perf(web): avoid CookieJar's overhead if no cookie

See also aio-libs/aiohttp#7583

Signed-off-by: Rongrong <[email protected]>

src/web/req.py

@@ -21,7 +21,7 @@
 from ..compat import nullcontext, ssl_create_default_context, AiohttpUvloopTransportHotfix
 from ..aio_helper import run_async
 from ..errors_collection import RetryInIpv4
-from .utils import WebResponse, proxy_filter, logger, sentinel
+from .utils import YummyCookieJar, WebResponse, proxy_filter, logger, sentinel
 
 DEFAULT_READ_BUFFER_SIZE: Final = 2 ** 16
 
@@ -132,7 +132,7 @@ async def _get(url: str, resp_callback: Callable, timeout: Optional[float] = sen
 
     async def _fetch():
         async with aiohttp.ClientSession(connector=proxy_connector, timeout=aiohttp.ClientTimeout(total=timeout),
-                                         headers=_headers) as session:
+                                         headers=_headers, cookie_jar=YummyCookieJar()) as session:
             async with session.get(url, read_bufsize=read_bufsize, read_until_eof=read_until_eof) as response:
                 async with AiohttpUvloopTransportHotfix(response):
                     status_url_history = [(resp.status, resp.url) for resp in response.history]

src/web/utils.py

@@ -2,6 +2,8 @@
 from typing import Union, Optional, AnyStr
 from typing_extensions import Final
 
+import aiohttp
+import aiohttp.abc
 import feedparser
 from contextlib import suppress
 from dataclasses import dataclass
@@ -24,6 +26,47 @@
 sentinel = object()
 
 
+class YummyCookieJar(aiohttp.abc.AbstractCookieJar):
+    """
+    A cookie jar that acts as a DummyCookieJar in the initial state.
+    Then it only switches to CookieJar when there is any cookie (``update_cookies`` is called).
+    In our use case, it is common that the response does not contain any cookie, as we mostly fetch RSS feeds and
+    multimedia files.
+    As a result, the cookie jar is mostly empty, and the overhead of CookieJar, which is expensive, is unnecessary.
+    So it is expected that YummyCookieJar will seldom switch to CookieJar, acting as a DummyCookieJar most of the time.
+
+    See also https://github.com/aio-libs/aiohttp/issues/7583
+    """
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.__real_cookie_jar = aiohttp.DummyCookieJar(*args, **kwargs)
+        self.__init_args = args
+        self.__init_kwargs = kwargs
+        self.__is_dummy = True
+
+    def update_cookies(self, *args, **kwargs):
+        if self.__is_dummy:
+            self.__real_cookie_jar = aiohttp.CookieJar(*self.__init_args, **self.__init_kwargs)
+            self.__is_dummy = False
+        return self.__real_cookie_jar.update_cookies(*args, **kwargs)
+
+    def __iter__(self):
+        return self.__real_cookie_jar.__iter__()
+
+    def __len__(self) -> int:
+        return self.__real_cookie_jar.__len__()
+
+    def clear(self, *args, **kwargs):
+        return self.__real_cookie_jar.clear(*args, **kwargs)
+
+    def clear_domain(self, *args, **kwargs):
+        return self.__real_cookie_jar.clear_domain(*args, **kwargs)
+
+    def filter_cookies(self, *args, **kwargs):
+        return self.__real_cookie_jar.filter_cookies(*args, **kwargs)
+
+
 class WebError(Exception):
     def __init__(self, error_name: str, status: Union[int, str] = None, url: str = None,
                  base_error: Exception = None, hide_base_error: bool = False, log_level: int = log.DEBUG):