Improve performance of path functions on cache miss

[bdraco]

There were a few places we called the quoter when we know its never going to do anything.

On cache miss for .query_string, this will help a bit. However its not possible to benchmark right now because of #1443 (comment)

We can avoid some of the property accesses internally as well which should speed these up a bit. Since the aiohttp url dispatcher calls path_safe there is a much larger chance we get a cache miss so it makes sense to optimize this one a bit.

[codspeed-hq]

CodSpeed Performance Report

Merging #1443 will improve performances by 84.24%

_{Comparing skip_quoting_empty_and_slash (1713ef4) with master (7585a96)}

Summary

⚡ 7 improvements
✅ 91 untouched benchmarks

Benchmarks breakdown

	Benchmark	master	skip_quoting_empty_and_slash	Change
⚡	test_empty_path_safe_uncached	166 µs	90.6 µs	+83.31%
⚡	test_empty_path_uncached	166.9 µs	90.6 µs	+84.24%
⚡	test_empty_query_string_uncached	104.6 µs	80.4 µs	+30.17%
⚡	test_empty_raw_path_qs_uncached	113.6 µs	97.1 µs	+17%
⚡	test_path_safe_uncached	182.1 µs	161.8 µs	+12.58%
⚡	test_raw_path_qs_uncached	114.1 µs	93.9 µs	+21.55%
⚡	test_raw_path_qs_with_query_uncached	145.2 µs	125.1 µs	+16.03%

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.14%. Comparing base (7585a96) to head (1713ef4).
Report is 24 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1443      +/-   ##
==========================================
+ Coverage   96.13%   96.14%   +0.01%     
==========================================
  Files          31       31              
  Lines        5948     5970      +22     
  Branches      362      364       +2     
==========================================
+ Hits         5718     5740      +22     
  Misses        204      204              
  Partials       26       26              

Flag	Coverage Δ
CI-GHA	96.14% <100.00%> (+0.01%)	⬆️
MyPy	49.71% <36.66%> (-0.15%)	⬇️
OS-Linux	99.56% <100.00%> (+<0.01%)	⬆️
OS-Windows	99.62% <100.00%> (+<0.01%)	⬆️
OS-macOS	99.31% <100.00%> (+<0.01%)	⬆️
Py-3.10.11	99.29% <100.00%> (+<0.01%)	⬆️
Py-3.10.15	99.52% <100.00%> (+<0.01%)	⬆️
Py-3.11.10	99.52% <100.00%> (+<0.01%)	⬆️
Py-3.11.9	99.29% <100.00%> (+<0.01%)	⬆️
Py-3.12.7	99.52% <100.00%> (+<0.01%)	⬆️
Py-3.13.0	99.52% <100.00%> (+<0.01%)	⬆️
Py-3.9.13	99.25% <100.00%> (+<0.01%)	⬆️
Py-3.9.20	99.48% <100.00%> (+<0.01%)	⬆️
Py-pypy7.3.16	99.54% <100.00%> (+<0.01%)	⬆️
Py-pypy7.3.17	99.56% <100.00%> (+<0.01%)	⬆️
VM-macos-latest	99.31% <100.00%> (+<0.01%)	⬆️
VM-ubuntu-latest	99.56% <100.00%> (+<0.01%)	⬆️
VM-windows-latest	99.62% <100.00%> (+<0.01%)	⬆️
pytest	99.56% <100.00%> (+<0.01%)	⬆️

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.

[bdraco]

Its not going to show any difference because the cached_property is already cached and it only does it once

type(url).query_string.wrapped(url) would work to be able to benchmark cache miss wrapped was public in propcache

[bdraco]

Probably should do a new release of propcache with wrapped marked as public but it will likely fail because of pypa/gh-action-pypi-publish#307

[bdraco]

Its actually unlikely to make much difference since with the caching improvements (unreleased due to the publish issue) its so much more likely to be cached anyways.

[bdraco]

Well I guess it is worth it