feat(mangler): mangle private class names

[camc314]

$ cat test.js
class Foo {
  constructor(k) {
    this.#kasd = k;
  }

  getK() {
    return this.#kasd;
  }
}
class Foo2 {
  constructor(k) {
    this.#kasd = k;
  }

  getK() {
    return this.#kasd;
  }
}

export { Foo, Foo2 };
$ cargo run -p oxc_minifier --example minifier -- --mangle
   Compiling oxc_mangler v0.51.0 (/Users/cameron/github/Boshen/oxc/crates/oxc_mangler)
   Compiling oxc_minifier v0.51.0 (/Users/cameron/github/Boshen/oxc/crates/oxc_minifier)
    Finished `dev` profile [unoptimized] target(s) in 0.96s
     Running `target/debug/examples/minifier --mangle`
class Foo {
        constructor(e) {
                this.#e = e;
        }
        getK() {
                return this.#e;
        }
}
class Foo2 {
        constructor(e) {
                this.#e = e;
        }
        getK() {
                return this.#e;
        }
}
export { Foo, Foo2 };

No diff on cargo misize. looks like non of the test suite has classes with private class members

[camc314]

• feat(mangler): mangle private class names #9279 👈 (View in Graphite)
• refactor(mangler): move base54 into seperate mod #9278
• main

---

How to use the Graphite Merge Queue

Add either label to this PR to merge it via the merge queue:

• 0-merge - adds this PR to the back of the merge queue
• hotfix - for urgent hot fixes, skip the queue and merge this PR next

You must have a Graphite account in order to use the merge queue. Sign up using this link.

An organization admin has enabled the Graphite Merge Queue in this repository.

Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.

This stack of pull requests is managed by Graphite. Learn more about stacking.

[codspeed-hq]

CodSpeed Performance Report

Merging #9279 will degrade performances by 88.4%

_{Comparing c/02-21-feat_mangler_mangle_private_class_names (9106d71) with main (4ad328b)}

Summary

❌ 3 regressions
✅ 30 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
❌	mangler[antd.js]	16.2 ms	139.3 ms	-88.4%
❌	mangler[react.development.js]	299.4 µs	1,766.4 µs	-83.05%
❌	mangler[typescript.js]	39.8 ms	280.4 ms	-85.79%

[Boshen]

This adds an additional AST pass, but how come it's not shown on the benchmark 🤔 ?

[camc314]

so i implemented this inside:

oxc/crates/oxc_minifier/src/lib.rs

Line 48 in 82c354a

pub fn build<'a>(self, allocator: &'a Allocator, program: &mut Program<'a>) -> MinifierReturn {

minifier::build. but that fn doesn't run when doing the benches. we run the mangler/minifier explicitly.

i think i could either:

1. move this directly into the mangler
2. move this as a minifier pass
   probably 1 is better?

[Boshen]

• move this directly into the mangler

Yeah this seems better. We can also add a flag (turned off by default) to enable this.

[camc314]

this will regress perf a lot as we are now building semantic inside mangler (inside runner.run, rather than in the prior phase)

I had to make this change because semantic requires a reference to the Program, but we need to pass a mutable reference into here for the private class members to mutate the AST.

i'm open to other ideas:

• we could have dedicted methods for mangling symbols vs private class members, and a .finish returning the symbol table
• something else?