mithwire 0.50.3__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.
- mithwire/__init__.py +32 -0
- mithwire/cdp/README.md +4 -0
- mithwire/cdp/__init__.py +6 -0
- mithwire/cdp/accessibility.py +668 -0
- mithwire/cdp/animation.py +494 -0
- mithwire/cdp/audits.py +1995 -0
- mithwire/cdp/autofill.py +292 -0
- mithwire/cdp/background_service.py +215 -0
- mithwire/cdp/bluetooth_emulation.py +626 -0
- mithwire/cdp/browser.py +821 -0
- mithwire/cdp/cache_storage.py +311 -0
- mithwire/cdp/cast.py +172 -0
- mithwire/cdp/console.py +107 -0
- mithwire/cdp/crash_report_context.py +55 -0
- mithwire/cdp/css.py +2750 -0
- mithwire/cdp/database.py +179 -0
- mithwire/cdp/debugger.py +1405 -0
- mithwire/cdp/device_access.py +141 -0
- mithwire/cdp/device_orientation.py +45 -0
- mithwire/cdp/dom.py +2257 -0
- mithwire/cdp/dom_debugger.py +321 -0
- mithwire/cdp/dom_snapshot.py +876 -0
- mithwire/cdp/dom_storage.py +222 -0
- mithwire/cdp/emulation.py +1779 -0
- mithwire/cdp/event_breakpoints.py +56 -0
- mithwire/cdp/extensions.py +238 -0
- mithwire/cdp/fed_cm.py +283 -0
- mithwire/cdp/fetch.py +507 -0
- mithwire/cdp/file_system.py +115 -0
- mithwire/cdp/headless_experimental.py +115 -0
- mithwire/cdp/heap_profiler.py +401 -0
- mithwire/cdp/indexed_db.py +528 -0
- mithwire/cdp/input_.py +701 -0
- mithwire/cdp/inspector.py +95 -0
- mithwire/cdp/io.py +101 -0
- mithwire/cdp/layer_tree.py +464 -0
- mithwire/cdp/log.py +190 -0
- mithwire/cdp/media.py +313 -0
- mithwire/cdp/memory.py +305 -0
- mithwire/cdp/network.py +5342 -0
- mithwire/cdp/overlay.py +1468 -0
- mithwire/cdp/page.py +3972 -0
- mithwire/cdp/performance.py +124 -0
- mithwire/cdp/performance_timeline.py +200 -0
- mithwire/cdp/preload.py +575 -0
- mithwire/cdp/profiler.py +420 -0
- mithwire/cdp/pwa.py +278 -0
- mithwire/cdp/py.typed +0 -0
- mithwire/cdp/runtime.py +1589 -0
- mithwire/cdp/schema.py +50 -0
- mithwire/cdp/security.py +518 -0
- mithwire/cdp/service_worker.py +401 -0
- mithwire/cdp/smart_card_emulation.py +891 -0
- mithwire/cdp/storage.py +1573 -0
- mithwire/cdp/system_info.py +327 -0
- mithwire/cdp/target.py +829 -0
- mithwire/cdp/tethering.py +65 -0
- mithwire/cdp/tracing.py +377 -0
- mithwire/cdp/util.py +18 -0
- mithwire/cdp/web_audio.py +606 -0
- mithwire/cdp/web_authn.py +598 -0
- mithwire/cdp/web_mcp.py +293 -0
- mithwire/core/_contradict.py +142 -0
- mithwire/core/browser.py +923 -0
- mithwire/core/cf_templates/cf_dark_checkbox.png +0 -0
- mithwire/core/cf_templates/cf_light_checkbox.png +0 -0
- mithwire/core/config.py +323 -0
- mithwire/core/connection.py +564 -0
- mithwire/core/element.py +1205 -0
- mithwire/core/tab.py +2202 -0
- mithwire/core/util.py +5063 -0
- mithwire-0.50.3.dist-info/METADATA +1049 -0
- mithwire-0.50.3.dist-info/RECORD +76 -0
- mithwire-0.50.3.dist-info/WHEEL +5 -0
- mithwire-0.50.3.dist-info/licenses/LICENSE.txt +619 -0
- mithwire-0.50.3.dist-info/top_level.txt +1 -0
mithwire/core/browser.py
ADDED
|
@@ -0,0 +1,923 @@
|
|
|
1
|
+
# Copyright 2024 by UltrafunkAmsterdam (https://github.com/UltrafunkAmsterdam)
|
|
2
|
+
# All rights reserved.
|
|
3
|
+
# This file is part of the mithwire package.
|
|
4
|
+
# and is released under the "GNU AFFERO GENERAL PUBLIC LICENSE".
|
|
5
|
+
# Please see the LICENSE.txt file that should have been included as part of this package.
|
|
6
|
+
|
|
7
|
+
from __future__ import annotations
|
|
8
|
+
|
|
9
|
+
import asyncio
|
|
10
|
+
import atexit
|
|
11
|
+
import http.cookiejar
|
|
12
|
+
import json
|
|
13
|
+
import logging
|
|
14
|
+
import os
|
|
15
|
+
import pathlib
|
|
16
|
+
import pickle
|
|
17
|
+
import urllib.parse
|
|
18
|
+
import urllib.request
|
|
19
|
+
import warnings
|
|
20
|
+
from collections import defaultdict
|
|
21
|
+
from typing import List, Tuple, Union
|
|
22
|
+
|
|
23
|
+
from .. import cdp
|
|
24
|
+
from . import tab, util
|
|
25
|
+
from ._contradict import ContraDict
|
|
26
|
+
from .config import Config, PathLike, is_posix
|
|
27
|
+
from .connection import Connection
|
|
28
|
+
|
|
29
|
+
logger = logging.getLogger(__name__)
|
|
30
|
+
|
|
31
|
+
|
|
32
|
+
class Browser(Connection):
|
|
33
|
+
"""
|
|
34
|
+
The Browser object is the "root" of the hierarchy and contains a reference
|
|
35
|
+
to the browser parent process.
|
|
36
|
+
there should usually be only 1 instance of this.
|
|
37
|
+
|
|
38
|
+
All opened tabs, extra browser screens and resources will not cause a new Browser process,
|
|
39
|
+
but rather create additional :class:`mithwire.Tab` objects.
|
|
40
|
+
|
|
41
|
+
So, besides starting your instance and first/additional tabs, you don't actively use it a lot under normal conditions.
