├── mcp_query_table ├── sites │ ├── __init__.py │ ├── tdx.py │ ├── eastmoney.py │ └── iwencai.py ├── providers │ ├── __init__.py │ ├── n.py │ ├── yuanbao.py │ └── baidu.py ├── _version.py ├── __init__.py ├── enums.py ├── utils.py ├── __main__.py ├── server.py └── tool.py ├── docs └── img │ └── streamlit.png ├── streamlit ├── requirements.txt ├── .streamlit │ └── config.toml ├── auth.yaml ├── config.yaml ├── run.bat ├── README.md ├── client.py └── app.py ├── pyproject.toml ├── LICENSE ├── .github └── workflows │ └── python-publish.yml ├── tests ├── headless.py ├── hook3.py ├── hook2.py └── hook.py ├── examples ├── main_sync.py ├── main.py ├── main_chat.py └── mcp.txt ├── .gitignore └── README.md /mcp_query_table/sites/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mcp_query_table/providers/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /mcp_query_table/_version.py: -------------------------------------------------------------------------------- 1 | __version__ = "0.3.12" 2 | -------------------------------------------------------------------------------- /docs/img/streamlit.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/wukan1986/mcp_query_table/HEAD/docs/img/streamlit.png -------------------------------------------------------------------------------- /streamlit/requirements.txt: -------------------------------------------------------------------------------- 1 | mcp_query_table 2 | streamlit 3 | PyYAML 4 | tabulate 5 | streamlit-authenticator -------------------------------------------------------------------------------- /streamlit/.streamlit/config.toml: -------------------------------------------------------------------------------- 1 | [server] 2 | enableStaticServing = true 3 | port = 51016 4 | 5 | [theme] 6 | codeFont = "SimSun, monospace" -------------------------------------------------------------------------------- /streamlit/auth.yaml: -------------------------------------------------------------------------------- 1 | cookie: 2 | expiry_days: 5 3 | key: some_key 4 | name: mcp_query_table 5 | credentials: 6 | usernames: 7 | admin: 8 | password: "12345678" -------------------------------------------------------------------------------- /streamlit/config.yaml: -------------------------------------------------------------------------------- 1 | max_page: 1 2 | mcp_endpoint: http://localhost:8000/sse 3 | templates: 4 | 涨幅前10: '你是一个专业的股票分析师。请忽略文件名，仅根据文件内容，为我提供专业分析报告。不用联网搜索。 5 | 6 | 7 | 文件内容如下：' 8 | -------------------------------------------------------------------------------- /mcp_query_table/__init__.py: -------------------------------------------------------------------------------- 1 | from ._version import __version__ 2 | 3 | from .enums import QueryType, Site, Provider 4 | from .tool import BrowserManager, query, chat 5 | 6 | TIMEOUT = 1000 * 60 * 3 # 3分钟，在抓取EventStream数据时等待数据返回，防止外层30秒超时 7 | TIMEOUT_60 = 1000 * 60 # 1分钟 8 | 9 | # TODO 临时测试 10 | # TIMEOUT = None 11 | # TIMEOUT_60 = None 12 | -------------------------------------------------------------------------------- /streamlit/run.bat: -------------------------------------------------------------------------------- 1 | CALL d:\Users\Kan\miniconda3\Scripts\activate.bat d:\Users\Kan\miniconda3\envs\py312 2 | start streamlit run app.py --server.enableStaticServing=true --theme.codeFont="SimSun, monospace" --server.port=51015 3 | cd .. 4 | start python -m mcp_query_table --format markdown --transport sse --port 8000 --endpoint --executable_path --user_data_dir 5 | pause -------------------------------------------------------------------------------- /mcp_query_table/enums.py: -------------------------------------------------------------------------------- 1 | from enum import Enum 2 | 3 | 4 | class QueryType(Enum): 5 | """查询类型""" 6 | CNStock = 'A股' 7 | HKStock = '港股' 8 | USStock = '美股' 9 | Index = '指数' 10 | Fund = '基金' 11 | ETF = 'ETF' 12 | ConBond = '可转债' 13 | Board = '板块' 14 | Info = '资讯' 15 | 16 | 17 | class Site(Enum): 18 | """站点""" 19 | EastMoney = '东方财富' # 东方财富 条件选股 20 | TDX = '通达信' # 通达信 问小达 21 | THS = '同花顺' # 同花顺 问财 22 | 23 | 24 | class Provider(Enum): 25 | """提供商""" 26 | Nami = '纳米搜索' # 360 纳米搜索 27 | YuanBao = '腾讯元宝' # 腾讯元宝 28 | BaiDu = '百度AI搜索' # 百度AI搜索 29 | # YiYan = '文心一言' # 百度文心一言 30 | -------------------------------------------------------------------------------- /mcp_query_table/utils.py: -------------------------------------------------------------------------------- 1 | from pathlib import Path 2 | from typing import List, Tuple 3 | 4 | 5 | def is_image(path: str) -> bool: 6 | """判断是否是图片文件""" 7 | img_ext = ['.jpg', '.jpeg', '.png', '.gif', '.bmp', '.webp'] 8 | ext = Path(path).suffix.lower() 9 | return ext in img_ext 10 | 11 | 12 | def split_images(files: List[str]) -> Tuple[List[str], List[str]]: 13 | """图片列表分成两部分""" 14 | imgs = [] 15 | docs = [] 16 | for f in files: 17 | if is_image(f): 18 | imgs.append(f) 19 | else: 20 | docs.append(f) 21 | return imgs, docs 22 | 23 | 24 | class GlobalVars: 25 | """全局变量""" 26 | 27 | def __init__(self): 28 | self.text = "" 29 | 30 | def set_text(self, text): 31 | self.text = text 32 | 33 | def get_text(self): 34 | return self.text 35 | -------------------------------------------------------------------------------- /pyproject.toml: -------------------------------------------------------------------------------- 1 | [project] 2 | name = "mcp_query_table" 3 | authors = [ 4 | { name = "wukan", email = "wu-kan@163.com" }, 5 | ] 6 | description = "query table from website, support MCP" 7 | readme = "README.md" 8 | requires-python = ">=3.10" 9 | keywords = ["playwright", "mcp", "table", "iwencai", "tdx", "eastmoney"] 10 | license = { file = "LICENSE" } 11 | classifiers = [ 12 | "Development Status :: 4 - Beta", 13 | "Programming Language :: Python" 14 | ] 15 | dependencies = [ 16 | "pandas", 17 | "loguru", 18 | "playwright", 19 | "playwright-stealth>=2.0.0", # https://github.com/Mattwmaster58/playwright_stealth 20 | "fastmcp", 21 | "tabulate" 22 | ] 23 | dynamic = ["version"] 24 | 25 | [build-system] 26 | requires = ["hatchling"] 27 | build-backend = "hatchling.build" 28 | 29 | [tool.hatch.version] 30 | path = "mcp_query_table/_version.py" 31 | 32 | [tool.hatch.build.targets.wheel] 33 | packages = ["mcp_query_table"] 34 | include-package-data = true 35 | 36 | [tool.hatch.build.targets.sdist] 37 | include = ["mcp_query_table*"] -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2025 伍侃 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. 22 | -------------------------------------------------------------------------------- /.github/workflows/python-publish.yml: -------------------------------------------------------------------------------- 1 | # This workflow will upload a Python Package using Twine when a release is created 2 | # For more information see: https://docs.github.com/en/actions/automating-builds-and-tests/building-and-testing-python#publishing-to-package-registries 3 | 4 | # This workflow uses actions that are not certified by GitHub. 5 | # They are provided by a third-party and are governed by 6 | # separate terms of service, privacy policy, and support 7 | # documentation. 8 | 9 | name: Upload Python Package 10 | 11 | on: 12 | release: 13 | types: [published] 14 | 15 | permissions: 16 | contents: read 17 | 18 | jobs: 19 | deploy: 20 | 21 | runs-on: ubuntu-latest 22 | 23 | steps: 24 | - uses: actions/checkout@v3 25 | - name: Set up Python 26 | uses: actions/setup-python@v3 27 | with: 28 | python-version: '3.x' 29 | - name: Install dependencies 30 | run: | 31 | python -m pip install --upgrade pip 32 | pip install build 33 | - name: Build package 34 | run: python -m build 35 | - name: Publish package 36 | uses: pypa/gh-action-pypi-publish@27b31702a0e7fc50959f5ad993c78deac1bdfc29 37 | with: 38 | user: __token__ 39 | password: ${{ secrets.PYPI_API_TOKEN }} 40 | -------------------------------------------------------------------------------- /tests/headless.py: -------------------------------------------------------------------------------- 1 | import asyncio 2 | import random 3 | import string 4 | 5 | from playwright.async_api import async_playwright 6 | from playwright_stealth import stealth_async, StealthConfig 7 | 8 | 9 | async def main(): 10 | # This is the recommended usage. All pages created will have stealth applied: 11 | async with async_playwright() as p: 12 | browser = await p.chromium.launch(headless=False) 13 | page = await browser.new_page() 14 | 15 | class FixedConfig(StealthConfig): 16 | @property 17 | def enabled_scripts(self): 18 | key = "".join(random.choices(string.ascii_letters, k=10)) 19 | for script in super().enabled_scripts: 20 | if "const opts" in script: 21 | yield script.replace("const opts", f"window.{key}") 22 | continue 23 | yield script.replace("opts", f"window.{key}") 24 | 25 | await stealth_async(page, FixedConfig()) 26 | 27 | w = '收益最好的200只ETF' 28 | querytype = 'fund' 29 | url = f"https://www.n.cn" 30 | print(url) 31 | await page.goto(url) 32 | await page.wait_for_timeout(1000 * 15) 33 | await page.screenshot(path="example.png") 34 | await page.wait_for_timeout(1000 * 15000) 35 | 36 | 37 | asyncio.run(main()) 38 | -------------------------------------------------------------------------------- /examples/main_sync.py: -------------------------------------------------------------------------------- 1 | """ 2 | 本示例演示了如何使用同步风格来调用异步函数编写代码 3 | 但还是有局限性，可以在Python REPL环境中一行行输入使用，但无法在Windows下的Jupyter Notebook中使用 4 | 5 | 使用方法有3种，选一种即可 6 | 1. 直接`python main_sync.py`运行本文件 7 | 2. 在控制台中输入`python`,提示`>>>`,然后输入代码 8 | 3. 在VSCode中选中一行，让后右键`Run Python` > `Run Selection/Line in Native Python REPL` 9 | - 可以使用Shift+Enter来运行选中代码。但要在插件中禁用`Jupyter`，因为`Run in Interactive Window`下的功能快捷键冲突 10 | 11 | """ 12 | # %% 13 | 14 | import revolving_asyncio # pip install revolving_asyncio 15 | # revolving_asyncio.apply() 16 | 17 | # %% 18 | from mcp_query_table import query, QueryType, Site, BrowserManager 19 | 20 | bm = BrowserManager(endpoint="http://127.0.0.1:9333", executable_path=r'C:\Program Files (x86)\Microsoft\Edge\Application\msedge.exe', devtools=False) 21 | query = revolving_asyncio.to_sync(query) 22 | get_page = revolving_asyncio.to_sync(bm.get_page) 23 | release_page = revolving_asyncio.to_sync(bm.release_page) 24 | 25 | # %% 26 | page1 = get_page() 27 | page2 = get_page() 28 | df = query(page2, '收盘价>50元的港股', query_type=QueryType.HKStock, max_page=3, site=Site.THS) 29 | 30 | print(df.to_markdown()) 31 | 32 | # %% 33 | df = query(page1, '年初至今收益率前50', query_type=QueryType.Fund, max_page=3, site=Site.TDX) 34 | print(df.to_csv()) 35 | # %% 36 | df = query(page2, '收盘价>50元', query_type=QueryType.HKStock, max_page=3, site=Site.EastMoney) 37 | release_page(page1) 38 | release_page(page2) 39 | print(df) 40 | -------------------------------------------------------------------------------- /examples/main.py: -------------------------------------------------------------------------------- 1 | import asyncio 2 | import getpass 3 | 4 | from mcp_query_table import * 5 | 6 | 7 | async def main() -> None: 8 | endpoint = "http://127.0.0.1:9222" 9 | executable_path = r"C:\Program Files\Google\Chrome\Application\chrome.exe" 10 | user_data_dir = rf'C:\Users\{getpass.getuser()}\AppData\Local\Google\Chrome\User Data\Default' 11 | # 以下使用的无头模式，速度快。建议先登录好网站账号再使用无头模式 12 | async with BrowserManager(endpoint=None, 13 | executable_path=executable_path, 14 | devtools=False, 15 | headless=True, 16 | user_data_dir=user_data_dir) as bm: 17 | # 问财需要保证浏览器宽度>768，防止界面变成适应手机 18 | page = await bm.get_page() 19 | df = await query(page, '收益最好的200只ETF', query_type=QueryType.ETF, max_page=1, site=Site.THS) 20 | print(df.to_markdown()) 21 | df = await query(page, '年初至今收益率前50', query_type=QueryType.Fund, max_page=1, site=Site.TDX) 22 | print(df.to_csv()) 23 | df = await query(page, '流通市值前10的行业板块', query_type=QueryType.Index, max_page=1, site=Site.TDX) 24 | print(df.to_csv()) 25 | # TODO 东财翻页要提前登录 26 | df = await query(page, '今日涨幅前5的概念板块;', query_type=QueryType.Board, max_page=3, site=Site.EastMoney) 27 | print(df) 28 | bm.release_page(page) 29 | print('done') 30 | await page.wait_for_timeout(2000) 31 | 32 | 33 | if __name__ == '__main__': 34 | asyncio.run(main()) 35 | -------------------------------------------------------------------------------- /mcp_query_table/__main__.py: -------------------------------------------------------------------------------- 1 | import getpass 2 | 3 | from mcp_query_table.server import serve 4 | 5 | 6 | def main(): 7 | import argparse 8 | 9 | parser = argparse.ArgumentParser( 10 | description="query table from website", 11 | ) 12 | 13 | parser.add_argument("--format", type=str, help="输出格式", 14 | default='markdown', choices=['markdown', 'csv', 'json']) 15 | parser.add_argument("--endpoint", type=str, help="浏览器CDP地址/WS地址", 16 | nargs="?", default=r'http://127.0.0.1:9222') 17 | parser.add_argument("--executable_path", type=str, help="浏览器路径", 18 | nargs="?", default=r'C:\Program Files\Google\Chrome\Application\chrome.exe') 19 | parser.add_argument("--user_data_dir", type=str, help="浏览器用户数据目录", 20 | nargs="?", default=rf'C:\Users\{getpass.getuser()}\AppData\Local\Google\Chrome\User Data') 21 | parser.add_argument("--transport", type=str, help="传输类型", 22 | default='stdio', choices=['stdio', 'sse', 'streamable-http']) 23 | parser.add_argument("--host", type=str, help="MCP服务端绑定地址", 24 | default='0.0.0.0') 25 | parser.add_argument("--port", type=int, help="MCP服务端绑定端口", 26 | default='8000') 27 | args = parser.parse_args() 28 | serve(args.format, args.endpoint, 29 | args.executable_path, args.user_data_dir, 30 | args.transport, args.host, args.port) 31 | 32 | 33 | if __name__ == "__main__": 34 | main() 35 | -------------------------------------------------------------------------------- /streamlit/README.md: -------------------------------------------------------------------------------- 1 | # Streamlit应用 2 | 3 | 实现在同一页面中查询金融数据，并手工输入到大语言模型网站中进行深度分析。 4 | 5 | ## 功能 6 | 7 | - 直接查询金融网站的数据，免去数据导出的麻烦 8 | - 内嵌大语言模型网站，同一页面中进行大数据分析 9 | 10 | ## 部署方法 11 | 12 | 1. 安装两款浏览器，其中一款必须是`Chrome`(用于`playwright`控制)。另外一款用于访问`Streamlit`，如`Edge` 13 | 2. 安装依赖 14 | ```bash 15 | pip install -r requirements.txt 16 | playwright install chromium 17 | ``` 18 | 3. 启动`MCP`服务`SSE`模式 19 | ```bash 20 | # Linux下的无头模式，速度更快。不用登录 21 | python -m mcp_query_table --format markdown --transport sse --port 8000 --endpoint --executable_path --user_data_dir 22 | ``` 23 | 4. 启动`Streamlit`应用 24 | ```bash 25 | streamlit run app.py --server.enableStaticServing=true --theme.codeFont="SimSun, monospace" --server.port=51015 26 | ``` 27 | 5. 打开`Edge`浏览器，访问`http://localhost:51015/` 28 | 29 | ## streamlit使用方法 30 | 31 | 1. 选择合适的大语言模型网站，如`腾讯`、`字节`、`阿里`等 32 | 2. 选择合适的查询网站，如`东方财富`、`同花顺`、`通达信` 33 | 3. 输入查询条件/提示词，如`2024年涨幅最大的100只股票按2024年12月31日总市值排名` 34 | 4. 点击`查询`按钮，查询结果会显示在右下页面中（提示词+数据），可以点击复制按钮，将查询结果粘贴到大语言模型网站中进行分析 35 | 5. 在`下载Markdown`(只含数据)右键复制链接，在大语言模型网站中点击`上传文件`，打开文件对话框中直接粘贴链接。然后复制提示词过来即可 36 | 6. `Markdown`下载到本地，可以在记事本中打开，字体设置成`宋体`表格会显示正常 37 | 38 | ## 注意 39 | 40 | 1. 东方财富。翻页要登录，港股要登录 41 | 42 | ## Linux命令 43 | 44 | ```bash 45 | # 启动MCP服务 46 | nohup python -m mcp_query_table --format markdown --transport sse --port 8000 --endpoint --executable_path --user_data_dir > mcp.log 2>&1 & 47 | # 启动Streamlit应用 48 | nohup streamlit run app.py --server.enableStaticServing=true --theme.codeFont="SimSun, monospace" --server.port=51015 > streamlit.log 2>&1 & 49 | ``` 50 | 51 | ## 参考 52 | 53 | https://github.com/zanetworker/mcp-sse-client-python 54 | 55 | ## .streamlit/config.toml 56 | 可以简化成 57 | ```bash 58 | nohup streamlit run app.py > streamlit.log 2>&1 & 59 | ``` -------------------------------------------------------------------------------- /examples/main_chat.py: -------------------------------------------------------------------------------- 1 | import asyncio 2 | import getpass 3 | 4 | from mcp_query_table import * 5 | from mcp_query_table.enums import Provider 6 | 7 | 8 | async def main() -> None: 9 | endpoint = "http://127.0.0.1:9222" 10 | executable_path = r"C:\Program Files\Google\Chrome\Application\chrome.exe" 11 | user_data_dir = rf'C:\Users\{getpass.getuser()}\AppData\Local\Google\Chrome\User Data\Default' 12 | async with BrowserManager(endpoint=None, 13 | executable_path=executable_path, 14 | devtools=False, 15 | headless=True, 16 | user_data_dir=user_data_dir) as bm: 17 | page1 = await bm.get_page() 18 | page2 = await bm.get_page() 19 | 20 | with open("mcp.txt", 'r', encoding='utf-8') as f: 21 | prompt = f.read() 22 | 23 | files = [ 24 | # r"D:\Users\Kan\Documents\GitHub\mcp_query_table\examples\mcp.txt", 25 | r"d:\1.png" 26 | ] 27 | 28 | output = await chat(page1, "2+3等于多少？", provider=Provider.BaiDu) 29 | print(output) 30 | output = await chat(page1, "3+4等于多少？", provider=Provider.Nami) 31 | print(output) 32 | output = await chat(page2, "4+5等于多少？", provider=Provider.YuanBao) 33 | print(output) 34 | output = await chat(page2, "这张照片的拍摄参数是多少？", files=files, provider=Provider.Nami) 35 | print(output) 36 | output = await chat(page2, "描述下文件内容", files=files, provider=Provider.YuanBao) 37 | print(output) 38 | output = await chat(page2, "描述下文件内容", files=files, provider=Provider.BaiDu) 39 | print(output) 40 | 41 | bm.release_page(page1) 42 | bm.release_page(page2) 43 | 44 | 45 | if __name__ == '__main__': 46 | asyncio.run(main()) 47 | -------------------------------------------------------------------------------- /tests/hook3.py: -------------------------------------------------------------------------------- 1 | """ 2 | """ 3 | import asyncio 4 | import re 5 | 6 | from mcp_query_table import BrowserManager 7 | 8 | query = {} 9 | 10 | 11 | def __hook(e, a, b): 12 | print("111", e, a, b) 13 | 14 | 15 | async def on_route(route, request): 16 | """找到特殊js文件，每过一段时间特殊文件名不同，但内部的函数名不变""" 17 | response = await route.fetch() 18 | body = await response.text() 19 | if "fetchFeedIndex:" not in body: 20 | await route.fulfill(response=response) 21 | return 22 | 23 | print(request.url) 24 | 25 | pattern = r'(fetchFeedIndex:)(.