Jarvis-Brain 0.1.6.1__py3-none-any.whl → 0.1.7.2__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: Jarvis_Brain
3
- Version: 0.1.6.1
3
+ Version: 0.1.7.2
4
4
  Summary: Jarvis brain mcp
5
5
  Requires-Python: >=3.10
6
6
  Requires-Dist: beautifulsoup4
@@ -0,0 +1,11 @@
1
+ mcp_tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
2
+ mcp_tools/dp_tools.py,sha256=LT2x9edWmmDqNoy66zPgwjUSLpYbMPUPluvDoAXs0FU,10421
3
+ mcp_tools/main.py,sha256=Fdt2N3oKGwvruuno_ywnuWSlm1BexE9ZY669H2LTo9w,1056
4
+ tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
5
+ tools/browser_manager.py,sha256=EXM7n-sDOtdQGpWkVTAZHWhepVU-7PAoUTDNgGF9_fQ,1938
6
+ tools/browser_proxy.py,sha256=yVTRpZA0IGLsL2Bl6zN_4hMdx2lX9UaMx5O6DPlrGRg,6333
7
+ tools/tools.py,sha256=TaWs-CNXy-py9BFmCnJrQ09ke938xXpImf-N2Qo_Rvc,4708
8
+ jarvis_brain-0.1.7.2.dist-info/METADATA,sha256=_j2fYANuNrAtnMwBnttjNSZ9z2vq7CDbjHS5LNnzKmQ,241
9
+ jarvis_brain-0.1.7.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
10
+ jarvis_brain-0.1.7.2.dist-info/entry_points.txt,sha256=YFQT4xpkUqt5dM5wlKPQQOqcjMuFrT9iuRAzIpAyH7U,51
11
+ jarvis_brain-0.1.7.2.dist-info/RECORD,,
mcp_tools/dp_tools.py CHANGED
@@ -7,6 +7,7 @@ from fastmcp import FastMCP
7
7
 
8
8
  from tools.browser_manager import BrowserManager
9
9
  from tools.tools import compress_html, requests_html, dp_headless_html, assert_waf_cookie, dp_mcp_message_pack
10
+ from tools.browser_proxy import DPProxyClient, DPProxyClientManager
10
11
 
11
12
  html_source_code_local_save_path = os.path.join(os.getcwd(), "html-source-code")
12
13
  waf_status_code_dict = {
@@ -17,7 +18,7 @@ waf_status_code_dict = {
17
18
  ont_turn_max_token = 25000
18
19
 
19
20
 
20
- def register_visit_url(mcp: FastMCP, browser_manager: BrowserManager):
21
+ def register_visit_url(mcp: FastMCP, browser_manager: BrowserManager, client_manager: DPProxyClientManager):
21
22
  @mcp.tool(name="visit_url",
22
23
  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")
23
24
  async def visit_url(url: str, use_mobile_user_agent=False) -> dict[str, Any]:
@@ -26,6 +27,7 @@ def register_visit_url(mcp: FastMCP, browser_manager: BrowserManager):
26
27
  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"
27
28
  port, _browser = browser_manager.create_browser(mobile_user_agent)
28
29
  tab = _browser.get_tab()
30
+ client_manager.create_client(tab)
29
31
  tab.get(url)
30
32
  tab_id = tab.tab_id
31
33
  return dp_mcp_message_pack(
@@ -35,12 +37,12 @@ def register_visit_url(mcp: FastMCP, browser_manager: BrowserManager):
35
37
  )
36
38
 
37
39
 
38
- def register_get_html(mcp: FastMCP, browser_manager):
40
+ def register_get_html(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
39
41
  @mcp.tool(name="get_html", description="使用Drissionpage获取某一个tab页的html")
40
42
  async def get_html(browser_port: int, tab_id: str) -> dict[str, Any]:
41
43
  _browser = browser_manager.get_browser(browser_port)
42
44
  tab = _browser.get_tab(tab_id)
43
- file_name_prefix = str(tab.title).replace("/", "_").replace(":", "_")
45
+ file_name_prefix = hashlib.md5(str(tab.title).encode('utf-8')).hexdigest()
44
46
  if not os.path.exists(html_source_code_local_save_path):
45
47
  os.makedirs(html_source_code_local_save_path)
46
48
  min_html, compress_rate = compress_html(tab.html)
@@ -56,18 +58,20 @@ def register_get_html(mcp: FastMCP, browser_manager):
56
58
  return dp_mcp_message_pack(message, tab_id=tab_id, htmls_local_path=html_file_list)
57
59
 