|
|
42
|
+
|
|
43
|
+
Tab objects will represent and control
|
|
44
|
+
- tabs (as you know them)
|
|
45
|
+
- browser windows (new window)
|
|
46
|
+
- iframe
|
|
47
|
+
- background processes
|
|
48
|
+
|
|
49
|
+
note:
|
|
50
|
+
the Browser object is not instantiated by __init__ but using the asynchronous :meth:`mithwire.Browser.create` method.
|
|
51
|
+
|
|
52
|
+
note:
|
|
53
|
+
in Chromium based browsers, there is a parent process which keeps running all the time, even if
|
|
54
|
+
there are no visible browser windows. sometimes it's stubborn to close it, so make sure after using
|
|
55
|
+
this library, the browser is correctly and fully closed/exited/killed.
|
|
56
|
+
|
|
57
|
+
"""
|
|
58
|
+
|
|
59
|
+
_process: asyncio.subprocess.Process
|
|
60
|
+
_process_pid: int
|
|
61
|
+
_http: HTTPApi = None
|
|
62
|
+
_cookies: CookieJar = None
|
|
63
|
+
|
|
64
|
+
config: Config
|
|
65
|
+
|
|
66
|
+
@classmethod
|
|
67
|
+
async def create(
|
|
68
|
+
cls,
|
|
69
|
+
config: Config = None,
|
|
70
|
+
*,
|
|
71
|
+
user_data_dir: PathLike = None,
|
|
72
|
+
headless: bool = False,
|
|
73
|
+
browser_executable_path: PathLike = None,
|
|
74
|
+
browser_args: List[str] = None,
|
|
75
|
+
sandbox: bool = True,
|
|
76
|
+
host: str = None,
|
|
77
|
+
port: int = None,
|
|
78
|
+
**kwargs,
|
|
79
|
+
) -> Browser:
|
|
80
|
+
"""
|
|
81
|
+
entry point for creating an instance
|
|
82
|
+
"""
|
|
83
|
+
if not config:
|
|
84
|
+
config = Config(
|
|
85
|
+
user_data_dir=user_data_dir,
|
|
86
|
+
headless=headless,
|
|
87
|
+
browser_executable_path=browser_executable_path,
|
|
88
|
+
browser_args=browser_args or [],
|
|
89
|
+
sandbox=sandbox,
|
|
90
|
+
host=host,
|
|
91
|
+
port=port,
|
|
92
|
+
**kwargs,
|
|
93
|
+
)
|
|
94
|
+
instance = cls(config)
|
|
95
|
+
await instance.start()
|
|
96
|
+
return instance
|
|
97
|
+
|
|
98
|
+
def __init__(self, config: Config, **kwargs):
|
|
99
|
+
"""
|
|
100
|
+
constructor. to create a instance, use :py:meth:`Browser.create(...)`
|
|
101
|
+
|
|
102
|
+
:param config:
|
|
103
|
+
"""
|
|
104
|
+
|
|
105
|
+
try:
|
|
106
|
+
asyncio.get_running_loop()
|
|
107
|
+
except RuntimeError:
|
|
108
|
+
raise RuntimeError(
|
|
109
|
+
"{0} objects of this class are created using await {0}.create()".format(
|
|
110
|
+
self.__class__.__name__
|
|
111
|
+
)
|
|
112
|
+
)
|
|
113
|
+
# weakref.finalize(self, self._quit, self)
|
|
114
|
+
self.config = config
|
|
115
|
+
|
|
116
|
+
"""current targets (all types"""
|
|
117
|
+
self.info = None
|
|
118
|
+
self._target = None
|
|
119
|
+
self._process = None
|
|
120
|
+
self._process_pid = None
|
|
121
|
+
self._keep_user_data_dir = None
|
|
122
|
+
self._is_updating = asyncio.Event()
|
|
123
|
+
self.connection: Connection = None
|
|
124
|
+
super().__init__("", auto_attach=False)
|
|
125
|
+
logger.debug("Session object initialized: %s" % vars(self))
|
|
126
|
+
|
|
127
|
+
@property
|
|
128
|
+
def main_tab(self) -> tab.Tab:
|
|
129
|
+
"""returns the target which was launched with the browser"""
|
|
130
|
+
return next(filter(lambda x: x.target.type_ == "page", self.targets))
|
|
131
|
+
|
|
132
|
+
@property
|
|
133
|
+
def targets(self) -> List[Connection]:
|
|
134
|
+
return self._targets
|
|
135
|
+
|
|
136
|
+
@property
|
|
137
|
+
def tabs(self) -> List[tab.Connection]:
|
|
138
|
+
return [x for x in self._targets if x.target.type_ == "page"]
|
|
139
|
+
|
|
140
|
+
# @property
|
|
141
|
+
# def tabs(self) -> List[tab.Tab]:
|
|
142
|
+
# """returns the current targets which are of type "page"
|
|
143
|
+
# :return:
|
|
144
|
+
# """
|
|
145
|
+
# tabs = filter(lambda item: item.type_ == "page", self.targets)
|
|
146
|
+
# return [tab.Tab(self, x) for x in tabs]
|
|
147
|
+
# # return list(tabs)
|
|
148
|
+
|
|
149
|
+
@property
|
|
150
|
+
def cookies(self) -> CookieJar:
|
|
151
|
+
if not self._cookies:
|
|
152
|
+
self._cookies = CookieJar(self)
|
|
153
|
+
return self._cookies
|
|
154
|
+
|
|
155
|
+
@property
|
|
156
|
+
def stopped(self):
|
|
157
|
+
if self._process and self._process.returncode is None:
|
|
158
|
+
return False
|
|
159
|
+
return True
|
|
160
|
+
# return (self._process and self._process.returncode) or False
|
|
161
|
+
|
|
162
|
+
async def wait(self, time: Union[float, int] = 0.1):
|
|
163
|
+
"""wait for <time> seconds. important to use, especially in between page navigation
|
|
164
|
+
|
|
165
|
+
:param time:
|
|
166
|
+
:return:
|
|
167
|
+
"""
|
|
168
|
+
try:
|
|
169
|
+
await asyncio.wait(
|
|
170
|
+
[
|
|
171
|
+
asyncio.create_task(self.update_targets()),
|
|
172
|
+
asyncio.create_task(asyncio.sleep(time)),
|
|
173
|
+
],
|
|
174
|
+
return_when=asyncio.ALL_COMPLETED,
|
|
175
|
+
)
|
|
176
|
+
except asyncio.TimeoutError:
|
|
177
|
+
pass
|
|
178
|
+
|
|
179
|
+
sleep = wait
|
|
180
|
+
"""alias for wait"""
|
|
181
|
+
|
|
182
|
+
async def get(
|
|
183
|
+
self, url="chrome://welcome", new_tab: bool = False, new_window: bool = False
|
|
184
|
+
) -> tab.Tab:
|
|
185
|
+
"""top level get. utilizes the first tab to retrieve given url.
