      console.error(`  ${pythonCmd} -m pip install -r ${requirementsPath}`);
+    console.error("Installing Python dependencies...");
+    try {
+      await spawnAsync(
+        venvPython,
+        ["-m", "pip", "install", "-r", requirementsPath],
+        {
+          stdio: "inherit",
+        },
+      );
+      console.error("Python dependencies installed successfully\n");
+    } catch (installError) {
+      console.error("\n❌ Failed to install Python dependencies");
+      console.error("Please run manually:");
+      console.error(`  ${venvPython} -m pip install -r ${requirementsPath}`);
       process.exit(1);
     }
-
-    console.error('✅ Python dependencies installed successfully\n');
   }
 }
 
@@ -68,28 +110,28 @@ async function ensureDependencies(pythonCmd) {
 function spawnAsync(command, args, options = {}) {
   return new Promise((resolve, reject) => {
     const child = spawn(command, args, {
-      stdio: options.stdio || 'pipe',
-      shell: process.platform === 'win32',
-      ...options
+      stdio: options.stdio || "pipe",
+      shell: process.platform === "win32",
+      ...options,
     });
 
-    let stdout = '';
-    let stderr = '';
+    let stdout = "";
+    let stderr = "";
     let code = null;
 
     if (child.stdout) {
-      child.stdout.on('data', (data) => {
+      child.stdout.on("data", (data) => {
         stdout += data.toString();
       });
     }
 
     if (child.stderr) {
-      child.stderr.on('data', (data) => {
+      child.stderr.on("data", (data) => {
         stderr += data.toString();
       });
     }
 
-    child.on('close', (exitCode) => {
+    child.on("close", (exitCode) => {
       code = exitCode;
       if (code === 0) {
         resolve({ stdout, stderr, code });
@@ -102,7 +144,7 @@ function spawnAsync(command, args, options = {}) {
       }
     });
 
-    child.on('error', (error) => {
+    child.on("error", (error) => {
       reject(error);
     });
   });
@@ -112,36 +154,36 @@ async function main() {
   try {
     // 检查 Python 脚本是否存在
     if (!fs.existsSync(PYTHON_SCRIPT)) {
-      console.error('Error: mcp_server.py not found at', PYTHON_SCRIPT);
+      console.error("Error: mcp_server.py not found at", PYTHON_SCRIPT);
       process.exit(1);
     }
 
-    const pythonCmd = await findPython();
-    await ensureDependencies(pythonCmd);
+    const systemPython = await findPython();
+    const venvPython = await ensureVenv(systemPython);
+    await ensureDependencies(venvPython);
 
     // 启动 MCP 服务器
-    console.error('巨潮资讯 MCP 服务器已启动，等待连接...');
-    const child = spawn(pythonCmd, [PYTHON_SCRIPT], {
-      stdio: 'inherit',
-      shell: process.platform === 'win32',
+    console.error("巨潮资讯 MCP 服务器已启动，等待连接...");
+    const child = spawn(venvPython, [PYTHON_SCRIPT], {
+      stdio: "inherit",
+      shell: process.platform === "win32",
       env: {
         ...process.env,
-        PYTHONPATH: path.join(__dirname, '..', 'python')
-      }
+        PYTHONPATH: path.join(__dirname, "..", "python"),
+      },
     });
 
     // 处理子进程退出
-    child.on('error', (error) => {
-      console.error('Failed to start MCP Server:', error.message);
+    child.on("error", (error) => {
+      console.error("Failed to start MCP Server:", error.message);
       process.exit(1);
     });
 
-    child.on('exit', (code) => {
+    child.on("exit", (code) => {
       process.exit(code || 0);
     });
-
   } catch (error) {
-    console.error('Error:', error.message);
+    console.error("Error:", error.message);
     process.exit(1);
   }
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@youhaozhao/cninfo-mcp",
-  "version": "1.0.5",
+  "version": "1.0.7",
   "description": "MCP Server for querying and downloading Chinese listed companies' annual reports from CNINFO (巨潮资讯网)",
   "keywords": [
     "mcp",

package/python/mcp_server.py

@@ -96,7 +96,9 @@ def query_annual_reports_tool(stock_code: str, year: Optional[int] = None) -> di
 
