fix(es/ast): Fix unicode lone surrogates handling

.changeset/metal-radios-swim.md

@@ -0,0 +1,5 @@
+---
+swc_ecma_ast: major
+---
+
+fix(es/ast): Fix unicode lone surrogates handling

bindings/binding_core_node/src/bundle.rs

@@ -281,6 +281,7 @@ impl swc_core::bundler::Hook for Hook {
                     span,
                     raw: None,
                     value: file_name.into(),
+                    lone_surrogates: false,
                 }))),
             },
             KeyValueProp {

crates/swc/tests/fixture/issues-7xxx/7678/output/1.js

@@ -1 +1 @@
-let str="\uD83D\uDC68\\u200D\uD83D\uDE80";let obj={"\uD83D\uDC68\\u200D\uD83D\uDE80":"wrong"};
+let str="\\uD83D\\uDC68\\u200D\\uD83D\\uDE80";let obj={"\\uD83D\\uDC68\\u200D\\uD83D\\uDE80":"wrong"};

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings10_ES5.1.normal.js

@@ -3,4 +3,4 @@
 //  2. Let cu1 be floor((cp – 65536) / 1024) + 0xD800.
 // Although we should just get back a single code point value of 0xD800,
 // this is a useful edge-case test.
-var x = "\u{D800}";
+var x = "\ud800";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings10_ES6.1.normal.js

@@ -3,4 +3,4 @@
 //  2. Let cu1 be floor((cp – 65536) / 1024) + 0xD800.
 // Although we should just get back a single code point value of 0xD800,
 // this is a useful edge-case test.
-var x = "\u{D800}";
+var x = "\ud800";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings11_ES5.1.normal.js

@@ -3,4 +3,4 @@
 //  2. Let cu2 be ((cp – 65536) modulo 1024) + 0xDC00.
 // Although we should just get back a single code point value of 0xDC00,
 // this is a useful edge-case test.
-var x = "\u{DC00}";
+var x = "\udc00";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings11_ES6.1.normal.js

@@ -3,4 +3,4 @@
 //  2. Let cu2 be ((cp – 65536) modulo 1024) + 0xDC00.
 // Although we should just get back a single code point value of 0xDC00,
 // this is a useful edge-case test.
-var x = "\u{DC00}";
+var x = "\udc00";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates10_ES5.1.normal.js

@@ -3,4 +3,4 @@
 //  2. Let cu1 be floor((cp – 65536) / 1024) + 0xD800.
 // Although we should just get back a single code point value of 0xD800,
 // this is a useful edge-case test.
-var x = "\u{D800}";
+var x = "\ud800";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates11_ES5.1.normal.js

@@ -3,4 +3,4 @@
 //  2. Let cu2 be ((cp – 65536) modulo 1024) + 0xDC00.
 // Although we should just get back a single code point value of 0xDC00,
 // this is a useful edge-case test.
-var x = "\u{DC00}";
+var x = "\udc00";

crates/swc/tests/vercel/full/utf8-1/output/index.js

@@ -5,7 +5,7 @@ import r from './on-demand-entries-client';
 import { addMessageListener as n, connectHMR as c } from './error-overlay/websocket';
 var o = JSON.parse(document.getElementById('__NEXT_DATA__').textContent);
 window.__NEXT_DATA__ = o;
-var s = o.assetPrefix, i = o.page, _ = null, u = __webpack_hash__, d = (s = s || '') + (s.endsWith('/') ? '' : '/') + '_next/static/webpack/';
+var s = o.assetPrefix, i = o.page, _ = null, d = __webpack_hash__, u = (s = s || '') + (s.endsWith('/') ? '' : '/') + '_next/static/webpack/';
 n(function(a) {
     if ('\uD83D\uDC93' !== a.data) try {
         var r = JSON.parse(a.data);
@@ -28,7 +28,7 @@ n(function(a) {
                                 5
                             ]), [
                                 4,
-                                fetch('undefined' != typeof __webpack_runtime_id__ ? "".concat(d).concat(u, ".").concat(__webpack_runtime_id__, ".hot-update.json") : "".concat(d).concat(u, ".hot-update.json"))
+                                fetch('undefined' != typeof __webpack_runtime_id__ ? "".concat(u).concat(d, ".").concat(__webpack_runtime_id__, ".hot-update.json") : "".concat(u).concat(d, ".hot-update.json"))
                             ];
                         case 2:
                             return [
@@ -38,7 +38,7 @@ n(function(a) {
                         case 3:
                             return e = t.sent(), a = '/' === i ? 'index' : i, (Array.isArray(e.c) ? e.c : Object.keys(e.c)).some(function(e) {
                                 return -1 !== e.indexOf("pages".concat(a.startsWith('/') ? a : "/".concat(a))) || -1 !== e.indexOf("pages".concat(a.startsWith('/') ? a : "/".concat(a)).replace(/\//g, '\\'));
-                            }) ? document.location.reload(!0) : u = _, [
+                            }) ? document.location.reload(!0) : d = _, [
                                 3,
                                 5
                             ];

