Jarvis-Brain 0.1.5.2__py3-none-any.whl → 0.1.7.1__py3-none-any.whl

{jarvis_brain-0.1.5.2.dist-info → jarvis_brain-0.1.7.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Jarvis_Brain
-Version: 0.1.5.2
+Version: 0.1.7.1
 Summary: Jarvis brain mcp
 Requires-Python: >=3.10
 Requires-Dist: beautifulsoup4

jarvis_brain-0.1.7.1.dist-info/RECORD

@@ -0,0 +1,11 @@
+mcp_tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+mcp_tools/dp_tools.py,sha256=ZOmHTIofSeuqiY3YZHNnK72-_MRo22tBAB9NJ4KyEbo,10416
+mcp_tools/main.py,sha256=Fdt2N3oKGwvruuno_ywnuWSlm1BexE9ZY669H2LTo9w,1056
+tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+tools/browser_manager.py,sha256=EXM7n-sDOtdQGpWkVTAZHWhepVU-7PAoUTDNgGF9_fQ,1938
+tools/browser_proxy.py,sha256=YbqXT6h0Zz_H1TMGBd8wwPABIGpE0nGLwYzDlA4t674,6320
+tools/tools.py,sha256=TaWs-CNXy-py9BFmCnJrQ09ke938xXpImf-N2Qo_Rvc,4708
+jarvis_brain-0.1.7.1.dist-info/METADATA,sha256=nLX_3KOUhtCzLP_qScoyY6sdH0CtAWekIlqtfM2RZR0,241
+jarvis_brain-0.1.7.1.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+jarvis_brain-0.1.7.1.dist-info/entry_points.txt,sha256=YFQT4xpkUqt5dM5wlKPQQOqcjMuFrT9iuRAzIpAyH7U,51
+jarvis_brain-0.1.7.1.dist-info/RECORD,,

mcp_tools/dp_tools.py

@@ -1,59 +1,77 @@
+import hashlib
 import json
 import os
 from typing import Any
 
-from DrissionPage._elements.none_element import NoneElement
 from fastmcp import FastMCP
 
+from tools.browser_manager import BrowserManager
 from tools.tools import compress_html, requests_html, dp_headless_html, assert_waf_cookie, dp_mcp_message_pack
+from tools.browser_proxy import DPProxyClient, DPProxyClientManager
 
 html_source_code_local_save_path = os.path.join(os.getcwd(), "html-source-code")
 waf_status_code_dict = {
     412: "瑞数",
     521: "加速乐"
 }
-
-
-def register_visit_url(mcp: FastMCP, browser_manager):
-    @mcp.tool(name="visit_url", description="使用Drissionpage打开url访问某个网站")
-    async def visit_url(url: str) -> dict[str, Any]:
-        port, _browser = browser_manager.create_browser()
+# 一轮最大输入，以免单个html最大长度超过ai最大输入
+ont_turn_max_token = 25000
+
+
+def register_visit_url(mcp: FastMCP, browser_manager: BrowserManager, client_manager: DPProxyClientManager):
+    @mcp.tool(name="visit_url",
+              description="使用Drissionpage打开url访问某个网站，并开始监听初始tab页的所有的XHR请求，当需要使用手机版浏览器Ua时use_mobile_user_agent为True")
+    async def visit_url(url: str, use_mobile_user_agent=False) -> dict[str, Any]:
+        mobile_user_agent = None
+        if use_mobile_user_agent:
+            mobile_user_agent = "Mozilla/5.0 (Linux; Android 6.0; Nexus 5 Build/MRA58N) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/143.0.0.0 Mobile Safari/537.36"
+        port, _browser = browser_manager.create_browser(mobile_user_agent)
         tab = _browser.get_tab()
+        client_manager.create_client(tab)
         tab.get(url)
         tab_id = tab.tab_id
         return dp_mcp_message_pack(
-            f"已在[{port}]端口创建浏览器对象，并已打开链接：{url}",
+            f"已在[{port}]端口创建浏览器对象，并已打开链接：{url}，打开的模式是：{'手机版' if use_mobile_user_agent else '电脑版'}",
             tab_id=tab_id,
             browser_port=port
         )
 