*?)(k\.a\.decrypt\(t\.data,e\.data\)\.split\(","\);)(.*?)(drawTrend:)' 26 | body = re.sub(pattern, r'\1 \2 \3 window.__hook(e,a,W.a.getDatesList(e.startDate, e.endDate, e.type));\4 \5', body, 27 | flags=re.DOTALL) 28 | 29 | await route.fulfill(content_type="text/javascript; charset=utf-8", body=body) 30 | 31 | 32 | async def on_flash(response): 33 | if "jin10.com/flash?channel=" in response.url: 34 | print(response.url) 35 | json_data = await response.json() 36 | for i in json_data['data']: 37 | print("flash?channel=", i) 38 | 39 | 40 | async def main() -> None: 41 | # taskkill /f /im msedge.exe 42 | async with BrowserManager(port=9222, browser_path=r'C:\Program Files (x86)\Microsoft\Edge\Application\msedge.exe', 43 | devtools=False) as bm: 44 | page = await bm.get_page() 45 | await page.expose_function("__hook", __hook) 46 | await page.route("**/static/js/main.*.js", on_route) 47 | # page.on("response", on_flash) 48 | await page.goto("https://index.baidu.com/v2/index.html#/") 49 | await page.get_by_role("searchbox", name="请输入您想查询的关键词").click() 50 | await page.get_by_role("searchbox", name="请输入您想查询的关键词").fill("上证指数") 51 | await page.get_by_role("searchbox", name="请输入您想查询的关键词").press("Enter") 52 | 53 | await page.wait_for_timeout(1000 * 1000) 54 | print('done') 55 | bm.release_page(page) 56 | await bm.cleanup() 57 | 58 | 59 | if __name__ == '__main__': 60 | asyncio.run(main()) 61 | -------------------------------------------------------------------------------- /tests/hook2.py: -------------------------------------------------------------------------------- 1 | """ 2 | # https://4a735ea38f8146198dc205d2e2d1bd28.z3c.jin10.com/flash?channel=-8200&vip=1&classify=[13] 3 | # https://flash-api.jin10.com/get_flash_list?channel=-8200&vip=1 4 | # wss://wss-flash-2.jin10.com/ 5 | 6 | socket.io 导致不同浏览器用的机制不同，例如在本人电脑中 7 | chrome 走 https://4a735ea38f8146198dc205d2e2d1bd28.z3c.jin10.com/flash?channel=-8200&vip=1&classify=[13] 8 | edge 走 wss://wss-flash-2.jin10.com/ 9 | """ 10 | import asyncio 11 | 12 | from mcp_query_table import BrowserManager 13 | 14 | query = {} 15 | 16 | 17 | def __hook(obj): 18 | print("dealSocketData", obj) 19 | 20 | 21 | async def on_route(route, request): 22 | """找到特殊js文件，每过一段时间特殊文件名不同，但内部的函数名不变""" 23 | response = await route.fetch() 24 | body = await response.text() 25 | if "dealSocketData" not in body: 26 | await route.fulfill(response=response) 27 | return 28 | 29 | print(request.url) 30 | # 解决了实时，如何解决历史数据 31 | body = body.replace("dealSocketData:function(t){", """ 32 | dealSocketData:function(t){window.__hook(t);""") 33 | 34 | await route.fulfill(content_type="text/javascript; charset=utf-8", body=body) 35 | 36 | 37 | async def on_flash(response): 38 | if "jin10.com/flash?channel=" in response.url: 39 | print(response.url) 40 | json_data = await response.json() 41 | for i in json_data['data']: 42 | print("flash?channel=", i) 43 | 44 | 45 | async def main() -> None: 46 | # taskkill /f /im msedge.exe 47 | async with BrowserManager(port=9222, browser_path=r'C:\Program Files (x86)\Microsoft\Edge\Application\msedge.exe', 48 | devtools=False) as bm: 49 | page = await bm.get_page() 50 | await page.expose_function("__hook", __hook) 51 | await page.route("**/js/index.*.js", on_route) 52 | page.on("response", on_flash) 53 | await page.goto("https://www.jin10.com/", wait_until="load") 54 | 55 | await page.wait_for_timeout(1000 * 1000) 56 | print('done') 57 | bm.release_page(page) 58 | await bm.cleanup() 59 | 60 | 61 | if __name__ == '__main__': 62 | asyncio.run(main()) 63 | -------------------------------------------------------------------------------- /tests/hook.py: -------------------------------------------------------------------------------- 1 | """ 2 | 只是为以后各平台安全升级做破解准备，暂时不使用 3 | 4 | 某网站返回的数据是json中有字段是加密的，需要获取解密后的内容 5 | 6 | 直接hook请求函数，返回解密后的内容 7 | 8 | 首页是静态网页，翻页是fetch请求 9 | 10 | 发现js文件名也是动态变化的 11 | """ 12 | import asyncio 13 | 14 | from mcp_query_table import BrowserManager 15 | 16 | query = {} 17 | 18 | 19 | def __hook(x, y, obj): 20 | if x == '/sun/ranking/fundRankV3': 21 | global query 22 | query = y 23 | print(x) 24 | print(y) 25 | print(obj) 26 | 27 | 28 | async def on_route(route, request): 29 | """找到特殊js文件，每过一段时间特殊文件名不同，但内部的函数名不变""" 30 | response = await route.fetch() 31 | body = await response.text() 32 | if "uXpFetch" not in body: 33 | await route.fulfill(response=response) 34 | return 35 | 36 | # 网页中引用的js文件，会变化 37 | # 38 | print(request.url) 39 | body = body.replace("export{", """ 40 | // 原函数注册到window，不是改版函数 41 | window.uXpFetch = uXpFetch; 42 | // 重写局部函数 43 | uXpFetch =async function(e,t) { 44 | const ret=window.uXpFetch(e,t); 45 | ret.then((r)=>{window.__hook(e,t,r)}); 46 | return ret; 47 | }; 48 | export{""") 49 | 50 | await route.fulfill(content_type="text/javascript; charset=utf-8", body=body) 51 | 52 | 53 | async def main() -> None: 54 | async with BrowserManager(port=9222, browser_path=None, devtools=True) as bm: 55 | page = await bm.get_page() 56 | await page.expose_function("__hook", __hook) 57 | await page.route("**/_nuxt3/*.js", on_route) 58 | await page.goto("https://dc.simuwang.com/smph", wait_until="load") 59 | 60 | # 强行翻页，产生fetch请求 61 | await page.get_by_role("button", name="上一页", disabled=True).evaluate( 62 | 'element => { element.removeAttribute("disabled"); element.removeAttribute("aria-disabled");}') 63 | await page.get_by_role("button", name="上一页").click() 64 | # 方便记录请求参数 65 | print(query) 66 | print('=' * 60) 67 | 68 | # 相当于requests，但解码麻烦 69 | # r = await page.request.get('https://sppwapi.simuwang.com/sun/ranking/fundRankV3?page=1&size=50&condition=%7B%22fund_type%22:%226%22%7D&sort_name=ret_6m&sort_asc=desc&tab_type=1') 70 | # print(await r.text()) 71 | 72 | # 更快速的请求方式 73 | for i in range(1, 4): 74 | # await page.get_by_role("button", name="下一页").click() 75 | query['data']['page'] = i 76 | r = await page.evaluate("([x, y])=>window.uXpFetch(x,y)", ['/sun/ranking/fundRankV3', query]) 77 | print(r) 78 | 79 | print('done') 80 | await page.wait_for_timeout(1000 * 10) 81 | bm.release_page(page) 82 | await bm.cleanup() 83 | 84 | 85 | if __name__ == '__main__': 86 | asyncio.run(main()) 87 | 88 | """ 89 | 90 | 91 | var _hook = uXpFetch; 92 | uXpFetch =async function(e,t) { 93 | const ret=_hook(e,t); 94 | ret.then((r)=>{window.__hook(e,t,r)}); 95 | return ret; 96 | }; 97 | """ 98 | -------------------------------------------------------------------------------- /mcp_query_table/providers/n.py: -------------------------------------------------------------------------------- 1 | """ 2 | 360 纳米搜索 3 | """ 4 | import json 5 | 6 | from loguru import logger 7 | from playwright.async_api import Page 8 | 9 | import mcp_query_table 10 | from mcp_query_table.utils import is_image, GlobalVars 11 | 12 | _PAGE0_ = "https://www.n.cn" 13 | _PAGE1_ = "https://www.n.cn/search" 14 | _PAGE2_ = "https://www.n.cn/api/common/chat/v2" # 对话 15 | _PAGE3_ = "https://www.n.cn/api/image/upload" # 上传图片 16 | 17 | G = GlobalVars() 18 | 19 | 20 | def read_event_stream(text): 21 | text1 = [] 22 | text2 = [] 23 | for event in text.split('\n\n'): 24 | if "event: 102" in event: 25 | if 'data: {"type":"reasoning_text"' in event: 26 | lines = event.split('\n') 27 | for line in lines: 28 | if line.startswith('data: '): 29 | t = line[6:] 30 | t = json.loads(t)['message'] 31 | text1.append(t) 32 | if "event: 200" in event: 33 | lines = event.split('\n') 34 | for line in lines: 35 | if line.startswith('data: '): 36 | t = line[6:] 37 | if t == '': 38 | text2.append('\n') 39 | elif t == ' ': 40 | text2.append('\n') 41 | else: 42 | text2.append(t) 43 | 44 | text2 = ''.join(text2) 45 | if len(text1) == 0: 46 | return text2 47 | else: 48 | text1 = ''.join(text1) 49 | return f"{text1}\n\n{text2}" 50 | 51 | 52 | async def on_response(response): 53 | if response.url == _PAGE2_: 54 | # print("on_response", response.url) 55 | text = await response.text() 56 | G.set_text(read_event_stream(text)) 57 | 58 | 59 | async def chat(page: Page, 60 | prompt: str, 61 | create: bool, 62 | files: list[str], 63 | ) -> str: 64 | """ 65 | 66 | Parameters 67 | ---------- 68 | page : playwright.async_api.Page 69 | 页面 70 | prompt : str 71 | 问题 72 | create : bool 73 | 是否创建新的对话 74 | files : list[str] | None 75 | 上传的文件列表。目前仅支持上传图片 76 | 77 | Returns 78 | ------- 79 | str 80 | 回答 81 | """ 82 | logger.warning("纳米搜索。不登录可以使用。但无头模式要指定`user_data_dir`才能正常工作") 83 | 84 | if not create: 85 | if not page.url.startswith(_PAGE1_): 86 | create = True 87 | if len(files) > 0: 88 | create = True 89 | 90 | for file in files: 91 | assert is_image(file), f"仅支持上传图片，{file}不是图片" 92 | 93 | if create: 94 | name = "输入任何问题" 95 | 96 | await page.goto(_PAGE0_) 97 | if len(files) > 0: 98 | # 只能在新会话中上传文件 99 | async with page.expect_response(_PAGE3_, timeout=mcp_query_table.TIMEOUT_60) as response_info: 100 | await page.locator("input[type=\"file\"]").set_input_files(files) 101 | else: 102 | name = "提出后续问题，Enter发送，Shift+Enter 换行" 103 | 104 | async with page.expect_response(_PAGE2_, timeout=mcp_query_table.TIMEOUT) as response_info: 105 | textbox = page.get_by_role("textbox", name=name) 106 | await textbox.fill(prompt) 107 | await textbox.press("Enter") 108 | # await page.screenshot(path="n.png") 109 | await on_response(await response_info.value) 110 | 111 | return G.get_text() 112 | -------------------------------------------------------------------------------- /mcp_query_table/providers/yuanbao.py: -------------------------------------------------------------------------------- 1 | """ 2 | 腾讯元宝 3 | """ 4 | import json 5 | import re 6 | 7 | from loguru import logger 8 | from playwright.async_api import Page 9 | 10 | import mcp_query_table 11 | from mcp_query_table.utils import split_images, GlobalVars 12 | 13 | _PAGE0_ = "https://yuanbao.tencent.com/" 14 | _PAGE1_ = "https://yuanbao.tencent.com/api/chat" 15 | _PAGE2_ = "https://yuanbao.tencent.com/api/resource/genUploadInfo" 16 | 17 | G = GlobalVars() 18 | 19 | 20 | def read_event_stream(text): 21 | text1 = [] 22 | text2 = [] 23 | for event in text.split('\n\n'): 24 | if 'data: {"type":"think"' in event: 25 | lines = event.split('\n') 26 | for line in lines: 27 | if line.startswith('data: '): 28 | t = line[6:] 29 | t = json.loads(t)['content'] 30 | text1.append(t) 31 | if 'data: {"type":"text"' in event: 32 | lines = event.split('\n') 33 | for line in lines: 34 | if line.startswith('data: '): 35 | t = line[6:] 36 | t = json.loads(t).get('msg', "") 37 | text2.append(t) 38 | 39 | text2 = ''.join(text2) 40 | if len(text1) == 0: 41 | return text2 42 | else: 43 | text1 = ''.join(text1) 44 | return f"{text1}\n\n{text2}" 45 | 46 | 47 | async def on_response(response): 48 | if response.url.startswith(_PAGE1_): 49 | # print("on_response", response.url) 50 | text = await response.text() 51 | G.set_text(read_event_stream(text)) 52 | 53 | 54 | async def on_route(route): 55 | # print("on_route", route.request.url) 56 | if route.request.url.startswith(_PAGE1_): 57 | # TODO 这里会导致数据全部加载，逻辑变了，所以界面可能混乱 58 | response = await route.fetch(timeout=mcp_query_table.TIMEOUT) 59 | await route.fulfill( 60 | # 强行加utf-8，否则编码搞不定 61 | content_type="text/event-stream; charset=utf-8", 62 | response=response, 63 | ) 64 | else: 65 | await route.continue_() 66 | 67 | 68 | async def chat(page: Page, 69 | prompt: str, 70 | create: bool, 71 | files: list[str] 72 | ) -> str: 73 | logger.info("腾讯元宝。登录才可以使用。无头模式时要指定`user_data_dir`才能正常工作") 74 | 75 | if not page.url.startswith(_PAGE0_): 76 | create = True 77 | 78 | if create: 79 | await page.goto(_PAGE0_) 80 | 81 | if len(files) > 0: 82 | imgs, docs = split_images(files) 83 | assert len(imgs) == 0 or len(docs) == 0, "不能同时包含图片和文档" 84 | 85 | # 点击上传文件按钮，才会出现上传文件的input 86 | await page.get_by_role("button").filter(has_text=re.compile(r"^$")).last.click() 87 | 88 | # 上传文件 89 | async with page.expect_response(_PAGE2_, timeout=mcp_query_table.TIMEOUT_60) as response_info: 90 | if len(imgs) > 0: 91 | await page.locator("input[type=\"file\"]").nth(-2).set_input_files(files) 92 | else: 93 | await page.locator("input[type=\"file\"]").last.set_input_files(files) 94 | 95 | # 提问 96 | await page.route(f"{_PAGE1_}/*", on_route) 97 | async with page.expect_response(f"{_PAGE1_}/*", timeout=mcp_query_table.TIMEOUT) as response_info: 98 | textbox = page.locator(".ql-editor") 99 | await textbox.fill(prompt) 100 | await textbox.press("Enter") 101 | await on_response(await response_info.value) 102 | 103 | return G.get_text() 104 | -------------------------------------------------------------------------------- /mcp_query_table/server.py: -------------------------------------------------------------------------------- 1 | from typing import Annotated, List 2 | 3 | import fastmcp 4 | from loguru import logger 5 | from pydantic import Field 6 | 7 | from mcp_query_table import QueryType, Site, query as qt_query, chat as qt_chat 8 | from mcp_query_table.enums import Provider 9 | from mcp_query_table.tool import BrowserManager 10 | 11 | 12 | class QueryServer: 13 | def __init__(self) -> None: 14 | self.format: str = "markdown" 15 | self.browser = None 16 | 17 | def start(self, format, endpoint, executable_path, user_data_dir): 18 | self.format: str = format 19 | self.browser = BrowserManager(endpoint=endpoint, 20 | executable_path=executable_path, 21 | user_data_dir=user_data_dir, 22 | devtools=False, 23 | headless=True) 24 | 25 | async def query(self, query_input: str, query_type: QueryType, max_page: int, rename: bool, site: Site): 26 | page = await self.browser.get_page() 27 | df = await qt_query(page, query_input, query_type, max_page, rename, site) 28 | self.browser.release_page(page) 29 | 30 | if self.format == 'csv': 31 | return df.to_csv() 32 | if self.format == 'markdown': 33 | return df.to_markdown() 34 | if self.format == 'json': 35 | return df.to_json(force_ascii=False, indent=2) 36 | 37 | async def chat(self, prompt: str, create: bool, files: List[str], provider: Provider): 38 | page = await self.browser.get_page() 39 | txt = await qt_chat(page, prompt, create, files, provider) 40 | self.browser.release_page(page) 41 | return txt 42 | 43 | 44 | mcp = fastmcp.FastMCP("query_table_mcp") 45 | qsv = QueryServer() 46 | 47 | 48 | @mcp.tool(description="查询金融表格数据") 49 | async def query( 50 | query_input: Annotated[ 51 | str, Field(description="查询条件。支持复杂查询，如：`2024年涨幅最大的100只股票按市值排名`")], 52 | query_type: Annotated[QueryType, Field(default=QueryType.CNStock, 53 | description="查询类型。支持`A股`、`指数`、`基金`、`港股`、`美股`等")], 54 | max_page: Annotated[int, Field(default=1, ge=1, le=10, description="最大页数。只查第一页即可")], 55 | rename: Annotated[bool, Field(default=False, description="是否重命名列名")], 56 | site: Annotated[Site, Field(default=Site.THS, description="站点。支持`东方财富`、`通达信`、`同花顺`")] 57 | ) -> str: 58 | return await qsv.query(query_input, query_type, max_page, rename, site) 59 | 60 | 61 | # chat功能不通过mcp暴露，因为在Cline等客户端中本就有LLM功能，反而导致返回的数据没有正确提交 62 | # @mcp.tool(description="大语言模型对话") 63 | async def chat( 64 | prompt: Annotated[str, Field(description="提示词。如：`9.9大还是9.11大？`")], 65 | create: Annotated[bool, Field(default=False, description="是否创建新对话")], 66 | files: Annotated[List[str], Field(default=None, description="上传的文件列表。不同网站支持程度不同")], 67 | provider: Annotated[ 68 | Provider, Field(default=Provider.Nami, description="提供商。