feat(es/minifier): support `reduce_escaped_newline` #10232

wjw99830 · 2025-03-19T16:58:35Z

Description: Add compress.experimental.reduce_escaped_newline option to convert StringLiteral with escaped newline \n to TemplateLiteral with newline character(only for compress.ecma >= 2015).

Default is true.

BREAKING CHANGE: None.

Related issue (if exists): #10229

changeset-bot · 2025-03-19T16:58:51Z

🦋 Changeset detected

Latest commit: ea90d46

The changes in this PR will be included in the next version bump.

Not sure what this means? Click here to learn what changesets are.

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

codspeed-hq · 2025-03-19T17:21:36Z

CodSpeed Performance Report

Merging #10232 will improve performances by 1.32%

_{Comparing wjw99830:feat/reduce_escaped_newline (ea90d46) with main (0d88041)}

Summary

⚡ 1 improvements
✅ 151 untouched benchmarks

Benchmarks breakdown

	Benchmark	`BASE`	`HEAD`	Change
⚡	`es/lints/libs/d3`	28.1 ms	27.7 ms	+1.32%

kdy1 · 2025-03-20T02:53:35Z

I think we should not add a new option for this

wjw99830 · 2025-03-20T03:27:29Z

I think we should not add a new option for this

You mean enable the feature by default and cannot be turned off ?

I have the same idea in fact. But considering the stability of the feature, I added that.

kdy1 · 2025-03-20T03:56:26Z

If so, can we have an experimental config instead? Like compress.experimental.reduce_escaped_newline? You may need to define CompressExperimentalOptions struct. (And please make it #[non_exhaustive])

wjw99830 · 2025-03-20T04:00:04Z

Good idea.I'll try it later.

wjw99830 · 2025-03-20T06:49:16Z

I find that swc_ecma_codegen will generate \n for \\n within TemplateElement.raw.

let unescape = match c {
    '\\' => match iter.next() {
        Some(c) => match c {
            'n' => Some('\n'),
            // ...
        },
        // ...
    },
    _ => Some(c),
};

match unescape {
    Some('\n') => buf.push('\n'),
    // ...
}

It’s same as what I do inside reduce_escaped_newline_for_tpl_element, so I will remove it to prevent redundant work.

It still makes sense to retain reduce_escaped_newline_for_str_lit, means that reduce_escaped_newline works only when compress.ecma >= 2015. Do you have any better ideas about the design of this option?

kdy1 · 2025-03-20T06:55:14Z

Maybe add an option to the codegen option?
It's already #[non_exhaustive], so you would need to add one method and call it from

swc/crates/swc/src/lib.rs

Lines 910 to 917 in b192dc8

    
           codegen_config: swc_ecma_codegen::Config::default() 
        
               .with_target(target) 
        
               .with_minify(true) 
        
               .with_ascii_only(opts.format.ascii_only) 
        
               .with_emit_assert_for_import_attributes( 
        
                   opts.format.emit_assert_for_import_attributes, 
        
               ) 
        
               .with_inline_script(opts.format.inline_script),

wjw99830 · 2025-03-20T07:18:08Z

OK. Maybe we should call it reduce_escaped_controls to including \t(or possible others in future).

…ine` for minifier and `reduce_escaped_newline` for codegen

crates/swc_ecma_codegen/src/config.rs

kdy1 · 2025-03-23T12:10:04Z

crates/swc_ecma_minifier/src/compress/optimize/strings.rs

@@ -137,4 +137,29 @@ impl Optimizer<'_> {
            _ => {}
        }
    }
+
+    pub(super) fn reduce_escaped_newline_for_str_lit(&mut self, expr: &mut Expr) {


Do we still need this, even with the codegen config?

This is used to transform string literal to template literal. Maybe should rename the method name, it's not clear enough.