 
-def register_get_html(mcp: FastMCP, browser_manager):
+def register_get_html(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
     @mcp.tool(name="get_html", description="使用Drissionpage获取某一个tab页的html")
     async def get_html(browser_port: int, tab_id: str) -> dict[str, Any]:
         _browser = browser_manager.get_browser(browser_port)
         tab = _browser.get_tab(tab_id)
-        file_name = tab.title + f"_{tab_id}.html"
+        file_name_prefix = str(tab.title).replace("/", "_").replace(":", "_")
         if not os.path.exists(html_source_code_local_save_path):
             os.makedirs(html_source_code_local_save_path)
-        abs_path = os.path.join(html_source_code_local_save_path, file_name)
-        with open(abs_path, "w", encoding="utf-8") as f:
-            min_html, compress_rate = compress_html(tab.html)
-            f.write(min_html)
-        return dp_mcp_message_pack(f"已保存tab页：【{tab_id}】的html源码", tab_id=tab_id, html_local_path=abs_path)
-
-
-def register_get_new_tab(mcp: FastMCP, browser_manager):
-    @mcp.tool(name="get_new_tab", description="使用Drissionpage创建一个新的tab页，在新的tab页中打开url")
+        min_html, compress_rate = compress_html(tab.html)
+        html_str_list = [min_html[i:i + ont_turn_max_token] for i in range(0, len(min_html), ont_turn_max_token)]
+        html_file_list = []
+        for index, html_str in enumerate(html_str_list):
+            file_name = file_name_prefix + f"_{tab_id}_segment{index}.html"
+            abs_path = os.path.join(html_source_code_local_save_path, file_name)
+            with open(abs_path, "w", encoding="utf-8") as f:
+                f.write(html_str)
+            html_file_list.append(abs_path)
+        message = f"已保存tab页：【{tab_id}】的html源码片段共{len(html_file_list)}个"
+        return dp_mcp_message_pack(message, tab_id=tab_id, htmls_local_path=html_file_list)
+
+
+def register_get_new_tab(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
+    @mcp.tool(name="get_new_tab",
+              description="使用Drissionpage创建一个新的tab页，在新的tab页中打开url，并开始监听新的tab页的所有XHR请求")
     async def get_new_tab(browser_port: int, url: str) -> dict[str, Any]:
         _browser = browser_manager.get_browser(browser_port)
-        tab = _browser.new_tab(url)
+        tab = _browser.new_tab()
+        client_manager.create_client(tab)
+        tab.get(url)
         _browser.activate_tab(tab)
         tab_id = tab.tab_id
         return dp_mcp_message_pack(f"已创建新的tab页，并打开链接：{url}", tab_id=tab_id)
 
 
-def register_switch_tab(mcp: FastMCP, browser_manager):
+def register_switch_tab(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
     @mcp.tool(name="switch_tab", description="根据传入的tab_id切换到对应的tab页", )
     async def switch_tab(browser_port: int, tab_id: str) -> dict[str, Any]:
         _browser = browser_manager.get_browser(browser_port)
@@ -61,7 +79,7 @@ def register_switch_tab(mcp: FastMCP, browser_manager):
         return dp_mcp_message_pack(f"已将tab页:【{tab_id}】切换至最前端")
 
 
-def register_close_tab(mcp: FastMCP, browser_manager):
+def register_close_tab(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
     @mcp.tool(name="close_tab", description="根据传入的tab_id关闭tab页", )
     async def close_tab(browser_port, tab_id) -> dict[str, Any]:
         _browser = browser_manager.get_browser(browser_port)
@@ -69,9 +87,10 @@ def register_close_tab(mcp: FastMCP, browser_manager):
         return dp_mcp_message_pack(f"已将tab页:【{tab_id}】关闭")
 
 
-def register_check_selector(mcp: FastMCP, browser_manager):
+def register_check_selector(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
     @mcp.tool(name="check_selector", description="查找tab页中是否包含元素，并返回元素attr_name所对应的值")
-    async def check_selector(browser_port: int, tab_id: str, css_selector: str, attr_name: str = "text") -> dict[str, Any]:
+    async def check_selector(browser_port: int, tab_id: str, css_selector: str, attr_name: str = "text") -> dict[
+        str, Any]:
         _browser = browser_manager.get_browser(browser_port)
         target_tab = _browser.get_tab(tab_id)
         css_selector = css_selector
@@ -95,7 +114,7 @@ def register_check_selector(mcp: FastMCP, browser_manager):
         )
 
 
-def register_quit_browser(mcp: FastMCP, browser_manager):
+def register_quit_browser(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
     @mcp.tool(name="quit_browser", description="退出浏览器会话，关闭浏览器")
     async def quit_browser(browser_port: int) -> dict[str, Any]:
         flag, _browser = browser_manager.remove_page(browser_port)
@@ -108,7 +127,25 @@ def register_quit_browser(mcp: FastMCP, browser_manager):
         )
 
