[ie/theguardian] Add extractors (#8535)

[yt-dlp.git] / yt_dlp / cookies.py

diff --git a/yt_dlp/cookies.py b/yt_dlp/cookies.py
index 80428c747bfaeb1f7ae6430b131e0ea4e5adb4ee..a71fbc28baa4ae589a8fa77b56b6bbc7b22ef3bb 100644 (file)
--- a/yt_dlp/cookies.py
+++ b/yt_dlp/cookies.py
@@ -33,7 +33,6 @@
 from .utils import (
     Popen,
     error_to_str,
-    escape_url,
     expand_path,
     is_path_like,
     sanitize_url,
@@ -42,6 +41,7 @@
     write_string,
 )
 from .utils._utils import _YDLLogger
+from .utils.networking import normalize_url
 
 CHROMIUM_BASED_BROWSERS = {'brave', 'chrome', 'chromium', 'edge', 'opera', 'vivaldi'}
 SUPPORTED_BROWSERS = CHROMIUM_BASED_BROWSERS | {'firefox', 'safari'}
@@ -138,7 +138,7 @@ def _extract_firefox_cookies(profile, container, logger):
         containers_path = os.path.join(os.path.dirname(cookie_database_path), 'containers.json')
         if not os.path.isfile(containers_path) or not os.access(containers_path, os.R_OK):
             raise FileNotFoundError(f'could not read containers.json in {search_root}')
-        with open(containers_path) as containers:
+        with open(containers_path, encoding='utf8') as containers:
             identities = json.load(containers).get('identities', [])
         container_id = next((context.get('userContextId') for context in identities if container in (
             context.get('name'),
@@ -1308,7 +1308,7 @@ def prepare_line(line):
 
     def get_cookie_header(self, url):
         """Generate a Cookie HTTP header for a given url"""
-        cookie_req = urllib.request.Request(escape_url(sanitize_url(url)))
+        cookie_req = urllib.request.Request(normalize_url(sanitize_url(url)))
         self.add_cookie_header(cookie_req)
         return cookie_req.get_header('Cookie')
 
@@ -1317,7 +1317,7 @@ def get_cookies_for_url(self, url):
         # Policy `_now` attribute must be set before calling `_cookies_for_request`
         # Ref: https://github.com/python/cpython/blob/3.7/Lib/http/cookiejar.py#L1360
         self._policy._now = self._now = int(time.time())
-        return self._cookies_for_request(urllib.request.Request(escape_url(sanitize_url(url))))
+        return self._cookies_for_request(urllib.request.Request(normalize_url(sanitize_url(url))))
 
     def clear(self, *args, **kwargs):
         with contextlib.suppress(KeyError):