Refactor parser to be lenient on values

[blakeembrey]

Building on #57 for benchmarking. Rewrites the parser to be incremental without regex validation, and more lenient for upstream dependents who need things like foo=bar=baz to be parsed correctly.

[blakeembrey]

Perf tests before:

 ✓ src/index.bench.ts > parse 4917ms
     name                                      hz     min     max    mean     p75     p99    p995    p999     rme  samples
   · basic                          10,545,863.30  0.0000  2.0555  0.0001  0.0001  0.0001  0.0002  0.0007  ±0.91%  5272932
   · simple parameters               2,414,131.39  0.0003  0.3282  0.0004  0.0004  0.0005  0.0005  0.0010  ±0.37%  1207066
   · quoted and escaped parameters   1,700,063.21  0.0005  0.3490  0.0006  0.0006  0.0007  0.0008  0.0012  ±0.33%   850032
   · OWS-heavy parameters            2,370,873.91  0.0003  0.4244  0.0004  0.0004  0.0005  0.0006  0.0010  ±0.45%  1185437

 ✓ src/index.bench.ts > format 4824ms
     name                                      hz     min     max    mean     p75     p99    p995    p999     rme  samples
   · basic                          15,954,499.11  0.0000  0.0858  0.0001  0.0001  0.0001  0.0001  0.0002  ±0.07%  7977250
   · simple parameters               2,152,850.11  0.0004  0.3426  0.0005  0.0005  0.0006  0.0009  0.0011  ±0.44%  1076426
   · quoted and escaped parameters   1,545,029.32  0.0005  0.3580  0.0006  0.0007  0.0008  0.0010  0.0013  ±0.33%   772515

After:

 ✓ src/index.bench.ts > parse 5031ms
     name                                      hz     min     max    mean     p75     p99    p995    p999     rme  samples
   · basic                          10,488,660.07  0.0000  1.8354  0.0001  0.0001  0.0001  0.0002  0.0007  ±1.16%  5244331
   · simple parameters               2,896,440.78  0.0002  0.4201  0.0003  0.0003  0.0005  0.0005  0.0009  ±0.22%  1448221
   · quoted and escaped parameters   1,650,035.57  0.0005  0.3441  0.0006  0.0006  0.0007  0.0009  0.0021  ±0.33%   825018
   · OWS-heavy parameters            2,673,445.24  0.0003  0.3925  0.0004  0.0004  0.0004  0.0005  0.0008  ±0.22%  1336723

 ✓ src/index.bench.ts > format 4775ms
     name                                      hz     min     max    mean     p75     p99    p995    p999     rme  samples
   · basic                          16,172,897.54  0.0000  0.0415  0.0001  0.0001  0.0001  0.0001  0.0002  ±0.10%  8086450
   · simple parameters               2,102,840.62  0.0004  0.3839  0.0005  0.0005  0.0006  0.0010  0.0012  ±0.50%  1051421
   · quoted and escaped parameters   1,518,375.90  0.0005  0.3014  0.0007  0.0007  0.0008  0.0010  0.0013  ±0.33%   759188

Arguably not much performance change but the changes aren't easy to do in a regex only.

[blakeembrey]

Realized there is an optimization available for getting rid of .trim(), new benchmark:

 ✓ src/index.bench.ts > parse 5026ms
     name                                      hz     min     max    mean     p75     p99    p995    p999     rme  samples
   · basic                          11,091,574.29  0.0000  3.7066  0.0001  0.0001  0.0001  0.0001  0.0005  ±1.51%  5545788
   · simple parameters               2,688,793.52  0.0003  0.3477  0.0004  0.0004  0.0005  0.0005  0.0008  ±0.20%  1344397
   · quoted and escaped parameters   1,758,222.89  0.0005  0.3292  0.0006  0.0006  0.0007  0.0007  0.0012  ±0.25%   879112
   · OWS-heavy parameters            2,349,556.00  0.0003  3.2109  0.0004  0.0004  0.0005  0.0005  0.0009  ±1.87%  1174778

 ✓ src/index.bench.ts > format 4797ms
     name                                      hz     min     max    mean     p75     p99    p995    p999     rme  samples
   · basic                          15,884,536.95  0.0000  0.0384  0.0001  0.0001  0.0001  0.0001  0.0002  ±0.08%  7942269
   · simple parameters               2,111,259.33  0.0004  0.3335  0.0005  0.0005  0.0006  0.0009  0.0011  ±0.43%  1055630
   · quoted and escaped parameters   1,517,261.44  0.0005  0.3292  0.0007  0.0007  0.0008  0.0010  0.0013  ±0.42%   758631

[bjohansebas]

Just one question before I continue reviewing, will this be a new major version? I assume so, or not? Just asking so I can better align with what you’re planning to do.

[blakeembrey]

Yes, sorry. I mentioned a major in the typescript PR so this plus follow ups beyond this to further optimize by removing old code paths will be a major. I tried to change minimal things in this PR on the API side to get a fair comparison on benchmarks.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (edaa362) to head (cc480a1).

Additional details and impacted files

@@               Coverage Diff               @@
##           be/typescript       #58   +/-   ##
===============================================
  Coverage         100.00%   100.00%           
===============================================
  Files                  1         1           
  Lines                 97       137   +40     
  Branches              40        55   +15     
===============================================
+ Hits                  97       137   +40     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[blakeembrey]

Accidentally deleted rebasing my 3rd branch from main due to the TypeScript PR changes. 😓