 
-def register_assert_waf(mcp: FastMCP, browser_manager):
+def register_pop_first_packet(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
+    @mcp.tool(name="pop_first_packet",
+              description="每调用一次就会弹出传入的tab页所监听到的数据包中的第一个packet_message")
+    async def pop_first_packet(browser_port: int, tab_id: str) -> dict[str, Any]:
+        _browser = browser_manager.get_browser(browser_port)
+        client = client_manager.get_client(tab_id)
+        packet_message = client.pop_first_packet()
+        message = f"tab页:【{tab_id}】，暂时没有监听到XHR数据包"
+        if packet_message:
+            message = f"tab页:【{tab_id}】，监听到XHR数据包",
+        return dp_mcp_message_pack(
+            message,
+            browser_port=browser_port,
+            tab_id=tab_id,
+            packet_message=packet_message
+        )
+
+
+def register_assert_waf(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
     @mcp.tool(name="assert_waf",
               description="通过对比requests、有头浏览器、无头浏览器获取到的html，判断网页是否使用了waf以及是否为动态渲染的网页")
     async def assert_waf(browser_port: int, tab_id: str) -> dict[str, Any]:
@@ -156,11 +193,3 @@ def register_assert_waf(mcp: FastMCP, browser_manager):
             raw_headless_rate_difference=h_hless_rate_diff,
             head_headless_rate_difference=h_hless_rate_diff
         )
-
-# def register_highlight_element_captcha(mcp: FastMCP, browser_manager):
-#     @mcp.tool(name="highlight_element_captcha",
-#               description="将传入的Selector在页面上高亮，并截屏")
-#     async def highlight_element_captcha(browser_port: int, tab_id: str, selector: str) -> dict[str, Any]:
-#         _browser = browser_manager.get_browser(browser_port)
-#         tab = _browser.get_tab(tab_id)
-#         tab.ele

mcp_tools/main.py

@@ -2,25 +2,27 @@ from fastmcp import FastMCP
 
 from mcp_tools.dp_tools import *
 from tools.browser_manager import browser_manager
+from tools.browser_proxy import client_manager
 
 mcp = FastMCP("Jarvis Brain Mcp Tools")
 
 # 根据环境变量加载模块
 enabled_modules = os.getenv("MCP_MODULES", "TeamNode-Dp").split(",")
-base_cwd = os.getenv("BASE_CWD", os.environ.get("PWD"))
+base_cwd = os.getenv("BASE_CWD", os.path.expanduser('~'))
 
 if "TeamNode-Dp" in enabled_modules:
     # 页面管理
-    register_close_tab(mcp, browser_manager)
-    register_switch_tab(mcp, browser_manager)
-    register_get_new_tab(mcp, browser_manager)
+    register_close_tab(mcp, browser_manager, client_manager)
+    register_switch_tab(mcp, browser_manager, client_manager)
+    register_get_new_tab(mcp, browser_manager, client_manager)
     # 功能
-    register_visit_url(mcp, browser_manager)
-    register_get_html(mcp, browser_manager)
-    register_check_selector(mcp, browser_manager)
+    register_visit_url(mcp, browser_manager, client_manager)
+    register_get_html(mcp, browser_manager, client_manager)
+    register_check_selector(mcp, browser_manager, client_manager)
+    register_pop_first_packet(mcp, browser_manager, client_manager)
 
 if "JarvisNode" in enabled_modules:
-    register_assert_waf(mcp, browser_manager)
+    register_assert_waf(mcp, browser_manager, client_manager)
 
 
 def main():

tools/browser_manager.py

@@ -3,6 +3,7 @@ import random
 from typing import Optional, Tuple
 import os
 from DrissionPage import ChromiumPage, ChromiumOptions
+import platform
 
 
 class BrowserManager:
@@ -15,18 +16,21 @@ class BrowserManager:
             cls._instance.browser_pool = {}
         return cls._instance
 
-    def create_browser(self) -> Tuple[int, ChromiumPage]:
+    def create_browser(self, user_agent: str = None) -> Tuple[int, ChromiumPage]:
         """创建新的浏览器实例"""
         random_port = random.randint(9223, 9934)
         while random_port in self.browser_pool:
             random_port = random.randint(9223, 9934)
 
         co = ChromiumOptions().set_local_port(random_port)
+        if user_agent:
+            co.set_user_agent(user_agent)
+        if platform.system() != 'Windows':
+            co.set_argument('--no-sandbox')
         custom_data_dir = os.path.join(os.path.expanduser('~'), 'DrissionPage', "userData", f"{random_port}")
-
-        # 2. 配置浏览器选项，指定该目录
-        # co = ChromiumOptions()
         co.set_user_data_path(custom_data_dir)  # 设置用户数据路径
+        # if not os.path.exists(custom_data_dir):
+        #     os.makedirs(custom_data_dir)
         self.browser_pool[random_port] = ChromiumPage(co)
         return random_port, self.browser_pool[random_port]
 

