Source code for nodriver.core.browser

# Copyright 2024 by UltrafunkAmsterdam (https://github.com/UltrafunkAmsterdam)
# All rights reserved.
# This file is part of the nodriver package.
# and is released under the "GNU AFFERO GENERAL PUBLIC LICENSE".
# Please see the LICENSE.txt file that should have been included as part of this package.

from __future__ import annotations

import asyncio
import atexit
import json
import logging
import os
import pathlib
import pickle
import urllib.parse
import urllib.request
import warnings
from collections import defaultdict
from typing import List, Tuple, Union

from .. import cdp
from . import tab, util
from ._contradict import ContraDict
from .config import Config, PathLike, is_posix
from .connection import Connection

logger = logging.getLogger(__name__)


[docs] class Browser: """ The Browser object is the "root" of the hierarchy and contains a reference to the browser parent process. there should usually be only 1 instance of this. All opened tabs, extra browser screens and resources will not cause a new Browser process, but rather create additional :class:`nodriver.Tab` objects. So, besides starting your instance and first/additional tabs, you don't actively use it a lot under normal conditions. Tab objects will represent and control - tabs (as you know them) - browser windows (new window) - iframe - background processes note: the Browser object is not instantiated by __init__ but using the asynchronous :meth:`nodriver.Browser.create` method. note: in Chromium based browsers, there is a parent process which keeps running all the time, even if there are no visible browser windows. sometimes it's stubborn to close it, so make sure after using this library, the browser is correctly and fully closed/exited/killed. """ _process: asyncio.subprocess.Process _process_pid: int _http: HTTPApi = None _cookies: CookieJar = None config: Config connection: Connection
[docs] @classmethod async def create( cls, config: Config = None, *, user_data_dir: PathLike = None, headless: bool = False, browser_executable_path: PathLike = None, browser_args: List[str] = None, sandbox: bool = True, host: str = None, port: int = None, **kwargs, ) -> Browser: """ entry point for creating an instance """ if not config: config = Config( user_data_dir=user_data_dir, headless=headless, browser_executable_path=browser_executable_path, browser_args=browser_args or [], sandbox=sandbox, host=host, port=port, **kwargs, ) instance = cls(config) await instance.start() return instance
def __init__(self, config: Config, **kwargs): """ constructor. to create a instance, use :py:meth:`Browser.create(...)` :param config: """ try: asyncio.get_running_loop() except RuntimeError: raise RuntimeError( "{0} objects of this class are created using await {0}.create()".format( self.__class__.__name__ ) ) # weakref.finalize(self, self._quit, self) self.config = config self.targets: List = [] """current targets (all types""" self.info = None self._target = None self._process = None self._process_pid = None self._keep_user_data_dir = None self._is_updating = asyncio.Event() self.connection: Connection = None logger.debug("Session object initialized: %s" % vars(self)) @property def websocket_url(self): return self.info.webSocketDebuggerUrl @property def main_tab(self) -> tab.Tab: """returns the target which was launched with the browser""" return sorted(self.targets, key=lambda x: x.type_ == "page", reverse=True)[0] @property def tabs(self) -> List[tab.Tab]: """returns the current targets which are of type "page" :return: """ tabs = filter(lambda item: item.type_ == "page", self.targets) return list(tabs) @property def cookies(self) -> CookieJar: if not self._cookies: self._cookies = CookieJar(self) return self._cookies @property def stopped(self): if self._process and self._process.returncode is None: return False return True # return (self._process and self._process.returncode) or False
[docs] async def wait(self, time: Union[float, int] = 0.1): """wait for <time> seconds. important to use, especially in between page navigation :param time: :return: """ try: await asyncio.sleep(time) except asyncio.TimeoutError: pass
