feat(oxc_codegen): preserve annotate comment (#3465)

1. Copy tests from
https://github.com/evanw/esbuild/blob/efa3dd2d8e895f7f9a9bef0d588560bbae7d776e/internal/bundler_tests/bundler_dce_test.go#L3833-L3971
2. Add option to preserve annotate comment like `/* #__NO_SIDE_EFFECTS__
*/` and `/* #__PURE__ */`

Cargo.toml

@@ -176,6 +176,7 @@ cfg-if              = "1.0.0"
 schemars            = "0.8.21"
 oxc-browserslist    = "0.16.1"
 criterion2          = { version = "0.9.0", default-features = false }
+daachorse                 = { version = "1.0.0" }
 
 [workspace.metadata.cargo-shear]
 ignored = ["napi", "oxc_traverse"]

crates/oxc_codegen/Cargo.toml

@@ -25,7 +25,9 @@ oxc_allocator = { workspace = true }
 oxc_syntax    = { workspace = true }
 oxc_sourcemap = { workspace = true }
 bitflags      = { workspace = true }
-
+once_cell     = { workspace = true }
+daachorse     = { workspace = true }
+rustc-hash    = { workspace = true }
 [dev-dependencies]
 oxc_parser = { workspace = true }
 base64     = { workspace = true }

crates/oxc_codegen/examples/codegen.rs

@@ -28,14 +28,16 @@ fn main() -> std::io::Result<()> {
     println!("Original:");
     println!("{source_text}");
 
-    let options = CodegenOptions { enable_source_map: false, enable_typescript: true };
+    let options =
+        CodegenOptions { enable_source_map: false, enable_typescript: true, ..Default::default() };
     let printed =
-        Codegen::<false>::new("", &source_text, options.clone()).build(&ret.program).source_text;
+        Codegen::<false>::new("", &source_text, options, None).build(&ret.program).source_text;
     println!("Printed:");
     println!("{printed}");
 
     let ret = Parser::new(&allocator, &printed, source_type).parse();
-    let minified = Codegen::<true>::new("", &source_text, options).build(&ret.program).source_text;
+    let minified =
+        Codegen::<true>::new("", &source_text, options, None).build(&ret.program).source_text;
     println!("Minified:");
     println!("{minified}");
 

crates/oxc_codegen/examples/sourcemap.rs

@@ -26,10 +26,11 @@ fn main() -> std::io::Result<()> {
         return Ok(());
     }
 
-    let codegen_options = CodegenOptions { enable_source_map: true, enable_typescript: true };
+    let codegen_options =
+        CodegenOptions { enable_source_map: true, enable_typescript: true, ..Default::default() };
 
     let CodegenReturn { source_text, source_map } =
-        Codegen::<false>::new(path.to_string_lossy().as_ref(), &source_text, codegen_options)
+        Codegen::<false>::new(path.to_string_lossy().as_ref(), &source_text, codegen_options, None)
             .build(&ret.program);
 
     if let Some(source_map) = source_map {

crates/oxc_codegen/src/annotation_comment.rs

@@ -0,0 +1,70 @@
+use std::usize;
+
+use daachorse::DoubleArrayAhoCorasick;
+use once_cell::sync::Lazy;
+use oxc_ast::{Comment, CommentKind};
+
+use crate::Codegen;
+static MATCHER: Lazy<DoubleArrayAhoCorasick<usize>> = Lazy::new(|| {
+    let patterns = vec!["#__NO_SIDE_EFFECTS__", "@__NO_SIDE_EFFECTS__", "@__PURE__", "#__PURE__"];
+
+    DoubleArrayAhoCorasick::new(patterns).unwrap()
+});
+
+pub fn get_leading_annotate_comment<const MINIFY: bool>(
+    node_start: u32,
+    codegen: &mut Codegen<{ MINIFY }>,
+) -> Option<(u32, Comment)> {
+    let maybe_leading_comment = codegen.try_get_leading_comment(node_start);
+    let (comment_start, comment) = maybe_leading_comment?;
+    let real_end = match comment.kind {
+        CommentKind::SingleLine => comment.end,
+        CommentKind::MultiLine => comment.end + 2,
+    };
+    let source_code = codegen.source_code;
+    let content_between = &source_code[real_end as usize..node_start as usize];
+    // Used for VariableDeclaration (Rollup only respects "const" and only for the first one)
+    if content_between.chars().all(|ch| ch.is_ascii_whitespace()) {
+        let comment_content = &source_code[*comment_start as usize..comment.end as usize];
+        if MATCHER.find_iter(&comment_content).next().is_some() {
+            return Some((*comment_start, *comment));
+        }
+        None
+    } else {
+        None
+    }
+}
+
+pub fn print_comment<const MINIFY: bool>(
+    comment_start: u32,
+    comment: Comment,
+    p: &mut Codegen<{ MINIFY }>,
+) {
+    match comment.kind {
+        CommentKind::SingleLine => {
+            p.print_str("//");
+            p.print_range_of_source_code(comment_start as usize..comment.end as usize);
+            p.print_soft_newline();
+            p.print_indent();
+        }
+        CommentKind::MultiLine => {
+            p.print_str("/*");
+            p.print_range_of_source_code(comment_start as usize..comment.end as usize);
+            p.print_str("*/");
+            p.print_soft_space();
+        }
+    }
+}
+
+pub fn gen_comment<const MINIFY: bool>(node_start: u32, codegen: &mut Codegen<{ MINIFY }>) {
+    if !codegen.options.preserve_annotate_comments {
+        return;
+    }
+    if let Some((comment_start, comment)) = codegen.try_take_moved_comment(node_start) {
+        print_comment::<MINIFY>(comment_start, comment, codegen);
+    }
+    let maybe_leading_annotate_comment = get_leading_annotate_comment(node_start, codegen);
+    if let Some((comment_start, comment)) = maybe_leading_annotate_comment {
+        print_comment::<MINIFY>(comment_start, comment, codegen);
+    }
+}

crates/oxc_codegen/src/gen.rs

@@ -1,3 +1,4 @@
+use crate::annotation_comment::{gen_comment, get_leading_annotate_comment};
 use oxc_allocator::{Box, Vec};
 #[allow(clippy::wildcard_imports)]
 use oxc_ast::ast::*;
@@ -29,6 +30,11 @@ where
     }
 }
 
