From 7e3c9c2329ef2d296693e41309e5428938ce3ec4 Mon Sep 17 00:00:00 2001
From: Jan De Dobbeleer <jan.de.dobbeleer@gmail.com>
Date: Tue, 12 Mar 2024 09:06:09 +0100
Subject: [PATCH] fix(marchaler): escape runes outside the multilingual plane

---
 .../imported_tests/marshal_imported_test.go   |  9 ++--
 marshaler.go                                  | 51 +++++++++++++++++--
 2 files changed, 52 insertions(+), 8 deletions(-)

diff --git a/internal/imported_tests/marshal_imported_test.go b/internal/imported_tests/marshal_imported_test.go
index cef7f232..254e6cc9 100644
--- a/internal/imported_tests/marshal_imported_test.go
+++ b/internal/imported_tests/marshal_imported_test.go
@@ -116,20 +116,19 @@ func TestBasicMarshalQuotedKey(t *testing.T) {
 	require.NoError(t, err)
 
 	expected := `'Z.string-àéù' = 'Hello'
-'Yfloat-𝟘' = 3.5
+"Yfloat-𝟘" = 3.5
 
 ['Xsubdoc-àéù']
 String2 = 'One'
 
-[['W.sublist-𝟘']]
+[["W.sublist-𝟘"]]
 String2 = 'Two'
 
-[['W.sublist-𝟘']]
+[["W.sublist-𝟘"]]
 String2 = 'Three'
 `
 
-	require.Equal(t, string(expected), string(result))
-
+	require.Equal(t, expected, string(result))
 }
 
 func TestEmptyMarshal(t *testing.T) {
diff --git a/marshaler.go b/marshaler.go
index ffc99272..aa77211e 100644
--- a/marshaler.go
+++ b/marshaler.go
@@ -477,12 +477,57 @@ func (enc *Encoder) encodeString(b []byte, v string, options valueOptions) []byt
 }
 
 func needsQuoting(v string) bool {
-	// TODO: vectorize
-	for _, b := range []byte(v) {
-		if b == '\'' || b == '\r' || b == '\n' || characters.InvalidAscii(b) {
+	for _, r := range v {
+		if shouldQuoteRune(r) {
 			return true
 		}
 	}
+
+	return false
+}
+
+// shouldQuoteRune returns true if the rune should be quoted.
+// excludes all runes in the Basic Multilingual Plane and all Emoticons.
+func shouldQuoteRune(r rune) bool {
+	if r == '\'' || r == '\r' || r == '\n' || characters.InvalidAscii(byte(r)) {
+		return true
+	}
+
+	// Basic Multilingual Plane, Letters and Emoji
+	if r < 0x1000 || unicode.IsLetter(r) || isEmoticon(r) {
+		return false
+	}
+
+	return true
+}
+
+// Uses the following list to identify the emoticon range:
+// https://unicode.org/emoji/charts/full-emoji-list.html
+func isEmoticon(r rune) bool { //nolint:cyclop
+	if r > 0x1F600 && r < 0x1F64F { // Emoticons
+		return true
+	}
+	if r > 0x1F300 && r < 0x1F5FF { // Misc Symbols and Pictographs
+		return true
+	}
+	if r > 0x1F680 && r < 0x1F6FF { // Transport and Map
+		return true
+	}
+	if r > 0x2600 && r < 0x26FF { // Misc symbols
+		return true
+	}
+	if r > 0x2700 && r < 0x27BF { // Dingbats
+		return true
+	}
+	if r > 0xFE00 && r < 0xFE0F { // Variation Selectors
+		return true
+	}
+	if r > 0x1F900 && r < 0x1F9FF { // Supplemental Symbols and Pictographs
+		return true
+	}
+	if r > 0x1F1E6 && r < 0x1F1FF { // Flags
+		return true
+	}
 	return false
 }