fix(codegen): fix unicode escaping in es5

crates/swc/tests/tsc-references/templateStringWhitespaceEscapes2_ES6_es5.1.normal.js

@@ -1,3 +1,3 @@
 //@target: es6
 // <TAB>, <VT>, <FF>, <SP>, <NBSP>, <BOM>
-"\t\v\f \xa0";
+"\t\v\f \xa0\uFEFF";

crates/swc/tests/tsc-references/templateStringWhitespaceEscapes2_es5.1.normal.js

@@ -1,2 +1,2 @@
 // <TAB>, <VT>, <FF>, <SP>, <NBSP>, <BOM>
-"\t\v\f \xa0";
+"\t\v\f \xa0\uFEFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings06_ES5_es5.1.normal.js

@@ -1,4 +1,4 @@
 // @target: es5
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  1. Assert: 0 ≤ cp ≤ 0x10FFFF.
-var x = "􏿿";
+var x = "\uDBFF\uDFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings06_ES6_es5.1.normal.js

@@ -1,4 +1,4 @@
 // @target: es6
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  1. Assert: 0 ≤ cp ≤ 0x10FFFF.
-var x = "􏿿";
+var x = "\uDBFF\uDFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings08_ES5_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (FFFF == 65535)
-var x = "￿";
+var x = "\uFFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings08_ES6_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (FFFF == 65535)
-var x = "￿";
+var x = "\uFFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings09_ES5_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (10000 == 65536)
-var x = "𐀀";
+var x = "\uD800\uDC00";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings09_ES6_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (10000 == 65536)
-var x = "𐀀";
+var x = "\uD800\uDC00";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings13_ES5_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es5
-var x = "󝷝";
+var x = "\uDB37\uDDDD";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings13_ES6_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es6
-var x = "󝷝";
+var x = "\uDB37\uDDDD";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings15_ES5_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es5
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings15_ES6_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es6
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings16_ES5_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es5
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInStrings16_ES6_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es6
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates06_ES5_es5.1.normal.js

@@ -1,4 +1,4 @@
 // @target: es5
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  1. Assert: 0 ≤ cp ≤ 0x10FFFF.
-var x = "􏿿";
+var x = "\uDBFF\uDFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates06_ES6_es5.1.normal.js

@@ -1,4 +1,4 @@
 // @target: es6
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  1. Assert: 0 ≤ cp ≤ 0x10FFFF.
-var x = "􏿿";
+var x = "\uDBFF\uDFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates08_ES5_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (FFFF == 65535)
-var x = "￿";
+var x = "\uFFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates08_ES6_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (FFFF == 65535)
-var x = "￿";
+var x = "\uFFFF";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates09_ES5_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (10000 == 65536)
-var x = "𐀀";
+var x = "\uD800\uDC00";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates09_ES6_es5.1.normal.js

@@ -2,4 +2,4 @@
 // ES6 Spec - 10.1.1 Static Semantics: UTF16Encoding (cp)
 //  2. If cp ≤ 65535, return cp.
 // (10000 == 65536)
-var x = "𐀀";
+var x = "\uD800\uDC00";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates13_ES5_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es5
-var x = "󝷝";
+var x = "\uDB37\uDDDD";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates13_ES6_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es6
-var x = "󝷝";
+var x = "\uDB37\uDDDD";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates15_ES5_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es5
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates15_ES6_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es6
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates16_ES5_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es5
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc/tests/tsc-references/unicodeExtendedEscapesInTemplates16_ES6_es5.1.normal.js

@@ -1,2 +1,2 @@
 // @target: es6
-var x = "ꯍ㑖碐";
+var x = "\uABCD\uEF12\u3456\u7890";

crates/swc_ecma_codegen/src/lib.rs

@@ -507,7 +507,8 @@ where
             }
             StrKind::Synthesized => {
                 let single_quote = false;
-                let value = escape_without_source(&node.value, self.wr.target(), single_quote);
+                let value =
+                    escape_without_source(&node.value, self.wr.target(), single_quote, false);
 
                 (single_quote, value)
             }
@@ -3070,7 +3071,12 @@ fn unescape_tpl_lit(s: &str, is_synthesized: bool) -> String {
     result
 }
 
-fn escape_without_source(v: &str, target: EsVersion, single_quote: bool) -> String {
+fn escape_without_source(
+    v: &str,
+    target: EsVersion,
+    single_quote: bool,
+    emit_non_ascii_as_unicode: bool,
+) -> String {
     let mut buf = String::with_capacity(v.len());
     let mut iter = v.chars().peekable();
 
@@ -3092,7 +3098,6 @@ fn escape_without_source(v: &str, target: EsVersion, single_quote: bool) -> Stri
                     buf.push_str("\\\\")
                 }
             }
-
             '\'' if single_quote => buf.push_str("\\'"),
             '"' if !single_quote => buf.push_str("\\\""),
 