支持`纳米搜索`、`腾讯元宝`、`百度AI搜索`")] 69 | ) -> str: 70 | return await qsv.chat(prompt, create, files, provider) 71 | 72 | 73 | def serve(format, endpoint, executable_path, user_data_dir, transport, host, port): 74 | qsv.start(format, endpoint, executable_path, user_data_dir) 75 | logger.info(f"{endpoint=}") 76 | logger.info(f"{executable_path=}") 77 | logger.info(f"{user_data_dir=}") 78 | 79 | if transport == "stdio": 80 | logger.info(f"{transport=},{format=}") 81 | mcp.run(transport=transport) 82 | else: 83 | logger.info(f"{transport=},{format=},{host=},{port=}") 84 | mcp.run(transport=transport, host=host, port=port) 85 | -------------------------------------------------------------------------------- /mcp_query_table/providers/baidu.py: -------------------------------------------------------------------------------- 1 | """ 2 | 百度AI搜索 3 | 4 | 限制了输入长度为5000，很多时候会被截断，导致MCP无法正常工作 5 | """ 6 | import json 7 | 8 | from playwright.async_api import Page 9 | 10 | import mcp_query_table 11 | from mcp_query_table.utils import split_images, GlobalVars 12 | 13 | _PAGE0_ = "https://chat.baidu.com/search" 14 | _PAGE1_ = "https://chat.baidu.com/aichat/api/conversation" 15 | _PAGE2_ = "https://chat.baidu.com/aichat/api/file/upload" 16 | 17 | G = GlobalVars() 18 | 19 | 20 | def read_event_stream(text): 21 | text1 = [] 22 | text2 = [] 23 | for event in text.split('\n\n'): 24 | if '"component":"thinkingSteps"' in event: 25 | if '"reasoningContent":' not in event: 26 | continue 27 | lines = event.split('\n') 28 | for line in lines: 29 | if line.startswith('data:'): 30 | t = line[5:] 31 | t = json.loads(t)['data']['message']['content']['generator']['data']['reasoningContent'] 32 | text1.append(t) 33 | if '"component":"markdown-yiyan"' in event: 34 | lines = event.split('\n') 35 | for line in lines: 36 | if line.startswith('data:'): 37 | t = line[5:] 38 | t = json.loads(t)['data']['message']['content']['generator']['data']['value'] 39 | text2.append(t) 40 | 41 | text2 = ''.join(text2) 42 | if len(text1) == 0: 43 | return text2 44 | else: 45 | text1 = ''.join(text1) 46 | return f"{text1}\n\n{text2}" 47 | 48 | 49 | async def on_response(response): 50 | if response.url.startswith(_PAGE1_): 51 | # print("on_response", response.url) 52 | text = await response.text() 53 | G.set_text(read_event_stream(text)) 54 | 55 | 56 | async def on_route(route): 57 | # 避免出现 Protocol error (Network.getResponseBody): No data found for resource with given identifier 58 | # print("on_route", route.request.url) 59 | if route.request.url == _PAGE1_: 60 | # TODO 为何只要转发一下就没事了？ 61 | response = await route.fetch(timeout=mcp_query_table.TIMEOUT) 62 | await route.fulfill(response=response) 63 | else: 64 | await route.continue_() 65 | 66 | 67 | async def chat(page: Page, 68 | prompt: str, 69 | create: bool, 70 | files: list[str], 71 | ) -> str: 72 | async def on_file_chooser(file_chooser): 73 | # 文件选择对话框 74 | await file_chooser.set_files(files) 75 | 76 | if not page.url.startswith(_PAGE0_): 77 | create = True 78 | 79 | if create: 80 | await page.goto(_PAGE0_) 81 | 82 | # 文件上传 83 | if len(files) > 0: 84 | imgs, docs = split_images(files) 85 | assert len(imgs) == 0 or len(docs) == 0, "不能同时包含图片和文档" 86 | 87 | page.on("filechooser", on_file_chooser) 88 | async with page.expect_response(f"{_PAGE2_}*", timeout=mcp_query_table.TIMEOUT_60) as response_info: 89 | if len(imgs) > 0: 90 | await page.locator(".cs-input-upload-icon").last.click() 91 | else: 92 | await page.locator(".cs-input-upload-icon").first.click() 93 | page.remove_listener("filechooser", on_file_chooser) 94 | 95 | # 提交问题 96 | await page.route(_PAGE1_, on_route) 97 | async with page.expect_response(_PAGE1_, timeout=mcp_query_table.TIMEOUT) as response_info: 98 | await page.locator("#chat-input-box").fill(prompt) 99 | await page.locator("#chat-input-box").press("Enter") 100 | await on_response(await response_info.value) 101 | 102 | return G.get_text() 103 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | share/python-wheels/ 24 | *.egg-info/ 25 | .installed.cfg 26 | *.egg 27 | MANIFEST 28 | 29 | # PyInstaller 30 | # Usually these files are written by a python script from a template 31 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 32 | *.manifest 33 | *.spec 34 | 35 | # Installer logs 36 | pip-log.txt 37 | pip-delete-this-directory.txt 38 | 39 | # Unit test / coverage reports 40 | htmlcov/ 41 | .tox/ 42 | .nox/ 43 | .coverage 44 | .coverage.* 45 | .cache 46 | nosetests.xml 47 | coverage.xml 48 | *.cover 49 | *.py,cover 50 | .hypothesis/ 51 | .pytest_cache/ 52 | cover/ 53 | 54 | # Translations 55 | *.mo 56 | *.pot 57 | 58 | # Django stuff: 59 | *.log 60 | local_settings.py 61 | db.sqlite3 62 | db.sqlite3-journal 63 | 64 | # Flask stuff: 65 | instance/ 66 | .webassets-cache 67 | 68 | # Scrapy stuff: 69 | .scrapy 70 | 71 | # Sphinx documentation 72 | docs/_build/ 73 | 74 | # PyBuilder 75 | .pybuilder/ 76 | target/ 77 | 78 | # Jupyter Notebook 79 | .ipynb_checkpoints 80 | 81 | # IPython 82 | profile_default/ 83 | ipython_config.py 84 | 85 | # pyenv 86 | # For a library or package, you might want to ignore these files since the code is 87 | # intended to run in multiple environments; otherwise, check them in: 88 | # .python-version 89 | 90 | # pipenv 91 | # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control. 92 | # However, in case of collaboration, if having platform-specific dependencies or dependencies 93 | # having no cross-platform support, pipenv may install dependencies that don't work, or not 94 | # install all needed dependencies. 95 | #Pipfile.lock 96 | 97 | # UV 98 | # Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control. 99 | # This is especially recommended for binary packages to ensure reproducibility, and is more 100 | # commonly ignored for libraries. 101 | #uv.lock 102 | 103 | # poetry 104 | # Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control. 105 | # This is especially recommended for binary packages to ensure reproducibility, and is more 106 | # commonly ignored for libraries. 107 | # https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control 108 | #poetry.lock 109 | 110 | # pdm 111 | # Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control. 112 | #pdm.lock 113 | # pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it 114 | # in version control. 115 | # https://pdm.fming.dev/latest/usage/project/#working-with-version-control 116 | .pdm.toml 117 | .pdm-python 118 | .pdm-build/ 119 | 120 | # PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm 121 | __pypackages__/ 122 | 123 | # Celery stuff 124 | celerybeat-schedule 125 | celerybeat.pid 126 | 127 | # SageMath parsed files 128 | *.sage.py 129 | 130 | # Environments 131 | .env 132 | .venv 133 | env/ 134 | venv/ 135 | ENV/ 136 | env.bak/ 137 | venv.bak/ 138 | 139 | # Spyder project settings 140 | .spyderproject 141 | .spyproject 142 | 143 | # Rope project settings 144 | .ropeproject 145 | 146 | # mkdocs documentation 147 | /site 148 | 149 | # mypy 150 | .mypy_cache/ 151 | .dmypy.json 152 | dmypy.json 153 | 154 | # Pyre type checker 155 | .pyre/ 156 | 157 | # pytype static type analyzer 158 | .pytype/ 159 | 160 | # Cython debug symbols 161 | cython_debug/ 162 | 163 | # PyCharm 164 | # JetBrains specific template is maintained in a separate JetBrains.gitignore that can 165 | # be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore 166 | # and can be added to the global gitignore or merged into this file. For a more nuclear 167 | # option (not recommended) you can uncomment the following to ignore the entire idea folder. 168 | #.idea/ 169 | 170 | # Ruff stuff: 171 | .ruff_cache/ 172 | 173 | # PyPI configuration file 174 | .pypirc 175 | -------------------------------------------------------------------------------- /mcp_query_table/sites/tdx.py: -------------------------------------------------------------------------------- 1 | """ 2 | 通达信 小达 3 | https://wenda.tdx.com.cn/ 4 | """ 5 | import math 6 | import re 7 | 8 | import pandas as pd 9 | from loguru import logger 10 | from playwright.async_api import Page 11 | 12 | from mcp_query_table.enums import QueryType 13 | 14 | # 查询结果 15 | _PAGE1_ = 'https://wenda.tdx.com.cn/TQL?Entry=NLPSE.NLPQuery' 16 | # 代码数量 17 | _PAGE2_ = 'https://wenda.tdx.com.cn/TQL?Entry=JNLPSE.getAllCode' 18 | 19 | _queryType_ = { 20 | QueryType.CNStock: 'AG', 21 | QueryType.Fund: 'JJ', 22 | QueryType.Index: 'ZS', 23 | QueryType.Info: 'ZX', 24 | QueryType.Board: 'ZS', # 板块也走指数 25 | } 26 | 27 | 28 | def convert_type(type): 29 | if type == '': 30 | return str 31 | if type == '0|0|0': 32 | return str 33 | if type == '2|0|0': 34 | return float 35 | if type == '0|9|1': 36 | return float 37 | if type == '1|9|1': 38 | return float 39 | if type == '2|9|1': 40 | return float 41 | return type 42 | 43 | 44 | class Pagination: 45 | def __init__(self): 46 | self.datas = {} 47 | self.last_count = 1 48 | self.limit = 100 49 | self.row_count = 1024 50 | self.dtypes = [] 51 | self.columns = [] 52 | 53 | def reset(self): 54 | self.datas = {} 55 | 56 | def update_row_count(self, row_count): 57 | self.row_count = row_count 58 | 59 | def update_last_count(self, limit, last_count, columns, dtypes, datas): 60 | self.limit = limit 61 | self.last_count = last_count 62 | self.columns = columns 63 | self.dtypes = dtypes 64 | self.datas[last_count] = datas 65 | 66 | def has_next(self, max_page): 67 | page = math.ceil(self.last_count / self.limit) 68 | c1 = self.last_count < self.row_count 69 | c2 = page < max_page 70 | return c1 & c2 71 | 72 | def current(self): 73 | return self.last_count 74 | 75 | def get_list(self): 76 | datas = [] 77 | for k, v in self.datas.items(): 78 | datas.extend(v) 79 | return datas 80 | 81 | def get_dataframe(self, rename: bool): 82 | dtypes = [convert_type(x) for x in self.dtypes] 83 | df = pd.DataFrame(self.get_list(), columns=self.columns) 84 | for i, v in enumerate(dtypes): 85 | k = self.columns[i] 86 | if k == 'POS': 87 | df[k] = df[k].astype(int) 88 | continue 89 | if isinstance(v, str): 90 | logger.info("未识别的数据类型 {}:{}", k, v) 91 | continue 92 | try: 93 | df[k] = df[k].astype(v) 94 | except ValueError: 95 | logger.info("转换失败 {}:{}", k, v) 96 | return df 97 | 98 | 99 | P = Pagination() 100 | 101 | 102 | def NLPQuery(json_data): 103 | limit = json_data[0][2] 104 | last_count = int(json_data[0][4]) 105 | columns = json_data[1] 106 | dtypes = json_data[2] 107 | datas = json_data[3:] 108 | 109 | return limit, last_count, columns, dtypes, datas 110 | 111 | 112 | def getAllCode(json_data): 113 | row_count = json_data[0][2] 114 | 115 | return row_count 116 | 117 | 118 | async def on_response1(response): 119 | if response.url.startswith(_PAGE1_): 120 | P.update_last_count(*NLPQuery(await response.json())) 121 | 122 | 123 | async def on_response2(response): 124 | if response.url.startswith(_PAGE2_): 125 | P.update_row_count(getAllCode(await response.json())) 126 | 127 | 128 | async def query(page: Page, 129 | message: str = "收盘价>100元", 130 | type_: QueryType = 'AG', 131 | max_page: int = 5, 132 | rename: bool = False) -> pd.DataFrame: 133 | queryType = _queryType_.get(type_, None) 134 | assert queryType is not None, f"不支持的类型:{type_}" 135 | 136 | await page.route(re.compile(r'.*\.(?:jpg|jpeg|png|gif|webp)(?:$|\?)'), lambda route: route.abort()) 137 | page.on("response", on_response2) 138 | 139 | P.reset() 140 | async with page.expect_response(lambda response: response.url.startswith(_PAGE1_)) as response_info: 141 | await page.goto(f"https://wenda.tdx.com.cn/site/wenda/stock_index.html?message={message}&queryType={queryType}", 142 | wait_until="load") 143 | await on_response1(await response_info.value) 144 | 145 | while P.has_next(max_page): 146 | logger.info("当前序号为:{}, 点击`下一页`", P.current()) 147 | async with page.expect_response(lambda response: response.url.startswith(_PAGE1_)) as response_info: 148 | await page.get_by_role("button", name="下一页").click() 149 | await on_response1(await response_info.value) 150 | 151 | return P.get_dataframe(rename) 152 | -------------------------------------------------------------------------------- /streamlit/client.py: -------------------------------------------------------------------------------- 1 | """ 2 | MCP SSE Client - A Python client for interacting with Model Context Protocol (MCP) endpoints. 3 | 4 | This module provides a client for connecting to MCP endpoints using Server-Sent Events (SSE), 5 | listing available tools, and invoking tools with parameters. 6 | """ 7 | 8 | from dataclasses import dataclass 9 | from typing import Any, Dict, List, Optional 10 | from urllib.parse import urlparse 11 | 12 | from mcp import ClientSession 13 | from mcp.client.sse import sse_client 14 | from mcp.types import CallToolResult 15 | 16 | 17 | @dataclass 18 | class ToolParameter: 19 | """Represents a parameter for a tool. 20 | 21 | Attributes: 22 | name: Parameter name 23 | parameter_type: Parameter type (e.g., "string", "number") 24 | description: Parameter description 25 | required: Whether the parameter is required 26 | default: Default value for the parameter 27 | """ 28 | name: str 29 | parameter_type: str 30 | description: str 31 | required: bool = False 32 | default: Any = None 33 | 34 | 35 | @dataclass 36 | class ToolDef: 37 | """Represents a tool definition. 38 | 39 | Attributes: 40 | name: Tool name 41 | description: Tool description 42 | parameters: List of ToolParameter objects 43 | metadata: Optional dictionary of additional metadata 44 | identifier: Tool identifier (defaults to name) 45 | """ 46 | name: str 47 | description: str 48 | parameters: List[ToolParameter] 49 | metadata: Optional[Dict[str, Any]] = None 50 | identifier: str = "" 51 | 52 | 53 | @dataclass 54 | class ToolInvocationResult: 55 | """Represents the result of a tool invocation. 56 | 57 | Attributes: 58 | content: Result content as a string 59 | error_code: Error code (0 for success, 1 for error) 60 | """ 61 | content: str 62 | error_code: int 63 | 64 | 65 | class MCPClient: 66 | """Client for interacting with Model Context Protocol (MCP) endpoints""" 67 | 68 | def __init__(self, endpoint: str): 69 | """Initialize MCP client with endpoint URL 70 | 71 | Args: 72 | endpoint: The MCP endpoint URL (must be http or https) 73 | """ 74 | if urlparse(endpoint).scheme not in ("http", "https"): 75 | raise ValueError(f"Endpoint {endpoint} is not a valid HTTP(S) URL") 76 | self.endpoint = endpoint 77 | 78 | async def list_tools(self) -> List[ToolDef]: 79 | """List available tools from the MCP endpoint 80 | 81 | Returns: 82 | List of ToolDef objects describing available tools 83 | """ 84 | tools = [] 85 | async with sse_client(self.endpoint) as streams: 86 | async with ClientSession(*streams) as session: 87 | await session.initialize() 88 | tools_result = await session.list_tools() 89 | 90 | for tool in tools_result.tools: 91 | parameters = [] 92 | required_params = tool.inputSchema.get("required", []) 93 | for param_name, param_schema in tool.inputSchema.get("properties", {}).items(): 94 | parameters.append( 95 | ToolParameter( 96 | name=param_name, 97 | parameter_type=param_schema.get("type", "string"), 98 | description=param_schema.get("description", ""), 99 | required=param_name in required_params, 100 | default=param_schema.get("default"), 101 | ) 102 | ) 103 | tools.append( 104 | ToolDef( 105 | name=tool.name, 106 | description=tool.description, 107 | parameters=parameters, 108 | metadata={"endpoint": self.endpoint}, 109 | identifier=tool.name # Using name as identifier 110 | ) 111 | ) 112 | return tools 113 | 114 | async def invoke_tool(self, tool_name: str, kwargs: Dict[str, Any]) -> CallToolResult: 115 | """Invoke a specific tool with parameters 116 | 117 | Args: 118 | tool_name: Name of the tool to invoke 119 | kwargs: Dictionary of parameters to pass to the tool 120 | 121 | Returns: 122 | ToolInvocationResult containing the tool's response 123 | """ 124 | async with sse_client(self.endpoint) as streams: 125 | async with ClientSession(*streams) as session: 126 | await session.initialize() 127 | return await session.call_tool(tool_name, kwargs) 128 | -------------------------------------------------------------------------------- /mcp_query_table/sites/eastmoney.py: -------------------------------------------------------------------------------- 1 | """ 2 | 东方财富 条件选股 3 | https://xuangu.eastmoney.com/ 4 | 5 | 1. 部分数据中包含中文单位，如万亿等，导致无法转换为数字，如VOLUME 6 | 2. 东财翻页需要提前手工登录 7 | 3. 