Skip filtering CookieJar when the jar is empty or all cookies hav…

…e expired (#7819)

The filtering itself and its preparation in `CookieJar.filter_cookies()`
is expensive. Sometimes there are no cookies in the jar or all cookies
have expired. Skip filtering and its preparation in this case.

Because the empty check is much cheaper than `_do_expiration()`, I think
it deserves to be duplicated before and after calling
`_do_expiration()`.

```console
$ python3.11 -m timeit -s 'from collections import defaultdict; d=defaultdict(foo="bar")' \
> 'if not d: pass'
50000000 loops, best of 5: 8.3 nsec per loop
$ python3.11 -m timeit -s 'from collections import defaultdict; d=defaultdict()' \
> 'if not d: pass'
50000000 loops, best of 5: 8.74 nsec per loop
$ python3.11 -m timeit -s 'from aiohttp import CookieJar; cj = CookieJar()' \
> 'cj._do_expiration()'
200000 loops, best of 5: 1.86 usec per loop
```

CHANGES/7819.feature

@@ -0,0 +1 @@
+Skip filtering ``CookieJar`` when the jar is empty or all cookies have expired.

CONTRIBUTORS.txt

@@ -282,6 +282,7 @@ Robert Lu
 Robert Nikolich
 Roman Markeloff
 Roman Podoliaka
+Rong Zhang
 Samir Akarioh
 Samuel Colvin
 Sean Hunt

aiohttp/cookiejar.py

@@ -234,7 +234,6 @@ def filter_cookies(
         self, request_url: URL = URL()
     ) -> Union["BaseCookie[str]", "SimpleCookie[str]"]:
         """Returns this jar's cookies filtered by their attributes."""
-        self._do_expiration()
         if not isinstance(request_url, URL):
             warnings.warn(
                 "The method accepts yarl.URL instances only, got {}".format(
@@ -246,6 +245,13 @@ def filter_cookies(
         filtered: Union["SimpleCookie[str]", "BaseCookie[str]"] = (
             SimpleCookie() if self._quote_cookie else BaseCookie()
         )
+        if not self._cookies:
+            # Skip do_expiration() if there are no cookies.
+            return filtered
+        self._do_expiration()
+        if not self._cookies:
+            # Skip rest of function if no non-expired cookies.
+            return filtered
         hostname = request_url.raw_host or ""
         request_origin = URL()
         with contextlib.suppress(ValueError):