crates/swc_bundler/examples/bundle.rs

@@ -195,6 +195,7 @@ impl swc_bundler::Hook for Hook {
                     span,
                     raw: None,
                     value: file_name.into(),
+                    lone_surrogates: false,
                 }))),
             },
             KeyValueProp {

crates/swc_bundler/src/util.rs

@@ -33,6 +33,7 @@ pub(crate) trait VarDeclaratorExt: Into<VarDeclarator> {
                         span: DUMMY_SP,
                         raw: None,
                         value: name.into(),
+                        lone_surrogates: false,
                     }
                     .assign_to(Ident::new_no_ctxt(atom!("INJECTED_FROM"), DUMMY_SP)),
                 ]
@@ -193,6 +194,7 @@ fn metadata(key: &str, value: &str) -> PropOrSpread {
             span: DUMMY_SP,
             value: value.into(),
             raw: None,
+            lone_surrogates: false,
         })
         .into(),
     })))

crates/swc_bundler/tests/deno.rs

@@ -1120,6 +1120,7 @@ impl swc_bundler::Hook for Hook {
                     span,
                     raw: None,
                     value: file_name.into(),
+                    lone_surrogates: false,
                 }))),
             },
             KeyValueProp {

crates/swc_bundler/tests/fixture.rs

@@ -153,6 +153,7 @@ impl swc_bundler::Hook for Hook {
                     span,
                     raw: None,
                     value: file_name.into(),
+                    lone_surrogates: false,
                 }))),
             },
             KeyValueProp {

crates/swc_ecma_ast/src/expr.rs

@@ -1199,6 +1199,15 @@ pub struct TplElement {
     /// You may need to perform. `.replace("\r\n", "\n").replace('\r', "\n")` on
     /// this value.
     pub raw: Atom,
+
+    /// The string value contains lone surrogates.
+    ///
+    /// `cooked` is encoded with `\u{FFFD}` to mark the lone surrogate as an
+    /// escaped value.
+    ///
+    /// For example, a "\uD808" is a lone surrogate, and it's encoded as
+    /// `\u{FFFD}D808`.
+    pub lone_surrogates: bool,
 }
 
 impl Take for TplElement {
@@ -1208,6 +1217,7 @@ impl Take for TplElement {
             tail: Default::default(),
             cooked: None,
             raw: Default::default(),
+            lone_surrogates: false,
         }
     }
 }
@@ -1219,12 +1229,14 @@ impl<'a> arbitrary::Arbitrary<'a> for TplElement {
         let span = u.arbitrary()?;
         let cooked = Some(u.arbitrary::<String>()?.into());
         let raw = u.arbitrary::<String>()?.into();
+        let lone_surrogates = u.arbitrary::<bool>()?.into();
 
         Ok(Self {
             span,
             tail: false,
             cooked,
             raw,
+            lone_surrogates,
         })
     }
 }

crates/swc_ecma_ast/src/lib.rs

@@ -571,7 +571,7 @@ mod rkyv_layout_assert {
     assert_size!(ArchivedTaggedTpl, 28);
     assert_size!(ArchivedThisExpr, 8);
     assert_size!(ArchivedTpl, 24);
-    assert_size!(ArchivedTplElement, 32);
+    assert_size!(ArchivedTplElement, 36);
     assert_size!(ArchivedUnaryExpr, 16);
     assert_size!(ArchivedUpdateExpr, 16);
     assert_size!(ArchivedYieldExpr, 20);
@@ -616,7 +616,7 @@ mod rkyv_layout_assert {
     assert_size!(ArchivedNull, 8);
     assert_size!(ArchivedNumber, 32);
     assert_size!(ArchivedRegex, 24);
-    assert_size!(ArchivedStr, 28);
+    assert_size!(ArchivedStr, 32);
 
     // Module declaration types
     assert_size!(ArchivedDefaultDecl, 36);
@@ -625,16 +625,16 @@ mod rkyv_layout_assert {
     assert_size!(ArchivedExportDefaultDecl, 44);
     assert_size!(ArchivedExportDefaultExpr, 12);
     assert_size!(ArchivedExportDefaultSpecifier, 24);
-    assert_size!(ArchivedExportNamedSpecifier, 80);
-    assert_size!(ArchivedExportNamespaceSpecifier, 40);
-    assert_size!(ArchivedExportSpecifier, 84);
+    assert_size!(ArchivedExportNamedSpecifier, 88);
+    assert_size!(ArchivedExportNamespaceSpecifier, 44);
+    assert_size!(ArchivedExportSpecifier, 92);
     assert_size!(ArchivedImportDecl, 36);
     assert_size!(ArchivedImportDefaultSpecifier, 32);
-    assert_size!(ArchivedImportNamedSpecifier, 72);
-    assert_size!(ArchivedImportSpecifier, 76);
+    assert_size!(ArchivedImportNamedSpecifier, 76);
+    assert_size!(ArchivedImportSpecifier, 80);
     assert_size!(ArchivedImportStarAsSpecifier, 32);
     assert_size!(ArchivedModuleDecl, 48);
-    assert_size!(ArchivedModuleExportName, 32);
+    assert_size!(ArchivedModuleExportName, 36);
     assert_size!(ArchivedNamedExport, 36);
 