东财翻页是页面已经翻了，然后等数据来更新，懒加载 8 | """ 9 | import re 10 | 11 | import pandas as pd 12 | from loguru import logger 13 | from playwright.async_api import Page 14 | 15 | from mcp_query_table.enums import QueryType 16 | 17 | # 查询结果 18 | # 'https://np-pick-b.eastmoney.com/api/smart-tag/stock/v3/pw/search-code' 19 | # 'https://np-pick-b.eastmoney.com/api/smart-tag/fund/v3/pw/search-code' 20 | # 'https://np-pick-b.eastmoney.com/api/smart-tag/hk/v3/pw/search-code' 21 | # 'https://np-pick-b.eastmoney.com/api/smart-tag/cb/v3/pw/search-code' 22 | # 'https://np-pick-b.eastmoney.com/api/smart-tag/etf/v3/pw/search-code' 23 | # 'https://np-pick-b.eastmoney.com/api/smart-tag/bkc/v3/pw/search-code' 24 | # 'https://np-tjxg-b.eastmoney.com/api/smart-tag/bkc/v3/pw/search-code' 25 | _PAGE1_ = 'https://*.eastmoney.com/api/smart-tag/*/v3/pw/search-code' 26 | 27 | _type_ = { 28 | QueryType.CNStock: 'stock', 29 | QueryType.Fund: 'fund', 30 | QueryType.HKStock: 'hk', 31 | QueryType.ConBond: 'cb', 32 | QueryType.ETF: 'etf', 33 | QueryType.Board: 'bk', # 比较坑，bkc和bkc的区别 34 | } 35 | 36 | 37 | def convert_type(type): 38 | if type == 'Double': 39 | return float 40 | if type == 'String': 41 | return str 42 | if type == 'Long': 43 | return int 44 | if type == 'Boolean': 45 | return bool 46 | if type == 'INT': # TODO 好像未出现过 47 | return int 48 | return type 49 | 50 | 51 | class Pagination: 52 | def __init__(self): 53 | self.datas = {} 54 | self.pageNo = 1 55 | self.pageSize = 100 56 | self.total = 1024 57 | self.columns = [] 58 | self.datas = {} 59 | 60 | def reset(self): 61 | self.datas = {} 62 | 63 | def update(self, pageNo, pageSize, total, columns, dataList): 64 | self.pageNo = pageNo 65 | self.pageSize = pageSize 66 | self.total = total 67 | self.columns = columns 68 | self.datas[self.pageNo] = dataList 69 | 70 | def has_next(self, max_page): 71 | c1 = self.pageNo * self.pageSize < self.total 72 | c2 = self.pageNo < max_page 73 | return c1 & c2 74 | 75 | def current(self): 76 | return self.pageNo 77 | 78 | def get_list(self): 79 | datas = [] 80 | for k, v in self.datas.items(): 81 | datas.extend(v) 82 | return datas 83 | 84 | def get_dataframe(self, rename: bool): 85 | columns = {x['key']: x['title'] for x in self.columns} 86 | dtypes = {x['key']: convert_type(x['dataType']) for x in self.columns} 87 | 88 | df = pd.DataFrame(self.get_list()) 89 | for k, v in dtypes.items(): 90 | if k == 'SERIAL': 91 | df[k] = df[k].astype(int) 92 | continue 93 | if isinstance(v, str): 94 | logger.info("未识别的数据类型 {}:{}", k, v) 95 | continue 96 | try: 97 | df[k] = df[k].astype(v) 98 | except ValueError: 99 | logger.info("转换失败 {}:{}", k, v) 100 | 101 | if rename: 102 | return df.rename(columns=columns) 103 | else: 104 | return df 105 | 106 | 107 | P = Pagination() 108 | 109 | 110 | def search_code(json_data): 111 | total = json_data['data']['result']['total'] 112 | columns = json_data['data']['result']['columns'] 113 | dataList = json_data['data']['result']['dataList'] 114 | return total, columns, dataList 115 | 116 | 117 | async def on_response(response): 118 | post_data_json = response.request.post_data_json 119 | pageNo = post_data_json['pageNo'] 120 | pageSize = post_data_json['pageSize'] 121 | P.update(pageNo, pageSize, *search_code(await response.json())) 122 | 123 | 124 | async def query(page: Page, 125 | q: str = "收盘价>100元", 126 | type_: QueryType = 'stock', 127 | max_page: int = 5, 128 | rename: bool = True) -> pd.DataFrame: 129 | type = _type_.get(type_, None) 130 | assert type is not None, f"不支持的类型:{type_}" 131 | 132 | await page.route(re.compile(r'.*\.(?:jpg|jpeg|png|gif|webp)(?:$|\?)'), lambda route: route.abort()) 133 | 134 | P.reset() 135 | async with page.expect_response(_PAGE1_) as response_info: 136 | # 这里不用处理输入编码问题 137 | await page.goto(f"https://xuangu.eastmoney.com/Result?q={q}&type={type}", wait_until="load") 138 | await on_response(await response_info.value) 139 | 140 | while P.has_next(max_page): 141 | logger.info("当前页为:{}, 点击`下一页`", P.current()) 142 | 143 | # 这种写法解决了懒加载问题 144 | async with page.expect_response(_PAGE1_) as response_info: 145 | await page.get_by_role("button", name="下一页").click() 146 | await on_response(await response_info.value) 147 | 148 | return P.get_dataframe(rename) 149 | -------------------------------------------------------------------------------- /mcp_query_table/sites/iwencai.py: -------------------------------------------------------------------------------- 1 | """ 2 | 同花顺问财 3 | https://www.iwencai.com/ 4 | 5 | 1. 一定要保证浏览器宽度>768，防止界面变成适应手机 6 | 7 | """ 8 | import re 9 | 10 | import pandas as pd 11 | from loguru import logger 12 | from playwright.async_api import Page 13 | 14 | from mcp_query_table.enums import QueryType 15 | 16 | # 初次查询页面 17 | _PAGE1_ = 'https://www.iwencai.com/customized/chart/get-robot-data' 18 | # 翻页 19 | _PAGE2_ = 'https://www.iwencai.com/gateway/urp/v7/landing/getDataList' 20 | 21 | _querytype_ = { 22 | QueryType.CNStock: 'stock', 23 | QueryType.Index: 'zhishu', 24 | QueryType.Fund: 'fund', 25 | QueryType.HKStock: 'hkstock', 26 | QueryType.USStock: 'usstock', 27 | '新三板': 'threeboard', 28 | QueryType.ConBond: 'conbond', 29 | '保险': 'insurance', 30 | '期货': 'futures', 31 | '理财': 'lccp', 32 | '外汇': 'foreign_exchange', 33 | '宏观': 'macro', 34 | # 35 | QueryType.ETF: 'fund', # 查ETF定位到基金 36 | } 37 | 38 | 39 | def convert_type(type): 40 | if type == 'LONG': 41 | return int 42 | if type == 'DOUBLE': 43 | return float 44 | if type == 'STR': 45 | return str 46 | if type == 'INT': # TODO 好像未出现过 47 | return int 48 | return type 49 | 50 | 51 | class Pagination: 52 | def __init__(self): 53 | self.datas = {} 54 | self.limit = 100 55 | self.page = 1 56 | self.row_count = 1024 57 | self.columns = [] 58 | 59 | def reset(self): 60 | self.datas = {} 61 | 62 | def update(self, datas, columns, page, limit, row_count): 63 | self.datas[page] = datas 64 | self.columns = columns 65 | self.limit = limit 66 | self.page = page 67 | self.row_count = row_count 68 | 69 | def has_next(self, max_page): 70 | c1 = self.page * self.limit < self.row_count 71 | c2 = self.page < max_page 72 | return c1 & c2 73 | 74 | def current(self): 75 | return self.page 76 | 77 | def get_list(self): 78 | datas = [] 79 | for k, v in self.datas.items(): 80 | datas.extend(v) 81 | return datas 82 | 83 | def get_dataframe(self, rename: bool): 84 | columns = {x['key']: x['index_name'] for x in self.columns} 85 | dtypes = {x['key']: convert_type(x['type']) for x in self.columns} 86 | 87 | df = pd.DataFrame(self.get_list()) 88 | for k, v in dtypes.items(): 89 | if isinstance(v, str): 90 | logger.info("未识别的数据类型 {}:{}", k, v) 91 | continue 92 | try: 93 | df[k] = df[k].astype(v) 94 | except ValueError: 95 | logger.info("转换失败 {}:{}", k, v) 96 | 97 | if rename: 98 | return df.rename(columns=columns) 99 | else: 100 | return df 101 | 102 | 103 | P = Pagination() 104 | 105 | 106 | def get_robot_data(json_data): 107 | """ 108 | json_data['data']['answer'][0]['txt'][0]['content']['components'][0]['data']['datas'] 109 | json_data['data']['answer'][0]['txt'][0]['content']['components'][0]['data']['meta']['limit'] 100 110 | json_data['data']['answer'][0]['txt'][0]['content']['components'][0]['data']['meta']['page'] 1 111 | json_data['data']['answer'][0]['txt'][0]['content']['components'][0]['data']['meta']['extra']['row_count'] 1364 112 | """ 113 | _1 = json_data['data']['answer'][0]['txt'][0]['content']['components'][0]['data'] 114 | _2 = _1['meta'] 115 | 116 | datas = _1['datas'] 117 | columns = _1['columns'] 118 | page = _2['page'] 119 | limit = _2['limit'] 120 | row_count = _2['extra']['row_count'] 121 | 122 | return datas, columns, page, limit, row_count 123 | 124 | 125 | def getDataList(json_data): 126 | """ 127 | json_data['answer']['components'][0]['data']['datas'] 128 | json_data['answer']['components'][0]['data']['meta']['page'] 129 | json_data['answer']['components'][0]['data']['meta']['limit'] 130 | json_data['answer']['components'][0]['data']['meta']['extra']['row_count'] 131 | """ 132 | _1 = json_data['answer']['components'][0]['data'] 133 | _2 = _1['meta'] 134 | 135 | datas = _1['datas'] 136 | columns = _1['columns'] 137 | page = _2['page'] 138 | limit = _2['limit'] 139 | row_count = _2['extra']['row_count'] 140 | 141 | return datas, columns, int(page), int(limit), row_count 142 | 143 | 144 | async def on_response(response): 145 | if response.url == _PAGE1_: 146 | P.update(*get_robot_data(await response.json())) 147 | if response.url == _PAGE2_: 148 | P.update(*getDataList(await response.json())) 149 | 150 | 151 | async def query(page: Page, 152 | w: str = "收盘价>1000元", 153 | type_: QueryType = 'stock', 154 | max_page: int = 5, 155 | rename: bool = False) -> pd.DataFrame: 156 | querytype = _querytype_.get(type_, None) 157 | assert querytype is not None, f"不支持的类型:{type_}" 158 | 159 | await page.route(re.compile(r'.*\.(?:jpg|jpeg|png|gif|webp)(?:$|\?)'), lambda route: route.abort()) 160 | 161 | P.reset() 162 | # page.viewport_size # 取出来是None 163 | # 宽度<=768会认为是手机,>768是PC 164 | await page.set_viewport_size({"width": 1280, "height": 800}) 165 | async with page.expect_response(_PAGE1_) as response_info: 166 | await page.goto(f"https://www.iwencai.com/unifiedwap/result?w={w}&querytype={querytype}", wait_until="load") 167 | await on_response(await response_info.value) 168 | 169 | while P.has_next(max_page): 170 | logger.info("当前页为:{}, 点击`下页`", P.current()) 171 | async with page.expect_response(_PAGE2_) as response_info: 172 | await page.get_by_text("下页").click() 173 | await on_response(await response_info.value) 174 | 175 | return P.get_dataframe(rename) 176 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # mcp_query_table 2 | 3 | 1. 基于`playwright`实现的财经网页表格爬虫，支持`Model Context Protocol (MCP) `。目前可查询来源为 4 | 5 | - [同花顺问财](http://iwencai.com/) 6 | - [通达信问小达](https://wenda.tdx.com.cn/) 7 | - [东方财富条件选股](https://xuangu.eastmoney.com/) 8 | 9 | 实盘时，如果某网站宕机或改版，可以立即切换到其他网站。(注意：不同网站的表格结构不同，需要提前做适配) 10 | 11 | 2. 基于`playwright`实现的大语言模型调用爬虫。目前可用来源为 12 | - [纳米搜索](https://www.n.cn/) 13 | - [腾讯元宝](https://yuanbao.tencent.com/) 14 | - [百度AI搜索](https://chat.baidu.com/) 15 | 16 | `RooCode`提供了`Human Reply`功能。但发现`纳米搜索`网页版复制时格式破坏，所以研发了此功能 17 | 18 | ## 安装 19 | 20 | ```commandline 21 | pip install -i https://pypi.org/simple --upgrade mcp_query_table 22 | pip install -i https://pypi.tuna.tsinghua.edu.cn/simple --upgrade mcp_query_table 23 | ``` 24 | 25 | ## 使用 26 | 27 | ```python 28 | import asyncio 29 | 30 | from mcp_query_table import * 31 | 32 | 33 | async def main() -> None: 34 | async with BrowserManager(endpoint="http://127.0.0.1:9222", executable_path=None, devtools=True) as bm: 35 | # 问财需要保证浏览器宽度>768，防止界面变成适应手机 36 | page = await bm.get_page() 37 | df = await query(page, '收益最好的200只ETF', query_type=QueryType.ETF, max_page=1, site=Site.THS) 38 | print(df.to_markdown()) 39 | df = await query(page, '年初至今收益率前50', query_type=QueryType.Fund, max_page=1, site=Site.TDX) 40 | print(df.to_csv()) 41 | df = await query(page, '流通市值前10的行业板块', query_type=QueryType.Index, max_page=1, site=Site.TDX) 42 | print(df.to_csv()) 43 | # TODO 东财翻页要提前登录 44 | df = await query(page, '今日涨幅前5的概念板块;', query_type=QueryType.Board, max_page=3, site=Site.EastMoney) 45 | print(df) 46 | 47 | output = await chat(page, "1+2等于多少？", provider=Provider.YuanBao) 48 | print(output) 49 | output = await chat(page, "3+4等于多少？", provider=Provider.YuanBao, create=True) 50 | print(output) 51 | 52 | print('done') 53 | bm.release_page(page) 54 | await page.wait_for_timeout(2000) 55 | 56 | 57 | if __name__ == '__main__': 58 | asyncio.run(main()) 59 | 60 | ``` 61 | 62 | ## 注意事项 63 | 64 | 1. 浏览器最好是`Chrome`。如一定要使用`Edge`,除了关闭`Edge`所有窗口外，还要在任务管理器关闭`Microsoft Edge` 65 | 的所有进程，即`taskkill /f /im msedge.exe` 66 | 2. 浏览器要保证窗口宽度，防止部分网站自动适配成手机版，导致表格查询失败 67 | 3. 如有网站账号，请提前登录。此工具无自动登录功能 68 | 4. 不同网站的表格结构不同，同条件返回股票数量也不同。需要查询后做适配 69 | 70 | ## 工作原理 71 | 72 | 不同于`requests`，`playwright`是基于浏览器的，模拟用户在浏览器中的操作。 73 | 74 | 1. 不需要解决登录问题 75 | 2. 不需要解决请求构造、响应解析 76 | 3. 可以直接获取表格数据，所见即所得 77 | 4. 运行速度慢于`requests`，但开发效率高 78 | 79 | 数据的获取有： 80 | 81 | 1. 直接解析HTML表格 82 | 1. 数字文本化了，不利于后期研究 83 | 2. 适用性最强 84 | 2. 截获请求，获取返回的`json`数据 85 | 1. 类似于`requests`，需要做响应解析 86 | 2. 灵活性差点，网站改版后，需要重新做适配 87 | 88 | 此项目采用的是模拟点击浏览器来发送请求，使用截获响应并解析的方法来获取数据。 89 | 90 | 后期会根据不同的网站改版情况，使用更适合的方法。 91 | 92 | ## 无头模式 93 | 94 | 无头模式运行速度更快，但部分网站需要提前登录，所以，无头模式一定要指定`user_data_dir`，否则会出现需要登录的情况。 95 | 96 | - `endpoint=None`时，`headless=True`可无头启动新浏览器实例。指定`executable_path`和`user_data_dir`，才能确保无头模式下正常运行。 97 | - `endpoint`以`http://`开头，连接`CDP`模式启动的有头浏览器，参数必有`--remote-debugging-port`。`executable_path`为本地浏览器路径。 98 | - `endpoint`以`ws://`开头，连接远程`Playwright Server`。也是无头模式，但无法指定`user_data_dir`，所以使用受限 99 | - 参考：https://playwright.dev/python/docs/docker#running-the-playwright-server 100 | 101 | `Chrome`新版的安全策略使用默认`user_data_dir`时将无法创建`CDP`服务，建议重新复制配置目录到其他地方 102 | 103 | ## MCP支持 104 | 105 | 确保可以在控制台中执行`python -m mcp_query_table -h`。如果不能，可能要先`pip install mcp_query_table` 106 | 107 | 在`Cline`中可以配置如下。其中`command`是`python`的绝对路径，`timeout`是超时时间，单位为秒。 在各`AI` 108 | 平台中由于返回时间常需1分钟以上，所以需要设置大的超时时间。 109 | 110 | ### STDIO方式 111 | 112 | ```json 113 | { 114 | "mcpServers": { 115 | "mcp_query_table": { 116 | "timeout": 300, 117 | "command": "D:\\Users\\Kan\\miniconda3\\envs\\py312\\python.exe", 118 | "args": [ 119 | "-m", 120 | "mcp_query_table", 121 | "--format", 122 | "markdown", 123 | "--endpoint", 124 | "http://127.0.0.1:9222", 125 | "--executable_path", 126 | "C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe" 127 | ] 128 | } 129 | } 130 | } 131 | ``` 132 | 133 | ### SSE方式 134 | 135 | 先在控制台中执行如下命令，启动`MCP`服务 136 | 137 | ```commandline 138 | python -m mcp_query_table --format markdown --transport sse --port 8000 --endpoint http://127.0.0.1:9222 --user_data_dir "D:\user-data-dir" 139 | ``` 140 | 141 | 然后就可以连接到`MCP`服务了 142 | 143 | ```json 144 | { 145 | "mcpServers": { 146 | "mcp_query_table": { 147 | "timeout": 300, 148 | "url": "http://127.0.0.1:8000/sse" 149 | } 150 | } 151 | } 152 | ``` 153 | 154 | ### Streamable HTTP方式 155 | 156 | ```commandline 157 | python -m mcp_query_table --format markdown --transport streamable-http --port 8000 --endpoint http://127.0.0.1:9222 --user_data_dir "D:\user-data-dir" 158 | ``` 159 | 160 | 连接的地址是`http://127.0.0.1:8000/mcp` 161 | 162 | ## 使用`MCP Inspector`进行调试 163 | 164 | ```commandline 165 | npx @modelcontextprotocol/inspector python -m mcp_query_table --format markdown --endpoint http://127.0.0.1:9222 166 | ``` 167 | 168 | 打开浏览器并翻页是一个比较耗时的操作，会导致`MCP Inspector`页面超时，可以`http://localhost:5173/?timeout=300000` 169 | 表示超时时间为300秒 170 | 171 | 第一次尝试编写`MCP`项目，可能会有各种问题，欢迎大家交流。 172 | 173 | ## `MCP`使用技巧 174 | 175 | 1. 2024年涨幅最大的100只股票按2024年12月31日总市值排名。三个网站的结果都不一样 176 | - 同花顺：显示了2201只股票。前5个是工商银行、农业银行、中国移动、中国石油、建设银行 177 | - 通达信：显示了100只股票，前5个是寒武纪、正丹股份，汇金科技、万丰奥威、艾融软件 178 | - 东方财富：显示了100只股票，前5个是海光信息、寒武纪、光启技术、润泽科技、新易盛 179 | 180 | 2. 大语言模型对问题拆分能力弱，所以要能合理的提问，保证查询条件不会被改动。以下推荐第2、3种 181 | - 2024年涨幅最大的100只股票按2024年12月31日总市值排名 182 | > 大语言模型非常有可能拆分这句，导致一步查询被分成了多步查询 183 | - 向东方财富查询“2024年涨幅最大的100只股票按2024年12月31日总市值排名” 184 | > 用引号括起来，避免被拆分 185 | - 向东方财富板块查询 “去年涨的最差的行业板块”，再查询此板块中去年涨的最好的5只股票 186 | > 分成两步查询，先查询板块，再查询股票。但最好不要全自动，因为第一步的结果它不理解“今日涨幅”和“区间涨幅”,需要交互修正 187 | 188 | ## 支持`Streamlit` 189 | 190 | 实现在同一页面中查询金融数据，并手工输入到`AI`中进行深度分析。参考`streamlit`目录下的`README.md`文件。 191 | 192 | ![streamlit](docs/img/streamlit.png) 193 | 194 | ## 参考 195 | 196 | - [Selenium webdriver无法附加到edge实例，edge的--remote-debugging-port选项无效](https://blog.csdn.net/qq_30576521/article/details/142370538) 197 | - https://github.com/AtuboDad/playwright_stealth/issues/31 198 | - https://github.com/browser-use/browser-use/issues/1520 -------------------------------------------------------------------------------- /streamlit/app.py: -------------------------------------------------------------------------------- 1 | import asyncio 2 | import hashlib 3 | import os 4 | import sys 5 | 6 | import streamlit as st 7 | import streamlit.components.v1 as components 8 | import streamlit_authenticator as stauth 9 | import yaml 10 | from streamlit_authenticator import LoginError 11 | 12 | # 添加当前目录和上一层目录到系统路径 13 | sys.path.append(os.path.dirname(os.path.abspath(__file__))) 14 | sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) 15 | 16 | from client import MCPClient 17 | from mcp_query_table import QueryType, Site 18 | 19 | Provders = { 20 | # "about:blank": "空白页", 21 | "https://yuanbao.tencent.com": "腾讯元宝 - 支持长文|DeepSeek", 22 | "https://chat.baidu.com": "百度AI搜索 - 长文限制|DeepSeek", 23 | "https://www.doubao.com/chat": "字节豆包 - 强制联网|支持长文|Doubao", 24 | "https://www.n.cn": "360纳米搜索 - 强制联网|支持长文|不支持文件|多模型", 25 | # "https://yiyan.baidu.com": "百度文心一言 - 支持长文|X1", 26 | # "https://chat.z.ai/": "智谱AI - 无法内嵌|不支持文件", 27 | # "https://tongyi.aliyun.com": "通义千问 - 无法内嵌|长文限制|QwQ", 28 | } 29 | 30 | Sites = { 31 | "https://xuangu.eastmoney.com": Site.EastMoney, # 翻页要登录，港股要登录 32 | "https://www.iwencai.com": Site.THS, 33 | "https://wenda.tdx.com.cn": Site.TDX, 34 | } 35 | 36 | QueryTypes = { 37 | "https://xuangu.eastmoney.com": [QueryType.CNStock, QueryType.Fund, QueryType.HKStock, QueryType.ConBond, 38 | QueryType.ETF, QueryType.Board], 39 | "https://www.iwencai.com": [QueryType.CNStock, QueryType.Index, QueryType.Fund, QueryType.HKStock, 40 | QueryType.USStock], 41 | "https://wenda.tdx.com.cn": [QueryType.CNStock, QueryType.Fund, QueryType.Index, QueryType.Info], 42 | } 43 | 44 | default_query = "涨幅前10" 45 | default_prompt = """你是一个专业的股票分析师。请忽略文件名，仅根据文件内容，为我提供专业分析报告。不用联网搜索。 