|
|
186
|
+
|
|
187
|
+
convenience function known from selenium.
|
|
188
|
+
this function handles waits/sleeps and detects when DOM events fired, so it's the safest
|
|
189
|
+
way of navigating.
|
|
190
|
+
|
|
191
|
+
:param url: the url to navigate to
|
|
192
|
+
:param new_tab: open new tab
|
|
193
|
+
:param new_window: open new window
|
|
194
|
+
:return: Page
|
|
195
|
+
"""
|
|
196
|
+
if new_tab or new_window:
|
|
197
|
+
# creat new target using the browser session
|
|
198
|
+
target_id = await self.send(
|
|
199
|
+
cdp.target.create_target(
|
|
200
|
+
url, new_window=new_window, enable_begin_frame_control=True
|
|
201
|
+
)
|
|
202
|
+
)
|
|
203
|
+
|
|
204
|
+
# connection = tab.Tab(target=target_id, parent=self, auto_attach=False)
|
|
205
|
+
# await connection.attach()
|
|
206
|
+
# self._targets.append(connection)
|
|
207
|
+
await self.update_targets()
|
|
208
|
+
connection = next(
|
|
209
|
+
filter(lambda x: x.target.target_id == target_id, self.targets)
|
|
210
|
+
)
|
|
211
|
+
await connection.attach()
|
|
212
|
+
|
|
213
|
+
else:
|
|
214
|
+
# first tab from browser.tabs
|
|
215
|
+
connection: tab.Tab = next(
|
|
216
|
+
filter(lambda item: item.target.type_ == "page", self.targets)
|
|
217
|
+
)
|
|
218
|
+
|
|
219
|
+
frame_id, loader_id, *_ = await connection.send(cdp.page.navigate(url))
|
|
220
|
+
await connection.attach()
|
|
221
|
+
await self.update_targets()
|
|
222
|
+
# await self
|
|
223
|
+
return connection
|
|
224
|
+
|
|
225
|
+
async def create_context(
|
|
226
|
+
self,
|
|
227
|
+
url: str = "chrome://welcome",
|
|
228
|
+
new_tab: bool = False,
|
|
229
|
+
new_window: bool = True,
|
|
230
|
+
dispose_on_detach: bool = True,
|
|
231
|
+
proxy_server: str = None,
|
|
232
|
+
proxy_bypass_list: List[str] = None,
|
|
233
|
+
origins_with_universal_network_access: List[str] = None,
|
|
234
|
+
proxy_ssl_context=None,
|
|
235
|
+
) -> tab.Tab:
|
|
236
|
+
"""
|
|
237
|
+
creates a new browser context - mostly useful if you want to use proxies for different browser instances
|
|
238
|
+
since chrome usually can only use 1 proxy per browser.
|
|
239
|
+
socks5 with authentication is supported by using a forwarder proxy, the
|
|
240
|
+
correct string to use socks proxy with username/password auth is socks://USERNAME:PASSWORD@SERVER:PORT
|
|
241
|
+
http/https proxies with authentication are also supported: http://USERNAME:PASSWORD@SERVER:PORT
|
|
242
|
+
|
|
243
|
+
dispose_on_detach – (EXPERIMENTAL) (Optional) If specified, disposes this context when debugging session disconnects.
|
|
244
|
+
proxy_server – (EXPERIMENTAL) (Optional) Proxy server, similar to the one passed to –proxy-server
|
|
245
|
+
proxy_bypass_list – (EXPERIMENTAL) (Optional) Proxy bypass list, similar to the one passed to –proxy-bypass-list
|
|
246
|
+
origins_with_universal_network_access – (EXPERIMENTAL) (Optional) An optional list of origins to grant unlimited cross-origin access to. Parts of the URL other than those constituting origin are ignored.
|
|
247
|
+
proxy_ssl_context – (Optional) Custom SSL context for HTTPS proxy connections. If None, a default context is used.
|
|
248
|
+
|
|
249
|
+
:param new_window:
|
|
250
|
+
:type new_window:
|
|
251
|
+
:param new_tab:
|
|
252
|
+
:type new_tab:
|
|
253
|
+
:param url:
|
|
254
|
+
:type url:
|
|
255
|
+
:param dispose_on_detach:
|
|
256
|
+
:type dispose_on_detach:
|
|
257
|
+
:param proxy_server:
|
|
258
|
+
:type proxy_server:
|
|
259
|
+
:param proxy_bypass_list:
|
|
260
|
+
:type proxy_bypass_list:
|
|
261
|
+
:param origins_with_universal_network_access:
|
|
262
|
+
:type origins_with_universal_network_access:
|
|
263
|
+
:param proxy_ssl_context:
|
|
264
|
+
:type proxy_ssl_context: ssl.SSLContext
|
|
265
|
+
:return:
|
|
266
|
+
:rtype:
|
|
267
|
+
"""
|
|
268
|
+
|
|
269
|
+
if proxy_server:
|
|
270
|
+
fw = util.ProxyForwarder(
|
|
271
|
+
proxy_server=proxy_server, ssl_context=proxy_ssl_context
|
|
272
|
+
)
|
|
273
|
+
proxy_server = fw.proxy_server
|
|
274
|
+
|
|
275
|
+
ctx: cdp.browser.BrowserContextID = await self.send(
|
|
276
|
+
cdp.target.create_browser_context(
|
|
277
|
+
dispose_on_detach=dispose_on_detach,
|
|
278
|
+
proxy_server=proxy_server,
|
|
279
|
+
proxy_bypass_list=proxy_bypass_list,
|
|
280
|
+
origins_with_universal_network_access=origins_with_universal_network_access,
|
|
281
|
+
)
|
|
282
|
+
)
|
|
283
|
+
target_id: cdp.target.TargetID = await self.send(
|
|
284
|
+
cdp.target.create_target(
|
|
285
|
+
url, browser_context_id=ctx, new_window=new_window, for_tab=new_tab
|
|
286
|
+
)
|
|
287
|
+
)
|
|
288
|
+
await self.sleep(0.5)
|
|
289
|
+
connection: tab.Tab = next(
|
|
290
|
+
filter(
|
|
291
|
+
lambda item: item.target.type_ == "page"
|
|
292
|
+
and item.target.target_id == target_id,
|
|
293
|
+
self.targets,
|
|
294
|
+
)
|
|
295
|
+
)
|
|
296
|
+
return connection
|
|
297
|
+
|
|
298
|
+
async def start(self=None) -> Browser:
|
|
299
|
+
"""launches the actual browser"""
|
|
300
|
+
|
|
301
|
+
if not self:
|
|
302
|
+
raise RuntimeError(
|
|
303
|
+
"use ``await Browser.create()`` to create a new instance"
|
|
304
|
+
)
|
|
305
|
+
|
|
306
|
+
if self._process or self._process_pid:
|
|
307
|
+
if self._process.returncode is not None:
|
|
308
|
+
return await self.create(config=self.config)
|
|
309
|
+
raise RuntimeError("ignored! this call has no effect when already running.")