58
60
 
59
- def register_get_new_tab(mcp: FastMCP, browser_manager):
61
+ def register_get_new_tab(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
60
62
  @mcp.tool(name="get_new_tab",
61
63
  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")
62
64
  async def get_new_tab(browser_port: int, url: str) -> dict[str, Any]:
63
65
  _browser = browser_manager.get_browser(browser_port)
64
- tab = _browser.new_tab(url)
66
+ tab = _browser.new_tab()
67
+ client_manager.create_client(tab)
68
+ tab.get(url)
65
69
  _browser.activate_tab(tab)
66
70
  tab_id = tab.tab_id
67
71
  return dp_mcp_message_pack(f"已创建新的tab页,并打开链接:{url}", tab_id=tab_id)
68
72
 
69
73
 
70
- def register_switch_tab(mcp: FastMCP, browser_manager):
74
+ def register_switch_tab(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
71
75
  @mcp.tool(name="switch_tab", description="根据传入的tab_id切换到对应的tab页", )
72
76
  async def switch_tab(browser_port: int, tab_id: str) -> dict[str, Any]:
73
77
  _browser = browser_manager.get_browser(browser_port)
@@ -75,7 +79,7 @@ def register_switch_tab(mcp: FastMCP, browser_manager):
75
79
  return dp_mcp_message_pack(f"已将tab页:【{tab_id}】切换至最前端")
76
80
 
77
81
 
78
- def register_close_tab(mcp: FastMCP, browser_manager):
82
+ def register_close_tab(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
79
83
  @mcp.tool(name="close_tab", description="根据传入的tab_id关闭tab页", )
80
84
  async def close_tab(browser_port, tab_id) -> dict[str, Any]:
81
85
  _browser = browser_manager.get_browser(browser_port)
@@ -83,7 +87,7 @@ def register_close_tab(mcp: FastMCP, browser_manager):
83
87
  return dp_mcp_message_pack(f"已将tab页:【{tab_id}】关闭")
84
88
 
85
89
 
86
- def register_check_selector(mcp: FastMCP, browser_manager):
90
+ def register_check_selector(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
87
91
  @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")
88
92
  async def check_selector(browser_port: int, tab_id: str, css_selector: str, attr_name: str = "text") -> dict[
89
93
  str, Any]:
@@ -110,7 +114,7 @@ def register_check_selector(mcp: FastMCP, browser_manager):
110
114
  )
111
115
 
112
116
 
113
- def register_quit_browser(mcp: FastMCP, browser_manager):
117
+ def register_quit_browser(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
114
118
  @mcp.tool(name="quit_browser", description="退出浏览器会话,关闭浏览器")
115
119
  async def quit_browser(browser_port: int) -> dict[str, Any]:
116
120
  flag, _browser = browser_manager.remove_page(browser_port)
@@ -123,7 +127,25 @@ def register_quit_browser(mcp: FastMCP, browser_manager):
123
127
  )
124
128
 
125
129
 
126
- def register_assert_waf(mcp: FastMCP, browser_manager):
130
+ def register_pop_first_packet(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
131
+ @mcp.tool(name="pop_first_packet",
132
+ 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")
133
+ async def pop_first_packet(browser_port: int, tab_id: str) -> dict[str, Any]:
134
+ _browser = browser_manager.get_browser(browser_port)
135
+ client = client_manager.get_client(tab_id)
136
+ packet_message = client.pop_first_packet()
137
+ message = f"tab页:【{tab_id}】,暂时没有监听到XHR数据包"
138
+ if packet_message:
139
+ message = f"tab页:【{tab_id}】,监听到XHR数据包",
140
+ return dp_mcp_message_pack(
141
+ message,
142
+ browser_port=browser_port,
143
+ tab_id=tab_id,
144
+ packet_message=packet_message
145
+ )
146
+
147
+
148
+ def register_assert_waf(mcp: FastMCP, browser_manager, client_manager: DPProxyClientManager):
127
149
  @mcp.tool(name="assert_waf",
128
150
  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")
129
151
  async def assert_waf(browser_port: int, tab_id: str) -> dict[str, Any]:
mcp_tools/main.py CHANGED
@@ -2,6 +2,7 @@ from fastmcp import FastMCP
2
2
 
3
3
  from mcp_tools.dp_tools import *
4
4
  from tools.browser_manager import browser_manager
5
+ from tools.browser_proxy import client_manager
5
6
 