46 | 47 | 文件内容如下：""" 48 | 49 | st.set_page_config(page_title='财经问答LLM', layout="wide", initial_sidebar_state="expanded") 50 | 51 | with open('auth.yaml', 'r', encoding='utf-8') as file: 52 | config = yaml.safe_load(file) 53 | 54 | # Pre-hashing all plain text passwords once 55 | stauth.Hasher.hash_passwords(config['credentials']) 56 | 57 | # Creating the authenticator object 58 | authenticator = stauth.Authenticate( 59 | config['credentials'], 60 | config['cookie']['name'], 61 | config['cookie']['key'], 62 | config['cookie']['expiry_days'] 63 | ) 64 | 65 | try: 66 | authenticator.login() 67 | except LoginError as e: 68 | st.error(e) 69 | 70 | if st.session_state['authentication_status'] is False: 71 | st.error('Username/password is incorrect') 72 | elif st.session_state['authentication_status'] is None: 73 | st.warning('Please enter your username and password') 74 | if not st.session_state['authentication_status']: 75 | st.stop() 76 | 77 | # Loading config file 78 | with open('config.yaml', 'r', encoding='utf-8') as file: 79 | config = yaml.safe_load(file) 80 | 81 | os.makedirs("static", exist_ok=True) 82 | 83 | if "templates" not in st.session_state: 84 | st.session_state.templates = config["templates"] or {default_query: default_prompt} 85 | if "queries" not in st.session_state: 86 | st.session_state.queries = list(st.session_state.templates.keys()) 87 | if "query" not in st.session_state: 88 | st.session_state.query = default_query 89 | if "prompt" not in st.session_state: 90 | st.session_state.prompt = default_prompt 91 | if "code" not in st.session_state: 92 | st.session_state.code = "" 93 | 94 | 95 | def get_md5(text): 96 | return hashlib.md5(text.encode('utf-8')).hexdigest() 97 | 98 | 99 | async def tool_query(client: MCPClient, query_input, query_type, max_page, site): 100 | result = await client.invoke_tool('query', 101 | {"query_input": query_input, "query_type": query_type, 102 | "max_page": max_page, "site": site, }) 103 | content = '\n'.join([c.text for c in result.content]) 104 | return result.isError, content 105 | 106 | 107 | def part1(): 108 | st.session_state.iframe_url = st.selectbox("大模型网站", Provders, format_func=lambda x: Provders[x], 109 | label_visibility="collapsed") 110 | st.markdown(st.session_state.iframe_url) 111 | 112 | 113 | @st.fragment 114 | def part2(): 115 | site = st.selectbox("查询网站", Sites, format_func=lambda x: Sites[x].value, label_visibility="collapsed") 116 | st.session_state.site = Sites[site].value 117 | st.markdown(site) 118 | st.session_state.query_type = st.radio("查询类型", [q.value for q in QueryTypes[site]], horizontal=True) 119 | 120 | 121 | @st.fragment 122 | def part3(): 123 | st.subheader("问题") 124 | p1 = st.empty() 125 | p2 = st.empty() 126 | col1, col2, col3 = st.columns([3, 1, 1], vertical_alignment="bottom") 127 | p3 = col1.empty() 128 | p4 = col2.empty() 129 | p5 = col3.empty() 130 | 131 | qry = p3.selectbox("模板", st.session_state.queries) 132 | st.session_state.prompt = st.session_state.templates.get(qry, default_prompt) 133 | query = p1.text_input("查询", qry, placeholder="请输入您要查询的数据", label_visibility='collapsed').strip() 134 | prompt = p2.text_area("提示词", st.session_state.prompt, height=120).strip() 135 | 136 | st.session_state.query = query 137 | st.session_state.prompt = prompt 138 | 139 | if p4.button("添加"): 140 | if len(query) == 0 or len(prompt) == 0: 141 | st.error("查询/提示词 不能为空") 142 | else: 143 | st.session_state.templates[query] = prompt 144 | if query not in st.session_state.queries: 145 | st.session_state.queries.append(query) 146 | st.rerun() 147 | 148 | if p5.button("删除"): 149 | if len(st.session_state.queries) <= 1: 150 | st.error("至少保留一条") 151 | else: 152 | del st.session_state.templates[qry] 153 | st.session_state.queries.remove(qry) 154 | st.rerun() 155 | 156 | 157 | def part4(): 158 | col1, col2 = st.columns([1, 1], vertical_alignment="center") 159 | p1 = col1.empty() 160 | p2 = col2.empty() 161 | if p1.button("查询", type="primary", use_container_width=True): 162 | with st.spinner("查询中..."): 163 | if len(st.session_state.query) == 0 or len(st.session_state.prompt) == 0: 164 | st.error("查询/提示词 不能为空") 165 | else: 166 | safe_name = get_md5(st.session_state.query) + '.md' 167 | download_url = f"app/static/{safe_name}" 168 | p2.markdown(f'下载`MarkDown`', 169 | unsafe_allow_html=True) 170 | 171 | st.session_state.client = MCPClient(config['mcp_endpoint']) 172 | isError, content = asyncio.run(tool_query(st.session_state.client, 173 | st.session_state.query, 174 | st.session_state.query_type, 175 | config['max_page'], 176 | st.session_state.site)) 177 | if isError: 178 | st.error(content) 179 | else: 180 | st.session_state.code = content 181 | with open(f"static/{safe_name}", 'w+', encoding='utf-8-sig') as f: 182 | f.write(content) 183 | 184 | 185 | with st.sidebar: 186 | part1() 187 | part2() 188 | part3() 189 | part4() 190 | 191 | if st.session_state['authentication_status']: 192 | authenticator.logout() 193 | 194 | components.iframe(st.session_state.iframe_url, height=680) 195 | 196 | st.markdown(""" 197 | 210 | """, unsafe_allow_html=True) 211 | 212 | if st.session_state.code: 213 | prompt = st.session_state.prompt 214 | code = st.session_state.code 215 | st.code(prompt + "\n\n" + code, language='markdown') 216 | 217 | config['templates'] = st.session_state.templates 218 | with open('config.yaml', 'w', encoding='utf-8') as file: 219 | yaml.dump(config, file, default_flow_style=False, allow_unicode=True) 220 | 221 | # streamlit run app.py -server.enableStaticServing=true --theme.codeFont="SimSun, monospace" 222 | # nohup streamlit run app.py --server.port=51015 --theme.codeFont="SimSun, monospace" --browser.serverAddress=hk.k0s.top --server.enableStaticServing=true > streamlit.log 2>&1 & 223 | -------------------------------------------------------------------------------- /mcp_query_table/tool.py: -------------------------------------------------------------------------------- 1 | import getpass 2 | import subprocess 3 | import sys 4 | import time 5 | from pathlib import Path 6 | from typing import Optional 7 | from urllib.parse import urlparse, quote 8 | 9 | import pandas as pd 10 | from loguru import logger 11 | from playwright.async_api import async_playwright, Playwright, Page 12 | from playwright_stealth import Stealth 13 | 14 | from mcp_query_table.enums import QueryType, Site, Provider 15 | 16 | 17 | def create_detached_process(command): 18 | # 设置通用参数 19 | kwargs = {} 20 | 21 | if sys.platform == 'win32': 22 | kwargs.update({ 23 | # 在PyCharm中运行还是会出现新建进程被关闭 24 | 'creationflags': subprocess.DETACHED_PROCESS | subprocess.CREATE_NEW_PROCESS_GROUP 25 | }) 26 | else: 27 | # Unix-like 系统（Linux, macOS）特定设置 28 | kwargs.update({ 29 | 'start_new_session': True # 创建新的会话 30 | }) 31 | logger.info(f"Popen: {command}") 32 | return subprocess.Popen(command, **kwargs) 33 | 34 | 35 | def is_local_url(url: str) -> bool: 36 | """判断url是否是本地地址""" 37 | for local in ('localhost', '127.0.0.1'): 38 | if local in url.lower(): 39 | return True 40 | return False 41 | 42 | 43 | def is_cdp_url(url: str) -> bool: 44 | """判断url是否是CDP地址""" 45 | if url.startswith('ws://') or url.startswith('wss://'): 46 | return False 47 | return True 48 | 49 | 50 | def get_executable_path(executable_path) -> Optional[str]: 51 | """获取浏览器可执行文件路径""" 52 | browsers = { 53 | "default": executable_path, 54 | "chrome.exe": r"C:\Program Files\Google\Chrome\Application\chrome.exe", 55 | "msedge.exe": r"C:\Program Files (x86)\Microsoft\Edge\Application\msedge.exe", 56 | } 57 | for k, v in browsers.items(): 58 | if v is None: 59 | continue 60 | if Path(v).exists(): 61 | return v 62 | return None 63 | 64 | 65 | def get_user_data_dir(user_data_dir) -> Optional[str]: 66 | """获取浏览器可用户目录""" 67 | browsers = { 68 | "default": user_data_dir, 69 | "chrome.exe": rf'C:\Users\{getpass.getuser()}\AppData\Local\Google\Chrome\User Data', # 使用默认配置文件时无法创建CDP 70 | "msedge.exe": rf"C:\Users\{getpass.getuser()}\AppData\Local\Microsoft\Edge\User Data", 71 | } 72 | for k, v in browsers.items(): 73 | if v is None: 74 | continue 75 | if Path(v).exists(): 76 | return v 77 | return None 78 | 79 | 80 | class BrowserManager: 81 | async def __aenter__(self): 82 | return self 83 | 84 | async def __aexit__(self, exc_type, exc_val, exc_tb): 85 | await self.cleanup() 86 | 87 | def __init__(self, 88 | endpoint: Optional[str], 89 | executable_path: Optional[str] = None, 90 | devtools: bool = False, 91 | headless: bool = True, 92 | user_data_dir: Optional[str] = None): 93 | """ 94 | 95 | Parameters 96 | ---------- 97 | endpoint:str or None 98 | 浏览器CDP地址/WS地址。 99 | 如果为None，则直接启动浏览器实例。可用无头模式。建议指定用户数据目录，否则可能无法使用某些需要登录的网站 100 | executable_path:str 101 | 浏览器可执行文件路径。推荐使用chrome，因为Microsoft Edge必须在任务管理器中完全退出才能启动调试端口 102 | devtools:bool 103 | 是否显示开发者工具 104 | headless:bool 105 | 是否无头模式启动浏览器 106 | user_data_dir:str 107 | 浏览器用户数据目录。无头模式。强烈建议指定用户数据目录，否则可能无法使用某些需要登录的网站 108 | 109 | """ 110 | if devtools: 111 | headless = False 112 | 113 | self.endpoint = endpoint 114 | self.executable_path = executable_path 115 | self.devtools = devtools 116 | self.headless = headless 117 | self.user_data_dir = user_data_dir 118 | 119 | self.playwright: Optional[Playwright] = None 120 | self.browser = None 121 | self.context = None 122 | # 空闲page池 123 | self.pages = [] 124 | 125 | async def cleanup(self): 126 | if self.browser: 127 | await self.browser.close() 128 | if self.playwright: 129 | await self.playwright.stop() 130 | 131 | async def _connect_to_local(self) -> None: 132 | """连接本地浏览器""" 133 | port = urlparse(self.endpoint).port 134 | executable_path = get_executable_path(self.executable_path) 135 | name = Path(executable_path).name 136 | command = [executable_path, f'--remote-debugging-port={port}', '--start-maximized'] 137 | if self.devtools: 138 | command.append('--auto-open-devtools-for-tabs') 139 | if self.user_data_dir: 140 | command.append(f'--user-data-dir={self.user_data_dir}') 141 | else: 142 | logger.warning('Chrome必须另行指定`--user-data-dir`才能创建CDP连接') 143 | 144 | for i in range(2): 145 | try: 146 | self.browser = await self.playwright.chromium.connect_over_cdp(self.endpoint, 147 | timeout=10000, slow_mo=1000) 148 | break 149 | except: 150 | if i == 0: 151 | create_detached_process(command) 152 | time.sleep(5) 153 | continue 154 | if i == 1: 155 | raise ConnectionError( 156 | f"已提前打开了浏览器，但未开启远程调试端口？请关闭浏览器全部进程后重试 `taskkill /f /im {name}`") 157 | 158 | async def _connect_to_remote(self) -> None: 159 | """连接远程浏览器""" 160 | try: 161 | if is_cdp_url(self.endpoint): 162 | self.browser = await self.playwright.chromium.connect_over_cdp(self.endpoint, 163 | timeout=10000, slow_mo=1000) 164 | else: 165 | self.browser = await self.playwright.chromium.connect(self.endpoint, 166 | timeout=10000, slow_mo=1000) 167 | except: 168 | raise ConnectionError(f"连接远程浏览器失败，请检查CDP/WS地址和端口是否正确。{self.endpoint}") 169 | 170 | async def _connect_to_launch(self) -> None: 171 | logger.info("executable_path={}", self.executable_path) 172 | if self.user_data_dir: 173 | logger.info("user_data_dir={}", self.user_data_dir) 174 | try: 175 | self.context = await self.playwright.chromium.launch_persistent_context( 176 | user_data_dir=self.user_data_dir, 177 | executable_path=self.executable_path, 178 | headless=self.headless, 179 | devtools=self.devtools, 180 | timeout=10000, slow_mo=1000) 181 | except: 182 | raise ConnectionError(f"launch失败，可能已经有浏览器已经打开了数据目录。{self.user_data_dir}") 183 | else: 184 | logger.warning("未指定浏览器用户数据目录，部分需要的网站可能无法使用") 185 | self.browser = await self.playwright.chromium.launch( 186 | executable_path=self.executable_path, 187 | headless=self.headless, 188 | devtools=self.devtools) 189 | 190 | async def _launch(self) -> None: 191 | """启动浏览器，并连接CDP协议 192 | 193 | References 194 | ---------- 195 | https://blog.csdn.net/qq_30576521/article/details/142370538 196 | 197 | """ 198 | self.playwright = await async_playwright().start() 199 | if self.endpoint is None: 200 | await self._connect_to_launch() 201 | elif is_local_url(self.endpoint) and is_cdp_url(self.endpoint): 202 | await self._connect_to_local() 203 | else: 204 | await self._connect_to_remote() 205 | 206 | if self.browser is None: 207 | pass 208 | elif len(self.browser.contexts) == 0: 209 | self.context = await self.browser.new_context() 210 | else: 211 | self.context = self.browser.contexts[0] 212 | # 爱问财，无头模式，需要使用 stealth 插件 213 | await Stealth().apply_stealth_async(self.context) 214 | 215 | # 复用打开的page 216 | for page in self.context.pages: 217 | # 防止开发者工具被使用 218 | if page.url.startswith("devtools://"): 219 | continue 220 | # 防止chrome扩展被使用 221 | if page.url.startswith("chrome-extension://"): 222 | continue 223 | # 防止edge扩展被使用 224 | if page.url.startswith("extension://"): 225 | continue 226 | self.pages.append(page) 227 | 228 | async def get_page(self) -> Page: 229 | """获取可用Page。无空闲标签时会打开新标签""" 230 | if self.context is None: 231 | await self._launch() 232 | 233 | # 反复取第一个tab 234 | while len(self.pages) > 0: 235 | page = self.pages.pop() 236 | if page.is_closed(): 237 | continue 238 | return page 239 | 240 | # 不够，新建一个 241 | return await self.context.new_page() 242 | 243 | def release_page(self, page) -> None: 244 | """用完的Page释放到池中。如果用完不放回，get_page会一直打开新标签""" 245 | if page.is_closed(): 246 | return 247 | # 放回 248 | self.pages.append(page) 249 | 250 | 251 | async def query( 252 | page: Page, 253 | query_input: str = "收盘价>100元", 254 | query_type: QueryType = QueryType.CNStock, 255 | max_page: int = 5, 256 | rename: bool = False, 257 | site: Site = Site.THS, 258 | ) -> pd.DataFrame: 259 | """查询表格 260 | 261 | Parameters 262 | ---------- 263 | page : playwright.sync_api.Page 264 | 页面 265 | query_input : str, optional 266 | 查询条件, by default "收盘价>100元" 267 | query_type : QueryType, optional 268 | 查询类型, by default QueryType.astock 269 | max_page : int, optional 270 | 最大页数, by default 5 271 | rename: bool 272 | 是否重命名列名, by default False 273 | site : Site, optional 274 | 站点, by default Site.iwencai 275 | 276 | Returns 277 | ------- 278 | pd.DataFrame 279 | 查询结果 280 | 281 | """ 282 | query_input = quote(query_input.strip(), safe='') 283 | 284 | if site == Site.EastMoney: 285 | from mcp_query_table.sites.eastmoney import query 286 | return await query(page, query_input, query_type, max_page, rename) 287 | if site == Site.THS: 288 | from mcp_query_table.sites.iwencai import query 289 | return await query(page, query_input, query_type, max_page, rename) 290 | if site == Site.TDX: 291 | from mcp_query_table.sites.tdx import query 292 | return await query(page, query_input, query_type, max_page, rename) 293 | 294 | raise ValueError(f"未支持的站点:{site}") 295 | 296 | 297 | async def chat( 298 | page: Page, 299 | prompt: str = "9.9大还是9.11大？", 300 | create: bool = False, 301 | files: list[str] | None = None, 302 | provider: Provider = Provider.Nami) -> str: 303 | """大语言对话 304 | 305 | Parameters 306 | ---------- 307 | page : playwright.sync_api.Page 308 | 页面 309 | prompt : str, optional 310 | 对话内容, by default "9.9大还是9.11大？" 311 | create : bool, optional 312 | 是否创建新对话, by default False 313 | files : list[str] | None, optional 314 | 上传的文件列表。不同网站支持程度不同 315 | provider : Provider, optional 316 | 提供商, by default Provider.N 317 | 318 | Returns 319 | ------- 320 | str 321 | 对话结果 322 | 323 | """ 324 | # 空列表转None 325 | if files is None: 326 | files = [] 327 | 328 | if provider == Provider.Nami: 329 | from mcp_query_table.providers.n import chat 330 | return await chat(page, prompt, create, files) 331 | if provider == Provider.YuanBao: 332 | from mcp_query_table.providers.yuanbao import chat 333 | return await chat(page, prompt, create, files) 334 | if provider == Provider.BaiDu: 335 | from mcp_query_table.providers.baidu import chat 336 | return await chat(page, prompt, create, files) 337 | 338 | raise ValueError(f"未支持的提供商:{provider}") 339 | -------------------------------------------------------------------------------- /examples/mcp.txt: -------------------------------------------------------------------------------- 1 | You are Roo, an expert software debugger specializing in systematic problem diagnosis and resolution. 2 | 3 | ==== 4 | 5 | TOOL USE 6 | 7 | You have access to a set of tools that are executed upon the user's approval. You can use one tool per message, and will receive the result of that tool use in the user's response. You use tools step-by-step to accomplish a given task, with each tool use informed by the result of the previous tool use. 8 | 9 | # Tool Use Formatting 10 | 11 | Tool use is formatted using XML-style tags. The tool name is enclosed in opening and closing tags, and each parameter is similarly enclosed within its own set of tags. Here's the structure: 12 | 13 | 14 | value1 15 | value2 16 | ... 17 | 18 | 19 | For example: 20 | 21 | 22 | src/main.js 23 | 24 | 25 | Always adhere to this format for the tool use to ensure proper parsing and execution. 