sleep = wait """alias for wait""" def _handle_target_update( self, event: Union[ cdp.target.TargetInfoChanged, cdp.target.TargetDestroyed, cdp.target.TargetCreated, cdp.target.TargetCrashed, ], ): """this is an internal handler which updates the targets when chrome emits the corresponding event""" if isinstance(event, cdp.target.TargetInfoChanged): target_info = event.target_info current_tab = next( filter( lambda item: item.target_id == target_info.target_id, self.targets ) ) current_target = current_tab.target if logger.getEffectiveLevel() <= 10: changes = util.compare_target_info(current_target, target_info) changes_string = "" for change in changes: key, old, new = change changes_string += f"\n{key}: {old} => {new}\n" logger.debug( "target #%d has changed: %s" % (self.targets.index(current_tab), changes_string) ) current_tab._target = target_info elif isinstance(event, cdp.target.TargetCreated): target_info: cdp.target.TargetInfo = event.target_info from .tab import Tab new_target = Tab( ( f"ws://{self.config.host}:{self.config.port}" f"/devtools/{target_info.type_ or 'page'}" # all types are 'page' internally in chrome apparently f"/{target_info.target_id}" ), target=target_info, browser=self, ) self.targets.append(new_target) logger.debug("target #%d created => %s", len(self.targets), new_target) elif isinstance(event, cdp.target.TargetDestroyed): current_tab = next( filter(lambda item: item.target_id == event.target_id, self.targets) ) logger.debug( "target removed. id # %d => %s" % (self.targets.index(current_tab), current_tab) ) self.targets.remove(current_tab) asyncio.create_task(self.update_targets())
[docs] async def get( self, url="chrome://welcome", new_tab: bool = False, new_window: bool = False ) -> tab.Tab: """top level get. utilizes the first tab to retrieve given url. convenience function known from selenium. this function handles waits/sleeps and detects when DOM events fired, so it's the safest way of navigating. :param url: the url to navigate to :param new_tab: open new tab :param new_window: open new window :return: Page """ if new_tab or new_window: # creat new target using the browser session target_id = await self.connection.send( cdp.target.create_target( url, new_window=new_window, enable_begin_frame_control=True ) ) # get the connection matching the new target_id from our inventory connection: tab.Tab = next( filter( lambda item: item.type_ == "page" and item.target_id == target_id, self.targets, ) ) connection._browser = self else: # first tab from browser.tabs connection: tab.Tab = next( filter(lambda item: item.type_ == "page", self.targets) ) # use the tab to navigate to new url frame_id, loader_id, *_ = await connection.send(cdp.page.navigate(url)) # update the frame_id on the tab connection.frame_id = frame_id connection._browser = self await self return connection
[docs] async def create_context( self, url: str = "chrome://welcome", new_tab: bool = False, new_window: bool = True, dispose_on_detach: bool = True, proxy_server: str = None, proxy_bypass_list: List[str] = None, origins_with_universal_network_access: List[str] = None, ) -> tab.Tab: """ creates a new browser context - mostly useful if you want to use proxies for different browser instances since chrome usually can only use 1 proxy per browser. socks5 with authentication is supported by using a forwarder proxy, the correct string to use socks proxy with username/password auth is socks://USERNAME:PASSWORD@SERVER:PORT dispose_on_detach – (EXPERIMENTAL) (Optional) If specified, disposes this context when debugging session disconnects. proxy_server – (EXPERIMENTAL) (Optional) Proxy server, similar to the one passed to –proxy-server proxy_bypass_list – (EXPERIMENTAL) (Optional) Proxy bypass list, similar to the one passed to –proxy-bypass-list 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. :param new_window: :type new_window: :param new_tab: :type new_tab: :param url: :type url: :param dispose_on_detach: :type dispose_on_detach: :param proxy_server: :type proxy_server: :param proxy_bypass_list: :type proxy_bypass_list: :param origins_with_universal_network_access: :type origins_with_universal_network_access: :return: :rtype: """ if proxy_server: fw = util.ProxyForwarder(proxy_server=proxy_server) proxy_server = fw.proxy_server ctx: cdp.browser.BrowserContextID = await self.connection.send( cdp.target.create_browser_context( dispose_on_detach=dispose_on_detach, proxy_server=proxy_server, proxy_bypass_list=proxy_bypass_list, origins_with_universal_network_access=origins_with_universal_network_access, ) ) target_id: cdp.target.TargetID = await self.connection.send( cdp.target.create_target( url, browser_context_id=ctx, new_window=new_window, for_tab=new_tab ) ) await self.sleep(0.5) connection: tab.Tab = next( filter( lambda item: item.type_ == "page" and item.target_id == target_id, self.targets, ) ) return connection