 
 @mcp.tool()
-def download_annual_reports_tool(stock_code: str, year: Optional[int] = None, save_path: Optional[str] = None) -> dict:
+def download_annual_reports_tool(
+    stock_code: str, year: Optional[int] = None, save_path: Optional[str] = None
+) -> dict:
     """
     Download annual reports for a Chinese listed company
 
@@ -170,7 +172,9 @@ def query_prospectus_tool(stock_code: str) -> dict:
                 "announcementTime": r.get("announcementTime", ""),
                 "secCode": r.get("secCode", ""),
                 "secName": r.get("secName", ""),
-                "adjunctUrl": base_url + r.get("adjunctUrl", "") if r.get("adjunctUrl") else "",
+                "adjunctUrl": base_url + r.get("adjunctUrl", "")
+                if r.get("adjunctUrl")
+                else "",
             }
             for r in reports
         ]

package/python/spider.py

@@ -3,9 +3,12 @@
 """
 
 import datetime
+import logging
 import os
 import random
+import re
 import time
+from typing import Optional, Union
 
 import requests
 
@@ -13,6 +16,7 @@ download_path = "https://static.cninfo.com.cn/"
 # 使用脚本所在目录的相对路径
 _saving_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "pdf")
 saving_path = _saving_path + "/"
+logger = logging.getLogger(__name__)
 
 User_Agent = [
     "Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Win64; x64; Trident/5.0; .NET CLR 3.5.30729; .NET CLR 3.0.30729; .NET CLR 2.0.50727; Media Center PC 6.0)",
@@ -25,7 +29,7 @@ User_Agent = [
 ]
 
 
-headers = {
+BASE_HEADERS = {
     "Accept": "application/json, text/javascript, */*; q=0.01",
     "Content-Type": "application/x-www-form-urlencoded; charset=UTF-8",
     "Accept-Encoding": "gzip, deflate",
@@ -37,10 +41,64 @@ headers = {
 }
 
 
+def _build_headers() -> dict:
+    """构造请求头，避免在并发场景下修改全局字典。"""
+    headers = BASE_HEADERS.copy()
+    headers["User-Agent"] = random.choice(User_Agent)
+    return headers
+
+
+def _date_range(start_date: str) -> str:
+    """构造查询时间区间，结束日期取当天，避免硬编码过期。"""
+    datetime.datetime.strptime(start_date, "%Y-%m-%d")
+    today = datetime.date.today().strftime("%Y-%m-%d")
+    return f"{start_date}~{today}"
+
+
+def _is_annual_report_title(
+    title: str, year_filter: Optional[Union[int, str]] = None
+) -> bool:
+    """
+    判断标题是否为“年度报告正文”。
+
+    支持常见变体：
+    - 2024年年度报告
+    - 2024年度报告
+    - 2024年报
+    """
+    compact_title = re.sub(r"\s+", "", title or "")
+
+    # 非正文公告关键词过滤
+    exclude_keywords = [
+        "摘要",
+        "确认意见",
+        "取消",
+        "更正",
+        "补充",
+        "说明",
+        "提示",
+        "致歉",
+        "修订",
+        "英文",
+    ]
+    if any(keyword in compact_title for keyword in exclude_keywords):
+        return False
+
+    year_expr = re.escape(str(year_filter)) if year_filter is not None else r"\d{4}"
+    suffix_expr = r"(?:[（(]更新后[)）])?"
+    patterns = [
+        rf".*{year_expr}年年度报告{suffix_expr}",
+        rf".*{year_expr}年度报告{suffix_expr}",
+    ]
+    if year_filter is not None:
+        patterns.append(rf".*{year_expr}年报{suffix_expr}")
+
+    return any(re.fullmatch(pattern, compact_title) for pattern in patterns)
+
+
 # 深市 年度报告
 def szseAnnual(page, stock):
     query_path = "http://www.cninfo.com.cn/new/hisAnnouncement/query"
-    headers["User-Agent"] = random.choice(User_Agent)  # 定义User_Agent
     query = {
         "pageNum": page,  # 页码
         "pageSize": 30,
@@ -52,10 +110,12 @@ def szseAnnual(page, stock):
         "plate": "sz",
         "category": "category_ndbg_szsh",  # 年度报告
         "trade": "",
-        "seDate": "2020-01-01~2026-02-15",  # 时间区间
+        "seDate": _date_range("2020-01-01"),  # 时间区间
     }
 