     // Operator types
@@ -700,17 +700,17 @@ mod rkyv_layout_assert {
     assert_size!(ArchivedTsConstructorType, 32);
     assert_size!(ArchivedTsEntityName, 28);
     assert_size!(ArchivedTsEnumDecl, 44);
-    assert_size!(ArchivedTsEnumMember, 48);
-    assert_size!(ArchivedTsEnumMemberId, 32);
+    assert_size!(ArchivedTsEnumMember, 52);
+    assert_size!(ArchivedTsEnumMemberId, 36);
     assert_size!(ArchivedTsExportAssignment, 12);
     assert_size!(ArchivedTsExprWithTypeArgs, 20);
-    assert_size!(ArchivedTsExternalModuleRef, 36);
+    assert_size!(ArchivedTsExternalModuleRef, 40);
     assert_size!(ArchivedTsFnOrConstructorType, 36);
     assert_size!(ArchivedTsFnParam, 36);
     assert_size!(ArchivedTsFnType, 28);
     assert_size!(ArchivedTsGetterSignature, 24);
-    assert_size!(ArchivedTsImportEqualsDecl, 76);
-    assert_size!(ArchivedTsImportType, 92);
+    assert_size!(ArchivedTsImportEqualsDecl, 80);
+    assert_size!(ArchivedTsImportType, 96);
     assert_size!(ArchivedTsIndexSignature, 28);
     assert_size!(ArchivedTsIndexedAccessType, 20);
     assert_size!(ArchivedTsInferType, 60);
@@ -725,9 +725,9 @@ mod rkyv_layout_assert {
     assert_size!(ArchivedTsMappedType, 84);
     assert_size!(ArchivedTsMethodSignature, 40);
     assert_size!(ArchivedTsModuleBlock, 16);
-    assert_size!(ArchivedTsModuleDecl, 92);
-    assert_size!(ArchivedTsModuleName, 32);
-    assert_size!(ArchivedTsModuleRef, 40);
+    assert_size!(ArchivedTsModuleDecl, 96);
+    assert_size!(ArchivedTsModuleName, 36);
+    assert_size!(ArchivedTsModuleRef, 44);
     assert_size!(ArchivedTsNamespaceBody, 44);
     assert_size!(ArchivedTsNamespaceDecl, 40);
     assert_size!(ArchivedTsNamespaceExportDecl, 32);
@@ -758,8 +758,8 @@ mod rkyv_layout_assert {
     assert_size!(ArchivedTsTypeParamDecl, 16);
     assert_size!(ArchivedTsTypeParamInstantiation, 16);
     assert_size!(ArchivedTsTypePredicate, 48);
-    assert_size!(ArchivedTsTypeQuery, 112);
-    assert_size!(ArchivedTsTypeQueryExpr, 96);
+    assert_size!(ArchivedTsTypeQuery, 116);
+    assert_size!(ArchivedTsTypeQueryExpr, 100);
     assert_size!(ArchivedTsTypeRef, 44);
     assert_size!(ArchivedTsUnionOrIntersectionType, 20);
     assert_size!(ArchivedTsUnionType, 16);

crates/swc_ecma_ast/src/lit.rs

@@ -190,6 +190,15 @@ pub struct Str {
     /// Use `None` value only for transformations to avoid recalculate escaped
     /// characters in strings
     pub raw: Option<Atom>,
+
+    /// The string value contains lone surrogates.
+    ///
+    /// `value` is encoded with `\u{FFFD}` to mark the lone surrogate as an
+    /// escaped value.
+    ///
+    /// For example, a "\uD808" is a lone surrogate, and it's encoded as
+    /// `\u{FFFD}D808`.
+    pub lone_surrogates: bool,
 }
 
 impl Take for Str {
@@ -198,6 +207,7 @@ impl Take for Str {
             span: DUMMY_SP,
             value: atom!(""),
             raw: None,
+            lone_surrogates: false,
         }
     }
 }
@@ -209,8 +219,14 @@ impl<'a> arbitrary::Arbitrary<'a> for Str {
         let span = u.arbitrary()?;
         let value = u.arbitrary::<String>()?.into();
         let raw = Some(u.arbitrary::<String>()?.into());
+        let lone_surrogates = u.arbitrary()?;
 
-        Ok(Self { span, value, raw })
+        Ok(Self {
+            span,
+            value,
+            raw,
+            lone_surrogates,
+        })
     }
 }
 
@@ -282,6 +298,7 @@ impl From<Atom> for Str {
             span: DUMMY_SP,
             value,
             raw: None,
+            lone_surrogates: false,
         }
     }
 }