[docs] async def start(self=None) -> Browser: """launches the actual browser""" if not self: warnings.warn("use ``await Browser.create()`` to create a new instance") return if self._process or self._process_pid: if self._process.returncode is not None: return await self.create(config=self.config) warnings.warn("ignored! this call has no effect when already running.") return # self.config.update(kwargs) connect_existing = False if self.config.host is not None and self.config.port is not None: connect_existing = True else: self.config.host = "127.0.0.1" self.config.port = util.free_port() if not connect_existing: logger.debug( "BROWSER EXECUTABLE PATH: %s", self.config.browser_executable_path ) if not pathlib.Path(self.config.browser_executable_path).exists(): raise FileNotFoundError( ( """ --------------------- Could not determine browser executable. --------------------- Make sure your browser is installed in the default location (path). If you are sure about the browser executable, you can specify it using the `browser_executable_path='{}` parameter.""" ).format( "/path/to/browser/executable" if is_posix else "c:/path/to/your/browser.exe" ) ) if getattr(self.config, "_extensions", None): # noqa self.config.add_argument( "--load-extension=%s" % ",".join(str(_) for _ in self.config._extensions) ) # noqa exe = self.config.browser_executable_path params = self.config() logger.info( "starting\n\texecutable :%s\n\narguments:\n%s", exe, "\n\t".join(params) ) if not connect_existing: self._process: asyncio.subprocess.Process = ( await asyncio.create_subprocess_exec( # self.config.browser_executable_path, # *cmdparams, exe, *params, stdin=asyncio.subprocess.PIPE, stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.PIPE, close_fds=is_posix, ) ) self._process_pid = self._process.pid self._http = HTTPApi((self.config.host, self.config.port)) util.get_registered_instances().add(self) await asyncio.sleep(0.25) for _ in range(5): try: self.info = ContraDict(await self._http.get("version"), silent=True) except (Exception,): if _ == 4: logger.debug("could not start", exc_info=True) await self.sleep(0.5) else: break if not self.info: raise Exception( ( """ --------------------- Failed to connect to browser --------------------- One of the causes could be when you are running as root. In that case you need to pass no_sandbox=True """ ) ) self.connection = Connection(self.info.webSocketDebuggerUrl, browser=self) if self.config.autodiscover_targets: logger.info("enabling autodiscover targets") self.connection.handlers[cdp.target.TargetInfoChanged] = [ self._handle_target_update ] self.connection.handlers[cdp.target.TargetCreated] = [ self._handle_target_update ] self.connection.handlers[cdp.target.TargetDestroyed] = [ self._handle_target_update ] self.connection.handlers[cdp.target.TargetCrashed] = [ self._handle_target_update ] await self.connection.send(cdp.target.set_discover_targets(discover=True)) await self.update_targets()
# await self # self.connection.handlers[cdp.inspector.Detached] = [self.stop] # return self
[docs] async def grant_all_permissions(self): """ grant permissions for: accessibilityEvents audioCapture backgroundSync backgroundFetch clipboardReadWrite clipboardSanitizedWrite displayCapture durableStorage geolocation idleDetection localFonts midi midiSysex nfc notifications paymentHandler periodicBackgroundSync protectedMediaIdentifier sensors storageAccess topLevelStorageAccess videoCapture videoCapturePanTiltZoom wakeLockScreen wakeLockSystem windowManagement """ permissions = list(cdp.browser.PermissionType) permissions.remove(cdp.browser.PermissionType.FLASH) permissions.remove(cdp.browser.PermissionType.CAPTURED_SURFACE_CONTROL) await self.connection.send(cdp.browser.grant_permissions(permissions))
[docs] async def tile_windows(self, windows=None, max_columns: int = 0): import math import mss m = mss.mss() screen, screen_width, screen_height = 3 * (None,) if m.monitors and len(m.monitors) >= 1: screen = m.monitors[0] screen_width = screen["width"] screen_height = screen["height"] if not screen or not screen_width or not screen_height: warnings.warn("no monitors detected") return await self distinct_windows = defaultdict(list) if windows: tabs = windows else: tabs = self.tabs for tab in tabs: window_id, bounds = await tab.get_window() distinct_windows[window_id].append(tab) num_windows = len(distinct_windows) req_cols = max_columns or int(num_windows * (19 / 6)) req_rows = int(num_windows / req_cols) while req_cols * req_rows < num_windows: req_rows += 1 box_w = math.floor((screen_width / req_cols) - 1) box_h = math.floor(screen_height / req_rows) distinct_windows_iter = iter(distinct_windows.values()) grid = [] for x in range(req_cols): for y in range(req_rows): num = x + y try: tabs = next(distinct_windows_iter) except StopIteration: continue if not tabs: continue tab = tabs[0] try: pos = [x * box_w, y * box_h, box_w, box_h] grid.append(pos) await tab.set_window_size(*pos) except Exception: logger.info( "could not set window size. exception => ", exc_info=True ) continue return grid