-    namelist = requests.post(query_path, headers=headers, data=query)
+    namelist = requests.post(
+        query_path, headers=_build_headers(), data=query, timeout=30
+    )
     result = namelist.json()
     if result and "announcements" in result and result["announcements"]:
         return result["announcements"]
@@ -65,7 +125,6 @@ def szseAnnual(page, stock):
 # 沪市 年度报告
 def sseAnnual(page, stock):
     query_path = "http://www.cninfo.com.cn/new/hisAnnouncement/query"
-    headers["User-Agent"] = random.choice(User_Agent)  # 定义User_Agent
     query = {
         "pageNum": page,  # 页码
         "pageSize": 30,
@@ -77,10 +136,12 @@ def sseAnnual(page, stock):
         "plate": "sh",
         "category": "category_ndbg_szsh",  # 年度报告
         "trade": "",
-        "seDate": "2020-01-01~2026-02-15",  # 时间区间
+        "seDate": _date_range("2020-01-01"),  # 时间区间
     }
 
-    namelist = requests.post(query_path, headers=headers, data=query)
+    namelist = requests.post(
+        query_path, headers=_build_headers(), data=query, timeout=30
+    )
     result = namelist.json()
     if result and "announcements" in result and result["announcements"]:
         return result["announcements"]
@@ -90,7 +151,6 @@ def sseAnnual(page, stock):
 # 深市 招股
 def szseStock(page, stock):
     query_path = "http://www.cninfo.com.cn/new/hisAnnouncement/query"
-    headers["User-Agent"] = random.choice(User_Agent)  # 定义User_Agent
     query = {
         "pageNum": page,  # 页码
         "pageSize": 30,
@@ -102,10 +162,12 @@ def szseStock(page, stock):
         "plate": "sz",
         "category": "",
         "trade": "",
-        "seDate": "2015-01-01~2026-02-15",  # 时间区间
+        "seDate": _date_range("2015-01-01"),  # 时间区间
     }
 
-    namelist = requests.post(query_path, headers=headers, data=query)
+    namelist = requests.post(
+        query_path, headers=_build_headers(), data=query, timeout=30
+    )
     result = namelist.json()
     if result and "announcements" in result and result["announcements"]:
         return result["announcements"]
@@ -115,7 +177,6 @@ def szseStock(page, stock):
 # 沪市 招股
 def sseStock(page, stock):
     query_path = "http://www.cninfo.com.cn/new/hisAnnouncement/query"
-    headers["User-Agent"] = random.choice(User_Agent)  # 定义User_Agent
     query = {
         "pageNum": page,  # 页码
         "pageSize": 30,
@@ -127,44 +188,27 @@ def sseStock(page, stock):
         "plate": "sh",
         "category": "",
         "trade": "",
-        "seDate": "2015-01-01~2026-02-15",  # 时间区间
+        "seDate": _date_range("2015-01-01"),  # 时间区间
     }
 
-    namelist = requests.post(query_path, headers=headers, data=query)
+    namelist = requests.post(
+        query_path, headers=_build_headers(), data=query, timeout=30
+    )
     result = namelist.json()
     if result and "announcements" in result and result["announcements"]:
         return result["announcements"]
     return []
 
 
-def Download(single_page, year_filter=None, save_path=None):
+def Download(
+    single_page,
+    year_filter: Optional[Union[int, str]] = None,
+    save_path: Optional[str] = None,
+):
     """下载公告列表中的 PDF 文件"""
     if single_page is None:
         return
 