6
7
  mcp = FastMCP("Jarvis Brain Mcp Tools")
7
8
 
@@ -11,16 +12,17 @@ base_cwd = os.getenv("BASE_CWD", os.path.expanduser('~'))
11
12
 
12
13
  if "TeamNode-Dp" in enabled_modules:
13
14
  # 页面管理
14
- register_close_tab(mcp, browser_manager)
15
- register_switch_tab(mcp, browser_manager)
16
- register_get_new_tab(mcp, browser_manager)
15
+ register_close_tab(mcp, browser_manager, client_manager)
16
+ register_switch_tab(mcp, browser_manager, client_manager)
17
+ register_get_new_tab(mcp, browser_manager, client_manager)
17
18
  # 功能
18
- register_visit_url(mcp, browser_manager)
19
- register_get_html(mcp, browser_manager)
20
- register_check_selector(mcp, browser_manager)
19
+ register_visit_url(mcp, browser_manager, client_manager)
20
+ register_get_html(mcp, browser_manager, client_manager)
21
+ register_check_selector(mcp, browser_manager, client_manager)
22
+ register_pop_first_packet(mcp, browser_manager, client_manager)
21
23
 
22
24
  if "JarvisNode" in enabled_modules:
23
- register_assert_waf(mcp, browser_manager)
25
+ register_assert_waf(mcp, browser_manager, client_manager)
24
26
 