async def _get_targets(self) -> List[cdp.target.TargetInfo]: info = await self.connection.send(cdp.target.get_targets(), _is_update=True) return info
[docs] async def update_targets(self): targets: List[cdp.target.TargetInfo] targets = await self._get_targets() target_ids = [t.target_id for t in targets] existing_target_ids = [t.target_id for t in self.targets] for t in targets: for existing_tab in self.targets: existing_target = existing_tab.target if existing_target.target_id == t.target_id: existing_tab.target.__dict__.update(t.__dict__) break else: self.targets.append( Connection( ( f"ws://{self.config.host}:{self.config.port}" f"/devtools/page" # all types are 'page' somehow f"/{t.target_id}" ), target=t, browser=self, ) ) await asyncio.sleep(0)
async def __aenter__(self): return self async def __aexit__(self, exc_type, exc_val, exc_tb): if exc_type and exc_val: raise exc_type(exc_val) def __iter__(self): self._i = self.tabs.index(self.main_tab) return self def __getitem__( self, item: Union[str, int, slice] ) -> Union[tab.Tab, List[tab.Tab]]: """ allows to get py:obj:`tab.Tab` instances by using browser[0], browser[1], etc. a string is also allowed. it will then return the first tab where the py:obj:`cdp.target.TargetInfo` object (as json string) contains the given key, or the first tab in case no matches are found. eg: `browser["google"]` gives the first tab which has "google" in it's serialized target object. :param item: :type item: :return: :rtype: tab.Tab """ if isinstance(item, int): return self.tabs[item] elif isinstance(item, slice): tabs: List[tab.Tab] = [] sta, sto, ste = item.start, item.stop, item.step if not ste: ste = 1 if not sto: sto = len(self.tabs) - 1 if not sta: sta = 0 for x in range(sta, sto, ste): try: tabs.append(self.tabs[x]) except IndexError: pass return tabs elif isinstance(item, tuple): r = range(*item) tabs: List[tab.Tab] = [] for i in r: try: tabs.append(self.tabs[i]) except IndexError: pass return tabs elif isinstance(item, str): for t in self.tabs: if item.lower() in str(t.target.to_json()).lower(): return t else: return self.tabs[0] def __reversed__(self): return reversed(list(self.tabs)) def __next__(self): try: return self.tabs[self._i] except IndexError: del self._i raise StopIteration except AttributeError: del self._i raise StopIteration finally: if hasattr(self, "_i"): if self._i != len(self.tabs): self._i += 1 else: del self._i
[docs] def stop(self): try: # asyncio.get_running_loop().create_task(self.connection.send(cdp.browser.close())) asyncio.get_event_loop().create_task(self.connection.disconnect()) logger.debug("closed the connection using get_event_loop().create_task()") except RuntimeError: if self.connection: try: # asyncio.run(self.connection.send(cdp.browser.close())) asyncio.run(self.connection.disconnect()) logger.debug("closed the connection using asyncio.run()") except Exception: pass for _ in range(3): try: self._process.terminate() logger.info( "terminated browser with pid %d successfully" % self._process.pid ) break except (Exception,): try: self._process.kill() logger.info( "killed browser with pid %d successfully" % self._process.pid ) break except (Exception,): try: if hasattr(self, "browser_process_pid"): os.kill(self._process_pid, 15) logger.info( "killed browser with pid %d using signal 15 successfully" % self._process.pid ) break except (TypeError,): logger.info("typerror", exc_info=True) pass except (PermissionError,): logger.info( "browser already stopped, or no permission to kill. skip" ) pass except (ProcessLookupError,): logger.info("process lookup failure") pass except (Exception,): raise self._process = None self._process_pid = None
def __await__(self): # return ( asyncio.sleep(0)).__await__() return self.update_targets().__await__() def __del__(self): pass