|
|
310
|
+
|
|
311
|
+
# self.config.update(kwargs)
|
|
312
|
+
connect_existing = False
|
|
313
|
+
if self.config.host is not None and self.config.port is not None:
|
|
314
|
+
connect_existing = True
|
|
315
|
+
else:
|
|
316
|
+
self.config.host = "127.0.0.1"
|
|
317
|
+
self.config.port = util.free_port()
|
|
318
|
+
|
|
319
|
+
if not connect_existing:
|
|
320
|
+
logger.debug(
|
|
321
|
+
"BROWSER EXECUTABLE PATH: %s", self.config.browser_executable_path
|
|
322
|
+
)
|
|
323
|
+
if not pathlib.Path(self.config.browser_executable_path).exists():
|
|
324
|
+
raise FileNotFoundError(
|
|
325
|
+
(
|
|
326
|
+
"""
|
|
327
|
+
---------------------
|
|
328
|
+
Could not determine browser executable.
|
|
329
|
+
---------------------
|
|
330
|
+
Make sure your browser is installed in the default location (path).
|
|
331
|
+
If you are sure about the browser executable, you can specify it using
|
|
332
|
+
the `browser_executable_path='{}` parameter."""
|
|
333
|
+
).format(
|
|
334
|
+
"/path/to/browser/executable"
|
|
335
|
+
if is_posix
|
|
336
|
+
else "c:/path/to/your/browser.exe"
|
|
337
|
+
)
|
|
338
|
+
)
|
|
339
|
+
|
|
340
|
+
if getattr(self.config, "_extensions", None): # noqa
|
|
341
|
+
self.config.add_argument(
|
|
342
|
+
"--load-extension=%s"
|
|
343
|
+
% ",".join(str(_) for _ in self.config._extensions)
|
|
344
|
+
) # noqa
|
|
345
|
+
|
|
346
|
+
exe = self.config.browser_executable_path
|
|
347
|
+
params = self.config()
|
|
348
|
+
|
|
349
|
+
logger.info(
|
|
350
|
+
"starting\n\texecutable :%s\n\narguments:\n%s", exe, "\n\t".join(params)
|
|
351
|
+
)
|
|
352
|
+
if not connect_existing:
|
|
353
|
+
self._process: asyncio.subprocess.Process = (
|
|
354
|
+
await asyncio.create_subprocess_exec(
|
|
355
|
+
# self.config.browser_executable_path,
|
|
356
|
+
# *cmdparams,
|
|
357
|
+
exe,
|
|
358
|
+
*params,
|
|
359
|
+
stdin=asyncio.subprocess.PIPE,
|
|
360
|
+
stdout=asyncio.subprocess.PIPE,
|
|
361
|
+
stderr=asyncio.subprocess.PIPE,
|
|
362
|
+
close_fds=is_posix,
|
|
363
|
+
)
|
|
364
|
+
)
|
|
365
|
+
self._process_pid = self._process.pid
|
|
366
|
+
|
|
367
|
+
self._http = HTTPApi((self.config.host, self.config.port))
|
|
368
|
+
util.get_registered_instances().add(self)
|
|
369
|
+
# Connect to the freshly-spawned Chrome's DevTools endpoint. The old
|
|
370
|
+
# loop budgeted only ~2.75s (5 attempts x 0.5s), which is tight: a
|
|
371
|
+
# cold Chrome typically binds its DevTools port in ~1-2s, leaving
|
|
372
|
+
# <1s of headroom. Any system hiccup (Spotlight scan, antivirus
|
|
373
|
+
# scan-on-execute, contended host, Chrome auto-update install) eats
|
|
374
|
+
# the margin and surfaces as a spurious "Failed to connect to
|
|
375
|
+
# browser" -- even though Chrome is fine and would have come up a
|
|
376
|
+
# moment later.
|
|
377
|
+
#
|
|
378
|
+
# Strategy: exponential backoff (50ms doubling, capped at 1s)
|
|
379
|
+
# against a wall-clock deadline of ~10s. Warm starts converge in a
|
|
380
|
+
# few probes; cold/contended starts get a real chance to finish;
|
|
381
|
+
# genuinely broken launches still fail in bounded time.
|
|
382
|
+
deadline = asyncio.get_running_loop().time() + 10.0
|
|
383
|
+
delay = 0.05
|
|
384
|
+
last_exc: BaseException | None = None
|
|
385
|
+
while True:
|
|
386
|
+
try:
|
|
387
|
+
self.info = ContraDict(await self._http.get("version"), silent=True)
|
|
388
|
+
break
|
|
389
|
+
except Exception as exc:
|
|
390
|
+
last_exc = exc
|
|
391
|
+
if asyncio.get_running_loop().time() + delay >= deadline:
|
|
392
|
+
logger.debug("could not start", exc_info=True)
|
|
393
|
+
break
|
|
394
|
+
await asyncio.sleep(delay)
|
|
395
|
+
delay = min(delay * 2, 1.0)
|
|
396
|
+
|
|
397
|
+
if not self.info:
|
|
398
|
+
raise Exception(
|
|
399
|
+
(
|
|
400
|
+
"""
|
|
401
|
+
---------------------
|
|
402
|
+
Failed to connect to browser
|
|
403
|
+
---------------------
|
|
404
|
+
One of the causes could be when you are running as root.
