From e5b7f72b719814ffa2748e8e8ed1c6713a24e1a6 Mon Sep 17 00:00:00 2001 From: Heiner Lohaus Date: Thu, 22 Feb 2024 00:16:58 +0100 Subject: Move some modules, create providers dir Set min version for duckduckgo Make duckduckgo search async Remove get_lastet_version --- g4f/requests/__init__.py | 56 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 56 insertions(+) create mode 100644 g4f/requests/__init__.py (limited to 'g4f/requests/__init__.py') diff --git a/g4f/requests/__init__.py b/g4f/requests/__init__.py new file mode 100644 index 00000000..d278ffaf --- /dev/null +++ b/g4f/requests/__init__.py @@ -0,0 +1,56 @@ +from __future__ import annotations + +from urllib.parse import urlparse + +try: + from curl_cffi.requests import Session + from .curl_cffi import StreamResponse, StreamSession + has_curl_cffi = True +except ImportError: + from typing import Type as Session + from .aiohttp import StreamResponse, StreamSession + has_curl_cffi = False + +from ..webdriver import WebDriver, WebDriverSession, bypass_cloudflare, get_driver_cookies +from ..errors import MissingRequirementsError +from .defaults import DEFAULT_HEADERS + +def get_args_from_browser(url: str, webdriver: WebDriver = None, proxy: str = None, timeout: int = 120) -> dict: + """ + Create a Session object using a WebDriver to handle cookies and headers. + + Args: + url (str): The URL to navigate to using the WebDriver. + webdriver (WebDriver, optional): The WebDriver instance to use. + proxy (str, optional): Proxy server to use for the Session. + timeout (int, optional): Timeout in seconds for the WebDriver. + + Returns: + Session: A Session object configured with cookies and headers from the WebDriver. + """ + with WebDriverSession(webdriver, "", proxy=proxy, virtual_display=False) as driver: + bypass_cloudflare(driver, url, timeout) + cookies = get_driver_cookies(driver) + user_agent = driver.execute_script("return navigator.userAgent") + parse = urlparse(url) + return { + 'cookies': cookies, + 'headers': { + **DEFAULT_HEADERS, + 'Authority': parse.netloc, + 'Origin': f'{parse.scheme}://{parse.netloc}', + 'Referer': url, + 'User-Agent': user_agent, + }, + } + +def get_session_from_browser(url: str, webdriver: WebDriver = None, proxy: str = None, timeout: int = 120) -> Session: + if not has_curl_cffi: + raise MissingRequirementsError('Install "curl_cffi" package') + args = get_args_from_browser(url, webdriver, proxy, timeout) + return Session( + **args, + proxies={"https": proxy, "http": proxy}, + timeout=timeout, + impersonate="chrome110" + ) \ No newline at end of file -- cgit v1.2.3