class CookieJar: def __init__(self, browser: Browser): self._browser = browser # self._connection = connection async def get_all( self, requests_cookie_format: bool = False ) -> List[Union[cdp.network.Cookie, "http.cookiejar.Cookie"]]: """ get all cookies :param requests_cookie_format: when True, returns python http.cookiejar.Cookie objects, compatible with requests library and many others. :type requests_cookie_format: bool :return: :rtype: """ connection = None for tab in self._browser.tabs: if tab.closed: continue connection = tab break else: connection = self._browser.connection cookies = await connection.send(cdp.storage.get_cookies()) if requests_cookie_format: import requests.cookies return [ requests.cookies.create_cookie( name=c.name, value=c.value, domain=c.domain, path=c.path, expires=c.expires, secure=c.secure, ) for c in cookies ] return cookies async def set_all(self, cookies: List[cdp.network.CookieParam]): """ set cookies :param cookies: list of cookies :type cookies: :return: :rtype: """ connection = None for tab in self._browser.tabs: if tab.closed: continue connection = tab break else: connection = self._browser.connection cookies = await connection.send(cdp.storage.get_cookies()) await connection.send(cdp.storage.set_cookies(cookies)) async def save(self, file: PathLike = ".session.dat", pattern: str = ".*"): """ save all cookies (or a subset, controlled by `pattern`) to a file to be restored later :param file: :type file: :param pattern: regex style pattern string. any cookie that has a domain, key or value field which matches the pattern will be included. default = ".*" (all) eg: the pattern "(cf|.com|nowsecure)" will include those cookies which: - have a string "cf" (cloudflare) - have ".com" in them, in either domain, key or value field. - contain "nowsecure" :type pattern: str :return: :rtype: """ import re pattern = re.compile(pattern) save_path = pathlib.Path(file).resolve() connection = None for tab in self._browser.tabs: if tab.closed: continue connection = tab break else: connection = self._browser.connection cookies = await self.get_all(requests_cookie_format=False) included_cookies = [] for cookie in cookies: for match in pattern.finditer(str(cookie.__dict__)): logger.debug( "saved cookie for matching pattern '%s' => (%s: %s)", pattern.pattern, cookie.name, cookie.value, ) included_cookies.append(cookie) break pickle.dump(cookies, save_path.open("w+b")) async def load(self, file: PathLike = ".session.dat", pattern: str = ".*"): """ load all cookies (or a subset, controlled by `pattern`) from a file created by :py:meth:`~save_cookies`. :param file: :type file: :param pattern: regex style pattern string. any cookie that has a domain, key or value field which matches the pattern will be included. default = ".*" (all) eg: the pattern "(cf|.com|nowsecure)" will include those cookies which: - have a string "cf" (cloudflare) - have ".com" in them, in either domain, key or value field. - contain "nowsecure" :type pattern: str :return: :rtype: """ import re pattern = re.compile(pattern) save_path = pathlib.Path(file).resolve() cookies = pickle.load(save_path.open("r+b")) included_cookies = [] connection = None for tab in self._browser.tabs: if tab.closed: continue connection = tab break else: connection = self._browser.connection for cookie in cookies: for match in pattern.finditer(str(cookie.__dict__)): included_cookies.append(cookie) logger.debug( "loaded cookie for matching pattern '%s' => (%s: %s)", pattern.pattern, cookie.name, cookie.value, ) break await connection.send(cdp.storage.set_cookies(included_cookies)) async def clear(self): """ clear current cookies note: this includes all open tabs/windows for this browser :return: :rtype: """ connection = None for tab in self._browser.tabs: if tab.closed: continue connection = tab break else: connection = self._browser.connection await connection.send(cdp.storage.clear_cookies()) class HTTPApi: def __init__(self, addr: Tuple[str, int]): self.host, self.port = addr self.api = "http://%s:%d" % (self.host, self.port) @classmethod def from_target(cls, target: "Target"): ws_url = urllib.parse.urlparse(target.websocket_url) inst = cls((ws_url.hostname, ws_url.port)) return inst async def get(self, endpoint: str): return await self._request(endpoint) async def post(self, endpoint, data): return await self._request(endpoint, data) async def _request(self, endpoint, method: str = "get", data: dict = None): url = urllib.parse.urljoin( self.api, f"json/{endpoint}" if endpoint else "/json" ) if data and method.lower() == "get": raise ValueError("get requests cannot contain data") if not url: url = self.api + endpoint request = urllib.request.Request(url) request.method = method request.data = None if data: request.data = json.dumps(data).encode("utf-8") response = await asyncio.get_running_loop().run_in_executor( None, lambda: urllib.request.urlopen(request, timeout=10) ) return json.loads(response.read()) atexit.register(util.deconstruct_browser)