tools/browser_proxy.py

@@ -0,0 +1,163 @@
+import threading
+from collections import deque
+
+from DrissionPage import ChromiumPage, ChromiumOptions
+from DrissionPage._pages.chromium_tab import ChromiumTab
+from DrissionPage._units.listener import DataPacket
+from typing import Tuple, Optional
+import json
+
+
+class DPProxyClient:
+    def __init__(self, driver: ChromiumTab, self_kill=False):
+        self.tab_id = driver.tab_id
+        self.driver = ChromePageProxy(driver, self)
+        self.thread = None
+        self.self_kill = self_kill
+        # self.packet_list = []
+        self.packet_queue = deque()
+
+    def get_driver(self, start_listen, count=None, timeout=10) -> ChromiumTab:
+        """
+        获取代理后的driver、tab
+        :param start_listen: 若你自己写的代码里已经使用自动化框架监听发包的功能了，则该值应该置为False。若没监听，则必须将该值置为True
+        :param count: 需捕获的数据包总数，为None表示无限
+        :param timeout: 两个数据包之间等待的最大时长（秒），为None表示无限，默认为10秒
+        :return:
+        """
+        if start_listen:
+            self.driver.listen.set_targets(res_type="XHR")
+            self.driver.listen.start()
+            self.thread = threading.Thread(target=self.start_listen, args=(count, timeout,))
+            self.thread.start()
+        return self.driver
+
+    def start_listen(self, count=None, timeout=10):
+        for _ in self.driver.listen.steps(count=count, timeout=timeout, gap=1):
+            pass
+
+    # 每次调用函数，都从队列的左端弹出一个数据包
+    def pop_first_packet(self):
+        if self.packet_queue:
+            result = self.packet_queue.popleft()
+            return json.dumps(result, ensure_ascii=False)
+        else:
+            return None
+
+
+class DPProxyClientManager:
+    """浏览器池管理器 - 使用单例模式"""
+    _instance = None
+
+    def __new__(cls):
+        if cls._instance is None:
+            cls._instance = super().__new__(cls)
+            cls._instance.tab_pool = {}
+        return cls._instance
+
+    def create_client(self, tab: ChromiumTab, self_kill=False) -> Tuple[str, DPProxyClient, ChromiumTab]:
+        """创建新的tab页面代理实例"""
+        client = DPProxyClient(tab, self_kill=self_kill)
+        tab = client.get_driver(True)
+        tab_id = tab.tab_id
+        self.tab_pool[tab_id] = {"client": client, "driver": tab}
+        return tab_id, client, tab
+
+    def get_client(self, tab_id: str) -> Optional[DPProxyClient]:
+        """根据端口获取浏览器实例"""
+        return self.tab_pool.get(tab_id).get("client", None)
+
+    def remove_client(self, tab_id: str) -> Tuple[bool, Optional[ChromiumPage]]:
+        """根据端口移除浏览器实例"""
+        client = self.tab_pool.pop(tab_id, None)
+        return client is not None, client
+
+    def list_clients(self) -> list[int]:
+        """列出所有活跃的浏览器端口"""
+        return list(self.tab_pool.keys())
+
+
+class ChromePageProxy:
+    def __init__(self, page, client=None):
+        self.__dict__['page'] = page
+        self.__dict__['client'] = client
+
+    def __getattr__(self, item):
+        attr = getattr(self.page, item)
+        print(item, attr)
+        if item == 'listen':
+            listen_proxy = DrissionPageListenerProxy(attr, self.__dict__['client'])
+            return listen_proxy
+        return attr
+
+
+class DrissionPageListenerProxy:
+    def __init__(self, listener, client=None):
+        self.listener = listener
+        self.client = client
+
+    def __getattr__(self, item):
+        attr = getattr(self.listener, item)
+        # 当监听到wait被调用的时候
+        if item == "wait":
+            def wrapper(*args, **kwargs):
+                result = attr(*args, **kwargs)
+                check_data_packet(result, self.client)
+                return result
+
+            return wrapper
+        # 当监听到steps被调用的时候
+        if item == "steps":
+            def wrapper(*args, **kwargs):
+                if kwargs.get("gap", 1) > 1:
+                    raise Exception("暂不支持多包监控")
+                result = attr(*args, **kwargs)
+                if attr.__name__ == "steps":
+                    for step in result:
+                        check_data_packet(step, self.client)
+                        yield step
+
+            return wrapper
+        return attr
+
+
+def check_data_packet(packet: DataPacket, client: DPProxyClient):
+    """
+    封装监听到的数据包，并将其存放在client的packet_queue中
+    :param packet:
+    :param client:
+    :return:
+    """
+    url = packet.url
+    method = packet.request.method
+    data = None
+    if packet.request.hasPostData:
+        data = packet.request.postData
+    temp_dict = {
+        "url": url,
+        "method": method,
+        "request_data": data,
+        "request_headers": packet.request.headers,
+        "response_headers": packet.response.headers,
+        "response_body": packet.response.body,
+    }
+    print(temp_dict)
+    client.packet_queue.append(temp_dict)
+
+
+client_manager = DPProxyClientManager()
+
+#
+# if __name__ == '__main__':
+#     co = ChromiumOptions().set_user_agent(
+#         "Mozilla/5.0 (Linux; Android 6.0; Nexus 5 Build/MRA58N) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/143.0.0.0 Mobile Safari/537.36")
+#     tab = ChromiumPage(co).latest_tab
+#     client = DPProxyClient(tab, self_kill=False)
+#     # client = CaptchaClient(tab, self_kill=True)
+#     tab = client.get_driver(True)
+#     url = "https://api.toutiaoapi.com/feoffline/hotspot_and_local/html/hot_list/index.html?client_extra_params=%7B%22custom_log_pb%22%3A%22%7B%5C%22style_id%5C%22%3A%5C%2240030%5C%22%2C%5C%22entrance_hotspot%5C%22%3A%5C%22search%5C%22%2C%5C%22location%5C%22%3A%5C%22hot_board%5C%22%2C%5C%22category_name%5C%22%3A%5C%22hotboard_light%5C%22%7D%22%7D&count=50&log_pb=%7B%22style_id%22%3A%2240030%22%2C%22entrance_hotspot%22%3A%22search%22%2C%22location%22%3A%22hot_board%22%2C%22category_name%22%3A%22hotboard_light%22%7D&only_hot_list=1&tab_name=stream&enter_keyword=%23%E7%BE%8E%E5%9B%BD%E9%80%80%E5%87%BA66%E4%B8%AA%E5%9B%BD%E9%99%85%E7%BB%84%E7%BB%87%23"
+#     tab.get(url)
+#     for _ in range(5056):
+#         new_packet = client.pop_first_packet()
+#         print(new_packet, "23")
+#         time.sleep(1)