|
|
405
|
+
In that case you need to pass no_sandbox=True
|
|
406
|
+
"""
|
|
407
|
+
)
|
|
408
|
+
)
|
|
409
|
+
|
|
410
|
+
self.websocket_url = self.info.webSocketDebuggerUrl
|
|
411
|
+
await self.attach()
|
|
412
|
+
await self.update_targets()
|
|
413
|
+
# await self
|
|
414
|
+
|
|
415
|
+
async def grant_all_permissions(self):
|
|
416
|
+
"""
|
|
417
|
+
grant permissions for:
|
|
418
|
+
accessibilityEvents
|
|
419
|
+
audioCapture
|
|
420
|
+
backgroundSync
|
|
421
|
+
backgroundFetch
|
|
422
|
+
clipboardReadWrite
|
|
423
|
+
clipboardSanitizedWrite
|
|
424
|
+
displayCapture
|
|
425
|
+
durableStorage
|
|
426
|
+
geolocation
|
|
427
|
+
idleDetection
|
|
428
|
+
localFonts
|
|
429
|
+
midi
|
|
430
|
+
midiSysex
|
|
431
|
+
nfc
|
|
432
|
+
notifications
|
|
433
|
+
paymentHandler
|
|
434
|
+
periodicBackgroundSync
|
|
435
|
+
protectedMediaIdentifier
|
|
436
|
+
sensors
|
|
437
|
+
storageAccess
|
|
438
|
+
topLevelStorageAccess
|
|
439
|
+
videoCapture
|
|
440
|
+
videoCapturePanTiltZoom
|
|
441
|
+
wakeLockScreen
|
|
442
|
+
wakeLockSystem
|
|
443
|
+
windowManagement
|
|
444
|
+
"""
|
|
445
|
+
permissions = list(cdp.browser.PermissionType)
|
|
446
|
+
permissions.remove(cdp.browser.PermissionType.FLASH)
|
|
447
|
+
permissions.remove(cdp.browser.PermissionType.CAPTURED_SURFACE_CONTROL)
|
|
448
|
+
await self.send(cdp.browser.grant_permissions(permissions))
|
|
449
|
+
|
|
450
|
+
async def tile_windows(self, windows=None, max_columns: int = 0):
|
|
451
|
+
import math
|
|
452
|
+
|
|
453
|
+
import mss
|
|
454
|
+
|
|
455
|
+
m = mss.mss()
|
|
456
|
+
screen, screen_width, screen_height = 3 * (None,)
|
|
457
|
+
if m.monitors and len(m.monitors) >= 1:
|
|
458
|
+
screen = m.monitors[0]
|
|
459
|
+
screen_width = screen["width"]
|
|
460
|
+
screen_height = screen["height"]
|
|
461
|
+
if not screen or not screen_width or not screen_height:
|
|
462
|
+
warnings.warn("no monitors detected")
|
|
463
|
+
return
|
|
464
|
+
await self
|
|
465
|
+
distinct_windows = defaultdict(list)
|
|
466
|
+
|
|
467
|
+
if windows:
|
|
468
|
+
tabs = windows
|
|
469
|
+
else:
|
|
470
|
+
tabs = self.tabs
|
|
471
|
+
for tab in tabs:
|
|
472
|
+
window_id, bounds = await tab.get_window()
|
|
473
|
+
distinct_windows[window_id].append(tab)
|
|
474
|
+
|
|
475
|
+
num_windows = len(distinct_windows)
|
|
476
|
+
req_cols = max_columns or int(num_windows * (19 / 6))
|
|
477
|
+
req_rows = int(num_windows / req_cols)
|
|
478
|
+
|
|
479
|
+
while req_cols * req_rows < num_windows:
|
|
480
|
+
req_rows += 1
|
|
481
|
+
|
|
482
|
+
box_w = math.floor((screen_width / req_cols) - 1)
|
|
483
|
+
box_h = math.floor(screen_height / req_rows)
|
|
484
|
+
|
|
485
|
+
distinct_windows_iter = iter(distinct_windows.values())
|
|
486
|
+
grid = []
|
|
487
|
+
for x in range(req_cols):
|
|
488
|
+
for y in range(req_rows):
|
|
489
|
+
num = x + y
|
|
490
|
+
try:
|
|
491
|
+
tabs = next(distinct_windows_iter)
|
|
492
|
+
except StopIteration:
|
|
493
|
+
continue
|
|
494
|
+
if not tabs:
|
|
495
|
+
continue
|
|
496
|
+
tab = tabs[0]
|
|
497
|
+
|
|
498
|
+
try:
|
|
499
|
+
pos = [x * box_w, y * box_h, box_w, box_h]
|
|
500
|
+
grid.append(pos)
|
|
501
|
+
await tab.set_window_size(*pos)
|
|
502
|
+
except Exception:
|
|
503
|
+
logger.info(
|
|
504
|
+
"could not set window size. exception => ", exc_info=True
|
|
505
|
+
)
|
|
506
|
+
continue
|
|
507
|
+
return grid
|
|
508
|
+
|
|
509
|
+
async def _get_targets(self) -> List[cdp.target.TargetInfo]:
|
|
510
|
+
info = await self.send(cdp.target.get_targets())
|
|
511
|
+
|
|
512
|
+
return info
|
|
513
|
+
|
|
514
|
+
async def update_targets(self):
|
|
515
|
+
|
|
516
|
+
targets = await self.send(cdp.target.get_targets())
|
|
517
|
+
#
|
|
518
|
+
# current_tabs_targets = [t.target for t in self.children]
|
|
519
|
+
#
|
|
520
|
+
for t in targets:
|
|
521
|
+
for ctab in self._targets:
|
|
522
|
+
if ctab.target.target_id == t.target_id:
|
|
523
|
+
ctab.target = t
|
|
524
|
+
break
|
|
525
|
+
else:
|
|
526
|
+
_t = tab.Tab(target=t, parent=self, auto_attach=False)
|
|
527
|
+
self._targets.append(_t)
|
|
528
|
+
|
|
529
|
+
for ctab in self._targets.copy():
|
|
530
|
+
if ctab.target not in targets:
|
|
531
|
+
self._targets.remove(ctab)
|
|
532
|
+
|
|
533
|
+
await asyncio.sleep(0)
|
|
534
|
+
|
|
535
|
+
def __iter__(self):
|
|
536
|
+
self._i = self.tabs.index(self.main_tab)
|
|
537
|
+
return self
|
|
538
|
+
|
|
539
|
+
def __getitem__(
|
|
540
|
+
self, item: Union[str, int, slice]
|
|
541
|
+
) -> Union[tab.Tab, List[tab.Tab], None]:
|
|
542
|
+
"""
|
|
543
|
+
allows to get py:obj:`tab.Tab` instances by using browser[0], browser[1], etc.