Document added.

kdy1 · 2025-03-23T12:49:15Z

crates/swc_ecma_minifier/src/compress/optimize/strings.rs

+                    quasis: vec![TplElement {
+                        span: s.span,
+                        cooked: Some(s.value.clone()),
+                        raw: s.value.clone(),


This may cause an issue.

The function for Tpl#raw => Str#value is at https://rustdoc.swc.rs/swc_core/ecma/utils/swc_ecma_ast/struct.Str.html#method.from_tpl_raw , but let me search for Str => Tpl function...

You should use functions at

swc/crates/swc_ecma_minifier/src/compress/pure/strings.rs

Lines 573 to 593 in f33b0bc

pub(super) fn convert_str_value_to_tpl_cooked(value: &Atom) -> Cow<str> {

value

.replace("\\\\", "\\")

.replace("\\`", "`")

.replace("\\$", "$")

.into()

}

pub(super) fn convert_str_value_to_tpl_raw(value: &Atom) -> Cow<str> {

value

.replace('\\', "\\\\")

.replace('`', "\\`")

.replace('$', "\\$")

.replace('\n', "\\n")

.replace('\r', "\\r")

.into()

}

pub(super) fn convert_str_raw_to_tpl_raw(value: &str) -> Atom {

value.replace('`', "\\`").replace('$', "\\$").into()

}

This function replace '\n' to "\\n" for Tpl#raw, but I just want to preserve the '\n'. It seems little bit redundant if replace "\\n" back to '\n'.

If so, can you add test cases for inputs that contain those strings, respectively?

One test per the search text in "", like one for "\$" in convert_str_value_to_tpl_cooked (you need 3 test cases for the function) and "$" for convert_str_value_to_tpl_raw?

In pure/strings.rs and pure/misc.rs, it appears to be invoked with Str#value...

I use a new fn to convert Str#value to Tpl#raw without replacing for \n and test for some edge cases.

Does the original function need to be fixed in this PR?

It’s fine to skip for this PR unless there’s a test failure

Any other suggestions for this review(or above)?

crates/swc_ecma_minifier/src/option/terser.rs

...ecma_minifier/tests/fixture/next/target-es2015/static/chunks/main-04b5934c26266542/output.js

…ed_newline

kdy1

Other things look good. Only one concern about consistency. Thanks!

kdy1 · 2025-03-30T09:51:43Z

crates/swc_ecma_minifier/src/option/terser.rs

+#[non_exhaustive]
+pub struct TerserExperimentalOptions {
+    #[serde(default = "true_by_default")]
+    pub reduce_escaped_newline: bool,


Suggested change

pub reduce_escaped_newline: bool,

pub reduce_escaped_newline: Option<bool>,

Sorry, I missed this one in the previous review.

true_by_default with bool means

experimental: {} => true, regardless of defaults

experimental: null => Uses defaults

I think those two should behave identically because the user didn't specify the reduce_escaped_newline

It's my oversight. Thanks to point out.

kdy1

Thank you!

