fix(es/codegen): Fix codegen of unicode surrogates (#7985)

**Description:**


x-ref: https://vercel.slack.com/archives/C02HY34AKME/p1695334071194139


Reproduction: https://github.com/kdy1/repro-test-mdx-korean



Regression of next.js: `v13.4.11`(swc_core@v0.79.13) => `v13.5.2`
(swc_core@v0.83.12)


 - `next@v13.4.19` works (swc_core@v0.79.59)
 - `next@v13.4.20-canary.3` works (swc_core@v0.79.70)
 -  **`next@v13.4.20-canary.32` fails** (swc_core@v0.83.12)

Commit range:
662f236...e67bf05

- `swc_core@v0.79.70`:
662f236
- `swc_core@v0.83.12`:
e67bf05

Diff: https://gist.github.com/kdy1/047e7e5537c34180d446cb3d5b95fce8

---

I did more investigation by monkey-patching the `next` package.
`.minify()` resolved without an exception.
It means that `.minify()` is producing an invalid ES code.

So... candidates are

 - #7890
 - #7876
 - #7858
 - #7856
 - #7853
 - #7832

crates/swc/tests/exec.rs

@@ -372,8 +372,11 @@ fn test_file_with_opts(
                     idx,
                     res.code
                 );
-                println!("external_helpers: {:?}", opts.config.jsc.external_helpers);
+
                 println!("target: {:?}", opts.config.jsc.target.unwrap());
+                println!("minify: {:?}", opts.config.jsc.minify.is_some());
+                println!("external_helpers: {:?}", opts.config.jsc.external_helpers);
+                println!("source_map: {:?}", opts.source_maps.is_some());
 
                 let actual_stdout = stdout_of(
                     &res.code,

crates/swc_ecma_codegen/src/lib.rs

@@ -3850,13 +3850,34 @@ fn get_quoted_utf16(v: &str, ascii_only: bool, target: EsVersion) -> String {
                                 inner_buf.push('}');
                             }
 
+                            let range = if is_curly {
+                                3..(inner_buf.len() - 1)
+                            } else {
+                                2..6
+                            };
+
+                            let val_str = &inner_buf[range];
+
                             if is_valid {
-                                buf.push_str(&inner_buf);
+                                let v = u32::from_str_radix(val_str, 16).unwrap_or_else(|err| {
+                                    unreachable!(
+                                        "failed to parse {} as a hex value: {:?}",
+                                        val_str, err
+                                    )
+                                });
 
-                                let end = if is_curly { 7 } else { 5 };
+                                if v > 0xffff {
+                                    buf.push_str(&inner_buf);
 
-                                for _ in 0..end {
-                                    iter.next();
+                                    let end = if is_curly { 7 } else { 5 };
+
+                                    for _ in 0..end {
+                                        iter.next();
+                                    }
+                                } else if (0xd800..=0xdfff).contains(&v) {
+                                    buf.push('\\');
+                                } else {
+                                    buf.push_str("\\\\");
                                 }
                             }
                         } else if is_curly {

crates/swc_ecma_codegen/src/tests.rs

@@ -1,11 +1,13 @@
 use std::{
     fmt::Debug,
     io::Write,
+    path::PathBuf,
     sync::{Arc, RwLock},
 };
 
 use swc_common::{comments::SingleThreadedComments, FileName, SourceMap};
 use swc_ecma_parser;
+use swc_ecma_testing::{exec_node_js, JsExecOptions};
 use testing::DebugUsingDisplay;
 
 use self::swc_ecma_parser::{EsConfig, Parser, StringInput, Syntax};
@@ -889,3 +891,39 @@ export default {
         },
     );
 }
+
+#[testing::fixture("tests/str-lits/**/*.txt")]
+fn test_str_lit(input: PathBuf) {
+    test_str_lit_inner(input)
+}
+
+/// Print text back using `console.log`
+fn run_node(code: &str) -> String {
+    exec_node_js(
+        code,
+        JsExecOptions {
+            cache: true,
+            module: false,
+            args: vec![],
+        },
+    )
+    .expect("failed to execute node.js")
+}
+
+fn test_str_lit_inner(input: PathBuf) {
+    let raw_str = std::fs::read_to_string(&*input).unwrap();
+    let input_code = format!("console.log('{raw_str}')");
+
+    let output_code = parse_then_emit(
+        &input_code,
+        Config::default()
+            .with_target(EsVersion::latest())
+            .with_minify(true),
+        Syntax::default(),
+    );
+
+    let expected = run_node(&input_code);
+    let actual = run_node(&output_code);
+
+    assert_eq!(actual, expected);
+}