|
|
544
|
+
a string is also allowed. it will then return the first tab where the py:obj:`cdp.target.TargetInfo` object
|
|
545
|
+
(as json string) contains the given key, or the first tab in case no matches are found. eg:
|
|
546
|
+
`browser["google"]` gives the first tab which has "google" in it's serialized target object.
|
|
547
|
+
|
|
548
|
+
:param item:
|
|
549
|
+
:type item:
|
|
550
|
+
:return:
|
|
551
|
+
:rtype: tab.Tab
|
|
552
|
+
"""
|
|
553
|
+
if isinstance(item, int):
|
|
554
|
+
return self.tabs[item]
|
|
555
|
+
elif isinstance(item, slice):
|
|
556
|
+
tabs: List[tab.Tab] = []
|
|
557
|
+
sta, sto, ste = item.start, item.stop, item.step
|
|
558
|
+
if not ste:
|
|
559
|
+
ste = 1
|
|
560
|
+
if not sto:
|
|
561
|
+
sto = len(self.tabs) - 1
|
|
562
|
+
if not sta:
|
|
563
|
+
sta = 0
|
|
564
|
+
for x in range(sta, sto, ste):
|
|
565
|
+
try:
|
|
566
|
+
tabs.append(self.tabs[x])
|
|
567
|
+
except IndexError:
|
|
568
|
+
pass
|
|
569
|
+
return tabs
|
|
570
|
+
elif isinstance(item, tuple):
|
|
571
|
+
r = range(*item)
|
|
572
|
+
tabs: List[tab.Tab] = []
|
|
573
|
+
for i in r:
|
|
574
|
+
try:
|
|
575
|
+
tabs.append(self.tabs[i])
|
|
576
|
+
except IndexError:
|
|
577
|
+
pass
|
|
578
|
+
return tabs
|
|
579
|
+
elif isinstance(item, str):
|
|
580
|
+
for t in self.tabs:
|
|
581
|
+
if item.lower() in str(t.target.to_json()).lower():
|
|
582
|
+
return t
|
|
583
|
+
else:
|
|
584
|
+
return self.tabs[0]
|
|
585
|
+
|
|
586
|
+
def __reversed__(self):
|
|
587
|
+
return reversed(list(self.tabs))
|
|
588
|
+
|
|
589
|
+
def __next__(self) -> tab.Tab | tab.Connection | None:
|
|
590
|
+
try:
|
|
591
|
+
return self.tabs[self._i]
|
|
592
|
+
except IndexError:
|
|
593
|
+
del self._i
|
|
594
|
+
raise StopIteration
|
|
595
|
+
except AttributeError:
|
|
596
|
+
del self._i
|
|
597
|
+
raise StopIteration
|
|
598
|
+
finally:
|
|
599
|
+
if hasattr(self, "_i"):
|
|
600
|
+
if self._i != len(self.tabs):
|
|
601
|
+
self._i += 1
|
|
602
|
+
else:
|
|
603
|
+
del self._i
|
|
604
|
+
|
|
605
|
+
def stop(self):
|
|
606
|
+
try:
|
|
607
|
+
# asyncio.get_running_loop().create_task(self.send(cdp.browser.close()))
|
|
608
|
+
|
|
609
|
+
asyncio.get_event_loop().create_task(self.aclose())
|
|
610
|
+
logger.debug("closed the connection using get_event_loop().create_task()")
|
|
611
|
+
except RuntimeError:
|
|
612
|
+
if self.connection:
|
|
613
|
+
try:
|
|
614
|
+
# asyncio.run(self.send(cdp.browser.close()))
|
|
615
|
+
asyncio.run(self.aclose())
|
|
616
|
+
logger.debug("closed the connection using asyncio.run()")
|
|
617
|
+
except Exception:
|
|
618
|
+
pass
|
|
619
|
+
|
|
620
|
+
for _ in range(3):
|
|
621
|
+
try:
|
|
622
|
+
self._process.terminate()
|
|
623
|
+
logger.info(
|
|
624
|
+
"terminated browser with pid %d successfully" % self._process.pid
|
|
625
|
+
)
|
|
626
|
+
break
|
|
627
|
+
except (Exception,):
|
|
628
|
+
try:
|
|
629
|
+
self._process.kill()
|
|
630
|
+
logger.info(
|
|
631
|
+
"killed browser with pid %d successfully" % self._process.pid
|
|
632
|
+
)
|
|
633
|
+
break
|
|
634
|
+
except (Exception,):
|
|
635
|
+
try:
|
|
636
|
+
if hasattr(self, "browser_process_pid"):
|
|
637
|
+
os.kill(self._process_pid, 15)
|
|
638
|
+
logger.info(
|
|
639
|
+
"killed browser with pid %d using signal 15 successfully"
|
|
640
|
+
% self._process.pid
|
|
641
|
+
)
|
|
642
|
+
break
|
|
643
|
+
except (TypeError,):
|
|
644
|
+
logger.info("typerror", exc_info=True)
|
|
645
|
+
pass
|
|
646
|
+
except (PermissionError,):
|
|
647
|
+
logger.info(
|
|
648
|
+
"browser already stopped, or no permission to kill. skip"
|
|
649
|
+
)
|
|
650
|
+
pass
|
|
651
|
+
except (ProcessLookupError,):
|
|
652
|
+
logger.info("process lookup failure")
|
|
653
|
+
pass
|
|
654
|
+
except (Exception,):
|
|
655
|
+
raise
|
|
656
|
+
self._process = None
|
|
657
|
+
self._process_pid = None
|
|
658
|
+
|
|
659
|
+
def __await__(self):
|
|
660
|
+
# return ( asyncio.sleep(0)).__await__()
|
|
661
|
+
return self.update_targets().__await__()
|
|
662
|
+
|
|
663
|
+
def __del__(self):
|
|
664
|
+
pass
|
|
665
|
+
|
|
666
|
+
|
|
667
|
+
class CookieJar:
|
|
668
|
+
def __init__(self, browser: Browser):
|
|
669
|
+
self._browser = browser
|
|
670
|
+
# self._connection = connection
|
|
671
|
+
|
|
672
|
+
async def get_all(
|
|
673
|
+
self, requests_cookie_format: bool = False
|
|
674
|
+
) -> List[Union[cdp.network.Cookie, "http.cookiejar.Cookie"]]:
|
|
675
|
+
"""
|
|
676
|
+
get all cookies
|
|
677
|
+
|
|
678
|
+
:param requests_cookie_format: when True, returns python http.cookiejar.Cookie objects, compatible with requests library and many others.