@@ -3110,16 +3115,33 @@ fn escape_without_source(v: &str, target: EsVersion, single_quote: bool) -> Stri
             '\u{7f}'..='\u{ff}' => {
                 let _ = write!(buf, "\\x{:x}", c as u8);
             }
-
             '\u{2028}' => {
                 buf.push_str("\\u2028");
             }
             '\u{2029}' => {
                 buf.push_str("\\u2029");
             }
-
             _ => {
-                buf.push(c);
+                if !emit_non_ascii_as_unicode || c.is_ascii() {
+                    buf.push(c);
+                } else if c > '\u{FFFF}' {
+                    // if we've got this far the char isn't reserved and if the callee has specified
+                    // we should output unicode for non-ascii chars then we have
+                    // to make sure we output unicode that is safe for the target
+                    // Es5 does not support code point escapes and so surrograte formula must be
+                    // used
+                    if target <= EsVersion::Es5 {
+                        // https://mathiasbynens.be/notes/javascript-encoding#surrogate-formulae
+                        let h = ((c as u32 - 0x10000) / 0x400) + 0xd800;
+                        let l = (c as u32 - 0x10000) % 0x400 + 0xdc00;
+
+                        let _ = write!(buf, "\\u{:04X}\\u{:04X}", h, l);
+                    } else {
+                        let _ = write!(buf, "\\u{{{:04X}}}", c as u32);
+                    }
+                } else {
+                    let _ = write!(buf, "\\u{:04X}", c as u16);
+                }
             }
         }
     }
@@ -3134,25 +3156,31 @@ fn escape_with_source(
     s: &str,
     single_quote: Option<bool>,
 ) -> String {
-    if target <= EsVersion::Es5 {
-        return escape_without_source(s, target, single_quote.unwrap_or(false));
-    }
-
     if span.is_dummy() {
-        return escape_without_source(s, target, single_quote.unwrap_or(false));
+        return escape_without_source(s, target, single_quote.unwrap_or(false), false);
     }
 
-    //
     let orig = cm.span_to_snippet(span);
     let orig = match orig {
         Ok(orig) => orig,
         Err(v) => {
-            return escape_without_source(s, target, single_quote.unwrap_or(false));
+            return escape_without_source(s, target, single_quote.unwrap_or(false), false);
         }
     };
 
+    if target <= EsVersion::Es5 {
+        let emit_non_ascii_as_unicode = orig.contains("\\u");
+
+        return escape_without_source(
+            s,
+            target,
+            single_quote.unwrap_or(false),
+            emit_non_ascii_as_unicode,
+        );
+    }
+
     if single_quote.is_some() && orig.len() <= 2 {
-        return escape_without_source(s, target, single_quote.unwrap_or(false));
+        return escape_without_source(s, target, single_quote.unwrap_or(false), false);
     }
 
     let mut orig = &*orig;
@@ -3162,7 +3190,7 @@ fn escape_with_source(
     {
         orig = &orig[1..orig.len() - 1];
     } else if single_quote.is_some() {
-        return escape_without_source(s, target, single_quote.unwrap_or(false));
+        return escape_without_source(s, target, single_quote.unwrap_or(false), false);
     }
 
     let mut buf = String::with_capacity(s.len());

crates/swc_ecma_codegen/src/tests.rs

@@ -554,7 +554,21 @@ CONTENT\r
 fn test_escape_without_source() {
     fn es2020(src: &str, expected: &str) {
         assert_eq!(
-            super::escape_without_source(src, EsVersion::Es2020, true),
+            super::escape_without_source(src, EsVersion::Es2020, true, false),
+            expected
+        )
+    }
+
+    fn es2020_nonascii(src: &str, expected: &str) {
+        assert_eq!(
+            super::escape_without_source(src, EsVersion::Es2020, true, true),
+            expected
+        )
+    }
+
+    fn es5(src: &str, expected: &str) {
+        assert_eq!(
+            super::escape_without_source(src, EsVersion::Es5, true, true),
             expected
         )
     }
@@ -576,6 +590,13 @@ fn test_escape_without_source() {
     es2020("\u{1000}", "\u{1000}");
     es2020("\u{ff}", "\\xff");
     es2020("\u{10ffff}", "\u{10ffff}");
+
+    es2020_nonascii("\u{FEFF}abc", "\\uFEFFabc");
+    es2020_nonascii("\u{10ffff}", "\\u{10FFFF}");
+
+    es5("\u{FEFF}abc", "\\uFEFFabc");
+    es5("\u{10ffff}", "\\uDBFF\\uDFFF");
+    es5("\u{FFFF}", "\\uFFFF");
 }
 
 #[test]
@@ -612,7 +633,7 @@ fn issue_1619_2() {
 #[test]
 fn issue_1619_3() {
     assert_eq!(
-        escape_without_source("\x00\x31", EsVersion::Es3, true),
+        escape_without_source("\x00\x31", EsVersion::Es3, true, false),
         "\\x001"
     );
 }
@@ -657,3 +678,22 @@ impl Write for Buf {
 fn issue_2213() {
     assert_min("a - -b * c", "a- -b*c")
 }
+
+#[test]
+fn issue3617() {
+    let from = r"// a string of all valid unicode whitespaces
+    module.exports = '\u0009\u000A\u000B\u000C\u000D\u0020\u00A0\u1680\u2000\u2001\u2002' +
+      '\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200A\u202F\u205F\u3000\u2028\u2029\uFEFF';";
+    let expected = r#"// a string of all valid unicode whitespaces
+module.exports = '\t\n\v\f\r \xa0\u1680\u2000\u2001\u2002' + '\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200A\u202F\u205F\u3000\u2028\u2029\uFEFF';"#;
+
+    let out = parse_then_emit(from, Default::default(), Syntax::default(), EsVersion::Es5);
+
+    dbg!(&out);
+    dbg!(&expected);
+
+    assert_eq!(
+        DebugUsingDisplay(out.trim()),
+        DebugUsingDisplay(expected.trim()),
+    );
+}