tools/tools.py

@@ -1,5 +1,6 @@
 import time
-
+import random
+import os
 import minify_html
 from DrissionPage import ChromiumPage, ChromiumOptions
 from bs4 import BeautifulSoup
@@ -21,8 +22,11 @@ def requests_html(url):
 def dp_headless_html(url):
     opt = ChromiumOptions().headless(True)
     opt.set_argument('--no-sandbox')
-    # 随机端口，固定9222端口的话，可能会被瑞数检测到
-    opt.auto_port()
+    """创建新的浏览器实例"""
+    random_port = random.randint(9934, 10034)
+    custom_data_dir = os.path.join(os.path.expanduser('~'), 'DrissionPage', "userData", f"{random_port}")
+    opt.set_user_data_path(custom_data_dir)  # 设置用户数据路径
+    opt.set_local_port(random_port)
     page = ChromiumPage(opt)
     tab = page.latest_tab
     tab.get(url)

jarvis_brain-0.1.5.2.dist-info/RECORD

@@ -1,10 +0,0 @@
-mcp_tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-mcp_tools/dp_tools.py,sha256=TsktupVxzQkDRH4w-niEqJba3LkeWxIeyyx24ehpqAs,8411
-mcp_tools/main.py,sha256=ockDfiMWaNroXf6sJ9dk-jU-_SvoGgP-0t-TbW2XWyw,827
-tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-tools/browser_manager.py,sha256=G0nUQSmMVWzS08IjO87reEyMCpy-IiM3BxwdNUvcmdI,1740
-tools/tools.py,sha256=qPa80xzQWCD8mSdg8isCgyei0XN8a7guGtR2J_MbQDU,4481
-jarvis_brain-0.1.5.2.dist-info/METADATA,sha256=s5BNa9hKn38Tk0_S8iGPpDxwZ7h-x2NO1iXLtV8kRzM,241
-jarvis_brain-0.1.5.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-jarvis_brain-0.1.5.2.dist-info/entry_points.txt,sha256=YFQT4xpkUqt5dM5wlKPQQOqcjMuFrT9iuRAzIpAyH7U,51
-jarvis_brain-0.1.5.2.dist-info/RECORD,,