fix(es/minifier): Handle toExponential(undefined)

[cuyl]

• Evaluate when parameters is empty or undefined.

Closes #11572

[changeset-bot]

⚠️ No Changeset found

Latest commit: c72e302

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[github-actions]

Binary Sizes

File	Size
swc.linux-x64-gnu.node	28M (28744264 bytes)

Commit: 035bdb0

[codspeed-hq]

Merging this PR will not alter performance

✅ 190 untouched benchmarks

---

_{Comparing cuyl:handle-undefined-args-for-toprecision (c72e302) with main (f11fd70)}

[cuyl]

Opps, let me see if I can find a way to remove clone.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0e614e347d

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[cuyl]

Do not reject all spread arguments in number call folding
This guard returns immediately whenever the first argument is spread, which makes get_first_arg unreachable for cases like (1.23).toExponential(...[]) and (1.23).toExponential(...[undefined]). Those calls are equivalent to passing no precision and are exactly the new cases added in tests/terser/compress/evaluate/number_method_call/input.js, so the current condition prevents the intended optimization from ever running.
Useful? React with 👍 / 👎.

We already have eval_array_spread, eval_spread_array_in_args and eval_spread_array_in_array runs before eval_number_method_call I belive most of cases like this are handled.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c72e302ce2

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[cuyl]

Does anything block this PR?
@kdy1

[kdy1]

AI coding is fine, but well... the diff in this PR does not look fine

[kdy1]

This function is very wrong

[cuyl]

Opps, Should I revert it? I just want to remove the clone as you requested.
At first time I tried to add lifetime to prevent the clone, but I got an error.

error: reached the recursion limit while instantiating `get_first_arg::<'_, std::iter::Map<std::slice::Iter<'_, std::option::Option<swc_ecma_ast::ExprOrSpread>>, {closure@...}>>`

The code is here.

fn get_first_arg<'a, I>(args: I, expr_ctx: &ExprCtx) -> Option<Option<f64>>

where

    I: IntoIterator<Item = Option<&'a ExprOrSpread>>,

{

    args.into_iter()

        .find_map(|arg| {

            if let Some(arg) = arg {

                if arg.spread.is_some() {

                    return match arg.expr.as_array() {

                        Some(args) => {

                            if args.elems.is_empty() {

                                // next argument is used if the first spread argument is empty

                                // so we can't evaluate the call.

                                None

                            } else {

                                Some(get_first_arg(

                                    args.elems.iter().map(|e| e.as_ref()),

                                    expr_ctx,

                                ))

                            }

                        }

                        // If the spread argument is not an array, we can't evaluate the call.

                        None => Some(None),

                    };

                } else if !arg.expr.is_undefined(*expr_ctx) {

                    return Some(Some(eval_as_number(*expr_ctx, &arg.expr)));

                }

            }

            Some(None)

        })

        .unwrap_or(None)

}

To resolve the error, I've using AI to refactor it to the non-recursion version.

[cuyl]

But AI can't make it right, so I changed some logic to make it right.