chore: Replace sync.Mutex with sync.Map

[piyushroshan]

Sync.Map provides a key level locking to allow concurrent key writes.

Thank you for contributing to Coraza WAF, your effort is greatly appreciated
Before submitting check if what you want to add to coraza list meets quality standards before sending pull request. Thanks!

Make sure that you've checked the boxes below before you submit PR:

• My code includes positive and negative tests.
• I have an appropriate description with correct grammar.
• I have read Contribution guidelines, maintainers note and Quality standard.
• My code is properly linted and passes pre-commit tests.

Thanks for your contribution ❤️

Benchmarks based on mutex vs sync.Map
sync.Mutex

goos: darwin
goarch: arm64
pkg: github.com/corazawaf/coraza/v3/internal/corazawaf
cpu: Apple M1 Pro
BenchmarkAddTransformationUnique-10        38994            282718 ns/op          766436 B/op          5 allocs/op
BenchmarkAddTransformationSame-10         777428              9162 ns/op             242 B/op          5 allocs/op
PASS
ok      github.com/corazawaf/coraza/v3/internal/corazawaf       19.642s

sync.Map

goos: darwin
goarch: arm64
pkg: github.com/corazawaf/coraza/v3/internal/corazawaf
cpu: Apple M1 Pro
BenchmarkAddTransformationUnique-10        32040            220037 ns/op          627028 B/op          9 allocs/op
BenchmarkAddTransformationSame-10         846592              6618 ns/op             366 B/op          9 allocs/op
PASS
ok      github.com/corazawaf/coraza/v3/internal/corazawaf       13.973s

[jcchavezs]

I think this might perform slightly better but I wonder if we could write a benchmark

[jptosso]

IMO it's a hard case to benchmark bc it requires a lot of concurrency and we cannot use gh actions for it
A sync mutex blocks read and write, and sync map just blocks at key level

[fzipi]

maybe something like https://medium.com/@nikolay.dimitrov/comparing-go-performance-of-mutex-sync-map-and-channels-for-syncing-a-map-of-counters-61810c201e17 ?

[fzipi]

Another option is to use sync.RWMutex and use RLock for the ones which are read-only.

[jcchavezs]

To be honest without evidence this being anyhow better I would not proceed with the change. sync.RWMutext an easier change to land.

[piyushroshan]

To be honest without evidence this being anyhow better I would not proceed with the change. sync.RWMutext an easier change to land.

Added the benchmark

[fzipi]

So the benchmark shows this is indeed faster for the tests. The question remains about what happens with real concurrency, but I think it will only behave better because there is no full blocking. @jcchavezs Do you want to make benchmakrs with sync.RWMutex and push them here also to compare? 🙏

[fzipi]

ping @jcchavezs

[jcchavezs]

I will do that tomorrow.

[jptosso]

LGTM

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 81.66%. Comparing base (244ba00) to head (c0d3b05).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1197      +/-   ##
==========================================
- Coverage   81.66%   81.66%   -0.01%     
==========================================
  Files         168      168              
  Lines        9655     9652       -3     
==========================================
- Hits         7885     7882       -3     
  Misses       1519     1519              
  Partials      251      251              

Flag	Coverage Δ
coraza.rule.case_sensitive_args_keys	81.62% <100.00%> (-0.01%)	⬇️
coraza.rule.multiphase_valuation	81.66% <100.00%> (-0.01%)	⬇️
coraza.rule.no_regex_multiline	81.59% <100.00%> (-0.01%)	⬇️
default	81.66% <100.00%> (-0.01%)	⬇️
examples+	16.46% <0.00%> (+<0.01%)	⬆️
examples+coraza.rule.case_sensitive_args_keys	81.62% <100.00%> (-0.01%)	⬇️
examples+coraza.rule.multiphase_valuation	81.49% <100.00%> (-0.01%)	⬇️
examples+coraza.rule.no_regex_multiline	81.51% <100.00%> (-0.01%)	⬇️
examples+memoize_builders	81.61% <100.00%> (-0.01%)	⬇️
examples+no_fs_access	80.93% <100.00%> (-0.01%)	⬇️
ftw	81.66% <100.00%> (-0.01%)	⬇️
memoize_builders	81.75% <100.00%> (-0.01%)	⬇️
no_fs_access	81.10% <100.00%> (-0.01%)	⬇️
tinygo	81.62% <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.

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests