From 68326a60ee20e2a8831ae86e9867b352e0f74ca6 Mon Sep 17 00:00:00 2001 From: Ben Muthalaly Date: Tue, 27 Feb 2024 15:30:31 -0600 Subject: [PATCH] Add cookies file to http request in `download_url` --- archivebox/util.py | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-) diff --git a/archivebox/util.py b/archivebox/util.py index 5321081c..2e1e4907 100644 --- a/archivebox/util.py +++ b/archivebox/util.py @@ -3,6 +3,7 @@ __package__ = 'archivebox' import re import requests import json as pyjson +import http.cookiejar from typing import List, Optional, Any from pathlib import Path @@ -164,13 +165,26 @@ def parse_date(date: Any) -> Optional[datetime]: @enforce_types def download_url(url: str, timeout: int=None) -> str: """Download the contents of a remote url and return the text""" - from .config import TIMEOUT, CHECK_SSL_VALIDITY, WGET_USER_AGENT + from .config import ( + TIMEOUT, + CHECK_SSL_VALIDITY, + WGET_USER_AGENT, + COOKIES_FILE, + ) timeout = timeout or TIMEOUT + + cookie_jar = http.cookiejar.MozillaCookieJar() + if COOKIES_FILE is not None: + cookie_jar.load(COOKIES_FILE, ignore_discard=True, ignore_expires=True) + else: + cookie_jar = None + response = requests.get( url, headers={'User-Agent': WGET_USER_AGENT}, verify=CHECK_SSL_VALIDITY, timeout=timeout, + cookies=cookie_jar, ) content_type = response.headers.get('Content-Type', '')