[PR #11054/e2eb1959 backport][3.12] Fix CookieJar memory leak in filter_cookies()

[patchback]

This is a backport of PR #11054 as merged into master (e2eb195).

Summary

This PR fixes a memory leak in CookieJar.filter_cookies() that was causing unbounded memory growth when making requests to different URL paths. The issue was introduced in version 3.10.0 and could cause memory usage to grow from ~100MB to ~5GB over weeks of operation.

The Problem

The filter_cookies() method was inadvertently creating empty cookie entries for every domain-path combination it checked. This happened because:

1. The method generates all possible domain-path combinations for a request URL
2. It then accesses self._cookies[p] for each combination
3. Since self._cookies is a defaultdict(SimpleCookie), accessing a non-existent key creates an empty SimpleCookie object
4. These empty objects accumulate over time, causing the memory leak

The Fix

Added a simple check before accessing the dictionary:

if p not in self._cookies:
    continue

This prevents the creation of empty cookie entries while maintaining all existing functionality.

Testing

• Added regression test test_filter_cookies_does_not_leak_memory() that verifies the internal storage doesn't grow when filtering cookies for different paths
• All existing cookie tests pass
• Manual testing confirms the memory leak is resolved

Impact

• Fixes Possible memory leak in the CookieJar #11052 memory leak issue
• No API changes
• No behavioral changes for existing code
• Minimal performance impact (one additional dictionary lookup per domain-path combination)

[codspeed-hq]

CodSpeed Performance Report

Merging #11068 will not alter performance

_{Comparing patchback/backports/3.12/e2eb1959237c8bc46a3f0e79a77c2086145d98cf/pr-11054 (1c0d613) with 3.12 (872cab6)}

Summary

✅ 59 untouched benchmarks

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.24%. Comparing base (872cab6) to head (1c0d613).
⚠️ Report is 95 commits behind head on 3.12.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             3.12   #11068      +/-   ##
==========================================
+ Coverage   98.21%   98.24%   +0.02%     
==========================================
  Files         130      130              
  Lines       41588    41612      +24     
  Branches     2281     2286       +5     
==========================================
+ Hits        40847    40881      +34     
+ Misses        563      557       -6     
+ Partials      178      174       -4     

Flag	Coverage Δ
CI-GHA	98.13% <100.00%> (+0.01%)	⬆️
OS-Linux	97.87% <100.00%> (+0.01%)	⬆️
OS-Windows	95.47% <100.00%> (+0.49%)	⬆️
OS-macOS	97.08% <100.00%> (+<0.01%)	⬆️
Py-3.10.11	96.96% <100.00%> (?)
Py-3.10.17	97.49% <100.00%> (+0.21%)	⬆️
Py-3.11.12	97.57% <100.00%> (-0.01%)	⬇️
Py-3.11.9	97.05% <100.00%> (?)
Py-3.12.10	97.92% <100.00%> (+0.86%)	⬆️
Py-3.13.3	97.92% <100.00%> (+0.23%)	⬆️
Py-3.9.13	96.86% <100.00%> (+<0.01%)	⬆️
Py-3.9.22	97.38% <100.00%> (+0.20%)	⬆️
Py-pypy7.3.16	86.69% <100.00%> (?)
VM-macos	97.08% <100.00%> (+<0.01%)	⬆️
VM-ubuntu	97.87% <100.00%> (+0.01%)	⬆️
VM-windows	95.47% <100.00%> (+0.49%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.