feat(es/minifier): support reduce_escaped_newline

f83f8fd

wjw99830 requested a review from a team as a code owner March 19, 2025 16:58

fix: add mangleOnly output

1a52251

wjw99830 added 2 commits March 23, 2025 14:58

feat(es/minifier): support `compress.experimental.reduce_escaped_newl…

3d3ec93

…ine` for minifier and `reduce_escaped_newline` for codegen

chore: fix ci failure

f5efd95

kdy1 self-assigned this Mar 23, 2025

kdy1 added this to the Planned milestone Mar 23, 2025

kdy1 requested changes Mar 23, 2025

View reviewed changes

chore: document

1c1b38b

kdy1 reviewed Mar 23, 2025

View reviewed changes

wjw99830 and others added 4 commits March 25, 2025 22:42

fix: cover more cases for converting Str#value to Tpl#raw

4652a34

chore: fmt

59bae0c

Merge branch 'main' of github.com:wjw99830/swc into feat/reduce_escap…

549379d

…ed_newline

fix: merge conflicts

1c17d58

kdy1 reviewed Mar 30, 2025

View reviewed changes

fix: consistency

2570efd

kdy1 previously approved these changes Mar 31, 2025

View reviewed changes

Create rotten-eggs-sparkle.md

ea90d46

kdy1 dismissed their stale review via ea90d46 March 31, 2025 02:27

kdy1 requested a review from a team as a code owner March 31, 2025 02:27

kdy1 merged commit 64fb286 into swc-project:main Mar 31, 2025
18 checks passed

kdy1 modified the milestones: Planned, v1.11.14 Mar 31, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(es/minifier): support `reduce_escaped_newline` #10232

feat(es/minifier): support `reduce_escaped_newline` #10232

wjw99830 commented Mar 19, 2025 •

edited

Loading

changeset-bot bot commented Mar 19, 2025 •

edited

Loading

codspeed-hq bot commented Mar 19, 2025 •

edited

Loading

kdy1 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025 •

edited

Loading

kdy1 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025

kdy1 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025

kdy1 Mar 23, 2025

wjw99830 Mar 23, 2025

wjw99830 Mar 23, 2025

kdy1 Mar 23, 2025

kdy1 Mar 23, 2025

kdy1 Mar 23, 2025

wjw99830 Mar 23, 2025

kdy1 Mar 23, 2025 •

edited

Loading

wjw99830 Mar 24, 2025

wjw99830 Mar 29, 2025

wjw99830 Mar 29, 2025

kdy1 Mar 29, 2025 •

edited

Loading

wjw99830 Mar 30, 2025

kdy1 left a comment

kdy1 Mar 30, 2025

kdy1 Mar 30, 2025

kdy1 Mar 30, 2025

wjw99830 Mar 30, 2025

kdy1 left a comment

	pub(super) fn convert_str_value_to_tpl_cooked(value: &Atom) -> Cow<str> {
	value
	.replace("\\\\", "\\")
	.replace("\\`", "`")
	.replace("\\$", "$")
	.into()
	}

	pub(super) fn convert_str_value_to_tpl_raw(value: &Atom) -> Cow<str> {
	value
	.replace('\\', "\\\\")
	.replace('`', "\\`")
	.replace('$', "\\$")
	.replace('\n', "\\n")
	.replace('\r', "\\r")
	.into()
	}

	pub(super) fn convert_str_raw_to_tpl_raw(value: &str) -> Atom {
	value.replace('`', "\\`").replace('$', "\\$").into()
	}

	pub reduce_escaped_newline: bool,
	pub reduce_escaped_newline: Option<bool>,

feat(es/minifier): support reduce_escaped_newline #10232

feat(es/minifier): support reduce_escaped_newline #10232

Conversation

wjw99830 commented Mar 19, 2025 • edited Loading

changeset-bot bot commented Mar 19, 2025 • edited Loading

🦋 Changeset detected

codspeed-hq bot commented Mar 19, 2025 • edited Loading

Merging #10232 will improve performances by 1.32%

Summary

Benchmarks breakdown

kdy1 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025 • edited Loading

kdy1 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025

kdy1 commented Mar 20, 2025

wjw99830 commented Mar 20, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kdy1 Mar 23, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kdy1 Mar 29, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kdy1 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kdy1 left a comment

Choose a reason for hiding this comment

feat(es/minifier): support `reduce_escaped_newline` #10232

feat(es/minifier): support `reduce_escaped_newline` #10232

wjw99830 commented Mar 19, 2025 •

edited

Loading

changeset-bot bot commented Mar 19, 2025 •

edited

Loading

codspeed-hq bot commented Mar 19, 2025 •

edited

Loading

wjw99830 commented Mar 20, 2025 •

edited

Loading

kdy1 Mar 23, 2025 •

edited

Loading

kdy1 Mar 29, 2025 •

edited

Loading