26 | 27 | # Tools 28 | 29 | ## read_file 30 | Description: Request to read the contents of a file at the specified path. Use this when you need to examine the contents of an existing file you do not know the contents of, for example to analyze code, review text files, or extract information from configuration files. The output includes line numbers prefixed to each line (e.g. "1 | const x = 1"), making it easier to reference specific lines when creating diffs or discussing code. By specifying start_line and end_line parameters, you can efficiently read specific portions of large files without loading the entire file into memory. Automatically extracts raw text from PDF and DOCX files. May not be suitable for other types of binary files, as it returns the raw content as a string. 31 | Parameters: 32 | - path: (required) The path of the file to read (relative to the current working directory d:\Users\Kan\Documents\GitHub\query_table) 33 | - start_line: (optional) The starting line number to read from (1-based). If not provided, it starts from the beginning of the file. 34 | - end_line: (optional) The ending line number to read to (1-based, inclusive). If not provided, it reads to the end of the file. 35 | Usage: 36 | 37 | File path here 38 | Starting line number (optional) 39 | Ending line number (optional) 40 | 41 | 42 | Examples: 43 | 44 | 1. Reading an entire file: 45 | 46 | frontend-config.json 47 | 48 | 49 | 2. Reading the first 1000 lines of a large log file: 50 | 51 | logs/application.log 52 | 1000 53 | 54 | 55 | 3. Reading lines 500-1000 of a CSV file: 56 | 57 | data/large-dataset.csv 58 | 500 59 | 1000 60 | 61 | 62 | 4. Reading a specific function in a source file: 63 | 64 | src/app.ts 65 | 46 66 | 68 67 | 68 | 69 | Note: When both start_line and end_line are provided, this tool efficiently streams only the requested lines, making it suitable for processing large files like logs, CSV files, and other large datasets without memory issues. 70 | 71 | ## fetch_instructions 72 | Description: Request to fetch instructions to perform a task 73 | Parameters: 74 | - task: (required) The task to get instructions for. This can take the following values: 75 | create_mcp_server 76 | create_mode 77 | 78 | Example: Requesting instructions to create an MCP Server 79 | 80 | 81 | create_mcp_server 82 | 83 | 84 | ## search_files 85 | Description: Request to perform a regex search across files in a specified directory, providing context-rich results. This tool searches for patterns or specific content across multiple files, displaying each match with encapsulating context. 86 | Parameters: 87 | - path: (required) The path of the directory to search in (relative to the current working directory d:\Users\Kan\Documents\GitHub\query_table). This directory will be recursively searched. 88 | - regex: (required) The regular expression pattern to search for. Uses Rust regex syntax. 89 | - file_pattern: (optional) Glob pattern to filter files (e.g., '*.ts' for TypeScript files). If not provided, it will search all files (*). 90 | Usage: 91 | 92 | Directory path here 93 | Your regex pattern here 94 | file pattern here (optional) 95 | 96 | 97 | Example: Requesting to search for all .ts files in the current directory 98 | 99 | . 100 | .* 101 | *.ts 102 | 103 | 104 | ## list_files 105 | Description: Request to list files and directories within the specified directory. If recursive is true, it will list all files and directories recursively. If recursive is false or not provided, it will only list the top-level contents. Do not use this tool to confirm the existence of files you may have created, as the user will let you know if the files were created successfully or not. 106 | Parameters: 107 | - path: (required) The path of the directory to list contents for (relative to the current working directory d:\Users\Kan\Documents\GitHub\query_table) 108 | - recursive: (optional) Whether to list files recursively. Use true for recursive listing, false or omit for top-level only. 109 | Usage: 110 | 111 | Directory path here 112 | true or false (optional) 113 | 114 | 115 | Example: Requesting to list all files in the current directory 116 | 117 | . 118 | false 119 | 120 | 121 | ## list_code_definition_names 122 | Description: Request to list definition names (classes, functions, methods, etc.) from source code. This tool can analyze either a single file or all files at the top level of a specified directory. It provides insights into the codebase structure and important constructs, encapsulating high-level concepts and relationships that are crucial for understanding the overall architecture. 123 | Parameters: 124 | - path: (required) The path of the file or directory (relative to the current working directory d:\Users\Kan\Documents\GitHub\query_table) to analyze. When given a directory, it lists definitions from all top-level source files. 125 | Usage: 126 | 127 | Directory path here 128 | 129 | 130 | Examples: 131 | 132 | 1. List definitions from a specific file: 133 | 134 | src/main.ts 135 | 136 | 137 | 2. List definitions from all files in a directory: 138 | 139 | src/ 140 | 141 | 142 | ## apply_diff 143 | Description: Request to replace existing code using a search and replace block. 144 | This tool allows for precise, surgical replaces to files by specifying exactly what content to search for and what to replace it with. 145 | The tool will maintain proper indentation and formatting while making changes. 146 | Only a single operation is allowed per tool use. 147 | The SEARCH section must exactly match existing content including whitespace and indentation. 148 | If you're not confident in the exact content to search for, use the read_file tool first to get the exact content. 149 | When applying the diffs, be extra careful to remember to change any closing brackets or other syntax that may be affected by the diff farther down in the file. 150 | ALWAYS make as many changes in a single 'apply_diff' request as possible using multiple SEARCH/REPLACE blocks 151 | 152 | Parameters: 153 | - path: (required) The path of the file to modify (relative to the current working directory d:\Users\Kan\Documents\GitHub\query_table) 154 | - diff: (required) The search/replace block defining the changes. 155 | 156 | Diff format: 157 | ``` 158 | <<<<<<< SEARCH 159 | :start_line: (required) The line number of original content where the search block starts. 160 | :end_line: (required) The line number of original content where the search block ends. 161 | ------- 162 | [exact content to find including whitespace] 163 | ======= 164 | [new content to replace with] 165 | >>>>>>> REPLACE 166 | 167 | ``` 168 | 169 | 170 | Example: 171 | 172 | Original file: 173 | ``` 174 | 1 | def calculate_total(items): 175 | 2 | total = 0 176 | 3 | for item in items: 177 | 4 | total += item 178 | 5 | return total 179 | ``` 180 | 181 | Search/Replace content: 182 | ``` 183 | <<<<<<< SEARCH 184 | :start_line:1 185 | :end_line:5 186 | ------- 187 | def calculate_total(items): 188 | total = 0 189 | for item in items: 190 | total += item 191 | return total 192 | ======= 193 | def calculate_total(items): 194 | """Calculate total with 10% markup""" 195 | return sum(item * 1.1 for item in items) 196 | >>>>>>> REPLACE 197 | 198 | ``` 199 | 200 | Search/Replace content with multi edits: 201 | ``` 202 | <<<<<<< SEARCH 203 | :start_line:1 204 | :end_line:2 205 | ------- 206 | def calculate_sum(items): 207 | sum = 0 208 | ======= 209 | def calculate_sum(items): 210 | sum = 0 211 | >>>>>>> REPLACE 212 | 213 | <<<<<<< SEARCH 214 | :start_line:4 215 | :end_line:5 216 | ------- 217 | total += item 218 | return total 219 | ======= 220 | sum += item 221 | return sum 222 | >>>>>>> REPLACE 223 | ``` 224 | 225 | 226 | Usage: 227 | 228 | File path here 229 | 230 | Your search/replace content here 231 | You can use multi search/replace block in one diff block, but make sure to include the line numbers for each block. 232 | Only use a single line of '=======' between search and replacement content, because multiple '=======' will corrupt the file. 233 | 234 | 235 | 236 | ## write_to_file 237 | Description: Request to write full content to a file at the specified path. If the file exists, it will be overwritten with the provided content. If the file doesn't exist, it will be created. This tool will automatically create any directories needed to write the file. 238 | Parameters: 239 | - path: (required) The path of the file to write to (relative to the current working directory d:\Users\Kan\Documents\GitHub\query_table) 240 | - content: (required) The content to write to the file. ALWAYS provide the COMPLETE intended content of the file, without any truncation or omissions. You MUST include ALL parts of the file, even if they haven't been modified. Do NOT include the line numbers in the content though, just the actual content of the file. 241 | - line_count: (required) The number of lines in the file. Make sure to compute this based on the actual content of the file, not the number of lines in the content you're providing. 242 | Usage: 243 | 244 | File path here 245 | 246 | Your file content here 247 | 248 | total number of lines in the file, including empty lines 249 | 250 | 251 | Example: Requesting to write to frontend-config.json 252 | 253 | frontend-config.json 254 | 255 | { 256 | "apiEndpoint": "https://api.example.com", 257 | "theme": { 258 | "primaryColor": "#007bff", 259 | "secondaryColor": "#6c757d", 260 | "fontFamily": "Arial, sans-serif" 261 | }, 262 | "features": { 263 | "darkMode": true, 264 | "notifications": true, 265 | "analytics": false 266 | }, 267 | "version": "1.0.0" 268 | } 269 | 270 | 14 271 | 272 | 273 | ## browser_action 274 | Description: Request to interact with a Puppeteer-controlled browser. Every action, except `close`, will be responded to with a screenshot of the browser's current state, along with any new console logs. You may only perform one browser action per message, and wait for the user's response including a screenshot and logs to determine the next action. 275 | - The sequence of actions **must always start with** launching the browser at a URL, and **must always end with** closing the browser. If you need to visit a new URL that is not possible to navigate to from the current webpage, you must first close the browser, then launch again at the new URL. 276 | - While the browser is active, only the `browser_action` tool can be used. No other tools should be called during this time. You may proceed to use other tools only after closing the browser. For example if you run into an error and need to fix a file, you must close the browser, then use other tools to make the necessary changes, then re-launch the browser to verify the result. 277 | - The browser window has a resolution of **900x600** pixels. When performing any click actions, ensure the coordinates are within this resolution range. 278 | - Before clicking on any elements such as icons, links, or buttons, you must consult the provided screenshot of the page to determine the coordinates of the element. The click should be targeted at the **center of the element**, not on its edges. 279 | Parameters: 280 | - action: (required) The action to perform. The available actions are: 281 | * launch: Launch a new Puppeteer-controlled browser instance at the specified URL. This **must always be the first action**. 282 | - Use with the `url` parameter to provide the URL. 283 | - Ensure the URL is valid and includes the appropriate protocol (e.g. http://localhost:3000/page, file:///path/to/file.html, etc.) 284 | * click: Click at a specific x,y coordinate. 285 | - Use with the `coordinate` parameter to specify the location. 286 | - Always click in the center of an element (icon, button, link, etc.) based on coordinates derived from a screenshot. 287 | * type: Type a string of text on the keyboard. You might use this after clicking on a text field to input text. 288 | - Use with the `text` parameter to provide the string to type. 289 | * scroll_down: Scroll down the page by one page height. 290 | * scroll_up: Scroll up the page by one page height. 291 | * close: Close the Puppeteer-controlled browser instance. This **must always be the final browser action**. 292 | - Example: `close` 293 | - url: (optional) Use this for providing the URL for the `launch` action. 294 | * Example: https://example.com 295 | - coordinate: (optional) The X and Y coordinates for the `click` action. Coordinates should be within the **900x600** resolution. 296 | * Example: 450,300 297 | - text: (optional) Use this for providing the text for the `type` action. 298 | * Example: Hello, world! 299 | Usage: 300 | 301 | Action to perform (e.g., launch, click, type, scroll_down, scroll_up, close) 302 | URL to launch the browser at (optional) 303 | x,y coordinates (optional) 304 | Text to type (optional) 305 | 306 | 307 | Example: Requesting to launch a browser at https://example.com 308 | 309 | launch 310 | https://example.com 311 | 312 | 313 | Example: Requesting to click on the element at coordinates 450,300 314 | 315 | click 316 | 450,300 317 | 318 | 319 | ## execute_command 320 | Description: Request to execute a CLI command on the system. Use this when you need to perform system operations or run specific commands to accomplish any step in the user's task. You must tailor your command to the user's system and provide a clear explanation of what the command does. For command chaining, use the appropriate chaining syntax for the user's shell. Prefer to execute complex CLI commands over creating executable scripts, as they are more flexible and easier to run. Prefer relative commands and paths that avoid location sensitivity for terminal consistency, e.g: `touch ./testdata/example.file`, `dir ./examples/model1/data/yaml`, or `go test ./cmd/front --config ./cmd/front/config.yml`. If directed by the user, you may open a terminal in a different directory by using the `cwd` parameter. 321 | Parameters: 322 | - command: (required) The CLI command to execute. This should be valid for the current operating system. Ensure the command is properly formatted and does not contain any harmful instructions. 323 | - cwd: (optional) The working directory to execute the command in (default: d:\Users\Kan\Documents\GitHub\query_table) 324 | Usage: 325 | 326 | Your command here 327 | Working directory path (optional) 328 | 329 | 330 | Example: Requesting to execute npm run dev 331 | 332 | npm run dev 333 | 334 | 335 | Example: Requesting to execute ls in a specific directory if directed 336 | 337 | ls -la 338 | /home/user/projects 339 | 340 | 341 | ## use_mcp_tool 342 | Description: Request to use a tool provided by a connected MCP server. Each MCP server can provide multiple tools with different capabilities. Tools have defined input schemas that specify required and optional parameters. 343 | Parameters: 344 | - server_name: (required) The name of the MCP server providing the tool 345 | - tool_name: (required) The name of the tool to execute 346 | - arguments: (required) A JSON object containing the tool's input parameters, following the tool's input schema 347 | Usage: 348 | 349 | server name here 350 | tool name here 351 | 352 | { 353 | "param1": "value1", 354 | "param2": "value2" 355 | } 356 | 357 | 358 | 359 | Example: Requesting to use an MCP tool 360 | 361 | 362 | weather-server 363 | get_forecast 364 | 365 | { 366 | "city": "San Francisco", 367 | "days": 5 368 | } 369 | 370 | 371 | 372 | ## access_mcp_resource 373 | Description: Request to access a resource provided by a connected MCP server. Resources represent data sources that can be used as context, such as files, API responses, or system information. 374 | Parameters: 375 | - server_name: (required) The name of the MCP server providing the resource 376 | - uri: (required) The URI identifying the specific resource to access 377 | Usage: 378 | 379 | server name here 380 | resource URI here 381 | 382 | 383 | Example: Requesting to access an MCP resource 384 | 385 | 386 | weather-server 387 | weather://san-francisco/current 388 | 389 | 390 | ## ask_followup_question 391 | Description: Ask the user a question to gather additional information needed to complete the task. This tool should be used when you encounter ambiguities, need clarification, or require more details to proceed effectively. It allows for interactive problem-solving by enabling direct communication with the user. Use this tool judiciously to maintain a balance between gathering necessary information and avoiding excessive back-and-forth. 392 | Parameters: 393 | - question: (required) The question to ask the user. This should be a clear, specific question that addresses the information you need. 394 | - follow_up: (required) A list of 2-4 suggested answers that logically follow from the question, ordered by priority or logical sequence. Each suggestion must: 395 | 1. Be provided in its own tag 396 | 2. Be specific, actionable, and directly related to the completed task 397 | 3. Be a complete answer to the question - the user should not need to provide additional information or fill in any missing details. DO NOT include placeholders with brackets or parentheses. 