crates/swc_ecma_codegen/tests/str-lits/pr-7985/full.txt

@@ -0,0 +1 @@
+****:한국어TürkçeTurkish\\n-**uk**:УкраїнськаUkrainian\\n-**vi**:TiếngViệtVietnamese\\n-**zh-cn**:简体中文SimplifiedChinese\\n-**zh-tw**:繁體中文TraditionalChinese\\n\\n\\n\\n##Updatetranslatedstrings\\n\\nIfyouwant`fr`locales|`Espa\\\\xF1ol(Latinoam\\\\xE9rica)Spanish(LatinAmerica)\\"}),`\\n`,(n\\"fi\\"}),\\":SuomiFinnish\\"}),`\\n`,(n\\"fr\\"}),\\":Fran\\\\xE7aisFrench\\"}),`\\n`,(n\\"fr-be\\"}),\\":Fran\\\\xE7ais(Belgique)French(Belgium)\\"}),`\\n`,(n\\"fr-ca\\"}),\\":Fran\\\\xE7ais(Canada)French(Canada)\\"}),`\\n`,(n\\"fr-ch\\"}),\\":Fran\\\\xE7ais(Suisse)French(Switzerland)\\"}),`\\n`,(n\\"fr-fr\\"}),\\":Fran\\\\xE7ais(France)French(France)\\"}),`\\n`,(n\\"hu\\"}),\\":MagyarHungarian\\"}),`\\n`,(n\\"id\\"}),\\":BahasaIndonesiaIndonesian\\"}),`\\n`,(n\\"it\\"}),\\":ItalianoItalian\\"}),`\\n`,(n\\"ja\\"}),\\":\\\\u65E5\\\\u672C\\\\u8A9EJapanese\\"}),`\\n`,(n\\"ko\\"}),\\":\\\\uD55C\\\\uAD6D\\\\uC5B4Korean\\"}),`\\n`,(n\\"nl\\"}),\\":NederlandsDutch\\"}),`\\n`,(n\\"nl-be\\"}),\\":Nederlands(Belgi\\\\xEB)Dutch(Belgium)\\"}),`\\n`,(n\\"no\\"}),\\":NorskNorwegian\\"}),`\\n`,(n\\"pl\\"}),\\":(n\\"th\\"}),\\":\\\\u0E44\\\\u0E17\\\\u0E22Thai\\"}),`\\n`,(n\\"tr\\"}),\\":T\\\\xFCrk\\\\xE7eTurkish\\"}),`\\n`,(n.li,{()(n.

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/1.txt

@@ -0,0 +1 @@
+****:한국어TürkçeTurkish\\n-**uk**:Українсь

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/10.txt

@@ -0,0 +1 @@
+(Belgium)\\"}),`\\n`,(n\\"fr-ca\\"}),\\":Fran\\\\xE7ais(Canada)

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/11.txt

@@ -0,0 +1 @@
+French(Canada)\\"}),`\\n`,(n\\"fr-ch\\"}),\\":Fran\\\\xE7ais(Suisse)

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/12.txt

@@ -0,0 +1 @@
+French(Switzerland)\\"}),`\\n`,(n\\"fr-fr\\"}),\\":Fran\\\\xE7ais(France)French

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/13.txt

@@ -0,0 +1 @@
+(France)\\"}),`\\n`,(n\\"hu\\"}),\\":MagyarHungarian\\"}),`\\n`,(n\\"id\\"}),\\":BahasaIndonesia

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/14.txt

@@ -0,0 +1 @@
+Indonesian\\"}),`\\n`,(n\\"it\\"}),\\":Italiano

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/15.txt

@@ -0,0 +1 @@
+Italian\\"}),`\\n`,(n\\"ja\\"}),\\":\\\\u65E5\\\\u672C\\\\u8A9E

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/16.txt

@@ -0,0 +1 @@
+\\\\uD55C\\\\uAD6D\\\\uC5B4

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/17.txt

@@ -0,0 +1 @@
+Thai\\"}),`\\n`,(n\\"tr\\"}),\\":T\\\\xFCrk\\\\xE7eTurkish\\"}),`\\n`,(n.li,{()(n.

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/18.txt

@@ -0,0 +1 @@
+Korean\\"}),`\\n`,(n\\"nl\\"}),\\":Nederlands

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/19.txt

@@ -0,0 +1 @@
+Dutch\\"}),`\\n`,(n\\"nl-be\\"}),\\":Nederlands(Belgi\\\\xEB)Dutch(Belgium)\\"}),`\\n`,(n\\"no\\"}),\\":Norsk

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/2.txt

@@ -0,0 +1 @@
+каUkrainian\\n-**vi**:TiếngViệtVietna

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/20.txt

@@ -0,0 +1 @@
+Norwegian\\"}),`\\n`,(n\\"pl\\"}),\\":(n\\"th\\"}),\\":\\\\u0E44\\\\u0E17\\\\u0E22

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/3.txt

@@ -0,0 +1 @@
+mese\\n-**zh-cn**:简体中文Simplified

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/4.txt

@@ -0,0 +1 @@
+Chinese\\n-**zh-tw**:繁體中文Traditional

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/5.txt

@@ -0,0 +1 @@
+Chinese\\n\\n\\n\\n##Updatetranslatedstrings\\n\\nIfyou

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/6.txt

@@ -0,0 +1 @@
+want`fr`locales|`Espa\\\\xF1ol(Lati

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/7.txt

@@ -0,0 +1 @@
+noam\\\\xE9rica)Spanish(Latin

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/8.txt

@@ -0,0 +1 @@
+America)\\"}),`\\n`,(n\\"fi\\"}),\\":Suomi

crates/swc_ecma_codegen/tests/str-lits/pr-7985/splitted/9.txt

@@ -0,0 +1 @@
+Finnish\\"}),`\\n`,(n\\"fr\\"}),\\":Fran\\\\xE7aisFrench\\"}),`\\n`,(n\\"fr-be\\"}),\\":Fran\\\\xE7ais(Belgique)French

crates/swc_ecma_minifier/scripts/next/add-test.sh

@@ -16,5 +16,5 @@ rm -rf .next
 # This is not useful data, as this is a script used only by me (@kdy1)
 npx next telemetry disable
 
-npx next build | grep '{ name:' | node "$SCRIPT_DIR/evaluate.js"
+npm run build | node "$SCRIPT_DIR/evaluate.js"
 touch "$SCRIPT_DIR/../../tests/compress.rs"