+/// the [GenComment] trait only generate annotate comments like `/* @__PURE__ */` and `/* @__NO_SIDE_EFFECTS__ */`.
+pub trait GenComment<const MINIFY: bool> {
+    fn gen_comment(&self, _p: &mut Codegen<{ MINIFY }>, _ctx: Context) {}
+}
+
 impl<'a, const MINIFY: bool, T> GenExpr<MINIFY> for Box<'a, T>
 where
     T: GenExpr<MINIFY>,
@@ -611,10 +617,24 @@ impl<'a, const MINIFY: bool> Gen<MINIFY> for VariableDeclaration<'a> {
         if p.options.enable_typescript && self.modifiers.contains(ModifierKind::Declare) {
             p.print_str(b"declare ");
         }
+
+        if p.options.preserve_annotate_comments
+            && matches!(self.kind, VariableDeclarationKind::Const)
+        {
+            if let Some(declarator) = self.declarations.first() {
+                if let Some(ref init) = declarator.init {
+                    if let Some(leading_annotate_comment) =
+                        get_leading_annotate_comment(self.span.start, p)
+                    {
+                        p.move_comment(init.span().start, leading_annotate_comment);
+                    }
+                }
+            }
+        }
         p.print_str(match self.kind {
-            VariableDeclarationKind::Const => b"const",
-            VariableDeclarationKind::Let => b"let",
-            VariableDeclarationKind::Var => b"var",
+            VariableDeclarationKind::Const => "const",
+            VariableDeclarationKind::Let => "let",
+            VariableDeclarationKind::Var => "var",
         });
         if !self.declarations.is_empty() {
             p.print_hard_space();
@@ -637,6 +657,7 @@ impl<'a, const MINIFY: bool> Gen<MINIFY> for VariableDeclarator<'a> {
 impl<'a, const MINIFY: bool> Gen<MINIFY> for Function<'a> {
     fn gen(&self, p: &mut Codegen<{ MINIFY }>, ctx: Context) {
         p.add_source_mapping(self.span.start);
+        self.gen_comment(p, ctx);
         if !p.options.enable_typescript && self.is_typescript_syntax() {
             return;
         }
@@ -847,6 +868,27 @@ impl<'a, const MINIFY: bool> Gen<MINIFY> for ExportNamedDeclaration<'a> {
         if !p.options.enable_typescript && self.is_typescript_syntax() {
             return;
         }
+        if p.options.preserve_annotate_comments {
+            match &self.declaration {
+                Some(Declaration::FunctionDeclaration(_)) => {
+                    gen_comment(self.span.start, p);
+                }
+                Some(Declaration::VariableDeclaration(var_decl))
+                    if matches!(var_decl.kind, VariableDeclarationKind::Const) =>
+                {
+                    if let Some(declarator) = var_decl.declarations.first() {
+                        if let Some(ref init) = declarator.init {
+                            if let Some(leading_annotate_comment) =
+                                get_leading_annotate_comment(self.span.start, p)
+                            {
+                                p.move_comment(init.span().start, leading_annotate_comment);
+                            }
+                        }
+                    }
+                }
+                _ => {}
+            };
+        }
         p.print_str(b"export ");
         if p.options.enable_typescript && self.export_kind.is_type() {
             p.print_str(b"type ");
@@ -1009,6 +1051,18 @@ impl<'a, const MINIFY: bool> GenExpr<MINIFY> for Expression<'a> {
     }
 }
 
+impl<const MINIFY: bool> GenComment<MINIFY> for ArrowFunctionExpression<'_> {
+    fn gen_comment(&self, codegen: &mut Codegen<{ MINIFY }>, _ctx: Context) {
+        gen_comment(self.span.start, codegen);
+    }
+}
+
+impl<const MINIFY: bool> GenComment<MINIFY> for Function<'_> {
+    fn gen_comment(&self, codegen: &mut Codegen<{ MINIFY }>, _ctx: Context) {
+        gen_comment(self.span.start, codegen);
+    }
+}
+
 impl<'a, const MINIFY: bool> GenExpr<MINIFY> for TSAsExpression<'a> {
     fn gen_expr(&self, p: &mut Codegen<{ MINIFY }>, precedence: Precedence, ctx: Context) {
         if p.options.enable_typescript {
@@ -1538,6 +1592,7 @@ impl<'a, const MINIFY: bool> Gen<MINIFY> for PropertyKey<'a> {
 impl<'a, const MINIFY: bool> GenExpr<MINIFY> for ArrowFunctionExpression<'a> {
     fn gen_expr(&self, p: &mut Codegen<{ MINIFY }>, precedence: Precedence, ctx: Context) {
         p.wrap(precedence > Precedence::Assign, |p| {
+            self.gen_comment(p, ctx);
             if self.r#async {
                 p.add_source_mapping(self.span.start);
                 p.print_str(b"async");