-    headers = {
-        "Accept": "application/json, text/javascript, */*; q=0.01",
-        "Content-Type": "application/x-www-form-urlencoded; charset=UTF-8",
-        "Accept-Encoding": "gzip, deflate",
-        "Accept-Language": "zh-CN,zh;q=0.9,en-US;q=0.8,en;q=0.7,zh-HK;q=0.6,zh-TW;q=0.5",
-        "Host": "www.cninfo.com.cn",
-        "Origin": "http://www.cninfo.com.cn",
-    }
-
-    # 按年份筛选允许下载的标题
-    allowed_list = []
-    if year_filter:
-        allowed_list = [
-            f"{year_filter}年年度报告（更新后）",
-            f"{year_filter}年年度报告",
-        ]
-    else:
-        # 默认下载 2015 年至当前年份
-        current_year = datetime.datetime.now().year
-        for year in range(2015, current_year + 1):
-            allowed_list.append(f"{year}年年度报告（更新后）")
-            allowed_list.append(f"{year}年年度报告")
-
     allowed_list_2 = [
         "招股书",
         "招股说明书",
@@ -181,20 +225,13 @@ def Download(single_page, year_filter=None, save_path=None):
         if "确认意见" in title or "取消" in title or "摘要" in title:
             continue
 
-        # 检查标题是否精确匹配（避免"摘要"等变体被误下载）
-        allowed = False
-        for item in allowed_list:
-            if title == item:
-                allowed = True
-                break
+        # 年报标题匹配：支持“2024年年度报告/2024年度报告/2024年报”等变体
+        is_annual_report = _is_annual_report_title(title, year_filter=year_filter)
 
         # 检查招股书
-        for item in allowed_list_2:
-            if item in title:
-                allowed = True
-                break
+        is_prospectus = any(item in title for item in allowed_list_2)
 
-        if allowed:
+        if is_annual_report or is_prospectus:
             download = download_path + i["adjunctUrl"]
             name = (
                 i["secCode"]
@@ -209,19 +246,19 @@ def Download(single_page, year_filter=None, save_path=None):
             file_path = output_dir + name
 
             # 显示下载进度
-            print(f"  ↓ {name}")
+            logger.info("↓ %s", name)
 
             # 确保目录存在
             os.makedirs(output_dir, exist_ok=True)
 
             time.sleep(random.random() * 2)
 
-            headers["User-Agent"] = random.choice(User_Agent)
-            r = requests.get(download)
-
-            f = open(file_path, "wb")
-            f.write(r.content)
-            f.close()
+            r = requests.get(
+                download, headers={"User-Agent": random.choice(User_Agent)}, timeout=30
+            )
+            r.raise_for_status()
+            with open(file_path, "wb") as f:
+                f.write(r.content)
             downloaded_count += 1
         else:
             continue
@@ -237,17 +274,18 @@ def query_prospectus(stock_code):
         announcements_sse = sseStock(1, stock_code)
         all_announcements.extend(announcements_sse)
     except Exception as e:
-        print(f"沪市招股书查询失败: {e}")
+        logger.warning("沪市招股书查询失败: %s", e)
 
     try:
         announcements_szse = szseStock(1, stock_code)
         all_announcements.extend(announcements_szse)
     except Exception as e:
-        print(f"深市招股书查询失败: {e}")
+        logger.warning("深市招股书查询失败: %s", e)
 
     prospectus_keywords = ["招股书", "招股说明书", "招股意向书"]
     filtered = [
-        a for a in all_announcements
+        a
+        for a in all_announcements
         if any(kw in a.get("announcementTitle", "") for kw in prospectus_keywords)
     ]
 
@@ -288,21 +326,30 @@ def query_annual_reports(stock_code, year=None):
         announcements_sse = sseAnnual(1, stock_code)
         all_announcements.extend(announcements_sse)
     except Exception as e:
-        print(f"沪市年报查询失败: {e}")
+        logger.warning("沪市年报查询失败: %s", e)
 
     # 查询深市
     try:
         announcements_szse = szseAnnual(1, stock_code)
         all_announcements.extend(announcements_szse)
     except Exception as e:
-        print(f"深市年报查询失败: {e}")
+        logger.warning("深市年报查询失败: %s", e)
 