|
|
679
|
+
:type requests_cookie_format: bool
|
|
680
|
+
:return:
|
|
681
|
+
:rtype:
|
|
682
|
+
|
|
683
|
+
"""
|
|
684
|
+
connection = None
|
|
685
|
+
for tab in self._browser:
|
|
686
|
+
if tab.target:
|
|
687
|
+
connection = tab
|
|
688
|
+
break
|
|
689
|
+
|
|
690
|
+
else:
|
|
691
|
+
connection = self._browser
|
|
692
|
+
cookies = await connection.send(cdp.storage.get_cookies())
|
|
693
|
+
if requests_cookie_format:
|
|
694
|
+
import requests.cookies
|
|
695
|
+
|
|
696
|
+
return [
|
|
697
|
+
requests.cookies.create_cookie(
|
|
698
|
+
name=c.name,
|
|
699
|
+
value=c.value,
|
|
700
|
+
domain=c.domain,
|
|
701
|
+
path=c.path,
|
|
702
|
+
expires=c.expires,
|
|
703
|
+
secure=c.secure,
|
|
704
|
+
)
|
|
705
|
+
for c in cookies
|
|
706
|
+
]
|
|
707
|
+
return cookies
|
|
708
|
+
|
|
709
|
+
async def set_all(self, cookies: List[cdp.network.CookieParam]):
|
|
710
|
+
"""
|
|
711
|
+
set cookies
|
|
712
|
+
|
|
713
|
+
:param cookies: list of cookies
|
|
714
|
+
:type cookies:
|
|
715
|
+
:return:
|
|
716
|
+
:rtype:
|
|
717
|
+
"""
|
|
718
|
+
connection = None
|
|
719
|
+
for tab in self._browser:
|
|
720
|
+
if tab.target:
|
|
721
|
+
connection = tab
|
|
722
|
+
break
|
|
723
|
+
|
|
724
|
+
else:
|
|
725
|
+
connection = self._browser
|
|
726
|
+
|
|
727
|
+
await connection.send(cdp.storage.set_cookies(cookies))
|
|
728
|
+
|
|
729
|
+
async def save(self, file: PathLike = ".session.dat", pattern: str = ".*"):
|
|
730
|
+
"""
|
|
731
|
+
save all cookies (or a subset, controlled by `pattern`) to a file to be restored later
|
|
732
|
+
|
|
733
|
+
:param file:
|
|
734
|
+
:type file:
|
|
735
|
+
:param pattern: regex style pattern string.
|
|
736
|
+
any cookie that has a domain, key or value field which matches the pattern will be included.
|
|
737
|
+
default (param not specified) = ".*" (all)
|
|
738
|
+
|
|
739
|
+
eg: the pattern "(cf|.com|nowsecure)" will include those cookies which:
|
|
740
|
+
- have a string "cf" (cloudflare)
|
|
741
|
+
- have ".com" in them, in either domain, key or value field.
|
|
742
|
+
- contain "nowsecure"
|
|
743
|
+
:type pattern: str
|
|
744
|
+
:return:
|
|
745
|
+
:rtype:
|
|
746
|
+
"""
|
|
747
|
+
import re
|
|
748
|
+
|
|
749
|
+
pattern = re.compile(pattern)
|
|
750
|
+
save_path = pathlib.Path(file).resolve()
|
|
751
|
+
connection = None
|
|
752
|
+
for tab in self._browser:
|
|
753
|
+
if tab.target:
|
|
754
|
+
connection = tab
|
|
755
|
+
break
|
|
756
|
+
else:
|
|
757
|
+
connection = self._browser
|
|
758
|
+
|
|
759
|
+
cookies = await self.get_all(requests_cookie_format=False)
|
|
760
|
+
included_cookies = []
|
|
761
|
+
for cookie in cookies:
|
|
762
|
+
for match in pattern.finditer(str(cookie.__dict__)):
|
|
763
|
+
logger.debug(
|
|
764
|
+
"saved cookie for matching pattern '%s' => (%s: %s)",
|
|
765
|
+
pattern.pattern,
|
|
766
|
+
cookie.name,
|
|
767
|
+
cookie.value,
|
|
768
|
+
)
|
|
769
|
+
included_cookies.append(cookie)
|
|
770
|
+
break
|
|
771
|
+
pickle.dump(cookies, save_path.open("w+b"))
|
|
772
|
+
|
|
773
|
+
async def load(self, file: PathLike = ".session.dat", pattern: str = ".*"):
|
|
774
|
+
"""
|
|
775
|
+
load all cookies (or a subset, controlled by `pattern`) from a file created by :py:meth:`~save_cookies`.
|
|
776
|
+
|
|
777
|
+
:param file:
|
|
778
|
+
:type file:
|
|
779
|
+
:param pattern: regex style pattern string.
|
|
780
|
+
any cookie that has a domain, key or value field which matches the pattern will be included.
|
|
781
|
+
default (param not specified) = ".*" (all)
|
|
782
|
+
|
|
783
|
+
eg: the pattern "(cf|.com|nowsecure)" will include those cookies which:
|
|
784
|
+
- have a string "cf" (cloudflare)
|
|
785
|
+
- have ".com" in them, in either domain, key or value field.