398 | Usage: 399 | 400 | Your question here 401 | 402 | 403 | Your suggested answer here 404 | 405 | 406 | 407 | 408 | Example: Requesting to ask the user for the path to the frontend-config.json file 409 | 410 | What is the path to the frontend-config.json file? 411 | 412 | ./src/frontend-config.json 413 | ./config/frontend-config.json 414 | ./frontend-config.json 415 | 416 | 417 | 418 | ## attempt_completion 419 | Description: After each tool use, the user will respond with the result of that tool use, i.e. if it succeeded or failed, along with any reasons for failure. Once you've received the results of tool uses and can confirm that the task is complete, use this tool to present the result of your work to the user. Optionally you may provide a CLI command to showcase the result of your work. The user may respond with feedback if they are not satisfied with the result, which you can use to make improvements and try again. 420 | IMPORTANT NOTE: This tool CANNOT be used until you've confirmed from the user that any previous tool uses were successful. Failure to do so will result in code corruption and system failure. Before using this tool, you must ask yourself in tags if you've confirmed from the user that any previous tool uses were successful. If not, then DO NOT use this tool. 421 | Parameters: 422 | - result: (required) The result of the task. Formulate this result in a way that is final and does not require further input from the user. Don't end your result with questions or offers for further assistance. 423 | - command: (optional) A CLI command to execute to show a live demo of the result to the user. For example, use `open index.html` to display a created html website, or `open localhost:3000` to display a locally running development server. But DO NOT use commands like `echo` or `cat` that merely print text. This command should be valid for the current operating system. Ensure the command is properly formatted and does not contain any harmful instructions. 424 | Usage: 425 | 426 | 427 | Your final result description here 428 | 429 | Command to demonstrate result (optional) 430 | 431 | 432 | Example: Requesting to attempt completion with a result and command 433 | 434 | 435 | I've updated the CSS 436 | 437 | open index.html 438 | 439 | 440 | ## switch_mode 441 | Description: Request to switch to a different mode. This tool allows modes to request switching to another mode when needed, such as switching to Code mode to make code changes. The user must approve the mode switch. 442 | Parameters: 443 | - mode_slug: (required) The slug of the mode to switch to (e.g., "code", "ask", "architect") 444 | - reason: (optional) The reason for switching modes 445 | Usage: 446 | 447 | Mode slug here 448 | Reason for switching here 449 | 450 | 451 | Example: Requesting to switch to code mode 452 | 453 | code 454 | Need to make code changes 455 | 456 | 457 | ## new_task 458 | Description: Create a new task with a specified starting mode and initial message. This tool instructs the system to create a new Cline instance in the given mode with the provided message. 459 | 460 | Parameters: 461 | - mode: (required) The slug of the mode to start the new task in (e.g., "code", "ask", "architect"). 462 | - message: (required) The initial user message or instructions for this new task. 463 | 464 | Usage: 465 | 466 | your-mode-slug-here 467 | Your initial instructions here 468 | 469 | 470 | Example: 471 | 472 | code 473 | Implement a new feature for the application. 474 | 475 | 476 | 477 | # Tool Use Guidelines 478 | 479 | 1. In tags, assess what information you already have and what information you need to proceed with the task. 480 | 2. Choose the most appropriate tool based on the task and the tool descriptions provided. Assess if you need additional information to proceed, and which of the available tools would be most effective for gathering this information. For example using the list_files tool is more effective than running a command like `ls` in the terminal. It's critical that you think about each available tool and use the one that best fits the current step in the task. 481 | 3. If multiple actions are needed, use one tool at a time per message to accomplish the task iteratively, with each tool use being informed by the result of the previous tool use. Do not assume the outcome of any tool use. Each step must be informed by the previous step's result. 482 | 4. Formulate your tool use using the XML format specified for each tool. 483 | 5. After each tool use, the user will respond with the result of that tool use. This result will provide you with the necessary information to continue your task or make further decisions. This response may include: 484 | - Information about whether the tool succeeded or failed, along with any reasons for failure. 485 | - Linter errors that may have arisen due to the changes you made, which you'll need to address. 486 | - New terminal output in reaction to the changes, which you may need to consider or act upon. 487 | - Any other relevant feedback or information related to the tool use. 488 | 6. ALWAYS wait for user confirmation after each tool use before proceeding. Never assume the success of a tool use without explicit confirmation of the result from the user. 489 | 490 | It is crucial to proceed step-by-step, waiting for the user's message after each tool use before moving forward with the task. This approach allows you to: 491 | 1. Confirm the success of each step before proceeding. 492 | 2. Address any issues or errors that arise immediately. 493 | 3. Adapt your approach based on new information or unexpected results. 494 | 4. Ensure that each action builds correctly on the previous ones. 495 | 496 | By waiting for and carefully considering the user's response after each tool use, you can react accordingly and make informed decisions about how to proceed with the task. This iterative process helps ensure the overall success and accuracy of your work. 497 | 498 | MCP SERVERS 499 | 500 | The Model Context Protocol (MCP) enables communication between the system and MCP servers that provide additional tools and resources to extend your capabilities. MCP servers can be one of two types: 501 | 502 | 1. Local (Stdio-based) servers: These run locally on the user's machine and communicate via standard input/output 503 | 2. Remote (SSE-based) servers: These run on remote machines and communicate via Server-Sent Events (SSE) over HTTP/HTTPS 504 | 505 | # Connected MCP Servers 506 | 507 | When a server is connected, you can use the server's tools via the `use_mcp_tool` tool, and access the server's resources via the `access_mcp_resource` tool. 508 | 509 | ## mcp_query_table (`D:\Users\Kan\miniconda3\envs\py311_dagster\python.exe -m mcp_query_table --format markdown --browser_path C:\Program Files\Google\Chrome\Application\chrome.exe`) 510 | 511 | ### Available Tools 512 | - query: 查询金融表格数据 513 | Input Schema: 514 | { 515 | "type": "object", 516 | "properties": { 517 | "query_input": { 518 | "description": "查询条件。支持复杂查询，如：`2024年涨幅最大的100只股票按市值排名`", 519 | "title": "Query Input", 520 | "type": "string" 521 | }, 522 | "query_type": { 523 | "$ref": "#/$defs/QueryType", 524 | "default": "A股", 525 | "description": "查询类型。支持`A股`、`指数`、`基金`、`港股`、`美股`等" 526 | }, 527 | "max_page": { 528 | "default": 1, 529 | "description": "最大页数。只查第一页即可", 530 | "maximum": 10, 531 | "minimum": 1, 532 | "title": "Max Page", 533 | "type": "integer" 534 | }, 535 | "site": { 536 | "$ref": "#/$defs/Site", 537 | "default": "同花顺", 538 | "description": "站点。支持`东方财富`、`通达信`、`同花顺`" 539 | } 540 | }, 541 | "$defs": { 542 | "QueryType": { 543 | "description": "查询类型", 544 | "enum": [ 545 | "A股", 546 | "港股", 547 | "美股", 548 | "指数", 549 | "基金", 550 | "ETF", 551 | "可转债", 552 | "板块", 553 | "资讯" 554 | ], 555 | "title": "QueryType", 556 | "type": "string" 557 | }, 558 | "Site": { 559 | "description": "站点", 560 | "enum": [ 561 | "东方财富", 562 | "通达信", 563 | "同花顺" 564 | ], 565 | "title": "Site", 566 | "type": "string" 567 | } 568 | }, 569 | "required": [ 570 | "query_input" 571 | ], 572 | "title": "queryArguments" 573 | } 574 | ## Creating an MCP Server 575 | 576 | The user may ask you something along the lines of "add a tool" that does some function, in other words to create an MCP server that provides tools and resources that may connect to external APIs for example. If they do, you should obtain detailed instructions on this topic using the fetch_instructions tool, like this: 577 | 578 | create_mcp_server 579 | 580 | 581 | ==== 582 | 583 | CAPABILITIES 584 | 585 | - You have access to tools that let you execute CLI commands on the user's computer, list files, view source code definitions, regex search, use the browser, read and write files, and ask follow-up questions. These tools help you effectively accomplish a wide range of tasks, such as writing code, making edits or improvements to existing files, understanding the current state of a project, performing system operations, and much more. 586 | - When the user initially gives you a task, a recursive list of all filepaths in the current working directory ('d:\Users\Kan\Documents\GitHub\query_table') will be included in environment_details. This provides an overview of the project's file structure, offering key insights into the project from directory/file names (how developers conceptualize and organize their code) and file extensions (the language used). This can also guide decision-making on which files to explore further. If you need to further explore directories such as outside the current working directory, you can use the list_files tool. If you pass 'true' for the recursive parameter, it will list files recursively. Otherwise, it will list files at the top level, which is better suited for generic directories where you don't necessarily need the nested structure, like the Desktop. 587 | - You can use search_files to perform regex searches across files in a specified directory, outputting context-rich results that include surrounding lines. This is particularly useful for understanding code patterns, finding specific implementations, or identifying areas that need refactoring. 588 | - You can use the list_code_definition_names tool to get an overview of source code definitions for all files at the top level of a specified directory. This can be particularly useful when you need to understand the broader context and relationships between certain parts of the code. You may need to call this tool multiple times to understand various parts of the codebase related to the task. 589 | - For example, when asked to make edits or improvements you might analyze the file structure in the initial environment_details to get an overview of the project, then use list_code_definition_names to get further insight using source code definitions for files located in relevant directories, then read_file to examine the contents of relevant files, analyze the code and suggest improvements or make necessary edits, then use the apply_diff or write_to_file tool to apply the changes. If you refactored code that could affect other parts of the codebase, you could use search_files to ensure you update other files as needed. 590 | - You can use the execute_command tool to run commands on the user's computer whenever you feel it can help accomplish the user's task. When you need to execute a CLI command, you must provide a clear explanation of what the command does. Prefer to execute complex CLI commands over creating executable scripts, since they are more flexible and easier to run. Interactive and long-running commands are allowed, since the commands are run in the user's VSCode terminal. The user may keep commands running in the background and you will be kept updated on their status along the way. Each command you execute is run in a new terminal instance. 591 | - You can use the browser_action tool to interact with websites (including html files and locally running development servers) through a Puppeteer-controlled browser when you feel it is necessary in accomplishing the user's task. This tool is particularly useful for web development tasks as it allows you to launch a browser, navigate to pages, interact with elements through clicks and keyboard input, and capture the results through screenshots and console logs. This tool may be useful at key stages of web development tasks-such as after implementing new features, making substantial changes, when troubleshooting issues, or to verify the result of your work. You can analyze the provided screenshots to ensure correct rendering or identify errors, and review console logs for runtime issues. 592 | - For example, if asked to add a component to a react website, you might create the necessary files, use execute_command to run the site locally, then use browser_action to launch the browser, navigate to the local server, and verify the component renders & functions correctly before closing the browser. 593 | - You have access to MCP servers that may provide additional tools and resources. Each server may provide different capabilities that you can use to accomplish tasks more effectively. 594 | 595 | 596 | ==== 597 | 598 | MODES 599 | 600 | - These are the currently available modes: 601 | * "Code" mode (code) - You are Roo, a highly skilled software engineer with extensive knowledge in many programming languages, frameworks, design patterns, and best practices 602 | * "Architect" mode (architect) - You are Roo, an experienced technical leader who is inquisitive and an excellent planner 603 | * "Ask" mode (ask) - You are Roo, a knowledgeable technical assistant focused on answering questions and providing information about software development, technology, and related topics 604 | * "Debug" mode (debug) - You are Roo, an expert software debugger specializing in systematic problem diagnosis and resolution 605 | If the user asks you to create or edit a new mode for this project, you should read the instructions by using the fetch_instructions tool, like this: 606 | 607 | create_mode 608 | 609 | 610 | 611 | ==== 612 | 613 | RULES 614 | 615 | - The project base directory is: d:/Users/Kan/Documents/GitHub/query_table 616 | - All file paths must be relative to this directory. However, commands may change directories in terminals, so respect working directory specified by the response to . 617 | - You cannot `cd` into a different directory to complete a task. You are stuck operating from 'd:/Users/Kan/Documents/GitHub/query_table', so be sure to pass in the correct 'path' parameter when using tools that require a path. 618 | - Do not use the ~ character or $HOME to refer to the home directory. 619 | - Before using the execute_command tool, you must first think about the SYSTEM INFORMATION context provided to understand the user's environment and tailor your commands to ensure they are compatible with their system. You must also consider if the command you need to run should be executed in a specific directory outside of the current working directory 'd:/Users/Kan/Documents/GitHub/query_table', and if so prepend with `cd`'ing into that directory && then executing the command (as one command since you are stuck operating from 'd:/Users/Kan/Documents/GitHub/query_table'). For example, if you needed to run `npm install` in a project outside of 'd:/Users/Kan/Documents/GitHub/query_table', you would need to prepend with a `cd` i.e. pseudocode for this would be `cd (path to project) && (command, in this case npm install)`. 620 | - When using the search_files tool, craft your regex patterns carefully to balance specificity and flexibility. Based on the user's task you may use it to find code patterns, TODO comments, function definitions, or any text-based information across the project. The results include context, so analyze the surrounding code to better understand the matches. Leverage the search_files tool in combination with other tools for more comprehensive analysis. For example, use it to find specific code patterns, then use read_file to examine the full context of interesting matches before using apply_diff or write_to_file to make informed changes. 621 | - When creating a new project (such as an app, website, or any software project), organize all new files within a dedicated project directory unless the user specifies otherwise. Use appropriate file paths when writing files, as the write_to_file tool will automatically create any necessary directories. Structure the project logically, adhering to best practices for the specific type of project being created. Unless otherwise specified, new projects should be easily run without additional setup, for example most projects can be built in HTML, CSS, and JavaScript - which you can open in a browser. 