     # 按年份过滤
     if year:
-        year_str = str(year)
+        year_expr = re.escape(str(year))
+        year_patterns = [
+            rf"{year_expr}年年度报告",
+            rf"{year_expr}年度报告",
+            rf"{year_expr}年报",
+        ]
         filtered = []
         for announcement in all_announcements:
-            if year_str in announcement.get("announcementTitle", ""):
+            title = re.sub(r"\s+", "", announcement.get("announcementTitle", ""))
+            # 这里故意使用宽松匹配作为“预筛选”以保留候选项。
+            # 真正的严格判定（fullmatch + 排除词）在 Download() 的
+            # _is_annual_report_title() 中执行，形成两层防线。
+            if any(re.search(pattern, title) for pattern in year_patterns):
                 filtered.append(announcement)
         all_announcements = filtered
 
@@ -337,17 +384,22 @@ def download_annual_reports(stock_code, year=None, save_path=None):
 
 
 def Run(page_number, stock):
+    annual_report = []
+    stock_report = []
+    annual_report_ = []
+    stock_report_ = []
+
     try:
         annual_report = szseAnnual(page_number, stock)
         stock_report = szseStock(page_number, stock)
         annual_report_ = sseAnnual(page_number, stock)
         stock_report_ = sseStock(page_number, stock)
     except Exception:
-        print(page_number, "page error, retrying")
+        logger.warning("%s page error, retrying", page_number)
         try:
             annual_report = szseAnnual(page_number, stock)
         except Exception:
-            print(page_number, "page error")
+            logger.warning("%s page error", page_number)
     Download(annual_report)
     Download(stock_report)
     Download(annual_report_)
@@ -355,9 +407,12 @@ def Run(page_number, stock):
 
 
 if __name__ == "__main__":
+    logging.basicConfig(
+        level=logging.INFO, format="%(asctime)s %(levelname)s %(message)s"
+    )
     with open("company_id.txt") as file:
         lines = file.readlines()
         for line in lines:
             stock = line
             Run(1, line)
-            print(line, "done")
+            logger.info("%s done", line.strip())

package/scripts/install-python-deps.js

@@ -7,6 +7,7 @@
 const { spawn } = require("child_process");
 const fs = require("fs");
 const path = require("path");
+const os = require("os");
 
 const REQUIREMENTS_FILE = path.join(
   __dirname,
@@ -15,6 +16,15 @@ const REQUIREMENTS_FILE = path.join(
   "requirements.txt",
 );
 
+const VENV_DIR = path.join(os.homedir(), ".cninfo-mcp", "venv");
+
+function getVenvPython() {
+  if (process.platform === "win32") {
+    return path.join(VENV_DIR, "Scripts", "python.exe");
+  }
+  return path.join(VENV_DIR, "bin", "python3");
+}
+
 async function findPython() {
   const pythonCommands = [
     "python3",
@@ -75,16 +85,31 @@ async function main() {
     return;
   }
 
+  // 创建虚拟环境（如果不存在）
+  const venvPython = getVenvPython();
+  if (!fs.existsSync(venvPython)) {
+    console.log("Creating Python virtual environment...");
+    try {
+      fs.mkdirSync(path.dirname(VENV_DIR), { recursive: true });
+      await spawnCommand(pythonCmd, ["-m", "venv", VENV_DIR]);
+      console.log("Virtual environment created");
+    } catch (venvError) {
+      console.warn("  Failed to create virtual environment during npm install");
+      console.warn("  It will be created automatically on first run");
+      return;
+    }
+  }
+
   try {
-    // 检查 mcp 是否已安装
-    await spawnCommand(pythonCmd, ["-c", "import mcp"]);
+    // 检查 mcp 是否已安装（用 venv 的 python）
+    await spawnCommand(venvPython, ["-c", "import mcp"]);
     console.log("✅ Python dependencies already installed");
   } catch (error) {
-    // 执行安装
+    // 执行安装（用 venv 的 pip）
     console.log("📦 Installing Python dependencies...");
     try {
       await spawnCommand(
-        pythonCmd,
+        venvPython,
         ["-m", "pip", "install", "-r", REQUIREMENTS_FILE],
         {
           stdio: "inherit",