25
27
 
26
28
  def main():
tools/browser_proxy.py ADDED
@@ -0,0 +1,163 @@
1
+ import threading
2
+ from collections import deque
3
+ import time
4
+ from DrissionPage import ChromiumPage, ChromiumOptions
5
+ from DrissionPage._pages.chromium_tab import ChromiumTab
6
+ from DrissionPage._units.listener import DataPacket
7
+ from typing import Tuple, Optional
8
+ import json
9
+
10
+
11
+ class DPProxyClient:
12
+ def __init__(self, driver: ChromiumTab, self_kill=False):
13
+ self.tab_id = driver.tab_id
14
+ self.driver = ChromePageProxy(driver, self)
15
+ self.thread = None
16
+ self.self_kill = self_kill
17
+ # self.packet_list = []
18
+ self.packet_queue = deque()
19
+
20
+ def get_driver(self, start_listen, count=None, timeout=10) -> ChromiumTab:
21
+ """
22
+ 获取代理后的driver、tab
23
+ :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
24
+ :param count: 需捕获的数据包总数,为None表示无限
25
+ :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
26
+ :return:
27
+ """
28
+ if start_listen:
29
+ self.driver.listen.set_targets(res_type="XHR")
30
+ self.driver.listen.start()
31
+ self.thread = threading.Thread(target=self.start_listen, args=(count, timeout,))
32
+ self.thread.start()
33
+ return self.driver
34
+
35
+ def start_listen(self, count=None, timeout=10):
36
+ for _ in self.driver.listen.steps(count=count, timeout=timeout, gap=1):
37
+ pass
38
+
39
+ # 每次调用函数,都从队列的左端弹出一个数据包
40
+ def pop_first_packet(self):
41
+ if self.packet_queue:
42
+ result = self.packet_queue.popleft()
43
+ return json.dumps(result, ensure_ascii=False)
44
+ else:
45
+ return None
46
+
47
+
48
+ class DPProxyClientManager:
49
+ """浏览器池管理器 - 使用单例模式"""
50
+ _instance = None
51
+
52
+ def __new__(cls):
53
+ if cls._instance is None:
54
+ cls._instance = super().__new__(cls)
55
+ cls._instance.tab_pool = {}
56
+ return cls._instance
57
+
58
+ def create_client(self, tab: ChromiumTab, self_kill=False) -> Tuple[str, DPProxyClient, ChromiumTab]:
59
+ """创建新的tab页面代理实例"""
60
+ client = DPProxyClient(tab, self_kill=self_kill)
61
+ tab = client.get_driver(True)
62
+ tab_id = tab.tab_id
63
+ self.tab_pool[tab_id] = {"client": client, "driver": tab}
64
+ return tab_id, client, tab
65
+
66
+ def get_client(self, tab_id: str) -> Optional[DPProxyClient]:
67
+ """根据端口获取浏览器实例"""
68
+ return self.tab_pool.get(tab_id).get("client", None)
69
+
70
+ def remove_client(self, tab_id: str) -> Tuple[bool, Optional[ChromiumPage]]:
71
+ """根据端口移除浏览器实例"""
72
+ client = self.tab_pool.pop(tab_id, None)
73
+ return client is not None, client
74
+
75
+ def list_clients(self) -> list[int]:
76
+ """列出所有活跃的浏览器端口"""
77
+ return list(self.tab_pool.keys())
78
+
79
+
80
+ class ChromePageProxy:
81
+ def __init__(self, page, client=None):
82
+ self.__dict__['page'] = page
83
+ self.__dict__['client'] = client
84
+
85
+ def __getattr__(self, item):
86
+ attr = getattr(self.page, item)
87
+ print(item, attr)
88
+ if item == 'listen':
89
+ listen_proxy = DrissionPageListenerProxy(attr, self.__dict__['client'])
90
+ return listen_proxy
91
+ return attr
92
+
93
+
94
+ class DrissionPageListenerProxy:
95
+ def __init__(self, listener, client=None):
96
+ self.listener = listener
97
+ self.client = client
98
+
99
+ def __getattr__(self, item):
100
+ attr = getattr(self.listener, item)
101
+ # 当监听到wait被调用的时候
102
+ if item == "wait":
103
+ def wrapper(*args, **kwargs):
104
+ result = attr(*args, **kwargs)
105
+ check_data_packet(result, self.client)
106
+ return result
107
+
108
+ return wrapper
109
+ # 当监听到steps被调用的时候
110
+ if item == "steps":
111
+ def wrapper(*args, **kwargs):
112
+ if kwargs.get("gap", 1) > 1:
113
+ raise Exception("暂不支持多包监控")
114
+ result = attr(*args, **kwargs)
115
+ if attr.__name__ == "steps":
116
+ for step in result:
117
+ check_data_packet(step, self.client)
118
+ yield step
119
+
120
+ return wrapper
121
+ return attr
122
+
123
+
124
+ def check_data_packet(packet: DataPacket, client: DPProxyClient):
125
+ """
126
+ 封装监听到的数据包,并将其存放在client的packet_queue中
127
+ :param packet:
128
+ :param client:
129
+ :return:
130
+ """
131
+ url = packet.url
132
+ method = packet.request.method
133
+ data = None
134
+ if packet.request.hasPostData:
135
+ data = packet.request.postData
136
+ body = packet.response.body
137
+ temp_dict = {
138
+ "url": url,
139
+ "method": method,
140
+ "request_data": data,
141
+ "request_headers": dict(packet.request.headers),
142
+ "response_headers": dict(packet.response.headers),
143
+ "response_body": body,
144
+ }
145
+ # print(temp_dict)
146
+ client.packet_queue.append(temp_dict)
147
+
148
+
149
+ client_manager = DPProxyClientManager()
150
+
151
+ if __name__ == '__main__':
152
+ co = ChromiumOptions().set_user_agent(
153
+ "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")
154
+ tab = ChromiumPage(co).latest_tab
155
+ client = DPProxyClient(tab, self_kill=False)
156
+ # client = CaptchaClient(tab, self_kill=True)
157
+ tab = client.get_driver(True)
158
+ 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"
159
+ tab.get(url)
160
+ for _ in range(5056):
161
+ new_packet = client.pop_first_packet()
162
+ print(new_packet, "23")
163
+ time.sleep(1)
@@ -1,10 +0,0 @@
1
- mcp_tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
2
- mcp_tools/dp_tools.py,sha256=hijWdWQ9fhAQXrtrtqkaUd3CUH5sORAoQaF7BVtljfY,9078
3
- mcp_tools/main.py,sha256=SiLF-tcEdApkzMjTZJEpVYX5d-LPW7aqy6A6lCBA7YQ,829
4
- tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
5
- tools/browser_manager.py,sha256=EXM7n-sDOtdQGpWkVTAZHWhepVU-7PAoUTDNgGF9_fQ,1938
6
- tools/tools.py,sha256=TaWs-CNXy-py9BFmCnJrQ09ke938xXpImf-N2Qo_Rvc,4708
7
- jarvis_brain-0.1.6.1.dist-info/METADATA,sha256=1CSoEWNLJzbP1L_4Gz3ElX2jANUxnsMh8EqtwM5D6Pw,241
8
- jarvis_brain-0.1.6.1.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
9
- jarvis_brain-0.1.6.1.dist-info/entry_points.txt,sha256=YFQT4xpkUqt5dM5wlKPQQOqcjMuFrT9iuRAzIpAyH7U,51
10
- jarvis_brain-0.1.6.1.dist-info/RECORD,,