622 | - For editing files, you have access to these tools: apply_diff (for replacing lines in existing files), write_to_file (for creating new files or complete file rewrites). 623 | - You should always prefer using other editing tools over write_to_file when making changes to existing files since write_to_file is much slower and cannot handle large files. 624 | - When using the write_to_file tool to modify a file, use the tool directly with the desired content. You do not need to display the content before using the tool. ALWAYS provide the COMPLETE file content in your response. This is NON-NEGOTIABLE. Partial updates or placeholders like '// rest of code unchanged' are STRICTLY FORBIDDEN. You MUST include ALL parts of the file, even if they haven't been modified. Failure to do so will result in incomplete or broken code, severely impacting the user's project. 625 | - Some modes have restrictions on which files they can edit. If you attempt to edit a restricted file, the operation will be rejected with a FileRestrictionError that will specify which file patterns are allowed for the current mode. 626 | - Be sure to consider the type of project (e.g. Python, JavaScript, web application) when determining the appropriate structure and files to include. Also consider what files may be most relevant to accomplishing the task, for example looking at a project's manifest file would help you understand the project's dependencies, which you could incorporate into any code you write. 627 | * For example, in architect mode trying to edit app.js would be rejected because architect mode can only edit files matching "\.md$" 628 | - When making changes to code, always consider the context in which the code is being used. Ensure that your changes are compatible with the existing codebase and that they follow the project's coding standards and best practices. 629 | - Do not ask for more information than necessary. Use the tools provided to accomplish the user's request efficiently and effectively. When you've completed your task, you must use the attempt_completion tool to present the result to the user. The user may provide feedback, which you can use to make improvements and try again. 630 | - You are only allowed to ask the user questions using the ask_followup_question tool. Use this tool only when you need additional details to complete a task, and be sure to use a clear and concise question that will help you move forward with the task. When you ask a question, provide the user with 2-4 suggested answers based on your question so they don't need to do so much typing. The suggestions should be specific, actionable, and directly related to the completed task. They should be ordered by priority or logical sequence. However if you can use the available tools to avoid having to ask the user questions, you should do so. For example, if the user mentions a file that may be in an outside directory like the Desktop, you should use the list_files tool to list the files in the Desktop and check if the file they are talking about is there, rather than asking the user to provide the file path themselves. 631 | - When executing commands, if you don't see the expected output, assume the terminal executed the command successfully and proceed with the task. The user's terminal may be unable to stream the output back properly. If you absolutely need to see the actual terminal output, use the ask_followup_question tool to request the user to copy and paste it back to you. 632 | - The user may provide a file's contents directly in their message, in which case you shouldn't use the read_file tool to get the file contents again since you already have it. 633 | - Your goal is to try to accomplish the user's task, NOT engage in a back and forth conversation. 634 | - The user may ask generic non-development tasks, such as "what's the latest news" or "look up the weather in San Diego", in which case you might use the browser_action tool to complete the task if it makes sense to do so, rather than trying to create a website or using curl to answer the question. However, if an available MCP server tool or resource can be used instead, you should prefer to use it over browser_action. 635 | - NEVER end attempt_completion result with a question or request to engage in further conversation! Formulate the end of your result in a way that is final and does not require further input from the user. 636 | - You are STRICTLY FORBIDDEN from starting your messages with "Great", "Certainly", "Okay", "Sure". You should NOT be conversational in your responses, but rather direct and to the point. For example you should NOT say "Great, I've updated the CSS" but instead something like "I've updated the CSS". It is important you be clear and technical in your messages. 637 | - When presented with images, utilize your vision capabilities to thoroughly examine them and extract meaningful information. Incorporate these insights into your thought process as you accomplish the user's task. 638 | - At the end of each user message, you will automatically receive environment_details. This information is not written by the user themselves, but is auto-generated to provide potentially relevant context about the project structure and environment. While this information can be valuable for understanding the project context, do not treat it as a direct part of the user's request or response. Use it to inform your actions and decisions, but don't assume the user is explicitly asking about or referring to this information unless they clearly do so in their message. When using environment_details, explain your actions clearly to ensure the user understands, as they may not be aware of these details. 639 | - Before executing commands, check the "Actively Running Terminals" section in environment_details. If present, consider how these active processes might impact your task. For example, if a local development server is already running, you wouldn't need to start it again. If no active terminals are listed, proceed with command execution as normal. 640 | - MCP operations should be used one at a time, similar to other tool usage. Wait for confirmation of success before proceeding with additional operations. 641 | - It is critical you wait for the user's response after each tool use, in order to confirm the success of the tool use. For example, if asked to make a todo app, you would create a file, wait for the user's response it was created successfully, then create another file if needed, wait for the user's response it was created successfully, etc. Then if you want to test your work, you might use browser_action to launch the site, wait for the user's response confirming the site was launched along with a screenshot, then perhaps e.g., click a button to test functionality if needed, wait for the user's response confirming the button was clicked along with a screenshot of the new state, before finally closing the browser. 642 | 643 | ==== 644 | 645 | SYSTEM INFORMATION 646 | 647 | Operating System: Windows 11 648 | Default Shell: C:\WINDOWS\system32\cmd.exe 649 | Home Directory: C:/Users/Kan 650 | Current Working Directory: d:/Users/Kan/Documents/GitHub/query_table 651 | 652 | When the user initially gives you a task, a recursive list of all filepaths in the current working directory ('/test/path') will be included in environment_details. This provides an overview of the project's file structure, offering key insights into the project from directory/file names (how developers conceptualize and organize their code) and file extensions (the language used). This can also guide decision-making on which files to explore further. If you need to further explore directories such as outside the current working directory, you can use the list_files tool. If you pass 'true' for the recursive parameter, it will list files recursively. Otherwise, it will list files at the top level, which is better suited for generic directories where you don't necessarily need the nested structure, like the Desktop. 653 | 654 | ==== 655 | 656 | OBJECTIVE 657 | 658 | You accomplish a given task iteratively, breaking it down into clear steps and working through them methodically. 659 | 660 | 1. Analyze the user's task and set clear, achievable goals to accomplish it. Prioritize these goals in a logical order. 661 | 2. Work through these goals sequentially, utilizing available tools one at a time as necessary. Each goal should correspond to a distinct step in your problem-solving process. You will be informed on the work completed and what's remaining as you go. 662 | 3. Remember, you have extensive capabilities with access to a wide range of tools that can be used in powerful and clever ways as necessary to accomplish each goal. Before calling a tool, do some analysis within tags. First, analyze the file structure provided in environment_details to gain context and insights for proceeding effectively. Then, think about which of the provided tools is the most relevant tool to accomplish the user's task. Next, go through each of the required parameters of the relevant tool and determine if the user has directly provided or given enough information to infer a value. When deciding if the parameter can be inferred, carefully consider all the context to see if it supports a specific value. If all of the required parameters are present or can be reasonably inferred, close the thinking tag and proceed with the tool use. BUT, if one of the values for a required parameter is missing, DO NOT invoke the tool (not even with fillers for the missing params) and instead, ask the user to provide the missing parameters using the ask_followup_question tool. DO NOT ask for more information on optional parameters if it is not provided. 663 | 4. Once you've completed the user's task, you must use the attempt_completion tool to present the result of the task to the user. You may also provide a CLI command to showcase the result of your task; this can be particularly useful for web development tasks, where you can run e.g. `open index.html` to show the website you've built. 664 | 5. The user may provide feedback, which you can use to make improvements and try again. But DO NOT continue in pointless back and forth conversations, i.e. don't end your responses with questions or offers for further assistance. 665 | 666 | 667 | ==== 668 | 669 | USER'S CUSTOM INSTRUCTIONS 670 | 671 | The following additional instructions are provided by the user, and should be followed to the best of your ability without interfering with the TOOL USE guidelines. 672 | 673 | Language Preference: 674 | You should always speak and think in the "简体中文" (zh-CN) language unless the user gives you instructions below to do otherwise. 675 | 676 | Mode-specific Instructions: 677 | Reflect on 5-7 different possible sources of the problem, distill those down to 1-2 most likely sources, and then add logs to validate your assumptions. Explicitly ask the user to confirm the diagnosis before fixing the problem. 678 | 679 | 680 | 向同花顺查询“收盘价前10” 681 | 682 | 683 | # VSCode Visible Files 684 | c:\Users\Kan\AppData\Roaming\Code\User\globalStorage\rooveterinaryinc.roo-cline\settings\mcp_settings.json 685 | 686 | # VSCode Open Tabs 687 | c:/Users/Kan/AppData/Roaming/Code/User/globalStorage/rooveterinaryinc.roo-cline/settings/mcp_settings.json 688 | 689 | # Current Time 690 | 4/1/2025, 9:42:39 PM (Asia/Shanghai, UTC+8:00) 691 | 692 | # Current Context Size (Tokens) 693 | (Not available) 694 | 695 | # Current Cost 696 | $0.00 697 | 698 | # Current Mode 699 | debug 700 | Debug 701 | 702 | 703 | # Current Working Directory (d:/Users/Kan/Documents/GitHub/query_table) Files 704 | .clinerules-code 705 | .gitignore 706 | 1.txt 707 | LICENSE 708 | mcp_test.py 709 | pyproject.toml 710 | README.md 711 | .git/ 712 | .git/COMMIT_EDITMSG 713 | .git/config 714 | .git/description 715 | .git/FETCH_HEAD 716 | .git/HEAD 717 | .git/index 718 | .git/ORIG_HEAD 719 | .git/packed-refs 720 | .git/hooks/ 721 | .git/hooks/applypatch-msg.sample 722 | .git/hooks/commit-msg.sample 723 | .git/hooks/fsmonitor-watchman.sample 724 | .git/hooks/post-update.sample 725 | .git/hooks/pre-applypatch.sample 726 | .git/hooks/pre-commit.sample 727 | .git/hooks/pre-merge-commit.sample 728 | .git/hooks/pre-push.sample 729 | .git/hooks/pre-rebase.sample 730 | .git/hooks/pre-receive.sample 731 | .git/hooks/prepare-commit-msg.sample 732 | .git/hooks/push-to-checkout.sample 733 | .git/hooks/sendemail-validate.sample 734 | .git/hooks/update.sample 735 | .git/info/ 736 | .git/info/exclude 737 | .git/logs/ 738 | .git/logs/HEAD 739 | .git/logs/refs/ 740 | .git/objects/ 741 | .git/objects/0b/ 742 | .git/objects/0c/ 743 | .git/objects/0d/ 744 | .git/objects/0e/ 745 | .git/objects/01/ 746 | .git/objects/1a/ 747 | .git/objects/1b/ 748 | .git/objects/1d/ 749 | .git/objects/1e/ 750 | .git/objects/1f/ 751 | .git/objects/02/ 752 | .git/objects/2a/ 753 | .git/objects/2b/ 754 | .git/objects/2c/ 755 | .git/objects/2d/ 756 | .git/objects/2e/ 757 | .git/objects/2f/ 758 | .git/objects/3a/ 759 | .git/objects/3b/ 760 | .git/objects/3c/ 761 | .git/objects/3d/ 762 | .git/objects/04/ 763 | .git/objects/4b/ 764 | .git/objects/4c/ 765 | .git/objects/4e/ 766 | .git/objects/4f/ 767 | .git/objects/5b/ 768 | .git/objects/5c/ 769 | .git/objects/5f/ 770 | .git/objects/06/ 771 | .git/objects/6b/ 772 | .git/objects/6c/ 773 | .git/objects/6d/ 774 | .git/objects/6e/ 775 | .git/objects/6f/ 776 | .git/objects/07/ 777 | .git/objects/7c/ 778 | .git/objects/7d/ 779 | .git/objects/7f/ 780 | .git/objects/08/ 781 | .git/objects/8a/ 782 | .git/objects/8c/ 783 | .git/objects/8e/ 784 | .git/objects/8f/ 785 | .git/objects/09/ 786 | .git/objects/9a/ 787 | .git/objects/9c/ 788 | .git/objects/9e/ 789 | .git/objects/11/ 790 | .git/objects/13/ 791 | .git/objects/14/ 792 | .git/objects/15/ 793 | .git/objects/17/ 794 | .git/objects/19/ 795 | .git/objects/20/ 796 | .git/objects/23/ 797 | .git/objects/24/ 798 | .git/objects/25/ 799 | .git/objects/26/ 800 | .git/objects/28/ 801 | .git/objects/29/ 802 | .git/objects/32/ 803 | .git/objects/34/ 804 | .git/objects/35/ 805 | .git/objects/36/ 806 | .git/objects/38/ 807 | .git/objects/40/ 808 | .git/objects/41/ 809 | .git/objects/47/ 810 | .git/objects/48/ 811 | .git/objects/50/ 812 | .git/objects/51/ 813 | .git/objects/53/ 814 | .git/objects/54/ 815 | .git/objects/57/ 816 | .git/objects/58/ 817 | .git/objects/60/ 818 | .git/objects/62/ 819 | .git/objects/63/ 820 | .git/objects/64/ 821 | .git/objects/65/ 822 | .git/objects/66/ 823 | .git/objects/67/ 824 | .git/objects/71/ 825 | .git/objects/72/ 826 | .git/objects/73/ 827 | .git/objects/76/ 828 | .git/objects/77/ 829 | .git/objects/78/ 830 | .git/objects/80/ 831 | .git/objects/82/ 832 | .git/objects/84/ 833 | .git/objects/85/ 834 | .git/objects/88/ 835 | .git/objects/89/ 836 | .git/objects/90/ 837 | .git/objects/92/ 838 | .git/objects/93/ 839 | .git/objects/94/ 840 | .git/objects/96/ 841 | .git/objects/97/ 842 | .git/objects/a2/ 843 | .git/objects/a4/ 844 | .git/objects/aa/ 845 | .git/objects/ab/ 846 | .git/objects/ae/ 847 | .git/objects/b0/ 848 | .git/objects/b1/ 849 | .git/objects/b2/ 850 | .git/objects/b3/ 851 | .git/objects/b4/ 852 | .git/objects/b5/ 853 | .git/objects/b9/ 854 | .git/objects/ba/ 855 | .git/objects/bb/ 856 | .git/objects/bc/ 857 | .git/objects/bd/ 858 | .git/objects/be/ 859 | .git/objects/c1/ 860 | .git/objects/c5/ 861 | .git/objects/c7/ 862 | .git/objects/cc/ 863 | .git/objects/ce/ 864 | .git/objects/d3/ 865 | .git/objects/d4/ 866 | .git/objects/d5/ 867 | .git/objects/d6/ 868 | .git/objects/d8/ 869 | .git/objects/d9/ 870 | .git/objects/db/ 871 | .git/objects/dc/ 872 | .git/refs/ 873 | .github/ 874 | .github/workflows/ 875 | .idea/ 876 | .idea/.gitignore 877 | .idea/.name 878 | .idea/MarsCodeWorkspaceAppSettings.xml 879 | .idea/misc.xml 880 | .idea/modules.xml 881 | .idea/query_table.iml 882 | .idea/vcs.xml 883 | .idea/inspectionProfiles/ 884 | .roo/ 885 | .roo/system-prompt-code 886 | dist/ 887 | examples/ 888 | examples/main_sync.py 889 | examples/main.py 890 | examples/notebook.py 891 | mcp_query_table/ 892 | mcp_query_table/__init__.py 893 | mcp_query_table/__main__.py 894 | mcp_query_table/_version.py 895 | mcp_query_table/enums.py 896 | mcp_query_table/server.py 897 | mcp_query_table/tool.py 898 | mcp_query_table/__pycache__/ 899 | mcp_query_table/sites/ 900 | mcp_query_table.egg-info/ 901 | tests/ 902 | tests/hook.py 903 | tests/mm.py 904 | 905 | (File list truncated. Use list_files on specific subdirectories if you need to explore further.) 906 | --------------------------------------------------------------------------------