|
|
786
|
+
- contain "nowsecure"
|
|
787
|
+
:type pattern: str
|
|
788
|
+
:return:
|
|
789
|
+
:rtype:
|
|
790
|
+
"""
|
|
791
|
+
import re
|
|
792
|
+
|
|
793
|
+
pattern = re.compile(pattern)
|
|
794
|
+
save_path = pathlib.Path(file).resolve()
|
|
795
|
+
cookies = pickle.load(save_path.open("r+b"))
|
|
796
|
+
included_cookies = []
|
|
797
|
+
connection = None
|
|
798
|
+
for tab in self._browser:
|
|
799
|
+
if tab.target:
|
|
800
|
+
connection = tab
|
|
801
|
+
break
|
|
802
|
+
|
|
803
|
+
else:
|
|
804
|
+
connection = self._browser
|
|
805
|
+
for cookie in cookies:
|
|
806
|
+
for match in pattern.finditer(str(cookie.__dict__)):
|
|
807
|
+
included_cookies.append(cookie)
|
|
808
|
+
logger.debug(
|
|
809
|
+
"loaded cookie for matching pattern '%s' => (%s: %s)",
|
|
810
|
+
pattern.pattern,
|
|
811
|
+
cookie.name,
|
|
812
|
+
cookie.value,
|
|
813
|
+
)
|
|
814
|
+
break
|
|
815
|
+
await connection.send(cdp.storage.set_cookies(included_cookies))
|
|
816
|
+
|
|
817
|
+
async def clear(self):
|
|
818
|
+
"""
|
|
819
|
+
clear current cookies
|
|
820
|
+
|
|
821
|
+
note: this includes all open tabs/windows for this browser
|
|
822
|
+
|
|
823
|
+
:return:
|
|
824
|
+
:rtype:
|
|
825
|
+
"""
|
|
826
|
+
connection = None
|
|
827
|
+
for tab in self._browser:
|
|
828
|
+
if tab.target:
|
|
829
|
+
# continue
|
|
830
|
+
connection = tab
|
|
831
|
+
break
|
|
832
|
+
|
|
833
|
+
else:
|
|
834
|
+
connection = self._browser
|
|
835
|
+
|
|
836
|
+
await connection.send(cdp.storage.clear_cookies())
|
|
837
|
+
|
|
838
|
+
|
|
839
|
+
class HTTPApi:
|
|
840
|
+
def __init__(self, addr: Tuple[str, int]):
|
|
841
|
+
self.host, self.port = addr
|
|
842
|
+
self.api = "http://%s:%d" % (self.host, self.port)
|
|
843
|
+
|
|
844
|
+
@classmethod
|
|
845
|
+
def from_target(cls, target: "Target"):
|
|
846
|
+
ws_url = urllib.parse.urlparse(target.websocket_url)
|
|
847
|
+
inst = cls((ws_url.hostname, ws_url.port))
|
|
848
|
+
return inst
|
|
849
|
+
|
|
850
|
+
async def get(self, endpoint: str):
|
|
851
|
+
return await self._request(endpoint)
|
|
852
|
+
|
|
853
|
+
async def post(self, endpoint, data):
|
|
854
|
+
return await self._request(endpoint, data)
|
|
855
|
+
|
|
856
|
+
async def _request(self, endpoint, method: str = "get", data: dict = None):
|
|
857
|
+
url = urllib.parse.urljoin(
|
|
858
|
+
self.api, f"json/{endpoint}" if endpoint else "/json"
|
|
859
|
+
)
|
|
860
|
+
if data and method.lower() == "get":
|
|
861
|
+
raise ValueError("get requests cannot contain data")
|
|
862
|
+
if not url:
|
|
863
|
+
url = self.api + endpoint
|
|
864
|
+
request = urllib.request.Request(url)
|
|
865
|
+
request.method = method
|
|
866
|
+
request.data = None
|
|
867
|
+
if data:
|
|
868
|
+
request.data = json.dumps(data).encode("utf-8")
|
|
869
|
+
|
|
870
|
+
response = await asyncio.get_running_loop().run_in_executor(
|
|
871
|
+
None, lambda: urllib.request.urlopen(request, timeout=10)
|
|
872
|
+
)
|
|
873
|
+
return json.loads(response.read())
|
|
874
|
+
|
|
875
|
+
|
|
876
|
+
class BrowserContext:
|
|
877
|
+
def __init__(
|
|
878
|
+
self,
|
|
879
|
+
config: Config = None,
|
|
880
|
+
*,
|
|
881
|
+
user_data_dir: PathLike = None,
|
|
882
|
+
headless: bool = False,
|
|
883
|
+
browser_executable_path: PathLike = None,
|
|
884
|
+
browser_args: List[str] = None,
|
|
885
|
+
sandbox: bool = True,
|
|
886
|
+
host: str = None,
|
|
887
|
+
port: int = None,
|
|
888
|
+
keep_open: bool = False,
|
|
889
|
+
**kwargs,
|
|
890
|
+
):
|
|
891
|
+
self._config = config
|
|
892
|
+
self._user_data_dir = user_data_dir
|
|
893
|
+
self._headless = headless
|
|
894
|
+
self._browser_executable_path = browser_executable_path
|
|
895
|
+
self._browser_args = browser_args
|
|
896
|
+
self._sandbox = sandbox
|
|
897
|
+
self._host = host
|
|
898
|
+
self._port = port
|
|
899
|
+
self._kwargs = kwargs
|
|
900
|
+
self._instance: Browser = None
|
|
901
|
+
self._keep_open = keep_open
|
|
902
|
+
|
|
903
|
+
async def __aenter__(self):
|
|
904
|
+
if not self._instance:
|
|
905
|
+
self._instance = await Browser.create(
|
|
906
|
+
self._config,
|
|
907
|
+
user_data_dir=self._user_data_dir,
|
|
908
|
+
headless=self._headless,
|
|
909
|
+
browser_executable_path=self._browser_executable_path,
|
|
910
|
+
browser_args=self._browser_args,
|
|
911
|
+
sandbox=self._sandbox,
|
|
912
|
+
host=self._host,
|
|
913
|
+
port=self._port,
|
|
914
|
+
**self._kwargs,
|
|
915
|
+
)
|
|
916
|
+
return self._instance
|
|
917
|
+
|
|
918
|
+
async def __aexit__(self, exc_type, exc_val, exc_tb):
|
|
919
|
+
if not self._keep_open:
|
|
920
|
+
await util.deconstruct_browser(self._instance)
|
|
921
|
+
|
|
922
|
+
|
|
923
|
+
atexit.register(util.deconstruct_browser)
|